Raman Spectroscopic Classification of Polyethylene Glycol Samples of Varying Molecular Weights Using Machine...

Area of Science:

• Polymer Science
• Spectroscopy
• Chemometrics

Background:

• Polyethylene glycol (PEG) is a versatile water-soluble polymer (WSP) with properties influenced by molecular weight.
• Accurate characterization of PEG molecular weight is crucial for its diverse applications.

Purpose of the Study:

• To investigate the efficacy of Raman spectroscopy combined with machine learning for differentiating PEG samples based on molecular weight.
• To identify spectral features indicative of molecular weight variations in PEG.

Main Methods:

• Confocal Raman microscopy was used to analyze eight PEG samples with molecular weights from 1000 to 35,000 g/mol under standardized conditions.
• Supervised machine learning algorithms, including Support Vector Machine (SVM) and Principal Component Analysis-Linear Discriminant Analysis (PCA-LDA), were employed for data analysis.

Main Results:

• A Support Vector Machine (SVM) classifier achieved 93.4% accuracy in cross-validation and 72.6% on an independent test set.
• Discriminative spectral information was primarily found in spectral line-shape and shoulder regions, indicating molecular-weight-dependent conformational changes.
• Raman spectroscopy demonstrated reproducibility and non-destructiveness in distinguishing PEG samples of varying molecular weights.

Conclusions:

• Raman spectroscopy, coupled with machine learning, offers a viable method for molecular weight differentiation of polyethylene glycol.
• The findings highlight the presence of molecular-weight-dependent vibrational signatures in PEG Raman spectra.
• This approach provides a foundation for advanced quantitative spectral analysis and broader industrial applications.