A Novel LIBS-Machine Learning Strategy for Multimetal Detection in Microsized PMMA Particles: Efficient...

Area of Science:

• Environmental Science
• Analytical Chemistry
• Spectroscopy

Background:

• Microplastics (MPs) are significant environmental pollutants, capable of adsoring heavy metals.
• Composite pollution from MPs and associated contaminants poses a growing ecological threat.

Purpose of the Study:

• To develop a novel, simultaneous quantitative detection method for heavy metals in microplastics.
• To combine laser-induced breakdown spectroscopy (LIBS) with machine learning for enhanced analytical accuracy.

Main Methods:

• Utilized laser-induced breakdown spectroscopy (LIBS) for elemental analysis of microplastic samples.
• Applied partial least-squares (PLS) regression models combined with preprocessing and variable selection techniques (SNV-CARS-PLS, WT-CARS-PLS).
• Investigated the impact of preprocessing and variable selection on model performance using metrics like R², RMSE, and MRE.

Main Results:

• Achieved high correlation coefficients for Cr (Rp² = 0.9750), Pb (Rp² = 0.9759), and Cu (Rp² = 0.9088) using optimized PLS models.
• Demonstrated significant reductions in prediction errors (RMSEp, MREp) and high RPD values (≥20.4) for all three metals.
• Established low limits of detection (LODs) for Cr, Pb, and Cu, all below 1.534 ppm.

Conclusions:

• The developed SNV/WT-CARS-PLS method significantly enhances the accuracy of quantitative heavy metal analysis in microplastics.
• This approach provides robust theoretical and technical support for monitoring and preventing composite pollution associated with microplastics.
• The findings are crucial for understanding and mitigating the environmental impact of microplastic contamination.