School of Engineering and Science - Jacobs University

error propagation during the complete sample decomposition and ICPMS analytical
procedure. However, for many elements, in particular Y, Ba, the REE, and U, limits
of precision appear to be controlled by the ICPMS measurement, and not by the
sample decomposition method.
For the discussion of precision, three types of analytical precision are defined:
1) sample precision, calculated from the 60 mass scans of a single solution and
representing raw, uncorrected data; (2) run precision, calculated from element
concentrations determined by repeated analyses of a single solution over the duration
of an ICPMS run (typically 6-12 hours); and (3) method precision, calculated from
element concentrations determined from repeated acid decompositions and ICPMS
analyses of a sample powder. The three types of precision and their relationships are
illustrated in Figure 5. Note that sample precision reflects uncorrected intensities
(cps), whereas run and method precision incorporate internal standard, interference,
and blank corrections. To provide the best estimate of precision for analyses of rocks
and minerals, these three types of precision are calculated using data obtained for the
certified reference materials listed in Table 1. The discussion of sample, run, and
method precision is limited to those analyses where the CRM was run repeatedly as
part of every sample batch (see Fig. 3). Data are discussed as percent relative standard
deviation (%RSD), though it must be stressed that standard deviations themselves
usually vary significantly (±50% relative), so that an average RSD of 2% is expected
to typically range as low as 1% and as high as 3%.
Sample precision in ICPMS determinations for common rock types is
presented in Figure 6, along with the average sample precision for the 10 μg/kg
calibration standard. As the 10 μg/kg calibration standard represents a solution free of
the matrix effects typical of dissolved rock samples, it is expected to display the best
RSD values, which are ~2% for almost all monitored isotopes. However, the 10 μg/kg
calibration standard sample precision is indistinguishable from the sample precision
determined for dissolved rock solutions, particularly for rock types that are not tracemetal
poor, such as basalt and shale. Only for rock types that are very low in trace
metals (e.g., JDo-1 dolomite), are sample precisions worse than those observed in the
10 μg/kg calibration standard, and only then for certain elements (e.g., Rb, Mo, Cs,
Hf, Th, U). The run precision, where measurements of a CRM solution are
periodically repeated over the course of an ICPMS run, is generally ~2%, similar to
18