Designing Anode Materials for Proton Exchange Membrane Electrolysis via Literature Data and Machine Learning

Proton exchange membrane (PEM) electrolysis is a crucial
technology
for sustainable hydrogen production, yet the discovery of efficient
and durable anode materials remains a challenge. In this work, literature
data on PEM electrolysis are systematically collected and analyzed
to identify trends in material selection, synthesis methods, and key
experimental conditions. Statistical analysis revealed that a limited
number of anode and cathode materials dominate the field primarily
due to their catalytic activity and stability in acidic environments.
Supervised machine learning is then employed to predict anode materials
aimed at reducing Ir content in the anode material by using compositional
and experimental descriptors to model the cell performance. Experimental
validation of Ir0.9M0.1O2 (M = Cr,
Zn, Sb, Ho) demonstrated that incorporating these predicted elements
preserved cell performance while enabling a reduction in the Ir content
at the anode. This study demonstrates that integrating data-driven
approaches with literature analysis can effectively guide the discovery
of anode materials, advancing the PEM electrolysis technology.