| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Innovations in Energy Materials: Unveiling Future Possibilities of Computational Modelling and Atomically Controlled Experiments
|
| Presentation Title |
Optimization of CO2 Reduction Reaction Using Nanoporous Copper Catalysts through Machine Learning-Driven Process Parameter Modeling |
| Author(s) |
Yu-Hung Lai, Jun-Yi Lok, Wen-Han Tsai, I-Chung Cheng |
| On-Site Speaker (Planned) |
Yu-Hung Lai |
| Abstract Scope |
Nanoporous copper (NPC) holds promise for the CO2 reduction reaction (CO2RR), but its performance varies with preparation and reaction conditions. This work used a machine learning-evolutionary algorithm framework to optimize NPC electrode preparation for enhanced C2 product Faradaic efficiency (FE). The extreme gradient boosting regression (XGBR) algorithm, coupled with the genetic algorithm, predicted optimal parameters, achieving a total C2-specific FE of 66.71% and a space-time yield of 4693 μmol/(h∗g), surpassing other Cu catalysts. Further tuning led to the NPC-33 model, combining XGBR, neural network (NN), and support vector regression (SVR), reducing the mean absolute error (MAE) from 2.764 to 2.5. Validated with swarm and particle swarm algorithms, this combined model confirmed optimized process parameters. This study demonstrates the efficacy of machine learning methods in identifying optimal NPC electrode preparation parameters for enhanced CO2RR performance, offering novel insights for developing high-performance CO2RR catalysts. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Nanotechnology, Machine Learning, Sustainability |