Ultrafast Preparation of Nonequilibrium FeNi Spinels by Magnetic Induction Heating for Unprecedented Oxygen Evolution Electrocatalysis

Carbon-supported nanocomposites are attracting particular attention as high-performance, low-cost electrocatalysts for electrochemical water splitting. These are mostly prepared by pyrolysis and hydrothermal procedures that are time-consuming (from hours to days) and typically difficult to produce a nonequilibrium phase. Herein, for the first time ever, we exploit magnetic induction heating-quenching for ultrafast production of carbon-FeNi spinel oxide nanocomposites (within seconds), which exhibit an unprecedentedly high performance towards oxygen evolution reaction (OER), with an ultralow overpotential of only +260 mV to reach the high current density of 100 mA cm−2. Experimental and theoretical studies show that the rapid heating and quenching process (ca. 103 K s−1) impedes the Ni and Fe phase segregation and produces a Cl-rich surface, both contributing to the remarkable catalytic activity. Results from this study highlight the unique advantage of ultrafast heating/quenching in the structural engineering of functional nanocomposites to achieve high electrocatalytic performance towards important electrochemical reactions.


Introduction
Design and engineering of low-cost, high-performance catalysts play an important role in the development and advancement of electrochemical energy technologies, such as fuel cells and water electrolyzers [1]. Up to now, a range of functional nanocomposites have been hailed as viable alternatives to the conventional, noble metal-based catalysts, such as doped carbon, and carbon-supported nanoparticles of metals, metal oxides, sulfides, phosphides, and selenides [2][3][4]. These materials are typically prepared via the "traditional" thermal methods based on pyrolysis and hydrothermal procedures [5][6][7][8][9][10][11][12]. While these methods are rather facile and effective in sample synthesis, they are energy and time-consuming [13,14], and the slow heating ramp makes it difficult to produce a nonequilibrium phase within the samples, which might be critical in regulating the electronic structure and hence the electrocatalytic activity [15][16][17][18]. In a prior study [19], Holewinski et al. demonstrated that whereas Ag and Co metals were immiscible at equilibrium, AgCo alloy nanoparticles could be obtained by removing the sample from the tube furnace while hot and letting it cool down in ambient instead of following a traditional programmed cooling process and exhibited a remarkable electrocatalytic activity towards oxygen reduction reaction (ORR) unseen for either Ag or Co metals alone. Li et al. [20,21] showed that the laser ablation method could be exploited for the synthesis of Ag and Ru nanoparticles for high-efficiency water splitting, due to the formation of plenty of stacking faults or grain boundaries that were difficult to produce via conventional methods. Yao et al. [22] developed carbothermal shock synthesis to prepare highentropy alloy nanoparticles within seconds. Chen et al. [23] adopted flash joule heating to synthesize metastable 1 T phase of MoS 2 with S vacancies and observed a high efficiency towards hydrogen evolution reaction (HER). In another study [24], Zhang et al. developed a laser scribing method where metal oxide/graphene composites were prepared by simply lasing metal precursors loaded onto a graphene scaffold and exhibited a high performance towards oxygen evolution reaction (OER) likely due to the formation of surface defects. Li et al. [20,25,26] successfully used the laser ablation method to prepare a series of water splitting electrocatalysts with oxygen vacancies, faults, and unique morphologies.
Despite the progress, the toolbox for such sample synthesis has been limited, and the range of materials that can be produced and the extent of structural engineering remain narrow. Thus, further development of effective protocols for the synthesis of materials with unprecedented structures and properties is of both fundamental and technological significance [27,28]. In electrochemical water splitting, OER has been recognized as a major bottleneck that limits the overall performance because of complex reaction pathways and sluggish electron-transfer kinetics [17], and FeNi (oxy) hydroxides and spinel oxides have been extensively studied as viable alternatives to the traditional, noble metal-based commercial catalysts [29][30][31][32][33][34], where manipulation of the occupation of the e g orbitals of the octahedral metals and/ or metal-oxygen covalency represents the leading strategies for further enhancement of the OER activity [30]. This is generally achieved by engineering the spinel components, heterometal doping, and introduction of oxygen vacancies [25,31,35,36]. Phase segregation of Fe and Ni in the spinels has been believed to be the leading cause of the apparent loss of the electrocatalytic activity [37]. Yet, such segregation is inevitable for samples prepared via a "tedious" conventional thermal procedure as it is energetically favorable. In addition, residual heteroanions (e.g., Cl) adsorbed on or doped into the surface of Fe, Co, and Ni (hydro/oxyhydro) oxides may play a significant role in OER electrocatalysis [38][39][40]. Yet the impacts of such anion impurities have remained largely ignored, although most pyrolytically prepared spinel oxides are derived from iron and nickel chlorides. These issues can be addressed with the development of appropriate synthetic methods where nonequilibrium structures can be produced with reduced phase segregation and a remarkable concentration of anion impurities. This is the primary motivation of the present study.
Herein, we report an ultrafast heating/quenching method based on magnetic induction heating/rapid quenching (MIHRQ) [41] to prepare carbon-supported FeNi spinel composites (within seconds), which exhibit a clear mixing of the Ni and Fe phases and a Cl-rich surface, in contrast to the control samples that are prepared by prolonged heating and/or natural cooling to the ambient. In electrochemical measurements, the former displays an outstanding electrocatalytic performance towards OER, with an ultralow overpotential of only +260 mV to reach the high current density of 100 mA cm -2 , due to the formation of a nonequi-librium structure that is optimal for the adsorption of key reaction intermediates and eventual production of oxygen. The enhanced performance of the catalyst is also confirmed by results of first principle calculations. These results highlight the unique potential of MIHRQ in the deliberate production of nonequilibrium features in composite electrocatalysts. Notably, such an unprecedented tool can be readily extended to the preparation of a wide range of functional nanocomposites for diverse applications [42][43][44][45][46][47].

Results and Discussion
2.1. Magnetic Induction Heating-Rapid Quenching. The homemade MIHRQ apparatus is shown in Figure 1(a). A four-turn induction solenoid was twisted at a diameter of 5 cm, under which was placed a beaker containing ethanol and dry ice (-78°C) as the quenching agent. Experimentally, a calculated amount of the metal precursors (i.e., FeCl 3 and NiCl 2 ) was dropcast onto a piece of carbon paper (1:5 cm × 0:5 cm), which was then sandwiched between two rectangular iron sheets (2:5 cm × 2:5 cm × 0:01 cm). An iron nail was inserted into the center of the iron sheets and clamped to hold the assembly, which was placed in the center of the induction solenoid (Figure 1(b)). When a high frequency (30 kHz) current was passed to the solenoid, a strong magnetic field was produced, which instantly generated a strong Eddy current in the iron sheets and thus heated the sample rapidly to a high temperature. The induction current and time can be varied to control the heating temperature ( Figure 1(c)). For example, a solenoid current of 200 A for a heating time of 4 s would generate a temperature of 200-300°C, which barely changed the color of the iron sheets (Movie S1). Yet, when the solenoid current was increased to 400 and 600 A, the temperature could reach ca. 600 and 1000°C, respectively, inducing a glowing color of the iron sheets from faint red to white (Figure 1(b), Movie S2). Such ultrafast heating (instant heating rate up to 10 3 K s -1 ) can not only drastically enhance the time efficiency of sample preparation but more importantly also be exploited for the kinetic control of the materials structures, in contrast to typical pyrolysis or hydrothermal synthesis, which is usually completed within the time frame of hours to even days. After a select period of heating (of the order of seconds), the sample would be dropped into the quenching solution below (Movie S1 and S2) or removed from the solenoid and cooled down naturally in the ambient. This offers an additional control of the materials structures, especially for the production of nonequilibrium features.
A series of samples were prepared with the MIHRQ setup at a controlled induction current (X) for a select period of time (Y) and denoted as FeNiO-X-Y (Figure 1(d)). Control samples were prepared in the same manner except for cooling in ambient, and referred to as FeNiO NC -X-Y. Notably, MIHRQ can be extended to the preparation of a wide range of electrocatalysts beyond the FeNi spinel oxides (vide infra).  Figure S1). The interatomic distance was estimated to be 0.816-0.824 nm (inset to Figure 2(b)), close to that of FeNi spinel oxide (0.835 nm) [48]. Additional STEM images acquired along the <112>, <100>, and<103> zone axes also confirmed the spinel lattice structure ( Figure S2-S4). Furthermore, energy-dispersive X-ray spectroscopy-(EDS-) based elemental mapping studies clearly show an even distribution of Fe and Ni within the lattice (Figure 2(c)), suggesting atomic mixing of the Fe and Ni elements and no phase segregation. Notably, spindle-like FeNi oxide nanocrystals can also be found around these particles (Figures 2(d)-2(f) and Figure S5-S7), which feature a chlorine-rich surface (Figure 2(g)), with a Cl concentration of 12% in comparison to under 2% within the particles (Table S1). In fact, the atomic ratio of Fe : Ni : O : Cl in the nanospindles is estimated to be 1.9 : 1 : 4.9 : 1.1, while the overall ratio is close to 4.1 : 1 : 5.8 : 0.22, indicating that the spinel particles were Fe-rich oxide, while the nanospindles likely represent an intermediate phase between the precursors (metal chlorides) and the final spinel crystal (Table S1). Furthermore, it is noticeable that the Fe : Ni ratio is higher than the feeding ratio. This is likely because part of Ni was not fully converted into Ni oxide and washed away during the rapid quenching process. For FeNiO-250-16 that was prepared via a longer heating time, the Fe : Ni : O : Cl ratio was estimated to be 6.6 : 1 : 8.6 : 0.028 (Table S1), indicative of the formation of a Fe-rich structure that was almost free of Cl. In fact, the nanospindle features, with the unique chlorine rich surface, can only be produced with a short heating time and rapid quenching process. Variations of temperature versus time of magnetic induction heating, traditional hydrothermal heating, and pyrolysis. The dash lines indicate the cases with the quenching process. 'IH,' 'TF,' and 'HT' are short for induction heating, tube furnace, and hydrothermal, respectively. (d) SEM images of the NiFeO-250-4 sample at different preparation stages. The insert images are the corresponding photographs of the electrodes. The carbon paper can be seen to become darkened after the deposition of the metal salt precursors, and subsequent induction heating and rapid quenching leads to a pitch-black appearance of the carbon paper, with particulates formed onto the carbon fibers.
The control sample, FeNiO NC -250-4 that was produced by similar heating but natural cooling in the ambient, exhibited an obviously different morphology (Figure 2(h)), consisting of aggregates of nanoparticles into large chunks. In addition, significant phase segregation occurred within the sample (Figure 2(i)), where the elements of Ni and O appeared to be evenly distributed across the sample, whereas Fe was mostly confined within a small region, suggesting the growth of FeNiO spinel nanocrystals on a nickel oxide scaffold. This is consistent with the sample atomic ratio of Fe : Ni : O : Cl = 0:15 : 1 : 2:3 : 0:036. The fact that the sample was markedly richer in nickel was likely due to higher thermal volatility of the iron compounds [49], where the enhanced loss of Fe was facilitated by the relatively slow cooling (about 10 K s -1 ). Similar phase segregation was also observed with a prolonged heating time (e.g., FeNiO-250-16), where the temperature could reach ca. 1000°C (Figures 2(j) and 2(k) and Figure S8). Note that such   [50], which can be rationally avoided by rapid heating and quenching as manifested with FeNiO-250-4. The material structures were further characterized by Xray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS) measurements. Figure 3(a) depicts the high-resolution XPS scan of the Ni 2p electrons of FeNiO-250-4, where two peaks can be resolved at 855.9 and 856.9 eV for the Ni(II) 3p 3/2 electrons, suggesting the formation of five and six oxygen-coordinated Ni atoms on the surface (i.e., Ni(OH) 2 ), respectively [51], since no NiO species (binding energy around 854.7 eV) was detected [52]. A single peak was resolved at 711.2 eV in the Fe 2p scan (Figure 3(b)), due to the Fe(III) 3p 3/2 electrons [53], whereas three peaks were deconvoluted in the O 1 s spectrum ( Figure 3(d)) at 531.5 eV for hydroxide, 529.9 eV for metal-O, and 533.0 eV for C-O [54]. These observations are consistent with the formation of FeNiO spinel lattices (vide infra). Deconvolution of the Cl 2p spectrum ( Figure S9) yields two peaks at 198.3 and 199.9 eV, due to the 2p 3/2 and 2p 1/2 electrons of metal-Cl, respectively [55]. Taken together, these results suggest that the FeNiO-250-4 sample surface was mostly terminated with OH and Cl groups, in good agreement with results from the TEM and EDS measurements.
The XPS spectra of the FeNiO NC -250-4 and FeNiO-250-16 samples are shown in Figure S10-S12. The Fe 2p spectra showed only a rather insignificant difference among the sample series ( Figure S11 and FeNiO-250-16, likely due to reduced carbothermal effects by the rapid heating and quenching process [22]. Further insights into the bonding configurations of the metal centers were obtained from the extended X-ray absorption fine structure (EXAFS) results. Fitting of the FT-EXAFS data (Figure 3(f), Figure S13, and Table S2-S4) show that FeNiO-250-4 actually possessed Ni-O bonds with a bond length of 2.03 Å, somewhat smaller than those of FeNiO NC -250-4 (2.05 Å) and rock salt NiO (2.09 Å) [56]. This is consistent with the phase segregation in and FeNiO NC -250-4 (3.04 Å), suggesting a possible transition from spinel structure to metallic Fe [57].
In summary, results from these characterization measurements show that prolonged heating and slow cooling facilitated the O and Cl loss for the spinel samples. Prolonged heating also promoted phase segregation of Ni into rock salt NiO or metallic form. With a deliberate control of the heating time and cooling rate, two key nonequilibrium features of the FeNiO spinel nanoparticles can be achieved, minimal FeNi phase segregation and formation of a Cl-rich surface, both critical in OER electrocatalysis (details below).  (Figure 4(b)). In addition, at 100% iR compensation, FeNiO-250-4 can even produce an exceedingly high current density of 1 A cm -2 at only +1.64 V (Figure S14), which represents an unprecedentedly high activity among the leading FeNi oxide-based OER electrocatalysts reported in recent studies (Table S5 and Movie S3).
To quantify the electrochemical surface area (ECSA), voltametric measurements were also carried out in the nonfaradaic region of +0.9 to +1.1 V (Figure S15), and FeNiO-250-4 can be seen to exhibit the highest ECSA of 23.06 cm 2 , as compared to 15.75 cm 2 for FeNiO NC -250-4 and 17.06 cm 2 for FeNiO-250-16 (Table S6). When the OER currents were normalized to the respective ECSA, FeNiO-250-4 remained the best among the series (Figure S16a), reaching a current density of 10 mA cm -2 at +1.64 V. The corresponding turnover frequency (TOF) is depicted in Figure S16b at up to 0.21 s -1 for FeNiO-250-4, which was at least one order of magnitude higher than those of FeNiO NC -250-4 and FeNiO-250- 16. In fact, the FeNiO-250-4 sample represents the optimal condition ( Figure S17). It also shows excellent stability. At the applied potential of +1.53 V, over 80% of the initial current was retained even after 10 h's continuous operation   , which is one of the best performances as compared to the state of the art [37]. The corresponding OER polarization curve showed an anodic shift of only 10 mV. When the electrode was subject to additional 1,000 cyclic voltammetric (CV) scans between +1.20 and +1.65 V, the subsequent polarization curve exhibited a further anodic shift of only 10 mV (Figure 4(c)).
Notably, XPS measurements of the electrocatalyst after the stability test ( Figure S18) showed that the Fe : Ni ratio decreased to 1 : 1.8, as compared to that of the as-prepared sample (4.1 : 1, Table S1). This indicates that part of Fe was leached away during the prolonged OER process, which is commonly observed with relevant catalysts [37]. However, Cl remained detectable, and the relative ratio to Fe and Ni (1 : 3) is consistent with that of the pristine sample. Furthermore, in EDS-based elemental mapping analysis ( Figure S19), a homogeneous distribution of Fe and Ni, along with Cl, can be readily observed after the stability tests; and the atomic ratio of Fe to Ni was estimated to be 1 : 2, consistent with that from XPS measurements. Taken together, these results suggest that the activity decay is largely due to the Fe loss instead of structural change.

Theoretical Study.
To unravel the mechanistic origin of the remarkable OER activity observed above with FeNiO-250-4, slab models were built by chlorine substitution of the surface oxygen atom originally located between Fe oct (octahedral site) and Ni td (tetrahedral site) in NiFe 2 O 4 (100) (Figure 5(a)), based on the structural features identified in the above experimental characterization. Free energy calculations indicate that adsorption of OH favors the Ni td sites over the Fe oct sites on the surface, consistent with the XPS results ( Figure 3). Consequently, the NiFe 2 O 4 (100) surface with Cl substituting O and Ni td binding an OH (Ni(OH) Fe 2 O 4 (Cl) was used as the model catalyst ( Figure 5(a)). Other structures, Ni(OH)Fe 2 O 4 ( Figure 5 A two-site ( * -#) model was adopted to study the OER mechanism (details in the Supporting Information). From the free energy diagram in Figure 5 (Figure 6(a)), there is no observable charge density redistribution at the Fe 1 site (adjacent to Cl) but a significant electron depletion is observed on Fe 2 (away from Cl). With the introduction of Cl (Figure 6

Conclusions
In summary, the MIHRQ method was successfully developed and exploited for the ultrafast fabrication of metal oxide spinel nanostructures. Using NiCl 2 and FeCl 3 as the precursors, FeNi oxide spinels were obtained by heating at controlled currents within seconds and exhibited an even mixing of the Ni and Fe elements and a Cl-rich surface, in sharp contrast to samples prepared at prolonged heating and/or natural cooling in the ambient. The best sample FeNiO-250-4 needed an overpotential of only +260 mV to reach the high current density of 100 mA cm -2 and exhibited significant stability in alkaline media. Such a remarkable activity was attributed to the unique metastable structure that facilitated the adsorption of key reaction intermediates and O-O coupling, a major limiting step in OER. Results from this study highlight the unique advantages of MIHRQ in the rapid production of unprecedented material structures that are unattainable in conventional thermal processes for enhanced electrocatalytic performance and potential applications in the structural engineering of a diverse range of materials ( Figure S21).

Sample Preparation. Carbon paper (TGP-H-060 from
Toray Industries, Inc.) was cut into 1 cm × 1:5 cm pieces. A solution was prepared by dissolving 40 mg of NiCl 2 •6H 2 O, 10 mg of FeCl 3 , and 0.8 g of urea into 10 mL of water (supplied with a Barnstead Nanopure Water System, 18.2 MΩ cm). 100 mL of the solution was dropcast onto the carbon paper, which was then dried at ambient temperature and sandwiched between two iron sheets of 2:5 cm × 2:5 cm × 0:01 cm. The assembly was then placed in the center of a four-turn induction coil with a diameter of 5 cm, and magnetic induction heating was carried out at a controlled current (X = 100 -600 A) for a select heating time (Y = 2 -16 s), when the sample was dropped into an ethanol-dry ice solution (-78°C) placed underneath the induction coil for rapid quenching (caution: the ethanol must be fully cooled down by dry ice, or it will catch fire). The temperature was measured with an industrial infrared laser thermometer (BTMETER, BT-1500). The corresponding sample was denoted as FeNiO-X-Y.
Control samples were also prepared by removing the sample assembly from the heater and being cooled down in ambient to room temperature. These samples were referred to as FeNiO NC -X-Y.  9 Research samples were first sonicated, dispersed in ethanol, and then deposited onto copper grids for TEM characterization. Scanning electron microscopy (SEM) studies were carried out on FEI Quanta 3D FEG dual beam instrument. X-ray photoelectron spectroscopy (XPS) measurements were performed with a Phi 5400/XPS instrument equipped with an Al K α source operated at 350 W and 10 -9 Torr. X-ray diffraction (XRD) patterns were acquired with a Bruker D8 Advance diffractometer with Cu K α radiation (λ = 0:15418 nm). Self-consistent multiplescattering calculations were performed using the FEFF6 program to obtain the scattering amplitudes and phase-shift functions used to fit various scattering paths with the Artemis program. In the fitting of each sample, the E 0 values were correlated together to minimize the number of independent values, allowing reliable fitting results to be obtained. The σ 2 values were also correlated for some samples.

Electrochemistry.
Electrochemical measurements were carried out with a CHI 700e electrochemical workstation in a three-electrode configuration. The prepared carbon paper was fixed onto a graphite electrode holder, with an exposed surface area of 1 cm 2 . A platinum wire was adopted as the counter electrode and a Ag/AgCl in saturated KCl as the reference electrode. The reference electrode was calibrated against a reversible hydrogen electrode (RHE), and all potentials in the present study were referenced to this RHE.

DFT computation.
Spin-polarized density functional theory (DFT) calculations were carried out using the VASP (Vienna Ab-Initio Simulation Package) code [58]. Projector-augmented wave (PAW) method [59] with the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional [60] was used in all calculations. On-site Coulomb interactions were corrected within the DFT+U framework based on Dudarev's approximation [61]. U eff = 4:20 [62] and 6.40 [63] for Fe and Ni, respectively, were used [64]. Plane-wave basis set with a 400 eV energy cutoff provides a balance of accuracy and computational cost. Either quasi-Newton scheme or conjugate gradient algorithm implemented in VASP was used to relax structure until forces are converged to less than -0.03 eV Å -1 on unconstrained atoms and selfconsistent convergence until 10 -5 eV. The Gaussian smearing with a σ value of 0.05 was used to minimize entropy contribution to free energy. The bulk structure of NiFe 2 O 4 was taken from JCPDS (JCPDS Card No. 10-0325) and optimized in (4 × 4 × 1) k-point grid sampling of the surface Brillouin zone. The optimized lattice constant of 8.37 Å was found in close agreement with the experimental value of 8.35 Å.
A supercell consisting of five layers of NiFe 2 O 4 with an exposed (100) surface was constructed from the optimized bulk structure. A vacuum space of 14 Å in z-direction was inserted between the slabs, and the atoms in the top three layers were allowed to relax while those in the bottom two layers were fixed at the corresponding bulk position during structural optimization.
Additional details are included in the Supporting Information.

Data Availability
The data used to support the findings of this study are included within the article.
Energy (DOE) Office of Science by Argonne National Laboratory and was supported by the DOE under contract No. DE-AC02-06CH11357 and the Canadian Light Source and its funding partners. This research also used resources of the Center for Functional Nanomaterials (CFN), which is a U.S. Department of Energy Office of Science User Facility, at Brookhaven National Laboratory under Contract No. DE-SC0012704. The authors also thank Mr. Jeremy Barnett for the assistance in sample preparation and data acquisition of X-ray diffraction measurements in the UCSC X-ray Facility which was funded by the National Science Foundation (MRI-1126845).

Supplementary Materials
Additional computational details. Figure S1: crystal structure of FeNi oxide spinel, where blue, pink, and red indicate Fe, Ni, and O atoms, respectively. Figure Figure S6: EDS-based elemental mapping analysis. Figure S7: EDS-based elemental mapping analysis. Figure S8: XRD patterns of the sample series. Figure S9: XPS spectrum of the Cl 2p electrons of FeNiO-250-4. Figure S10: high-resolution XPS spectra of the Ni 2p electrons of (a) FeNiO NC -250-4 and (b) FeNiO-250-16. Figure Figure S19: SEM image and the corresponding EDS-based elemental mapping analysis of FeNiO-250-4 after stability test. Figure S20: PDOS of the Fe1 atom. Figure S21: additional materials prepared by MIHRQ. Table S1: elemental contents of the sample series based on EDS measurements. Table S2: fitting results of the EXAFS data of FeNiO-250-4. Table S3: fitting results of the EXAFS data of FeNiO NC -250-4. Table S4: fitting results of the EXAFS data of FeNiO-250-16.