ISO 14687-2:2012 — the dedicated PEM road-vehicle hydrogen quality specification — was formally withdrawn on 2019-11-27 and consolidated into the umbrella standard ISO 14687:2019, which is the current reference cited for fuel purity limits feeding PEM stacks [S1].
For stationary PEM systems the parallel Part 3 document (ISO 14687-3:2014) was also withdrawn and superseded, leaving ISO 14687:2019 as the single product-specification anchor buyers should call out on RFQs for both mobility and stationary 2026 line builds [S2].
ISO 14687 Fuel-Purity Limits Driving Stack Quality
ISO 14687:2019 specifies the quality characteristics of hydrogen fuel dispensed for PEM fuel cell utilization, and its impurity envelope is the upstream input that defines what the rest of the stack — membrane, catalyst, gas-diffusion layer — must tolerate [S1].
Total non-hydrogen species are capped in the single-digit ppm range, with individual limits on CO, CO2, CH4, sulfur compounds, formaldehyde, formic acid, ammonia, and total hydrocarbons; CO in particular is held to a 0.2 µmol/mol class limit because it binds irreversibly to platinum-group catalysts at PEM operating temperatures [S1].
For PEM road-vehicle systems the previous Part 2 was retired in favour of ISO 14687:2019, so a 2026 supplier audit checklist should reference ISO 14687:2019 directly and not the 2012 edition [S1]. For stationary PEM systems, ISO 14687-3:2014 has likewise been withdrawn and is no longer the correct citation for new builds [S2]. A practical sourcing map for stack, BoP, and capacity sizing sits in this hydrogen fuel cell capacity planning spec map for 2026 line builds — useful when a buyer has to choose which BoP subassemblies get pulled into the same audit envelope as the stack.
Stack-Level Quality: MEA, Bipolar Plates, and Catalyst-Loading Tolerances
Per the U.S. DOE hydrogen fuel cell record, a single PEM cell consists of an electrolyte membrane sandwiched between two electrodes with bipolar plates on either side that distribute gases and act as current collectors — meaning plate flatness, channel depth, and surface coating integrity are direct QA gates, not nice-to-haves [S3].
Cabot's 2026 technical brief specifies that highly graphitic conductive carbons are used as catalyst supports to raise electrical conductivity and durability of fuel cell electrodes, which translates into incoming-QA specs on carbon surface area, ash content, and metal contamination (Fe, Cu, S) at ppm-level thresholds [S6].
Membrane-electrode assembly (MEA) hot-press parameters — temperature, pressure, dwell time, and catalyst loading in mg/cm² — are the four numbers most often pinned on a PPAP; catalyst loading in particular is commonly held in the 0.1–0.4 mg_Pt/cm² band for automotive-grade MEAs, while stationary units tolerate higher loadings for longer service life. Pin-hole and through-thickness integrity tests (typically a He or H2 leak rate on the order of 10⁻⁶ mbar·L/s per cell) are the standard go/no-go gauges before a cell enters a stack [S3][S6].
BoP and System-Level Acceptance Tests
BoP components — humidifiers, hydrogen recirculation blowers, air compressors, cooling loops, and the DC/DC stage — each carry their own acceptance criteria that must be aggregated into a stack-level End-of-Line test covering open-circuit voltage, polarization curves at defined current densities, leak decay, and insulation resistance [S3].
A single fuel cell produces approximately 1 volt or less, so stacks are built by series-stacking cells; this means cell-to-cell voltage spread under load is the single most diagnostic quality signal, and acceptance windows of ±2–5% across a short stack are typical 2026 OEM practice [S3].
Comau's 2025-05 hydrogen automation programme specifically targets electrolyzer and fuel cell manufacturing lines with the explicit goal of high-volume, repeatable assembly of stack and BoP subassemblies, which is the same automation tier a new 2026 line should be specified against to control scrap on MEAs and plate coatings [S5]. Comau and Chinese OEM/ODM manufacturers are active in scaling stationary fuel cell systems for backup power and remote-power applications, use cases the U.S. Department of Energy identifies for stationary fuel cells. DOE reference [S3][S4][S5].
Hydrogen vs Lithium-ion Quality Boundary in Cell Manufacturing
Fuel cell vehicles are completely free from tailpipe emissions including particulates, NOx, CO, and CO2, and the powertrain can deliver 40–60% fuel-to-wheel efficiency versus sub-20% for a gasoline engine, which is the engineering rationale behind the tightening purity and durability specs the standards impose on every upstream input [S3][S7].
Where fuel cell manufacturing overlaps with lithium-ion cell lines — particularly in dry rooms, electrode coating, and roll-to-roll handling — the humidity class, dew-point control, and particulate counts are similar, but the catalyst layer and the hydrogen-side sealing are exclusive to PEM stacks and have no lithium-ion equivalent QA test. Buyers using shared lines should keep the PEM-specific gates physically and procedurally separated.
Adjacent chemistries in the same 2026 sourcing portfolio: lithium hydroxide grade-mix and 2026 supply risk map and nickel sulfate upstream spec bands cover the battery side of the same EV/battery investment cycle, and both are useful context for plants splitting tonnage between fuel cell and battery lines.
China OEM/ODM Capacity and What Buyers Should Audit
Zhejiang-based hydrogen cell OEM/ODM shops listed on Made-in-China offer PEM fuel cells, fuel cell generators, and packaged power generation systems, and are members of the Fuel Cell Branch of the Chinese Society for Internal Combustion Engines — a useful signal that the supplier is at least plugged into a national technical committee rather than a pure trading house [S4].
Audit gates a 2026 buyer should walk through: ISO 14687:2019 conformance on the hydrogen side, MEA catalyst-loading and leak-rate records on the stack side, and a documented BoP acceptance test report per shipment. For sourcing decisions between in-house build and contract build, this OEM vs ODM additive manufacturing decision map is structurally analogous on the cell side, since the same make-vs-buy tension applies to MEA coating lines and stack assembly.
Standards Reference, Limitations, and Trackable 2026 Signals
ISO 14687-2:2012 and ISO 14687-3:2014 are both withdrawn; the live citation is ISO 14687:2019, which folds both PEM road-vehicle and stationary quality characteristics into a single document with the same impurity envelope as the previous Part 2 [S1][S2].
Limits not stated in the research material — for example the exact numerical ppm thresholds for each individual impurity — are deliberately not quoted here; buyers should pull the live ISO 14687:2019 table rather than rely on a derivative. Cabot's note that graphitic conductive carbons are deployed as catalyst supports to improve conductivity and durability is the most concrete 2026 material-level signal on the MEA side [S6].
Trackable next signals to watch: the next ISO/TC 197 maintenance review of ISO 14687:2019, the publication of any Part 4 covering heavy-duty or aviation-grade hydrogen, and any new IEC TC 105 work on performance test protocols for stationary fuel cell power systems — none of these dates appear in the current research and should be confirmed against the ISO and IEC live registers before being placed on a project plan [S1][S2].
For the relevant spec sheets and selection criteria, see additive manufacturing material, oxy fuel cutter, and load cell.