Volume 2, Issue 2 (2026) – 7 articles
Cover Picture: Methane pyrolysis via molten catalysts offers a transformative route for coke-free hydrogen production and high-value carbon capture. However, the development of molten catalysts is hindered by a vast compositional space and the disordered atomic structure of the molten state, which makes traditional trial-and-error experimentation inefficient. Here, we introduce an artificial intelligence-empowered digital catalysis platform (DigMethpy) to accelerate the development of molten catalysts. This platform integrates experimental and computational data with machine learning models, literature-based knowledge bases, and large language models, forming a closed-loop workflow of “data → model → prediction → validation”. It provides a data-centric framework for intelligent catalyst design by iteratively refining prediction models and intelligent agents through data feedback. The platform is poised to evolve from a single-agent workflow toward multi-agent collaboration and a self-driving system, offering a scalable digital infrastructure to connect the research community and accelerate the industrialization of methane pyrolysis.
view this paper 
