ScholarWorks@성균관대학교

초록

Atmospheric concentrations of CO2 must be lowered to mitigate climate change and rising global temperatures. CO2 utilization is the most promising approach for the sustainable reduction of CO2 emissions. Interdisciplinary research is gaining increasing attention due to its broader application potential and the promising results of combining various fields. Computational approaches in catalytic research could be cost-effective and environmentally friendly. Machine Learning (ML) and 3D printing technologies may soon be able to produce nanoscale raw materials to synthesize the catalyst for commercial-scale applications. In this review article, recent advances in catalyst synthesis using 3D printing technologies and ML-based catalytic reactions, particularly those in CO2 hydrogenation, are critically analyzed, with a focus on the function of ML model prediction. ML approaches with high prediction accuracies are discussed comprehensively. Based on the literature Gray-box models can provide useful insights by revealing the essential catalytic traits, factors, and circumstances that affect the results. They can also provide a practical solution by fusing the benefits of black-box algorithms, such as ensemble models and NNs, with feature importance analysis. Finally, suggestions and recommendations for the potential applications of ML in chemical science, especially in heterogeneous catalysis, are provided along with future research directions. © 2024

키워드