CORE FROM ORGANOMETALLIC PRECURSORS FOR OXYGEN EVOLUTION AND

HYDROGEN EVOLUTION CATALYSIS

FIELD OF INVENTION

[0001] Catalysts for electrocatalytic oxygen and/or hydrogen evolution, and, more particularly, for catalysts utilized for photoelectrochemical or photocatalytic water splitting.

GOVERNMENT INTEREST

[0002] This invention was made with Government support under Grant No. CHE 1411495 and Grant No. DGE1450681 awarded by the National Science Foundation. The Government has certain rights in the invention.

CROSS REFERENCE TO RELATED APPLICATIONS

[0003] This application is related to provisional United States Patent Application Serial No. 62/474,346, filed March 21, 2017, entitled "Thin Films Of Metal Phosphides On Semiconductors From Organometallic Precursors For Oxygen Evolution And Hydrogen Evolution Catalysis," and United States Patent Application Serial No. 62/527394, filed June 30, 2017, entitled "Single- Source Precursors For The Chemical Vapor Deposition Growth Of Metal Phosphide Electrocatalysts On Conductive Substrates And Methods To Make And Use Thereof." These provisional patent applications (and the attachments thereto) are incorporated herein in their entirety for all purposes.

BACKGROUND OF INVENTION

[0004] Solar-driven water splitting, which cleanly converts solar energy to chemical energy in the form of hydrogen, holds the promise of meeting the world's clean energy demands. [Armstrong 2016]. There are three approaches to use solar energy for the production of fuels from water: photovoltaic-electric (PV-E), photoelectrochemical (PEC), and photocatalytic water splitting [Shaner 2016], although many semiconductors alone can in principle perform PEC or photocatalytic water splitting. [J. Li 2015]. It is well understood now that semiconductors should be integrated with hydrogen and oxygen evolution catalysts to achieve stable and efficient solar water splitting. [J. Yang 2013]. As such, the development of highly active catalysts to pair with semiconductors is essential to achieving efficient solar water splitting. Recent years have witnessed great progress in the development of water-splitting electrocatalysts composed of earth- abundant elements[g. Li 2014; Sun 2016], predominantly for the oxygen evolution reaction (OER) as it is more kinetically demanding than the hydrogen evolution reaction (HER). [Xiang 2016; Meng 2014] (Water-splitting is the conversion of water to hydrogen and oxygen. It includes the two-half reactions of OER and HER.)

[0005] Transition metal phosphides (TMPs) have gained eminence as the next-generation of water splitting catalysts with high activity demonstrated for both HER and OER. [Kibsgaard 2015; Xu 2015; Callejas 2016; Shi 2016]. Binary phosphides such as Ni ₂ P [Stem 2015], FeP [X. Yang 2015; Z Zhang 2014; Jiang 2014], CoP [Zh 2015; Tang 2016; X. Yu 2016], WP and MoP [X. Wang 2016], and recently ternary bimetallic phosphides such as CoMnP [D. Li 2016], (Coo.52Feo. ₈ )2P [Y. Tan 2016], MoWP [X. Wang 2016], and NiCoP [J. Yu 2016; C. Wang 2016] have shown higher activities than transition metal chalcogenides and double-metal hydroxides with ternary TMPs exhibiting lower overpotentials than binary TMPs. To date, there has been no report of the integration of TMPs with a semiconductor light absorber for PEC-OER. Previous reports are mainly focused on the integration of metal oxide/hydroxide OER catalysts with a semiconductor photoanode [Ran 2014; Zhong 2010; Xu 2015; Sun 2012]. In these reports, the performance is unsatisfactory because of the relatively low OER catalytic activity, and the poor electrical conductivity of metal oxide/hydroxide limits the charge transfer between the semiconductor and the OER catalyst, resulting in high interfacial recombination of photocarriers. TMPs typically have a metallic nature and the introduction of TMPs to the surface of semiconductors should promote the interfacial charge transfer and charge transportation from the semiconductor to the active surface, which leads to high PEC performance. However, current synthetic methods are not suitable for pairing a TMP with a semiconductor.

[0006] Though wind and solar are the fastest growing power sources in the world, a significant fraction of the power generated from those means is lost to resistive heating during power transmission from remote locations. A strategy to lose less of this energy is being pursued by Shell and the National Renewable Energy Laboratories wherein wind and solar energy is converted through electrolysis on site to chemical energy in the form of hydrogen gas. The hydrogen gas can be used as a chemical form of shippable, storable energy which can be piped to city centers with minimal loss. The hydrogen gas can then be converted to electricity with fuel cells producing water as the byproduct. Alternatively, the hydrogen gas can be used in industrial processes. The viability of the electrolysis process is contingent upon the development of efficient electrocatalysts.

[0007] The electrolysis of water into hydrogen fuel and oxygen offers a convenient route to store solar and wind energy chemically, an ideal solution for channeling off-peak power production and minimizing energy losses incurred in power transmission from often remote locations. The utilization of active, stable electrocatalysts with lower overpotentials will increase efficiency and stability, enabling commercial realization. Over the past several decades, tremendous progress has been made in the development of highly active catalysts composed of earth-abundant elements including transition metal phosphides (TMPs), oxides/hydroxides, carbides, nitrides, and chalcogenides as well as carbon-based nanomaterials for OER and HER. Catalysts active towards both reactions in the same electrolyte are logistically preferred. However, the vast majority of existing catalysts are unsuitable for use in the same electrolyte due to the mismatch of pH ranges in which the electrocatalysts are both stable and sufficiently active.

[0008] Since the first report of TMPs as electrocatalysts for water splitting with Ni ₂ P in 2013, TMPs have emerged as premier electrocatalysts for OER, HER, and in some cases overall water splitting. For the TMPs that demonstrate overall water splitting, the bifunctionality arises from the respective TMP's ability to catalyze the HER and serve as a precatalyst for the OER; indeed, TMPs are known to oxidize rapidly to highly active metal oxyhydroxides at their surface upon OER operation, yet retain the conductive TMP core. With several examples of bifunctional TMPs, there is now an impetus for the development of new strategies to fabricate these high-performance materials. The current preparation methods for TMPs can be grouped into four main routes: thermal phosphidation of films or nanostructured bulk alloys, electrochemical deposition, metallurgical synthesis, and solvothermal methods. These methods require harsh conditions with toxic, gaseous chemicals, and offer little control over the metal-phosphorus stoichiometry, phase purity, and conductivity. For example, while (Coo.52Feo. ₄ 8) ₂ P can be used to obtain the state-of-the- art cell voltage of 1.53 V when used as both anode and cathode, the material fabrication process, which involves a combination of arc-melting Co ₂ P, Co, and Fe followed by selective electrochemical etching, is not amenable to scaled-up production.

[0009] Many groups around the world have proven metal phosphides to be most promising electrocatalysts for water splitting. It should be noted that the metal phosphides perform better than the respective free metals toward both OER and HER, largely due to their superior corrosion resistance and the nanostructured morphologies their surfaces naturally adopt. The corrosion resistance and high, stable surface areas make them suitable candidates for commercial electrolyzers.

[0010] Electrodeposition processes are non-ideal for use in electrolyzers because the metal phosphides synthesized this way are typically spongy, exhibit phase inhomogeneity, cannot guarantee full surface coverage, and have poor surface adhesion.

[0011] Some substrates, such as nickel foam, can have their surfaces converted to metal phosphides by phosphidation (i.e., heating the substrate to high temperature and treating with phosphine gas or an alkylphosphine). But, these methods cannot guarantee phase purity as there should be a phosphorus gradient between the pristine metal and the surface post-phosphidation. [0012] Other techniques include those employed in which of metal is deposited onto the conductive substrate by sputtering (expensive) and then the surface phosphidized in vacuum at relatively high temperatures. Importantly, such method does not place the metal phosphide directly on the conductive substrate. Rather, a thick layer of the given metal exists between the surface and the conductive substrate. Moreover, as with the simple phosphidation method, there is a phosphorus gradient from bulk to surface.

[0013] Phosphidation is also performed on metal oxides, but this leads to materials with a metal oxide core after phosphidation leading to poor electrical conductivity, or materials with a low surface area which is determined by the morphology of the substrate before phosphidation. Some of them are prepared in the form of nanoparticles which are not amenable to the fabrication of binder-free electrode. As such the cost will increase and the performance will deteriorate.

[0014] Accordingly, the need remains for improved catalysts that can be utilized for electrochemical, photoelectrochemical or photocatalytic water splitting

SUMMARY OF INVENTION

[0015] It has been discovered that designer organometallic precursors can be converted to specific metal phosphide materials by metal-organic chemical vapor deposition (MOCVD) under mild conditions, a strategy that is ideally suited to grow specific TMPs directly on semiconductors without compromising the substrate. Moreover, because the stoichiometry is built into the precursor, the method can scale without the inhomogeneity of phases expected from existing approaches.

[0016] A new ternary TMP catalyst, FeMnP, has been discovered synthesized via MOCVD from the single-source precursor FeMn(CO) ₈ ^-PH ₂ ). [See Colson 2011]. Using a single-source precursor to prepare a metal-phosphide by chemical vapor deposition for electrocatalysis is new, as well as using a single-source precursor to coat a semiconductor by chemical vapor deposition. Metal-organic chemical vapor deposition using single-source organometallic precursors is referred to herein as the "SSP-MOCVD" method.

[0017] FeMnP was deposited on a three-dimensional Ti0 ₂ nanorod array photoanode to form a Ti0 ₂ /FeMnP core/shell nanostmcture exhibiting dramatically enhanced PEC performance. The high PEC performance is ascribed to the high OER catalytic activity, high conductivity of FeMnP, and the excellent interface with Ti0 ₂ , providing rapid charge transfer and separation without creating non-radiative recombination centers. As a result, the theoretical maximum photocurrent for Ti0 ₂ is achieved. [J. Li 2015]. Furthermore, the method can be considered general; other precursors sourced from the vast catalog of known main group element-containing homo- and heterometallic carbonyl clusters can be used to deposit specific electrocatalysts onto semiconductors. [See Whitmire 1998]. This work opens up a new avenue of device and material engineering for highly efficient solar water splitting.

[0018] The SSP-MOCVD method was used to grow FeMnP on Ti0 ₂ -nanorod arrays for photoelectrochemical OER, ultimately finding that the FeMnP/Ti0 ₂ core-shell structure could be stably operated at the theoretical photocurrent density of Ti0 ₂ .

[0019] A variety of catalysts have been developed for electrocatalytic oxygen evolution, but very few of them can be readily integrated with semiconducting light absorbers for photoelectrochemical or photocatalytic water splitting. The present invention is an efficient core/shell photoanode with a highly active oxygen evolution electrocatalyst shell (FeMnP) and semiconductor core (rutile T1O2) for the photoelectrochemical oxygen evolution reaction (PEC- OER). Metal-organic chemical vapor deposition (MOCVD) from a single-source precursor can be used to ensure good contact between the FeMnP and the Ti0 ₂ . The Ti0 ₂ /FeMnP core/shell photoanode reaches the theoretical photocurrent density for rutile Ti0 ₂ of 1.8 mA cm ^"2 at 1.23 V vs RHE under simulated 100 mW cm ^"2 (1 sun) irradiation. The dramatic enhancement is a result of the synergistic effects of the high OER activity of FeMnP (delivering an overpotential of 300 mV with a Tafel slope of 65 mV dec ^"1 in 1 M KOH) and the conductive interlayer between the surface active sites and semiconductor core which boosts the interfacial charge transfer and photocarrier collection. The facile fabrication of the Ti0 ₂ /FeMnP core/shell nanorod array photoanode offers a new strategy for preparing highly efficient photoelectrochemical solar energy conversion devices.

[0020] Furthermore, the uniform metal phosphide layer can be grown across a 3 -dimensional substrate, which was previously unachievable in the prior art.

[0021] The present invention can be used to produce hydrogen or oxygen from sunlight. Hydrogen produced this way can be used as a clean fuel for locomotion, a feedstock for industrial chemicals, or to generate power (and pure water). This present invention can also be used in other catalytic applications such that sunlight provides the driving energy for the reaction.

[0022] The SSP-MOCVD method can be used to deposit nanostructured ternary and bimetallic FeMnP from the volatile precursor FeMn(CO) ₈ ^-PH ₂ ) onto graphene-protected nickel foam (GNF) for electrochemical evaluation. FeMnP exhibits superior electrocatalytic activity as a bifunctional catalyst for efficient and stable overall water splitting.

[0023] The SSP-MOCVD method can be used to deposit nanostructured FeP from the volatile precursor Fe(CO) ₄ PH ₂ ^t Bu or [{(CO)4Fe}P(H) ^t Bu] ₂ onto FTO for electrochemical evaluation. FeP exhibits electrocatalytic activity for HER.

[0024] The SSP-MOCVD method can be used to deposit nanostructured Fe ₂ P from the volatile precursor Fe(CO) ₄ PH3 onto FTO for electrochemical evaluation. Fe ₂ P exhibits good electrocatalytic activity for HER.

[0025] The SSP-MOCVD method can be used to deposit nanostructured Fe ₃ P from the volatile precursor H ₂ Fe ₃ (CO)9P ^t Bu onto FTO for electrochemical evaluation. Fe ₂ P exhibits very good electrocatalytic activity for HER.

[0026] The present invention further includes a process by which transition metal phosphides can be grown directly on conductive substrates, the resulting material from which can serve as electrocatalysts for the oxygen evolution reaction (OER) and/or hydrogen evolution reaction (HER) of water electrolysis.

[0027] The catalyst preparation of the present invention (growing metal phosphide films on conductive substrates by the single-source MOCVD process) can be utilized in this.

[0028] In general, in one embodiment, the invention features a method to make a device that includes the step of preparing a semiconductor substrate. The method further includes introducing the substrate and suitable single-source precursor into a chemical-vapor deposition apparatus. The method further includes applying a vacuum while heating, such as at a temperature of 350 °C (or greater). The method further includes conducting water electrolysis to produce oxygen and/or hydrogen with and/or without light irradiation.

[0029] In general, in one embodiment, the invention features a method to adjust the composition of the metal phosphide thin film by co-decomposing similar precursors with different transition metals instead of a single precursor.

[0030] In general, in one embodiment, the invention features a method to dope the metal phosphide thin film with other main group elements by co-decomposing similar precursors with different transition metals and/or main group elements.

[0031] In general, in one embodiment, the invention could be applied to other semiconducting materials besides titanium dioxide.

[0032] In general, in one embodiment, the titanium dioxide could be modified to adjust the band gap prior to metal phosphide thin film deposition.

[0033] In general, in one embodiment, the semiconductor array can be altered by derivatization with conductive layers such as carbon nanotubes, reduced graphene oxide, graphene nanoribbons, etc.

[0034] In general, in one embodiment, a semiconductor nanorod array is grown on fluorine-doped tin oxide (FTO) glass. [0035] In general, in one embodiment, a semiconductor nanorod array is grown on other transparent conducting substrates.

[0036] Implementations of the invention can include one or more of the following features: (1) a precursor or blend of precursors used as feedstock for chemical vapor deposition, (2) different decomposition temperatures can be employed, (3) different annealing periods of the device post- deposition can be used, and (4) metal phosphide thin film thicknesses can be controlled by varying the amount of precursor employed in the deposition.

[0037] Implementations of the invention can include one or more of the following features: (1) the operating voltage can be varied, (2) the wavelength of radiation can be varied, and (3) the intensity of the irradiation can be varied.

[0038] In general, in one embodiment, the invention features a device that includes a semiconductor core and a transition metal phosphide shell deposited upon the semiconductor core to form a core/shell structure. The core/shell structure is operable for use in a photoelectrochemical or photocatalytic water splitting process.

[0039] In general, in another embodiment, the invention features a method that includes selecting a single source organometallic precursor. The method further includes utilizing the single source organometallic precursor as the chemical feedstock in a deposition process to deposit a transition metal phosphide uniformly on a semiconductor core to form a core/shell structure.

[0040] Implementations of the device and method of the invention can include one or more of the following features:

[0041] The semiconductor core can include a semiconductor material selected from the group consisting of heterometallic bismuth oxides, metal chalcogenides, tungsten oxides, titanium dioxide, copper oxide, and combinations thereof.

[0042] The semiconductor core can include a derivatized semiconductor material, wherein the derivatized semiconductor material comprises a conductive layer selected from the group consisting of carbon nanotubes, graphene oxide, and graphene nanoribbons.

[0043] The transition metal phosphide shell can include a transition metal phosphide that is selected from a group consisting of FeMnP, CoMnP, CoFeP, NiMnP, NiFeP, and WMnP

[0044] The transition metal phosphide shell can include a transition metal phosphide that is

FeMnP.

[0045] The semiconductor core can include Ti0 ₂ .

[0046] The semiconductor core can include Ti0 ₂ nanorod arrays. The core/shell structure can be a Ti0 ₂ FeMnP core/shell nanorod array.

[0047] The semiconductor core can be a semiconductor nanorod array.

[0048] The single source organometallic precursor can be H2Fe3(CO)9^-P-t-Bu).

[0049] The single source organometallic precursor can be

[0050] The single source organometallic precursor can be Fe(CO)4PH ₃ .

[0051] The single source organometallic precursor can be FeiCO^PE^Bu.

[0052] The single source organometallic precursor can be [{(CO) ₄ Fe}P(H) ^t Bu] ₂ .

[0053] The single source organometallic precursor can be Co ₃ (CO)9^-P-t-Bu).

[0054] The single organometallic precursor can be H2Fe ₃ (CO)9E in which E is S, Se, or Te.

[0055] The semiconductor core can include Ti0 ₂ nanorod arrays. The core/shell structure can be a Ti0 ₂ FeMnP core/shell nanorod array.

[0056] The deposition process can be a metal-organic chemical vapor deposition process.

[0057] The semiconductor core can include a semiconductor nanorod array. The method can further include growing the semiconductor array on a transparent conducting substrate.

[0058] The transparent conducting substrate can be a fluorine-doped tin oxide glass.

[0059] The method can further include annealing the core/shell structure.

[0060] The semiconductor core can include a coating of graphene.

[0061] The semiconductor core can include nickel. [0062] The semiconductor core can be multilayer graphene protected nickel.

[0063] The semiconductor core can include a nickel foam.

[0064] The nickel foam can be a multilayer graphene protected nickel foam.

[0065] The semiconductor core can include a metal foam.

[0066] The metal foam can be a multilayer graphene protected metallic foam.

[0067] The single source organometallic precursor can be FeMn(CO)8^-PH2).

[0068] The single source organometallic precursor can be Fe(CO) ₄ PH2 ^t Bu or

[{(03) ₄ Ρε}Ρ(Η) ^ι Βυ] ₂ with nanostructured FeP deposited on the semiconductor core.

[0069] The single source organometallic precursor can be Fe(CO) ₄ PH ₃ with nanostructured Fe ₂ P deposited on the semiconductor core.

[0070] The single source organometallic precursor is F^FesiCO^P^u with nanostructured Fe ₃ P deposited on the semiconductor core.

[0071] The semiconducting core can include fluorine-doped tin oxide.

[0072] The core/shell structure can be operable for use in a photoelectrochemical or photocatalytic water splitting process.

[0073] In general, in another embodiment, the invention features a device made by at least one of the above-described methods.

[0074] In general, in another embodiment, the invention features a method that includes selecting a device from the above-described devices. The method further includes utilizing the device in a photoelectrochemical or photocatalytic water splitting process.

[0075] Implementations of the invention can include one or more of the following features:

[0076] The device can be utilized in a photoelectrochemical water splitting process.

[0077] The photoelectrochemical water splitting process can include an oxygen evolution reaction. The device can be a catalyst in the oxygen evolution reaction.

[0078] The device can have a core/shell structure that includes FeP. The FeP can have an electrocatalytically active surface area of at least about 18 mF/cm ² and a Faradaic efficiency greater than 95%.

[0079] The device can have a core/shell structure comprising Fe ₂ P. The Fe ₂ P can have an electrocatalytically active surface area of at least about 19 mF/cm ² and a Faradaic efficiency greater than 95%.

[0080] The device can have a core/shell structure comprising Fe ₃ P. The Fe ₃ P can have an electrocatalytically active surface area of at least about 23 mF/cm ² and a Faradaic efficiency greater than 95%.

[0081] In general, in another embodiment, the invention features a photoanode that includes a Ti0 ₂ nanorod array core and a FeMnP shell coating the Ti0 ₂ nanorod array core. The photoanode is operable for photoelectrochemical oxygen evolution.

[0082] Implementations of the invention can include one or more of the following features:

[0083] The ratio of (i) the highest photoconversion efficiency of a photoanode comprising the Ti0 ₂ nanorod array core without the FeMnP shell to (ii) the photoconversion efficiency of the photoanode comprising the Ti0 ₂ nanorod array core and FeMnP shell at the same potential is at most 1:2.

[0084] The ratio can be at most 1:2.5.

[0085] In general, in another embodiment, the invention features a composition that is {Fe(CO) ₄ P(H)¾u} ₂ .

[0086] In general, in another embodiment, the invention features a method that includes selecting {Fe(CO) ₄ P(H) ⁱ Bu} ₂ The method further includes performing a Metal-organic chemical vapor deposition process using the {Fe(CO) ₄ P(H)¾u} ₂ as a single-source organometallic precursor.

[0087] The foregoing has outlined rather broadly the features and technical advantages of the invention in order that the detailed description of the invention that follows may be better understood. Additional features and advantages of the invention will be described hereinafter that form the subject of the claims of the invention. It should be appreciated by those skilled in the art that the conception and the specific embodiments disclosed may be readily utilized as a basis for modifying or designing other structures for carrying out the same purposes of the invention. It should also be realized by those skilled in the art that such equivalent constructions do not depart from the spirit and scope of the invention as set forth in the appended claims.

[0088] It is also to be understood that the invention is not limited in its application to the details of construction and to the arrangements of the components set forth in the following description or illustrated in the drawings. The invention is capable of other embodiments and of being practiced and carried out in various ways. Also, it is to be understood that the phraseology and terminology employed herein are for the purpose of the description and should not be regarded as limiting.

BRIEF DESCRIPTION OF THE DRAWINGS

[0089] FIGS. 1A-1E show structural characterization of a Ti0 ₂ FeMnP core/shell nanorod array.

[0090] FIGS. 1A-1B are SEM images (at different magnifications) of the Ti0 ₂ /FeMnP core/shell nanorod array.

[0091] FIG. 1C is an HRTEM image showing the thickness of the FeMnP shell and the crystalline lattice on Ti0 ₂ .

[0092] FIG. ID is a TEM image showing the uniform coating of FeMnP on a Ti0 ₂ nanorod.

[0093] FIG. IE is an EDS mapping over a single T1O2 nanorod showing the distribution of Fe, Mn, and P.

[0094] FIGS. 2A-2C are graphs showing chemical composition and states of the FeMnP shell in the Ti0 ₂ /FeMnP core/shell nanostructure showing the XPS spectra for: Fe 2p ₃ / ₂ (FIG. 2A), Mn 2pv2 (FIG. 2B), and P 2p (FIG. 2C) orbitals. In FIG. 2A, curves 201-203 are, respectfully, (201) surface, (202) 3 min sputter, and (203) 6 min sputter. In FIG. 2B, curves 204-206 are, respectfully, (204) surface, (205) 3 min sputter, and (206) 6 min sputter. In FIG. 2C, curves 207-209 are, respectfully, (207) surface, (208) 3 min sputter, and (209) 6 min sputter.

[0095] FIGS. 3A-3D show the morphology of FeMnP anode on FTO and its OER characterizations.

[0096] FIG. 3A is an SEM image of FeMnP on FTO.

[0097] FIG. 3B is an HRTEM image of FeMnP on FTO.

[0098] FIG. 3C is a graph showing polarization curves of FeMnP/FTO and bare FTO electrodes in 1.0 M KOH at 10 mV s" ¹ . Insert 303 is a graph referring to the corresponding Tafel plot of FeMnP/FTO.

[0099] FIG. 3D is a graph showing the CV curves of FeMnP/FTO electrode for the 1 ^st cycle and the 1000 ^th cycle at scan rate of 100 mV sr ¹ .

[0100] FIGS. 4A-4D show PEC-OER characterization.

[0101] FIG. 4A is a graph showing J-V curves of Ti0 ₂ nanorod and Ti0 ₂ /FeMnP core/shell nanorod photoanodes in the dark and under irradiation. Curves 401-404 are, respectively, for (401) Ti0 ₂ in the dark, (402) Ti0 ₂ with light, (403) Ti0 ₂ /FeMnP in the dark, and (404) Ti0 ₂ /FeMnP with light.

[0102] FIG. 4B is a graph showing photoconversion efficiency as a function of applied voltage. Curves 405-406 are, respectively, for (405) Ti0 ₂ , and (406) Ti0 ₂ /FeMnP.

[0103] FIG. 4C is a graph showing normalized plots of the photocurrent-time dependence for the Ti02 nanorod and Ti02/FeMnP core/shell nanorod. Curves 407-408 are, respectively, for (407) Ti0 ₂ , and (408) Ti0 ₂ /FeMnP. Insert 409 is a graph referring to the on-off J-t curves of the Ti0 ₂ and Ti0 ₂ /FeMnP core/shell nanorod at 1.23 V vs RHE.

D = (It - Isteady) I (linitiai - Isteady), With

It, linitiai, and Isteady are the time-dependent, the initial photocurrent, and the steady state photocurrent, respectively. Curves 410-411 are, respectively, for (410) Ti0 ₂ , and (411) Ti0 ₂ /FeMnP. [0104] FIG. 4D is a graph showing photochemical stability of the Ti02/FeMnP core/shell nanorod photoanode at 1.23 V vs RHE.

[0105] FIG. 5A is a graph showing Mott-Schottky plots of Ti0 ₂ nanorod and Ti0 ₂ /FeMnP core/shell nanorod photoanodes in the dark. White square 501 are for Ti0 ₂ 1962 Hz. Black squares 502 are for Ti0 ₂ 2962 Hz. Circles 503 are for Ti0 ₂ /FeMnP 1962 Hz. Inset 504 shows the Mott-Schottky plot of Mott-Schottky plot for FIG. 5A are for Ti0 ₂ /FeMnP (circles 503) on a different vertical scale.

[0106] FIG. 5B is an illustration showing band alignment of Ti0 ₂ FeMnP core/shell nanorod structure. EF-FTO and EF-FeMnP refer to the Fermi level of FTO and FeMnP.

[0107] FIG. 5C is an illustration showing core/shell structure 516 having a thin film of FeMnP 511 deposited from a single-source precursor FeMn(CO) ₈ ^-PH ₂ ) onto Ti0 ₂ core 511 having an active surface 512 of (Mni- _x Fe _x )OOH. When exposed to sunlight 513, core/shell structure 516 was found to be highly active for the photoelectrochemical oxygen evolution reaction (PEC-OER) 514.

[0108] FIG. 6 is an illustration of an embodiment of an MOCVD apparatus employed.

[0109] FIG. 7 is an SEM image of bare T1O2 nanorod array on FTO substrate.

[0110] FIG.8 is a graph showing XRD of TiO^eMnP core/shell nanorod array on FTO substrate (curve 805). (Square 802 and circles 803 indicates FeMnP and Ti0 ₂ , respectively) For comparison, the spectra of bare Ti0 ₂ nanorod array on FTO (curve 803) and bare FTO (curve 804) are also illustrated on this graph.

[0111] FIG. 9 is a graph showing XRD of FeMnP on FTO substrate. Curve 901 is for FTO, and curve 902 is for FeMnP-FTO.

[0112] FIG. 10 is a graph showing a Nyquist plot (plot 1001) of FeMnP FTO electrode at overpotential of 300 mV in 1.0 M KOH. Insert 1002 refers to the corresponding equivalent circuit having the equivalent series resistance (Rs) 1003 from the leads, solution resistance and wires, charge transfer resistance (Rt) 1004, and the double-layer capacitance (Ct) 1005. [0113] FIGS. 11A-11D show measurements of the electrochemically active surface area (ECSA).

[0114] FIGS. 11A-11B are graphs showing the CV scans of (FIG. 11 A) FTO and (FIG. 11B) FeMnP/FTO at different scan rates in 1.0 M KOH in the non-Faradaic potential range. Scan rates are (a) 10 mV s ^"1 (curves 1101 and 1106), (b) 30 mV s ^"1 (curve 1102 and 1107), (c) 50 mV s ^"1 (curve 1103 and 1108), (d) 70 mV s ^"1 (curve 1104 and 1109), and (a) 90 mV s ^"1 (curve 1105 and 1109) for FIGS. 11A-11B, respectively.

[0115] FIGS. 11C-11D are graphs that show the corresponding current density difference between the anodic current and the cathodic current at 0.675 V vs RHE as a function of scan rate for the (FIG. 11C) FTO and (FIG. 11D) FeMnP/FTO electrodes, respectively. The slope of the linear part is the electrochemical double-layer capacitance, which has a positive linear relationship with the ECSA.

[0116] FIG. 12 is a graph showing the produced H ₂ and 0 ₂ signal recorded by GC after 60 min electrolysis. Curve 1201 is before electrolysis. Curve 1202 and 1203 are, respectively, Pt and FeMnP.

[0117] FIG. 13A is a graph showing the UV-VIS spectra of Ti0 ₂ nanorod array on FTO glass showing the light absorption edge of 400 nm.

[0118] FIG. 13B is a graph showing the corresponding Tau plot of Ti0 ₂ giving the band gap of 3.0 eV.

[0119] FIG. 14 is a graph showing the transmittance of Ti0 ₂ (curve 1401) and Ti0 ₂ FeMnP (curve 1402).

[0120] FIG. 15A-15C are graphs showing Sputtering assisted XPS analysis of the chemical state of Ti0 ₂ FeMnP core/shell nanorod after stability test with the graphs for: Fe 2^ _{3 2} (FIG. 15A), Mn ¾¾/2 (FIG. 15B), and P 2p (FIG. 15C) orbitals. In FIG. 15A, curves 1501-1503 are, respectfully, (1501) surface, (1502) 3 min sputter, and (1503) 6 min sputter. In FIG. 15B, curves 1504-1506 are, respectfully, (1504) surface, (1505) 3 min sputter, and (1506) 6 min sputter. In FIG. 15C, curves 1507-1509 are, respectfully, (1507) surface, (1508) 3 min sputter, and (1509) 6 min sputter.

[0121] FIG. 16 is a graph that shows the Nyquist plots of the Ti0 ₂ and Ti0 ₂ FeMnP core/shell nanorod photoanodes at 1.23 V vs RHE under light irradiation. Curves 1601-1602 are, respectively, for (1601) T1O2, and (1602) Ti0 ₂ /FeMnP. The curves can also be modeled using the same equivalent circuit shown in insert 1002 in FIG 10. The much smaller radius of the semicircle indicated the significantly decreased charge transfer resistance.

[0122] FIG. 17 is a graph showing the work function of FeMnP/FTO and bare FTO. CPD refers to the contact potential difference between the gold probe and the sample. Square 1701 is for FTO, and circles 1702 are FeMnP/FTO.

[0123] FIGS. 18A-18B are SEM images (at different magnifications) of Ti0 ₂ /FeMnP core/shell nanorod array after stability test.

[0124] FIGS. 19A-19H are SEM images of Ti0 ₂ /FeMnP core/shell nanorod array with different FeMnP loading amount by changing the amount of precursor with (i) 2 mg precursor for FIGS. 19A and 19E, (ii) 5 mg precursor for FIGS. 19B and 19F, (iii) 10 mg precursor for FIGS. 19C and 19G, and (iv) 20 mg precursor for FIGS. 19D and 19H. The scale bar for FIGS. 19A-19D is 1 um. The scale bar for FIGS. 19E-19H is 100 nm.

[0125] FIG. 20 is an illustration of an embodiment of a second MOCVD apparatus employed.

[0126] FIGS.21A-21E are images and illustrations showing morphology characterization. FIGS. 21A-21B are SEM images of FeMnP on NF ad GNF, respectively. FIG. 21C is an HRTEM of FeMnP. FIG. 21D is a selected area electron diffraction (SAED) pattern image of FeMnP. FIG. 21E is an illustration of a Crystalline structure of hexagonal FeMnP (with the spheres illustrating P atoms and the differently shaded polyhedra are statistically mixed Fe and Mn atoms).

[0127] FIGS. 22A-22E are images and graphs showing composition and chemical states characterization of FeMnP. FIG. 22A is a TEM-EDS result of a single FeMnP platelet showing the existence and distribution of Fe (2201), Mn (2202), and P (2203) (with the scale bar being 1 μιη). FIG. 22B is a graph showing EDS composition analysis. FIGS. 22C-2E are graphs of elementary XPS scanning of FeMnP showing the chemical states of Fe, Mn, and P, respectively, before catalysis. In FIG. 22C, curves 2204-206 are, respectfully, (2204) surface, (2205) 3 min sputter, and (2206) 6 min sputter. In FIG.22D, curves 2207-2209 are, respectfully, (2207) surface, (2208) 3 min sputter, and (2209) 6 min sputter. In FIG. 22E, curves 2210-2212 are, respectfully, (2210) surface, (2211) 3 min sputter, and (2212) 6 min sputter.

[0128] FIGS. 23A-23D are graphs showing electrocatalytic activity characterization. FIG. 23A is a graph of the OER polarization curves in 0.1 M KOH at a scan rate of 5 mV s ^"1 . FIG. 23B is a graph of the OER Tafel plots. FIG.23C is a graph of the HER polarization curves in 0.5 M H2SO4 and 0.1 M KOH at a scan rate of 5 mV s ^"1 . FIG. 23D is a graph of the corresponding HER Tafel plots.

[0129] FIGS. 24A-24D are graphs showing stability characterization and overall water splitting. FIG. 24A is a graph of the OER polarization curves of the FeMnP/NF and FeMnP/GNF electrodes in 0.1 M KOH at scan rate of 100 mV s ^"1 , showing the 1 ^st cycle and the 1000 ^th cycle. FIG. 24B is a graph of the HER polarization curves of the FeMnP/NF and FeMnP/GNF electrodes in 0.5 M H2SO4 at scan rate of 100 mV s ^"1 , showing the 1 ^st cycle and the 1000 ^th cycle. FIG. 24C is a graph of the I-V curves of the two-electrode water splitting using FeMnP as bifunctional catalyst in 0.1 M KOH at scan rate of 5 mV s ^"1 . FIG. 24D is a graph showing the long-term stability at a constant cell voltage of 1.60 V for 75 hours using FeMnP/NF and 1.55 V for 75 hours using FeMnP/GNF.

[0130] FIG. 25A is a graph of recorded I-t curves in overall water splitting using two FeMnP/NF electrodes and two FeMnP/GNF electrodes, respectively, at the cell voltage of 1.60 V in 0.1 M KOH solution. Curves 2501-2502 are for FeMnP NF and FeMnP/GNF, respectively.

[0131] FIG. 25B is a graph of the gas amount of 0 ₂ and ¾ versus electrolysis time for the water splitting using two FeMnP/NF electrodes and two FeMnP/GNF electrodes. The Faradaic efficiency was determined by comparing the measured gas amount with the amount of gas calculated using the recorded current. Dark diamonds 2503 are FeMnP/NF measured 0 ₂ ; dark sphere 2504 are FeMnP/NF measured H ₂ ; light diamonds 2505 are FeMnP/GNF measured 0 ₂ ; light sphere 2506 are FeMnP/GNF measured H ₂ ; lines 2507-2510 are FeMnP/NF calculated 0 ₂ , FeMnP/NF calculated H ₂ , FeMnP/GNF calculated 0 ₂ , and FeMnP/GNF calculated H ₂ , respectively.

[0132] FIGS. 26A-26B are images taken under SEI model in WDS characterization of the tested FeMnP.

[0133] FIGS. 26C-26D are images taken under COMPO mode in WDS characterization of the tested FeMnP.

[0134] FIGS. 27A-27D are illustrations and a graph showing coverage-dependent hydrogen binding on (100) and (OOl)-Mn facets of FeMnP. FIGS. 27A-27B depict hydrogen adsorbed on FeMnP (100) at 1 ML and 7/6 ML, respectively. FIGS. 27C-27D depict hydrogen adsorbed on FeMnP (OOl)-Mn surface at 1 ML and 9/8 ML, respectively. FIG. 27E is a graph that shows the calculated differential free binding energy of hydrogen as a function of coverage. ML: monolayer.

[0135] FIGS. 28A-28C are illustrations of the crystal structure of Compound 3. FIG. 28A shows the crystal structure of the major component of Compound 3 (R,R isomer). FIG. 28B shows the crystal structure of the major component of Compound 3 viewed down the P-P bond. FIG. 28C shows the crystal structure of the minor component of Compound 3 viewed down the P-P bond. Thermal ellipsoids are given at 50% probability.

[0136] FIG. 29 is an illustration of an embodiment of a third MOCVD apparatus employed.

[0137] FIGS. 30A-30B are SEM Images of FeP/Fe ₂ P mixture on quartz at two magnifications.

[0138] FIGS. 31A-31D are graphs of XRD, XPS, andXPS-Depth Profile of FeP on Quartz. FIG. 31A is an XRD; FIG.31B is a Surface XPS Spectra; FIG.31C is an XPS Depth Profile; and FIG. 31D is an Interior XPS Spectra. Peaks denoted by * in the XRD pattern arise from the substrate. In FIG. 31C, curves 3101-3103 are for O, P, and Fe, respectively.

[0139] FIGS. 32A-32B are SEM images of FeP from Composition 2 on quartz at two magnifications.

[0140] FIGS. 33A-33B are SEM images of FeP from Composition 3 on quartz at two magnifications.

[0141] FIGS. 34A-34B are SEM images of Fe ₂ P on quartz at two magnifications.

[0142] FIGS.35A-35D are graphs of XRD, XPS, and XPS-Depth Profile of Fe ₂ P on Quartz. FIG. 35A is an XRD; FIG.35B is a Surface XPS Spectra; FIG.35C is an XPS Depth Profile; and FIG. 35D is an Interior XPS Spectra. In FIG.35C, curves 3501-3503 are for O, P, and Fe, respectively.

[0143] FIGS. 36A-36F are SEM images of FeP, Fe ₂ P, and Fe ₃ P on FTO. FIGS. 36A and 36D are of FeP; FIGS. 36B and 36E are of Fe ₂ P; and FIGS. 36C and 36F are of Fe ₃ P.

[0144] FIGS. 37A-37D are graphs showing electrochemical characterization. FIG. 37A shows polarization curves. In FIG. 37A, curves 3701-3705 are for FTO, FeP, Fe ₂ P, Fe ₃ P, and Pt. FIG. 37B shows corresponding Tafel slopes. In FIG. 37B, curves 3706-3709 are for FeP, Fe ₂ P, Fe ₃ P, and Pt. FIG. 37C shows charge transfer Tafel slopes. In FIG. 37C, curves 3710-3712 are for FeP, Fe ₂ P, and Fe ₃ P. FIG.37D shows half of the current density differences as a function of scan rate. The Cdi (the electrochemical double-layer capacitance), is the slope of the fitted lines 3713-3715 for FeP, Fe^, and Fe ₃ P, respectively, in FIG. 37D.

[0145] FIGS. 38A-38C are graphs showing curves recorded in a non-Faradaic reaction potential range of FeP, Fe ₂ P and Fe ₃ P, respectively. In FIGS. 38A-38C, curves 3801, 3806, and 3811 are for 10 mV/s; curves 3802, 3807, and 3812 are for 20 mV/s; curves 3803, 3808, and 3813 are for 30 mV/s; curves 3804, 3809, and 3814 are for 40 mV/s; and curves 3805, 3810, and 3815 are for 50 mV/s.

[0146] FIG.39A is a graph showing time dependence of current density under static overpotential of 120 mV for FeP, Fe ₂ P and Fe ₃ P for the evaluation of the long-term stability. Curves 3901-3903 are for FeP, Fe ₂ P, and Fe ₃ P, respectively.

[0147] FIG. 39B is a graph showing H ₂ gas amount versus electrolysis time. The Faradaic efficiency was determined by comparing measured H ₂ amount to the amount calculated from the current. Squares 3904 are FeP produced; circles 3905 are Fe ₂ P produced; triangles 3906 are Fe ₃ P produced; and lines 3907-3909 are for FeP, Fe ₂ P, and Fe ₃ P calculated, respectively.

[0148] FIG. 40 is a graph showing the HER polarization curves of the Fe ₃ P film electrode on FTO with different counter electrodes at 100 mV s in acid. Two Fe ₃ P electrodes were separately tested for 600 cycles at 100 mV s in acid with Pt plate and graphite rod as the counter electrode, respectively. . Curve 4001 is for the 1 ^st cycle for graphite rod, curve 4002 is for the 600 ^th cycle of graphite rod, curve 4003 is for the 1 ^st cycle for Pt plate, and curve 4004 is for the 600 ^th cycle of Pt plate.

[0149] FIG. 41 is a graph showing the normalized polarization curves (of FIG. 37A) by ECSA of the FeP, Fe ₂ P, and Fe ₃ P films' electrodes. Curves 4101-4103 are for FeP, Fe ₂ P, and Fe ₃ P, respectively.

DETAILED DESCRIPTION

TiC /FeMnP Core/Shell Nanorod Array Photoanode

[0150] The present invention is a device composed of a semiconductor coated uniformly with a transition metal phosphide that was deposited by chemical vapor deposition of a single source organometallic precursor. It has been found that metal phosphides can be grown, which show good performance towards the oxygen-evolution and hydrogen evolution reactions of water-splitting, by employing single-source organometallic precursors as the chemical feedstock in chemical vapor deposition for growth on a range of substrates, even 3-dimensional ones, as uniform coatings. When the substrate on which the metal phosphide is grown is a semiconductor, a large enhancement in the photocurrent and catalytic rate has been seen.

[0151] In some embodiments, the present invention is an efficient core/shell photoanode with a highly active oxygen evolution electrocatalyst shell (FeMnP) and semiconductor core (rutile Ti0 ₂ ) for photoelectrochemical oxygen evolution reaction (PEC-OER).

[0152] The present invention includes a new ternary, metallic phosphide FeMnP electrocatalyst prepared by single-source MOCVD method. FeMnP has been demonstrated to be an efficient OER electrocatalyst that can deliver an overpotential of 300 mV to reach a current density of 10 mA cm ^"2 with a low Tafel slope of 65 mV dec ^"1 . FeMnP also displayed excellent stability in strong base for OER. FeMnP was then uniformly deposited onto the 3D Ti0 ₂ nanorod array to form a core/shell nanostructure photoanode. The as-fabricated Ti0 ₂ /FeMnP core/shell nanorod photoanode displayed significantly enhanced PEC performance. The photocurrent density is close to the theoretical value of rutile Ti0 ₂ under 100 mW cm ^"2 irradiation. FeMnP also negatively shifted the photocurrent onset potential of Ti0 ₂ by about 200 mV. The high PEC performance is attributed to the high catalytic activity of FeMnP. The formation of a high quality interfacial junction between the Ti0 ₂ light absorber and the FeMnP electrocatalyst can promote interfacial hole transfer to the active metal oxyhydroxide surface. It has also been found that metallic FeMnP interlayer between the semiconductor core and surface active sites plays important role for the suppression of charge recombination and promotion of charge transfer. The present MOCVD method provides a new and efficient strategy to couple efficient electrocatalysts with semiconductor photoanodes for photoelectrochemical solar energy conversion.

Semiconductor Core

[0153] The device of the present invention includes a semiconductor material (which is also referred to as the semiconductor core of the device). The choice of semiconductor material can be varied to be suitable for the oxygen-evolution and/or the hydrogen evolution (depending on the positions of a given semiconductor's valence and conduction bands). Examples of semiconductor materials that can be utilized in the present invention include heterometallic bismuth oxides, metal chalcogenides, tungsten oxides, etc. [0154] In some embodiments, Ti0 ₂ nanotube arrays are utilized as the semiconductor material.

[0155] Ti0 ₂ nanorod arrays on fluorine doped tin oxide (FTO) were grown by a hydrothermal method. [Hwang 2012]. For example, 0.25 mL of titanium butoxide was added to 7.5 mL of concentrated HC1 (35%), followed by the addition of 7.5 mL deionized water. The above solution was then transferred into an autoclave with a capacity of 50 mL. Two pieces of FTO substrate were placed into the autoclave at an angle with the conductive side facing down. The autoclave was then heated to 150 °C for 24 hours in an electric oven. After the growth, the FTO substrates were rinsed copiously with deionized water and ethanol. Then the FTO with Ti0 ₂ nanorod array was dried at 70 °C for 6 hours, followed by calcination at 500 °C for 2 hours at a rate of 5 °C min ^"1 .

Transition Metal Phosphide Shell

[0156] The present invention further includes a transition metal phosphide uniformly coated on the semiconductor. This coating is sometimes referred to as the shell. This coating can be deposited by chemical vapor deposition of a single source organometallic precursor, such as FeMn(CO)8^-PH2), H2Fe ₃ (CO)9^3-P-t-Bu), etc. All transition metal phosphides can be potentially be prepared from suitable precursors where the transition metal forms carbonyl clusters as well as phosphorus-based ligands.

[0157] For example, the metal phosphide precursor can be a single-source precursor

FeMn(CO) ₈ ^-PH ₂ ).

Deposition Process

[0158] The single source organometallic precursor is deposited by a chemical vapor deposition process. In some embodiments, the chemical deposition process is a metal-organic chemical vapor deposition (MOCVD) process.

[0159] Using the Ti0 ₂ nanorod arrays, a Ti0 ₂ /FeMnP core/shell nanorod array was fabricated by a MOCVD method using single-source precursor FeMn(CO) ₈ ^-PH2). Such deposition of FeMnP was a modified process disclosed and taught in Colson 2010. For example, this MOCVD method was performed by employing the apparatus shown in FIG. 6.

[0160] The apparatus 601 of FIG. 6 includes a quartz tube 602 of inner diameter 22 mm equipped with a Kimble-Chase high vacuum valve 603 at the end furthest from the precursor 604 and capped with a size 25 o-ring joint/cap assembly 605. 15 mm ² substrates 606 (Ti0 ₂ /FTO) were affixed to the stainless steel heating stage 607 with silver paste and heated at 130 °C in air to cure the silver paste and provide good thermal contact. (The dimensions of stainless steel heating stage 607 are shown in the figure 607a). The precursor FeMn(CO) ₈ ^-PH ₂ ) was loaded in a glove box under a nitrogen atmosphere. The substrate-affixed heating stage was placed in the apparatus and sealed. It was transferred to a high-vacuum manifold and the end of the apparatus 601 with the precursor 604 was submerged in a liquid nitrogen bath. (The precursor was positioned in Zone 2 609 of apparatus 601. While the precursor 604 was kept at 77 K, the apparatus 601 was evacuated until a cold-cathode ionization vacuum gauge stabilized at approximately 2 x 10 ^"6 Torr. The position of the heating stage in the apparatus was adjusted to achieve optimum substrate coverage. Zone 1 608 of the apparatus 601, where the substrates were located, was then heated to 350 °C for 30 minutes using heating tape. The nitrogen bath was removed from the precursor 604 and the precursor 604 was allowed to warm to room temperature. After 20 minutes, a metallic film had formed on the substrates and walls surrounding the substrate assembly. The apparatus 601 was disassembled under nitrogen and the material stored in air. FeMnP could be deposited onto bare FTO with the same procedure by replacing the T1O2/FTO substrate with bare FTO with 20 mg of precursor 604.

Characterization

[0161] The core/shell nanostructure was examined with scanning electron microscopy (SEM). See FIGS. 1A-1B.

[0162] After FeMnP deposition, the smooth surface of T1O2 nanorod (FIG. 7) had been uniformly coated by a FeMnP film consisting of numerous nanocrystals. The strong peak located at 2Θ of 39.9° in the X-ray diffraction (XRD) pattern of Ti0 ₂ /FeMnP is assigned to the (111) plane of hexagonal FeMnP (See FIG. 8), and the peaks located at 33.5°, 51.4° and 62.5° belong to the rutile Ti0 ₂ nanorods (JCPDS No. 21-1276) [Hwang 2012], confirming the successful deposition of FeMnP on Ti0 ₂ . The broadness of the FeMnP (111) peak indicates the crystallite size is small, consistent with the size of the observed nanocrystals of FeMnP on the Ti0 ₂ surface from the SEM image (FIG. IB).

[0163] Furthermore, a high-resolution transmission electron microscopy (HRTEM) was used to examine the crystalline structure. As can be seen from FIG. IE, the FeMnP shell is clearly seen to be crystalline with an exposed (111) plane given the d-spacing of 0.23 nm of the crystalline lattice. The FeMnP shell thickness is measured to be 18 nm. The TEM image in FIG. ID also clearly shows a core/shell structure.

[0164] To probe the elemental composition of the core/shell, energy dispersive spectroscopy (EDS) mapping under TEM was conducted over a single Ti0 ₂ /FeMnP core/shell nanorod. As show in FIG. IE, Ti (101) and O (102) are located in the central part of the core/shell nanostructure, while the elements Fe (103), Mn (104), and P (105) are homogeneously distributed across the whole nanorod, confirming the core/shell nanostructure.

[0165] Sputtering assisted X-ray photoelectron spectroscopy (XPS) is an effective method to analyze the depth dependent element composition. Sputtering assisted XPS analysis was conducted with a fresh Ti0 ₂ /FeMnP core/shell sample (FIGS. 2A-2C). At the surface (shown in curves 201 of FIGS. 2A-2B), Fe and Mn were in the zero-oxidation (2^ ₃ / ₂ = 706.7 eV) and 2+ oxidation state (2p ₃ / ₂ = 641.4 eV), [Powell 2012; Di Castro 1989] respectively. There is a small amount of zero-oxidation state of Mn at the surface, giving the weak peak at 638.7 eV. As shown in FIG. 2C, the P 2p core level spectrum showed two different oxidation states with one stronger peak at 129 eV corresponding to the phosphide component, and one weaker peak at 132.6 eV corresponding to the oxidized phosphorous component, which is common in metal phosphides. [Z. Zhang 2014]. A large shift in the Mn 2pm binding energy toward 638.7 eV indicated the Mn was mostly zero-valent after three minutes of sputtering. All of the Fe and P were zero-valent. Increasing the sputtering time to 6 min, the peak intensity at 638.7 eV increased, corresponding to an increase of metallic Mn. The atomic ratio of Fe:Mn:P was 1.4:1:1 with a slight increase of Fe suggesting preferential sputtering of the Mn and P. [Mahan 2000; Ho 1976; Hoffman 1998; Hoffman 1980; Coburn 1976; Werner 1976]. The results of sputtering-assisted XPS analysis indicated the FeMnP shell is metallic.

Electrocatalytic Activity

[0166] The electrocatalytic activity of FeMnP supported on FTO (fluorine-doped tin oxide glass) by the same MOCVD method was studied via a standard three-electrode configuration in 1.0 M KOH. Before the electrochemical measurements, FeMnP on FTO was characterized by SEM, XRD and TEM. The SEM image in FIG. 3A shows that FeMnP on FTO exhibits a uniform nanoplate structure. XRD data also confirmed FeMnP is in its hexagonal setting with the strong peak at 20 of 39.9° corresponding to the (111) plane. See FIG. 9. The HRTEM image in FIG. 3B shows a clear crystalline lattice with a d-spacing of 0.23 nm, which is indexed to the (111) plane. It is therefore concluded that the FeMnP nanoplates on FTO substrate have the same crystalline structure as the FeMnP shell on the Ti0 ₂ nanorod.

[0167] FIG.3C shows the polarization curves after iR-correction of the FeMnP/FTO (curve 301) and bare FTO electrodes (curve 302). The FeMnP nanoplates exhibit excellent OER activity, as demonstrated by the low onset overpotential (240 mV, J= 1.0 mA cm" ² ) and small overpotential of 300 mV to reach an electrocatalytic current density of 10 mA cm ^"2 .

[0168] The high electrocatalytic OER activity was also confirmed by a low Tafel slope of 65 mV dec ^"1 . In addition, as shown in FIG. 3D, FeMnP has excellent stability given that the CV profile in the 1 ^st cycle (curve 305) is identical with that of the 1000 ^th cycle (curve 306) at the high scan rate of 100 mV s ^"1 . Compared to other catalysts (TABLE 1), the OER performance of FeMnP is much better than manganese oxide-based OER catalysts [Meng 2014; Ramirez 2014; Fekete 2013], nanostructured cobalt oxide/selenide OER catalysts [Ma 2014; Tuysuz 2012; Gao 12014], and NiO _x and CoPi [McCrory 2013] . The high OER performance of FeMnP is helped by the small charge transfer resistance (FIG. 10) and large electrocatalytically active surface area (ECSA) which is linearly related to the double-layer capacitance (Cdi). The Cdi of FeMnP/FTO is calculated to be 15.9 mF cm" ² , almost 192-fold higher than that of bare FTO (0.0829 mF cm" ² ) See FIGS. 11C-11D. The Faradaic efficiency for FeMnP electrode was measured after 60 min electrolysis at 300 mV overpotential in 1 M KOH. Before the measurement, the electrochemical cell was calibrated with two Pt plates (assuming 100% Faradaic efficiency) as the working and the counter electrode, respectively. The integrated area of the 0 ₂ signal (FIG. 12) for FeMnP (from curve 1203) was 0.01832 and for Pt (from curve 1202) was 0.01908, giving value for the Faradaic efficiency of 96%.

TABLE 1

Comparison Of OER Activity Of Single Phase Catalysts

nanop a es

[0169] The optical properties of the Ti0 ₂ nanorods on FTO and the TiO^eMnP core/shell nanostructures on FTO were analyzed by UV-vis spectrometer. The light absorption edge of rutile Ti0 ₂ nanorods was found to be about 400 nm, giving a 3.0 eV band gap. See FIGS. 13A-13B. The Ti0 ₂ /FeMnP core/shell nanostructure displays very low transmittance at all wavelengths. See FIG. 14. The PEC measurements were conducted with the light irradiation from the uncoated backside.

PEC-OER Characterization

[0170] The PEC OER performance of Ti0 ₂ /FeMnP core/shell nanostructure photoanode was evaluated in a three-electrode configuration in 1.0 M NaOH under simulated 1 Sun irradiation using an AM 1.5G filter (100 mW cm ² ) from the backside. It was found that the Ti0 ₂ /FeMnP core/shell photoanode displayed a significantly enhanced photocurrent density (FIG. 4A). The Ti0 ₂ nanorod photoanode displayed a current density of 0.7 mA cm at 1.23 V vs RHE under irradiation, similar to previous reports [Hwang 2012]. Surprisingly, the Ti0 ₂ /FeMnP core/shell nanostructure photoanode displayed a much higher current density of about 2.9 mA cm ^"2 at 1.23 V vs RHE, an almost 4-fold enhancement. By subtracting the corresponding dark current density, the photocurrent density was about 1.8 mA cm ^"2 , very close to the theoretical photocurrent density of rutile Ti0 ₂ under 100 mW cm ^"2 light irradiation [J. Li 2015].

[0171] The photocurrent density of the present Ti0 ₂ FeMnP core/shell photoanode is higher than those of all reported T1O2 based photoanodes in TABLE 2.

TABLE 2

Comparison Of Photoanodes Under Simulated Light Irradiation (100 mW cm' ² )

[0172] Notably, The Ti0 ₂ /FeMnP core/shell nanostructure photoanode showed a large negative shift of about 200 mV of the onset potential of the photocurrent. The highest photoconversion efficiency (FIG. 4B) of the Ti0 ₂ nanorod photoanode was 0.25% at 0.65 V vs RHE, while at the same potential the photoconversion efficiency of the Ti0 ₂ /FeMnP core/shell photoanode increased to 0.65%, an almost 2.6-fold enhancement. The maximum photoconversion efficiency was 1.33% at 0.95 V vs RHE for the Ti0 ₂ /FeMnP core/shell photoanode, while at the same potential the photoconversion efficiency of bare Ti0 ₂ nanorod was only 0.17%. The photocurrent density is strongly dependent on the charge recombination rate. The transient photocurrent curve was obtained by chronoamperometry to investigate the charge recombination behavior (insert 409 of FIG. 4C). Both the Ti0 ₂ nanorod and Ti0 ₂ /FeMnP core/shell nanorod photoanodes showed fast responses to the switching of the light ON-OFF signal.

[0173] A normalized parameter (D) was derived from the transient photocurrent curve to quantitatively measure the charge recombination behavior. [J. Li 2013]. FIG. 4C shows the normalized plots of In D as a function of time (t). The transient time constant (τ) is defined as the time when In D = -1. τ was estimated to be 4.6 sec and 10.2 sec for Ti0 ₂ and the Ti0 ₂ FeMnP core/shell nanorod, respectively, which confirmed the suppression of the charge recombination.

[0174] The photochemical stability of the Ti0 ₂ FeMnP core/shell nanorod photoanode was also investigated by conducting a chronoamperometry measurement at 1.23 V vs RHE under light irradiation (FIG. 4D). There was no obvious decay of the photocurrent density after 90 min.

[0175] Sputtering assisted XPS was used to analyze the chemical state of the Ti0 ₂ /FeMnP core/shell nanorod after the 90 min stability test (FIG. 15). The surface Fe and Mn are heavily oxidized to Fe ³⁺ (711 eV) and Mn ⁴⁺ (642.9 eV). [Paparazzo 1987; B. Tan 1991]. It is well accepted that metal phosphides can convert to metal hydroxides and oxyhydroxides, which are also highly active OER catalysts. After several minutes of sputtering to remove the oxide layer, metallic FeMnP exists supported by the peak at 706.9 eV in the Fe 2pm spectra and the peak at 129.4 eV in the P 2p spectra, as well as the obvious negative shift from 642.9 eV to 641.9 eV of the Mn oxidation peak in its 2/> ₃ / ₂ spectra.

[0176] The presence of a metallic FeMnP interlayer promoted the charge transportation because of its high electrical conductivity and hence enhanced the charge separation, confirmed by the much smaller charge transfer resistance (FIG. 16). As a result, the carrier density was significantly increased for the Ti0 ₂ /FeMnP core/shell nanorod compared to the bare Ti0 ₂ nanorod. [0177] FIG. 5A shows the Mott-Schottky plots of the Ti0 ₂ nanorod and Ti0 ₂ /FeMnP core/shell nanorod photoanodes in the dark. Using the slopes of the linear parts of the Mott-Schottky plots, the donor densities of the Ti0 ₂ and Ti0 ₂ FeMnP were calculated to be 1.04 x 10 ¹⁸ cm ^"3 and 1.77 x 10 ²⁰ cm ^"3 , respectively. The enhanced donor density contributed to the high PEC performance of the Ti02/FeMnP core/shell nanorod. The flatband potential of Ti0 ₂ nanorod was determined to be -0.81 V vs Ag|AgCl (-0.61 V vs NHE) at pH 13.6 in NaOH solution from the Mott-Schottky plots in FIG 5A. Considering 3.0 eV as the band gap of Ti0 ₂ , the valence band maximum (VBM) and conduction band minimum (CBM) were calculated to be -7.71 eV and -4.71 eV versus vacuum level, respectively. The measured work function of bare FTO and FeMnP FTO was -5.17 eV and -5.25 eV (FIG. 17), respectively.

Oxygen Evolution/Hydrogen Evolution Catalysis

[0178] The measured work function of FTO matches with the value previously reported in the literature. [Helander 2011]. The work function of FeMnP is very close to those of metal, confirming its metallic nature. The Fermi level of FTO is lower than the CBM of Ti0 ₂ , and on the other hand, the Fermi level of FeMnP is closer to the VBM of Ti0 ₂ . Based on these considerations, the band alignment is illustrated in FIG. 5B. Under illumination, the photogenerated electrons in the CB of Ti0 ₂ migrate to FTO (layer 506) and then to the counter electrode Pt for the water reduction reaction. The photogenerated holes in the VB of Ti0 ₂ (layer 507) will flow to the metallic FeMnP interlayer 508 and then to the oxide surface (509) active sites for the water oxidation reaction.

[0179] Metal phosphides are known to convert to metal at their surfaces under aqueous conditions. [Kibsgaard 2015]. In turn, these species are believed to be the active catalysts for the oxygen- evolution reaction. [Dutta 2017]. From XPS studies, it was determined that the surface after testing was primarily manganese (IV) oxo and hydroxo species with some iron(III) present coupled with a structural surface reorganization observed with SEM (FIGS. 18A-18B). The metal responsible for the catalysis is not plainly clear.

[0180] A recent study of 3d transition metal hydroxide/oxyhydroxide (M = Mn, Fe, Co, Ni) species towards OER found Fe-OOH to be most active with Mn-OOH being the least active [Burke 2015]; thus, iron would appear to be the active species in the present study. In fact, the addition of iron to NiOOH was found to improve the OER current density by 500-fold. [Friebel 2015]. However, the contribution of manganese to the catalysis is non-trivial. D. Li et al. prepared Co ₂ P and CoMnP nanoparticles for OER catalysts finding that substituting Mn for Co lowered the overpotential for a current density of 10 mA cm ^"2 from 0.37 V to 0.33 V and decreased the Tafel slope from 128 mV dec ^" l to 61 mV dec ¹ . [D. Li 2016]. Substitution of Mn for Fe in Fe ₃ 0 ₄ was also found to greatly improve the material's electrocatalytic OER performance. [Singh 1996].

[0181] Moreover, Mn0 ₂ as an OER electrocatalyst was reported to "self-heal" during operation in acidic and alkaline conditions with no observed mass loss. [Hu nh 2014]. Thus, an exact determination of which metal is active is elusive, but synergy likely arises from the bimetallicity of the surface layer with a further contribution to stability arising from the presence of the manganese.

[0182] This conclusion is further supported by comparison with the electrochemical performance of Fe ₂ P, which requires an overpotential of 390 mV for 10 mA cm ^"2 for OER as reported by Read et al., a full 90 mV higher than that required by FeMnP for the same current density. [Read 2016]. Furthermore, D. Li et al. found MnP nanoparticles to quickly fade in performance concomitant with oxidation. [D. Li 2016].

[0183] Importantly, however, catalysis at the metal hydroxide/oxyhydroxide surface layer is fed by EC or PEC current supplied from the pristine, conductive FeMnP layer from which it seamlessly grew. It should be noted that the overpotentials exhibited in this work are much lower than even the lowest overpotentials of the metal layered-double hydroxides, likely due to the thinness of the oxidized layer through which the charge must traverse to perform the catalysis. [0184] Although many new materials have been developed and identified as active electrocatalysts, it remains challenging to couple them with suitable semiconductors for PEC or photocatalytic water splitting. This is particularly true for TMPs. The common routes for TMPs like the solvothermal method using P _re d, P4, or Na ₃ P and metal chloride [Brock 2004], or thermal phosphidation using either in situ generated P¾ or trioctylphosphine [Kibsgaard 2015; Xiao 2015; Shi 2016], are too destructive for semiconductor substrates. These methods have limited phase control over the metal-phosphorus stoichiometries resulting in inhomogeneities. Moreover, existing methods cannot guarantee complete coverage of the metal phosphide over the semiconductor surface. The most similar technique in terms of substrate coverage is Atomic-Layer Deposition (ALD) [Mohan 2000], but metal phosphide deposition has not been achieved to date by ALD, and such films would likely also suffer from poor phase control.

[0185] Past work in which Fe ₃ P, (Fei- _x Co _x ) ₃ P, and Fe ₃ (Pi- _x Te _x ) thin films were grown on quartz substrates by MOCVD using organometallic precursors has established a simple route to growing TMP films at mild temperature.f Colson 2012; Leitner 2016].

[0186] In the present work, the method of the present invention was applied to grow a TMP film directly on a semiconductor surface. The MOCVD method employed for FeMnP demonstrates its advantageousness in the fabrication of a highly active photoanode with semiconductor/ electrocatalyst three-dimensional core/shell architecture. The fast, low temperature MOCVD process has no negative effects on the T1O2 substrate. The volatility of the specifically designed and synthesized single-source precursor ensures the uniform coating of FeMnP on Ti0 ₂ .

[0187] Results demonstrate the great potential of the present single-source MOCVD method in the integration of semiconductor and electrocatalyst for solar water splitting. Moreover given the availability of other main group element-containing homo- and heterometallic carbonyl clusters [Whitmire 1998; Schipper 2016], the MOCVD technique should prove generally applicable for deposition of other electrocatalysts from single-source precursors onto semiconductors for the development of various highly active photoelectrodes.

Bifunctional Metal Phosphide FeMnP From Single-Source

Metal Organic Chemical Vapor Deposition

[0188] Furthermore, FeMnP was grown on NF and GNF by a facile MOCVD method in which the stoichiometry was controlled by use of an atomically precise single-source precursor. FeMnP is an efficient bifunctional catalyst showing remarkable activity towards the HER and the OER individually and in tandem with robust stability. The electrocatalytic performance of FeMnP can be enhanced by growing it on the more conductive multilayer-graphene protected NF. The ability of the MOCVD method to grow uniform films of a phase-pure, active catalyst on 3 -dimensional structures (such as with NF/GNF) reveals it is a practical route to preparing electrode materials. This provides for new bifunctional catalysts based on transition metal phosphides. [0189] The present invention thus additional encompasses a process by which transition metal phosphides can be grown directly on conductive substrates, the resulting material from which can serve as electrocatalysts for the oxygen evolution reaction (OER) and/or hydrogen evolution reaction (HER) of water electrolysis.

[0190] It has been found that films of metal phosphides can be grown from a single source precursor with a metal-organic chemical vapor deposition approach. Specifically, FeMnP has been grown from the molecular precursor FeMn(CO) ₈ ^-PH ₂ ) on a variety of substrates. Notably, the heavy element stoichiometry of Fe, Mn, and P is conserved in the process. The catalytic ability of FeMnP grown on nickel foam and graphene-wrapped nickel foam, two examples of conductive substrates, has been evaluated finding that the resulting electrodes can serve as binder-free electrodes and stably perform both the HER and OER at state-of-the-art values. [0191] Transition metal phosphides other than FeMnP can also be synthesized in like fashion. For example, it has been found that Fe ₃ P can be grown on quartz using H2Fe ₃ P(CO)9P¾u, that (FQI- _x Cox) ₃ P and Fe ₃ (Pi _-x Te _x ) can be grown from blends of isolobal (structurally similar) molecular precursors on quartz, and several more examples. It was confirmed that these additional metal phosphides could also be grown on nickel foam, conductive in comparison to the published reports, and that these electrodes could perform water electrolysis.

[0192] The value proposition of the present invention includes that one can create state-of-the-art electrode materials for efficient electrolysis by designing a molecular precursor with the desired stoichiometry and then use that molecular source to place the catalyst on a conductive substrate on which it can perform the OER/HER with electricity.

[0193] Nickel foam is a 3 -dimensional substrate. Because the precursor is a gas under the MOCVD conditions, it diffuses through the 3-dimensional network and allows an even coating to form with a small amount of precursor. A range of other conductive substrates can be utilized; moreover, the success obtained with the primitive MOCVD apparatus employed in the work points to easy scalability for practical industrial utilization.

[0194] The present invention is directed to a MOCVD method that can be used to deposit metal phosphide on conductive substrates for the binder-free electrodes. Different metals have different catalytic activity to different reactions. By controlling the metal composition in the precursor, one can achieve either high active HER catalyst, or highly active OER catalysts, or even highly active bifunctional catalysts for the overall water splitting.

[0195] For example, an embodiment of the present invention is a process that includes the following steps:

(1) Loading 10-15 mg of a suitable precursor into a tube in a glovebox.

(2) A stainless steel stage (block of metal) with the substrate attached is introduced and the tube is connected with the metal organic chemical vapor deposition apparatus.

(3) The tube is evacuated.

(4) The end of the tube where the precursor rests is kept low temperature or at room temperature, while the stainless steel stage is heated to ~350°C and holds at 350°C for 10 min.

(5) Then, the precursor is allowed to warm, turn to a gas, and decompose to yield the metal phosphide catalyst when it encounters the hot surface of the substrate.

[0196] The method of deposition could be varied. Instead of requiring the precursor be in its pure form, one could dissolve it in a carrier solvent like hexane and introduce the solution slowly to the vacuum chamber with the stage hot. One could also carry out the deposition at ambient or under mild vacuum with heating of the precursor (the precursor does not necessarily require vacuum to volatilize; heat may also accomplish the task).

[0197] High performance from FeMnP has been demonstrated. Alternatively, synthesis of precursors to other heterometallic transition metal phosphides includes, but not limited to, CoMnP, CoFeP, NiMnP, NiFeP, WMnP, etc., which are believed to be catalysts that also can exhibit high performance.

Growth of Graphene

[0198] Nickel foam was used as the substrate for the chemical vapor deposition (CVD) growth of graphene according to Chen 2011. Before graphene growth, the nickel foam was cut into pieces of 10 mm x 30 mm, and ultrasonicated for 5 minutes in each of deionized water, diluted HC1 solution, ethanol, and acetone. The clean nickel foam was placed in the center of a quartz boat, which was placed in the center of a quartz tube. The quartz tube was then evacuated and filled with Ar gas. The vacuum-filling process was repeated twice. The nickel foam was heated to 1000 °C under Ar gas (600 seem). H ₂ (200 seem) was then introduced into the tube to clean and eliminate the oxidized surface of nickel foam. After 30 min of hydrogen flow, CH ₄ (3 seem) was introduced for the growth of graphene. After 5 min of reaction, the feeding of CH ₄ stopped and the sample was rapidly moved out from the heating zone and cooled to room temperature under Ar (600 seem) and H ₂ (200 seem). Deposition of FeMnP

[0199] The organometallic precursor were synthesized and used as a single source precursor for the deposition of FeMnP nanoplatelets on nickel foam (NF) and GNF using a MOCVD setup, such as the apparatus 2001 shown in FIG.20, which is similar to apparatus 601 in FIG. 6.

[0200] FeMnP was deposited onto the surface of bare nickel foam and graphene protected nickel foam. Before the deposition, the nickel foam was also cleaned with the same procedure as used in the growth of graphene. The organometallic precursor FeMn(CO) ₈ ^-PH ₂ ) was synthesized according to Cohort 2010, and used as the precursor 2004 for the deposition of FeMnP with the metal-organic chemical vapor deposition (MOCVD) apparatus 2001 (with the precursor 2004 positioned in Zone 2 2002). [Cohort 2012]. The substrates 2006 (nickel foam or graphene- protected nickel foam) were oriented vertically and affixed to end of a stainless steel heating stage 607 with silver paste and heated at 130 °C in air to cure the silver paste and provide good thermal contact.

[0201] 20 mg of the precursor FeMn(CO) ₈ ^-PH ₂ ) 2004 was loaded into the end of the tube apparatus 2001 in a glove box under a nitrogen atmosphere. The substrate-affixed heating stage was placed in the apparatus and sealed. It was transferred to a high-vacuum manifold and the end of the apparatus 2001 with the precursor 2004 was submerged in a liquid nitrogen bath. While the precursor 2004 was kept at 77 K, the apparatus 2001 was evacuated until a cold-cathode ionization vacuum gauge stabilized at approximately 2.0 x 10 ^"6 Torr. The position of the heating stage in the apparatus was adjusted to achieve optimum substrate coverage. Zone 1 608 of the apparatus 2001, where the substrates 2006 were located, was then preheated to 350 °C for 30 minutes using a heating tape. The nitrogen bath was removed from the precursor 2004 which was allowed to warm to room temperature. After 20 minutes, a metallic film had formed on the substrates and walls surrounding the substrate assembly. The apparatus 2001 was disassembled under nitrogen and the material stored in air.

Characterization of FeMnP

[0202] The quality of the CVD-grown graphene on nickel foam was established with Raman analysis. The Raman results showed the presence of monolayer and multilayer graphene where the intensity ratios of IG I2D was 0.82 and 1.24 in two representative, respective regions. There was no D band in the Raman spectra indicating a highly ordered structure. The morphologies of FeMnP on NF and FeMnP on GNF were observed with scanning electron microscopy (SEM). Before the deposition, both the NF and GNF possess smooth surfaces. SEM images at lower magnification showed both the NF and GNF were uniformly covered by FeMnP nanoplatelets with a film thickness of about 5 μιη. Closer observation at higher magnification showed that FeMnP on NF and FeMnP on GNF had similar platelet-like structures with thicknesses of about 50 nm. See FIGS. 21A-21B. The X-ray diffraction pattern (XRD) showed two diffraction peaks located at 40° and 42° 2Θ, which can be indexed to the (111) and (201) planes, respectively, of the hexagonal phase according to the PDF Card (04-006-1275) refined with GSAS software to yield lattice constants of a = b = 5.923 A, c = 3.527 A. The diffraction peaks at 44.5°, 51.8° and 76.4° 2Θ were assigned to the (111), (200) and (220) planes of NF (JCPDS no. 65-2865).

[0203] The high resolution transmission electron microscopy (HRTEM) image of FIG. 21C confirmed the crystalline lattice with spacing distances of 2.28 nm and 2.16 nm, consistent with the respective d-spacings of the (111) and (201) planes. The selected area electron diffraction (SAED) image of FIG. 2 ID showed a characteristic polycrystalline ring pattern, the rings of which can be indexed to the (111), (201), (210) and (300) planes, further confirming the hexagonal phase of FeMnP. FIG. 21E shows the crystal structure of FeMnP in polyhedral view. FeMnP produced using FeMn(CO) ₈ ^-PH ₂ ) has been shown to be in a hexagonal P62m space group with the metals occupationally disordered over two sites.

[0204] The elemental composition was investigated by TEM energy dispersive spectroscopy (TEM-EDS) shown in FIG. 22A. The Fe, Mn and P were homogeneously distributed across the whole FeMnP nanoplatelet. The elemental composition of FeMnP was further confirmed by energy-dispersive X-ray spectroscopy (EDS) (see FIG. 22B), which shows the K line signals of Fe, Mn and P. An image taken under secondary electron imaging (SEI) mode confirmed the uniform phase purity of FeMnP.

[0205] It was found that the atomic ratio of Fe:Mn:P was 1 :1 :1, which is consistent with the atomic ratio of Fe:Mn:P in the precursor compound, indicating perfect translation of the atomic ratio of Fe:Mn:P from its precursor to the catalyst. The trace amount of oxygen and Ni may come from surface oxide and the NF substrate, respectively.

[0206] The chemical states of the as-deposited FeMnP were examined by sputtering assisted X- ray photoelectron spectroscopy (XPS) of FIGS. 22C-22E. Before sputtering, the strong peak in Fe 2p ₃ / ₂ spectra at 706.9 eV indicates the surface Fe was in the zero-oxidation state. After sputtering away the surface elements for 3 min and 6 min, the Fe was still zero-valent. The surface Mn appeared to be divalent given the peak at 641.9 eV in its 2p ₃ / ₂ spectra; broadening of the Mn 2p ₃ / ₂ peak was observed coupled with a shift of the binding energy toward 638.7 eV after 3 min and 6 min sputtering, implying the presence of zero-valent Mn. The surface P showed two broad peaks at 129.5 eV and 133.4 eV in its 2p spectra, which are assigned to the phosphide and oxidized phosphorous components, respectively. The latter is due to surface oxidation. After 3 min and 6 min sputtering, the peak at 133.4 eV disappeared and a doublet peak appeared at 129.4 eV and 130.2 eV indexed to phosphide P, suggesting all the P is in its phosphidic state below the surface. The atomic ratio of Fe: Mn: P from XPS analysis is 1 : 1: 1, which is consistent with the results of WDS and EDS measurements.

Electrocatalvsis

[0207] The FeMnP/NF and FeMnP/GNF prepared by the SSP-MOCVD method can be directly used as binder-free electrodes for water splitting. The overpotential and Tafel slope are the two most important parameters for the evaluation of the catalytic activity of electrocatalysts for water splitting.

[0208] For the OER measurements, the polarization curves and corresponding Tafel plots in 0.1 M KOH aqueous solution are shown in FIGS. 23A-23B. (In these figures, curves 2301 and 2305 are for FeMnP/GNF; curves 2302 and 2306 are for FeMnP/NF; curves 2303 and 2307 are for NF; and curves 2304 and 2308 are for GNF). The overpotential is defined as the potential at which the current density reaches 10 mA cm" ² . Both the bare NF and GNF electrodes show negligible OER performance in base, which is consistent with literature reports. The FeMnP/NF electrode generated a current density of 10 mA cm ^"2 at a potential of 1.51 V versus RHE with an overpotential of as low as 280 mV and a Tafel slope of 57 mV dec ^"1 .

[0209] The OER overpotential and Tafel slope of the present FeMnP are lower than most recently developed TMPs, including CoMnP nanoparticles (330 mV and 61 mV dec ^"1 ), NiCoP microspheres (340 mV and 86 mV dec ^"1 ), CoP nanorods (290 mV and 65 mV dec ^"1 ), and Ni ₂ P (290 mV and 59 mV dec ^"1 ), which are among the highest activity TMPs OER catalysts. Previous electrodes with FeMnP on fluorine-doped tin oxide (FTO) delivered an impressive overpotential of 300 mV with a Tafel slope of 65 mV dec ^"1 despite the high resistivity of FTO. By depositing FeMnP on GNF, the overpotential decreased to 230 mV with a Tafel slope of only 35 mV dec ^"1 , making the FeMnP/GNF electrode among the best graphene-based OER catalysts. Turnover frequency (TOF) reflects the intrinsic catalytic activity of the catalyst. The calculated TOF in OER for FeMnP/GNF at an overpotential of 280 mV was 0.28 s ^"1 , reflecting the high intrinsic OER catalytic activity of FeMnP.

[0210] Besides the outstanding OER activity, the FeMnP/NF and FeMnP/GNF electrodes are also highly active towards HER in acid. FIGS. 23C-23D show the polarization curves and corresponding Tafel slopes in 0.5 M H ₂ S0 ₄ . (In these figures, curves 2309 and 2012 are for FeMnP/NF; curves 2310 and 2013 are for FeMnP/GNF; and curve 2311 is for FeMnP/GNF after 75 hours). The FeMnP/NF electrode shows high HER activity with an overpotential of 125 mV and Tafel slope of 60 mV dec ^"1 .

[0211] The HER performance of FeMnP is better than or comparable to the reported HER catalysts, including transition metal sulfides MoS ₂ , and Fe^sNU.sSo.s, transition metal phosphides FeP, Ni ₂ P, and WP. A much lower overpotential toward HER was achieved by depositing FeMnP on GNF. The FeMnP/GNF electrode needed an overpotential of as low as 57 mV with the Tafel slope of 54 mV dec . The extraordinarily low HER overpotential places FeMnP/GNF among the best HER catalysts. Notably, the exchange current density reflected the intrinsic activity of catalyst itself. The exchange current density increased to 1.0 mA cm ^"2 from 0.14 mA cm ^"2 of FeMnP/NF, only lower than the highest recorded number of 1.2 mA cm ^"2 by graphitic carbon supported nickel.

[0212] The HER activity in base was tested with the objective being total water splitting from one type of electrode in the same cell. FeMnP shows high HER activity 0.1 M KOH. The HER overpotentials of the FeMnP/NF and FeMnP/GNF electrodes in base were 249 mV and 84 mV, respectively, with corresponding Tafel slopes of 123 mV dec ^"1 and 78 mV dec ^"1 . The exchange current densities for FeMnP/NF and FeMnP/GNF in base were 0.14 mA cm ^"2 and 0.78 mA cm ^"2 . The calculated turnover frequency (TOF) for the HER at an overpotential of 100 mV for FeMnP/GNF was 0.14 s ^"1 .

[0213] The OER stability was evaluated by cycling the FeMnP/NF and FeMnP/GNF electrodes in 0.1 M KOH for 1000 cycles at 100 mV s ^"1 , as shown in FIG. 24A. (Curves 2401-2404 are FeMnP/NF 1 ^st cycle, FeMnP/NF 1000 ^th cycle, FeMnP/GNF 1 ^st cycle, and FeMnP/GNF 1000 ^th cycle, respectively). Upon cycling, the overpotential for the FeMnP/NF electrode to reach an anodic current density of 20 mA cm ² decreased from the initial 320 mV (1 ^st cycle) to 300 mV (the 1000 ^th cycle), indicating an improvement of the OER catalytic activity. A similar increase was also observed for the FeMnP/GNF electrode with the overpotential at 20 mA cm ^"2 decreasing from 240 mV to 220 mV.

[0214] The HER stability was determined by cycling the FeMnP/NF and FeMnP/GNF electrodes in 0.5 M H2SO4 over 1000 cycles at 100 mV s ^"1 , as shown in FIG. 24B. (Curves 2405-2484 are FeMnP NF 1 ^st cycle, FeMnP/NF 1000 ^th cycle, FeMnP/GNF 1 ^st cycle, and FeMnP/GNF 1000 ^th cycle, respectively). For the FeMnP/NF electrode, the HER performance showed a slight decay with the overpotential at 20 mA cm ^"2 increasing from the initial 150 mV to 168 mV. However, for the FeMnP/GNF electrode, the profile of the polarization curve at the 1 ^st cycle was the same as that at the 1000 ^th cycle, indicating good HER stability in acid.

[0215] The overall water splitting was conducted in 0.1 M KOH solution by pairing two FeMnP/NF electrodes or two FeMnP/GNF electrodes. The cell voltage for FeMnP/NF was 1.60 V to reach 10 mA cm ^"2 current density, while a lower cell voltage of 1.55 V at 10 mA cm ^"2 was obtained by using two FeMnP/GNF electrodes. See FIG. 24C. (Curve 2409 is for FeMnP/NF; curve 2410 is for FeMnP/GNF; and curve 2411 is for FeMnP/GNF after 75 hours). The cell voltage of the present FeMnP NF was better than or comparable to previously reported bifunctional electrocatalysts, such as NiCo ₂ S ₄ (1.63 V), Ni ₂ P (1.63 V), NiCoP (1.64 V), and NiCo ₂ 0 ₄ (1.65 V). The cell voltage of 1.55 V of FeMnP/GNF for 10 mA cm ^"2 was also much lower than those of other reported electrocatalysts. Both the FeMnP/NF and FeMnP/GNF electrodes showed extraordinary long-term stability for overall water splitting in base. As shown in FIG. 24D, the current density of the FeMnP/NF electrode at 1.60 V shows no decay. (Curve 2412 is for FeMnP/NF; and curve 2413 is for FeMnP/GNF). For the FeMnP/GNF, at the same cell voltage of 1.60 V after 75 hours of testing, the current density slightly decreased from the initial 27.5 mA cm ^"2 to 25.0 mA cm ^"2 .

[0216] As shown in FIG. 24C, the polarization curve after 75 hours of testing was almost the same as that before testing. FIGS. 25A-25B show the produced 0 ₂ and H ₂ amounts measured by gas chromatography (GC) match the theoretically calculated amounts of 0 ₂ and H ₂ during the overall water splitting using two FeMnP/NF electrodes or two FeMnP/GNF electrodes, respectively. The molar ratio of H ₂ to 0 ₂ are close to 2, suggesting almost 100% Faradaic efficiency for FeMnP/NF and FeMnP/GNF, respectively.

Catalysis

[0217] A good electrocatalyst with high activity is defined by good conductivity and a large electrochemically active surface area (ECSA). Nyquist plots for the electrodes showed that the ohmic resistance decreased after the nickel foam was coated by CVD grown multilayer graphene, coupled with a decrease in the charge transfer resistance for both the OER and HER. Graphene has high electrical conductivity and surface area, is very stable in acid and strong base, and can also to protect the surface of the nickel foam from oxidation. It is suggested that graphene, with its superior electron pathway and excellent conductivity, provides a strongly coupled interface between the active phase and current collector. The ECSA of FeMnP/NF and FeMnP/GNF was estimated from the double-layer capacitance (Cdi). The FeMnP/GNF was calculated to have an ECSA of 71 mF cm ^"2 , about 20% higher than the ECSA of 57 mF cm ^"2 of FeMnP/NF, and both of them are much higher than that of the bare NF of 1.4 mF cm ^"2 . The greater ECSA after FeMnP deposition on NF contributes to the activity of FeMnP.

[0218] A chronoamperometry measurement of the FeMnP/NF was conducted at an overpotential of 300 mV in 0.1 M KOH, which showed a stable current density over 20 hours. The electrodes were and then analyzed by WDS and sputter-assisted XPS. FIGS. 26A-26B are images taken under SEI mode, which show the structural homogeneity of the tested FeMnP which retained the nanoplatelet morphology. FIGS. 26C-26D are images taken under COMPO mode that show the phase uniformity of the tested FeMnP. Further composition analysis by WDS confirmed the existence of the presence of Fe, Mn and P with an atomic ratio of 1:1:1. XPS analysis indicated the surface of the OER tested electrode was oxidized, given the binding energies of surface Fe, Mn, and P were 710.9 eV, 641.9 eV, and 138.2 eV, corresponding to the FeO _x , Μηθχ, and phosphate species. The binding energy at 530.0 eV in the 01 s spectra also confirmed the existence of FeOx and MnO _x .

[0219] Sputter-assisted XPS was used to remove the surface species and it was found that the Fe, Mn and P were still in the zero-valence states below the surface. Therefore, it is concluded that over the course of the testing, the surface of FeMnP was oxidized to form active sites with the pristine underlying metallic FeMnP layer providing a highly conductive electron pathway, a configuration responsible for the high stability and performance of the FeMnP catalyst. Sputter- assisted XPS analysis on the electrode after long-term testing for HER in acid found that the surface of FeMnP had been oxidized, and under the oxidized surface metallic FeMnP still existed.

Mechanism

[0220] These experimental results suggested FeMnP exhibited high electrocatalytic activity toward HER. To elucidate the underlying mechanism, a series of DFT calculations were performed. Predictions from the Bravais-Friedel-Donnay-Harker (BFDH) algorithm indicated that (100) and (001) low index facets were the most probable terminations for FeMnP. A calculation of the surface formation energy of a FeMnP (100) resulted in 0.11 eV/A ² , while the cleavage of FeMnP along (001) planes required 0.42 eV/A ² to yield Fe- and Mn-terminated surfaces. The low energy required to form (100) facets was a good indication of dominant FeMnP (100) surfaces that may serve as active facets for electrocatalytic hydrogen evolution. Of the two different terminations exhibited by the less stable (001) facet, the Mn-terminated surface, referred to as (OOl)-Mn, were concentrated upon. The Fe-terminated FeMnP (001) facet strongly resembles Fe ₂ P (001), which was already shown to bind hydrogen too exergonic while Fe sites are available, and exhibits strongly endergonic binding after the preferred adsorptions sites are occupied.

[0221] A necessary, but not sufficient, criterion for optimal HER performance is a differential Gibbs free binding energy for hydrogen, AG _H , close to zero. This criterion can be rationalized in terms of the Sabatier principle and the competing hydrogen binding requirements for facile proton adsorption via the Volmer step and rapid H ₂ evolution by either the Tafel or Heyrovsky reaction. FIG. 27E shows AG _H as a function of coverage Θ on the (100) and (OOl)-Mn facets (curves 2701-2702, respectively). The preferred hydrogen adsorption site is between a Fe-Fe bridge site and a Mn atom on the (100) facet, FIG.27A, and threefold Mn sites on the (001) facet, FIG. 27C. With increasing hydrogen coverage up to 1 ML hydrogen atoms continue to occupy their preferred adsorption sites, but small repulsive interactions contribute to gradually weaker binding. Above 1 ML the differential AG _H on both surfaces changes from exergonic to endergonic, which corresponds to the expected surface coverage under electrocatalytic operating conditions.

[0222] For the (OOl)-Mn facet at 1 ML coverage, the smallest magnitude of AG _H = -0.24 eV was obtained, which indicates overbinding. Under these conditions, a subsequent hydrogen adsorption at 9/8 ML is highly unstable and requires +0.63 eV. This behavior is reminiscent of pure Fe ₂ P and corresponds to the occupation of an unfavorable binding site near a Mn atom, which also causes a small displacement of two H atoms in the preferred threefold Mn site (FIG. 27D). In contrast, hydrogen binding to the (100) facet is almost thermoneutral with AG _H = 0.06 eV at 7/6 ML coverage. In this case, hydrogen adsorbs between a Mn-Mn bridge site and an Fe atom, while forcing two other hydrogen atoms onto Fe-Fe bridge positions (FIG. 27B). Furthermore, AG _H of the (100) facet shows a weak coverage-dependence and remains less than 0.23 eV up to 2 ML.

[0223] This behavior suggests good tolerance to coverage variations and the possibility of favorable entropy contributions stemming from a large number of surface configurations that hydrogen atoms can assume in the shallow binding energy potential. Overall, the hydrogen binding characteristics of the FeMnP (100) surface fulfil the necessary requirement for efficient hydrogen evolution at its optimal coverage and further suggest that the favorable characteristics extend to even higher coverages. It is noted that the relevant hydrogen binding sites on FeMnP (100) are comprised of mixed Fe and Mn atoms, which points to a synergistic effect between these constituents and is fully consistent with the electrocatalytic characterization results here.

Preparation of Phase-Pure Films of FeP, Fe ₂ P, and Fe3P and Relative Catalytic Activities

[0224] The single-source precursor MOCVD method employed for the deposition of iron phosphides has permitted the formation of phase pure thin films of FeP and Fe ₂ P. Elimination of phosphine Fe(CO) ₄ L (L = PH ₃ , PH ₂ ¾u) leads to films that are metal-rich rather than possessing the stoichiometry of the SSP. This can be overcome by varying the temperature of decomposition or altering the structure of the SSP so that the phosphorus ligands are more strongly bound to iron. The use of a precursor that contains a diphosphane unit circumvents this problem and allows for production of the films with the desired stoichiometry at comparatively lower temperatures. The synthesis of these films allowed electrocatalytic activities of the iron phosphides to be analyzed, which show that their relative catalytic activity increases with the proportion of metal.

[0225] Assessments of the comparative catalytic activity of transition metal phosphides, Fe _x P (x = 1-3), have been made possible by use of chemical vapor deposition (CVD) of single-source organometallic precursors (SSPs) to grow phase-pure FeP and Fe ₂ P thin films for the first time. Films on quartz of Fe ₂ P were grown using Fe(CO) ₄ PH ₃ (Compound 1), while films of FeP were prepared using either Fe(CO)4P ⁱ BuH2 (Compound 2) or a new molecule {Fe(CO)4P(H)¾u}2 (Compound 3). Films of FeP, Fe2P and Fe ₃ P, prepared from H ₂ Fe ₃ (CO)9P¾u (Compound 4) were also deposited on fluorine-doped tin oxide (FTO), and these films were evaluated for their activities towards the hydrogen evolution reaction (HER) of the water splitting in 0.5 M H ₂ S0 ₄ . HER activity follows the trend FeP < Fe ₂ P < Fe ₃ P, with Fe ₃ P having the lowest overpotential of 49 mV at 10 mA cm ^"2 . The analysis revealed a clear trend of activity with metal-rich phosphide phases outperforming phosphorus poor phases for hydrogen evolution.

Compounds

[0226] As used herein, Compounds 1-4 are reflected in TABLE 3 below. TABLE 3

Compounds

Synthesis of Compounds 1-4

[0227] Compounds 1, 2, and 4 were prepared according to literature methods [Colson 2010, Miiller 1983, and Huttner 1980, respectively] and standard Schlenk technique.

[0228] Compound 3, which is a rare example of metallated organodiphosphane, can be prepared by treating an equal molar mixture of Na[HFe(C0 ₄ ] and Na ₂ [Fe(CO) ₄ ] with ¾uPCl ₂ in THF. The methodology is based on nucleophilic displacement of halide ions from ¾uPCl ₂ by Na[HFe(C0 ₄ ] with the proposed intermediacy of TBuPitTjCl FetCCXi} that would undergo subsequent reductive coupling by Na ₂ [Fe(CO)4].

[0229] For example, to a chilled (-10 °C) suspension of sodium hydroxide (1.9 g, 46 mmol) in 10 mL of methanol was added 2 mL of iron pentacarbonyl (15 mmol) and the resulting solution stirred at 1 h at this temperature before warming to room temperature after which the solution was allowed to stir for 20 h. The solvent was then removed in vacuo and the solids thoroughly dried. To the resulting solids were then added 80 mL of tetrahydrofuran to extract Na[HFe(CO)4], the solution of which was filtered into a flask containing 0.1865 NaH (7.8 mmol). Strong bubbling signaling the release of hydrogen concomitant with the formation of [Fe(CO) ₄ ] ^2" was observed which subsided after 30 m at which point an infrared spectrum of the solution indicated the presence of both [Fe(CO) ] ^2" and [HFe(CO) ₄ ]-. Then, 14 mL of a 1 M solution of P¾uCl (14.0 mmol) in diethyl ether was added in 2 mL aliquots spread over five minutes accompanied by a change in color from brown-orange to yellow-orange. After stirring overnight, the solvent was removed in vacuo and the solid treated with hexane (80 mL) and filtered. The filtered solution was then reduced to an oil in vacuo and left to crystallize at -10 °C. After three days at this temperature, large masses of crystals were found in the oil which was filtered off. The crystals were washed with 10 mL of hexane to yield 0.41 g (9.4% yield) of crystalline Compound 3. The filtered oil continued to produce crystals of Compound 3 upon standing at room temperature and after two weeks of standing yielded another 0.47 g of product for a total yield of 23.6%. Elemental analysis calc: C: 37.54%, H: 3.54%, N: 0%. Found: C: 36.79%, H: 3.44%, N: <0.50%. vco (hexanes): 2067.78(m), 2054.63(s), 2017.63(m), 1987.78(s), 1979.28(vs), 1974.15(vs), 1963.15(vs), 1945.20(vw) cm ^"1 . M.pt. 117-125 °C. ESI-MS Data for crystalline lb: mlz (%) 512.9 (76) [lb-H ⁺ ]-, 484.9 (32) [lb-H ⁺ - CO] ^" , 456.9 (66) [lb-H ⁺ -2 CO] ^" , 428.9 (33)[lb-H ⁺ - 3 CO] ^" , 400.9 (100)[lb-H ⁺ - 4 CO] ^" , 372.9 (14) [lb-H ⁺ - 5 CO] ^" . Ή-NMR data (CeDe, ppm): Compound 3 exhibits a complex second-order spectrum consistent with the two components which could arise from hindered rotation about the P-P bond. The first P-H multiplet is centered at 4.81; the second P-H multiplet is centered at 4.73 ppm. The t-butyl regions overlap with the envelope centered at 0.92 ppm. The P-H and t-butyl regions integrate to 1:9, consistent with the crystal structure.

[0230] A summary of X-ray data collection and refinement parameters for the compounds is given in TABLE 4.

TABLE 4

[0231] A summary of selected bond lengths and angles for the reported compounds is given in TABLE 5.

TABLE 5

[0232] The ³¹ P and ^l R spectra for Compound 3 were not only second-order in nature but suggested Compound 3 possesses hindered rotation about the P-P bond. Given that the crystal is racemic there are equal proportions of (R,R) and (5,5) enantiomers overall, even though each takes its turn being the minor component of one of the, the minor component seen in the crystal structure is a conformer, distinguished from the major component by the significantly different rotational orientation of its substituent groups about the phosphorus atoms, a situation which gives rise to two sets of overlapping signals in the Ή-ΝΜΡν spectrum with the major envelope for the P-H shift centered at 4.81 ppm and the minor component centered at 4.73 ppm. The minor conformer has its t-butyl groups eclipsed while the major conformer has them well separated, leading to the differing P-H environment. The racemic nature of the crystal makes the (R,R) designation for the major conformer arbitrary. Both enantiomers have equal proportions of the minor conformer.

[0233] The ³¹ P spectrum showed a similar pattern to the Ή pattern, however, with only one set of peaks. A 'H-decoupled ³¹ P experiment revealed only one phosphorus signal, suggesting that only one phosphorus environment exists. This is consistent with the two conformers having a similar phosphorus environment. Variable temperature ³¹ P and ^] H studies of Compound 3 (RT to 80 °C) were undertaken in attempts to see if coalescence of the ^! H patterns was possible. However, the ratio of the two P-H ¾ envelopes did not change although the envelopes migrated away from one another. In the phosphorus spectrum, the P signal migrated and appeared to flatten. MOCVD Growth of Phase Pure Thin Films of FeP and Fe P

[0234] The SS-MOCVD method developed for the growth of Fe ₃ P has been applied to production of thin films of FeP and Fe ₂ P on quartz and fluorine-doped tin-oxide (FTO). Such film deposition was carried out in a similar manner as discussed above. Key differences are that the distances of the tip of the metal stage to the end of the tube (distance "X" of apparatus 2901 shown in FIG. 29) and the start of the heating zone (distance "Y" of apparatus 2901) were varied according to film deposition parameters utilized in apparatus 2901 as show in the TABLE 6. Generally, the X Y ratio is in the range of 1.3:1 to 1.7:1.

TABLE 6

Film Deposition Parameters

[0235] Another key difference is that small boats/cups made out of aluminum foil (1 cm deep, 1 cm in diameter) were used to weigh and transfer the precursor 2904 to the bottom of the tube, which made it easier to get the precursor 2904 to the bottom of the tube without having any to stick to the walls. This was especially helpful when working with Compound 1 as it is a liquid. Substrates 2906 were cut into ~20.5mm x 11 mm sections and affixed to the stage with silver paste.

[0236] After deposition, the heating zone 608 was shifted forward to envelop the stage 607 and the material on the substrate 2906 allowed to anneal for two hours at the deposition temperature before slow cooling to room temperature. Note: Fe ₃ P was annealed for 24 h at 550 °C to get a satisfactorily crystalline material, which could then be used for HER testing. An aluminum foil jacket was placed over the end of the tube for Fe ₃ P deposition to encourage volatilization of the precursor.

[0237] Compound 1 has been demonstrated to yield nanoparticles of FeP by solution-based decomposition [Hunger 2013] and was explored as a CVD precursor; however, it was found that Fe ₂ P was the only material that deposited on a quartz substrate at 350 °C. At temperatures > 400 °C, a mixture of FeP and Fe ₂ P was obtained, which are shown in FIGS. 30A-30B. The ability to form phase pure Fe ₂ P showed that there was a very clean rearrangement process occurring at the lower temperatures to give a precise Fe ₂ P precursor with stoichiometric loss of P. A believed intermediate step is provided in equation (1).

2 Fe(CO) ₄ PH ₃ → ^-H)Fe ₂ (CO) ₆ ^-CO)^-PH ₂ ) + PH ₃ (1)

[0238] ³¹ P and ^! H NMR spectra of the off-gases from the decomposition of 1 under CVD conditions confirmed the elimination of PH ₃ .

[0239] It was believed that a *Bu in place of a hydrogen could slow, and potentially stop, the release of phosphorus as PH ₃ during decomposition as PH ₂ ¾u is a stronger donor and should be bound more tightly. At 350 °C, Fe ₂ P was again the product. At 450 °C, the decomposition of Compound 2 product was phase-pure FeP. Although both Compounds 1 and 2 lost phosphorus at 350 °C, the loss of phosphorus from Compound 1 was slowed by increasing the decomposition temperature to 450 °C and eliminated completely using Compound 2 at 450 °C, revealing that decomposition occurs at that temperature before rearrangement can take place with the loss of phosphorus.

[0240] While both precursors gave Fe ₂ P as the sole product at lower decomposition temperatures (350 °C), more crystalline Fe ₂ P was obtained from Compound 1. In contrast, the new derivative Compound 3 contains a P-P bond and was found to give pure FeP at 350 °C, although the deposition time required (8 h vs. 15 m) was longer due to the lower volatility of Compound 3 compared to that of Compound 2. The films of FeP derived from Compound 2 or Compound 3, however, were indistinguishable.

[0241] The Compound 3 molecule crystallized in the chiral space group P4i2i2 as a racemic twin (FIGS. 28A-28C) possessing two crystallographically distinct, dimeric "Fe(CO) ₄ (H)P ⁱ Bu" units joined by a P-P bond. The diphosphine (P'BuH)2 has never before been observed as a ligand, and only six examples of a diphosphine with formula = PR(H) have been previously structurally characterized. [Jones 2014; Duffy 2012; Tian 2013; Bartlett 1987; Huttner 2014]. Notably, the diphosphane coordinates at the equatorial rather than axial positions on Fe, an uncommon configuration usually associated with π-acceptor ligands. The molecule is C ₂ symmetric and both enantiomers are observed in an 85:15 ratio in the structure, although the crystals themselves are chiral and refined as a racemic twin. Interestingly, the two forms in the crystal structure (FIGS. 28B-28C) show dramatically different rotational conformations. The negative ion ESI-MS shows the [P-H ⁺ ] ^" ion very cleanly with successive loss of five CO ligands. The ³¹ P and ^! H NMR spectra showed complex second order behavior with the overall pattern consistent with those observed for the few existing related compounds. [Duffy 2012; Tian 2013].

Film Characterization

[0242] The films on quartz were characterized by (1) powder-XRD to establish phase, (2) XPS depth-profiling to ascertain homogeneity, and (3) SEM for morphology. See FIGS. 31A-31D, 32A-32B, and 33A-33B. The powder pattern of FeP indexed to its orthorhombic Pnma setting. See FIG.31 A. XPS depth profiling confirmed a 1 : 1 Fe:P ratio with only trace oxygen and carbon. See FIG. 31D. The surface was partially oxidized, but the interior was composed solely of Fe and P whose binding energies of 706.9 eV and 129.65 eV closely match the literature values of 706.9 eV and 129.34 eV, respectively. The SEM images of FIGS. 32A-32B of FeP from Compound 2 on quartz show that the film consists of rectangular crystallites approximately 500 nm in length. Films of FeP on quartz from Compound 3 shown in FIGS. 33A-33B were morphologically similar, although preferred orientation is apparent in the powder spectrum. [0243] Similarly, Fe ₂ P was characterized by SEM (as shown in FIGS. 34A-34B), PXRD (as shown in FIG. 35A), and XPS analysis with depth profiling (as shown in FIGS. 35B-35D). The XRD pattern was indexed to the hexagonal P 62m setting for Fe ₂ P. XPS depth profiling demonstrated the film is homogeneous with only trace oxygen and carbon. Like the FeP, the surface was partially oxidized, but the interior was pristine. The binding energies for Fe and P were 706.8 eV and 129.55 eV, which are close to the reported values of 706.8 eV and 129.31 eV. The SEM images of FIGS. 34A-34B shows the film to have contiguous hexagonal towers, reflecting the hexagonal symmetry of the Fe ₂ P crystal system.

[0244] FeP, Fe ₂ P, and Fe ₃ P were grown on FTO analogously to the depositions on quartz. The films were characterized by XRD, SEM, and XPS. The XRD patterns of the respective materials confirm the phase identity and purity of the respective iron phosphides. SEM images of FIGS. 36A-36F reveal clear morphological differences among the iron phosphides. FIGS.36A and 36D are of FeP; FIGS. 36B and 36E are of Fe ₂ P; and FIGS. 36C and 36F are of Fe ₃ P.

[0245] FeP has rectangular prisms, Fe ₂ P has stacked hexagonal sheets, and Fe ₃ P has cauliflower- like growths. These shapes can be attributed to the crystal structures of the respective iron phosphides. The rectangular blocks observed for FeP are consistent with an orthorhombic space group, the hexagonal sheets observed for Fe ₂ P are consistent with a hexagonal space group, and the cauliflower-like shape for the Fe ₃ P can be attributed to its tetragonal crystal system which has a tendency to twin along the {011 } faces.

Catalytic Activity of the Iron Phosphide Films

[0246] For HER characterization, films of FeP, Fe ₂ P and Fe ₃ P, whose synthesis on quartz has been detailed and described above starting from Compound 4, were grown on FTO. The ability to use a single-method to grow films of phase-pure, high quality (trace oxygen and carbon) materials cleanly on a conductive substrate facilitates direct comparison of FeP, Fe ₂ P, and Fe ₃ P. The sequence of HER activity has been shown to follow the series Fe ₃ P > Fe ₂ P > FeP with stability following the same trend. These shows metal-rich phosphides, particularly the M ₃ P phase, as being superior to metal-poor phosphides for HER.

[0247] The HER activity was evaluated with a three-electrode configuration in 0.5 M H ₂ S04, and the respective as-deposited iron phosphide on FTO was directly used as the working electrode. FIGS. 37A-37B shows the polarization curves after zR-correction and the corresponding Tafel slopes for the iron phosphides. The overpotential is defined as the potential to reach a current density of 10 mA-cm ^"2 . The performance of Pt was used as a reference and is in good agreement with reported values; the bare FTO showed negligible HER performance. The overpotentials for the FeP, Fe ₂ P and Fe ₃ P electrodes are 116 mV, 83 mV and 49 mV, respectively, as shown in FIG. 37A. The corresponding Tafel slopes for FeP, Fe ₂ P, and Fe ₃ P are 79 mV dec, 66 mV dec and 57 mV dec.

[0248] The HER performance increased in the order FeP, Fe ₂ P and Fe ₃ P, with the increase of iron content or with the decrease of P content. The exchange current density (jo) was obtained by extrapolating the linear part of the Tafel plots to intersect with the x-axis. Fe ₃ P has the highest intrinsic catalytic activity of the tested iron phosphides as a result of having the highest exchange current density of 1.32 mA-cm ^'2 as shown in FIG 37B.

[0249] Nyquist plots derived from electrochemical impedance spectroscopy (EIS) were employed to investigate the HER kinetics. All electrodes show a similar semicircle profile without Warburg impedance in the low frequency range, indicating the mass transport is rapid and is kinetically controlled. The active sites at the surface can be easily accessed by the electrolyte ions. The Nyquist plots can be fitted with the equivalent circuit consisting of the equivalent series resistance (Rs) and the charge transfer resistance (Ret) with the constant element referring to the double layer capacitance. All the electrodes have similar R _s values of about 5 Ω. Fe ₃ P exhibits the lowest charge transfer resistance of 25 Ω at the same overpotential of 160 mV, compared to FeP and Fe ₂ P, indicating that Fe ₃ P has the fastest charge transfer rate during the HER. The charge transfer Tafel slope was determined from the slope of the linear fitting of the plot of log R _c t versus overpotential. See FIG. 37C.

[0250] It was found that the values of the slopes fall between 39 mV/dec and 118 mV/dec, indicating the charge transfer is the rate determining step. The Fe ₃ P electrode displayed the lowest charge transfer Tafel slope of 64 mV/dec, further confirming that its charge transfer kinetics in the HER process are faster than those of Fe ₂ P and FeP.

[0251] The electrocatalytically active surface area (ECSA) was obtained by measuring the electrochemical double layer capacitance using cyclic voltammetry at a non-Faradaic reaction potential range, as shown in FIG. 37D and FIGS. 38A-38C. The ECSA for FeP, Fe ₂ P and Fe ₃ P was 18.3 mF/cm ² , 19.4 mF/cm ² and 23.6 mF/cm ² , respectively. The Fe ₃ P possessed the largest ECSA, in good agreement with it exhibiting the highest HER performance.

[0252] The long-term stability of the as-deposited catalysts by chronoamperometry measurements at an overpotential of 120 mV in 0.5 M H ₂ S04 solution was also investigated. See FIG.39A. The current density of Fe ₃ P is almost 2.7 times higher and six times higher than that of Fe ₂ P and FeP, respectively, further confirming the superior HER performance of Fe ₃ P compared to Fe ₂ P and FeP. All the electrodes showed stable current density over 20 hours operation. The Faradaic efficiencies of FeP, Fe ₂ P and Fe ₃ P were evaluated by comparing the measured amount of hydrogen by gas chromatography to the calculated amount of hydrogen according to the recorded current. As shown in FIG. 39B, the Faradaic efficiency was 96%, 98% and 97% for FeP, Fe ₂ P and Fe ₃ P, respectively.

[0253] The oxidation states of the tested Fe ₃ P film were determined with XPS. The surface XPS spectra indicated the existence of metallic Fe and phosphide components given the peaks located at 707.6 eV for Fe and 129.4 eV for P, respectively. That both Fe and P were found at the surface following testing in their zero-valent state is consistent with catalysis of HER by the iron phosphide. Furthermore, it was found that there was no detectable Pt at the surface of the tested Fe ₃ P sample. In order to eliminate this potential source of contamination, cyclic voltammetry was conducted at a scan rate of 100 mV/s for 600 cycles using Pt and a graphite rod as the counter electrode, respectively, as shown in FIG. 40. It was found that the profiles of the polarization curves are very close to each other. Compared to the 1 ^st cycle, the curves at the 600 ^th cycle shift to more negative potentials for the Pt and graphite rod counter electrodes. Therefore, it was safely conclude that there was no Pt contamination of the working electrode.

[0254] The as-deposited Fe ₃ P on FTO shows better than, or comparable, HER performance to Fe ₂ P and FeP in acidic media. TABLE 7 summarizes the HER performance of recently developed transition metal phosphides for comparison. Previous theoretical calculations have suggested that transition metal atoms function as the catalytic reaction center, while the negatively charged P atoms assist in trapping protons and promoting the dissociation of H ₂ . [Kibsgaard 2015; Shi 2016]. Therefore, the activity trend Fe ₃ P > Fe ₂ P > FeP is somewhat unexpected.

[0255] It is noted that, as summarized in TABLE 7, the overpotentials of some FeP nanostructures range from 31 mV to 88 mV, lower than the overpotential of some Fe ₂ P nanostructures (88 mV to 101 mV). TABLE 8 reflects the HER performance comparison of transition metal phosphides in 0.5 M H ₂ S0 ₄ solution. The overpotential is defined as the overpotential to reach a current density of 10 mA cm ^"2 . FTO: Fluorine-doped Tin oxide glass; GCE: glass carbon electrode; rGO: reduced graphene oxide.

TABLE 7

Carbon

FeP nanowires 31 53 Lv 2016 paper

FeP nanoparticles GCE 154 65 Tian 2016

Fe ₂ P Y. Zhang

GCE 88 49

nanoparticle/carbon 2015

Fe ₂ P

GCE 101 55.2 Liu 2017 nanoparticle/rGO

FeP nanowire Fe foil 96 39 0.17 Son 2016

FeP ₂ nanowire Fe foil 61 37 0.55 Son 2016

Callejas

CoP nanoparticle Ti 75 50

2015

Cu ₃ P nanowire Cu foam 143 67 0.18 Tian 2015

Popczun

Ni ₂ P nanoparticle GCE > 100 46 0.033

2013

[0256] Also, a sample of FeP ₂ nanowires has been reported to possess a lower overpotential of 61 mV than that of 96 mV of FeP nanowires, both on Fe foil. [Son 2016]. In the present invention, there is precise control over the stoichiometry of the iron and phosphorus content, and the catalysts are deposited onto planar FTO substrates to form uniform thin films, meaning that the FTO substrate has less effect on the geometric current density than common 3D substrates like Cu foam and Ni foam. The normalized current density by ECSA in FIG. 41 indicates that the FTER activity obeys the order of Fe ₃ P > Fe ₂ P > FeP.

[0257] It would seem that higher P content results in a lower overpotential. However, for the literature reports, the substrate and catalyst nanostructure morphology has a large effect on the HER performance of FeP and Fe ₂ P. In general, the wide variability of metal phosphides tested to date for HER can be understood as a result of synthetic differences in the preparation of metal phosphides such as surface stabilizing agents (a particular problem for nanoparticle samples), difficulty in controlling the surface chemistry, and variabilities in Ohmic contact. [0258] While embodiments of the invention have been shown and described, modifications thereof can be made by one skilled in the art without departing from the spirit and teachings of the invention. The embodiments described and the examples provided herein are exemplary only, and are not intended to be limiting. Many variations and modifications of the invention disclosed herein are possible and are within the scope of the invention. Accordingly, other embodiments are within the scope of the following claims. The scope of protection is not limited by the description set out above.

[0259] The disclosures of all patents, patent applications, and publications cited herein are hereby incorporated herein by reference in their entirety, to the extent that they provide exemplary, procedural, or other details supplementary to those set forth herein.

REFERENCES

[0260] Ai, G., et al. "Cobalt phosphate modified Ti0 ₂ nanowire arrays as co-catalysts for solar water splitting," Nanoscale 2015, 7(15), 6722-6728 ("Ai 2015").

[0261] Armstrong, R. C, et al, "The frontiers of energy," Nature Energy 2016, 1, 15020 ("Armstrong 2016").

[0262] Barlett, R. A. et al. "Reaction of Bulky Monosubstituted phosphorus(III) Halides with Disodium Pentacarbonylchromate. Steric and Electronic Factors in the Synthesis of Cr(CO) ₅ Complexes of Diphosphenes, Phosphinidenes, Phosphanes, Diphosphanes, and Cyclopolyphosphanes," J. Am. Chem. Soc. 1987, 109 (19), 5693-5698 ("Bartlett 198T).

[0263] Brock, S. L., et al, "Chemical Routes for Production of Transition - Metal Phosphides on the Nanoscale: Implications for Advanced Magnetic and Catalytic Materials," Chemistry -A European Journal IWA, 10 (14), 3364-3371 ("Brock 2004").

[0264] Burke, M. S., et al, "Revised oxygen evolution reaction activity trends for first-row transition-metal (oxy) hydroxides in alkaline media," The journal of physical chemistry letters 2015, 6 (18), 3737-3742 ("Burte 2015").

[0265] Callejas, J. F., et al, "Synthesis, Characterization, and Properties of Metal Phosphide Catalysts for the Hydrogen-Evolution Reaction," Chemistry of Materials 2016, 28 (17), 6017- 6044 ("Callejas 2016").

[0266] Callejas, J. F., et al, "Nanostructured Co ₂ P Electrocatalyst for the Hydrogen Evolution Reaction and Direct Comparison with Morphologically Equivalent CoP," Chem. Mater. 2015, 27 (10), 3769-3774 ("Callejas 2015").

[0267] Callejas, J. F., et al, "Electrocatalytic and Photocatalytic Hydrogen Production from Acidic and Neutral-pH Aqueous Solutions Using Iron Phosphide Nanoparticles," ACSNano 2014, 8 (11), 11101-11107 ("Callejas 2014").

[0268] Chae, S. Y., et al. "Improved photoelectrochemical water oxidation kinetics using a Ti0 ₂ nanorod array photoanode decorated with graphene oxide in a neutral pH solution," Physical Chemistry Chemical Physics 2015, 17 (12), 7714-7719 ("Chae 2015").

[0269] Chemelewski, W. D., et al. "Amorphous FeOOH oxygen evolution reaction catalyst for photoelectrochemical water splitting," Journal of the American Chemical Society 2014, 136 (7), 2843-2850 ("Chemelewski 2014").

[0270] Chen, Z., et al, "Three-dimensional flexible and conductive interconnected graphene networks grown by chemical vapour deposition," Nat. Mater., 2011, 10, 424-428 ("Chen 2011").

[0271] Coburn, J., "Sputtering in the surface analysis of solids: A discussion of some problems," Journal of Vacuum Science and Technology 1976, 13 (5), 1037-1044 ("Coburn 7976").

[0272] Colson, A. C, et al, "Synthesis of Phase - Pure Ferromagnetic Fe ₃ P Films from Single - Source Molecular Precursors," Advanced Functional Materials 2012, 22 (9), 1850-1855 ("Colson 2012").

[0273] Colson, A. C, et al., "Synthesis of Fe2-xMnxP Nanoparticles from Single-Source Molecular Precursors," Chemistry of Materials 2011, 23 (16), 3731-3739 ("Colson 2011").

[0274] Colson, A. C, et al, "Synthesis, Characterization, and Reactivity of the Heterometallic Dinuclear μ-ΡΗ ₂ and μ-ΡΡηΗ Complexes FeMn(CO) ₈ ^-PH ₂ ) and FeMn (∞) ₈ (μ-ΡΡ1ιΗ)," Organometallics 2010, 29 (20), 4611-4618 ("Colson 2010").

[0275] Cooper, G., et al., "Mott - Schottky Plots and Flatband Potentials for Single Crystal Rutile Electrodes," Journal of The Electrochemical Society 1982, 129 (9), 1973-1977 ("Cooper 1982"). [0276] Di Castro, V., et al, "XPS study of MnO oxidation," Journal of Electron Spectroscopy and Related Phenomena 1989, 48 (1), 117-123 ("Di Castro 1989").

[0277] Duffy, M. P.et al, "F. Reaction of Terminal Phosphinidene Complexes with Dihydrogen," Organometallics 2012, 31 (7), 2936-2939 ("Duffy 2012").

[0278] Dutta, A., et al. , "Developments on Metal Phosphides as Efficient OER Pre-catalysts," The Journal of Physical Chemistry Letters 2017 ("Dutta 201 ).

[0279] Fekete, M., et al. , "Highly active screen-printed electrocatalysts for water oxidation based on p-manganese oxide," Energy Environ Sci 2013, 6, 2222 ("Fekete 2013").

[0280] Friebel, D., et al, "Identification of highly active Fe sites in (Ni,Fe)OOH for electrocatalytic water splitting," Journal of the American Chemical Society 2015, 137 (3), 1305- 1313 ("Friebel 2015").

[0281] Gao, M. R., et al. , "Nitrogen-doped graphene supported CoSe ₂ nanobelt composite catalyst for efficient water oxidation," ACSNano 2014, 8 (4), 3970-3978 ("Gao 12014").

[0282] Gao, M., et al. "Efficient water oxidation using nanostructured a-nickel-hydroxide as an electrocatalyst," Journal of the American Chemical Society 2014, 136 (19), 7077-7084 ("Gao II

2014").

[0283] Helander, M., et al, "Work function of fluorine doped tin oxide," Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films 2011, 29 (1), 011019 ("Helander 2011").

[0284] Ho, P., et al., "Auger study of preferred sputtering on binary alloy surfaces," Surface Science 1976, 57 (1), 393-405 ("Ho 1916").

[0285] Hoang, S., et al. "Coincorporation of N and Ta into Ti0 ₂ nanowires for visible light driven photoelectrochemical water oxidation," The Journal of Physical Chemistry C 2012, 116 (44), 23283-23290 ("Hoang 2012").

[0286] Hofmann, S., Sputter depth profile analysis of interfaces," Reports on Progress in Physics 1998, 61 (7), 827 ^Hoffman 1998"). [0287] Hofmann, S., "Quantitative depth profiling in surface analysis: A review," Surface and Interface Analysis 1980, 2 (4), 148-160 ("Hoffman 1980").

[0288] Hunger, C, et al. "Stoichiometry-Controlled FeP Nanoparticles Synthesized from a Single Source Precursor," Chem. Commun. 2013, 49 (100), 11788-11790 ("Hunger 2013").

[0289] Huttner, G., et al, "Die Struktur von Bis(cyclopentadienyldicarhonylmangan)-1.2- Diphenyldiphosphan, [C ₅ H5(CO) ₂ Mn]2PhP(H)— (H)PPh," Z Fur Naturforschung B 2014, 31 (9), 1166-1169 ("Huttner 2014").

[0290] Huttner, G., et l., "R— P-Verbriickte Eien-Cluster-Hydride," J. Organomet. Chem. 1980, 191 (1), 161-169 ("Huttner 1980").

[0291] Huynh, M., et al., "A functionally stable manganese oxide oxygen evolution catalyst in acid," Journal of the American Chemical Society 2014, 136 (16), 6002-6010 ("Huynh 2014").

[0292] Hwang, Y., et al., "Photoelectrochemical Properties of Ti0 ₂ Nanowire Arrays: A Study of the Dependence on Length and Atomic Layer Deposition Coating," ACSNano 2012, 6 (6), 5060- 5069 ("Hwang 2012").

[0293] Jiang, P., et al., "A cost-effective 3D hydrogen evolution cathode with high catalytic activity: FeP nanowire array as the active phase," Angew Chem Int Ed Engl 2014, 53 (47), 12855- 9 ("Jiang 2014").

[0294] Jones, P. G., et al., "Oxidative Kniipfung Einer Phosphor-Phosphor-Bindung Unter Einwirkung von Ag(I)- Bzw. Cu(II)-Ionen: Synthese Und Struktur von [(C ₆ Hs)PH ₂ Ag^- (C6H ₅ PH) ₂ }]2(AsF6) ₂ , Einem Sechsgliedrigen Silber-Phosphor-Ring / The Oxidative Formation of a Phosphorus-Phosphorus Bond in the Presence of Ag(I) and Cu(II) Ions: Synthesis and Structure of [(C ₆ H ₅ )PH ₂ Ag^-(C ₆ H ₅ PH) ₂ }] ₂ (AsF ₆ ) ₂ , Containing a Six-Membered Silver- Phosphorus Ring," Z. Fur Naturforschung 52014, 40 (5), 590-593 ("Jones 2014"). [0295] Kibsgaard, J., et al, "Designing an improved transition metal phosphide catalyst for hydrogen evolution using experimental and theoretical trends. Energy Environ. Sci. 2015, 8 (10), 3022-3029 ("Kibsgaard 2015").

[0296] Kuo, C.-H., et al. "Facet-dependent catalytic activity of MnO electrocatalysts for oxygen reduction and oxygen evolution reactions," Chemical Communications 2015, 51 (27) 5951-5954 ("Kuo 2015").

[0297] Leitner, A. P., et al., "Thin Films of (Fei- _x Co _x ) ₃ P and from the Co- Decomposition of Organometallic Precursors by MOCVD," Chemistry of Materials 2016, 28 (19), 7066-7071 ("Leitner 201$

[0298] Li, D. , et al, "Efficient Water Oxidation Using CoMnP Nanoparticles," J Am Chem Soc 2016, 138 (12), 4006-4009 ("D. U 2016").

[0299] Li, J., et al, "Semiconductor-based photocatalysts and photoelectrochemical cells for solar fuel generation: a review," Catalysis Science & Technology 2015 ("J. Li 2015").

[0300] Li, 1, et al, "Plasmon-induced photonic and energy-transfer enhancement of solar water splitting by a hematite nanorod array. Nat Commun 2013, 4, 2651 ("J. Li 2013").

[0301] Li, Q., et al, "Nanocarbon electrocatalysts for oxygen reduction in alkaline media for advanced energy conversion and storage," Advanced Energy Materials 2014, 4 (6) ("Q. Li 2014").

[0302] Li, Y., et al. "Efficient and stable photoelectrochemical seawater splitting with Ti0 ₂ @g-

C ₃ N ₄ nanorod arrays decorated by Co-Pi," The Journal of Physical Chemistry C 2015, 119 (35),

20283-20292 ("7. Li 2015").

[0303] Liu, M., et al, "Fe ₂ P/Reduced Graphene Oxide/Fe ₂ P Sandwich-Structured Nanowall Arrays: A High-Performance Non-Noble-Metal Electrocatalyst for Hydrogen Evolution," J. Mater. Chem. A 2017, 5 (18), 8608-8615 ("Liu 201Γ). [0304] Lv, C, et al, "The Hierarchical Nanowires Array of Iron Phosphide Integrated on a Carbon Fiber Paper as an Effective Electrocatalyst for Hydrogen Generation," J. Mater. Chem. A 2016, 4 (4), 1454-1460 ("Lv 2016").

[0305] Ma, T., et al, "Metal-Organic Framework Derived Hybrid Co ₃ 0 ₄ -Carbon Porous Nanowire Arrays as Reversible Oxygen Evolution Electrodes," Journal of the American Chemical Society 2014, 736 (39), 13925-13931 ("Ma 2014").

[0306] Mahan, J. E., et al., "Trends in sputter yield data in the film deposition regime," Physical Review B 2000, 61 (12), 8516 ("Mahan 2000").

[0307] McCrory, C. C, et al., "Benchmarking heterogeneous electrocatalysts for the oxygen evolution reaction. Journal of the American Chemical Society 2013, 135 (45), 16977-16987 ("McCrory 2013").

[0308] Meng, Y, et al, "Structure-Property Relationship of Bifunctional Mn0 ₂ Nanostructures: Highly Efficient, Ultra-Stable Electrochemical Water Oxidation and Oxygen Reduction Reaction Catalysts Identified in Alkaline Media. J Am Chem Soc 2014, 136 (32), 11452-11464 ("Meng 2014").

[0309] Miiller, M., et al. "Schrittweiser Aufbau von μ3-ΡΟ ^, -Τππΐ6ΐ3ΐ1-αυ8ί6πι TJber P-H- Verbindungen," Chem. Ber. 1983, 116 (6), 2311-2321 ("Miiller 1983").

[0310] Ning, E., et al. "Ti0 ₂ /graphene/NiFe-layered double hydroxide nanorod array photoanodes for efficient photoelectrochemical water splitting," Energy & Environmental Science 2016, 9 (8), 2633-2643 ("Ning 2016").

[0311] Paparazzo, E., "XPS and auger spectroscopy studies on mixtures of the oxides Si0 ₂ , A1 ₂ 0 ₃ , Fe ₂ 0 ₃ and Cr ₂ 0 ₃ . Journal of Electron Spectroscopy and Related Phenomena 1987, 43 (2), 97-112 ^Paparazzo 198Γ).

[0312] Parker, R. A., "Static Dielectric Constant of Rutile (Ti0 ₂ ), 1.6-1060 K," Physical Review 1961, 124 (6), 1719 ("Parker 1961"). [0313] Pei, Z., et al. "Dramatically improved energy conversion and storage efficiencies by simultaneously enhancing charge transfer and creating active sites in MnO _x /Ti0 ₂ nanotube composite electrodes," Nano Energy 2016, 20, 254-263 ("Pei 2016").

[0314] Popczun, E. J., et al, "Nanostructured Nickel Phosphide as an Electrocatalyst for the

Hydrogen Evolution Reaction," J. Am. Chem. Soc. 2013, 135 (25), 9267-9270 ("Popczun 2013").

[0315] Powell, C, "Recommended Auger parameters for 42 elemental solids," Journal of

Electron Spectroscopy and Related Phenomena 2012, 185 (1), 1-3 ("Powell 2012").

[0316] Ramirez, A., et al, "Evaluation of MnO _x , Mn ₂ 0 ₃ , and Mn ₃ 0 ₄ Electrodeposited Films for the Oxygen Evolution Reaction of Water," JPhys Chem C2014, 118 (26), 14073-14081 ("Ramirez

2014").

[0317] Ran, J., et al., "Earth-abundant cocatalysts for semiconductor-based photocatalytic water splitting," Chemical Society Reviews 2014 ("Ran 2014").

[0318] Read, C. G., et al, "General strategy for the synthesis of transition metal phosphide films for electrocatalytic hydrogen and oxygen evolution," ACS applied materials & interfaces 2016, 8 (20), 12798-12803 ("Read 2016").

[0319] Schipper, D. E., et al., "Transformations in Transition-Metal Carbonyls Containing Arsenic: Exploring the Chemistry of [Et4N] ₂ [HAs{Fe(CO)4} ₃ ] in the Search for Single-Source Precursors for Advanced Metal Pnictide Materials," Organometallics 2016, 35 (4), 471-483 ("Schipper 2016").

[0320] Shaner, M. R., et al., "A comparative technoeconomic analysis of renewable hydrogen production using solar energy," Energy & Environmental Science 2016, 9 (7), 2354-2371 ("Shaner 2016").

[0321] Shi, Y., et al., "Recent advances in transition metal phosphide nanomaterials: synthesis and applications in hydrogen evolution reaction," Chem Soc Rev 2016, 45 (6), 1529-41 ("Shi 2016"). [0322] Singh, N., et al, "Electrocatalytic properties of spinel-type Mn _x Fe3 _-x 04 synthesized below 100 C for oxygen evolution in KOH solutions," Journal of the Chemical Society, Faraday Transactions 1996, 92 (13), 2397-2400 ("Singh 7996").

[0323] Son, C. Y., et al, "FeP and FeP ₂ Nanowires for Efficient Electrocatalytic Hydrogen Evolution Reaction," Chem. Commun. 2016, 52 (13), 2819-2822 ("Son 2016").

[0324] Song, E., et al. "Exfoliation of layered double hydroxides for enhanced oxygen evolution catalysis," Nature communications 2014, 5 ("Song 2014").

[0325] Stem, L.-A. , et al., "Ni ₂ P as a Janus catalyst for water splitting: the oxygen evolution activity of N12P nanoparticles," Energy & Environmental Science 2015, 8 (8), 2347-2351 ("Stem 2015").

[0326] Su, J., et al. "Ultrasmall graphitic carbon nitride quantum dots decorated self-organized Ti0 ₂ nanotube arrays with highly efficient photoelectrochemical activity," Applied Catalysis B: Environmental 2016, 186, 127-135 ("Su 2016").

[0327] Sun, K., et al., "A Stabilized, intrinsically Safe, 10% Efficient, Solar - Driven Water - Splitting Cell Incorporating Earth - Abundant Electrocatalysts with Steady - State pH Gradients and Product Separation Enabled by a Bipolar Membrane," Advanced Energy Materials 2016, 6 (13) ("Sun 2016").

[0328] Sun, K., et al. , "Nickel oxide functionalized silicon for efficient photo-oxidation of water," Energy & Environmental Science 2012, 5 (7), 7872-7877 ("Sun 2012").

[0329] Tan, B. J., et al., "XPS studies of solvated metal atom dispersed (SMAD) catalysts. Evidence for layered cobalt-manganese particles on alumina and silica," Journal of the American Chemical Society 1991, 113 (3), 855-861 ("B. Tan 1991").

[0330] Tan, Y., et al., "Versatile nanoporous bimetallic phosphides towards electrochemical water splitting," Energy Environ. Sci. 2016, 9 (7), 2257-2261 ("7. Tan 2016").

[0331] Tang, C, et al., "Fe-Doped CoP Nanoarray: A Monolithic Multifunctional Catalyst for Highly Efficient Hydrogen Generation," Adv Mater 2016 ("Tang 2016").

[0332] Tian, J., et al. , "Self-Supported Cu ₃ P Nanowire Arrays as an Integrated High-Performance Three-Dimensional Cathode for Generating Hydrogen from Water," Angew. Chem. Int. Ed. 2014, 53 (36), 9577-9581 ("Tian 2015").

[0333] Tian, L., et al., "Electrochemical Activity of Iron Phosphide Nanoparticles in Hydrogen Evolution Reaction/ CS Catal. 2016, 6 (8), 5441-5448 ("Tian 2016").

[0334] Tian, R., et al. "Simple Access to Tungsten-Stabilized Disecondary Diphosphines," Organometallics 2013, 32 (19), 5615-5618 ("Tian 2013").

[0335] Tuysiiz, H., et al., "Mesoporous Co ₃ 0 ₄ as an electrocatalyst for water oxidation," Nano Research ing 6, 47-54 ("Tuysuz 2012").

[0336] Wang, C, et al, "Monodisperse Ternary NiCoP Nanostructures as a Bifunctional Electrocatalyst for Both Hydrogen and Oxygen Evolution Reactions with Excellent Performance," Advanced Materials Interfaces 2016, 3 (4), 1500454 ("C. Wang 2016").

[0337] Wang, X.-D., et al, "Novel porous molybdenum tungsten phosphide hybrid nanosheets on carbon cloth for efficient hydrogen evolution," Energy Environ. Sci. 2016, 9 (4), 1468-1475 ("X. Wang 2016").

[0338] Wang, Z., et al. "Synergetic Effect of Conjugated Ni(OH) ₂ Ir0 ₂ Cocatalyst on Titanium- Doped Hematite Photoanode for Solar Water Splitting, The Journal of Physical Chemistry C 2015, 119 (34), 19607-19612 ("Z Wang 2015").

[0339] Werner, H., et al., "The influence of selective sputtering on surface composition," Surface Science 1976, 57 (2), 706-714 ("Werner 1976").

[0340] Whitmire, K. H., "Main group-transition metal cluster compounds of the group 15 elements," Advances in organometallic chemistry 1998, 42, 2-146 'Whitmire 1998°).

[0341] Xiang, C, et al, "Principles and implementations of electrolysis systems for water splitting," Materials Horizons 2016, 3 (3), 169-173 ("Xiang 2016"). [0342] Xiao, P., et al, "A Review of Phosphide - Based Materials for Electrocatalytic Hydrogen Evolution," Advanced Energy Materials 2015, 5 (24) ("Xiao 2015").

[0343] 27. Xu, D., et al, "Zn-Co layered double hydroxide modified hematite photoanode for enhanced photoelectrochemical water splitting," Applied Surface Science 2015, 358, 436-442 ("Xu 2015").

[0344] Yan, P., et al. "Photoelectrochemical Water Splitting Promoted with a Disordered Surface Layer Created by Electrochemical Reduction," ACS applied materials & interfaces 2015, 7 (6), 3791-3796 ("P. Yan 2015").

[0345] Yan, Y.; Xia, B. Y.; Ge, X.; Liu, Z.; Fisher, A.; Wang, X. A Flexible Electrode Based on Iron Phosphide Nanotubes for Overall Water Splitting. Chem. - Eur. J. 2015, 21 (50), 18062- 18067 ("7. Yan 201 J").

[0346] Yang, J., et al, "Roles of Cocatalysts in Photocatalysis and Photoelectrocatalysis," Accounts of Chemical Research 2013, 46, 1900-1909 ("J. Yang 2013").

[0347] Yang, W., et al "Ferroelectric polarization-enhanced photoelectrochemical water splitting in Ti0 ₂ -BaTi0 ₃ core-shell nanowire photoanodes," Nano letters 2015, 15 (11), 7574-7580 ("W. Yang 2015").

[0348] Yang, X., et al, "Rugae-like FeP nanocrystal assembly on a carbon cloth: an exceptionally efficient and stable cathode for hydrogen evolution, Nanoscale 2015, 7 (25), 10974-81 ("X. Yang 2015").

[0349] Yu, J., et al, "Ternary Metal Phosphide with Triple-Layered Structure as a Low-Cost and Efficient Electrocatalyst for Bifunctional Water Splitting," Advanced Functional Materials 2016, 26 (42), 7644-7651 ("J. Yu 2016").

[0350] Yu, X., et al, "Hollow CoP nanoparticle/N-doped graphene hybrids as highly active and stable bifunctional catalysts for full water splitting," Nanoscale 2016, 8 (21), 10902-10907 ("X. Yu 2016"). [0351] Zhang, J., et al. "A metal-free bifunctional electrocatalyst for oxygen reduction and oxygen evolution reactions," Nature nanotechnology 2015, 10 (5), 444-452 ("J. Zhang 2015").

[0352] Zhang, Y., et al., "Unique Fe ₂ P Nanoparticles Enveloped in Sandwichlike Graphited Carbon Sheets as Excellent Hydrogen Evolution Reaction Catalyst and Lithium-Ion Battery Anode," ACS Appl. Mater. Interfaces 2015, 7 (48), 26684-26690 ("Y. Zhang 2015").

[0353] Zhang, Z., et al., "FeP nanoparticles grown on graphene sheets as highly active non- precious-metal electrocatalysts for hydrogen evolution reaction," Chem Commun (Camb) 2014, 50 (78), 11554-7 ("Z. Zhang 2014").

[0354] Zhao, Y., et al. "Nitrogen-doped carbon nanomaterials as non-metal electrocatalysts for water oxidation," Nature communications 2013, 4 ("Zhao 2013").

[0355] Zhong, D. K., et al., "Photoelectrochemical Water Oxidation by Cobalt Catalyst ("Co - Pi")/a-Fe ₂ 0 ₃ Composite Photoanodes: Oxygen Evolution and Resolution of a Kinetic Bottleneck," Journal of the American Chemical Society 2010, 732 (12), 4202-4207 ("Zhong 2010").

[0356] Zhu, Y.-P., et al., "Self-Supported Cobalt Phosphide Mesoporous Nanorod Arrays: A Flexible and Bifunctional Electrode for Highly Active Electrocatalytic Water Reduction and Oxidation," Advanced Functional Materials 2015, 25 (47), 7337-7347 ("Zhu 2015").