Machine learning accelerated calculation and design of electrocatalysts for CO₂ reduction

Abstract

In the past decades, machine learning (ML) has impacted the field of electrocatalysis. Modern researchers have begun to take advantage of ML-based data-driven techniques to overcome the computational and experimental limitations to accelerate rational catalyst design. Hence, significant efforts have been made to perform ML to accelerate calculation and aid electrocatalyst design for CO 2 reduction. This review discusses recent applications of ML to discover, design, and optimize novel electrocatalysts. First, insights into ML aided in accelerating calculation are presented. Then, ML aided in the rational design of the electrocatalyst is introduced, including establishing a data set/data source selection and validation of descriptor selection of ML algorithms validation and predictions of the model. Finally, the opportunities and future challenges are summarized for the future design of electrocatalyst for CO 2 reduction with the assistance of ML.

Machine Learning Assisted Understanding and Discovery of CO2 Reduction Reaction Electrocatalyst

“…Different types of catalyst have been used for CO 2 RR, and the most popular ones include transition metals, metal alloys with two or more elements, and single metal catalysts . Some recent reviews focused on the methodologies and functions of ML in such research , yet pay much attention to the difference between application on different types of catalysts. In this paper, we fill in the gap by introducing the latest advances according to the type of catalyst.…”

Enhanced Piezoelectric Properties Enabled by Engineered Low-Dimensional Nanomaterials

“…104 The heterostructure can show enhanced energyharvesting performances compared to the individual components alone in the composite, 105−110 which are due to the electrostatic field in one component and residual tensile stress and complex interfaces between two components in the composites, resulting in the nonzero charges and high modulus of the composites for enhanced energy-harvesting devices. 111,112 The formation of LD materials-based composites can convert nonpiezoelectric materials to piezoelectric. When two entities in a composite are combined, piezoelectricity can be induced by breaking the initial centrosymmetric crystal structures.…”

“…The main advantages of the heterostructure combine different components with the interaction of each other, resulting in a higher degree of polarization and improved piezoelectric properties . The heterostructure can show enhanced energy-harvesting performances compared to the individual components alone in the composite, − which are due to the electrostatic field in one component and residual tensile stress and complex interfaces between two components in the composites, resulting in the nonzero charges and high modulus of the composites for enhanced energy-harvesting devices. , …”

Machine-learned digital phase switch for sustainable chemical production

“…For bio-based chemical production, machine learning has been particularly beneficial for real-time, dynamic control applications of biochemical processes such as bio-fermentation and anaerobic digestion (Pomeroy et al, 2022). Recent challenges in photochemical reaction multiscale modelling (Kovačič et al, 2020) can also be tackled using machine learning models to speed up computation (Sun et al, 2022). Despite such successes for machine learning in chemical reaction analysis, Kovács et al (2021) recently argued that model interpretation is a stumbling block for black-box machine learning models on reaction predictions.…”