minutes with 2.00 mL of isooctane. Following
ples (1 L for water and 40 g for wet soil) were pre-
phase separation, the isooctane extract was ana-
pared by adding an aqueous solution of white
lyzed.
phosphorus to yield concentrations near the meth-
Wet sediment samples were extracted by plac-
od detection limits for the solvent extraction meth-
ods (0.01 g/L for water and 1 g/kg for soil).
ing a 40-g subsample into a 120-mL jar containing
10.0 mL of reagent-grade water. Then, 10.0 mL of
Headspace SPME was performed and peak height
isooctane was added. Each jar was tightly sealed
data were obtained.
with a Teflon-lined cap and vortex-mixed for one
For those water samples that had white phos-
phorus concentrations near 0.01 g/L, the peak
minute, and then placed horizontally on a plat-
form shaker for 18 hours. The sample then was
heights obtained by headspace SPME were small,
allowed to stand undisturbed for 15 minutes, and,
but white phosphorus was consistently detectable
if necessary, centrifuged for five minutes, to per-
in all spiked samples (signal-to-noise ratios were
mit phase separation. Extracts were analyzed with-
greater than 7). Therefore the detection capability
in a few hours.
of the SPME appeared to be comparable to solvent
extraction. In addition, the SPME response factors
(means of the peak heights normalized to spiked
Gas chromatograph
concentration for each water matrix) were simi-
white phosphorus was determined by injecting a
lar. The similarity in response with different wa-
1.0-L aliquot on-column into an SRI Model 8610
ter matrices was further studied in terms of cali-
gas chromatograph equipped with a nitrogen
bration, as described below.
phosphorus detector. SPME fibers were thermally
In contrast, SPME peak heights varied with the
desorbed at 200C in the injection port of the same
different soil matrices. Peak heights were lowest
GC, and, for convenience, the SPME fiber was left
in the sample with the highest organic content and
in the injection port for the entire run time (five
grain size distribution. However, in all cases, the
minutes).
peak heights were much larger than those obtained
The methylsilicone fused silica column (J and
by solvent extraction, indicating that detection
W DB1, 0.53-mm-i.d., 15-m, 3.0-m film thickness)
capability of the headspace SPME might be better
was maintained at 80C. The carrier gas was ni-
than that for solvent extraction. Further studies
trogen set at 30 mL/minute. Under these condi-
were performed with field-contaminated samples
tions, white phosphorus eluted at 2.7 minutes.
as described below.
Depletion of total analyte present
RESULTS AND DISCUSSION
White phosphorus may be present in sediment
samples as heterogeneously distributed particles
Detection capability
We evaluated the detection capability of SPME
of different masses. Due to potential loss of white
while gathering validation data for analytical
phosphorus by sublimation and oxidation, tradi-
methods for white phosphorus by solvent extrac-
tional homogenization methods involving drying,
tion and gas chromatography (Walsh et al. 1995).
grinding, sieving, mixing, and subsampling are
For each matrix (Table 1), ten replicate spiked sam-
not applicable for white phosphorus-contaminat-
Table 1. Mean peak heights obtained following headspace SPME/GC for
water and soil samples spiked at white phosphorus concentrations near
the detection limit for solvent extraction methods.
Peak height
Spiked
Standard
RSD
Response
Matrix
conc.
n
Mean
deviation
(%)
factor*
0.012 g/L
Reagent-grade water
10
1,465
252
17
122,000
0.0097 g/L
Well water
10
1,119
106
9.5
115,000
0.010 g/L
Pond water
10
1,013
235
23
101,000
1.9 g/kg
Sand
9
270,481
20,356
7.5
142,000
0.97 g/kg
Lebanon (Sandy silt)
10
97,063
10,486
11
100,000
0.84 g/kg
USAEC (Loam)
9
74,381
6,157
8.3
88,500
* (Peak height)/(Spiked concentration [g/L]).
4