Heat-induced magnetic transition accelerates redox couple mediated electrocatalytic water oxidation in alkaline media


 The redox couple oxidation as an initial step for oxygen evolution reaction (OER) may be key for high electricity consumption in electrochemical redox couple mediated water splitting. Here, we report a heat-induced magnetic transition strategy to speed up the oxidation kinetics of redox couples. The activation energy of Ni2+/Ni3+ redox couple oxidation was significantly decreased by heating the Ni0.5Fe0.5OxHy OER catalyst above Néel temperature (TN) of 70oC. In such a strategy, the heat in stead of electricity to overcome the spin flipping of Ni2+/Ni3+ oxidation through the heat-sensitive ferrimagnetic-to-paramagnetic spin state changes. The efficient heat-electricity coupling enables Ni0.5Fe0.5OxHy to produce the lowest OER overpotential of 170 mV at 100 mA cm-2 at 90 oC in alkaline electrolyte, outperforming the benchmark IrO2 catalyst. Our findings demonstrate the application potential of heat-sensitive magnetic materials in the field of electrocatalysis, which may inspire insights into designing of multi-energy complementary OER devices.

Ni 2+ /Ni 3+ oxidation (Fig. 1c). And the K OER is restricted by PCSRET during electron injection from OER intermediates ( * OH, * O, * OOH) to the active catalytic species (Fig. 1d). A few theoretical results have discussed the spin-related K OER on magnetic catalysts 13,[15][16][17] . Although the redox couples used in water splitting, such as Ni, Fe, or Co ions-containing compounds 18 , are magnetic, to date no attention has been paid to accelerate K M by altering spin states. In particular, the spin-related K M also provided a chance to construct multi-energy coupling system if coupled non-electric energy to the electron spin changes of redox couple oxidation.
Here, we conceptually show a heat-induced magnetic transition strategy to accelerate the redox couple oxidation. We found that the activation energy of redox couple oxidation sharply decreased by contribution of heat in stead of electricity to the spin ipping. Our results show that the materials with magnetic transition sensitive to heat can promote the spin-related water splitting and the heat coupling to spin transition provides a solid physical basis to designing of multi-energy complementary OER devices.

OOH redox couple
We select the magnetic Ni x Fe 1−x (OH) 2 with inherently high K OER and frustrated K M as model material to show the magnetic transition awakened heat-electricity coupling strategy. We rst optimized the Fe content relative to Ni ( Supplementary Fig. 1). A highest OER current density was obtained in α-Ni x Fe 1−x (OH) 2 with Ni:Fe = 1:1 [denoted as Ni 0.5 Fe 0.5 (OH) 2 ], in good agreement with the previous report 19 .
The electrochemically deposited low-crystallinity α-Ni 0.5 Fe 0.5 (OH) 2 onto Ni foam (see Methods for the detailed procedure) was con rmed by XRD and high-resolution TEM and exhibited highest OER activity ( Supplementary Fig. 2). The α-Ni 0.5 Fe 0.5 (OH) 2 /γ-Ni 0.5 Fe 0.5 OOH dual-phase mixture (Ni 0.5 Fe 0.5 O x H y ) was created at 1.41 V at 30 o C to describe the Ni 2+ /Ni 3+ redox couple. We rst performed the spin-polarized density functional theory (DFT) calculations to disclose the magnetic ground states of α-Ni 0.5 Fe 0.5 (OH) 2 and γ-Ni 0.5 Fe 0.5 OOH. The calculated total density of states (DOS) suggested that Ni 0.5 Fe 0.5 (OH) 2 is semiconductor-like, which hosts a valence band near the Fermi level dominated by the spin-down Fe 3d orbits and a conductive band originating from asymmetric spin-up Ni 3d and spin-down Fe 3d states as seen from the projected DOS (Fig. 2a). In contrast, Ni 0.5 Fe 0.5 OOH is metal-like with asymmetric spin-down Ni 3d and spin-up Fe 3d orbits around the Fermi level, implying an easier electron transfer in Ni 0.5 Fe 0.5 OOH. Integration of the spin projected DOS of Fe and Ni indicated that the two materials are ferrimagnetic at 0 K with a magnetic moment of 2 µB per formula unit for Ni 0.5 Fe 0.5 (OH) 2 and 4.62 µB per formula unit for Ni 0.5 Fe 0.5 OOH. The ferrimagnetic states are originated from double exchange interaction of Fe-O-Ni con gurations.
The magnetic properties were further analyzed by using a superconducting quantum interface device (SQUID) magnetometer in their thermally stable region of 200-500K ( Supplementary Fig. 3). The zero-eld cooled (ZFC) and eld cooled (FC) scanning under an applied magnetic eld H = 500 Oe showed a moment as a function of temperature and a splitting between the ZFC and FC traces at low temperature for both Ni 0.5 Fe 0.5 (OH) 2 and Ni 0.5 Fe 0.5 O x H y , identifying paramagnetic-to-ferrimagnetic transition with a T N of 104 o C for Ni 0.5 Fe 0.5 (OH) 2 and 70 o C for Ni 0.5 Fe 0.5 O x H y (Fig. 2b), in agreement with the predictions of DFT calculations. The T N (53 o C) of pure Ni(OH) 2 is obviously lower than that of Ni 0.5 Fe 0.5 (OH) 2 ( Supplementary Fig. 4), revealing that Fe doping enhances the spin antiparallel interaction (ferrimagnetism) and so makes the spin state change di cult with heating. This veri ed that the Fe doping stabilized the high-spin Ni 2+ with of electronic con guration t 2g 6 e g 2 in Ni 0.5 Fe 0.5 (OH) 2 Supplementary Fig. 6b). The nearly constant slope of current-voltage curves ( Fig.3a and Supplementary   Fig. 7a) suggested that the Fe doping signi cantly promoted the OER kinetics at such a rate that is completely diffusion-limited, revealing that the low-grade heat does not impose on K OER but on the K M . A noticeable phenomenon is that for Ni 0.5 Fe 0.5 (OH) 2 the △V was gradually minimized with increasing temperature, suggesting that K M and K OER gradually tend to be equal. The V 100 was parallelly moved from 1.46 V at 30 o C to 1.40 V at 90 o C, corresponding to a low η 100 of 170 mV. The 60 mV decrease in OER overpotential would be a result of accelerated Ni 2+ /Ni 3+ oxidation kinetics, K M whose activation energy was successfully awakened by heat eld. In contrast, the Tafel slope of pure Ni(OH) 2 is temperature-dependent, varying from 105 mV dec -1 at 30 o C to 74 mV dec -1 at 90 o C ( Supplementary Fig. 7b), con rming that the water oxidation on Ni(OH) 2 without the help of Fe is mainly limited by its sluggish OER kinetics.  Table 2 and Supplementary  Fig. 8). At 1.45V ( Supplementary Fig. 9), the OER region, R ct,OER and C OER are nearly constant independently of temperature change, verifying the high OER activity of Ni 0.5 Fe 0.5 OOH, as well con rmed by the constant electrochemical avtive surface area ( Supplementary Fig. 10). However, at T T N , the closely equal R ct,M and R ct,OER con rmed that the K M and K OER keep nearly same rate. The slightly high R ct,M would be the origination of the small residual △V due to the semiconducting Ni 0.5 Fe 0.5 (OH) 2 with large resistance. Indeed, owing to the limited diffusion depth of OER intermediates in catalyst particles, the core parts of Ni 0.5 Fe 0.5 (OH) 2 particles are not able to participate OER process, thus contributing the partial potential drop.
Owing to the structural distortion and low structural symmetry, the total energy difference between ferrimagnetic and paramagnetic Ni 0.5 Fe 0.5 OOH is small (about 30 meV by DFT calculations). And the free energy change for oxidation of ferrimagnetic Ni 0.5 Fe 0.5 (OH) 2 into ferrimagnetic or paramagnetic Ni 0.5 Fe 0.5 OOH was similar ( Supplementary Fig. 11), implying that the paramagnetic Ni 0.5 Fe 0.5 OOH could be expected during OER, although exhibiting ferrimagnetic ground state at 0 K. Indeed, the free energy pro les demonstrated that the overpotential (η) of rate-determining step to form * O via * OH is signi cantly lower for paramagnetic Ni 0.5 Fe 0.5 OOH (0.56 eV) than the ferrimagnetic Ni 0.5 Fe 0.5 OOH (1.64 eV, Fig. 3d).
This result would further suggest the oxidation of ferrimagnetic Ni 0.5 Fe 0.5 (OH) 2 to produce the paramagnetic Ni 0.5 Fe 0.5 OOH, a magnetic state affording high enough K OER independently of temperature change. That is, the frustrated K M is a result of spin ipping when T < T N . Indeed, the DFT calculations indicated that the phase transition from paramagnetic Ni 0.5 Fe 0.5 (OH) 2 to paramagnetic Ni 0.5 Fe 0.5 OOH is thermodynamically favourite ( Supplementary Fig. 11), clearly concluding that the spin ipping is high electricity consumption process. In fact, at T T N (Fig. 3a) Fig. 12). As shown in Fig. 4a and b, Ni 2p core-level XPS analysis on raw Ni 0.5 Fe 0.5 (OH) 2 indicated the binding energy at 855.5 eV with a satellite peak at 861.3 eV for Ni 2+ 2p 3/2 in α-Ni(OH) 2 20 .
Raman bands (Fig. 4c)  observed at 473 cm -1 (depolarized E g mode, bending) and 557 cm -(polarized A1 g mode, stretching) 24 , and the 455 cmfor Ni 2+ -OH species (A 1g ) was also visible due to the Ni 0.5 Fe 0.5 (OH) 2 /Ni 0.5 Fe 0.5 OOH coexistence, suggesting the sluggish ferrimagnetic Ni 2+ / paramagnetic Ni 3+ oxidation kinetics at low temperatures. Evidently, after introducing Fe, the Ni 2+ /Ni 3+ redox peak shifted by ∼60 mV from +1.34 V to +1.40 V ( Supplementary Fig. 6) at 30 o C, in good agreement with the previous reports 19 . A projected density of states (pDOS) analysis ( Fig. 2 and Supplementary Fig. 13 paramagnetic Ni 2+ /Ni 3+ redox cycling without spin ipping. In this situation, the energy requirement in spin ipping was completely provided by heat, thus largely accelerating the K M to achieve the nearly same rate with K OER due to the effective heat-electricity coupling. We also detected the Ni 2+ species at the OER potential region at T < T N ( Supplementary Fig. 14), further demonstrating that accelerating K M is a result of overcoming the high Ni 2+ /Ni 3+ oxidation barrier by heat-induced spin ipping or increasing applied potential.

Nature of high OER activity and physical basis of heat-electricity coupling
In particular, at 1.45 V and 90 o C, the OER process occurs and about 1.0 eV increase in binding energy of Fe 2p showed that Fe sites may be the OER active sites. Many efforts have been paid on distinguishing whether nickel 25,26 , iron [27][28][29] , or the synergy of nickel and iron sites 30 are the active centers, mainly focusing on detection of high-valence Fe or Ni species without attention to quantum spin(-orbital) exchange interactions (QSEI). Spin-polarized DFT results exhibited that all the OER intermediates ( * OH, * O, * OOH) tend to adsorb onto Fe sites (Fig.3d). The OER activity difference between paramagnetic and ferrimagnetic Ni 0.5 Fe 0.5 OOH originated from the downshift of Fermi level of paramagnetic Ni 0.5 Fe 0.5 OOH ( Fig. 5a), hence reducing the electron transfer barrier due to narrowing of the energy difference between Fermi level of catalyst and the bonding orbital level composing of d-p orbital hybrid of Fe 3d active species and O2p OER intermediates. Charge population analysis indicated that the Fe sites in both ferrimagnetic and paramagnetic Ni 0.5 Fe 0.5 OOH are more effective electron transfer pathway (Fig. 5b).
The underperformance in electron transfer for adsorption of * OH implied that the OER rate-determining step to form * O on Ni 0.5 Fe 0.5 OOH resulted from the weak interactions between Fe sites and * OH species.
For adsorption of * O, the effective electron transfer from * O to metal sites may suggest a fact that the electron from OER intermediates rst injected to Fe 3d via the Fe 3d-O2p bonding orbitals and then extracted to external circuit via Fe-O-Ni double exchange interactions.
The EPR spectra were recorded to examine the spin states of Fe 3+ ions (Fig. 5c). The total spectrum was simulated by using a standard line-shape model with anisotropic g-factor (WINEPR). The simulated results indicated that the paramagnetic EPR signal originates from the low-spin Fe 3+ (S = 1/2) with g 1 = 2.9790, g 2 = 2.0052, and g 3 = 1.9312 for Ni 0.5 Fe 0.5 (OH) 2 and g 1 = 2.9720, g 2 = 2.0051, and g 3 = 1.8510 for Ni 0.5 Fe 0.5 O x H y 31 . This evidence implied that at least partial Fe 3+ sites in Fe-O-Ni con gurations are lowspin paramagnetic at room temperature (ions in ferrimagnetic state only contribute to widening of the EPR line width due to the strong spin-spin interaction). The low spin state of Fe 3+ exhibited the spin distribution composing of the unoccupied e g orbitals and the incompletely occupied t 2g with one unpaired single-spin electron (Fig. 5d). In contrast, the low spin con guration of Ni 3+ only has incompletely occupied e g orbitals. The more electron transfer channels enable the Fe 3+ sites to be a strong electron acceptor, thus exhibiting a strong interactions with the OER intermediates.
Here, as shown in Fig. 5e, we can conclude that the Ni 0.5 Fe 0.5 O x H y is ferrimagnetic-to-paramagnetic transition sensitive to low-grade heat, thus achieving a low electron transfer barrier for Ni 2+ /Ni 3+ oxidation by paramagnetic Ni 0.5 Fe 0.5 (OH) 2 /Ni 0.5 Fe 0.5 OOH phase transition at above T N . The heat is consequently coupled to be responsible for spin ipping, effectively replacing electricity consumation. In addition, in the paramagnetic situation without spin ipping, the low-spin Fe 3+ with more electron transfer channels are more bene cial to the double exchange interaction of Fe-O-Ni con gurations in Ni 2+ /Ni 3+ oxidation and electron transfer in OER process, thus signi cantly accelerating the redox couple mediated water splitting with nearly same K M and K OER .
In summary, the heat-induced magnetic transition was conceptually veri ed to be effective to construct a heat-electricity synergistic water splitting system. As an example, the Ni 2+ /Ni 3+ cycling kinetics in the Nibased redox couple mediated water splitting, an initial OER bottleneck step, was signi cantly accelerated by coupling heat to the electricity-driven Ni 2+ /Ni 3+ redox cycling. Based on the spectroscopic and magnetic tests, we showed here a clear physical mechanism of heat-electricity coupling water splitting with the assistance of magnetic phase transition. The low-grade thermal eld (< 100 o C) instead of electricity was attributed to the thermally sensitive ferrimagnetic-to-paramagnetic spin state change of Ni 2+ /Ni 3+ redox cycling. Our results provide a new possibility to utilize various heat sources, such as that produced via industrial, solar-thermal or geothermal processes, by introducing magnetic state modulation, to design multi-energy complementary OER devices.

Methods
Theoretical calculations. The structural model of α-Ni(OH) 2 was derived from Ref. 32 32 which belongs to a space group of P-3m1. The structural model of γ-NiOOH was constructed by deprotonating β-NiOOH to produce Ni oxidation state of +3 (taken from Ref. 33 33 ). And Fe doping concentration of 50% was set for Fe 0.5 Ni 0.5 (OH) 2 and Fe 0.5 Ni 0.5 OOH phases based on the experimental results. As for the OER calculations, the edge models were constructed and the exposed oxygen atoms at surface were passivated by hydrogen atoms All calculations were performed with Vienna Ab initio Simulation Package (VASP) based on the densityfunctional theory (DFT) 34 . The generalized gradient approximation (GGA) with PBE functional was used for the exchange-correlation energy, and DFT-D2 method proposed by Grimme was adopted for van der Waals interactions 35 . A plane-wave expansion for the basis set with a cutoff energy of 400 eV was employed here. The gamma centered 1×1×1 k-point meshes was used for the Brillouin-zone integration of supercell models. All atoms are fully relaxed until the energy convergence and residual force was less than 10 -5 eV and 0.02 eV/Å, respectively. To eliminate interactions between the neighboring cells, vacuum regions of 16 Å were used for the OER surface model. Moreover, considering the strong correlation effect of d orbital in transition metals, the U value of 5.0 eV and 6.6 eV were chosen for d electrons of Ni and Fe atoms.
To determine the free energy of intermediates at the surface of catalyst (G * ) during the OER process, equations below were employed here: where E, ZPE, and S are the total energy, zero-point energy, and entropy of intermediates, respectively.
Electrodepositing of α-Ni 0.5 Fe 0.5 (OH) 2 on foam nickel (NF). The NF (1 cm x 2 cm) was ultrasonically cleaned with 1 M HCl solution for 30 min to remove the oxide layer on the surface and then rinsed with deionized water and absolute ethanol for several times, followed by dring at 60 o C for later use. The α-Ni 0.5 Fe 0.5 (OH) 2 on NF electrode was prepared by one-step electrochemical deposition (Fig. 1a). The threeelectrode system was employed, the working electrode was the NF electrode to be deposited, the counter electrode was platinum electrode, and the reference electrode was Ag/AgCl electrode. The microstructure and high-resolution crystal image were characterized by Tecnai G2 F20 transmission electron microscope (TEM). Before the test, the sample is ultrasonically dispersed in ethanol, and then dropped on the special copper mesh. After being purged and dried by air, it can be tested under vacuum with the acceleration voltage at 15 kV. The phase composition was characterized by X-ray diffraction (XRD, Rigaku Ultima III diffractometer, Cu Kα radiation). A Thermo ESCALAB 250 X-ray photoelectron spectrometer (XPS) was employed to analyze the composition and element valence of the samples. The binding energy was calibrated using C 1s with a binding energy of 284.6 eV as an internal standard. Raman spectra were examined by Horiba T64000@514 nm with an argon ion as laser light source. The electron paramagnetic resonance (EPR) spectra were obtained using a Bruker (Model EMX-10/12 X-band, Bruker, Germany) electron paramagnetic resonance spectrometer at room temperature (25 o C). The settings used were a center eld of 2670.0 G for Ni 0.5 Fe 0.5 (OH) 2 and 2900 G for Ni 0.5 Fe 0.5 O x H y , a microwave frequency of 9.84 GHz, and a power of 2 mW. The magnetic measurements were carried out on a superconducting quantum interference device (MPMS SQUID XL 7, Quantum Design, USA) with a moment detection limitation of 5x10 -8 emu, a magnetic eld strength of±7T, and the temperature range of 2-1000 K.
Electrochemical measurements. All electrochemical experiments were performed on Electrochemical Analyzer (CHI 760E) using a standard three-electrode con guration. The working electrode was Ni foam (thickness, 1.8 mm) or carbon paper (TGP-H-060, TORAY). The reference electrode was saturated Ag/AgCl, and the counter electrode was Pt foil. The electrocatalytic performances were tested in 1 M N 2 -saterated KOH solution as ambient conditions. All measured potentials in this work were converted to reversible hydrogen potential (RHE) according to Nernst equation of E RHE =E Ag/AgCl +0.059 pH+0.197 V. In the heating experiment, the effect of temperature on the reference electrode potential has been deducted, and the correction of temperature coe cient is as shown in Supplementary Fig. 15. Linear sweep voltammetry (LSV) was obtained with the scan rate of 10 mV s -1 with or without iR compensation within the cell. The electrochemical double layer(C dl ) were calculated by the non-Faradaic cycle voltammetry at different scan rates. The electrochemical impedance spectra (EIS) were recorded with a frequency range from 100 kHz to 0.01 Hz. The quantitative analysis of O 2 and H 2 are obtained by using gas chromatograph (GC8890, Agilent Corp., 5A zeolite column and Ar as carrier) to determine the faradaic e ciency. The test was performed in a gas-sealed electrolytic cell using owing nitrogen to prevent the accumulation of high temperature water vapor. The apparent electrochemical activation energy (Q M ) for Ni 2+ /Ni 3+ oxidation can be determined using the Arrhenius relationship: where J p is the kinetic peak current for Ni 2+ /Ni 3+ oxidation, T is the absolute temperature in Kelvin , and R is the universal gas constant. From the slope of the Arrhenius plot, the Q M can be extracted.
Declarations Figure 1 The main kinetic obstacles in redox couple mediated water splitting. a,Typical current-voltage curve of OER on redox-mediated catalysts. KM: oxidation rate constant of redox couple; KOER: OER rate constant.
b, An ideal situation is that the onset of OER started at the oxidation potential of redox couple, resulting from minimizing the ∆V by accelerating the KM and KOER to nearly same rate. c, The limit of KM originates from the proton-coupled spin-related electron transfer (PCSRET) during the oxidation of hydroxides/oxyhydroxides redox couple. d, The limit of KOER results from the PCSRET during evolution of OER intermediates (*OH, *O, *OOH).

Figure 2
The magnetic properties of Ni0.5Fe0.5(OH)2 and Ni0.5Fe0.5OxHy. a,Total DOS and projected DOS. b, ZFC and FC magnetizations as a function of temperature with applied magnetic eld H = 500 Oe. c, M-H loops.

Figure 3
The heat-electricity coupling electrochemical kinetics for Ni0.5Fe0.5(OH)2 at 30-90 oC. a, The temperature-dependent OER polarization curves at a scan rate of 10 mV s-1. Inset shows the temperaturedependent ∆V with a sharp decrease at T > TN. b, Arrhenius plot of inverse temperature versus log of Ni2+/Ni3+ oxidation peak current. c, The temperature-dependent EIS spectra at 1.4V. d, Free energy diagram for ferrimagnetic and paramagnetic Ni0.5Fe0.5OOH models, with calculated structures and ratedetermining steps.

Figure 5
Nature of high OER activity and physical basis of heat-electricity coupling. a Fermi level EF, referring to the vacuum level, for ferrimagnetic and paramagnetic Ni0.5Fe0.5OOH. b, Charge density difference for *OH and *O species adsorbed on the ferrimagnetic and paramagnetic Ni0.5Fe0.5OOH. The yellow and green charge densities correspond to charge accumulation and depletion, respectively. The electron transfer number (Δρ) was obtained by Bader charge analysis. The + and -indicate gaining electrons and losing electrons, respectively. c, Experimental and simulated EPR spectra for Ni0.5Fe0.5(OH)2 and Ni0.5Fe0.5OxHy. d, Low-spin Fe3+ signi cantly enlarging the PCSRET of OER by increasing the electron transfer pathway. e, Heat coupling to promote PCSRET of Ni2+/Ni3+ oxidation via heat-driven magnetic transition.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.