PEM Electrolyzer Design: The Engineering Trade-Offs

What is a PEM Electrolyzer?

A PEM (proton-exchange-membrane) electrolyzer splits water into hydrogen and oxygen using electricity and a solid polymer membrane that conducts protons from the anode to the cathode. It produces high-purity, pressurized hydrogen and is valued for its fast response and compact design.

How a PEM cell is built

A PEM cell is a sandwich: a proton-exchange membrane in the middle, an anode and cathode catalyst on each side, porous gas-diffusion layers, and flow fields that move water and gas in and out — all clamped between end plates.

The torque on the end-plate bolts does two jobs at once: it seals the cell against leaking pressurized hydrogen, and it presses the gas-diffusion layers, catalyst and membrane into tight electrical contact. Get that compression uneven and you get uneven current, local hot spots, and faster degradation — a theme that runs through the whole design.

The IV curve: how current density sets efficiency, CAPEX and OPEX

There is no single “efficiency” for an electrolyzer. You pick an operating point on the cell’s IV curve, and that point trades efficiency against capital cost.

Push the current density higher and the cell voltage rises, so efficiency falls and operating cost (OPEX) goes up. But each stack now makes more hydrogen, so you need fewer stacks — capital cost (CAPEX) drops. Designing a PEM system is largely about choosing where on that line to sit for a given customer and electricity price.

Temperature adds a second trade-off. Running warmer lifts efficiency, but it also accelerates the side reactions that shorten membrane life. A handy rule: each 1 °C of temperature change shifts cell voltage by about 4 mV. Lay a straight line against the low-current part of a real IV curve and the deviation higher up tells you how much the cell is self-heating — and whether it is under-cooled.

PEM vs. alkaline – which is more efficient?

On paper PEM wins; in a real plant they are about the same.

At the same current density a PEM cell runs roughly half a volt below alkaline. But PEM uses far more expensive materials, so nobody runs it at the low current density where that advantage shows. Run at realistic densities and both technologies end up near 1.9–2.0 V — and therefore near-identical efficiency.

Hydrogen crossover – the #1 design constraint

Crossover — hydrogen migrating from the cathode to the oxygen side — is, in this expert’s view, the single most underestimated factor in PEM design.

Safety

Hydrogen mixing with oxygen over a platinum catalyst is dangerous. The mixture becomes explosive at 4%, so the standards require staying below 2%. Because higher temperature and higher current density both increase crossover, they have to be managed together.

Membrane degradation

Hydrogen reaching the anode reacts to form hydrogen peroxide, which — via Fenton’s reaction with iron ions — produces hydroxyl radicals that attack the membrane’s fluorocarbon backbone. Damage concentrates into pinholes, pinholes raise crossover further, and a single cell failure can cascade into stack failure. In mature stacks, crossover is the number-one cause of failure.

What it forces on the design

To contain crossover, PEM electrolyzers use membranes four to five times thicker than fuel cells, run cooler (~55–60 °C rather than the chemically ideal ~80 °C), and can’t turn down below roughly 15–20% load before the gas mixture gets unsafe — which costs production at the start and end of every solar day.

Catalysts

The headline catalysts are iridium oxide on the anode and platinum on the cathode — but the “hidden” catalysts matter too.

Anode loadings run around 1–2.5 mg/cm² of iridium oxide; the cathode needs far less platinum. Crucially, a cell needs a reservoir of catalyst beyond the bare minimum: as catalyst poisons over thousands of hours and voltage rises, previously inactive catalyst takes over. Papers that report ultra-low loadings often ignore this — and ignore the extra platinum used in recombination and protective coatings, which can exceed the cathode catalyst itself.

Degradation, durability testing and lifetime

Degradation comes from two places — mechanical design (uneven pressure and flow) and water chemistry (metal ions, organics) — and proving a cell’s lifetime takes months.

Targets are brutal: PEM aims for under ~5 µV/hour of voltage rise, yet a single 1 °C swing moves voltage by ~4 mV — a thousand times larger. That’s why durability testing needs tight temperature control and many months of data, and why a “shielding voltage” of ~1.5 V is held during power loss so the iridium catalyst doesn’t dissolve. End of life is typically capped at 2.4 V per cell; choosing a lower current density extends runway and can add ~30% to lifetime.

Stack design and scaling

Good stacks deliver even pressure and even flow across every cell — and scaling them up is far harder than it looks.

Rigid end plates fight the bowing that starves the cell centre of pressure; wide manifolds and well-designed flow fields keep a low temperature spread (a few degrees) across the cell. Circular stacks handle pressure better and can be turned on a lathe; rectangular stacks waste less membrane from the roll and flow better, but bow under pressure. The prize is size: today’s PEM stacks sit around 2–2.5 MW, and the maximum has only roughly doubled in five to seven years — a 10 MW stack would be a genuine game-changer because it collapses sensors, cabling and rectifiers

Balance of plant: water, drag and electrics

The system around the stack exists to keep the plant safe, keep the stack happy, and meet the customer’s hydrogen spec.

Water purity is non-negotiable — aim for ASTM Type 1 (18 MΩ, <50 ppb organics). Electro-osmotic drag pulls about four water molecules across with every proton, so a 1 MW system moves on the order of 30 tonnes of water a day that must be repatriated. Cooling and uneven-flow margins can push pump flow toward ~4,000 L/min, and because pump power scales steeply with flow, sizing it right is a major efficiency lever. On the electrical side, keep the stack under 600 V (usually two stacks of ≤120 cells in series) to avoid high-voltage classification, and keep the control system as simple as safety allows — sensors are the most common cause of nuisance shutdowns.

Applications: oxygen, off-grid and the limits

By-product oxygen rarely pays for medical use, but works well for wastewater and aquaculture; true off-grid operation is possible but expensive.

Cleaning oxygen to medical grade adds cost to a free by-product that competes with already-cheap air-separation oxygen — wastewater treatment and fish farming, which tolerate low pressure and trace hydrogen, are the realistic uses. Off-grid systems must keep power running at all times: lose power and you lose safety monitoring, and a water-filled stack can freeze. Batteries or fuel cells solve it, but the cost climbs fast with how long the outage might last.

Watch a clip from the Webinar

Read next

Article based on a Mission Hydrogen webinar. Technical figures are the speaker’s field estimates, presented for general understanding rather than as design specifications.