Laser-induced breakdown spectroscopy and digital image data fusion for determination of the Al, Ca, Fe, Mg, and P in mineral fertilizer: overcome matrix effects in solid direct analysis

Abstract

Laser-induced breakdown spectroscopy (LIBS) and digital images were evaluated in the modeling for the prediction of Al, Ca, Fe, Mg, and P contents in mineral fertilizer samples. For modeling, univariate [matrix-matching calibration (MMC)] and multivariate [multiple linear regression (MLR) using only LIBS data, and data fusion (LIBS + digital image)] calibration strategies were evaluated. The predictive capacity of the models was increased in the following order: MMC<MLR (LIBS) < data fusion. Compared with the MMC and MLR (LIBS data only), the root mean square error (data fusion) values were 17% to 80% lower, demonstrating the improvement in accuracy.