Domain packs

Live now Molecular energy

PEM electrolysis catalysts

The live pack. Screen acidic oxygen-evolution catalysts and see what each candidate does to cost per kilogram.

Get early access Try the free calculator →

The problem

Green hydrogen from PEM water electrolysis needs a catalyst that drives oxygen evolution efficiently, survives acid at potential, and uses very little iridium. Those three pull against each other, and the third is a hard physical constraint: world primary iridium production is roughly seven to eight tonnes a year, as a by-product of platinum mining, so supply does not respond to demand.

The candidate space runs to millions of compositions, structures and surfaces. A laboratory can synthesise and test perhaps five to twenty a year. Almost everything that could be tried, never is.

How the pack works

  1. Relaxation and surface generation with machine-learned interatomic potentials, fine-tuned for acidic oxide surfaces rather than general bulk crystals.
  2. OH*, O* and OOH* adsorption energies across candidate sites, converted into a thermodynamic overpotential descriptor with uncertainty.
  3. A dissolution-risk estimate under acidic OER conditions, because stability is what actually kills PEM catalysts in service.
  4. Propagation through a PEM cell and system model to kWh/kg H2, iridium g/kW and levelised cost, so candidates rank on the commercial objective.
Candidate space
Compositions, structures and surfaces of acidic oxide catalysts
What one experiment costs
Synthesis and electrochemical testing — 2–3 months per batch
Predictor
Fine-tuned machine-learned interatomic potentials
Decides on
Levelised cost of hydrogen (€/kg)

Questions

Is this a replacement for DFT?

No. It is the screening layer that decides which candidates deserve converged DFT or a laboratory slot. Triage, not a verdict.

How much iridium reduction is meaningful?

Ask for grams per kilowatt rather than a percentage — percentages need a baseline and baselines vary enormously. Widely discussed targets for large-scale deployment sit well below 0.1 g/kW.

What if my composition is unusual?

If it falls outside the validated domain the run is refused with an explanation rather than returning a number we cannot defend.