Green hydrogen depends on a metal almost no one can name. A proton exchange membrane electrolyzer, which is a device that uses electricity and an acidic membrane to split water into hydrogen and oxygen, needs iridium to survive its own chemistry. Iridium is one of the rarest metals on Earth, with annual mine output near 7 to 7.5 metric tons. An iridium-free electrolyzer catalyst is a material that can drive the oxygen side of that reaction with no iridium at all, which would remove a hard physical cap on how many electrolyzers the world can build.
This article explains the constraint in grams per gigawatt, why acid makes the problem so stubborn, what low iridium and zero iridium results have actually shown, and why laboratory hours are still far from plant lifetime. Every durability claim below pairs lab time with the industrial bar in the same sentence, so a beaker result never reads as a shipping product.
How Much Iridium Does One Gigawatt Need
The math is the story. Catalyst loading means the weight of the catalyst metal spread over each square centimeter of the electrode. Commercial proton exchange membrane systems today use about 1 to 2 milligrams of iridium per square centimeter at the oxygen electrode, which converts to roughly 400 to 800 kilograms of iridium per gigawatt of capacity, depending on cell design and operating voltage.
Against that demand sits a tiny supply. Global iridium output is about 7 to 7.5 tons per year, largely as a byproduct of platinum mining in South Africa, according to the IRENA assessment of green hydrogen cost and materials. The arithmetic is direct: 7,000 kilograms divided by 400 kilograms per gigawatt supports about 17.5 gigawatts, and divided by 800 kilograms per gigawatt supports about 8.8 gigawatts, before any other industry takes its share. At older loadings near 0.7 to 2.5 grams per kilowatt, which equal 700 to 2,500 kilograms per gigawatt, the same 7 to 7.5 ton supply supports only about 3 to 11 gigawatts per year, a range reported in the World Bank review of electrolyzers for hydrogen production.
The U.S. Department of Energy sets the direction of travel in its technical targets for proton exchange membrane electrolysis. Total platinum group metal content for both electrodes was about 3.0 milligrams per square centimeter, or 0.8 grams per kilowatt, in the 2022 status case. The 2026 target is 0.5 milligrams per square centimeter, or 0.1 grams per kilowatt. The ultimate target is 0.125 milligrams per square centimeter, or 0.03 grams per kilowatt, alongside 80,000 hours of stack lifetime. Those loadings are published development targets for researchers to aim at, not current commercial performance. Iridium at the oxygen electrode accounts for much of that total, so hitting the metal target without losing performance or life is the central task.
Why PEM Electrolyzers Depend on Iridium
A PEM electrolyzer works in acid. Protons move through a thin polymer membrane while oxygen forms at the positive electrode and hydrogen forms at the negative electrode. The oxygen side runs the oxygen evolution reaction, which is the step that releases oxygen gas from water. Overpotential means the extra voltage above the theoretical minimum that must be paid to get hydrogen at a useful rate, and lower overpotential means less electricity wasted as heat.
Acid dissolves most cheap metals. Iron, nickel, and cobalt can be excellent catalysts in alkaline equipment, where the liquid is strongly basic, as shown by a nickel-iron-based next-generation electrolyzer and by earlier work on hydrogen from iron, nickel, and water. In the acidic interior of a PEM cell, those metals corrode or lose activity quickly. Iridium oxide survives because it stays largely intact while oxygen bubbles off its surface, and ruthenium oxide is active but dissolves faster unless iridium stabilizes it.
That stability is why current PEM stacks use about 2 milligrams of iridium per square centimeter, or about 0.5 to 0.8 grams per kilowatt at 1.9 volts and 2 amperes per square centimeter, as summarized in the U.S. Hydrogen Shot water electrolysis technology assessment. Current density means work rate per patch of electrode, so 2 amperes per square centimeter is a common industrial pace. The same assessment notes that replacing iridium with transition metal alternatives remains further from commercial use, which is why the field pursues two tracks at once: use far less iridium per cell and prove materials that use none.
What Iridium-Free Really Means
The phrase gets stretched, so precision helps. Low iridium means the catalyst still contains iridium but at a much smaller share or loading. Reduced iridium includes designs that mix a small amount of iridium into ruthenium oxide or spread atoms thinly on a stable support. “Iridium-free” means zero iridium in the catalyst formula, which is the strict sense used in this article and in the focus phrase “iridium-free electrolyzer catalyst.”
Zero iridium does not mean zero scarce metal. Some laboratory iridium-free formulas still contain palladium, platinum, cobalt, or rare additives that carry their own supply or cost questions. A fair reading asks three follow-up questions for every claim:
- What is the full formula. Iridium-free can still depend on another limited metal.
- At what current and voltage. A catalyst that looks efficient at 10 milliamps per square centimeter may face different stress at 2 amperes per square centimeter.
- For how long, and where. Hours in a glass cell with liquid acid are not the same as hours in a full membrane stack.
This distinction keeps comparisons honest. The sections below keep the test conditions next to each number, and the closing section places those lab hours beside the 40,000- to 80,000-hour lifetime that grid-scale stacks must reach.
A Low Iridium Bridge That Lasted 1,500 Hours
The nearest term help may come from using far less iridium rather than none. Researchers at Rice University reported an iridium-stabilized ruthenium oxide with an iridium-to-ruthenium atomic ratio of only 1 to 6, called Ru6IrOx, which cut iridium use by about 80 percent relative to conventional PEM anodes in their tests.
The design point matters. Modeling suggested that iridium placed just below the surface could stabilize the ruthenium oxide lattice with less than 50 atomic percent iridium. The team then showed continuous water electrolysis for more than 1,500 hours at 2 amperes per square centimeter, which is an industrial pace, although 1,500 lab hours are still far short of the 40,000 to 80,000 hour lifetime expected of commercial stacks. Stability was also checked in a 25 square centimeter PEM electrolyzer under industrial test conditions, which is larger than a coupon but still smaller than a commercial stack. The work was published in Nature Nanotechnology on 7 October 2025, volume 20, pages 1787 to 1795.
This is a bridge, not a destination. Ruthenium itself has supply limits, and an 80 percent cut still leaves some iridium in each cell. Yet the result is useful because it held industrial current for a long single run instead of reporting only a brief efficiency peak. It shows how far loading can fall while staying inside a real membrane assembly.
Two Iridium-Free Paths That Survived Acid
Two laboratory systems deserve attention because they removed both iridium and ruthenium and still ran in acid for hundreds of hours. Neither is a product, neither is available for purchase, and neither has run in a commercial stack. They are also different kinds of evidence: the cobalt spinel below ran inside a real membrane assembly at industrial current, while the palladium oxides below ran in liquid acid half-cells at a gentle current about two hundred times lower, so the two results cannot be read as comparable device options.
AI screened palladium oxides that reached 1,000 hours in acid
A multi-institution team used an AI-guided closed-loop platform to screen 2,942 catalysts across 53 material systems and 26 elements, then validated leads in glass half-cells holding 1 molar sulfuric acid at 10 milliamps per square centimeter. That setup tests catalyst films in liquid acid, not in a membrane electrolyzer, and its current is about two hundred times lower than the 2 amperes per square centimeter used in the Rice device test above. The screen surfaced complex oxides such as InMnPdOx and NiTaPdOx, which the authors note conventional design logic would not have predicted. The preprint was posted on 24 September 2026 and revised on 25 September 2026 as arXiv 2609.30133 on iridium and ruthenium free palladium oxide catalysts.
During long-term testing, plain palladium oxide passed 0.5 volts of overpotential after about 200 hours, NiTaPdOx passed the same marker after about 470 hours, while InMnPdOx kept overpotential below 0.5 volts for 1,000 hours of operation, although 1,000 hours at the gentle pace of 10 milliamps per square centimeter are still far short of the 40,000 to 80,000 hour lifetime needed at industrial current. The authors link performance to a nanostructure that supports activity while slowing palladium corrosion, and they note that palladium is more available than iridium or ruthenium, which could ease supply pressure even though palladium is itself a precious metal.
Lanthanum and manganese doped cobalt spinel that ran in a PEM cell
A second path modifies cobalt oxide, which is a crystal family called spinel, with small amounts of lanthanum and manganese. In work highlighted in an Electrochemical Society meeting abstract, the doped cobalt spinel reached an overpotential of 353 millivolts at 10 milliamps per square centimeter in acid and then ran in a PEM assembly at 2.47 volts at 2 amperes per square centimeter for 360 hours using Nafion 115 membrane, although 360 hours in one test cell are still far short of the 40,000 to 80,000 hour lifetime needed for commercial service. The result is notable because the test used a membrane electrode assembly rather than only a beaker cell and because it points to earth-abundant cobalt as the host instead of a platinum group metal. Cobalt brings its own sourcing questions, a pattern also seen in alkaline work, where a nickel-cobalt electrode matched platinum voltage in alkaline tests while still depending on cobalt supply.
Taken together, the two paths show different kinds of progress. The palladium oxide result is a materials screen that ran longest but only as a low-current acid film, while the cobalt spinel result is the closer device test because it ran at industrial current inside a membrane assembly for a shorter time. The scores cannot be ranked against each other, and current density must stay attached to every hour count.
Lab Hours Are Not Plant Lifetime
A durable catalyst is necessary but not sufficient. Commercial stacks must hold voltage within a narrow band for tens of thousands of hours while coping with starts and stops, impurities in water, pressure swings, heat, gas crossover through the membrane, and corrosion of porous layers, plates, and seals. The DOE table allows about 10 percent voltage loss from beginning of life to end of life, with an average degradation rate near 2.0 to 2.3 millivolts per thousand hours on the path to 80,000 hours.
The table below translates the featured results into that context. It keeps lab time and plant need in one view on purpose.
| System | Iridium content | Reported run | What it does not prove |
|---|---|---|---|
| Rice Ru6IrOx | Low, Ir to Ru ratio 1 to 6 | More than 1,500 hours at 2 amperes per square centimeter in lab and 25 square centimeter cells | 80,000 hour stack life with full balance of plant |
| InMnPdOx | Zero iridium and zero ruthenium | 1,000 hours below 0.5 volts overpotential at 10 milliamps per square centimeter in a liquid-acid half-cell, not a PEM device | Stability at 1 to 3 amperes per square centimeter in a large stack |
| La and Mn doped cobalt spinel | Zero iridium | 360 hours at 2 amperes per square centimeter in a PEM assembly | Multi thousand hour life, manufacturing yield and cobalt supply at scale |
| Platinum-free AEM cathode for comparison | No platinum group metals at cathode | More than 1,000 hours at 2.0 amperes per square centimeter in an anion exchange membrane cell, as covered in Intelligent Living reporting on a platinum-free catalyst that sustained industrial current for over 1,000 hours | Transfer to acidic PEM anodes, where materials face harsher corrosion |
Cost needs the same care. Catalyst metal is only part of stack cost, and the stack is only about 30 percent of system cost in some analyses, with power electronics, membranes, porous transport layers, bipolar plates, water treatment, compression, buildings and financing around it. Electricity price and how many hours per year the plant runs at high output usually move hydrogen cost more than grams of catalyst. A membrane-free route such as H2Pro decoupled electrolysis tackling the 1 dollar per kilogram challenge attacks a different cost center, which shows why catalyst progress must slot into full systems rather than stand alone.
Frequently Asked Questions
What is a PEM electrolyzer used for in hydrogen production
It uses renewable or grid electricity to split water into hydrogen and oxygen through an acidic polymer membrane. The hydrogen can feed steel, ammonia, refining, heavy transport and storage, where steady supply and purity matter more than small batch demos.
Which catalyst is used for electrolysis of water
Commercial PEM cells commonly pair platinum at the hydrogen electrode with iridium oxide at the oxygen electrode. Alkaline cells can use nickel, iron or cobalt based materials because the basic environment is less corrosive, but those metals usually struggle in PEM acid without protective design.
Why is iridium so hard to replace
Few materials combine high oxygen activity with slow dissolution in hot acid while bubbles, voltage swings, and impurities attack the surface. Ruthenium is active but dissolves faster. Earth-abundant oxides often corrode or need too much extra voltage, which wastes electricity.
What is the cost of 1 kilogram of green hydrogen
Estimates vary by power price, utilization and location, and most developed market figures sit well above fossil hydrogen near 1 to 2 dollars per kilogram. Catalyst savings help the equipment slice, but electricity and operating hours dominate the final kilogram price.
Can hydrogen be produced efficiently without iridium
Early results say possibly, but not yet at commercial scale. Laboratory cells have held hundreds to 1,000 hours without iridium under stated test conditions, although each of those runs is still far short of the 40,000 to 80,000 hour lifetime expected of industrial stacks. Larger cells, longer runs and independent repeats are the next tests.
How long does an iridium-free catalyst last
There is no single answer because current and cell type change the result. The longest zero iridium and zero ruthenium acid run cited here kept overpotential below 0.5 volts for 1,000 hours at low current in sulfuric acid, while a cobalt spinel ran 360 hours at industrial current in a PEM assembly, and both results are still far short of the 40,000 to 80,000 hour lifetime needed for plant service.
Bottom Line
Iridium turns green hydrogen scale-up into a materials problem. At hundreds of kilograms per gigawatt against mine supply near 7 to 7.5 tons per year, conventional loadings cannot support a hundred-gigawatt buildout without deep cuts in metal per cell, recycling, or new chemistry. An 80 percent cut that held 1,500 hours at industrial current shows the bridge, while 1,000-hour and 360-hour zero iridium runs show where the destination might lie, although each lab total is still far short of the 40,000 to 80,000-hour lifetime that financing and maintenance plans assume. None of the three featured catalysts is available for purchase or operating in a commercial stack. The next credible milestone is not a lower overpotential for a day but a named catalyst that holds industrial current in a larger stack for thousands of hours while reporting loading, voltage, degradation rate, and full formula together.