Comparison of three multivariate calibration methods as an approach to arsenic speciation by HG-AAS

This paper compares three multivariate calibration methods, namely classical least squares (CLS), inverse least squares (ILS) and Kaiman filter, applied to continuous-flow hydride generation with sodium tetrahydroborate(III) reducing agent and AAS detection for the purposes of speciation of As(III), As(V), monomethylarsonic acid (MMA), and dimethylarsinic acid (DMA). The precision of the three multivariate methods was compared in the calibration and prediction steps by standard error of prediction (SEP) and relative error of prediction (REP), respectively, for each analyte and no significant differences have been found for As(III), As(V), MMA, and DMA when the F-test was applied to compare the three multivariate calibration methods in pairs at the 95% confidence level. Determination of the As species was carried out in spiked drinking and sea water in the range 7–35 g 1^–1. Recoveries were in all cases around 100% and the detection limit for the least sensitive species was close to 5 g 1^–1.