Tuning OER Electrocatalysts toward LOM Pathway through the Lens of Multi-Descriptor Feature Selection by Artificial Intelligence-Based Approach

Abstract

From the thermodynamic and kinetic viewpoints, the oxygen evolution reaction (OER) is central to the production of hydrogen through electrocatalytic water splitting process. As a result, extensive research is carried out with an aim to develop (preferably low cost) electrocatalysts for efficient OER reaction. Two main mechanisms, the lattice oxygen mediated mechanism (LOM) and the adsorbate evolution mechanism (AEM), are proposed for the OER process, but the former seems to be favored as the latter is surrounded by a number of discrepancies. The LOM is based on the oxidation and reduction chemistry of lattice oxygen anions. Both from fundamental and application perspectives, identification and selection of crucial electronic and structural features to rationally tune the LOM for OER process and determine the rate of the OER reaction are essential but remain a challenge. As yet, this has largely been attempted by trial and error (synthesis and performance evaluation) approaches and, hence, has been tied up by tediousness and inefficiency. Considering the availability of large chemical space and huge probabilities in fine-tuning of electronic and structural attributes, the use of artificial intelligence (AI) can be efficacious. To predict promising material features that enable rational design of OER electrocatalysts, electrocatalytic performance maps have been developed based on reactivity multiple descriptors. The reactivity multiple descriptors are numerous and are classified into several families. Therefore, modern materials informatics technique is urgently required to enable large-scale data mining to rapidly screen and select the best physical features from multiple descriptors for a rational design of LOM-induced OER electrocatalysts. In this context, feature selection by artificial intelligence-based approach can be a solution to such challenge. Yet to the best of our knowledge, no attempt has been made before to tune electrocatalysts for the OER through LOM reaction mechanism following feature selection by artificial intelligence-based approach.