Highly efficient decomposition of ammonia using high-entropy alloy catalysts

If you care about hydrogen as a clean fuel, ammonia is one of those quietly brilliant ideas. It's a liquid under a modest pressure—about 8 bar at room temperature—and it packs around 4.25 kilowatt-hours into every liter. You can move it, store it, and ship it like a chemical commodity. But to use it as hydrogen, you have to crack it, and for years the gold standard catalyst for that step has been ruthenium. Fantastic performance, sure. Also, it's scarce and pricey. That bottleneck has kept a lot of good ideas stuck on the drawing board. Xie and colleagues set out to dodge that bottleneck by doing something counterintuitive: instead of using one precious metal, they blend five earth-abundant ones into a single, well-mixed alloy at the nanoscale. Their recipe—cobalt, molybdenum, iron, nickel, and copper—forms what's called a high-entropy alloy, essentially a solid solution where no single element dominates. The particles sit on carbon nanofibers, they're roughly 22 nanometers across, and the support offers a generous surface area in the 150 to 160 square meters per gram range. Metal loadings hover around 8 to 9 percent by weight. The trick is not just the mix; it's the control. They can dial the cobalt-to-molybdenum ratio across a broad range without the system separating into the usual, less active cobalt-molybdenum phases. The way they make these things is almost cinematic. They soak the nanofibers with metal salts and then hit the whole film with a burst of Joule heating that lasts about 55 milliseconds. In that blink, the temperature leaps into the 2000 to 2300 Kelvin range with a ramp rate on the order of one hundred thousand degrees per second, then cools back down. You get tiny liquid droplets, which crystallize into a single face-centered cubic alloy as they freeze. Five compositions spanning cobalt-15 molybdenum-55 to cobalt-55 molybdenum-15 fall out of this same shock process, all in the same phase, and microscopy shows crisp lattice planes with a spacing of about 2.18 angstroms. In plain English: uniform structure, well-mixed chemistry. What do they do with ammonia? A lot. At 500 degrees Celsius, the best two compositions post mass-specific rates of 22.1 and 16.7 grams of ammonia decomposed per gram of metal per hour. In head-to-head runs under the same conditions, that's roughly nineteen- and fourteen-fold faster than a bimetallic cobalt-molybdenum control, and they also outpace a ruthenium on carbon benchmark. If you care about rates per surface area, the top performer delivers about 0.74 grams per square meter per hour, which translates to an advantage of roughly 24 times over cobalt-molybdenum and about 19 times over ruthenium. For a mix of cheap, abundant metals, that's a seismic shift. The temperature profile tells the same story. These high-entropy catalysts wake up around 300 degrees Celsius. The most active composition hits 50 percent conversion at about 422 degrees, while the least active needs closer to 558. The star of the series drives ammonia all the way to full conversion at around 525 degrees, whereas the cobalt-rich tail isn't fully across the line until roughly 600. The spacing between those numbers matters; it's a way of seeing that composition really is a tuning knob, not just a label. Under the hood, the chemistry hinges on how nitrogen sticks to the surface. Too weak, and you can't activate ammonia. Too strong, and the surface clogs with nitrogen atoms that won't leave. Xie's team captures this balance with a simple descriptor: the adsorption energy of nitrogen, which they call delta E sub N. If you plot the reaction rate against that energy, you get a volcano—low at the extremes, high in the middle. Their best alloy, cobalt-25 molybdenum-45, sits right at the sweet spot with a delta E sub N of about 79 kilojoules per mole. That's close to what people report for ruthenium, around 84, and it also lines up with theory that put the optimum near 74 under the same 5 percent ammonia feed. Change the feed to pure ammonia, and the predicted optimum shifts; models suggest something like 39, reminding us the "right" binding strength depends on the gas you're actually running. How do they obtain delta E sub N from an experiment? With temperature-programmed desorption, or TPD, on nitrogen. You dose the surface, then heat it up at different rates and watch when nitrogen leaves. The key equation relates the heating rate, called beta, and the peak desorption temperature, T sub m. If you plot the natural log of beta divided by T sub m squared against one over T sub m, the slope gives you the desorption energy. Do that across the series, and two things stand out. There are two main desorption peaks, one tied to flat, ordered facets and another to more undercoordinated sites. As you move from cobalt-rich to molybdenum-rich, the energy rises sharply—from about 42 to 44 kilojoules per mole at the most cobalt-rich end to about 121 to 125 at the most molybdenum-rich. That's the right-hand side of the volcano: great at binding nitrogen, not so great at letting it go. The kinetic fingerprints back this up. On cobalt-lean, molybdenum-rich alloys, hydrogen becomes a stronger inhibitor, which you see as the hydrogen reaction order trending more negative. Shift toward cobalt, and ammonia activation starts to pinch the rate, reflected in an ammonia order that climbs toward about one. The apparent activation barrier tells the same tale in a single number: as low as 64.7 kilojoules per mole for the best alloy and up at 123.1 for the least active. Put all of that together, and you're looking at classic Sabatier behavior—a Goldilocks zone for nitrogen binding that you can reach not by swapping metals entirely, but by nudging composition inside a single solid solution. Now, all of this relies on a very specific surface—one where cobalt and molybdenum are thoroughly intermixed so that "sites" are actually little ensembles of neighbors, not islands of one metal next to another. That's where the computational work matters. The team modeled the standout composition, cobalt-25 molybdenum-45 with iron, nickel, and copper each around 10 percent, as a face-centered cubic nanoparticle and ran Monte Carlo sampling for ten million steps. Across three temperatures—573, 750, and 1000 Kelvin—the atoms shuffle but do not segregate. There's no long-range chemical ordering. You do see modest short-range tendencies, especially around molybdenum, but overall the surface stays well mixed. A simple thermodynamic screen points the same way: a size-mismatch parameter near 5.4 percent and a small, positive mixing enthalpy combine with the entropy of mixing to favor a stable solid solution under reaction conditions. That's the structural foundation for a "site-mixing" mechanism. Why harp on structure? Because stability is the other half of the story. Run the best alloy at 500 degrees Celsius for about 50 hours, and the activity barely budges. Post-run analysis finds only around 1.4 atomic percent nitrogen on the catalyst, with no sign of the bulk turning into a nitride. That's the kind of result that keeps engineers in the room. High activity is exciting; high activity that lasts is what makes it into a reactor. There's another way to see the magnitude of this shift. In addition to those head-to-head comparisons at 500 degrees, Xie and colleagues report even larger wins in some formulations and normalizations—improvements on the order of twenty-fold over ruthenium, and in one set of comparisons, more than a hundred-fold over a bimetallic cobalt-molybdenum. Different metrics tell slightly different stories, but the through line is clear: these five-element alloys can match or outstrip ruthenium without leaning on scarce elements. Let's stitch the mechanics and the materials into one design principle. By moving along the cobalt to molybdenum axis inside a single, thermally stable high-entropy alloy, you tune how strongly nitrogen binds. That one knob—delta E sub N—predicts where you'll land on the volcano of activity. The surface itself is an ensemble, not a patchwork, so you're adjusting the average "feel" of many neighboring atoms at once. And because the alloy remains a single phase across that range, your tuning is continuous, not stepwise. That coherence, from synthesis to structure to kinetics, is what makes this work by Xie, Yao, Huang, Liu, Zhang, Li, Wang, Hu, and Shahbazian-Yassar feel like more than a one-off. A quick word on how this could play out in practice, keeping our feet on the ground. The authors suggest you could even stage the catalyst in layers—start with a composition that binds nitrogen a bit more strongly where ammonia is plentiful at the reactor inlet, then gradually shift to weaker binding deeper in the bed where nitrogen atoms have to recombine and leave. That's a concept, not a field trial. But it's the kind of idea you can only entertain once you have a handle on a descriptor you can actually control. We're still a step away from a drop-in unit bolting onto every ammonia pipeline. Real-world feeds change, impurities matter, thermal cycling can surprise you, and reactors impose their own constraints. But the essentials here are solid. Ammonia is an attractive hydrogen carrier. Ruthenium has been a standout catalyst with a supply problem. And now there's a family of earth-abundant, single-phase, high-entropy nanoparticles that not only hold their own, but in many cases beat ruthenium on its own turf, while letting you steer the surface chemistry like a dimmer switch rather than a light switch. That's a promising place to build from.