Advance Nanozymes with High Catalytic Active Sites for Biosensors and Energy Storage Devices | AIChE

Advance Nanozymes with High Catalytic Active Sites for Biosensors and Energy Storage Devices

Authors 

Tripathi, A. - Presenter, University of Alberta
Accelerating deployment of decarbonization technologies is desperately needed to limit global warming to less than 2 °C. The petrochemical industry is particularly difficult to decarbonize because CO2 intensive thermocatalytic processes are entrenched in our infrastructure. To enable rapid decarbonization, we aim to accelerate design and testing of catalysts by redefining catalyst representation and structure-property relationships with large language models (LLMs). The result is a new research direction where we represent catalysts by the synthesis method, reaction conditions, and performance. The method is therefore verifiable and actionable – we can immediately synthesize and test the outcomes generated by the model. We are targeting CO2 hydrogenation to showcase how a complicated and temporally relevant reaction can be cheaply and easily represented via natural language processing. Our initial efforts are focused on tungsten carbide catalysts because of the existence of two distinct carbide phases, tungsten carbide (WC) and tungsten semi-carbide (W2C). With this example, we demonstrate how we can apply Bayesian optimization to a limited data set to identify the top performing WC-based catalysts for reverse water-gas shift with a minimum number of experiments.

In our approach, we are leveraging the LLM to better understand how WC structure affects CO2 hydrogenation properties. W2C has emerged as a promising candidate for CO2 hydrogenation, likely because the electronic structure have similar features to the active Mo2C catalyst when WC and MoC have same bulk structures. Through carefully controlling the carburization conditions, particle size and composition of the tungsten-based catalysts, and representing the catalysts with the synthesis procedures and reaction conditions, we aim to uncover a new classification of structure-property relationships that will expand the library of active, selective, and earth-abundant catalysts for CO2 hydrogenation.