(3dq) Algorithmic Exploration of “Building-Block” Chemistry

In the last decade a wealth of novel materials has been synthesized through what has come to be called “modular” or “building-block” chemistry. This new approach leverages the self-assembly of rigid molecular building blocks that only connect in very specific orientations and stoichiometries, which constrain the resulting assemblies into discrete geometries (e.g., polyhedra, tubes, spheres) or crystalline structures (e.g., metal-organic frameworks (MOFs)).

A remarkable feature of building-block chemistry is that the products are typically monodisperse, and their geometry and composition can be predicted a priori. Due to this predictability and the abundance of known building blocks (drawn from the vast organic chemistry literature), the field of MOF chemistry alone has reported thousands of novel structures in the past few years. While these numerous reports showcase the success of the building-block approach, they also belie the underlying combinatorial difficulty of finding the best material for a given application. There are millions of possible MOFs: trying to find the optimal material for a specific application by experimental trial-and-error is impractical and inefficient at best.

In my doctoral research at Northwestern University I developed algorithms that systematically enumerate all of the hypothetical MOFs that could be made from a given library of building blocks. Over 150,000 novel MOF structures were screened in a high throughput manner via molecular simulations. The scope of this computational screening effort was without precedent in the MOF field. For each MOF I obtained a range of material properties such as surface area, pore volume, pore size distribution, powder x-ray diffraction pattern, and gas adsorption. In addition to rapidly identifying MOFs that had higher gas storage performance than any previously known materials, I illuminated hitherto unidentified structure-property relationships that could only have been recognized by taking a global view of MOF structures. Working with synthetic chemists, we synthesized some of the highest performing MOFs predicted by my algorithms, and their measured gas adsorption agreed well with the simulations. Furthermore, by making our database of MOFs publicly available online, this research has already made a measurable, global impact in both academic and industrial labs, as well as stimulated significant commercial interest (>$1 million in venture capital). This disruptive approach to exploring the space of MOFs will significantly accelerate the discovery of better materials for applications such as hydrogen and natural gas storage.

More generally, the algorithmic investigation of materials self-assembled from building blocks can be usefully applied not only to MOFs but to discrete structures and other bulk material classes. Working closely with my current and anticipated experimental collaborator, Dr. Omar K. Farha, my future research will continue to interface closely with chemical synthesis and empirical measurements as I strive to develop predictive computational tools. My future research will also build on my anticipated post-doctoral research by incorporating modern high throughput quantum chemistry techniques (e.g., quantum Monte Carlo) to search for optimal catalysts and yield useful structure-property insights to create new catalysis design-rules. This algorithmic approach for exploring new chemical spaces, combined with the predictive power of quantum chemistry, will accelerate the pace of materials discovery and expand the scope of investigations in chemical separations, catalysis, and self-assembly.

PhD Advisor: Professor Randall Q. Snurr