Slit LIBS: a novel strategy to improve the efficiency of soil nutrient measurement from a stand-off plasma collection distance
摘要
A black slit-supported Remote - Laser Induced Breakdown Spectroscopy (R-LIBS) method, combined with a machine learning algorithm, is utilized to determine the concentration of available soil nutrients (N, P, K, Ca, Mg, S, B, Zn, Fe, Cu, and Mn). The experiments are conducted by varying the stand-off collection distance (SoD) from 0.6 m to 2 m. The machine learning algorithms employed in the study include Multilinear Regression (MLR), Support Vector Regression (SVR), Partial Least Squares Regression (PLSR), Least Absolute Shrinkage and Selection Operator (LASSO), and Gaussian Process Regression (GPR). To reduce LIBS data variability, a black slit setup is positioned over the soil to block plasma reflections from the surface. Additionally, data fluctuations are minimized by calculating the ratio of the strength of the elemental peak to that of the 553 nm (Mo I) emission line. The proposed approach enables precise estimation of soil nutrients with the assistance of the machine learning algorithms. After extensive experimentation, emission peaks corresponding to neutral and singly ionized atoms associated with soil nutrients are detected in the LIBS spectrum, both with and without the slit arrangement, regardless of variations in stand-off collection distance. The amount of plasma emission collected by the detector diminishes with an increase in SoD. The relative standard deviation (RSD) of the plasma emission detected decreases as the stand-off distance (SoD) increases, going from 12