A stack is cells in series, held at the right pressure and fed the same electrolyte. Most of what separates a good stack from a bad one is not the chemistry inside the cells but the mechanical engineering between them.
A single electrolysis cell produces hydrogen at a voltage of roughly two volts and whatever current its active area can carry. That is not a useful industrial machine. A stack is what you get when you put many cells in series, feed them all from one electrolyte loop and hold them together under compression.
It sounds like a packaging problem. It is not. Two stacks built from identical MEAs can produce noticeably different curves, and the difference lives in the parts that are not the MEA.
How a stack is put together
Only the first of those is chemistry. The other three are mechanical engineering, and they are where two stacks built from the same MEA start to differ.
The MEA is the membrane with a catalyst layer on each face, and it sets the efficiency ceiling of the whole machine. The porous transport layer between the catalyst layer and the plate has to conduct current into the catalyst and let gas out of it at the same time, which are competing requirements. On the anode side of an AEM cell it is typically nickel foam.
The bipolar plate separates one cell from the next, carries current from the back of one to the front of the next, and holds the flow field that distributes electrolyte across the active area. In our HXS stacks these are nickel-plated steel and the end plates are SUS316L.
What a stack designer actually chooses
- Active area. Sets the current a cell carries at a given current density, and how hard the electrolyte is to distribute evenly. A large plate is not a small plate scaled up, because flow uniformity gets worse faster than area grows.
- Cell count. Sets stack voltage. Our HXS-2 is a 23-cell stack, and the taller a stack gets the more a single weak cell drags the total.
- Compression. A genuine optimum rather than a maximum. Too little and contact resistance rises, which shows up directly as lost efficiency. Too much and the MEA and porous layer are damaged.
- Flow and sealing. Decides whether every cell sees the same electrolyte at the same temperature. Shunt currents and leaks live here, and both are design problems rather than manufacturing defects.
What matters about compression is uniformity across the plate rather than the absolute figure, because a stack compressed unevenly has cells operating at different conditions inside one housing.
Why this is where scale-up problems appear
A result in a single test cell is a materials result. Reproducing it in a 23-cell stack is a different problem, and reproducing it across every stack a factory ships is a third one.
The failure modes differ at each step. A test cell fails through its MEA. A stack fails through sealing, compression, flow distribution and the one cell that is not like the others. A production line fails through variation that was invisible when there was only one unit to look at.
This is the practical reason we design and build the cell, the stack and the system ourselves rather than assembling purchased components. When the MEA, the plate and the compression are three suppliers' decisions, improving any one of them lands in somebody else's design and comes back as a constraint.
What a stack datasheet should tell you
- Active area and cell count
- Rated current and voltage
- Electrolyte concentration and temperature
- Pressure range, and crossover across it
- Measured degradation, with the conditions of the run attached
If a supplier will not answer the second list, the first list cannot be compared with anyone else's.
What is usually missing is more informative than what is there: whether a quoted performance figure is one cell or the whole stack, the spread between the best and worst cell, the current the durability run was actually held at, and what the stack does at low load and during shutdown.
Our stacks
| Product | What it is |
|---|---|
| HXS-0 | Test cell, 1 to 25 cm² active area |
| HXS-2 | 2 kW stack, 23 cells |
| HXS-30 | 30 kW stack, from Q4 2026 |
| End plates | SUS316L, same across the three |
| Bipolar plates | Nickel-plated steel, same across the three |
| Published HXS-2 run | 2,281 h at 50 A, 0.3 M KOH, 40 °C, ambient pressure |
The HXS-0 exists so that a customer evaluating their own membrane or catalyst can measure it in the same cell architecture we use, rather than in a fixture that behaves differently from a stack. Datasheets for all of these are on the site in English without a form.
- What is an MEA→The component inside every cell in that stack, and what decides whether it is any good.
- What pressurised operation costs→28 mV per cell at 50 A, and the crossover measurements that set the real lower bound on load.
- Reading 2,281 hours of stack durability data→What a 23-cell stack does over 2,281 hours, including the cell-to-cell spread.
- Datasheets→Six products in English, no form.
Looking at AEM for a specific application?