The sample was shaken vigorously by hand; then
one subsample (Table 1). Samples with lower con-
the SPME phase was exposed to the headspace
centrations apparently did not contain white phos-
for 5 min. The SPME phase was immediately
phorus particles large enough to produce a visi-
transferred to a heated (200C) injection port of a
ble flame or leave orange residue. Based on these
portable gas chromatograph (SRI Model 8610)
results, we concluded that the field test provided
equipped with a nitrogen-phosphorus detector.
a means to quickly detect large (millimeter size)
The polydimethylsiloxane fused-silica column
WP particles if they were present in the subsam-
(J&W DB-1, 15 m 0.53 mm i.d., 3-m film thick-
ple taken, but several subsamples may be required
ness) was maintained at 80C and the carrier gas
to obtain a positive result for samples with few
was nitrogen set at 30 mL/min.
particles.
We then questioned how many subsamples
Laboratory method
should be tested if the first subsample yielded a
White phosphorus concentration (g/g) was
negative result. A 500-mL jar contains sufficient
determined using isooctane extraction of a 40-g
material for approximately 20 subsamples; how-
wet subsample followed by gas chromatography
ever, testing so many subsamples per sample
(Walsh and Taylor 1993).
would be extremely tedious. Next we tried pre-
concentrating what remained of samples 28
(Table 1) by sieving through a 30-mesh (0.59-mm)
sieve to remove most of the fine-grain silts and
RESULTS AND DISCUSSION
clays and reduce the volume of the sample so that
only one test need be run. No more particles were
Initial tests
found in samples 26; however, four more parti-
Initially, we collected samples from sites where,
cles were found in sample 7 and two more in
using the laboratory method, we previously had
sample 8.
detected WP and from two sites where WP was
not detected (Racine et al. 1993a and b). At each
sample site, a 500-mL jar was filled with sediment
Composite sampling
Since sieving provides a means to reduce the
and subsamples were smeared across the bottom
of aluminum pie pans. Samples were tested for
volume of a sample, we tested sieving in the field
WP by heating each pan on a camp stove.
as a means to create composite samples from a
For each sample where we had detected WP
large area (Fig. 3). Dabbling ducks at Eagle River
above 1 g/g by the laboratory method, we ob-
Flats have proved to be efficient samplers of white
tained a positive result by the field test for at least
phosphorus particles as evidenced by their high
mortality. We reasoned that by simulat-
ing the way ducks feed, by sieving sever-
Table 1. Comparison of white phosphorus concentrations
al small sediment samples over a large
found by laboratory method and number of white phospho-
area, we might increase the efficiency of
rus particles detected by field method.
sampling. Our concern over a sampling
Number
Number
method stems from the way in which the
WP conc.
subsamples
WP particles
marsh was contaminated. Projectiles con-
(g/g)
Sample
Sample ID
tested
found
taining white phosphorus produce dis-
1
1248
ND
2
0
crete and relatively small areas of con-
2
AEHA BT1
ND
10
0
tamination. For example, following the
3
AEHA PB1
0.0143
10
0
detonation of 81-mm mortar rounds the
4
AEHA D2
0.079
10
0
areas containing the majority of the white
5
AEHA DUP1
0.205
10
2
phosphorus residue were only 0.5 m in di-
6
AEHA D1
0.43
10
0
ameter (0.2 m2) (Walsh and Collins 1993).
7
AEHA C2
1.6
10
1
8
240
2.32
10
1
To sample for such a small hot spot with
9
53
10.2
2
1
90% confidence (β = 0.10) using a grid
10
1247
88
1
23
pattern of sampling would require a 0.9-
11
1245
168
2
7
m grid spacing (Gilbert 1987). When we
12
110
590
2
2
13
AEHA PB2
1740
1
12
consider that Eagle River Flats contains
14
1246
3,071
1
68
700,000 m2 of ponds, the grid approach
15
MHB site
5600
1
>100
to sampling would be unrealistically cost-
ly, even where multiple hot spots exist.
ND = not detected
3
Previous Page
