- This event has passed.
Leveraging Thermokinetics Data and Bayesian Statistics to Build Accurate Microkinetic Models
August 4, 2020 from 10:00 am — 11:00 am
Zoom Webinar with Srinivas Rangarajan
Department of Chemical & Biomolecular Engineering
Lehigh University, Bethlehem, PA
Ab initio microkinetic modeling, parameterized using density functional theory (DFT) energies, is a common tool to quantify reaction rates and analyze reaction mechanisms a priori in heterogeneous catalysis. Such models, however, often have large prediction errors even if they include plausible reaction steps and correctly model the active sites; this is partially due to the intrinsic inaccuracies of the chosen DFT functional. We here show that these errors can be systematically and statistically corrected through data fusion from experiments or a higher level of theory via multifidelity modeling. Specifically, we build transferable data-driven corrections to DFT energies in the form of Gaussian process models trained on transition metal single-crystal adsorption calorimetry data. When these corrections are applied to microkinetic models of reactions occurring on transition metal single-crystal surfaces, we see that the new models are 1 – 3 orders of magnitude superior to those built only with DFT. Further, these Gaussian process models serve as excellent priors for a full-fledged Bayesian inference to further improve the model (and parameters) based on kinetic data points.
This talk will present the underlying theory and illustrative examples and discuss data-driven techniques to extend the idea to larger reaction systems and other classes of catalysts with relatively scarce experimental data.
Srinivas is a P.C. Rossin Assistant Professor in the Department of Chemical & Biomolecular Engineering at Lehigh University, Bethlehem, PA. He previously was a postdoctoral research scholar working with Profs. Manos Mavrikakis and Christos Maravelias. He obtained his PhD in Chemical Engineering at the University of Minnesota in 2013 under the supervision of Profs. Prodromos Daoutidis and Aditya Bhan. Srinivas is originally from India and did his undergraduate studies at the Indian Institute of Technology, Madras. His research interests include complex network analysis, microkinetic modeling, computational heterogeneous catalysis, and applications of machine learning and optimization in reaction mechanism discovery and modeling. His research has been recognized with the ACS Petroleum Research Fund Doctoral New Investigator Award and the AIChE CAST division’s David Smith Jr Graduate Publication Award.
For Zoom information please contact email@example.com