Highly active catalyst derived from a 3D foam ofFe(PO3)2/Ni2P for extremely efficient water oxidationHaiqing Zhoua,1, Fang Yua,1, Jingying Suna, Ran Hea, Shuo Chena,2, Ching-Wu Chua,b,2, and Zhifeng Rena,2

aDepartment of Physics and Texas Center for Superconductivity, University of Houston, Houston, TX 77204; and bLawrence Berkeley National Laboratory,Berkeley, CA 94720

Contributed by Ching-Wu Chu, April 21, 2017 (sent for review January 30, 2017; reviewed by Hui-Ming Cheng and Song Jin)

Commercial hydrogen production by electrocatalytic water split-ting will benefit from the realization of more efficient and lessexpensive catalysts compared with noble metal catalysts, espe-cially for the oxygen evolution reaction, which requires a currentdensity of 500 mA/cm2 at an overpotential below 300 mV withlong-term stability. Here we report a robust oxygen-evolving elec-trocatalyst consisting of ferrous metaphosphate on self-supportedconductive nickel foam that is commercially available in largescale. We find that this catalyst, which may be associated withthe in situ generated nickel–iron oxide/hydroxide and iron oxyhydr-oxide catalysts at the surface, yields current densities of 10 mA/cm2

at an overpotential of 177 mV, 500 mA/cm2 at only 265 mV, and1,705 mA/cm2 at 300 mV, with high durability in alkaline electrolyteof 1 M KOH even after 10,000 cycles, representing activity enhance-ment by a factor of 49 in boosting water oxidation at 300 mV rel-ative to the state-of-the-art IrO2 catalyst.

iron | electrocatalytic water splitting | ferrous metaphosphate |oxygen evolution reaction | commercial utilization

Severe deterioration of the environment caused by the con-sumption of vast amounts of fossil fuels is driving us to ex-

plore renewable energy sources (1). Hydrogen (H2) producedfrom water splitting by an electrochemical process called waterelectrolysis has been considered to be a clean and sustainableenergy resource to replace fossil fuels and meet the rising globalenergy demand, because water is both the sole starting materialand byproduct when clean energy is produced by converting H2back to water (2, 3). To realize the overall water electrolysis forH2 production, oxygen evolution reaction (OER), also namedwater oxidation, plays another key role (4). OER is also an im-portant oxidative reaction in obtaining carbon fuels from CO2reduction (5) and achieving rechargeable metal–air batteries (6),and it has been meticulously studied for more than half a cen-tury. However, owing to the sluggish four-proton-coupled elec-tron transfer and rigid oxygen–oxygen bonding, this key processremains a major bottleneck in the water-splitting process. State-of-the-art OER catalysts, such as iridium dioxide (IrO2) andruthenium dioxide (RuO2), still require overpotentials of around300 mV to achieve current densities on the order of 10 mA/cm2,not to mention their scarcity and high costs, which severelyhinder the substantial market penetration of this technique (7).Thus, it is highly desirable and imperative to develop robust andstable oxygen-evolving electrocatalysts from earth-abundant andcost-effective elements.Commercial water electrolyzers require a competent electro-

catalyst that can efficiently deliver an oxidative current densityabove 500 mA/cm2 with long-term stability at overpotentials< 300 mV (8, 9). Although various earth-abundant materialshave been proven to be efficient catalysts for oxygen evolution,such as transition metal oxides (8, 10, 11), hydroxides (12, 13),oxyhydroxides (14, 15), phosphates (16–18), phosphides (19, 20),metal–organic frameworks (21), and carbon nanomaterials (22),many of them cannot meet the aforementioned commercial cri-teria for water–alkali electrolyzers, and, most importantly, theymay not survive long in high-current operation. To this end, we

report a highly efficient electrocatalyst for oxygen evolution re-action yielding current densities of 10 and 500 mA/cm2 at over-potentials of 177 and 265 mV, respectively, both of which are thelowest overpotential values for the corresponding current densitiesever reported, and showing excellent long-term stability.

Results and DiscussionElectrocatalytic Oxygen Evolution Reaction. We found that thecatalyst responsible for the large current density is Fe(PO3)2,which was formed on the surface of the conductive Ni2P/Ni foamscaffold (SI Appendix, Figs. S1–S4). We first investigated theOER activity of this Fe(PO3)2 catalyst and the correspondingreference materials (Ni foam, Ni2P, and IrO2) in 0.1 M KOHelectrolyte, as shown in SI Appendix, Fig. S5A. Among theseelectrodes, the Fe(PO3)2 electrode exhibits the highest OERcatalytic activity, where an overpotential as low as 218 mV vs.reversible hydrogen electrode (RHE) is required to yield ageometric current density of 10 mA/cm2; whereas, the Ni foam,Ni2P, and the state-of-the-art IrO2 electrodes require 377, 330,and 297 mV vs. RHE, respectively, as shown in SI Appendix, Fig.S5B. This overpotential of 218 mV is much smaller than those ofmost currently reported OER electrocatalysts (23), which clearlyindicates that Fe(PO3)2 is an outstanding OER catalyst in 0.1 MKOH electrolyte (SI Appendix, Table S1). In addition to OERactivity, stability is another critical criterion for evaluating thecatalysts. We found that Fe(PO3)2 is very stable at 10 mA/cm2 forover 20 h in 0.1 M KOH electrolyte, as shown in SI Appendix,Fig. S5C.

Significance

The oxygen evolution reaction (OER) is a sluggish reaction withpoor catalytic efficiency, which is one of the major bottlenecksin realizing water splitting, CO2 reduction, and rechargeablemetal–air batteries. In particular, the commercial utilizationof water electrolyzers requires an exceptional electrocatalystthat has the capacity of delivering ultra-high oxidative currentdensities above 500 mA/cm2 at an overpotential below 300 mVwith long-term durability. Few catalysts can satisfy such strictcriteria. Here we report a promising oxygen-evolving catalystwith superior catalytic performance and long-term durability;to the best of our knowledge, it is one of the most active OERcatalysts reported thus far that satisfies the criteria for large-scale commercialization of water–alkali electrolyzers.

Author contributions: S.C., C.-W.C., and Z.R. led the project and conceived of the originalidea; H.Z. and F.Y. designed research; H.Z., F.Y., J.S., and R.H. performed research; H.Z.and F.Y. analyzed data; and H.Z., F.Y., S.C., C.-W.C., and Z.R. wrote the paper.

Reviewers: H.-M.C., Institute of Metal Research, Chinese Academy of Sciences; and S.J.,University of Wisconsin–Madison.

The authors declare no conflict of interest.1H.Z. and F.Y. contributed equally to this work.2To whom correspondence may be addressed. Email: cwchu@uh.edu, schen34@uh.edu, orzren@uh.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1701562114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1701562114 PNAS | May 30, 2017 | vol. 114 | no. 22 | 5607–5611

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020

With an eye toward commercial applications, we also mea-sured the catalysts in 1 M KOH electrolyte. The higher OH−

concentration leads to higher electrical conductivity and poten-tially superior catalytic performance. The catalytic performanceamong the different electrodes shows the same trend (Fig. 1A) asin 0.1 M KOH, but with much higher current densities. Appar-ently, an overpotential larger than 380 mV is necessary to yield acurrent density of 500 mA/cm2 for the IrO2 electrode, consid-erably higher than the 265 mV overpotential needed for theFe(PO3)2 electrode to reach the same current density. When weapply an overpotential of 300 mV to activate oxygen evolution,this Fe(PO3)2 catalyst yields a current density of 1,705 mA/cm2,which is 341, 30, and 49 times better than that of the Ni foam,Ni2P, and IrO2 catalysts, respectively, suggesting a huge im-provement mainly originated from the Fe(PO3)2 itself, ratherthan from the Ni2P/Ni foam support. The steady-state electro-chemical analysis shown in Fig. 1B reveals that this electrodepossesses a small Tafel slope of 51.9 mV/dec, from which we canexpect that a 177-mV overpotential is needed to reach a currentdensity of 10 mA/cm2 for this electrode. Upon cyclic voltam-mogram testing, the anodic current density of this electrodeexhibits an obvious increase after 1,000 cycles compared with itsinitial state, and it then maintains nearly the same activity up to10,000 cycles (Fig. 1C), corroborating the finding that the cata-lyst is extremely durable to withstand accelerated degradation.Notably, the Fe(PO3)2 electrode sustains at steady catalyticcurrent densities of 100 and 500 mA/cm2 with very low over-potentials of 221 and 273 mV, respectively, even after chro-noamperometry testing for 20 h (Fig. 1D). To the best of ourknowledge, Fe(PO3)2 is one of the best performing catalysts sofar reported and it most successfully fulfills the strict criteria forlarge-scale commercialization of water–alkali electrolyzers (SIAppendix, Table S2).

Structural Characterization and Possible Mechanisms. To determinethe possible origins of such extraordinary performance, it is essen-tial to elucidate the crystalline structures and surface composition

of the electrocatalyst before and after electrochemical OERtesting. As mentioned above, Ni foam was used as the conductivesupport because of its low cost, good conductivity, and 3Dmacroporous feature (24, 25). Typical SEM images (SI Appendix,Fig. S3) show that the Fe(PO3)2 catalysts are uniformly distrib-uted on the Ni2P/Ni foam surface where a good electrical contactis formed between the (FePO3)2 catalysts and the highly con-ductive Ni2P, ensuring quick charge transfer between the catalystand the electrode as indicated by electrochemical impedancespectroscopy (EIS) measurements, discussed below. On theother hand, the crystalline features of these as-grown Fe(PO3)2particles can be well resolved from the high-resolution TEMimage (Fig. 2A), in which many nanocrystals show distinct latticefringes with 0.461-, 0.280-, 0.241-, and 0.198-nm lattice spacingsmarked by parallel lines, matching well with the interplanarspacings of the (�202), (221), (�132), and (�241) crystal planes,respectively, of Fe(PO3)2. The fast-Fourier transform (FFT)pattern taken from Fig. 2A consisting of discrete spots is anothersolid piece of evidence to confirm the crystalline nature of theseparticles. In contrast, such crystalline material undergoes somestructure changes after 10,000-cycle OER testing (SI Appendix,Fig. S6), and evolves into mainly an amorphous material asconfirmed by the TEM image and FFT pattern (Fig. 2B), whichcan be further verified by the Raman spectra due to differentvibration modes from different materials (Fig. 2C). It should benoted that no Raman vibration modes are detected on the Ni2Pcrystals at the excitation laser of 633 nm (SI Appendix, Fig. S7);however, two prominent peaks are located at 682 and 1,156 cm−1

for Fe(PO3)2, which can be attributed to the symmetric PO2−

stretching vibration modes related to the inequivalent P–Onbbonds and the symmetric stretching vibration modes associatedwith the P–O–P bonds, respectively. Both are unique to theFe(PO3)2 crystal. Other Raman peaks below 600 cm−1 are re-lated to network bending modes (26). After OER testing, nosuch vibration modes of the Fe(PO3)2 crystal are detected, butsome other distinctive peaks are observed and are associatedwith the unique Raman features of amorphous iron oxides (27),

Fig. 1. Electrocatalytic water oxidation activity. (A) Polarization curves recorded on different electrodes with a three-electrode configuration in 1 M KOHelectrolyte. (B) The relevant Tafel plots of the catalysts studied in A. (C) Polarization curves of the Fe(PO3)2 catalyst at its initial state and after 1,000 and10,000 cycles. (D) Chronoamperometric measurements of the OER at high current densities of 100 and 500 mA/cm2 for the Fe(PO3)2 electrode.

5608 | www.pnas.org/cgi/doi/10.1073/pnas.1701562114 Zhou et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020

rather than those of Ni2P-derived nickel oxides (SI Appendix,Fig. S7), further indicating structure changes in this Fe(PO3)2crystal during the OER electrocatalysis. In addition, elaborateXPS analysis (Fig. 2 D–F) helps us confirm that the originalmaterial is indeed Fe(PO3)2 based on the binding energies of Fe2p3/2, satellite peak, P 2p, and O 1s core levels extracted from theXPS data (28) and the atomic ratio (SI Appendix, Table S3), andthe final compound after OER testing of 10,000 cycles is possiblyamorphous FeOOH, judging from the binding energies Fe 2pand O 1s as well as the disappearance of P signals (14, 29). Theseresults suggest that the real catalytic sites may originate from theFe(PO3)2-derived amorphous FeOOH, which may explain whythe catalytic activity improves during the first 1,000 cycles.Whether as-synthesized amorphous FeOOH on Ni foam canperform as well as the Fe(PO3)2-derived amorphous FeOOH isnow being studied and will be reported accordingly (Fig. 1E). Infact, when FeOOH catalysts are electrodeposited onto some 3Dconductive scaffolds, such as CeO2 or metallic cobalt nanotubearrays on Ni foam, FeOOH has been experimentally proven tobe a good OER catalyst with an overpotential of 300 mV foraround 100 mA/cm2 current density (14, 30). Structural changesof nonoxide catalysts like selenides and phosphides have beenreported during OER operation (11, 19). Finally, it should benoted that there is an additional O 1s XPS peak located ataround 529.6 eV. By carrying out Raman measurements on thepost-OER Ni2P (SI Appendix, Fig. S7), and XPS analysis on theNi 2p3/2 binding energies of the original and post-OER Fe(PO3)2

electrode (SI Appendix, Fig. S8), we conclude that a small amountof nickel oxide/hydroxide is evolved from the Ni2P support at thesurface during OER electrocatalysis. This observation is consis-tent with the results reported previously by Hu and coworkers(19), who also reported Ni2P crystals as the starting materials forOER testing. Given that nickel oxide/hydroxide is in situ gener-ated from Ni2P during OER electrocatalysis, it should be notedthat a trace amount of iron impurities may incorporate into nickeloxide/hydroxide catalysts as reported previously (31, 32), resultingin the possible formation of Ni–Fe oxide/hydroxide in the post-OER catalysts. The incorporation of a small amount of Fe im-purities plays a positive role in boosting the catalytic activity ofnickel oxide/hydroxide by improving the electrical conductivityand affecting the electronic structure. Consequently, the finalcompounds derived from the Fe(PO3)2/Ni2P hybrid after OERtesting are probably a mixture of amorphous FeOOH and Ni–Feoxide/hydroxide, which are responsible for the superior catalyticperformance toward water oxidation.

High Intrinsic Catalytic Activity. Electrochemically active surfacearea is an important contributor to boosting the catalytic activityof OER catalysts. To verify this point, a simple cyclic voltammetrymethod (SI Appendix, Fig. S9) was introduced to determine thedouble-layer capacitance (Cdl), which has been deemed to beproportional to the effective surface area of the electrode (25, 33,34). By comparing the capacitance values among different cata-lysts, we can derive that the Fe(PO3)2 electrode has a capacitance

Fig. 2. Structural characterization. (A and B) High-resolution TEM images and FFT patterns (Insets) of the Fe(PO3)2 catalyst: (A) As prepared, displaying goodcrystallization, and (B) post-OER (after 10,000 cycles), mainly in an amorphous state. (C) Raman spectra of as-prepared and post-OER (after 10,000 cycles)catalysts. (D) XPS spectra of P 2p binding energies before and after OER tests (after 10,000 cycles). The P 2p peak in the original samples can be deconvolutedinto two components, 2p3/2 at 133.9 eV and 2p1/2 at 134.7 eV, confirming the formation of PO3

− compounds, and no P signal is detected in post-OER samples,suggesting structure changes on the catalyst surface. (E) XPS spectra of O 1s binding energies before and after OER tests (after 10,000 cycles). The originalsamples have two components of 531.8 eV for PO3

− and 533.4 eV for adsorbed H2O, and post-OER samples show O 1s core-level features consisting of FeOOHand nickel oxide, meaning that amorphous FeOOH is probably the dominant active site for water oxidation. (F) XPS spectra of Fe 2p3/2 and 2p1/2 bindingenergies before and after OER tests (after 10,000 cycles). It is apparent that the valence state of the Fe element is +2 for the as-synthesized samples, and it is graduallyconverted to +3 at the surface during water oxidation. The black and red curves in D–F are the original and fitted data, respectively. a.u., arbitrary unit.

Zhou et al. PNAS | May 30, 2017 | vol. 114 | no. 22 | 5609

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020

that is a 1.4- and a 3.5-fold increase over those measured for theNi2P and the Ni foam substrate, respectively (Fig. 3A), indicatingsome contribution from the highly active surface area of the as-prepared catalysts. However, there is a 30- and 341-fold im-provement in the current density of the Fe(PO3)2 electrodecompared with those of Ni2P and Ni foam (Fig. 3B), respectively.This suggests that its superior performance cannot be merelyattributed to the change of active surface area (Fig. 3A), but to ahigher intrinsic catalytic activity for water oxidation reaction thanin Ni2P catalysts and Ni foam substrate (11). To gain furtherinsight into the high intrinsic catalytic activity, we evaluated therelevant turnover frequencies (TOFs) of this Fe(PO3)2 catalyst,which can be derived from the equation (35, 36). TOF = j × A/(4 × F × m), where j, A, F, and m are the current density, surfacearea, Faraday constant, and number of moles of the active cat-alysts, respectively. This catalyst exhibits a TOF value around0.12 s−1 per 3d Fe atom at 300 mV in 1 M KOH, assumingthat all of the Fe ions in the catalyst are electrochemically activein catalytic water oxidation. This value is very much under-estimated, because not every metal atom could be catalyticallyactive in the OER process due to the 3D architecture. Even so, itis larger than many reported OER catalysts like NiFe layereddouble hydroxides (35, 36). On the other hand, such superiorcatalytic performance may also be related to the improvedelectrical conductivity, which has a significant impact on therelevant electron transfer between the catalyst and the support.To clarify this, we performed EIS measurements to check theelectrode kinetics of different catalysts (Fig. 3C). It is noteworthythat each Nyquist plot can be fitted by a semicircle with thesimplified Randle circuit model shown in the inset, from whichwe can extract the series resistance (Rs) and charge-transfer re-sistance (Rct). Indeed, the Fe(PO3)2 electrode possesses a muchsmaller Rct compared with other catalysts, suggesting facilitatedcharge transfer between the catalyst and the electrode. Thus, wethink that such outstanding catalytic performance may be asso-ciated with the high intrinsic catalytic activity, highly electro-chemically active surface area, and efficient charge transfer fromthe electrode.

ConclusionsA robust and stable Fe(PO3)2 catalyst supported on commercialNi foam is found to be highly efficient for water oxidation inelectrocatalytic water splitting. It only requires an overpotentialof 265 mV to yield a current density of 500 mA/cm2 in 1 M KOHwith excellent electrochemical durability for 10,000 cycles duringOER testing, and can survive at 500 mA/cm2 operation for over

20 h. The fabrication process for such an exceptional electro-catalyst is compatible with industrial standards and is economi-cally viable for large-scale production. We believe our finding isa giant step toward practical and economic production of hy-drogen by water splitting, which will significantly contribute tothe effort to reduce the consumption of fossil fuels.

Materials and MethodsChemicals. Iron(III) nitrate nonahydrate [Fe(NO3)30.9H2O, ≥99.95%; Sigma-Aldrich], Nafion 117 solution (5%; Sigma-Aldrich), sodium hypophosphitemonohydrate (NaH2PO2·H2O; Alfa Aesar), iridium oxide powder (IrO2, 99%;Alfa Aesar), potassium hydroxide (KOH, 50% wt/vol; Alfa Aesar), and Nifoam (1.5 mm, areal density 320 g/cm2) were used as received.

Growth of Ferrous Metaphosphate Catalyst on 3D Ni Foam. Ferrous meta-phosphate [Fe(PO3)2] was fabricated by a direct thermal phosphidationprocess in a tube furnace under Ar atmosphere. First, a 1.6 cm × 0.5 cm pieceof commercial Ni foam was dipped into an iron nitrate solution and slowlydried in air, and was then thermally phosphatized in argon gas at 450 °C for1 h to form Fe(PO3)2 crystals. The phosphorus source was 500 mg sodiumhypophosphite monohydrate (NaH2PO2∙H2O), which was put at the upstreamat around 400 °C. After thermal phosphidation, the sample was cooled downto room temperature under the protection of argon. The sample was thenimmersed into the iron nitrate precursor again, followed by a second thermalphosphidation under the same conditions. The as-prepared sample was usedas the working electrode directly. The iron nitrate solution was prepared bydissolving 0.75g Fe(NO3)3∙9H2O precursor in 5 mL deionized water with aresistivity of 18.3 MΩ cm. During the thermal phosphidation, the Ni2P wasformed on the surface of Ni foam simultaneously with the ferrous meta-phosphate. Therefore, the arrangement of our sample is Fe(PO3)2/Ni2P/Nifoam. The catalyst Fe(PO3)2 loading is around 8 mg/cm2.

Synthesis of Ni2P Catalyst on 3D Ni Foam. For comparison, the Ni2P was syn-thesized under the same conditions as those for Fe(PO3)2 preparation. Theonly difference was that the Ni foam was thermally phosphatized directlywithout iron nitrate solution.

Preparation of IrO2 Electrode on 3D Ni Foam. To prepare the IrO2 workingelectrode, 40 mg IrO2, 60 μL Nafion, 540 μL ethanol, and 400 μL deionizedwater (18.3 MΩ·cm resistivity) were ultrasonicated for 30 min to obtain ahomogeneous dispersion. After dip coating IrO2 dispersion onto Ni foam,this sample was placed in the fume hood overnight, followed by heating at120 °C for 3 h in Ar gas in a tube furnace. The loading of IrO2 catalyst on Nifoam is ∼8 mg/cm2.

Electrochemical Testing. The electrochemical tests were performed in a three-electrode system in 1 M or 0.1 M KOH electrolyte purged with high-purityoxygen gas continuously. A Pt wire and mercury/mercurous oxide (Hg/HgO)reference were used as the counter and reference electrodes, respectively. Thecatalysts on Ni foam were used as the working electrode directly. The OER

Fig. 3. Double-layer capacitance (Cdl) and EIS measurements. (A) Capacitive△J (= Ja − Jc) versus the scan rates for the Fe(PO3)2 electrode compared with Ni2Pand Ni foam. (B) Comparison of the current density of the Fe(PO3)2 electrode with those of the benchmark IrO2, Ni2P, and Ni foam at 300 mV. The Inset isthe plot of the current density in logarithmic scale. The error bars represent the range of the current density values from three independent measurements.(C) Nyquist plots of different oxygen evolution electrodes at the applied 300 mV overpotential. Inset is simplified Randle circuit model. All measurementswere performed in 1 M KOH electrolyte.

5610 | www.pnas.org/cgi/doi/10.1073/pnas.1701562114 Zhou et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020

catalytic activity was evaluated using linear sweep voltammetry with a sweeprate of 2 mV/s, while the stability of the catalysts was studied by chro-noamperometry testing and cyclic voltammetry (CV) with a sweep rate of50 mV/s. The electrochemical properties were studied normally after theactivation by 50 CV cycles. Electrochemical impedance spectroscopy wasmeasured at an overpotential of 300 mV from 0.1 Hz to 100 KHz with anamplitude of 10 mV. All of the measured potentials vs. the Hg/HgO wereconverted to an RHE by the Nernst equation (ERHE = EHg/HgO + 0.0591 pH +0.098). The equilibrium potential (E0) for OER is 1.23 V vs. RHE, so the po-tential difference between ERHE and 1.23 V is the overpotential.

Raman Measurements. The Raman spectra were carried out under a RenishawinVia Raman Spectroscope with He–Ne laser at 633 nm as an excitation

source. In general, the acquisition time was 20 s and we did three accumu-lations. Before measurements, the spectrometer was calibrated using theRaman peak of silicon at 520 cm−1. To avoid oxidation or any structuralchanges of the samples due to laser irradiation, the laser power was setaround 0.2 mW during measurements.

ACKNOWLEDGMENTS. This project was supported by the US DefenseThreat Reduction Agency under Grant FA7000-13-1-0001 and the USDepartment of Energy under Contract DE-SC0010831, as well as by US AirForce Office of Scientific Research Grant FA9550-15-1-0236, the T. L. L.Temple Foundation, the John J. and Rebecca Moores Endowment, and theState of Texas through the Texas Center for Superconductivity at theUniversity of Houston.

1. Lewis NS, Nocera DG (2006) Powering the planet: Chemical challenges in solar energyutilization. Proc Natl Acad Sci USA 103:15729–15735.

2. Dresselhaus MS, Thomas IL (2001) Alternative energy technologies. Nature 414:332–337.3. Gray HB (2009) Powering the planet with solar fuel. Nat Chem 1:7.4. Walter MG, et al. (2010) Solar water splitting cells. Chem Rev 110:6446–6473.5. Gao S, et al. (2016) Partially oxidized atomic cobalt layers for carbon dioxide elec-

troreduction to liquid fuel. Nature 529:68–71.6. Liang Z, Lu YC (2016) Critical role of redox mediator in suppressing charging insta-

bilities of lithium-oxygen batteries. J Am Chem Soc 138:7574–7583.7. Carmo M, Fritz DL, Merge J, Stolten D (2013) A comprehensive review on PEM water

electrolysis. Int J Hydrogen Energy 38:4901–4934.8. Smith RDL, et al. (2013) Photochemical route for accessing amorphous metal oxide

materials for water oxidation catalysis. Science 340:60–63.9. Lu X, Zhao C (2015) Electrodeposition of hierarchically structured three-dimensional

nickel-iron electrodes for efficient oxygen evolution at high current densities. NatCommun 6:6616.

10. Louie MW, Bell AT (2013) An investigation of thin-film Ni-Fe oxide catalysts for theelectrochemical evolution of oxygen. J Am Chem Soc 135:12329–12337.

11. Xu X, Song F, Hu X (2016) A nickel iron diselenide-derived efficient oxygen-evolutioncatalyst. Nat Commun 7:12324.

12. Gong M, et al. (2013) An advanced Ni-Fe layered double hydroxide electrocatalyst forwater oxidation. J Am Chem Soc 135:8452–8455.

13. Long X, et al. (2014) A strongly coupled graphene and FeNi double hydroxide hybridas an excellent electrocatalyst for the oxygen evolution reaction. Angew Chem Int EdEngl 53:7584–7588.

14. Feng JX, et al. (2016) FeOOH/Co/FeOOH hybrid nanotube arrays as high-performanceelectrocatalysts for the oxygen evolution reaction. Angew Chem Int Ed Engl 55:3694–3698.

15. Zhang B, et al. (2016) Homogeneously dispersed multimetal oxygen-evolving cata-lysts. Science 352:333–337.

16. Yang Y, Fei H, Ruan G, Tour JM (2015) Porous cobalt-based thin film as a bifunctionalcatalyst for hydrogen generation and oxygen generation. Adv Mater 27:3175–3180.

17. Cobo S, et al. (2012) A Janus cobalt-based catalytic material for electro-splitting ofwater. Nat Mater 11:802–807.

18. Ahn HS, Tilley TD (2013) Electrocatalytic water oxidation at neutral pH by a nano-structured Co(PO3)2 anode. Adv Funct Mater 23:227–233.

19. Stern LA, Feng LG, Song F, Hu XL (2015) Ni2P as a Janus catalyst for water splitting:The oxygen evolution activity of Ni2P nanoparticles. Energy Environ Sci 8:2347–2351.

20. Wang XG, Li W, Xiong DH, Petrovykh DY, Liu LF (2016) Bifunctional nickel phosphidenanocatalysts supported on carbon fiber paper for highly efficient and stable overallwater splitting. Adv Funct Mater 26:4067–4077.

21. Zhao SL, et al. (2016) Ultrathin metal-organic framework nanosheets for electro-

catalytic oxygen evolution. Nat Energy 1:16184.22. Li JC, et al. (2016) A 3D bi-functional porous N-doped carbon microtube sponge

electrocatalyst for oxygen reduction and oxygen evolution reactions. Energy Environ

Sci 9:3079–3084.23. Han L, Dong S, Wang E (2016) Transition-metal (Co, Ni, and Fe)-based electrocatalysts

for the water oxidation reaction. Adv Mater 28:9266–9291.24. Chen Z, et al. (2011) Three-dimensional flexible and conductive interconnected gra-

phene networks grown by chemical vapour deposition. Nat Mater 10:424–428.25. Zhou H, et al. (2016) Efficient hydrogen evolution by ternary molybdenum sulfose-

lenide particles on self-standing porous nickel diselenide foam. Nat Commun 7:12765.26. Zhang LY, Brow RK (2011) A Raman study of iron-phosphate crystalline compounds

and glasses. J Am Ceram Soc 94:3123–3130.27. Chemelewski WD, Lee HC, Lin JF, Bard AJ, Mullins CB (2014) Amorphous FeOOH ox-

ygen evolution reaction catalyst for photoelectrochemical water splitting. J Am Chem

Soc 136:2843–2850.28. Tong YY, et al. (2015) Three-dimensional astrocyte-network Ni-P-O compound with

superior electrocatalytic activity and stability for methanol oxidation in alkaline en-

vironments. J Mater Chem A Mater Energy Sustain 3:4669–4678.29. Liu JQ, Zheng MB, Shi XQ, Zeng HB, Xia H (2016) Amorphous FeOOH quantum dots

assembled mesoporous film anchored on graphene nanosheets with superior elec-

trochemical performance for supercapacitors. Adv Funct Mater 26:919–930.30. Feng JX, Ye SH, Xu H, Tong YX, Li GR (2016) Design and synthesis of FeOOH/CeO2

heterolayered nanotube electrocatalysts for the oxygen evolution reaction. Adv

Mater 28:4698–4703.31. Chen JYC, et al. (2015) Operando analysis of NiFe and Fe oxyhydroxide electro-

catalysts for water oxidation: Detection of Fe4+ by Mössbauer spectroscopy. J Am

Chem Soc 137:15090–15093.32. Trotochaud L, Young SL, Ranney JK, Boettcher SW (2014) Nickel-iron oxyhydroxide

oxygen-evolution electrocatalysts: The role of intentional and incidental iron in-

corporation. J Am Chem Soc 136:6744–6753.33. Cabán-Acevedo M, et al. (2015) Efficient hydrogen evolution catalysis using ternary

pyrite-type cobalt phosphosulphide. Nat Mater 14:1245–1251.34. Zhou H, et al. (2016) Highly efficient hydrogen evolution from edge-orientedWS2(1–x)Se2x

particles on three-dimensional porous NiSe2 foam. Nano Lett 16:7604–7609.35. Song F, Hu X (2014) Exfoliation of layered double hydroxides for enhanced oxygen

evolution catalysis. Nat Commun 5:4477.36. Ma W, et al. (2015) A superlattice of alternately stacked Ni-Fe hydroxide nanosheets

and graphene for efficient splitting of water. ACS Nano 9:1977–1984.

Zhou et al. PNAS | May 30, 2017 | vol. 114 | no. 22 | 5611

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020