Revisiting Electrocatalyst Design by a Knowledge Graph of Cu-Based Catalysts for CO2 Reduction

Electrocatalysis takes a significant role in the productionofsustainable fuels and chemicals. The combination of artificial intelligenceand catalytic science is exhibiting great potential to extract, analyze,and predict electrocatalysts. However, the currently developed machinelearning approach usually requires a mass of data from density functionaltheory calculations to train and optimize models. In contrast, a knowledgegraph has the potential to extract useful information from a largeamount of the literature without referring to density functional theory.Herein, a knowledge graph of Cu-based electrocatalysts for electrocatalyticCO(2) reduction is constructed based on a linguisticallyenriched SciBERT-based framework. This framework retrieves multipletypes of entities including material, regulation method, product,Faradaic efficiency, etc. from 757 scientific literature, generatesrepresentations with abundant domain-specific semantic information,and exhibits the capability to deal with electrocatalysts for CO2 reduction. The obtained graph shows the development historyof related catalysts, builds relationships between the factors associatedwith catalysis, and provides intuitive charts for researchers to gainuseful information. Furthermore, we propose a deep learning-basedprediction model, which integrates the semantic information from thescientific literature (word embedding) with the correlation of knowledgetriples (graph embedding) and realizes the prediction of the Faradaicefficiency for a targeted case. This work paves the way for catalystdesign in the manner of merging artificial intelligence with catalyticscience.