78.3 mg/kg (mean of duplicates), while the later
Table 6. Analysis of airborne dust from Fort Ord
sample size study gave a mean estimate of 41.8
and CFB-Valcartier soils.
mg/kg. A similar comparison for sample 1-44-11
HMX concentration (mg/kg)
yields 307 vs. 135 mg/kg.
by SW-846 Method 8330
Below, we suggest possible explanations for
these inconsistencies. First, the RSDs from the
Sample
Soil
Dust
sample size study might be expected to fall on the
low side because the two samples used have mid-
Fort Ord (1-44-composite
262
860
grid A 015 cm)
range concentrations where reproducibility should
be optimal. Further, all the subsampling and analy-
CFB-Valcartier (surface
382
739
sis for this study were done over a short time, i.e.,
composite)
all five replicates were removed at one time. This
invariably improves reproducibility, often greatly.
With respect to decreases in concentration esti-
types of soils, it is important to further investigate
mates, we acknowledge that there is a wide range
these issues and develop an improved homogeni-
in both the original values and the sample size
zation and subsampling procedure. Possibly, a
study values. Referring to Table 5 and excluding
misting device could be used to increase the soil
the 0.5-g samples, we note that the other 20 analy-
moisture level prior to subsampling, thereby
ses for sample 1-44-8 range from 113 to 171
reducing dust losses. Separate aliquots could be
mg/kg, compared with values of 204 and 409
used for moisture determination, thereby allow-
mg/kg for the original analyses. The probability
ing correction of concentration estimates to a dry
of both of the original analyses being substantially
weight basis. In a later section of this report, we
higher than any of the 20 later analyses is unlikely
will provide further evidence that the problem of
to be explained as a random occurrence. Further-
poor reproducibility resides in sample processing
more, exactly the same pattern holds true for
rather than extract analysis.
sample 1-44-11. In addition, data to be presented
later in this report (see Table 9) show this same
Evaluation of various on-site methods for use
pattern for four additional samples. Consequently,
with soils from the inland ranges
we feel confident that a real concentration decrease
If additional site characterization at Fort Ord is
has been demonstrated for the later analyses.
needed, the use of on-site methods and com-
It was noted earlier that we observed a large
positing could substantially reduce costs and time
amount of fine dust when these very dry bulk
(Crockett et al. 1996). The two explosives com-
samples were homogenized and subsampled.
pounds of most significance at the Fort Ord
Analysis of two samples of dust showed HMX
inland ranges are HMX and RDX. Two on-site
concentrations two to three times higher than for
methods have been developed for RDX. The first
the bulk soil (Table 6). Apparently, a substantial
is an enzyme immunoassay method by Strategic
portion of the explosives residues are associated
Diagnostics, called the DTECH method. The sec-
with very fine particles that are not bound to the
ond is a colorimetric method by Jenkins and Walsh
bulk sands. Clearly, there is danger of loss of these
(1992), which is commercially available from
fine particles during mixing and subsampling.
EnSys Corporation (now Strategic Diagnostics).
Also, the surface layers of remaining bulk samples
To compare the utility of these two methods, two
could be enriched as dust settles after mixing. Since
experiments were run. In the first experiment, the
HMX is known to be very stable under aerobic
extraction kinetics were evaluated. Three depth
conditions (Grant et al. 1993), and particularly
increments (015, 3045, and 105120 cm) of
when soils are dry (Bauer et al. 1989), physical
sample 1-44-1 were selected for this evaluation.
losses of dust may explain the much lower con-
Each sample was extracted with acetone for 3, 10,
centrations found in the sample size study. We
and 30 minutes, and an aliquot of the resulting
could also speculate that analyte not lost as fine
extract was analyzed for HMX using RP-HPLC.
dust may be associated with the larger particles,
The results show that extracts collected after only
and this could improve the reproducibility ob-
a 3-minute extraction period provide concentra-
served in the sample size study, since these
tions of HMX that are greater than 97% of those
samples had been previously manipulated.
obtained after 30 minutes (Table 7). These results
Since any further site characterization at the in-
agree with assessments made during initial devel-
land firing ranges will require analysis of these
17
Previous Page
