WIXLIB

The Mechanism of Water OxidationIn the future, water electrolysis could play a key role in sustainable energy conversion and storage infrastructure.[2] Firstobserved in 1789 in a transient discharge experiment by vanTroostwijk,[3] then possibly observed by Volta using one of hispiles[4] and finally made known to a wider audience in 1800 byNicolson and Carlisle,[5] it largely remained a scientific curiositythroughout the 19th century. During the early decades of the20th century, water electrolysis was industrialized and served toproduce very pure hydrogen for industrial uses, such as ammonia production or early petroleum refining.[6] Later in the20th century, materials and system improvements resulted insignificant increases in net efficiencies of electrolyzers above50 %. Alkaline water electrolysis, where hydroxide anions arethe major charge carrier in the liquid ion-conducting electrolyte, became a technology of choice for production of oxygenand hydrogen for life-support systems in anaerobic environments, such as submarines. The use of separator materials (diaphragms) aided in the design of more-compact electrolyzercells with improved separation of hydrogen and oxygen gas.[7]Later on, high-pressure alkaline electrolyzers were designedand commercialized. Work by Beer resulted in the technologically important dimensionally stable anode (DSA),[8] which remains the basis of electrolyzer anodes to date. In the DSA, catalysts are supported directly on metallic titanium as a thin film.Space exploration drove the development of polymer electrolyte membrane (PEM) electrolyzers for acidic media thatshowed very similar system architectures as hydrogen PEMfuel cells (PEMFCs).[9] Membrane materials such as perfluorosulfonic acid (Nafion ) allow the conduction of protons across athin polymer electrolyte layer. Advantages of proton-conducting PEM devices over liquid alkaline systems include highersafety and reliability, because no caustic soda is circulated inthe cell, greater energy efficiency, sustainability to high differential gas pressures, higher hydrogen production rates, and amore-compact cell design.[9a] Drawbacks relate to the use ofmore expensive noble metal anode catalyst materials, such asRu or Ir. Current developments in the field of electrolysis include the coupling of water electrolysis with photovoltaic solarenergy conversion[10] into an integrated direct photoelectrocatalytic (PEC) water-splitting technology.[11] PEC devices promisemore-compact electrode and possibly device architectures, yetrequire multifunctional hybrid electrode materials. Low solarconversion efficiencies combined with electrocatalytic overpotential losses largely limit the efficiencies of today’s PEC devices to below 5 %. Another current development centers on thesynthesis of practical alkaline polymer electrolyte membranes.This technology would combine the advantages of the PEMtechnology with the benefits of a mature alkaline electrolyzerindustry, in particular with regards to the use of less expensivenon-noble metal electrode materials. Finally, the integration ofelectrolyzer and fuel-cell functionalities into one single unitizedregenerative fuel-cell device is the subject of current researchand development efforts.[12] A regenerative fuel cell/electrolyzersystem offers the prospect of a two-way conversion of chemical and electrical energy using bifunctional electrodes with theadvantage of further increased system simplicity and compactness.ChemCatChem 2010, 2, 724 – 761Today, water electrolysis is frequently cited as a clean cornerstone production technology in a largely hydrogen-based sustainable energy infrastructure (hydrogen economy[13]). Electrolysis could serve to liberate hydrogen from a range of chemicals, not only from water. Also, being a scalable technology,electrolysis would allow the initial deployment of smaller-scaleelectrolysis units at comparable efficiencies to large-scale electrolyzers.[13b–d] The vision of a broader hydrogen-based energyeconomy, however, still faces a large number of serious technical and nontechnical road blocks.[13d] Among the scientific andtechnical challenges, the need for advanced materials is criticaland will likely require a large amount of additional basicenergy materials and catalysis research.Hydrogen is an important industrial feedstock for petroleumrefining and fertilizer industry. However, electrolysis accountsonly for about 4 % of global hydrogen production[9b, 13b–d, 14] .The recent surge of interest and investments in renewableelectricity through solar–thermal, solar–electric (photovoltaic),or wind farms, combined with the emerging interest in electromobility concepts for urban centers, has raised a substantialamount of awareness about pressurized electrolytic hydrogenfor storage of excess renewable electricity. As a results, hydrogen–wind hybrid power plants are currently being commercialized at a number of locations worldwide.[15] On a smaller scale,more and more consumer wind and photovoltaic units havebeen deployed in accordance with energy policies that offergenerous subsidies for renewable electricity fed back into thegrid. Once these subsidies are cut back and grid electricityprices continue to rise, a growing interest in home energy storage units, possibly based on water electrolysis and compressedhydrogen, will result.The overall electrochemical water splitting is divided intotwo half-cell redox reactions. The reduction process at thecathode (hydrogen evolution reaction, HER) proceeds according to2 Hþ þ2 e ! H2ð2Þwhile the oxidation process (oxygen evolution reaction, OER)at the anode of the electrolyzer is[16]H2 O ! 1 2 O2 þ2 Hþ þ2 e ð3ÞIn the total process, two electrons are transferred per formulaor a total charge of 192 928 C ( 2 F) crosses the electrocatalytic interface per mole of water converted.We will focus herein on the mechanistic aspects of watersplitting (electrolysis) itself. Photoelectrochemical water splitting, whereby a photon-absorbing component is directly coupled with a gas-evolution electrocatalyst, is beyond the scopeof this review. A recent article has highlighted photoelectrochemical water splitting in the context of photoelectrochemical solar fuels.[11] 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheimwww.chemcatchem.org727

H. Dau, C. Limberg, P Strasser et al.1.1. The thermodynamic efficiency of electrochemical watersplittingThe equilibrium thermodynamics of galvanic and electrolysiscells extends the set of thermodynamic variables of a chemicalprocess that does not involve free charges, an electric potential E, or the difference of two potentials, the voltage V.[17] Cellvoltages at constant pressure P and temperature T are relatedto thermal energy quantities by the amount of electricity transferred nF, where n denotes the number of electrons transferredand F is the Faraday constant.Water splitting (ws) according to Equation (1) at standardtemperature and pressure (STP) is associated with a molar enthalpy change of reaction, herein denoted as DHows,298, which isequal to DHof,H2 OðlÞ ,298, the heat of formation of one mole ofliquid water at temperature 298 K. The absolute value of thesequantities is 286 kJ mol 1 (water) or 2.96 eV. The quantityDHows,298 is also referred to as the higher heating value DHoHHV,298of one mole hydrogen. The change in Gibbs free energy of reaction (1) DGows,298 is 237 kJ mol 1 (water) or 2.46 eV. Note: Different energy units are used in the reviewed fields and we donot use consistently the same units but rather those most convenient in the specific context. For a rough transformation ofenergy units: 1 eV 100 kJ mol 1 25 kcal mol 1.At equilibrium, the minimum electrical energy, 2 F Vrev, required to split one mole of water under STP is related to theGibbs energy asDGows,298 ¼ 2 F V orev,298ð4ÞConsidering only absolute values of cell voltages, V orev,298 1.23 V at STP and is referred to as the reversible electrolysis cellvoltage at STP.The total energy required to split one mole of water at STPis given byDHows,298¼ DHoHHV,298¼ DHof,H2 OðlÞ ,298¼ DGows,298þ T DSows,298ð6ÞoHHV,298orev,298where V 1.48 V. The voltage difference between Vand V oHHV,298 implies that water electrolysis below 1.48 V andabove 1.23 V is an endothermic process at STP. Hence, neglecting ohmic and other loss processes, an electrolysis cell operated below 1.48 V would cool down during electrolysis or wouldrequire a heat supply to operate isothermally.At elevated temperatures, V orev,T drops based on Equations (4)and (5), whereas DHoHHV,T and V orev,T increase according to[18–19]DHoHHV,T ¼ ¼ DHof,H2 OðlÞ ,T þ ðHoH2 OðlÞ ,T of,H2 OðlÞ ,298Þ ¼ 2 F V oHHV,T728www.chemcatchem.orgDHTN;T;p ¼DHHHV;T;ðp pw Þ þ 1:5 pw HH2 OðlÞ ;T;pw HHo 2 OðlÞ ;298ð p pw Þð8Þ¼ 2 F VTN;T;pThe first term of the central expression is the higher heatingvalue of the electrolysis process at temperature T, and gaspressure (p pw), while the second term accounts for the stoichiometric evaporation of water from STP to operating conditions.Owing to a variety of energy and efficiency losses, cell voltages in practical electrolysis cells[9b] are generally above boththe thermoneutral voltage VTN[18] and the higher heating valuevoltage VHHV. That is, excess thermal energy generally leads toheating of the electrolysis cell.The efficiency of an isothermal electrolysis cell operated at acell voltage V oop,T relates its chemical energy output to its electrical energy input, 2 F Voop,T. The energy output may be considered in two separate ways. Firstly, the maximum reversiblework (Gibbs free energy) DG from reconverting hydrogen andoxygen, for instance in a fuel cell, may be of interest andhence a faradaic (voltage) efficiency[9b] of electrolysis, ho298;Faradaic ,can be written asho298;Faradaic ¼oDGows;298 Vrev;298¼ o2 F VopVop;298ð9Þð5Þwhere the last term at STP is 49 kJ mol 1 and corresponds tothe heat required from the surroundings. Thus, a higher-heating-value voltage, VHHV,[18] or an equivalent amount of electricity, 2 F VHHV, can be related to the water splitting energy according toDHoHHV,298 ¼ 2 F V oHHV,298where HoH2 OðlÞ ,T is the molar enthalpy of water at standard pressure and temperature T. As a result, the voltage range of endothermic water splitting broadens.[18]The formalism above assumed the electrolytic production ofperfectly dry hydrogen and oxygen gas. Taking into accountthe presence of a water-vapor partial pressure pw at a totalpressure p of hydrogen and oxygen, a total energy for splittingof one mole of water into humidified gases can be derived,which is associated with a thermo-neutral voltage VTN according to[18–19]ð7ÞThis quantity essentially represents the reciprocal value ofthe voltage efficiency of a fuel cell[20] with Vop replaced by theexperimentally observed output cell voltage of a fuel cell. Thefaradaic efficiency can never be larger than one. Secondly, hydrogen and oxygen may be viewed in terms of their heatingvalue and hence a thermal efficiency of an electrolysis cell,ho298;thermal , is written as[18]ho298;thermal ¼ooDHHHV;298VHHV;298¼ o2 F VopVop;298ð10ÞThis efficiency has become widely accepted even though itwould never be possible to recover the entire higher heatingvalue of hydrogen in an energy application. In principle,ho298;thermal can be larger than unity.[6, 18–19] It is the reciprocalvalue of the reversible thermodynamic efficiency of a fuelcell.[20] Typical values of hoT;thermal of industrial electrolyzers rangebetween 60 % to 80 % at a given current density (typically1 A cm 2) and at 363 K (90 8C) and 0.1 MPa. 2010 Wiley-VCH Verlag GmbH & Co. KGaA, WeinheimChemCatChem 2010, 2, 724 – 761

The Mechanism of Water OxidationEquations (9) and (10) provide a measure of the efficiency ofoverall water splitting. They are generally valid, regardless ofwhether it is an electrocatalytic process at an inorganic electrode, a biocatalytic process, or a catalytic pathway involvinghomogeneous organometallic catalysts. In a ‘translation’ of theelectrochemical terminology, anode and cathode potentialscorrespond to the midpoint redox potentials of the oxidant(anode potential) and reductant (cathode potential) that areused to drive the OER or HER. Similarly, in electrochemical, homogeneous, and biological catalysis, research typically considers only half-cell reactions, as is also the case in the mechanistic discussions of water oxidation (section 1.2). For half-cell reactions, however, Equations (9) and (10) are not directly applicable.Equation (9) may be reformulated asho298;Faradaic ¼orev;298VoVrev;298þ OPanode þ OPcathodeð11Þwhere OPanode and OPcathode are the absolute values of the overpotentials of the OER (anodic reaction) and HER (cathodic reaction), respectively. These overpotentials can be calculated inhomogeneous and biocatalysis as the difference between themidpoint potential of the oxidant (or reductant) and the OER(or HER) equilibrium potential. To quantify the efficiency of thehalf-cell reaction, we now assume for OER (or HER) thatOPcathode (or OPanode) equals zero. The thereby-obtained value ofhoOER (hoHER ) represents a meaningful measure of the energeticefficiency of the catalysis of water oxidation (hydrogen evolution) in electrochemical, homogeneous, and biocatalysis. For acombination of two half-cell reactions, calculation of the overall efficiency ho from the two half-cell efficiencies is straightforward. However, ho is not equal to the product of hoOER and hoHERbut to [(hoOER ) 1 (hoHER ) 1 1] 1.1.2. Reaction mechanisms of electrochemical water splittingAs described above, the electrochemical splitting of water involves two concurrent catalytic half-cell reactions accordingEquation (2) (HER) and Equation (3) (OER). Also shown aremechanistic schemes for either reaction process, which havebeen proposed in literature in the past. Clearly, oxygen evolution being a 4 electron process is more complex as comparedto the evolution of hydrogen and involves several surface adsorbed intermediates. As either process may contribute to voltage and efficiency losses, a discussion of the electrocatalysis ofwater splitting should in principle touch upon both the HERand OER. Work on biocatalytic water splitting has focusedeither on photosystem II (oxygen evolution), as discussed insection 4, or hydrogenase enzymes (hydrogen evolution),which are not discussed herein. Similarly, research concerninghomogeneous catalysis of water splitting has involved systemsliberating either hydrogen or oxygen (or, in rare cases, both;see section 3.2 for comparison). In electrochemical environments, HER on selected noble metals shows extremely high exchange current densities, j0, which is a measure of the intrinsicChemCatChem 2010, 2, 724 – 761turnover frequency of an electrochemical half-cell process. OnPt electrodes (Figure 1), for instance, HER shows values of j0 ofup to several hundred A cm 2,[21] resulting in a very reversiblehalf-cell electrode with high forward and backward reactionFigure 1. Catalytic reactions associated with electrochemical water electrolysis. An external voltage, symbolized by the battery symbol, drives the evolution of hydrogen at the cathode (HER) and the evolution of oxygen at theanode (OER). Protons are migrating from anode to cathode. Both mechanisms are depicted assuming a sequence of proton-coupled single-electronredox reactions. While HER shows one adsorbed intermediate, OER hasthree. The spheres symbolize a metal electrode, for example, Pt, coveredwith an oxide layer. The numbers in parentheses refer to chemical equationsgiven in the text.rates. However, OER on Pt, the catalytically most active element for this reaction, exhibits values of j0 on the order of10 9 A cm 2, which is why much recent work has shiftedtoward OER kinetics. Therefore, the OER is our focus here andalso in upcoming sections.In electrocatalysis, a fundamental concept in characterizingcatalytic processes on surfaces is the overpotential.[22] Overpotential is the deviation of the observed output cell voltage Vop(galvanic cell) or the applied cell voltage Vop (electrolysis) usedin Equations (9) and (10) from the thermodynamic reversiblecell voltage Vrev [Equation (4)]. The overpotential is generallyconsidered a kinetic phenomenon and splits into various contributions such as the ohmic voltage loss (ohmic overpotential)or mass-transport limitations (transport overpotential) duringthe flow of charge.[22] In the context of catalysis, “reaction overpotentials” relate to a kinetic hindrance of individual elementary reaction steps due to activation barriers. On that basis, reaction overpotential is a function of the flowing current and isthought to be linked to the slowest of the elementary reactionsteps, the rate-determining step (RDS), with the highest kineticactivation barrier.Despite much work on the OER, mechanistic models and hypotheses of the sequence of elementary steps based onatomic-scale experiments have remained scarce. Over thecourse of five decades, proposed mechanisms, conclusions asto the RDS, and experimental approaches have changed onlylittle. A severe hurdle to more direct atomic-scale insight intoOER is the fact that reactive intermediates have largely eludeddirect spectroscopic observation. Most early mechanistic workrelies on classical current–potential–time analyses and the derived Tafel slopes. Tafel slopes are derived from semi-logarithmic potential current plots and have been frequently used toobtain information on the mechanism and the rate-determin- 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheimwww.chemcatchem.org729

H. Dau, C. Limberg, P Strasser et al.ing step of the overall half-cell reaction. An early mechanisticscheme of electrochemical water splitting based on experiments on RuO2-based electrodes in acidic environments is outlined by Equations (12)—(15):[23]SþH2 O ! S OHþHþ þe ð12ÞS OH ! S OþHþ þe ð13ÞS OHþS OH ! S OþSþH2 Oð14ÞS OþS O ! O2 þ2 Sð15ÞIn this mechanistic hypothesis, S denotes a catalyticallyactive surface site. Equations (12), (13), and (15), on the onehand, and (12), (14), and (15), on the other, represent parallelreaction pathways. This mechanism and very similar schemeswere proposed for both well-defined Ru(110) single-crystaloxide surfaces[23d, 24] and compact RuO2 films.[23d] RuO2 singlecrystals were reported with a near 60 mV decade 1 and a120 mV decade 1 Tafel slope at low and high overpotentials, respectively.[23d, 24] On that basis, reaction (12) was thought to bethe RDS of the sequence (12)–(13)–(15) at high electrode potentials. An additional second chemical step, such as the hypothetical rearrangement of the S OH species according toEquation (16):[24]S OH ! S OH*de 1.[23d, 27c, d] Similar to the single crystal work, reaction (16) wasproposed for these films as a second elementary step. Lowtemperature Ru oxide films showed high BET surface areavalues combined with higher values of q*. The OER activitywas high for these films and the Tafel slope ranged in the40 mV decade 1 range. Hence, an electrochemical reaction (13)was now proposed as the rate-determining step and the fullmechanistic sequence thus becomes (12)–(13)–(15).Binary and ternary mixed oxides containing Ir, Sn, and Rushowed their own distinct complex Tafel slope patterns withno clear trend or controlling parameter.[25, 27c, d, g, 29] A number ofrecent studies focused on the effect of particle size on the OERmechanism for RuO2 and Ru1 xCoxO2 mixed oxides.[30] The authors used differentially pumped mass spectrometry (DEMS) todirectly correlate the evolution of molecular oxygen and its faradaic current profile. They

structure and function of the natural paragon—the manga-nese-calcium complex in photosystem II—for which ideas con-cerning redox-potential leveling, proton removal, and O O bond formation mechanisms are discussed. The last part high-lights common themes and unifying concepts. [a] Prof. Dr. H. Dau, M. Risch