Catalyst loading is decided in the lab, but compression and conditioning decide whether it survives

A stack engineer reads the first polarisation curve and decides whether the assembly has a future. Voltage climbs, temperature settles, gas flows match the plan. The failure, when it comes, starts inside a joint nobody measured on the drawing.

Catalyst loading sets the current the cell can produce. Compression sets how much of that catalyst is actually reachable, and conditioning sets whether the membrane survives its first week. Suppliers quote the loading and stay quiet about the other two.

Most early failures are mechanical. A gasket that compresses unevenly, a plate with a burr, a torque sequence improvised under time pressure: each leaves the electrode pair with different contact pressure, and the lowest point sets the behaviour of the whole cell.

Catalyst loading numbers

Loading is quoted in milligrams per square centimetre, and the figure hides as much as it reveals. A cathode with high loading and poor dispersion performs like a lean electrode, because only the catalyst touching the ionomer and the gas path does any work.

Buyers comparing quotes need loading, platinum group metal content and active area in one view, which is why procurement teams read the membrane electrode assemblies market by application and electrode geometry rather than by price per sheet.

Loading also drives the water balance. A thick catalyst layer holds more product water and resists drying at the outlet, while a thin layer responds faster but punishes any humidity excursion. The choice follows the application, not the datasheet ranking.

Compression set

Compression is the variable most often set by feel. Enough force closes the gas path and lowers contact resistance; too much crushes the gas diffusion layer, blocks the pores and creates a path for membrane creep that shows up months later.

The number to control is the compression of the diffusion layer, not the torque on the bolt. Torque spreads through the stack and depends on thread friction, so two stacks tightened to the same value can hold different pressures and age differently.

Pressure mapping on a witness sheet costs an afternoon and answers the question directly. It shows the low spots before they become leaks, and it tells you whether the fixture, the plate flatness or the gasket is responsible.

Conditioning protocol

Conditioning is where the membrane hydrates, the ionomer network rearranges and residual impurities leave the electrode. Rush it and the cell reaches its operating point in an unstable state; the damage stays invisible until the first load cycle after a shutdown.

The protocol should specify temperature, humidity, current steps and the time spent at each. Suppliers publish a recommended ramp, and it is a starting point: the same membrane in a different architecture needs a different schedule and its own record.

Recording the conditioning run matters as much as performing it. A cell that enters production as the reference unit needs a written history, otherwise every later drift argument ends with two engineers describing different starting points in the same meeting.

Water management

Water is produced at the cathode and dragged through the membrane by protons, so the anode dries while the cathode floods. Both failures look similar on a voltage trace, which is why humidity diagnosis needs pressure drop and impedance, not voltage alone.

Microporous layers and the diffusion media underneath decide how quickly product water leaves. A layer that is too hydrophobic pushes water to the outlet and dries the inlet; too little treatment and the pores fill under high current.

Operating temperature buys margin. Warmer cells carry vapour away more easily and tolerate wider humidity windows, but they also accelerate membrane degradation and chemical attack on the catalyst. The setpoint is a compromise that belongs in the test plan.

Gas diffusion layers

The diffusion layer controls the triple phase boundary as much as the catalyst does. Porosity, thickness and the binder treatment determine how gas reaches the catalyst and how water leaves, and a change in supplier alters all three at once.

Compressibility is the property most buyers ignore. A stiff layer needs higher stack pressure to make contact, which loads the membrane and the seals, while a soft layer compresses into the flow field and chokes the channels at the landing points.

Thickness tolerance across a roll decides how uniform the stack behaves. A layer that varies along its width gives one cell a different compression from its neighbour, and the stack then runs with a temperature profile nobody designed.

Plate and seal interfaces

Most field leaks start at the interface, not in the electrode. Seal grooves that are overfilled, plates with coating defects and coolant ports that sit too close to the gas path all create paths that no amount of membrane quality can close.

Seal material has to match the coolant and the gas chemistry. Ethylene propylene systems behave differently from silicone in the presence of glycol, and a seal that swells by a fraction of a millimetre changes compression across the whole active area.

Stack design should leave room for relaxation. Compression loss over time is normal, and a design with no margin turns a routine change into a re-torque campaign that disturbs every joint and every seal.

Stress test limits

Accelerated protocols compress years of operation into weeks, and the compression changes the failure mode. A test that cycles humidity faster than the cell can respond rewards designs that would fail slowly in the field, or punishes ones that would last.

Load cycling and start-stop cycling age the electrode differently. Start-stop conditions attack the catalyst support through potential excursions, while steady load cycling mostly loads the membrane and the seals, so one protocol cannot stand for both.

Report what the test actually measured. Durability claims that omit the cycle definition, the compression fixture and the conditioning history cannot be compared across suppliers, which is exactly why they keep appearing in marketing material.

Cost per kilowatt

Cost per kilowatt is the only comparison that survives a change of application. A cheap assembly with high loading and short life can cost more per operating hour than an expensive one that holds voltage across its whole service interval.

The calculation needs the platinum group metal content, the scrap value at end of life, the expected replacement interval and the balance of plant. Most buyers have the first figure and guess the rest, which is how a cheap sheet wins a tender.

Volume changes the answer. A small fleet justifies a higher unit price for consistency, while a large programme benefits from a supplier who can hold tolerance across many rolls and document every batch without being chased.

Supplier qualification

Qualification starts with samples and ends with a process audit. Sourcing teams that already buy membrane elements and separation media for water treatment tend to apply the same logic here, which helps on documentation and misleads on electrochemical testing.

Ask for the loading distribution across the roll, not the nominal figure. Ask how the catalyst is dispersed and what the ionomer ratio is. Suppliers who answer precisely are usually the ones who control the coating line and the drying ovens.

Second sourcing protects the programme but doubles the validation. Two suppliers making the same specification can deliver different compression behaviour, so a second source needs its own conditioning run and its own stack test before it ships to a customer.

Field returns

A returned stack carries more information than any bench test. The leak location, the compression set of the gasket and the state of the outlet diffusion layer tell you whether the failure came from assembly, from operation or from the material itself.

Cut the cell open before shipping it back. Photographs of the electrode surface, the membrane edge and the seal groove are cheap and settle arguments that otherwise run for months between supplier and customer with no shared evidence.

Feed the findings into the next build. A design change that never reaches the assembly instruction is a change nobody makes, and the same joint fails again on the next stack with the same complaint attached in the service report.

FAQ: questions buyers ask

Does higher loading help?

Only up to the point where the extra catalyst can be reached by gas and ions. Beyond that, thicker layers add mass transport resistance and hold water at the outlet. Loading is a design decision tied to the application, not a quality badge.

What compression should a stack use?

Follow the diffusion layer specification and confirm it with a pressure map. Bolt torque is a poor proxy because thread friction absorbs much of it. Start from the recommended range, then check that every cell sees similar pressure on a fully built stack.

How long should conditioning take?

Long enough for the cell to reach a stable voltage at each current step, which varies with membrane thickness and cell design. A fixed schedule copied from another stack is a guess, so record the ramp that worked here and reuse it.

Why does voltage drift after shutdown?

Shutdown lets the membrane dry and the catalyst surface change, so the first cycles after restart run at different humidity. If the drift does not recover within a defined window, look at sealing, compression and gas supply before blaming the electrode itself.

Can two suppliers be mixed?

Not without validation. Two sheets meeting the same specification can differ in compressibility and water transport, and a stack tuned to one will not behave identically with the other. Run a full stack test before mixing batches from two suppliers in production.