Abstract
The utilization of solar energy on a large scale requires its storage. In natural photosynthesis, energy from sunlight is used to rearrange the bonds of water to oxygen and hydrogen equivalents. The realization of artificial systems that perform “water splitting” requires catalysts that produce oxygen from water without the need for excessive driving potentials. Here we report such a catalyst that forms upon the oxidative polarization of an inert indium tin oxide electrode in phosphate-buffered water containing cobalt (II) ions. A variety of analytical techniques indicates the presence of phosphate in an approximate 1:2 ratio with cobalt in this material. The pH dependence of the catalytic activity also implicates the hydrogen phosphate ion as the proton acceptor in the oxygen-producing reaction. This catalyst not only forms in situ from earth-abundant materials but also operates in neutral water under ambient conditions.
Sunlight is the only renewable and carbon-neutral energy source of sufficient scale to replace fossil fuels and meet rising global energy demand (1). The diurnal variation in local insolation, however, demands a cost-effective storage of solar energy for its large-scale utilization. Of the possible storage methods, nature provides the blueprint for storing sunlight in the form of chemical fuels (1, 2). The primary steps of natural photosynthesis involve the absorption of sunlight and its conversion into spatially separated electron/hole pairs. The holes of this wireless current are then captured by the oxygen-evolving complex (OEC) to oxidize water to oxygen and the electrons are captured by photosystem I to reduce NADP+ (nicotinamide adenine dinucleotide phosphate) to NADPH (the reduced form of NADP+), nature's form of hydrogen (3). Thus, the overall primary events of photosynthesis store solar energy in a fuel by rearranging the chemical bonds of water to form H2 (i.e., NADPH) and O2.
An approach to duplicating photosynthesis outside of a photosynthetic membrane is to convert sunlight into spatially separated electron/hole pairs within a photovoltaic cell and then capture the charges with catalysts that mediate “water splitting” (1, 4). The four holes are captured by a catalyst at the anode to produce oxygen, and the four electrons are captured by a separate catalyst at the cathode to produce hydrogen. The net result is the storage of solar energy in the chemical bonds of H2 and O2.
A key determinant of energy storage in artificial photosynthesis is the efficiency of the water-splitting catalysts. Electrocatalysts that are efficient for solar-to-fuels conversion must operate close to the Nernstian potentials (E) for the H2O/O2 and H2O/H2 half-cell reactions shown in Scheme 1 (half-cell potentials given in the convention of reduction potentials).
The voltage in addition to E that is required to attain a given catalytic activity—referred to as overpotential—limits the efficiency of converting light into catalytic current. Of the two reactions, the H2O/O2 reaction is considerably more complex (5). This reaction requires a four-electron oxidation of two water molecules coupled to the removal of four protons to form a relatively weak oxygen-oxygen bond. In addition to controlling this proton-coupled electron transfer (PCET) (6, 7), a catalyst must tolerate prolonged exposure to oxidizing conditions. Even at the thermodynamic limit, water oxidation requires an oxidizing power that causes most chemical functional groups to degrade. Accordingly, the generation of oxygen from water presents a substantial challenge toward realizing artificial photosynthesis (8).
The fine-tuned molecular machinery of the OEC oxidizes water at a low overpotential using a Mn4O4Ca cluster (9–12). Outside the OEC, examples of water oxidation catalysts include first-row spinel and perovskite metal oxides, which require concentrated basic solutions (pH > 13) and moderate overpotentials (<400 mV), and precious metals and precious metal oxides, which operate with similar efficiencies under acidic conditions (pH < 1) (13–15). However, few catalysts operate in neutral water under ambient conditions. Neutral water is oxidized at Pt electrodes, and some precious metal oxides have been reported to operate electrocatalytically in neutral or weakly acidic solutions (16). The development of an earth-abundant, first-row catalyst that operates at pH 7 at low overpotential remains a fundamental chemical challenge. Here, we report an oxygen-evolving catalyst that forms in situ upon anodic polarization of an inert electrode in neutral aqueous phosphate solutions containing Co2+. Oxygen generation occurs under benign conditions: pH = 7, 1 atm, and room temperature.
Cobalt ions in the presence of chemical oxidants such as Ru(bpy)33+ (bpy, bipyridine; Eo = 1.26, where Eo is the standard potential) catalyze the oxidation of water to O2 in neutral phosphate solutions (17, 18). Oxygen yields drop in these reactions when oxidized Co species precipitate from solution because the catalytically active species is removed from the solution-phase reaction. However, an oxidation-induced precipitation may be exploited to prepare electrocatalysts in situ if the precipitated material remains catalytically active and can be oxidized at an electrode surface. To explore this possibility for Co-catalyzed water oxidation, we examined electrochemical oxidations of aqueous solutions containing phosphate and Co2+. Cyclic voltammetry of a 0.5 mM solution of Co(NO3)2 in 0.1 M potassium phosphate pH 7.0 (KPi electrolyte) exhibits an oxidation wave at Ep = 1.13 V (where Ep is the peak potential) versus the normal hydrogen electrode (NHE), followed by the onset of a strong catalytic wave at 1.23 V (Fig. 1A). A broad, relatively weak reduction wave is observed on the cathodic scan. The presence of a catalytic wave prompted us to examine the electrode activity during controlled-potential electrolysis.
(A) Cyclic voltammagram in 0.1 M KPi electrolyte at pH 7.0 with no Co2+ ion present (black line) and with 0.5 mM Co2+ present (red line). The potential was measured against a Ag/AgCl reference and converted to NHE potentials by using E(NHE) = E(Ag/AgCl) + 0.197 V. (B) Current density profile for bulk electrolysis at 1.29 V (versus NHE) in 0.1 M KPi electrolyte at pH 7.0 containing 0.5 mM Co2+. (Inset) Profile in the absence of Co2+.
Indium tin oxide (ITO) was used as the electrode for bulk electrolysis to ensure a minimal background activity for O2 production. An electrolysis at 1.29 V without stirring in neutral KPi electrolyte containing 0.5 mM Co2+ exhibits a rising current density that reaches a peak value >1 mA/cm2 after 7 to 8 hours (Fig. 1B). During this time, a dark coating forms on the ITO surface, and effervescence from this coating becomes increasingly vigorous (19). The same results are observed with either CoSO4, Co(NO3)2, or Co(OTf)2 (where OTf = triflate) as the Co2+ source, which indicates that the original Co2+ counterion is unimportant and that this activity does not depend on an impurity found in a specific source. The amount of charge passed during the course of an 8-hour electrolysis far exceeds what could be accounted for by stoichiometric oxidation of the Co2+ in solution (20). These observations are indicative of the in situ formation of an oxygen-evolving catalyst. Catalyst formation also proceeds on a fluorine tin oxide electrode and if KPi is replaced by NaPi electrolyte. In a control experiment, the current density during bulk electrolysis under identical conditions in the absence of Co2+ rapidly drops to a baseline level of ∼25 nA/cm2 (inset in Fig. 1B).
The morphology of the electrode coating formed during electrolysis in the presence of Co2+ was examined by scanning electron microscopy (SEM). The electrodeposited material consists of particles that have coalesced into a thin film and individual micrometer-sized particles on top of the film (Fig. 2A). The ITO substrate can be seen through cracks in the film that form upon drying, as evidenced by particles that are split into complementary pieces. The film thickness gradually increases over the course of the electrodeposition (see fig. S4 for additional images). At maximum activity under these electrolysis conditions, the film is >2 μm thick. The x-ray powder diffraction pattern of an electrodeposited catalyst shows broad amorphous features and no peaks indicative of crystalline phases other than the peaks associated with the ITO layer (fig. S1).
(A) SEM image (30° tilt) of the electrodeposited catalyst after 30 C/cm2 were passed in 0.1 M KPi electrolyte at pH 7.0, containing 0.5 mM Co2+. The ITO substrate can be seen through cracks in the dried film. (B) Typical EDX histogram acquired at 12 kV. cps, counts per second.
In the absence of detectable crystallites, the composition of the electrodeposited material was analyzed by three complementary techniques. Energy-dispersive x-ray analysis (EDX) spectra were obtained from multiple 100-to-300–μm2 regions of several independently prepared samples. These spectra identify Co, P, K, and O as the principal elemental components of the material (Fig. 2B). Although the material's morphology is not ideally suited for quantitative EDX, the analyses consistently indicate a Co:P:K ratio between ∼2:1:1 and 3:1:1. To obtain an independent determination of elemental composition, electrolysis was performed with several larger ITO electrodes; the deposited material was scraped off and combined for a total yield of ∼3 mg. Microanalytical elemental analysis of the combined material indicates 31.1% Co, 7.70% P, and 7.71% K, corresponding to a 2.1:1.0:0.8 Co:P:K ratio. Finally, the surface of an electrodeposited catalyst on the ITO substrate was analyzed by x-ray photoelectron spectroscopy (XPS). All peaks in the XPS spectra are accounted for by the elements detected above, in addition to In and Sn from the ITO substrate. The high-resolution P 2p peak at 133.1 eV is consistent with phosphate. The Co 2p peaks at 780.7 and 795.7 eV are in a range typical of Co2+ or Co3+ bound to oxygen (fig. S2) (21). Together, the x-ray diffraction and analytical results indicate that electrolysis of a Co2+ solution in neutral KPi electrolyte results in the electrodeposition of an amorphous Co oxide or hydroxide incorporating a substantial amount of phosphate anion at a stoichiometric ratio of roughly 2:1:1 for Co:P:K.
Three experiments were performed to establish that the catalytic activity observed with this material corresponds to authentic water oxidation. Each of these experiments was performed in neutral KPi electrolyte in the absence of Co2+. Catalyst coatings (∼1.3 cm2) were prepared in a preliminary step as described above and stored under ambient laboratory conditions until they were used.
To confirm that water is the source of the O2 produced, electrolysis was performed in helium-saturated electrolyte containing 14.6% 18OH2 in a gas-tight electrochemical cell in line with a mass spectrometer. The helium carrier gas was continuously flowed through the headspace of the anodic compartment into the mass spectrometer, and the relative abundances of 32O2, 34O2, and 36O2 were monitored at 2-s intervals. Within minutes of initiating electrolysis at 1.29 V, the signals for the three isotopes began to rise above their background levels as the O2 produced by the catalyst escaped into the headspace. Upon terminating the electrolysis 1 hour later, these signals slowly returned to their background levels (Fig. 3A). The 32O2, 34O2, and 36O2 isotopes were detected in the statistical ratio (72.9, 24.9, and 2.1% relative abundances, respectively) (Fig. 3B).
(A) Mass spectrometric detection of isotopically labeled 16,16O2 (black line), 16,18O2 (blue line), and 18,18O2 (red line) during electrolysis of a catalyst film on ITO in a KPi electrolyte containing 14.6% 18OH2. The green arrow indicates the initiation of electrolysis at 1.29 V (NHE), and the red arrow indicates the termination of electrolysis. (Inset) Expansion of the 18,18O2 signal. (B) Percent abundance of each isotope over the course of the experiment. Average observed abundance of ±2σ is indicated above each line. Statistical abundances are 72.9, 24.9, and 2.1%. (C) O2 production measured by fluorescent sensor (red line) and theoretical amount of O2 produced (blue line), assuming a Faradic efficiency of 100%. The green arrow indicates the initiation of electrolysis at 1.29 V, and the red arrow indicates the termination of electrolysis.
The Faradaic efficiency of the catalyst was measured with a fluorescence-based O2 sensor. Electrolysis was performed in KPi electrolyte in a gas-tight electrochemical cell under an N2 atmosphere with the sensor placed in the headspace. After initiating electrolysis at 1.29 V, the percentage of O2 detected in the headspace rose in accord with what was predicted by assuming that all of the current was caused by 4e– oxidation of water to produce O2 (Fig. 3C). The amount of O2 produced (95 μmol, 3.0 mg) greatly exceeds the amount of catalyst (∼0.2 mg), which shows no perceptible decomposition during the course of the experiment.
The stability of phosphate under catalytic conditions was assayed by 31P nuclear magnetic resonance (NMR). Electrolysis in a two-compartment cell with 10 mL of KPi electrolyte (1 mmol phosphate) on each side was allowed to proceed until 45 C had been passed through the cell (0.46 mmol electrons). Electrolysis solutions from both chambers show single, clean 31P resonances, which indicate that the electrolyte is robust under these conditions (fig. S3). Together, the mass spectrometry, Faradaic efficiency, and 31P NMR results demonstrate that the electrodeposited catalyst cleanly oxidizes H2O to O2 in neutral KPi solutions.
The current density of a catalyst on ITO was measured as a function of the overpotential (η) in KPi electrolyte without Co2+ (black circles in Fig. 4A). At pH 7.0, appreciable catalytic current is observed beginning at η = 0.28 V, and a current density of 1 mA/cm2 (corresponding to 9 μmol O2 cm–2 h–1) requires η = 0.41 V. The Tafel plot deviates slightly from linearity, possibly reflecting an uncompensated iR drop caused by the surface resistivity of the ITO (8 to 12 ohms per square). Substantial improvements in the activity profile may be attainable without changing the catalyst composition by depositing on alternative substrates or improving ohmic contact to the ITO.
(A) Tafel plot (black circles), η = (Vappl – iR) – E(pH 7) (where Vappl is the applied potential), of a catalyst film on ITO in 0.1 M KPi electrolyte pH 7.0, corrected for the iR drop of the solution. pH data were converted into a Tafel plot (red circles), η = (Vappl + 0.059ΔpH – iR) – E(pH 7), assuming Nernstian behavior and correcting for the iR drop of the solution. The pH = 5 and pH = 8 data points are indicated by arrows. (B) Current density dependence on pH in 0.1 M KPi electrolyte. The potential was set at 1.24 V (versus NHE) with no iR compensation.
The catalyst used to obtain the Tafel plot at pH 7 was subsequently transferred to KPi electrolyte at pH 4.6, and the current density was measured at a constant applied potential (1.24 V) while the pH was increased incrementally to 9.4 by adding aliquots of concentrated KOH. A plot of the log of current density versus pH exhibits a steep initial rise that levels off in the high-pH range such that increasing the pH from 8 to 9.4 at this applied potential has little effect (Fig. 4B). These data can be converted to a Tafel plot by using Eq. 1 (Scheme 1) and accounting for iR drop (see Fig. 4 legend). A comparison to the Tafel plot obtained at pH 7 indicates that the catalyst exhibits approximately Nernstian behavior from pH 5 to 8: Increasing the pH by one unit at constant applied potential (1.24 V) has nearly the same effect as increasing the overpotential by 0.059 V at pH 7 (red circles in Fig. 4A). This result implicates a reversible ne–, nH+ removal before the rate-determining step for O2 evolution in this pH range (here, n is the number of equivalents). Thus, an important component of the activity at pH 7 with this catalyst is the existence of one or more intermediates preceding O2 formation that are deprotonated reversibly by HPO42– in a PCET event (22). The pH-independent behavior above pH 8 at the applied potential may indicate a change in mechanism, most likely involving a deprotonated intermediate.
In addition to mediating the deprotonation required for catalysis, the KPi electrolyte provides a medium for in situ catalyst formation. Given that phosphate is a structural element and that the catalyst forms under oxidizing conditions, it is plausible that deposition is driven by the interaction of phosphate and Co3+. By judicious choice of other metal-anion pairs or combinations of multiple metals and anions, it may be possible to access other oxygen-evolving catalysts that form in situ and operate in neutral solutions. In situ formation is advantageous because, in principle, it enables catalyst deposition on a variety of substrates, including those that are too delicate to tolerate traditional catalyst preparation techniques. This attribute is important for interfacing a catalyst with a variety of electrochemical or photoelectrochemical cell designs.
In situ formation also implies a self-repair mechanism. Proposed molecular mechanisms involving O2/H2O cycles at Co centers suggest that catalytic reactions cycle among Co2+-, Co3+-, and Co4+-oxo oxidation states (18, 23). The propensity of metal ion dissolution has been shown to correlate with ligand substitution (24). Given that Co3+ is substitutionally inert relative to Co2+, a dynamic equilibrium between Co2+-HPO42– in solution and Co3+-HPO42– on the anodically poised electrode may be established. More generally, if a catalytic cycle involves an oxidation state that is prone to dissolution, this process can be countered by continual catalyst formation by establishing an equilibrium with the judicious choice of an anion.
The results reported herein highlight a new area of exploration for the development of easily prepared, earth-abundant catalysts that oxidize water. If artificial photosynthesis is to enable the storage of solar energy commensurate with global demand, water-splitting chemistry will need to be performed at a daunting scale. Storing the equivalent of the current energy demand would require splitting more than 1015 mol/year of water, which is roughly 100 times the scale of nitrogen fixation by the Haber-Bosch process. The conditions under which water splitting is performed will determine how solar energy is deployed. The catalyst reported here has many elements of natural photosynthesis, including (i) its formation from earth-abundant metal ions in aqueous solution, (ii) a plausible pathway for self-repair, (iii) a carrier for protons in neutral water, and (iv) the generation of O2 at low overpotential, neutral pH, 1 atm, and room temperature.
Supporting Online Material
www.sciencemag.org/cgi/content/full/1162018/DC1
Materials and Methods
Figs. S1 to S4
Movies S1 and S2
