High-resolution continuum source quartz tube atomic absorption spectrometry for the determination of As, Sb, Bi, Hg, Se and Te in food and environmental matrices after chemical vapor generation†

This study presents a broadly applicable spectrometric method for the determination of As, Sb, Bi, Hg, Se, and Te in food and environmental matrices based on chemical vapor generation high-resolution continuum source quartz tube atomic absorption spectrometry as a sequential method. The samples were subjected to microwave-assisted digestion, followed by the pre-reduction of As(V) and Sb(V) with 0.05 mol L^–1 thiourea in a 0.5 mol L^–1 HCl medium and Se(VI) and Te(VI) in a 7 mol L^–1 HCl medium. Chemical vapor was generated from an aliquot volume of 5 mL sample in 0.5 mol L^–1 HCl for As, Sb, Bi and Hg and 7 mol L⁻¹ for Se and Te by the addition of 3.5 and 2 mL of 2.5% (m/v) NaBH₄ solution stabilized in 0.1% (m/v) NaOH for Hg, Se, and Te and As, Bi, and Sb, respectively. Three pretreatment methods, namely, (i) addition of 1% (m/v) sulfamic acid; (ii) N₂ purging of the solution for 20 min; and (iii) addition of 1% (m/v) sulfamic acid followed by 10 min N₂ purging, were investigated for the elimination of nitrite and NO_x non-spectral interferences in the Se and Te determination. However, it was observed that pre-washing the reaction cell and the quartz tube atomizer with 6 L min^–1 argon for 20 s (As, Bi, Se, Te) and 30 s (Sb), after sample introduction into the reaction cell and before NaBH₄ solution addition, was crucial for the elimination of spectral interferences from residual NO_x and O₂, regardless of the sample pretreatment method. An increase in pre-washing time resulted in a decrease in the signal for all elements, indicating that the presence of O₂ traces is beneficial for high sensitivity. The limits of detection were (mg kg⁻¹) 0.031 (Hg); 0.016 (As); 0.015 (Bi); 0.008 (Sb); 0.084 (Se); and 0.030 (Te). The analysis of certified reference materials indicated recoveries of 98–103% and an expanded uncertainty of ±(17–18)% (k = 2, 95% confidence level). The z′ or z scores indicated a satisfactory performance of the method. The precision, evaluated from extended uncertainty (k = 2, n = 3) by analysis of real samples, was in the range of 4–10.7%.