Inverse Design of Nanoporous Materials for Gas Separation and Energy Storage via Physics-Based and Data-Driven Modeling

Nanoporous materials (e.g., activated carbon, zeolite and metal-organic framework) have attracted great interest in recent years because of their excellent structural and chemical properties including large specific surface area, good mechanic strength and tailorable pore size and local chemical environment. Such materials are promising to serve as better alternatives to conventional materials such as porous electrodes in energy storage or adsorbents and membranes in separating molecules with similar properties. Because of the almost infinite design space, not only is the identification of best nanoporous materials with target performance practically infeasible with traditional experimental trail-and-error methods, but also it imposes theoretical and computational challenges for the computational modeling of nanoporous materials in gas separation and energy storage.vii To accomplish the inverse design of nanoporous materials in gas separation and energy storage, this dissertation aims to establish physics-based and data-driven models that can be used to fast and accurately evaluate the performance of nanoporous materials. Toward that end, I developed classical density functional theory (cDFT) to predict the adsorption of multicomponent gas mixtures in nanoporous materials. The adsorption isotherms predicted by cDFT were in excellent agreement with grand canonical Monte Carlo simulation and experimental measurement. In addition, I extended the simplified string method to calculate the minimum energy path (MEP) of rigid polyatomic molecules in nanoporous materials. The diffusion coefficients predicted from MEP via the transition-state theory agreed quantitatively well with those from molecular dynamics simulation. Furthermore, I implemented the physics-based models with massively parallel GPU-acceleration, which leaded to orders of magnitude speedup compared with conventional molecular simulation. Moreover, I combined the data-driven models and evolutionary algorithm with physics-based models to case study the inverse design of nanoporous materials for the separation of D2/H2 and of C2H4/C2H6. For energy storage, I established excellent correlations between the structural and chemical features of nanoporous materials and their in-operando electrochemical performance in supercapacitors using data-driven models and proposed useful guidelines for the inverse design. The computational framework developed in this dissertation demonstrated the feasibility for the inverse design of nanoporous materials for gas separation and energy storage with the combination of physics-based and data-driven models