(509cp) First-Principles Calculations to Data-Driven Discovery and Materials Design of Mxene Electrocatalysts | AIChE

(509cp) First-Principles Calculations to Data-Driven Discovery and Materials Design of Mxene Electrocatalysts

Authors 

Johnson, L. - Presenter, University of Pennsylvania
Burghardt, B. - Presenter, University of Pennsylvania
Morency, Y. - Presenter, University of Pennsylvania
Vojvodic, A., University of Pennsylvania
In the search for cheaper materials for sustainable applications, there is a desire to catalyze reactions at room temperature and pressure with high active surface area. Designing new catalysts can in turn improve photovoltaics, chemicals, and oil and gas industries. MXenes, a new class of 2D materials similar to graphene, are tunable through physical and chemical modifications and have been considered as electrocatalysts for a decade now.1–5 In this timeframe, learning methods have been used to predict and infer chemical and physical properties through analyzing datasets and building atomistic potentials.6

Using first-principles calculations, we generate a dataset from 350 unique MXenes interacting with 7 adsorbates corresponding to intermediates of the HER and NRR over up to 4 unique symmetries consisting of 3475 entries. We develop a data science pipeline used to clean, scale, and reduce the feature to the most important features. Both linear and decision-based models provided discernment between bare and functionalized MXenes, different adsorbates, and the specific terminating atoms as subsets to formalize the decision tree ensembles. Features belonging to the terminating atom and the specific *NxHy adsorbate were identified as crucial parameters needed to predict the adsorption energy. Further exploration of the data demonstrates generalizability through testing metrics and the visualization of the materials landscape through principal component analysis. We successfully applied the model to predict and capture trends in reactivity across 700 tuned MXene materials of interest, identifying sulfidation and supporting/straining as strategies to improve their nitrogen reduction activity.

This study provides insight for future research into developing models to capture reactivities using accessible DFT descriptors from featurizing the local atomic environment near the adsorption site of the materials. Furthermore, invaluable data and crucial analyses for designing materials and substances with high chemical activity are contributed to the 2D materials and catalysis communities.