Fe/Cu diatomic catalysts for electrochemical nitrate reduction to ammonia

If you’ve ever worried about nitrates in drinking water and wondered how we might turn that pollution into something useful, here’s the hook: there’s a way to clean the water and make fertilizer at the same time. Electrochemistry can pull nitrate down to ammonia at room temperature and near atmospheric pressure. That’s elegant. The catch is, it’s slow and messy on most electrodes. Nitrate doesn’t like to stick. Hydrogen does, so you get the hydrogen evolution reaction muscling in and stealing electrons. Even when nitrate does react, it can wander off into nitrite, nitric oxide, or other nitrogen byproducts instead of the ammonia you want. Zhang and colleagues tackled those two bottlenecks—weak nitrate adsorption and too many side roads—with a deceptively simple idea: put two different metal atoms right next to each other and cradle them in the right carbon-nitrogen scaffold. Their catalyst is a iron-copper heterodimer anchored in the edge holes of nitrogen-doped graphene, which they call iron-copper H N G. Picture a perforated graphene sheet whose hole edges are laced with nitrogen atoms. Into that “holed” rim, they tuck an iron atom and a copper atom so close that the two metals bond to each other, each also coordinated by roughly two nitrogens. Chemists would call it a Y-type M L 3 site: two nitrogen ligands and a metal–metal pair. Why go to all that trouble? Because two unlike metals can split the work. One can grab and activate nitrate; the other can ease the later hydrogenation steps. The nitrogen-doped, holey graphene does two more things. It offers a lot of edge sites where atoms bind strongly, and it moves reactants in and out fast. The surface area here is big—about 858 square meters per gram—with pores clustered around two to three nanometers. That’s a lot of accessible edge to host a lot of active pairs. You always want to check whether the catalyst is what you think it is. The microscopy reads like a roll call of atomic-scale evidence. High-angle annular dark-field images show speckles you’d expect for single atoms, but you also see paired bright dots—diatomic neighbors—again and again. Energy-dispersive line scans across those pairs give a metal–metal separation centered around 2.3 angstroms. There’s no sign of metal nanoparticles in the diffraction patterns, just broad carbon peaks, which is what you want if the metals are truly isolated. The Raman spectrum shows more defect signal than pristine graphene, consistent with nitrogen doping and the added holes. X-ray absorption seals it. At the copper edge, the spectra land between copper one plus and copper two plus references; at the iron edge, between iron two plus and iron three plus. More telling, the iron edge shifts slightly higher in energy when iron is paired with copper versus iron alone, a fingerprint of electron transfer from iron to copper. Extended fine-structure analysis picks out metal–nitrogen paths and a distinct iron-copper path, with coordination numbers near two nitrogens per metal—right in line with that M N 2 edge-site picture. The distribution of iron and copper loadings—about 3.3 and 2.8 weight percent—also matches a dense field of atomic sites rather than large particles. How do they build it? The recipe is straightforward once you know the destination. Start from graphene oxide, carve more holes into it chemically, then add iron and copper salts and a nitrogen source under hydrothermal conditions. A final anneal in ammonia at high temperature fixes the nitrogen into the carbon lattice and locks the metal pair into those edge sites. The “holey” part is not cosmetic; those rim nitrogens are the anchors that let the two metals sit side by side without migrating and clumping. Now, the fun part—what it does. In alkaline electrolyte with nitrate present, the catalyst takes off. At a modest potential of minus 0.3 volts versus the reversible hydrogen electrode, you see around 38.5 milliamps per square centimeter. The headline, though, is selectivity. At that same potential, the Faradaic efficiency to ammonia hits 92.51 percent. That’s a significant improvement over controls. Iron alone on the same graphene framework, copper alone, even a physical mixture of the two—none of them get close to that peak. Push the potential to minus 0.5 volts and you trade a bit of efficiency for speed: the ammonia production rate reaches about 1.08 millimoles per hour per milligram of catalyst. If you factor in the energy it takes to make each gram of ammonia under these conditions, you get about 8.76 watt-hours per gram—leaner than many lab-scale electrochemical routes. Selectivity numbers are great, but you always want to be sure the ammonia you’re measuring actually came from nitrate and not a stray contaminant. The team ran the gold-standard isotope check. When they fed in nitrate labeled with nitrogen-15, the ammonium peak in the proton nuclear magnetic resonance, or N M R, moved exactly where nitrogen-15 N H 4 plus should be. With unlabeled nitrate, you saw the nitrogen-14 N H 4 plus signatures. The ultraviolet-visible colorimetry tracking those products aligned with the N M R, so the quantification was consistent across methods. If you ask why this iron-copper pair performs so well, theory and operando mass spectrometry point to the same thing: you need to grab nitrate strongly at the start and then not hold on too tightly to the intermediates. On the computer, nitrate’s binding energy on the heterodimer is about minus 1.19 electronvolts. That’s stronger than iron alone on the same support, at around minus 0.89, and much stronger than copper alone, at minus 0.56. If you stick with homodimers, you go off balance—iron–iron binds too hard at roughly minus 1.89, copper–copper too weak at about minus 0.21. On the iron-copper site, the d orbitals of iron and copper line up with the oxygen p orbitals of nitrate in a way that stabilizes the initial adsorbate and, crucially, weakens the N–O bonds once nitrate is on the surface. A bond analysis that chemists call the integrated crystal orbital Hamilton population puts that N–O weakening in stark numbers: around minus 14.04 electronvolts for N O star on iron-copper H N G, which is a signature of a very activated N–O bond. What does that look like in the reaction sequence? Nitrate lands and takes its first electron-proton hit to become an adsorbed nitrate species, then nitrite, then nitric oxide, then progressively hydrogenated species like hydroxylamine on its way to ammonia. When the team watched products evolve during cycling with differential electrochemical mass spectrometry, they saw clear transient signals for nitric oxide and hydroxylamine, and a strong, sustained signal for ammonia. Compared with iron–iron or copper–copper pairs, the iron-copper sites shift the balance toward more complete deoxygenation—one measure is that the ratio of nitrite to nitric oxide signals is lower relative to those homodimers—lining up with the idea that the heterodimer gets past the sticky intermediates more cleanly. Another way to say it: the dual site choreographs two hard tasks. Iron, a bit more electron rich, can hook nitrate and start pulling those oxygen atoms off. Copper, a bit more electron poor after the iron-to-copper charge transfer, doesn’t overbind the later intermediates, so the hydrogenation steps flow. That division of labor matters in practice. It pushes the reaction away from the hydrogen evolution side path and away from routes that end in nitrogen gas, both of which siphon off current without providing you ammonia. Stability is where a lot of clever site designs fall apart. Here, the numbers are boring in the best possible way. Run the cell for twenty-four hours at minus 0.3 volts and the Faradaic efficiency hovers around ninety percent the whole time. The geometric production rate stays near 2400 micrograms per hour per square centimeter. When the team took the catalyst back under the microscope and back under the X-ray beam after the marathon, the diatomic iron-copper signatures were still there. The voltammograms before and after almost trace each other. Does this hold outside the comfort of a single, sweet-spot concentration? Across a range of nitrate levels, the peak efficiency to ammonia at minus 0.3 volts sits in a tight band—roughly 83 to 93 percent. In a fifty-milliliter batch with a typical polluted-water nitrate load of 200 milligrams per liter as nitrogen, the cell nearly cleared the nitrate in three hours and delivered about 189 milligrams per liter as ammonia-nitrogen. The accounting didn’t sum perfectly—the total measurable nitrate plus ammonia after the run was a bit lower than what you started with—so there are minor byproducts. But the dominant path went where they wanted it. Controls make or break claims of a “pairing effect,” and Zhang’s team ran the obvious ones. Iron single atoms on the same holey, nitrogen-doped graphene? Good, but not great. Copper single atoms? Worse on both counts—adsorption and selectivity. A simple physical mixture of iron and copper single-atom catalysts? Still no. The co-localized heterodimer is doing something you don’t get by proximity on the electrode; you need the two metals sharing that local nitrogen-rich environment and sharing charge. If you prefer the picture to the numbers, it’s this: the active site is a tiny, polarized see-saw. Iron donates a bit of electron density to copper. Nitrate feels that polarization and nestles in, making the first N–O bond easier to cleave. As the intermediate morphs into nitric oxide and then hydroxylamine, the site doesn’t clutch so hard that these species get stuck. At the operating potential where the device really sings—minus 0.3 volts—computations suggest each step rolls downhill in free energy. That’s why the current is high and the selectivity stays in the nineties. The path of least resistance is the ammonia path. There’s also a materials story hiding in plain sight. The holey graphene isn’t just a metal host; it’s an active part of the stabilization strategy. Edge nitrogens give you M N 2 sites that are strong enough to hold single atoms and even diatomic pairs through long electrolysis. The high surface area and mesopores feed reactants and clear products quickly, which, together with the alkaline electrolyte, keeps the competing hydrogen reaction at bay when nitrate is abundant. Stepping back, the study gives a crisp design rule you can carry forward. When a reaction demands both strong initial adsorption and smooth release of partially reduced intermediates, a hetero-atomic pair can give you both knobs on one site. Iron-copper H N G is a concrete instance: a iron-copper bond flanked by two nitrogens on a conductive carbon scaffold delivers 92.51 percent efficiency to ammonia at modest bias, high current density, and day-scale stability. The operando mass spectrometry and the density functional theory are not just box-checking; they show why the numbers look like that. Could this logic travel? Probably. Not every metal pair will hit the same sweet spot—iron–iron may overbind and stall, while copper–copper may let nitrate slip by—but the framework is general. Choose two metals whose d orbitals and electronegativities split the workload you care about, then seat them in a nitrogen-rich, conductive host that keeps them close without letting them sinter. Nitrate to ammonia is only one playground. The same balancing act might tune other multi-electron, multi-proton transformations where early activation and late-stage release both matter. But that’s for the next paper. For this one, the payoff is right here: a microscopic, two-atom partnership that turns a water contaminant into a commodity chemical with high selectivity, reasonable energy input, and a structure that holds up under load. It’s a small site with a big lesson—sometimes, chemistry’s best compromises come in pairs.