113 / 390
Table 7:
Reference values of boron an phosphorus determination as result of laboratory
comparison after critical assessment [26,27] according to results of new microwave assisted
procedure quantified by MMC, standard deviations include the error propagation over the
whole method, n = 4
reference values
results of lab comparison
found values
by MMC
reference values
results of lab comparison
found values
by MMC
boron contents
phosphorus contents
p' n'
′
Z
/ μg g
-1
Z
/ μg g
-1
p' n'
′
Z
/ μg g
-1
Z
/ μg g
-1
Ref. 2A
15 60
53.0
± 2.6
53.2
± 0.9 13 52
22.6
± 2.1
22.0
± 2.9
Ref. 2B
14 56
42.7
± 2.6
41.5
± 0.2 14 56
34.2
± 2.6
34.0
± 1.6
Ref. 2C
14 56
71.7
± 4.2
71.1
± 1.4 13 52
25.6
± 2.4
26.3
± 1.2
Ref. 3A
19 76
31.4
± 6.1
33.0
± 0.4 15 60
23.1
± 2.0
23.6
± 0.9
Ref. 3B
18 72
53.1
± 5.4
53.6
± 1.1 15 60
14.0
± 1.6
14.6
± 0.5
Ref. 3C
19 76
11.0
± 1.5
10.7
± 0.4 14 56
42.0
± 4.8
44.5
± 1.4
Table 7 shows an excellent agreement between our values and the reference values
obtained in the interlaboratory comparisons. It demonstrates that this new method
provides highly accurate and precise analytical results. The fact of slightly higher
boron contents found in Table 6 is not confirmed by the results in Table 7. The higher
contents of boron of the silicon samples and the lower uncertainty of the reference
values contribute to the better agreement with our found values.
In addition to B and P all interlaboratory comparisons tested several other trace
element impurities, which substantially coincides with the scope of testing (cf. Table
3). None of the element contents determined by the new method had been identified
as an outlier in the interlaboratory comparisons. Our analytical values passed all
statistical tests and therefore represent partial results of the reference values.
Conclusion
The present paper describes a new and fast method for the determination of trace
elements in MG-Si. The silicon samples are digested in a microwave-assisted high-
pressure system and then analyzed without prior removal of matrix components
directly by ICP-OES. Digestion in a closed vessel system minimizes losses of
phosphorous during the dissolution of silicon. The avoidance of the time-consuming
critical step of matrix evaporation prevents significant losses of volatile boron species.
The newly developed method is statistically validated, whereby its proof of
practicality for the determination of trace elements in silicon, particularly boron and
phosphorus, is demonstrated. All analytical results are supplied with an uncertainty
that is calculated by a consequent application of the Gaussian error propagation
including uncertainty contributions that include the weighing errors for each dilution
step and the initial sample masses, the confidence intervals of calculated
concentrations for each selected emission line and the average concentration per
element over the selected emission lines with their individual confidence intervals.
The new method allows a precise and accurate determination of impurities in silicon
with a purity of 99.996%.
Acknowledgement
The authors thanks especially Dr. Gilles Humbert from FerróPem (France) for the
initiation of the interlaboratory comparisons for determining boron and other trace
elements in silicon, as well as all participants of the round robin tests for the
permission to share their results.
105