Quantitative trace element analyses of silicate reference materials and a stainless steel using the proton microprobe

The accuracy of and detection limits for trace element analysis by the proton microprobe were determined using a stainless steel sample Gk-19 and glass samples prepared from whole rock reference powders JA-1 andesite, JB-1a basalt, JF-1 feldspar and JGb-1 gabbro. In the steel sample, the measured concentrations of the major transition metals (Cr, Mn, Fe, Co, Ni and Cu) agreed with the recommended values within an average uncertainty of 3%. Thus, the analytical software for the proton microprobe showed an excellent correction for the secondary fluorescence effect and executed appropriate theoretical calculations for accurate quantification. The measured concentrations of all the other trace elements also agreed with recommended values within an average uncertainty of ±7%. For the glass samples, the measured contents of trace Ni in JB-1a and JGb-1 agreed with the recommended values within uncertainties of 14% and 3%, but the contents of trace Mn and Cu in all the samples were an average of 30% and 40% lower than the recommended values, respectively. The discrepancy in Mn and Cu contents is considered to be due to an effect of overlap of the low-energy side tail of the intense FeKα peak for the MnKα peak and a loss of Cu during preparation of the glasses, respectively. The concentrations of all the trace elements, except Cu, agreed with the recommended values within ±10% (average) for JA-1, JB-1a and JF-1, and ±15% for JGb-1. In addition, almost all trace elements were detected at levels of only a few micrograms per gram.