The species of Cr(III) and Cr(VI) in water samples were determined by flow injection on-line preconcentration and separation on two-microcolumn system-derivative flame atomic absorption spectrometry during a collaborative analysis for certification. The Cr(III) and Cr(VI) in water samples were retained on two microcolumns with ion exchange resin and were eluted directly to nebulizer by 15% HNO3 and 8% NH4NO3, respectively. The characteristic concentration (at the sensitivity grade of 2 mV min-1 for 1 min of preconcentration time) for Cr(III) and Cr(VI) were 0.130 and 0.0985 μg l-1, in the order which were 332- and 431-fold better than those of FAAS, and 45- and 47-fold better than those of FI-FAAS, respectively. The relative standard deviations were 3.27% and 3.66% with corresponding detection limits (3σ) of 0.244 and 0.235 μg l-1, respectively. The linear ranges of determinations for Cr(III) and Cr(VI) were 0~100 μgm l-1 with correlation coefficients of 0.9984 to 0.9996. The satisfactory recovery of 94.4%~106% for Cr(III) and Cr(VI) could be obtained from water samples.

Hydride generation atomic fluorescence spectrometry (HG-AFS) has been used for determination of hydride-forming elements because of its high sensitivity, simplicity, and low costs, but most of such work has been concentrated on single element analysis, and reports dealing with multielement determination by HG-nondispersive (ND)AFS are rare. In this work, a sensitive HGNDAFS method was developed for simultaneous determination of trace cadmium and arsenic in biological materials. The conditions for the generation of volatile cadmium and arsenic species from the reaction with KBH4 in aqueous solution were investigated using a double-channel AFS integrated with an intermittent flow reactor. Like thiourea and Co(II), ascorbic acid was found to significantly enhance the generation efficiency of volatile Cd and As species. The interferences of coexisting ions were evaluated. Under optimal conditions, the detection limits for Cd and As were determined to be 10 and 150 ng L-1, respectively. The precision for 11 replicate determinations at the 1 μg L-1 Cd level and the 10 μg L-1 As level were 3.5 and 2.7% (RSD), respectively. The recoveries of spike analytes in the biological samples studied ranged from 94 to 109%. The proposed method was successfully applied to the simultaneous determination of Cd and As in a variety of biological samples.