Pb when the certified concentration was above one
Table 5. Treatment scheme for spiking soil
subsamples with metals.
of the estimates of detection listed in Table 6. In
only two cases were the values established by the
Sets
Metal groups
RF/Comp. Kα normalization method off by more
than 50% from the certified values. An apparently
1
Cr
Cu
Zn
As
Pb
2
Ni
Se
Hg
Tl
Co
high concentration was established for Cu in the
3
Sb
Ag
Ba
Cd
Sn
CRM 021 soil, and a low one for the SRM 2704 river
sediment. Fundamental parameter analysis with
Treatment concentrations
the Spectrace 9000 XRF failed to establish concen-
Subsample
trations within 50% six times, twice each for Cu,
S1
1000*
125
0
500
250
As, and Pb. No values were obtained for Cu in the
S2
500
250
1000
0
125
SRM 2704 river sediment and SRM 2711 soil, while
S3
250
0
125
1000
500
S4
125
1000
500
250
0
low determinations of As occurred for both the
S5
0
500
250
125
1000
SRM 2710 soil and the CRM 020 soil, and high es-
timates were obtained for Pb in the CRM 013 paint
Matrix
chips and the CRM 014 baghouse dust. The FP de-
blank
NF
NF
NF
NF
NF
terminations made with the X-Met 920 failed to
* g/g
meet this criterion in only three cases. A low value
NF Not fortified
was established for As in the CRM 020 and high
values for Pb in both the CRM 013 paint chips and
Table 6. Detection limit estimates and intensity
the CRM 014 baghouse dust.
counts.
The high values obtained for the CRM refer-
Detection limits (g/g)
ence materials by these two methods of XRF analy-
sis are not necessarily incorrect, since the certified
Spectrace X-Met
RF/
Peak
†
Compt.** intensity*†
Source
Metal
9000*
920
value is based on an acid extraction that does not
necessarily represent the total amount present.
FE-55
Cr
180
--
--
--
However, a low determination for these standards,
Cd-109
Cr
525
325
270
1.18
or one that fails to be within 50% of the value
Cd-109
Mn
410
225
--
--
stated for the NIST reference materials, would be
Cd-109
Fe
225
200
--
--
aberrant. The low As concentrations determined
Cd-109
Co
205
180
--
1.33
by FP analysis had previously been identified as a
Cd-109
Ni
125
175
--
3.82
Cd-109
Cu
90
175
54
6.58
problem when samples contain much larger (>10
Cd-109
Zn
70
160
90
6.47
times) quantities of Pb (Harding 1991). The false-
Cd-109
As
50
140
42
3.18
negative Cu determinations were only for samples
Cd-109
Se
35
140
--
18.0
with certified concentrations very close to the es-
Cd-109
Hg
60
--
--
5.33
timates of detection. Overall, these two rapid meth-
Cd-109
Tl
--
--
--
7.83
Cd-109
Pb
30
--
48
8.34
ods of analysis showed that they were fairly insen-
sitive to this wide variety of particulate matrices
Am-241
Ag
--
70
--
25.9
by establishing concentrations that would be ap-
Am-241
Cd
180
100
--
24.6
propriate for the data quality objectives stated.
Am-241
Sn
100
80
--
27.1
Am-241
Sb
65
80
--
33.5
Table 9 shows XRF concentration estimates ob-
Am-241
Ba
20
100
--
31.4
tained for Cu, Zn, and Pb along with the values
obtained by acid-extraction/ICP analysis. The FP
* Minimum detection limit
†
analysis was only performed with the X-Met 920
Minimum determination limit
** Method detection limit
XRF analyzer. In those cases where the values ob-
*† Matrix-corrected peak intensity for 1000-ppm spiked
tained by acid-extraction/ICP analysis were above
RMA soil
the appropriate estimate of detection (Table 6), only
Cu in sample A and Zn in sample F had XRF esti-
mates that were off by more than 50%. A high Cu
metals should be easily quantitated below 1000
g/g by XRF analysis.
value was established by RF/Comp. Kα normal-
Tables 7 and 8 show the concentration estimates
ization analysis, and a high Zn value was estab-
obtained for the commercial reference materials
lished by FP analysis. Again, since these reported
by these two methods of rapid sample analysis.
estimates were higher than those obtained after
These two tables show results for Cu, Zn, As, and
acid extraction and ICP analysis, which is not nec-
5
Previous Page
