CDS&E: Novel Hybrid Modeling Paradigm for Multi-Component Water Vapor Adsorption

University Of Notre Dame, Notre Dame IN

Investigators

Abstract

This project will use state-of-the-art computer simulations and machine learning algorithms to advance fundamental understanding of water vapor and water vapor mixture adsorption in porous materials. These are key to developing technologies like atmospheric water harvesting and carbon capture in the presence of water vapor. These are crucial for society in the context of water security, climate change, and other grand challenges. This research project uniquely combines powerful computational modeling and machine learning tools to produce new models that can describe multicomponent water vapor adsorption in porous materials. The research will lead to fundamental insights into the complex interactions between the materials and the adsorbed components of the mixture, which are crucial for technological advancements in areas of national importance including climate change and water security. Outreach and education components within this project include investigating commercialization opportunities for dehumidification processes through the Notre Dame Engineering, Science, and Technology Entrepreneurship Excellence Masters (ESTEEM) program and undergraduate and graduate student training via research experiences and education modules in existing electives. This research program will develop a hybrid modeling paradigm for multicomponent adsorption involving water vapor. The new modeling approach will integrate molecular modeling, statistical mechanics, and machine learning techniques to systematically determine new models capable of describing mixture adsorption with water vapor. The development of these technologies requires fundamental understanding of water vapor and its multicomponent mixtures in confinement and a framework that allows for proper material and technological evaluations in this scenario. A major bottleneck in this context is the lack of models allowing for proper prediction and evaluation. The research goal of this proposal is to produce new hybrid models capable of accurately and efficiently describing multicomponent water vapor adsorption in porous materials. This will be accomplished through a unique combination of machine learning, statistical mechanics, molecular modeling, active learning, and autonomous experiment design. This combination will yield a novel methodological framework for the development of mixture adsorption models that deal with water vapor. These goals are supported by two research objectives to (1) combine active learning with statistical mechanics for molecular modeling of multicomponent water vapor adsorption and (2) develop hybrid models that combine thermodynamic theory and machine learning. These models will be developed using mixtures relevant to atmospheric water harvesting and carbon capture in humid streams. The proposed research seeks to establish new computational methods that are generalizable and transferable to other scenarios, with water vapor serving as a challenging and societally important class of demonstration problems. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.