(2ht) Rigorous Statistical Mechanics and Rare Events Tools to Model Catalyst Site Ensembles

Research Interests

Several tailor-made catalysts are being developed to address energy and sustainability challenges in hydrocarbon feedstock conversion, electrocatalysis, polymerization, carbon capture and other areas. Many of these catalysts have a distribution of catalytic site environments, like sub-nanometer clusters, high-entropy alloys, amorphous catalysts, etc. Often, these active site distributions are synthesis dependent and extremely sensitive to operation conditions. Ab initio computational methods and, recently, machine learning (ML) methods have made tremendous progress in understanding some of these catalysts. However, the development of rigorous statistical mechanics tools to calculate experimental observables for several such catalysts has been limited. Most computational studies make ad hoc assumptions about active site structure and catalyst mechanisms. Thus, precluding precise comparison to experiments.

My research program will develop rigorous theoretical and computational tools to model the structure and reactivity of complex catalysts with a distribution of active sites. My training in catalysis, statistical mechanics, rare events methods, ML, and population balance modeling provides a unique combination to develop this research program. Specifically, my group will 1) model the synthesis dependent structure of complex catalysts like supported clusters, high-entropy alloys, and amorphous catalysts, 2) develop rigorous rare events tools to efficiently calculate site-averaged reactivity of catalyst site-distributions, and 3) develop theoretical methods to infer catalyst structure and dynamics from spectroscopy and microscopy measurements.

Prior work

The overarching goal of my PhD research was to develop methods for modeling the synthesis and reactivity of atomically dispersed amorphous catalysts. I developed ab initio parameterized population balance models to simulate the grafting of metal complexes onto amorphous supports (a method to synthesize amorphous catalysts). I developed an importance learning algorithm using ML and non-Boltzmann sampling to efficiently calculate site-averaged kinetics of amorphous catalysts. I developed probabilistic models informed by spectroscopic measurements to describe the distribution of surface hydroxyl groups on amorphous silica supports. I also had the opportunity to work on rare events methods where I demonstrated that the accuracy and efficiency of infrequent metadynamics (a popular rare events method to calculate rates of several processes including, protein folding, catalysis, nucleation, etc.) is severely limited in the presence of fluxional reactants.

In my postdoctoral research I have developed actively trained ML potentials with basin-hopping to discover ensembles of low-energy metal clusters on non-porous and porous supports. My framework is being used by several group members for different applications. Currently, I am modeling the dynamics of supported clusters under reaction conditions, developing structure-property ML models for sub-nanometer metal clusters, and exploring the effect of site-structure on spin-crossing.

Teaching Interests

I believe mentorship and teaching are one of the most rewarding aspects of a faculty career. As a postdoc I have had the opportunity to closely mentor an undergraduate student and a graduate student on modeling bi-metallic catalysts and developing structure-property ML models, respectively. I have also had to opportunity to teach a few classes. I taught a class on Special Topics in Energy at the University of Delaware. I also attended a two weeklong NSF teaching workshop here. As a part of the workshop, I designed and delivered a lesson on the second law of thermodynamics from an information theory perspective. In addition, I am also interested in developing alternative methods of teaching. In my free time, I have been developing animations on mathematics and reaction rate theory.

I am confident in teaching all core chemical engineering courses, both at undergraduate and graduate levels. I am particularly interested in teaching thermodynamics, statistical mechanics, mathematical methods, reaction rate theory, and chemical kinetics. I want to develop a new upper-level undergraduate/graduate course on statistical mechanics taught from an information theory perspective and a graduate course on reaction rate theory and rare events.

Selected Publications: (* indicates equal contribution)

1. Salman A. Khan, Stavros Caratzoulas, and Dionisios G. Vlachos. "Cluster catalyst induced support reconstructions", in review.
2. Salman A. Khan, Sahan Godahewa, Pubudu Wimalasiri, Ward Thompson, Susannah L. Scott, and Baron Peters. "Grafting TiCl4 onto amorphous silica: modeling effects of silanol heterogeneity." Chem. Mater., 2022, 34 (9), 3920–3930.
3. Salman A. Khan, Bradley M. Dickson, Baron Peters, "How fluxional reactants limit the accuracy/efficiency of infrequent metadynamics." J. Chem. Phys., 2020, 153 (5), 054125.
4. Craig A. Vandervelden,* Salman A. Khan,* Susannah L. Scott, and Baron Peters. "Site-averaged kinetics for catalysts on amorphous supports: an importance learning algorithm." React. Chem. Eng., 2020, 5 (1), 77-86.
5. Salman A. Khan,* Craig A. Vandervelden,* Susannah L. Scott, and Baron Peters. "Grafting metal complexes onto amorphous supports: from elementary steps to catalyst site populations via kernel regression." React. Chem. Eng., 2020, 5 (1), 66-76.

Education

University of California, Santa Barbara, U.S.A.
PhD in Chemical Engineering (September 2016-August 2021)

Indian Institute of Technology, Kanpur, India
B.tech-M.tech dual degree in Chemical Engineering (August 2011-May 2016)