期刊:Physics of Fluids [American Institute of Physics] 日期:2024-12-01卷期号:36 (12)
标识
DOI:10.1063/5.0237448
摘要
This study examined the distributions of pseudo-phases of supercritical CO2 (scCO2) at the scCO2-subcritical water (subH2O) interface. Molecular dynamics and grand canonical Monte Carlo simulations were used for phase identification with a revised molecular deep neural network (DNN) model that includes van der Waals interactions and Coulomb forces. Near the interface, scCO2 can exist in dissolved-liquid, gas-like, two-phase-like, or liquid-like phases, significantly affecting its density distributions and physical properties, such as compressibility. The locations of the Widom line and Widom delta were identified, and their crossings altered the phase distributions and physical properties of scCO2. Using the unified law λ* = (P + a)/bT, the combined effects of temperature and pressure on scCO2 phase distribution were studied: crossing the Widom line changes the phase distributions from a five-layer to a three-layer structure, eliminates the GL phase, shifts the state from compressive to hard-compressive, and alters the interfacial tension from a linear-decrease stage to a pseudo-plateau stage. Crossing the Widom delta results in a transition from seven-layer to three-layer structures, indicating a second-order phase transition. The two-phase-like scCO2 has a density between 0.3 and 0.7 g/ml, fluctuating as a buffer between liquid-like and gas-like phases. The molecular DNN model effectively distinguished supercritical phases and tracked phase distribution changes at interfaces across a wide range of pressures and temperatures.