A comparison between chemical and photochemical vapor generation techniques for mercury determination using univariate and multivariate optimization

In this work, analytical methods for Hg determination in fish, using cold vapor (CV) and photochemical vapor generation (PVG), were optimized and compared. Both techniques were coupled to the same atomic absorption spectrometry (AAS) detection system: a compact and low cost flow injection mercury analyzer. Parameters such as carrier solution flow rate, carrier gas flow rate (PVG and CV), formic acid concentration (PVG), HCl and NaBH4 concentrations (CV) were optimized by univariate and multivariate strategies. Under the univariate optimized conditions, the figures of merit were established using CV and PVG techniques and the PVG efficiency (compared with CV) was estimated as 133%. Since this result is not in agreement with those reported in literature, a multivariate optimization was also performed for both techniques. The two-level full factorial design showed that the investigated variables have interaction effects between them, indicating that a multivariate strategy is more adequate for the optimization procedure. Using the Doehlert matrix design new experimental conditions were obtained for CV, leading to an increase of 40% in the Hg sensitivity while the best conditions for PVG did not changed. Using the new optimized conditions for CV, the PVG efficiency was of 96%, which is in agreement with those reported by other authors. Moreover, an improvement in the figures of merit for CV was achieved. Three fish reference materials were evaluated by both techniques and the results showed that the obtained Hg concentrations were not statistically different from certified values, according to Student t-test with 95% of confidence. The optimized methods allowed simple and accurate determination of Hg in fish by CV-AAS and PVG-AAS techniques. Both methods are simple and reliable, being appropriated for routine analysis; however, the PVG should be preferred because of its lower LOD and cost of analysis.