# Chemometric analysis of ethoxylated polymer products using extracted MALDI-TOF-MS peak distribution features

## TL;DR

This paper introduces a chemometric method to analyze ethoxylated polymers using MALDI-TOF-MS by extracting and weighting key peak distribution features for better interpretation and comparison.

## Contribution

The novel approach uses ratio-scaled extracted features from MALDI-TOF-MS peak distributions to improve chemometric analysis of ethoxylated products.

## Key findings

- Ratio-scaled features outperform full spectra analysis in identifying subtle compositional differences.
- The method enables auto-assignment of peak distributions to possible compounds, improving interpretability.
- The approach is demonstrated on Polysorbate 80, PEG castor oil, and Oleth-20 with clear results.

## Abstract

MALDI-TOF MS (matrix-assisted laser desorption/ionization time-of-flight mass spectrometry) of ethoxylate products produces spectra with distributions of regularly spaced peaks resulting from the addition of monomer units of ethylene oxide to the oligomer. We show that overlapping peak distributions from the different ethoxylated constituents of natural raw materials can be resolved, so that features of the individual distributions (m/z at distribution maximum, intensity at the distribution maximum, width of the distribution at half height, and ratio of the distribution to the major peak distribution) can be extracted and used with statistical pattern recognition techniques to study ethoxylated products. Crucially, we weight the extracted features, so that features from a distribution with a high ratio to the main distribution are given more importance (‘ratio-scaled’). We exemplify the method by characterizing the structural variation between types of compositionally diverse Polysorbate 80, PEG castor oil and Oleth-20, and compare the chemometric analysis using our extracted features with analysis of the full spectra. We demonstrate that using ratio-scaled extracted features gives superior results to the full spectrum, both in terms of identifying subtle compositional differences that would otherwise be missed, and in interpretability. Importantly, the integrated auto-assignment of peak distributions to possible compounds allows the results to be reported in terms of the most abundant oligomers of the raw material constituents. This simplification facilitates interpretation of the results and allows the comparison of closely related products.