(147j) Bridging Thermal and Electrochemical Catalysis: Rational Catalyst Design at Atomic Scales through Physical and Machine-Learning Based Insights

I am interested in working on the intersection between chemistry and material sciences, specifically focusing on the technical aspects of utilizing machine learning (ML) for materials discovery in various applications such as thin-film photovoltaics, photo/electro catalysis, batteries, and novel material synthesis. My goal is to contribute to the development of sustainable processes that leverage renewable resources.

My expertise lies in employing computational techniques to investigate the relationship between the structure and properties of materials, particularly their chemical and electronic properties. Additionally, I am skilled in implementing ML-based screening methods to create predictive models for material performance and stability. I have extensive experience using cutting-edge automation and computational tools in the fields of solid-state chemistry and biopharmaceutics. I enjoy working collaboratively with chemists, engineers, and product managers to design, simulate, and optimize active materials and processes. Additionally, my prior three-year industrial experience as a team leader in the Catalytic Reforming Unit at the Indian Oil Corporation Limited, Digboi Refinery, India has equipped me with valuable leadership skills, strong work ethics, and strategic planning abilities that enable me to thrive in challenging work environments. During my PhD and postdoctoral work, I also had the opportunity to mentor undergraduate and graduate students on research projects, which has honed my ability to independently formulate and lead projects to successful completion.

RESEARCH OVERVIEW

The overarching goal of my research is to accelerate materials discovery via a wide range of computational methods. I strive to combine physics-based models, high-throughput simulations, ab initio simulation methods like density functional theory (DFT), and machine learning (ML) to discover functional materials and design catalysts for thermal and electrocatalytic reactions. Such initiatives target applications that enable use of renewable resources and sustainable processes. The same basic principles and approaches, borrowed from catalysis, are applied to screen solid polymer electrolytes (SPE) in pursuit of safer and more durable low-cost lithium-ion batteries. These goals have been accomplished in collaboration with experimentalists and our findings will be stored in online databases (theory & experiment) for easy use or reproduction by other researchers.

RESEARCH SUMMARY

In the realm of heterogeneous catalysis, whether thermal or electrochemical, physical models supplemented by machine learning (ML) can accelerate the discovery of inexpensive and abundant catalysts with high activity, selectivity, and stability. Computationally predicting effective materials is essential because the chemical space of candidate materials can be quite large, and coupled with challenges in targeted synthesis, experimental exploration of this material space is prohibitively time-taking. Therefore, my efforts encompass physical insights complemented ML approaches for rational catalyst materials design, spanning new classes of materials like high entropy oxides and alloys, dual and single metal atom catalysts. Fundamental insights into complex material systems (through interface and defect engineering) and complex reaction chemistries deliver sustainable solutions to the energy crisis and the use of renewable energy sources. Materials modeling are also integrated with big-data optimization approaches using collaborator’s experimental data. Bayesian optimization via experimental data reduces the high dimensionality of descriptors space for materials search. These algorithms are also employed towards designing highly conductive polymer electrolytes for the development of next-generation safe lithium-ion batteries.

RESEARCH HIGHLIGHTS

While conducting my postdoctoral research at Lawrence Livermore National Laboratory, I focused on various aspects of energy conversion (H₂ generation from liquid organic hydrogen carriers), carbon capture, and the analysis of associated processes within these technologies. My work primarily involved employing classical and ab-initio simulations to understand the mechanisms behind carbon dioxide methanation. Additionally, I conducted thorough thermodynamic and kinetic evaluations of the reaction pathways, developed microkinetic models, and performed sensitivity analyses across different operating parameters. Throughout this project, I collaborated with a multidisciplinary team to establish a framework for integrating modeling approaches at different scales, ranging from surface design to reactor design and eventual deployment.

During my postdoctoral work at SUNCAT (Stanford University), I successfully developed physical insights complemented ML approaches towards swift predictions of metal ad-atom diffusion barriers and surface/segregation energies for enhanced understanding of sintering and catalyst durability. Moreover, I successfully integrated experimental and theoretical data (which will be uploaded into an online database cathub, https://www.catalysis-hub.org) for high quality predictions of material performance towards electrochemical reactions (e.g., Oxygen reduction reaction, ORR) using ML techniques. Moreover, my PhD dissertation investigated reaction chemistry (hydrodeoxygenation and CO oxidation) at oxide/metal interfaces that take advantage of the properties of the interfacial components. We used DFT calculations, microkinetic modelling and a Bayesian statistical inference approach to identify dominant reaction networks, form a descriptor-based design model, and predict optimal catalytic activity. I have also worked on a collaborative project to tune the conductivity in PEO₆ electrolyte-based Li-ion batteries where I achieved an order of magnitude increase in room temperature ionic Li⁺ conductivity via defects and interstitial migrations.

On a bigger picture, ML extracted human-interpretable models along with derived physical insights can accelerate materials discovery and design as well as foster new collaborations through the sharing of ideas via easy access to online database of theory and experimental physicochemical properties.

RESEARCH EXPERTISE

Heterogeneous Catalysis

Computational Chemistry & modelling

Electrochemical systems

Applied Surface Science

Lithium-ion batteries

Liquid Organic Hydrogen Carriers (LOHC)

SELECTED PUBLICATIONS

My publications can be found via my Google Scholar page (https://scholar.google.com/citations?hl=en&user=ZHEdfbsAAAAJ&view_op=list_works&sortby=pubdate). Below follows a selection of publications:

1. L. Wang ^#, S. Deo ^# (^#equal contribution), A. Mukhopadhyay, NA Pantelis, M. J. Janik, R. Rioux, Emergent Behavior in Oxidation Catalysis over Single-Atom Pd on a Reducible CeO₂ Support via Mixed Redox Cycles, ACS Catalysis 12 (20), 12927-12941

2. S. Deo, M. J. Janik, Predicting an optimal oxide/metal catalytic interface for hydrodeoxygenation chemistry of biomass derivatives, Catalysis Science & Technology, 2021, 11, 5606 - 5618

3. J. Zhang, S. Deo, M.J. Janik, J. Medlin, Control of molecular bonding strength on metal catalysts with organic monolayers for CO₂ reduction J. Am. Chem. Soc. 2020, 142, 11, 5184-5193

4. L. Wang, S. Deo, K. Dooley, M.J. Janik, R. Rioux, Influence of metal nuclearity and physicochemical properties of ceria on the oxidation of carbon monoxide, Chin. J. Catal., 2020, 41: 951–962

5. S. Deo, W. Medlin, E. Nikolla, M.J. Janik, Reaction paths for hydrodeoxygenation of furfuryl alcohol at TiO2/Pd interfaces, Journal of Catalysis, 377 (2019) 28-40

6. D. Gao, I.T. McCrum, S. Deo, Y.-W. Choi, F. Scholten, W. Wan, J.G. Chen, M.J. Janik, B. Roldan Cuenya, Activity and selectivity control in CO2 electroreduction to multicarbon products over CuOx catalysts via electrolyte design, ACS Catal., (2018) 10012-10020

Figure. Materials optimization methodology using a machine-learning (ML) based approach and Bayesian optimization via integration between theory and experiments