Statistical methods for the description of the structural transition in supercritical fluids

Abstract

Supercritical fluids have a wide range of applications in the chemical industries. This versatility is believed to come from its instantaneous tunability

a small change of pressure makes it possible to change various thermophysical properties of supercritical fluids such as the isobaric heat capacity, thermal expansivity, solubility, and diffusivity. This drastic change of thermophysical properties is believed to stem from the structural transition in supercritical fluids near the supercritical gas-liquid boundary. The existence of supercritical gas-liquid boundary casts us the following profound questions. What is liquid? What is gas? Can we define them based on the mathematical expressions? How can the structural transition be modeled and applied to the prediction of thermophysical properties of supercritical fluids? This dissertation attempts to answer these questions based on molecular simulations in conjunction with data science methodologies. This dissertation comprises of five chapters. In Chapter 1, some important concepts and previous works about the supercritical gas-liquid transition are introduced. In Chapter 2, two kinds of data science methodologies (statistical mixture models and probabilistic classification) are designed and tested for the quantitative examination of the structural transition in supercritical Lennard-Jones (LJ) fluid. The characteristics of structural transition are analyzed based on the theory of random percolation and critical phenomena. In Chapter 3, the classification algorithm is used to find out the quasi-universality of the structural transition in simple fluid models. Supercritical gas-liquid boundaries of these simple and non-polar substances follow the same curve on the reduced density-temperature diagram, and the validity of three-parameter corresponding state principle is proven. In Chapter 4, the classification algorithm is applied to a dilute supercritical carbon dioxide mixture to explain the solvation structure and the origin of the negative partial molar volume in the supercritical mixture. The designed algorithm makes it possible to harmonize the conflicting viewpoints on the solvation structure of supercritical fluids around a solute. In Chapter 5, the academic significance of the proposed algorithm and the expected utilization of statistical methodology is discussed. Overall, the quasi-universality of the structural transition and the harmonization of conflicting viewpoints on the solvation structure substantiates the adequateness of the data science methodologies proposed in this dissertation. This method will be extended to explain the liquid-liquid criticality, rigid/non-rigid transition near the Frenkel line and critical Casimir force. Ultimately, the statistical methods proposed in this work would provide a theoretical basis for the theory of fluid polyamorphism and be applied to design and optimization of the supercritical fluid processes.