United States
          Environmental Protection
          Agency
                           Office of
                           Research and
                           Development
                                                            Office of Solid
                                                            Waste and
                                                            Emergency
                                                            Response
EPA/600/S-99/002
May 19, 1999
          Federal   Facilities  Forum   Issue
         Field  Sampling And  Selecting  On-site Analytical
         Methods For Explosives In Water
         A. B. Crockett1, H. D. Craig2, and T. F. Jenkins3

           The Federal Facilities Forum is a group of
         U.S. Environmental Protection Agency (EPA)
         scientists and  engineers who represent EPA
         regional  offices  and  are  committed  to the
         identification and resolution of issues affecting
         the characterization and remediation of federal
         facility Superfund, Resource Conservation and
         Recovery Act, and Base  Realignment and
         Closure sites.   Current forum members are
         identified at the end of this  paper. The forum
         members identified a need to provide remedial
         project managers and other federal,  state, and
         private personnel working on hazardous waste
         sites with a technical issue paper that identifies
         screening  procedures  for   characterizing
         groundwater and surface water contaminated
         with explosive and propellant compounds. Some
         Forum members  provided technical guidance
         and direction in the development of this issue
         paper, and other members provided comments.

           This paper was prepared by A. B. Crockett,
         H. D. Craig, and T. F. Jenkins. Support for this
         project was  provided  by  the EPA National
         Exposure Research Laboratory, Environmental
         Sciences Division with the assistance of the
         Superfund Project's Technology Support Center
         for Monitoring and Site Characterization. For
         further information,   contact Ken  Brown,
         Technology Support Center Director, at (702)
         798-2270, Alan B. Crockett at (208)  526-1574,
         or Harry D. Craig at (503) 326-3689.
                                           It is imperative that any persons working
                                         on sites believed to be contaminated with
                                         explosive residues thoroughly familiarize
                                         themselves with the physical  and toxic
                                         properties  of  the  materials  potentially
                                         present and take all measures as may be
                                         prudent and/or prescribed by law to protect
                                         life, health, and property. This publication is
                                         not intended to include discussions of the safety
                                         issues associated with sites contaminated  with
                                         explosive residues. Examples of safety issues to
                                         be considered include but are not limited to
                                         geophysical  detection  methods,  explosion
                                         (detonation) hazards,  toxicity  of secondary
                                         explosives, and personal protective equipment.
                                         Information pertaining to toxicity concerns can
                                         be found in Roberts and Hartley  (1992)  and
                                         Yinon (1990).   Specifically,  this paper is not
                                         intended to serve as a guide  for sampling and
                                         analysis of unexploded ordnance (UXO),  bulk
                                         high explosives, or secondary explosives in soil
                                         where concentrations  exceed 100,000 mg/kg
                                         (10%).    These  conditions  present a
                                         potential  detonation hazard;  therefore,
                                         explosive safety procedures  and  safety
                                         precautions  should  be identified before
                                         initiating site characterization activities in
                                         such environments.  It also does not serve as
                                         a guide to installation of groundwater wells in
                                         areas in which such hazards exist.
-*
echnology
upport
reject

1 Idaho National Engineering and Environmental Laboratory, Lockheed Martin Idaho Technologies Company
2 U.S. Environmental Protection Agency, Region 10
3 U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory
          Technology Support Center for
          Monitoring and Site Characterization
          National Exposure Research Laboratory
          Environmental Sciences Division
          Las Vegas, NV 89193-3478
                              Technology Innovation Office
                              Office of Solid Waste and Emergency Response
                              U.S. EPA, Washington, B.C.
                              Walter W. Kovalick, Jr., Ph.D., Director
                                            Printed on Recycled Paper

-------
Purpose and Scope
  This issue paper provides guidance to remedial
project managers on field sampling and selecting
on-site  analytical methods  for detecting  and
quantifying secondary explosive compounds in
water (see Table  1).  A similar issue paper was
previously  prepared   on  explosives  in  soils
(Crockett et al. 1996), and updated as all.S. Army
Corps of Engineers Cold Regions Research and
Engineering   Laboratory   (CRREL)   report
(Crockett et al. 1998).  The paper also includes a
brief discussion of the reference analytical method
for  the  determination of  14  explosives  and
co-contaminants in water, soil, and sediments,
EPA Method 8330 (EPA 1998).
 Table 1.   Analytical Methods for Commonly Occurring Explosives, Propellants, and
           Impurities/Degradation Products.
Acronym
Compound Name
Nitroaromatics
TNT
TNB
DNB
2,4-DNT
2,6-DNT
Tetryl
2AmDNT
4AmDNT
NT
NB
Nitramines
RDX
HMX
NQ
2,4,6-trinitrotoluene
1,3,5-trinitrobenzene
1,3-dinitrobenzene
2,4-dinitrotoluene
2,6-dinitrotoluene
Methyl-2,4,6-trinitrophenylnitramine
2-amino-4,6-dinitrotoluene
4-amino-2,6-dinitrotoluene
Nitrotoluene (three isomers)
Nitrobenzene

Hexahydro-1 ,3,5-trinitro-1 ,3,5-triazine
Octahydro-1 ,3,5,7-tetranitro-1 ,3,5,7-tetrazocine
Nitroguanidine
Field Laboratory
Method Method
Cs
Cp, Ip, CFIp, FOBp
Cs, Is, CFIs
Cs
Cs
Cs, Is
Cs
Is

Cs
Cp, Ip, CFIp, FOBp
Cp
Cs
N
N
N
N
N
N
N
N
N
N
N
N
N
N
Q
 Nitrate Esters
 NC         Nitrocellulose
 NG         Nitroglycerin
 PETN      Pentaerythritol tetranitrate
 Ammonium Picrate/Picric Acid
 AP/PA     Ammonium 2,4,6-trinitrophenoxide/2,4,6-trinitrophenol
                        Cs
                        Cs
                        Cs
                        Cs

                      Cp, Is
 *L
G,*P
 *p
    A  =  Ammonium Picrate/Picric Acid (Thorne and Jenkins 1995a)
    C  =  Colorimetric field method(s)
  CFI  =  Continuous flow immunosensor
  FOB  =  Fiber-optic biosensor
    Q  =  Nitroguanidine (Walsh 1989)
     I  =  Immunoassay field method(s)
    L  =  Nitrocellulose (Walsh unpublished CRREL method)
    N  =  EPA SW-846, Nitroaromatics and Nitramines by HPLC, Method 8330 (EPA 1998)
    G  =  EPA SW-846, Nitroglycerin by HPLC, Method 8332 (EPA 1998)
    P  =  PETN and NG (Walsh unpublished CRREL method)
    p  =  Primary target analyte
    s  =  Secondary target analyte
 * The performance of a number of field methods has not been assessed using "approved" laboratory methods.  It is
   recommended that verification of the performance of any analytical method be an integral part of a sampling/analysis
   projects quality assurance program.

-------
  This issue paper is divided into the following
major sections:  (1) purpose and scope, (2) back-
ground, (3) objectives of water sampling and
monitoring water, (4) an overview of sampling
and analysis for explosives in water, (5) proce-
dures  for  statistically comparing  on-site and
reference analytical methods, (6) a summary of
on-site analytical methods for explosives in water,
and (7) a summary of the EPA reference method
for explosive compounds in water, Method 8330.
While some sections may be used independently,
joint  use  of the  field sampling  and  on-site
analytical methods sections is recommended to
develop a sampling and analytical approach that
achieves project objectives.

  Many of the explosives listed in Table 1 are not
specific target compounds of on-site methods, yet
they may  be detected by one  or more  on-site
methods because of their similar chemical struc-
ture.  The explosive  and propellant  compounds
targeted by high-performance liquid chromato-
graphy (HPLC) methods such  as EPA Method
8330 also are listed in the table.

Background

  Evaluating sites potentially contaminated with
explosives  is   necessary  to  carry   out  EPA,
U.S. Department of Defense, and U.S. Department
of Energy policies for site characterization and
remediation under  the  Superfund,  Resource
Conservation and Recovery  Act,  Installation
Restoration, Base Realignment and Closure, and
Formerly  Used  Defense  Site  environmental
programs.   Facilities that may be contaminated
with explosives include, for example, active and
former manufacturing  plants, ordnance  works,
Army ammunition plants, Naval ordnance plants,
Army  depots,  Naval ammunition  depots, and
Army  and  Naval  proving grounds,  burning
grounds,  artillery  impact ranges,  explosive
ordnance disposal sites, bombing ranges, firing
ranges, and ordnance test and evaluation facilities.

  Historical disposal practices  from manufac-
turing, spills, ordnance demilitarization, lagoon
disposal of explosives-contaminated wastewater,
and open burn and/or open detonation (OB/OD)
of explosive  sludge, waste  explosives,  excess
propellants,  and  unexploded  ordnance  often
resulted in soil and groundwater contamination.
Common munitions fillers and their associated
secondary explosives (indicated in parentheses
[see Table 1 for definitions of acronyms used in
the  following  paragraphs])  include  Amatol
(ammonium  nitrate/TNT),   Baratol  (barium
nitrate/TNT), Cyclonite  or  Hexogen  (RDX),
Cyclotols (RDX/TNT), Composition A-3 (RDX),
Composition B  (TNT/RDX), Composition C-4
(RDX), Explosive D or Yellow D  (AP/PA),
Octogen (HMX), Octols (HMX/TNT), Pentolite
(PETN/TNT), Picratol (AP/TNT), tritonal (TNT),
tetrytols (tetryl/TNT), and Torpex (RDX/TNT).

  Propellant  compounds include  DNTs  and
single-base (NC),  double-base  (NC/NG), and
triple-base (NC/NG/NQ) smokeless powders.  In
addition, NC is frequently  spiked with other
compounds  (e.g., TNT,  DNT, and  DNB)  to
increase its explosive properties. Explosive D or
Yellow D is used primarily in Naval  munitions
such as mines, depth charges, and medium to large
caliber projectiles.  Tetryl is used primarily as a
booster charge, and PETN is used in detonation
cord.

  Although on-site waste disposal of munitions-
related compounds was discontinued 20 to 50
years ago, a number of munitions facilities have
high levels of soil and groundwater  contamin-
ation. Under ambient environmental conditions,
explosives are highly persistent in groundwater
and  soil,  exhibiting  a resistance to naturally
occurring volatilization, biodegradation, and hy-
drolysis.  Talmage et al. (1999)  reviewed the
environmental fate of  several explosive com-
pounds  as discussed below.  Data indicate that
explosives  in weathered,  contaminated soils
exhibit slower degradation and desorption kinetics
than explosive residues  in spiked soil samples
(Grant et al.  1995).  Desorption of explosives
from  soil depends  on  environmental  factors
including  soil chemistry, contaminant concen-
tration, and the number of pore volumes leached
through the soil  (Pennington et al. 1995; EPA
1995).

-------
  Biological degradation products of TNT in
water,  soil, or  sediments  include  2AmDNT;
4AmDNT; and 2,6-diamino-4-nitrotoluene;  and
2,4-diamino-6-nitrotoluene. Photolysis of TNT in
water results in formation of 1,3,5-trinitrobenzene
(TNB) and several other compounds.  The com-
pound  TNB biologically degrades  into  3,5-
dinitroanaline, which has been recommended as
an  additional  analyte  for EPA  Method 8330
(Grant et al. 1993).  In surface waters, TNT is
degraded by photolysis and has a half-life of 0.5
to many hours. The biological half-life of TNT is
much longer,  ranging  from several  weeks to
6 months.   Spanggord et al. (1980) reviewed
studies on the  sorption of TNT by soils  and
sediments and reported the soil-water partitioning
coefficients (Kp) to range from 5.5 to 19.3 ([(ig
chemical in soil/g of soil]/[(ig chemical in water/g
of water]).  Recent data  show that irreversible
binding may be  a significant long-term  fate of
TNT that has sometimes not been considered in
older studies (Brannon and Myers 1997; Comfort
et al. 1995). Studies of compost residues (Thome
and Leggett 1997) and C-14 labeled TNT spiked
into soil (Comfort et al.  1995; Hundal et al. 1997)
show that, overtime, solvent-extractable TNT and
metabolic products decrease, but not all of the
original TNT  can be  accounted for.  As the
solvent-extractable TNT decreases, the concen-
tration of hydrolyzable TNT degradation products
increases. Acid hydrolysis is able to break some
chemical  bonds  between   TNT  degradation
products and humus or soils. However, overtime,
those bonds seem to become even stronger and
cannot be  chemically broken to recover TNT
degradation products (Hundal et al. 1997).

  Although the water solubility of RDX  is only
low to moderate, the compound is moderately to
highly mobile in the environment.  When released
to the environment, RDX can be expected to leach
to and persist in groundwater (Talmage et al.
1999).  In  surface water, RDX is degraded by
photolysis  to formaldehyde, nitrate and nitrite
ions, and nitroso compounds, for which the half-
lives range from hours to many days, depending
on the  environmental conditions.  As shown by
measured soil-water Kp ranging from 0.80 to 4.15
for sandy loam, clay loam, and organic clay, RDX
does not strongly partition to sediments.  In soils,
RDX is quite persistent and is biodegraded very
slowly aerobically, and about an order of magni-
tude faster anaerobically (Brannon  and Myers
1997).   The limited biodegradation  of RDX in
water has been accompanied with the identifi-
cation of hexahydro-l-nitroso-3,5-dinitro-l,3,5-
triazine (MNX), hexahydro-l,3-dinitroso-5-nitro-
1,3,5-triazine  (DNX),  and  hexahydro-1,3,5-
trinitroso-l,3,5-triazine (TNX), which are RDX
intermediates formed by sequential reductions of
the nitro groups to nitroso groups (McCormick et
al. 1981; Kitts et al. 1994; Sikora et al. 1997).
These mono-, di-, and trinitroso intermediates of
RDX are environmentally undesirable. Additional
products formed were hydrazine,  1,1-dimethyl -
hydrazine, 1,2-dimethylhydrazine, formaldehyde,
and methanol.

  The compound HMX has a low to moderate
affinity for soil and suspended  material, which
accounts for the  ready migration of HMX to
groundwater.  However,  the  low solubility of
HMX limits migration of HMX to groundwater.
The primary mechanism of removal of HMX from
surface water is through photolysis. The photol-
ysis half-life of HMX is from 2 to 17 days and
adsorption to suspended material and  biosorption
is not significant. While  aerobic and anaerobic
degradation of HMX to  1,1-dimethyl-hydrazine
has  been  demonstrated   in  enriched  media,
biodegradation  is  not expected to contribute
significantly to the loss of HMX under ambient
conditions.  While HMX  contamination is not
detected as commonly as TNT or RDX, military-
grade RDX contains approximately 10% HMX as
a manufacturing impurity (Army 1984).

  Tetryl is  primarily degraded by hydrolysis in
groundwater in which it is sometimes detected and
by photolysis in  surface water in which  it is
seldom detected. Photolysis is about an order of
magnitude faster than hydrolysis, and the latter
rate has been estimated at about 300 days at 20 °C
with a pH of 6.8.  The solubility of tetryl in water
is 75 ppm at 20 °C, which may impede leaching to
groundwater. The primary hydrolysis product,
picric acid, has a solubility of 11,000 mg/L and
may leach to groundwater.

-------
  The frequency of occurrence of specific explo-
sives in groundwater was assessed by Walsh et al.
(1993), who compiled analytical data on water
samples collected from 32 military installations.
Of the 812 samples analyzed by EPA  Method
8330 (EPA 1998), a total of 114 samples (14%)
contained detectable levels  of explosives.  The
frequency of occurrence  and  the  maximum
concentrations detected  are shown in Table 2.
The  most commonly occurring compound  in
contaminated samples, RDX, was detected in 61 %
of the  contaminated samples.   The compound
TNT was detected in 56%  of the contaminated
samples.   Overall, RDX or TNT or both were
detected  in   94% of the  samples containing
explosive residues. Thus, by analyzing for RDX
and TNT at the facilities sampled, 94% of the
contaminated samples could have been identified.
This demonstrates the feasibility of screening for
one or two compounds or classes of compounds to
identify the  extent of groundwater and surface
water contamination at munitions sites assuming
that the method detection limits are adequate. At
locations in which  RDX-contaminated wastewater
has been disposed of in lagoons or in which spills
have occurred, there is  a significant  likelihood
that the groundwater has  been contaminated with
explosives.

  The U.S. Army conducted a study from 1984 to
1985 to evaluate the impact of selected open burn
and/or open detonation facilities on groundwater
quality under various  site-specific conditions
(AEHA 1986). A total of 109 wells were sampled
at 17 individual facilities.  The facilities were
selected to represent a reasonably large cross-
section of OB/OD  sites with fairly  diverse
environmental settings.  Samples were analyzed
for TNT, 2,4-DNT, 2,6-DNT, RDX, HMX, and
tetryl.   The  results  (see Table 3) show that
explosives were detected in groundwater at 9 of
19 (47%) sites. The compound most frequently
detected was TNT, but RDX was  detected at
considerably higher levels. Detected in just two
wells, HMX was detected at high concentration,
and  tetryl  was  never  detected.    The  study
examined  factors  potentially contributing  to
groundwater  contamination  including  soil
permeability, depth to groundwater, temperature,
the level of surface soil contamination, the size
and age of the facility, and annual precipitation
and evaporation rates.  The conclusions were that
(1) in the  eastern half  of  the  country,  the
"predominant  factor  precluding  significant
contamination is low soil permeability" and (2) in
the West,  the  major  factor  that precludes
groundwater contamination is "apparently the
significant  excess  of  evaporation  over
precipitation". With the exception of the level of
surface  soil contamination,  the  other  factors
showed little or no association  with resultant
groundwater quality.  Recent  studies by  AEHA
(1994) and Jenkins et al.  (1997) also have
identified explosives groundwater contamination
at OB/OD and target impact areas resulting from
active firing range activities.

  Other recent studies have shown that explosives
in surface   water   may  migrate  considerable
distances from moderate to highly contaminated
disposal areas (LANL 1996; Murphy and Wade
1998). Elevated levels of explosives in surface
waters and sediments have been detected from 1.0
to 1.5  mi  downstream  from  source  areas.
Moderately to highly contaminated soils often
leach  explosives  into  groundwater,  and
contaminated groundwater may re-emerge into
surface water, particularly for nitramines such as
RDX and HMX.  Contaminated sediments also
may  serve  as  a source of recontamination to
surface water because of the low affinity of most
explosives to sediments.

-------
  Table 2.  Occurrence of Analytes Detected in Groundwater Contaminated
           with Explosives.
Compound
Samples with
Analyte Present
(%)
Maximum and
(Median Levels)
(M9/L)
Nitroaromatics
TNT
1,3,5-TNB
2-AmDNT
2,4-DNT
4-AmDNT
DNB
Tetryl
2,6-DNT
56
28
23
21
15
13
13
9
981
46
218
6.7
217
8.7
12
29
(3.5)
(1.5)
(11)
(1.2)
(4.6)
(0.78)
(0.92)
(0.10)
Nitramines
RDX
HMX
TNT and/or RDX
61
14
94
1400
673

(3.0)
(76)

  Derived from Walsh et al. (1993).
Table 3.  Occurrence and Concentration of Explosive Residues in Groundwater
         at Open Burning Open Detonation Sites.
Type
Explosive
TNT
RDX
2,4-DNT
2,6-DNT
HMX
Tetryl
Facilities
(%)
41
35
35
18
12
0
Wells
(%)
12
10
6
4
2
0
Maximum Geometric Mean
(ug/L) (ug/L)
306
1195
1788
651
583
NA
32
168
14
13
365
NA
Derived from AEHA (1986)
17 facilities, 109 wells total

-------
Objectives of Sampling and Monitoring
Water

Data Quality Objectives

  The EPA data quality objective (DQO) process
is   designed to  facilitate  the  planning  of
environmental  data  collection  activities  by
specifying the intended use of the data (i.e., the
decision that is to be made), the decision criteria
(i.e., the action level), and the tolerable error rates
(EPA 1994a; ASTM D 5792, "Standard Practice
for Generation of Environmental Data Related to
Waste Management Activities:  Development of
Data Quality Objectives"). Integrated use of on-
site  and  laboratory methods  for  explosives in
water facilitates achieving objectives such  as
determining the horizontal and vertical extent of
contamination, obtaining data to conduct a risk
assessment, identifying  candidate  waste   for
treatability studies and pumping tests, identifying
the amount of groundwater or surface water to be
remediated, monitoring and optimizing treatment
systems,  and  determining  whether  remedial
actions have met cleanup criteria.

Objectives in Sampling Water

  The frequency of occurrence and the coefficient
of variation of a  contaminant determine the
number of samples  required to adequately char-
acterize exposure pathways, and both are essential
in designing sampling plans. Low frequencies of
occurrence and high coefficients of variation, such
as with explosives, require more samples to char-
acterize the exposure pathways of interest. Sam-
pling variability typically contributes much more
to total error than analytical variability.  Under
these conditions, the major effort should be to re-
duce sampling variability by taking more samples
using less expensive methods (EPA 1992).

  Environmental data such as the rates of occur-
rence, average concentrations, and coefficients of
variation  are  typically highly  variable   for
contaminants associated with  explosive sites.
Solidatambienttemperatures, explosives dissolve
slowly and  sparingly in aqueous  solutions and
have low  vapor  pressures.   These  chemical
properties limit the modes of mobility compared
to other contaminants such as fuels or solvents.
The  differences between explosives and  most
other organic contaminants are a function of
contaminant fate and transport properties, occur-
rence in different media, interactions with  other
chemicals, and use and disposal practices. Areas
of high concentrations that serve as sources for
contamination of groundwater remain at or near
the surface  where  deposited,  unless  the  soils
themselves are moved (Jenkins et al. 1996a).

  The  EPA  guidance  for data usability in risk
assessment (EPA  1992) indicates that on-site
methods can produce  legally defensible data if
appropriate method quality control is available
and if documentation is adequate. Field analyses
can be used to decrease cost and analytical time as
long as supplemental data are available from an
analytical method capable of quantifying multiple
explosive  analytes  (e.g., Method  8330) (EPA
1992). Significant quality assurance oversight of
field analysis is  recommended to ensure data
usability.  The accuracy (i.e., the correctness of
the concentration value and a combination of both
systematic error [bias] and random error [preci-
sion]) of on-site measurements may not be as high
as in fixed  laboratories, but the  quicker  turn-
around and the possibility of analyzing a larger
number of samples more than compensates for
this potential lack in accuracy. Remedial project
managers, in  consultation  with chemists and
quality assurance personnel, should set accuracy
levels for each method and proficiency standards
for the on-site analyst.

Drinking Water Health Advisories and Water
Quality Criteria for Explosives

  In  1985, the EPA and the Department of the
Army  established a  Memorandum of Under-
standing (MOU) to develop EPA Drinking Water
Health  Advisories  for  Army  environmental
contaminants (Roberts and  Hartley 1992).   The
(MOU) memo  resulted in a  review  of the
toxicological database  for selected  munitions
chemicals and the development of recommended
exposure limits for specific durations (1 day, 10
days, longer term  [7 years], and  lifetime [70

-------
years]) (Roberts and Hartley 1992; Roberts et al.
1993; EPA 1996b). Both cancer and noncancer
toxicity endpoints were considered in the assess-
ment.  The EPA  Drinking Water Health Advi-
sories values for lifetime exposure to selected
explosives or 1E-04 lifetime excess cancer risk
levels for EPA Group B (probable human) carcin-
ogens are presented in Table 4.

  The  EPA has not established water quality
criteria for munitions compounds, but a series of
unpublished  reports by Oak  Ridge  National
Laboratory have been compiled (Talmage et al.
  1999) in which acute and chronic water quality
  criteria were calculated for TNT in accordance
  with EPA guidelines (Stephan et al. 1985). How-
  ever, the available data on other  explosive
  compounds  were not  sufficient to meet these
  guidelines so Tier II or secondary values were
  calculated in accordance with EPA guidance for
  the Great Lakes System (EPA  1993c).  These
  water quality criteria are summarized in Table 4
  along with sediment quality criteria (Talmage et
  al. 1999) normalized to organic carbon (milligram
  explosive/kilogram organic carbon).
  Table 4. Water Quality Criteria for Munitions-Related Chemicals
                        Drinking Water
Water and Sediment Quality Criteria/Screening
               Benchmarks3
Compound
TNT
RDX
HMX
1,3-DNB
1,3,5-TNB
2,4-DNT
2,6-DNT
NC
NG
NQ
Health Advisories
(ug/L)
2d
2d
400d
1d

5e
5e
Nontoxic
5d
700d
Acute"
(ug/L)
570
700
1880
110
30





Chronicb
(ug/L)
90
190
330
20
10





Sediment0
(mg/kgoc)
9.2
1.3
0.47
0.67
0.24















        a Talmage et al. (1999)
        b Calculated in accordance with EPA Tier I (TNT) or Tier II guidelines (other chemicals) (EPA 1993c)
        c Milligrams chemical/kg organic carbon in the sediment; calculated in accordance with EPA
         guidelines (Stephan et al. 1985)
        d Lifetime exposure (EPA 1996b)
        e Lifetime excess carcinogenic risk of 1E-04 (EPA 1996b).

-------
Advantages of On-Site Analytical Methods
Monitoring Remediation Measures
  On-site methods may be useful for analysis of
water treatment processes for explosives, such as
granular activated carbon (GAC) or chemical and
ultraviolet  (UV)  oxidation  treatment  systems
(EPA 1993a, 1996a; AEC 1997).  However, on-
site methods should be evaluated against labor-
atory methods on a site and matrix-specific basis
because of the possibility of matrix interferences.
Treatability studies may be used to evaluate the
potential  of different treatment technologies to
remove and degrade target and intermediate ex-
plosive compounds and to evaluate whether clean-
up levels can be achieved for site remediation.
Treatability study waste for explosives-contam-
inated  waters should be of higher than average
concentration to evaluate removal rates for target
and  intermediate  compounds.   The  potential
effects from compounds related to treatment pro-
cesses, such as TNB from chemical and UV oxi-
dation systems (AEC 1997) and MNX, DNX, and
TNX  from  biological  and  phytoremediation
systems (Sikora et al. 1997),  also should  be
evaluated.

  On-site analytical methods are a valuable, cost-
effective tool to assess the nature and extent of
contamination (EPA 1997b). Because costs per
sample are lower, more samples can be analyzed.
In addition,  the  availability of near-real-time
results permits redesign of the sampling scheme
while in the field.  On-site analysis also facilitates
more effective use of off-site laboratories using
more robust analytical methods.  Even if on-site
methods are only used to determine the presence
or absence of contamination and the contaminated
samples are sent off-site for laboratory analysis,
total analytical costs can be reduced considerably,
provided that the on-site  methods  have  low
enough detection limits to meet site DQOs. Be-
cause  on-site methods  provide near-real-time
feedback, they can be used to focus additional
sampling on areas of known contamination, thus
possibly  saving  additional  mobilization  and
sampling efforts.
  During site  remediation, such as Superfund
remedial actions,  data  may  be needed on  a
near-real-time  basis to  assess the progress  of
pump-and-treat remedial actions (EPA  1994b;
Craig et al. 1996). These treatment systems are
often estimated to operate for a period of 10 to 30
years.   On-site  methods  can be  used  during
remediation to monitor individual extraction wells
and combined influent explosives concentrations,
as well as  to  evaluate GAC  breakthrough and
determine when to replace the GAC bed. Final
attainment of groundwater or surface water clean-
up levels should be determined  by an approved
method, such as EPA Method 8330  (EPA 1998).

  Figure 1 shows the time  series extraction well
concentrations of TNT and RDX for a GAC treat-
ment system for a 110-acre groundwater plume at
the  U.S. Naval  Submarine  Base in Bangor,
Washington.  The influent for a typical  single-
stage fixed bed GAC system enters the top of the
carbon column, and the explosives are adsorbed as
the waste stream flows through the column. The
treated liquid stream (effluent) exits the bottom of
the column. Once the effluent no longer meets the
treatment criterion, the  spent carbon is reacti-
vated,  regenerated,  or replaced.   As the GAC
system  continues to operate,  the  mass-transfer
zone moves down the column. Figure 2 shows the
adsorption pattern and the corresponding effluent
breakthrough curve. The breakthrough curve is a
plot of the ratio of effluent concentration (Ce) to
influent concentration (C0) as a function of the
water volume treated per unit time.  When a pre-
determined concentration appears in the effluent
(CB), breakthrough has occurred. At this point,
the effluent quality no  longer meets treatment
objectives.  When the carbon becomes so satur-
ated with explosives that they can no longer be
adsorbed, the carbon is said to be spent (Ce = C0).
Alternate design  arrangements may allow indi-
vidual adsorbers  in multi-adsorber systems to be
operated beyond the breakpoint as far as complete
exhaustion. This condition of operation is defined
as the operating limit (Ce = CL).

-------
  On-site colorimetric methods for system moni-
toring and determination of breakthrough curves
are being used at a Superfund remedial action for
an explosives washout lagoon groundwater GAC
pump-and-treat system (ACOE  1998).  Figure 3
shows the RDX breakthrough curve for between
bed samples in a two-bed GAC system in series
for a 350-acre groundwater plume at the Umatilla
Chemical Depot in Hermiston, Oregon. Influent
concentrations into the system are 97 (ig/L of
TNT, 29 ng/L of TNB, 710 ng/L RDX, and 63
(ig/L of HMX. Final effluent concentrations also
are monitored  using the on-site colorimetric
methods.   The only compound detected in the
between bed samples and final effluent samples
was RDX. The GAC system exhibits preferential
adsorption  for explosives compounds (TNT  >
TNB > HMX > RDX) in the same waste stream.
Other explosives compounds are progressively
displaced in favor of TNT adsorption.  The pres-
ence of multiple explosives will reduce the carbon
bed life in relation to single compound isotherms,
particularly  for the  breakthrough  of RDX
(Vlahakis  1974;  Lee  and  Stenstrom  1996).
Typical  loading rates  achieve  1  to  4% total
explosives loading onto the lead GAC bed before
breakthrough occurs.

Overview of Sampling and Analysis
for Explosives in Water

Explosive Hazards During Well Installation

  The explosives safety procedures necessary for
geophysical detection, handling, and disposal of
UXO and geotechnical operations such as well
installation in areas that potentially contain high
levels of explosives in soil are beyond the scope
of this document.  These conditions  present a
potential detonation hazard; therefore, explosive
safety procedures and safety precautions should
be identified before initiating site characterization
activities in these environments (EPA 1993a). A
qualified explosives safety expert should be con-
sulted in preparing field sampling procedures for
operations under these conditions (ACOE 1996a,
1996b).
Water Sampling

  Except for the significant hazards of installing
wells and working in areas that may contain UXO,
bulk high explosives, or highly contaminated
surface soils such as explosives washout lagoons,
procedures for sampling groundwater and surface
water for explosive residues are similar to sam-
pling for other semi volatile organic compounds.
The EPA guidance on groundwater sampling can
be found in  Subsurface Characterization and
Monitoring Techniques,  Volume 1:  Solids and
Ground Water (EPA 1993b, Chapter 5).  The
American Society for Testing and  Materials
(ASTM) provides guidance for sampling surface
and   groundwater:   "Guide  for  Sampling
Groundwater Monitoring Wells," Standard D
4448; "Guide for Planning and Implementing a
Groundwater Monitoring Program," Standard D
5851; "Practice for Sampling Wastes from Pipes
and Other Discharge Points," Standard D 5013;
and "Practice for Sampling with a Dipper or Pond
Sampler," Standard D 5358.  Other  standard
procedures can be located on the ASTM World
Wide Web site, http://www.astm.org.

Well Screens and Casing Materials

  Parker and Ranney  (1993) and  Parker et al.
(1989) demonstrated that none of the  explosives
evaluated sorb to well casings.  There were no
significant   differences  among  polytetra-
fluoroethylene,  rigid  polyvinyl chloride,  and
stainless steel used as well casing  materials  for
RDX, TNT, HMX, TNB, DNB, NB,  2AmDNT,
DNT, or NTs.

Containers, Holding Times, and Preservation
Methods

  The EPA guidance (EPA 1998) on sampling
containers for semivolatile organic compounds
specifies 1-gal, two 0.5-gal, or four  1-L amber
glass containers with Teflon-lined lids.  These
containers  and  volumes  were  designed  for
laboratory procedures such as Method 8330  for
which significant sample concentration may be
required. Similar bottles, of adequate volume for
the method, should be  satisfactory for on-site
analytical methods.
                                            10

-------
       TNT Concentration Changes Over Time

        T              T              T
     Wells Located <350 Feet Downgjadfent of Former Lagoon
              ^••^.4EOI^-^-4I.4E0lf^-^-4E4Ea^-a-4E0lf»^-^4Ea^--^-COI
                              Hate
RDX Concentration Changes Over Time
              Near-Field Extraction Wdls
                      Date
   Wells Located 350-700 Feet Down gradient of Farmer Lagoon
       MO-
 I
 s
   t»
  C !
 CJ"
                              Date
                                                                 14X
              Far-Field Extraction Wdls
                      Date
Figure 1. Extraction Well Concentrations of TNT and RDX for a GAC Treatment System for a 110-acre Groundwater Plume at the U.S.

         Naval Submarine Base in Bangor, Washington.

-------
                                       c,
                                       I
                                      ~
Cr
                                      T
                                     Ce 100
g 50
(
Granular Activated Charcoal System

:- 	 - -Jf~
: 	 ^^X 	
> 5 1D152U253D35404& 5O 55 «0 G& 7O 75 £0 85 90 9
Days




&
                              Figure 3.  RDX Breakthrough Curve
  The EPA-specified holding time for nitroaro-
matic compounds in water is 7 days until extrac-
tion, and extracts must  be analyzed  within the
following 40 days (EPA 1998).   The specified
sample preservation procedure is cooling to 4» C.
This criterion was based on professional judgment
rather than experimental data.   While recent,
scientifically based data have been generated on
improved preservation procedures for explosives
  in water (see below), the ramifications of using
  such procedures for legally defensible data should
  be considered during the DQO process. Deviation
  from  EPA procedures may require the user to
  justify such changes and might result in the data
  being deemed unfit for the intended use.

     Because of the short holding times between
  sample  collection  and analysis  using  on-site
                                             12

-------
analytical methods, sample preservation is typically
not a concern. However, if samples will be held
before analysis, sample preservation may need to
be considered. For split samples sent to an off-site
analytical laboratory, sample  holding  times  and
improved  preservation  methods   become  an
important consideration.

  Two recent studies, by Maskarinec et al. (1991)
and Grant et al. (1993), have shown that nitramines
are  stable  in water,  without  any  form  of
preservation for 30 and 50 days.  However, both
studies also demonstrated that nitroaromatics can
undergo  significant  degradation  within  days.
Maskarinec recommended a maximum holding time
of 4 days for DNT in groundwater while Grant
found DNT relatively stable and  recommended a
holding time of 30 days for surface water.  Grant's
work on TNB and TNT showed losses of 55% and
35%, respectively, in  7 days  for spiked surface
water samples stored under refrigeration.  Jenkins
et al. (1995a) showed that tetryl also can  degrade
rapidly in surface waters  with 73% being lost in
7 days despite refrigeration.   Degradation  was
much  faster  in  surface  water than it  was in
groundwater or reagent water.

  Because  nitroaromatics can degrade rapidly in
water  samples, Jenkins et al. (1995a) evaluated
possible sample preservation procedures.  Sample
acidification to a pH of 2 with sodium bisulfate was
demonstrated  to  retard  microbiological   and
chemical transformations, is  relatively  easy to
conduct in  the field, and does not  interfere with
solid phase extraction preconcentration procedures
in Method  3535A (EPA 1998).   Acidification of
spiked surface water samples eliminated losses of
TNT and TNB for 64 days and at least 28 days for
tetryl. Nitramines (RDX and HMX) were stable in
spiked, refrigerated surface water, with or without
preservation, for at least 64 days.  Small losses of
aminodinitro-toluenes  were  observed  for  both
acidified and unacidified samples,  and the loss rate
was initially higher for acidified samples. Acidified
samples must be neutralized if the samples will be
concentrated  using the salting-out  procedure in
Method 8330.
Procedures For Statistically Comparing
On-site   And  Reference  Analytical
Methods

  When on-site  methods are used, their  perfor-
mance needs to be evaluated, which is commonly
done by analyzing the splits of some water samples
by both the on-site method and a reference method
(commonly Method 8330). The performance of the
on-site method is then statistically compared to the
reference method using  a  variety of criteria,
depending   upon  the   objective   and   the
characteristics  of the data.   In most   cases,
measures of precision and bias are determined.
Precision refers to the agreement among a set of
replicate measurements and is commonly reported
as  the  relative  standard deviation   (standard
deviation divided by the mean and expressed as a
percent), the coefficient of  variation  (standard
deviation divided by the mean), or the relative
percent  difference  (sample  value minus  the
reference method value divided by the mean  and
expressed as a percent). Bias refers to systematic
deviation from the true value.

  The following discussion of statistical methods
applies to comparisons of analytical results based
on paired sample data (e.g., duplicate or split water
samples are analyzed by both an on-site method
and a reference method or water sample extracts
are analyzed by a reference and on-site method).

Precision and Bias Tests for Measurements of
Relatively   Homogenous   Material—When
multiple  splits of well-homogenized samples  are
analyzed using  different  analytical   methods,
statistical procedures summarized in Grubbs (1973),
Blackwood and Bradley (1991), and Christensen
and Blackwood (1993) can be used to compare the
precision and  bias of the  methods.    Grubbs
described a statistical  approach appropriate for
comparing the precision of two methods that takes
into  account  the high correlation between  the
measurements from each method.  An advantage
of Grubbs'  approach is that it provides unbiased
estimates of the precision of each method by
partitioning the variance of the measurement results
                                             13

-------
into its component parts (e.g., variance caused by
subsampling  and by the  analytical  method).
Blackwood and Bradley (1991) extended Grubbs'
approach to a simultaneous test for equal precision
and bias of two methods. Similar tests are provided
in  Christensen   and  Blackwood  (1993)   for
evaluating more than two methods.

  For comparisons involving bias alone, t-tests or
analysis of variance may be performed.  For com-
paring two methods, paired t-tests are appropriate
for assessing relative bias in normally distributed
data (otherwise data are transformed to achieve
normality or  nonparametric tests are used).  A
paired t-test can be used to test whether the con-
centration as  determined by an on-site method is
significantly  different from Method 8330 or any
other reference method. For comparing multiple
methods, a randomized complete block analysis of
variance can be used in which the methods are the
treatments and each  set of split samples consti-
tutes a block.

  These  tests are best applied when the concen-
trations of explosives are all of approximately the
same magnitude.  As the variability in the sample
concentration increases, the capability of these tests
for detecting differences  in  precision or  bias
decreases. The variability in the true quantities in
the samples  is of concern, and high variability in
sample results caused by poor precision rather than
variability in the true concentration is handled well
by these  methods.

  Precision and  Bias Tests for Measurements
over  Large   Value   Ranges—When   the
concentrations of explosives cover a large range
of  values,  regression  methods  for  assessing
precision  and  accuracy   become  appropriate.
Regression analysis  is useful  because it  allows
characterization of nonconstant precision and bias
effects   and   predicts   intervals   for  new
measurements (e.g.,  the  results  of an  on-site
method can be used to predict the concentration if
the samples were analyzed by a reference method).

  In a regression  analysis, the less precise on-site
method is generally treated  as  the  dependent
variable and the more precise reference analytical
method (e.g.,  Method 8330) as the independent
variable.  A linear  relationship and a slope that
differs from a value  of 1.0 indicates  a constant
relative bias in the on-site method (i.e., the two
methods differ by a fixed percentage).  Similarly,
an intercept value significantly different from zero
indicates a  constant absolute bias (i.e., the two
methods differ by a fixed absolute quantity).  Of
course, both fixed and relative bias components
may be present.

  When uncertainty is associated with the concen-
tration of an explosive as measured by the refer-
ence  method, standard least  squares  regression
analysis can produce misleading results.  Standard
least squares regression incorporates the assump-
tion that the independent variable values are known
exactly as in standard reference material.  When
the reference method  results contain appre-ciable
error compared to the on-site method, regression
and variability  estimates are biased. Furthermore,
the interpretation of  R-squared and uncertainty
intervals are affected, which is known as an errors-
in-variables problem.

  Because of  the errors-in-variables problem, the
slope coefficient in the regression of the on-site
data on the reference data generally will be biased
low.  Hence, a standard regression test to deter-
mine  whether  the  slope is significantly different
from 1 can result in rejection of the null hypothesis
even when there is no  difference in the true bias of
the two methods.  A  similar argument applies to
tests of the intercept value being equal to zero.

  To perform  a proper errors-in-variables regres-
sion requires  consideration  of the measurement
errors in both the dependent and  independent
variables.   The  appropriate  methods for  per-
forming the regression (including  some guidance
about  how large the error in  the reference
analytical method can  be before a problem is
encountered) are outlined in Mandel (1984).  These
methods require estimating the ratio of the random
error variance  for the on-site method to that of the
reference analytical method.  With  split sample
data,  suitable  estimates  of the  variance  ratio
generally  can  be obtained  by  using variance
                                              14

-------
estimates from Grubbs' test or the related tests
mentioned above.

  If the variance ratio is  not  constant over the
range under study, more complicated models than
those analyzed in Mandel (1984) must be employed.
Alternatively, transformations of the data could
stabilize the variance ratio.  Note that it is the
variance ratio, not the individual variances, that
must remain constant.  For example, the  ratio of
variances  for two  methods  with nonconstant
absolute variances but constant relative variances
will still have a constant variance ratio.

  It should be noted that performing regressions on
data sets in which samples with concentrations
below the detection limit (for one or both methods)
have been  eliminated also may  result in biased
regression estimates, regardless of the regression
analysis method that is used.

  Comparison to Regulatory Thresholds and
Action Limits—When the purpose of sampling is
to make a  decision based on the comparison of
results to a specific value such as an action level
for  cleanup,  on-site  and reference  analytical
method results may be compared simply on the
basis of the degree of agreement between the two
methods. The appropriate statistical tests are based
on  the binomial  distribution and include  tests of
equality  of  proportions  and  chi-square  tests
comparing the sensitivity and specificity (i.e., false
positive and false negative rates) of the on-site
method relative to the reference analytical method.
Note that any measure of consistency between the
two methods is  affected by how close the true
values in the samples are to the action level. The
closer the true values are to the action level, the
less the two methods will agree, even if they are of
equal accuracy. For example, if the action level is
2 • g/L and most samples have levels of above 100
• g/L, the agreement between the on-site  method
and reference should be very good.  If, however,
the concentration in most samples is 0.5 to 10 • g/L,
the two methods will be much more likely to
disagree.   This  must be  kept  in mind  when
interpreting results, especially when  comparing
across different  studies for which samples may
have  been  collected at  considerably  different
analyte levels.
Summary of On-site Analytical Methods
for Explosives in Water

  There is significant interest in field methods for
rapidly and economically determining the presence
and  concentration  of  secondary explosives in
groundwater and  the  influent  and effluent to
groundwater  remediation  facilities.      Such
procedures  allow  much  greater  flexibility in
mapping the extent of  contamination,  designing
pumping strategies based on near-real-time data,
accruing more detailed characterization for a fixed
cost,  and guiding continuous  remedial  actions.
Ideally,  on-site analytical methods would  provide
high-quality data on a near-real-time basis at low
cost and of sufficient quality to meet all intended
uses including risk assessments and final remedial
action objectives  without  the  need  for more
rigorous procedures. While the currently  available
on-site methods are not ideal  (i.e.,  they  are not
capable   of   providing   compound-specific
concentrations  of  multiple compounds  simul-
taneously as might be desired in risk assessment),
they have proved to be  very valuable during the
characterization  and remediation monitoring of
some sites.  Currently available field methods that
have  been  evaluated against standard  analytical
methods  and demonstrated  in  the field  include
colorimetric, immunoassay, and biosensor methods
(see  Table  5).    Each  method  has  relative
advantages and disadvantages, so that no method is
optimal for all  applications.   To assist in  the
selection of one or more screening methods for
various  users needs, Table 6 (modified and ex-
panded from EPA 1997a) provides information on
on-site test  methods for detecting explosives in
water. The selection criteria are discussed in the
following sections.

  The three types of on-site methods, colorimetric,
immunoassay, and biosensor, are fundamen-tally
quite different. The CRREL  colorimetric methods
were developed by Jenkins for TNT (Jenkins 1990)
                                             15

-------
and RDX (Walsh and Jenkins 1991) in soils. Later
Jenkins et al. (1994a, 1995b) developed a solid-
phase extraction method for TNT and RDX in
water  in  which the  extraction disks could  be
extracted with acetone and analyzed by the  soil
analytical procedures.  The same methods are now
used in the Strategic Diagnostics,  Inc., EnSys
procedure for extraction of TNT and RDX + HMX
from groundwater, and the EnSys TNT and RDX
soil test kits are used to complete the analysis (see
below). (Note that the EnSys procedure  refers to
the RDX + HMX kit  as simply the RDX kit while
the draft EPA Method 8510 refers to it as RDX +
HMX.  The latter designation is used throughout
this document.) The commercial versions of the
methods are the most commonly used. Therefore,
the tables and the  text refer only to the EnSys
procedures for TNT and RDX + HMX but the
CRREL methods can provide equivalent results.

  Researchers  at  CRREL  also  developed  a
colorimetric  analytical procedure for quantifying
ammonium picrate and picric acid in soil and water
(Thorne and  Jenkins 1995a,  1995b).    In  the
procedure, 2 L of water are drawn  through an
anion extraction disk under vacuum and the disk is
washed to remove interferences. Picrate ions are
converted to picric acid and are eluted from the
disk, and  absorbance measurements  are  made
before  and after conversion to the yellow picrate
ion.

  In the EnSys colorimetric method for water, solid-
phase extraction is used to remove and concentrate
analytes from water.   A 2-L water  sample is
passed through a  stack  of two membranes to
preconcentrate  TNT on the top membrane  and
RDX on the bottom membrane.  Acetone is used to
elute RDX + HMX from the bottom disk, and a
chemical reaction is induced that causes a color
change indicative of RDX in the solution.  The
RDX + HMX concentration is estimated from the
absorbance   at  510  nm   on   a   portable
spectrophotometer.  The  top disk is eluted with
acetone,  and a different chemical reaction is
induced causing a color change indicative of TNT.
The  TNT concentration  is  estimated from the
absorbance at 540 nm.
         Table 5.   Available On-Site Analytical Methods for Explosives in Water.
Analyte(s)
Nitroaromatics
1.




2.


3.
4.
TNT




TNB


DNT
Tetryl
Nitramines
1.


2.
3.
RDX


HMX
NQ
Nitrate Esters
1.
2.
3.
NC
NG
PETN
AP/PA
Type Test
Colorimetric
Colorimetric
Immunoassay

Biosensor

Colorimetric
Immunoassay
Biosensor
Colorimetric
Colorimetric
Colorimetric
Colorimetric
Immunoassay
Biosensor
Colorimetric
Colorimetric
Colorimetric
Colorimetric
Colorimetric
Colorimetric
Colorimetric
Developer/Test Kit
EnSys - TNT
EnSys - TNT
D TECH - TNT
RaPID Assay
Continuous Flow Immunosensor - TNT
Fiber-Optic Biosensor - TNT
CRREL, EnSys - TNT
RaPID Assay
Continuous Flow Immunosensor - TNT
EnSys - TNT
EnSys - TNT
EnSys - RDX + HMX
EnSys - RDX + HMX
D TECH - RDX
Continuous Flow Immunosensor - RDX
Fiber-Optic Biosensor - RDX
EnSys - RDX + HMX
EnSys - RDX + HMX
EnSys - RDX + HMX
EnSys - RDX + HMX
EnSys - RDX + HMX
EnSys - RDX + HMX
CRREL 1
            1  U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory
                                             16

-------
Table 6. Comparative Data for Selecting On-Site Analytical Methods for Explosives in Water.3
Method/Kit
CRREL
EnSys
(Commercial
version
CRREL, TNT
and RDX
methods)
DTECH
RaPID Assay
Continuous
Flow
Immunosensor
Fiber-Optic
Biosensor with
Fluidics Unit
Criteria
Method Type,
Analytes, and EPA
Method #
Colorimetric
Ammonium Picrate
/Picric Acid
Colorimetric
TNT
RDX + HMX
Draft Method 8510
Immunoassay - ELISA
TNT
RDX
Immunoassay - ELISA
Magnetic particle/tube kit
TNT
Immunosensor
TNT
RDX
Proposed Method 4655
Immunosensor
TNT
RDX
Proposed Method 4656
Detection Range and
Range Factor
AP/PA: 3.6 to 200 • g/L (56
X)
TNT: 1 to 30 • g/L (30 X)
RDX: 5 to 200 • g/L (40 X)
H MX: 15 to 300- g/L (20 X)
TNT & RDX:
5 to 45 • g/L (9 X)
with DETECHTOR
TNT: 0.07 to 5 • g/L (71 X)
TNT and RDX:
10to1,000-g/L(100X)
TNT: 10 to 150- g/L (15 X)
RDX: 10 to 100- g/L (10 X)
Type of
Results
Quantitative
Quantitative
Semiquantitative
(concentration
range)
Quantitative
Quantitative
Quantitative
Samples per
Batch
AP/PA: Single or
batched
Single
Four (single or
batch)
Batch up to
51 samples
Sequential
Single up to a
batch of four
Water
Sample
Size
2L
2L
1 mL
100|jL
150|jL
1.7mL
for four
fiber
analyses
Sample Preparation
and Extraction
Filtration if the sample
is cloudy, solid-phase
extraction using anion
extraction disk, eluted
with methanol and
sulfuric acid.
Solid-phase extraction
using two membranes
filters, elution of filters
with acetone.
None
Filter (0.2- m) if gross
particulates are
present
To 955 |jL sample,
add 25 |jL ethanol and
20 |jL of SOX buffer
To 1.7 mL of sample,
add 200 |jL buffer and
100 |jL acetone.
Analysis Time/
Production Rate
(one person)
20 minutes to hours to filter, faster
per sample if batched;
recommended only for low turbidity
waters. 20 mins./ sample to
analyze.
20 minutes to a few hours for
filtering, recommended only for low
turbidity waters.
TNT : 35 mins./10 samples
RDX: 50 mins./10 samples
40 minutes for eight samples for
TNT and RDX.
10 to 15 mins. for single sample
70 minutes for 51 samples
3 to 4 minutes per sample, plus 3 to
4 minutes for internal standard, plus
1 minute peak analysis.
Total time < 20 minutes for typical 2-
3 analyses/sample
TNT: 8 minutes per quadruplicate
sample or batch of four.
RDX: 16 minutes per quadruplicate
sample or batch of four.
Double times to run reference
analysis. Typically each sample is
analyzed 2 to 4 times.
' Expanded and modified from EPA 1997a

-------
     TableG. (Continued)
Method/Kit
CRREL
EnSys
DTECH
RaPID Assay
Continuous
Flow
Immunosensor
Fiber-Optic
Biosensor with
Fluidics unit
Criteria
Interferences and Cross-reactivities > 1% based on IC50
Relatively free of humic and nitroaromatic interferences.
TNT = TNT + TNB + DNB + DNTs + tetryl;
RDX + HMX = RDX + HMX + PETN + NQ + NC + NG
Humics interfere with TNT and RDX; nitrate and nitrite interfere
with RDX.
TNT interferes with RDX method only when both are present.
Cross-reactivity:
TNT: tetryl = 35%; TNB = 23%; 2AmDNT = 1 1 %; 2,4-DNT = 4%;
AP/PA unknown but -100% at lower limit of detection
RDX: HMX = 3%
Cross-reactivity:
TNB = 65%; 2,4-Dinitroaniline = 6%; tetryl = 5%; 2,4-DNT = 4%;
2AmDNT = 3%;
DNB = 2%
TNT Method: TNB = 600%, tetryl = 38%, 2-AmDNT = 21 %, 2,4-DNT=
20%, NB = 16%, 2-NT = 9%, HMX = 5%, 2,6-DNT = 4%
RDX: 1 ,2-dinitroglycerin = 18%, HMX = 5%, TNB = 4%, 2,4-DNT =
3%, 1,3-DNB = 3%
TNT Method: TNB = 9%. All other tested explosives <4%
RDX Method: no explosive related interferences, (<3%)
no nitrate/nitrite interference
Supplier Recommended QA/QC '
Blank and spiked water samples analyzed daily.
Method and water blanks and a control sample
daily, one duplicate/20 samples. Some positive
field results (1:10) should be confirmed.
Samples testing positive should be confirmed
using standard methods.
Duplicate standard curves; positive control
sample supplied. Positive results requiring action
may need confirmation by another method.
Internal standards used for quantification, blank
matrix sample for background subtraction.
Reference analysis every other sample, run
blank once per set of fiber probes.
Storage Conditions and
Shelf Life of Kit or Reagents
Store at room temperature.
Store at room temperature.
Shelf life:
TNT = 2 to 24 months at 27- C
RDX = 2 to 12 months at 27- C
Store at room temperature or
refrigerate; do not freeze or
exceed 37- C for prolonged
period. Shelf life: 9 months at
room temperature.
Refrigerate reagents 2 to 8- C.
Do not freeze.
Shelf life 3 to 12 months.
Store activated membranes
moist at 4- C and away from
light.
Shelf life ~ 1 month
Fiber probes: Shelf life > 1 year
when stored < 27- C
Skill Level
Medium high
Medium
Low
Med-high,
initial training
recommended
Medium
Medium
00
     a Expanded and modified from EPA 1997a
     1 Site specific DQOs should always be used to select appropriate QA/QC

-------
Table 6. (Continued)
Method or
Kit
CRREL
EnSys
DTECH
RaPID Assay
Continuous
Flow
Immunosensor
Fiber-Optic
Biosensor
with Fluidics
unit
Criteria
Training Availability
None
Applicable video on
CRREL soil method
available only,
address in text.
Training available.
Applicable video on
CRREL soil and
groundwater methods
are available,
addresses in text.
Training available
Training available
No formalized training
available at this time.
No formalized training
available at this time.
Costs
(not including labor)
$15/sample plus $1,500
for Hach spectrometer.
Vacuum filtration
apparatus needed.
$21/sampleforTNT,
$25/sample for RDX
plus$175/day or
$450/wk, $800/mo for
lab station. Lab station
cost = $1 ,950.
Vacuum filtration
apparatus needed.
$32.50/sample for TNT
or RDX plus $300 for
DTECHTOR (optional).
$13to$20/sample plus
$4,000 for equip.
(purchase), 175/day,
$450/wk or $800 for
first month, $400 each
additional month
(rental).
$50/coupon which lasts
for -20 to 30 samples
plus $21 ,000 for
instrument
(FAST 2000).
$3 to 5/sample plus
$18,000 for instrument
(Analyte 2000 from
Research International)
and $-8,000 for Fluidics
unit.
Comparisons to
Method 8330 References
Thorne and Jenkins 1995a
Craig etal. 1996; EPA
1997a; Jenkins and
Schumacher 1990; Jenkins
eta!1994b
Craig etal. 1996; EPA
1997a;
Teaney and Hudak 1994
Thorne and Myers 1997
Craig etal. 1996; EPA
1997a; Rubioetal. 1996
Craig etal. 1996; EPA
1997a; Bart etal. 1997 a,
1997b;ESTCP1998
Craig etal. 1996; EPA
1997a; Shriver-Lake et al.
1995, 1997; ESTCP 1998
Other
References
Thorne and
Jenkins
1995b
Jenkins et
al. 1995b
USAGE
1999
Calif. EPA
1996a,
1996b
Calif. EPA
1996c
Narang et al
1998,
Whelan et
al. 1993
Shriver-
Lake et al.
1998,
Golden et al.
1997
Developer Information
Tom Jenkins
CRREL
72 Lyme Road
Hanover, NH 03755-1290
(603) 646-4385
Strategic Diagnostics
1 1 1 Pencader Dr.
Newark, DE 19702-3322
(800) 544-8881
www.sdix.com
Strategic Diagnostics
1 1 1 Pencader Dr.
Newark, DE 19702-3322
(800) 544-8881
www.sdix.com
Strategic Diagnostics
1 1 1 Pencader Dr.
Newark, DE 19702-3322
(800) 544-8881
www.sdix.com
Anne Kusterbeck
Naval Research Lab.
4555 Overlook Ave. SW
Washington, D.C. 20375
(202) 404-6042
Lisa Shriver-Lake
Naval Research Lab.
4555 Overlook Ave. SW
Washington, D.C. 20375
(202) 404-6045
Additional Considerations
Large work area (two large desks);
requires the most setup; electricity required;
deionized, methanol, and sulfuric acid
required; must assemble materials;
glassware must be rinsed between
analyses; vacuum filtration apparatus
needed.
Large work area (desk size) power supply
required to charge Hach spectrometer;
possible TNB interference; color indication
of other compounds; requires acetone and
deionized water; cuvettes must be rinsed
between analyses. Nitrate and nitrite
interferences with RDX kit can be corrected
using alumin-a-cartridges. Vacuum filtration
apparatus needed.
Small working area; few setup
requirements; no electricity or refrigeration
required; temperature dependent
development time (effect can be reduced by
changing DTECHTOR setting); significant
amount of packing; relatively narrow range;
no check on test; easy to transport or carry;
kits can be customized. Out-of- range
reruns require use of another kit.
Large work area (desk); requires setup
time, electricity and refrigeration; less
temperature dependent; low detection limit;
all reagents supplied; reagents and kit need
refrigeration. Out-of-range reruns require
dilution and full reanalysis.
Desk size work area. Less packaging
waste, requires electricity and refrigeration.
Instrumentation available from: Research
International, 18706 142nd Ave. NE,
Woodinville, WA 98072, (206) 486-7831
Desk size work area, can be operated
without fuidics unit, requires electricity,
refrigeration recommended. Less packaging
waste. Quantification requires sample
dilution when percent inhibition is >60% for
TNT or > 80% for RDX
' Expanded and modified from EPA 1997a

-------
  The steps of the various immunoassay methods
differ considerably. The simplest of the methods
is the D TECH method.  In the D TECH kit,
antibodies specific for TNT  and closely related
compounds are linked to solid particles. The TNT
molecules in water samples are captured by the
solid particles and collected on the membrane of a
cup  assembly.   A color-developing solution is
added to the cup assembly, and the presence (or
absence) of TNT is determined by comparing the
solution in the assembly cup to a color card or by
using the simple field test meter.  The  color is
inversely proportional to the concentration of TNT.

  The continuous flow immunosensor (CFI) and
the  fiber-optic  biosensor (FOB) methods  were
developed by the Naval Research  Laboratory
(NRL).   The CFI method (Bart  et  al.  1997a,
1997b; EPA  1997a)  is  an antibody-based bio-
sensor capable of detecting low molecular weight
molecules in  aqueous solutions.  With  the  CFI
method, TNT or RDX antibodies are immobilized
onto a solid phase (beads, membrane or glass
capillary) saturated with a fluorescent-dye labeled
antigen. The solid phase is placed in a support and
an aqueous buffer solution is pumped through the
support to establish flow. Samples are prepared
in the buffer solution and injected upstream of the
support. When a sample containing TNT or RDX
is introduced,  the TNT  or RDX  binds to the
immobilized  antibody, displacing  some  of the
labeled antigen, which is subsequently detected by
a fluorometer. The concentration is proportional to
the fluorometer  signal. A portable version of the
CFI  (FAST  2000)  has been engineered  by
Researchlnternational, Inc. which incorporates the
solid phase, sample injection, pump, fluidics control,
and a fluorometer into a single instru-ment, with
associated  software  for data  acquisition  and
analysis using a lap top computer.

  The FOB method  (Shriver-Lake et al. 1995,
1997; EPA  1997a) is based on a competitive
immunoassay using  a  fluorescent dye  as the
reporter molecule. Fluorescent-dye labeled TNB
is used  as the competitor on the surface of an
optical probe. The labeled TNB is exposed to an
antibody-coated  optical  fiber  for  4  minutes,
generating a specific signal that corresponds to the
100% signal.  The reference signal is defined as
the signal change associated with the labeled TNB
alone.  Inhibition of this signal is observed when
TNT is present in a sample. The percent inhibition
observed is proportional to the TNT concentration
in the sample.  The reference signal is determined
both before and  after running  the  sample to
normalize  for the gradual  decrease in antibody
activity. A portable version of the FOB (Analyte
2000) has  been engineered by Research Interna-
tional, Inc.  Originally developed only for TNT, the
FOB  method  has been modified and is  now
available for RDX as well.

Method Type, Analytes, and  EPA Method
Number

  The first column of the criteria listed in Table 6
identifies the type of water screening method, the
analytes it detects,  and the EPA draft method
number.
  The CRREL colorimetric method for Explosive
D or Yellow D (AP/PA) has been formally docu-
mented (Thorne and Jenkins 1995 a) but it is not
under consideration for incorporation into SW-846
(EPA  1998),  nor is it being evaluated by  any
method certification organization.

  Commercially available colorimetric kits  mar-
keted under the  EnSys trade name  and manu-
factured   by  Strategic  Diagnostics,   Inc.,  are
available for determining nitroaromatics (TNT) and
nitramines (RDX  +  HMX) in soils.   The same
analytical methods can be used to analyze acetone
extracts of filter  disks  that  have  extracted
nitroaromatics and nitramines from 2 L of water
via vacuum filtration.  The water-extraction step
requires at least a small field laboratory.  There-
fore, Strategic Diagnostics has not promoted the
EnSys  TNT and RDX + HMX  procedures for
analysis of water although the company  will
provide procedures  upon request. The EnSys
TNT colorimetric method  detects nitroaromatics
(i.e., TNT, TNB, DNB, DNTs, and terry 1), and the
RDX + HMX method detects nitramines (RDX,
HMX, and NQ), and nitrate esters (NC, NG, and
PETN).   While  NC is detected by  the  actual
analytical method, it is not clear that acetone will
elute NC from the membrane filter disks with the
                                            20

-------
other explosives.   The EnSys RDX +  HMX
method is draft EPA Method 8510 and is designed
for soil and water.  The EPA Method 8515 is the
EnSys TNT method specific to soil (summarized in
soils  issue paper [Crockett et al.  1996, 1998]).
However, the water extraction step of the RDX +
HMX method (draft Method 8510) extracts TNT
on the first filter of a two-filter stack, and the first
filter can be extracted with acetone and analyzed
using the TNT soils method (Method 8515). The
EPA currently has no plans to revise Method 8515
to include analysis of water samples.

  Strategic Diagnostics, Inc., also manufacturers
commercial enzyme-linked immunosorbent assay
(ELISA) kits, including the D TECH and RaPID
Assay  kits  to detect TNT in water.  D TECH
immunoassay kits also are available  for  RDX.
Other explosives compounds  can sometimes be
detected using immunoassay kits because of their
cross-reactivity (see the Interferences and Cross-
Reactivity section).  The California Environmental
Protection Agency has certified the D TECH kits
for TNT (California  EPA  1996a)  and  RDX
(California EPA 1996b) for both water and soil.
The California EPA also has certified the RaPID
Assay  kit for TNT (California EPA 1996c) for
water and soil.

  The  Naval Research  Laboratory's two bio-
sensor-based methods  have   been  evaluated
although  they  are  not yet fully  commercially
available. The FAST 2000  and Analyte 2000
instruments are  commercially available but  the
coated membranes  and optical fibers are not.
However, the draft EPA methods describe how to
prepare membranes and fibers.  Both the CFI and
FOB methods are capable of determining TNT and
RDX.  A recent report on both methods (ESTCP
1998) has been submitted to EPA with the intent of
establishing new methods to be incorporated into
SW-846 (EPA 1998) and EPA has assigned draft
method numbers, Method 4655  for the CFI and
4656 for the FOB method. The draft methods are
written for both TNT and RDX in water only but
a soils method may  follow.

Detection Range and Range Factor

  The lower detection  limits of the on-site  meth-
ods for water range considerably, from 0.07 ug/L
to 15 ug/L.  The detection range of a test kit can
be important and a broad range is generally more
desirable.  The importance of the range of the test
kit  depends  on  the  range  of concentrations
expected in samples, the ability to estimate the
approximate concentration  from the sample ex-
tract, the amount of effort required to dilute and
rerun  a sample, and the sampling and analytical
objective.   Some test kits have a range  factor
(upper limit of range /lower limit) of just one order
of magnitude (10X), while other methods span two
orders   of  magnitude  (100X).   Because  the
concentration of explosives may  range widely,
reanalysis  of many out-of-range samples may be
necessary if  the  objective  is  to determine the
concentration of an  explosive  in  water.   The
D TECH  immunoassay  methods  require  an
additional  complete analysis  for each sample
dilution.  Other immunoassay methods can run
multiple dilutions in the same analytical run, but the
dilutions  must  be prepared  without knowing
whether they are needed. The EnSys colorimetric
procedure for RDX provides sufficient reagent to
allow running several dilutions at no additional cost.
For the EnSys kits, the analyzed sample can simply
be diluted and reread in the spectrophotometer.
Research results indicate that dilution ratios of as
high as 1 to 10,000 may be  necessary to keep
concentrations in the linear range of the tests, and
these  dilutions can be conveniently made in  one
step  using glass microliter  syringes (Jenkins
et al 1996b). The procedures that the test methods
use for sample dilution should be considered during
method selection.

  The detection range of a kit becomes much less
relevant when  the  objective is  to determine
whether a sample is above or below a single action
limit because the same dilution can be used for all
samples. In some cases, changing the range  of a
kit may be desirable to facilitate decision making.
Cleanup  levels for explosives  in  water vary
considerably depending on, for example, the site
conditions, compounds present and their relative
concentration, use of the water, results  of  risk
assessments, and the selected remedial technology
(EPA 1993a,  1996a; Craig et  al. 1996). Typical
remediation goals for water  are less than  one
microgram/liter for DNTs; low micrograms/liter for
RDX, TNT, TNB, DNB, and NG; and hundreds of
micrograms/liter for HMX and NQ (EPA 1996a).
                                            21

-------
Type of Results

  The type  of results provided by  the  various
screening methods are, depending on the concen-
tration range,   quantitative or semiquantitative.
The CRREL 2,4-DNT, EnSys, RaPID Assay, the
CFI, and the FOB are quantitative methods that
provide a numerical result.  The D TECH kits are
semiquantitative and indicate that the  concen-
tration level of an analyte is within one of several
ranges. For example, the D TECH RDX water kit,
without dilution, indicates a concentration within
one of the following ranges: less than 5,5 to 15,15
to 25, 25 to 45, 45 to 60, and greater than 60 ug/L.

Samples per Batch

  Several of the available test kits are designed to
run batches of samples, single samples, or both.
However, using a test kit designed for analyzing a
large batch to analyze one or two samples may not
be cost-effective or efficient. For methods requir-
ing filtration to concentrate the analyte, multiple
samples can be simultaneously extracted using a
filtration manifold.

Water Sample Size

  The test methods use samples of either 1 mL or
less of water, or 2 L of water.

Sample Preparation and Extraction

  Sample preparation and extraction only applies to
the colorimetric methods, which require filtration
and extraction of a 2-L water sample.

Analysis Time

  The filtration and extraction step associated with
both the CRREL and EnSys methods requires a
minimum of 20 minutes exclusive of extracting the
filter and may take well over an hour depending on
the amount of particulate matter in the sample.  For
this reason, these methods  are not recommended
for turbid waters.

  Actual sample analysis time for a single sample
ranges from 3 minutes to about  70  minutes
although  as  many  as   10  samples  can  be
batched and analyzed in the same 70 minutes. For
the NRL methods, eight or more subsamples from
each sample were analyzed and the results were
averaged for the recent method validation study
(ESTCP  1998).  However,  for routine use,  it is
expected that  two  to  four subsamples  would
typically be analyzed from each  sample.   The
effective production rate also depends  on the
number of reruns required for samples out of the
detection range.

Interferences and  Cross-Reactivity

  One  of the major differences among the field
methods is  interference for colorimetric methods
and cross-reactivity for the immunoassay and bio-
sensor  methods.  The colorimetric methods for
TNT and RDX are broadly class sensitive—that is,
they are not only able to detect the presence of the
target analyte  but also respond to many other
similar compounds (nitroaromatics, and nitra-mines
and nitrate  esters, respectively). For colorimetric
methods, interference is defined as the positive
response of the  method to  secondary  target
analytes or  co-contaminants similar to the primary
target analyte.   The immunoassay  and biosensor
methods are relatively  specific for the primary
target analytes that  they are designed to detect.
For the immunoassay and  biosensor meth-ods,
cross-reactivity is defined as the positive response
of the method to secondary target analytes or
co-contaminants similar to  the primary target
analyte.

  Depending on the sampling objectives, broad
sensitivity or specificity can  be an  advantage or a
disadvantage.  If the objective is to determine
whether explosive compounds are present,  then
broad sensitivity is an advantage. Another advan-
tage of the broad response of colorimetric meth-
ods is that they may  be used to detect compounds
other than the primary target analyte.  For exam-
ple, the colorimetric  RDX +HMX method may be
used to determine PETN when RDX + HMX
levels are relatively low or absent.  If a secondary
target analyte  is  present at  only low  concen-
trations in a sample, the effect on the analytical
result is minimal.  If the objective is to determine
the concentration of TNT when relatively high
levels  of  other nitroaromatics or RDX when
                                            22

-------
elevated levels of other nitramines or nitrate esters
are  present,  other  methods  may   be more
appropriate.

  Extremes of temperature and pH can interfere
with on-site analytical methods. According to the
California Military  Environmental  Coordination
Committee, physical  conditions comprising tem-
peratures outside the range of 4 to 32* C and pH
levels less than  3 or greater  than  11  (CMECC
1996) are generally not recommended for both
colorimetric and immunoassay methods.  Specific
product literature should be consulted for more
information.

  Colorimetric  Methods—For TNT methods,
the  primary  target  analyte  is  TNT  and  the
secondary target  analytes are other nitroaromatics
such as  TNB, DNB, 2,4-DNT, 2,6-DNT,  and
tetryl.   For  RDX methods,  the primary target
analyte is RDX + HMX, and the secondary target
analytes  include  PETN,  NG, and NQ.  If the
primary  target analyte is the only  compound
present,  the colorimetric methods  measure the
concentration of that compound.  If multiple
analytes  are  present, the field methods measure
the primary target  analyte plus the secondary
target analytes.   The response of colorimetric
methods to the secondary target analytes is similar
to the response of the primary target analyte and
remains  constant throughout the concentration
range of the methods although the observed colors
may be different.

  If multiple analytes are present in water, colori-
metric field results can be roughly compared with
EPA Method 8330 results (EPA 1998). For exam-
ple, if a  water sample  (as analyzed by Method
8330) contains 100 ug/L each of TNT, TNB, RDX,
and  HMX, the EnSys  colorimetric methods for
TNT would measure approximately 200 ug/L (100
TNT +  100  TNB), and the RDX test kit would
measure approximately 200 ug/L (100 RDX+ 100
HMX).   This example is somewhat  simplistic
because  each   compound  has  a somewhat
different response factor.

  While colorimetric kits are not compound speci-
fic, the color development of the extracts often can
provide an indication of the types of compounds
that may be present. For example, with the TNT
kit, TNT and TNB turn red; DNB turns purple;
2,4-DNT turns blue; and 2,6-DNT turns pink and
tetryl turns orange. For the EnSys RDX + HMX
kit, RDX and HMX turn pink as do nitroglycerine,
PETN,  and  nitrocellulose.    An orange color
indicates a mixture of TNT and nitramines  or
nitrate esters.

  Immunoassay Methods —For TNT immuno-
assay kits, the primary target analyte is TNT, and
the secondary target analytes are nitroaromatics
TNB, DNTs, AmDNTs, and tetryl.  For the RDX
kit, the primary target analyte is RDX, and there is
but little cross-reactivity with HMX (3%). If the
primary target  analyte  is  the only  compound
present in water, the immunoassay methods mea-
sure the concentration of that compound.

  If multiple analytes are present in water, the
immunoassay kits  measure the  primary target
analyte plus some percentage of the cross-reactive
secondary target analytes. The response of immu-
noassay kits to the secondary target analytes is not
equivalent to that of the primary target analyte and
does not remain constant throughout the concen-
tration range of the kits.  In addition, different
immunoassay kits have different cross-reactivities
to secondary target analytes based on the anti-
bodies used to develop each method.  Cross-
reactivities for  immunoassay  kits  are usually
reported at the 50% response level (IC50), typically
the midpoint of the concentration range of the kits.

  Table 7 shows the reported cross-reactivities at
IC50 for the immunoassay kits. A complete cross-
reactivity curve for the entire concentration range
should be obtained from the manufacturers for the
immunoassay kits being considered. Where mul-
tiple  analytes exist in water samples, immuno-
assay results may not directly compare with EPA
Method 8330 (EPA 1998) results. For example, an
immunoassay kit may have cross-reactivities of
23% for TNB for the TNT test kit and 3% HMX
cross-reactivity for the RDX test kit.  The follow-
ing simple example illustrates  cross-reactivity;
however, in practice, it is not practical to calculate
contaminant concentrations in this manner because
of synergistic  effects and cross-reactivity  is
nonlinear.    Using  the same  sample as  the
                                            23

-------
colorimetric example above, if a water sample (as
analyzed by Method 8330) contains 100 ug/L each
of TNT, TNB, RDX,  and  HMX, the TNT field
immunoassay kit would measure approximately
123 ug/L (100 TNT + 23 TNB), and the RDX field
method would measure approximately  103 ug/L
(100 RDX + 3 HMX).

  Biosensor Methods—For  the CFI method,
relative to TNT at 100%, the cross-reactivities of
other explosive compounds are TNB  600%;tetryl
38%; 2-AmDNT 21%; 2,4-DNT 20%; NB 16%;
2-NT 9%; HMX 5%; 2,6-DNT 4%; 4-AmDNT
1%; and RDX 1%. The RDX method is much
more compound specific with 18% cross-reac-
tivity for 1,2-dinitroglycerin, 5% for HMX, 4% for
TNB, and about 3% for DNB, 3-NT, and 2-NT.

  For the FOB TNT method, TNB is 9% cross-
reactive, and for the RDX method, all of the 17
explosive-related compounds tested are less than
3% cross-reactive.

  Matrix Interferences—Colorimetric, immun-
oassay, and biosensor methods may be subject to
positive or negative  matrix interferences  from
organic and inorganic substances in  water.  For
colorimetric methods, through careful visual anal-
ysis noted by a positive red or pink color change in
the sample before colorimetric  analysis,  these
interferences can be evaluated.  Inorganic nitrate
and  nitrite  in  water  samples  interfere  with
colorimetric methods unless special procedures are
used to remove these compounds during analysis.
It is important to note that high levels of humic
organics can impart a yellowish coloration to the
acetone extracts.  An increase in the intensity of
the yellow color upon reaction with the reagent is
not a positive response for the TNT test, and the
development of a reddish hue to the solution is
necessary before a detection is claimed. Analysis
of a field matrix blank may be useful in identifying
such interference.

  Many of the  immunoassay  methods  use  a
reverse-coloration process, and nontarget analyte
organic matrix interference results in less color
development. Therefore, on-site method results
are biased high compared to laboratory results.
  Both the CFI and FOB NRL methods have been
used at several different sites, but interferents
other  than  other explosives  have not  been
determined.

Supplier Recommended
Quality Assurance/Quality Control

  The manufacturers or developers recommended
quality  assurance/quality   control  (QA/QC)
procedures vary  considerably  with the on-site
method.   Some  test  methods do  not specify
QA/QC procedures and leave to the investigator
the  determination of  the numbers  of blanks,
duplicates, replicates, and standards that are run.
During field application of these methods, it  is
common to send at least 10 to 20% of the positive
samples to an off-site  laboratory for analysis by
EPA Method 8330.   A smaller fraction of the
nondetect samples also may be  verified.  In  some
cases, field methods are used to identify samples
containing explosive residues,  and all such samples
are sent to an off-site laboratory for analysis.   In
any case, the QC  samples recommended by the
method developer should be used. However, it is
up to the user to determine how much and what
types of QA/QC are needed to achieve the DQOs.

  While it is essential to ensure that field methods
perform as intended, requiring laboratory type QC
may  be  inappropriate for  on-site  analytical
methods.  Because  site characterization efforts
may be  cost constrained, excess  QC  samples
reduce the number of field samples that can be
analyzed.  Good sample handling procedures and
correlation of the field methods with the laboratory
HPLC method  over the concentration  range of
interest should  be the  primary  performance
criteria. Documentation of procedures and results
must be emphasized.

  During the initial  evaluation of on-site and
off-site analytical  methods, it may be desirable to
analyze a variety of QC samples  to determine
sources of error. The methods can then be modi-
fied to minimize error as efficiently as practical to
include, for example, the collection and analysis of
duplicates, replicates, splits of samples, and  splits
of extracts.
                                            24

-------
              Table 7.   On-Site Analytical Methods for Explosives in Water, Percent Interference, or Cross-Reactivity.
Is)
Test Method
Nitroaromatics
TNT
TNB
DNB
2,4-DNT
2,6-DNT
2AmDNT
4AmDNT
Tetryl
Nitramines
RDX
HMX
Other
PETN
TNT
EnSys a
DTEChT
RaPID Assay3
Flow Immunosensor
Fiber Optic Biosensor a
100
100
100
100
100
100
23
65
600
9
100
_b
2
-
<3
100
4
4
20
<3
100
-
<1
4
<3
NC
11
3
21
<3
NC
<1
1
1
<3
100
35
5
38
<4
NC
<1
<1
1
<1
NC
<1
<1
5
<4
-
-
-
-
-
RDX
EnSys a
DTECH3
Flow Immunosensor
Fiber Optic Biosensor a
NC
<1
2
<3
NC
<1
4
<3
NC
<1
3
<3
NC
<1
3
<3
NC
<1
1
<3
NC
<1
1
<3
NC
<1
2
<3
NC
<1
1
<3
100
100
100
100
100
3
5
<3
100
<1
-
-
  a  Interference for colorimetric methods, cross-reactivity for immunoassay methods at 50% response
  b  No data
NC  No Color Development
                                                                                                           (IC50)

-------
Storage Conditions and Shelf Life of Kit or
Reagents

  Storage conditions and the shelf life of irnmuno-
assay kits are more critical than with colorimetric
methods. The reagents for some immunoassay
kits should be refrigerated but not frozen or ex-
posed to high temperatures. Their shelf life can
vary from 3 months to more than 1 year.  Color-
imetric reagents can be stored at room temper-
ature. The EnSys colorimetric kits have shelf lives
of at least  2 months and up to 1 or 2 years.
Before ordering test kits, it is important to know
when they will be used to ensure that they will be
used before the expiration date.

  For immunoassay  kits, D TECH may be stored
at room  temperature while the RaPID Assay
reagents should be refrigerated. Neither kit should
be subjected to freezing.

  The CFI membranes need to be stored moist at
4» C, and away from light.  They have a shelf life
of about  1  month.  It is recommended that the
FOB be operated out of direct sunlight and recom-
mended but not required, that  stock solutions be
refrigerated. Stock solutions can be freeze dried
for storage up to a year  and rehydrated and held
for up to a  month unrefrigerated.  The antibody-
coated fibers may be preserved and stored for
more than a year  at room temperature if freeze
dried or be placed in a buffer solution at 4* C.
These procedures  may become simplified if  the
CFI and FOB methods become commercialized.

Skill Level

  The skill level necessary or required to run these
tests varies from low to high (Table 6), requiring a
few hours to a day of training.  The manufacturer
of the commercial kits generally provides  on-site
training. A free training videotape on the CRREL
version of the TNT and  RDX procedures  (which
also is useful for the EnSys colorimetric kits) is
available by submitting a written request to:

Commander U.S. Army Environmental  Center
Attn:  SFIM-AEC-ETT/Martin H. Stutz
Aberdeen Proving Ground
MD  21010
E-mail: mstutz@,aec2.apgea.army.mil.
A training  video  on the  USAGE  Standard
Operating Procedures  for Analysis of TNT and
RDX (USAGE 1999) will  be available from:

Kira Lynch
Seattle District Corps
PO Box 3755 (EN-TB-ET)
Seattle WA 98124-2255
E-mail: Kira.P.Lynch(g),usace.army.mil.

Training videos are also available for some kits.

Cost

  As shown in Table 6, routine sample costs vary
by method.  The cost  per sample is affected by
the  costs  of consumable  items,  analytical
instruments, and reusable apparatuses required to
run the method. In figuring the cost per sample, it
is  important to estimate  the costs of possible
reruns for out-of-range analyses. With the EnSys
colorimetric kits, or the CRREL AP/PA method,
the color-developed extracts may be simply diluted
and reread with the spectrophotometer. It should
also be noted that the  CRREL TNT and RDX +
HMX methods should become more economical
than the  EnSys kits as the number of samples
increases.  With the other methods, the original
water sample  must be diluted and reanalyzed,
which for immunoassay methods requires the use
of an additional kit. Colorimetric methods typically
have sufficient extra acetone for dilution to rerun
samples with no  increase in material  cost.  It
should be noted that the per-sample costs shown
in Table 6 are only  for supplies plus equipment.
Labor, data management, data review and data
reporting are not included.

  In contrast to the previous methods which have
relatively low initial costs  and higher per sample
costs, the two NRL biosensor-based methods,
have high initial capital costs and low per sample
costs. Eventually  there is a break-even  point at
which, with high numbers of samples, the NRL
methods  become  more economical than  colori-
metric and immunoassay methods.

Comparisons to Laboratory Method 8330

  The objectives of the study or investigation, the
site-specific contaminants  of concern, the concen-
tration ranges encountered or expected, and their
relative concentration ratios affect the selection of
                                            26

-------
a particular on-site method.  The accuracy of an
on-site method is another selection criterion but
care must be used in interpreting accuracy results
from comparisons between reference analytical
methods and on-site methods.

  Colorimetric methods actually measure classes
of compounds (i.e., nitroaromatics or nitramines)
and immunoassay methods are more compound
specific.  Therefore, the reported accuracy of a
method may  depend on the mix  of explosives
compounds present in the water sample and the
reference method data used for the comparison
(i.e.,  data  on   specific  compounds  or  total
nitroaromatics or nitramines).

  The precision and bias of the screening methods
are most appropriately assessed by comparison to
established laboratory  methods such  as EPA
Method 8330. Methods of comparison that have
been  used  include  relative  percent difference
(RPD), linear regression, correlation, and percent
of false positive and false negative results.  If
precision and bias are of critical importance, it is
recommended that the reports referenced in this
section be consulted directly.  Statistical  results
can be misleading when outliers or extreme values
are present. For example, in one linear regression
comparing field and laboratory methods, the slope
reported was  11  (i.e., measured against an ideal
slope  of 1) while the correlation coefficient was
0.99 (i.e., measured against an ideal of 1.0) all
because of one  very high concentration value
being  included with the remaining low values. It
also should be remembered that the contribution of
analytical error may be small compared to total
error.  Field error is usually the major contributor
to total error.  In comparing results, it should be
noted that all  CFI and FOB statistical results are
usually based on the means of several  analyses
per sample whereas the regression lines and RPD
for the other on-site methods are based on single
measurements.

  Several studies have been conducted comparing
the performance  of two or more on-site methods
with Method  8330.  Thorne and Myers (1997)
evaluated several immunoassay methods including
the D  TECH TNT and RDXkits, and the RaPID
Assay TNT kit.   Craig et al. (1996) and EPA
(1997) evaluated (1) the EnSys TNT and RDX +
HMX colorimetric kits,  (2) D TECH TNT and
RDX immunoassay kits, (3) the RaPID Assay
TNT immunoassay kit, (4) the CFI methods for
TNT and RDX, and (5) the FOB for TNT.  The
results presented include estimates of method bias
as determined by calculated RPDs and linear
regression analysis. Another study was conducted
by NRL to obtain more comprehensive data on the
FOB and CFI methods (ESTCP 1998). The Army
Corps of Engineers compared the EnSys TNT and
RDX + HMX method with Method 8330 during
monitoring  at  the Umatilla Groundwater site
(ACOE 1998).  The  results from each of these
studies are  summarized below and in Tables 8
and 9.

  The  Thorne and Myers (1997) study inves-
tigated TNT and RDX levels in  44 groundwater
wells from three sites:  the Umatilla Army Depot
in Hermiston, Oregon; the Naval Submarine Base
in Bangor,  Washington; and the Naval  Surface
Warfare Center in Crane, Indiana. The capability
of immunoassay kits was evaluated to determine
whether groundwater samples exceeded the EPA
lifetime health advisory of 2 ug/L (Roberts and
Hartley 1992; Roberts et al. 1993) and whether
the RPDs (the difference between the field and
reference method concentration divided by  the
mean value and expressed as a percent) were
within ± 50% of Method 8330 results.  The results
"were disappointing" and "none of the test kits
performed  as well as advertised," Thorne and
Myers (1997) reported. "The quantitative assays
were neither accurate nor precise enough to re-
place  Method 8330 although they could be used
adequately  as screening tools".  The D TECH
RDX test "failed badly" by producing 24% false
negative and 18%  false positive results relative to
the drinking water advisory limit of 2 ug/L. The D
TECH TNT kit produced 30% false positive and
no false negatives.  The detection limit of both D
TECH kits  are above the 2 ug/L drinking water
advisory limits.  The RaPID Assay method for
TNT  demonstrated no  false positive  or false
negative results. Thorne and Myers also looked at
the percent  of sample results within ± 50% of the
Method 8330 results.  For D TECH TNT and
RDX  kits,  32%  and  58%  of the  results,
                                            27

-------
respectively, were within acceptable limits.  For
the RaPID Assay method,  85% of the results
were   acceptable.    Finally  they  conducted
regression analyses comparing the RaPID Assay
TNT kit performance with Method 8330 results on
groundwater samples.  The ideal regression line
would be Y = mX + b where the slope, m, would
equal 1 and the intercept, b, would equal 0.  The
dependent variable is Y (on-site method estimate),
and the independent variable is X (Method 8330
result). A correlation coefficient (r) is typically
calculated that shows the degree of association
between the on-site method and Method 8330 and
can range between -1 and +1. For a perfect posi-
tive correlation r =  1.  The Thorne  and Myers
RaPID Assay results were Y = 1.48X + 0.0  with
a correlation  coefficient of r =  0.93, which is
highly significant (99% probability level).

  Results from the EPA study (Craig et al. 1996;
EPA 1997a) are summarized in Tables 8 and 9 for
the groundwater  samples from the Umatilla
Chemical Depot in Hermiston, Oregon, and the
Naval  Submarine  Base  in Bangor, Washington.
Groundwater at Umatilla has high nitrates and low
turbidity  while groundwater  and  leachate  at
Bangor has relatively high  organic carbon and
higher turbidity. Tables 8 and 9 includes the slope
of regression  lines for TNT and RDX  data
respectively, the correlation coefficient  (r),  the
mean and median of the absolute RPD values
(indication of precision), and mean of the RPDs
(indication of bias). The mean RPD closest to 0
shows  the greatest average agreement with the
reference laboratory method. The study concluded
that no on-site analytical method out performed
the other methods in all comparisons.  For the
TNT methods, the EnSys and CFI had the highest
accuracy  followed by the FOB,  RaPID Assay,
and D TECH methods.  All TNT method results
were biased high based on the net RPDs at both
sites and were generally biased slightly low for
RDX.  For RDX, the EnSys and CFI methods
showed the highest  accuracy  followed by  D
TECH. In general, the RDX on-site analytical
methods performed better than the TNT method.
The performance  may  have resulted from the
higher levels of RDX, which necessitated sample
dilution  and,  thereby,  also  reduced  matrix
interferences.
  The EnSys TNT kit accuracies were similar for
both sites and for the RDX kit, results were
slightly more accurate at the Bangor site.  Using
an RPD acceptance criterion of ± 50%  of the
Method 8330 result, 89%  of the EnSys TNT
results were acceptable and 78% of the RDX
results were acceptable. Overall accuracy of the
TNT and RDX EnSys colorimetric methods were
acceptable.

  The D-TECH methods were more accurate at
Umatilla   than   Bangor   because  of  lower
interference from organics and particulate matter.
The  majority of the  TNT RPD values were
positive and linear regression slopes were greater
than 1.0, thereby indicating a high bias for the on-
site  methods,  possibly  resulting  from  TNB
interference or cross-reactivity.  Using an RPD
acceptance criterion of ± 50% of the Method 8330
result,  70% of the  D  TECH  results  were
acceptable while 56% of the RDX results were
acceptable.

  The CFI TNT method performed about  the
same  at  both  sites while  the  FOB method
performed better at the Umatilla site.  For RDX,
the CFI performed well at both sites and was
similar in accuracy to the EnSys method.

  A recent report (ESTCP 1998) documented the
performance of the NRL CFI and FOB methods
at the Umatilla Army Depot, the Bangor Naval
Submarine Base, and the Crane, Naval Surface
Weapons Center.  Both methods were modified
significantly between the Craig et al.(1995) study
and the ESTCP (1998) study. For the CFI, the
instrument was changed and the small columns
were replaced by membranes.   The FOB TNT
antibody was  changed as  well  as  the fiber
geometry.  The statistics provided in Table 8 and
discussed  below  were  calculated   after  the
analytical results below 10 ug/L, or listed as below
the detection limit, were replaced with  5 ug/L
(one-half  of  the  method  detection   limit).
However, data were not included if both the NRL
and Method 8330 data would have been replaced
with 5 ug/L. This approach was taken to  permit
more data to be included in the analyses yet avoid
producing extreme RPDs near the detection limit.
Because a few very high or low values relative to
                                            28

-------
most of the data can have a misleading impact on
linear regression data, some regression equations
were  recalculated  with high  or  low samples
deleted based on the Method 8330 concentration.
The recalculated results are discussed in the text
below.

  The CFI data for TNT showed highly significant
correlations with Method 8330 data at the Umatilla
and Bangor sites; however, the mean  and median
RPDs  were  high  (Table 8).   This  apparent
contradiction results   from  several  high
concentration samples.  If the  three samples at
Umatilla with Method 8330 results greater than
200 ug/L are deleted (367, 846, and 1160 ug/L),
the regression equation changes to Y = 0.61X + 9
and the correlation coefficient drops to 0.51, which
is not significant (95% probability level). For the
Bangor  site,  two of the seven sample  results
showed concentrations greater than 200  ug/L and
most of the remaining Method 8330 values were
below 5 ug/L. Only four samples from the Crane
facility  were analyzed  by  the  CFI  method;
therefore, no regression line was calculated.  The
mean and median RPDs were about 100.

  The FOB results for TNT showed regression
slopes  of  0.41 and 0.35 for the  Umatilla and
Bangor  sites, respectively.  The corresponding
correlation coefficients were 0.52 and 0.70, which
were significant and not significant, respectively.
For all of the TNT samples analyzed by the FOB
method, the TNT levels were below 200 ug/L. No
Crane samples were analyzed for TNT using the
FOB method.

  For RDX,  the  CFI  method showed  highly
significant and significant regressions with Method
8330  data  collected at Umatilla and Bangor,
respectively   (ESTCP   1998).   Net  RPDs
demonstrated a low bias  at both sites  (-63 and
-42), and the means and medians of the absolute
RPDs ranged from 68 to 78.  At Crane, the CFI
method  showed a highly significant regression,
mean and median absolute RPDs of 64 and 42
with a net RPD of -6.  If the one especially high
or low 8330 value is deleted from the Bangor and
Crane data sets, the resulting regression lines are
still significant,  as is the regression for Umatilla
when the three samples with 8330 values greater
than 200 ug/L are deleted.

  The FOB results on RDX  at Umatilla and
Bangor  showed highly  significant regressions,
mean and median RPDs generally below 50, and
low net  RPDs.  All of the Umatilla 8330 results
were below 200 ug/L, and if two high  samples
(356 and 562 ug/L) at Bangor are deleted, the
regression is still significant although the  slope of
the regression line changes to 2.5.

  The  Army   Corps  of  Engineers  collected
numerous groundwater samples at Umatilla that
were analyzed for multiple compounds during an
effort to document the conditions in groundwater
wells and  the  effectiveness  of  the granular
activated carbon treatment system (ACOE 1998).
After eliminating nonrepresentative data, EnSys
and Method 8330 data were available for 40 RDX
and 36 TNT samples (Table 8).  These  regression
results for these data show Y = 0.69X  + 132 with
r = 0.90 for RDX and Y = 1.3X  -15 with r = 0.97
for the TNT data. The averages of the absolute
value of the RPDs were 31  for  RDX and 44 for
TNT. For these data sets, the net RPDs (average
when sign is considered) were -6.1 for RDX and
22 for TNT and are relatively unbiased compared
with other  study results presented in Table  9.
Ideally,  the net  RPD should balance out to zero
indicating no bias.

  Several projects comparing Method 8330 results
with on-site analytical  methods  are underway so
additional published data will become available
from EPA and the Corps of Engineers. Also, see
the section on  emerging technologies  about a
planned demonstration for the summer  of 1999 on
current  and  emerging on-site  methods  for
explosives in water and soil.
                                            29

-------
           Table 8. Comparison of On-Site Analytical Methods for TNT to EPA Method 8330
Method
EnSys1
DTECH1
RaPID Assay1
Flow Immunosensor1
Fiber-Optic Biosensor1
Flow Immunosensor2'4
Fiber-Optic Biosensor2'4
EnSys3
Method
Flow Immunosensor2'4
TNT
Regression
Umatilla
Y = 1.4X+191
Y = 2.0X+73
Regression
Bangor
Y=1.1X+51
Y=11X-558
Y=1.0X+140
Y = 1.2 + 46
Y=1.7X-267
Y = 0.71X-18
Y = 0.41X + 24
Y=1.3X-15
Y=1.2X + 242
Y = 0.71X + 285
Y=1.6X-7
Y = 0.35X+10

Regression Crane
NA
Correlation
Coefficient (r)
Umatilla/Bangor
0.98**/1.0**
0.88**/1.0**
0.99**combined
0.70*70.84*
0.91**/0.76*
0.92**/0.98**
0.52*70.70
0.97**
Correlation
Coefficient (r)
NA
Mean RPD
(absolute value)
Umatilla/Bangor
66/58
64/143
78 combined
47/52
33/107
114/100
85/55
44
Mean RPD
(absolute value)
103
Median RPD
(absolute value)
Umatilla/Bangor
45/63
48/152
87 combined
47/38
25/116
147/89
74/52
30
Median RPD
(absolute value)
103
Net RPD
Umatilla/Bangor
66/58
58/143
78 combined
32/51
30/100
-41/87
67/40
22
Net RPD
36
Number of
Samples
Umatilla/
Bangor
15/9
15/7
7
combined
11/7
12/8
14/7
16/6
36
Number of
Samples
4
OJ
o
             1  EPA1997a
             2  ESTCP1998
             3  ACOE1998
             4
               Statistics based on means of usually eight or more analyses of each sample.
               Statistically significant at the 95% probability level.
               Statistically significant at the 99% probability level.

-------
Table 9. Comparison of On-Site Analytical Methods for RDX to EPA Method 8330
Method
EnSys1
DTECH1
Flow Immunosensor1
Flow Immunosensor2'4
Fiber-Optic Biosensor2'4
EnSys3
Method
Flow Immunosensor2'3
Fiber-Optic Biosensor2'3
RDX
Regression
Umatilla
Y = 0.81X+ 135
Y=1.3X-269
Y = 0.92X+203
Y = 0.72X-30
Y = 0.53X+13
Y = 0.69X+ 132
Regression
Bangor
Y = 0.96X + 6
Y = 1.7X+172
Y = 0.72X+1.1
Y = 0.67X - 3
Y = 0.61X+38

Regression Crane
Y = 0.75X + 40
Y = 0.44X-5
Correlation
Coefficient (r)
Umatilla/Bangor
0.86**/0.92**
0.96**/0.61*
0.72**/0.92**
0.91**/0.69*
0.64**/0.82**
0.90**
Correlation
Coefficient (r)
0.77**
0.94**
Mean RPD
(absolute value)
Umatilla/Bangor
33/21
53/67
26/30
78/76
37/56
31
Mean RPD
(absolute value)
64
100
Median RPD
(absolute value)
Umatilla/Bangor
27/21
32/56
19/23
78/68
33/40
32
Median RPD
(absolute value)
42
104
Net RPD
Umatilla/Bangor
-11/-7.7
-36/61
-11/-30
-6S/-42
10/14
-6.1
Net RPD
-6
-100
Number of
Samples
Umatilla/Bangor
23/12
23/12
20/12
20/11
20/10
40
Number of
Samples
13
11
 1  EPA1997a
 2  ESTCP1998
 3
    ACOE1998
 4  Statistics based on means of usually eight or more analyses of each sample.
 *  Statistically significant at the 95% probability level.
 **  Statistically significant at the 99% probability level.

-------
Additional Considerations

  Other important factors in the selection of an
on-site method include, for example, the size and
type of working area required, the temperature of
the working area, the need for electricity and
refrigeration, the amount of waste produced, the
need  to transport solvents, and the degree  of
portability.   Immunoassay  methods  are  more
sensitive than colorimetric methods to freezing and
elevated  temperatures,  and  the   ambient
temperature affects  the speed at  which  color
development takes place on some immunoassay
methods. Most tests are best run protected from
the weather, for example, in a van, field trailer, or
nearby building.

Emerging Methods

  A  GC-nitrogen phosphorus detector  method
currently under  development at CRREL appears
to offer the ability to provide on-site analysis for
the common  suite of nitroaromatics and  nitra-
mines in water (Hewitt and Jenkins in press).  In
this method, the analytes of interest are precon-
centratedby passing 1 L of water through Empore
SDB-RPS extraction membranes and eluting the
retained compounds with 5  mL of acetone.  An
aliquot of the acetone extract is then determined
on a field portable gas chromatograph equipped
with a nitrogen-phosphorus  detector.   Method
detection limits were demonstrated to be below 1
ppb for TNT, RDX, 4-amino-DNT, 2-amino-DNT,
and 2,4-DNT.

  Two German companies are now manufacturing
immunoassay kits for explosives in water but no
literature was found comparing on-site results to
standard methods. The TNT kit by Coring System
uses a 96 well microplate to which TNT conjugate
is bound. Samples, or TNT standards and a TNT
specific  antibody (rabbit) are pipetted into  wells
and the plate is incubated for  an hour.  After
rinsing, a rabbit specific antibody is added, the
plate is incubated for an hour, rinsed, and substrate
is added to  the  wells.  After  20 minutes  of
incubation, blue  color development is stopped and
the resulting yellow color is read at 450 nm using
a photometer. The TNT concentration is inversely
proportional to the color. The method detection
limit is reported to be 0.5 ug/L for water. The
method is cross reactive with 1,3-DNT (650%),
and 2,4-DNT (60%). For more information send
e-mail to:  info(g),coring.de.

Coring currently has no plans to market their kit in
the U.S.  No information was provided by Bio-
Genes on their immunoassay procedure.

  In December, 1998, EPA issued a Notice of an
Intention to Conduct a Demonstration and Per-
formance  Verification  Study of Explosives Field
Analytical Devices as  part of the Environmental
Testing and Verification  program. The  demon-
stration is planned for the summer of 1999 and will
include  analysis  of both water and  soils. For
further information on the demonstration, contact:

Eric Koglin
National Exposure Research Laboratory
Environmental Sciences Division
P.O. Box 93478, Las Vegas, NV 89193-3478
Phone: 702-798-2432
E-mail: koglin.eric(g),epa. gov

Summary of the EPA Reference
Methods for Explosive Compounds
in Water

Properties of Secondary Explosives
  The  two  secondary  explosives  used  to the
greatest extent by the U.S. military over the past
70 years are TNT and RDX.  With their manu-
facturing  impurities  and  environmental  trans-
formation  products, the two compounds  account
for a large part of the explosives contamination at
active  and  former U.S.  military  installations.
While all explosive compounds can all be classi-
fied as semivolatile  organic  chemicals, their
physical and chemical properties require different
analytical  approaches than normally is used for
other semivolatiles.

  Table  10 presents  some of the  important
physical and chemical properties for TNT and
                                            32

-------
RDX, and some of their commonly encountered
manufacturing  impurities  and  environmental
transformation products.  The unique properties
that  differentiate  these chemicals from  other
semivolatiles  such as polychlorinated biphenyls
(PCBs) and polynuclear aromatic hydrocarbons
(PNAs) are their thermal lability  and polarity.
Many of these compounds thermally degrade or
explode at temperatures  below 300* C.   Thus,
methods based on gas chromatography (GC) have
not gained wide acceptance. However, methods
developed by Hable et al. (1991) and Walsh  and
Ranney (1998a, 1998b) have shown that the gas-
chromatography electron-capture detector (GC-
ECD) method can be used successfully for nitro-
aromatics and nitramines in water. In addition, log
KOW values range from 0.06 to 2.01 compared with
values of 4 to 5 for PCBs and PNAs, indicating
that these compounds are quite polar and that the
nonpolar solvents used for other  semi-volatile
organics are not the best choice for extraction of
nitroaromatics and nitramines from water.  For
most  routine  analyses,  environmental  water
samples are  extracted with  either  salting-out
solvent extraction with acetonitrile or using solid-
phase extraction with a styrene-divinyl-benzene-
based solid phase (Method 3535A  [EPA 1998]).
The sample extracts are analyzed using reversed-
phase high-performance liquid chromatography
(RP-HPLC),  often  using Method  8330  (EPA
1998) or a recently adopted GC-ECD method,
Method 8095 (EPA 1998).

Water Extraction

  High  concentration   water  samples  have
generally been analyzed by diluting an aliquot of
the water  1:1 with methanol, then filtering  the
sample  through a  0.45 to  0.50-um  filter,  and
analyzing  a 100-uL aliquot  of the  filtrate  by
RP-HPLC-UV.  Quantitation limits for this direct
water method (Method 8330) range from 5.7 ug/L
for 2,4-DNT to 14 ug/L for RDX.

  Often detection limits that can be obtained using
the direct  water method are  not sufficient  for
project-specific DQOs.  In these cases, the target
analytes must be extracted from the  water and
preconcentrated  before  either  RP-HPLC-UV
(Method 8330) or GC-ECD (Method 8095) deter-
mination.  Extraction is accomplished using either
salting-out solvent extraction (Leggett et al. 1990)
followed  by   nonevaporative  preconcentration
(Jenkins and  Miyares 1991), or by solid-phase
extraction, EPA Method 3535A (Jenkins et al.
1995b, 1995c; EPA 1998).

  A  direct comparison of salting-out solvent
extraction and solid-phase extraction with RP-
HPLC-UV was  conducted by Jenkins et al.
(1994a, 1995b, 1995c) using groundwater samples
from the Crane, Indiana, Naval Surface Warfare
Center.  The results indicate that  excellent
extraction efficiency  was  achieved using both
procedures  (recoveries were  generally greater
than 90%).    Quantitation limits using these
approaches were  similar and  ranged  from less
than 0.1  ug/L for some target analytes to  0.84
ug/L for RDX. The authors cautioned, however,
that carefully cleaned solid phases must be used or
interferences  will  be released from  the solid
phases by some  water matrices (Jenkins et al.
1994a, 1995b, 1995c). A small residual peak that
interfered with RDX was found even with highly
cleaned solid-phase  materials.  The GC-ECD
method, which was  recently given preliminary
approval by the EPA (Method 8095, [EPA 1998]),
specifies  that  solid-phase  extraction should be
used  when  samples  are  to  be  analyzed by
GC-ECD.  The  salting-out solvent  extraction
method was not evaluated for use with GC-ECD.
Method detection limits (MDLs) for the GC-ECD
method range from 0.04 ug/L to 0.4 ug/L for the
various target analytes when a 500-mL sample is
used  and  preconcentrated  into  5   mL  of
acetonitrile.

Reversed-Phase High-Performance Liquid
Chromatography Determination

  Generally, detection of the analyte within the
proper retention time window on two columns with
different   retention   orders  is  required  for
confirmation of the presence of these explosives.
Method 8330 specifies primary analysis on an
LC-18 (octadecylsilane) column with  confirma-
tion  on a cyanopropylsilane  (LC-CN) column
(Jenkins et al. 1989).
                                            33

-------
Table 10. Physical and Chemical Properties of Predominant Nitroaromatics and Nitramines.
Compound
TNT
TNB
2,4-DNT

Tetryl
RDX
HMX
Molecular
Weight
227
213
182

287
222
296
Melting Point
(-C)
80.1 to 81. 6
122.5
69.5 to 70.5

129.5
204.1
286
Boiling Point
(-C)
240 (explodes)
315
300
(decomposes)
(decomposes)
(decomposes)
(decomposes)
Water Solubility
(mg/L at 20- )
130
385
270

80
42
5 at 25-
Vapor Pressure
(to rr at 20-)
4.4E-06
2.2E-04
1.1E-04

5.7E-09
4.1E-09
3.3E-14
iogKow
1.86
1.18
2.01

1.65
0.86
0.061

-------
  Walsh, Chalk, and Merritt (1973) were the first to
report on the use of RP-HPLC for the analysis of
nitroaromatics  in  munitions  waste.   Most sub-
sequent HPLC methods for these compounds rely
on UV detection  because of its sensitivity and
ruggedness.  Initially, determination was specified at
254  nm because  of the availability  of  fixed
wavelength detectors  at this wavelength based on
the mercury vapor lamps and a significant absor-
bance of all target analytes.  Current instruments
are  generally  equipped  with   either   variable
wavelength detectors  or diode array detectors, and
wavelengths  of maximum  absorption  can  be
selected to optimize detection.  However, 254 nm is
still often used because it is  specified in Method
8330  and  because  of the   low incidence  of
interference at this wavelength.

Gas-Chromatography Electron-Capture
Detector Determination

  The earliest use of gas chromatography to deter-
mine nitroaromatics dates from the  early 1960s
(Parsons et al. 1961).  The early methods used the
relatively insensitive flame ionization detector, and it
was not until the early 1970s that the selectivity and
sensitivity of the  BCD for  nitroaromatics was
realized (Murrmann et al. 1971). The first GC-ECD
method for nitroaromatics and nitramines in water
was developed by Hoffsommer and Rosen (1972).
The introduction of fused silica columns in the early
1980s   reduced  the   problems  with   thermal
degradation of these thermally labile compounds and
permitted routine  determination  of nitramines  by
GC. Routine analytical methods were developed for
nitroaromatics  by  Belkin et  al.  (1985)  and for
nitroaromatic and nitramines by Hable et al. (1991)
at the  Army  Environmental Hygiene Agency.
Hable's method used solvent extraction to extract
and  preconcentrate  the  target analytes  and
GC-ECD for determination.   Unfortunately, two
separate extractions  were   needed,  one  for
nitroaromatics  using  toluene  and a second for
nitramines using isoamylacetate.  More recently,
Walsh and Ranney (1998b) combined solid-phase
extraction with GC-ECD, and the results were used
to establish EPA Method 8095 (EPA 1988).  One
advantage of Method  8095 compared with Method
8330 is the  ability to quantify nitroglycerine and
PETN  in  the  same   determination  as  the
nitroaromatics and nitramines.

  Method 8095 specifies  that detection of peaks
from the BCD  within the proper retention time
window on two columns with different polarity is
required for confirmation of the  presence of the
target analytes. Method 8095 specifies DB1 as the
primary analytical column (although DB5 provides
better resolution for target analytes)  and either
RTX200 or RTX225 as the confirmation column
(Walsh and Ranney 1998a, 1998b).

  It  is  important that  the  injector  and  oven
temperatures, column lengths,  and linear velocities
specified in Method 8095 are used for GC-ECD
analysis. Otherwise, poor recovery, particularly for
HMX, will result. The injection port liner must be
thoroughly deactivated and changed frequently, or
performance will be degraded for HMX, RDX, and
the aminodinitrotoluenes.  Particular attention also
must be given to thorough drying of the solid phase
used for solid-phase extraction before elution with
acetonitrile.

  A comparison of the performance of GC-ECD
with RP-HPLC-UV for  these target  analytes  in
water is presented by Walsh and Ranney (1998a,
1998b).  Analysis of extracts by both RP-HPLC-
UV and GC-ECD results in  excellent analytical
confirmation, particularly when target analytes are
present at very low concentrations.

Method Specifications and Validation

  Based on the research described above,  EPA
Method 8330 (EPA 1998) and Method 8095 (EPA
1998) specify the following:

A. Salting-out the Solvent  Extraction
   1.  Place 251.3 g of sodium chloride in a 1-L
      (round) volumetric  flask.   Add a 770-mL
      aliquot of the water sample, and stir the flask
      with a stirring bar until the salt is dissolved.
   2.  While stirring the solution,  add  164 mL of
      acetonitrile to the flask.  Stir for at least 15
      minutes (30 minutes is safer) to dissolve as
      much acetonitrile as possible. Turn  off the
                                              35

-------
     stirring bar, and allow the phases to separate
     for at least 10 minutes.
   3. Remove the upper acetonitrile layer (about 8
     mL) with a Pasteur pipette and transfer it to
     a 100-mL (round) volumetric flask. Add  10
     mL of fresh acetonitrile to the water sample
     in the 1-L flask and stir for an additional  15
     minutes, followed by  10 minutes to allow the
     phases  to  separate.  Remove the  upper
     acetonitrile layer and combine with the initial
     acetonitrile extract in the 100-mL flask.
   4. Add 84 mL of salt water (325 g of NaCl per
     1,000 mL of reagent-grade water) to the 100-
     mL flask, and stir for 15 minutes, followed by
     10  minutes for phase separation.  Carefully
     transfer the top acetonitrile layer to a 10-mL
     graduated cylinder using a Pasteur pipette.
     Add an additional 1.0 mL of acetonitrile to the
     100-mL flask and stir for 15 minutes followed
     by  10 minutes for phase separation. Combine
     the second extract with the first in the 10-mL
     graduated  cylinder. Record  the volume  of
     extract  and  then   dilute   it  1:1   with
     reagent grade water.  This extract is analyzed
     using RP-HPLC.
B. Cartridge Solid-Phase Extraction
   1. Obtain prepacked  solid-phase extraction
     cartridges (Porapak RDX or Sep-Pak, 6 cc,
     500 mg, or equivalent). Clean the cartridges
     by  placing them on a solid-phase extraction
     manifold and passing 15  mL of acetonitrile
     through each using gravity flow.  Then flush
     the acetonitrile from the cartridges using 30
     mL of reagent-grade water. Ensure that the
     cartridges are never allowed to dry after the
     initial cleaning.
   2. Place a connector on the top of each car-
     tridge and fit the connector with a length  of
     one-eighth-in.-diameter Teflon tubing.  Place
     the other end of the  tubing in  a 1-L beaker
     containing 500 mL of sample.  Turn on the
     vacuum and  set the flow rate through each
     cartridge at about 10  mL per minute. Adjust
     the  flow rate  if  it  declines  significantly
     because  of  partial   plugging  from the
     suspended material.    After extracting the
     sample, remove the top plug containing the
     fitted tubing from each cartridge and pass 10
     mL  of reagent-grade water  through the
     cartridge using gravity flow unless the car-
     tridges are sufficiently plugged to require a
     vacuum.  Use a 5-mL aliquot of acetonitrile to
     elute  the  retained analytes from the  car-
     tridges under gravity  flow.  Measure the
     volume of the recovered acetonitrile, and
     either use  directly for GC-ECD determin-
     ation  (Method 8095) or dilute  1:1  with
     reagent-grade water  for  RP-HPLC-UV
     determination.
C. Membrane Solid-Phase Extraction
   1. Preclean styrene-divinylbenzene membranes
     (47 mm, Empore or equivalent) by centering
     on a 47-mm vacuum  filter apparatus and add
     several milliliters  of  acetonitrile to swell the
     membrane  before clamping the reservoir in
     place. Add a 15-mL  aliquot of acetonitrile to
     soak into the membrane for 3 minutes.  Then
     turn on the vacuum and pull most (but not all)
     of the solvent through the membrane.
   2. Add a 30-mL aliquot of reagent-grade water
     and resume the vacuum.  Just before the last
     of the water is pulled through the membrane,
     remove the vacuum,  fill the reservoir with a
     500-mL sample,  and resume the vacuum.
     The  sample  extraction  will take from 5
     minutes to an hour depending on the amount
     of suspended matter present. Once the water
     is eluted, draw air through the membrane for
     1  minute to remove  excess water. Place a
     40-mL vial below the outlet of the membrane,
     and add a 5-mL aliquot of acetonitrile on top
     of the membrane.  Allow the acetonitrile to
     soak into the membrane for 3 minutes.  Then
     apply the vacuum to  pull  the acetonitrile
     through  the  membranes  into  the   vials.
     Remove each resulting extract with a Pasteur
     pipette, and measure  the volume in a 10-mL
     graduated cylinder.  Measure the volume of
     the recovered acetonitrile, and either use
     directly for GC-ECD determination (Method
     8095) or diluted 1:1 with reagent-grade water
     for RP-HPLC-UV determination.
                                              36

-------
                                        Summary
   A  number  of  defense-related  sites  are
contaminated with elevated levels of secondary
explosives in groundwater and  surface water.
Levels  of contamination range from  barely
detectable (approximately  1 ug/L) to more than
10,000 ug/L.   On-site analytical methods are
essential to  more  economical  and improved
characterization and remediation.  What they lack
in accuracy and multi-compound specificity, they
more than make up for in the increased number of
samples that can be analyzed and the utility of
near-real-time data for making decisions on-site.
While verification using a standard laboratory
analytical method such as EPA Method 8330 or
8095  should  be part  of any  quality assurance
program,  reducing  the  number  of samples
analyzed by more expensive methodology can
result  in reduced costs and more efficient use of
limited resources while still achieving the DQOs.

   Two basic types of on-site analytical methods
are in use for explosives in water:  colorimetric
and immunoassay. Colorimetric methods genera-
ly detect broad classes of compounds such  as
nitroaromatics or nitramines while immun assay
methods are more compound specific. Prototype
biosensor methods for TNT and RDX have been
               field tested and are emerging methods for explo-
               sives analysis  in water (Rogers  and Gerlach
               1996).  Because TNT or RDX is usually present
               in explosive-contaminated groundwater or surface
               water, the use of field procedures designed  to
               detect these or similar  compounds can be very
               effective.

               Selection of an on-site analytical method involves
               evaluation of many factors including the specific
               objectives of the study and DQOs, compounds  of
               interest, explosives present at the site, the number
               of samples to be run, the sample  analysis rate,
               interferences or cross-reactivity of the method, the
               skill required, the analytical cost per sample, and
               the need for and availability of support facilities  or
               services.  Other factors  that should be considered
               are  the precision and accuracy of the on-site
               analytical method and the required detection limits.
               It should be remembered that analytical error may
               be small  compared  to  field error and that the
               precision and bias of a method is dependent on the
               site-specific conditions (compounds present and
               relative concentration) as well  as the skill of the
               analyst.
                           Federal Facility Forum Members
Region 1
U.S. EPA
JFK Federal Building
Boston, MA 02203

Megan Cassidy, Co-Chair
(617) 573-5785

Region 2
U.S. EPA
290 Broadway
New York, NY  10007-1866

Bill Roach
(212) 637-4335
Region 3
U.S. EPA
341 Chestnut Bldg.
Philadelphia, PA 19107

Steve Hirsh, Co-Chair
(215) 566-3052

Paul Leonard
(215) 566-3350
Region 4
U.S. EPA
345 Courtland Street, N.E.
Atlanta, GA 30303-3415

Carl Froede
(404) 562-8550

Jim Barksdale
(404) 562-8537

Olga Perry
(404) 562-8534
                                             37

-------
Region 5
U.S. EPA
77 W. Jackson Boulevard
Chicago, IL 60604

Craig Thomas
(312) 886-5907

Carol Witt  Smith
(312) 886-6146

Region 6
U.S. EPA
1445 Ross  Avenue
Dallas, TX 75202-2733

Nancy Morlock
(214) 665-6650

Chris Villarreal
(214) 665-6758
Ruby Williams
(214) 665-6733
Region 7
U.S. EPA
726 Minnesota Avenue
Kansas City, KS 66101

Scott Marquess
(913)551-7131

Region 8
U.S. EPA
999 18th Street
Denver, CO 80202-2413

Floyd Nichols
(303) 312-6983

Jim Kiefer
(303) 312-6907

Judith McCulley
(303) 312-6667

Region 9
U.S. EPA
75 Hawthorne Street
San Francisco, CA 94105-3901

Glenn Kistner
(415) 744-2210

Sheryl Lauth
(415) 744-2387
Region 10
U.S. EPA
1200 Sixth Avenue
Seattle, WA 98101

Harry Craig
(503) 326-3689

Kathy Stryker
(206)553-1171

Headquarters
U.S. EPA/OSWER
401 M Street, SW
Washington, DC 20460

Allison Abernathy
(202) 260-5646

Doug Bell
(202) 260-8716
Lance Elson
(202) 564-2577

Remi Langum
(202) 260-2457

Diane Lynne
(202) 564-2587
                                          38

-------
                                        Notice
   The U.S.  Environmental Protection Agency,    publication. Mention of trade names or cornmer-
through its Office of Research and Development,    cial products does not constitute endorsement or
funded and prepared this Issue Paper. It has been    recommendation by the EPA for use.
peer reviewed by the EPA and approved for
                                  Acknowledgment

   Work partly performed under the auspices of    wishes to thank the U.S. Army Environmental
the U.S. Department of Energy, Office Contract    Center and Cold Regions Research and Engineer-
No. DE-AC07-94ID13223,  through Interagency    ing Laboratory for assisting in the preparation of
Agreement No. DW89937192-01-2 with the U.S.    this document.
Environmental  Protection Agency.  The EPA
                                           39

-------
                                      References
ACOE. 1996a. Safety Concepts and Basic Considerations for Unexploded Ordnance (UXO) Operations,
ETL-385-1-1, U.S. Army Corps of Engineers, Huntsville Center, Alabama.

ACOE. 1996b, Generic Scope of Work for Ordnance Avoidance Operations, ETL-385-1-2, U.S. Army
Corps of Engineers, Huntsville Center, Alabama.

ACOE. 1998. Chemical  Report No. 1, Contaminated Groundwater Remediation Treatment System
Startup Phase, Umatilla Army Depot Activity, Hermiston, Oregon, Contract No. DACA67-95-C-0101,
prepared by ICF Kaiser Engineers, Inc. for U.S. Army Corps of Engineers, Seattle District, Seattle,
Washington.

AEC. 1997. Technology Application  Analysis, Demonstration/Validation of Ultraviolet Oxidation
(UV/Ox)  Methods for the Remediation of Explosives Contaminated Groundwater at Savanna Army
Depot Activity, Savanna, Illinois, SFIM-AEC-ET-CR-97007, U.S. Army Environmental Center, Aberdeen
Proving Ground, Maryland. 23 pp.

AEHA. 1986. Ground-Water Monitoring Study No.  38-26-0457-86,  AMC  Open  Burning/Open
Detonation Facilities, U.S. Army Environmental Hygiene Agency, Aberdeen Proving Ground, Maryland.

AEHA. 1994. Wastewater Management Study No. 32-24-HP16-94, Effects of Active Firing Range
Activities  on  Environmental  Media,  Aberdeen  Proving  Ground-Aberdeen  Area,  Maryland,
31 January-30 November 1993, U.S. Army Environmental Hygiene Agency, Aberdeen Proving Ground,
Maryland.

Army. 1984. Military Explosives, TM-9-1300-214, Department of the Army, Washington, D.C.

ASTM. American Society for Testing and Materials, Conshohocken, Pennsylvania, http://www. astm. org

Bart, J.C., L.L. Judd, K.E. Hoffman, AM. Wilkins, and AW. Kusterbeck. 1997a. Application of a
portable immunosensor to detect the explosives TNT and RDX in groundwater samples. Environ. Sci.
Technol. 31(5): 1505-1511.

Bart, J.C., L.L. Judd, K.E. Hoffman,  AM. Wilkins,  P.T. Charles, and AW. Kusterbeck.  1997b.
Detection and Quantitation of the Explosives 2,4,6-Trinitrotoluene and Hexahydro-l,3,5-trinitro-l,3,5-
triazine in Groundwater Using a Continuous Flow Immunosensor, In: Immunochemical Technology for
Environmental Applications, ACS Symposium Series 657, ed. D.S. Aga and E.M. Thurman, pp. 210-220.

Belkin, F., R.W. Bishop, and M.V. Sheely.  1985. Analysis  of explosives in water by  capillary gas
chromatography. J. Chromatog. Sci. 24: 532-534.

Blackwood, L.G, and E. L. Bradley. 1991. An omnibus test for comparing two measuring devices. J.
Qual. Tech. 23(1): 12-16.

Brannon, J.M., and T.E. Myers. 1997. Review of Fate and Transport Processes of Explosives, Technical
Report IRRP-97-21, U.S. Army Engineer Waterways Experiment Station, Vicksburg, Mississippi. 24 pp.
                                           40

-------
California EPA. 1996a. D TECH™ TNT kit evaluation report. Technology Certification Program,
available from Office of Pollution Prevention and Technology Development, 400 P Street, Sacramento,
California 95814.

California EPA. 1996b. D TECH™ RDX kit evaluation report. Technology Certification Program,
available from Office of Pollution Prevention and Technology Development, 400 P Street, Sacramento,
California 95814.

California EPA. 1996c. Ohmicron TNT RaPID Assay® evaluation report. Technology Certification
Program, available from Office of Pollution Prevention and Technology Development, 400 P Street,
Sacramento, California 95814.

Christensen, R, and L.G. Blackwood. 1993. Tests for precision and accuracy of multiple measuring
devices. Technometrics. 35:411-420.

CMECC.  1996. Field Analytical Measurement Technologies, Applications, and Selection, California
Military Environmental Coordination Committee, State of California Water Resources Control Board.
http: //www. epa. gov/region09/qa/measure-technol. pdf

Comfort, S.D., P.J. Shea, L.S. Hundal, Z. Li, B.L. Woodbury, J.L. Martin, and W.L. Powers. 1995. TNT
transport and fate in contaminated soil. J. Environ. Qual. 24:1174-1182.

Craig, H., G. Ferguson, A. Markos, A. Kusterbeck, L. Shriver-Lake, T. Jenkins, and P. Thorne. 1996.
Field Demonstration of On-Site Analytical Methods for TNT and RDX in Groundwater, In: Proceedings
of the  Great  Plains  Rocky  Mountain  Hazardous Substance Research  Center (HSRC)AVaste
Management  Education and  Research  Center (WERC) Joint Conference  on the  Environment,
Albuquerque, New Mexico, May 21-23, 1996. http://www. engg.ksu. edu/HSRC/96Proceed/craig. pdf

Crockett, A.B., H.D. Craig, T.F.  Jenkins, and W.E. Sisk. 1996. Field Sampling and Selecting On-Site
Analytical Methods for Explosives  in  Soil, Federal  Facilities Forum  Issue,  EPA/540/R-97/501,
U.S. Environmental Protection Agency, 32 pp. http://www.epa.gov/crdlvweb/pdf/fld-smpl.pdf

Crockett, A.B., T.F. Jenkins, H.D. Craig, and W.E. Sisk. 1998. Overview of On-Site Analytical Methods
for Explosives in Soil, Special Report 98-4, U.S. Army Corps of Engineers, Cold Regions Research and
Engineering Laboratory, 30 pp.

http://www.crrel.usace.army.mil/techpub/ RREL_Reports_web/reports/SR98_04.pdf

EPA. 1991. Engineering Bulletin: Granular Activated Carbon Treatment, EPA/540/2-91-024, Office of
Emergency and Remedial Response, Washington, D.C. and Office of Research and  Development,
Cincinnati, Ohio, 7 pp.

EPA. 1992. Guidance for Data Useability in Risk Assessment (Part A). Final Report, OSWER Directive
9285.7-09A, U.S. Environmental Protection Agency, Office of Emergency and Remedial Response,
Washington, D.C., 290 pp.

EPA.  1993a. Handbook: Approaches for the Remediation of Federal Facility  Sites Contaminated with
Explosive or Radioactive Wastes, EPA/625/R-93/013, U.S. Environmental Protection Agency, Office of
Research and Development, Washington, D.C., 116 pp.
                                            41

-------
EPA. 1993b. Subsurface Characterization and Monitoring Techniques. Volume 1:  Solids and Ground
Water, EPA/625/R-93/003a, U.S.  Environmental Protection  Agency, Office of  Research and
Development, Washington, D.C.

EPA. 1993c. Water Quality Guidance for the Great Lakes System and Correction: Proposed Rules. Fed
Regist. 58:20802-21047.

EPA. 1994a. Guidance for the Data Quality Objectives Process, EPA QA/G-4, Quality Assurance
Management Staff, U.S. Environmental Protection Agency, Washington, D.C., 68 pp.

EPA. 1994b.  Methods for Monitoring Pump-and-Treat Performance, EPA/600/R-94/123, Office of
Research and Development, Washington, D.C. 102 pp.

EPA. 1995. Umatilla Army Depot Activity Soils Desorpti on Study, Project No. 71370, prepared by Black
& Veatch Waste Science, Inc., for U.S. EPA Region 10, Seattle, Washington.

EPA. 1996a. Final Guidance: Presumptive Response Strategy and Ex-Situ Treatment Technologies for
Contaminated Groundwater at CERCLA Sites, EPA 540/R-96/023, PB96-963508, Office of Solid Waste
and Emergency Response, Washington, D.C.

EPA. 1996b. Drinking Water Regulations and Health Advisories, EPA 822-B-96-002, U. S. Environmental
Protection Agency, Office of Water, Washington, D.C.

EPA. 1997a. Explosives in Water Screening Technologies, UMDA and SUBASE Bangor Final Report,
Project No. 71370, prepared by  Black  & Veatch Special Projects Corp. for U.S. EPA Region 10,
Seattle, Washington.

EPA. 1997b. The Use of Field Methods to Support RFI Streamlining, EPA Corrective Action Workgroup
Quality Assurance Subcommittee, unpublished, 13 pp. http://www.epa.gov/reg5rcra/wptdiv/cars/field.pdf

EPA. 1998. Test Methods for Evaluating Solid Waste, Physical/Chemical Methods, SW-846, Revision
0, Office of Solid Waste and Emergency Response, U.S. Environmental Protection Agency, Washington
D.C.

ESTCP.  1998. Field Demonstration and Method Validation of NRL  Environmental Immunosensors,
prepared by Center for Bio/Molecular Science and Engineering, Naval Research Laboratory, for
Environmental  Security Technology Certification Program.

Golden, IP.,  E.W. Saaski,  L.C. Shriver-Lake, GP. Anderson, and  F.S.  Ligler.  1997.  Portable
multichannel fiber optic biosensor for field detection. Opt. Eng. 36:1008-1013.

Grant, C.L., T.F. Jenkins, and S.M. Golden.  1993. Evaluation of Pre-extraction Analytical Holding Times
for Nitroaromatics and Nitramine Explosives in Water,  Special  Report 93-24, U.S. Army Corps of
Engineers, Cold Regions Research and Engineering Laboratory, 13 pp.

Grant, C.L., T.F. Jenkins,  K.F. Meyers,  and E.F. McCormick. 1995. Holding time estimates for soils
containing explosives residues: comparison of fortification vs. field contamination. Environ. Toxicol.
Chem. 14:1865-1874.
                                            42

-------
Grubbs, F.E. 1973. Errors of measurement,  precision, accuracy and the statistical  comparison of
measuring instruments. Technometrics. 15:53-66.

Hable,M., C. Stern, C. Asowata, andK. Williams. 1991. Determination of nitroaromatics and nitramines
in ground and drinking water by wide-bore capillary gas chromatography. J. Chromatog. Sci. 29:131-135.

Hewitt, A.D., and T.F. Jenkins, (in press). On-Site Method for Nitroaromatic andNitramine Explosives
in Soil and Groundwater. Special Report 99-X. U.S. Army Corps of Engineers, Cold Regions Research
and Engineering Laboratory.

Hofsommer, J.C., and J.M. Rosen.  1972. Analysis  of explosives in sea water. Bull. Environ. Contam.
Toxicol. 7:177-181.

Hundal, L.S., P.J. Shea, S.D. Comfort, W.L. Powers, and J. Singh. 1997. Long-term TNT sorption and
bound residue formation in soil. J. Environ. Qual. 26:896-904.

Jenkins, T.F., M.E. Walsh, P.W.  Schumacher, P.H.  Miyares, C.F. Bauer, and C.L. Grant. 1989. Liquid
chromatographic  method for the determination of extractable nitroaromatic and nitramine residues in soil.
J. Assoc. of Official Anal Chem. 72:890-899.

Jenkins, T. F., 1990. Development of a Simplified Field Screening Method for the Determination of TNT
in Soil, Special Report 90-38, U.S. Army Corps of Engineers, Cold Regions Research and Engineering
Laboratory.

Jenkins, T.F., and P.W. Schumacher. 1990. Evaluation of a Field Kit for Detection of TNT in Water and
Soils, Special Report 90-20, U.S. Army Corps of Engineers,  Cold Regions Research and Engineering
Laboratory, 14 pp.

Jenkins, T.F., and P.H. Miyares. 1991. Non-evaporative preconcentration technique for volatile  and
semi-volatile solutes in certain polar solvent. Anal. Chem. 63:1341-1343.

Jenkins, T.F., P.H.  Miyares, K.F. Myers, E.F. McCormick, and A.B. Strong.  1994a. Comparison of
solid-phase extraction and salting-out solvent extraction for preconcentration of nitroaromatic  and
nitramine explosives from water. Anal. Chim. Acta. 289: 69-78.

Jenkins, T.F., P.O.  Thorne, and  M.E. Walsh.  1994b. Field Screening Method for TNT and RDX in
Groundwater, Special Report 94-14, U.S. Army Corps of  Engineers, Cold Regions Research  and
Engineering Laboratory, 12 pp.

Jenkins, T.F., P.G Thorne, E.F. McCormick, and K.F. Myers. 1995a. Preservation of Water Samples
Containing Nitroaromatics and Nitramines, Special Report 95-16, U.S. Army Corps of Engineers, Cold
Regions Research and Engineering Laboratory, 31 pp.
http://www.crrel.usace.army.mil/valliere/CRREL_Reports_web/reports/SR95_16.pdf

Jenkins, T.F., P.G. Thorne, K.F. Myers, E.F.  McCormick,  D.E. Parker, and B.L. Escalon.  1995b.
Evaluation of Clean Solid Phases for Extraction of Nitroaromatics and Nitramines from Water, Special
Report 95-22, U.S.  Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, 15
pp. http://www. crrel.usace. army.mil/valliere/CRREL_Reports_web/reports/SR95_22.pdf
                                            43

-------
Jenkins, T.F., P.G. Thorne, K.F. Myers, andE.F. McCormick. 1995c. Evaluation of the New Clean Solid
Phases for Extraction of Nitroaromatics and Nitramines from Water, Presented at the Eleventh Annual
Waste Testing and Quality Assurance Symposium, Washington, pp. 128-142, DC, 23-28.

Jenkins, T.F., C.L. Grant, G.S. Brar, P.G.  Thorne,  T.A.  Ranney, and P.W.  Schumacher.  1996a.
Assessment of  Sampling Error Associated with  Collection and  Analysis  of Soil  Samples  at
Explosives-Contaminated Sites, Special Report 96-15, U.S.  Army  Corps of Engineers, Cold Regions
Research and Engineering Laboratory, 38 pp.
http://www.crrel.us ace.army.mil/techpub/CRREL_Reports_web/reports/SR96_15.pdf

Jenkins, T.F., P.W. Schumacher, J.G Mason, and P.G. Thorne.  1996b. On-Site Analysis for High
Concentrations of Explosives in Soil: Extraction Kinetics and Dilution Procedures. U.S. Army Corps of
Engineers, Cold Regions Research and Engineering Laboratory Special Report 96-10.
http: //www. crrel. us ace, army. mil/techpub/CRREL_Reports_web/reports/SR96_l 0. pdf

Jenkins, T.F., M.E. Walsh, P.G. Thorne, S. Thiboutot, G Ampleman, T.A. Ranney, and C.L. Grant. 1997.
Assessment of Sampling Error Associated With Collection and Analysis of Soil Samples at a Firing Range
Contaminated with HMX, Special Report 97-22, U.S. Army Corps of Engineers, Cold Regions Research
and Engineering Laboratory.
http: //www. crrel. us ace, army. mil/techpub/CRREL_Reports_web/reports/SR97_22. pdf

Kitts, C.L., D.P. Cunningham, and P.J. Unkefer.  1994. Isolation of three hexahydro-l,3,5-trinitro-
1,3,5-triazine degrading species of the family Enterobacteriaceae fromnitramine explosive-contaminated
soil. Appl. Environ. Microbiol. 60:4608^1611.

LANL 1996. RFI Report for  Potential  Release Sites in TA-16,  Los Alamos National Laboratory,
Environmental Restoration Project, LA-UR-96-3191, Los Alamos, New Mexico.

Lee,  C.K.M., and M.K. Stenstrom. 1996. Competitive Adsorption of Cyclotrimethylenetrinitramine
(RDX) and Cyclotetramethylenetetranitramine (HMX), UCLA ENG 96-152, Civil and Environmental
Engineering Department, University of California at Los Angeles, Los Angeles, California.

Leggett, D.C., T.F. Jenkins, and P.H. Miyares. 1990. Salting-out solvent extraction for preconcentration
of neutral, polar organics from water. Anal. Chem. 62:1355-1356.

Mandel, J. 1984. Fitting straight lines when both variables are subject to error. J. Environ. Qual. 16:1-14.

Maskarinec, M.P., C.K. Bayne, L.H. Johnson, S.K. Holladay, RA. Jenkins, and B.A. Tomkins. 1991.
Stability of Explosives in Environmental Water and Soil Samples, ORNL/TM-11770, Oak Ridge National
Laboratory, Oak Ridge,  Tennessee, 98 pp.

McCormick, N.G,J.H. Cornell, and A.M. Kaplan. 1981. Biodegradation of hexahydro-l,3,5-trinitro-l,3,5-
triazine. Appl. Environ. Microbiol. 42: 817-823.

Murphy, W.L.,  and R.  Wade.  1998. Final Report: RCRA Facility Investigation, Phase II  Release
Assessment for Surface  Water SWMU 03/10 Ammunition Burning Ground, Technical Report GL-98-2,
U.S.  Army Corps of Engineers,  Waterways Experiment Station, Vicksburg, Mississippi.
                                            44

-------
Murrmann, R.P., T.F. Jenkins, and D.C. Leggett. 1971. Composition and Mass Spectra of Impurities in
Military Grade TNT Vapor. U.S. Army Corps of Engineers, Cold Regions Research and Engineering
Laboratory Special Report 158, Hanover, New Hampshire.

Narang, U., P. R. Gauger, A. W. Kusterbeck, and F. S.  Ligler. 1998. Multianalyte detection using a
capillary-based flow immunosensor. Anal. Biochem. 255:13-19.

Parker, L.V., T.F. Jenkins,  and P.B.  Black. 1989. Evaluation of Four  Well Casing Materials for
Monitoring Selected Trace Level Organics in Ground Water, Special Report 89-18, U.S. Army Corps of
Engineers, Cold Regions Research and Engineering Laboratory, 29 pp.

Parker, L.V., and T. A. Ranney. 1993. Effect of Concentration on Sorption of Dissolved Organics by Well
Casings, Special Report 93-8, U.S. Army Corps of Engineers, Cold Regions Research and Engineering
Laboratory, 17 pp.

Parsons, G.S.,  S.M.  Tsang, M.P.  DiGiamo, R.  Feinland, R.A.L. Paylor.  1961. Separation and
determination of mono- and dinitrotoluene isomers by gas  chromatography.  Anal. Chem. 33:1858-1859.

Pennington, J.C., T.E. Myers, W.M. Davis, T.J. Olin, T.A. McDonald, C.A. Hayes, and D.M. Townsend.
1995. Impacts of Sorption on In-Situ Bioremediation of Explosives Contaminated Soils, Technical Report
IRRP-95-1, U.S. Army Waterways Experiment Station, Vicksburg, Mississippi.

Roberts, W.C., and W.R Hartley. 1992. Drinking Water Health Advisory: Munitions. Lewis Publishers,
Boca Raton, Florida.

Roberts, W.C., B.J. Commons, H.T. Bausum, C.O.  Abernathy, J.J. Murphy, K. Khanna, and E.V.
Ohanian. 1993. Overview of the Health Effects of Selected Munitions Chemicals, EPA 822-R-93-022,
Office of Water, U.S.  Environmental Protection Agency,  Washington, D.C., 94 pp.

Rogers, K.R., and C.L. Gerlach. 1996. Environmental biosensors, a status report. Environ. Sci. Technol.
30(11):486A-491A.

Rubio, F.R, T.S. Lawruk, A.M. Gueco, D.P. Herzog, and J.R Fleeker. 1996. Determination of TNT in
soil and water by a magnetic particle-based enzyme immunoassay system.  Proceedings of 11th Annual
Waste Testing and Quality Assurance Symposium, American Chemical Society, July 23-28, 1995.

Shriver-Lake, L.C., K.A. Breslin, P.T. Charles, D.W. Conrad, J.P.  Golden,  and F.S. Ligler.  1995.
Detection  of  TNT  in  water using an  evanescent  wave fiber-optic biosensor. Anal.  Chem.
67(14):2431-2435.

Shriver-Lake, L.C., B.L. Dormer, and F.S. Ligler. 1997. On-site detection of TNT with a portable fiber
optic biosensor. Environ. Sci. Technol. 31(3):837-841.

Shriver-Lake, L.C., N.A. Naz, and F.S. Ligler.1998. A fiber optic biosensor for the detection of
TNT/RDX in  environmental  samples,  In:  Current   Protocols  in  Field  Analytical  Chemistry,
ed. V. Lopez-Avila, John Wiley & Sons, pp 2E.l-2E.il.

Sikora, F.J., L.L. Behrends, H.S. Coonrod, W.D. Phillips, and D.F. Bader. 1997. Phytoremediation of
explosives in groundwater using innovative wetlands-based treatment technologies, In: Proceedings of
                                            45

-------
the 12th Annual Conference on Hazardous Waste Research, Great Plains-Rocky Mountain Hazardous
Substance Research Center, Kansas City, Missouri, May 19-22, 1997, pp. 168-178.

Spanggord, R.J., T. Mill, T.W. Chou, W.R. Mabey, J.H. Smith, and S. Lee. 1980. Environmental Fate
Studies on certain munition wastewater constituents, Final Report, Phase I - Literature Review, Contract
No. DAMD 17-78-C-8081, AD-A082372,  prepared by  SRI International for U.S.  Army Medical
Research and Development Command, Ft. Derrick, Frederick, Maryland.

Stephan,  C.E., D.I. Mount, D.J. Hansen, J.H.  Gentile,  C.A.  Chapman,  and W.A.  Brungs.  1985.
Guidelines for Deriving Numerical Water Quality Criteria for the Protection of Aquatic Organisms and
Their Uses. PB85-227049, U.S. Environmental Protection Agency, Washington, D.C.

Swanson, A., H.E. Canavan, L.A.  Kelly, and J.B. Roberts. 1996. Comparison of mobile laboratory
screening methods for high explosive with  EPA  SW-846 Method 8330. Proceedings of Fourth
International Conference On-Site Analysis, January 21-24, 1996.

Talmage, S.S., D.M. Opresko, C.J. Maxwell, C.J.E. Welsh, P.M. Cretella, P.S. Hovatter, and F.B.
Daniel. 1999. Nitroaromatic munition compounds: environmental effects and screening values, Rev.
Environ. Contain. Toxicol.  161:1-157.

Teaney, G.B., and RT. Hudak. 1994. Development of an enzyme immunoassay based field screening
system for the detection of RDX in soil and water. Proceedings of 87th Annual Meeting and Exhibition,
Air & Waste Management Association, Cincinnati, Ohio, 94-RP143.05, 15 pp.

Thorne, P.G, and T.F.  Jenkins. 1995a.  Development of a Field Method  for Ammonium Picrate/Picric
Acid in Soil and Water, Special Report 95-20, U.S. Army  Corps of Engineers, Cold Regions Research
and Engineering Laboratory, Hanover, New Hampshire, 22 pp.
http://www.crrel.usace.army.mil/valliere/CRREL_Reports_web/reports/SR95_20.pdf

Thorne, P.G, and T.F. Jenkins. 1995b. Field screening method for picric  acid/ammonium picrate in soil
and water. In: Field Screening Methods for Hazardous Wastes and Toxic Chemicals, VIP-47, Air &
Waste Management Association, Pittsburgh, Pennsylvania, 2:942-947.

Thorne, P.G, and D.C. Leggett. 1997. Hydrolytic  release of bound residues from composted soil
contaminated with 2,4,6-trinitrotoluene. Environ. Toxicol. Chem. 16:1132-1134.

Thorne, P.G, and K.F. Myers. 1997. Evaluation  of Commercial Enzyme Immunoassays for the Field
Screening of TNT and RDX in Water, Special Report 97-32, U.S. Army Cold Regions Research and
Engineering Laboratory, Hanover, New Hampshire, 15 pp.
http://www.crrel.usace.army.mil/valliere/CRREL_Reports_web/reports/SR97_32.pdf

USAGE.  1999.  Standard Operating Procedures for Analysis of TNT and RDX, SOP No. TNT/RDX-
1 (draft), ICF Kaiser Engineers, available from Kira Lynch, Seattle District Corps, PO Box 3755 (EN-TB-
ET), Seattle WA 98124-2255.

Vlahakis, J.G 1974. A laboratory study of RDX adsorption by carbon, Naval Facilities Engineering
Command, Army Research and Development Command, Alexandria, Virginia.
                                            46

-------
Walsh, J.T., R.C. Chalk, and C. Merritt. 1973. Application of liquid chromatography to pollution
abatement studies of munitions wastewater. Anal. Chem. 45:1215-1220.

Walsh, M.E.  1989. Analytical Methods for Determining Nitroguanidine in Soil and Water,  Special
Report 89-35, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory.

Walsh, M.E., and T.F. Jenkins. 1991. Development of A Field Screening Method for RDX in Soil, Special
Report 91-7, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory.

Walsh, M.E., T.F. Jenkins, P.S. Schnitker, J.W. Elwell, and M.H. Stutz. 1993.  Evaluation of SW-846
Method 8330 for Characterization of Sites Contaminated with Residues of High Explosives,  Special
Report 93-5, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, 17 pp.

Walsh, M.E., and T.A. Ranney. 1998a. Determination of Nitroaromatic, Nitramine, and Nitrate Ester
Explosives in Water Using Solid-Phase Extraction and GC-ECD:  Comparison with HPLC.  Special
Report 93-2, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, 28 pp.
http: //www. crrel. us ace, army. mil/techpub/CRREL_Reports_web/reports/CR98_02. pdf

Walsh, M.E., and T.A. Ranney. 1998b. Determination of nitroaromatic,  nitramine, and nitrate  ester
explosives in water using solid-phase extraction and gas chromatography-electron capture detection:
comparison with high-performance liquid chromatography. J. Chromatog. Sci. 36:406-416.

Whelan, J.P., A.W. Kusterbeck, G.A. Wemhoff, R. Bredehorst, and F.S. Ligler. 1993. Continuous-flow
immunosensor for detection of explosives. Anal. Chem. 65:3561-3565.

Yinon, J. 1990. Toxicity and Metabolism of Explosives. CRC Press, Boca Raton, Florida.
                                            47

-------