2025 FEMS Master Thesis Award – Finalists

Antonios Sarikas | 3rd place

HSSTCM, Greece – University of Crete

From potential energy surface to gas adsorption via deep learning

A generalized framework for predicting gas adsorption properties of porous materials was developed using the potential energy surface as the sole input. The latter is treated as a 3D energy image and processed via a 3D convolutional neural network. The proposed approach was applied to MOFs for predicting CO₂ uptake, where it outperformed conventional models in both accuracy and data efficiency. Additionally, the transferability of the approach to other host–guest systems was demonstrated by examining CH₄ uptake in COFs.

Antonios Sarikas is a Chemistry graduate with a Master’s degree in Physical & Computational Chemistry, and currently a PhD candidate at the University of Crete, focusing on the application of machine learning for studying gas adsorption in metal-organic frameworks (MOFs). His research involves developing predictive models to estimate adsorption properties and leveraging inverse-design approaches to develop tailor-made MOFs for adsorption-related applications.