Analysis of Heterogeneity in Polyglycerol Ester Surfactants

Doctoral thesis, 2026

The properties and performance of materials are strongly influenced by their structure, yet many natural and synthetic materials consist of complex mixtures that vary in composition, size, or structure, collectively referred to as heterogeneity. Understanding this heterogeneity is essential for establishing reliable structure-property relationships, yet it is often difficult to characterize in detail, and its influence on material performance is therefore not always well understood. This thesis investigates heterogeneity from several perspectives, with a primary focus on polyglycerol ester surfactants, in which heterogeneity comes from both the polyglycerol headgroup and the fatty acid tail attachment.

A methodology based on quantitative and multidimensional NMR spectroscopy was developed to characterize and quantify the constitutional heterogeneity of commercial polyglycerol materials. The results showed that polyglycerols consist of mixtures of molecules containing structurally distinct glycerol subunits connected through different linkage patterns, and that their constitutional complexity increases with oligomer size. A simplified approach based on quantitative ¹H NMR was also demonstrated, offering a more rapid alternative for routine characterization.

The influence of heterogeneity on surfactant properties was investigated for polyglycerol ester surfactants, where a varying degree of esterification was considered. Heterogeneity was found to significantly affect solution behavior, adsorption kinetics, and foaming performance, while equilibrium surface tension was comparatively less sensitive to composition. In a complementary study, structurally well-defined branched triglycerol ester surfactants were used to establish structure-property relationships while minimizing molecular heterogeneity. This revealed unexpected effects of molecular architecture on self-assembly and interfacial behavior, highlighting the importance of headgroup structure and intermolecular interactions.

To explore data-driven approaches for characterization of heterogeneity more broadly, machine learning was applied to colloidal gold nanoparticles. Convolutional neural networks were trained to predict particle size distributions directly from UV–vis spectra, successfully extracting information about both average particle dimensions and polydispersity, and demonstrating the potential of machine learning as a tool for characterization of heterogeneous materials.

Overall, this thesis demonstrates that heterogeneity can be quantified using advanced experimental and data-driven approaches, and that heterogeneity can strongly influence surfactant performance.