Trace and ultratrace analysis of gallium arsenide by different mass spectrometric techniques

Abstract

The capability of different solid-state mass spectrometric methods [spark source mass spectrometry (SSMS), laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), radiofrequency glow discharge mass spectrometry (rf GDMS) and secondary ion mass spectrometry (SIMS)] in comparison with inductively coupled plasma mass spectrometry (ICP-MS) was investigated by the trace analysis of GaAs. For trace analysis using solid-state mass spectrometry, a semiconducting laboratory GaAs standard (using high-purity GaAs) doped with Zn, B, Si, Ge, Sn, Sb, P, S, Se and Te (in the µg g^–1 concentration range) was prepared by the liquid encapsulation vertical Bridgeman technique. A selected piece of the synthetic laboratory GaAs standard was investigated directly by SIMS, SSMS, rf GDMS and LA-ICP-MS. For the quantification of SIMS measurements single element ion-implanted GaAs certified reference standards were used. After dissolution of the GaAs sample in high-purity HNO₃–H₂O₂, the concentrations of the doped elements were measured by ICP-MS and inductively coupled plasma atomic emission spectrometry (ICP-AES). By using the results of SIMS, ICP-MS and ICP-AES for the selected piece of the synthetic laboratory GaAs standard, relative sensitivity coefficients (RSCs) of the elements in SSMS, rf GDMS and LA-ICP-MS were determined. The experimentally determined RSCs were used for correcting measured concentrations in an unknown GaAs sample.

In order to reduce matrix effects in ICP-MS, a procedure for complete GaAs matrix separation in a chlorine–argon stream at 280 °C was evaluated. The recoveries of 24 elements after the chlorination of GaAs were determined to be nearly 100% (except for Sn and Ta). Ultratrace analysis of semiconducting GaAs after matrix separation was carried out by double-focusing sector field ICP-MS. The detection limits of ultratrace elements in GaAs after matrix separation (in the low ng g–1 concentration range) were better by about one order of magnitude compared with measurements without matrix separation and were comparable to those obtained by solid-state mass spectrometry.