Table 3. Comparison of recoveries from fortified samples.
Percent recovery*
Sample
HMX RDX TNB DNB DNA TNT 24D
4A
2A
B (spike)
Dir
99.5
98.5 95.6
96.5
98.1
SOE
94.2
91.2 92.9
83.2
92.1
SPE-C
99.0
101.0 96.6
94.1
95.1
SPE-M
92.5
95.6 89.3
88.6
86.9
B (spike)
Dir
98.8
98.2 95.9
97.2
99.2
SOE
91.0
95.0 89.0
81.0
89.0
SPE-C
93.5
100.0 89.8
89.1
89.3
SPE-M
88.0
102.0 83.0
78.0
82.0
29
Dir
95
95.5 95.2
92.8 93
SOE
107
89
85
89
65
SPE-C 103
107 104
105 102
SPE-M 80
78
76
78
77
4
Dir
105.5
105 103
104 105
191† 76.0
SOE
23†
83
76
SPE-C 351†
95† 92.2
91.1 93.7
SPE-M 308†
49.5† 87.4
85.6 90.8
* % recovery calculated by subtracting the value obtained from the analysis of
the unspiked sample from the value obtained for analysis of the fortified
sample, dividing by the concentration spiked and multiplying by 100.
† For an explanation of these unreliable estimates, see text.
samples were all analyzed by the direct RP-HPLC
spike added is greater than that already present
method (without preconcentration) and by RP-
in the matrix. If, on the other hand, the spike con-
HPLC after preconcentration using salting-out
centration is small relative to the matrix concen-
solvent extraction (SOE), cartridge solid-phase
tration, a situation encountered here, the result is
extraction (SPE-C), and membrane solid-phase
obtained by subtraction of two similar concentra-
extraction (SPE-M). The following target analytes
tions to obtain a much smaller value. This can re-
were detected in these samples (the number of
sult in widely ranging percent recoveries. This
samples where the analytes were detected in at
problem has been discussed in detail by Provost
least one of the four analyses is given in paren-
and Elder (1983) who stated "Regardless of how
theses): HMX (19), RDX (22), TNB (4), DNB (5), 3,5-
percent recovery is defined, it can be shown that
DNA (6), TNT (11), 2,4-DNT (2), 4ADNT (15), and
percent recovery data tend to be unreliable when
2ADNT (15). Concentrations measured for HMX
the spike/background ratio is small." In sample
4, for example, the concentrations determined in
erally much higher than for the nitroaromatics and
the unspiked sample for HMX were 45.7, 48.0, and
40.8 g/L for the SOE, SPE-C, and SPE-M meth-
aminonitroaromatics.
Although results from the direct method are
ods, respectively. Sample fortification added only
another 2 g/L and hence it is not surprising that
certainly not error free, they are subject to many few-
er sources of error than methods employing a
percent recoveries, calculated as described above,
were 23, 351, and 308% for SOE, SPE-C, and SPE-M,
respectively. As shown by Provost and Elder (1983)
Table 4. Ratio of concentrations obtained for the
various preconcentration methods relative to that
this is not a surprising result and does not indi-
from the direct method.
cate that the methods were out of control. The
excellent results for the other QC samples pro-
Concentration-preconc./Concentration-direct
vide convincing evidence that both the direct and
Analyte
n
SOE
SPE-C
SPE-M
preconcentration procedures were in-control and
0.870 0.188
0.957 0.147
0.833 0.129
HMX
11
providing very reliable concentration estimates.
0.800 0.184
0.975 0.192
0.882 0.158
RDX
13
1.010 0.252
1.143 0.331
1.015 0.244
TNT
4
0.909 0.128
0.996 0.106
0.925 0.095
Comparison of results using groundwater
4ADNT
5
0.865 0.106
1.0210.066*
0.871 0.057
samples from Naval Surface Warfare Center
2ADNT
5
Analytical results for 33 groundwater samples
n = number of ratios in each mean.
from NSWC are presented in Appendix A. These
* Value significantly different at the 95% confidence level.
6
Previous Page
