Data-driven discovery of electrocatalysts for CO2 reduction using active motifs-based machine learning

Abstract

The electrochemical carbon dioxide reduction reaction (CO2RR) is an attractive approach for mitigating CO2 emissions and generating value-added products. Consequently, discovery of promising CO2RR catalysts has become a crucial task, and machine learning (ML) has been utilized to accelerate catalyst discovery. However, current ML approaches are limited to exploring narrow chemical spaces and provide only fragmentary catalytic activity, even though CO2RR produces various chemicals. Here, by merging pre-developed ML model and a CO2RR selectivity map, we establish high-throughput virtual screening strategy to suggest active and selective catalysts for CO2RR without being limited to a database. Further, this strategy can provide guidance on stoichiometry and morphology of the catalyst to researchers. We predict the activity and selectivity of 465 metallic catalysts toward four expected reaction products. During this process, we discover previously unreported and promising behavior of Cu-Ga and Cu-Pd alloys. These findings are then validated through experimental methods.

Designing Surface and Interface Structures of Copper-Based Catalysts for Enhanced Electrochemical Reduction of CO2 to Alcohols

“…For example, Cu(100) was found to be easier for C-C coupling by combining electrochemical tests and DFT calculations [31]. In situ Raman was performed recently, confirming that higher surface coverage of adsorbed *CO on the Cu (110) surface promotes the formation of the *OCCO and *CH 2 CHO intermediates to generate C 2 products; comparatively, the Cu (111) surface possessed low *CO coverage to produce CH 4 [32]. In recent research, a product-specific active site for ECR was concluded by means of detailed analysis on nine single-crystal copper surfaces.…”

“…Due to the distinct arrangement of surface atoms and the resulting interaction with reaction molecules, different crystal facets of the catalysts tend to present varied performance toward ECR [28]. The first ECR on single-crystal Cu was performed by Frese, who found increasing CH 4 generation on Cu(100), Cu (110), and Cu (111) surfaces [29]. In 2002, Hori et al systematically studied the important impact of Cu facets toward specific ECR products, including CH 4 , C 2 H 4 , CH 3 COOH, CH 3 CHO, and C 2 H 5 OH [30].…”

“…In recent research, a product-specific active site for ECR was concluded by means of detailed analysis on nine single-crystal copper surfaces. The functions of lattice facet, coordination number, and step-terrace angle were taken into consideration for specific ECR performance, and Cu (110), which possesses a coordination number of seven and a larger step-terrace angle, was found to be able to promote ethanol production [33]. Additionally, some high-index copper facets have been found to prefer C 2+ production in ECR [34,35].…”

Electrified CO2 valorization in emerging nanotechnologies: a technical analysis of gas feedstock purity and nanomaterials in electrocatalytic and bio-electrocatalytic CO2 conversion

“…This iterative process could be accelerated using new machine learning approaches that combine DFT models with material science libraries to predict catalyst designs with desirable properties. 86 Once optimal catalysts have been identified, approaches such as permselective coatings may be applied overtop of the CO 2 -RR catalysts to block impurities while allowing efficient transfer of reactants ( i.e. CO 2 ) to the electrode interface.…”

Redefining the Stability of Water Oxidation Electrocatalysts: Insights from Materials Databases and Machine Learning

“…Comprehensive studies across various TM X-ide compositions have provided valuable insights. , Controlling variables such as the catalytic support, morphology, and catalyst loading allows for focused investigation of specific parameters, for example, the impact of nonmetal size on the reconstruction and OER performance. Furthermore, researchers can draw on previous efforts to develop materials databases for other classes of materials, such as halide perovskites, transition-metal dichalcogenides and oxides, catalysts for electrochemical CO 2 reduction, , and the Open Catalyst 2022 (OC22) data set for oxide electrocatalysts …”