Desenvolvimento de software e modelos para análise de nutrientes em material vegetal utilizando espectroscopia de emissão óptica com plasma induzido por laser (LIBS)

Abstract

With the increasing population and demand for food, increasing crop productivity without expanding agricultural land is a crucial challenge. Soil chemical analysis and foliar diagnosis are important tools to ensure healthy plant development, assisting in the decision-making process for the appropriate and sustainable application of inputs, especially fertilizers, amendments, and pesticides. In this context, the LIBS technique stands out for allowing quick and multi-elemental analysis with minimal or no sample preparation. LIBS also has a wide range of configurations, with the most common being single-pulse (SP-LIBS) and double-pulse (DP-LIBS) configurations. However, analyzing data generated by LIBS is complex, requiring automation. This work presents the development of a software called LIBSsa, which automates the analysis of LIBS data to develop quantification models of macro and micronutrients in sugarcane, coffee, and soybean leaves. The software was created in Python, a widely used language in both scientific and corporate environments, mainly for being open-source and offering a wide range of scientific tools. In addition, protocols for foliar analysis with LIBS were proposed, using two sets of samples: Set 1, measured in a DP-LIBS system, and Set 2, measured in an SP-LIBS system. Promising results were obtained, where for Set 1 samples, quantification models were obtained for Ca, Mg, Mn, and P, with calibration and validation correlations above 83%, while for Set 2 samples, quantification models were obtained for Al, Cu, Fe, K, and Zn, with calibration and validation correlations above 70%. All results were obtained using LIBSsa, which proved to be intuitive and computationally efficient, in addition to having open-source code, offering a promising path for quick and accurate nutrient analysis in plants using LIBS.