Multivariate quantitative analysis of metal elements in steel using laser-induced breakdown spectroscopy

Area of Science:

• Analytical Chemistry
• Materials Science
• Spectroscopy

Background:

• Laser-induced breakdown spectroscopy (LIBS) offers rapid, in situ elemental analysis of metal samples without preparation.
• Severe spectral interference poses a significant challenge for accurate quantitative analysis in LIBS.
• Existing methods struggle to fully account for multivariate factors affecting spectral data.

Purpose of the Study:

• To evaluate and compare the effectiveness of different quantitative analysis methods for metal samples using LIBS.
• To investigate the performance of linear (PLSR) and nonlinear (SVR) models in mitigating spectral interference.
• To determine the optimal method for accurate elemental quantification in steel analysis via LIBS.

Main Methods:

• Quantitative analysis of Manganese (Mn), Chromium (Cr), Vanadium (V), and Titanium (Ti) in metal samples.
• Application of single-variable calibration, Partial Least Squares Regression (PLSR), and Support Vector Regression (SVR) models.
• Inclusion of spectral interference lines within the PLSR (linear) and SVR (nonlinear) models.

Main Results:

• The nonlinear SVR model demonstrated superior quantitative analysis performance.
• SVR achieved high R-squared values (0.990-0.995) and low root-mean-squared errors (0.011-0.045) for the analyzed elements.
• SVR exhibited minimal element bias, outperforming PLSR and single-variable calibration.

Conclusions:

• Nonlinear quantitative analysis models, specifically SVR, effectively suppress spectral interference, background noise, and self-absorption in LIBS.
• SVR provides a robust and accurate method for elemental quantification in the steel industry using LIBS.
• The study highlights the advantage of nonlinear modeling for complex spectral data in LIBS applications.