APM-ML
Advanced porous materials for hydrogen separations – smart development and characterisation approaches supported by machine learning (APM-ML) addresses a key challenge in the hydrogen transition: how to distribute hydrogen efficiently using existing natural gas infrastructure, and then recover it where it is needed. The project focuses on developing advanced membrane materials that can selectively separate hydrogen from hydrogen/natural gas mixtures, offering a potentially scalable and lower-energy alternative to conventional purification technologies. 
The research combines expertise in porous materials, membrane engineering and digital experimentation through a collaboration between the University of Luxembourg and the National University of Singapore. A particular emphasis is placed on structured MOF- and COF-based materials, as well as polymer-based composite membranes, with the aim of achieving both high hydrogen selectivity and high permeance under realistic operating conditions. In parallel, the project develops robust supported membrane architectures that are relevant not only for fundamental studies, but also for future scale-up and industrial application.  
A distinctive feature of APM-ML is its integration of machine learning into the experimental workflow. The project includes the development of a fully automated membrane testing platform capable of generating high-quality, reproducible performance data and using those data to guide smarter experimental design. This is paired with an open cloud-based repository for materials and membrane characterisation results, supporting international collaboration, open science and, ultimately, faster discovery of high-performance hydrogen separation materials.  
At the University of Luxembourg, the project contributes to broader efforts in hydrogen technologies and clean energy systems by linking advanced materials development with digital tools, automation and international research training. Beyond scientific publications, APM-ML is intended to create new platforms, data resources and testing capabilities that can strengthen future research and accelerate the deployment of hydrogen in industry and energy applications.
This work is supported by the Luxembourg National Research Fund (FNR) under Grant AFR Bilateral/17112420.
