Collaborative Research: Computational Design of Metal-Organic Framework Materials

Oregon State University, Corvallis OR

Investigators

Abstract

This award supports research in computational methods for the automatic exploration of the search space of metal-organic frameworks to obtain desired combinations of mechanical, thermal and chemical properties. A metal-organic framework is a repeating three-dimensional crystal lattice with a large open space-frame structure composed of organic linking molecules bonded to inorganic nodal units. The interconnected pore-spaces in these materials endow them with a rich variety of unusual properties that can be exploited for gas storage, gas separation, and catalysis. Given the constraints of chemical bonds and bond angles, one cannot easily customize the lattice for a particular size and shape. Thankfully there is an astronomically large number of permutations for combining various atomic elements together. However, human designers are confounded on how to find one for their particular problem domain. What is needed is a computational process to search the space for a best solution. By formalizing the molecular design as a decision tree, the developed computer algorithms will invent new materials whose functional behavior is defined by its chemical makeup and the resulting geometry and movement of the lattice structure. As part of testing the design approach, the project will address two technologically important problems: designing new metal organic frameworks optimized for gas storage, and materials for separating isomers in industrially important chemical feed stocks. STEM outreach activities to high school students and teachers will also be performed. Specifically, the project seeks materials that exhibit highly chemically selective adsorption or permeability of gases through mechanisms that arise from chemical, steric, and vibrational behavior of the frameworks. The computational search for this incorporates a unique graph transformation approach that mimics correct stoichiometric reactions and leads to a large search tree that is amenable to recent advances in artificial intelligence planning algorithms. Furthermore, machine learning methods will establish a link between the structure and function of organic frameworks by leveraging data from complex molecular simulations. This will lead to more efficient search of the decision tree so that meaningful results can be obtained. Through detailed simulations of the resulting metal organic frameworks, the researchers will publish how their new materials can be used to tackle challenging problems in energy storage, high-tech manufacturing, and the creation of new sensitive sensor equipment.