AEM electrolysis gets a durability slope: under 2% over 3,000 hours in a three-cell stack

Less than 0.67% of performance lost per 1,000 hours ^[14] is the form in which this result becomes usable to anyone building a maintenance budget. Run that line across a calendar year of unbroken operation and it comes to under 5.9% ^[15]. The linearity is where the honesty goes: degradation is not obliged to be a straight line, and 3,000 hours is 125 days ^[18], not the service life a plant is underwritten against.

The hardware is the part that lifts this above a lab curve. An active area of about 99 square inches is roughly 639 square centimetres ^[16], which is stack territory rather than a button cell. But the account does not report current density, cell voltage, operating temperature or platinum loading ^[11], and it does not say whether that area is per cell or the whole assembly ^[12]. A 2% figure with no operating point cannot be converted into cost per kilogram of hydrogen. It can only be compared with other 2% figures.

The failure mode being addressed is compositional. As nickel leaves a conventional platinum-nickel catalyst, the composition and electronic structure change and activity drops ^[4]. The Korea Institute of Materials Science group's answer relies on modelling that says nickel is harder to pull out of an ordered lattice ^[3][1], and the measured gap is roughly six times more nickel lost from the disordered material ^[17]. That comparison comes from durability testing rather than from the 3,000-hour stack run ^[13], so nickel retention and the sub-2% figure cannot be paired as cause and measured effect, even though they point the same way.

There is a mismatch between the demonstration and the cost argument attached to it. The researchers note that replacement and maintenance add significantly to operating costs, particularly at larger scale and where renewable supply introduces fluctuating operating conditions ^[10]. The 3,000 hours were continuous ^[2]. Steady load is the easier case, and it says nothing about what the ordered structure does under exactly the conditions the cost argument identifies as expensive. The claimed benefits, longer catalyst life and lower replacement cost ^[9], are therefore demonstrated in the regime that was never the problem.

Then there is the platinum. AEM electrolysis draws interest partly because it can cut reliance on expensive precious metals compared with some conventional routes ^[7], while the alkaline hydrogen evolution reaction is slow enough to demand real catalytic activity in the first place ^[8]. The durability fix here still contains platinum, and whether that is affordable depends on grams per kilowatt, which is the one number the account omits ^[11]. Durability has moved from a single-cell curve to a stack-level slope, which is a genuine step. The disclosure that would change a model, though, is loading and duty cycle, not more hours at steady state.