A mechanism study on laser-induced breakdown spectroscopy and machine learning-based characterization method for...

Area of Science:

• Analytical Chemistry
• Materials Science
• Environmental Science

Background:

• Laser-induced breakdown spectroscopy (LIBS) offers rapid characterization of waste organic polymers (WOP).
• Lack of mechanistic interpretability in LIBS models raises concerns about practical reliability for WOP analysis.
• Understanding fundamental chemical correlations is crucial for advancing LIBS applications in waste management.

Purpose of the Study:

• To systematically investigate chemical correlations between WOP fuel properties and LIBS spectral features.
• To enhance the reliability and practical application of LIBS for WOP characterization.
• To provide theoretical validation for LIBS-based systems in energy recovery and circular economy initiatives.

Main Methods:

• Feature selection to identify key LIBS spectral peaks related to WOP composition.
• Machine learning interpretability analysis to understand spectral feature impacts.
• Model construction using selected key peaks versus raw spectra or principal components.

Main Results:

• High prediction accuracy achieved for carbon (97.74%), hydrogen (91.22%), oxygen (91.28%), and lower heating value (LHV) (97.02%).
• Models using 10 selected key peaks outperformed those using raw LIBS spectra or principal components.
• C2 swan bands were critical for carbon, oxygen, and LHV prediction; H I line was essential for hydrogen prediction.

Conclusions:

• Mechanistic investigation provides theoretical validation for LIBS-based WOP characterization.
• The methodology supports practical implementation in energy recovery and sustainable waste management.
• This work advances efficient resource utilization for a circular economy.