14
preconcentration step. For that reason, we treat-
ed the results from the direct analysis as "true val-
SPE-C vs. SOE
ues" for purposes of comparison with results from
12
SPE-M vs. SOE
the three preconcentration techniques. Table 4
summarizes results for samples where analytes
10
were detected by the direct RP-HPLC method. Of
SPE-C
vs. SOE
the 33 groundwater samples analyzed, 11 had de-
SPE-M
8
tectable HMX using direct analysis, with concen-
vs. SOE
trations ranging from 25 to 325 g/L. Likewise,
RDX was detected in 13 groundwaters using the
6
direct method with concentrations ranging from
13 to 608 g/L; TNT in four samples with concen-
4
trations ranging from 14 to 180 g/L; 4ADNT and
2ADNT in five samples with concentrations rang-
ing from 9 to 59 g/L and 7 to 65 g/L, respec-
2
tively; and TNB was detected in two samples at 5
and 8 g/L. For a given analyte, the ratio of the
0
concentration obtained for each preconcentration
2
4
6
8
10
12
14
HMX Concentration by SOE (g/L)
technique relative to that for the direct method was
Figure 2. Plot of HMX concentrations determined for
computed, and the mean and standard deviation
groundwater samples using SOE with those using SPE-C
was obtained (Table 4). Mean ratios ranged from
and SPE-M.
0.800 for RDX using the SOE method to 1.143 for
TNT using the SPE-C method. Only for 2ADNT
the direct method. In Figures 2, 3, and 4 the con-
was a significant difference among methods de-
centrations of HMX, RDX, and TNT, determined
using SPE-C and SPE-M, are plotted against the
(SPE-C was different from SOE and SPE-M, which
concentrations obtained using SOE. In the absence
were not significantly different from each other).
of bias, the plots should have a slope of 1.00 and
The results of this analysis indicate that, for rela-
an intercept of 0. Regression analyses were con-
tively high concentrations, all three preconcentra-
ducted for the SPE-C vs. SOE and SPE-M vs. SOE
tion techniques produced concentrations similar
individually for each analyte, and the resulting
to that from the direct analysis method, with ana-
lyte recoveries in all cases at or above 80%. These
14
results demonstrate a marked improvement in the
recovery of HMX and RDX using the SDB-RPS
SPE-C vs. SOE
12
SPE-M vs. SOE
membrane relative to that observed in our origi-
nal study where the SDB membrane was used
SPE-C
(Jenkins et al. 1992, 1994). This improvement is
vs. SOE
10
particularly striking for HMX, where recoveries
SPE-M
improved from about 49% to 83%, and appears to
vs. SOE
8
be due to an improvement in retention for polar
compounds resulting from sulfonation of the sty-
rene-divinylbenzene. Recovery of HMX and RDX
6
using the Porapak RDX cartridge remains excel-
lent at 96 and 98%, respectively.
4
Since the value of a preconcentration technique
is to enable determination at concentrations be-
low those that can be determined directly, it is im-
2
portant to evaluate its performance when concen-
trations are below the detection limits of the di-
0
2
4
6
8
10
12
14
rect method. Since the SOE method is the proce-
RDX Concentration by SOE (g/L)
dure currently recommended in SW846 Method
8330, results for SPE-C and SPE-M were compared
Figure 3. Plot of RDX concentrations determined for
with those obtained for SOE for samples with an-
groundwater samples using SOE with those using SPE-C
alyte concentrations below the detection limits of
and SPE-M.
7
Previous Page
