Avaliação de modelos de calibração para determinação direta de zinco em grãos de feijão por pXRF

Abstract

Zn is an essential element for human health. However, its deficiency is observed in millions of individuals around the world, especially in underdeveloped countries, causing symptoms such as diarrhea, hair loss, and even leading to death. Therefore, strategies to mitigate this public health issue have been studied and implemented, mainly focused on the biofortification of staple food crops. One of these crops is the common beans (Phaseolus vulgaris L.), one of the most consumed grains in the world, recognized as a natural source of proteins and micronutrients. To achieve this task, analytical techniques are employed for Zn quantification. The aim of this study was to develop and evaluate calibration models using pXRF for direct analysis of grains of common beans, avoiding expensive sample preparation steps. For this purpose, grains of common beans (n = 17) were ground, freeze-dried and pressed into pellets. Operating conditions were optimized, such as the type of filter and the irradiation time. A test sample, with the highest Zn content, were submitted to acid extraction with diluted HCl and HNO3, as a way to prepare the standards of the calibration curves. The ground common bean samples, the sample obtained after extractions and mixing in different proportions to prepare the calibration curves, were also subjected to acid digestion and were analyzed by ICP OES, used as a reference method. The Zn content in the common bean samples analyzed by the reference method varied between 25 ± 1 to 33 ± 5 mg kg-1, which is in accordance with the literature. The optimized operating conditions for pXRF measurements were as follows: use of the Ti/Al filter and irradiation time of 10 s. Three calibration models were built and the obtained results were statistically compared with the Zn reference data. With 95% of confidence level using the Student's t test, the data from the models using the HNO3-extracted samples and the calibration model using certified reference materials presented significant differences with reference values. However, the calibration model built using HCl-extracted samples showed no statistically different data with the reference values. For this model, the linear working range varied between 3 and 35 mg kg-1 Zn with a linear correlation coefficient (r) greater than 0.99. Summing up, depending on the calibration model, pXRF can be recognized as a high potential method for the direct Zn determination in grains of common beans, in a fast and in a non-destructively way.