licate, and was nil again for the silt. We next spiked
show poor recovery (41 to 74%) of the spiked
10 replicates of Ottawa sand to obtain an estimate
analytes. However, grain size alone did not ac-
of MDL (Table 2, Table A7) for extraction of 25 g of
count for low recovery in one of the Fort Ord
soil with 50 mL of acetonitrile, and we found that
samples, each of which was sandy. The reason for
the MDL were very similar to those obtained us-
the difference in stability of spiked versus aged,
ing just 2 g of soil and 10 mL of acetonitrile. The
field-introduced analytes is beyond the scope of
expectation of improvement in detection capabil-
this study, but this difference greatly reduces our
ity by using more soil in proportion to the volume
ability to judge method accuracy.
of extraction solvent was negated by the small in-
terfering peaks introduced by the matrix.
Drying of samples
To further explore matrix effects on analyte re-
More artifacts associated with spiking were re-
covery, we spiked 5 replicate 25-g samples of Ot-
vealed when we attempted to fortify soils with
tawa sand with 5 mL of spike solution to yield 10-
analytes from the vapor phase. We wanted to
g/kg nitroaromatic and 40- g/kg RDX
know whether soil samples collected from a
concentrations. We used a larger volume of spike
minefield should be extracted field-moist to pre-
solution to wet as much soil as possible, and we
vent loss of the more volatile analytes. Air-drying
allowed the spiked samples to age uncapped in
is desirable because it facilitates homogenization
the fume hood for 24 hours. Recoveries were
and prevents introduction of water into the gas
around 80% for the nitroaromatics (except TNT
chromatograph where it can degrade the deacti-
due to co-elution of an unknown interference) and
vated injection port liner.
90% for RDX (Table A8).
In the first experiment, we placed 50 g of AEC
Next we spiked five replicate samples of AEC
standard soil in a porous nylon bag and suspended
soil, silt, and wet silt (5 mL of water added to 25 g
the bag above crystals (1 g) of TNT (Kodak). We
of silt) using the same procedure as for sand. We
wetted another 50 g of AEC soil with 10 mL of
also added spike solution to two empty test tubes
distilled water, and placed the damp soil in an-
and left the tubes uncapped for 24 hours, as for
other nylon bag and suspended it over another 1
the tubes containing the spiked soils. Recoveries
g of TNT. In both cases, the soil and TNT were
were around 90% for the samples without soil, but
sealed in 1-gallon glass jars for one week in the
were inconsistent and low from each of the spiked
dark. A shallow dish of distilled water was added
soils (Table A9), showing that these soils are ei-
to the jar containing the damp soil to prevent the
ther actively sequestering or destroying the
soil from drying during exposure to the TNT va-
analytes added at such a low concentration (10 g/
por.
kg). The AEC soil has been used many times be-
The analytes found in the TNT vapor were de-
fore for spike recovery studies that used HPLC
termined by headspace solid-phase micro extrac-
analysis, and several years ago it was used to ex-
tion (SPME) (polyacrylate) (Jenkins et al. [in prep]).
amine the stability of spiked soils (Bauer et al.
Major peaks in the chromatogram for which we
1989). In this study 2-g soil samples were spiked
had standards were (in decreasing order) 2,4-DNT,
with 1 mL of spike solution to yield a target con-
3,5-DNT, 2,5-DNT, 1,3-DNB, 2,3-DNT, 2,6-DNT,
centration of 4000 g/kg, the solvent was allowed
2,4,6-TNT, 1,2-DNB, and 3,4-DNT.
to evaporate, and the spiked soils were stable for
After one week of exposing the soils to the TNT
up to 62 days. Now, when we use a small volume
vapor, each soil was split by fractional shoveling.
of spike solution in proportion to the amount of
Five replicate 2-g subsamples were taken from half
soil spiked, and spike at a much lower concentra-
of each soil sample and each subsample placed
tion (10 g/kg), the soil matrix effects on spiked
immediately in 10 mL of acetonitrile. The other
analytes become pronounced, resulting in low and
half of each soil was spread on an aluminum pie
inconsistent recoveries of the spiked analytes from
pan and placed in a fume hood overnight. The next
some matrices.
morning, five replicate 2-g subsamples were taken
Looking back on the MS/MSD of field-contami-
from each air-dried soil and each subsample
nated soils (Table A2), we note variable recover-
placed in 10 mL of acetonitrile. All samples were
ies. Lacking extensive characterization of each of
sonicated overnight, then filtered and analyzed by
these soils, the cause of the variable recoveries is
GC-ECD.
open to speculation. One difference that was vi-
Analytes found in the soil (Table 3) that was
sually obvious was grain size. The finest-grained
not wetted were the same as those found in the
soil was that from Chickasaw, and this soil did
vapor, in roughly the same order of abundance,
18
Previous Page
