This is the peer reviewed version of the following ar3cle: Nature Energy 2019, 4, 519-525, which has been published in final form at hEps://www.nature.com/ar3cles/s41560-019-0404-4 This ar3cle may be used for non-commercial purposes in accordance with Macmillan Terms and Condi3ons for Self-Archiving. ARTICLES https://doi.org/10.1038/s41560-019-0404-4 Direct magnetic enhancement of electrocatalytic water oxidation in alkaline media Felipe A. Garcés-Pineda1, Marta Blasco-Ahicart1, David Nieto-Castro1, Núria López and José Ramón Galán-Mascarós 1,2* 1 text Industrially profitable water splitting is one of the great challenges in the development of a viable and sustainable hydrogen economy. Alkaline electrolysers using Earth-abundant catalysts remain the most economically viable route to electrolytic hydrogen, but improved efficiency is desirable. Recently, electron spin polarization was described as a potential way to improve water-splitting catalysis. Here, we report the significant enhancement of alkaline water electrolysis when a moderate magnetic field (≤450 mT) is applied to the anode. Current density increments above 100% (over 100 mA cm−2) were found for highly magnetic electrocatalysts, such as the mixed oxide NiZnFe4Ox. Magnetic enhancement works even for decorated Ni–foam electrodes with very high current densities, improving their intrinsic activity by about 40% to reach over 1 A cm−2 at low overpotentials. Thanks to its simplicity, our discovery opens opportunities for implementing magnetic enhancement in water splitting. W ater electrolysis is widely considered the most promising hydrogen source for the establishment of a clean and sustainable hydrogen economy powered by renewable energy sources1–3. In particular, the storage of photovoltaic energy as a fuel via water splitting could represent a dominant contribution to societal energy demands in the near future4, combining the excellent efficiency of photovoltaics with the easy storage and transport of fuels. Unfortunately, electrolyser technologies are still too expensive compared with the cheap hydrogen obtained from steam reforming5. Because of this, improving these technologies is still a major research challenge, and the subject of several public and private funding schemes. Nowadays, two electrolyser technologies are commercially available, either based on alkaline liquid electrolytes or polymer electrolyte membranes (PEMs). PEM electrolysers are more efficient and allow for higher production rates (current densities up to 2 A cm−2), but this comes with the unsolved issue of very expensive parts6. This includes the Nafion membrane itself, the titanium metal bipolar plates and the noble metal catalysts Pt/C and IrO27. These noble metals are expensive and extremely scarce, making them a limiting factor for the scale-up of these devices when targeted for mass production. The record for solar to hydrogen efficiency, at an impressive 30%8, was recently reported via the combination of a photovoltaic cell with two PEM electrolysers connected in series. The complete system is very efficient and achieved almost the maximum theoretical limit. However, its viability for large-scale application will require further study since it is based in noble metal catalysts. In contrast, alkaline liquid electrolysers are a very mature technology, and have been commercially available for over 50 years. In alkaline media, Earth-abundant electrocatalysts are stable enough to run both half-reactions. High-surface-area Ni-based electrodes and catalysts represent the state of the art, providing currents up to 0.5 A cm−2 at less than 70% efficiency9–11. Despite its moderate performance, this technology offers the most economic electrolytic hydrogen due to the overall lower cost of its components12. However, little room is apparently left for improvement. Benchmarking catalytic studies have demonstrated that many Earth-abundant metals are able to offer very good performance under alkaline conditions13. A very interesting opportunity to improve water-splitting kinetics comes from the oxygen evolution reaction (OER), which is typically considered the bottleneck for overall water splitting, as a slow and energy-demanding, four-electron process. The formation of the O–O bond, on breaking two water molecules must proceed via spin conservation to yield the paramagnetic triplet state of molecular oxygen. Thus, spin polarization of the active catalyst surface may favour parallel spin alignment of oxygen atoms during the reaction to improve the efficiency of the process, as suggested by theoretical studies14,15. The positive effect of spin polarization in OER was recently confirmed16,17. Experimental results with chiral catalysts suggested that spin polarization, as induced by the chiral structure of the active centre, was responsible for a superior electrocatalytic activity. In practical terms, the development of chiral OER electrocatalysts is certainly a plausible strategy, but it still represents a clear challenge in the field of affordable/scalable OER catalysts for wide and accessible implementation18. Theoretically, it was also proposed that magnetic electrodes may offer analogous positive effects19,20. Particularly, the gate effect of spin control for photosystem II21, and generally the effects towards oxygen electrochemistry decomposing the different energy and entropic terms have been suggested14. It is worth mentioning that the use of magnetic fields in water electrolysis has been briefly studied in the past, although from very different perspectives. On the one hand, a positive effect can be found associated with the influence of a magnetic field on the Lorentzian movement for diffusion of reagents and gas bubbles, therefore improving mass transport22–25. On the other hand, magnetic improvement has also been achieved by applying high-frequency alternating magnetic fields on a flow cell equipped with electrodes modified with magnetic nanoparticles, in a hyperthermia-like process26. In both cases, magnetic fields induce indirect effects on performance, either by improving gas/liquid diffusion or increasing local temperature, respectively. Additionally, both approaches need special electrolyser designs for their convenient exploitation. Institute of Chemical Research of Catalonia (ICIQ), The Barcelona Institute of Science and Technology (BIST), Tarragona, Spain. 2Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Spain. *e-mail: jrgalan@iciq.es 1 NATURE ENERGY | VOL 4 | JUNE 2019 | 519–525 | www.nature.com/natureenergy 519 ARTICLES NATURE ENERGY b a 30 Percentage of magnetocurrent Current density (mA cm–2) 70 60 50 40 H OFF H ON 30 20 10 pH 14 pH 13 pH 12 pH 11 20 10 0 0 1.52 1.56 1.60 1.64 1.68 Potential (V) versus RHE 1.55 1.60 1.65 1.70 Potential (V) versus RHE Fig. 1 | Magnetic effect on water electrolysis with a bare Ni–foam anode. a, Polarization curve (5 mV s−1) in 1 M KOH (pH 14) with (ON, open circles) and without (OFF, filled circles) an applied ≤450 mT magnetic field. b, Magnetocurrent (see equation (4) in Methods) as a function of pH. Here, we describe how an external magnetic field, easily applied by approaching a permanent magnet, enhances the electrocatalytic activity of magnetic anodes in alkaline conditions. Our experimental data indicate that the magnetic field is affecting the reaction pathway, resulting in a net positive effect, and not related to indirect effects such as those previously described. This magneto-enhancement appears to be proportional to the magnetic nature of the catalysts, and is particularly useful for the highly magnetic iron–nickel oxides, which are the preferred OER catalysts under these conditions. Magnetic field effect We turned our attention to alkaline water electrolysis to investigate the response of a bare Ni–foam anode in a liquid electrolyte (KOH; 1 M) cell equipped with a platinum mesh cathode and an Ag/AgCl 3.5 M KCl reference electrode under an applied magnetic field. This was easily implemented by approaching a commercial neodymium permanent magnet. Ni–foam is not an innocent support because a thin catalytic Fe-doped Ni oxide layer rapidly evolves on the surface under working conditions, taking up Fe from electrolyte impurities11. In our case, after Ni–foam electrode conditioning, the polarization curve (linear sweep voltammetry (LSV)) was measured with and without a magnetic field of ~450 mT (Supplementary Fig. 1 and Methods). The data showed a significant positive effect on spin polarization (Fig. 1a). The onset potential and precatalytic basal current were identical in both regimes. The appearance of a magnetocurrent component above the onset yielded a lower Tafel slope under the applied magnetic field, reaching higher currents at any given potential (Supplementary Fig. 2). The magnetic enhancement, normalized by the basal electrocatalytic performance (equation (4)), reached a maximum 25% at ~1.60 V versus the reversible hydrogen electrode (RHE; Fig. 1b). After this threshold, the relative magnetocurrent decreased, which we assigned to diffusion limitations provoked by very intense O2 gas bubbling. In other words, the efficiency of this alkaline electrolysis cell is boosted just by the implementation of a magnetic field (applied from a convenient distance), without any modification of the electrolyser architecture. The absence of any effect in the precatalytic current indicated that the magnetic field was not affecting the electrode capacitance (related to the total number of redox active sites) or bulk electron transport processes. The different Tafel slope also supported either a different rate-limiting step or an alternative mechanism, suggesting that the magnetic field was speeding up the electrocatalytic reaction kinetics. The genuine mechanistic effect of the spin polarization induced by the permanent magnet in our work was further supported by analogous experiments as a function of pH (Fig. 1b). The magnetocurrent effect was maximized at pH 14, decreasing as 520 the pH decreased and becoming negligible below pH 11. An indirect non-mechanistic enhancement, perhaps related to mass transport or local heat, cannot justify such a dramatic pH dependence. Additionally, this observation suggests that the dominant OER mechanistic pathway changed from alkaline media to neutral conditions27, since no magnetic effect is observed below pH 11. This also suggests that the rate-limiting step under neutral conditions was not spin restricted, in contrast with alkaline conditions. We studied a variety of OER catalysts under identical working conditions: deposited on two-dimensional Ni–foil anodes as Fumatech FAA-3 ionomer inks (see Methods for the detailed procedure). We selected some of the best state-of-the-art OER catalysts (Raney Ni, NiFe2Ox, FeNi4Ox and Ni2Cr2FeOx28–32) as well as some catalysts30,33–35 with very different magnetic features (the non-magnetic IrO2, the antiferromagnet NiO, the spinel ZnFe2Ox, and the highly magnetic ferrites NiZnFe4Ox and NiZnFeOx3,34,36–40). Polarization data (Fig. 2a–i) showed negligible magnetic enhancement for the only non-magnetic catalyst, IrO2, and the appearance of magnetocurrent in all other cases. However, the effect was very different in magnitude, depending on the catalyst. We measured the magnetization curves (Supplementary Fig. 3) for all of these catalysts, looking for a correlation between magnetocurrent and magnetization data. We plotted the magnetocurrent density as a function of bulk magnetization for each catalyst at 450 mT (Supplementary Fig. 4). Given the very different base currents, we did not find any significant trend with absolute values. However, when we analysed the relative magnetocurrent as the percentage of variation in current density (Fig. 3a and Supplementary Table 1), a clear relationship between both parameters was found (Fig. 3b) that appeared to be linear. There was one exception to the general trend: NiFe2Ox exhibited a lower magnetocurrent than expected from the magnetization data. This may have been due to surface states, since bulk magnetization does not necessarily need to be that of the surface/active sites for all materials. The maximum relative effect was observed for NiZnFe4Ox, where current doubled from 24 to 40 mA cm−2 at 1.65 V and higher. We modified the working conditions for this set-up to confirm the true magnetic enhancement observed. An identical and consistent magnetocurrent was found when the electrolysis was carried out under a turbulent regime forced by mechanical stirring (Supplementary Fig. 5). This confirmed that the observed magnetic field enhancement was not due to improved mass transport effects, as in multiple previous studies22–25. At the same time, the effect was very sensitive to the relative position of the permanent magnet with respect to the electrode, also supporting the directional effect on electrode surface magnetization (Supplementary Fig. 6). NATURE ENERGY | VOL 4 | JUNE 2019 | 519–525 | www.nature.com/natureenergy ARTICLES NATURE ENERGY 15 30 H OFF H ON 20 10 10 5 20 5 10 Potential (V) versus RHE Current density (mA cm–2) 30 H OFF H ON 20 10 10 5 1.56 1.60 1.64 20 20 10 0 1.50 h 10 1.55 1.60 Potential (V) versus RHE 0 1.65 Current density (mA cm–2) 30 H OFF H ON 40 30 H OFF H ON 10 20 5 10 1.55 1.60 25 40 30 30 H OFF H ON 20 10 10 1.55 1.60 0 1.65 50 (c) 40 15 30 H OFF H ON 10 20 5 10 1.50 1.55 1.60 0 1.65 Potential (V) versus RHE i 50 0 1.60 20 0 0 1.65 40 20 1.55 0 1.65 Potential (V) versus RHE 100 50 (h) 40 80 60 30 H OFF H ON 20 40 10 20 0 1.50 1.55 1.60 Magnetocurrent density (mA cm–2) 20 f 50 Magnetocurrent density (mA cm–2) 40 30 10 Potential (V) versus RHE 50 (g) 20 5 0 1.50 0 30 H OFF H ON Potential (V) versus RHE (b) 0 1.50 0 Magnetocurrent density (mA cm–2) Current density (mA cm–2) 40 1.70 15 Potential (V) versus RHE g 1.65 10 Magnetocurrent density (mA cm–2) 20 1.52 1.60 Magnetocurrent density (mA cm–2) 40 Magnetocurrent density (mA cm–2) 25 0 e 50 30 15 1.55 50 (a) 40 Potential (V) versus RHE Current density (mA cm–2) d 30 H OFF H ON 0 1.50 0 1.60 1.55 10 Current density (mA cm–2) 1.50 40 Current density (mA cm–2) 0 1.45 15 15 Magnetocurrent density (mA cm–2) 40 Magnetocurrent density (mA cm–2) 20 c 50 Current density (mA cm–2) b 50 (f) Current density (mA cm–2) 25 Magnetocurrent density (mA cm–2) Current density (mA cm–2) a 0 Potential (V) versus RHE Fig. 2 | Polarization data under an applied magnetic field for different OER catalysts. Polarization curves (5 mV s−1) in 1 M KOH electrolyte (pH 14) for Ni–foil electrodes decorated with OER catalysts (OFF, filled circles), and under an applied ≤450 mT magnetic field (ON, open circles). OER catalysts were drop-casted on Ni–foil as FAA-3 ionomer inks. a, IrO2. b, NiO. c, Raney Ni. d, Ni2Cr2FeOx. e, NiFe2Ox. f, FeNi4Ox. g, ZnFe2Ox. h, NiZnFe4Ox. i, NiZnFeOx. The magnetocurrent component (that is, the difference in current density with and without the magnetic field) is represented in red for each catalyst. NiZnFe4Ox has another advantage, since it can be magnetically attached to an Ni metal support due to its high magnetization. We decorated an Ni–foam substrate with NiZnFe4Ox via simple onestep sonication, optimizing the deposition time to 15 min according to the electrocatalytic performance (Supplementary Fig. 7). The electrode surface was analysed by powder X-ray diffraction, Raman spectroscopy and environmental scanning electron microscopy (Supplementary Figs. 8–10), confirming the presence of the magnetic phase attached to the Ni–foam surface. Excellent longterm stability over 24 h at 10 mA cm−2 was found for these electrodes, despite using exclusively magnetic binding (Supplementary Fig. 11). On application of the magnetic field, the current density NATURE ENERGY | VOL 4 | JUNE 2019 | 519–525 | www.nature.com/natureenergy roughly doubled (Fig. 3c), reaching over 150 mA cm−2 at 1.65 V versus RHE. We carried out a chronoamperometry experiment by successively moving the magnet next to the electrode (magnet ON) and then removing it (magnet OFF). As shown in Fig. 3d, the effect of the magnetic perturbation was immediate, as the current was enhanced consistently under the presence of the magnetic field. Supplementary Videos 1–3 illustrate this effect at basal, 10 mA cm−2 and 100 mA cm−2 current densities. Although glass corrosion may occur in alkaline media electrocatalysis41, we did not find any effect on reproducibility and stability during our experiments using glassware. The use of Pt mesh counter electrodes in alkaline conditions may also produce migration 521 ARTICLES NATURE ENERGY Potential (V) versus RHE a 1.30 1.35 1.40 1.45 1.50 1.55 1.60 1.65 b NiZnFe4Ox 120 NiZnFeOx 100 Magnetocurrent (%) ZnFe2Ox NiO Ni2Cr2FeOx FeNi4Ox Raney Ni NiFe2Ox FeNi4Ox IrO2 NiO Ni2Cr2FeOx ZnFe2Ox NiZnFe4Ox NiZnFeOx 80 60 40 20 NiFe2Ox 0 Raney Ni 0 20 40 60 80 100 120 0 20 c 60 80 d 200 600 400 H OFF 100 H ON 300 200 50 100 160 Current density (mA cm–2) 500 150 Magnetocurrent density (mA cm–2) Current density (mA cm–2) 40 Magnetization (A m–2 kg–1) Maximum magnetocurrent (%) H ON 140 120 100 80 H OFF 0 1.55 1.60 1.65 0 Potential (V) versus RHE 60 10 20 30 40 50 60 70 Time (min) Fig. 3 | Magnetic enhancement of water electrolysis under an applied magnetic field. a, Bar diagram with the maximum magnetocurrent observed for the various magnetic OER catalysts expressed as the relative percentage of the base current, and corresponding applied potential (blue dots). b, Correlation between the maximum relative magnetocurrent (at 1.67 V versus RHE) and bulk magnetization. The dashed line is a guide to illustrate the trend. c, Polarization data (5 mV s−1) for Ni–foam electrodes magnetically decorated (see Methods) with NiZnFe4Ox particles (OFF, filled circles), and under an applied magnetic field (ON, open circles). d, A pulsed magneto-chronoamperometry experiment was performed at a constant potential of ~1.67 V versus RHE for the Ni–foam electrodes magnetically decorated (see Methods) with NiZnFe4Ox particles. All data were collected in 1 M KOH electrolyte (pH 14). Error bars represent s.d. (n = 4). and deposition of Pt on other parts of the electrochemical set-up42. This is crucial in hydrogen evolution experiments, where Pt is a highly active catalyst, in contrast with OERs, where Pt is considered an inert metal43. We repeated the NiZnFe4Ox key experiments with a carbon counter electrode and a Hg/HgO reference electrode (Supplementary Fig. 12). The magnetocurrent results were consistent, independent of counter electrode or reference, and thus confirmed the negligible effect of the Pt mesh counter electrode. Additionally, we repeated the same experiments using a fluorinedoped tin oxide (FTO) glass anode support, to rule out the possible participation of the electrode support. We drop-casted the NiZnFe4Ox ink on this diamagnetic electrode and observed an even stronger magnetocurrent enhancement, reaching an improvement of 150% (Supplementary Fig. 13a). This confirmed that the magnetocurrent effect originated at the active sites of the catalysts and was not due to a support effect. Still, the absolute gain in magnetocurrent was significantly higher on the Ni supports (Supplementary Fig. 13b,c), thanks to their metallic character. The appearance of a magnetocurrent in water oxidation catalysis may be considered analogous to the effect reported for chiral systems. Chiral OER catalysts restrict hydrogen peroxide production17 (a significant side reaction at close-to-neutral pH), precluding anti522 parallel HO–OH formation. However, under alkaline conditions, we could not detect the presence of H2O2 during the OER above pH 11 (Supplementary Fig. 14). This was a major difference between both effects. Apparently, during alkaline water splitting, the H2O2 side reaction was already minimized, and the applied magnetic field sped up the dominant reaction pathway. Thus, we associate this effect with the spin alignment of oxygen atoms during the reaction, as favoured by the magnetic field (external and local)14. Indeed, computational studies suggest that this spin-aligned pathway is thermodynamically favoured in alkaline media (see Supplementary Discussion). Magnetic enhancement at high current densities For technological applications, high geometric current densities are needed. Among the OER catalysts we tested, NiZnFeOx delivered the highest absolute current densities on magneto-enhancement (Fig. 2i and Supplementary Fig. 4). To maximize geometric performance, we sprayed nanoparticles of this spinel (Supplementary Fig. 15) onto Ni–foam electrodes as FAA-3 inks. For these electrodes, the magnetic enhancement in the catalytic activity became particularly significant above 100 mA cm−2, allowing 20–30 mV savings for a given current, reaching 300 mA cm−2 at an overpotential of NATURE ENERGY | VOL 4 | JUNE 2019 | 519–525 | www.nature.com/natureenergy ARTICLES NATURE ENERGY b 500 600 450 Current density (mA cm–2) 300 500 H OFF 400 H ON 250 300 200 200 150 100 100 500 1,000 H OFF 800 400 H ON 600 300 400 200 200 Magnetocurrent density (mA cm–2) 350 Magnetocurrent density (mA cm–2) 400 600 1,200 Current denstity (mA cm–2) a 100 50 0 1.45 1.50 1.55 0 1.60 Potential (V) versus RHE 0 1.40 1.45 1.50 1.55 1.60 0 Potential (V) versus RHE Fig. 4 | Polarization data for surface-modified Ni–foam anodes. a,b, Polarization curves (5 mV s−1) for Ni–foam electrodes decorated with FeOx (a; ref. 44) and FeOx/Ni2P (b; ref. 46). Data were collected with (open circles) or without (filled circles) an applied ≤450 mT magnetic field. All data were collected in 1 M KOH electrolyte (pH 14). just 341 mV (Supplementary Fig. 16a). These electrodes were robust over 24 h, with and without magnetic enhancement (Supplementary Fig. 16b). Online O2 detection also confirmed quantitative OER Faradaic efficiency with and without the presence of a magnetic field (Supplementary Fig. 17). Surface-decorated Ni–foam electrodes have been reported with stable and lower overpotentials at very high current densities: those decorated with FeOx44 or FeOx/Ni2P45, as grown in  situ. We prepared these two electrodes following the published procedures. The electrocatalytic activity was consistent with previous reports (Fig. 4 and Supplementary Table 2). Both electrodes significantly improved their performance under a magnetic field, reducing by ~15 mV the overpotential (η) required to reach current densities above 100 mA cm−2. Although these potential reductions may seem small, given the very low Tafel slope for these catalysts, the result is an extraordinary current boost, increasing several hundreds of mA cm−2 just on spin polarization (Supplementary Table 2). For instance, at 1.64 V versus RHE (η = 415 mV), the current density promoted by an FeOx/Ni2P-decorated Ni–foam anode increases from 920 to 1,300 mA cm−2 under a magnetic field (Fig. 4). Conclusion In summary, we have demonstrated the positive effect of external magnetic fields to speed up electrochemical water oxidation—a key step towards the production of electrolytic hydrogen or other solar fuels. Being a spin-restricted reaction, the magnetic field favours the parallel alignment of oxygen radicals during the formation of the O–O bond—the dominant mechanistic pathway in alkaline conditions. We found a trend in the magnetic nature of the catalysts, with a negligible effect for non-magnetic catalysts but maximum enhancement for highly magnetic ones, including the Ni–Fe oxide series. For some of these catalysts, current densities doubled just by moving a permanent magnet next to the anodic compartment of the glass cell. This magnetic enhancement also occurred at very high currents, opening interesting possibilities for its implementation in alkaline electrolyser technologies. The low fields that promote magneto-enhancement (<0.4 Tesla) can be achieved with low-cost ceramic magnets. Methods All chemicals were commercially available (Sigma–Aldrich) and were used without further purification. The Ni–foil substrate (NI000550; Goodfellow) was of 2.0 mm NATURE ENERGY | VOL 4 | JUNE 2019 | 519–525 | www.nature.com/natureenergy thickness and 99.0% purity. The Ni–foam substrate (NI003852; Goodfellow) was of 1.6 mm thickness, 95% porosity and 99.5% purity. The FTO-coated glass slides were 12–14 Ω per square surface resistivity (NSG TEC; Pilkington; 15 A; 2.2 mm). The Nd–magnet ring (IMA) had the following dimensions: diameter, A = 76 mm; internal diameter, B = 35 mm; thickness, C = 6 mm. Oxide synthesis. The metal oxides NiO, NiFe2Ox, FeNi4Ox and Ni2Cr2FeOx were obtained using a combustion method46. A starting aqueous solution (50 ml) of the metal nitrate salts in the desired ratio was prepared, with the iron concentration fixed to 50 mM. Glycine was added to the aqueous solution in a glycine-to-metal molar ratio of 1.20 and stirred until total dissolution. Afterwards, the solution was heated to 200 °C until the solvent was totally evaporated and the glycine had combusted. This flamy combustion process was accompanied by the vigorous emission of gasses (CO2, N2 and water vapour). The resulting porous dark solids were recovered and sintered at 1,100 °C in a tubular oven for 1 h in air. Finally, the sintered materials were mechanically milled in an agate ball mill for 15 min at 25 Hz for a final particle size of ~150–250 nm. The metal oxide NiZnFe4Ox was obtained via a coprecipitation method. The desired amounts of NiCl2⋅6H2O (5 mmol), ZnCl2 (5 mmol) and FeCl3⋅6H2O (20 mmol) were dissolved in 100 ml of distilled water. Then, precipitation was induced with 15 mmol of NaOH. The reactive solution was kept at 85 °C for 1 h. The precipitate was thoroughly washed with distilled water several times to remove sodium and chlorine excess from the solid. Then, the wet solid was dried at 110 °C overnight, followed by thermal ramp annealing at 550 °C for 2 h and 1,000 °C for 1 h in a tubular oven. The final product was milled in an agate ball mill for 15 min at 25 Hz for a final particle size of ~100–200 nm. The metal oxide NiZnFeOx was obtained via a hydrothermal method. Equimolecular amounts of the metal nitrates were dissolved in water (metal concentration: 50 mM) and the solution was hydrolysed with diluted aqueous ammonia until pH 8.5 was achieved. The solution was introduced to a Teflon cup and mounted in an autoclave at 140 °C for 2 h. After the hydrothermal treatment, the pressure vessel was cooled in air, the product was washed with H2O and CH3CH2OH, and the nanoparticles were collected by centrifugation (final particle size: ~8 nm). Electrode preparation. Ni–foil and FTO electrodes. Catalyst inks were prepared following previously reported procedure47. The catalyst (10 mg) was mixed with a 10 wt% of ionomer (FAA-3 ionomer; FUMATEC), and the ink was completed with a liquid CH3CH2OH:H2O mixture (3:1 in volume) to a final volume of 1 ml. The electrode substrates were cleaned in acetone and Milli-Q water before the ink deposition. Then, 84 μl of ink was drop-casted on 1 cm2 of clean Ni–foil or FTO surface. The electrodes were dried at 60 °C to obtain a total loading of 0.84 mg catalyst cm−2. Blank electrodes were prepared following the same procedure without any catalyst content in the ink formulation. Ni–foam. NiZnFe4Ox was deposited on Ni–foam by direct magnetic interaction between the materials. First, Ni–foam was treated with thermal annealing at 450 °C for 30 min to generate a passivated layer on the surface. Then, 100 mg of the ferrite was sonicated for 20 min in 20 ml of Milli-Q water until a homogeneous suspension was obtained. Then, the previously treated Ni–foam was dipped in the suspension 523 ARTICLES NATURE ENERGY and sonicated in an ultrasonic bath for different depositions times of 15, 30 or 60 min. The final coated Ni–foam electrode was air dried. NiZnFeOx nanoparticles were deposited on Ni–foam electrodes using the spray technique with analogous FAA-3 ink, as described above. The electrode support was kept at 60 °C during the spraying process to ensure homogeneous deposition of the catalyst ink. Electrochemistry. All electrochemical experiments were performed in borosilicate glassware with a Bio-Logic SP-150 potentiostat, Ag/AgCl (3.5 M KCl) reference electrode (ALS) and a Pt mesh counter electrode. Unless otherwise stated, the solution electrolyte used for all electrochemical tests was prepared with 90% KOH and Milli-Q water. The desired pH was measured and adjusted with an 877 Titrino Plus pH probe (Metrohm). All potentials reported in this manuscript were converted to the RHE reference scale. The exact pH value was measured for each experiment and used to convert the measured potential to the RHE electrode using the equation: ERHE = EAg∕AgCl + 0.059 × pH + 0.205 (1) 0 considering E NHE = EAg∕AgCl + 0.205 V for our Ag/AgCl (3.5 M KCl) reference electrode. The water oxidation overpotential (η) was calculated by subtracting the 0 thermodynamic water oxidation potential, E O ∕H O = 1.229(V) versus RHE 2 2 (pH 14), from the experimental potential (ERHE) at pH 14: η = ERHE − 1.229 (2) Linear sweep voltammetry experiments were performed at a 5 mV s−1 scan rate in 1 M KOH. iR-compensation was applied to all polarization curves based on resistance data. The magnetic field was applied by moving an Nd–magnet as close as possible to the electrochemical cell (~1 mm glass thickness) (see Supplementary Video 2). The magnetocurrent density was calculated by substracting the LSV values obtained with the applied magnetic field (magnet ON) from the LSV data without the magnetic field (magnet OFF): Magnetocurrent density = J (HON) − J (HOFF) (3) The magnetocurrent was calculated with the following equation: Magnetocurrent(%) = J (HON) − J (HOFF) × 100 J (HOFF) (4) Oxygen evolution was detected with an Ocean Optics NeoFox oxygen-sensing system equipped with a FOXY probe. The FOXY probe was calibrated with two-point calibration, fixing 0% O2 under N2 flow and 20.9% O2 in air. The threeelectrode configuration experiment was performed in a 50 ml two-neck roundbottom flask filled with 1 M KOH solution (Vgas space = ~18 ml) with the FOXY probe placed in the gas space. The system was purged with an N2 flow for 30 min before starting the electrochemical experiment. After removing the N2 flow, a baseline of 6 min was recorded before starting the chronoamperometry measurement. The moles of O2 generated during the electrochemical experiment were calculated with the following equation, considering ideal gas behaviour: n O2 = %O2 × Ptotal × Vgas space R×T (5) where %O2 is given by the FOXY probe, Ptotal is 1 atm, Vgas space (l) is measured for each experiment, R is 0.082 L atm K−1 mol−1 and T is 298 K. The Faradaic oxygen production curve was calculated taking into account the charge passed through the system during the chronoamperometry experiment, as described in equation (6): n O2 = Q ne × F (6) where Q (C) is the charge passed through the system, ne = 4 is the number of electrons needed to generate one molecule of O2, and F is the Faraday constant (96,485 C mol−1). Hydrogen peroxide was detected by colorimetric titration, with o-tolidine as the redox indicator, following an Elmms–Hauser procedure48. An indicator solution was added to 4 ml of the electrolyte solution obtained from the electrochemical cell after 1 h of chronoamperometry at different potentials, and left to react for 30 min. In the presence of H2O2, a yellow colour appeared with an ultraviolet–visible absorption peak at around 436 nm. To quantify the H2O2 concentration, a calibration curve was determined using 30% w/w commercial H2O2 (Supplementary Fig. 14). No H2O2 was detected in any experiment carried out above pH 11. Physical methods. Powder X-ray diffraction was obtained with a Bruker AXS D8-Discover diffractometer (40 kV and 40 mA). Environmental scanning electron microscopy data were obtained using Quanta 600 equipment from FEI under high-vacuum conditions with a Large-Field Detector at 20 kV. Transmission 524 electron microscopy was performed with a JEOL JEM1011 microscope operating at 80 or 100 kV, equipped with a high-contrast 2k × 2k AMT mid-mount digital camera. Fourier transform infrared spectroscopy–Raman spectra were collected with a Renishaw inVia Reflex Raman confocal microscope, equipped with a diode laser emitting at 785 nm at a nominal power of 300 mW, and a Peltier-cooled charge-coupled device detector (−70 °C) coupled to a Leica DM2500 microscope. Calibration was carried out daily by recording the Raman spectrum of an internal Si standard. Rayleigh scattered light was appropriately rejected using edge-type filters. Laser power was used at nominal 1% to avoid sample damage. Spectra were recorded with the accumulation of at least three scans of 30 s each. Thermal treatment was carried out using a tubular Nabertherm P330 furnace. The magnetic field calibration of the Nd–magnet was performed using an analogue Hall sensor (HE144; Asensor Technology AB) (Supplementary Fig. 1). The 450 mT quoted in the article corresponds to the maximum magnetic field measured when the Hall sensor was in direct contact with the magnet. During experiments, the magnet was between 1 and 2 mm from the magnet. Therefore, the magnetic field applied during the experiments was always lower than the maximum limiting value of 450 mT. We quote this maximum limit due to the difficulty in estimating the average magnetic field applied to the electrode during operation. The magnetic properties of the bulk materials were determined using an MPMS-XL SQUID magnetometer (Quantum Design) at 300 K. Data availability All experimental data generated or analysed during this study are included in this published article and its Supplementary Information files. Received: 6 October 2018; Accepted: 30 April 2019; Published online: 10 June 2019 References 1. Zhang, B. et al. Homogeneously dispersed multimetal oxygen-evolving catalysts. Science 352, 333–337 (2016). 2. Xu, Y.-F. et al. In situ formation of zinc ferrite modified Al-doped ZnO nanowire arrays for solar water splitting. J. Mater. Chem. A 4, 5124–5129 (2016). 3. Cady, C. W. et al. Tuning the electrocatalytic water oxidation properties of AB2O4 spinel nanocrystals: A (Li, Mg, Zn) and B (Mn, Co) site variants of LiMn2O4. ACS Catal. 5, 3403–3410 (2015). 4. McKone, J. R., Lewis, N. S. & Gray, H. B. Will solar-driven water-splitting devices see the light of day? Chem. Mater. 26, 407–414 (2014). 5. Schmidt, O., Gambhir, A., Staffell, I., Nelson, J. & Few, S. Future cost and performance of water electrolysis: an expert elicitation study. Int. J. Hydrogen Energy 42, 30470–30492 (2017). 6. Carmo, M., Fritz, D. L., Mergel, J. & Stolten, D. A comprehensive review on PEM water electrolysis. Int. J. Hydrogen Energy 38, 4901–4934 (2013). 7. Yuan, W., Zhao, M., Yuan, J. & Li, C. M. Ni foam supported threedimensional vertically aligned and networked layered CoO nanosheet/ graphene hybrid array as a high-performance oxygen evolution electrode. J. Power Sources 319, 159–167 (2016). 8. Jia, J. et al. Solar water splitting by photovoltaic-electrolysis with a solar-tohydrogen efficiency over 30%. Nat. Commun. 7, 13237 (2016). 9. Song, F. et al. Transition metal oxides as electrocatalysts for the oxygen evolution reaction in alkaline solutions: an application-inspired renaissance. J. Am. Chem. Soc. 140, 7748–7759 (2018). 10. Guo, C. X. & Li, C. M. Room temperature-formed iron-doped nickel hydroxide on nickel foam as a 3D electrode for low polarized and highcurrent-density oxygen evolution. Chem. Commun. 54, 3262–3265 (2018). 11. Trotochaud, L., Young, S. L., Ranney, J. K. & Boettcher, S. W. Nickel–iron oxyhydroxide oxygen-evolution electrocatalysts: the role of intentional and incidental iron incorporation. J. Am. Chem. Soc. 136, 6744–6753 (2014). 12. Zeng, K. & Zhang, D. Recent progress in alkaline water electrolysis for hydrogen production and applications. Prog. Energy Combust. Sci. 36, 307–326 (2010). 13. McCrory, C. C. L., Jung, S., Peters, J. C. & Jaramillo, T. F. Benchmarking heterogeneous electrocatalysts for the oxygen evolution reaction. J. Am. Chem. Soc. 4, 16977–16987 (2013). 14. Gracia, J. Spin dependent interactions catalyse the oxygen electrochemistry. Phys. Chem. Chem. Phys. 19, 20451–20456 (2017). 15. Gracia, J., Sharpe, R. & Munarriz, J. Principles determining the activity of magnetic oxides for electron transfer reactions. J. Catal. 361, 331–338 (2018). 16. Mtangi, W., Kiran, V., Fontanesi, C. & Naaman, R. Role of the electron spin polarization in water splitting. J. Phys. Chem. Lett. 6, 4916–4922 (2015). 17. Mtangi, W. et al. Control of electrons’ spin eliminates hydrogen peroxide formation during water splitting. J. Am. Chem. Soc. 139, 2794–2798 (2017). 18. Zhang, W., Banerjee-Ghosh, K., Tassanari, F. & Naaman, R. Enhanced electrochemical water splitting with chiral molecule-coated Fe3O4 nanoparticles. ACS Energy Lett. 3, 2308–2313 (2018). NATURE ENERGY | VOL 4 | JUNE 2019 | 519–525 | www.nature.com/natureenergy ARTICLES NATURE ENERGY 19. Chretien, S. & Metiu, H. O2 evolution on a clean partially reduced rutile TiO2 (110) surface and on the same surface precovered with Au1 and Au2: the importance of spin conservation. J. Chem. Phys. 129, 074705 (2008). 20. Torum, E., Fang, C. M., de Wijs, G. A. & de Groot, R. A. Role of magnetism in catalysis: RuO2 (110) surface. J. Phys. Chem. C 117, 6353–6357 (2013). 21. Jiao, J., Sharpe, R., Lim, T., Niemantsverdriett, J. W. H. & Gracia, J. Photosystem II acts as a spin-controlled electron gate during oxygen formation and evolution. J. Am. Chem. Soc. 139, 16604–16608 (2017). 22. Elias, L. & Hegde, C. Effect of magnetic field on HER of water electrolysis on NiÐW alloy. Electrocatalysis 8, 375–382 (2017). 23. Zheng, Z. et al. Magnetic field-enhanced 4-electron pathway for well-aligned Co3O4/electrospun carbon nanofibers in the oxygen reduction reaction. Chem. Sus. Chem. 11, 580–588 (2018). 24. Katz, E., Lioubashevski, O. & Willner, I. Magnetic field effects on bioelectrocatalytic reactions of surface-confined enzyme systems: enhanced performance of biofuel cells. J. Am. Chem. Soc. 127, 3979–3988 (2005). 25. Monzon, L. M. A., Rode, K., Venkatesan, M. & Coey, J. Electrosynthesis of iron, cobalt, and zinc microcrystals and magnetic enhancement of the oxygen reduction reaction. Chem. Mater. 24, 3878–3885 (2012). 26. Niether, C. et al. Improved water electrolysis using magnetic heating of FeC–Ni core–shell nanoparticles. Nat. Energy 3, 476–483 (2018). 27. Giordano, L. et al. pH dependence of OER activity of oxides: current and future perspectives. Catal. Today 262, 2–10 (2016). 28. McCrory, C. C. L. et al. Benchmarking hydrogen evolving reaction and oxygen evolving reaction electrocatalysts for solar water splitting devices. J. Am. Chem. Soc. 137, 4347–4357 (2015). 29. Trotochaud, L., Ranney, J. K., Williams, K. N. & Boettcher, S. W. Solution-cast metal oxide thin film electrocatalysts for oxygen evolution. J. Am. Chem. Soc. 134, 17253–17261 (2012). 30. Gerken, J. B., Shaner, S. E., Massé, R. C., Porubsky, N. J. & Stahl, S. S. A survey of diverse Earth abundant oxygen evolution electrocatalysts showing enhanced activity from Ni–Fe oxides containing a third metal. Energy Environ. Sci. 7, 2376–2382 (2014). 31. Smith, R. D. L. et al. Photochemical route for accessing amorphous metal oxide materials for water oxidation catalysis. Science 340, 60–63 (2013). 32. Seitz, L. C. et al. A highly active and stable IrOx/SrIrO3 catalyst for the oxygen evolution reaction. Science 353, 1011–1014 (2016). 33. Liu, G., Wang, K., Gao, X., He, D. & Li, J. Fabrication of mesoporous NiFe2O4 nanorods as efficient oxygen evolution catalyst for water splitting. Electrochim. Acta 211, 871–878 (2016). 34. Liu, G., Gao, X., Wang, K., He, D. & Li, J. Uniformly mesoporous NiO/ NiFe2O4 biphasic nanorods as efficient oxygen evolving catalyst for water splitting. Int. J. Hydrogen Energy 41, 17976–17986 (2016). 35. Dez-Garca, M. I., Lana-Villarreal, T. & Gómez, R. Study of copper ferrite as a novel photocathode for water reduction: improving its photoactivity by electrochemical pretreatment. ChemSusChem 9, 1504–1512 (2016). 36. Ng, J. W. D. et al. Gold-supported cerium-doped NiOx catalysts for water oxidation. Nat. Energy 1, 16053 (2016). 37. Islam, M. U. et al. Electrical behaviour of fine particle, co-precipitation prepared Ni–Zn ferrites. Solid State Commun. 130, 353–356 (2004). 38. Wang, X. et al. A novel NiZn ferrite integrated magnetic solenoid inductor with a high quality factor at 0.7–6 GHz. AIP Adv. 7, 056606 (2017). 39. He, Q. et al. Electrically controllable spontaneous magnetism in nanoscale mixed phase multiferroics. Nat. Commun. 2, 225 (2011). NATURE ENERGY | VOL 4 | JUNE 2019 | 519–525 | www.nature.com/natureenergy 40. Su, Y.-Z. et al. One dimensionally spinel NiCo2O4 nanowire arrays: facile synthesis, water oxidation, and magnetic properties. Electrochim. Acta 174, 1216–1224 (2015). 41. Mayrhofer, K. J. J., Wiberg, G. K. H. & Arenz, M. Impact of glass corrosion on the electrocatalysis on Pt electrodes in alkaline electrolyte. J. Electrochem. Soc. 155, P1–P5 (2008). 42. Rodriguez, P., Tichelaar, F. D., Koper, M. T. M. & Yanson, A. I. Cathodic corrosion as a facile and effective method to prepare clean metal alloy nanoparticles. J. Am. Chem. Soc. 133, 17626–17629 (2011). 43. Chen, R. et al. Use of platinum as the counter electrode to study the activity of nonprecious metal catalysts for the hydrogen evolution reaction. ACS Energy Lett. 2, 1070–1075 (2017). 44. Dong, G. et al. In situ formation of highly active NiFe based oxygen-evolving electrocatalysts via simple reactive dip-coating. J. Mater. Chem. A 5, 11009–11015 (2017). 45. Zhou, H. et al. Highly active catalyst derived from a 3D foam of Fe(PO3)2/ Ni2P for extremely efficient water oxidation. Proc. Natl Acad. Sci. USA 114, 5607–5611 (2017). 46. Chen, J. Y. C., Miller, J. T., Gerken, J. B. & Stahl, S. S. Inverse spinel NiFeAlO4 as a highly active oxygen evolution electrocatalyst: promotion of activity by a redox-inert metal ion. Energy Environ. Sci. 7, 1382–1386 (2014). 47. Carmo, M., Doubek, G., Sekol, R. C., Linardi, M. & Taylor, A. D. Development and electrochemical studies of membrane electrode assemblies for polymer electrolyte alkaline fuel cells using FAA membrane and ionomer. J. Power Sources 230, 169–175 (2013). 48. Ellms, J. W. & Hauser, S. J. Ortho-tolidine as a reagent for the colorimetric estimation of small quantities of free chlorine. J. Ind. Eng. Chem. 5, 915–917 (1913). Acknowledgements This work was funded by: the European Union’s Horizon 2020 research and innovation programme under grant agreement CREATE number 721065; FEDER/Ministerio de Ciencia, Innovación y Universidades – Agencia Estatal de Investigación/RTI2018095618-B-I0; and the Generalitat de Catalunya (2017-SGR-1406 and the CERCA Programme). The authors also acknowledge BSC-RES for computational resources. Author contributions J.R.G.-M. and N.L. proposed the concept. F.A.G.-P. and J.R.G.-M. designed the experiments. M.B.-A. and F.A.G.-P. performed the synthesis, processing and electrochemical experiments. D.N.-C. performed the magnetic measurements and analyses. N.L. performed the computational studies. All authors wrote the manuscript. Competing interests The authors declare no competing interests. Additional information Supplementary information is available for this paper at https://doi.org/10.1038/ s41560-019-0404-4. Reprints and permissions information is available at www.nature.com/reprints. Correspondence and requests for materials should be addressed to J.R.G. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. © The Author(s), under exclusive licence to Springer Nature Limited 2019 525