RESUMO
The conceptual expansion, fast development, and general acceptance of flow analysis are consequence of its adherence to the principles of green and white analytical chemistry, and chemical derivatization plays an essential role in this context. Through the flow analysis development, however, some of its potentialities and limitations have been overlooked. This is more evident when the involved modifications in flow rates, timing and/or manifold architecture deteriorate the analytical signals. These aspects have not always been systematically investigated, and are addressed here in relation to flow analyzers with UV-Vis spectrophotometric detection. Novel strategies for solution handling, guidance for dealing with the aforementioned analytical signal deterioration, and an alternative possibility for exploiting differential aspiration are presented. The concept of blank reagent carrier stream is proposed.
RESUMO
Multivariate calibration involving partial least squares was exploited in the flow-based spectrophotometric determination of molybdenum in river waters relying on the Mo(VI)-catalyzed iodide oxidation by H2O2 under acidic conditions. Two sample aliquots were simultaneously inserted into the carrier stream, and differential pumping was accountable for in-line addition of sulfuric acid to one of them. Pronounced gradients (acidity and reagent concentrations) were established along the complex sample zone formed, and the absorbance-time function was characterized by local maximum and minimum values. As these values were intrinsically more precise, they were used for implementing the PLS multivariate calibration. Mo(VI) and Fe(III) were jointly determined, and Fe(III) interference was straightforwardly circumvented. Influence of reagent concentrations, acidity, available time for reaction development, and nature of the acid was investigated, and this later parameter manifested itself as relevant for discriminating purposes. The calibration set consisted of 6.2 - 50.0µgL-1 Mo(VI) plus 0.5 - 7.0mgL-1 Fe(III) solutions. The PLS model was characterized by good prediction ability [RMSEP = 0.67µgL-1 for Mo(VI)]. The innovation was applied to spiked river waters, and analytical precision, sampling rate, recovery, detection limit and reagent consumption were estimated as 0.5 - 2.4%, 31h-1, 98-114%. 0.88µgL-1 Mo(VI), and 54.0mg KI per determination, respectively. Results were in agreement with ICP OES.