Abstract: To improve the detection accuracy of laser-induced breakdown spectroscopy (LIBS) for heavy metal elements in water, LIBS technique is combined with single variable calibration curve (SVCCLIBS) method and partial least squares regression (PLS-LIBS) method respectively to quantitative analyze Cr, Mn, and Ca in mixed aqueous solutions. The influence of coexisting elements on the detection accuracy of analytical elements is studied by PLS-LIBS. The results show that the detection accuracy of analytical elements is greatly influenced by coexisting elements, and the total relative errors of the prediction of the analyzed element concentrations are significantly reduced when the analytical line intensities of analytical elements and coexisting elements are taken as the input variables of PLS model. The total relative errors of concentration prediction of Cr, Mn, and Ca elements obtained by SVCCLIBS method are 14.3%, 8.46%, and 6.35% respectively, while the relative errors of PLS-LIBS method are improved to 2.30%, 0.74%, and 0.03%, respectively. The linearity R2 of the concentration prediction correlation curve for Mn element is improved from 0.985 for SVCC-LIBS method to 0.999 for PLSLIBS method. The research results indicate that the PLS-LIBS method can effectively improve the detection accuracy of trace metal elements in mixed aqueous solutions.

Key words: spectroscopy, laser-induced breakdown spectroscopy, mixed aqueous solution, metal element; partial least-squares regression, coexisting element, detection accuracy