Laser-induced breakdown spectroscopy analysis of alloying elements in steel: Partial least squares modeling based on the low-resolution spectra and their first derivatives

Abstract

One of the unique laser-induced breakdown spectroscopy (LIBS) platforms, a compact device, has been considered for use in various applications because of portability and cost effectiveness. The performance of a simple LIBS device assembled with a compact, low-power, diode-pumped solid-state laser and a miniature, non-gated spectrometer was investigated for analysis of low-alloy steel samples. The low spectral resolution (∼0.9 nm) precluded observation of any single resolved analyte lines for Mn, Cr, Ni, and Ti. However, partial least squares-regression modeling extracted the unresolved spectral features correlated to the analyte concentrations successfully. For the highly congested wavelength regions selected for modeling Ni and Ti concentrations, finding reliable location of baseline to be subtracted was very difficult. This could be circumvented by using the first derivatives of LIBS spectra. Our results indicate that the first derivation of LIBS spectra can be used to further improve the analytical performance of a simple, low-resolution LIBS device. Additionally, this implies that appropriate data preprocessing can assist in downsizing a LIBS spectrometer, especially for the sample with matrix elements showing dense rich emission-line structure.