bility of removing the explosives contamination as well.
fiber was conditioned following the recommendation
The top and bottom surfaces of the mines were
of the manufacturer. Just before use, each fiber was ther-
mally desorbed at 250C in a GC-injection port to
independently sampled as follows. A number of filter
paper disks (1.5 cm) (Whatman) that had been pre-
assure that it was clean.
soaked in methanol were placed randomly, using stain-
To sample the headspace, the vial cap was removed
less steel forceps, on the mine surface, and the metha-
and the aluminum foil was punctured by the needle
nol was allowed to evaporate. The filter disks were then
housing of the fiber. The fiber was extended into the
transferred to 15-mL silanized amber vials (Supelco)
headspace for a sampling period of 5 to 20 minutes.
containing 3 mL of AcN. The samples were immedi-
After the sampling was complete, the fiber was with-
ately placed on ice in the dark for storage and shipment
drawn into its protective needle and the needle was
to the laboratory. Ten filter disks each were used to
removed from the vial. The outside of the needle was
sample the top and bottom surfaces of the anti-tank
wiped with a tissue wetted with acetone to eliminate
mines (TMM-1, TMA-5, and Type 72) and, likewise,
any vapors sorbed to the metal. The needle was then
three disks were used for each surface of the anti-
inserted into the injection port of the gas chromatograph
for thermal desorption and analysis by GC-ECD. Analy-
personal mines (PMA-1A and PMA-2). Thus, the top
and bottom area sampled was 17.7 cm2 for the anti-
sis conditions were the same as for soil extracts. The
tank mines and 5.31 cm2 for the anti-personal mines.
mass of analytes desorbed was determined by use of
Two TMM-1 mines, three Type 72 and PMA-2 mines,
reference standards (SARMS) in acetonitrile. Concen-
and four TMA-5 and PMA-1A mines were sampled.
trations were then calculated by dividing the desorbed
Additional vials containing only AcN were shipped
mass by the volume of vapor sampled by the fiber. A
to the laboratory to serve as solvent blanks. At various
description of how the effective sampling volume of
times throughout the sampling session, several filter disks
this fiber was determined is presented elsewhere
(Jenkins et al. 1999). The soilair (Ks/a) partitioning
were allowed to air dry and then placed into acetonitrile
coefficient was calculated by dividing the soil concen-
and shipped to the laboratory to serve as sampling blanks.
tration in nanograms per gram, as determined previ-
When the samples arrived at the laboratory, they
ously, by the vapor concentration in nanograms per
were stored under the same conditions as the soil
milliliter. The resulting partition coefficient has the units
samples until analyzed. In preparation for analysis, the
milliliters of air per gram of soil.
samples were allowed to warm to room temperature
and the acetonitrile was filtered to remove soil particles,
using the same method as filtering soil extracts. The
RESULTS AND DISCUSSION
AcN was then analyzed by GC-ECD in a manner simi-
lar to the soil extracts.
Physical soil characterization
Analysis of soil and soil headspace vapor
Soils at the DARPA Electronic Nose Testing Facil-
to estimate soilair partition coefficients
ity at Fort Leonard Wood are classified as Plato silt
loam (fine, mixed, mesic Aquic Fragiudalfs) according
Soil samples were collected above, around, and
to the USDA soil survey of Pulaski County, Missouri
below buried land mines at Fort Leonard Wood in April
1999 and analyzed for munitions as previously
(U.S. Department of Agriculture Soil Conservation
described. Twelve of these samples, containing high
Service 1989). The pH values (5.76.4) in samples taken
levels of analytes, were selected for headspace vapor
in the first sampling set (sites 19) were within the range
analysis to enable computation of soilair partition
reported by USDA for this location (3.67.3) (Table
coefficients. Several grams of each soil selected were
13), but organic carbon values (0.580.84%) were
placed in 40-mL glass vials, the tops of the vials were
slightly lower than reported by USDA (12%). CEC,
covered with aluminum foil, and plastic screw caps were
exchangeable iron, and TKN were moderate to low.
Particle size distribution was typical of the soil type
placed on top to hold the foil in place. The sides of the
described by USDA, except that samples tended to be
vials were covered with aluminum foil to exclude light
slightly more sandy than those reported by the USDA
and they were stored at room temperature for several
(Table 14). Visual observation of soil samples in the
days to equilibrate with headspace vapor. At the end of
second and third sets suggests that these samples were
the equilibration period, the headspace vapor was
similar to samples in the first set and that soil proper-
sampled using a solid phase micro extraction (SPME)
ties were relatively uniform across the site.
device. The SPME fiber used for this study was coated
with a 65-m film of polydimethylsiloxane/
Soilwater partition coefficients (Kd) for the Fort
Leonard Wood soils were relatively low for both TNT
divinylbenzene (PDMS/DVB). The manufacturer
and 2,4-DNT, ranging from 1.9 to 16.2 for 2,4,6-TNT
(Supelco) calls this their pink fiber. Prior to use, the
12
to contents
Previous Page
