Chemometrics Approach Based on Wavelet Transforms for the Estimation of Monomer Concentrations from FTIR Spectra

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Chemometrics

Background:

• Fourier-transform infrared (FTIR) spectroscopy is valuable for identifying functional groups and molecules in organic mixtures.
• Quantitative analysis of FTIR spectra is challenging due to overlapping peaks of varying widths, hindering accurate concentration prediction in chemical reactions.

Purpose of the Study:

• To develop a chemometric approach for accurate and interpretable prediction of component concentrations in chemical reactions using FTIR spectra.
• To address the limitations of traditional methods in handling overlapping spectral peaks.

Main Methods:

• A novel chemometric method combining wavelet transform for spectral decomposition and sparse linear regression for quantitative analysis.
• Wavelet transform decomposes FTIR spectra into peaks of different widths, followed by building a sparse linear regression model using wavelet coefficients.
• Model interpretability is achieved through regression coefficients visualized on Gaussian distributions, linking spectral regions to predictions.

Main Results:

• The proposed method demonstrated superior predictive ability for monomer concentrations in copolymerization reactions compared to various conventional linear and non-linear regression techniques.
• Rigorous validation confirmed the enhanced predictive performance.
• Visualization of regression coefficients provided insights consistent with other chemometric and qualitative evaluations.

Conclusions:

• The developed chemometric approach effectively predicts monomer concentrations in copolymerization reactions using FTIR data.
• The method offers improved interpretability, aiding in the understanding of spectral data relationships.
• This technique is a valuable tool for quantitative analysis and spectral interpretation in complex chemical systems.