Carbon quantification in soils with different textures using laser-induced breakdown spectroscopy: spectral interference correction and use of a 3D plane model.

Soil carbon (C) determinations have been widely studied due to soil C sequestration that contributes to the mitigation of greenhouse gas emissions and improves soil quality. However, traditional chemical processes for large-scale analysis generate waste, are time-consuming, and have a high cost per measurement. Laser-induced breakdown spectroscopy (LIBS) is a multi-element spectroanalytical technique that allows fast and low-cost analysis, almost no sample preparation is required, and does not generate hazardous chemical waste. Two emission lines are commonly used for LIBS C determination, 193.03 and 247.85 nm. However, Brazilian soils have a high concentration of aluminum (Al) and iron (Fe), directly interfering in those C emission lines. Furthermore, multiple soil textures increase the difficulty of building calibration models due to matrix effects. In the present work, a mathematical model is proposed to quantify the total C in soil samples having different textures bypassing spectral interferences. A LIBS-specific method for removing outliers has been developed with 6% spectrum removal. From the univariate analysis, it was noticed that some results were projections of a 3D surface in a 2D space, so a 3D plane model was obtained with good fits for the evaluated C emission lines, R2 > 0.91, with limits of detection of 0.11% and 0.13% and limits of quantitation of 0.11% and 0.32% for lines 193.03 and 247.85 nm, respectively. Three repetitions were used to test the robustness of the methods and presented an R2 of 0.95 and 0.93, a mean error of about 20.38% and 24.12% for lines 193.03 and 247.85 nm, respectively, and a root mean square error of prediction lower than 0.40% for both lines.

Direct determination of nutrient elements in plant leaves by double pulse laser-induced breakdown spectroscopy: evaluation of calibration strategies using direct and inverse models for matrix-matching.

This study aims to develop a single calibration model to determine nutrient elements directly (Ca, Mg, Mn, and P) in soybean and sugar cane leaf samples by double pulse laser-induced breakdown spectroscopy (DP LIBS). Matrix-matching calibration (MMC) was evaluated using direct and inverse models. Forty-five samples were used to build the calibration model (23 soybean leaves and 22 sugar cane leaves), and fifteen were used for the prediction test (8 soybean leaves and 7 sugar cane leaves) models. In the direct model, the analyte concentration in the sample is the independent variable, and the analytical signal is the dependent variable. In the inverse model, the analytical signal is the independent variable, and the analyte concentration in the sample is the dependent variable. In general, both models presented satisfactory results; however, the inverse model performed better. Emission lines used to propose calibration models were selected using a linear Pearson's correlation (R) strategy between each spectral point and the Ca, Mg, Mn, and P concentration measured by reference methods using inductively coupled plasma optical emission spectrometry (ICP OES). The root mean square errors of prediction (RMSEP) for the direct models were 0.60 g kg-1 to (Ca), 0.47 g kg-1 (Mg), 9.3 mg kg-1 to (Mn), and 0.28 g kg-1 to (P); for inverse model was 0.55 g kg-1 to (Ca), 0.39 g kg-1 (Mg), 10.5 mg kg-1 to (Mn), and 0.21 g kg-1 to (P). The calibration strategies proposed in this study may minimize matrix effects in direct solid analysis in soybean and sugar cane leaf samples, performing the determination of Ca, Mg, Mn, and P by DP LIBS using a single calibration model.