Comparison of nanosecond and femtosecond laser-induced breakdown spectroscopy for determination of U and Th in tantalum–niobium ores

Abstract

The measurement of uranium (U) and thorium (Th) is a critical step in the mining process of tantalum–niobium (Ta–Nb) ores. However, the quantitative analysis of high-Z (atomic number) elements in U polymetallic ores by laser-induced breakdown spectroscopy (LIBS) is usually very difficult because of their complex spectra. To overcome this limitation, a comparison of the quantitative analysis capability between nanosecond LIBS (ns-LIBS) and femtosecond LIBS (fs-LIBS) was performed in this study. A multivariate calibration method based on the partial least squares regression (PLSR) algorithm was coupled with LIBS to solve the interference of the analytical lines and enhance the performance of LIBS analysis. Due to the different excitation sources, ns-LIBS and fs-LIBS exhibit significantly different spectral characteristics, thereby affecting the performance of quantitative analysis. With the assistance of data pretreatment, optimization of the number of principal component (PC) procedures and the selection of the spectral region, the performance of U and Th determination for both ns-LIBS and fs-LIBS was provided by the PLSR algorithm. Compared to ns-LIBS, fs-LIBS provided enhanced predictive capability and accuracy, with the root mean square error of calibration (RMSEC) decreased from 56 ppm to 30 ppm for U and from 192 ppm to 175 ppm for Th, the fitting coefficients (R²) increased from 0.9859 to 0.9959 for U and from 0.9989 to 0.9991 for Th, and all mean relative errors (MREs), root mean square errors of prediction (RMSEPs) and most of the relative standard deviations (RSDs) for U and Th improved for three validation samples. The results illustrate that fs-LIBS combined with PLSR can effectively enhance the accuracy of U and Th determination in Ta–Nb ores.