Methanol-to-olefins (MTO) is a crucial non-petroleum route to produce light olefins like ethene and propene, which are essential for manufacturing plastics and other chemicals. However, traditional catalyst development relies heavily on trial and error, making it challenging to balance product selectivity, catalyst lifetime, and production costs.
To address this, the researchers constructeda comprehensive MTO database covering 41 types zeolite topologies and over 6,000 data points. They trained and evaluated more than 20 machine learning(ML) models, with tree-based ensemble methods achieving an accuracy ofover 90% in predicting catalyst performance. Through interpretable ML analysis, they extracted key design rules for high-selectivity catalysts: the largest ring size (LRS) ≤ 9-membered ring, the maximum entrapped sphere diameter (MDi) between 7.35–7.71 Å, and acid density (A/T) ≤ 0.01.
Guided by these rules, the team synthesized a high-silica SSZ-13 catalyst (Si/Al = 100). Under optimized reaction conditions (450 °C, WHSV = 1.0 h⁻¹), the catalyst exhibited extraordinary performance: 100% methanol conversion, 87.6% combined ethene and propene selectivity, and an unprecedented 61.1% initial ethene selectivity—surpassing most previously reported MTO catalysts.To validate thereliability of the ML models', the team also tested a newly synthesized STT zeolite that was not included in the original database. The experimental results closely matched the model predictions, confirming the method's strong generalization ability.
The research was jointly conducted by the team led by Academician Prof. Dr. Zhongmin Liu and Prof. Dr.Miao Yang from Dalian Institute of Chemical Physics, Chinese Academy of Sciences, and the groupled by Associate Professor Dr. Xiaoguang Wang from Dalian University of Technology. They achieved a major breakthrough in the field of methanol-to-olefins (MTO) catalysis: by leveraging machine learning (ML) technology, they successfully discovered a zeolite catalyst with ultra-high initial ethene selectivity. Beyond developing a high-performance MTO catalyst, this studyestablished a data-driven framework for accelerated catalyst development, demonstrating the great potential of artificial intelligence in revolutionizing catalytic materials research. The results were published in Chinese Journal of Catalysis (DOI: 10.1016/S1872-2067(25)64903-5 )
About the Journal
Chinese Journal of Catalysis is co-sponsored by Dalian Institute of Chemical Physics, Chinese Academy of Sciences and Chinese Chemical Society, and it is currently published by Elsevier group. This monthly journal publishes in English timely contributions of original and rigorously reviewed manuscripts covering all areas of catalysis. The journal publishes Reviews, Accounts, Communications, Articles, Highlights, Perspectives, and Viewpoints of highly scientific values that help understanding and defining of new concepts in both fundamental issues and practical applications of catalysis. Chinese Journal of Catalysis ranks among the top one journals in Applied Chemistry with a current SCI impact factor of 17.7. The Editors-in-Chief are Profs. Can Li and Tao Zhang.
At Elsevier http://www.journals.elsevier.com/chinese-journal-of-catalysis
Manuscript submission https://mc03.manuscriptcentral.com/cjcatal
Chinese Journal of Catalysis
Machine-learning-aided discovery of methanol-to-olefins zeolite catalysts with ultra-high initial selectivity
3-Feb-2026