Comparing laser induced breakdown spectroscopy, near infrared spectroscopy, and their integration for simultaneous multi-elemental determination of micro- and macronutrients in vegetable samples.

Laser-induced breakdown spectroscopy (LIBS) is an appealing analytical technique for simultaneous multi-elemental analysis. Near-infrared spectroscopy (NIRS) has also been suggested for the same purpose, mainly for vegetable samples. However, LIBS has failed to provide adequate results in many cases due to sample matrix complexity, and NIRS performance is harmed because of its lack of sensitivity and indirect correlation with inorganic elemental species. In this work, the performance of these two techniques are compared for the determination of micro- and macroelements in vegetable samples (Brachiaria forages) using multivariate regression. In addition, a data fusion scheme, in which spectral data sourced by NIRS is integrated with LIBS, is proposed to improve elemental content determination in those samples. The information of the molecular composition detected by NIR vibrational spectroscopy was consistently selected by recursive partial least squares to yield quantitative multivariate models for K, Ca, Mg, Mn and Fe in forage plants that are superior to models based on the use of individual NIRS and LIBS spectral information. While all data fusion models showed better predictive accuracy than any of the two individual techniques, best results were observed for Ca. This suggests that matrix composition affects each element determination by LIBS distinctively and supports the idea that a successful quantitative data fusion strategy for LIBS requires a technique such as NIRS which is sensitive to this variability.