Research Proposals
Machine learning aided catalysis for finding out descriptors
Recently, data science and machine learning (ML) are being exploited to screen vast catalysts space for various reactions. However, there are still lack in proper interpretation in decision making from the provided properties. With domain knowledge and impacting key features in hand for any ML predicted properties and associating with the structural catalyst conditions we can obtain the useful structure property relationships. Therefore, I will investigate the structure property relationship using ML and understand the obtained results using physics-based approaches to obtain efficient and effective descriptors for catalytic reaction.
Oriented external electric field driven catalysis for selectivity
Accelerating the catalytic process is crucial in catalyst discovery, alongside directing reactions towards specific products. The conventional catalyst discovery approaches face limitations in enhancing selectivity. To address this challenge, I propose using oriented external electric field approach to modulate specific bonds, particularly the transition state intermediates crucial for desired product formation. By applying oriented external electric field aligned with the transition state structure, one can selectively weaken the transition state bond towards targeted products, enhance selectivity by lowering the activation energies.
Real-time dynamics of catalysis under operating condition
The synthesis of an efficient catalyst is not sufficient to address energy-related challenges. The catalyst might restructure during the reaction due to various factors related to the reaction. The restructure of a catalyst during the reaction can be driven by various factors such as temperature, pressure, pH, reaction intermediates and so on. These aspects are critical to consider when trying to develop practical and reliable solutions for energy-related problems using catalytic conversion. To understand this into more detail, we will consider the dynamics of catalyst with respect to the above-mentioned factors.
Modelling of real-time catalyst under operating condition
The discovery of efficient catalysts remains a time-consuming process, both experimentally and theoretically. Most computational catalysis studies focus on idealized clean surfaces, which often do not represent the actual conditions present during reactions. In reality, various experimentally relevant factors can significantly influence catalytic behavior. This disparity hinders direct comparison between theoretical predictions and experimental observations. To address these challenges, I will focus on real-time computational modeling for various reactions that bridges experimental findings with reaction conditions and materials used in practice.
Development of physics-based descriptor for catalysis
Catalytic reactions present an enormous number of possible combinations of catalytic materials and operating conditions, which must be systematically understood and generalized through fundamental physics-based modeling. While several machine learning (ML) models have been developed to tackle this complexity, to get more fundamental insight into catalytic behavior and reaction mechanisms, I propose to shoot the problem from one another direction. Hence, I will be focused on physics-based descriptor findings for accurate and efficient prediction of catalytically relevant properties.
Mechanism of multi-step and multi-product reactions
For well-studied reactions such as HER, ORR, and OER, ORR the key reaction steps are relatively well understood. However, for more complex reactions (like COâ‚‚RR and NRR, hydrocarbon combustion and so on) which involve multiple possible products and pathways, uncovering the detailed reaction mechanisms is essential. Therefore, another key direction of my proposed research will focus on elucidating reaction mechanisms for various reactions to better understand product selectivity and catalytic performance. In this context, source of proton effect of other experimentally relevant parameters will be considered during mechanistic investigation.
Teaching statement
Teaching Philosophy
Clear presentation of material
Detailed explanations
Well-defined goals
Delivering accessible and interpretable contents
Help students visualize the atomic world
Fostering a deeper comprehension
Goals and Methods
Peer-to-peer learning
In person and online teaching
Analogies to everyday experiences
Problem-solving approach
Extra doubt solving classes
Assignments as well as timed, closed-book exams
Laboratory Mentoring
Detailed laboratory reports
Self-driven (intellectual curiosity) research
Suggesting way to find solutions instead of giving solutions instantly (up to a certain time period)
Communicating openly with students about expectations