Automatic LIBS baseline correction by physical-constrained airPLS method: a case of equivalence ratio measurement in high temperature after-burn gas

When laser-induced breakdown spectroscopy (LIBS) was used in online equivalence ratio measurements, the spatial–temporal temperature gradient would lead to significantly varied baselines, which require real-time corrections. In this study, the correlation relationship between spectral intensity and broadband radiation was utilized as the physical constraint in automatic baseline correction. The pixel-to-pixel correlation coefficient on the wavelength ( \({R}^{2}\) $R^2$ ) was introduced as the “Correlation Spectrum”, and the summation of all correlation coefficients ( \(sum{R}^{2}\) $sum{R}^2$ ) was used as an objective function. By maximizing \(sum{R}^{2}\) $sum{R}^2$ using Particle Swarm Optimization algorithm, the key parameters in the adaptive iteratively reweighted Penalized Least Squares (airPLS) method were automatically optimized. The baselines fitted by airPLS could adapt to the plasma excitation variations in a temperature range from 1458 to 1705 K, which improved accuracy in equivalence ratio measurements. The potential advantages of Correlation Spectrum in weak peak identification and broadening effect evaluation were also demonstrated. The benchmarked results suggested that the proposed approach can be conveniently employed in online LIBS measurements and improve its quantification performance.