High-Efficiency Prediction of Surfactant Properties through Automated Hydrophilic-Hydrophobic Partitioning

Predicting surfactant properties from structurally diverse molecules remains challenging and limits the rational design of new surfactants. This difficulty arises from the complex interplay between hydrophilic head and hydrophobic tail domains, which govern self-assembly and interfacial behavior. This work presents an automated decomposition-based framework that identifies surfactant head and tail domains and computes interpretable, domain-specific structural descriptors. These descriptors are used in data-driven models to predict key properties, namely critical micelle concentration (CMC), surface tension, and adsorption efficiency, across a wide range of surfactants, including Gemini architectures. Accurate predictions with reduced computational cost and improved extrapolation are achieved. The same descriptors enable clustering analysis of head and tail structures to identify combinations that yield optimal property profiles, supporting the rational design of new surfactants with targeted functionalities.