A Q-switched solid state Nd:YAG laser operating at a third harmonic (355 nm) wavelength and an echelle spectrograph coupled with an ICCD system were used to study the plasma on a glass target. The present work is mainly focused on the investigation of multivariate calibration methods like principal component regression (PCR) and partial least squares regression (PLSR) for the analysis of Mn in complex matrices like glass. The glass studied has Mn as an analyte of interest whose doping concentration in the matrix varies from 0.77% to 11.61%. The performance of univariate and multivariate methods has been presented in this paper through their figures of merit. Improved prediction accuracy, limit of detection (LOD) and regression coefficients (R2) have been reported when the data were analyzed using PCR and PLSR. The calibration curves of six emission lines of Mn have been analyzed using a univariate method that resulted in R2 values varying from 0.85 to 0.98. This method resulted in a correlation uncertainty of 10% and a LOD of 0.20 wt%. R2 values of 0.98 to 0.99 have been obtained for the multivariate calibration curves of Mn analyzed in three selected regions of the LIBS spectrum. The optimum LOD and root mean square error of prediction (RMSEP) using PCR and PLSR were found to be 0.02 wt% and 0.54 wt%, respectively. The significant improvement in the analytical performances of multivariate calibration methods for the investigation of LIBS data is evident from the aforementioned results. Finally, the results of PCR and PLSR were confirmed by PCA classification.
|Number of pages||8|
|Publication status||Published - 21-10-2016|
All Science Journal Classification (ASJC) codes
- Analytical Chemistry
- Chemical Engineering(all)