Presentation description

Recent molecular dynamics simulations have shown that supercooled aqueous solutions with low enough ion concentrations (<20%) can undergo nanophase segregation into regions of solute-having high-density liquid water (HDL) and regions of solute-less low-density water liquid (LDL). This phenomenon is a result of water preference to form the four coordinated LDL expelling the solutes into HDL regions, has been observed with the coarse-grained water model monoatomic water (mW). Because the mW model, based on the Stillinger-Weber potential, does not represent well the behavior of water under pressure, in this work we embarked on a journey to implement nanophase segregation of a supercooled aqueous solution in the machine learning coarse-grained water model ML-BOP, which reproduces well the pressure dependence of the melting point of ice and other water properties. Modeling water with ML-BOP and the solutes with the Stillinger-Weber potential at an ion concentration of 5% we are able to showcase how nanophase segregation is a robust feature of supercooled aqueous solutions, and how the extent of phase segregation depends on the strength of water-ion interactions and the cooling rate. Our final goal is to exploit the ability of ML-BOP to accurately model water at high pressures up to 0.2GPa, to study the interplay between the liquid-liquid phase separation observed in pure ML-BOP at high pressures with the formation of the two different density liquids characteristic of nanophase segregated solutions.