See all Hide authors and affiliations

Edited by Thomas E. Mallouk, University of Pennsylvania, Philadelphia, PA, and approved June 8, 2021 (received for review January 29, 2021)

The efficiency with which renewable fuels and feedstocks are synthesized from electrical sources is largely limited by the sluggish water oxidation reaction. We show that the optimal water oxidation catalyst could be achieved by systematically modulating the coordination of the Ir active sites using an in situ cryogenic–photochemical reduction synthesis method. We achieved a highly oxidized Ir single site (Ir+5.3) in the best atom utilization by single-atom catalysts on electrochemically stable supports. The origin of water oxidation activity in an Ir single-atom catalyst is revealed experimentally and theoretically. The concept and strategy of this work are expected to pioneer novel approaches to engineer single-atom catalysts.

The efficiency of the synthesis of renewable fuels and feedstocks from electrical sources is limited, at present, by the sluggish water oxidation reaction. Single-atom catalysts (SACs) with a controllable coordination environment and exceptional atom utilization efficiency open new paradigms toward designing high-performance water oxidation catalysts. Here, using operando X-ray absorption spectroscopy measurements with calculations of spectra and electrochemical activity, we demonstrate that the origin of water oxidation activity of IrNiFe SACs is the presence of highly oxidized Ir single atom (Ir5.3+) in the NiFe oxyhydroxide under operating conditions. We show that the optimal water oxidation catalyst could be achieved by systematically increasing the oxidation state and modulating the coordination environment of the Ir active sites anchored atop the NiFe oxyhydroxide layers. Based on the proposed mechanism, we have successfully anchored Ir single-atom sites on NiFe oxyhydroxides (Ir0.1/Ni9Fe SAC) via a unique in situ cryogenic–photochemical reduction method that delivers an overpotential of 183 mV at 10 mA ⋅ cm−2 and retains its performance following 100 h of operation in 1 M KOH electrolyte, outperforming the reported catalysts and the commercial IrO2 catalysts. These findings open the avenue toward an atomic-level understanding of the oxygen evolution of catalytic centers under in operando conditions.

Efficient and cost-effective electrocatalysts play critical roles in energy conversion and storage and the societal pursuit of sustainable energy (1⇓ –3). The water oxidation reaction, also known as oxygen evolution reaction (OER), in particular, is an enabling process for diverse clean energy technologies, including water splitting (4⇓ –6), solar fuels (2), CO2 reduction (7), and rechargeable metal–air batteries (8). Unfortunately, the kinetics of the OER are sluggish, which limits the power conversion efficiency and the overall efficiency.

Very recently, higher valence transition metal ions such as Co4+ (9⇓ –11), Ni4+ (12⇓ –14), and Fe4+/5+ (15, 16) generated through a potential-dependent deprotonation reaction have been incorporated into metal oxides/hydroxides, resulting in enhanced water oxidation activity. Incorporating precious metals such as Ir, Ru, and Pt is much less explored, but they offer greater opportunities due to their tendency to form single-atom sites. Computational work has predicted either direct substitution of Ni4+ and Fe4+ by Ir4+ and Ru4+ or that the metal would be close to its most stable +4 oxidation state based on the high stability of the rutile phase (17⇓ ⇓ –20). On the other hand, the high activity of Sr-leached SrIrO3/IrOx and Li-removed LixIrO3 catalysts (5, 21, 22) and a recent prediction of highly active and high-oxidation homogeneous water oxidation systems (23) indicate that increased oxidation may lead to improved activity if the active site can be stabilized under operating conditions.

Single-atom catalysts (SACs) have offered an ideal system to precisely control the local coordination environments and oxidation states of the single-site centers (24⇓ ⇓ ⇓ ⇓ –29). The single-atom nature of these active centers leads to well-defined coordination environments and enhanced metal–support interactions, which provide remarkable catalytic performance in a number of heterogeneous reactions, including in the water–gas shift reaction (24), heterogeneous reduction of CO2 processes (27), CO oxidation (28), and oxygen reduction reaction (29). We adopted this strategy and sought to incorporate high-oxidation Ir metal sites into the support to enhance the water oxidation activity.

Here, we developed an in situ cryogenic–photochemical reduction method for anchoring Ir single sites on the NiFe oxyhydroxide support. Density functional theory (DFT) calculations predict unusually stable IrO6 octahedral SAC anchored atop the NiFe oxyhydroxide layers. In operando Ir L3-edge X-ray absorption spectroscopy (XAS) combined with spectra simulations of Ir L3-edge of such structures revealed high oxidation of Ir (+5.3) in the NiFe oxyhydroxides under operating conditions, in agreement with DFT predictions and spectra calculations. The highly oxidized Ir single-site exhibits exceptional OER performance, with a 183 mV overpotential at 10 mA ⋅ cm−2, outperforming the precious metal oxide IrO2. These findings are further corroborated by our calculations of the theoretical OER overpotentials, which show that increasing the Ir oxidation state leads to improved activity, and an overpotential of 0.184 V is obtained for the most oxidized NiFeIr:3O SAC. Such structural and compositional heterogeneity poses a key obstacle to unambiguously identifying the exact atomistic structure of the active sites and to further establishing a definitive correlation with the catalytic properties that can guide the subsequent design of future generations of SACs.

In order to test our hypothesis, we set out to explore how Ir is incorporated into NiFe oxyhydroxides. Using DFT, we have compared the free energy of Ir, substituting either Ni or Fe within the NiFeOOH layered structure (NiFe) to the free energy of anchoring oxidized Ir atop of the NiFe layer (Fig. 1A). For NiFe, we used a widely adopted γ-NiFeOOH bulk model with interlaced and solvated K+ cations and protons since 1 M KOH was used as an electrolyte in the experiment (see SI Appendix for computational details) (18, 30, 31). Out of a large number of tested cases, only two possibilities had significant negative free energy of Ir doping (Fig. 1A and SI Appendix, Fig. S1). The two stable configurations are 1) a conventional in-layer substitutional doping of Ir4+ replacing Ni4+ (NiFeIr) (17, 18) and 2) single IrO6 octahedron bonded directly to the NiFe layer, which we refer to as the SAC (NiFeIr SAC) configuration. A similar Ir–SAC configuration was recently proposed by Zhao et al. utilizing a Fe2O3 support (32). The protonation of the three outermost oxygens and the resulting Ir high-oxidation state in the NiFeIr SAC is a direct function of applied voltage (SAC NiFeIr:3OH versus NiFeIr:3O), but its stability is always greater than that of NiFeIr (>28 kJ/mol) (Fig. 1A and SI Appendix, Fig. S1). The remarkable ability of Ir to sustain a high-oxidation state, the availability to form multiple oxygen bonds at the NiFe layer, and the stabilization by nearby solvents allow for the presence of such a stable SAC configuration.

The preparation route to Ir single-atom on NiFe oxyhydroxides and atomic structure characterizations of NiFeIr by HAADF-STEM. (A) A DFT prediction of a preferred Ir atom embedding within the NiFeOOH layered structure under operating conditions. The conventional in-layer substitution of Ni4+ by Ir4+ (NiFeIr) (Left) is compared to a more preferred single IrO6 octahedron bonded to NiFe layer in SAC configuration (NiFeIr SAC) (Right). The energies with respect to given references are at zero bias. Note that NiFeIr SAC maintains its overall stability even under applied voltage (SI Appendix, Fig. S1). (B) In situ cryogenic–photochemical reduction synthesis of Ir0.1/Ni9Fe samples. (C) HAADF-STEM image of Ir single atoms on Ni9FeOOH supports (Ir0.1/Ni9Fe). Representative Ir single atoms are shown as bright spots with yellow circles.

In light of the theoretical prediction, we sought to synthesize Ir single atoms on NiFe oxyhydroxides (Ir0.1/Ni9Fe SAC) via the in situ cryogenic–photochemical reduction method (Fig. 1B). In this method, we first synthesized highly nanoporous Ni9Fe oxyhydroxides as the support (6). Next, a mixed aqueous solution of IrCl3(H2O)3 and Ni9Fe oxyhydroxides was frozen quickly in liquid nitrogen (−196.5 °C) to form a thin layer of ice. Then, the frozen solution was exposed to ultraviolet (UV) irradiation on a cooling stage with a temperature of −35 °C for 1 h. After the UV treatment, the frozen solution was kept for 24 h in dark conditions at 4 °C. After centrifuging, washing, and freeze-drying the samples multiple times, the samples from the in situ cryogenic–photochemical process were stabilized on Ni9Fe oxyhydroxide supports (SI Appendix). Aberration-corrected high-angle annular dark-field–scanning transmission electron microscopy (HAADF-STEM) showed Ir dispersed on Ni9Fe oxyhydroxides as individual atoms (Fig. 1C). From inductively coupled plasma optical emission spectrometry (ICP-OES) analysis, the molar ratio of Ni:Fe:Ir to be 9:1:0.1 is determined. Clusters were formed when increasing the Ir loading on Ni9FeOOH supports (SI Appendix, Fig. S2). Using the theoretical prediction and in situ cryogenic–photochemical synthesis, we have achieved stabilized and well-distributed Ir single-atom sites in the NiFe oxyhydroxides.

To investigate the local electronic configuration and atomic structures of the Ir0.1/Ni9Fe under OER operating conditions, we employed an in situ electrochemical cell for XAS measurements combined with absorption spectra simulations and extended X-ray absorption fine structure (EXAFS) fitting (Fig. 2). Electrochemistry was performed in a homemade cell using catalysts on carbon paper as a working electrode, an Ag/AgCl reference electrode, a platinum wire counter electrode, and 1 M KOH solution as the electrolyte (SI Appendix, Fig. S3).

In operando XAS characterization of the Ir0.1/Ni9Fe catalyst coupled to DFT simulated spectra. (A) Experimental Ir-L3 edge XANES spectra of Ir0.1/Ni9Fe SAC at OCV and after applying 1.35 V, 1.45 V, and 1.55 V versus RHE. (B) WL position of Ir0.1/Ni9Fe SAC during water oxidation reactions. Ir foil and IrO2 were used as references for the oxidation states of 0 and +4. For comparison, the normalized WL intensity from DFT simulated spectra is also shown on the separate y-axis. Calculated Ir Bader charges indicating higher oxidation states for SAC structure are also shown on secondary x-axis. (C) Theoretical Ir L3-edge XANES spectra of Ir foil, rutile-IrO2, Ir doped sites (NiFeIr), and Ir anchored sites (NiFeIr SAC) (see also Figs. 1 and 4). Insets show the corresponding differences in the structural models. (D) Experimental Ir L3-edge EXAFS spectra of Ir foil, Ir0.1/Ni9Fe at OCV, and at 1.4 V versus RHE.

The white lines (WLs) of Ir L3-edge X-ray absorption near edge structure (XANES) spectra for NiFeIr SAC under different applied potentials were compared with the ones from standard Ir foil and commercial IrO2 (Fig. 2A). For each condition, the Ir valence state was estimated from the position of the WL, which is expected to shift toward higher energy as the oxidation state increases. For Ir0.1/Ni9Fe SAC at open circuit, the energy position of the WL is very close to that of IrO2 reference, indicating a state close to Ir4+. When the applied potential was raised to 1.35 V versus the reversible hydrogen electrode (RHE), the WL gradually shifted toward higher energies, in agreement with a prior study on the IrOx/RuO2 system (33). Using the WL position of Ir and IrO2 as references and interpolating the rest of the samples to a linear dependency WL position versus valence (34), the XAS spectra indicated an Ir valence of +4 for Ir0.1/Ni9Fe SAC at open circuit voltage (OCV) and +5.3 for Ir0.1/Ni9Fe SAC at 1.45 V versus RHE, respectively (Fig. 2B). The position of the L3 WL shifts toward higher energy and is in agreement with the expected electronic state: 5d5 (Ir4+) for IrO2 and Ir0.1/Ni9Fe SAC at OCV and 5d3.7 (Ir5.3+) for Ir0.1/Ni9Fe SAC at 1.45 V versus RHE.

The OCEAN (Obtaining Core Excitations from Ab initio electronic structure and NIST core-level Bethe-Salpeter Equation) code (35, 36) simulation of the XANES region of the Ir L3-edge (Fig. 2C; see also SI Appendix for details) on DFT-optimized structures introduced in Fig. 1A is in very good agreement with the experimental data (Fig. 2A and SI Appendix, Fig. S4). The increase in the WL intensity from IrO2 to the calculated NiFeIr SAC structures in the simulated data implies an increase in the oxidation state of the Ir atoms that well matches the experimental trend (Fig. 2C). This correlation is further supported when comparing the WL intensity of the calculated spectra versus DFT Bader charge (Fig. 2B). Particularly, the WL intensity obtained experimentally for Ir0.1/Ni9Fe SAC under operating conditions is consistent with the simulated spectrum of NiFeIr:3OH SAC (Fig. 2C) and lies generally in between NiFeIr:3OH and NiFeIr:3O SAC (SI Appendix, Fig. S5), proving that Ir is present in the high-oxidation state under applied bias. Conversely, when the Ir atom substitutes one of the surface Ni atoms (NiFeIr), the Ir oxidation state is much closer to that of IrO2 as evidenced by a peak height (Fig. 2C) and the decrease in WL intensity in the simulated spectrum (Fig. 2B). However, the maximum theoretical oxidation state is not reached in the experiment, which is indicative of a possible mixture of the NiFeIr SAC and NiFeIr-type coordination in the Ir0.1/Ni9Fe SAC sample. Judging from the DFT-calculated Bader charge results in optimized bulk structures, the electronic structures and charge of Ni and Fe atoms with and without Ir doping are very similar.

To further verify that Ir0.1/Ni9Fe SAC contains only atomically dispersed individual Ir atoms, in situ EXAFS spectra were obtained for NiFeIr SAC at open circuit and at operating conditions (Fig. 2D). The Ir–Ir contribution at about 2.3 Å is absent in the k3-weighted EXAFS at the Ir L3-edge for both NiFeIr SAC at open circuit and at 1.4 V versus RHE, strongly indicating that Ir exists as isolated atoms during operating conditions. The only prominent shell located in between 1.5 Å and 2 Å for Ir0.1/Ni9Fe SAC at open circuit and at 1.4 V versus RHE, respectively, originated from the Ir-O contribution.

Next, we explored the water oxidation activity of Ir0.1/Ni9Fe SAC and relevant controls on Ni foam in 1 M KOH electrolyte. The backward sweep was taken for the calculation of overpotential to avoid any nonfaradaic contribution to the total current. Although NiFe oxyhydroxides are indeed the best nonprecious metal electrocatalysts, further modification is needed to achieve better performance. The beneficial role of Au on NiFe LDH has been reported recently (37). The Ir0.1/Ni9Fe SAC achieved the optimal water oxidation catalytic activity with the low overpotential of 183 mV at 10 mA ⋅ cm−2 (Fig. 3A), superior to that of the best commercial IrO2 (∼280 mV) (see also SI Appendix, Table S1 for detailed comparisons). The activity of Ir0.1/Ni9Fe SAC is superior to that of the Ir0.01/Ni9Fe owing to the formation of more Ir active sites (Fig. 3 A and B). However, upon incorporating higher Ir loading to 10%, the activity of Ir1/Ni9Fe catalyst decreased due to the formation of Ir clusters (SI Appendix, Fig. S2). The Ir0.1/Ni9Fe SAC has the least Tafel slope of 49 mV/dec, implying more favorable reaction kinetics toward water oxidation reaction than the control samples (38) (Fig. 3B). The electrochemical stability of the Ir0.1/Ni9Fe SAC was tested at a constant overpotential of 200 mV on Ni foam for 100 h (Fig. 3C). We observed no appreciable decrease in current density in this time interval. We used HAADF-STEM characterizations to prove that no Ir–Ir bond was formed in the Ir0.1/Ni9Fe SAC catalyst after electrochemical reaction, revealing the Ir atoms were dispersed at the atomic level (SI Appendix, Fig. S7). Obviously, the structure of Ir single sites after stability testing remained stable.

Performance of Ir0.1/Ni9Fe SAC and controls in the three-electrode system. (A) The OER polarization curves and (B) overpotential at 10 mA ⋅ cm−2 and Tafel slopes of Ir0.1/Ni9Fe SAC and controls loaded on Ni foam in 1 M KOH. (C) Chronoamperometry measurement of Ir0.1/Ni9Fe SAC on Ni foam electrode at a constant potential of 1.43 V versus RHE for 100 h. (D) Current density of Ir0.1/Ni9Fe SAC and Ni9FeOOH control plotted versus scan rates. The slopes obtained from a linear fit were used to represent ECSA. The inset shows the ECSA normalized current density at 1.45 V versus RHE for Ir0.1/Ni9Fe SAC and Ni9FeOOH, respectively.

Lastly, we investigated whether the water oxidation activity of Ir0.1/Ni9Fe SAC originates from intrinsic catalytic activity of Ir active sites or exclusively from an enhanced surface area. We analyzed the electrochemically active surface area (ECSA) and ECSA-normalized current density at 1.45 V versus RHE for Ir0.1/Ni9Fe SAC and Ni9FeOOH control (Fig. 3D). Based on ECSA results, the intrinsic activity of Ir0.1/Ni9Fe SAC is higher than that of Ni9FeOOH control. Ir0.1/Ni9Fe SAC shows the advantages not only in improving OER activity and stability but also lowering Ir usage and cost.

To complete our analysis of Ir0.1/Ni9Fe SAC system, we evaluated theoretical OER overpotentials (SI Appendix, Computational Details) of the obtained stable systems using Ir at edge sites, subsurface, and as single-atom near Fe sites [adapted from an Au single-atom study (37)] in the (100) surface of NiFe labeled as NiFeIr and octahedral NiFeIr SAC anchored at the NiFe basal planes [(001) facet]. These are compared to the (110) surface of rutile IrO2 and to the Fe site in NiFe (100) surface obtained previously (31). First, we show that the calculated ΔGOOH to ΔGOH scaling (Fig. 4A) for Ir sites is generally improved over the universal scaling, where OOH* is stabilized with nearby surface oxygens, and is also necessary in order to obtain theoretical overpotentials lower than 0.3 V. Additionally, the sites with Ir in higher oxidation states bind the adsorbates progressively, which further increases the OOH* stabilization. Theoretically, adding Ir into NiFe oxyhydroxides significantly improves the second step (OH* to O*) in the OER mechanism for Ir sites (Fig. 4B and SI Appendix, Fig. S9). The theoretical OER overpotentials, shown as a two-dimensional (2D) OER volcano plot, are summarized in Fig. 4B.

Tuning the OER energetics of NiFeOOH via Ir site engineering. (A) Improved ΔGOOH to ΔGOH (OH-OOH) scaling (ΔGOOH = ΔGOH + 2.686 eV) for the D–G models relative to the universal and ideal scaling lines. (B) Two-dimensional activity volcano plot with optimal region (shown in yellow) based on the (A) scaling and with positions of the thermodynamis for D–G systems. (C) Free energy diagram of the OER mechanism of our models relative to ideal catalyst (at U = 0 V), highlighting the close similarity with model G. (D–G) Structural models: (D) (110) surface of rutile-IrO2 (IrO2), (E) (100) surface of NiFe, (F) Ir site in the (100) surface of NiFeIr, and (G) (001) basal plane of NiFeIr SAC. Results for additional tested structures are shown in SI Appendix, Fig. S8. All structures are shown with the O* as adsorption intermediate. The atom colors are Ir, light blue; Fe, yellow; Ni, light gray; O, red; and H, light pink.

The 2D volcano plot further shows that the Ir site in the IrO2(110) binds significantly stronger (ΔGOH is the most negative in Fig. 4 A and B) than any other model considered and appears in the lower-left part of the 2D plot. On the other hand, the Fe site in the NiFe (100) binds significantly weaker due to the presence the Fe-O oxo intermediate (ΔGOH is the most positive in Fig. 4 A and B) and appears in the upper-right part of the 2D plot. Additionally, we have tested the effect captured by Zhang et al. for Au/NiFe LDHs on Fe sites (37). Our expanded calculations (SI Appendix, Fig. S9) show that the effect of adding Ir is different than that of adding Au in Ni9FeOOH. Adding single-atom Ir next to the Fe-active site [SI Appendix, Fig. S9, structure (i)] results in very reactive O* at the Ir-O-Fe bridge, which leads to unfavorable O* to OOH* OER step with overpotential of 1.8 V. For completeness, we have also tested the effect of subsurface doping by Ir next to the Fe-active site [SI Appendix, Fig. S9, structure (h)]. In all tested cases, adding Ir next to Fe did not improve the activity of the Fe-active site, indicating no synergetic interaction of Ir on Fe sites in the NiFe oxyhydroxides. Interestingly, all Ir sites in the NiFeIr and NiFeIr SAC models straddled two extrema set by two structures (Fig. 4 D and E) in a nearly linear fashion. This trend is well-correlated with the formal and calculated oxidation states of the Ir sites in the NiFeIr (+4.4) and NiFeIr SAC (+4.8 and +5.6) models (Fig. 2B). Improved OH-OOH scaling and a gradual decrease of 5d electrons at the Ir atoms to bond with active oxygen 2p states—indicative of weaker adsorbate bonding—in turn shifts the position of the NiFeIr SAC model inside the optimal OER region. A similar system on Ir-anchored NiFe oxyhydroxides was developed, in which they discovered that the isolated Ir sites undergo a deprotonation process to form the multiple Ir+3.57 active sites during OER, promoting the O–O coupling (39). The final calculated theoretical overpotentials are 0.184 V for NiFeIr:3O SAC (Fig. 4G), 0.333 V for the Ir site in NiFeIr(100) (Fig. 4F), 0.381 V for the Fe site in NiFe(100) (Fig. 4E), and, finally, 0.664 V for IrO2 (110) (Fig. 4D). The very low overpotential of 0.184 V for NiFeIr SAC is obtained only for well-stabilized OOH intermediates and agrees well with the experimental overpotential of 183 mV for Ir0.1/Ni9Fe SAC catalyst. A detailed list of DFT energies and structures is provided via the catalysis-hub.org repository (40, 41).

In summary, we established a protocol in which Ir single-atom sites are stabilized on NiFe oxyhydroxides via an in situ cryogenic–photochemical reduction method, which leads to Ir0.1/Ni9Fe SAC catalysts with superior water oxidation performance. By combining XAS analyses with simulated spectra and calculated OER activity, we unambiguously identify that Ir species adopt a single-atom configuration in an oxidation state greater than +5. The Ir0.1/Ni9Fe SAC delivers an overpotential of 183 mV at 10 mA ⋅ cm−2, which halves the 0.3 to 0.4 V OER overpotential wall and is superior to most Ir-based catalysts known to date. The above experimental findings are further corroborated by the DFT calculations, which show that the improved activity is due to weaker binding and improved OH-OOH scaling of the highly oxidized Ir–SAC sites. Importantly, the utilization of NiFe oxyhydroxides as catalytic support provides an ideal model system to probe and quantitatively establish the correlation between the atomistic structure of the single-metal centers and their catalytic properties for a diverse set of electrocatalytic reactions. Our findings demonstrate unique electrochemical properties of SACs, with general implications for the design of materials employing oxygen redox chemistry.

Our general in situ cryogenic–photochemical reduction synthesis of Ir single atoms/Ni9FeOOH is illustrated in Fig. 1B. First, a controlled ratio of 0.1 mg ⋅ mL−1 IrCl3(H2O)3 and Ni9FeOOH in aqueous solution was prepared in a beaker, and the beaker was frozen quickly in a liquid nitrogen (−196.5 °C) bath in dark conditions, keeping the beaker flat. The mixed solution can be frozen quickly to form ultrathin ice, which ensures that the homogeneous distribution of Ir precursor molecules in the solution can be maintained during the solidification process. Second, the frozen ice was exposed to UV irradiation on a cooling stage with a temperature of −35 °C for 1 h in an environmental chamber (Associated Environmental Systems), avoiding ice melting and forming Ir single atoms on Ni9FeOOH support. The powder density of the UV light near ice was measured (0.7 mW ⋅ cm−2) using a radiometer (Newport Model PMKIT-05-01). Then, the Ir single atoms on Ni9FeOOH samples were filtered, cleaned, and dried in vacuo in a Labconco FreeZone 6 freeze dryer. Finally, the Ir0.1/Ni9Fe SACs were collected.

HAADF-STEM was taken on a TEAM 0.5 transmission electron microscope operated at 300 kV at the National Center for Electron Microscopy. The samples were prepared by dropping catalyst powder dispersed in ethanol onto carbon-coated copper TEM grids (Ted Pella) using micropipettes and were dried under ambient conditions. The atomic composition of Ir/NiFe catalyst was determined using an ICP-OES (Agilent 720 with a hermetically sealed charge-coupled device (CCD) detector).

In the homemade cell, electrochemistry was performed using catalysts on carbon paper as a working electrode, an Ag/AgCl reference electrode, a platinum wire counter electrode, and 1 M KOH solution as the electrolyte (SI Appendix, Fig. S3). The in situ Ir L3-edge XAS data were collected at the X-Ray Fluorescence Microprobe beamline 10.3.2 of the Advanced Light Source (ALS) and beamline 4-1 at the Stanford Synchrotron Radiation Lightsource. The ALS data were preprocessed (deadtime corrected, calibrated, and deglitched) using custom LabVIEW software available at the beamline. The spectra were further aligned, calibrated, and normalized using Athena software (Demeter version 0.9.26). To extract EXAFS oscillations, the background was removed in k-space using a five-domain cubic spline. Ir L3-edge EXAFS were fitted in R-space using artemis with the DFT-optimized NiFeIr:3Os model (SI Appendix, Fig. S8).

Electrochemical measurements were performed using a three-electrode system connected to an electrochemical workstation (Bio-Logic VMP3) with a built-in electrochemical impedance spectroscopy analyzer. The working electrode was catalysts deposited on Ni foam. Ag/AgCl (with saturated KCl as the filling solution) and platinum foil were used as reference and counter electrodes, respectively. To load the catalyst on a Ni foam, 20 mg of catalyst was dispersed in a mixture containing 2 mL of water and 2 mL ethanol, followed by the addition of 100 μL Nafion solution. The suspension was sonicated for 30 min to prepare a homogeneous ink. Ni foam coated with water-resistant silicone glue was drop-casted with the catalyst ink. Unless otherwise stated, all experiments were performed at ambient temperature (22 ± 2 °C). Electrode potentials were converted to the RHE based on the following equation: E(RHE) = E(Ag/AgCl) + 0.197 V + 0.059 × pH.

The ECSA of each catalyst was calculated based on measuring their electrochemical capacitances. We measured the CVs in a narrow non-Faradaic potential window in which the change of the current is principally due to the charging of the double-layer, which is expected to be linearly proportional to the active surface area. The measured capacitive current densities at the average potential in the selected range were plotted as a function of scan rates. The slope of the linear fit was calculated as the double-layer capacitance (Cdl). The ECSA of the catalyst is calculated from the following equation:ECSA=CdlCscmECSA2 ECSA=CdI/Cs.

Cs is the specific capacitance in 1 M KOH electrolyte. The intrinsic activity was obtained by normalizing the current to the ECSA to exclude the effect of surface area on catalytic performance.

The density functional calculations were performed within the Vienna ab initio simulation package using the projector-augmented wave potentials. We employed the Perdew-Burke-Ernzerh (PBE) functional together with the Hubbard-U correction method applied for the d-electrons of Ni (U-J = Ueff = 6.45 eV) and Fe (Ueff = 4.3 eV) atoms. For Ni, the Ueff parameter was chosen according to the Materials Project Database, while for Fe, it was taken from recent NiFeOOH paper of Martirez and Carter (42). For NiFeOOH system, other authors have used similar values for NiFeOOH that are within +/−1 eV. The bulk optimization calculations were performed at the energy cutoff of 600 eV on the 3 × 3 × 4 k-point mesh per 3 × 3 × 2 unit cell of the NiFeOOH [K8(Ni22Fe2)O48 + 16H2O] used in our recent work. In this cell, a single Ir atom was placed either in the conventional in-layer substitutional doping (replacing Ni4+, Ni3+, and Fe4+ sites) or in the SAC configuration relative to Ni and Fe and variable OH*/O* groups (see results in SI Appendix, Fig. S1).

For surfaces, we have employed 3 × 1 symmetric (100) slabs of four layers of M–O (keeping the bottom two fixed) and 3 × 2 × 2 (001) basal plane (bottom layer fixed) and 15 Å of a vacuum. Here, we used 500 eV energy cutoff and the 4 × 4 × 1 and (3 × 3 × 1) k-point meshes, respectively. Full relaxation below a minimum threshold force of 0.02 eV/Å2 was performed in all cases. To obtain the theoretical overpotential for each surface, the standard OER mechanism and which has been applied to many types of oxides [* → OH*, OH* → O*, O* → OOH*, OOH* → O2(g)] was assumed. The Gibbs free energies of the OER intermediates calculated via the computational hydrogen electrode method include room temperature corrections, zero-point energy, and the vibrational enthalpy and entropy contributions [relative to H2(g) and H2O(l)] obtained by means of the harmonic approximation. The combined effect of total energies of adsorption is ΔGcorr. (OH*) = 0.302 eV, ΔGcorr. (O*) = −0.015 eV, and ΔGcorr. (OOH*) = 0.345 eV, respectively. The fully optimized bulk models, reference structures, and their respective energies are included as part of the catalysis-hub.org repository.

The Ir L3-edge for the different structures was calculated using the ocean code. Electron wave functions were generated using the quantumespresso code. The norm-conserving pseudopotentials for Ir and O generated using the FHI98PP package using Trouiller–Martins type, PBE-generalized gradient approximation (GGA) was utilized. For the simulation of rutile IrO2 and Ir metal, experimental crystallographic structures were used (Inorganic Crystal Structure Database [ICSD] #84577 and #640730, respectively) with a k-points sampling of (4 × 4 × 4), 1,000 screening bands, and (2 × 2 × 2) k-points sampling for the screening calculation. For the DFT generated structures, the k-points sampling was decreased to (2 × 2 × 1) with 2,000 screening bands. The same core-offset energy, plane-wave energy cutoff (120 Rydberg), and broadening (3.5 eV) were utilized for all the simulations.

The simulated spectra were normalized and aligned to the experimental results. The WL was numerically integrated after fitting the data using Lorentzian and arctangent fit function (SI Appendix, Fig. S4). The WL intensity was normalized by the WL intensity of the simulation for rutile IrO2. The WLs of the experimental data were normalized in the same way using the experimental IrO2 as standard.

A detailed list of DFT energies and structures is provided via the catalysis-hub.org repository (40, 41) (https://www.catalysis-hub.org/publications/ZhengOrigin2020). The fully optimized bulk models, reference structures, and their respective energies are also included as part of the catalysis-hub.org repository. All other study data are included in the article and/or supporting information.

This work was supported by the Department of Energy (DOE), Office of Basic Energy Sciences, Division of Materials Sciences and Engineering (Contract DE-AC02-76SF00515) and by DE-SC0008685 to the SUNCAT Center of Interface Science and Catalysis. The Advanced Light Source and Molecular Foundry are supported by the Director, Office of Science, Office of Basic Energy Sciences of the US DOE under Contract DE-AC02-05CH11231 and the use of the computer time allocation for the “Transition metal-oxide and metal surfaces: applications and reactivity trends in catalysis” at the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the US DOE under Contract DE-AC02-05CH11231. Use of the Stanford Synchrotron Radiation Light Source, SLAC National Accelerator Laboratory, is supported by the US DOE, Office of Science, Office of Basic Energy Sciences under Contract DE-AC02-76SF00515. We acknowledge the support on electron microscopy from Stanford Nano Shared Facilities.

↵1X.Z. and J.T. contributed equally to this work.

Author contributions: X.Z., J.T., M.B., and Y.C. designed research; X.Z., J.T., X.Y., C.J.A., H.M., S.C.F., C.S., and R.C.D. performed research; X.Z., J.T., A.G., J.A.G.T., E.M.B., P.E., S.C.F., C.S., R.C.D., J.A.R., J.V., and Y.C. contributed new reagents/analytic tools; X.Z., J.T., and Y.C. analyzed data; and X.Z., J.T., A.G., M.B., and Y.C. wrote the paper.

The authors declare no competing interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2101817118/-/DCSupplemental.

This open access article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).

Thank you for your interest in spreading the word on PNAS.

NOTE: We only request your email address so that the person you are recommending the page to knows that you wanted them to see it, and that it is not junk mail. We do not capture any email address.

Feedback    Privacy/Legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID, and Research4Life.