Amphiphilic Hollow Microgels at the Liquid-Liquid Interface: Studying Soft Colloids' Shape and Interior via Computer Simulations and Machine Learning.

Single amphiphilic network-like polymer particles with random distribution of hydrophilic (A) and hydrophobic (B) segments in the polymer network and internal cavity were studied at an oil-water interface by means of mesoscopic computer simulations. Effects of the cavity size, interfacial tension between the liquids, their selectivity as solvents toward species A and B, and the degree of incompatibility between the A and B units on the internal microgel structure and distribution of the liquids are considered. Similar to the regular (cavity-free) counterparts, the intrashell mixing of two immiscible liquids was found for the hollow particles. However, the presence of a cavity filled with one of the liquids caused some unique differences, and for the symmetric composition (equal fractions of A and B segments), the adsorption pathway of the hollow microgels appeared to predefine their position with respect to the interface and the sort of the liquid inside the cavity. In turn, the machine learning has been used to establish structure-property relationships of the simulated systems. As a result, most notable parameters and their combinations affecting (i) lateral and normal sizes of the adsorbed microgels, (ii) their position along the normal to the interface, and (iii) the integrity of the cavity were established. Subsequently, the developed framework was used to predict the presented values of the microgels suitable for different interfacial scenarios. Notably, a boundary case with a periodically collapsing cavity has been found.