Artificial Photosynthesis: Molecular Systems for ... - ACS Publications


Artificial Photosynthesis: Molecular Systems for...

0 downloads 146 Views 39MB Size

This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Review pubs.acs.org/CR

Artificial Photosynthesis: Molecular Systems for Catalytic Water Oxidation Markus D. Kar̈ kas̈ ,* Oscar Verho, Eric V. Johnston, and Björn Åkermark* Department of Organic Chemistry, Arrhenius Laboratory, Stockholm University, SE-106 91 Stockholm, Sweden 3.2. Iridium Catalysts Based on CyclopentadieneType Ligands 3.3. Development of Iridium N-Heterocyclic Carbene and Other Iridium Catalysts 3.4. Decomposition of Homogeneous Iridium Catalysts − Generation of Catalytically Active Heterogeneous Iridum Species 4. Artificial Catalysts Housing Earth-Abundant FirstRow Transition Metals 4.1. Early Biomimetic Manganese-Based Photosynthetic Mimics 4.2. Manganese-Based Catalysts Capable of Generating Oxygen 4.3. Molecular Water Oxidation Catalysts Based on Iron 4.4. Oxygen Formation in Copper-Based Systems 4.5. Cobalt-Based Catalysts Capable of Oxidizing Water 5. Synthetic Metal−Oxo Mimics of the Oxygen Evolving Complex 5.1. Polyoxometalates − The Molecular, Inorganic Alternative 5.1.1. Ruthenium-Based Polyoxometalates 5.1.2. Cobalt-Based Polyoxometalates − The Ambiguity of Distinguishing between Homogeneous and Heterogeneous Catalysis 5.1.3. Other Metal-Based Polyoxometalates 5.2. Metal Cubanes Based on Earth-Abundant Metals as Water Oxidation Catalysts 5.2.1. The Self-Assembling Cobalt−Phosphate (Co−Pi) Catalyst 5.2.2. Cobalt Cubanes and Related Cobalt Oxide Materials as Oxygen Evolving Catalysts 5.2.3. Manganese Cubanes as Water Oxidation Catalysts 6. Conclusions Author Information Corresponding Authors Notes Biographies Acknowledgments Abbreviations References

CONTENTS 1. Introduction 1.1. The Oxygen Evolving Complex − The Site Where Water Is Oxidized 1.2. Artificial Photosynthesis − Architectural Complexity 1.2.1. Light Absorption by a Chromophore − The Photosensitizer 1.2.2. Reduction of Protons for the Generation of Solar Fuels − The Reduction Catalyst 1.2.3. The Water Oxidation Catalyst − Catalytic Complexity at the Frontier 1.3. Experimental Design for Evaluating Water Oxidation Catalysts 1.3.1. Commonly Used Chemical Oxidants 1.3.2. Light-Driven Water Oxidation 1.3.3. (Photo)electrochemical Water Oxidation 1.3.4. Homogeneous versus Heterogeneous Catalysis − The Conversion of Molecular Entities to Heterogeneous Catalytically Active Species 1.4. Mechanistic Considerations Associated with Artifical Water Oxidation Catalysts 2. Molecular Ruthenium-Based Water Oxidation Catalysts 2.1. Dinuclear Ruthenium-Based Systems for Water Oxidation 2.2. Single-Site Ruthenium-Based Catalysts − Simple and Efficient 2.2.1. Mononuclear Ruthenium Catalysts Comprising Neutral Ligand Frameworks 2.2.2. Mononuclear Ruthenium Catalysts Containing Anionic Ligand Backbones 3. Iridium-Based Systems for Water Oxidation 3.1. Cyclometalated Iridium Complexes

© 2014 American Chemical Society

11864 11865 11865 11866 11866 11867 11868 11868 11869 11869

11870 11871 11871 11871 11883

11883 11895 11924 11924

11924 11928

11930 11934 11934 11937 11947 11950 11955 11962 11962 11962

11968 11974 11975 11976

11979 11982 11987 11988 11988 11988 11988 11989 11989 11990

Received: October 11, 2013 Published: October 29, 2014 11863

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 1. Schematic representation of the processes involved in photosynthesis and the photosynthetic machinery, located in the membranes of the chloroplasts. Reprinted with permission from ref 18. Copyright 2012 InTech.

Scheme 1. Kok Cycle in Which the OEC Cycles through Five Redox States, S0−S4, via Consecutive Photo-oxidation

1. INTRODUCTION The sustainable production of clean energy constitutes one of the most important scientific challenges of the 21st century. Currently, the major part of the global energy supply is provided by carbon-based energy sources, which are connected to severe environmental issues, such as air pollution and the greenhouse effect. Consequently, there exists a strong demand for clean and environmentally friendly carbon-neutral alternatives that have the potential to meet the needs of present and future generations. In fact, every hour the sun provides our planet with more energy than what is consumed during a whole

year. This fact has mesmerized researchers and makes the conversion of solar energy into fuel and electricity one of the most promising solutions to this problem.1−13 Through billion years of evolution, nature has devised a system that has the capability to harness the energy provided by the sun. The photosynthetic apparatus (Figure 1) uses the solar energy to reduce CO2 to carbohydrates (here represented as C6H12O6) according to eq 1.14−17 The natural system constitutes an excellent source of inspiration for how to design a system that can utilize solar energy for fuel production. 11864

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews 6CO2 + 12H 2O → C6H12O6 + 6O2 + 6H 2O

Review

(1)

In photosystem (PS) II, a central pair of chlorophylls, P680, is excited and transfers an electron to the acceptor system, which subsequently reduces CO2. The oxidized form, P680•+, which is a strong oxidant with an oxidation potential of ca. +1.2 V versus the normal hydrogen electrode (vs NHE),19 then recovers the electron from a Mn4Ca-cluster in the oxygen-evolving complex (OEC). After four consecutive electron abstractions from the OEC, two molecules of H2O are oxidized to generate one molecule of O2 and four protons. Examination of the photosynthetic machinery is a prime example of research that requires interplay between numerous different disciplines, such as biochemistry, biophysics, molecular and structural biology, quantum mechanics, as well as physiology and ecology.14,17,20 However, for an artificial system, a more practical approach would be to use solar energy to split H2O into molecular oxygen and hydrogen (eqs 2−4). This process requires the coupling of the two half-reactions: (i) oxidation of H2O to generate the reducing equivalents (electrons) (eq 2) and (ii) reduction of protons to molecular hydrogen (eq 3).1−10 2H 2O → O2 + 4H+ + 4e−

(2)

4H+ + 4e− → 2H 2

(3)

2H 2O → O2 + 2H 2

(4)

Figure 2. Structure of the manganese calcium cubane, Mn4Ca, located in the OEC. Reprinted with permission from ref 25. Copyright 2011 Macmillan Publishers Ltd.

will continue as long as the sun shines!”34 Realizing Ciamician’s vision, the creation of artificial systems for generation of fuels from sunlight, based on the priciples of natural photosynthesis, can be described as the creation of an “artificial leaf”.35 This is a photodriven system, where the energy is stored as, for example H2, which is produced from the reduction of protons generated from the oxidation of H2O.36,37 One of the main difficulties in realizing light-induced splitting of H2O is that the two half-reactions (eqs 2 and 3) are multielectron processes. Harnessing solar energy for the photoproduction of renewable fuels therefore requires interfacing several fundamental but challenging photophysical steps with complicated catalytic transformations.38−40 In an artificial photosynthetic system, it is thus essential to devise an efficient process that (1) can absorb a photon by a chromophore (photosensitizer), (2) form a charge-separated state by transferring the electron to a reduction catalyst, usually via a primary acceptor, (3) accept and accumulate two consequtive electrons at the reduction catalyst to subsequently use these to reduce two protons to molecular hydrogen, (4) allows regeneration of the chromophore by transfer of an electron from the oxidation catalyst, generally via a primary donor, and (5) after transfer of four consequtive electrons, one by one, from the oxidation catalyst, the oxidation catalyst is regenerated by a formal one-step transfer of four electrons from two molecules of H2O to generate one molecule of O2 and four protons.41,42 The construction of an artificial photosynthetic system thus requires that one-electron-transfer events can be coupled to catalytic units that are able of mediating the four-electron-fourproton splitting of H2O. These catalysts may be interfaced with the surface of semiconductor materials to enhance the photoelectrochemical H2O splitting by two different approaches: (i) the catalysts for the two half reactions (H2O oxidation and proton reduction) can be coattached to a lightharvesting material, which has the appropriate band structure for overall H2O splitting, or (ii) the catalysts can be separately attached to two different semiconductors, connected by a socalled “hard wire”. For discussions covering semiconductor materials that mediate photocatalytic H2O splitting, we refer to

1.1. The Oxygen Evolving Complex − The Site Where Water Is Oxidized

The core of the OEC has been studied both by crystallography and by other spectroscopic techniques, such as X-ray absorption spectroscopy. It is comprised of four manganese and one calcium atom, and is held together by surrounding μoxo and μ-hydroxo ligands in a cubane-like arrangement, where three of the manganese centers are capped by the calcium atom, while the fourth dangling manganese atom is connected to the cubane via two of its oxo-groups (Figure 2). The structural motif of the OEC is preserved in essentially all species of photosynthetic organisms, and the absence of natural diversity of this catalytically competent cubane core demonstrates its fundamental importance.21−30 The mechanism by which the OEC catalyzes H2O oxidation has been extensively studied.31,32 A general opinion is that the OEC cycles through five oxidation states, denoted S0−S4 (Kok cycle). In this scheme, the S0 state represents the most reduced state, and the resting state of the catalytic cycle. The fourelectron photo-oxidized state, S4, is the most oxidized state, and it is through the S4 → S0 transition that molecular O2 is liberated (Scheme 1).33 1.2. Artificial Photosynthesis − Architectural Complexity

The Italian photochemist Giacomo Ciamician predicted about a century ago that solar energy could be used to solve a future energy crisis. In his 1912 article in Science, he stated: “On the arid lands there will spring up industrial colonies without smoke and without smokestacks; forests of glass tubes will extend over the plains and glass buildings will rise everywhere; inside of these will take place the photochemical processes that hitherto have been the guarded secret of the plants, but that will have been mastered by human industry which will know how to make them bear even more abundant fruit than nature, for nature is not in a hurry and mankind is. And if in a distant future the supply of coal becomes completely exhausted, civilization will not be checked by that, for life and civilization 11865

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

a number of recent reviews that cover this topic thoroughly.43−50 In an artificial photosynthetic system, a chromophore (photosensitizer) can first be attached to a semiconductor. Upon illumination a charge-separated state is formed, by electron transfer from the photoexcited chromophore to the conduction band of the semiconductor. It is important that this charge-separated state is sufficiently long-lived to allow the chromophore to abstract an electron from the catalyst, thus regenerating the chromophore and a one-electron oxidized catalyst. After four successive charge-separations and concomitant electron abstractions, the catalyst refills the holes by abstracting electrons from H2O, thereby oxidizing it to molecular O2. The electrons abstracted from H2O are ultimately delivered to a reduction catalyst, which mediates the reduction of protons (generated from H2O oxidation) to solar fuels, such as H2. The generated H2 could be either used directly as a source of energy or be used to reduce CO2 to other types of fuels of higher complexity, such as methane or methanol. A schematic overview of how a device for H2O splitting can be constructed is depicted in Figure 3.6,41,42

The most well-studied molecular photosensitizers for electron-transfer processes are the ruthenium(II) polypyridyl, [Ru(bpy)3]2+-type complexes (1, Figure 4; bpy = 2,2′-

Figure 4. Structure of the versatile ruthenium polypyridyl, [Ru(bpy)3]2+, type chromophore 1.

bipyridine). These complexes have been studied since the early 1970s when it was discovered that they could undergo electron transfer from their excited states to a sacrificial electron acceptor.51,52 This opened new avenues of research, where these complexes have been used in various applications and have recently received significant attention in the field of photoredox catalysis.53 In light of this, it is instructive to consider the photochemistry of the [Ru(bpy)3]2+-type complexes. These photosensitizers have a distinct absorption in the visible region (λmax ≈ 450 nm), and upon illumination, one electron in a metalcentered t2g orbital is excited to a ligand-centered π* orbital. In this metal-to-ligand charge transfer state (MLCT), the ruthenium center has formally been photo-oxidized to RuIII, while the bpy ligand backbone has undergone a single electron reduction. The initially photogenerated state, the lowest excited singlet state (1[Ru(bpy)3]2+*), is short-lived and undergoes rapid intersystem crossing54 to yield a long-lived triplet state, 3 [Ru(bpy)3]2+*, that has the potential to engage in singleelectron transfer. The electronic configuration of this excited state can thus be described as a 3[(dπ)5(πbpy*)1] with almost exclusively triplet character. This excited triplet state has a long lifetime because the conversion to the singlet state is spinforbidden and decays to the ground state on the microsecond time-scale. The excited state is thus sufficiently long-lived to partake in bimolecular electron-transfer reactions, which can compete with deactivation pathways. The photoexcited state has the ability of being both more oxidizing and more reducing than the ground state of [Ru(bpy)3]2+, and this dual nature of the excited state enables it to serve either as a single-electron oxidant or as a reductant, as depicted in Scheme 2.55−59 This unique property of the [Ru(bpy)3]2+-type complexes coupled with their chemical stability, redox properties, excitedstate reactivity, luminescence emission, excited-state lifetime, and synthetic flexibility makes them highly versatile and attractive for mediating both photo-oxidation and photoreduction. However, coupling light sensitization to the generation of the desired oxidant, [Ru(bpy)3]3+, requires efficient electron transfer to an acceptor, a reduction catalyst, which should ultimately use the electron for the production of solar fuels. 1.2.2. Reduction of Protons for the Generation of Solar Fuels − The Reduction Catalyst. The prospect is that H2 will become a significant energy source in the upcoming years, by virtue of its clean combustion characteristics in fuel

Figure 3. Schematic picture of a molecular assembly for overall H2O splitting consisting of a photosensitizer (PS), a water oxidation catalyst (WOC), and a hydrogen evolving catalyst (HEC), for the production of solar fuels.

Although the steps discussed above seem straightforward, the combination of all of them to enable efficient light sensitization, fast electron transfer, and ultimately efficient catalysis is not easily accomplished. To date, the main focus has been to design and synthesize molecular systems comprising electron donors and acceptors that mimic the light-driven charge separation events occurring in the natural photosynthetic reaction centers. The essential challenge in a man-made device for solar-driven H2O splitting is to develop a molecular arrangement that can orchestrate and enable all of the discussed events to occur efficiently with marginal energy losses. Although significant progress has been made in the field of artificial photosynthesis, there still remains much to be done before realizing its full potential. 1.2.1. Light Absorption by a Chromophore − The Photosensitizer. The first step in an artificial photosynthetic system is light-harvesting by a chromophore, that is, a lightabsorbing component, analogous to the photosynthetic pigments. This chromophore should efficiently absorb and convert the incoming solar energy into an excited state that can transfer an electron to an acceptor for the creation of a charge-separated state, thus generating the required thermodynamic driving force for the desired chemical reactions. Both photoactive molecular dyes and semiconductors can induce electron transfer and have the potential of being used as light-harvesting chromophores in a future artificial device for H2O splitting.4,41,46 11866

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 2. Simplified Picture of the Photochemistry of [Ru(bpy)3]2+

cells. Currently, molecular hydrogen can be produced from fossil fuels employing well-established industrial processes. Clearly, this is not “green” and calls for technologies that realize a sustainable large-scale H2 economy.60,61 The reduction of protons to H2 has a large activation barrier, resulting in a great overpotential and thus a low rate. It is therefore necessary to use a catalyst, which will lower the barrier and allow for efficient generation of H2. As previously described, the electrons abstracted from the excited photosensitizer are ultimately delivered to an acceptor, a hydrogen evolving catalyst (HEC), which catalyzes the reduction of protons to H2 (or carbon-based fuels). This reduction constitutes one half-reaction (eq 3) in an artificial photosynthetic H2O splitting system.62−65 Platinum-based complexes have proved to be excellent catalysts for the reduction of protons.66,67 However, the scarcity and high cost associated with platinum limit its use in large-scale applications. It is therefore of great importance to develop catalysts for solar fuel production that are based on abundant and inexpensive metals. Nature utilizes so-called hydrogenase enzymes that are comprised of iron and/or nickel cofactors as the catalytically active sites.68−73 The hydrogenases are a family of metallosulfur enzymes responsible for mediating both the oxidation of molecular H2 and the reduction of protons to H2. In nature, these reactions proceed in neutral aqueous solutions near the thermodynamic potential with extremely high turnover frequencies (TOFs; defined as moles of produced product per mole of catalyst per unit of time) of 100−10 000 mol of produced H2 per mole of catalyst per second.74

Research has been devoted to developing nonbiological, structurally simple iron-based mimics that are functional models of the active sites of hydrogenases for H2 production, such as complex 2 (Figure 5).75,76 Complementary to this,

Figure 5. Structural examples of artificial [FeFe]-hydrogenase (2) and cobalt diglyoxime complex (cobaloxime, 3).

other first-row transition metal catalysts for H2 evolution include the macrocyclic cobalt77,78 (3, Figure 5) and nickel79,80 complexes, which have been studied for several decades.81 1.2.3. The Water Oxidation Catalyst − Catalytic Complexity at the Frontier. At a first glance, the oxidation of H2O seems to be a straightforward transformation, due to the structural simplicity of starting material and products. However, this task is definitely not easily accomplished, and the majority of redox processes in nature are single- or two-electron processes. The oxidation of H2O is an exception because it 11867

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

1.3. Experimental Design for Evaluating Water Oxidation Catalysts

requires the transfer of four electrons, rearrangement of multiple bonds, and finally formation of the O−O bond. The basic thermodynamic requirements for the splitting of H2O, according to eq 5, are associated with a Gibbs free energy (ΔG) of ∼237 kJ mol−1. Converting this to an electrochemical potential, by the use of the Nernst equation (eq 6), where n is the number of electrons mol−1 transferred in the reaction, F is the Faraday constant (96 485 Coulomb mol−1), translates into a minimum potential of +1.229 V vs NHE at STP (STP is standard temperature and pressure).82 H 2O(l) → 1/2O2(g) + H 2(g)

(5)

ΔG = −nFE

(6)

When examining if a potential complex exhibits WOC activity, the catalytic experiments are usually carried out by use of a sacrificial oxidant, which is advantageous as it enables studies in bulk solution. Although this does not exactly mimic the conditions employed in a potential artificial photosynthetic cell, the use of sacrificial oxidants enables rapid screening and tuning of WOCs. A prerequisite of these sacrificial oxidants is that they have an oxidation potential more positive than the investigated WOC for H2O oxidation to be thermodynamically favorable.85 1.3.1. Commonly Used Chemical Oxidants. When assessing the catalytic activity of a potential WOC, the oxidant is either added to an aqueous solution containing the metal complex of interest or vice versa. The generated O2 is then measured directly in solution by a so-called Clark electrode, or in the gas phase by techniques such as mass spectrometry or gas chromatography. The most common oxidant for screening WOCs is the strong one-electron oxidant ceric ammonium nitrate (CAN, Ce(NH4)2(NO3)6, CeIV), which has also been widely used for the oxidation of organic compounds.86 CeIV has a redox potential of approximately +1.70 V vs NHE87 and has only weak absorption in the UV−vis region, making it applicable for mechanistic studies using various spectroscopic techniques. Moreover, its commercial availability and stability in aqueous acidic solutions make CAN a suitable choice as sacrificial oxidant for the examination of WOCs (eq 7).85

This implies that light with a wavelength shorter than 1000 nm has enough energy to split the H2O molecule into H2 and O2 and thus includes the use of the entire visible solar spectrum and part of the near-infrared spectrum, which together accounts for ∼80% of the total solar radiation. However, in electrolysis, the splitting of H2O is associated with a large overpotential,83 a kinetic phenomenon that increases the potential required to carry out the specific reaction (i.e., the deviation of the applied potential from the thermodynamic). In the context of catalysis, this overpotential translates to a high activation barrier for the reaction in question, which requires the involvement of catalysts to promote the splitting of H2O by lowering the activation barrier. Because of the complexity and the high oxidation potential required to oxidize H2O, this half reaction is currently considered to be the bottleneck in the development of a sustainable artificial solar fuel system. This calls for a catalyst capable of accumulating four oxidizing equivalents and operating close to the thermodynamic potential of H2O oxidation. Also, the formation of damaging high-energy intermediates needs to be avoided as they can reduce the longevity of the WOC. An ideal WOC should thus be fast, amenable to interfacing with photosensitizing materials, and stable to oxidative, hydrolytic, and thermal degradation during turnover.82 Multielectron-transfer catalysis can be facilitated by synchronizing proton and electron-transfer events, via proton-coupled electron transfer (PCET), and is essential in many biological and chemical processes. PCET permits the total charge of a chemical species to remain unchanged, whereas just single electron transfer, without the loss of a proton, leads to charge accumulation and high-energy intermediates. Coupling electron transfer to proton transfer may thus allow the accumulation of multiple redox equivalents and influence reaction pathways and energetics of a certain chemical reaction. This is an essential feature in realizing the four-electron oxidation of H2O in artificial photosynthesis.84 The search for novel and improved WOCs has led to the development of a number of homogeneous and heterogeneous WOCs. In the sections to follow, this Review aims at summarizing the recent development in the rapidly growing field of artificial molecular WOCs. This Review focuses on covering synthetic molecular systems and highlights the different approaches that have been explored in the development of artificial WOCs. Heterogeneous catalysts, such as metal oxide-based WOCs, which may be generated by decomposition of the studied homogeneous WOCs, will also briefly be highlighted and discussed.

cat.

4Ce IV + 2H 2O ⎯→ ⎯ 4Ce III + O2 + 4H+

(7)

However, there are some disadvantages associated with CeIV; it requires strongly acidic solutions (pH < 1), which excludes its use when examining a potential WOC that is acid sensitive. Recently, it has also been brought to light that the nitrate anion in CAN can be involved in promoting oxygen atom transfer (OAT). This can affect the O2 evolution mechanism as one of the oxygen atoms may originate from the nitrate ions of CAN instead of H2O and thus preclude “real” H2O oxidation.88 By contrast, potassium peroxymonosulfate (Oxone) is a powerful two-electron oxidant with an oxidation potential of 1.82 V vs NHE.89 The main advantages of Oxone are its stability in solutions up to pH ∼6 and its ability to act as a twoelectron oxidant. However, its two-electron oxidizing nature allows it to act as an OAT reagent and questions its relevance to PECs where catalysis usually occurs through one-electrontransfer events.85,90 Sodium periodate, NaIO4, is also a two-electron oxidant that has been used to study WOCs.91 It has an oxidation potential of ∼1.60 V vs NHE (at pH 1) and can potentially be used in solutions of pH up to 7.5,92 allowing for determination of H2O oxidation activity under neutral conditions. In addition to the two-electron oxidants Oxone and NaIO4, sodium hypochlorite has been examined as a primary two-electron oxidant at alkaline pH’s.85 Peroxides have been used to drive H2O oxidation, mostly employing first-row transition metal WOCs. However, as with all two-electron oxidants, concern needs to be given to the origin of the evolved O2, and this ambiguity questions their usefulness in the evaluation of a potential WOC. As a complement to the primary oxidants discussed above, the mild one-electron oxidant [Ru(bpy)3]3+ has been explored. The [Ru(bpy)3]3+ oxidant delivers a potential of 1.26 V vs NHE, and the main advantage of employing this oxidant is its activity at near-neutral conditions. Although [Ru(bpy)3]3+ 11868

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 6. Molecular structures of [Ru(bpy)3]2+-based photosensitizers and their oxidation potentials.

possesses some attractive properties for evaluating potential WOCs, the use of this oxidant to drive H2O oxidation is still fairly uncommon. This is mainly due to the low redox potential of [Ru(bpy)3]3+, which is insufficient to activate a majority of the developed WOCs. Conducting routine characterizations with [Ru(bpy)3]3+ might also be intimidating because this oxidant is relatively expensive when compared to other commonly used oxidants. Another reason is that [Ru(bpy)3]3+ is rapidly decomposed at pH > 4, in a reaction that results in only minor formation of O2. Electron transfer from the WOC must therefore compete with decomposition of [Ru(bpy)3]3+ to allow for efficient H2O oxidation.93 1.3.2. Light-Driven Water Oxidation. To avoid the problems associated with [Ru(bpy)3]3+ as a primary chemical oxidant, a three-component light-driven system has been developed, which is based on in situ formation of [Ru(bpy)3]3+ from [Ru(bpy)3]2+.94 One advantage of this light-driven system is that different [Ru(bpy)3]2+-type derivatives can be used with oxidation potentials ranging from 1.10 < E < 1.54 V vs NHE (Figure 6). In this light-driven system, a sacrificial electron acceptor is needed, and the most commonly used acceptor is sodium persulfate (Na2S2O8). This system is well-studied and believed to proceed via quenching of the photoexcited state [Ru(bpy)3]2+* by S2O82−, to produce [Ru(bpy)3]3+, sulfate, and a sulfate radical (SO4•−). Because the sacrificial acceptor undergoes irreversible bond cleavage after electron transfer, the problem with recombination, or back-donation occurring, can be ruled out. The sulfate radical is a strong oxidant by itself and has sufficient potential (E° > 2.40 V vs NHE95) to directly oxidize the [Ru(bpy)3]2+ to generate a second equivalent of the oxidant [Ru(bpy)3]3+. Four equivalents of the photogenerated [Ru(bpy)3]3+ then sequentially oxidize the WOC, which in turn oxidizes H2O after four consecutive electron transfers. The chemistry of the persulfate-based light-driven system is shown in eqs 8−11.94 2[Ru(bpy)3 ]2 + + 2hν → 2[Ru(bpy)3 ]2 + *

However, it is important to note that the generated sulfate radical with its high oxidation potential has the possibility of oxidizing the WOC directly. To ensure that the sulfate radical preferentially reacts with [Ru(bpy)3]2+, it is vital to conduct the photocatalytic H2O oxidation experiments with a significantly higher concentration of the photosensitizer than the WOC being studied. As an alternative to using persulfate as the electron acceptor, CoIII has also been explored.52,96 1.3.3. (Photo)electrochemical Water Oxidation. Complementary to evaluating potential WOCs with a primary oxidant, electrochemical methods can also be employed. Although assembling a catalyst on a conducting material might seem relatively straightforward, it requires considerable expertise and equipment, which explains the limited number of examples where WOCs have been functionalized on a conducting surface for electrochemical or photoelectrochemical H2O oxidation.97 The catalyst of interest can be assembled on the conducting surface by different methods, such as physisorption or covalent attachment. The functional groups that have been used for covalently interfacing the catalyst with different conducting materials are either carboxylate (−COO−) or phosphonate (−PO3).98,99 An important factor to consider for electrochemical halfreactions is the so-called overpotential. In homogeneous H2O oxidation catalysis, the overpotential can simply be considered as the difference between the redox potential of the oxidant employed and the equilibrium potential for the oxygen evolving reaction (OER; eq 2). The fundamental key to achieving efficient, electrochemical H2O splitting is to drastically lower the overpotential required for H2O oxidation. The overpotential is related to the activation barriers for the individual steps of a reaction pathway and can be described as a kinetic obstacle for the studied process. However, in electrocatalysis, the term overpotential is the potential (voltage) that has to be applied in an electrolytic cell in addition to the thermodynamically determined potential for a given half-reaction.19b Here, the overpotential is also considered as a kinetic factor that is compartmentalized into ohmic voltage loss (ohmic overpotential) and mass-transport limitations (transport overpotential), and used to quantify the efficiency of the electrocatalytic cell.100−102 The overpotential is a feature of the individual cell design and is therefore expected to differ between cells and the operational conditions. It is an essential concept that can be used to characterize catalytic processes on surfaces. In this regard, an important relationship is the Tafel equation (eq 12), which relates current density (kinetic rate) to the overpotential.100−102

(8)

2[Ru(bpy)3 ]2 + * + 2S2 O82 − → 2[Ru(bpy)3 ]3 + + 2SO4•− + 2SO4 2 −

(9)

2[Ru(bpy)3 ]2 + + 2SO4•− → 2[Ru(bpy)3 ]3 + + 2SO4 2 − (10)

4[Ru(bpy)3 ]2 + + 2S2 O82 − + 2hν → 4[Ru(bpy)3 ]3 + + 4SO4 2 −

(11) 11869

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews RT RT ln i0 − ln i αnF αnF 2.303RT 2.303RT ln i0 − ln i = αnF αnF

Review

decomposition pathway. Another classical sign that suggests the transformation of the initial molecular species is an unexplained induction period before catalysis is initiated. However, a lag phase does not necessarily have to be associated with in situ generation of a catalytically active heterogeneous species either. Instead, it could be an indication that the molecular complex used is only a precursor that is slowly converted to a homogeneous catalytically active species that enters the catalytic cycle. It is also tempting to believe that reisolation of the molecular catalyst from a postcatalytic reaction solution means that the catalyst operates as a homogeneous catalyst.103 In recent years, the clear distinction between homogeneous and heterogeneous catalysis has been blurred by the use of different metal clusters, metal nanoparticles, or nanomaterials. Nanoparticles refer to clusters that, usually, are within the size range of 1−10 nm. Solutions of these can appear clear by the naked eye and even have a color that resembles the studied metal complex, which makes it difficult to determine what really constitutes the catalytically active species. In addition to this, the catalytically active species might be derived from the transformation of only a small fraction of the initial metal complex, leaving most of the starting complex unchanged. Because nanoparticles can display exceptionally high catalytic activities, conversion of even a small amount of the initial molecular species to nanoparticles can have a significant effect on the observed catalytic reaction. It should be noted that the specific reaction conditions and time scales might heavily influence the nature of true catalytic species. A catalyst may thus operate under one reaction mechanism under one set of conditions but a completely different mechanism under another set of reaction conditions. Further problems arise if the true catalyst is not reproducibly generated, as can be the case with nanoparticles where even minor differences in the experimental procedure can result in particles with varying size and shape, which can translate into diverse catalytic behavior.104−108 A toolbox of tests, based on a variety of physical techniques, has been developed to assess whether the starting complex is the real catalyst or just a precursor. The mere presence of metal particles can be confirmed by X-ray photoelectron spectroscopy (XPS), which allows for determination of the different elements that are present in the investigated material and their respective oxidation states. Energy-dispersive X-ray spectroscopy (EDX) can also be used for elemental analysis and chemical characterization of a certain surface. Additional and useful analytical techniques are transmission electron microscopy (TEM) and dynamic light scattering (DLS). TEM is an imaging technique that can be used to detect particles as small as a few nanometers in size; however, subnanometer clusters, especially those comprised of light first-row transition metals, are usually beyond the detection limit of this technique. The new extension of TEM, high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM), is a promising tool for detecting small metal particles in the subnanometer range. Finally, DLS is a powerful technique for identifying the presence, or absence, of nanoparticles in solution. DLS permits for the determination of the mean radius of the suspended particles and has also the advantage of direct, in operando, analysis of the catalytic solutions, in contrast to TEM.

η=

η = a + b log i

(12) (13)

where η is the overpotential, R is the universal gas constant, T is the absolute temperature, α is the so-called charge transfer coefficient, n is the number electrons involved in the charge transfer controlled reaction, F is Faraday constant, i0 is the exchange current density, and i is the current density. Equation 12 can also be rewritten in a simplified form to yield eq 13, where a is the Tafel equation constant, and b is the so-called Tafel slope. From eq 12, the intercept at η = 0 gives the exchange current density, i0, from which the charge transfer coefficient, α, can be derived, and this value is usually close to 0.5. Thus, for a given reaction with n = 1 or n = 2, this translates to Tafel slopes of 120 and 60 mV, respectively. In other words, the Tafel equation is essential for the evaluation of electrocatalytic systems as it describes the overpotential as a function of the current density, and the so-called Tafel slopes (derived from semilogarithmic potential current plots) are employed frequently to obtain mechanistic insight and information regarding the rate-determining step of the overall half-cell reaction.100−102 1.3.4. Homogeneous versus Heterogeneous Catalysis − The Conversion of Molecular Entities to Heterogeneous Catalytically Active Species. Oxidation of H2O requires harsh oxidative conditions to be converted to O2. As will be demonstrated in this Review, this makes the screened WOCs susceptible to irreversible structural modifications, including transformation of the initially molecular species into heterogeneous metal oxides/hydroxides. A recurring and usually intricate question that arises is if the true catalytic species is homogeneous in nature, consisting of a well-defined molecular metal complex, or if a heterogeneous catalytically active species is produced in situ. Because these oxide materials are robust and efficient catalysts for mediating H2O oxidation, considerable doubts exist as to whether the initial metal complex is the true catalytic species. Resolving this homogeneity−heterogeneity issue is vital if one wants to gain fundamental knowledge of the catalytic mechanism by which a metal complex mediates a certain reaction. This is a prerequisite if one intends to design improved catalytic protocols by, for example, introducing subtle ligand modifications. The topic of distinguishing between homogeneous catalysis by metal complexes and heterogeneous catalysis by in situ generated metal oxide materials is certainly not only relevant for the area of H2O oxidation but is also a central question in all fields of metal catalysis. One of the strongest indications that a certain process does not proceed through a homogeneous pathway is the visible deposition of a metallic precipitate or mirror on the reaction vessel, which can easily be isolated and screened for the desired catalytic activity. Although the deposited material is not an active catalyst, it is still possible that the initial metal complex is decomposed to a catalytically active nanomaterial that remains in solution and that is subsequently transformed to the observed inactive metallic deposit. However, it should be stressed that the apparent occurrence of a precipitate does not need to be associated with the generation of a catalytically active heterogeneous phase, and can simply be a catalyst 11870

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

1.4. Mechanistic Considerations Associated with Artifical Water Oxidation Catalysts

There exists a requirement on WOCs to produce stable highvalent metal−oxo species at low redox potentials, to be able to partake in the complicated process of multielectron oxidation of H2O. To enable the rational design of molecular complexes that can fulfill this requirement, extensive research has been directed toward elucidating and understanding the fundamental steps of H2O oxidation. One step in particular that needs to be better understood is the mechanism(s) for the O−O bond formation. Despite the experimental difficulties in determining the reaction mechanisms, some progress has been made, which has yielded two major mechanistic pathways for H2O oxidation (Scheme 3): (i) solvent water nucleophilic attack (WNA) and (ii) interaction of two M−O units (I2M).109−119

Figure 7. Molecular structure of the so-called “blue dimer”, cis,cis[(bpy)2(H2O)Ru(μ-O)Ru(H2O)(bpy)2]4+ (8).

proved that the difficult multielectron oxidation of H2O to O2 was indeed possible. The presence of the μ-oxo bridge promotes a strong electronic coupling between the two metal centers in the blue dimer 8. This strong electronic coupling is advantageous because it facilitates the stabilization of the complex at high oxidation states by electronic delocalization.125,126 A structural analogue to the blue dimer, [(tpy)(H2O) 2Ru(μ-O)Ru(H2O)2(tpy)]4+ (tpy = 2,2′;6′,2″-terpyridine), has also been reported to catalyze H2O oxidation.127,128 The blue dimer and structurally related derivatives containing phosphonate groups for stable anchoring onto different oxide-based surfaces have also been reported.129−131 The low overpotential of the blue dimer and its derivatives enabled the use of light-driven protocols employing photogenerated [Ru(bpy)3]3+.132,133 Multistep electron-transfer events are common in many biological systems, including photosynthesis, and enable controlled kinetics of the redox reactions. An interesting method of enhancing the catalytic rates of H2O oxidation is adding kinetically rapid electron-transfer mediators (ETMs) to the catalytic systems as depicted in Scheme 4c. For the “blue dimer” 8, this approach has resulted in rate enhancements by factors of up to ∼30.134 Similar rate enhancements have been observed in catalytic oxidations of organic substrates by molecular oxygen and hydrogen peroxide, that is, the reverse reaction of H2O oxidation, by employing biomimetic mediators. In these oxidation protocols, the direct oxidation of an organic substrate by either O2 or H2O2 is rare and associated with a high energy barrier, which translates into unfavorable kinetics for the direct electron transfer from the organic substrate to the oxidant (O2). By introducing ETMs, the high kinetic barrier is divided into several smaller ones that enable low-energy electron transfer and translates into a higher rate and selectivity (Scheme 4).135,136 In addition to the seminal work on the “blue dimer”, extensive mechanistic insight has been gained through structural and kinetic studies,132,137−153 together with detailed calculations.154−158 However, the results from the H218O isotopic labeling experiments performed by the groups of Meyer and Hurst were rather controversial. The two groups obtained conflicting results regarding the isotopic distribution, which consequently led to different interpretation of the mechanism. The mechanism proposed by Meyer and coworkers involved four PCET steps that give access to a highvalent RuV,V intermediate (9). This intermediate is subsequently attacked by an ancillary H2O molecule, generating hydroperoxido species 11, which is intramolecularly oxidized by the second ruthenium center, ultimately resulting in the release of O2 (Scheme 5).

Scheme 3. Schematic Representation of the Two Mechanistic Pathways for O−O Bond Formation for SingleSite and Dinuclear Artificial WOCs

In the WNA mechanism, H2O acts as the nucleophile by attacking the electrophilic high-valent metal−oxo species, which leads to cleavage of the metal−oxo π-bond and the concomitant generation of the crucial oxygen−oxygen bond. This results in the formal two-electron reduction of the metal center to form a metal hydroperoxide species (M−OOH), which can subsequently undergo further oxidation to liberate O2. The other mechanistic pathway, the I2M, involves the radical coupling of two metal−oxo species that hold significant radical character, generating a [M−O−O−M] species that may undergo further oxidation to ultimately release O2. Generally, systems proceeding via the I2M pathway contain flexible ligand scaffolds with large bite angles (>90°) that can promote the O−O bond formation.109,115,120 However, despite this mechanistic progress, there still exists a need for the development of a better understanding regarding the ligand-dependent preferences of the two mechanisms, and the parameters that favor one mechanism over the other.

2. MOLECULAR RUTHENIUM-BASED WATER OXIDATION CATALYSTS 2.1. Dinuclear Ruthenium-Based Systems for Water Oxidation

In 1982, Meyer’s group reported on the first complex capable of mediating the four-electron-four-proton oxidation of H2O.121,122 This was the dinuclear μ-oxo-bridged ruthenium complex cis,cis-[(bpy)2(H2O)Ru(μ-O)Ru(H2O)(bpy)2]4+ (8, Figure 7), more commonly known as the “blue dimer” due to its characteristic blue color. Although the turnover number (TON; defined as moles of produced product per mole of catalyst, nO2/ncat), 13.2,123 and TOF, 4.2 × 10−3 s−1,124 were moderate when using CeIV as oxidant, this seminal study 11871

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 4. Comparison of (a) Direct Reoxidation by Molecular Oxygen, (b) Reoxidation via Low-Energy Electron Transfer by the Use of ETMs, and (c) Principle of a System Using ETMs in H2O Oxidation and the Redox Mediator Effect

Scheme 5. Proposed Pathway of O2 Evolution for the “Blue Dimer”, cis,cis-[(bpy)2(H2O)Ru(μ-O)Ru(H2O)(bpy)2]4+ (8), by Meyer and Co-workers

The mechanism proposed by Hurst and co-workers also involves the generation of a high-valent RuV,V intermediate, but instead of nucleophilic attack by a H2O molecule on one of the ruthenium-oxo species, it was suggested that the bipyridine ligand is attacked at the α-position and subsequently oxidized to give intermediate 14. The generated radical is then quenched by the attack of a second H2O molecule at the β-position, to yield 15, from which subsequent reduction of the ruthenium centers leads to O−O bond formation (16). This is finally followed by O2 release and the recovery of the aromaticity of the bpy ring (Scheme 6).132

The comprehensive mechanistic studies have also generated significant information concerning the decomposition pathways of the blue dimer. The general cause of the low activity is believed to be related to the instability of the μ-oxo bridge, which upon cleavage results in the breakdown of the dimeric structure into nonactive monomeric ruthenium complexes. Continued research therefore focused on finding more stable organic ligands capable of bringing two ruthenium centers together in close proximity, which would enable more efficient H2O oxidation catalysis. 11872

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 6. Previously Proposed Mechanism for O−O Bond Formation for the “Blue Dimer”, cis,cis-[(bpy)2(H2O)Ru(μO)Ru(H2O)(bpy)2]4+ (8), Where the bpy Ring Is Actively Involved in the Catalytic Mechanisma

a

This mechanism has been proven improbable for O2 generation, but the initial steps might be important for decomposition of the ligand frameworks.

oxidizing H2O to O2. In this complex, the two ruthenium metal centers have deliberately been placed in close proximity, oriented in a cis fashion to one another, by the introduction of a rigid pyrazole ligand backbone as the bridge between the two metal centers.160 Extensive mechanistic studies on the dinuclear complex 17 demonstrated that oxygen−oxygen bond formation, in contrast to the “blue dimer” 8, proceeds solely via the I2M pathway. Moreover, as compared to the blue dimer, the rate of O2 evolution is more than 3 times faster for the Ru− Hbpp complex 17 under similar conditions. This significant increase in activity results from the more favorable ligand matrix of complex 17, which preorientates the metal centers and facilitates intramolecular coupling of the two ruthenium(IV)-oxos.161,162 Complex 17 benefits from the absence of the μ-oxo bridge, thus avoiding decomposition by reductive cleavage and a lower degree of competing anation side reactions,163 which was a limitation of the blue dimer. Although complex 17 lacks the μ-oxo bridge, it still suffers from a low TON of 17.5, and displayed only an overall efficiency of 70% with regard to the oxidant CeIV. This could partially be alleviated by anchoring catalyst 17 onto a solid support,164

Another well-studied ruthenium WOC, the Ru−Hbpp catalyst [Ru2(OH2)2(bpp)(tpy)2]2+, 17 (Figure 8; Hbpp =

Figure 8. Structure of the Ru−Hbpp catalyst, [Ru2(OH2)2(bpp)(tpy)2]2+ (17), developed by Llobet and co-workers.

2,2′-(1H-pyrazole-3,5-diyl)dipyridine), was reported by the group of Llobet in 2004.159 Complex 17 was the first dinuclear ruthenium complex lacking a Ru−O−Ru motif capable of

Figure 9. Molecular structures of the two ruthenium complexes [(py-SO3)(OH2)Ru(μ-Mebbp)Ru(OH2)(py-SO3)]3− (18) and [(py)2Ru(μMebbp)(μ-OAc)Ru(py)2]2+ (19). 11873

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

whereupon its performance improved dramatically. Under optimized conditions, the catalyst proved capable of oxidizing H2O to O2 with a TON of 250.165 The Ru−Hbpp catalyst has also been structurally modified to allow for electrostatic interaction with several solid supports, such as SiO2 and FTO-TiO2 (FTO = fluorine-doped tin oxide).166 Llobet and co-workers also synthesized the two structurally related ruthenium complexes: the diaqua complex [(pySO3)(OH2)Ru(μ-Mebbp)Ru(OH2)(py-SO3)]3− (18) and the acetate bridged complex [(py)2Ru(μ-Mebbp)(μ-OAc)Ru(py)2]2+ (19), depicted in Figure 9. They successfully demonstrated how subtle changes in the ligand architecture can be used to control the O−O bond formation pathway. An interesting observation was that all attempts to convert ruthenium complex 19 to the diaqua complex 18 failed. This is in contrast to the related Ru−Hbpp catalyst 17 where dissolving the acetate bridged version in acidic pH at room temperature immediately resulted in the formation of the aqua complex [Ru2(OH2)2(bpp)(tpy)2]2+ (17).167 This dramatic difference in reactivity was found to be related to a structural effect exerted by the Mebbp ligand, where ruthenium complexes containing the Mebpp ligand have all of the equatorial atoms from the Mebpp and acetate ligands closely separated in the equatorial plane. In contrast to complex 17 housing the bpp ligand, the acetate ligand in complex 19 is confined in a syn fashion to produce a more stable acetate bridged ruthenium complex. These variations in coordination also translate into a distinct difference in the observed catalytic activity, where access to free sites for coordination of aqua ligands is not granted in the acetate complex 19. This prevents complex 19 from participating in H2O oxidation, and, as a consequence, only the diaqua complex 18 turned out to be an active WOC.167 By use of isotopically labeled H2O, the authors managed to provide evidence that the diaqua complex 18 mediated the O− O bond formation via a WNA-type mechanism. This is in contrast to the Ru−Hbpp catalyst in which there exists a distinct “through space interaction” of the two ruthenium−oxo moieties, which subsequently couple to generate the O−O bond via the so-called I2M mechanism. These findings highlight that a small structural modification of the ligand scaffolds can induce subtle geometrical changes in the arrangement of the two Ru−OHx units, which in turn leads to a switch in the mechanism for O−O bond formation. It can also be concluded that the mechanistic landscape for the WNA and I2M pathways can be closely related and that subtle variations in the ligand environment might lead to an alteration of the activation barriers for the two processes, thus favoring one mechanism over the other.167 The Ru−Hbpp complex 17 has also been modified to yield a dinuclear ruthenium complex that has a trans-disposition of the Ru−OH units. In this complex, the rigid meridionally coordinating tpy ligand was substituted with the more flexible tridentate bpea ligand (bpea = N,N-bis(pyridin-2-ylmethyl)ethanamine) that can coordinate in both a facial and a meridional fashion. This ligand exchange resulted in the dinuclear ruthenium complex trans,fac-{[Ru(OH)(bpea)]2(μbpp)}3+ (20) where the two OH-groups are trans to each other (Figure 10). The cis complex did not form due to steric constraints. Complex 20 was found to be active in catalyzing H2O oxidation, when using CeIV as oxidant. However, the rate of O2 formation was significantly reduced as compared to the Ru−Hbpp catalyst 17 where the two aqua ligands point toward

Figure 10. Molecular structures of the trans,fac-{[Ru(OH)(bpea)]2(μbpp)}3+ (20) and trans,fac-{[Ru(OH2)(tpym)]2(μ-bpp)}3+ (21) complexes.

each other in a favorable in,in-fashion, which allows for intramolecular coupling to take place. On the other hand, trans complex 20 with its electronic constraints most likely operates through a different mechanism, either via nucleophilic attack of a solvent H2O molecule to a high-valent rutheniumoxo intermediate or via a bimolecular mechanism where interaction of two RuO units from two different molecules leads to O−O bond formation.168 A subsequent paper by the authors utilized the facial tpym ligand (tpym = tris(2pyridyl)methane) for construction of trans complex 21. H218O isotopic labeling experiments showed that the trans dinuclear ruthenium complex 21 operated via a ratedetermining step involving a bimolecular O−O bond formation step, precluding that an intramolecular I2M mechanism is operating.169 Later, Thummel and co-workers synthesized a series of dinuclear complexes (22−34) that also contained rigid polypyridyl-based ligands as the bridge between the two ruthenium centers (Figure 11). The authors were interested in evaluating the effect that the axial ligands exerted on the properties and catalytic activity of the corresponding dinuclear ruthenium complexes. Kinetic analysis revealed that the initial rate of O2 evolution was first-order in catalyst concentration, suggesting cooperation between the two ruthenium centers. As compared to Meyer’s “blue dimer” and Llobet’s Ru−Hbpp catalyst, the polypyridyl-based complexes developed by Thummel displayed significantly higher TONs (up to 600 when CeIV was used as the chemical oxidant).170,171 More recently, Llobet and co-workers have prepared, isolated, and thoroughly characterized a family of tetranuclear ruthenium complexes consisting of two subunits containing the bpp-ligands linked by a meta-, para-, or ortho-substituted xylyl moiety.172 In this study, the catalytic performances of the four aqua complexes, the three tetranuclear ruthenium complexes 35−37, and their dinucleating counterpart 38, proved to be different (Figure 12). It was found that when employing an excess of CeIV, all of the complexes generated significant amounts of both O2 and CO2. Examination of the tetranuclear complexes 35−37 revealed that they were similar to those of their dinuclear counterpart 38. This provided evidence that the xylylic spacer does not provide any electronic coupling of the ruthenium centers housed in the two ligand subunits in 35−37. However, the relative ratio of [O2]/[CO2] proved to be dependent on both the substitution pattern of the xylyl moiety and the concentration of the starting complexes. The generation of CO2 proved to be significantly different for each of the four ruthenium−aqua complexes and can be rationalized in terms of the accessibility of the easily oxidized methylenic moieties in each complex. 11874

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 11. Dinuclear ruthenium complexes developed by the group of Thummel.

tBu2qui)2(btpyan)]2+ (39, Figure 13; 3,6-tBu2qui = 3,6-di-tbutyl-1,2-benzoquinone; btpyan = 1,8-bis(2,2′:6′,2″terpyridyl)anthracene) where the quinone ligand is an essential and highly active component of the metal complex, with its ability to function as an electron reservoir during catalysis. The electronic absorption spectrum of complex 39 displayed a strong band at 576 nm. Addition of a base led to the disappearance of this band, with the appearance of a new band at 850 nm. Treatment of the resulting solution with acid restored the band at 576 nm. On the basis of these results, the absorption bands at 576 and 850 nm were assigned to the MLCT transitions of the RuII → quinone and RuII → semiquinone moieties, respectively. The reversibility of the system invoked the acid−base equilibrium between the corresponding quinone and semiquinone ruthenium complexes (eq 14): [Ru 2(OH)2 (Q)2 (btpyan)]2 + → [Ru 2(O•)2 (SQ)2 (btpyan)] (14) Figure 12. Structures of the tetranuclear ruthenium complexes 35−37 together with their dinucleating counterpart 38.

It should be noted that the dissociation of two protons from the hydroxo groups triggers the reduction of the quinone moieties to semiquinones. This reduction can be explained by intramolecular electron transfer from the oxo ligands to the quinone units.176 By contrast, in the analogous bpy complex [Ru2(OH)2(bpy)2(btpyan)]2+ 40, no change in the electronic absorption spectra was observed upon addition of base.174 The dinuclear ruthenium complex 39 was shown to be insoluble in aqueous solutions and was therefore successfully deposited on an indium tin oxide (ITO) electrode. Electrochemical measurements of the complex 39 on the modified

In the preceding ruthenium complexes, only the metal centers constituted the redox-active part. However, an interesting approach by Tanaka and co-workers made use of a ruthenium complex containing a redox-active ligand.173−175 Dioxolenes that are coordinated to metal centers may exist in three different redox states: (i) quinone (Q), (ii) semiquinone (SQ), or (iii) catechol (cat). The authors synthesized the dinuclear ruthenium complex [Ru2(OH)2(3,611875

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 13. Structure of the [Ru2(OH)2(3,6-tBu2qui)2(btpyan)]2+ complex 39 containing the redox-active quinone ligand, the analogous bpy complex [Ru2(OH)2(bpy)2(btpyan)]2+ 40, the single-site analogue of complex 39, [Ru(OH2)(3,6-tBu2qui)(tpy)]2+ (41), and the [Ru2(μCl)(bpy)2(btpyan)]3+ complex 42.

Figure 14. Representaion of the related Ru complexes 43−45.

(OH2)(3,6-tBu2qui)(tpy)]2+ (41), which was catalytically inactive, it was demonstrated that the presence of two ruthenium centers in close vicinity was essential for obtaining an active WOC and provided evidence that the two ruthenium−hydroxo moieties in 39 are key structural features.173,182,183 The electronic effects of the dioxolene ligands have also been studied by introduction of various substituents.184 It was found that the presence of electronwithdrawing substituents in the dioxolene units shifted the resonance equilibrium from the RuII(SQ) complex to the RuII(cat). These substituents could also inhibit O2 evolution activity, most likely due to the increased redox potentials.185 A structural analogue of the [Ru2(OH)2(3,6-tBu2qui)2(btpyan)]2+ complex 39, [Ru2(μ-Cl)(bpy)2(btpyan)]3+ (42), has also been reported to mediate H2O oxidation.186 It should also be pointed out that the groups of Berlinguette and Nocera have developed related Ru complexes (Figure 14) housing different spacer ligands for studying the O−O formation process.187,188 Most of the developed WOCs suffer from a serious drawback: for the oxidation to be thermodynamically favorable, they need to be driven by a strong oxidant such as CeIV. Because CeIII has low absorption in the visible region, CeIV cannot be photochemically generated from CeIII, thus making its use in large-scale applications questionable. An attractive approach would therefore be to drive H2O oxidation with [Ru(bpy)3]3+-type oxidants, which can be photogenerated from the corresponding well-studied [Ru(bpy)3]2+-type complexes. However, for this to be realized, the redox potentials of the WOCs need to be reduced. Thus far, the WOCs containing

ITO electrode displayed two peaks at 0.60 and 1.40 V vs NHE, which were assigned to the [Ru2II(O•)2(Q)2(btpyan)]2+/ [Ru2II(O•)2(SQ)2(btpyan)] and [Ru2III(O•)2(Q)2(btpyan)]2+/ [Ru2III(O•)2(SQ)2(bt-pyan)] redox couples, respectively. Hence, the first redox process is a ligand-centered oxidation, while the event occurring at 1.40 V corresponds to a metallocalized oxidation of RuII to RuIII. At potentials >1.70 V, there existed a strong increase in the anodic current that was assigned to the oxidation of H2O. The modified ITO electrodes with the deposited complex 39 proved to be highly active catalytic systems, capable of reaching TONs as high as 33 500. After 40 h of electrochemical catalysis, the evolution of O2 ceased, probably as a result of gradual exfoliation of the catalyst from the electrode surface. The analogous [Ru2(OH)2(bpy)2(btpyan)]2+ complex 40 did not generate O2 to the same extent as complex 39, which suggested that the presence of the redox-active quinone units in 39 was fundamental for the catalytic activity.174 The redox noninnocence of the quinone ligand in complex 39 has also been verified by quantum chemical calculations. It was suggested that the quinone ligands were of fundamental importance by accommodating electrons, thereby favoring catalysis by lowering important barriers during the reaction pathway for H2O oxidation.177−181 The use of a bridging ligand to bring the two ruthenium centers in close proximity is essential for the intramolecular coupling to generate the crucial O−O bond. The btpyan ligand fixes the ruthenium centers at a suitable distance to allow for O−O bond formation. By synthesizing the single-site analogue of complex 39, [Ru11876

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

the new ligand with the ruthenium precursor Ru(DMSO)4Cl2.194 Complex 47 proved to be a robust and efficient catalyst both in chemical and light-driven H2O oxidation. Under optimized conditions, an impressive TON value of >10 000 and a TOF of 1.2 s−1 could be reached when CeIV was used as oxidant at acidic pH.194 Electrochemical measurements confirmed that complex 47 displayed a pronounced catalytic current toward H2O oxidation at pH 1 when the potential was higher than 1.50 V vs NHE. Interestingly, under neutral pH, the catalytic potential was significantly lower than that of complex 46, with an onset potential of ∼1.20 V, thereby indicating that the oxidation could be driven by photogenerated [Ru(bpy)3]3+. Subsequently, complex 47 was evaluated under neutral conditions in an artificial light-driven H2O oxidation system consisting of catalyst 47, the sacrificial electron acceptor (S2O82−), and the photosensitizer [Ru(bpy)3]2+. Upon irradiation with visible light, O2 production was immediately triggered, and a TON of ∼60 and a TOF value of 0.1 s−1 were obtained for complex 47. Furthermore, by employing the more strongly oxidizing photosensitizers [Ru(bpy)2(deeb)]2+ and [Ru(bpy)(deeb)2]2+ (deeb = diethyl 2,2′-bipyridine-4,4′-dicarboxylate), both the initial rate and the TON of the light-driven H2O oxidation could be significantly increased.194 During light-driven H2O oxidation, the pH of the reaction mixture was greatly decreased, to pH ∼3. This was the major reason for the cease of O2 evolution, similar to what was also found for complex 46.193 Neutralization of the reaction pH and addition of more [Ru(bpy)3]2+ gave rise to renewed O2 generation upon continued irradiation, clearly demonstrating that the dinuclear catalyst was photostable in aqueous solutions.194 Recently, the group of Åkermark reported on the synthesis of a dimeric ruthenium complex 48 based on a new type of bioinspired ligand (Figure 16). It was originally proposed that

neutral polypyridyl ligand scaffolds require a potential of >1.40 V vs NHE to carry out the four-electron oxidation of H2O. Appropriate electronic modifications of the ligand backbones by, for example, introduction of electron-donating groups could lower the redox potentials.189 However, a more effective way of achieving this was accomplished by the group of Åkermark and Sun, who demonstrated that lowering of the oxidation potential of the catalysts could be accomplished by introducing anionic ligands into the ligand frameworks of the WOCs. Generally, the redox potentials of metal complexes can be tuned by ligand modification, and the incorporation of negatively charged ligands into metal complexes has previously been shown to stabilize metal centers at higher oxidation states and result in complexes with significantly lowered redox potentials.190,191 With this in mind, the group of Åkermark and Sun recently designed a new type of ligand containing two carboxylic groups in the 6-positions of the pyridyl motifs. Interestingly, complexation with Ru(DMSO)4Cl2 under basic conditions resulted in a dinuclear ruthenium complex (46, Figure 15) with

Figure 15. Molecular structures of the dinuclear ruthenium complexes 46 and 47.

an “anti” structure, in which the two ruthenium centers occupy positions on opposite sides of the central pyridazine ring.192 In the presence of CeIV as oxidant at pH 1, the dinuclear complex 46 could promote H2O oxidation, giving a TON of ∼1700 that is considerably higher than those reported for the previously related complexes comprising neutral ligand scaffolds.123,159,170 In an aqueous solution at pH 7, at potentials higher than 1.30 V vs NHE, complex 46 showed a catalytic current ascribed to H2O oxidation. Although the oxidation potentials of complex 46 were significantly lower as compared to those of complexes bearing neutral ligand backbones, the overpotential toward H2O oxidation was not sufficiently low to allow photogenerated [Ru(bpy)3]3+ to drive H2O oxidation. However, [Ru(bpy)3]2+ derivatives that have oxidation potentials spanning from 1.37 to 1.54 V vs NHE were thus employed as photosensitizers for light-driven H2O oxidation, with TONs reaching up to ∼1300 in the presence of persulfate as electron acceptor. It was also shown that the catalyst was still active when O2 evolution had ceased. This was demonstrated by neutralization of the reaction mixture (pH 7) and addition of extra photosensitizer and electron acceptor, which resulted in resumed production of O2 upon illumination.193 It was believed that complex 46 would benefit from having the two ruthenium centers in a cis-fashion. By installing an aromatic ring at the 4,5-positions of the pyridazine moiety in the ligand backbone in complex 46, the cis-dimeric ruthenium complex 47 (Figure 15) was obtained upon complexation of

Figure 16. Structure of the dimeric ruthenium complex 48 containing the bioinspired ligand (L = 4-picoline).

the ligand would be capable of accommodating two ruthenium centers with the phenolic part acting as the bridge between the two metal centers. However, upon complexation, the expected dinuclear ruthenium complex was not obtained, but instead the dimeric complex 48 formed exclusively. Although the desired dinuclear complex was not afforded, the authors examined the catalytic activity of complex 48 for H2O oxidation. Preliminary catalytic experiments employed CeIV as oxidant in acidic solution (pH 1). It was found that an immediate evolution of O2 could be triggered upon addition of the oxidant to an aqueous solution of 48, resulting in a high TON of 1000.195 Electrochemical measurements in aqueous phosphate buffer at pH 7.2 showed that complex 48 displayed an electrochemical 11877

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 1. Summary and References to Available Data for Dinuclear Ruthenium Water Oxidation Catalysts

11878

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 1. continued

11879

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 1. continued

11880

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 1. continued

11881

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 1. continued

11882

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 1. continued

a

Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole of catalyst per s−1. cUsing Ce(NH4)2(NO3)6 (CAN, CeIV) as the chemical oxidant. dElectrolysis at a potential of 1.90 V vs NHE in H2O/trifluoroethanol (1:9, v/v) solutions. eDeposited onto an ITO electrode with subsequent electrolysis at a potential of 1.90 V vs NHE in H2O at pH 4.0. fPhotochemical oxidation using [Ru(bpy)2(deeb)](PF6)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. gPhotochemical oxidation using [Ru(deeb)2(dmbpy)](PF6)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. h Photochemical oxidation using [Ru(deeb)2(bpy)](PF6)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. iPhotochemical oxidation using [Ru(bpy)2(deeb)](PF6)2 as photosensitizer and [Co(NH3)5Cl]Cl2 as the sacrificial electron acceptor. jPhotochemical oxidation using [Ru(deeb)2(dmbpy)](PF6)2 as photosensitizer and [Co(NH3)5Cl]Cl2 as the sacrificial electron acceptor. kPhotochemical oxidation using [Ru(deeb)2(bpy)](PF6)2 as photosensitizer and [Co(NH3)5Cl]Cl2 as the sacrificial electron acceptor. lPhotochemical oxidation using [Ru(bpy)3]Cl2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. mUsing [Ru(bpy)3](PF6)3 as the chemical oxidant. nPhotochemical oxidation using [Ru(bpy)3](PF6)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. bpy = 2,2′-bipyridine, deeb = diethyl 2,2′-bipyridine-4,4′dicarboxylate, dmbpy = 4,4′-dimethyl-2,2′-bipyridine.

2.2. Single-Site Ruthenium-Based Catalysts − Simple and Efficient

onset potential for H2O oxidation starting at 1.20 V vs NHE (Table 1). This indicated that [Ru(bpy)3]3+-type complexes, photogenerated from the corresponding [Ru(bpy)3]2+ complex, could be used to drive H2O oxidation, and as expected evolution of O2 could be detected when an aqueous solution containing 48 was added to [Ru(bpy)3]3+. Initially, an artificial system containing a sacrificial electron acceptor (S2O82−), the photosensitizer [Ru(bpy)3]Cl2, and catalyst 48 was used for the photochemical H2O oxidation experiments, but when using these conditions no oxygen evolution could be detected. However, when the anion of the photosensitizer was changed from Cl− to PF6−, oxygen evolution was observed with a TON of ca. 30.195 A similar effect has also been observed for a singlesite ruthenium-based WOC when using [Ru(bpy)3]Cl2 as the photosensitizer.196 Furthermore, substituting [Ru(bpy)3](PF6)2 with the more strongly oxidizing photosensitizer [Ru(bpy)2(deeb)](PF6)2 resulted in an increase of the TON to 250. Interestingly, addition of more photosensitizer led to the generation of more O2, indicating that decomposition of photosensitizer was the major limitation of this photocatalytic system. To demonstrate that both of the oxygen atoms in the O2 produced originated from H2O, isotopic labeling experiments were conducted. By using 4.5% 18OH2 together with mass spectrometry, it could be established that both of the oxygen atoms originated from solvent H2O. Altogether, this study constitutes an illustrative example of how incorporation of negatively charged functional groups into the ligand backbones can lead to WOCs that can be driven by light.195

Because of the multimetallic core of the OEC, it was long envisioned that artificial molecular WOCs must accommodate multiple metal centers to cope with the accumulation of the four oxidizing equivalents needed for H2O oxidation. The initial lack of reports on single-site WOCs led to the creation of a paradigm, which claimed that at least two metal centers were required for H2O oxidation to occur. However, this early belief has now been disproved, and today there exists a variety of single-site catalysts that can mediate the four-electron oxidation of H2O. Although there were some early reports on single-site ruthenium complexes being active in H2O oxidation,197,198 these claims were not properly supported by experiments. Specifically, the reported redox potentials of these complexes did not match the thermodynamic potential required to drive H2O oxidation at the investigated pH. 2.2.1. Mononuclear Ruthenium Catalysts Comprising Neutral Ligand Frameworks. Credible proof that the fourelectron oxidation of H2O could occur on single-site metal complexes was provided by the group of Thummel in 2005. The vital ligand backbone in these ruthenium complexes consisted of a tridentate polypyridyl type ligand, 2,6-di(1,8naphthyridin-2-yl)pyridine, with uncoordinated naphthyridine nitrogens (Figure 17). The uncoordinated nitrogens interact with the aqua ligand through hydrogen bonding, thus stabilizing the single-site aqua complexes.170 11883

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 17. Molecular structures of the single-site ruthenium complexes 49−51.

The electronic absorption data of complexes 49−51, in an acetone solution, showed a shift of the MLCT band, a result of the excitation of an electron from a metal-centered d-orbital to a π*-orbital of the most electronegative ligand in the complex. When the electron-donating ability of the axial ligands was varied, a red-shift was observed. Of the synthesized complexes, complex 49, containing the most electron-rich axial ligand (NMe2 substituted), exhibited the lowest energy-transition (longest wavelength). The same general trend was observed in electrochemical studies, where complex 49 proved to be more easily oxidized to the RuIII state and more stabilized at this state as compared to complexes 50 and 51, as a result of more pronounced electron donation from the ligand. However, in the catalytic H2O oxidation experiments, the dimethylaminosubstituted complex 49 turned out to be the least active catalyst, most likely as a result of protonation of the amino group under the acidic conditions.199 Further elaboration of complexes 49−51 provided an extensive library of related single-site complexes, which were examined for their activity toward catalytic H2O oxidation (Figures 18 and 19).200 Within the [Ru(tpy)(bpy)Cl]+ series 52−56, the electronic absorption spectra displayed the same trend as for complexes 49−51, with a shift in the MLCT band toward longer wavelengths for the electron-withdrawing substituted complexes. A similar trend was also observed for the redox properties of these complexes, making it possible to correlate the electron-donating property of the ligands to the RuIII/RuII redox potentials. Complexes containing electronwithdrawing ligands caused a destabilization of the RuIII state, which resulted in higher redox potentials. Measurements of O2 evolution, conducted on complexes 52−56, revealed that the more electron-rich complexes 53 and 54 displayed a higher catalytic rate for O2 evolution than the complexes with electron-withdrawing ligands, which on the other hand gave higher TONs. Hence, it appeared for catalysts 52−56 that there existed a reverse relationship between catalytic rate and TON. In the second group of molecular catalysts (57−62), the bpy ligand backbone was replaced by benzo- or diaza-analogues of bpy. Complex 57, housing the phenantroline ligand, was the only complex in this group that showed catalytic activity. The reason for this was believed to be related to the steric hindrance caused by the benzo-derivative ligands in complexes 58−60. However, it could also originate from their decreased ability to exchange chloride by H2O.200 The influence of steric strain in ruthenium-based WOCs has recently been thoroughly investigated by the group of Thummel.201

Figure 18. Molecular structures of the [Ru(tpy)(NN)Cl]+ complexes 52−62 (where NN = bidentate polypyridyl ligand).

For complexes 61 and 62, the bpm and bpz ligands (bpm = 2,2′-bipyrimidine; bpz = 2,2′-bipyrazine) are most likely protonated under the acidic conditions. Protonation would lead to an increase of the redox potentials and might thus make it thermodynamically unfavorable to drive H2O oxidation when using CeIV as the chemical oxidant. Interestingly, complexes 65, 66, and 68 were found to generate O2, while complexes 1, 63, 64, and 67 were unable to produce any detectable amount of O2.200 The discovery of the highly important feature that H2O oxidation could occur at a single metal center represented a paradigm shift for the field of H2O oxidation catalysis and quickly resulted in an outburst of reports on active single-site ruthenium WOCs. Molecular single-site catalysts offer the possibility of straightforward ligand design, synthesis, and characterization. The relative ease with which the ligand environment can be tuned in these molecular systems, both electronically and sterically, makes them amenable for structure−activity relationship studies. Advantages are also offered when conducting detailed mechanistic studies, from both an experimental and a theoretical point of view. However, the relative ease with which these molecular mononuclear catalysts can be developed and thoroughly studied is not solely the fundamental objective to be pursued. The incorporation and/or attachment of these catalysts to viable supramolecular assemblies, for example, linking them to a chromophore for light-absorption, for H2O splitting, is also greatly facilitated. Although the group of Thummel did not conduct mechanistic studies to verify the unimolecular nature of their catalystic reactions, unambiguous evidence that H2O oxidation proceeded through single-site catalysis was provided by Meyer and co-workers in 2008.202 By replacing the chloride in the inactive complexes 61 and 62, the authors were able to conduct 11884

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 19. Various single-site ruthenium complexes studied by Thummel and co-workers.

detailed mechanistic studies of catalytic H2O oxidation on the two single-site ruthenium−aqua WOCs [Ru(tpy)(bpm)(OH2)]2+ (69) and [Ru(tpy)(bpz)(OH2)]2+ (70) (Figure 20).

Scheme 7. Proposed Catalytic Mechanism, Proceeding via the So-Called WNA-Mechanism, for the Single-Site Ruthenium-Based WOCs [Ru(tpy)(bpm)(OH2)]2+ (69) and [Ru(tpy)(bpz)(OH2)]2+ (70)

Figure 20. Single-site ruthenium−aqua complexes [Ru(tpy)(bpm)(OH2)]2+ (69) and [Ru(tpy)(bpz)(OH2)]2+ (70) studied by Meyer and co-workers.

On the basis of kinetic and thermodynamic evidence, it was proposed that these single-site ruthenium WOCs catalyzed the oxidation of H2O through a seven-coordinated ruthenium center, involving a η2-peroxido intermediate, [RuIV-η2-OO]2+. Starting from [RuII−OH2]2+, the addition of 3 equiv of CeIV resulted in the formation of the key intermediate [RuVO]3+, which triggered nucleophilic attack of a solvent H2O molecule to generate a peroxide intermediate [RuIII−OOH]2+, via the WNA mechanism. Further oxidation of this species, through proton-coupled oxidation, supposedly formed the [RuIV− OO]2+, which finally liberated O2 (Scheme 7). Kinetic experiments (in 0.1 M HNO3) revealed that the rate-limiting step in the catalytic cycle of H2O oxidation was the release of O2 from the peroxide intermediate [RuIV−OO]2+ to regenerate [RuII−OH2]2+.202 It should be noted that a later study provided

conclusive evidence that the catalytic mechanism by which ruthenium complex 50 oxidizes H2O also is of unimolecular nature.203 However, complexes 49−51 appear to prefer to undergo a strict PCET pathway, avoiding the formation of the “key intermediate” [RuVO]. The results suggested a mechanism where catalytic H2O oxidation, under both neutral and basic conditions, is triggered by a one-proton-one-electron pathway involving a [RuIVO]2+ species. This, together with the fact that the onset potentials for catalytic H2O oxidation in the investigated pH region are lower than the [RuVO]/ [RuIVO] redox couple, implies that catalytic H2O oxidation 11885

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

proceeds by a thermodynamically more favorable “direct pathway” via a [RuIVO]2+, thus avoiding the higher valent [RuVO]3+ complex.204 For the [Ru(tpy)(NN)Cl]+-type complexes, substitution of the chloride ligand by a solvent H2O molecule generates the corresponding aqua complex [Ru(tpy)(NN)(OH2)]+, which is considered to be the active species during H2O oxidation. To validate this, and to develop a more active catalyst, the group of Sakai carried out further studies with the [Ru(tpy)(bpy)Cl]+ complex, where the Cl− in 52 was substituted by H2O, to generate [Ru(tpy)(bpy)(OH2)]2+ (71). The authors were able to confirm that this modification led to a WOC with higher activity and improved stability. Monitoring of the MLCT band of 71 at 476 nm suggested that the initial RuII species was not involved in the catalytic cycle of O2 evolution. Addition of a reductant (ascorbic acid) to a postcatalytic solution regenerated the MLCT band of the aqua complex 71, thus confirming the robustness of the catalyst under the catalytic conditions. The induction period for the chloro complex 52 and the suppression of H2O oxidation in the presence of NaCl clearly demonstrated the negative effect of Cl−, confirming that the aqua species, [Ru(tpy)(bpy)(OH2)]2+, was the active intermediate.205 The chloride effect also demonstrated the importance of easy access to a vacant coordination site for H2O at the metal center, to obtain active catalysts for the oxidation of H2O (cf., complexes 1 and 67, which are catalytically inactive). Berlinguette and co-workers conducted a detailed study of the [Ru(tpy)(bpy)(Cl)]+-type family and their corresponding aqua complexes, [Ru(tpy)(bpy)(OH2)]2+ (Figure 21). By

the tpy ligand did not have an equally strong influence on the redox potentials as when the bpy ligand was modified (Table 2 Table 2. Electrochemical Data for [Ru(tpy)(bpy)Cl]+ (48) and the [Ru(tpy)(bpy)(OH2)]2+-Type Complexes 67−76a E/V vs NHE

OMe

Cl

COOH

complex

RuIII/RuII

52 71 72 73 74 75 76 77 78 79 80

1.05 1.04 0.99 0.91 0.87 1.07 1.11 1.16 1.11 1.16 1.20

RuIV/RuIII

RuV/RuIV

1.15 1.19 1.18 1.11 1.13 1.16 1.21

1.73 1.69 1.63 1.65 1.74 1.71 1.74 1.80 1.80 1.82

a

Adapted with permission from ref 206. Copyright 2010 American Chemical Society. Conditions: Electrochemical data were obtained in 0.1 M HNO3 aqueous solutions (pH 1.07) with a 1 mM concentration of the complex, measured with a scan rate of 50 mV s−1.

and Figure 22). Different effects of the ligand substituents on the catalytic activity of the catalysts were noticed for the bpy and tpy substituted complexes. Thus, the introduction of electron-withdrawing substituents on the bpy ligands led to decreased catalytic rates, while the same modification of the tpy ligand resulted in increased catalytic rate. This demonstrates that subtle modifications can have an unpredicted influence on the catalytic properties.206 To elucidate possible deactivation pathways, a solution containing complex 71 was examined after the addition of 1000 equiv of CeIV. The oxidation product 2,2′-bipyridine N,N′dioxide was characterized by NMR spectroscopy and mass spectrometry. However, no evidence for the dissociation of the tpy ligand could be found. It was also revealed that the introduction of substituents on the ligand located trans to the coordinating aqua ligand (Ru−O bond) in these complexes had a stronger influence on the chemical properties than the cis ligands. When comparing the group of complexes 72, 75, and 78 with 73, 76, and 79, it was observed that modification at the bpy ligand had a more powerful effect on the properties of the corresponding complexes. Density functional theory (DFT) calculations further revealed that the Ru−N bond situated trans to the Ru−O was weakened at higher oxidation states of the ruthenium center. This was manifested as an elongation of the trans Ru−Nbpy bond, whereas the Ru−Ntpy bonds were kept rather unperturbed.206 Subsequent, extensive mechanistic studies on the [Ru(tpy)(bpy)(OH2)]2+-type complexes 71, 73, and 79 were also carried out by the group of Berlinguette.207,208 The bpy ligand was selected as the position of modification based on the previous observation that the bpy ligand located trans to the Ru−O bond exerted a stronger influence on the properties of the corresponding WOCs. Spectrophotometric techniques and electrochemistry indicated that slight changes of the reaction conditions and the electron density on the ligand frameworks had a tremendous impact on catalytic activity and stability, opening different reaction pathways. The key intermediate in these catalytic reactions was proposed to be a high-valent RuVO, as suggested by Sakai,209 Yagi,210 and Meyer and co-

Figure 21. Structures of the [Ru(tpy)(bpy)(OH2)]2+-type complexes 71−80 studied by the groups of Berlinguette, Sakai, and Yagi.

installing a range of substituents in the periphery of both the bpy and the tpy ligand frameworks, they were able to get specific insight into the influence of the electronic properties on the activity of the single-site ruthenium complexes.206 Furthermore, they were able to confirm the findings of Sakai and co-workers, that the aqua complexes exhibited a higher catalytic activity and shorter induction period than the chloride ligated complexes. By the use of 1H NMR and UV−vis spectroscopy, they monitored the chloride−aqua ligand exchange, that is, the conversion of 52 into 71, and found that the Ru−Cl bond was preserved in [D6]DMSO, but upon dissolution in D2O, significant quantities of the aqua complex [Ru(tpy)(bpy)(OD2)]2+ were formed. The rate of chloride− aqua exchange proved to be dependent on the electronic properties of the bpy ligand, where electron-withdrawing groups slowed Cl− displacement.206 Electrochemical data for the complexes 71−80 were recorded in aqueous media, and the trend was consistent with the electronic nature of the bpy and tpy ligands. Modifying 11886

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 22. (a) Diagram displaying how modifications introduced to the tpy ligand (72, 75, and 78) and bpy ligand (73, 76, and 79) affect the redox potential for the RuIII/RuII couple in the [Ru(tpy)(bpy)(OH2)]2+-type complexes. This is further visualized in (b), where a further modification on the tpy ligand (73 to 74) does not have a similar effect.

Figure 23. Monomeric ruthenium−aqua complexes studied by Meyer and co-workers.

workers202 for the related single-site ruthenium complexes 69 and 70. Labeling experiments, with H218O, gave compelling evidence for a competing OAT pathway involving the oxidant CeIV.207,208 It was concluded that a substantial amount of the evolved O2 contained an oxygen atom originating from the nitrate anion of CeIV. This OAT pathway of CeIV has also been proposed by other groups,211,212 thus highlighting the

importance of establishing the origin of the oxygen atoms in the evolved O2 when studying WOCs. In light of the fact that oxidation of H2O could occur on a single metal site, Meyer and co-workers synthesized a series of monomeric ruthenium−aqua complexes (69, 70, 71, and 81− 91, Figure 23) and evaluated their performance in H2O oxidation catalysis. The cyclic voltammograms of these complexes displayed peaks in the regions 0.51−1.13 V and 1.24−1.48 V vs NHE, which were assigned to the RuIII/RuII 11887

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

and RuIV/RuIII redox couples, respectively. An additional RuV/ RuIV peak was observed between 1.40 and 1.72 V, as a shoulder at the onset of a catalytic current belonging to the electrocatalytic oxidation of H2O. This clearly demonstrated the impact of σ-donor and π-acceptor ability of the different ligands in stabilizing the ruthenium center at different oxidation states. Remarkably, all of the synthesized complexes were capable of promoting O2 generation when CeIV was employed as chemical oxidant, which highlights the generality of mediating H2O oxidation by single-site ruthenium complexes.213 In a subsequent paper, the carbene-based ruthenium complex [Ru(tpy)(Mebim-py)(OH2)]2+ 81 (Mebim-py = 3-methyl-1pyridylbenzimidazol-2-ylidene) was further studied as a catalyst for H2O oxidation.214 Electrochemical measurements showed that the peak current for the RuIII/II couple (id) for the Ru carbene catalyst was proportional to the square root of the scan rate (υ1/2), as expected for a diffusional redox couple. The catalytic peak currents (icat) at different pH were found to vary linearly with [RuII−OH2]2+, and a linear relationship could also be established between the square root of the scan rate (i/υ1/2) and the catalytic peak currents with decreasing scan rate resulting in increased catalytic peak currents.19b,214 These observations are consistent with a single-site mechanism for H2O oxidation with a contribution to rate-limiting behavior from a step prior to electron transfer to the electrode. This supports a mechanistic pathway similar to that of the [Ru(tpy)(bpy)(OH2)]2+-type catalysts, where PCET events generate a high-valent [RuVO] intermediate, with O−O bond formation occurring through H2O nucleophilic attack. Subsequent oxidation of the generated ruthenium−peroxo species results in the liberation of O2. The most distinctive feature of the carbene catalyst was the drastically enhanced rates of H2O oxidation, as compared to catalyst 84.214 A general synthetic approach for coupling of a chromophore to a WOC, to create molecular assemblies for overall H2O splitting, is to link these two units together through an amide bond. This approach provides a robust linkage and was used to construct a ruthenium assembly, revealing that the amide bridge did not interfere with the photophysical or redox properties of the constituent chromophore and WOC units.215 This method has also been utilized for coupling a phosphonatefunctionalized [Ru(bpy)3]2+-type chromophore to rutheniumbased WOCs.216,217 Attachment of the molecular assembly to a TiO2 surface (92, Figure 24) enabled electron injection from the ruthenium center of the chromophore to TiO2 (eq 15). This was subsequently followed by intra-assembly electron transfer to give [TiO2(e−)−RuaII−RubIII−OH2] (eq 16) on the subnanosecond time scale. These results verified that the preparation of chromophore−catalyst assemblies based on amide bridges offers synthetic tuning and great flexibility in the nature of chromophore, WOC, and the amide linkage.217 In addition to this strategy, electropolymerization has also been used for preparing spatially controlled surface assemblies.218

Figure 24. Structure of the molecular assembly 92 attached to TiO2.

The ruthenium catalyst 84 ([Ru(Mebimpy)(bpy)(OH2)]2+), containing the tridentate Mebimpy ligand (Mebimpy = 2,6bis(1-methylbenzimidazol-2-yl)-pyridine), has also been successfully modified for attachment onto semiconducting oxide surfaces.219−222 By introducing a slight modification in the bpy ligand, the resulting complex 93 ([Ru(Mebimpy)(4,4′((HO)2OPCH2)2bpy)(OH2)]2+) could be anchored to different metal oxides through stable binding of the phosphonate groups (Figure 25). The resulting electrodes were subsequently

Figure 25. Representation of the [Ru(Mebimpy)(4,4′((HO)2OPCH2)2bpy)(OH2)]2+ complex 93 attached to a metal oxide electrode.

evaluated in electrochemical H2O oxidation. With applied potentials >1.85 V vs NHE, it was found that electrocatalytic currents for H2O oxidation were sustained for several hours, thus demonstrating the robustness of the catalytic system. However, more importantly, it was realized that the catalytic peak current at 1.85 V varied linearly with the surface coverage and that the scan-rate-normalized catalytic peak current increased with decreasing scan rate. This is consistent with a rate-limiting step prior to electron transfer to the electrode and with a single-site mechanism for H2O oxidation by the surfacebounded catalyst, thus suggesting that the catalyst retains its solution mechanism when immobilized on the surface.219

[TiO2 −Rua II−Rub II−OH 2] → [TiO2 (e−)−Rua III−Rub II−OH 2]

(15)

[TiO2 (e−)−Rua III−Rub II−OH 2] → [TiO2 (e−)−Rua II−Rub III−OH 2]

(16) 11888

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

An important discovery was also made with the [Ru(Mebimpy)(bpy)(OH2 )] 2+ complex 84 and its related phosphonate derivative [Ru(Mebimpy)(4,4′((HO)2OPCH2)2bpy)(OH2)]2+ 93, both in solution and on oxide surfaces. By employing a strategy called base-assisted concerted atom-proton transfer (APT), a significant acceleration of the O−O bond forming step was achieved. The attack of a solvent H2O molecule on the high-valent RuVO intermediate, the rate-determining step of the catalytic cycle, was dramatically increased upon addition of bases as the result of concerted proton transfer to the base. The concerted APT effect avoids the generation of the high energy peroxide intermediate [Ru(Mebimpy)(bpy)(OOH2)]3+, where the rate constant for O−O bond formation (kO−O) now contains two terms with a base-dependent constant kB (eq 17). Quantum mechanical/molecular mechanical minimal free energy path calculations were also performed on the systems and qualitatively verified the observed experimental trends in reactivity.223 It should be pointed out that a similar APT effect has also been observed for the [Ru(tpy)(bpz)(OH2)]2+ catalyst 70.224

k O − O = kH2O + kB[B]

When comparing assembly 94 to solely the WOC part, the assembly demonstrated enhanced catalytic rates in H2O oxidation catalysis when CeIV was used as oxidant, and provided evidence for a redox mediator effect. The group of Thummel conducted a detailed structure− activity study by preparing and characterizing a wide range of single-site ruthenium complexes of two different types: [Ru(NNN)(NN)(X)]2+ and [Ru(NNN)(pic)2(X)]2+ (Figure 27; NNN = tridentate polypyridyl ligand; NN = bidentate polypyridyl ligand; X = H2O or halide).226 The [Ru(NNN)(NN)(X)]2+-type complexes contain an aqua or halide ligand in the axial position, whereas the [Ru(NNN)(pic)2(X)]2+-type complexes have this ligand in the equatorial position. When using CeIV as the chemical oxidant, a majority of the investigated complexes displayed activity, with complexes 58, 103, 105, and 108 as the exceptions. A general trend that could be observed was that [Ru(NNN)(pic)2(X)]2+-type complexes with picoline ligands were more active than the complexes housing the bpy-type ligands. For the halide containing complexes, the halogen must likely be replaced with an aqua ligand before catalysis can take place. This ligand exchange will enable PCET and give access to ruthenium species of higher valency, which is essential for the oxidation of H2O. For the bromo- and chloro-containing ruthenium complexes, an induction period of several minutes was required before O2 could be detected, indicating that these complexes only served as precatalysts. However, a peculiar observation was that the iodo-containing ruthenium complexes displayed a first-order behavior with an unusual high initial rate for O2 formation, and essentially no induction period. This might imply that substitution with a solvent H2O molecule is not always a requirement for catalysis and that a different in situ generated iodide-containing intermediate accounts for the observed O2 evolution.226 Thummel and co-workers recently reported the synthesis of a ruthenium-based chromophore−catalyst dyad capable of driving H2O oxidation by visible light, when persulfate was used as the sacrificial electron acceptor.227 The dyad was prepared by the use of a pyrazine-based linker and is depicted in Figure 28. Comparing the catalytic activity to the analogous intermolecular component system revealed that the intramolecular system displayed a far superior catalytic activity in H2O oxidation. In a related study,228 the group of Sun also investigated the importance of having access to a vacant coordination site for H2O, to acquire efficient WOCs. The authors synthesized the three analogous complexes [Ru(tpy)(pic)3]2+ (65), [Ru(tpy)(pic)2Cl]+ (98), and [Ru(tpy)(pic)2(OH2)]2+ (97), and evaluated these complexes under both chemical and photocatalytic conditions. The complexes were capable of driving H2O oxidation with CeIV and photocatalytically when using the strong photosensitizer [Ru(bpy)(deeb)2]2+, although with modest activity (TON ∼2, 43, and 84, respectively). In contrast to the chemical oxidations, the photocatalytic reactions showed no induction period and could be explained by a fast ligand exchange induced by light and/or the high pH that facilitated the generation of the catalytically active ruthenium− aqua complex. A structurally related phosphonate functionalized analogue of ruthenium complex 97 has also been loaded on ITO electrodes for electrochemically driven oxidation of H2O.229 To distinguish electronic and hydrogen-bonding effects and to quantify them, fluorine was introduced into the bpy backbone of complexes related to [Ru(tpy)(bpy)Cl]2+ (52)

(17)

Table 3 summarizes the base-dependent constant kB for various bases and illustrates that the kB increases with the Table 3. Rate Constants for the Base-Assisted Oxygen Atom Transfer from RuVO to H2O for Various Bases base

pKa (HB)

kB (M−1 s−1)

H2PO4− −

2.15 4.75 7.20 −1.74

3.8 ± 0.5 10.3 ± 1 48 ± 5 ∼3.5 × 10−3

OAc HPO42− H2O

proton-accepting ability of the base. The addition of HPO42− causes a rate enhancement of ∼1500 as compared to having H2O as the base.223 As noted previously, the surface-bound catalyst [Ru(Mebimpy)(4,4′-((HO)2OPCH2)2bpy)(OH2)]2+ 93 maintained its chemical and physical properties with a similar rate enhancement upon addition of bases.219 The fact that the APT also influences the surface-bound catalyst 93 highlights that enhanced APT reactivity has the possibility of being transferred to semiconductor solution interfaces in an artificial device for solar fuel production.223 The redox mediator effect observed for the “blue dimer”134 has also been employed to the ruthenium-based chromophore−catalyst assembly 94 (Figure 26) in H2O oxidation.225

Figure 26. Representation of the ruthenium-based chromophore− catalyst assembly 94. 11889

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 27. Molecular structures of ruthenium complexes studied by Thummel and co-workers.

showed an unprecedented hydrogen bond between the fluorine atom and one of the hydrogen atoms of the aqua ligand.

and [Ru(tpy)(bpy)(OH2)]2+ (71), affording complexes 120 and 121 (Figure 29).230 The crystal structure of complex 121 11890

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

catalytic PCET events.230 Further work that focused on installing pendant electron-donating groups that could potentially stabilize the different intermediates involved in H2O oxidation resulted in the synthesis of Ru complexes 122− 124 (Figure 30). In the chloride precursors, the installment of the proximal methoxy group on the bpy moiety was found to promote ionization of the choride ligand, presumably by forming a hydrogen-bonding interaction with the incoming aqua ligand. The 6,6′-dimethoxy substituted complex 124 showed higher TONs and TOFs than the 4,4′-isomer, suggesting that the introduction of hydrogen-bonding units proximal to the active site in WOCs provided a way of increasing the efficiencies of the catalysts.231 The design of novel WOCs with improved efficiencies requires insight into the pathways that are responsible for mediating O2 evolution. However, of equal importance is gaining knowledge of the processes that ultimately lead to unproductive pathways and decomposition of the investigated WOC. Also, recognition of routes that are connected to advantageous parallel catalytic pathways would be of particular value. In this regard, the discovery by the group of Llobet that the dinuclear ruthenium WOCs 126 and 127 could be generated from the corresponding [Ru(tpy)(5,5′-X2-bpy)(OH2)]2+ complex (X = H for 71 and X = F for 120) and [Ru(tpy)(O)2(OH2)]2+ (125) in the presence of CeIV under acidic conditions (Scheme 8).232 The improved robustness of the dinuclear ruthenium system was attributed to several factors: (i) electronic coupling of two ruthenium centers through an oxo bridge grants fast intramolecular electron transfer within the different species in the catalytic pathway and allows for distributing oxidizing equivalents over several metal centers, thus alleviating the accumulation of multiple redox equivalents at a single metal center, (ii) the incorporation of the trans-dioxo unit stabilizes the generated dinuclear species at higher oxidation states, in contrast to the catalytically inactive single-site complex 125, and (iii) the oxo bridge and the terminal oxo moieties in complexes 126 and 127 have the potential of promoting hydrogen-bonded intermediates, as was illustrated by calculations, which is critical for reducing the formation of highenergy intermediates. This work clearly showed that single-site ruthenium complexes can in some cases in fact react via dinuclear metal systems, involving generation of dinuclear complexes from the initial single-site complexes.232 Llobet and co-workers have also prepared single-site ruthenium complexes, which are structurally related to the dinuclear Ru−Hbpp catalyst. The four aqua complexes are depicted in Figure 31 and are based on two analogous ligand scaffolds. As a consequence of the meridionally coordinating

Figure 28. Structure of the ruthenium assembly 119 housing the pyrazine-based linker.

Figure 29. Structures of ruthenium complexes 120 and 121.

The presence of fluorine was shown to dramatically influence the electronic properties of the ruthenium center in these complexes. Instead of detecting a stepwise oxidation from RuII → RuIII → RuIV, a two-electron wave was found (eq 18), which is different from complex 71 where only one-electron processes occur.230 [Ru IV O]2 + + 2e− + 2H+ → [Ru II−OH 2]2 +

(18)

This demonstrates that the presence of the fluorine atom increases the RuIII/RuII redox couple but has an opposite effect on the RuIV/RuIII couple. It was further shown that the presence of fluorine in 120 decreased the pKa by 0.8 log units as compared to 52. However, in complex 121, the pKa is increased by 0.6 log units, which was attributed to the hydrogen bonding in 121 that stabilizes the structure. The fact that the two ruthenium complexes 120 and 121 have practically the same electron density at the metal center, according to electrochemical measurements, signals that the hydrogenbonding effect observed in complex 121 is responsible for the dramatic difference in pKa. A significantly lower TON was also seen for complex 121 and was ascribed to the stabilizing hydrogen-bonding effect, which might inhibit important

Figure 30. Structures of Ru complexes 122−124. 11891

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 8. Formation of Ruthenium Complexes 126 and 127

ing ruthenium complexes based on the tmtacn ligand (tmtacn = 1,4,7-trimethyl-1,4,7-triazacyclononane).234 This macrocyclic polyamine ligand coordinates in a facial fashion and is an excellent complement to the polypyridyl based tpy ligand. To explore the effect the bpy ligand has on the redox potentials, complexes 132−134 were synthesized (Figure 32).

Figure 31. Structures of the in- and out-[Ru(tpy)(L)(OH2)]2+ complexes 128−131.

tpy ligand, two isomeric ruthenium complexes are generated, referred to as the in and out isomers. Electrochemical measurements in organic solvents of the corresponding chloro complexes showed only a single redox wave that was assigned to the RuIII/RuII couple. However, all four aqua complexes 128−131 displayed two peaks associated with the RuIII/RuII and RuIV/RuIII redox couples, emphasizing the importance of accessing PCET for the aqua complexes. Further scanning to more anodic potentials resulted in a large anodic current, which was ascribed to the formation of a RuV species that triggers electrocatalytic oxidation of H2O. Moreover, the out isomers (129 and 131) were found to display lower redox potentials than the corresponding in isomers.233 As a result, the out isomers were also more efficient as catalysts in H2O oxidation when using CeIV as sacrificial oxidant. To gain insight into the process of O2 generation, kinetic analyses were conducted by the use of UV−vis spectroscopy. This indicated that release of O2 from a RuIV− OO species was the rate-determining step (eq 19).233 Ru IV −OO + H 2O → Ru II−OH 2 + O2

Figure 32. [Ru(tmtacn)(bpy)(OH2)]2+-type complexes 132−134.

Electrochemical measurements revealed that the tmtacn ligand had a pronounced effect on the redox chemistry of the [Ru(tmtacn)(bpy)(OH2)]2+-type complexes 132−134. The two redox couples RuIII/RuII and RuIV/RuIII were strongly shifted to lower potentials relative to the polypyridyl analogues. This was ascribed to the enhanced σ-donating character and the lack of π-accepting ability of the facial tmtacn ligand as opposed to the meridional tpy ligand. Within the [Ru(tmtacn)(bpy)(OH2)]2+ series, the methoxy substituted complex 134 with the most electron-donating ligand was found to have the lowest redox potentials. However, the potential for the RuV/RuIV redox couple, which was directly connected to the electrocatalytic H2O oxidation, was less affected by the tmtacn ligand. All complexes were examined in CeIV-driven H2O oxidation and were demonstrated to be competent WOCs. Consistent with the electrochemical data, the most electron-rich complex 134 exhibited the highest activity. The stability of the [Ru(tmtacn)(bpy)(OH2)]2+-type complexes was also evaluated by monitoring the MLCT band spectrophotometrically. Addition of CeIV to a solution containing complex 132 resulted in an instantaneous bleaching of the MLCT band, coupled with formation of O2. After O2 evolution had ceased, an excess of ascorbic acid was added to the reaction mixture to reduce ruthenium WOC 132 back to the RuII state. This led to the quantitative regeneration of the MLCT band and indicated that the studied ruthenium WOCs were stable under the catalytic conditions, at least for a couple of turnovers.234 Single-site ruthenium complexes based on the tetradentate ligand TPA (TPA = tris(2-pyridylmethyl)amine) have also been synthesized and evaluated for H2O oxidation.235 The TPA ligand and its derivative have previously been utilized

(19)

It was noticed that complexes 128 and 129, in addition to O2 formation, also generated CO2. This reactivity difference suggests that the labile site for oxidation and decomposition is the phenyl ring in 128 and 129. In the Hbpp-containing complexes 130 and 131, the noncoordinating pyridyl group is most likely protonated at pH 1 under which the catalytic experiments were conducted. This protonation might protect the pyridyl ring from being oxidized, as opposed to complexes housing the phenyl ligand.233 To investigate if H2O oxidation catalyzed by single-site ruthenium complexes is a general event or whether it is necessary for the WOCs to house a meridionally coordinating ligand framework, Sakai and co-workers focused on construct11892

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 33. Molecular structures of ruthenium complexes 135−138 based on the tetradentate ligand TPA.

relationship between initial rates and catalyst concentration, indicating that catalysis occurred at single ruthenium centers. To demonstrate the unique effect of the abnormal carbene ligand, the catalytic activity was compared to a ruthenium complex housing a normal N-heterocyclic carbene (NHC). The ruthenium complex containing the normal NHC ligand displayed low TONs and TOFs in combination with substantial CO2 production, which demonstrated the superior catalytic effect arising from the abnormal triazolylidene carbene.241 These results illustrate the unique properties of abnormal carbenes and their ability to give highly active ruthenium-based WOCs. The simplicity by which these catalytic systems can be tuned by modifications of the substituent at the triazolylidene nitrogens could become useful when constructing devices for overall H2O splitting. Because polypyridyl-based ruthenium complexes are frequently used as photosensitizers, their photophysical and photochemical properties have been extensively studied. Phenomena such as photoredox,53 photosubstitution,242,243 photoluminescence,244 and photoisomerization245 are processes that have been observed and involve the excited 3MLCT state. While [Ru(bpy)3]2+-type complexes can act as oxidants in H2O oxidation, but do not seem to catalyze the reaction, a number of related complexes have been found to be active catalysts. A notable example is complex 148. In an early study, this was found to undergo cis/trans photoisomerization to give the trans isomer 147 (Scheme 9).246

extensively as model systems of the active site of mono- and dinuclear nonheme metalloproteins for activation of molecular O2.236,237 All of the ruthenium complexes 135−137 (Figure 33) were shown to be capable of mediating H2O oxidation when employing CeIV as chemical oxidant. The catalytic activity for complex 137 was found to be far superior to that of complexes 135 and 136, highlighting the positive effect of the introduced bpy ligand. However, the investigated complexes also underwent oxidative decomposition resulting in simultaneous CO2 formation during H2O oxidation catalysis, which was invoked as an explanation for the low efficiency.235 A subsequent study involved the synthesis of analogous Ru complex 138, housing an isoquinoline-based ligand. However, the substitution of the pyridines with isoquinoline units resulted in reduced catalytic activity. This negative effect could stem from the higher redox potential for the RuV/RuIV redox couple. Also, noncovalent interactions between the isoquinoline groups could contribute to the decreased activity.238 Carbenes,239 and especially abnormal (triazolylidene) carbenes,240 are well-known for their pronounced σ-donor ability and their capability to form stable complexes with several different transition metals. Consequently, they can be used as ligands to stabilize metal centers in high oxidation states. Albrecht and co-workers utilized the strong electron-donating ability of abnormal carbenes to synthesize a series of single-site ruthenium complexes (139−146; Figure 34) based on

Scheme 9. Cis/Trans Photoisomerization of the [Ru(bpy)2(OH2)]2+ Complex

Figure 34. Structures of the single-site ruthenium complexes 139− 146.

This monomeric ruthenium complex houses two aqua ligands and has the potential to lose four protons and four electrons to produce a formal RuVI bis-oxo species that can participate in the four-electron oxidation of H2O.247 The cis/ trans photoisomerization of the [Ru(bpy)2(OH2)2]2+ complex and its impact on H2O oxidation catalysis has recently been revisited.248,249 It was revealed that the cis isomer 148 was more efficient than the corresponding trans isomer 147 in promoting H2O oxidation and that cis/trans isomerization did not occur to

pyridine-functionalized abnormal triazolylidene carbene ligands.241 Electrochemical analysis revealed a correlation between the redox potentials and the electron-donation ability of the wing-tip group, where electron-donating groups were shown to decrease the redox potentials of the ruthenium complexes. The ruthenium complexes 139−146 were subsequently evaluated as WOCs and found to be capable of catalyzing the oxidation of H2O. Kinetics revealed a linear 11893

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 10. Solvent H2O Nucleophilic Attack versus Intramolecular O−O Bond Formation Pathways for cis-[RuVi(Bpy)2(O)2]2+ Together with the Calculated Free Energies (kcal mol−1)

a significant extent in the absence of light. Experimental evidence and quantum chemical calculations support a mechanism in which the O−O bond is generated by solvent H2O nucleophilic attack on a RuVIO, instead of intramolecular coupling of the two oxo moieties in cis[RuVI(bpy)2(O)2]2+ (149, Scheme 10). Detailed insight into the photoisomerization process was also gained through UV− vis, electron paramagnetic resonance (EPR), X-ray absorption spectroscopy (XAS), together with quantum chemical calculations. This illustrated that decoordination of the aqua ligand occurs in the excited state of the cis isomer. This generates a coordinatively unsaturated metal complex that undergoes structural rearrangement followed by recoordination of an aqua ligand to yield the trans isomer. Another example where photoisomerization has been demonstrated to control the redox properties and H2O oxidation catalysis involves the two RuII aqua complexes 154 and 155, housing the pynap ligand (pynap = 2-(2-pyridyl)-1,8naphthyridine).250,251 Complexes d-[Ru(tpy)(pynap)(OH2)]2+ (154; d = distal) and p-[Ru(tpy)(pynap)(OH2)]2+ (155; p = proximal) only deviate in the orientation of the pynap ligand, and are geometrical isomers (Figure 35). It was shown that the isomer 154 was photoisomerized to generate isomer 155, and that, although these complexes are structurally related, they exhibit dramatically different catalytic properties. When comparing bpy-containing complexes with complexes comprised of the pynap ligand, it was found that the redox potentials for the RuIII/RuII couple were lower in the pynapcontaining complexes. The difference in the electrochemical properties and the pKa values of the two geometric isomers provided evidence for a favorable hydrogen-bonding interaction between the hydrogen on the aqua ligand and the

Figure 35. Molecular structures of the two isomers d-[Ru(tpy)(pynap)(OH2)]2+ (154) and p-[Ru(tpy)(pynap)(OH2)]2+ (155).

uncoordinated pynap nitrogen in 155. The Pourbaix diagrams of the isomers also provided detailed insight into the difference in the electrochemical properties of the two isomers, with 154 exhibiting electrochemical and pKa features comparable to those of the related [Ru(tpy)(bpy)(OH2)]2+ WOC (67, Figure 21). The catalytic activity of the two geometrical isomers was also examined, and isomer 155 turned out to be a poor WOC, while 154 displayed remarkable activity in H2O oxidation. Another surprising difference was the observation of a catalytic current for proton reduction to molecular H2 with complex 155, but not for 154.251 Insight into the mechanism of photoisomerization has also been obtained through the use of transient absorption spectroscopy and quantum chemical calculations.252 It was revealed that the rate of photoisomerization of 154 drastically decreased at pH > 7, and did not occur at all at pH > 11. This effect was attributed to the formation of the corresponding hydroxo complex (pKa = 9.7), which renders it inert toward photoisomerization. It was believed that upon excitation of the d-[Ru(tpy)(pynap)11894

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(OH2)]2+, the aqua ligand dissociates from the ruthenium center to generate a pentacoordinated structure, [Ru(tpy)(pynap)]2+*, which subsequently gets converted to the proximal structure. In contrast to the excited state of the other isomer, p-[Ru(tpy)(pynap)(OH2)]2+, the aqua ligand stays coordinated to the metal center, probably as a result of the stabilizing hydrogen-bonding interactions that exist in the p[Ru(tpy)(pynap)(OH2)]2+ isomer. Back-isomerization to the distal isomer was therefore predicted to be unfavorable. However, for both hydroxo isomers, the coordination of the hydroxo ligand is sustained and no photoisomerization occurs.252 A structurally related monomeric ruthenium complex has also been reported to undergo photoisomerization, where the two isomers displayed different H2O oxidation activity.253 These findings clearly demonstrate the catalytic diversity displayed by complexes 154 and 155 and the impact that photoisomerization can exert on H2O oxidation catalysis. Finally, Milstein and co-workers reported on the intriguing ruthenium pincer complex 156 with a dearomatized ligand scaffold, which in the presence of H 2 O can liberate stoichiometric amounts of H2 and O2 via a stepwise process involving consecutive thermal- and light-driven reactions (Figure 36). In the first step, complex 156 reacts rapidly with

Finally, exposure of 158 to light for 2 days caused O2 evolution with concomitant formation of the Ru0 species 159, from which regeneration of 156 subsequently occurred via ligand dearomatization. The liberation of O2 was suggested to originate from a two-step process involving a reductive elimination from complex 158 to release H2O2, which is then disproportionated to O2 and H2O. The alternative intermolecular and binuclear mechanism for O−O bond formation was ruled out on the basis of mechanistic experiments involving isotopically mixed-labeled dihydroxy complex 158-18O16O (Table 4).254,255 2.2.2. Mononuclear Ruthenium Catalysts Containing Anionic Ligand Backbones. As previously noted, the major drawback of WOCs containing neutral ligands is their incompatibility with photosensitizer-based oxidants, which restricts their use in light-driven H2O oxidation systems. As a consequence of this incompatibility, CeIV is instead commonly used as a sacrificial oxidant to chemically drive H2O oxidation. Unfortunately, this approach is not viable for practical use in future solar fuel cells, which should ultimately be driven by sunlight. To enable light-driven systems, it is important to match the oxidation potential of the WOC with that of a chromophore, a photosensitizer, so that they become compatible. This can be realized by lowering the oxidation potential of the WOC to below that of the photosensitizer, thus making it thermodynamically favorable to drive oxidation of H2O (Figure 37). One way to accomplish this would be to introduce a ligand with lower or no π-accepting capability and more pronounced σ-donating capability.234 Another strategy, which is more efficient, would be to introduce ligands with negatively charged functional groups, and in this way increase the electron density of the ruthenium center, resulting in a lower oxidation potential of the corresponding metal complex. The positive results obtained when applying this strategy for dinuclear ruthenium complexes193−195 motivated the groups of Sun and Åkermark to develop a number of single-site systems containing negatively charged groups. The group of Sun thus synthesized the mononuclear ruthenium complex [Ru(bda)(pic)2] (160, Figure 38; H2bda = 2,2′-bipyridine-6,6′-dicarboxylic acid), containing a tetradentate version of the bpy-type ligand with two extra coordinating and negatively charged carboxylate functionalities.256 Because of the strong electron-donating ability of the carboxylate moieties in the ligand, complex 160 showed dramatically lower redox potentials than ruthenium WOCs comprised of neutral ligand backbones. By the use of a Clark-type oxygen electrode, it could be demonstrated that ruthenium complex 160 was able to generate O2. Interestingly, the process of O2 evolution was second order with respect to catalyst, suggesting the involvement of a dinuclear catalytic species (vide infra). This was further supported by NMR and mass spectrometry (MS) investigations, where analysis of the NMR spectrum showed two different patterns of proton resonances, and where two major products were observed by high-resolution mass spectrometry (HRMS) measurements. One pattern was ascribed to the RuIII species [Ru(bda)(pic)2]+, while the other belonged to another species that had a mass value corresponding to a seven-coordinated RuIV species, [Ru(bda)(pic)2(OH)]+.256 On the basis of these results and the large bite angle observed for complex 160 (122.99°), the authors suggested that one of

Figure 36. Proposed mechanism for the formation of H2 and O2 from H2O by complex 156.

H2O at room temperature, leading to ligand aromatization with quantitative formation of the trans hydrido−hydroxo complex 157. The authors proposed that formation of complex 157 occurred by a mechanism involving coordination of H2O at the vacant coordination site situated trans to the hydride ligand, followed by proton migration to the side arm. This H2O activation process is unique in that it involves the cooperation between the ligand framework and the metal with no change in the oxidation state of the metal center.254 Complex 157 proved to be sufficiently stable to allow isolation, which enabled full characterization. The crystal structure revealed that the complex had a distorted octahedral coordination geometry at the ruthenium center. Furthermore, the nonlinear Ru−O−H bond angle indicated repulsion between the oxygen lone pairs and the metal d electrons. Evolution of H2 could be triggered from 157 by refluxing it in H2O for 3 days, regenerating the cis-dihydroxo complex 158. 11895

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. Summary and References to Available Data for Mononuclear Ruthenium Water Oxidation Catalysts Containing Neutral Ligand Scaffolds

11896

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11897

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11898

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11899

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11900

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11901

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11902

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11903

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11904

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11905

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11906

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

11907

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued

a

Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole of catalyst per s−1. cUsing Ce(NH4)2(NO3)6 (CAN, CeIV) as the chemical oxidant. dPhotochemical oxidation 11908

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 4. continued using [Ru(bpy)3]Cl2 as photosensitizer and K2S2O8 as the sacrificial electron acceptor. ePhotochemical oxidation using [Ru(deeb)2(bpy)]2+ as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. fA slightly higher TON value of 2.4 has been reported but with a lower TOF value of 0.00048 (see ref 252). gThe complex splits H2O into H2 and O2 via consecutive thermal and photochemical treatment. bpy = 2,2′-bipyridine, deeb = diethyl 2,2′-bipyridine-4,4′-dicarboxylate.

Figure 37. Schematic representation of a coupled chromophore−WOC assembly attached to a semiconductor, for overall H2O splitting (top) and the related energies for these processes (lower).

ligands interact with ruthenium WOCs at the higher oxidation states. The relatively low redox potentials of ruthenium complex 160 encouraged the authors to attempt to isolate such an exotic high-valent ruthenium species. Interestingly, they were able to isolate and obtain the crystal structure of an uncommon seven-coordinated dimeric RuIV complex (161) comprised of two [Ru(bda)(pic)2(OH)]+ units with a central bridging ([HOHOH]−) group (Figure 39).256 The crystal structure revealed a seven-coordinated ruthenium center, which resides in a highly disordered pentagonal bipyramidal configuration, where the angles in the pentagonal plane were all close to the ideal value of 72°. Furthermore, it was concluded that the seven-coordinated RuIV complex 161 was capable of promoting H2O oxidation, indicating that this species is indeed an intermediate in the catalytic cycle. This work demonstrated that ligand exchange on the octahedral ruthenium center was not a prerequisite for creating an open coordination site for an aqua ligand. Also, the resulting sevencoordinate RuIV−OH intermediate did not bring about any significant structural changes and was therefore found to be

Figure 38. Representation of the [Ru(bda)(pic)2] ruthenium complex 160.

the intermediates in the catalytic cycle was a seven-coordinated ruthenium species,256 an intermediate earlier considered both by Meyer and Thummel.200,202 However, at the time no decisive experimental evidence had been provided that supported the existence of such a seven-coordinated intermediate, which restricted the knowledge of how aqua 11909

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 39. Crystal structure of the dimeric ruthenium(IV) complex 161. Adapted with permission from ref 256. Copyright 2009 American Chemical Society.

associated with a low reorganization energy pathway for the oxidation of H2O.256 The low oxidation potential of ruthenium complex 160 made it compatible with commonly used photosensitizers, and thereby suitable for use in a light-driven H2O oxidation system.257,258 When a three-component mixture, consisting of complex 160, a [Ru(bpy)3]2+-type photosensitizer, and [Co(NH3)5Cl]Cl2 or sodium persulfate (Na2S2O8) as sacrificial electron acceptors, was irradiated with visible light, O2 evolution was immediately triggered. However, it was found that both catalytic systems were pH sensitive and suffered from fast deactivation as a result of the rapidly decreasing pH of the solution. Following a similar approach, the group of Sun prepared two supramolecular assemblies 162 and 163 consisting of [Ru(bda)(pic)2] as the WOC unit with two photosensitizers incorporated into the axial picoline ligands (Figure 40). The success of this coupled system for light absorption and subsequent H2O oxidation was made possible by the strongly electron-donating bda ligand, which lowered the onset potential of the H2O oxidation process to approximately 1.00 V vs NHE (at pH 7), making it compatible with many ruthenium-based chromophores (e.g., E1/2([Ru(bpy)3]3+/[Ru(bpy)3]2+) = 1.26 V vs NHE). Photocatalytic experiments employing assemblies 162 and 163 were conducted in degassed phosphate buffer solutions, using sodium persulftate as the sacrificial electron acceptor. Upon irradiation with visible light, assembly 162 was observed to rapidly evolve O2 with a TOF of 4.7 min−1 as determined by Clark electrode measurements, while assembly 163 proved to be inactive.259 The significantly shorter excitedstate lifetime of the [Ru(tpy)2]2+-units in 163 in comparison to that of the [Ru(bpy)3]2+-units in 162 was invoked as an explanation by the authors for the observed inactivity, because it translates into poor electron transfer between the WOC and the photoexcited photosensitizer.58 Control experiments affirmed that all three components of the reaction, light, assembly, and electron acceptor, were required for triggering O2 evolution. Comparison of assembly 162 to the corresponding separate component system consisting of [Ru(bda)(pic)2], [Ru(bpy)3]2+, and sodium persulfate revealed an almost 5-fold difference in activity,

Figure 40. Representations of the molecular assemblies 162 and 163.

TON 38 versus 8, in favor of the coupled system. The authors also investigated the degradation pathway for assembly 162 by the use of MS, which showed a correlation between the slow ligand dissociation of one of the two axial photosensitizer units of the assembly and the decrease in H2O oxidation rate over time. Although this decomposition process imposed limitations on the overall efficiency of this simple system, it still constituted an important proof-of-principle for realizing photocatalytic O2 evolution, which could help in facilitating future research on visible-light driven dye-sensitized PECs comprised of molecular catalysts.259 A series of [Ru(bda)(pic)2]+-type catalysts have also been prepared, in which the axial monodentate picoline ligand was substituted with a variety of electron-donating and electronwithdrawing groups (Figure 41).260 Electrochemistry revealed three redox waves, corresponding to RuIII−OH2/RuII−OH2, RuIV−OH/RuIII−OH2, and RuVO/RuIV−OH, at pH 1. A catalytic current started after the RuVO/RuIV−OH redox wave, assuring that H2O oxidation was triggered by the generation of RuVO. The redox potentials for the RuIII− OH2/RuII−OH2 redox couple were strongly dependent on the nature of the axial ligand in the studied ruthenium complexes 164−169, spanning from 0.45 to 0.75 V, where electrondonating groups expectedly led to a decrease in the redox potential. However, in the higher oxidation states RuIV and RuV, this difference was dramatically equated, and the electronic effect was less pronounced. Electron-withdrawing substituents were shown to enhance the catalytic activity of the [Ru(bda)(pic)2]+-type complexes when examined in CeIV-driven oxidation of H2O. 11910

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 11. In Situ Transformation of [Ru(κ3bda)(DMSO)(L)2]-Type Ruthenium Complexes 170−173 to [Ru(κ4-bda)(DMSO)(L)]-Type Ruthenium Complexes 176−179 in Acidic Aqueous Solutions

Figure 41. Structures of the [Ru(bda)(pic)2]+-type catalysts 164−169.

A subsequent report concerned the development of [Ru(bda)(L)2] complexes in which the axial ligands were not based solely on nitrogen donor atoms.261 Here, DMSO and imidazole were employed as axial ligands to generate a vast variety of complexes with nonequivalent and equivalent axial ligands, respectively. This strategy resulted in ruthenium complexes 170−175 (Figure 42), in which the octahedral RuII centers had

DMSO. However, catalytic CeIV-driven H2O oxidation experiments indicated that complexes 176−179 with the axially coordinated DMSO ligand were more efficient in catalyzing H2O oxidation than complexes 174 and 175. Quantum chemical calculations made it possible to correlate the structure and coordination mode of the axial ligands with accessibility of the terminal oxygen atom, which relates to the energetics of the radical coupling pathway; that is, the structure and electronics of the axial ligands can either facilitate the I2M pathway or block it by means of steric hindrance (vide infra). The axial ligands in 176−179 allow unhindered coupling between the terminal oxygen atoms, which explains their higher efficiency.261 Previous studies implied that the preferred pathway for O−O bond formation for complex 160 involved radical coupling of two seven-coordinated RuIV−O• units, [RuIV−O•···O•−RuIV]. Release of O2 then occurs from a RuIV−OO−RuIV intermediate after subsequent oxidation (Scheme 12, left).262 Further studies explored the exchange of the flexible bda ligand in complex 160 to the rigid pda ligand of the structurally related [Ru(pda)(pic)2] complex 180 (Figure 43; H2pda = 1,10-phenanthroline2,9-dicarboxylic acid). It was noticed that the kinetics of CeIVdriven catalytic H2O oxidation changed from a binuclear process for complex 160 to a mononuclear process (Scheme 12, right) for complex 180.263 This result revealed that a small alteration in the ligand structure can have a major impact on the catalytic pathway. DFT calculations showed that the O− Ru−O bite angle of complex 180 is slightly increased as compared to complex 160, indicating that a hydroxide ligand could bind to the RuIV center, making complex 180 capable of forming a seven-coordinated RuIV intermediate. The calculations also indicated that the WNA pathway was favored over the I2M pathway for complex 180.263 This mechanistic difference has also been observed when complexes 160 and 180 were confined in the nanocage of SBA-16.264 Furthermore, from DFT calculations it could be concluded that the carboxylate groups could contribute to a proton-coupled nucleophilic attack on the RuVO species by preorienting the reactant H2O molecule and perhaps even promoting cleavage of the O−H bond. The nucleophilic attack results in a RuIII− OOH species that is oxidized via PCET to yield the formal peroxo complex RuIV−OO, from which O2 release finally occurs.263

Figure 42. Molecular structures of the [Ru(κ3-bda)(DMSO)(L)2]and [Ru(κ4-bda)(L)2]-type complexes, 170−173 and 174,175, respectively.

different coordination environment. It was anticipated to yield ruthenium complexes of the type [Ru(κ4-bda)(L)2] with C2v symmetry; however, instead complexes 170−173 were isolated as the [Ru(κ3-bda)(DMSO)(L)2] derivatives. It was revealed that the [Ru(κ3-bda)(DMSO)(L)2]-type complexes 170−173 readily lost the equatorial imidazole ligand in solution, giving complexes 176−179 that all have a seventh accessible coordination site (Scheme 11). This ligand exchange process was shown to be rapid in acidic aqueous solutions, which clearly resemble the conditions of CeIV-driven H2O oxidation.261 Electrochemical measurements showed that large catalytic currents started after the RuV/RuIV redox couple, implying that RuV promotes H2O oxidation similarly to the previously synthesized Ru−bda catalysts. The redox potentials of the bisimidazole complexes 174 and 175 were found to be lower than those of the mixed imidazole−DMSO complexes 176−179, presumably because imidazole is a better electron donor than 11911

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 12. Schematic Representation of the I2M and WNA Pathways and the Impact That the bda and pda Ligand Backbones in [Ru(bda)(pic)2] (160) and [Ru(pda)(pic)2] (180) Have on These Catalytic Pathways

Figure 43. Molecular structure of the [Ru(pda)(pic)2] complex 180. Figure 44. Structure of the [Ru(bda)(isoq)2] complex 187 (left) and the noncovalent interactions between the isoquinolines, which promotes dimerization and radical coupling of the two RuVO units to generate the O−O bond (right).

A similar approach focused on replacing the axial picoline ligand in the [Ru(bda)(pic)2] (160) complex with two isoquinoline units, to generate the structural analogous [Ru(bda)(isoq)2] complex 187 (Figure 44, left; isoq = isoquinoline).265 This modification had a major impact on the catalytic rate and resulted in an impressive TOF of ∼300 s−1, which is comparable to that of the natural photosystem. This extremely fast catalyst was studied by a variety of different techniques, including spectroscopy, electrochemistry, and kinetics. Quantum chemical modeling was used to get a mechanistic understanding of the underlying reasons for the high catalytic activity. Electrochemical studies of catalyst 187 showed three peaks, which were ascribed to the formal oxidations RuII → RuIII → RuIV. The Pourbaix diagram revealed that these processes corresponded to [RuIII−OH2]/ [RuII−OH2], [RuIV−OH]/[RuIII−OH2], and [RuVO]/ [RuIV−OH], respectively, where generation of the RuV state triggers O2 evolution. Kinetic measurements of the initial reaction rate demonstrated that it was second order in relation to the concentration of catalyst 187, similar to what was

observed for the related [Ru(bda)(pic)2] catalyst 160. This finding implied that H2O oxidation occurred through a common reaction mechanism for the two structurally related catalysts 160 and 187. Spectroscopy and quantum chemistry provided evidence for a mechanism that involved formation of a formal high-valent RuVO species. This species undergoes a fast radical coupling to yield the peroxo intermediate RuIV− OO−RuIV, which is further oxidized to the formal Ru2IV,IV superoxo species [RuIV−O•−O−RuIV] from which facile liberation of O2 occurs. The dramatic rate enhancement for the [Ru(bda)(isoq)2] complex 187 was proposed to be due to noncovalent interactions between the isoquinolines, which lower the barrier for radical coupling of the two RuVO units and thus facilitate dimerization as well as O−O bond formation (Figure 44, right).265 11912

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 45. Schematic illustration of the molecular device where complex 160 and [Ru(bpy)2(4,4′-(PO3H2)2bpy)]2+ were coimmobilized onto a TiO2 sintered FTO electrode. Reprinted with permission from ref 269. Copyright 2013 American Chemical Society.

photosensitizer, which generated negligible photocurrents that rapidly decayed. Furthermore, the authors could demonstrate a correlation between the pH at which the Nafion membranes had been prepared and the photocurrent decay of the corresponding PEC. It was found that the system constructed from the most alkaline Nafion polymer solution (pH 9.8) displayed the most stable photocurrent. This pH dependence suggests the involvement of a rapid proton release in the H2O oxidation mechanism, which has a profound influence on the catalytic properties of complex 188. Moreover, it could be demonstrated by the use of GC that the device was capable of producing H2 simultaneously with O2, when a small bias of −0.13 V vs NHE was applied.266 Subsequent work by the group of Sun focused on designing general methods for supporting WOCs on semiconductive or conductive surfaces of PEC devices, with the aim of enabling the construction of more efficient and robust coupled systems for H2O splitting. A promising approach was to anchor the pyrene derivative of complex [Ru(bda)(pic)2] 160 onto a MWCNT-coated ITO glass electrode via strong noncovalent π−π stacking interactions.267 This assembly was evaluated as a working electrode in electrolysis experiments, carried out in neutral aqueous solutions. Subjecting the functionalized electrode to a constant applied potential of 1.40 V vs NHE resulted in the immediate evolution of both O2 and H2 from the working electrode and the counter electrode, respectively. Impressively, the long-term electrolysis over 10 h yielded a TON of 11.000 and a TOF of 0.3 s−1 for the O2 evolution, demonstrating both the efficiency and the robustness of the system. Later, the group of Sun explored the copper(I) catalyzed azide−alkyne cycloaddition (so-called CuAAc or “click reaction”) as a universal method of covalently attaching WOCs to conductive surfaces. Although a complete PEC device was not prepared in this study, the authors could show that this methodology constituted a viable option for attaching WOCs on a conducting carbon surface without affecting their

Although the introduction of the isoquinoline ligands in catalyst 187 caused a dramatic enhancement in the catalytic rate, negligible amounts of CO2 were observed, indicating that oxidative decomposition was not the major decomposition pathway. Analysis of the postcatalytic reaction mixtures validated that the isoquinoline ligand had dissociated from the [Ru(bda)(isoq)2] complex. On the basis of this observation, the authors therefore concluded that dissociation of the axial nitrogen-based ligands was the major deactivation pathway for the Ru−bda catalyst.265 Sun and co-workers have created several functional and lightdriven PECs based on ruthenium WOCs, capable of producing H2 with low external potential bias. In 2010, the group reported on the use of a light-driven PEC utilizing complex 160 as the immobilized WOC unit with a dye-sensitized TiO2 nanomaterial as the anode and platinum metal as the cathode.266 As the photosensitizer, [Ru(bpy)2(4,4′-(PO3H2)2bpy)]2+ was chosen because it displayed several attractive properties, such as (i) suitable oxidation potential, (ii) efficient charge separation, (iii) absorption in the visible region, and (iv) that it can be strongly attached to TiO2 surfaces. Immobilization of complex 160 was carried out by first transforming it to the corresponding cationic species 188 (160+) by one-electron oxidation with CeIV and then adsorbing it onto a Nafion membrane. Although, the Nafion polymer possesses desirable electrical conducting properties, in combination with high chemical and thermal stability, it has the drawback of being comprised of sulfonic groups that make it acidic and increase the onset potential for H2O oxidation. However, the authors managed to circumvent this issue by adjusting the pH of the Nafion polymer solution before it was used for the construction of the Nafion membranes in the PECs. Gratifyingly, the adjusted TiO2−RuP/Nafion−188 system was found to give rise to a significant photocurrent upon illumination, confirming that efficient electron transfer occurred from complex 188 to the photo-oxidized photosensitizer. This was in sharp contrast to the corresponding systems lacking either complex 188 or the 11913

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

intrinsic activity.268 Following this study, Sun and co-workers also reported on the direct covalent coimmobilization of both complex 160 as the WOC and [Ru(bpy) 2 (4,4′(PO3H2)2bpy)]2+ as the photosensitizer onto a TiO2 sintered FTO electrode.269,270 Noteworthy in this respect was the silane-based linker with the flexible propyl chain that was used for the immobilization of the WOC, which was chosen to provide increased mobility. The added mobility was believed to promote the catalysis in two ways: (i) by improving the efficiency of the electron transfer between the photosensitizer and the WOC, and (ii) bringing two WOCs together into close proximity to facilitate the binuclear radical coupling of the O− O bond forming step. A three-electrode PEC for the visible-light (λ > 400 nm) driven H2O splitting was constructed with the functionalized photoanode as working electrode (Figure 45). For the lightdriven electrolysis experiments, an external bias of 0.20 V vs NHE was applied, resulting in very high and pH-dependent photocurrents. Even though the best result was obtained when the electrolysis was carried out in phosphate buffer (pH 8.0), the authors chose to conduct further experiments under neutral conditions as it constituted a better representation of the natural processes occurring in PS II. Gratifyingly, lowering of the pH to 6.8 still afforded a high photocurrent density when the system was subjected to an external bias of 0.20 V vs NHE over 100 s of illumination. In contrast, the corresponding functionalized photoanode lacking either the WOC or photosensitizer unit only yielded negligible photocurrents, emphasizing the necessity of both components for the function of the device. Quantitative analysis over ca. 500 s of illumination revealed the generation of 0.75 μmol of O2 and 1.34 μmol of H2, which corresponded to a TON of 498 and an average TOF of 1.0 s−1. The Faradaic efficiencies for the O2 and H 2 generation were calculated to be 83% and 74%, respectively.269 However, for a WOC to be successfully implemented in a commercial device for large-scale applications, it has to be extremely robust and efficient. Under the highly oxidizing conditions required to mediate the four-electron oxidation of H2O, the developed artificial WOCs decompose and/or are deactivated after a certain time period, which constitutes a severe and general issue for WOCs. It is therefore surprising that little attention has been dedicated to identifying and obtaining insight into the decomposition and deactivation pathways. The current information regarding these detrimental processes of artificial WOCs is quite dispersed, where oxidative decomposition206 and ligand dissociation265 have been pointed out as the main deactivation pathways. On the basis of the observations on deactivation processes for the [Ru(bda)(isoq)2] complex 187,265 the group of Sun envisioned that the longevity of Ru−bda type catalysts could be enhanced by increasing the binding affinity of the axial ligands. DFT calculations were therefore performed in attempts to establish a correlation between the bond strength of the axial ligand and the robustness of the WOCs.271 A series of Ru−bda catalysts (189−191, Figure 46) were synthesized where inspection of the energies of the axial ligands made it possible to experimentally correlate this to the stability of the investigated WOCs. Because of the bimolecular reaction of the [Ru(bda)(pic)2] catalyst 160, the performance might be improved by forcing the key RuVO intermediate into facile intramolecular radical coupling. It was therefore proposed that directly linking two

Figure 46. Structures of the Ru−bda catalysts 189−191.

monomeric [Ru(bda)(pic)2] units through the use of flexible bidentate axial ligand frameworks could enhance the catalytic activity of the [Ru(bda)(pic)2] system. This hypothesis was shown to be valid by synthesis of the dimeric ruthenium complexes 192−194 (Figure 47), which showed enhanced catalytic activity as compared to their monomeric precursors.272 Sun and co-workers have also reported on the synthesis of the two RuII WOCs, [Ru(pdc)(pic)3] (195) and [Ru(pdc)(bpy)(pic)] (196) (Figure 48), containing the negatively charged tridentate pdc ligand (H2pdc = 2,6-pyridinedicarboxylic acid).273 This ligand increased the electron density at the ruthenium center, thus lowering the oxidation potential of the WOC, to make it compatible with commonly used [Ru(bpy)3]2+-type photosensitizers. Interestingly, despite their high structural similarity, the two complexes showed significant difference in activity, both in chemical and in photochemical H2O oxidation. For instance, in chemically driven H2O oxidation using CeIV as oxidant under acidic conditions, complex 195 was found to be a rather effective catalyst with a TON of 553 and a rate of 0.23 s−1, which can be compared to complex 196 that only gave a TON of 17 and a rate of 0.0072 s−1. A linear dependence of the initial rate of the complexes was observed for complexes 195 and 196, indicating that the catalysts most likely follow a mechanism where WNA occurs on a high-valent ruthenium−oxo species, generated from H2O−picoline exchange in the precatalysts 195 and 196. The authors ascribed the difference in activity to the relative ease of this H2O−picoline ligand exchange, which was supported by mechanistic studies utilizing electrochemistry and 1 H NMR spectroscopy. Complex 195 proved to rapidly exchange the equatorially bound picoline with a solvent molecule in mixed H2O/acetonitrile solutions under acidic conditions, thus generating a [Ru(pdc)(pic)2(H2O)]+ species, which the authors claimed was responsible for the catalytic activity. In contrast, complex 196 with bpy occupying the equatorial position and only one axially bound picoline did not exhibit such a behavior, and consequently no free picoline could be observed. These observations are fundamental and highlight the catalytic mechanistic diversity of ruthenium catalysts structurally related to complexes 195 and 196.273 To explore the effects of different axial ligands, complexes 197−201 were synthesized (Figure 49).274 It was found that the electron-donating ability of the monodentate ligands increased in the order 4-methoxypyridine > pyridine > pyrazine. Electrochemical measurements verified that strong catalytic currents corresponding to H2O oxidation occurred for complexes 197−199, while complexes 200 and 201 only showed negligible currents. Plotting the catalytic current at 1.70 11914

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 47. Representation of the dimeric ruthenium complexes 192−194.

Figure 48. Representation of the two ruthenium complexes [Ru(pdc)(pic)3] (195) and [Ru(pdc)(bpy)(pic)] (196), housing the negatively charged tridentate pdc ligand.

Figure 50. Molecular structures of the two ruthenium complexes [Ru(bpc)(pic)3]+ (202) and [Ru(bpc)(bpy)(OH2)]+ (203).

Figure 49. Structurally related ruthenium complexes 197−201.

resemble the previously investigated [Ru(tpy)(bpy)(OH2)]2+type complexes (71−80). The properties of the complexes 202 and 203 were studied by a variety of different techniques, including stopped-flow measurements, mass spectrometry, electrochemistry, and UV−vis spectroscopy. This allowed the authors to demonstrate the advantages of introducing a strongly donating carboxylate functionality into the ligand backbone of the WOCs, which included: (i) drastic lowering of the redox potentials, granting better accessibility to the higher valent oxidation states of the ruthenium center, (ii) reduced overpotential for catalytic H2O oxidation, (iii) enhancement of the picoline−H2O ligand exchange, (iv) facilitation of O2 release from the ruthenium metal center, and (v) a possible interaction between the carboxylate functionality and a solvent H2O molecule, which preorients the reactant H2O molecule and facilitates nucleophilic attack on the [RuVO] intermediate to form the essential O−O bond. However, a potential disadvantage of these complexes is the retardation of proton transfer, which compromises release of protons from the complexes. Hence, the introduction of the negatively charged bpc ligand promotes electron-transfer events but not PCET processes, which is in contrast to the neutral tridentate tpy ligand. Recently, the group of Åkermark reported on the use of two anionic ligands containing imidazole, phenol, and carboxlyate moieties for generation of the two RuIII-based WOCs [Ru(hpbc)(pic)3] (204; H3hpbc = 2-(2-hydroxyphenyl)-1H-

V vs the redox potential of the RuIII/RuII couple for the investigated complexes afforded a linear correlation, which implied that the electrochemical activity was directly related to the electron-donating ability of the monodentate ligand. A similar trend of the catalytic activity was observed in the CeIVdriven oxidation of H2O for complexes 197−201 and was in agreement with the electrochemical measurements. In a subsequent paper, the two complexes [Ru(bpc)(pic)3]+ (202) and [Ru(bpc)(bpy)(OH2)]+ (203; Hbpc = 2,2′bipyridine-6-carboxylic acid), depicted in Figure 50, were synthesized.275 With the anionic bpc ligand these complexes 11915

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

higher turnovers could be obtained by replacing [Ru(bpy)3]2+ with the more strongly oxidizing photosensitizer [Ru(bpy)2(deeb)]2+ (1.40 V vs NHE).276 In this study,276 substantial efforts were also dedicated toward elucidating the mechanism of H2O oxidation, catalyzed by complexes 204 and 205. In accordance with earlier observations made by Sun and co-workers,273 a catalytically active aqua-species (206) is obtained from ligand exchange of a picoline ligand by a solvent H2O molecule. Evidence for such a ligand exchange was provided by UV−vis spectrophotometric titration experiments, from which it was also possible to determine the solvolysis equilibrium constants for complexes 204 and 205. The catalytically active ruthenium species (206) is presumed to be in equilibrium with 207, which subsequently undergoes two PCET events to ultimately furnish a RuVO species (210) that could be observed by HRMS and is believed to be the key intermediate during catalysis (Scheme 13). Evidence for the dependence of proton removal on each oneelectron redox step between the RuIII and RuV state was confirmed by electrochemical measurements. The Pourbaix diagram of complex 204 demonstrated that all three oxidation steps, from the RuII state to RuV, were pH-dependent with a slope of −59 mV per pH unit, indicating that PCET plays a fundamental role during the four-electron oxidation of H2O.276 The authors had previously managed to find support for a high-valent RuV−oxo species by the use of HRMS for the related dimeric ruthenium complex 48,277 and also in this study the methodology proved to be successful. Upon the addition of 10 equiv of the oxidant [Ru(bpy)3]3+ to a solution containing complex 205, it was possible to distinguish a signal by HRMS that corresponded to [RuVO + H+]+ in positive mode, and with a isotope pattern matching the theoretically expected.276 This suggests that appropriate design of the ligand scaffold offers molecular catalysts that have the ability of stabilizing the redox-active metal center(s) in high-valent states. Combining both redox and proton transfer mediator motifs into the catalytic units should facilitate the simultaneous transfer of electrons and protons, thus avoiding high-energy intermediates, and give access to new reaction pathways. Catalysts such as complexes 204 and 205 offer tremendous potential in terms of synthetic ease, cost, high catalytic activity, and the possibility of designing molecular assemblies where they can be coupled to cocatalysts or chromophores in future light-harvesting devices for the production of solar fuels.

benzo[d]imida-zole-7-carboxylic acid) and [Ru(hpb)(pic)3] (205; H3hpb = 2-(2-hydroxyphenyl)-1H-benzo[d]imidazol-7ol), depicted in Figures 51 and 52.276 The two single-site

Figure 51. Structures of ruthenium complexes 204 and 205 (L = 4picoline).

ruthenium complexes were easily synthesized from commercially available starting materials and were shown to have low redox potentials as a result of the strong electron-donating properties of the negatively charged tridentate ligand scaffolds. Furthermore, these ligand motifs allowed the incorporation of redox and proton transfer mediator motifs into the WOCs, which facilitated the simultaneous transfer of electrons and protons, via PCET. This approach was inspired by nature, where imidazole, carboxylate, and phenol are vital units in the OEC. These functional groups facilitate PCET events that are fundamental in many biological processes, because new reaction pathways are opened, which do not involve highenergy intermediates. Complexes 204 and 205 were successfully employed in catalytic H2O oxidation, using either pregenerated or photogenerated [Ru(bpy)3]3+-type complexes as oxidants. From the catalytic experiments, it could be concluded that complex 204 was more efficient than 205. In the chemical H2O oxidation experiments with complex 204, it was possible to reach a TON of up to 4000 and an initial turnover frequency of >7 s−1, which is the highest reported for a metal-based WOC when employing [Ru(bpy)3]3+ as oxidant. The two bioinspired ruthenium complexes were also evaluated in light-driven H2O oxidation, under neutral conditions (pH 7.2) using [Ru(bpy)3]2+-type photosensitizers and Na2S2O8 as sacrificial electron acceptor. Moderate catalytic activity was observed when [Ru(bpy)3]2+ was utilized as photosensitizer. This was ascribed to the small difference in redox potential between the WOCs and [Ru(bpy)3]3+, which results in a low thermodynamic driving force for the oxidation of H2O. Gratifyingly,

Figure 52. DFT-calculated structures of the two single-site ruthenium complexes [Ru(hpbc)(pic)3] 204 (left) and [Ru(hpb)(pic)3] 205 (right). Distances shown are in angstroms. 11916

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 13. Schematic Representation of the Possible Ligand Involvement during H2O Oxidation Catalysis Mediated by the Bioinspired Ligands in Ruthenium Complex 206 (L = 4-Picoline)

polypyridyl ligands.278 Furthermore, previous mechanistic studies based upon monomeric ruthenium-based WOCs have suggested that release of O2 from the metal center constitutes the rate-limiting step for the overall O2 evolution.202,207 On the basis of these results, it is reasonable to suggest that the more electron-donating phenolic ligand facilitates the whole process of H2O oxidation, by destabilizing the Ru−O bond in a fashion similar to how it destabilizes the ruthenium−picoline bond. Ligand architecture and catalyst optimization are of primary importance for developing WOCs with a longer lifetime. The amide bond is a bond of great importance in biological systems as it is found in peptides, enzymes, and a vast array of natural products.279 Transition metal complexes housing amines (−NR3) or amides (−NR2) as ligands are well explored, in contrast to complexes containing N-amidato (deprotonated amide) ligands. A key feature of N-amidato ligands is the strong σ-donor ability from the coordinated nitrogen(s), which stabilizes the metal center and allows access to high-valent oxidation states.280−283 The Åkermark group envisioned that a tailored catalyst design based upon the tetradentate ligand H2bpb (H2bpb = N,N′-1,2-phenylene-bis(2-pyridine-carboxamide)) would afford active single-site ruthenium WOCs.284 This amide ligand contains two sites with dissociable protons, and the ligand framework offers an easily tunable environment, a key factor for enabling the incorporation of WOCs into devices for overall H2O splitting. The tetradentate ligand backbone was readily synthesized from low cost and readily available starting materials, and was subsequently reacted with [Ru(DMSO)(Cl)2] to yield the ruthenium complex [Ru(bpb)(pic)2]+ (212, Figure 54.). Ruthenium complex 212 was first evaluated for H2O oxidation, using the mild one-electron oxidant [Ru(bpy)3]3+. Upon addition of an aqueous solution of the complex to the oxidant, evolution of O2 occurred, as confirmed by real-time mass

In a subsequent study, the group of Sun reported on the related RuII complex [Ru(hqc)(pic)3] (211, Figure 53; H2hqc =

Figure 53. Molecular structure of ruthenium complex [Ru(hqc)(pic)3] 211.

8-hydroxyquinoline-2-carboxylic acid), housing the tridentate, phenol-containing, hqc ligand.278 Electrochemical studies revealed that complex 211 displayed a significantly lower oxidation potential for the RuIII/RuII transition (0.23 V as compared to 0.50 V vs NHE in CH2Cl2) than that of the pdcbased ruthenium complex 195, which was explained by a better π electron-donating ability of the phenolate group as compared to that of the carboxylate group. As a result of the increased electron-donating properties of this ligand, the coordination between the ruthenium center and the equatorial picoline is further weakened, resulting in a faster picoline−H2O ligand exchange, and an overall more reactive complex. In this study, complex 211 was also successfully employed in chemical and visible-light driven H2O oxidation and appeared to be far superior to ruthenium WOCs containing neutral 11917

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 54. Molecular structures of ruthenium complexes [Ru(bpb)(pic)2]+ (212) and [Ru(bpb)(CO)(OH2)] (213).

spectrometry measurements. A kinetic study was performed, where the initial rate of O2 formation was demonstrated to be first-order in catalyst concentration, implying that bimolecular reactions are not contributing to the catalytic H2O oxidation. Detailed studies were also conducted to investigate the stability of complex 212 toward ligand dissociation and oxidative decomposition. To provide support that complex 212 is the active catalyst during H2O oxidation and that it retains its structural identity after catalytic cycling, a solution of the reaction mixture after about 20 turnovers was analyzed by HRMS, which verified that the structure of the catalyst was preserved.284 However, analysis of the reaction mixture after O2 evolution had ceased revealed a peak at m/z = 462.9985, in the negative mode. This corresponds to formation of the [Ru(bpb)(CO)(OH2) (213) − H+]− species. In addition to the evolution of O2, CO could also be detected as one of the gaseous byproducts (originating from decomposition of the catalyst itself or the employed oxidant [Ru(bpy)3]3+) during the catalytic oxidation experiments and could be a potential source for the formation of complex 213 from 212. A plausible intermediate in this conversion was also found at m/z = 539.0548, in positive mode, corresponding to a complex with the structure [Ru(bpb)(pic)(CO)]+.284 This suggested that the CO-containing ruthenium complex 213 was not catalytically active, and to verify this hypothesis it was independently synthesized. The crystal structure of complex 213 demonstrates that the ruthenium atom is in an octahedral configuration housing the tetradentate amide-based ligand in the equatorial plane, together with a H2O molecule and carbon monoxide in the axial positions. The angle of N1− Ru1−N4 is 115.08°, which is larger than the ideal 90° of a complex in an octahedral configuration and may provide the ruthenium complex with a seventh coordination site. All attempts to use the CO-containing complex 213 to catalyze H2O oxidation with [Ru(bpy)3]3+ as oxidant failed, with no detectable amounts of evolved O2.284 To explain the striking difference in the reactivity between complexes 212 and 213 and to get a more comprehensive insight into the catalytically active species, electrochemical measurements were performed on the two single-site ruthenium complexes. The Pourbaix diagram of complex 212 showed three peaks at pH 7, assigned to the formal redox couples RuIII−OH/RuII−OH2, RuIV−OH/RuIII−OH, and RuVIO/RuIV−OH, respectively (Figure 55). Interestingly, this suggests that a high-valent [RuVIO]2+ or [RuV−O•]2+

Figure 55. Pourbaix diagram of (a) ruthenium complex [Ru(bpb)(pic)2]+ (212) and (b) [Ru(bpb)(CO)(OH2)] (213).

species is generated and triggers oxidation of H2O. This is in contrast to a majority of the reported WOCs, which are oxidized to a formal RuVO that constitutes the catalytic key intermediate. For the CO-containing [Ru(bpb)(CO)(OH2) complex 213, the Pourbaix diagram only revealed a single peak at pH 7, which involved a three-electron redox process to directly transform [RuII−OH2] to [RuVO] (or [RuIV− O•]).284 The electrochemical difference between the two structurally related single-site ruthenium complexes 212 and 213 is intriguing and highlights the strong impact exerted by the axial ligands on the stability of the ruthenium center at different redox levels and their catalytic activity. A possible explanation for the lack of catalytic activity of ruthenium complex 213 could be that this particular system needs to reach the high-valent RuVI state to be able to act as a catalyst for H2O oxidation. The fact that the CO-containing complex 213 is generated from 212 under the employed catalytic conditions is an important feature because it indicates a novel mode of deactivation for ruthenium-based WOCs. Because it is difficult to predict which molecular structures that are expected to catalyze H2O oxidation at low overpotentials, it was a major breakthrough when it was realized that single-site ruthenium complexes could mediate the oxidation of H2O to molecular O2. This discovery has resulted in an increased research interest in molecular single-site metal complexes, not solely based upon the rare earth element ruthenium, but also iridium-based, as well as more earthabundant elements such as iron and cobalt (vide infra) (Table 5). Although there is a great number of reported artificial WOCs, the design of molecular structures that are capable of mediating the four-electron oxidation of H2O at low over11918

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 5. Summary and References to Available Data for Mononuclear Ruthenium Water Oxidation Catalysts Containing Negative Ligand Scaffolds

11919

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 5. continued

11920

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 5. continued

11921

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 5. continued

11922

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 5. continued

a Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole of catalyst per s−1. cUsing Ce(NH4)2(NO3)6 (CAN, CeIV) as the chemical oxidant. dPhotochemical oxidation using [Ru(bpy)3]Cl2 as photosensitizer and [Co(NH3)5Cl]Cl2 as the sacrificial electron acceptor. ePhotochemical oxidation using [Ru(bpy)3]Cl2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. fPhotochemical oxidation using [Ru(dmbpy)3](PF6)2 as photosensitizer and [Co(NH3)5Cl]Cl2 as the sacrificial electron acceptor. gPhotochemical oxidation using [Ru(deeb)2(bpy)]Cl2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. hPhotochemical oxidation using [Ru(deeb)3]Cl2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. i Using [Ru(bpy)3](PF6)3 as the chemical oxidant. jPhotochemical oxidation using [Ru(bpy)3](PF6)2 as photosensitizer and Na2S2O8 as the sacrificial

11923

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 5. continued electron acceptor. kPhotochemical oxidation using [Ru(bpy)2(deeb)](PF6)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. l Photochemical oxidation using [Ru(deeb)2(bpy)](PF6)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. bpy = 2,2′-bipyridine, deeb = diethyl 2,2′-bipyridine-4,4′-dicarboxylate, dmbpy = 4,4′-dimethyl-2,2′-bipyridine.

by an inert iridium species, [Ir(ppy)2(bpy)]+, lacking free coordination sites, resulted in no O2 generation.285 Electrochemical measurements confirmed that there existed a correlation between the IrIV/IrIII redox potentials and the electrochemical onset potentials. The oxidation waves of complexes 214−218 ranged from 1.20 to 1.74 V vs NHE. DFT calculations on the cyclometalated iridium complexes verified that the highest occupied molecular orbitals (HOMOs) displayed mixed metal−ligand character, that is, that the ligand is interacting with the iridium metal center through d−π interactions. This confirms the high tunability of this system, where the addition of electron-withdrawing and electrondonating substituents will have a direct impact on the electronic nature of the metal center. The finding that these relatively simple single-site iridium complexes displayed such a high activity and stability made them an appealing alternative to the existing ruthenium WOCs.285 A subsequent computational study was also performed on the iridium complex [Ir(ppy)2(H2O)2]+ 214 (Table 6). The aim was to investigate the mechanism of the nucleophilic attack of H2O on the generated iridum−oxo species to produce the crucial O−O bond.286 The authors wanted to rationalize the cooperation between Lewis acidity and Brønsted basicity in catalytic H2O oxidation, and for this different species were studied, [Ir(ppy)2(O)(H2O)]+, [Ir(ppy)2(O)(OH)], and [Ir(ppy)2(O)2]−, differing only in the protonation state of the aqua ligand. In the absence of a basic ancillary ligand, it was suggested that the proton was transferred from the attacking H2O molecule to the oxo-moiety in the iridium WOC. However, the calculations showed that the proton is preferably accepted by the ancillary ligand, thus suggesting that H2O oxidation is promoted by an internal base (Scheme 14). Deprotonation of the incoming H2O enhances its nucleophilicity and promotes O−O bond formation. The study demonstrated a correlation between a high basicity of the ancillary ligand and a low energy barrier for the O−O bond formation. In accordance, the [Ir(ppy)2(O)2]− species containing the less electrophilic IrV O unit but the most basic ancillary ligand was shown to have the lowest energy barrier for O−O bond formation.286

potentials continues to be challenging. Clearly, an understanding of the fundamental principles is the key to developing more efficient WOCs.

3. IRIDIUM-BASED SYSTEMS FOR WATER OXIDATION 3.1. Cyclometalated Iridium Complexes

A vital finding that had a major impact on the field of H2O oxidation catalysis was published in 2008 by Bernhard and coworkers.285 This essential paper highlighted that single-site cyclometalated iridium complexes 214−218 could mediate H2O oxidation (Figure 56). The relative simplicity by which

Figure 56. Molecular structures of the cyclometalated iridium complexes reported by Bernhard and co-workers.

these complexes could be designed and tuned made them highly attractive from a synthetic point of view. The cyclometalated phenylpyridine (ppy) was employed as a bidentate ligand, which takes advantage of the strong carbon−iridium bond in the final iridium complexes and afforded the robustness that is required for H2O oxidation. The iridium−aqua complexes were easily obtained by reacting the appropriate chloride-bridged IrIII dimer, [(ppy)2IrCl]2, with AgOTf. The catalytic activities of the synthesized iridium complexes 214−218 were evaluated by CeIV-driven H2O oxidation, and in all cases O2 evolution occurred. The cyclometalated iridium complexes displayed catalytic activity for several days, demonstrating the robustness of the catalytic system. In addition to the evolved O2, traces of CO2 were observed, indicating at least partial oxidation of the ligand framework. To ensure that the generated O2 originated from H2O and not from the nitrate anion of CAN, control experiments using Ce(OTf)4 as the chemical oxidant were conducted. These experiments confirmed that H2O was the single source of the produced O2. However, a reduction in the O2 production rate was observed, suggesting that the OTf− ion has a negative effect on the O2 evolution process. Addition of acetonitrile and dimethyl sulfoxide, which are well-known coordinating ligands, was also found to impede O2 generation. Replacing the catalysts

3.2. Iridium Catalysts Based on Cyclopentadiene-Type Ligands

Inspired by the work of Bernhard, the group of Crabtree and Brudvig developed a novel class of single-site iridium WOCs bearing a pentamethylcyclopentadiene (Cp*) ligand (219− 221, Figure 57). The idea behind these iridium WOCs was to introduce a more electron-donating ligand, the Cp* ligand, anticipating that this effect would result in more active catalysts. Production of O2 from these catalysts was shown to be 1 order of magnitude faster than that of the previously developed cyclometalated iridium catalysts 214−218. Cyclic voltammetry of the chloro containing complex 219 displayed three irreversible oxidation peaks, whereas the triflate complex 220 only showed two.287 The authors suggested that this difference, together with the higher rate observed for 219, originated from 11924

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 6. Summary and References to Available Data for Cyclometalated Iridium-Based Water Oxidation Catalysts

a

Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole catalyst per s−1. cUsing Ce(NH4)2(NO3)6 (CAN, CeIV) as the chemical oxidant. dIt should be noted that a higher TOF of 0.042 has been reported under similar reaction conditions but with a lower catalyst concentration.

because: (i) the catalyst could be fully recovered after the reaction, (ii) of the absence of lag time, (iii) no deposition could be observed, and (iv) the reaction proceeded with a firstorder rate with regard to catalyst concentration.287 Further studies by Brudvig and Crabtree yielded a variety of half-sandwich iridium complexes, containing either the Cp* or the cyclopentadiene (Cp) ligand (Figure 58).288 In complexes 222−230, the cyclometalated ligand was replaced by bipyridine, phenantroline, bipyrimidine, and tetramethylenediamine (tmeda), respectively. The activity in H2O oxidation was

chloride oxidation, which had previously been observed for WOCs.122 Complex 221, containing the pyrimidine ligand, displayed a lower catalytic activity than complex 219. This is most likely due to the protonation that occurs in acidic media and limits the donor ability of these ligands and reduces their ability to promote H2O oxidation. Isotopic labeling experiments verified that H2O was the source of the evolved O2. Moreover, it was proposed that the formation of heterogeneous IrO2 particles did not take place under the applied catalytic conditions 11925

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

bond formation, which was shown to proceed via a low-energy pathway involving intermolecular nucleophilic attack of a H2O molecule on the high-valent IrVO species. This attack was accompanied by proton transfer that generated an iridium− hydroperoxide species.288,289 Liberation of O2 then took place from this intermediate after subsequent oxidation, which resembles the mechanism suggested for single-site ruthenium complexes.109−115 It was proposed that the presence of benzylic carbons in the catalysts could result in modifications of the ligand framework during catalysis. Oxidative modification of the Cp* ligand has previously been shown to occur with iodine(III) reagents to yield acetoxylated or hydroxylated products.290 The Cp* ligand was therefore replaced by the Cp ligand, which should result in complexes lacking potentially oxidazable benzylic carbons. It was revealed that complex 226 housing the Cp ligand, an analogue of complex 222, gave a lower reaction rate. The two triphenylphosphine and carbon monoxide-containing iridium complexes 227 and 228 were also evaluated as potential WOCs because these coordinatively labile ligands were thought to undergo irreversible oxidation, thus creating an open site for coordination of H2O. These complexes exhibited excellent activity in H2O oxidation, in contrast to ruthenium complex 213. Kinetic studies at low catalyst loadings demonstrated a first-order dependence in iridium for catalysts 219, 222, and 229, implying that catalytic H2O oxidation occurred at a single iridium center. However, at high catalyst loadings, complex 229 was found to exhibit zero-order behavior due to a possible deactivation process.288 Considering the peculiar changes in rate in the initial phase of catalysis concerning the Cp*-containing catalysts, it was suggested that this was caused by oxidative decomposition and/ or deactivation of the catalysts, either by structural modifications of the bidentate ligand or of the CH3 groups in the Cp* ligand. Decomposition of only the bidentate ligands is assumed to result in strucurally related complexes and similar kinetics as the tris-aqua iridium complex 229. If this is accompanied by loss of the Cp* ligand, it seems reasonable to assume that catalytically active iridium nanoparticles can be generated, which could be responsible for the observed catalysis. Although TEM did not reveal any formation of

Scheme 14. Proposed O−O Bond Formation for the [Ir(ppy)2(O)(X)]n Species (Where X = H2O (n = +1), OH (n = 0), or O (n = −1))

Figure 57. Molecular structures of the Cp*- and Cp-containing iridium WOCs developed by Brudvig and Crabtree.

assessed with CeIV as chemical oxidant. Among the analogous complexes 222−230, complex 222 was found to be the most active. The tmeda containing complex 225 displayed a lower activity and appeared to lose the tmeda ligand due to protonation of the amino group under the highly acidic reaction conditions, resulting in the same catalyst precursor as complex 229. DFT calculations and electrochemical measurements suggested that the key intermediate during catalysis was an IrVO species. The fundamental step in the catalytic cycle is the O−O

Figure 58. Cp* and Cp iridium catalysts for H2O oxidation. 11926

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 7. Summary and References to Available Data for Cyclopentadiene-Type Iridium-Based Water Oxidation Catalysts

11927

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 7. continued

a

Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole catalyst per s−1. cUsing Ce(NH4)2(NO3)6 (CAN, CeIV) as the chemical oxidant. dA higher TON of 1240 has also been reported (see ref 324). eUsing NaIO4 as the chemical oxidant. fA higher TON of ∼4000 has been reported (see ref 324). gIt should be noted that a higher TOF value of 1.52 s−1 has been reported (see ref 118).

nanoparticulate species,288 further experiments are warranted to rule out their participation in the catalytic activity (vide infra). There is also an example where a Cp* iridium-based WOCs has been interfaced with a chromophore for photodriven H2O oxidation (Table 7). The iridium complex 231 and the fluorosubstituted zinc-based porphyrin chromophore 232 were codeposited on TiO2 nanoparticles (Figure 59).291

Illumination of untreated TiO2 anodes resulted in negligible photocurrents. However, anodes treated with codeposited iridium WOC 231 and zinc porphyrin 232 produced a high and sustained photocurrent, consistent with photodriven oxidation of H2O. Anodes that lacked any of the chromophore or iridium catalyst components only gave rise to unsustained photocurrents, demonstrating that the presence of both components is necessary for maintaining catalytic currents. 3.3. Development of Iridium N-Heterocyclic Carbene and Other Iridium Catalysts

Crabtree and Brudvig also devised a method to stabilize the iridium centers in the iridium WOCs in high oxidation states.292 This approach utilized NHCs, which are wellknown for binding tightly to metal centers and stabilizing them at high-valent states, and resulted in the preparation of complex 233, Cp*Ir(κ2,C2,C2-NHC)(Cl) (Figure 60). As anticipated, the NHC-containing iridum complex 233 was capable of oxidizing H2O, when driven by the chemical oxidant CeIV. A kinetic survey established that the reaction had an order close to 2, indicating that the reaction was bimolecular in iridium. The electrochemistry of complex 233 was therefore studied, and this revealed several oxidation peaks at less positive

Figure 59. Molecular structures of the iridium-based WOC 231 and the zinc-based porphyrin chromophore 232.

Figure 60. Molecular structure of the iridium−NHC complex Cp*Ir(κ2,C2,C2-NHC)(Cl) (233). 11928

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 61. Structures of the NHC iridium complexes 234−237.

absent, and (iii) no increase in catalytic current was seen upon successive scans.293 Dicarbene ligands have also been employed to stabilize the iridium center as exemplified by the two iridium WOCs 236 and 237. Unfortunately, no beneficial effects from these ligand architectures could be observed in the catalytic H2O oxidation experiments, and complexes 236 and 237 were found to exhibit activities comparable to that of previously reported iridium WOCs.294 An interesting approach was taken by Bernhard and Albrecht who made use of abnormal (mesoionic) carbene ligands.295 The pyridinium-functionalized salt was employed as the ligand in the complexation with [Cp*IrCl2]2. This induced a double C−H bond activation to afford the two structurally related iridium complexes 238 and 239, both accommodating the abnormal carbene ligand but differing in the coordination mode (Figure 62). The C−H activation was further confirmed by the

potentials than the cyclometalated iridium WOC 219. This discovery suggested that the introduced NHC ligand could indeed stabilize the iridium metal center at high oxidation states. Additional support for this was also obtained from computational analysis, which verified a lower IrIV/IrIII redox potential for complex 233, than for 219. This finding suggested that the oxidized species is relatively stable and that there is a probability of observing and characterizing this intermediate. The ruthenium(III) oxidant [Ru(bpy)3]3+ with its redox potential of 1.26 V vs NHE should have sufficient oxidizing power to generate the quasi-reversible electrochemical wave that occurs at 0.90 V, and that corresponds to the formation of the IrIV species. At the same time, while [Ru(bpy)3]3+ provides the IrIV species, [Ru(bpy)3]3+ does not possess the oxidizing power to generate higher valent species. Indeed, with [Ru(bpy)3]3+ it was possible to generate the one-electron oxidized species of the Ir−NH C comp lex 2 33, [Cp*IrIV(κ2,C2,C2-NHC)(Cl)]+, where low temperature EPR measurement showed that it was a low-spin d5 complex with S = 1/2.292 These results imply that the NHCs ligands should indeed stabilize high-valent oxidation states and allow preparation of more robust iridium WOCs than those previously reported. Instead of using a C,C-chelating carbene, Hetterscheid and Reek employed N-dimethylimidazolin-2-ylidene (a “normal carbene”) as a ligand to stabilize the iridium center in the WOCs.293 By starting from [Cp*IrCl2]2, the dichloro NHC iridium complex 234 was obtained, which upon treatment with excess Ag2O was transformed into the dihydroxo NHC iridium complex 235 (Figure 61). Both complexes catalyzed the oxidation of H2O, where complex 234 displayed a significant incubation time, indicating that chloride displacement is essential for the catalytic activity. Kinetics of the catalytic process afforded support for a mechanism where catalysis occurs at a single iridium center. Complex 235 showed impressive TONs (>2000), and no significant decrease in rate could be observed after the reaction was halfway from completion (1 h). The dihydroxo complex 235 also turned out to be an efficient catalyst for the disproportionation of hydrogen peroxide. Both the disproportionation of H2O2 and H2O oxidation are thought to proceed via similar catalytic intermediates, which implicates that H2O oxidation could in principle occur via initial peroxide formation that is subsequently disproportionated to liberate O2. Electrochemical analysis of complex 235 (in an acetonitrile solution) indicated an irreversible oxidation wave at 1.43 V vs NHE. In aqueous solutions, a distinct catalytic current corresponding to catalytic H2O oxidation could be observed. It was proposed that the catalysis was of homogeneous nature because: (i) no deposition could be observed on the electrodes, (ii) the characteristic features exhibited by IrO2 deposited on a carbon electrode were

Figure 62. Representations of the two iridium complexes 238 and 239 housing an abnormal carbene ligand.

crystal structures of the two complexes, which showed the anticipated piano-stool geometry, together with a five- or sixmembered metallacycle, respectively. The water-soluble, abnormal carbene complexes were evaluated as WOCs using CeIV as chemical oxidant, where addition of the complexes to the oxidant triggered immediate O2 evolution. It could be concluded that catalysts 238 and 239 were significantly more active than the previously reported catalyst [Ir(ppy)2(H2O)2](OTf) (214). Impressively, TONs of ∼10 000 were reached with the iridium complexes 238 and 239, which displayed a first-order dependence on catalyst concentration. The authors speculated that these remarkable numbers originated from the adjustable electronic nature of the abnormal carbene ligand(s). The zwitterionic character of the ligand may enhance the access to high-valent iridium(V)−oxo species, which are believed to be key intermediates in the catalytic cycle of H2O oxidation. Electrochemical analysis of the complexes signaled that the ligand may be involved during the catalysis. In the range 0.95 < E < 1.25 V vs SCE, multiple oxidation waves were observed, which could not be assigned 11929

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

the dinucleating ligand bpi (Figure 64). From the crystal structure, it was, surprisingly, revealed that the imine

solely to oxidations of the iridium center in complexes 238 and 239.295 This finding suggests that cooperative catalysis296 occurred, where both the iridium metal center and the abnormal carbene(s) play an active role in the catalytic process. In later studies, the use of triazolylidene compounds as ligands in iridium-based WOCs was demonstrated.297 The triazolylidene ligand was easily obtained from the corresponding triazolium salt by the use of Ag2O to abstract the proton. In situ treatment with [Cp*Ir(Cl)2]2 yielded the corresponding iridium complex 240 (Figure 63), which is air-stable both in solution and in the solid state.

Figure 64. Structure of the dinuclear [(cod)(Cl)Ir(μ-bpi)Ir(cod)]+ complex 242.

functionality is bridging through π-coordination. The first iridium center (Ir2) is coordinated to the imine nitrogen via σcoordination, which most likely activates the imine moiety toward η2 coordination to the other iridium center (Ir1). The imine bond in complex 242 is dramatically elongated as compared to C−N bonds in imines that are only coordinated via σ-coordination, thus demonstrating the unique interactions present in this complex. Treating the complex with CeIV resulted in O2 evolution, where the kinetics on the dinuclear iridium complex 242 revealed a first-order dependency in catalyst concentration. The authors wanted to ascribe the activity of complex 242 to its dinuclear molecular nature, because a related mononuclear iridium complex did not produce O2 as efficiently. However, the lag time in the catalytic experiments made it difficult to draw any conclusions regarding the homogeneity or whether the active species originated from the dinuclear catalyst. As for the ruthenium-based WOCs, the catalytic activity for a majority of the developed iridium-based WOCs has been assessed with the strong one-electron oxidant CeIV (Table 8). However, there exist alternative oxidants that could be used to establish if a complex is active in H2O oxidation. One of the oxidants that has been employed is NaIO4, a two-electron oxidant that does not require as acidic conditions as CeIV, making it an excellent alternative to CeIV for distinguishing and evaluating new WOCs.299−301

Figure 63. Synthesis of iridium complex 240 and its reversible cyclometalation to generate complex 241.

Treatment of complex 240 with a base, such as NaOAc, promoted cyclometalation and the formation of complex 241. Although this transformation was shown to be relatively slow at room temperature, it could be signficantly accelerated at elevated temperatures. In 1 M HCl solutions, the cyclometalation process is reversible and effectively regenerates iridium complex 240. This peculiar reversibility could be of importance during catalysis by contributing to redox catalysis via site-specific protonation/deprotonation, thus enabling PCET through ligand cooperative catalysis. Evaluating complex 240 for H2O oxidation in the presence of CeIV demonstrated that the complex was active for more than a week, giving a TON > 20 000. This catalytic system is remarkable and constitutes one of the most active nonelectrochemical systems; however, the involvement of heterogeneous iridium nanoparticles cannot fully be excluded.297 Photoelectrochemical O2 evolution was also evaluated for complex 240, and, as anticipated, the complex was able to enhance the photocurrent density. During the experiments, deposition on the photoelectrode occurred, and it was reasoned that this deposition impedes electron transfer from the dissolved iridium complex to the light-absorbing component. However, this problem could in part be resolved by changing the pH of the solution. This pH modification was believed to affect the stability and lead to dissociation of the electrodeposited material, thus regenerating the activity of the photoelectrode. Because the electron transfer from the iridium WOC to the light absorber (hematite) seemed to be ineffective in the current photosetup, it was suggested that the photocatalytic performance of the PEC could perhaps be improved by physi- or chemisorption of the iridium WOC on the photoelectrode.297 Because dinuclear metal complexes could be more stable than mononuclear ones, by distributing the oxidizing equivalents over several metal centers, de Bruin and co-workers attempted to synthesize a dinuclear iridium complex.298 This resulted in the dinuclear iridium complex, [(cod)(Cl)Ir(μbpi)Ir(cod)]+ (242; bpi = (pyridin-2-ylmethyl)(pyridin-2ylmethylene)-amine; cod = 1,5-cyclooctadiene), containing

3.4. Decomposition of Homogeneous Iridium Catalysts − Generation of Catalytically Active Heterogeneous Iridum Species

Because of the ambigious results regarding some of the developed iridium WOCs, the question concerning their homogeneity has been raised. Under the highly oxidizing conditions required for achieving H2O oxidation, there is a possibility that the WOCs decompose and give rise to active heterogeneous materials.302−304 It is well established that iridium nanoparticles, IrOx, are highly active WOCs,305 where the different particle sizes307 and the method of preparation306 influence the overall catalytic activity. This could explain that a certain characteristic behavior is observed in the beginning,307,308 followed by a switch to a different one. The task of distinguishing homogeneous catalysis from heterogeneous can be challenging. This ambiguity was addressed by Crabtree and Brudvig, who examined the anodic deposition of several iridium-based WOCs.309−312 In their first study,309 they examined the two iridium−aqua complexes 229 and 230. Electrochemical measurements of complex 229 displayed a catalytic wave with 11930

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 8. Summary and References to Available Data for Iridium-Based Water Oxidation Catalysts

a Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole catalyst per s−1. cUsing Ce(NH4)2(NO3)6 (CAN, CeIV) as the chemical oxidant. dUsing NaIO4 as the chemical oxidant. ePhotochemical oxidation using [Ru(bpy3)]Cl2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor.

11931

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

EQCN to the tris−aqua complex 229 revealed that when reaching a potential of 1.20 V vs NHE, a mass deposit was observed.310 In accordance with previous results,309 complex 229 serves merely as a precatalyst that decomposes during catalysis to generate a material that is deposited on the electrode as a heterogeneous BL. This mass deposit continues on the return cathodic scan until a potential of 1.10 V was reached and the mass on the electrode stayed basically constant. EQCN affirmed that, unlike complex 229, complex 244 does not give any deposition during voltammetry. These differences in EQCN behavior acknowledge complex 244 as a homogeneous catalyst and signal an alternation in catalysis for complexes 229 and 244, when driven by an electrochemical potential. Although these observations strongly suggest that complex 244 was stable during catalytic cycling, it did not totally exclude that suspended iridium oxide nanoparticles were generated.310 It could also be concluded that the organometallic precursor had an impact on the deposition process and ultimately on the properties on the electrodeposited material.311 This difference in catalyst precursor was manifested in a difference in deposition rate and electrochemical response of the deposited material, but more importantly, this effect was not noted when using nonorganometallic precursors. In contrast to the organometallic catalyst precursors, the use of preformed IrOx nanoparticles resulted in a deposited material that was not stable on the anodic surface when applying moderately oxidizing potentials. Grotjahn and co-workers studied the decomposition of several iridium WOCs and managed to show that iridium nanoparticles are being formed already after the addition of a few equivalents of the oxidant (CeIV). During O2 evolution, it was observed that the reaction mixtures quickly changed color to violet, blue, or olive-green, depending on the specific complex that was used. Because CeIII is colorless, the authors speculated that the different colors were connected to the various iridium complexes.313 It has been reported that depending on the additives, size, or aggregation, iridium nanoparticles have similar colors314,315 due to characteristic UV−vis absorptions in the region 550−700 nm.316 The formation of nanoparticles was subsequently confirmed by the use of scanning tunnelling electron microscopy (STEM) and powder X-ray diffraction (PXRD).313 Analysis of the reaction mixtures after H2O oxidation confirmed the formation of iridium nanoparticles during catalysis. These solutions were also studied by UV−vis, where an increase of the absorption in the 550−650 nm region was found,313 further pointing to the formation of nanoparticles.317 This increase in absorbance at 580 nm was detectable within a few minutes after the iridium complex was added to CeIV. Because of the instability of the catalysts, it was of interest to study the early stages of the catalysis. This was achieved by addition of limited equivalents of the oxidant (CeIV) to the catalyst solution. It was concluded that several equivalents of the oxidant were needed before O2 was generated, and these equivalents were proportional to the number of organic ligands in the catalyst precursor. The tris−aqua iridium complex 229, for instance, required ∼15 equiv of oxidant before evolution of O2 was detected, whereas complexes 219, 238, or 239 demanded as much as 30 equiv to give rise to O2 generation. Authentic iridium nanoparticles only required 5 equiv to trigger O2 evolution, thus emphasizing that decomposition of the

an onset at 1.15 V vs NHE. Upon successive scans, a blue layer (BL) was formed on the electrode corresponding to an iridium oxide material, which was accompanied by the appearance of a redox process at 0.88 V. In line with the deposition of the BL, the onset wave at 1.10 V also increased in intensity. Removing, rinsing, and transferring the electrode to a fresh solution confirmed the presence of the catalytic wave and the reversible feature. Ideally, the peak separation (ΔE) is zero; however, for the reversible process, the peak separation was found to be ∼15 mV, which is compatible with a nondiffusional, surface-bound species. During the electrochemical measurements, O2 bubbles were observed at the electrode surface, verifying that the deposited BL material was responsible for the catalysis. The deposited BL was shown to be a robust catalyst that was active for several hours. It could also be concluded that the presence of the Cp* ligand in the iridium precursor was necessary for deposition. For instance, the use of iridium trichloride hydrate as the iridium precursor did not result in any detectable deposition nor any waves corresponding to catalytic H2O oxidation. For the related Cp* iridium WOCs, [Cp*Ir(κ2-ppy)Cl] (219) or [Cp*Ir(bpy)Cl]+ (222), no deposition was observed. BL formation was also not electrode-dependent and could be observed on several types of electrodes, such as carbon, platinum, gold, FTO, and ITO electrodes. The authors also examined the overpotential for O2 generation and found that a relatively low overpotential of ≤300 mV existed for the BL deposited material, thus highlighting a low kinetic barrier for the O−O bond formation. These findings support the fact that an inorganic iridium deposit is responsible for mediating the catalytic oxidation of H2O when applying the two iridium−aqua precursors 229 and 230.309 In their subsequent work, Crabtree and Brudvig focused on implementing gravimetry, by using an electrochemical quartz crystal nanobalance (EQCN) to probe the homogeneity of the iridium complexes depicted in Figure 65.310 Complexes 243

Figure 65. Molecular structures of Cp* iridium complexes 243 and 244.

and 244, housing the pyridyl-alkoxy ligand, were chosen because they are completely water-soluble, thus allowing for an extensive electrochemical characterization in aqueous solutions. All three complexes exhibited activity in H2O oxidation, and the evolved O2 was determined by rotating ring-disk electrode (RRDE) electrochemistry and by a Clark electrode. After being used in a solution containing complex 244, the electrode was rinsed and transferred to a fresh solution (free of 244). The electrode showed essentially the same activity as the original background, suggesting that no catalytically active deposited material was present on the electrode. EQCN was employed to detect any changes in mass at the surface of the working electrode under the electrochemical measurements. This is an extremely sensitive method that has the possibility to detect changes down to nanograms per square centimeter. Applying 11932

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 66. 4,4′-Substituted and 6,6′-substituted bpy-type [Cp*Ir] complexes 245−251.

organic ligand framework of the complexes was involved in the activation of the iridium WOCs.313 Fukuzumi and co-workers introduced substituents into the bpy ligand framework to study the effects of these modifications.318 By synthesizing different 4,4′-substituted bpy ligands, housing either electron-donating or electron-withdrawing functional groups, they assessed the catalytic reactivity, and more importantly, the susceptibility of the corresponding iridium complexes toward nanoparticle formation (Figure 66). Examination of the catalytic reactivity of the different [Cp*Ir] complexes in H2O oxidation was performed with CeIV as the chemical oxidant. By monitoring the spectral change of CeIV at 420 nm, it was revealed that all catalysts except the OH-functionalized catalyst 245 exhibited constant reaction rates that decreased toward the end of the catalytic experiments (due to depletion of the oxidant). By contrast, catalyst 245 displayed an increase in rate with reaction time, suggesting that the catalyst was transformed into a catalytically more active species. Electrochemical properties, that is, onset potential and anodic current, for the substituted complexes 245−248 matched the order of catalytic reactivity when driven by CeIV. The data obtained for 245 suggested that catalyst degradation occurred and that nanoparticle formation was responsible for the observed catalytic activity. To monitor if nanoparticle formation occurred, DLS measurements were conducted on the solutions after catalysis for the different catalysts. As suspected, nanoparticles could be detected with catalyst 245, whereas for catalysts 246−248 no direct evidence for nanoparticle formation could be obtained.318 The property of the OH-substituted complex to decompose to nanoparticles might arise from the relative ease of oxidizing the phenolic OH to a quinoid-type structure, which could explain the accelerated decomposition and nanoparticle formation. The 6,6′-substituted iridium catalysts 249−251 have also been reported to generate iridium nanoparticles/clusters under similar catalytic conditions.319 However, according to a recent study, even complex 245 functions as a homogeneous catalyst when periodate is used as the oxidant.320 These studies highlight the delicate balance between maintaining a stable ligand framework and the production of iridium nanoparticles, which can act as highly active catalysts. It should be noted that the iridium-based catalyst 248 and its phosphonate analogue have also been covalently anchored onto ITO electrodes for electrochemical H2O oxidation.321 This approach resulted in functionalized electrodes that were shown to be highly robust and efficicent in mediating H2O oxidation. The group of Macchioni chose to investigate the decomposition of three iridium-based WOCs depicted in Figure 67: 222, 252, and the tris−aqua complex 229. Following

Figure 67. A series of iridium WOCs studied by the group of Macchioni.

up on their previous studies,322,323 where it was concluded that catalysts 222 and 252 exhibited a first-order dependence in catalyst concentration, the authors were now interested in delineating possible decomposition pathways for the catalysts 222, 252, and 229.324,325 It was revealed that the order for catalyst 229 was 1.5,324,325 thus verifying the finding by the group of Brudvig and Crabtree.288 The robustness of the catalysts was evaluated by two different methods: (i) a second aliquot of oxidant (CeIV), containing the same equivalents of oxidant, was added to the reaction before the first aliquot was completely consumed, and (ii) a delay was introduced between complete consumption of the first aliquot of oxidant and before the addition of the second aliquot. Complexes 222 and 252 exhibited a drop in performance (∼20% for each run) in both experiments. By contrast, the tris−aqua complex 229 did not display any decrease in catalytic performance during several runs in the absence of delays. However, implementation of delays between the runs was shown to result in a decrease of the catalytic activity. This implied that, in contrast to catalysts 222 and 252, catalyst 229 is deactivated mainly after the reaction had ceased.324,325 In situ NMR studies were conducted to establish if the active species were derived from the decomposition. For complex 229, it was shown that the Cp* ancillary ligand decomposed to acetic acid, and presumably also to carbon dioxide,324,325 which has previously been observed for a rhodium complex.326 Remarkably, after adding only a few equivalents (20 equiv) of CeIV, less than 3% of complex 252 remained intact in solution. By using both 1D and 2D NMR experiments, the authors elucidated the probable decomposition pathways for the Cp*containing iridium WOCs. It was indicated that the CH3 moieties in the Cp* were oxidized to CH2OR (where R = H or OH), which have the possibility of being further oxidized to CHO or COOH. Moreover, it seemed like the expulsion of one carbon moiety from the Cp* ligand (in the form of CH2OHCOOH or HCOOH) triggered further decomposition processes. The proposed decomposition pathway of the Cp* iridium WOCs is depicted in Scheme 15. DFT calculations suggest that the decomposition pathway is mediated by an 11933

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 15. Proposed Decomposition Pathway for Cp* Iridium WOCs (R = H or OH)

importance of measuring more than O2 evolution when determining the molecularity of new iridium catalysts.

intramolecular attack of a superoxide ligand on the carbons in the Cp* ring, and the carbons of the CH3 groups. This destructive pathway seems to be favored by an η1-coordination of the O2 moiety, that is, the formation of an [Cp*Ir(η1-O2)] species.324,325 The question of the stability of the developed iridium-based WOCs is of fundamental importance. The high catalytic activity associated with IrOx materials and the formation of small clusters can cause severe problems in attempts to distinguish between homogeneous and heterogeneous catalysis. The obvious reason is that the decomposition of only a small fraction of the initial homogeneous molecular complex into heterogeneous materials might be enough to account for the observed catalytic activity. The organic ligand frameworks might also affect the degradative process and even remain in the heterogeneous material, thus influencing the properties of the generated heterogeneous species, in terms of both structure and catalytic activity. The transformation of a homogeneous complex into a heterogeneous catalytically active material may thus be strongly affected by the reaction conditions. As a result, catalysis might be homogeneous under a certain set of conditions but heterogeneous under another similar set. The extensive studies of the iridium-based WOCs comprising cyclopentadiene-type ligand scaffolds have shown that these types of WOCs have a tendency to be transformed into a catalytically active heterogeneous phase (Table 9). This causes severe challenges when studying these WOCs and raises questions regarding their topicity, where the specific ligand, reaction conditions, or time scale have to be considered. This Review does not intend to address if iridium nanoparticles are the sole active WOC responsible for all of the observed catalytic activity under the employed reaction conditions. However, it highlights the

4. ARTIFICIAL CATALYSTS HOUSING EARTH-ABUNDANT FIRST-ROW TRANSITION METALS 4.1. Early Biomimetic Manganese-Based Photosynthetic Mimics

Inspired by the Mn4Ca cluster in the natural OEC, chemists have dedicated major efforts toward the synthesis of manganese-based WOCs. Manganese has a high natural abundance, low cost, and is especially interesting because it is the metal chosen by nature. Furthermore, it has a rich redox chemistry, granting access to a wide range of oxidation states with strong oxidizing power. Unfortunately, this also brings about an intrinsic instability of these high-valent species, which calls for effective ligand stabilization to prevent them from decomposing. In addition, the strongly oxidizing property of such manganese species requires that the ligand scaffolds are robust and stable against oxidative degradation. In nature, the Mn4Ca cluster is kept together by several bridging μ-oxo ligands, and is further stabilized by surrounding amino acids containing imidazole and carboxylate functionalities. This efficient stabilization allows the Mn4Ca cluster to cycle up to 106 times between the five oxidation states of the Kok cycle during the oxidation of H2O. However, due to the presence of highly oxidizing species throughout this cycle, damage is regularly inflicted on the surrounding peptide structures, resulting in a rapid turnover of the polypeptide constituents of the PS II.4 Initial synthetic work focused heavily on the construction of simplified artificial mimics containing structural features present 11934

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 9. Summary and References to Available Data for bpy-Containing and Related Iridium-Based Water Oxidation Catalysts

a Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole catalyst per s−1. cUsing NaIO4 as the chemical oxidant. dUsing Ce(NH4)2(NO3)6 (CAN, CeIV) as the chemical oxidant. eA higher TOF value 0.50 s−1 has also been reported (see ref 323).

11935

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 68. Coupled ruthenium−manganese assemblies 260−267.

various redox-active manganese complexes (Figure 68). This strategy permitted construction of more accurate models of the PS II, which mimicked the light-harvesting process performed by the natural chlorophylls and the subsequent coupled electron-transfer events occurring in the OEC. By subjecting these assemblies to flash photolysis, it proved to be possible to trigger electron transfer from the excited RuII center to a sacrificial electron acceptor. The generated hole at the ruthenium center was then rapidly refilled by an intramolecular electron transfer from the adjacent manganese complexes. Clear evidence that the electron transfer was indeed intramolecular could be obtained by comparing the rates of regeneration of the RuII center in the absence and presence of a covalently linked manganese complexes. In all cases, the rates in the linked assemblies proved to be several orders of magnitude faster than those in the intermolecular systems. Further support for an intramolecular electron transfer was given by EPR measurements, which confirmed that oxidation of MnII to MnIII took place in the presence of the photo-oxidized RuII unit.6 The assemblies 260−267 provided insight into the parameters that influence the rate and efficiency of the excited-state quenching and the intramolecular electron transfer. For instance, it was demonstrated that the quenching

in the OEC, which could be utilized as model systems for gaining fundamental insight in the mechanism of H2O oxidation. This research yielded a myriad of structural mimics, differing in oxidation state, ligand scaffolds, and nuclearity.327 Dinuclear complexes comprised of at least one bridging μoxo constitute the most common and well-studied group of mimics, and in this extensive work a vast variety of different nitrogen-based ligand frameworks have been explored, ranging from bipyridines,328 terpyridines,329 and phenantrolines330 to more elaborate multidentate ligands.331 Although these dinuclear mimics proved useful in elucidating the basic concepts of the coordination chemistry of manganese with biomimetic ligands, these models were not sufficient to mimic the delicate tetranuclear structure of the Mn4Ca cluster in the OEC. Consequently, this motivated the design of mimics with higher nuclearity, which led to the construction and characterization of several tri-332 and tetranuclear333 complexes that opened up for a deeper understanding of the processes occurring in the active site of the OEC. Taking this concept a step further, research initiated by the Åkermark group during the 1990s focused on the construction of a series of heterobimetallic assemblies in which the photosensitizer (a [Ru(bpy)3]2+-unit) was covalently linked to 11936

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

rate constant decreased exponentially with the metal-to-metal distance and that it was possible to modulate the lifetime of the excited state by appropriate tuning of the ligand framework.6 Early examples also exist of molecular systems where photochemical charge separation was coupled with electron transfer from a metal complex to an external electron acceptor. These were based on the interfacing of manganese-based catalyst entities to photoactive ruthenium chromophores.334−337 Upon photoexcitation, electron transfer occurred from the excited state of [Ru(bpy)3]2+ to an external electron acceptor. The photo-oxidized ruthenium chromophore subsequently abstracted an electron from the manganese center(s) intramolecularly. However, in most of the cases, these coupled ruthenium− manganese assemblies were only capable of yielding monooxidized catalytic centers, and further electron-transfer processes most likely resulted in catalyst decomposition rather than H2O oxidation. In one case, in the coupled ruthenium− manganese assembly 268, light excitation and electron transfers led to the accumulation of three oxidizing equivalents at the dinuclear manganese moiety, thereby converting it from Mn2II,II to Mn2III,IV (Figure 69).338 During this process, the acetate

Figure 70. Molecular structures of the three manganese porphyrin complexes 269−271.

vs NHE, a clear sign of H2O oxidation activity. In contrast, analogous measurements of the anodic oxidation of the aqueous acetonitrile solutions made in the absence of complexes 269−271 did not afford any O2 evolution in the potential range up to +2.50 V vs NHE. Interestingly, on switching to the related monomeric manganese porphyrin complexes, the O2 evolution ceased, emphasizing the need of having two manganese centers in close proximity. Among the three complexes 269−271 evaluated in this study, complex 271 exhibited the highest activity, with a maximum TON of 9.2, although it was found to require a significantly higher oxidation potential than the other two. Moreover, O2 evolution was found to be proportional to the concentration of the used manganese complex, suggesting its participation in the ratedetermining step. Control experiments for all complexes with 18 O labeled H2O established that both oxygens in the evolved O2 originated from the solvent H2O. The mechanism involving 269−271 was postulated to begin with the oxidation of two MnIII−OH units into the corresponding MnVO species, which then as the result of the close proximity of the two Mn centers proceeds with the formation of a MnIV−O−O−MnIV complex. Finally, decomposition of this complex, triggered by a OH− ion replacement of the peroxy bridge, leads to the release of O2 and regeneration of the starting Mn2III dimer.339 In a later mechanistic study using meta-chloroperoxybenzoic acid (mCPBA) as the chemical oxidant, the authors were able to thoroughly characterize the dimeric MnVO species and confirm its key role in the O2 evolution process.340 The next breakthrough was made in 1999 by the group of Crabtree and Brudvig with the development of the dimeric complex [(H2O)(tpy)Mn(μ-O)2Mn(tpy)(OH2)]3+ (272, Figure 71), which was the first manganese catalyst capable of promoting chemically driven H2O oxidation.341−343 Since then,

Figure 69. Coupled ruthenium−manganese assembly 268, where the Mn2II,II unit is capable of mediating three consecutive electron transfers to the photogenerated RuIII moiety to generate a Mn2III,IV complex.

ligands that were coordinated to the manganese centers were substituted by aqua ligands, thus enabling PCET events and access to the high-valent Mn2III,IV state. This work is related to the processes occurring in PS II and was a major step toward coupling light-sensitized single-electron transfer with multielectron catalysis. 4.2. Manganese-Based Catalysts Capable of Generating Oxygen

The first steps toward realizing O2 evolution by a molecular manganese complex were taken by the group of Naruta in 1994.339 They prepared the three analogous dimeric face-toface manganese triphenylporphyrin complexes 269−271 that were demonstrated to catalyze electrochemical H2O oxidation (Figure 70). In aqueous acetonitrile solutions (5% v/v H2O in acetonitrile) containing nBu4NOH, all three complexes exhibited a similar irreversible increase in current at ≥1.40 V

Figure 71. Molecular structure of the [(H2O)(tpy)Mn(μ-O)2Mn(tpy)(OH2)]3+ complex 272. 11937

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 16. Plausible Mechanism for O2 Evolution Mediated by the [(H2O)(terpy)Mn(μ-O)2Mn(terpy)(OH2)]3+ Catalyst 272

several derivatives of this complex, containing substituted bpy ligands, have been synthesized and evaluated for their activity in O2 evolution.344,345 Examination of the crystal structure of this bis(μ-oxo) MnIIIMnIV bridged complex 272 revealed that it was a mixed valence dimer, in which the two manganese centers

were crystallographically identical because the dimer is situated in an inversion center. Remarkably, the complex also accommodated two exchangeable aqua ligands, one on each Mn center. These exchangeable aqua ligands are of fundamental importance for the reactivity of this complex as 11938

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

involving 272 is intricate and depends greatly on the exact reaction conditions. Furthermore, heterogenization of complex 272 onto materials such as kaolin,353,357 mica,357−360 montmorillonite,353,361 TiO2,362,363 or even into a metal−organic framework (MOF)364 has proved to make its utilization in CeIV-driven H2O oxidation more reproducible. The explanation for this effect was primarily ascribed to two factors: (i) heterogenization of 272 improves its stability and consequently suppresses permanganate formation, and (ii) a higher effective concentration of 272 may allow it to operate via an alternative mechanism involving cooperation of two catalyst molecules. However, it has also been suggested that the improved activity originates from transformation of the molecular manganese complex into layered manganese oxide particles, which are the true catalytic species.365 In 2003, McKenzie and co-workers reported on the preparation and characterization of two related dimeric Mn2II,II complexes comprising the monoanionic and pentadentate ligands N-methyl-N′-carboxymethyl-N,N′-bis(2pyridylmethyl)ethane-1,2-diamine (Hmcbpen) and N-benzylN′-carboxymethyl-N,N′-bis(2-pyridylmethyl)ethane-1,2-diamine (Hbcbpen).366 Although the authors were initially motivated to use these complexes as model systems for manganese lipoxygenases, the mechanistic work conducted in the study later provided the foundation for the application of the [Mn2II,II(mcbpen)2(H2O)2]2+ complex 278 as a catalyst for O2 evolution (Figure 72).367,368 The crystal structure revealed

they allow the complex to take part in H2O oxidation when using sodium hypochlorite (NaOCl) as the chemical oxidant.341 In this study,341 an initial rate of O2 evolution of 12 mol O2 per h per mol of 272 and four catalytic turnovers over 6 h were observed when the reaction was carried out in 0.07 M aqueous NaOCl (pH 8.6). The modest catalytic turnover of 272 was ascribed to a rather facile oxidative breakdown of the ligand, ultimately resulting in the liberation of the manganese ions and the formation of inactive permanganate. The formation of permanganate could be monitored by UV−vis spectroscopy, using its characteristic isosbestic points at 497 and 583 nm. Additional evidence for the oxidative decomposition of 272 was provided by measurements of the consumed ClO−, which was found to be approximately 10 times greater than the amount needed to evolve the observable amounts of O2. Isotopic labeling studies using H218O confirmed that 75% of the oxygen content in the evolved O2 was derived from the solvent H2O, while the rest was found to come from the oxidant ClO−. On the basis of the isotopic distribution, the authors proposed the mechanism depicted in Scheme 16. In this mechanism, complex 272 undergoes a one-electron oxidation to the corresponding Mn2IV,IV species 273, which subsequently exchanges one of its aqua ligands with ClO− to form intermediate 274. This is followed by a two-electron oxidation of 274 to generate the bis(μ-oxo) MnVMnV16O species 275, expulsing Cl− in the process.341,346 From here, it was suggested that species 275 could rapidly exchange with a 18 OH− or H218O on the other manganese center of the dimer to give a bis(μ-oxo) 18OMnVMnV species (276). This is consistent with the isotopic content of 75% 18O and 25% 16 O, which was indeed observed in the evolved O2 during the isotopic labeling experiments. Any of the two MnVO species (MnV16O (275) or MnV18O (276)) can then react with an outer-sphere H218O molecule to produce O2 and the bis(μ-oxo) MnIIIMnIII dimer 277.341 Subsequent studies demonstrated that replacing ClO− with Oxone (KHSO5) resulted in a more efficient O2 evolution.342,347−349 The increased efficiency was ascribed to the higher reactivity of Oxone toward the bis(μ-oxo) MnVMnVO species 275. However, the performance of this system proved to be highly dependent on the ratio of 272 and Oxone. When a 1:500 ratio was used, the majority of complex 272 was converted into permanganate, and less than 10% of the oxygen atoms in the evolved O2 originated from the solvent H2O. Lowering the ratio to 1:100 prevented breakdown of catalyst and allowed for incorporation of 50% of the oxygen atoms from H2O into the evolved O2. Unfortunately, the major limitation of 272 was the necessity of using oxygen transfer reagents to drive the oxidation of H2O. Although a later study by the group of Brudvig and Crabtree claimed that it was possible to use homogeneous 272 together with the one-electron oxidant CeIV to drive H2O oxidation,350 several other reports seemed to dismiss this finding.351−353 It has also been shown that oxidation of the dimeric manganese complex 272 resulted in the formation of a tetranuclear manganese complex, a so-called “dimer of dimers” complex.354,355 An early report claimed that this tetranuclear complex could not oxidize H2O electrochemically,352 while a more recent report presented contradictory results where the “dimer of dimers” complex in fact was able of performing electrochemical H2O oxidation with a TON of 2.8.356 Collectively, these ambiguous results show that H2O oxidation

Figure 72. Structure of the [Mn2II,II(mcbpen)2(H2O)2]2+ complex 278.

some notable features of complex 278, such as the sevencoordinate geometry of the two manganese centers and the dual purpose of the carboxylate groups to bridge the dimeric complex and to hydrogen bond to the aqua ligands. Complex 278 was subjected to tert-butyl hydroperoxide (TBHP) in acetonitrile or alcoholic solvent with the aim of investigating whether it was possible to access higher oxidation states of the manganese centers, which would have long enough lifetime to allow identification by electrochemical, spectroscopic (EPR and UV−vis), or spectrometric methods (MS). These experiments provided detailed mechanistic insight by showing the appearance of several high-valent metal species. For instance, treating 278 with TBHP in H2O or MeOH led to an oxidative cleavage of the dimeric structure, generating two monomeric MnIII−OH or MnIII−OR species. This caused an observable change of the color of the reaction solution from beige to brown, and when H2O was used as solvent, gas 11939

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 17. Proposed Mechanism for O2 Evolution for Complex [Mn2II(mcbpen)2(H2O)2]2+ (278)

resulted in the evolution of 1.06 mmol O2, which corresponded to a TON of 6.2. Control experiments performed under identical conditions, in which Mn(ClO4)2 was used as catalyst instead, yielded no O2 evolution and supported the unique reactivity of 278. The hepta-coordination was invoked to explain why the complex displayed a remarkable stability toward oxidation both in the solid state and in solution, which is different from a majority of related Mn complexes. In contrast to 272, complex 278 did not decompose to permanganate after prolonged reaction times.367 On the basis of the results from the earlier study,366 the authors proposed a mechanism for how complex 278 mediated H 2 O oxidation (Scheme 17).367 As determined by a combination of ESI−MS and UV−vis spectroscopy, the reaction of 278 and TBHP yielded the mono(μ-oxo)-bridged Mn2III,III-species 280, most likely via a pathway involving a MnIII−OH intermediate (279) where the solvent H2O constitutes the source of the OH-ligand. Subsequent oxidation of 280 by another molecule of TBHP affords the bis(μ-oxo)bridged Mn2IV,IV-complex 281, which was previously observed

evolution could be observed. When the reaction was performed in an acetonitrile solution, two other species were formed, which were identified as the mono-μ-oxo-bridged Mn2III,III complex and the bis-μ-oxo-bridged MnIV,IV complexes. These complexes are most likely formed in the H2O/MeOH reactions as well, but under those conditions they have an insufficient lifetime to allow detection by the techniques used.366 The structure of 278 and the discovery that it could yield high-valent manganese species made the authors aware of the fact that it could be considered as a model for the OEC rather than manganese lipoxygenases. By the use of a Clark electrode and membrane inlet mass spectrometry (MIMS), they could confirm that the gas evolved from the reaction between 278 and TBHP in aqueous media was indeed O2. Using 18O labeled H2O, significant formation of 16O18O was observed, arising from incorporation of one oxygen from H2O and one from the oxidant TBHP. Instigated by complex 278’s ability to function as a WOC, the authors pursued more elaborate catalytic experiments in which a 0.5 mM solution of 278 (0.17 mmol) was subjected to addition of four times 20 equiv of TBHP. This 11940

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 73. Manganese complexes studied for H2O oxidation.

Anderlund and co-workers synthesized the tetranuclear manganese complex 286 and related tetranuclear complexes,351,371,372 which are closely related to McKenzie’s dinuclear complex 278. Both complexes 278 and 286, together with the related complexes 283−285, and complex 272 were then studied under H2O oxidation conditions. The overall goal was to conduct a systematic comparison of a number of potential manganese WOCs (Figure 73) under identical conditions, using several different oxidants where the wellstudied ruthenium-based WOC 8 (“blue dimer”) and heterogeneous RuO2 were used as references (Table 10).351

experimentally in the acetonitrile reactions. Collapse of 281 via the proposed species 282 liberates O2 and regenerates 278, thereby closing the catalytic cycle.367 Further DFT calculations also indicated the involvement of monomeric MnII centers that are oxidized to MnIV-oxos.369 An interesting feature of this mechanism is the active involvement of the carboxylate moiety of the ligand framework that is suggested to play an important stabilizing role by alternating between various nonbinding and binding modes. This mechanistic role is reminiscent of the so-called “carboxylate shift” that is proposed to be involved in the mechanisms of several natural nonheme iron enzymes.370 Complex 278 was also evaluated in CeIV-driven H2O oxidation to determine if it could function as a WOC when nonoxygen transfer oxidants were used. Upon addition of (NH4)2[Ce(NO3)6] to an aqueous solution of 278, immediate O2 evolution was triggered, although it turned out to be significantly less pronounced than when TBHP was employed as oxidant. In contrast to the experiments with TBHP, the use of CeIV led to a rapid decrease in pH to 1. This was invoked by the authors to explain the decrease in catalytic efficiency, because a low pH would make the formation of high-valent oxo-bridged manganese species unfavorable and allow for ligand dissociation by protonation. However, the most surprising result from this study was found when examining the isotopomeric ratio of the generated O2. These data revealed that also when CeIV was used as the oxidant to drive H2O oxidation, only one oxygen atom in the evolved O2 originated from H2O, suggesting the occurrence of an oxygen transfer reaction. The only source of oxygen besides the H2O in these experiments was the nitrate ion in the oxidant, and therefore it was concluded that 278 was reactive enough to abstract an oxygen atom from the nitrate ion.367 However, it is important to point out that the authors did not find any evidence of the reduction of nitrate ions, and re-examination of these reactions may be warranted.

Table 10. Oxygen Evolution Rates (mMO2 min−1 Mmetal−1) for a Variety of Catalysts Capable of Promoting O2 Evolutiona oxidant compound

H2O2b

TBHP

HSO5−

ClO−

CeIV

[Ru(bpy)3]3+

278 272 283 284 285 286 blue dimer (8) RuO2

1.8 7.5 33 320 200 >500 traces

36 traces ∼1 nd nd 34 traces

>500 38 16 nd nd 105 2.5

7 traces nd nd nd nd >500

nd nd nd nd nd nd >500

nd nd nd nd nd nd traces

66

nd

47

2.2

145

22

a

Rates were measured by a Clark electrode after 2 min, after addition of 50 equiv of the oxidant. bOnly 2 equiv of H2O2 was added.

For the reactions, anaerobic aqueous solutions containing the complexes were placed in the reaction chamber of a Clark-type polarographic oxygen electrode and excess of the different oxidants was subsequently added, after which the O2 evolution was monitored during for 2 min. The final concentration of the metal in all experiments was 2 mM, and to this was then added 11941

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

The group of Åkermark and Sun has also dedicated extensive research toward the development of manganese-based WOCs and initial work focused on corrole-based systems due to their well-known ability to stabilize coordinated metal-centers in high oxidation states. This led the group to prepare the cationic mono- and bis-manganese corrole complexes 289 and 290.377 In analogy to complexes 269−271 prepared by Naruta and coworkers, the two ligands of the bis-manganese corrole complex 290 are situated in a face-to-face manner that locks the relative position of the two metal centers and enables the stabilization of high-valent states (Figure 75). However, in contrast to complexes 269−271, a xanthene moiety was chosen as a bridge instead of an aryl one, because it was anticipated to provide higher flexibility. It was also envisioned that the free arm on the xanthene moiety of complex 289 could participate in H2O oxidation by facilitating the coordination of aqua ligands to the metal center via hydrogen bonding. Both complexes were thoroughly characterized by a combination of 1H NMR, MS, and UV−vis spectroscopy, which enabled structure assessment. Although 1H NMR interpretation proved to be difficult due to the paramagnetic properties of manganese corroles 289 and 290, clear support for their structure was provided by MS. For 289, a peak in the negative mode could be observed that corresponded to the molecular ion [M − H+]−, while 290 gave rise to a molecular ion peak in the positive mode ([M + H+]+). To gain additional proof of the structure of 290, it was further characterized by HRMS (positive mode), which allowed for the detection of the three characteristic ions: 808.5800 [(M − 2Cl)/2]+, 826.0650 [(M − Cl + H+)/2]+, and 1652.5498 [M − Cl]+. By comparison of the measured and calculated results, the manganese centers in complex 292 were both confirmed to be in the MnIV oxidation state. In agreement with the HRMS studies, the UV−vis spectra also verified that the metal centers in both corrole complexes were in the MnIV oxidation states.377 Electrochemical measurements of the two manganese corrole complexes revealed that high-valent oxidation states could be reached at relatively low potentials. Interestingly, it was also shown that there existed an intramolecular interaction between the two metal centers in complex 290, which could be important in H2O oxidation catalysis. The catalytic activity of both corrole complexes was assessed by electrochemistry where scanning to anodic potentials (E > 1.00 vs NHE) revealed a reduction peak at −1.09 V. This peak was assigned to the reduction of O2, thus providing strong support that the manganese complexes could access high-valent states at low redox potentials, making them promising candidates for artificial WOCs.377 Encouraged by these results, the authors conducted a followup study that was facilitated by insight gained from DFT calculations.378 The calculations were found to favor a mechanism proceeding via nucleophilic attack by OH−/H2O on a MnVO intermediate. To support this mechanistic proposal experimentally, the related monomeric MnIII corrole complex 291 (Figure 76) was synthesized as it was hypothesized to give rise to a MnVO species with sufficient stability to allow direct detection.379 Treating 291 with the oxidant TBHP generated the expected MnVO, which after addition of nBu4NOH resulted in rapid O2 evolution. The processes could be followed by real-time MS, and the catalytic species were identified by a combination of UV−vis and HRMS.

50 equiv of the oxidant (except for H2O2 where only 2 equiv was used due to a rapid disproportionation reaction).351 Regarding the general activity of the different oxidants, the study established an efficiency trend among the oxygen transfer reagents for the manganese complexes (H2O2 > Oxone > TBHP > ClO−). Essentially, O2 evolution driven by ClO− seemed to be a unique feature for the ruthenium-based complexes, with the exception of 272 that displayed minor evolution of O2 on about 7 mM min−1 Mmetal−1. However, because no isotope experiments were conducted in this study, no information regarding the varying extent of oxygen incorporation from the different oxidants into the evolved O2 and how this was affected by the different catalysts could be obtained. Because it is possible that O2 evolution can occur through disproportionation of the oxidant instead of through H2O oxidation, caution must be taken when interpreting these results. Another important conclusion that was drawn from these experiments was that none of the manganese complexes was capable of mediating O2 evolution when the one-electron oxidants CeIV or photogenerated [Ru(bpy)3]3+ were used. In contrast, both of the ruthenium-based controls were able to catalyze H2O oxidation, when driven by these one-electron oxidants.351 In a subsequent study, the combination of isotopic-labeling experiments (H218O) and EPR was used together with photogenerated [Ru(bpy)3]3+ to study O2 evolution and catalytic intermediates.373 Here, it was found that among complexes 272, 283, and 286, only complex 283 yielded O2, albeit in substoichiometric amounts. It is also noteworthy to point out that related dinuclear manganese complexes as those studied by the group of Styring has been explored by Kurz and co-workers.374,375 Brudvig and co-workers have recently investigated the two single-site manganese complexes 287 and 288 (Figure 74), and

Figure 74. Structures of the two single-site manganese complexes 287 and 288 containing pentadentate ligands.

compared their catalytic activities in chemical-driven H2O oxidation using Oxone and H2O2.376 The complexes were comprised of pentadentate ligand frameworks containing primarily pyridyl motifs as the coordinating groups. The authors could show that the incorporation of a carboxamido moiety trans to the labile coordination site allowed the formation of a high-valent manganese species capable of mediating O2 formation. This was ascribed to the strong σdonor ability of the carboxamido ligand that stabilizes the crucial MnIII resting state and promotes further oxidation to higher oxidation states at lower redox potentials. This study highlighted that ligand scaffolds containing negatively charged groups stabilize the manganese center(s) at high oxidation states and that this feature might constitute a useful approach for the construction of robust WOCs. 11942

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 75. Structures of the manganese corrole complexes 289 and 290.

loss of a proton afforded intermediate 294 that could be detected by MS. From this intermediate, O2 production could be triggered directly by disproportionation of intermediate 294 or via a two-step process that involves oxidation of 294 to a MnV-species, which then releases O2.379 Continued work by the group of Åkermark focused on improving the stability of the manganese moiety in assembly 268, which had earlier been reported to be capable of delivering three electrons to a photogenerated ruthenium(III) center.338 A fourth electron transfer would most likely have afforded a species capable of oxidizing H2O to O2, but decomposition of the ligand via oxidative degradation of the labile benzylic amines presumably occurred instead. In attempts to alleviate this problem, a biomimetically inspired and a more oxidatively stable ligand containing both imidazole and carboxylate groups was prepared.380 An

Figure 76. Structure of MnIII corrole complex 291.

The mechanism based on these results begins with the reaction between 291 and TBHP to generate the crucial MnVO species (292, Scheme 18). This is followed by the attack of OH−/H2O (for simplicity, only OH− is depicted in Scheme 18) to generate a MnIII−OOH species 293, which after

Scheme 18. Proposed Mechanism for O2 Evolution for MnIII Corrole Complex 291

11943

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 77. Molecular structure of the dimeric manganese complex 295 (left) and the crystal structure of the tetranuclear manganese complex 296 generated from complex 295. Right: Reprinted with permission from ref 380. Copyright 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

Figure 78. Molecular structure of the dinuclear Mn2II,III complex 297 containing a distal carboxylate group (left) and the calculated structure of the complex in its formal Mn2IV,V state, showing the hydrogen-bonding interaction between the distal carboxyl group and the Mn bounded hydroxide(s). Right: Reproduced with permission from ref 382. Copyright 2014 PCCP Owner Societies.

efficient system that reached a higher TON of 4. Isotopic labeling experiments using H218O were conducted, and the isotopomeric ratio measured by MS confirmed that both oxygen atoms in the evolved O2 were derived from the solvent H2O. This makes 295 the first manganese-based complex capable of mediating true H2O oxidation driven by a singleelectron oxidant. Moreover, the nature of the single-electron oxidant allowed the catalytic system to be driven by visible light, which provide the first step toward realizing an artificial photosynthetic device housing a manganese complex as the WOC unit.380 Further work focused on linking the dinuclear Mn2II,III complex 295 to a [Ru(bpy)3]2+-type photosensitizer.381 These efforts resulted in a Ru−Mn2 assembly that was capable of catalyzing chemical H2O oxidation; however, photochemical H2O oxidation in the presence of an external electron acceptor failed. From photophysical measurements, it could be shown that the designed assembly suffered from a short-lived excitedstate lifetime. By the use of time-resolved emission studies, it was demonstrated that the decay of the excited state was intricate and fast, and could not be modeled with a singleexponential decay function. The photophysical properties were also influenced by pH, where neutral and high pH significantly shortened the lifetime of the excited state. Calculations revealed the existence of an intricate excited-state manifold together with different protonation states, which could contribute to the complicated photophysical behavior. This work highlights the

interesting feature of this ligand scaffold is the negatively charged carboxylate moieties that lower the oxidation potential of the metal complex. This made it compatible with singleelectron oxidants, such as photogenerated [Ru(bpy)3]3+-type complexes. Complexation of the ligand (H5L) with Mn(OAc)2 in MeOH afforded the dimeric [Mn2II,III(H2L)(OAc)(OMe)] complex 295 (Figure 77, left), which according to the crystal structure adopts a tetranuclear structure (296) in the solid state (Figure 77, right) that bears resemblance to the structure of the tetranuclear manganese core in the natural OEC. The catalytic experiments were commenced by subjecting 295 to a 480-fold excess of [Ru(bpy)3]3+ in phosphate buffer (0.1 M, pH 7.2), which immediately resulted in O2 evolution that could be detected in real-time by the use of MS. After about an hour, O2 evolution had ceased, giving a TON of ∼25 and an initial TOF of ∼0.027 s−1. Control experiments where 295 was either replaced with Mn(OAc)2 or the free ligand verified that 295 is indeed responsible for the observed catalytic activity. With these results in hand, the logical continuation was to extend this protocol to a light-driven system, where only catalytic amounts of [Ru(bpy)3]2+-type photosensitizers were used together with persulfate as a stoichiometric electron acceptor. Initial experiments using the photosensitizer [Ru(bpy)3]2+ resulted in only one turnover, which was ascribed to the rather low oxidation potential of the photosensitizer (1.26 V vs NHE). However, replacing it by [Ru(bpy)2(deeb)]2+, which has a higher oxidation potential and consequently grants a higher thermodynamic driving force, resulted in a more 11944

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 11. Summary and References to Available Data for Manganese-Based Water Oxidation Catalysts

11945

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 11. continued

11946

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 11. continued

a

Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole of catalyst per s−1. cElectrolysis at a potential of 2.2 V vs NHE in acetonitrile/H2O (95:5, v/v) solutions containing nBu4NOH. dUsing NaClO as the chemical oxidant. eUsing Oxone as the chemical oxidant. fUsing Ce(NH4)2(NO3)6 (CAN, CeIV) as the chemical oxidant. gUsing TBHP as the chemical oxidant. hUsing H2O2 as the chemical oxidant. iUsing [Ru(bpy)3](PF6)3 as the chemical oxidant. j Photochemical oxidation using [Ru(bpy3)](PF6)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. kPhotochemical oxidation using [Ru(bpy2)(deeb)](PF6)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. bpy = 2,2′-bipyridine, deeb = diethyl 2,2′bipyridine-4,4′-dicarboxylate.

4.3. Molecular Water Oxidation Catalysts Based on Iron

challenging task of efficiently linking sequential one-electron transfer events to the four-electron oxidation of H2O. Following the work on complex 295, the group of Åkermark prepared a series of analogous dimeric Mn2II,III complexes bearing various substituents in the para position of the phenol motif.382 All complexes had a sufficiently low redox potential to allow H2O oxidation to be driven by [Ru(bpy)3]3+. Among the investigated WOCs, complex 297 (Figure 78) containing a long alkyl chain with a terminal carboxylate group was found to be the most efficient catalyst. Interestingly, the higher catalytic efficiency of complex 297 could not be attributed to a lower onset potential, which would otherwise translate into a higher thermodynamic driving force for mediating H2O oxidation. Instead, as supported by DFT calculations, the distal carboxylate group was shown to assist the catalytically active manganese centers by preorienting the incoming H2O nucleophile as well as stabilizing the high-valent manganeseoxo/hydroxy species through hydrogen bonding (Table 11). This yields a unique hydrogen-bonded scaffold that might facilitate H2O oxidation and thus promote the crucial O−O bond formation through PCET. The possibility of comprising a noninnocent distal group within ligand frameworks in metal complexes is an interesting feature that has perhaps not received sufficient attention in H2O oxidation catalysis and could be a general strategy for promoting catalytic H2O oxidation.

During the past decade, iron chemistry has become the subject of extensive research in the field of organic chemistry and other disciplines. It constitutes an ideal element for incorporating into catalytic systems because of its low cost, low toxicity, and high natural availability.385−389 The access to higher oxidation states promotes the use of iron-based complexes in a variety of redox reactions.390,391 Consequently, iron has also been considered as a promising metal to utilize for the construction of artificial WOCs. However, to make the development of iron-based WOCs practically feasible, special ligand architectures are required that permit the generation and stabilization of highvalent iron species that are robust enough to allow H2O oxidation to occur. This challenge was perhaps the reason it took until 2010 before the first example of an artificial iron-based WOC was reported. The group of Bernhard and Collins demonstrated that the tetraamido macrocyclic ligands (TAMLs) could be used to to give high-valent iron complexes.392 Thus, five different FeIII complexes (298−302) based on this ligand scaffold with substituents of varying electronic properties were prepared (Figure 79) and evaluated in CeIV-driven H2O oxidation at pH 0.7. Complexes 299−302 were found to give rise to rapid O2 evolution at varying rates, while complex 298 proved to be inactive, perhaps because of its lower stability in acidic environment. The authors also found that the rate of O2 correlated well with the electronic properties of the ligands and that electron-withdrawing substituents increased the rate. Of the studied complexes, 302 showed the highest activity with a 11947

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 79. Structure of the Fe−TAML complexes 298−302.

TOF of >1.3 s−1 and first-order kinetics in the rate of O2 evolution. The corresponding control experiments conducted with simpler iron precursors, such as Fe2O3, Fe(acac)3, [Fe(bpy)3](NO3)2, and Fe(NO3)3 both with and without the TAML, resulted in no evolution of O2, thus confirming the unique reactivity of the Fe−TAML complexes. Another interesting finding in this study was that it was possible to replace the chemical oxidant CeIV by NaIO4, although it resulted in a slower reaction. However, NaIO4 possesses some important features as a chemical oxidant, such as being transparent in the visible part of the spectrum. Unlike CeIV, it does not require the use of strongly acidic media, thus allowing for straightforward mechanistic investigations by UV−vis spectrometry under milder reaction conditions.392 Quantum chemical calculations support a mechanism for O−O bond formation that involves generation of a TAML•−FeVO species, which can be attacked by either H2O or the nitrate ion of CAN.393,394 Three competing pathways for catalyst modification were also found, involving H2O or nitrate attack on the ligand scaffold, as well as oxidation of the amide units within the ligand. These unproductive routes could perhaps explain the fast deactivation of the catalyst under homogeneous conditions.394 Subsequent work has demonstrated that the Fe−TAML complexes can be immobilized on carbon-based supports to afford stable catalytic materials for H2O oxidation. The immobilized systems were shown to produce O2 with higher efficiencies as compared to their homogeneous counterpart, highlighting that immobilization can yield more robust electrode materials for oxidation of H2O.395 Shortly later, the group of Fillol and Costas evaluated the catalytic performance of a series of structurally diverse tetradentate and pentadentate iron complexes for H2O oxidation (303−309, Figure 80). Also here, CeIV was utilized as the chemical oxidant at pH 1.396 Among these, complexes 303−307, containing different ligand frameworks, were identified as competent WOCs. Interestingly, all of the active complexes shared the common structural feature of having two available coordination sites in a cis fashion, whereas the inactive complexes had them oriented either in a trans fashion (complex 308) or had only one available coordination site (complex 309). Of the developed complexes, complex 304 was demonstrated to be one of the most efficient homogeneous WOCs based on the first-row transition metals to date and was found to function both when using CeIV and NaIO4 as oxidant, resulting in excellent TONs and TOFs of 360 and 0.23 s−1 (at pH 1) or >1050 and 0.062 s−1 (at pH 2). Moreover, in contrast to the earlier reported FeIII TAMLs where the major part of the

Figure 80. Iron complexes 303−309 housing tetra- or pentadentate nitrogen ligands studied by Fillol and co-workers.

catalytic activity was terminated in a matter of seconds, this complex remained active for a couple of hours. Analyses by GC and MS confirmed that when H2O oxidation was carried out using istopically labeled H 2 18 O, the isotopomeric ratio of the evolved O2 reflected the expected theoretical values for a complete solvent origin. In addition to this, O2 was found to be the only gaseous product formed in the reaction. By the use of DLS and nanoparticle-tracking analysis (NTA), the authors could rule out the in situ formation and involvement of iron nanoparticles, which otherwise could have been a potential source for the catalytic activity.396 To elucidate the mechanism of H2O oxidation for these iron complexes, the reaction involving complex 303 and CeIV as oxidant was chosen as a model and was examined using UV−vis titration. These experiments provided support for a mechanism in which a [FeIV(O)(OH2)(LN4)]2+ species is generated (where LN4 = tetradentate nitrogen ligand), which is suggested to constitute the resting state of the catalytic cycle. This [FeIV(O)(OH2)(LN4)]2+ species could also be detected by MS measurements. The high-valent [FeIV(O)(OH2)(LN4)]2+ species is proposed to undergo oxidation by CAN, to generate a highly reactive FeVO that is rapidly attacked by a solvent H2O molecule, leading to the formation of the crucial O−O bond followed by liberation of O2.396 To shed light on the mechanism of the final part of the reaction, that is, the generation of O2 from the [FeIV(O)(OH2)(LN4)]2+ species, the authors decided to address this in a subsequent study.397 In this study, a family of structural analogues of complex 303 (Figure 81) hosuing different 11948

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

catalytic activity in borate buffer at pH 8.5, using [Ru(bpy)3]3+ as the oxidant.402 Initially, the cis-Fe(mcp)Cl2 complex was evaluated (mcp = N,N′-dimethyl-N,N′-bis(2-pyridylmethyl)-cyclohexane-1,2-diamine), which is the dichloro-analogue of complex 304 that earlier had been used by Costas and co-workers. However, cisFe(mcp)Cl2 proved to be inactive as a WOC. A series of related iron complexes with bpy, tpy, cyclen, or tmc ligands were therefore studied instead, and surprisingly all of these were found to catalyze H2O oxidation to different extent. Even the simple iron precursor Fe(ClO4)3 turned out to be catalytically active. Also, in the light-driven oxidation experiments using [Ru(bpy)3]2+ as photosensitizer, all complexes, including the dichloro-analogue of 304, were found to be active WOCs.402 These results indicated that all complexes included in the study served as precatalysts and decomposed to a common species, which was the true catalyst. By the use of a combination of techniques such as DLS, EDX, and XPS, this catalytic species was identified as α-Fe2O3 nanoparticles in the size range of 15−70 nm. Furthermore, it was shown that the observed difference in activity between the investigated iron complexes could be correlated to the size of the formed Fe2O3 nanoparticles, where the smallest nanoparticles displayed the highest activity. These findings raise serious concerns about the commonly used argument in the field of H2O oxidation that a significant change in activity upon the structural modulation of the ligand framework of a complex is a strong indication of a homogeneous mechanism.402 Although the ligand is not directly involved in the catalytic mechanism, it can have a pronounced effect on the size and structure of the nanoparticles and indirectly affect their activity. On the basis of the results obtained by Costas and Lau, it can be concluded that there might exist a profound pH-dependency, which at lower pH favors a molecular iron-oxo pathway with no evidence of nanoparticle formation, while higher pH’s favor the formation of catalytically active iron nanoparticles.403 The redox potential of the oxidant can also play a key role in determining the dominating mechanism. For instance, CeIV that was used in the first study was not capable of oxidizing FeIII to Fe2O4 at pH 1, which is a prerequisite for the nanoparticle formation. Under these conditions, any amounts of [Fe(OH2)6]3+ released from dissociated complexes will therefore persist in solution unchanged and will be incapable of participating in H2O oxidation. Meyer and co-workers recently reported the synthesis of the mononuclear FeIII complex [Fe(dpaq)(OH2)]2+ (315, Figure 82; dpaq = 2-[bis(pyridine-2-ylmethyl)]amino-N-quinolin-8-ylacetamido). This Fe III complex is comprised of the pentadentate macrocyclic ligand dpaq and mediates electrocatalytic H2O oxidation in propylene carbonate with added H2O as the limiting reagent. Mechanistic experiments suggested a catalytic single-site mechanism that involved the generation of an FeVO species via PCET, which is subsequently attacked

substituents in the ortho and para positions of the pyridine moiety in the ligand was synthesized and evaluated in H2O oxidation.

Figure 81. Structure of iron complexes 303 and 310−314.

This work gave additional insight into the mechanism of the crucial O−O bond formation and also revealed that structural modifications of the ligand backbone had a significant effect on the catalytic performance. Carrying out the reaction with 12.5 μM of complex 312 and 10.000 equiv of CAN at pH 0.7 resulted in the most efficient reaction, affording a TON of 180 over 2 h and a maximum TOF of 0.23 s−1. This clearly emphasized the beneficial effect of having a strongly electronwithdrawing substituent in the para position. Interestingly, introduction of a fluorine or methyl substituent into the ortho position led to a dramatic decrease of the catalytic activity of the corresponding complexes 313 and 314, suggesting that these substituents obstruct coordination to the iron center. Also, the ortho-substituted complexes displayed lower stability with termination of O2 evolution after only 1 h.397 Extensive mechanistic studies performed on complexes such as 303 using UV−vis spectroscopy allowed the authors to propose the mechanism depicted in Scheme 19. In this mechanism, CeIV interacts with the coordinated aqua ligand of the [FeIV(O)(OH2)(LN4)]2+ intermediate as a Lewis acid, releasing a proton in the process. This process results in the formation of a [FeIV(O)(OH)−CeIV]5+ resting state, which could be verified experimentally by UV−vis spectroscopy. The resting state exists in equilibrium with the corresponding [FeV(O)(OH)−CeIII]5+ species, which in turn constitutes the key intermediate undergoing the nucleophilic attack by H2O prior to O−O bond formation. From here, direct one-electron oxidation of the peroxo intermediate [FeIII(OOH)(OH2)]2+ to [FeIV(OOH)(OH2)]+ by CeIV presumably facilitates the release of O2 and regeneration of the catalyst. By comparing the TOFs of complexes 303 and 310−314 with the various rate constants and conducting Hammett plot analysis, it was clearly demonstrated that the nucleophilic attack by H2O constituted the rate-determining step. This finding suggests that the reaction is favored by electron-withdrawing substituents, which increase the electrophilicity of the FeVO center.397 Although it has been proposed that the active species is a FeIV intermediate,398 several studies support that O−O bond formation proceeds from the FeV state.399−401 The complexity of the iron-based WOCs is elegantly illustrated by a recent study of Lau and co-workers. In this study, a series of iron complexes were screened for their

Scheme 19. Proposed Mechanism for Oxidation of H2O by Iron Complexes 303 and 310−314

11949

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

cally driven H2O oxidation (Figure 84).410 It was realized that a readily available copper bipyridine complex (323) could selfassemble from simple starting materials in aqueous solutions, giving rise to an active WOC. Electrochemical measurements of the copper−bipyridine complex revealed that it exhibited a large, pH-dependent, and irreversible wave at 1.30−1.50 V vs NHE. Over multiple scans, bubble formation could be seen on the electrode surface. Scanning to cathodic potentials, after anodic scanning, showed an irreversible peak at −0.30 V, which was assigned to the reduction of O2 and indicated that O2 had been generated via oxidation of H2O. On the basis of the Faradaic efficiency, it could be concluded that 30 equiv of O2 was generated based on the total amount of copper. However, it was assumed that only the copper species in close proximity of the electrode surface were catalytically active. These catalytically active species seemed to be robust and displayed high rates, although a high overpotential of ∼0.75 V was required at high pH (pH 12.5). In situ generation of the catalyst gave results similar to those when the isolated catalyst was added to the solutions, thus indicating that the catalyst selfassembles from inexpensive and abundant materials at appropriate pH. This behavior is a valuable feature and could be used to explain the robustness of the catalytic system. By EPR measurements, it could be shown that in solution the copper complex 323, [(bpy)Cu(μ-OH)]22+, was in equilibrium with the mononuclear species 324 and 325, with 325 dominating under catalytic conditions at high pH. Lowering of the pH resulted in the formation of the antiferromagnetically coupled dimeric species [(bpy)Cu(μ-OH)]22+ (323), while at pH < 8, the bis-aqua complex [(bpy)Cu(OH2)2]2+ (324) appeared to be the most stable species (Figure 85).410 Plotting the catalytic current, ic, against the concentration of copper resulted in linear correlation, thus revealing that the catalysis occurred at a single copper metal center. This was unexpected because copper complexes readily form dimeric copper−peroxo complexes (Cu−O−O−Co). Several observations suggested that the investigated system was homogeneous under the catalytic conditions: (i) no deposited material could

Figure 82. Structure of the FeIII complex [Fe(dpaq)(OH2)]2+ 315.

by a solvent H2O molecule, in analogy to the previously studied ruthenium polypyridyl WOCs.404 Yang and co-workers evaluated another series of iron complexes 316−322 comprised of ligand scaffolds with hydrogen-bonding functionalities situated in the second coordination sphere. These ligands were anticipated to facilitate the H2O oxidation by allowing PCET events (Figure 83).405 Unfortunately, only two out of these seven complexes (316 and 317) were able to reach the crucial FeIVO state and catalyze H2O oxidation, as seen by absorption spectra. Moreover, the choice of CeIV as an oxidant was found to indirectly counteract the purpose of the hydrogen-bonding functionalities by lowering the pH of the reaction, leading to the protonation of these groups. Consequently, the active iron-based WOCs among this series did not display any significant improvement relative to previously reported systems (Table 12).396 However, the concept of preparing ligand frameworks, which are strategically functionalized with moieties that promote PCET, is a promising method for the development of future iron-based WOCs. 4.4. Oxygen Formation in Copper-Based Systems

The investigation of copper-based WOCs is emerging as a novel research area with great potential within the field of H2O oxidation. The group of Mayer recently reported on the first example of a homogeneous copper catalyst for electrochemi-

Figure 83. Iron complexes 316−322 containing functional groups for promoting PCET. 11950

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 12. Summary and References to Available Data for Iron-Based Water Oxidation Catalysts

11951

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 12. continued

11952

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 12. continued

a

Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole of catalyst per s−1. cUsing Ce(NH4)2(NO3)6 (CAN, CeIV) as the chemical oxidant. dUsing NaIO4 as the chemical oxidant. eIt should be noted that a higher TOF of 0.12 s−1 has been obtained, but with a lower TON of 70. fPhotochemical oxidation using [Ru(bpy)3]Cl2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. gUsing [Ru(bpy)3](ClO4)3 as the chemical oxidant. hElectrolysis at a potential of 1.58 V vs NHE in propylene carbonate/H2O (100:8) solutions containing 0.5 M LiClO4. ikcat value obtained for electrolysis at a potential of 1.58 V vs NHE in propylene carbonate/H2O (92:8, v/v) solutions containing 0.5 M LiClO4. bpy = 2,2′-bipyridine. 11953

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

invokes a rate-limiting step that involes the coupling of two copper−oxo moieties to generate a μ-peroxo or a bis(μ-oxo) bridged copper species. However, slight modification of the reaction conditions alters the mechanism to involve a catalytic pathway that is first-order in copper concentration and suggests nucleophilic H2O attack on a high-valent metal−oxo species (cf., single-site ruthenium WOCs).411 In a follow-up study, the group of Meyer reported on the H2O oxidation catalyzed by a copper complex (326, Figure 86)

Figure 84. Copper-based WOC 323 developed by Mayer and coworkers.

be observed, (ii) no discoloration of the electrodes was noted, (iii) no increase in catalytic current or change in wave shape was seen over multiple scans in the electrochemical measurements, (iv) optical spectra during electrolysis resulted in no new bands that might correspond to generated nanoparticles, and (v) rinsed electrodes that had been used in the catalytic experiments and cycled in fresh solutions, in the absence of copper, exhibited no catalytic wave.410 These observations together with the linear dependence of catalytic current on the copper concentration implied that the electrocatalyst was of molecular nature, although it was not possible to rule out the participation of a supported or colloidal material. Recently, Chen and Meyer reported that simple CuII salts were electrochemically active in catalyzing H2O oxidation.411 Addition of Cu II salts to aqueous buffered solutions dramatically enhanced the current, an increase that was assigned to catalytic H2O oxidation. The onset potential for this catalytic current appeared at 1.05 V vs NHE at pH 10.8, corresponding to an overpotential of ∼0.45 V. The catalytic current, ic, was found to be second order in copper concentration and stable for at least 6 h. When exposed to long-term electrolysis, the deposition of a film on the electrode was observed. The UV−vis spectrum of this material had absorption bands at λ = 360 and 530 nm, and the film could be redissolved in Na2CO3 solutions (pH 10.8), thus generating CuII. However, in the absence of any buffer, at pH 10.8, the deposited material was found to be stable. The solid material was found to be active in H2O oxidation and contributed to the overall catalytic current. XPS confirmed that the solid material consisted of copper and oxygen, together with insignificant amounts of carbon. SEM and XPS measurements revealed that the deposited material only appeared at high copper concentrations (>1 mM), suggesting a concentration-dependent formation of a high-valent electroactive film.411 A distinct feature was that when the electrocatalytic experiments were performed in NaHCO3 solutions at pH 6.7, the catalytic current varied linearly with [CuII], thus pointing toward single-site copper catalysis. This is in contrast to the second-order dependence in copper that was observed in Na2CO3 solutions at pH 10.8. This study highlights the dual nature of copper-catalyzed oxidation of H2O, where the secondorder dependence on the copper concentration at high pH

Figure 86. Structure of the copper complex 326 housing the triglycylglycine macrocyclic ligand.

containing a triglycylglycine macrocyclic ligand (H4TGG).412 Isolation of the copper complex 326 demonstrated that it could self-assemble in solution upon mixing of the macrocyclic ligand and Cu(OH)2, at pH 11, which could be monitored by its characteristic d−d transition at λ = 530 nm. Cyclic voltammetry of complex 326 at pH 11 showed a welldefined, reversible wave at 0.58 V vs NHE. This wave was found to be pH-dependent with a slope of −59 mV per pH unit, corresponding to PCET from the CuII center to generate a CuIII species, shown in eq 20.412 [(TGG)Cu II−OH 2]2 − → [(TGG)Cu III−OH]2 −

(20)

The onset potential for electrochemical H2O oxidation occurred at ∼1.10 V vs NHE, corresponding to an overpotential of ∼0.52 V. Catalytic activity was shown to be stable for at least 5 h, after which the activity slightly decreased as a result of the lowering of pH. The evolved O2 was measured and equaled ∼13 catalytic turnovers (based on the initial amount of copper complex 326 in solution), with a Faradaic efficiency of 99%. Multiple lines of evidence supported that the catalyst was homogeneous in nature: (i) no spectroscopic change was observed during long-term electrolysis, (ii) insignificant changes in peak current or wave shapes were seen over multiple scans in the cyclic voltammograms, (iii) an electrode subjected to electrolysis for 2 h with complex 326 (at pH 11) gave no catalytic current when transferred to a fresh solution, in the absence of 326, and (iv) SEM and XPS demonstrated the absence of electrode deposited material after electrolysis for 2 h. An additional control experiment where CuSO4 was added to a 0.25 M phosphate buffer solution resulted in immediate precipitation of Cu3(PO4)2, verifying that the electrocatalysis is

Figure 85. Assumed solution equilibrium of the [(bpy)Cu(μ-OH)]22+ complex 323 at different pH values. 11954

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

thus promoting oxidation of the metal center and influencing catalytic activity. Electrochemical measurements of the copper complex 327 displayed large, irreversible oxidative waves that corresponded to catalytic H2O oxidation. This catalytic process was found to occur at a low overpotential of ∼510−560 mV at pH 12−14, which is significantly lower than the previously reported Cubpy WOC 323. During electrolysis, a deposit could be noticed; however, it was established that the generated film consisted of oligomers or polymers of copper complex 327 and not of copper oxides. The authors successfully obtained single crystals for X-ray diffraction of the copper complex 327 in solution. The acquired crystals were found to be of the same phase as single crystals of a coordination polymer grown from a mixed methanol/water solution as indicated by the identical PXRD patterns. In the solid state, the copper center and the ligand form a 1D coordination polymer with the general formula [Cu(HL)(μ2-OH)]n, where the copper adopts a distorted square planar coordination environment together with a monodeprotonated ligand and two bridging hydroxo groups.413 When comparing with the structurally related copper complex 328, bearing the isomeric 4,4′-dihydroxy-2,2′-bipyridine, it could be shown that this complex exhibited a higher overpotential for H2O oxidation than complex 327, thus indicating that the low overpotential associated with complex 327 was not merely due to the electron-donating nature of the substituents. For the copper complex 327, electrochemistry revealed two peaks, where the first one was assigned to the CuIII/CuII redox couple. Interestingly, by comparing with the analogous zinc system, the second peak could be attributed to a ligand centered oxidation, demonstrating that the ligand participates in the catalytic cycle and has a vital role in the generation of a low overpotential for H2O oxidation. Calculations verified this by demonstrating that the spin density associated with the second oxidation is located mainly on the ligand, highlighting that the generated formal CuIV species is better described as a CuIII species with a ligand radical. The redox-noninnocent character of the 6,6′-dihydroxy2,2′-bipyridine ligand in copper complex 327 thus translates into stabilization of the high-valent intermediates and lowering of the overpotential for H2O oxidation.413 Collectively, this work is yet another demonstration that incorporation of redoxaccessible ligand frameworks into metal complexes could be a viable strategy for constructing efficient WOCs (Table 13).

mediated by the copper complex and not simply by uncomplexed CuII present in solution.412 The catalytic current for H2O oxidation exhibited a linear dependence on the concentration of the copper catalyst, indicating single-site catalysis. On the basis of these observations, a mechanism for H2O oxidation by copper complex 326 was outlined (Scheme 20). The proposed Scheme 20. Proposed Mechanism for H2O Oxidation by Copper Complex 326, [(TGG)CuII−OH2]2−

mechanism begins with proton-coupled oxidation of the CuII−aqua complex to generate the corresponding CuIII− hydroxo complex. A further oxidation leads to the high-valent [(TGG)CuIV(O)]2− species (or the oxyl species [(TGG)CuIII(O•)]2−), which is analogous to the highly reactive RuV− oxo (or RuIV−oxyl) intermediate. This high-valent copper species undergoes nucleophilic attack by a solvent H2O molecule to generate a CuII hydroperoxo intermediate ([(TGG)CuII−OOH]3−). Subsequent oxidations of this hydroperoxo intermediate result in [(TGG)CuIII−OO•]2−, from which O2 is liberated, thereby closing the catalytic cycle by regenerating the starting complex, [(TGG)CuII−OH2]2−.412 This mechanism is similar to the well-established catalytic mechanism for single-site ruthenium WOCs. The group of Lin recently developed a “biomimetic” copperbased WOC with a low overpotential for H2O oxidation.413 The authors wanted to imitate the natural PCET processes involving Tyrz in PS II and envisioned that this would be possible by the use of a ligand containing suitable pendant groups. A copper-based WOC (327, Figure 87) housing the ligand 6,6′-dihydroxy-2,2′-bipyridine was designed as it was thought that this ligand would provide a suitable redoxnoninnocent environment that could facilitate PCET events,

4.5. Cobalt-Based Catalysts Capable of Oxidizing Water

Although simple cobalt salts have been known to catalyze the oxidation of H2O since the 1980s,414−418 the interest in molecular cobalt-based WOCs was negligible until a couple of years ago when Nocera and co-workers showed that an in situ generated cobalt-based WOC (Co−Pi) was capable of operating under neutral conditions (pH ∼7) with a low overpotential (vide infra). Since then, the development of molecular cobalt-based WOCs has progressed rapidly, in conformity with other firstrow transition metals, such as manganese and iron. Unfortunately, the research has been hampered by the intrinsic instability of the CoII metal center. This has hindered the preparation of structures analogous to those of ruthenium- and iridium-based complexes, and required particular care in the design of the ligand frameworks. In 2011, the group of Berlinguette reported on the first example of a well-defined and stable coordination compound

Figure 87. Molecular structure of copper complex 327 housing the redox-noninnocent 6,6′-dihydroxy-2,2′-bipyridine ligand. 11955

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 13. Summary and References to Available Data for Copper-Based Water Oxidation Catalysts

a

Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole of catalyst per s−1. cElectrolysis at a potential of 1.35 V vs NHE in aqueous solutions at pH 12.5. dElectrolysis at a potential of 1.30 V vs NHE in aqueous phosphate buffer solutions (0.25 M, pH 11). eElectrolysis at a potential of 1.13 V vs NHE in aqueous 0.1 M NaOH/NaOAc solutions at pH 12.4. The turnover number (TON) is calculated on the basis of the total amount of Cu in the solution. fEstimated from the slope of the plot of icat/id versus ν−1/2.

originated from in situ generated Co−phosphate films (Co− Pi).419 Related to this work, the group of Chang reported on a structurally related analogue of the [Co(Py5)(OH 2 )] 2+ complex capable of promoting electrocatalytic production of H2.420 Concerns regarding the homogeneity of 329 were raised when subsequent studies on a cobalt-based POM catalyst revealed that low concentrations of homogeneous CoII could spawn CoOx nanoparticles that display a similar catalytic wave at 1.40 V vs NHE.421 Instigated by this finding, Berlinguette and co-workers revisited the catalytic behavior of 329 to ascertain its claimed activity. Extensive mechanistic work by electrochemistry made it possible to distinguish between the electrochemical profiles of the homogeneous and heterogeneous species. For complex 329, support was found for a molecular pathway for the catalytic activity, thus securing its position as a homogeneous WOC. However, the participation of CoOx nanoparticles in H2O oxidation could not be ruled out entirely. This study clearly demonstrated the challenges and potential pitfalls that exist when evaluating molecular cobaltbased systems, where as little as 1−2% of free CoII from dissociated complexes can give rise to an observable catalytic effect that mistakenly can be ascribed to the parent molecular complexes.422 Following this, Nocera and co-workers reported on the use of cobalt hangman corrole complexes as highly active oxygen reduction catalysts.423 Interestingly, this family of complexes was later found to be capable of catalyzing the reverse reaction, that is, H2O oxidation, under electrochemical conditions as demonstrated with the complex bearing meso-pentafluorophenyl substituents (330) and its β-octafluorinated congener (331) (Figure 89, left).424 The rather unique structure of these complexes is believed to play an important role in their observed reactivity as it allows H2O molecules to be positioned in close proximity to the cobolt−oxo group, which promotes the crucial O−O bond forming reaction (Figure 89, right).425

based on cobalt, using the oxidatively stable pentadentate ligand 2,6-(bis(bis-2-pyridyl)methoxy-methane)-pyridine (Py5). The resulting complex, [Co(Py5)(OH2)]2+ (329, Figure 88), which

Figure 88. Molecular structure of the [Co(Py5)(OH2)]2+ complex 329.

was capable of mediating electrochemical H2O oxidation, displayed a reversible redox wave at 0.75 V vs NHE (pH 2.2), corresponding to the [CoIII−OH]2+/[CoII−OH2]2+ redox couple. This redox couple exhibited a pH dependence of ca. −59 mV per pH unit up to pH 11.7, revealing that the redox event is connected to the removal of a proton. This observation showed that complex 329 operates through a PCET process, which grants access to high oxidation states at lower redox potentials. A pH-independent redox wave was also identified at 1.40 vs NHE between pH 7.6−10.3, corresponding to the [CoIV−OH]3+/[CoIII−OH]2+ couple. This redox process was succeeded by a significant increase in current, characteristic of a catalytic process, and this was assigned to catalytic H2O oxidation. At pH > 10.3, this process becomes pH-dependent (−59 mV per pH unit), which is consistent with a shift toward a PCET process involving the [CoIVO]2+/[CoIII−OH]2+ couple. On the basis of these observations, O−O bond formation was proposed to result from a nucleophilic attack by H2O/OH− on the corresponding CoIV-hydroxide/oxo intermediates. Control experiments performed in the original study excluded that the catalytic current at pH 7.6−10.3 11956

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 90. Molecular structure of cobalt corrole complex [Co(tpfc)(py)2] (332).

Figure 89. Structures of the two meso-pentafluorophenyl-substituted cobalt hangman corrole complexes 330 and 331 (left) and the ability of the carboxylic to preorganize H2O inside the hangman cleft, which is beneficial for the O−O bond formation (right).

sufficient oxidizing power to function as a potential WOC, which urged the authors to further study its electrochemical properties in aqueous solutions. In these experiments, complex 332 was deposited on an ITO-coated glass electrode to give a catalyst film of 2.5 nmol cm−3, and the electrode was subsequently immersed into an aqueous phosphate buffer solution (0.1 M, pH 7.0). A substantial rise of the current began at ∼1.36 V vs NHE, which was attributed to catalytic H2O oxidation and occurred at an overpotential of ∼0.54 V. At 1.61 V, the TOF for the H2O oxidation process was determined to 0.20 s−1. Moreover, the onset potential for H2O oxidation for complex 332 was found to be pH-dependent with a slope of −67 mV per pH unit between pH 6.0−14.0, indicating the involvement of a PCET event in the catalytic mechanism. In addition to the observed pH dependency, the rate of H2O oxidation for complex 332 proved to be enhanced by increasing phosphate concentration, which suggests that a concerted APT pathway played a critical role in the O−O bond formation process. The phosphate can thus act as a base to catalyze the nucleophilic attack by the H2O molecule by abstracting one proton during the O−O bond formation step. Theoretical studies indicated that this base-assisted O−O bond formation process constituted the rate-determining step of the catalytic cycle. The authors also dedicated significant efforts to verify that complex 332 operated under homogeneous conditions, that is, that complex 332 was responsible for the observed electrocatalytic O2 evolution activity, and that it was stable under the catalytic conditions. Multiple lines of support for this were presented, involving: (i) no decay of the catalytic current was observed during 250 min of electrolysis, (ii) electrochemical measurements of an ITO electrode coated with complex 332 before and after electrolysis showed to be identical, (iii) the electrode displayed no electrocatalytic activity after extensive washing with H2O and CH2Cl2, (iv) scanning electron microscopy (SEM) and EDX analysis of the washed ITO electrode after electrolysis showed features similar to those of bare ITO surface, (v) analysis of recovered catalyst by UV−vis spectroscopy and mass spectrometry identified the presence of complex 332, and (vi) a bare ITO electrode immersed into a 0.1 M aqueous phosphate buffer solution containing Co(OAc)2 did not give rise to any catalytic current.427 Recently, the group of Groves reported on a series of singlesite cobalt porphyrin complexes functioning as electrochemical WOCs under neutral conditions (333−335, Figure 91).428 Among these, CoIII-5,10,15,20-tetrakis(1,3-dimethylimidazolium-2-yl)porphyrin ([CoIII-TDMImP], 333) with a highly electron-deficient ligand structure proved to be the most

Complex 331 was found to be the most efficient catalyst of the two, displaying a catalytic onset at 1.45 V vs NHE under neutral pH with a TOF of 0.8 s−1. This was ascribed to the increased oxidizing power of 331 provided by the additional fluoro substituents. Moreover, complex 331 showed high stability throughout the reaction with no detectable formation of CO2, arising from ligand degradation. Unfortunately, the authors were unable to conclusively establish the identity of the involved homogeneous species in the catalytic reaction, due to uncertainties in the exact electronic structure of the starting complex 331.424 However, this issue was addressed in a subsequent computational study by the group of Lai, which helped to shed light on the catalytic mechanism and answer why cobalt is the ideal metal for hangman complexes in favor of iron, manganese, iridium, or ruthenium.426 The calculations pointed out two theoretically possible precatalytic oxo-intermediates that could be subjected to a nucleophilic attack by H2O or OH−. The first intermediate was identified as a [(corrole)CoIVO]2− species possessing two unpaired metal−oxo centered electrons with no radicaloid character on the corrole ring, while the other species was suggested to be a CoV-species with a corrole ring cation radicaloid character, best described by a [(corrole•+)CoIVO]− electronic structure. Of these two, the latter CoV-species was shown to have a transition state for the O−O bond formation of significantly lower energy. Therefore, it was concluded to be the most likely candidate to undergo the nucleophilic attack by H2O, which was in agreement with the first suggestion provided by the group of Nocera in their original work.424 The main reason as to why cobalt is the most suitable metal for corrole hangman complexes is its relative ease of undergoing two-electron reduction and the high OH• affinity of the cobalt−oxo species. In this regard, cobalt was shown to be superior to the other studied metals, by enabling highly favorable thermodynamics for the overall H2O oxidation event.426 Another interesting and bifunctional cobalt corrole complex [Co(tpfc)(pyr) 2] (332, Figure 90; tpfc = 5,10,15-tris(pentafluorophenyl)corrole; pyr = pyridine) for both electrochemical O2 evolution and H2 production was very recently reported by Lei et al.427 Electrochemical measurements of cobalt corrole complex 332 in an acetontrile solution at a glassy carbon electrode revealed two reversible waves at E1/2 = 1.01 and 1.57 V vs NHE. From theoretical studies, the authors could ascribe the first oxidation to a corrole-centered event. The electrochemical results confirmed that complex 332 possessed 11957

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 92. Molecular structures of the cobalt porphyrin complexes 336−338. Figure 91. Structures of cobalt porphyrin complexes 333−335.

components were omitted, confirming that all are necessary for the catalysis to occur. The involvement of active Co−oxides in the catalytic reaction was ruled out by DLS experiments, which did not show any signs of nanoparticle formation. Moreover, to establish that the complex did not decompose into other soluble catalytically active cobalt species, analysis of the decomposition products was carried out by MS. The major decomposition product was shown to result from oxidative cleavage of the porphyrin ring at the meso-positions. However, this species was still capable of holding onto the cobalt center and preventing it from leaching into solution, which ruled out that simpler cobalt species were the source of the observed catalytic activity. Kinetic measurements, where the initial rate was plotted against catalyst concentration, revealed a second-order dependence, suggesting either a mechanism involving the coupling of two CoIV-oxyl radicals or a mechanism proceeding via the disproportionation of two CoIV-species to generate a CoV- and a CoIII-species. The CoIII-species generated in the latter mechanism could then form the corresponding CoIV−oxyl complex, which can undergo a similar radical coupling event as in the former mechanism. Unfortunately, attempts to distinguish between these two plausible mechanisms by either EPR experiments or DFT calculations were unsuccessful.429 Joint efforts by the groups of Llobet and Stahl led to the creation of two binuclear CoIIICoIII-(μ-peroxo) complexes comprised of the bpp ligand (339 and 340, Figure 93).430

effective catalyst, with a O2 formation rate of 170 nmol cm−2 min−1 and a Faradaic efficiency close to 90%. The electrochemical experiments with complex 333 were conducted in a 0.2 M aqueous sodium phosphate solution at pH 7 and gave rise to a strong catalytic current with an onset potential of ∼1.41 V vs NHE and sustained O2 evolution over several hours without any observable loss of catalytic current. Careful mechanistic investigation by electrochemistry revealed that the diamagnetic CoIII−TDMImP species, formed after a one-electron oxidation of the corresponding CoII-species at an applied potential of 300 mV, was the active form of the catalyst. This catalytically active CoIII-species was found to undergo a second anodic oxidation, beyond that of the ligandbased oxidation of the porphyrin ring, which appeared to coincide with H2O oxidation. This second wave was suggested to correspond to a one-electron oxidation of the CoIII-species with concomitant loss of one proton to the buffer anion to give a CoIV−O porphyrin radical cation. Consequently, this can be considered as a formal CoV-species that has acquired two oxidation equivalents above CoIII and constitutes the catalytically competent species for H2O oxidation catalysis. Furthermore, the authors found evidence that O−O bond formation occurred at a single Co-site and constituted the rate-limiting chemical step of the catalytic cycle. This was indicated by a scan-rate dependency of the catalytic current and a linear dependence of the catalytic current on the catalyst concentration. Several lines of evidence were presented in this study to support the homogeneous nature of this catalytic system. For example, it was shown that the resting CoIII complex was stable for hours and that the catalytic activity of the system was not hampered by either cobalt ion sequestration with EDTA or a three-phase test with Chelex resin. Taking these findings into account, together with the indications that O−O bond formation is rate-limiting and the linear dependence of catalytic current on catalyst concentration, it seems highly probable that the catalysis is occurring at a single catalytic molecular site and not by free CoII ions or a Co film.428 A similar series of complexes that are capable of catalyzing light-driven H2O oxidation is the cobalt−porphyrin complexes 336−338 (Figure 92), which were recently reported by the group of Sakai.429 The three complexes are similar in structure, and all contain a CoIII center stabilized by a porphyrin ligand that varies only in the aryl groups (N-methylpyridine, benzoic acid, benzenesulfonic acid) located at the four meso-positions. Catalytic H2O oxidation was performed at pH 11, in phosphate buffer, with [Ru(bpy)3]2+ as photosensitizer and Na2S2O8 as sacrificial electron acceptor, which resulted in TOFs ranging between 0.118 and 0.170 s−1 and TONs between 89 and 122 (measured after 30 min) for the three complexes. No O2 evolution was observed upon irradiation when any of the four

Figure 93. Dinuclear CoIII,III peroxo complexes 339 and 340.

Inspiration for this WOC structure originated from a previous computational study, which demonstrated that dinuclear CoIIICoIII-bridging peroxo intermediates are involved in the H2O oxidation mechanism catalyzed by heterogeneous Co− oxides.431 At pH 2.1 (in 0.1 M phosphate buffer), both complexes were able to mediate electrochemical H2O oxidation and proved to be stable over a period of several hours without degrading to Co−oxide nanoparticles. Moreover, electrochemical evidence implied that 339 and 340 acted through similar mechanisms, where two single-electron oxidations were required to generate the CoIVCoIV state before O2 evolution could occur. For complex 339, a first quasi-reversible wave was 11958

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

found at 1.56 V vs NHE, corresponding to the CoIVCoIII/ CoIIICoIII redox couple, which was followed by a catalytic wave at 1.91 V vs NHE. Electrochemical experiments revealed a second oxidation wave at 1.82 V belonging to the CoIVCoIV/ CoIVCoIII redox couple, thus confirming that oxidation of H2O is triggered when the CoIVCoIV state is reached. Complex 340 exhibited a shift of the two redox waves toward lower redox potentials, 1.35 and 1.84 V vs NHE, respectively. The involvement of CoOx nanoparticles or films in the catalytic mechanism was ruled out on the basis of the reaction pH, which was not sufficiently alkaline to allow for their formation. Additional support for the molecular pathway was obtained from the fact that no enhancement of the catalytic current could be observed at prolonged reaction times, which would otherwise have indicated decomposition of the complex into more active heterogeneous cobalt species.430 To the best of our knowledge, the first molecular cobaltbased WOC reported to mediate H2O oxidation without the need of electrochemical conditions was the trans-[Co(qpy)(OH2)2]2+ complex (qpy = 2,2′:6′,2″:6″,2‴-quaterpyridine) prepared by the group of Lau (341, Figure 94). Remarkably,

Figure 95. Structures of the single-site cobalt salophen complex 342 and dyad 343.

electrochemical and photochemical H2O oxidation.435 In the solid state, the cobalt center of complex 342 was found to adopt a square planar geometry, which changes to a square pyramidal geometry in aqueous solution, as a result of an extension of the cobalt coordination sphere by an additional aqua ligand at the apical site. Spectrophotometric titration experiments determined the pKa of this H2O molecule to 6.4, indicating that it is sufficiently acidic to be deprotonated into a hydroxyl moiety at neutral pH. In the electrocatalytic studies, significant differences could be observed between the voltammogram of 342 and that belonging to ligand-free cobalt aqua complex, indicating that the investigated catalyst can function as a true molecular WOC. Furthermore, the electrocatalytic experiments revealed that complex 342 exhibited a very low operating overpotential, η = 0.3 V at 0.7 mA cm−2 current density, which was markedly better than previously reported cobalt WOCs that typically display overpotentials in the range 0.5−0.6 V.419,422,424 In photocatalytic experiments performed in buffered aqueous solutions (phosphate buffer, pH 7), employing [Ru(bpy)3]2+ and persulfate, a considerable amount of O2 could be detected over a time frame of ca. 2 h. Under the best conditions, complex 342 was found to give a TON of 17 and a quantum yield of 15.8%. The stability of 342 under the photocatalytic conditions was assessed by combined use of UV−vis, DLS, and EPR, which suggested that no free CoII ions or heterogeneous CoOx were formed during the course of the H2O oxidation catalysis.435 The cobalt salophen complex 342 has also been attached to a [Ru(bpy)3]2+-type photosensitizer unit to generate dyad 343. Unfortunately, dyad 343 was found to suffer from leaching of cobalt under the studied photochemical experiments, which greatly reduced its practical utility.436 Subsequently, the group of Ding and Ma reported on a highly efficient system for photocatalytic H2O oxidation involving a related single-site cobalt salen complex 344 (Figure 96), which was shown to achieve TONs and TOFs of up to 854 and 6.4 s−1, respectively.437 However, the authors could prove by the use of DLS and SEM that complex 344 only served as a precursor for in situ generated catalytically active cobalt nanoparticles under the employed photocatalytic conditions ([Ru(bpy)3]2+/persulfate in borate buffer at pH 9). This study suggests that also salen and salophen types of cobalt complexes

Figure 94. Structure of the single-site molecular cobalt complex trans[Co(qpy)(OH2)2]2+ 341, which is capable of catalyzing both H2O oxidation and proton reduction.

complex 341 proved to be compatible with the mild photosensitizer [Ru(bpy)3]2+, allowing H2O oxidation to be driven by light when using persulfate as electron acceptor in borate buffer (pH 8). Measurements of O2 evolution recorded a TON of 335, which was obtained after irradiating the reaction (λ = 457 nm) for ca. 1.5 h. Moreover, to allow the direct comparison with a previously reported WOC, the cobalt-based POM [Co4(OH2)2(PW9O34)2]10−, H2O oxidation with 341 was also carried out using stoichiometric amounts of [Ru(bpy)3]3+.432 The observed TOF of 4 s−1 for 341 is comparable to that of [Co4(OH2)2(PW9O34)2]10−,433 demonstrating that molecular cobalt complexes can reach the catalytic efficiencies of carbon-free WOC systems. The commonly used collection of control experiments, involving 18O-labeling studies, MS, DLS, and reactions with simpler cobalt precursors, confirmed that a definite and unique molecular mechanism was operating for 341 where both oxygen atoms in the evolved O2 originated from solvent H2O.432 However, the authors did not provide any detailed information regarding the exact structure of the catalytic intermediates, although they proposed the involvement of metal−oxo species in accordance with earlier literature reports.109−115 Complex 341 also showed interesting versatility by catalyzing H2O reduction when using an iridium-based photosensitizer as electron donor.434 Irradiation of complex 341 for 20 h at λ > 420 nm resulted in a TON of 1020 of the evolved H2. Recently, Pizzolato et al. demonstrated that even a single-site cobalt salophen complex (342, Figure 95) could catalyze both

Figure 96. Molecular structure of the single-site cobalt salen complex 344. 11959

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 14. Summary and References to Available Data for Cobalt-Based Water Oxidation Catalysts

11960

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 14. continued

a

Turnover numbers (TONs) are defined as moles of produced product per mole of catalyst, nO2/ncat. bTurnover frequencies (TOFs) are defined as moles of produced product per mole of catalyst per s−1. cElectrolysis at a potential of 1.59 V vs NHE in aqueous solutions. dElectrolysis at a potential 11961

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Table 14. continued of 1.60 V vs NHE in aqueous solutions at pH 7. eThe catalyst was coated onto an ITO electrode and subjected to electrolysis at a potential of 1.60 V vs NHE in aqueous phosphate buffer solutions at pH 7. fElectrolysis at a potential of 1.50 V vs NHE in aqueous phosphate buffer solutions at pH 7. g Photochemical oxidation using [Ru(bpy)3](NO3)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. hUsing [Ru(bpy)3](ClO4)3 as the chemical oxidant. iPhotochemical oxidation using [Ru(bpy)3]Cl2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor. j Photochemical oxidation using [Ru(bpy)3](ClO4)2 as photosensitizer and Na2S2O8 as the sacrificial electron acceptor.

This dinuclear POM was reported in 2004 by Shannon and co-workers.439 They found that POM 345 was capable of electrochemically generating O2 when subjected to a potential of 1.01 V vs NHE (at pH 8.0). Cyclic voltammetry of the POM showed three waves at −0.10, +0.28, and +0.76 V vs NHE, which was assigned to the formal oxidations of RuII → RuIII → RuIV → RuV. Interestingly, the authors also developed a POM containing a single ruthenium center ([RuIII(H2O)PW9O34]4−) and showed that it did not catalyze the oxidation of H2O. This result highlighted the necessity of having at least two ruthenium centers in close proximity to allow the generation of O2.439 A major breakthrough was made in 2008 when the groups of Hill440 and Bonchio441 independently reported that the POM [{Ru4(μ-O)4(μ-OH)2(H2O)4}(γ-SiW10O36)2]10− 346, comprising a tetraruthenate core, was oxidatively and hydrolytically stable enough to carry out the oxidation of H2O. The ruthenium centers are bridged by oxo ligands with an overall D2d symmetry, where the tetraruthenate core exists in an adamantane-like rearrangement (Figure 97). Several lines of

are susceptible to degradation into catalytically active materials, providing yet another example of the importance of carefully establishing the homogeneity/heterogeneity of the catalytic protocol under the investigated reaction conditions (Table 14).

5. SYNTHETIC METAL−OXO MIMICS OF THE OXYGEN EVOLVING COMPLEX 5.1. Polyoxometalates − The Molecular, Inorganic Alternative

A crucial factor for the development of a viable artificial device for the splitting of H2O is a WOC that is robust and manages to withstand the highly oxidizing conditions required for the oxidation of H2O. Because a majority of the developed WOCs contain organic ligand frameworks, they suffer from the drawback of being degraded after prolonged reaction times. In this regard, polyoxometalates (POMs) containing noncarbonaceous (inorganic) ligands constitute an excellent alternative to WOCs comprised of organic ligands. The versatile nature of POMs, in terms of structure, size, redox chemistry, photochemistry, and charge distribution, makes them one of the most rapidly developing areas of chemistry today.438 POMs are polyanionic clusters consisting of early transition metals and oxygen where the majority is based on the transition metals V, W, Nb, Mo, or Ta. These are usually found in their highest oxidation state, that is, d0 electronic configuration making the POMs oxidatively inert and excellent catalysts. The synthesis of POMs is relatively straightforward and can usually be performed in one pot, where several parameters can be modulated for the fine-tuning of the synthetic protocol: (i) concentration and type of metal oxide anion (MoO42−, VO43−, WO42−), (ii) the pH and the type of acid that is used, (iii) introduction of heteroatom(s), (iv) concentration and type of electrolyte, (v) addition of other ligands, (vi) type of reducing agent that is employed, (vii) temperature, and (viii) solvent.438 Another advantage of using POMs in H2O oxidation catalysis is that they exhibit stability similar to heterogeneous supports, while at the same time being molecular in nature and highly soluble. These favorable properties allow a combination of the advantages associated with heterogeneous and homogeneous catalysts, such as high stability and being amenable to mechanistic and computational studies. Although the POMs tend to be oxidatively inert, they can be hydrolytically sensitive. This issue can be solved by introducing one or more heteroatoms, giving so-called “heteropolyanions”, which can be stable over a broader pH range. Other important features that affect the properties of the POMs, and which are significant for catalysis, are the ionic strength and the type of countercation. These factors influence the reactivity and the redox potentials of the corresponding POMs and are essential for carrying out multielectron catalysis.438 5.1.1. Ruthenium-Based Polyoxometalates. The first POM able to catalyze the oxidation of H2O was the dinuclear ruthenium-based Na14[RuIII2Zn2(H2O)2(ZnW9O34)2] 345.

Figure 97. Structure of the tetraruthenate POM [{Ru4(μ-O)4(μOH)2(H2O)4}(γ-SiW10O36)2]10− 346. Reprinted with permission from ref 441. Copyright 2008 American Chemical Society.

evidence indicated that the ruthenium centers were all in the +IV oxidation state: (i) the magnetic properties, (ii) bond valence sum calculations, (iii) the fact that POM 346 was EPR silent, (iv) elemental analysis calculations of the number of counterions, and finally (v) the potentials of the cyclic voltammograms verified that the core was a [RuIV4]. Interestingly, the two groups carried out the catalytic experiments under different conditions. Hill and co-workers employed [Ru(bpy)3]3+ as the chemical oxidant under neutral conditions (pH 7),440 while the group of Bonchio instead used CeIV as the oxidant under acidic conditions (pH 0.6).441 Both groups were particularly careful to demonstrate that the tetraruthenate POM is stable under the catalytic conditions and that the POM structure with its ruthenium core was 11962

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

POM 346, indicating an efficient quenching by the photogenerated [Ru(bpy)3]3+ by electron transfer from the catalyst 346. Under the catalytic turnover conditions, the overall quantum yield (Φ = moles of generated O2/moles of absorbed photons) was measured to be as high as ∼0.045. It was also found that the efficiency by which [Ru(bpy)3]3+ was generated and its reaction with POM 346 were the limiting factors of the applied photocatalytic system. It was suggested that higher quantum yields could be obtained if better systems for [Ru(bpy)3]3+ generation would be employed.442 This hypothesis was verified by Bonchio and co-workers, who utilized the dendrimeric [Ru{(μ-dpp)Ru(bpy)2}3]8+ (347; dpp = 2,3-bis(2′-pyridyl)pyrazine).443 This had a broader absorption in the visible region than standard [Ru(bpy)3]2+, allowing it to absorb a greater portion of the light in the visible region (Figure 99).

maintained during catalysis. This was shown by (i) the reversibility of the acid−base titrations, (ii) the reproducibility and reversibility of the electrochemical measurements, (iii) the superimposable resonance Raman experiments, (iv) UV−vis spectroscopy, and (v) mass spectrometry measurements.440,441 Because the group of Hill used the mild one-electron oxidant [Ru(bpy)3]3+, the accumulated [Ru(bpy)3]2+ could be spectrophotometrically monitored, and the evolved O2 was determined by gas chromatography. In the presence of POM 346, the reaction time for the conversion of [Ru(bpy)3]3+ to [Ru(bpy)3]2+ was significantly increased. Additional experiments were carried out to verify that catalysis was indeed accomplished by the tetraruthenate POM and not by any possible decomposition products. Analyzing the activity of RuCl3, at an equimolar ruthenium concentration, resulted in longer reaction times than catalyst 346, affirming that catalysis is mediated by the POM 346 and that decomposition products did not contribute to the observed effect. To confirm that the oxygen atoms in the generated O2 originate from H2O, experiments were performed with isotopically labeled H2O (H 2 18 O). These experiments showed that the ratio 18,18 O2/18,16O2 was in close agreement to the theoretical ratio, thus confirming that both oxygen atoms in the O2 were derived from H2O.440 Bonchio and co-workers used CeIV as oxidant, which resulted in a high TON of ∼500 and a moderate TOF of ∼0.01 s−1. A linear dependence on the initial rate of O2 evolution on the catalyst concentration affirmed that it occurred within the tetraruthenium core. After depletion of the oxidant, the addition of more CeIV triggered the evolution of more O2, thus highlighting the robustness of the ruthenium POM.441 In addition to being able to mediate H2O oxidation driven by a chemical oxidant, the ruthenium POM also displayed remarkable catalytic performance in light-induced oxidation of H2O. Hill and co-workers employed a photocatalytic system consisting of a [Ru(bpy)3]2+-type photosensitizer and persulfate as the sacrificial electron acceptor (Figure 98).442

Figure 99. Structure of the tetranuclear ruthenium photosensitizer [Ru{(μ-dpp)Ru(bpy)2}3]8+ 347.

By using a three-component system consisting of catalyst 346, photosensitizer 347, and persulfate in a phosphate buffer solution, a quantum yield of 0.30 was obtained. This is impressive, considering that the theoretical value is 0.5 (recall eqs 8−11), and means that 60% of the photons were utilized in the interconversion of H2O to O2. Comparing the electronic absorption spectra of the supramolecular photosensitizer before and after catalysis revealed that only ∼5% of the photosensitizer had decomposed during the experiments. These numbers, together with the fact that a larger fraction of the visible light (50% higher in the absence of 11963

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Scheme 21. Summary of the Processes Taking Place during Photocatalytic Experiments with POM 346, [Ru(bpy)3]2+ as Photosensitizer, and Persulfate as Sacrificial Electron Acceptor

observed, which signaled that [Ru(bpy)3]3+ was generated. The bleaching effect remained constant over the entire time scale of the photochemical experiment, implying that charge recombination between the photo-oxidized photosensitizer and the electron acceptor did not take place. However, in the presence of POM 346, there was a distinct recovery of the bleach, with a rate depending on the concentration of POM 346. This result clearly demonstrated that fast electron transfer occurred from the ruthenium POM to the photo-oxidized photosensitizer to generate the one-electron oxidized POM 346. The bimolecular rate constant for the electron transfer between POM 346 and the photosensitizer was estimated to be as high as 2.1 × 109 M−1 s−1, which is only within 1 order of magnitude from a diffusion controlled rate.444 Evaluation of the electron-transfer processes was also performed under heterogeneous conditions, in which [Ru(bpy)2(dpb)]2+ (dpb = 4,4′-diphosphonic-2,2′-bipyridine acid) was employed as the photosensitizer and adsorbed onto nanocrystalline TiO2. This assembly was then subjected to laser flash photolysis (λex = 532 nm). Ultrafast electron injection into the semiconductor occurred as monitored by bleaching of the MLCT absorption, which was subsequently restored by charge recombination. The photosensitizer loaded TiO2 was then charged with different concentrations of the POM 346. The uptake of 346 was monitored spectrophotometrically and was thought to proceed merely by electrostatic interaction of the negatively charged POM with the positively charged photosensitizer. Exposing the catalyst-doped samples to laser flash photolysis resulted in faster charge recombination of the photo-oxidized photosensitizer, thus providing evidence that the POM 346 was able of reducing the photogenerated [Ru(bpy)3]3+ species. The authors ascribed the fast electron transfer to the photogenerated [Ru(bpy)3]3+ to (i) the highly negatively charged POM, which would facilitate close contact with the positively charged photosensitizer, and (ii) the low reorganization energy of the POM network.444 Further studies focused on the ion pairing between the [Ru(bpy)3]2+ photosensitizer and the tetraruthenium POM 346, and on the kinetics of the hole transfer from the oxidized photosensitizer to 346.445 Fluorescence and conductometric titrations of [Ru(bpy)3]2+ with 346 were performed to monitor the ion pairing in aqueous solution. The results indicated formation of adducts with a 346/[Ru(bpy)3]2+ ratio of 1:4. This aggregation caused complications in the laser flash

photolysis experiments due to extremely fast (2 ps) electron transfer to POM 346 via oxidative quenching, which was followed by charge recombination (15−150 ps). The tetrametallic ruthenium core in 346 had several accessible reduced states (vide infra), making the oxidative quenching of the excited state of the photosensitizer thermodynamically favorable. Scheme 21 summarizes the different processes occurring during the photocatalytic experiments. After some optimization, it was discovered that catalyst 346 could partake in as many as 45 turnovers of [Ru(bpy)3]3+ reduction in 40 ms (at [346] = 0.5 × 10−6 M). These numbers are remarkable because if one would assume that one molecule of O2 is generated after four electron transfers, it would correspond to an extremely high TOF of 280 s−1 (TOF = 1/ 4Δ[[Ru(bpy)3]3+]/(Δt·[346])).445 The two groups of Hill and Bonchio subsequently addressed this issue to gain fundamental mechanistic insight into how ruthenium POM 346 oxidizes H2O to O2.446−448 Bonchio and co-workers performed a mechanistic study by combining electrochemical, UV−vis, EPR, and resonance Raman measurements to identify the intermediates during H2O oxidation. The electrochemical properties of POM 346 (with lithium as the countercation) recorded in an aqueous solution (pH 0.6) displayed three reversible waves and one anodic wave in a narrow potential range (0.60 < E < 1.40 V vs NHE). The three reversible redox couples were observed at E1/2 = +0.73, +0.86, and +1.11 V vs NHE, while the last anodic peak was observed at +1.31 V. These results suggest a stepwise oxidation of the Ru 4 IV,IV,IV,IV to ultimately yield a high-valent Ru 4 V,V,V,V intermediate. The four-step oxidation could also be monitored spectroscopically by plotting the absorption at λ = 443 nm as a function of the titration with SnCl2 (reductant) or CeIV (oxidant). This plot gives rise to a line with four distinct changes in the slope, according to the oxidation state of the POM 346. It was revealed that the resting state of POM 346 under aerobic conditions is the one-electron oxidized state RuV,Ru3IV,IV,IV. Aerobic oxidation yielded a paramagnetic species, which could be detected by cryogenic EPR, with gx, gy, and gz values of 1.97, 1.67, and 1.43, respectively. This indicates that there is a metal-centered S = 1/2 ground state, corresponding to a RuVRu3IV,IV,IV intermediate. Further addition of 1 equiv of CeIV transforms this paramagnetic RuVRu3IV,IV,IV species into the diamagnetic EPR inactive Ru2V,VRu2IV,IV species.446 11964

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

calculations, which revealed that the HOMOs of the POM 346 are close in energy.449 These findings support the observed Coulomb staircase in the electrochemical experiments and may be one key feature for successful four-electron oxidation of H2O in a narrow potential window. It could also be concluded that the four subsequent one-electron oxidations of POM 346, thus generating Ru4V, all involve the {Ru4} core orbitals. It was initially believed that catalytic H2O oxidation at pH 1 was triggered upon reaching the Ru2VRu2IV state and that cycling occurred through the Ru 2 IV Ru 2 III → Ru 2 V Ru 2 IV transition. However, this does not afford enough power because the redox potential is below the potential for oxidation of H2O to O2. Therefore, it seemed more likely that catalytic H2O oxidation was triggered upon reaching the Ru4V state. The redox couple Ru4V/Ru3VRuIV has an estimated potential of +1.45 V vs NHE and is well accessible if CeIV is used as the oxidant. At neutral pH, when [Ru(bpy)3]3+ is employed as oxidant, it is also likely that H2O oxidation is triggered when the Ru4V oxidation state is reached. Figure 100 depicts the

As previously mentioned, the kinetics of H2O oxidation is first-order with regard to the catalyst concentration, implying that the tetra-ruthenium core of POM 346 manages to carry out the four-electron oxidation of H2O to O2. Because bimolecular pathways do not contribute to the catalytic activity, H2O oxidation most likely occurs through nucleophilic attack of H2O on the high-valent ruthenium centers. Intramolecular coupling of two neighboring ruthenium centers can presumably be disregarded due to geometrical restrictions. Hill and co-workers also carried out an extensive mechanistic study to describe how O2 was evolved from POM 346.447 In this work, they were able to isolate the one-electron oxidized state of POM 346 after subjecting it to 2.2 equiv of CeIV. A related dinuclear ruthenium POM [{Ru2III,III(μ-O)2(H2O)2}(γSiW10O36)]6− was also synthesized, and under aerobic conditions this dinuclear POM could be oxidized to yield 346, according to eq 21. 2[{Ru 2 III,III(μ‐O)2 (H 2O)2 }(γ ‐SiW10O36 )]6 − + O2 + 2H+ → [{Ru4 IV (μ‐O)4 (μ‐OH)2 (H 2O)4 }(γ ‐SiW10O36 )2 ]10 − (21)

The crystal structure of the one-electron oxidized species of POM 346 is similar to 346 itself (see Figure 97), but there are clear differences in the bond lengths and the angles. Bond valence sum (BVS) calculations on the four ruthenium centers give an average value of 4.27, which is consistent with a oneelectron oxidation of the tetra-ruthenium core in POM 346 to form [RuVRu3IV].447 It was discovered that the electrochemical properties of POM 346 were dependent on pH and ionic strength. At acidic pH, five quasi-reversible waves were observed in the positive region of the voltammogram. At pH 2, four of the peaks were wellresolved in the positive region, while at pH 1 these peaks were shifted toward more positive values, indicating that they are coupled with the removal of protons. These waves were assigned to the Ru2VRu2IV/RuVRu3IV, RuVRu3IV/Ru4IV, Ru4IV/ Ru3IVRuIII, Ru3IVRuIII/Ru2IVRu2III, and Ru2IVRu2III/Ru4III redox couples, respectively (see Table 15).447

Figure 100. H2O oxidation catalysis by ruthenium POM 346. Reprinted with permission from ref 447. Copyright 2009 American Chemical Society.

processes occurring during H2O oxidation catalysis by POM 346. Recent work brought additional support for the presence of these high-valent oxidation processes via use of the Fourier transformed alternating current voltammetric technique.450 It should also be noted that the POM [{Ru4IV(μ-O)4(μOH)2(H2O)4}(γ-SiW10O36)2]10‑ 346 has been evaluated as an inorganic synzyme for the disproportionation of H2O2 to O2 and H2O.451 Upon the addition of H2O2 to a solution of POM 346, vigorous O2 evolution took place with quantitative peroxide decomposition within 20 min. The applied system was able of producing TONs up to 3000 and initial TOFs of ≥1.27 s−1. UV−vis spectroscopy verified that the structure of the catalyst remained unchanged under the catalytic conditions. Further work in the laboratory of Bonchio focused on attaching the ruthenium POM 346 to multiwalled carbon nanotubes (MWCNTs). Advantages associated with the use of MWCNTs are that they: (i) control the material morphology, (ii) increase the surface area, (iii) possess good mechanical properties, (iv) have thermal stability, and (v) are easy to functionalize. To attach the POM to the CNTs, the CNTs were decorated with polyamidoamine ammonium dendrimers. Subjecting the decorated CNTs to POM 346 at pH 5 yielded 346@MWCNT through electrostatic interaction between the positively charged dendrimer and the negatively charged POM 346 (see Figure 101). Exposing 346 to pristine MWCNTs resulted in negligible catalyst loading, thus clearly showing the influence of the attached dendrimer moiety. The functionalized material, 346@MWCNT, was characterized by a variety of different spectroscopic techniques. Resonance Raman spectroscopy showed that the structure of 346 was maintained after

Table 15. Potentials of the Anodic (Ea) and Cathodic (Ec) Peaks of POM 346a pH 2

pH 1

E1/2

Ea

Ec

E1/2

Ea

Ec

redox couple

1.17 0.95 0.73 0.57 0.23

1.21 1.03 0.78 0.63 0.44

1.14 0.87 0.68 0.51 0.02

1.21 1.01 0.78 0.62 0.35

1.25 1.06 0.82 0.66 0.47

1.18 0.96 0.75 0.59 0.24

Ru2VRu2IV/RuVRu3IV RuVRu3IV/Ru4IV Ru4IV/Ru3IVRuIII Ru3IVRuIII/Ru2IVRu2III Ru2IVRu2III/Ru4III

a

Adapted with permission from ref 447. Copyright 2009 American Chemical Society. All potentials are given in V vs NHE. Reaction conditions: Cyclic voltammograms of POM 346 (0.7 mM) were obtained under Ar, in an aqueous 0.2 M lithium sulfate buffer (pH 2) or an aqueous 0.1 M HCl solution (pH 1).

Plotting the reduction potentials of the redox couples as a function of the number of electrons (n) added or removed from POM 346 resulted in a linear dependence. The behavior exerted by POM 346 is ascribed to a Coulomb staircase and is rarely observed in molecular systems, where electronic coupling usually brings about unevenly spaced redox potentials.447 To explain this peculiar property, the authors carried out DFT 11965

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

surface properties of the two anodes made it possible to relate the electrocatalytic activity to the one-dimensional anisotropy and the electronic properties of the carbon supports. The catalytic activity of the electrodes was studied by cyclic voltammetry in a neutral aqueous solution (phosphate buffer, pH 7). Relevant background experiments verified that the POM-free support and the pristine MWCNTs did not catalyze H2O oxidation. By contrast, the POM-functionalized ITO electrodes generated an oxidation wave at 1.10 V vs NHE. This wave was followed by a catalytic current for H2O oxidation and translated into an overpotential of merely 0.35 V. Moreover, the 346@MWCNT material was found to display superior activity as compared to [email protected] These results show that interfacing WOCs with MWCNTs can afford new materials with improved stabilities and activities. Continued work in this area has resulted in the manufacturing of an assembly where POM 346 is attached to graphene via favorable electrostatic interactions (by using the same dendrimeric principle as with the MWCNTs). This novel material was evaluated by cyclic voltammetry and displayed an even lower overpotential (0.3 V) than the MWCNT functionalized material and proved to be stable over 4 h of testing with negligible loss of activity.453 A related approach has also been reported by Hill and co-workers for attaching Ru POM 346 to graphene.454 A detailed computational study of the tetraruthenium POM 346 was conducted by Bonchio and co-workers, and revealed that the tetraruthenium core in 346 mimicked RuO2 surfaces (Figure 102). X-ray absorption near edge structure (XANES) spectra of POM 346 and hydrous RuO2 displayed identical edge and line shape. DFT calculations also established the equivalency of the ruthenium core in POM 346 and the RuO2 crystalline surfaces. These calculations highlight that all of the ruthenium centers (Ru IV −OH 2 ) are present in a d 4 configuration with the electrons formally being centered on the ruthenium atom. According to their calculations, the first oxidation affords a single RuV−OH moiety (with three unpaired electrons on the ruthenium center) consistent with previous experimental studies of the ruthenium POM 346. However, the calculations pointed to the fact that, for the catalyst to reach its active state, it was necessary for the ruthenium core to be oxidized to Ru2VIRu2V. In this state, two of the ruthenium centers are present as RuVI−oxos (or RuV− oxyls). Reaching this high-valent intermediate will trigger O−O bond formation via nucleophilic attack of H2O on a ruthenium−oxo moiety with simultaneous loss of a proton, to generate a ruthenium−hydroperoxo intermediate,455 in accord-

Figure 101. Schematic overview of a H2O splitting cell containing the 346@MWCNT O2 evolving anode. Reprinted with permission from ref 452. Copyright 2010 Macmillan Publishers Ltd.

attachment to the MWCNT. Small-angle X-ray scattering (SAXS) and scanning transmission electron microscopy (STEM) experiments verified that the POM 346 was deposited on the MWCNT support as single molecular units and that minimal aggregation took place. This feature is essential for bridging heterogeneous and homogeneous catalysis and could lead to high turnover rates.452 With the aim of validating the importance of the CNTs, POM 346 was also attached to polyamidoamine ammonium dendrimer functionalized amorphous carbon (AC), to furnish 346@AC. Drop casting the hybrid materials onto ITO electrodes resulted in oxygen evolving anodes that were tested for their activity. The resemblance of the morphology and

Figure 102. (A) Structure of the ruthenium POM 346 in the resting state (Ru4IV). (B−F) The octahedral environment around a single ruthenium center in POM 346 together with the electronic structures of dfferent intermediates. Reprinted with permission from ref 455. Copyright 2013 The National Academy of Sciences. 11966

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

ance with many molecular single-site ruthenium WOCs.109−115 Concurrently, with formation of the Ru−OOH, the other ruthenium−oxo is converted to a Ru−OH through proton transfer. The Ru−OOH is further oxidized to a RuV−OO•, via reduction of one of the RuV−OH to a RuIV−OH2, from which O2 is finally liberated.455 This study illustrates that the presence of multimetal centers does not implicate that multisite mechanisms have to be involved in the reaction pathway. On the other hand, the presence of several metal centers, although not directly involved, could provide an electronically stabilizing factor for H2O oxidation catalysis. An interesting computational study was performed by Musaev and co-workers, where O2 and H2O activation (i.e., reduction of molecular O2 and H2O oxidation) were reviewed to compare these fundamental processes for ruthenium substituted POMs with the “blue dimer” 8.456 It was previously observed that the tetranuclear ruthenium POM 346 [{Ru4(μO)4(μ-OH)2(H2O)4}(γ-SiW10O36)2]10− was formed by oxidation of the dinuclear POM 348 [{Ru 2 III (μ-OH) 2 }(γSiW10O36)]4− by molecular O2.440 Although the dinuclear POM 348 can activate O2, experiments revealed that it was incapable of catalyzing H2O oxidation. Previous calculations by Yang and Baik showed that O2 activation by the structurally similar “blue dimer” 8, [(bpy)2(H2O)RuIII(μ-O)RuIII(H2O)(bpy)2]4+, to generate [(bpy)2(O)RuV(μ-O)RuV(O)(bpy)2]4+ is endothermic by 22.8 kcal mol−1 and has a high energy barrier of 48.7 kcal mol−1.155 The reason for this fundamental difference in reactivity between the bridged diruthenium systems in the structurally related POM 348 and the “blue dimer” 8 is of considerable interest. From the extensive study by Musaev and co-workers, it appears that this difference is due to the different energetics in the final step in the oxidation of H2O, eqs 22 and 23.456

BVS calculations validated that all ruthenium centers were in the +IV oxidation state. As compared to POM 346, the BVS suggests that only one of the oxygen atoms bridging the two ruthenium centers within one {PW10O37Ru2} moiety is protonated in the new POM 349. The electrochemistry of POM 349 resembled that of the previously studied 346, and the redox potentials also exhibited a linear dependence on the redox states. Applying the synthesized POM 349 in photocatalytic H2O oxidation, in a system consisting of [Ru(bpy)3]2+ as photosensitizer and persulfate as electron acceptor, resulted in lower TONs than for 346, which was ascribed to the slightly lower driving force for the reaction with 349.457 Fukuzumi and co-workers reported the synthesis of the two single-site ruthenium POMs [RuIII(H2O)SiW11O39]5− (350) and [RuIII(H2O)GeW11O39]5− (351), depicted in Figure 103.458 Employing these POMs in catalytic H2O oxidation

Figure 103. Structures of single-site ruthenium POMs [RuIII(H2O)SiW11O39]5− (350, left) and [RuIII(H2O)GeW11O39]5− (351, right). Adapted with permission from ref 458. Copyright 2011 American Chemical Society.

with CeIV as oxidant resulted in evolution of O2. Background reactions, performed by combining different ruthenium sources and K8[SiW11O39] (or K8[GeW11O39]), did not display any catalytic activity, thus verifying that POMs 350 and 351 are responsible for the observed O2 evolution. The use of 18O enriched H2O confirmed that the generated O2 originated exclusively from H2O and not from any other source. To gain insight into the catalytic mechanism, the Pourbaix diagrams of the two single-site POMs in the range 0 < pH < 8 were constructed (Figure 104). Three one-electron redox processes were shown to occur, involving the RuIII/RuII, RuIV/ RuIII, and RuV/RuIV redox couples. At pH 1, the Pourbaix diagrams revealed that the first redox process is coupled to the removal of one proton (because E1/2 is decreased by −59 mV per pH unit) and presumably involves the transition (HL)RuII−OH2 → (L)RuIII−OH2 (L = [SiW11O39]8− for 350 or [GeW11O39]8− for 351). The next step, which is not coupled with the loss of a proton, suggests that (L)RuIV−OH2 is generated. From this intermediate, the high-valent (L)RuVO is formed through a two-proton-one-electron event and is thought to be the active species in H2O oxidation for the singlesite ruthenium POMs. As can be seen from the Pourbaix diagrams, the pKa values of 351 were lower than for 350. This could be explained by the fact that germanium is more electronegative than silicon and affects the acidity of the ruthenium−aqua ligand, which renders the ruthenium center less electron-rich. UV−vis spectrophotometric titration of POMs 350 and 351 also confirmed the formation of (L)RuVO upon addition of stoichiometric equivalents of CeIV to an aqueous solution (pH 1) containing the catalyst.458 Cryogenic EPR experiments of 350 displayed a two-axis anisotropic signal with g⊥ = 2.4 and g∥ = 2.0, which was assigned to a RuIII species with S = 1/2. Oxidation of POM 350

[{(O•)Ru(μ‐OH)2 Ru(O•)}(γ ‐SiW10O36 )]4 − → [{Ru(μ‐OH)2 Ru}(γ ‐SiW10O36 )]4 − + O2

(22)

[(bpy)2 (O•)Ru(μ‐O)Ru(O•)(H 2O)(bpy)2 ]4 + → [(bpy)2 Ru(μ‐O)Ru(bpy)2 ]4 + + O2

(23)

The calculated energies for eqs 22 and 23 were found to be 70.4 and 21.5 kcal mol−1, respectively. This large difference may originate from several factors, including the ruthenium−oxygen and ruthenium−L (where L = (SiW10O36) or bpy) bond strengths. It was revealed that neither the geometries of the L ligands nor the two ruthenium cores change dramatically during these processes. However, the RuO bonds are stronger in POM 348 than in the “blue dimer” 8, and the Ru−L bond lengths are substantially affected for the POM 348 (when L = (SiW10O36)). Thus, the electron-rich nature of the SiW10O36 moiety, with its better σ-donor and π-acceptor abilities as compared to bpy, may explain the difference in reactivity for the POM 348 and the “blue dimer” 8.456 Hill and co-workers have also reported on another isostructural ruthenium-based POM that displayed activity as a WOC. Because the catalytic activity is strongly dependent on the redox potentials of the catalyst, they speculated that changing the central heteroatom from Si4+ to P5+ would result in a change in the overall charge of the POM, altering its redox properties. Therefore, the [γ-SiW10O36]8− moiety in POM 346 was replaced with the analogous [γ-PW10O36]7− moiety to yield the POM [{Ru4(μ-O)5(μ-OH)(H2O)4}(γ-PW10O36)2]9− 349. 11967

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

Figure 104. Pourbaix diagram of (a) POM 350 and (b) POM 351 (pKa values are denoted by the vertical dashed lines; L = [SiW11O39]8− for 350 or [GeW11O39]8− for 351). Conditions: Electrochemical measurements of 350 and 351 (2.0 mM) were obtained in 0.1 M Britton−Robinson buffer solutions in the range 1.5 < pH < 8.0 and in 0.1 Na2SO4 aqueous solutions in the range 0 < pH < 1.5 by using a glassy carbon electrode as working electrode, a saturated calomel electrode (SCE) as reference electrode, and Pt wire as auxiliary electrode. The pH of the solution was changed by using 0.36−3.6 M NaOH aqueous solutions or HNO3 (69%). Reprinted with permission from ref 458. Copyright 2011 American Chemical Society.

by addition of 2 equiv of CeIV generated a new anisotropic signal with g⊥ = 2.1 and g∥ = 1.9 and was indicative of a RuV O species. When CeIV was used as oxidant, the driving force for the oxidation to generate the (L)RuVO is large and expected to proceed with diffusion-controlled rate. Kinetic experiments resulted in a rate law that was first-order and denotes that the O−O bond formation probably occurs through nucleophilic attack of H2O on the RuVO species to produce a RuIII− OOH intermediate. Saturation kinetics suggested that the successive oxidation of the RuIII−OOH, with final release of O2, competes with the back reaction, thus regenerating the RuVO species. Collectively, all experiments pointed to a mechanism in which a RuVO intermediate is formed via fast oxidation of RuIII−OH2. Attack by H2O on this high-valent RuVO generates RuIII−OOH, which is further oxidized and ultimately leads to the liberation of O2 (Scheme 22).458 Scheme 22. Plausible Mechanism of H2O Oxidation Mediated by Single-Site Ruthenium POMs 350 and 351a

Figure 105. Structure of cobalt POM 352 ([Co4(H2O)2(PW9O34)2]10−). Reprinted with permission from ref 433. Copyright 2010 American Association for the Advancement of Science (AAAS).

catalytic activity was assessed using [Ru(bpy)3]3+ as the chemical oxidant. When using a concentration of 0.12 μM of the POM and 2.4 mM of the oxidant (at pH 8), a high TON of 1000 and TOF > 5 s−1 were achieved. The rate of H2O oxidation was found to be highly dependent on the pH of the reaction medium. The authors were careful to substantiate the apparent stability of cobalt POM 352 under the applied turnover conditions.433 Seven lines of evidence were provided to address the stability of the synthesized cobalt POM: (i) 31P NMR spectra and (ii) UV−vis spectroscopy of a pH 8 solution of POM 352 showed no significant structural changes, (iii) addition of free bpy to the reaction solution did not inhibit H2O oxidation (if free CoII ions are formed during catalysis and are responsible for the catalysis, the bpy would coordinate to the dissolved CoII ions to generate [Co(bpy)3]2+ and thereby inhibit the H2O oxidation), (iv) a postreaction solution containing POM 352 was analyzed by 31P NMR and displayed only peaks belonging to 352 and free phosphate from the buffer, (v) IR characterization, (vi) reuse of the cobalt POM after a catalytic run, and (vii) electrochemical measurements provided proof that the active

a

Adapted with permission from ref 458. Copyright 2011 American Chemical Society.

5.1.2. Cobalt-Based Polyoxometalates − The Ambiguity of Distinguishing between Homogeneous and Heterogeneous Catalysis. An important discovery was made in 2010 when Hill and co-workers found that the cobalt-based POM [Co4(H2O)2(PW9O34)2]10‑ 352 was capable of mediating H2O oxidation (Figure 105).433 In fact, this cobalt POM was originally synthesized by Weakley and co-workers in the early 1970s, but was not studied for H2O oxidation at the time.459 Cyclic voltammetry of the cobalt POM revealed a catalytic current for H2O oxidation occurring close to the thermodynamic potential. Other cobalt-based POMs with different cobalt cores did not display this catalytic feature, highlighting that the structural core of the POM is important. For POM 352, the 11968

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

species was retained after catalytic turnover. These findings suggest, at least under these conditions, that the active catalytic species is homogeneous and that dissolved CoII is not responsible for the observed catalytic activity.433 Light-driven H2O oxidation with POM 352 has also been achieved in a photocatalytic system consisting of POM 352, [Ru(bpy)3]2+, and persulfate.460,461 Control experiments validated that O2 generation required the presence of all four components: photons, POM 352, [Ru(bpy)3]2+, and persulfate. It was also noted that photosensitizer decomposition was substantially higher in the absence of POM 352, suggesting that efficient electron transfer takes place from the catalyst to the oxidized photosensitizer, thus protecting it from decomposition. The addition of more persulfate after ceased evolution of O2 led to resumed H2O oxidation and confirmed that the structural integrity of the catalyst core was retained during catalysis. Additional spectroscopic experiments, recycling of the catalyst, and DLS data testified that catalyst degradation did not occur during catalysis.460 Elaborate studies aiming at revealing the limiting factor of this photocatalytic system showed that the chemical yield (ΦCY = 2 [evolved O2]/[Na2S2O8]0) is a crucial component that limits the quantum yield (ΦQY = 2 [evolved O2]/amount of photons absorbed). The ΦCY increased linearly with catalyst concentration, suggesting that there existed a noncatalytic pathway for the photogenerated [Ru(bpy)3]3+ besides the preferred catalytic pathway. The initial quantum yield was determined to 0.30 for the photosystem with catalyst 352 and is thus higher than that of ruthenium POM 346. It was found that also the chemical yield and TON were higher for 352 than for the ruthenium POM 346, which hints that POM 352 has a better selectivity for mediating H2O oxidation. By using stopped-flow technique, the authors measured the concentration of [Ru(bpy)3]3+ by monitoring the absorption at 670 nm. This made it possible to compare the efficiencies of 346 and 352 in catalytic H2O oxidation. The results obtained from these measurements suggested that POM 352 is more efficient than POM 346.460 As previously discussed in this Review, it is difficult to prove if a catalyst operates as a homogeneous catalytic entity or if the initial molecular species is acting as a precursor for in situ generated metal oxide particles that are responsible for the observed catalytic activity. The investigated catalyst may operate under one mechanistic pathway under a certain set of reaction conditions but by a completely disparate mechanism under different, or even slightly deviating, conditions. The question whether Co POM 352 truly is a homogeneous catalyst or merely acts as a precursor for a heterogeneous material was spurred by a report by Finke (vide infra) and has been under debate ever since. Stracke and Finke investigated electrochemical H 2O oxidation by the tetracobalt POM 352 and found that the actual catalytic species responsible for the observed activity might not be of homogeneous nature. Instead, several lines of evidence were presented that indicated that in situ generated heterogeneous CoOx particles were responsible for the oxidation of H2O.421 The authors conducted detailed electrochemical studies of POM 352 under “relevant turnover conditions” (i.e., phosphate buffer, pH 8). A 500 μM aqueous phosphate buffer solution of POM 352 was carefully studied by linear-sweep voltammetry, which resulted in a catalytic wave for H2O oxidation with an onset potential of 1.25 V vs NHE. Aging of this solution, over a 3 h period, resulted in an increase (∼10

times) of the catalytic wave and a decrease of the onset potential. This aging effect suggested that the dominant active species was not initially present and that POM [Co4(H2O)2(PW9O34)2]10− (352) was transformed in situ to a more active derivative under these conditions. Application of a constant potential of 1.30 V vs NHE led to a rapid increase in the current with formation of O 2 . Accompanying the increase in current, the authors observed a deposition of a film on the working electrode. This film was analyzed by a toolbox of techniques consisting of UV−vis, SEM, and EDX, which identified it as CoOx (Scheme 23). Scheme 23. Proposed Pathway for Formation of Heterogeneous CoOx Catalyst from the Tetracobalt POM 352a

a

Reprinted with permission from ref 421. Copyright 2011 American Chemical Society.

Removing the electrode from the solution, rinsing it with H2O, and positioning it in an aqueous solution consisting only of a 0.1 M phosphate buffer solution (pH 8.0) resulted in sustained catalytic activity in the electrochemical experiments.421 SEM measurements of the generated electrode film showed that it did not contain any tungsten, which indicated that none of the POM was deposited on the electrode. UV−vis measurements showed that the absorbance band at 580 nm of an aqueous phosphate buffer solution (pH 8.0) containing 500 μM [Co4(H2O)2(PW9O34)2]10− (352) decreased by 4.3 ± 0.6% over a 3 h period, corresponding to a decomposition of 21.5 μM of 352. By constructing a calibration curve of aq [CoII], the authors were able to calculate the time-dependent concentration of CoII for a 500 μM [Co4(H2O)2(PW9O34)2]10− (352) solution. It was shown that over a 3 h period the CoII concentration increased from 1 ± 1 to 58 ± 2 μM. Further experiments were conducted to determine how much of the catalytic activity could be attributed to the [CoII]. By comparing a 3 h aged 500 μM solution of [Co4(H2O)2(PW9O34)2]10− (352) with a 58 μM Co(NO3)2 solution, Stracke and Finke were able to establish that the amount of leached CoII could account for the observed catalytic H2O oxidation activity.421 It should be noted that there exists a great difference in the physicochemical properties of electrochemical and chemical H2O oxidation (either by using a pregenerated oxidant or in situ formation of the oxidant as in photocatalytic systems). Stracke and Finke also emphasized that the study421 did not examine either the system or the precise conditions that were utilized by Hill433,460 and co-workers. Further studies were therefore necessary to determine to what extent CoOx particle formation occurs with chemical oxidants and in photochemical H2O oxidation experiments, and to what degree these particles are responsible for the observed catalytic activity. 11969

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

samples, the prereaction electrodes had significantly higher amounts of cobalt surface coverage, thus emphasizing that not even the electrodeposited CoOx species is stable under the reactions conditions used in this study.462 Although Stracke and Finke were unable to definitely distinguish between a homogeneous and a heterogeneous mechanism (or from a combination of the both), it could be concluded by the means of this study that multiple, complementary methods are necessary to resolve the homogeneity−heterogeneity issue associated with cobalt POM 352 under a certain set of reaction conditions. Subsequent work by the group of Geletti and Hill463 therefore focused on examining the active species for the cobalt POM 352, under the originally investigated catalytic reaction conditions.433 By the use of techniques such as cathodic adsorptive stripping voltammetry and inductively coupled plasma mass spectrometry, the authors could quantify the amount of leached cobalt into solution from POM 352 and found that O2 generation by POM 352 was dependent on buffer, pH, and catalyst concentration. Collectively, it was proposed that the amount of released cobalt was not enough to account for the observed H2O oxidation activity,463 thus suggesting that POM 352 operates as a molecular WOC under the originally investigated conditions.433 The homogeneity issue of the tetracobalt POM 352 in photochemical systems was also investigated by Sartorel and Scandola by use of laser flash photolysis.464 The bleaching of the [Ru(bpy)3]2+ MLCT absorption (λ = 450 nm) was used to monitor the electron-transfer processes. In a system consisting of POM 352, [Ru(bpy)3]2+, and persulfate, flash photolysis was shown to lead to bleaching of the MLCT absorption with subsequent bleach recovery (occurring from [Ru(bpy)3]3+ reduction via electron transfer). Omitting POM 352 resulted in persistence of the bleach, confirming that 352 efficiently quenched the photo-oxidized photosensitizer. However, the most striking finding was that electron transfer increased with the aging of the solutions containing POM 352. By plotting the amount of reduced [Ru(bpy)3]3+ versus the aging time of the solutions of 352, the authors demonstrated that pristine 352 barely contributes to the electron-transfer processes of [Ru(bpy)3]3+. Instead, another species originating from POM 352 was presumably responsible for the observed electrontransfer processes. An obvious candidate for the in situ generated species was free CoII ions. However, control experiments with Co(NO3)2 resulted in negligible amounts of reduced [Ru(bpy)3]3+. This, in combination with light-scattering experiments, showed that generation of colloidal oxides did not occur either.464 Instead, a previously reported465 decomposition product with a Co:POM stoichiometry of 2:1 was found. Cyclic voltammetry (in 0.2 M phosphate buffer, pH 8.0) was also employed to monitor the in situ formation of an active WOC and revealed that a small catalytic wave at 1.40 V vs NHE appeared immediately upon dissolution of POM 352. This confirmed that [Ru(bpy)3]2+, which has a redox potential of 1.26 V vs NHE for the RuIII/RuII redox couple, was not thermodynamically capable of promoting H2O oxidation with pristine 352. Solutions that are aged showed a buildup of a catalytic wave starting at 1.26 V, which explained the behavior observed in laser flash photolysis experiments.464 This work suggests that POM 352 is incapable of promoting H2O oxidation with photochemically generated [Ru(bpy)3]3+ as the oxidant and that the in situ formation of

The continued work of Stracke and Finke provided further evidence that the exact reaction conditions employed have a direct impact on the nature of the active WOC. In this study, they used a wide range of experiments to support that cobalt POM 352 does not decompose to CoOx when lower catalyst concentrations and higher potentials are employed.462 In their previous work,421 Stracke and Finke used a concentration of POM 352 of 500 μM, and electrochemically driven H2O oxidation was performed at 1.30 V vs NHE. However, in the latter investigation,462 a POM concentration of 2.5 μM and potentials ≥1.50 V vs NHE were applied. Cyclic voltammetry of solutions containing low concentrations of the tetracobalt POM 352 (2.5 μM) revealed several differences as compared to deposited heterogeneous CoOx: (i) the onset potential for H2O oxidation of the cobalt POM was found to be a couple of hundred millivolts more positive than CoOx, and (ii) the pH and Tafel features for CoOx differed from the cobalt POM. Repeated cycling of the POM 352 demonstrated that there was no increase in waves belonging to in situ generated CoOx species. Rinsing of the electrode after bulk electrolysis at 1.60 V vs NHE of a 2.5 μM solution of 352 demonstrated that the electrode showed only background activity.462 Altogether, these findings pointed against the suggestion that the catalytic activity under these exact conditions originated from heterogeneous CoOx, generated from aqueous Co2+, which is the opposite to the results of their former study421 where the dominant catalytic species was proposed to be electrodeposited CoOx. Stracke and Finke also studied the hydrolytic stability of the apparent tetracobalt POM in aqueous phosphate buffer solutions (0.1 M, pH 8), and they could conclude that aging of these solutions resulted in the release of cobalt ions from the POM core. Solutions that had been aged for 1 h showed a cobalt release corresponding to a “decomposition” of 2.5% of POM 352, thus indicating that the POM was not completely stable in solutions in contrast to the earlier findings of Hill and co-workers.462 To determine to what degree the released Co atoms contributed to the observed O2 evolution activity, the authors compared aged solutions of cobalt POM 352 with Co(NO3)2, which showed that the latter could not account for the observed catalytic O2 evolution.462 This differs from their previous finding421 where the dissociation of cobalt at higher POM concentrations (500 μM of 352) could quantitatively account for the observed O2 generation. Although direct hydrolysis cannot account for the observed activity, this does not rule out the direct oxidative (electrochemical) decomposition of POM 352 into highly active CoOx particles when a potential is applied. This issue was investigated by electrodeposition of CoOx from Co(NO3)2 solutions with further evaluation of the O2 formation activity. The activity of these CoOx coated electrodes suggested that as little as 4−8% POM decomposition could account for the observed catalytic activity of cobalt POM 352. Postreaction solutions of POM 352 were therefore analyzed and revealed that transformation of POM 352 occurred to such an extent that the possible formation of colloidal CoOx could not be neglected as the active WOC under the specific reaction conditions. These postreaction measurements suggest that even if POM 352 indeed is the active WOC, the POM is not completely stable under the reaction conditions employed in the study.462 When the deposited material on glassy carbon electrodes was treated with either POM 352 or CoOx solutions and subsequently analyzed by XPS, it showed the presence of traces of cobalt only. However, as compared to the postreaction 11970

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

2.085 Å) suggests that the oxo ligands present in 352 weaken the Co−O bond. However, as highlighted by NMR experiments, this weakening does not have a major impact on the H2O exchange rate in POM 352. This could be explained by cluster-based hydrogen bonding, through protonation of the oxo ligands in the surroundings of the coordinated aqua ligand, which stabilizes it and slows H2O exchange. High-resolution mass spectrometry measurements confirmed that the POM structure was intact in the pH region. In POM 353, the H2O exchange rate was found to be pH independent and 1 order of magnitude slower than that of 352 and [Co(H2O)6]2+. This is caused by an elongated Co−O bond trans to the aqua ligand that causes the cobalt center to become quasi-five coordinated and induces a contraction of the Co−OH2 bond.467 Thus, the fast H2O exchange in POM 352 excludes this process from being rate-limiting and responsible for the maximum catalytic activity observed at pH 8.433 Xie and co-workers synthesized anodes functionalized with POM [Co4(H2O)2(PW9O34)2]10− 352, where mesoporous carbon nitride (MCN) was chosen as the supporting material due to its uniform nanochannels, high surface area, and high conductivity.470 In addition to these properties, MCN also offers −NH2 and −NHR groups on its walls that can be used as possible sites for immobilization of the negatively charged POMs. The functionalized MCN was assembled by immersing protonated MCN in an aqueous solution of the cobalt POM 352 for 24 h. During this time, the POM penetrated the channels of the MCN and was assembled at the positively charged inner sphere of the mesopores of the MCN. TEM was used to determine the morphology and distribution of the POM. This affirmed that the POM clusters were well-dispersed inside the nanochannels and that the shape and diameter of the nanochannels of the POM functionalized MCN were similar to those of pristine MCN. Catalytic activity in H2O oxidation was assessed by loading the functionalized MCN onto an ITO electrode through casting. In aqueous phosphate buffer at pH 7.0, these electrodes showed a catalytic wave, with an onset potential of 1.20 V vs NHE, assigned to H2O oxidation. As anticipated, the POM functionalized material displayed a higher electrocatalytic activity than that of the POM 352 itself, suggesting a synergistic effect between the POM and the MCN material. UV−vis spectroscopy and XRD of the POM/MCN functionalized ITO electrode did not display any distinct changes before and after the electrolysis, showing that the structure of the POM/MCN functionalized ITO electrode was stable during the electrochemical measurements. These results also demonstrate that anchoring of the cobalt POM to MCN improved the electrical contact between the redox-active cobalt centers and the electrode surface.470 Hill and co-workers recently reported the synthesis and catalytic activity of a silicon analogue of the tetracobalt POM 352, the POM [{Co 4 (μ-OH)(H 2 O) 3 }(Si 2 W 19 O 70 )] 11− (354).471 They isolated the new silicon-based POM in two different isomeric forms, which cocrystallized as a 1:1 mixture. The activity of POM 354 was tested in photocatalytic H2O oxidation with the [Ru(bpy)3]2+−persulfate system. With this system, a TON of 80 and an initial TOF of 0.1 s−1 could be reached. The stability of POM 354 was then examined under the applied turnover conditions by measuring the spectral changes of 354 over time, at different pH values. The absorbance band increased with time, and this trend was visible in a variety of buffer solutions, such as borate buffer,

molecularly active species from POM 352 are most likely responsible for the fast electron transfer to [Ru(bpy)3]3+. Collectively, these studies signal that attempts to discriminate between homogeneous and heterogeneous catalysis have to be done with great care and that the presence or absence of a deposited film is in itself not sufficient evidence of a certain type of catalysis.466 It is obvious that the exact nature of reaction conditions can govern whether the true identity of the catalytic species is of homogeneous or heterogeneous nature and subtle changes of these conditions, such as the concentrations of the different components, can dramatically affect the outcome. It is thus clearly difficult to determine whether the catalytic activity is derived from the initial molecular species, from small amounts of a heterogeneous nanocrystalline decomposition product, or even from small amounts of a secondary homogeneous complex. This ambiguity has become a hot topic for discussion; however, the problem of distinguishing between homogeneous and heterogeneous catalysis is not limited to the field of WOCs but is a general issue in catalysis. A study performed by the group of Ivanović−Burmazović focused on analyzing the H2O exchange in tetracobalt POM 352 and in an inactive CoII POM ([CoII(H2O)(SiW11O39)]6−, 353) (Figure 106, right).467 There are two equivalent aqua

Figure 106. Left: Polyhedral representation of POM 352 ([Co4(H2O)2(PW9O34)2]10−). Center: Top view of the central Co4 core, highlighting the two aqua ligands (large red spheres). Right: Polyhedral representation of the inactive POM 353 ([CoII(H2O)(SiW11O39)]6−). Reprinted with permission from ref 467. Copyright 2011 American Chemical Society.

ligands within the structure of the tetracobalt POM 352, located on two cobalt centers, on opposite sides of the POM cluster, which are considered to be important for the catalytic activity of 352. This fact motivated the authors to carry out a study that aimed to provide insight into the kinetics of the H2O exchange at the cobalt centers. Previous data support that H2O exchange occurs faster at CoII centers than at CoIII centers.468 In aqueous solutions, this H2O exchange parameter can thus be used to determine the redox state of cobalt. The kinetics of the H2O exchange processes for POMs 352 and 353 were determined by the use of temperature- and pressure-dependent 17O NMR spectroscopy in aqueous solutions with varying pH values (6.1 < pH < 10).467 Measurements revealed that the rate constant for the exchange was of the same order as that of [Co(H2O)6]2+, thus supporting a previous study.469 It was found that at different pH’s the rates and activation parameters did not change significantly. These results imply that the coordinated aqua ligands are not deprotonated under the studied pH range. Inspection of the average Co−OH2 bond distances in POM 352 (dCo−O = 2.116(6) Å) and in [Co(H2O)6]2+ (dCo−O = ≤ 11971

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

intermediate was not involved in the catalytic cycle. A decline in the catalytic activity was observed at high POM concentrations, which could be explained by the deposition of a double salt consisting of photosensitizer and the cobalt POMs. Electrochemical measurements that were carried out on the two POMs showed catalytic currents starting at 1.45−1.55 V vs NHE, which could be ascribed to H2O oxidation, confirming that they are indeed active as WOCs. It was also noted that CeIV could not be used to drive H2O oxidation with POMs 355 and 356. Mixing the cerium oxidant and the POM resulted in deposition of a dark brown solid. Analysis of the solid showed that cerium, molybdenum, and cobalt constituted the major components. This study clearly demonstrated that the tetracobalt core of POM 352 was not a necessary structural motif for the preparation of active WOCs.472 Interestingly, a mixed-valence dinuclear Co2II,III POM has just been reported by Song et al.473 Although there are many POMs that are capable of oxidizing H2O, no special attention has been given to the Keggin-type POMs. The Keggin-type POMs have the general structural formula [XM12O40]n− (where X can be P5+, Si4+, B3+, etc.), which is the most stable structural motif of POM catalysts. The authors reported the isolation of a Keggintype POM, [CoIIICoII(H2O)W11O39]7− (357), built up by the exotic mixed-valence Co2II,III core. The crystal structure of the mixed-valence POM revealed that the average bond length of Co−O in one cobalt center was ∼1.822 Å, while the Co−O distance in the other metal center was determined to ∼2.075 Å. A difference in coordination was also observed for the two cobalt centers; the CoIII center is situated in a tetrahedrally coordinated environment, while the CoII center is octahedrally coordinated. The catalytic activity was initially assessed in light-driven H2O oxidation with [Ru(bpy)3]2+ as photosensitizer and persulfate as sacrificial electron acceptor. Furthermore, a detailed study was conducted where several other Keggintype POMs were evaluated as WOCs. The library of screened Keggin POMs, in addition to 357, consisted of [CoIIW12O40]6− (358), [CoIIIW12O40]5− (359), [CoIICoII(H2O)W11O39]8− (360), [SiCoII(H2O)W11O39]6− (361), and [PCoII(H2O)W11O39]5− (362). The experiments revealed that the POM analogues 358 and 359 exhibited slight activity, while 361 and 362 were inactive. Only the mixed-valence POM 357 displayed good activity, signaling that the nature of the central heteroatom is important for the catalytic activity. POM 360 has the possibility of mediating H2O oxidation as well; however, it proved not to be stable under the catalytic conditions. Judging from these experiments, it seems plausible to claim that the octahedral CoII site is responsible for the observed H2O oxidation activity, while the central tetrahedral CoIII center has a profound effect by modulating the activity of the other cobalt center (i.e., the catalytically active CoII center). The use of isotopically labeled H2O, H218O, validated that the generated O2 originated exclusively from solvent H2O. Electrochemical measurements of POM 357 displayed a catalytic wave for H2O oxidation, with an onset potential of ∼1.17 V vs NHE, affirming that [Ru(bpy)3]3+ has thermodynamic power to drive H2O oxidation.473 Because it has previously been argued that cobalt-based POMs might decompose, resulting in a more active WOC (of either homogeneous or heterogeneous nature), the authors were particularly motivated to address this issue. Multiple lines of evidence indicated that POM 357 was stable under the photocatalytic experiments: (i) DLS measurements confirmed

sodium acetate buffer, and phosphate buffer. Increasing the pH resulted in a faster increase of the absorbance changes. To explain these findings, an aged solution of POM 354 was slowly evaporated to yield X-ray quality crystals for structure determination. It could be concluded that 354 was converted to [{Co(H 2 O)}(μ-H 2 O) 2 K{Co(H 2 O) 4 }(Si 2 W 18 O 66 )] 11− , which was subsequently transformed to the thermodynamically stable species [Co(H2O)SiW11O39]6−.471 As a result of the hydrolytic decomposition of POM 354, the question arose if 354 was really the actual catalyst mediating the H2O oxidation. Addition of bpy, which functions as a poisoning agent of free [Co(H2O)6]2+,433 resulted in a precipitate that was shown to contain [Co(H2O)SiW11O39]6− with [Co(bpy)3]2+ as countercation(s). This finding emphasized the apparent instability associated with POM 354. Furthermore, aged solutions of POM 354 displayed lower activity than that of freshly prepared solutions of 354, while freshly prepared solutions of the decomposition products had less or no activity in H2O oxidation catalysis. This suggests that POM 354 does act as an active WOC under the investigated photocatalytic conditions.471 The question that arose at this point was whether it was necessary to have a tetranuclear core, or if catalytic activity could be achieved with “low-nuclearity” cobalt POMs. This question was assessed by the group of Sakai by the development of two cobalt POMs: a mononuclear POM with the structural formula [CoMo6O24H6]3− (355) and a dinuclear POM, [Co2Mo10O38H4]6− (356). These POMs were evaluated in light-driven H2O oxidation with the [Ru(bpy)3]2+−persulfate system. Both POMs were found to generate O2, and DLS confirmed that no formation of colloidal particles took place. This constituted the first example of highly active O2 evolving POMs with cores lower than tetranuclear. Addition of extra persulfate after the cessation in O2 evolution resulted in restoration of the catalytic activity, thereby highlighting the stability of the catalysts.472 Furthermore, it was shown that the O2 evolving reaction was competing with O2 nonevolving reactions. A reasonable explanation was the decomposition of [Ru(bpy)3]n+ (n = 2 or 3), resulting from nucleophilic attack of H2O on the bpy rings. This H2O attack would lead to the generation of hydroxylated bpy radicals (eqs 24−27) that would ultimately cause degradation of [Ru(bpy)3]n+ and generation of CO2.93 In the presence of the POMs, the degradation of the photosensitizer becomes much slower, thus emphasizing that the efficient electron transfer from the POM to the photo-oxidized photosensitizer protects it from degradation.472 [Ru(bpy)3 ]2 + + S2 O82 − hν

→ [Ru(bpy)3 ]3 + + SO4 2 − + SO4•−

[Ru(bpy)3 ]2 + + SO4•− → [Ru(bpy)3 ]3 + + SO4 2 −

(24) (25)

[Ru(bpy)3 ]n + + SO4•− → [Ru(bpy)2 (bpy •+)](n + 1) + + SO4 2 −

(26)

[Ru(bpy)2 (bpy •+)](n + 1) + + H 2O → [Ru(bpy)2 (bpy •−OH)]n + + H+

(27)

Kinetic experiments concluded that the reaction was first order in POM concentration, suggesting that a dimeric 11972

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

that nanoparticles were not produced, (ii) electrochemical experiments of 357 and Co(NO3)2 (from which nanoparticles are produced during catalysis) in aqueous borate buffer solutions displayed distinct differences, thus providing support for the stability of 357, (iii) a comparison of POM 357 ([CoIIICoII(H2O)W11O39]7−) with its structural analogue [CoIICoII(H2O)W11O39]8− (360) revealed that it was not hydrolytically stable in borate buffer solutions (borate buffer solutions of 360 generated a blue-purple precipitate after only 1 h of aging, which was determined to be CoOx), (iv) the recycled catalyst 357 was studied by Fourier transform infrared spectroscopy (FT-IR), DLS, EDX, and its activity was tested in H2O oxidation, thus verifying that the structural integrity of 357 remained intact, and (v) UV−vis indicated no spectral changes in borate buffer within the 10 min duration of the reaction. In addition, solutions of POM 357 that were aged for 24 h exhibited no change in H2O oxidation activity. The vital differences between POM 357 and its structural analogue 360 suggested that the central cobalt center with its +III oxidation state plays an essential part in the preservation of the structural features of POM 357.473 Flash photolysis was subsequently performed with POM 357, and it was concluded that it was capable of quenching the photogenerated [Ru(bpy)3]3+. Aging of solutions of POM 357 did not result in any changes in the electron-transfer activity,473 the opposite to what Sartorel and Scandola found for POM 352,464 signifying that 357 is stable over the investigated time period. It was found that having a high concentration of either the catalyst 357 or the photosensitizer resulted in precipitation of a solid material, supposedly consisting of the photosensitizer and POM 357. Carrying out the photochemical experiments in different buffers verified the effect they had on the catalytic activity. By measuring the pH after O2 formation had ceased, it was revealed that with some buffers the pH had dropped significantly. The authors discovered that borate buffer was most efficient in buffering the pH of the reaction solution, thus providing an explanation why a higher catalytic activity was observed for the reactions performed in this buffer. Consequently, the decrease in the catalytic activity for the other buffers may partially originate from the lower thermodynamic driving force for H2O oxidation at lower pH values.473 Cobalt POMs of higher nuclearity have also been prepared, such as the nonanuclear cobalt POM {Co9(H2O)6(OH)3(HPO4)2(PW9O34)3}16− (363, Figure 107), synthesized by the group of Galán−Mascarós.474 The nonanuclear POM was prepared and proved to be more stable at higher pH than the tetracobalt POM 352. Initial electrochemical experiments showed that a brown film, characterized as cobalt oxide, was deposited on the anode. By UV−vis measurements, it could be determined that ∼15% of the initial POM 363 decomposed during electrolysis. On addition of bpy to a solution of the POM to trap free CoII, no material was deposited on the electrode, affirming that POM 363 could mediate H2O oxidation. The catalytic activity for POM 363 was subsequently evaluated with the two-electron and potential oxygen-atom transfer oxidant NaClO (in 0.9 M phosphate buffer, pH 8.0). The efficiency of this catalytic system was found to be low, suggesting that competing reaction pathways occurred, such as disproportionation of hypochlorite to chloride and chlorite. Generation of CoOx under the chemical oxidations was ruled out by means of UV−vis spectroscopy and DLS, which showed

Figure 107. Molecular structures of the polyanions [Co4(H2O)2(PW 9 O 34 ) 2 ] 10− (352, left) and {Co 9 (H 2 O) 6 (OH) 3 (HPO 4 ) 2 (PW9O34)3}16− (363, right). Reprinted with permission from ref 474. Copyright 2012 American Chemical Society.

neither change in absorbance nor formation of heterogeneous nanoparticles. Isolation of POM 363 from the reaction media verified that the structural integrity was maintained, confirming its stability under the catalytic reactions conditions. By following a strategy similar to that for the electrochemical studies, the authors repeated the experiments by adding a 10fold excess of bpy per Co atom. No significant changes in the presence or absence of bpy, in terms of TON or TOF, could be observed and indicates that the cobalt core in the POM remained intact. The successive addition of the oxidant to the reaction mixture, after termination of O2 generation, initiated O2 evolution again with essentially identical TONs and TOFs. The authors could further demonstrate the robustness of the nonnuclear POM by showing that solutions of 363 that were aged for several weeks remained active without any loss in catalytic activity or decomposition.474 Further work focused on attaching the nonanuclear POM 363 to amorphous carbon paste electrodes. By synthesizing an insoluble salt of the POM (the cesium salt), the authors were able to incorporate it into the solid-state matrix. Cyclic voltammetry of the POM-functionalized carbon electrodes gave rise to a wave for catalytic H2O oxidation, which could also be observed by the formation of gas bubbles (O2) on the electrode. H2O electrolysis with the functionalized electrodes could be achieved for more than 8 h, demonstrating the longterm stability of the POM electrodes under the catalytic conditions. However, the question that still remained was if the catalytic activity was caused by the POM-functionalized electrodes or by the formation of CoOx nanoparticles. This was addressed by performing the electrolysis at pH 1, where cobalt oxide is unstable. Under these conditions, 100 times higher current densities were obtained for the POM-functionalized electrodes than for the cobalt oxide functionalized electrodes, thus providing evidence that cobalt oxide formation did not adventitiously participate in the catalysis.475 A recent report by Han et al. also focused on the synthesis of a series of high-nuclearity cobalt-based POMs and their application in light-driven H2O oxidation. Here, one of the synthesized POMs was found to contain a Co4O4 cubane motif, which was structurally analogous to the core of the OEC.476 In a recent study, the group of Patzke reported on the synthesis and throughout characterization of a small series of isostructural tungstobismutate-based POM-WOCs comprising [{Co(H 2 O) 3 } 2 {CoBi 2 W 19 O 66 (OH) 4 }] 10− (364), [Co 2.5 (H2O)6{Bi2W19.5O66(OH)4}]8− (365), and [Mn1.5(H2O)6{Bi211973

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

W20.5O68(OH)2}]6− (366).477 The POMs were evaluated in visible-light-driven H2O oxidation using [Ru(bpy)3]2+ as the photosensitizer and S2O82− as the sacrificial electron acceptor in buffered systems. In particular, three criteria were emphasized in this study: (i) establishing SAR for POM core/shell construction, (ii) investigating the POM/photosensitizer interactions, and (iii) confirming the stability of the POMs under the catalytic conditions. Unexpectedly, only the Cobased POM 364 proved to be catalytically active, while no H2O oxidation occurred in the presence of Co POM 365 or Mn POM 366. This unique reactivity could not be ascribed to a higher stability of POM 364 under the employed catalytic conditions (20 mM Na2SiF6/NaHCO3 buffer, pH 5.8), as all POMs were found to be highly stable under these circumstances. Instead, the authors invoked subtle differences in the Co or Mn vacancies at the terminal positions of POMs 365 and 366 as an explanation for the lack of catalytic activity, where lower metal occupancies were proposed to translate into inactive POMs. Interestingly, the authors also found evidence that the catalytically active POM 364 formed a stable complex with the [Ru(bpy)3]2+ photosensitizer, which could be recycled in multiple O2 evolution experiments. 5.1.3. Other Metal-Based Polyoxometalates. In addition to the extensively studied ruthenium and cobalt POMs, there have been reports on manganese,478,479 molybdenum,480 iridium,481 and nickel-based POMs482 as active WOCs. As previously discussed in this Review, manganese constitutes perhaps the most attractive metal in WOC design because of its low cost, high natural abundance, and its relevance to the natural photosynthetic machinery. Unfortunately, the construction of functional Mn-based POMs capable of oxidizing H2O has for a long time remained a challenging topic. However, one example is the high-nuclearity POM [Mn19(OH)12(SiW10O37)6]34− (367), that was recently reported by the group of Kortz. It was isolated as a hydrated sodium salt, Na34[Mn19(OH)12(SiW10O37)6]·115H2O (Na367), and found to be a promising catalyst for electrochemical H2O oxidation.478 From single-crystal X-ray diffraction, it was found that POM 367 was comprised of a cationic {Mn19(OH)12}26+ cluster stabilized by six dilacunary [αSiW10O37]10− units, giving a structure with S6 point-group symmetry. All 19 MnII are positioned in the same plane and form a hexagonal structure consisting of edge-sharing MnO6 octahedra. By using BVS calculations, it was found that the MnII ions in the Mn19 unit were connected by 12 μ3-hydroxy bridges. The electrochemistry of POM 367 was assessed in an aqueous solution (pH 5) and showed two well-defined waves assigned to tungsten at −0.610 and −0.720 V vs NHE. Controlled potential coulometry provided support that already the reduced form of the first tungsten redox couple was active for H2 evolution. The first step of the oxidation of the MnII ions was found to proceed to the MnIV state, via MnIII. Controlled potential coulometry also indicated that all of the manganese centers were electro-active and in the +II oxidation state. Interestingly, POM 367 was found to give rise to a strong oxidation wave at more positive potential values, which was assigned to the oxidation of H2O. POM 367 also displayed kinetic parameters for O2 evolution from H2O,478 which compared favorably to that of previously reported Co-Pi catalyst (vide infra) and Ru-based POM 346. The high efficiency of POM 367 was ascribed to the multiple μ-hydroxo/ oxo bridging units connecting the adjacent metal centers, which

bear substantial structural resemblance with the CaMn4O4 cluster of the OEC.478 Following this study, the group of Kortz synthesized a lownuclearity Mn-based POM (368) consisting of three MnIII ions bridging two trilacunary [A-α-PW9O34)2]9− Kegging units in a sandwich-type assembly.479 Each MnIII ion is connected to two edge-shared WO6 octahedra at the lacunary sites of each Keggin unit via μ2-oxo bridges. Two of the manganese centers (Mn1 and Mn3) further coordinate to two trans-related and terminal H2O molecules in a distorted octahedral geometry, while the third manganese center (Mn2) is only linked to one H2O molecule in a square-pyramidal geometry. From electrochemical experiments conducted in aqueous solutions (pH 5), it was found that POM 368 was also a competent electrocatalyst for H2O oxidation when using potentials exceeding ∼1.29 V vs NHE. Interestingly, the group of Zhang decided to utilize the intermolecular interactions between light-absorbing units and POMs for the conctruction of supramolecular assemblies.480 Such interactions have been previously observed for different metal-based POMs.477,483 As a synthetic concept, three highly negatively charged POMs ([Mo6O19]2−, [Mo5S2O23]4−, and [Mo8O26]4−) with different sizes were combined with the positively charged ruthenium chromophore [Ru(phen)3]2+ to construct the three POM−PS hybrids 369−371, respectively. In these hybrids, the catalytically active metal responsible for mediating H2O oxidation is Mo. The ability of the synthesized POM−PS hybrids to oxidize H2O was investigated in a photocatalytic setup using persulfate as the sacrificial electron acceptor. It was shown that all of the hybrids were capable of evolving O2, with the following efficiency order 371 > 370 > 369. These results are in agreement with the number of terminal MoO sites in the utilized POMs480 and demonstrate that the use of POM−PS hybrids gives stable molecular WOCs, which could eventually be used for the development of new POM−PS type hybrids. The group of Hill reported the first structurally characterized iridium-based POM, [(IrCl4)KP2W20O72]14− (372).481 The crystal structure of the iridium POM revealed that the IrCl4 fragment was dangling and only connected to the polytungstate unit through two oxo groups (Figure 108). POM 372 was tested for its activity in H2O oxidation with the one-electron oxidant [Ru(bpy)3]3+ under neutral condition (pH 7.2). Quantification of the evolved O2 showed that POM 372 was indeed an active WOC.

Figure 108. Structure of the iridium-based POM [(IrCl 4 )KP2W20O72]14− 372. Adapted with permission from ref 481. Copyright 2009 American Chemical Society. 11974

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

nm as determined by HRTEM. By the use of SEM-EDX, the Ru loading could be estimated to ca. 0.5 wt %. Initial evaluation of the RuO2 nanocatalyst for H2O oxidation was conducted using CeIV as the chemical oxidant, and a Clark-type electrode was used for determination of the amount of evolved O2. Interestingly, close to stoichiometric amounts (97%) of the expected O2 level were observed after 2 h. Moreover, the RuO2 nanocatalyst displayed negligible decrease in rate upon five repeating inoculations with CeIV, indicating the stability of the catalytic system. The authors also demonstrated that the RuO2 nanoparticles could be photodeposited on other semiconductor materials, such as N-doped titania (Kronos VLP 7001), showing the generality of the synthetic method. The RuO2 deposited on N-doped titania was found to be an efficient catalyst for UV-light driven H2O oxidation in 0.1 M NaOH when using sodium persulfate as the sacrificial electron acceptor. The group of Ren has in a series of studies investigated the performance of RuO2 nanoparticles, confined within mesoporous SBA-15 silica support, in H2O oxidation catalysis. In their first study, they demonstrated that nanostructured RuO2/ SBA-15 could function as an efficient and robust WOC when driven by CeIV.491 Following this work, the authors showed that the RuO2/SBA-15 nanocatalyst, as well as an analogous version of the catalyst, prepared from a slightly improved synthetic procedure both worked as WOCs when driven photochemically by Ru(bpy)32+. Under optimal reaction conditions, the second nanocatalyst, synthesized via the improved synthetic method, achieved a high TOF of 0.027 s−1 with an O2 yield of 95%. This nanocatalyst also displayed impressive stability and performance in a recycling test, where it proved to be capable of evolving 86−95% of the theoretical amount of O2 over five repeating cycles.492 Recently, the group of Wang reported on the development of a NiRuOx coated n-type Si photoanode and its application as an efficient and stable catalytic system for electrochemically driven H2O oxidation.493 The NiRuOx coating, which was introduced using a sputtering technique, was found to have dual functions, where it in addition to being responsible for the catalytic activity also had a protective effect of the Si photoanode. Normally, unprotected Si-based photoanodes are not compatible with the conditions of H2O oxidation catalysis because their higher thermodynamic oxidation potential as compared to H2O makes them susceptible to side reactions that lead to passivation of the Si surface. The electrocatalytic evaluation of the NiRuOx coated n-type Si photoanode was carried out in a buffered aqueous solution under neutral conditions with a Xe lamp (100 mW cm−2) as the light source. Under these conditions, the NiRuOx coated n-type Si photoanode displayed highly promising electrocatalytic properties, including a low onset potential, a high on/off ratio, and excellent long-term stability, which all compared favorably to the other photoanode structures that were investigated in this study. Also, iridium oxides are active catalysts for generation of O2 from H2O. This was recently disclosed by Mallouk and coworkers, who interfaced iridium particles with a [Ru(bpy)3]2+type chromophore for photocatalytic H2O oxidation.494,495 The group used citric acid stabilized IrO2 nanoparticles that previously have been shown to be useful for obtaining IrO2 particles for electrocatalytic H2O oxidation.496,497 A functionalized ruthenium polypyridyl chromophore was prepared that contained phosphate groups for straightforward attachment to TiO2, and a malonate group for stabilization of the IrO2

A possible decomposition product of the iridium POM is IrO2, which is a well-known and efficient WOC. Control experiments with IrCl3 were therefore conducted and revealed that the catalytic activity of IrCl3 was slightly higher than for the iridium POM. Subsequent UV−vis spectroscopy experiments also indicated that complete dissociation (>99%) of POM 372 occurred in 24 h aged solutions. These findings further highlight the possibility of partial decomposition of iridiumbased WOCs to generate catalytically active IrO2 nanoparticles and the difficulties associated when studying these types of WOCs.481 Hill and co-workers have also presented a nickel-containing POM, [Ni5(OH)6(H2O)3(Si2W18O66)]12− (373). It was discovered that the addition of POM 373 to a solution containing either [Ru(bpy)3]3+ (in the chemical oxidations) or [Ru(bpy)3]2+ (in the light-driven oxidations) resulted in the immediate precipitation of a [Ru(bpy)3]n+-POM 373 salt. Although salt formation took place, the equilibrium between POM 373 in solution and in the [Ru(bpy)3]n+-POM 373 salt affirmed that O2 generation from POM 373 was viable.482 Under the catalytic experiments, the nickel POM might be subjected to decomposition, which could lead to the formation of catalytically active NiOx particles.484 To address this issue, the solution was filtered to remove the insoluble [Ru(bpy)3]n+POM salt 373 before the photocatalytic experiment, and this allowed the authors to verify that the supernatant solution was inactive (in O2 generation). Similar control experiments with Ni(NO3)2 confirmed that no nanoparticles were generated when POM 373 was used and that leaching of NiII from 373 did not occur prior to the catalytic experiments. DLS measurements, UV−vis spectroscopy, and IR all showed that the nickel POM 373 was stable in buffered aqueous solutions, prior to the addition of [Ru(bpy)3]n+. This suggests that POM 373 is stable under the catalytic conditions and does not give rise to a significant amount of nickel hydroxide nanoparticles.482 5.2. Metal Cubanes Based on Earth-Abundant Metals as Water Oxidation Catalysts

Catalysts for H2O oxidation constructed from metal oxides constitute an excellent complement to molecular WOCs containing organic ligand scaffolds.485,486 Metal oxides have tremendous potential as catalysts by being oxidatively inert toward decomposition under the harsh conditions required for H2O oxidation in contrast to molecular WOCs containing organic ligands. Colloidal metal oxide particles have been evaluated as active WOCs since the 1970s,487 especially those based on noble metals. In particular, iridium315,488 and ruthenium oxides489 are known to be highly efficient catalytic O2 evolving materials. H2O oxidation catalyzed by metal oxides is clearly a vast subject, which must be treated separately from this Review. However, because of the frequent suspicion that metal oxides may inadvertently be formed by decomposition of the homogeneous catalysts, we would like to add a brief discussion of oxide catalysts. One example of the use of nanostructured RuO2 for chemical and light-driven H2O oxidation was recently reported by Mills et al.490 The RuO2 nanocatalyst was prepared by using a photodeposition method, where an aqueous solution of K2RuO4 was stirred together with powdered TiO2 and irradiated with a Xe or Hg arc lamp. The irradiation triggered the photoreduction of the RuO42− to RuO2, which deposited on the TiO2 support as nanoparticles in the size range of 2−3 11975

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

transfer mediator. Illumination of TiO2 electrodes containing codeposited iridum-mediator assembly and ruthenium chromophore gave rise to a stable photocurrent, consistent with the occurrence of catalytic H2O oxidation. Flash photolysis experiments showed that the presence of the biomimetic electron-transfer mediator decreased the lifetime of the photooxidized ruthenium chromophore, demonstrating that efficient electron-transfer quenching occurred from the IrIV center to the photogenerated RuIII center of the chromophore. The acceleration of electron transfer between the iridium WOC and the photo-oxidized ruthenium chromophore and the ability of controlling these events should enable tuning and improvement of the H2O splitting efficiency in future devices for solar fuel production. An intriguing observation was the ability of succinic acidstabilized IrO2 nanoparticles to evolve O2 from aqueous solutions, in the absence of any chromophores.499 The processes for this catalytic system were thought to involve visible-light excitations among the Ir-d (t2g) and Ir-d (eg) bands. This discovery might be applicable in future catalytic systems for H2O splitting and could become applicable for other nanocrystalline transition metal oxides with valence and conduction band energies similar to those of IrO2. Although the use of iridium and ruthenium oxides gives potentially viable WOCs, these metals are among the least abundant in the Earth’s crust. The development of earthabundant, first-row transition metal WOCs that are capable of operating at neutral conditions (pH ∼7) and with a low overpotential is therefore a fundamental challenge for the scientific community. Recently, the group of Nocera reported that a cobalt film, generated by electrodeposition of neutral aqueous solutions containing CoII salts and phosphate ions, was an efficient WOC when anodic potentials were applied (vide infra). This highly important finding has resulted in revival of the use of metal oxides as catalysts for H2O splitting. Moreover, it strongly suggests that H 2 O oxidation catalyzed by homogeneous cobalt complexes in phosphate buffer must be interpreted with great caution. The following parts of this Review will cover and highlight recent breakthroughs in studies on cobalt- and manganesebased oxides as catalysts for H2O oxidation. Studying these systems is especially interesting because of the potential formation of oxides from homogeneous cobalt and manganese catalysts. 5.2.1. The Self-Assembling Cobalt−Phosphate (Co− Pi) Catalyst. Early work focusing on cobalt oxide materials includes the work by El Wakkad and Hickling,500 where the anodic deposition of cobalt oxide(s) was investigated. When starting with a clean anode deposited with a cobalt film, in neutral phosphate buffer solutions, they discovered that the initial dissolution of cobalt ions into solution could be reverted. This suggested that a self-repair mechanism occurred in phosphate buffer solutions. More importantly, but not further discussed by the authors, was the ability of the cobalt films to oxidize H2O to O2. Benson, Briggs, and Wynne-Jones also investigated the deposition of cobalt oxide/hydroxide films from alkaline solutions, but no investigation of the O2 evolution was reported in those studies.501,502 A major deactivation pathway for molecular cobalt catalysts is the precipitation of cobalt species from solution, which depletes the solution of the catalytically competent species. Although this process might be harmful for the catalytic activity, it has the potential of generating an electrodeposited catalytically active

particles (374, Figure 109). This tailored chromophore functioned as an effective stabilizer for IrO2 nanoparticles and

Figure 109. Schematic picture of the TiO2 electrode functionalized with the ruthenium chromophore−IrO2 nanoparticle assembly 374. Reprinted with permission from ref 494. Copyright 2009 American Chemical Society.

when this assembly was adsorbed onto TiO2, rapid electron injection into the semiconductor could be observed. The functionalized TiO2 electrode was used in a PEC for overall H2O splitting. By using a platinum electrode, this cell was capable of producing O2 and H2 at the anode and cathode of the cell, respectively. However, the cell suffered from a decay in photocurrent, which was most likely due to competitive oxidative decomposition of the photo-oxidized chromophore. Tuning of the rates of electron-transfer processes occurring in the cell should enable the construction of more efficient devices based on this concept.494,495 The redox mediator effect134 that was observed for the ruthenium-based catalytic systems has also been employed in the biomimetic system 375, containing a nanoparticulate IrOxbased WOC (Figure 110).498 On the basis of principles similar to those for the molecular ruthenium−manganese assembly 296, a tyrosine−histidine unit was incorporated as an electron-

Figure 110. Electronic processes in the biomimetic electron-transfer mediator assembly 375. 11976

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

material in situ, thus resulting in a catalyst liable to self-repair. However, the potential of CoOx materials as WOCs was not realized until 2008. That year, the group of Nocera reported that applying a potential to aqueous phosphate buffer solutions containing simple CoII salts resulted in the in situ formation of an electrodeposited film (i.e., the Co−Pi catalyst), which was highly active towards catalyzing H2O oxidation.503 This catalyst has since then received much attention due to its ability to selfassemble under neutral conditions, its self-repair mechanism, and the fact that the catalyst is comprised of low-cost and earthabundant materials. Electrochemical measurements displayed an oxidation wave at 1.13 V vs NHE, which was followed by a strong catalytic current attributed to catalytic H2O oxidation, with an onset potential of 1.23 V. When an ITO electrode was subjected to a phosphate buffer solution containing CoII, the formation of a dark film could be observed during bulk electrolysis. The morphology of the electrodeposited material was investigated by means of SEM and PXRD, and revealed that it was amorphous and consisted of coalesced particles of micrometer size on top of the film. The deposited film exhibited maximal catalytic activity at a thickness of >2 μm, where prolonged time of electrolysis led to increased film thickness. EDX analysis was also performed to determine the elemental composition of the deposited material. Cobalt, phosphorus, and potassium were found to be the principal components. The Tafel plot indicated that the catalytic film exhibited a Nernstian behavior in the range 5 < pH < 8, with a decrease in the overpotential by 59 mV per pH unit. This suggested the existence of catalytic species that underwent a PCET event, probably via HPO42−, which preceded the generation of O2. This Co−Pi catalyst is extraordinary in the sense that it is comprised of only inorganic, earth-abundant materials, works at near-neutral pH, and is spontaneously self-assembled under the environment required for H2O oxidation.503 The self-assembling process was studied in different electrolytes, and it could be shown that catalyst growth was not solely limited to the use of phosphate buffer solutions.504,505 In fact, electrodeposition of the cobalt catalyst took place in both methylphosphonate (MePi) and borate (Bi) electrolytes. Subjecting these systems to a single electrochemical scan, with subsequent removal and rinsing of the electrodes, and placing them in fresh electrolyte solutions (in the absence of CoII) resulted in a sustained catalytic wave. This confirmed that electrodeposition of a catalytically active species occurred at modest potentials. SEM measurements of the catalytic films generated from the different electrolytes showed that progressively thicker films were obtained upon prolonged electrolysis. As with the Co−Pi catalyst, these films were amorphous in nature and displayed no detectable crystallites. Evaluating the activity by construction of Tafel plots of the different catalytic films unveiled that the films exhibited similar slopes, which implied that all of the examined electrolytes were equally effective at shuffling protons during the catalysis. The proton-accepting ability of the electrolyte was further examined by the use of electrolytes based on SO42−, NO3−, and ClO4−, that is, nonbuffering electrolytes, and it could be concluded that these poor proton acceptor electrolytes suppressed catalyst function and that a good PCET environment was essential for the catalyst growth. It was also concluded that the Co−Pi system maintained a stable catalytic current during electrolysis, in contrast to the catalyst operating in a sulfate-based electrolyte.504

Performing the electrolysis in an electrolyte containing 0.5 M NaCl or seawater gave sustained catalytic currents, demonstrating that chloride anions did not inhibit the catalytic activity of the O2 evolving films. The electrolyte is of significance for the formation of an active catalyst film, presumably through the maintenance of a stable pH environment with no proton buildup. In nonbuffering electrolytes, preservation of bulk pH is reached by protonation of the Co−Pi catalyst, due to the lack of other bases, and induces corrosion of the otherwise robust and functional catalyst.504 The ability of the Co−Pi catalyst to undergo self-assembly and self-repair was evaluated by radioactive labeling experiments, in which electrodeposition of the Co−Pi was performed with radiolabeled 57CoII ions.506 These experiments revealed that leaching of 57CoII ions occurred from the deposited film. When keeping the electrode at an open circuit potential, ∼1.5% of the labeled 57Co could be detected in solution (after 39 h), and the release occurred continuously from the catalyst film. However, when the potential was kept at 1.30 V, no cobalt was detected in the electrolyte. In contrast, terminating the potential bias (after 4 h) resulted in leaching of cobalt into the reaction solution. The dissolved cobalt could be reabsorbed by reapplying a potential at the electrode (at 15 and 25 h), and after 14 h of continuous bias application only 0.002% dissolved cobalt could be detected. This shows that in the absence of an applied potential the electrodeposited film is slowly depleted of CoII. However, when a high potential (1.30 V) is reapplied, the catalyst film is regenerated. The other major component of the catalyst, phosphate, was also monitored by employing labeled 32Pphosphate. Catalytic films, deposited from labeled 32P solutions, which were kept at an open circuit potential displayed twice as much leaching of 32P-phosphate as films that were held at a potential of 1.30 V. Also, the rate of phosphate reabsorption was higher in open circuit potential films. The films were also composed of an alkali cation (Na+ or K+), and it was discovered that alkali cation exchange was significantly faster (>90% exchange after 10 min) than phosphate exchange. These data indicate that phosphate is coordinated to cobalt centers and that these centers are engaged in a more substantial metal− oxygen catalyst framework (due to the observed higher exchange rate of phosphate, as compared to cobalt). When a potential was applied to solutions based on nonprotonaccepting electrolytes, cobalt was released and corrosion occurred. Increasing the potential resulted in increased corrosion, which stresses the importance of having a protonaccepting electrolyte to mediate a self-repair mechanism for the Co−Pi catalyst. While different oxidation states might require disparate coordination environments to be stabilized, a selfrepair mechanism could potentially result in a more viable catalytic system.506 Knowledge of the exact chemical structure and the oxidation states of a certain catalyst during catalysis is essential for understanding the mechanism of a given reaction and would also enable the study of structure−activity relationships. To resolve the structural issue for the Co−Pi catalyst, several important papers have been published.507−511 Although there has been some debate regarding the exact structure of the Co− Pi catalyst film, the two independent studies based on XAS and extended X-ray absorption fine structure (EXAFS) measurements indicated that the catalyst film contained discrete cobalt−oxo structural motifs. These consisted of edge-sharing CoO6 octahedra with a mean cobalt valency of +III (Figure 11977

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

2.27, for the first film prepared at 1.03 V where H2O oxidation is expected to be negligible. The remaining films, which were subjected to higher potentials, had significant decreases in the CoII signal while the intensity of the signal at geff = 2.27 was increased, as the potential was increased. The increase in signal intensity at geff = 2.27 and its resemblance to the cubane model complex ([Co4O4(C5H5N)4(OAc)4](ClO4)) where one of the cobalt centers is in the high-valent CoIV oxidation state strongly indicated that this characteristic signal belonged to an electrochemically generated CoIV species. These results highlight that a high-valent CoIV species might indeed be generated and involved in the catalytic cycle of H2O oxidation.512 Mechanistic studies of the Co−Pi catalyst, by means of electrokinetic measurements, have been fruitful and provided a proposed pathway by which the Co−Pi oxidizes H2O to O2.513−515 Tafel plots were constructed for Co−Pi catalyst films with different thickness and all proved to be linear over 3 orders of magnitude. All films exhibited Tafel slopes of ∼60 mV/ decade, which suggested a single mechanistic pathway consisting of a reversible one-electron transfer prior to the chemical rate-limiting step. From Tafel plots, obtained from Co−Pi films deposited on Pt rotating disk electrodes, it was concluded that the reaction was not limited by mass transport to the catalyst films (over the studied current/potential range).513,514 The examined films exhibited TOFs of ∼2 × 10−3 s−1 at an overpotential of 410 mV. However, because the majority of the cobalt centers are expected to contribute solely to the structural integrity of the film, the catalytic activity is underestimated as it assumes that all cobalt centers in the deposited film are equally catalytically active. The pH dependence of the catalytic activity was investigated by the use of both potentiostatic and galvanostatic techniques and revealed an inverse first-order relationship to the proton activity, which is in accordance with loss of a single proton in a pre-equilibrium step preceding the rate-limiting step. Although no dependence on phosphate was revealed, it is thus likely that phosphate is involved as a proton acceptor in PCET mediated processes. Electrochemical measurements indicated that the predominant oxidation state of the cobalt centers presumably was CoIII or higher during catalysis. From the electrokinetic measurements and previous studies of the Co−Pi catalyst, the resting state of the catalyst is most likely comprised of a mixed-valence oxidation state (CoIIICoIV). Oxidation of this mixed-valence cluster involves a one-electron PCET event, where the phosphate has a crucial role in mediating rapid proton shuffling. The CoIV species participates in a proton-independent step that generates the O−O bond and ultimately results in liberation of O2 (Figure

111). The data could also exclude the presence of direct cobalt−phosphorus bonds, thus ruling out phosphate oxygens

Figure 111. Structure of the Co−Pi catalyst with its edge-sharing molecular cobaltate cluster. Bridging oxo/hydroxo ligands are shown in gray; nonbridging oxygen ligands (water, hydroxide, or phosphate) complete the octahedral coordination geometry of each peripheral Co ion and are shown in light gray. Adapted with permission from ref 508. Copyright 2010 American Chemical Society.

as μ-oxo bridges between two cobalt centers. However, phosphate coordination to the outer parts of the cobalt−oxo network could not be excluded. The structural studies of the Co−Pi film suggest that it consists of molecular-sized clusters of ∼14 cobalt atoms with a cobalt cubane-type structural motif. The structural analogy of the cubane-type structure (M4O4) of the Co−Pi catalyst to the tetranuclear manganese cluster in the natural photosynthetic system is intriguing and suggests a bridging of the gap between biological and artificial systems. This unification could potentially open for the development of a novel generation of WOCs with increased efficiency. Further studies lead to the proposal that a CoIV intermediate is present in the cycle of electrocatalytic oxidation of H2O.512 EPR measurements on the Co−Pi catalyst film displayed a broad resonance at geff ≈ 5 and at geff = 2.27. The resonance at geff ≈ 5 is a characteristic feature of other CoII species, such as Co3O4 and Co3(PO4)2, and was assigned to a CoII species in the catalytic films, with S = 3/2. As previously discussed, the Co−Pi is proposed to consist of cobalt−oxido clusters in a CoO6 octahedral conformation. This arrangement should favor a low-spin CoIII species, which cannot be observed by EPR due to its diamagnetism. Upon subsequent oxidation, a CoIV species should be generated with S = 1/2. Films that had been subjected to potentials of 1.03, 1.14, and 1.34 V vs NHE were examined by EPR. The characteristic CoII feature was observed, in conjunction with a small signal at geff =

Figure 112. Proposed H2O oxidation pathway by the Co−Pi catalyst, where a PCET event is proceeded by a rate-limiting O−O bond-forming step. Curved lines denote phosphate, or OH, terminal or bridging ligands. Reprinted with permission from ref 514. Copyright 2010 American Chemical Society. 11978

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

112).514 A soluble model complex of the CoPi catalyst has also been developed and studied by the group of Nocera.516 Theoretical studies of the H2O oxidation mechanism revealed that a probable catalytic pathway proceeds via initial oxidation of two adjacent CoIII−OH moieties to generate two CoIV−oxo centers by PCET. The formed CoIV−oxos then undergo a direct coupling to create an O−O bond, which proceeds through a reaction pathway with a low kinetic barrier and a strong thermodynamic driving force. Further PCET processes led to O2 evolution and afford a viable pathway for catalytic turnover.431 A similar PCET effect was also found in related nickel−borate films.517,518 In addition to the widely used technique of electrodepositing the Co−Pi catalyst from aqueous solutions of CoII salts, an interesting and alternative approach has been developed where the Co−Pi can be deposited on nonconductive glass material by sputtering of cobalt metal.519 Evaluation of the catalytic films revealed that no differences existed between the sputtered films and films generated by the “conventional” electrodeposition method. This sputtering method allows the direct use of cobalt metal, eliminating the requirement of using aqueous solutions for materials that otherwise would be subjected to corrosion. Yet another approach that has been reported focuses on the in situ formation of CoOx species by the use of cobalt-containing polyoxotitanates as reservoirs for cobalt ions.520 Several different methods have been used to anchor the Co− Pi catalyst to different photoanodic materials. Hematite (αFe2O3) holds great potential as a photoanode and offers several benefits, such as low cost, robustness under oxidative conditions, environmental nontoxicity, and ability to absorb visible light. Photocatalytic H2O oxidation by pure hematite does not occur to a significant extent as a result of slow kinetics. For this reason, tandem PEC-photovoltaic (PV) devices have been envisioned to provide the necessary input of energy needed to overcome these obstacles. Gamelin and co-workers interfaced the Co−Pi catalyst with hematite to achieve photoelectrochemical H2O splitting.521,522 They found that the electrodeposited α-Fe2O3 PECs were highly dependent on several parameters, such as the surface morphology, silicon doping level, growth temperature, and method of deposition.523 In these composite materials, the photogenerated holes in the α-Fe2O3 are thus transferred and stored as oxidizing equivalents in the Co−Pi catalyst, which subsequently oxidizes H2O to O2 (Scheme 24). The photogenerated electrons, which are being abstracted from H2O, are transferred to a catalyst, such as Pt, for proton reduction. It could be concluded from the differences between the dark and photocurrrent responses that the manufactured Co-Pi/αFe2O3 photoanodes reduced the bias voltage by >350 mV in

the light-driven H2O oxidation. Electronic absorption spectra showed that the Co-Pi/α-Fe2O3 photoanodes absorb light throughout the whole visible spectral region and display a maximum incident photon-to-current efficiency (IPCE) of 18% at 450 nm. Further measurements revealed that the composite materials could also be used in salt water under mild conditions (pH 8) without any loss of activity. Several control experiments were performed to exclude the possibility that the measured photocurrents were derived from decomposition or other unwanted processes: (i) the addition of CoII ions to the electrolyte did not have any effect on the photocurrent, (ii) replacing the PEC electrolyte solutions with fresh solutions did not change the photocurrents or induce any induction period, and (iii) photoelectrochemical measurements at 1.00 V were unaltered for >10 h.521 An alternative approach was demonstrated by Choi and Steinmiller, where the Co−Pi catalyst was photodeposited on ZnO photoanodes. This resulted in anodes that were sufficiently chemically and photochemically stable to allow generation of photocurrents upon illumination. This study highlights a general method for photodeposition of WOCs for electron transfer to photogenerated holes in ZnO to realize light-driven H2O oxidation.524 Silicon provides an excellent alternative to metal−oxide semiconductor materials and is widely used in photovoltaic systems. A variety of attractive properties are associated with the use of silicon: (i) it absorbs a substantial fraction of the solar spectrum, (ii) it is associated with low losses in bulk and interfacial carrier transfer, and (iii) its low cost and natural abundance. Deposition of the Co−Pi catalyst onto a siliconbased electrode for the construction of integrated photoanodes resulted in a significant photocurrent generation, and allowed sufficient lowering of the onset potentials for carrying out catalytic H2O oxidation. More importantly, the functionalized electrodes could be operated at neutral pH, thus resulting in long-term stability of the developed devices.525,526 The Co−Pi catalyst has also been deposited on silicon photoanodes for use in wireless solar H2O splitting. The developed devices were capable of accomplishing 2.5% efficiency upon illumination with 1 sun.527 This wireless approach might open new avenues of exploration and result in the development of inexpensive solar-to-fuels photosystems. These results clearly show that by integrating the Co−Pi catalyst with light-absorbing semiconductors, photoanodes with reduced external power demands can be obtained, which are active under neutral conditions. The developed photoanodes may thus offer strategies for a sustainable and efficient production of solar fuels. In general, the Co−Pi catalyst contains the essential elements of the OEC in the natural photosynthetic machinery. It is comprised of an earth-abundant transition metal and has the ability to undergo self-assembly, and self-healing at neutral conditions, and to operate in natural H2O, as well as seawater, which demonstrates its potential as a WOC for use in future H2O splitting devices.528 5.2.2. Cobalt Cubanes and Related Cobalt Oxide Materials as Oxygen Evolving Catalysts. The discovery that the cubane core, M4O4, is an active structural component in H2O oxidation catalysis and the fact that it exhibits a remarkable resemblance to the catalytic core of the OEC has triggered substantial research toward developing catalysts based on this important motif. Although there are some early examples of cobalt cubanes, including the two complexes Co 4 O 4 (pyr) 4 (OAc) 4 (376) 5 2 9 and [Co 4 O 4 (bpy) 4 -

Scheme 24. Overview of the Processes Occurring in the Co− Pi/α-Fe2O3 Photoanodesa

a

Reprinted with permission from ref 521. Copyright 2009 American Chemical Society. 11979

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(OAc)2]2+,530,531 experiments related to catalytic H 2 O oxidation were not performed at that time. However, the groups of Dismukes and Sartorel investigated the cubane complex Co4O4(pyr)4(OAc)4 (376), and its activity in H2O oxidation.532−534 The complex was easily prepared from the simple building blocks Co(NO3)2, sodium acetate (NaOAc), and pyridine. Cubane 376 exhibits D2d point-group symmetry and is comprised of a [Co4O4]4+ core with four bidentate acetate ligands and four monodentate pyridine ligands (Figure 113). In acetonitrile, a reversible one-electron

conditions and that [Ru(bpy)3]2+ photodecomposition limited the catalytic TON measurements.532 The core in cobalt cubane 376 is related to the catalytically active core in the Co−Pi catalyst. Detailed electronic structural studies in combination with DFT calculations on the oneelectron oxidized intermediate of cubane 376, [Co4O4(pyr)4(OAc)4]+ (376+), revealed a highly delocalized unpaired electron in 376+.535 By combining several magnetic measurements and calculations, the authors were able to convincingly show that the one-electron oxidized 376+ has an unpaired electron in the SOMO that is almost equally shared between the four cobalt centers and the four μ-oxos of the cuboidal core, resulting in a formal [Co+3.1254O4] species. As a result of this, the octahedral cobalt centers adopt a low-spin electron configuration with delocalization of the electron hole over the whole cuboidal core of 376+. Further kinetic studies of the cobalt cubane 376 resulted in the identification of distinct mechanisms for the PCET event for the oxidation of CoIII−OH to CoIV−O.536 The Pourbaix diagram for the CoIV/CoIII conversion at low pH values (4), this redox process was pHindependent, and thus no proton was removed upon oxidation. Extraction of the rate constants, k, for the oxidation process and use of kinetic isotope effects enabled the construction of two different mechanistic proposals for the redox process. At pH 1, no observable difference in the kPCET was noted, which is consistent with a stepwise PCET mechanism, where an equilibrium proton transfer is followed by rate-limiting electron transfer, that is, a proton−electron-transfer (PET) mechanism. From NMR spectroscopy and stopped-flow experiments, it could be concluded that the unidirectional pathway, when the electron and proton are transferred to the same molecule, is a concerted PCET, whereas the bidirectional pathway is a stepwise PCET process involving the transfer of electron and proton to different molecules. This work demonstrated that in addition to the importance of PCET in activating H2O, PCET is also important for diffusion of charge through the Co−Pi catalyst. Harriman and co-workers showed already in 1988 that spinel-type Co3O4 was an active catalyst for H2O oxidation.305a More recently, spinel-type Co3O4 was studied by Singh and coworkers,537 and somewhat later, Jiao and Frei explored the use of nanostructured Co3O4 clusters embedded in mesoporous SBA-15 silica support538 as candidates for efficient WOCs.539 The TEM images of these materials confirmed that the integrity of the silica channel structure was maintained in the functionalized materials. The Co3O4 clusters were found to consist of parallel bundles of nanorods with a structure regulated by the silica channels. When the silica network was removed, the structure of the Co3O4 clusters was maintained. The use of selected area electron diffraction (SAED) made it possible to assess the crystalline nature of the Co3 O 4 nanoclusters, and a combination of PXRD and EXAFS confirmed that the developed nanoclusters exhibited a Co3O4 spinel structure.539 The O2 evolution activity was established in a photocatalytic system, comprised of [Ru(bpy)3]2+/persulfate, which revealed that the mesoporous materials indeed could function as competent WOCs. Readdition of photosensitizer and persulfate to the system after O2 generation had ceased restarted O2 evolution and highlighted the robustness of these nanoclusters.

Figure 113. Structure of the cobalt cubane Co4O4(pyr)4(OAc)4 376.

redox process was observed at 0.91 V vs NHE, which was assigned to the formal generation of [Co3IIICoIV]. Performing the electrochemical experiments in aqueous solutions at pH 6.8 produced a catalytic current corresponding to electrocatalytic H2O oxidation.532 The catalytic activity of cubane 376 in H2O oxidation was studied photochemically using the [Ru(bpy)3]2+/S2O82−-photosystem. Within this system, the cobalt cubane 376 was capable of oxidizing H2O to O2, with a TON of 40 after 1 h of illumination. Unfortunately, facile photodecomposition of the [Ru(bpy)3]2+ chromophore was observed under the employed conditions, which limited the reaction time to 1 h. Moreover, the specific kinetics of the separate components of this photosystem was complex and prevented the authors from pursuing any further mechanistic studies. Control experiments verified that no O2 was evolved if any of the individual components (photons, [Ru(bpy)3]2+, persulfate, or catalyst) was excluded from the reaction solution. 1H NMR spectroscopic studies on the pre- and postreaction solutions revealed only 1.0 V as compared to the manganese-only containing complex 379.606 This highlights that the presence of a nonredox-active metal center in metal complexes can facilitate the formation of species of higher oxidation states, thereby leading to new accessible catalytic pathways for carrying out novel transformations. In subsequent work, a series of tetranuclear mixed-metal complexes were prepared, housing three manganese centers, [MMn3(μ4-O)(μ2-O)] (where M = Na, Ca, Sr, Zn, Y; Figure 119). All of the clusters contain a redox-inactive metal center that is a central part in these multimetal complexes.607 A linear dependence, with a slope of ∼100 mV per pKa unit, was observed between the Lewis acidity (pKa values) of the

Figure 118. X-ray structures of (a) complex 378 containing the [Mn 3 IV CaO 4 ] 6+ core and (b) complex 379 containing the [Mn2IVMn2IIIO4]6+ structural motif. Thermal ellipsoids are drawn at 50% probability. Hydrogen atoms and solvent molecules are not shown for clarity. Adapted with permission from ref 606. Copyright 2011 American Association for the Advancement of Science (AAAS).

Figure 119. Structures of the tetrametallic trimanganese dioxido complexes [Mn3M(μ4-O)(μ2-O)]. Reprinted with permission from ref 607. Copyright 2013 Macmillan Publishers Ltd. 11985

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

redox-inactive metal in these trimaganese complexes and the redox potentials. It was revealed that the stronger Lewis acids had a higher redox potential, which is anticipated because they are expected to withdraw more electron density from the oxido ligands. As a result, they make the manganese centers less electron-rich and cause a destabilization of the higher valence states of the manganese atoms. An interesting observation was that the Sr and Ca metal complexes had similar redox potentials, a behavior that correlates with previous studies on the OEC.607 Also, heterometallic complexes of the type [MMn3O4] (M = Sr2+, Zn2+, Sc3+, Y3+) were prepared, containing a structural motif that was more structurally relevant to the OEC than the core in the previously synthesized dioxido complexes (Figures 120 and 121).608

These intriguing results indicated that tuning of these mixed metal−oxide clusters, and related systems, can be achieved by simple substitution of a redox-inactive metal center. Modifying the nature of this redox-inactive metal center without causing a structural alteration of the clusters permits a significant degree of thermodynamic tuning and access to new reactivity patterns. These studies also highlight the potential involvement of calcium as a redox-modulating metal center in the OEC for efficient O2 production from H2O.610 The structural relevance of these artificial metal−oxido clusters to the active site in the OEC provides a valuable foundation for the development and insight into future biologically relevant or heterogeneous metal−oxide materials for H2O splitting. A paper that contributes to a better understanding of H2O oxidation catalysis by manganese oxide materials was recently published by the group of Dau and Driess.611,612 In this report, the authors described that highly active managense oxide WOCs can be easily synthesized from well-defined, but catalytically inactive, nanostructured manganese(II) oxide particles through the partial oxidation (corrosion) by CAN. The nanostructured manganese(II) oxide precursor was synthesized from Mn(OAc)2 in the presence of tri-noctylphosphine oxide and benzylamine (Scheme 25). This

Figure 120. Structures of the [CaMn4O5] active site (mirror image) in the OEC (left), the previously studied dioxido complexes [MMn3O2] (middle), and the structurally more relevant [MMn3O4] cubane-type complexes (right).

Scheme 25. Synthesis of the Active MnOx WOC Material through Partial Oxidation (Corrosion) of the Nanostructured MnO with CAN as Oxidant

It was of interest to see if these newly developed cubane-type complexes exhibited a correlation between the Lewis acidity of the redox-inert metal and the redox potential. Plotting the Lewis acidity of the redox-inactive metal against the redox potentials for the [MMn 3 IV O 4 ] complexes revealed a correlation, with a slope of ∼100 mV per pKa unit. Additionally, it could be shown that the presence of a higher oxido content (cf., [MMn3O2] and [MMn3O4]) led to a remarkable negative shift of the redox potentials. For the cubane [MMn3O4] complexes, the redox potentials were decreased by almost 1000 mV as compared to the previously characterized dioxido complexes [MMn3O2].608

well-defined manganese(II) oxide was evaluated as a WOC in chemically driven H2O oxidation with CAN and resulted in evolution of O2. However, close inspection of the initial O2 evolution kinetics revealed an induction period before O2 was produced, which suggested that the initially employed nanostructured manganese oxide material was transformed in situ

Figure 121. Structures of the cubane-type heterometallic [MMn3O4] complexes. Reprinted with permission from ref 608. Copyright 2013 The National Academy of Sciences. 11986

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

In a broader perspective, this work also highlights the obstacles and pitfalls when engineering new WOCs. Care must be taken in the attempts to determine the structure of the catalytically active species, which might not necessarily be the initially characterized complex. Instead, the starting complex might only function as a precursor that is transformed in situ, thus making the identification of the real catalyst challenging. In this regard, it is essential to apply multiple experimental approaches and techniques in the investigation of newly engineered WOCs. Another important conclusion that can be drawn from these studies is that subtle differences in the reaction conditions may cause alterations in the mechanistic pathway of a specific catalytic reaction.

into a catalytically active MnOx material. The as-prepared MnOx material (through corrosion) did not display a lag phase, implying that the material was sufficiently active to promote H2O oxidation and that the observed induction time for the catalytically inactive MnO precursor can be attributed to the time required to transform the inactive precursor to the active MnOx WOC. Detailed analysis by use of several different techniques, such as EDX, EXAFS, TEM, SEM, XPS, PXRD, and nitrogen adsorption and desorption measurements, revealed that there occurs a significant structural change when the precursor is transformed into the active state. TEM and SEM images of the two manganese oxide materials showed that the precursor is comprised of crystalline cubical-shaped particles of 40−80 nm in size with a surface that is coated with an amorphous layer. Treating the precursor with CAN generated a material consisting of amorphous hollow cubes of MnOx. PXRD measurements showed that the crystallinity was lost when the initial manganese(II) nanoparticles were treated with CAN. According to nitrogen adsorption and desorption measurements, there was also a remarkable increase of the surface area, from 28 to 88 m2 g−1. XPS studies also revealed an increase in the Mn 2p3/2 and 2p1/2 binding energies, suggesting an enhancement of the mean oxidation state of the manganese from +2 to +2.5. EXAFS analysis confirmed that treatment with CAN resulted in severe structural rearrangement of the initial manganese oxide. The active MnOx catalyst was found to be comprised of MnIII and MnIV ions linked by di-μ-oxido bridges (edgesharing MnO6 octahedra),611,612 which strongly resembles the structures of the previously reported birnessite-like calcium manganese oxide WOCs.598,610 On the basis of these similarities, a structural model was proposed for the catalytically active MnOx WOC, which consists of layers of bridged di-μ-oxido manganese with H2O molecules and cations occupying the interlayer space. However, in contrast to calcium manganese oxide WOCs, the interlayer calcium cations are replaced by manganese ions, which thus appear to have the same function as the calcium ions in the mixed calcium manganese oxides (Figure 122).611,612 This suggests that the mixed valence character in the developed MnOx catalyst in combination with the high degree of structural disorder could provide the required structural flexibility for H2O oxidation catalysis.

6. CONCLUSIONS The projected future energy crisis is likely to constitute a severe threat to the continued growth of our society. In this perspective, there exists a major demand for the production of sustainable and carbon-neutral energy that can cover the needs of future generations. Among the available technologies, the utilization of solar energy to split H2O holds great potential for supplying the needed amount of storable fuels to prevent a future energy crisis. Mimicking the activity of the OEC with an artificial WOC constitutes a significant scientific challenge, which will require a multidisciplinary approach that bridges the knowledge of numerous research fields. Synchronizing photochemical charge separation to H2O oxidation requires the successive transfer of four electrons from the WOC, which is definitely not an easy task. In this Review, we have examined the efforts to develop robust and efficient catalysts for H2O oxidation, to provide the necessary reducing equivalents for the production of solar fuels. Particular attention has been devoted to the design of active artificial WOCs and recognizing the criticial elements in such designs. This is now a tremendously active area of research, which includes practitioners from a wide array of disciplines. As a result, substantial progress has been made in understanding the underlying chemical principles of catalytic H2O oxidation, which has led to the development of new WOCs. Unfortunately, the efficiency of the constructed catalysts thus far has not reached the level that is needed for the incorporation into a commercial cell for solar fuel generation. However, in contrast to nature, which is restricted to using relatively unstable protein frameworks and earthabundant metals, mankind has access to more durable materials, which certainly increases the possibilities for constructing more efficient artificial WOCs. During three decades of research, a great variety of strategies have been explored for the construction of WOCs, ranging from molecular complexes to different kinds of metal−oxo mimics. Molecular catalysts have the advantage of offering straightforward design, synthesis, and characterization. The ligand scaffold of molecular catalysts also enables simple tuning, making them amenable for structure−activity relationship studies. Because of this, the research on homogeneous molecular complexes has been essential for the current understanding of the mechanistic principles behind H2O oxidation catalysis, both for the natural and for artificial systems. Among the molecular WOCs, those based on ruthenium have shown the greatest promise by exhibiting high catalytic efficiencies and good compatibility with photosensitizer-type oxidants (which is a prerequisite for light-driven protocols). However, the cost and scarcity of ruthenium could

Figure 122. Structure of the crystalline MnO precursor and the active MnOx catalyst. Right: The Mn atoms in violet form a defect-rich layer of di-μ-oxido bridged MnIII,IV ions, where the MnII,III ions in green are not part of this layer and may interconnect layer fragments. In the previously reported calcium manganese oxides, calcium ions occupy the interlayer positions indicated in green. Reprinted with permission from ref 611. Copyright 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. 11987

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

AUTHOR INFORMATION

potentially hinder its application in the future, and, consequently, significant work has been dedicated to the development of catalytic systems based on earth-abundant metals, such as manganese, iron, and cobalt. Interesting complements to molecular WOCs are the inorganic metal oxide materials, which can either be accessed directly through rational synthesis or be generated in situ from the degradation of homogeneous precursors. These materials possess high thermal and redox stability coupled with the stable current density under the operating reaction conditions, which makes them promising catalytic components for incorporation into viable integrated solar fuel systems. Recent work on these materials has shown that a cubical oxide core appears to be an important structural feature in a range of developed WOCs that are comprised of 3d transition metals, including the cobalt and manganese cubanes and the spinel-phase materials. These reports suggest that this topology has a key role in H2O oxidation catalysis, both in the natural photosynthetic system and in bioinspired mimics. An advantage of these metal oxide materials is that the active catalytic species can be formed in situ and enables the catalyst to be deposited on a variety of conducting materials. Furthermore, these catalysts can be promoted into undergoing self-repair, thus resulting in dynamic equilibrium with constant buildup of the catalytically active species, in analogy with the natural photosystem. The recent breakthroughs with cobalt and manganese oxides have resulted in WOCs that have the potential to operate under mild conditions, in terms of both pH and temperature. As a result, these newly developed WOC materials constitute an important step toward converting solar energy into storable fuels. Development of novel procedures for producing metal oxide materials is therefore important because this could yield materials with lower overpotentials and ability to self-assemble. Here, the recently developed technique613 by the group of Berlinguette and Trudel might open new pathways for producing low cost and highly active metal oxide materials614,615 for H2O oxidation catalysis or, even more excitingly, for manufacturing of hybrid metal materials with unique catalytic activities.616−618 Integration of WOCs into photovoltaic cells seizes the basic elements of energy conversion and storage by an “artificial leaf”.619 Although a plethora of molecular catalysts and materials have been reported to produce O2 and/or H2, the overall efficiency when these catalysts and/or materials are combined into a single device for splitting of H2O is still rather low. This splitting of H2O requires interfacing several crucial events, such as light sensitization, charge separation, oxidation of H2O, and reduction of the generated protons to H2. At present, this interfacing is not sufficiently efficient to make H2O splitting, using solar energy, the entry to a sustainable and carbon-neutral future. Moreover, prediction of molecular structures and/or oxide materials that have the possibility to catalyze the four-electron oxidation of H2O at low overpotentials continues to be intricate and thus constitutes a major obstacle. Gaining fundamental understanding of structure− activity relationships will certainly offer the key to developing more efficient WOCs and realizing the construction of an “artificial leaf” for efficient solar energy conversion.

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Notes

The authors declare no competing financial interest. Biographies

Markus D. Kärkäs obtained his Master of Science degree in Chemistry in 2008 from Stockholm University, where he conducted undergraduate research in the laboratory of Professor Åkermark. In the same year, he started his graduate studies under the direction of Professors Björn Åkermark and Jan-Erling Bäckvall. Markus received his Ph.D. degree in 2013 after conducting research focusing on the development and mechanistic insight of artificial water oxidation catalysts. Dr. Kärkäs is currently working as a researcher in the group of Professor Åkermark and is engaged in the studies and development of homogeneous, as well as heterogeneous, water oxidation catalysts.

Oscar Verho earned his Masters degree in 2010 at the Department of Organic Chemistry, Stockholm University, Sweden. He soon joined the group of Professor Jan-Erling Bäckvall as a graduate student, from which he recently graduated. During his Ph.D., he worked on both homogeneous and heterogeneous transition-metal catalysis. In addition to this, he has been involved in several collaboration projects with the group of Professor Björn Åkermark, which has aimed at developing new water oxidation catalysts and applying them for organic transformations. 11988

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

the Bror Holmberg Medal 2009, both from the Swedish Chemical Society, and Ulla and Stig Holmquists Prize 2009 from Uppsala University.

ACKNOWLEDGMENTS Financial support from the Knut and Alice Wallenberg Foundation, the Carl Trygger Foundation, and the Swedish Energy Agency is gratefully acknowledged. We would also like to express our gratitude to the referees for their valuable comments, which have helped us to improve this Review. ABBREVIATIONS AC amorphous carbon APT atom-proton transfer bpea N,N-bis(pyridin-2-ylmethyl)ethanamine bpi (pyridin-2-ylmethyl)(pyridin-2-ylmethylene)amine bpm 2,2′-bipyrimidine bpy 2,2′-bipyridine bpz 2,2′-bipyrazine btpyan 1,8-bis(2,2′:6′,2″-terpyridyl)anthracene BVS bond valence sum CAN ceric ammonium nitrate cat catechol cod 1,5-cyclooctadiene Cp cyclopentadiene Cp* pentamethylcyclopentadiene d distal deeb diethyl 2,2′-bipyridine-4,4′-dicarboxylate DFT density functional theory DLS dynamic light scattering dpaq 2-[bis(pyridine-2-ylmethyl)]amino-N-quinolin-8-yl-acetamido dpb 4,4′-diphosphonic-2,2′-bipyridine acid dpp 2,3-bis(2′-pyridyl)pyrazine EDX energy-dispersive X-ray spectroscopy EPR electron paramagnetic resonance EQCN electrochemical quartz crystal nanobalance η overpotential ETM electron-transfer mediator EXAFS extended X-ray absorption fine structure FT-IR Fourier transform infrared spectroscopy FTO fluorine-doped tin oxide HAADF-STEM high-angle annular dark field scanning transmission electron microscopy Hbcbpen N-benzyl-N′-carboxymethyl-N,N′-bis(2pyridylmethyl)ethane-1,2-diamine Hbpc 2,2′-bipyridine-6-carboxylic acid Hbpp 2,2′-(1H-pyrazole-3,5-diyl)dipyridine HEC hydrogen evolving catalyst Hmcbpen N-methyl-N′-carboxymethyl-N,N′-bis(2pyridylmethyl)ethane-1,2-diamine HOMO highest occupied molecular orbital HRMS high-resolution mass spectrometry HRTEM high-resolution transmission electron microscopy H2bda 2,2′-bipyridine-6,6′-dicarboxylic acid H2bpb N,N′-1,2-phenylene-bis(2-pyridine-carboxamide H2hqc 8-hydroxyquinoline-2-carboxylic acid H2pda 1,10-phenanthroline-2,9-dicarboxylic acid H2pdc 2,6-pyridinedicarboxylic acid

Eric V. Johnston completed his Masters studies in 2008 at the Department of Organic Chemistry, Stockholm University, Sweden. In the same year, he began his graduate studies under the supervision of Professors Jan-Erling Bäckvall and Björn Åkermark. During his Ph.D., he worked on several different projects in the area of catalytic oxidations and artificial photosynthesis. In 2012, he obtained his Ph.D. degree in organic chemistry and is currently doing a postdoc at Memorial Sloan-Kettering Cancer Center under the supervision of Professor Samuel J. Danishefsky, where he is engaged in the chemical synthesis of glycolsylated proteins that play important roles in modern cancer treatment.

Björn Åkermark is Professor Emeritus in Organic Chemistry at the Royal Institute of Technology, Stockholm, and since his retirement is employed at Stockholm University as Guest Professor and research leader. He received his doctorate from the Royal Institute of Technology with Professors Holger Erdtman and Carl-Axel Wachtmeister in 1967 and became Assistant Professor at the Royal Institute the same year. He spent the year 1967/68 as visiting scholar with Professor Eugen van Tamelen at Stanford University. He then returned to the Royal Institute of Technology, where he became Associate Professor in 1972 and subsequently Professor in 1980. Professor Åkermark and his colleagues have authored more than 260 scientific publications. He introduced organometallic chemistry and homogeneous catalysis to Sweden in the mid 1960s and started research on artificial photosynthesis in 1973, during the first energy crisis. Professor Åkermark is still active in the field of homogeneous catalysis, but has since 1989 focused his research on artificial photosynthesis, where he cofounded the Swedish Network for artificial photosynthesis in 1995. He is a member of the Swedish and American Chemical Societies as well as the World Academy of Art and Science. He is Dr. h.c. at the University of Aix-Marseille and has received several prestigious awards, such as the Zorn Fellowship from the Sweden-America Foundation in 1977, the Arrhenius Medal 1978 and 11989

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews H3hpb H3hpbc IPCE isoq ITO I2M MCN mcp Mebimpy Mebim-py mCPBA MIMS MLCT MOF MS MWCNT NHC NHE NTA OAT OEC OER Q qpy p PAH PCET PEC PET POM ppy PS PV PXRD pyr Py5 pynap RRDE SAED SAXS SEM SQ STEM STP TAML TBHP TDMImP TEM tmeda tmtacn TOF TON TPA tpfc tpy tpym vs WNA

Review

WOC XANES XAS XPS 3,6-tBu2qui

2-(2-hydroxyphenyl)-1H-benzo[d]imidazol-7ol 2-(2-hydroxyphenyl)-1H-benzo[d]imida-zole7-carboxylic acid incident photon-to-current efficiency isoquinoline indium tin oxide interaction of two M-O units mesoporous carbon nitride N,N′-dimethyl-N,N′-bis(2-pyridylmethyl)-cyclohexane-1,2-diamine 2,6-bis(1-methylbenzimidazol-2-yl)-pyridine 3-methyl-1-pyridylbenzimidazol-2-ylidene meta-chloroperoxybenzoic acid membrane inlet mass spectrometry metal-to-ligand charge transfer metal−organic framework mass spectrometry multiwalled carbon nanotube N-heterocyclic carbene normal hydrogen electrode nanoparticle-tracking analysis oxygen atom transfer oxygen evolving complex oxygen evolving reaction quinone 2,2′:6′,2″:6″,2‴-quaterpyridine proximal poly(allylamine) proton-coupled electron transfer photoelectrochemical cell proton−electron transfer polyoxometalate phenylpyridine photosystem photovoltaic powder X-ray diffraction pyridine 2,6-(bis(bis-2-pyridyl)methoxy-methane)-pyridine 2-(2-pyridyl)-1,8-naphthyridine rotating ring-disk electrode selected area electron diffraction small-angle X-ray scattering scanning electron microscopy semiquinone scanning transmission electron microscopy standard temperature and pressure tetraamido macrocyclic ligand tert-butyl hydroperoxide 5,10,15,20-tetrakis(1,3-dimethylimidazolium2-yl)porphyrin transmission electron microscopy tetramethylenediamine 1,4,7-trimethyl-1,4,7-triazacyclononane turnover frequency turnover number tris(2-pyridylmethyl)amine 5,10,15-tris(pentafluorophenyl)corrole 2,2′;6′,2″-terpyridine tris(2-pyridyl)methane versus water nucleophilic attack

water oxidation catalyst X-ray absorption near edge structure X-ray absorption spectroscopy X-ray photoelectron spectroscopy 3,6-di-t-butyl-1,2-benzoquinone

REFERENCES (1) Chow, J.; Kopp, R. J.; Portney, P. R. Science 2003, 302, 1528. (2) Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 15729. (3) Balzani, V.; Credi, A.; Venturi, M. ChemSusChem 2008, 1, 26. (4) Sun, L.; Hammarström, L.; Åkermark, B.; Styring, S. Chem. Soc. Rev. 2001, 30, 36. (5) Tachibana, Y.; Vayssieres, L.; Durrant, J. R. Nat. Photonics 2012, 6, 511. (6) Kärkäs, M. D.; Johnston, E. V.; Verho, O.; Åkermark, B. Acc. Chem. Res. 2014, 47, 100. (7) Cook, T. R.; Dogutan, D. K.; Reece, S. Y.; Surendranath, Y.; Teets, T. S.; Nocera, D. G. Chem. Rev. 2010, 110, 6474. (8) See also prefaces to recent collections of reviews: (a) Brudvig, G. W. Coord. Chem. Rev. 2008, 252, 231. (b) Eisenberg, R.; Gray, H. B. Inorg. Chem. 2008, 47, 1697. (c) Hammarström, L.; Hammes-Schiffer, S. Acc. Chem. Res. 2009, 42, 1859. (9) Tollefson, J. Nature 2011, 473, 134. (10) Service, R. F. Science 2011, 334, 925. (11) Dasgupta, S.; Brunschwig, B. S.; Winkler, J. R.; Gray, H. B. Chem. Soc. Rev. 2013, 42, 2213. (12) Morton, O. Nature 2006, 443, 19. (13) Allakhverdiev, S. I.; Casal, J. J.; Nagata, T. Photochem. Photobiol. Sci. 2009, 8, 137. (14) Nelson, N.; Ben-Shem, A. Nat. Rev. Mol. Cell Biol. 2004, 5, 971. (15) McEvoy, J. P.; Brudvig, G. W. Chem. Rev. 2006, 106, 4455. (16) Barber, J. Biochem. Soc. Trans. 2006, 34, 619. (17) Barber, J. Chem. Soc. Rev. 2009, 38, 185. (18) Vengadesh, P. In Nanocrystals - Synthesis, Chacterization and Applications; Nerella, S., Ed.; InTech: Rijeka, 2012; pp 41−60. (19) Throughout this Review, redox potentials are presented against the normal hydrogen electrode (NHE). Occasionally the referenced potentials have been determined versus different electrodes and have been converted to NHE for the purpose of comparison. For conversion constants, see: (a) Pavlishchuk, V. V.; Addison, A. W. Inorg. Chim. Acta 2000, 298, 97. (b) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, 2nd ed.; John Wiley & Sons Inc.: New York, 2001. (20) Siegbahn, P. E. M. Acc. Chem. Res. 2009, 42, 1871. (21) Ferreira, K. N.; Iverson, T. M.; Maghlaoui, K.; Barber, J.; Iwata, S. Science 2004, 303, 1831. (22) Loll, B.; Kern, J.; Saenger, W.; Zouni, A.; Biesiadka, J. Nature 2005, 438, 1040. (23) Yano, J.; Kern, J.; Irrgang, K.-D.; Latimer, M. J.; Bergmann, U.; Glatzel, P.; Pushkar, Y.; Biesiadka, J.; Loll, B.; Sauer, K.; Messinger, J.; Zouni, A.; Yachandra, V. K. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 12047. (24) Yano, J.; Kern, J.; Sauer, K.; Latimer, M. J.; Pushkar, Y.; Biesiadka, J.; Loll, B.; Saenger, W.; Messinger, J.; Zouni, A.; Yachandra, V. K. Science 2006, 314, 821. (25) Umena, Y.; Kawakami, K.; Shen, J.-R.; Kamiya, N. Nature 2011, 473, 55. (26) Kern, J.; Alonso-Mori, R.; Tran, R.; Hattne, J.; Gildea, R. J.; Echols, N.; Glöckner, C.; Hellmich, J.; Laksmono, H.; Sierra, R. G.; Lassalle-Kaiser, B.; Koroidov, S.; Lampe, A.; Han, G.; Gul, S.; DiFiore, D.; Milathianaki, D.; Fry, A. R.; Miahnahri, A.; Schafer, D. W.; Messerschmidt, M.; Seibert, M. M.; Koglin, J. E.; Sokaras, D.; Weng, T.-C.; Sellberg, J.; Latimer, M. J.; Grosse-Kunstleve, R. W.; Zwart, P. H.; White, W. E.; Glatzel, P.; Adams, P. D.; Bogan, M. J.; Williams, G. J.; Boutet, S.; Messinger, J.; Zouni, A.; Sauter, N. K.; Yachandra, V. K.; Bergmann, U.; Yano, J. Science 2013, 340, 491. (27) Cox, N.; Messinger, J. Biochim. Biophys. Acta 2013, 1827, 1020. 11990

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(28) Glatzel, P.; Schroeder, H.; Pushkar, Y.; Boron, T., III; Mukherjee, S.; Christou, G.; Pecoraro, V. L.; Messinger, J.; Yachandra, V. K.; Bergmann, U.; Yano, J. Inorg. Chem. 2013, 52, 5642. (29) Rapatskiy, L.; Cox, N.; Savitsky, A.; Ames, W. M.; Sander, J.; Nowaczyk, M. M.; Rögner, M.; Boussac, A.; Neese, F.; Messinger, J.; Lubitz, W. J. Am. Chem. Soc. 2012, 134, 16619. (30) Kulik, L. V.; Epel, B.; Lubitz, W.; Messinger, J. J. Am. Chem. Soc. 2007, 129, 13421. (31) Haumann, M.; Liebisch, P.; Müller, C.; Barra, M.; Grabolle, M.; Dau, H. Science 2005, 310, 1019. (32) Renger, G.; Renger, T. Photosynth. Res. 2008, 98, 53. (33) Kok, B.; Forbush, B.; McGloin, M. Photochem. Photobiol. 1970, 11, 457. (34) Ciamician, G. Science 1912, 36, 385. (35) Nocera, D. G. Acc. Chem. Res. 2012, 45, 767. (36) Hurst, J. K. Science 2010, 328, 315. (37) Meyer, T. J. Acc. Chem. Res. 1989, 22, 163. (38) Bard, A. J.; Fox, M. A. Acc. Chem. Res. 1995, 28, 141. (39) Amouyal, E. Sol. Energy Mater. Sol. Cells 1995, 38, 249. (40) Frischmann, P. D.; Mahata, K.; Würthner, F. Chem. Soc. Rev. 2013, 42, 1847. (41) Gust, D.; Moore, T. A.; Moore, A. L. Acc. Chem. Res. 2009, 42, 1890. (42) Concepcion, J. J.; House, R. L.; Papanikolas, J. M.; Meyer, T. J. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 15560. (43) Walter, M. G.; Warren, E. L.; McKone, J. R.; Boettcher, S. W.; Mi, Q.; Santori, E. A.; Lewis, N. S. Chem. Rev. 2010, 110, 6446. (44) Yang, J.; Wang, D.; Han, H.; Li, C. Acc. Chem. Res. 2013, 46, 1900. (45) Chen, X.; Shen, S.; Guo, L.; Mao, S. S. Chem. Rev. 2010, 110, 6503. (46) Osterloh, F. E. Chem. Soc. Rev. 2013, 42, 2294. (47) Tran, P. D.; Wong, L. H.; Barber, J.; Loo, J. S. C. Energy Environ. Sci. 2012, 5, 5902. (48) Kudo, A.; Miseki, Y. Chem. Soc. Rev. 2009, 38, 253. (49) Kitano, M.; Hara, M. J. Mater. Chem. 2010, 20, 627. (50) Park, Y.; McDonald, K. J.; Choi, K.-S. Chem. Soc. Rev. 2013, 42, 2321. (51) Demas, J. N.; Adamson, A. W. J. Am. Chem. Soc. 1971, 93, 1800. (52) Gafney, H. D.; Adamson, A. W. J. Am. Chem. Soc. 1972, 94, 8238. (53) For reviews on photoredox catalysis, see, for example: (a) Prier, C. K.; Rankic, D. A.; MacMillan, D. W. C. Chem. Rev. 2013, 113, 5322. (b) Narayanam, J. M. R.; Stephenson, C. R. J. Chem. Soc. Rev. 2011, 40, 102. (c) Xuan, J.; Xiao, W.-J. Angew. Chem., Int. Ed. 2012, 51, 6828. (d) Xi, Y.; Yi, H.; Lei, A. Org. Biomol. Chem. 2013, 11, 2387. (54) Forster, L. S. Coord. Chem. Rev. 2006, 250, 2023. (55) Juris, A.; Balzani, V.; Barigelletti, F.; Campagna, S.; Belser, P.; von Zelewsky, A. Coord. Chem. Rev. 1988, 84, 85. (56) Kalyanasundaram, K. Coord. Chem. Rev. 1982, 46, 159. (57) Campagna, S.; Puntoriero, F.; Nastasi, F.; Bergamini, G.; Balzani, V. Top. Curr. Chem. 2007, 280, 117. (58) Sauvage, J.-P.; Collin, J.-P.; Chambron, J.-C.; Guillerez, S.; Coudret, C.; Balzani, V.; Barigelletti, F.; De Cola, L.; Flamigni, L. Chem. Rev. 1994, 94, 993. (59) Thompson, D. W.; Ito, A.; Meyer, T. J. Pure Appl. Chem. 2013, 85, 1257. (60) For discussions on the hydrogen economy, see, for example: (a) Lubitz, W.; Tumas, B. Chem. Rev. 2007, 107, 3900. (b) Palo, D. R.; Dagle, R. A.; Holladay, J. D. Chem. Rev. 2007, 107, 3992. (c) Marbán, G.; Valdés-Solís, T. Int. J. Hydrogen Energy 2007, 32, 1625. (61) Nocera, D. G. Inorg. Chem. 2009, 48, 10001. (62) Esswein, A. J.; Nocera, D. G. Chem. Rev. 2007, 107, 4022. (63) Thoi, V. S.; Sun, Y.; Long, J. R.; Chang, C. J. Chem. Soc. Rev. 2013, 42, 2388. (64) Morris, A. J.; Meyer, G. J.; Fujita, E. Acc. Chem. Res. 2009, 42, 1983. (65) Du, P.; Eisenberg, R. Energy Environ. Sci. 2012, 5, 6012.

(66) Fukuzumi, S.; Yamada, Y.; Suenobu, T.; Ohkuboa, K.; Kotani, H. Energy Environ. Sci. 2011, 4, 2754. (67) Sakai, K.; Ozawa, H. Coord. Chem. Rev. 2007, 251, 2753. (68) Frey, M. ChemBioChem 2002, 3, 153. (69) Darensbourg, M. Y.; Lyon, E. J.; Smee, J. J. Coord. Chem. Rev. 2000, 206, 533. (70) Tillmann Stiebritz, M.; Reiher, M. Chem. Sci. 2012, 3, 1739. (71) Peters, J. W.; Lanzilotta, W. N.; Lemon, B. J.; Seefeldt, L. C. Science 1998, 282, 1853. (72) Volbeda, A.; Charon, M.-H.; Piras, C.; Hatchikian, E. C.; Frey, M.; Fontecilla-Camps, J. C. Nature 1995, 373, 580. (73) For a discussion on the origin of hydrogenases, see: Martin, W.; Müller, M. Nature 1998, 392, 37. (74) Pershad, H. R.; Duff, J. L. C.; Heering, H. A.; Duin, E. C.; Albracht, S. P. J.; Armstrong, F. A. Biochemistry 1999, 38, 8992. (75) For reviews on artificial hydrogenases, see: (a) Vignais, P. M.; Billoud, B. Chem. Rev. 2007, 107, 4206. (b) Sun, L.; Åkermark, B.; Ott, S. Coord. Chem. Rev. 2005, 249, 1653. (c) Lubitz, W.; Reijerse, E. J.; Messinger, J. Energy Environ. Sci. 2008, 1, 15. (d) Fontecilla-Camps, J. C.; Volbeda, A.; Cavazza, C.; Nicolet, Y. Chem. Rev. 2007, 107, 4273. (e) Tard, C.; Pickett, C. J. Chem. Rev. 2009, 109, 2245. (f) Gloaguen, F.; Rauchfuss, T. B. Chem. Soc. Rev. 2009, 38, 100. (76) For selected examples of artificial hydrogenases for H2 production, see, for example: (a) Ott, S.; Kritikos, M.; Åkermark, B.; Sun, L. Angew. Chem., Int. Ed. 2003, 42, 3285. (b) Silakov, A.; Olsen, M. T.; Sproules, S.; Reijerse, E. J.; Rauchfuss, T. B.; Lubitz, W. Inorg. Chem. 2012, 51, 8617. (c) Na, Y.; Wang, M.; Pan, J.; Zhang, P.; Åkermark, B.; Sun, L. Inorg. Chem. 2008, 47, 2805. (d) Ott, S.; Kritikos, M.; Åkermark, B.; Sun, L.; Lomoth, R. Angew. Chem., Int. Ed. 2004, 43, 1006. (e) Gao, W.; Liu, J.; Åkermark, B.; Sun, L. Inorg. Chem. 2006, 45, 9169. (77) For recent reviews on cobaloxime complexes for H2 production, see: (a) Dempsey, J. L.; Brunschwig, B. S.; Winkler, J. R.; Gray, H. B. Acc. Chem. Res. 2009, 42, 1995. (b) Artero, V.; Chavarot-Kerlidou, M.; Fontecave, M. Angew. Chem., Int. Ed. 2011, 50, 7238. (c) Losse, S.; Vos, J. G.; Rau, S. Coord. Chem. Rev. 2010, 254, 2492. (78) For selected examples of cobalt-based complexes for H2 production, see, for example: (a) Zhang, P.; Wang, M.; Gloaguen, F.; Chen, L.; Quentel, F.; Sun, L. Chem. Commun. 2013, 49, 9455. (b) Stubbert, B. D.; Peters, J. C.; Gray, H. B. J. Am. Chem. Soc. 2011, 133, 18070. (c) Marinescu, S. C.; Winkler, J. R.; Gray, H. B. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 15127. (d) Du, P.; Schneider, J.; Luo, G.; Brennessel, W. W.; Eisenberg, R. Inorg. Chem. 2009, 48, 4952. (79) For recent reviews on nickel-based complexes for H2 evolution, see: (a) DuBois, M. R.; DuBois, D. L. Chem. Soc. Rev. 2009, 38, 62. (b) Wang, M.; Chen, L.; Sun, L. Energy Environ. Sci. 2012, 5, 6763. (c) Dubois, M. R.; Dubois, D. L. Acc. Chem. Res. 2009, 42, 1974. (80) For selected examples of nickel-based complexes for H2 production, see, for example: (a) Le Goff, A.; Artero, V.; Jousselme, B.; Tran, P. D.; Guillet, N.; Métayé, R.; Fihri, A.; Palacin, S.; Fontecave, M. Science 2009, 326, 1384. (b) Helm, M. L.; Stewart, M. P.; Bullock, R. M.; DuBois, M. R.; DuBois, D. L. Science 2011, 333, 863. (c) Jacques, P.-A.; Artero, V.; Pécaut, J.; Fontecave, M. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 20627. (81) Fisher, B.; Eisenberg, R. J. Am. Chem. Soc. 1980, 102, 7361. (82) Inoue, H.; Shimada, T.; Kou, Y.; Nabetani, Y.; Masui, D.; Takagi, S.; Tachibana, H. ChemSusChem 2011, 4, 173. (83) Hamann, C. H.; Hamnett, A.; Vielstich, W. Electrochemistry; Wiley-VCH: Weinheim, 2007. (84) For excellent reviews on proton-coupled electron transfer (PCET), see: (a) Weinberg, D. R.; Gagliardi, C. J.; Hull, J. F.; Fecenko Murphy, C.; Kent, C. A.; Westlake, B. C.; Paul, A.; Ess, D. H.; Granville McCafferty, D.; Meyer, T. J. Chem. Rev. 2012, 112, 4016. (b) Gagliardi, C. J.; Vannucci, A. K.; Concepcion, J. J.; Chen, Z.; Meyer, T. J. Energy Environ. Sci. 2012, 5, 7704. (c) Wenger, O. S. Chem.Eur. J. 2011, 17, 11692. (d) Meyer, T. J.; Huynh, M. H. V.; Holden Thorp, H. Angew. Chem., Int. Ed. 2007, 46, 5284. (e) Huynh, M. H. V.; Meyer, T. J. Chem. Rev. 2007, 107, 5004. 11991

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(85) Parent, A. R.; Crabtree, R. H.; Brudvig, G. W. Chem. Soc. Rev. 2013, 42, 2247. (86) Nair, V.; Deepthi, A. Chem. Rev. 2007, 107, 1862. (87) Wadsworth, E.; Duke, F. R.; Goetz, C. A. Anal. Chem. 1957, 29, 1824. (88) Ikeda-Ohno, A.; Tsushima, S.; Hennig, C.; Yaitab, T.; Bernhard, G. Dalton Trans. 2012, 41, 7190. (89) Robert, A.; Meunier, B. New J. Chem. 1988, 12, 885. (90) Limburg, J.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chim. Acta 2000, 297, 301. (91) Hetterscheid, D. G. H.; Reek, J. N. H. Eur. J. Inorg. Chem. 2014, 742. (92) Weavers, L. K.; Hua, I.; Hoffmann, M. R. Water Environ. Res. 1997, 69, 1112. (93) Ghosh, P. K.; Brunschwig, B. S.; Chou, M.; Creutz, C.; Sutin, N. J. Am. Chem. Soc. 1984, 106, 4772. (94) Bolletta, F.; Juris, A.; Maestri, M.; Sandrini, D. Inorg. Chim. Acta 1980, 44, L175. (95) Gau, B. C.; Chen, H.; Zhang, Y.; Gross, M. L. Anal. Chem. 2010, 82, 7821. (96) Navon, G.; Sutin, N. Inorg. Chem. 1974, 13, 2159. (97) Joya, K. S.; Joya, Y. F.; Ocakoglu, K.; van de Krol, R. Angew. Chem., Int. Ed. 2013, 52, 10426. (98) Swierk, J. R.; Mallouk, T. E. Chem. Soc. Rev. 2013, 42, 2357. (99) Joya, K. S.; Vallés-Pardo, J. L.; Joya, Y. F.; Eisenmayer, T.; Thomas, B.; Buda, F.; de Groot, H. J. M. ChemPlusChem 2013, 78, 35. (100) Bagotsky, V. S. Fundamentals of Electrochemistry, 2nd ed.; John Wiley & Sons, Inc.: Hoboken, NJ, 2006. (101) Hamann, C. H.; Hamnett, A.; Vielstich, W. Electrochemistry; Wiley-VCH: Weinheim, 2007. (102) Bockris, J. O.; Reddy, A. K. N. Modern Electrochemistry 2B: Electrodics in Chemistry, Engineering, Biology, and Environmental Science; Kluwer Academic: New York, 2000; Vol. 2B. (103) For excellent reviews discussing the homogeneity of chemical reactions, see: (a) Crabtree, R. H. Chem. Rev. 2012, 112, 1536. (b) Artero, V.; Fontecave, M. Chem. Soc. Rev. 2013, 42, 2338. (c) Fukuzumi, S.; Hong, D. Eur. J. Inorg. Chem. 2014, 645. (104) An, K.; Somorjai, G. A. ChemCatChem 2012, 4, 1512. (105) Astruc, D.; Lu, F.; Aranzaes, J. R. Angew. Chem., Int. Ed. 2005, 44, 7852. (106) Hamlin, J. E.; Hirai, K.; Millan, A.; Maitlis, P. M. J. Mol. Catal. A: Chem. 1980, 7, 543. (107) Lewis, L. N. Chem. Rev. 1993, 93, 2693. (108) Aiken, J. D.; Finke, R. G. J. Mol. Catal. A 1999, 145, 1. (109) Privalov, T.; Åkermark, B.; Sun, L. Chem.Eur. J. 2011, 17, 8313. (110) Wang, L.-P.; Wu, Q.; van Voorhis, T. Inorg. Chem. 2010, 49, 4543. (111) Romain, S.; Vigara, L.; Llobet, A. Acc. Chem. Res. 2009, 42, 1944. (112) Concepcion, J. J.; Tsai, M. K.; Muckerman, J. T.; Meyer, T. J. J. Am. Chem. Soc. 2010, 132, 1545. (113) Lin, X.; Hu, X.; Concepcion, J. J.; Chen, Z.; Liu, S.; Meyer, T. J.; Yang, W. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 15669. (114) Vigara, L.; Ertem, M. Z.; Planas, N.; Bozoglian, F.; Leidel, N.; Dau, H.; Haumann, M.; Gagliardi, L.; Cramer, C. J.; Llobet, A. Chem. Sci. 2012, 3, 2576. (115) Romero, I.; Rodríguez, M.; Sens, C.; Mola, J.; Kollipara, M. R.; Francàs, L.; Mas-Marza, E.; Escriche, L.; Llobet, A. Inorg. Chem. 2008, 47, 1824. (116) Clark, A. E.; Hurst, J. K. In Mechanisms of Water Oxidation Catalyzed by Ruthenium Coordination Complexes, in Progress in Inorganic Chemistry; Karlin, K. D., Ed.; John Wiley & Sons, Inc.: Hoboken, NJ, 2011; Vol. 57, doi: 10.1002/9781118148235.ch1. (117) Kikuchi, T.; Tanaka, K. Eur. J. Inorg. Chem. 2014, 607. (118) Savini, A.; Bucci, A.; Bellachioma, G.; Rocchigiani, L.; Zuccaccia, C.; Llobet, A.; Macchioni, A. Eur. J. Inorg. Chem. 2014, 690. (119) Sala, X.; Romero, I.; Rodríguez, M.; Escriche, L.; Llobet, A. Angew. Chem., Int. Ed. 2009, 48, 2842.

(120) Sala, X.; Maji, S.; Bofill, R.; García-Antón, J.; Escriche, L.; Llobet, A. Acc. Chem. Res. 2014, 47, 504. (121) Gersten, S. W.; Samuels, G. J.; Meyer, T. J. J. Am. Chem. Soc. 1982, 104, 4029. (122) Gilbert, J. A.; Eggleston, D. S.; Murphy, W. R., Jr.; Geselowitz, D. A.; Gersten, S. W.; Hodgson, D. J.; Meyer, T. J. J. Am. Chem. Soc. 1985, 107, 3855. (123) Collin, J. P.; Sauvage, J. P. Inorg. Chem. 1986, 25, 135. (124) Nagoshi, K.; Yamashita, S.; Yagi, M.; Kaneko, M. J. Mol. Catal. A: Chem. 1999, 144, 71. (125) Doppelt, P.; Meyer, T. J. Inorg. Chem. 1987, 26, 2027. (126) Aguiló, J.; Naeimi, A.; Bofill, R.; Mueller-Bunz, H.; Llobet, A.; Escriche, L.; Sala, X.; Albrecht, M. New J. Chem. 2014, 38, 1980. (127) Lebeau, E. L.; Adeyemi, S. A.; Meyer, T. J. Inorg. Chem. 1998, 37, 6476. (128) Yang, X.; Baik, M.-H. J. Am. Chem. Soc. 2008, 130, 16231. (129) Jurss, J. W.; Concepcion, J. J.; Norris, M. R.; Templeton, J. L.; Meyer, T. J. Inorg. Chem. 2010, 49, 3980. (130) Gallagher, L. A.; Meyer, T. J. J. Am. Chem. Soc. 2001, 123, 5308. (131) Liu, F.; Cardolaccia, T.; Hornstein, B. J.; Schoonover, J. R.; Meyer, T. J. J. Am. Chem. Soc. 2007, 129, 2446. (132) Cape, J. L.; Hurst, J. K. J. Am. Chem. Soc. 2008, 130, 827. (133) Rotzinger, F. P.; Munavalli, S.; Comte, P.; Hurst, J. K.; Grätzel, M.; Pern, F.-J.; Frank, A. J. J. Am. Chem. Soc. 1987, 109, 6619. (134) Conception, J. J.; Jurss, J. W.; Templeton, J. L.; Meyer, T. J. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 17632. (135) For a review on multistep electron transfer, see: Piera, J.; Bäckvall, J.-E. Angew. Chem., Int. Ed. 2008, 47, 3506. (136) For examples of synthetic protocols based on ETMs with O2 as terminal oxidant, see, for example: (a) Johnston, E. V.; Karlsson, E. A.; Lindberg, S. A.; Åkermark, B.; Bäckvall, J.-E. Chem.Eur. J. 2009, 15, 6799. (b) Johnston, E. V.; Karlsson, E. A.; Tran, L.-H.; Åkermark, B.; Bäckvall, J.-E. Eur. J. Org. Chem. 2010, 1971. (c) Endo, Y.; Bäckvall, J.E. Chem.Eur. J. 2011, 17, 12596. (d) Babu, B. P.; Endo, Y.; Bäckvall, J.-E. Chem.Eur. J. 2012, 18, 11524. (e) Endo, Y.; Bäckvall, J.-E. Chem.Eur. J. 2012, 18, 13609. (f) Gigant, N.; Bäckvall, J.-E. Chem. Eur. J. 2013, 19, 10799. (137) Geselowitz, D.; Meyer, T. J. Inorg. Chem. 1990, 29, 3894. (138) Chronister, C. W.; Binstead, R. A.; Ni, J.; Meyer, T. J. Inorg. Chem. 1997, 36, 3814. (139) Binstead, R. A.; Chronister, C. W.; Ni, J.; Hartshorn, C. M.; Meyer, T. J. J. Am. Chem. Soc. 2000, 122, 8464. (140) Liu, F.; Concepcion, J. J.; Jurss, J. W.; Cardolaccia, T.; Templeton, J. L.; Meyer, T. J. Inorg. Chem. 2008, 47, 1727. (141) Moonshiram, D.; Alperovich, I.; Concepcion, J. J.; Meyer, T. J.; Pushkar, Y. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 3765. (142) Moonshiram, D.; Jurss, J. W.; Concepcion, J. J.; Zakharova, T.; Alperovich, I.; Meyer, T. J.; Pushkar, Y. J. Am. Chem. Soc. 2012, 134, 4625. (143) Hurst, J. K.; Zhou, J.; Lei, Y. Inorg. Chem. 1992, 31, 1010. (144) Lei, Y.; Hurst, J. K. Inorg. Chem. 1994, 33, 4460. (145) Lei, Y.; Hurst, J. K. Inorg. Chim. Acta 1994, 226, 179. (146) Yamada, H.; Hurst, J. K. J. Am. Chem. Soc. 2000, 122, 5303. (147) Yamada, H.; Koike, T.; Hurst, J. K. J. Am. Chem. Soc. 2001, 123, 12775. (148) Yamada, H.; Siems, W. F.; Koike, T.; Hurst, J. K. J. Am. Chem. Soc. 2004, 126, 9786. (149) Hurst, J. K.; Cape, J. L.; Clark, A. E.; Das, S.; Qin, C. Inorg. Chem. 2008, 47, 1753. (150) Cape, J. L.; Siems, W. F.; Hurst, J. K. Inorg. Chem. 2009, 48, 8729. (151) Stull, J. A.; Stich, T. A.; Hurst, J. K.; Britt, R. D. Inorg. Chem. 2013, 52, 4578. (152) Stull, J. A.; Britt, R. D.; McHale, J. L.; Knorr, F. J.; Lymar, S. V.; Hurst, J. K. J. Am. Chem. Soc. 2012, 134, 19973. (153) Polyansky, D. E.; Hurst, J. K.; Lymar, S. V. Eur. J. Inorg. Chem. 2014, 619. (154) Yang, X.; Baik, M.-H. J. Am. Chem. Soc. 2004, 126, 13222. 11992

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(155) Yang, X.; Baik, M.-H. J. Am. Chem. Soc. 2006, 128, 7476. (156) Batista, E. R.; Martin, R. L. J. Am. Chem. Soc. 2007, 129, 7224. (157) Bianco, R.; Hay, P. J.; Hynes, J. T. J. Phys. Chem. A 2011, 115, 8003. (158) Jurss, J. W.; Concepcion, J. J.; Butler, J. M.; Omberg, K. M.; Baraldo, L. M.; Graff Thompson, D.; Lebeau, E. L.; Hornstein, B.; Schoonover, J. R.; Jude, H.; Thompson, J. D.; Dattelbaum, D. M.; Rocha, R. C.; Templeton, J. L.; Meyer, T. J. Inorg. Chem. 2012, 51, 1345. (159) Sens, C.; Romero, I.; Rodríguez, M.; Llobet, A.; Parella, T.; Benet-Buchholz, J. J. Am. Chem. Soc. 2004, 126, 7798. (160) Roeser, S.; Ertem, M. Z.; Cady, C.; Lomoth, R.; BenetBuchholz, J.; Hammarström, L.; Sarkar, B.; Kaim, W.; Cramer, C. J.; Llobet, A. Inorg. Chem. 2012, 51, 320. (161) Romain, S.; Bozoglian, F.; Sala, X.; Llobet, A. J. Am. Chem. Soc. 2009, 131, 2768. (162) Bozoglian, F.; Romain, S.; Ertem, M. Z.; Todorova, T. K.; Sens, C.; Mola, J.; Rodríguez, M.; Romero, I.; Benet-Buchholz, J.; Fontrodona, X.; Cramer, C. J.; Gagliardi, L.; Llobet, A. J. Am. Chem. Soc. 2009, 131, 15176. (163) Ellis, C. D.; Gilbert, J. A.; Murphy, W. R., Jr.; Meyer, T. J. J. Am. Chem. Soc. 1983, 105, 4842. (164) Francàs, L.; Sala, X.; Benet-Buchholz, J.; Escriche, L.; Llobet, A. ChemSusChem 2009, 2, 321. (165) Mola, J.; Mas-Marza, E.; Sala, X.; Romero, I.; Rodríguez, M.; Viñas, C.; Parella, T.; Llobet, A. Angew. Chem., Int. Ed. 2008, 47, 5830. (166) Aguiló, J.; Francàs, L.; Liu, H. J.; Bofill, R.; García-Antón, J.; Benet-Buchholz, J.; Llobet, A.; Escriche, L.; Sala, X. Catal. Sci. Technol. 2014, 4, 190. (167) Neudeck, S.; Maji, S.; López, I.; Meyer, S.; Meyer, F.; Llobet, A. J. Am. Chem. Soc. 2014, 136, 24. (168) Mola, J.; Dinoi, C.; Sala, X.; Rodríguez, M.; Romero, I.; Parella, T.; Fontrodona, X.; Llobet, A. Dalton Trans. 2011, 40, 3640. (169) Maji, S.; Vigara, L.; Cottone, F.; Bozoglian, F.; Benet-Buchholz, J.; Llobet, A. Angew. Chem., Int. Ed. 2012, 51, 5967. (170) Zong, R.; Thummel, R. P. J. Am. Chem. Soc. 2005, 127, 12802. (171) Deng, Z.; Tseng, H.-W.; Zong, R.; Wang, D.; Thummel, R. Inorg. Chem. 2008, 47, 1835. (172) Francàs, L.; Sala, X.; Escudero-Adán, E.; Benet-Buchholz, J.; Escriche, L.; Llobet, A. Inorg. Chem. 2011, 50, 2771. (173) Wada, T.; Tsuge, K.; Tanaka, K. Angew. Chem., Int. Ed. 2000, 39, 1479. (174) Wada, T.; Tsuge, K.; Tanaka, K. Inorg. Chem. 2001, 40, 329. (175) For a recent review on redox-active ligands, see: Luca, O. R.; Crabtree, R. H. Chem. Soc. Rev. 2013, 42, 1440. (176) Intramolecular electron transfer from the oxo ligands (O2−) to the quinones could lead to the generation of two RuII−O• units or potentially result in the formation of a RuII−O−O−RuII species. (177) Muckerman, J. T.; Polyansky, D. E.; Wada, T.; Tanaka, K.; Fujita, E. Inorg. Chem. 2008, 47, 1787. (178) Ghosh, S.; Baik, M.-H. Inorg. Chem. 2011, 50, 5946. (179) Ghosh, S.; Baik, M.-H. Angew. Chem., Int. Ed. 2012, 51, 1221. (180) Tanaka, K.; Isobe, H.; Yamanaka, S.; Yamaguchi, K. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 15600. (181) Isobe, H.; Tanaka, K.; Shen, J.-R.; Yamaguchi, K. Inorg. Chem. 2014, 53, 3973. (182) Kobayashi, K.; Ohtsu, H.; Wada, T.; Kato, T.; Tanaka, K. J. Am. Chem. Soc. 2003, 125, 6729. (183) Tsai, M.-K.; Rochford, J.; Polyansky, D. E.; Wada, T.; Tanaka, K.; Fujita, E.; Muckerman, J. T. Inorg. Chem. 2009, 48, 4372. (184) Wada, T.; Yamanaka, M.; Fujihara, T.; Miyazato, Y.; Tanaka, K. Inorg. Chem. 2006, 45, 8887. (185) Wada, T.; Muckerman, J. T.; Fujita, E.; Tanaka, K. Dalton Trans. 2011, 40, 2225. (186) Wada, T.; Ohtsu, H.; Tanaka, K. Chem.Eur. J. 2012, 18, 2374. (187) Wasylenko, D. J.; Ganesamoorthy, C.; Koivisto, B. D.; Berlinguette, C. P. Eur. J. Inorg. Chem. 2010, 3135.

(188) Betley, T. A.; Wu, Q.; Van Voorhis, T.; Nocera, D. G. Inorg. Chem. 2008, 47, 1849. (189) Xu, Y.; Duan, L.; Åkermark, T.; Tong, L.; Lee, B.-L.; Zhang, R.; Åkermark, B.; Sun, L. Chem.Eur. J. 2011, 17, 9520. (190) Lomoth, R.; Huang, P.; Zheng, J.; Sun, L.; Hammarström, L.; Åkermark, B.; Styring, S. Eur. J. Inorg. Chem. 2002, 2965. (191) Lee, B.-L.; Kärkäs, M. D.; Johnston, E. V.; Inge, A. K.; Tran, L.H.; Xu, Y.; Hansson, Ö .; Zou, X.; Åkermark, B. Eur. J. Inorg. Chem. 2010, 5462. (192) Xu, Y.; Åkermark, T.; Gyollai, V.; Zou, D.; Eriksson, L.; Duan, L.; Zhang, R.; Åkermark, B.; Sun, L. Inorg. Chem. 2009, 48, 2717. (193) Xu, Y.; Duan, L.; Tong, L.; Åkermark, B.; Sun, L. Chem. Commun. 2010, 46, 6506. (194) Xu, Y.; Fischer, A.; Duan, L.; Tong, L.; Gabrielsson, E.; Åkermark, B.; Sun, L. Angew. Chem., Int. Ed. 2010, 49, 8934. (195) Kärkäs, M. D.; Johnston, E. V.; Karlsson, E. A.; Lee, B.-L.; Åkermark, T.; Shariatgorji, M.; Ilag, L.; Hansson, Ö .; Bäckvall, J.-E.; Åkermark, B. Chem.Eur. J. 2011, 17, 7953. (196) Lewandowska-Andralojc, A.; Polyansky, D. E.; Zong, R.; Thummel, R. P.; Fujita, E. Phys. Chem. Chem. Phys. 2013, 15, 14058. (197) Khan, M. M. T.; Ramachandraiah, G.; Mehta, S. H.; Abdi, S. H. R.; Kumar, S. J. Mol. Catal. 1990, 58, 199. (198) Pramanik, N. C.; Bhattacharya, S. Transition Met. Chem. 1997, 22, 524. (199) Zhang, G.; Zong, R.; Tseng, H.-W.; Thummel, R. P. Inorg. Chem. 2008, 47, 990. (200) Tseng, H.-W.; Zong, R.; Muckerman, J. T.; Thummel, R. Inorg. Chem. 2008, 47, 11763. (201) Kaveevivitchai, N.; Kohler, L.; Zong, R.; El Ojaimi, M.; Mehta, N.; Thummel, R. P. Inorg. Chem. 2013, 52, 10615. (202) Concepcion, J. J.; Jurss, J. W.; Templeton, J. L.; Meyer, T. J. J. Am. Chem. Soc. 2008, 130, 16462. (203) Polyansky, D. E.; Muckerman, J. T.; Rochford, J.; Zong, R.; Thummel, R. P.; Fujita, E. J. Am. Chem. Soc. 2011, 133, 14649. (204) Badiei, Y. M.; Polyansky, D. E.; Muckerman, J. T.; Szalda, D. J.; Haberdar, R.; Zong, R.; Thummel, R. P.; Fujita, E. Inorg. Chem. 2013, 52, 8845. (205) Masaoka, S.; Sakai, K. Chem. Lett. 2009, 38, 182. (206) Wasylenko, D. J.; Ganesamoorthy, C.; Koivisto, B. D.; Henderson, M. A.; Berlinguette, C. P. Inorg. Chem. 2010, 49, 2202. (207) Wasylenko, D. J.; Ganesamoorthy, C.; Henderson, M. A.; Koivisto, B. D.; Osthoff, H. D.; Berlinguette, C. P. J. Am. Chem. Soc. 2010, 132, 16094. (208) Wasylenko, D. J.; Ganesamoorthy, C.; Henderson, M. A.; Berlinguette, C. P. Inorg. Chem. 2011, 50, 3662. (209) Kimoto, A.; Yamauchi, K.; Yoshida, M.; Masaoka, S.; Sakai, K. Chem. Commun. 2012, 48, 239. (210) Yagi, M.; Tajima, S.; Komi, M.; Yamazaki, H. Dalton Trans. 2011, 40, 3802. (211) Yoshida, M.; Masaoka, S.; Abe, J.; Sakai, K. Chem.Asian J. 2010, 5, 2369. (212) Ikeda-Ohno, A.; Tsushima, S.; Hennig, C.; Yaita, T.; Bernhard, G. Dalton Trans. 2012, 41, 7190. (213) Concepcion, J. J.; Jurss, J. W.; Norris, M. R.; Chen, Z.; Templeton, J. L.; Meyer, T. J. Inorg. Chem. 2010, 49, 1277. (214) Chen, Z.; Concepcion, J. J.; Meyer, T. J. Dalton Trans. 2011, 40, 3789. (215) Ashford, D. L.; Stewart, D. J.; Glasson, C. R. K.; Binstead, R. A.; Harrison, D. P.; Norris, M. P.; Concepcion, J. J.; Fang, Z.; Templeton, J. L.; Meyer, T. J. Inorg. Chem. 2012, 51, 6428. (216) Concepcion, J. J.; Jurss, J. W.; Hoertz, P. G.; Meyer, T. J. Angew. Chem., Int. Ed. 2009, 48, 9473. (217) Ashford, D. L.; Song, W.; Concepcion, J. J.; Glasson, C. R. K.; Brennaman, M. K.; Norris, M. R.; Fang, Z.; Templeton, J. L.; Meyer, T. J. J. Am. Chem. Soc. 2012, 134, 19189. (218) Ashford, D. L.; Lapides, A. M.; Vannucci, A. K.; Hanson, K.; Torelli, D. A.; Harrison, D. P.; Templeton, J. L.; Meyer, T. J. J. Am. Chem. Soc. 2014, 136, 6578. 11993

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(219) Chen, Z.; Concepcion, J. J.; Jurss, J. W.; Meyer, T. J. J. Am. Chem. Soc. 2009, 131, 15580. (220) Chen, Z.; Concepcion, J. J.; Hull, J. F.; Hoertz, P. G.; Meyer, T. J. Dalton Trans. 2010, 39, 6950. (221) Chen, Z.; Concepcion, J. J.; Luo, H.; Hull, J. F.; Paul, A.; Meyer, T. J. J. Am. Chem. Soc. 2010, 132, 17670. (222) Hoertz, P. G.; Chen, Z.; Kent, C. A.; Meyer, T. J. Inorg. Chem. 2010, 49, 8179. (223) Chen, Z.; Concepcion, J. J.; Hu, X.; Yang, W.; Hoertz, P. G.; Meyer, T. J. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 7225. (224) Tamaki, Y.; Vannucci, A. K.; Dares, C. J.; Binstead, R. A.; Meyer, T. J. J. Am. Chem. Soc. 2014, 136, 6854. (225) Norris, M. R.; Conception, J. J.; Harrison, D. P.; Binstead, R. A.; Ashford, D. L.; Fang, Z.; Templeton, J. L.; Meyer, T. J. J. Am. Chem. Soc. 2013, 135, 2080. (226) Kaveevivitchai, N.; Zong, R.; Tseng, H.-W.; Chitta, R.; Thummel, R. P. Inorg. Chem. 2012, 51, 2930. (227) Kaveevivitchai, N.; Chitta, R.; Zong, R.; El Ojaimi, M.; Thummel, R. P. J. Am. Chem. Soc. 2012, 134, 10721. (228) Duan, L.; Xu, Y.; Tong, L.; Sun, L. ChemSusChem 2011, 4, 238. (229) Kiyota, J.; Yokoyama, J.; Yoshida, M.; Masaoka, S.; Sakai, K. Chem. Lett. 2010, 39, 1146. (230) Maji, S.; López, I.; Bozoglian, F.; Benet-Buchholz, J.; Llobet, A. Inorg. Chem. 2013, 52, 3591. (231) Marelius, D. C.; Bhagan, S.; Charboneau, D. J.; Schroeder, K. M.; Kamdar, J. M.; McGettigan, A. R.; Freeman, B. J.; Moore, C. E.; Rheingold, A. L.; Cooksy, A. L.; Smith, D. K.; Paul, J. J.; Papish, E. T.; Grotjahn, D. B. Eur. J. Inorg. Chem. 2014, 676. (232) López, I.; Ertem, M. Z.; Maji, S.; Benet-Buchholz, J.; Keidel, A.; Kuhlmann, U.; Hildebrandt, P.; Cramer, C. J.; Batista, V. S.; Llobet, A. Angew. Chem., Int. Ed. 2014, 53, 205. (233) Roeser, S.; Farràs, P.; Bozoglian, F.; Martinez-Belmonte, M.; Benet-Buchholz, J.; Llobet, A. ChemSusChem 2011, 4, 197. (234) Yoshida, M.; Masaoka, S.; Sakai, K. Chem. Lett. 2009, 38, 702. (235) Radaram, B.; Ivie, J. A.; Singh, W. M.; Grudzien, R. M.; Reibenspies, J. H.; Webster, C. E.; Zhao, X. Inorg. Chem. 2011, 50, 10564. (236) Suzuki, M. Acc. Chem. Res. 2007, 40, 609. (237) Que, L., Jr.; Dong, Y. Acc. Chem. Res. 1996, 29, 190. (238) Vennampalli, M.; Liang, G.; Webster, C. E.; Zhao, X. Eur. J. Inorg. Chem. 2014, 715. (239) For recent reviews on carbenes, see: (a) Díez-González, S.; Marion, N.; Nolan, S. P. Chem. Rev. 2009, 109, 3612. (b) Garrison, J. C.; Youngs, W. J. Chem. Rev. 2005, 105, 3978. (c) Herrmann, W. A. Angew. Chem., Int. Ed. 2002, 41, 1290. (d) Hahn, F. E.; Jahnke, M. C. Angew. Chem., Int. Ed. 2008, 47, 3122. (240) For recent reviews on abnormal carbenes, see: (a) Schuster, O.; Yang, L.; Raubenheimer, H. G.; Albrecht, M. Chem. Rev. 2009, 109, 3445. (b) Donnelly, K. F.; Petronilho, A.; Albrecht, M. Chem. Commun. 2013, 49, 1145. (c) Crabtree, R. H. Coord. Chem. Rev. 2013, 257, 755. (241) Bernet, L.; Lalrempuia, R.; Ghattas, W.; Müller-Bunz, H.; Vigara, L.; Llobet, A.; Albrecht, M. Chem. Commun. 2011, 47, 8058. (242) Durham, B.; Walsh, J. L.; Carter, C. L.; Meyer, T. J. Inorg. Chem. 1980, 19, 860. (243) Pinnick, D. V.; Durham, B. Inorg. Chem. 1984, 23, 1440. (244) Balzani, V.; Juris, A. Coord. Chem. Rev. 2001, 211, 97. (245) Porter, G. B.; Sparks, R. H. J. Photochem. 1980, 13, 123. (246) Durham, B.; Wilson, S. R.; Hodgson, D. J.; Meyer, T. J. J. Am. Chem. Soc. 1980, 102, 600. (247) Dobson, J. C.; Meyer, T. J. Inorg. Chem. 1988, 27, 3283. (248) Sala, X.; Ertem, M. Z.; Vigara, L.; Todorova, T. K.; Chen, W.; Rocha, R. C.; Aquilante, F.; Cramer, C. J.; Gagliardi, L.; Llobet, A. Angew. Chem., Int. Ed. 2010, 49, 7745. (249) Planas, N.; Vigara, L.; Cady, C.; Miró, P.; Huang, P.; Hammarström, L.; Styring, S.; Leidel, N.; Dau, H.; Haumann, M.; Gagliardi, L.; Cramer, C. J.; Llobet, A. Inorg. Chem. 2011, 50, 11134. (250) Yamazaki, H.; Hakamata, T.; Komi, M.; Yagi, M. J. Am. Chem. Soc. 2011, 133, 8846.

(251) Boyer, J. L.; Polyansky, D. E.; Szalda, D. J.; Zong, R.; Thummel, R. P.; Fujita, E. Angew. Chem., Int. Ed. 2011, 50, 12600. (252) Hirahara, M.; Ertem, M. Z.; Komi, M.; Yamazaki, H.; Cramer, C. J.; Yagi, M. Inorg. Chem. 2013, 52, 6354. (253) Padhi, S. K.; Fukuda, R.; Ehara, M.; Tanaka, K. Inorg. Chem. 2012, 51, 5386. (254) Kohl, S. W.; Weiner, L.; Schwartsburd, L.; Konstantinovski, L.; Shimon, L. J. W.; Ben-David, Y.; Iron, M. A.; Milstein, D. Science 2009, 324, 74. (255) Yang, X.; Hall, M. B. J. Am. Chem. Soc. 2010, 132, 120. (256) Duan, L.; Fischer, A.; Xu, Y.; Sun, L. J. Am. Chem. Soc. 2009, 131, 10397. (257) Duan, L.; Xu, Y.; Zhang, P.; Wang, M.; Sun, L. Inorg. Chem. 2010, 49, 209. (258) Wang, L.; Duan, L.; Tong, L.; Sun, L. J. Catal. 2013, 306, 129. (259) Li, F.; Jiang, Y.; Zhang, B.; Huang, F.; Gao, Y.; Sun, L. Angew. Chem., Int. Ed. 2012, 51, 2417. (260) Duan, L.; Wang, L.; Inge, A. K.; Fischer, A.; Zou, X.; Sun, L. Inorg. Chem. 2013, 52, 7844. (261) Wang, L.; Duan, L.; Stewart, B.; Pu, M.; Liu, J.; Privalov, T.; Sun, L. J. Am. Chem. Soc. 2012, 134, 18868. (262) Nyhlén, J.; Duan, L.; Åkermark, B.; Sun, L.; Privalov, T. Angew. Chem., Int. Ed. 2010, 49, 1773. (263) Tong, L.; Duan, L.; Xu, Y.; Privalov, T.; Sun, L. Angew. Chem., Int. Ed. 2011, 50, 445. (264) Li, B.; Li, F.; Bai, S.; Wang, Z.; Sun, L.; Yang, Q.; Li, C. Energy Environ. Sci. 2012, 5, 8229. (265) Duan, L.; Bozoglian, F.; Mandal, S.; Stewart, B.; Privalov, T.; Llobet, A.; Sun, L. Nat. Chem. 2012, 4, 418. (266) Li, L.; Duan, L.; Xu, Y.; Gorlov, M.; Hagfeldt, A.; Sun, L. Chem. Commun. 2010, 46, 7307. (267) Li, F.; Zhang, B.; Li, X.; Jiang, Y.; Chen, L.; Li, Y.; Sun, L. Angew. Chem., Int. Ed. 2011, 50, 12276. (268) Tong, L.; Göthelid, M.; Sun, L. Chem. Commun. 2012, 48, 10025. (269) Gao, Y.; Ding, X.; Liu, J.; Wang, L.; Lu, Z.; Li, L.; Sun, L. J. Am. Chem. Soc. 2013, 135, 4219. (270) Gao, Y.; Zhang, L.; Ding, X.; Sun, L. Phys. Chem. Chem. Phys. 2014, 16, 12008. (271) Duan, L.; Moyses Araujo, C.; Ahlquist, M. S.G.; Sun, L. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 15584. (272) Jiang, Y.; Li, F.; Zhang, B.; Li, X.; Wang, X.; Huang, F.; Sun, L. Angew. Chem., Int. Ed. 2013, 52, 3398. (273) Duan, L.; Xu, Y.; Gorlov, M.; Tong, L.; Andersson, S.; Sun, L. Chem.Eur. J. 2010, 16, 4659. (274) An, J.; Duan, L.; Sun, L. Faraday Discuss. 2012, 155, 267. (275) Tong, L.; Inge, A. K.; Duan, L.; Wang, L.; Zou, X.; Sun, L. Inorg. Chem. 2013, 52, 2505. (276) Kärkäs, M. D.; Åkermark, T.; Johnston, E. V.; Karim, S. R.; Laine, T. M.; Lee, B.-L.; Åkermark, T.; Privalov, T.; Åkermark, B. Angew. Chem., Int. Ed. 2012, 11589. (277) Verho, O.; Dilenstam, M. D. V.; Kärkäs, M. D.; Johnston, E. V.; Åkermark, T.; Bäckvall, J.-E.; Åkermark, B. Chem.Eur. J. 2012, 18, 16947. (278) Tong, L.; Wang, Y.; Duan, L.; Xu, Y.; Cheng, X.; Fischer, A.; Ahlquist, M. S. G.; Sun, L. Inorg. Chem. 2012, 51, 3388. (279) Valeur, E.; Bradley, M. Chem. Soc. Rev. 2009, 38, 606. (280) Berggren, G.; Betul Kaynak, F.; Anderlund, M. F.; Eriksson, L.; Åkermark, B. Acta Crystallogr. 2007, E63, m2672. (281) Collins, T. J.; Gordon-Wylie, S. W. J. Am. Chem. Soc. 1989, 111, 4511. (282) Singh, A. K.; Balamurugan, V.; Mukherjee, R. Inorg. Chem. 2003, 42, 6497. (283) Belda, O.; Moberg, C. Coord. Chem. Rev. 2005, 249, 727. (284) Kärkäs, M. D.; Åkermark, T.; Chen, H.; Sun, J.; Åkermark, B. Angew. Chem., Int. Ed. 2013, 52, 4189. (285) McDaniel, N. D.; Coughlin, F. J.; Tinker, L. L.; Bernhard, S. J. Am. Chem. Soc. 2008, 130, 210. 11994

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(286) Vilella, L.; Vidossich, P.; Balcells, D.; Lledós, A. Dalton Trans. 2011, 40, 11241. (287) Hull, J. F.; Balcells, D.; Blakemore, J. D.; Incarvito, C. D.; Eisenstein, O.; Brudvig, G. W.; Crabtree, R. H. J. Am. Chem. Soc. 2009, 131, 8730. (288) Blakemore, J. D.; Schley, N. D.; Balcells, D.; Hull, J. F.; Olack, G. W.; Incarvito, C. D.; Eisenstein, O.; Brudvig, G. W.; Crabtree, R. H. J. Am. Chem. Soc. 2010, 132, 16017. (289) The formation of an iridium(III) hydroperoxide (IrIII−OOH) intermediate, resulting from H2O (or hydroxide) nucleophilic attack on a IrVO species, has previously been observed on a metal oxide surface, see: Sivasankar, N.; Weare, W. W.; Frei, H. J. Am. Chem. Soc. 2011, 133, 12976. (290) Park-Gehrke, L. S.; Freudenthal, J.; Kaminsky, W.; DiPasquale, A. G.; Mayer, J. M. Dalton Trans. 2009, 1972. (291) Moore, G. F.; Blakemore, J. D.; Milot, R. L.; Hull, J. F.; Song, H.-E.; Cai, L.; Schmuttenmaer, C. A.; Crabtree, R. H.; Brudvig, G. W. Energy Environ. Sci. 2011, 4, 2389. (292) Brewster, T. P.; Blakemore, J. D.; Schley, N. D.; Incarvito, C. D.; Hazari, N.; Brudvig, G. W.; Crabtree, R. H. Organometallics 2011, 30, 965. (293) Hetterscheid, D. G. H.; Reek, J. N. H. Chem. Commun. 2011, 47, 2712. (294) Volpe, A.; Sartorel, A.; Tubaro, C.; Meneghini, L.; Di Valentin, M.; Graiff, C.; Bonchio, M. Eur. J. Inorg. Chem. 2014, 665. (295) Lalrempuia, R.; McDaniel, N. D.; Müller-Bunz, H.; Bernhard, S.; Albrecht, M. Angew. Chem., Int. Ed. 2010, 49, 9765. (296) Hetterscheid, D. G. H.; van der Vlugt, J. I.; de Bruin, B.; Reek, J. N. H. Angew. Chem., Int. Ed. 2009, 48, 8178. (297) Petronilho, A.; Rahman, M.; Woods, J. A.; Al-Sayyed, H.; Müller-Bunz, H.; MacElroy, J. M. D.; Bernhard, S.; Albrecht, M. Dalton Trans. 2012, 41, 13074. (298) Dzik, W. I.; Calvo, S. E.; Reek, J. N. H.; Lutz, M.; Ciriano, M. A.; Tejel, C.; Hetterscheid, D. G. H.; de Bruin, B. Organometallics 2011, 30, 372. (299) Codolà, Z.; Cardoso, J. M. S.; Royo, B.; Costas, M.; LloretFillol, J. Chem.Eur. J. 2013, 19, 7203. (300) Hintermair, U.; Hashmi, S. A.; Elimelech, M.; Crabtree, R. H. J. Am. Chem. Soc. 2012, 134, 9785. (301) Parent, A. R.; Brewster, T. P.; De Wolf, W.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chem. 2012, 51, 6147. (302) Parent, A. R.; Blakemore, J. D.; Brudvig, G. W.; Crabtree, R. H. Chem. Commun. 2011, 47, 11745. (303) Graeupner, J.; Hintermair, U.; Huang, D. L.; Thomsen, J. M.; Takase, M.; Campos, J.; Hashmi, S. M.; Elimelech, M.; Brudvig, G. W.; Crabtree, R. H. Organometallics 2013, 32, 5384. (304) Hintermair, U.; Sheehan, S. W.; Parent, A. R.; Ess, D. H.; Richens, D. T.; Vaccaro, P. H.; Brudvig, G. W.; Crabtree, R. H. J. Am. Chem. Soc. 2013, 135, 10837. (305) (a) Harriman, A.; Pickering, I. J.; Thomas, J. M. J. Chem. Soc., Faraday Trans. 1 1988, 84, 2795. (b) Nahor, C. S.; Hapiot, P.; Neta, P.; Harriman, A. J. Phys. Chem. 1991, 95, 616. (c) Hara, M.; Waraksa, C. C.; Lean, J. T.; Lewis, B. A.; Mallouk, T. E. J. Phys. Chem. A 2000, 104, 5275. (306) Kushner-Lenhoff, M. N.; Blakemore, J. D.; Schley, N. D.; Crabtree, R. H.; Brudvig, G. W. Dalton Trans. 2013, 42, 3617. (307) Junge, H.; Marquet, N.; Kammer, A.; Denurra, S.; Bauer, M.; Wohlrab, S.; Gärtner, F.; Pohl, M.-M.; Spannenberg, A.; Gladiali, S.; Beller, M. Chem.Eur. J. 2012, 18, 12749. (308) Marquet, N.; Gärtner, F.; Losse, S.; Pohl, M.-M.; Junge, H.; Beller, M. ChemSusChem 2011, 4, 1598. (309) Blakemore, J. D.; Schley, N. D.; Olack, G. W.; Incarvito, C. D.; Brudvig, G. W.; Crabtree, R. H. Chem. Sci. 2011, 2, 94. (310) Schley, N. D.; Blakemore, J. D.; Subbaiyan, N. K.; Incarvito, C. D.; D’Souza, F.; Crabtree, R. H.; Brudvig, G. W. J. Am. Chem. Soc. 2011, 133, 10473. (311) Blakemore, J. D.; Schley, N. D.; Kushner-Lenhoff, M. N.; Winter, A. M.; D’Souza, F.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chem. 2012, 51, 7749.

(312) Blakemore, J. D.; Mara, M. W.; Kushner-Lenhoff, M. N.; Schley, N. D.; Konezny, S. J.; Rivalta, I.; Negre, C. F. A.; Snoeberger, R. C.; Kokhan, O.; Huang, J.; Stickrath, A.; Tran, L. A.; Parr, M. L.; Chen, L. X.; Tiede, D. M.; Batista, V. S.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chem. 2013, 52, 1860. (313) Grotjahn, D. B.; Brown, D. B.; Martin, J. K.; Marelius, D. C.; Abadjian, M.-C.; Tran, H. N.; Kalyuzhny, G.; Vecchio, K. S.; Specht, Z. G.; Cortes-Llamas, S. A.; Miranda-Soto, V.; van Niekerk, C.; Moore, C. E.; Rheingold, A. L. J. Am. Chem. Soc. 2011, 133, 19024. (314) Wöhler, L.; Witzmann, W. Z. Anorg. Chem. 1908, 57, 323. (315) Harriman, A.; Thomas, J. M.; Millward, G. R. New. J. Chem. 1987, 11, 757. (316) For examples of IrOx nanoparticles formation, see, for example: (a) Hoertz, P. G.; Kim, Y.-I.; Youngblood, W. J.; Mallouk, T. E. J. Phys. Chem. B 2007, 111, 6845. (b) Nakagawa, T.; Bjorge, N. S.; Murray, R. W. J. Am. Chem. Soc. 2009, 131, 15578. (c) Zhao, Y.; HernandezPagan, E. A.; Vargas-Barbosa, N. M.; Dysart, J. L.; Mallouk, T. E. J. Phys. Chem. Lett. 2011, 2, 402. (317) It should be noted that the appearance of absorption bands at ∼580 nm is not necessarily indicative of iridium nanoparticle formation, and in situ formation of iridium nanoparticles is strongly dependent on the exact reaction conditions in accordance with ref 300. This electronic transition can arise from d−d transitions in molecular octahedral IrIV complexes, see, for example: (a) Uemura, S.; Spencer, A.; Wilkinson, G. J. Chem. Soc., Dalton Trans. 1973, 2565. (b) CastilloBlum, S. E.; Richens, D. T.; Sykes, A. G. J. Chem. Soc., Chem. Commun. 1986, 1120. (c) Castillo-Blum, S. E.; Richens, D. T.; Sykes, A. G. Inorg. Chem. 1989, 28, 954. (d) Savini, A.; Bellachioma, G.; Bolaño, S.; Rocchigiani, L.; Zuccaccia, C.; Zuccaccia, D.; Macchioni, A. ChemSusChem 2012, 5, 1415. (318) Hong, D.; Murakami, M.; Yamada, Y.; Fukuzumi, S. Energy Environ. Sci. 2012, 5, 5708. (319) Lewandowska-Andralojc, A.; Polyansky, D. E.; Wang, C.-H.; Wang, W.-H.; Himeda, Y.; Fujita, E. Phys. Chem. Chem. Phys. 2014, 16, 11976. (320) DePasquale, J.; Nieto, I.; Reuther, L. E.; Herbst-Gervasoni, C. J.; Paul, J. J.; Mochalin, V.; Zeller, M.; Thomas, C. M.; Addison, A. W.; Papish, E. T. Inorg. Chem. 2013, 52, 9175. (321) Joya, K. S.; Subbaiyan, N. E.; D’Souza, F.; de Groot, H. J. M. Angew. Chem., Int. Ed. 2012, 51, 9601. (322) Savini, A.; Bellachioma, G.; Ciancaleoni, G.; Zuccaccia, C.; Zuccaccia, D.; Macchioni, A. Chem. Commun. 2010, 46, 9218. (323) Bucci, A.; Savini, A.; Rocchigiani, L.; Zuccaccia, C.; Rizzato, S.; Albinati, A.; Llobet, A.; Macchioni, A. Organometallics 2012, 31, 8071. (324) Savini, A.; Belanzoni, P.; Bellachioma, G.; Zuccaccia, C.; Zuccaccia, D.; Macchioni, A. Green Chem. 2011, 13, 3360. (325) Zuccaccia, C.; Bellachioma, G.; Bolaño, S.; Rocchigiani, L.; Savini, A.; Macchioni, A. Eur. J. Inorg. Chem. 2012, 1462. (326) Hirai, K.; Nutton, A.; Maitlis, P. M. J. Mol. Catal. 1981, 10, 203. (327) Mukhopadhyay, S.; Mandal, S. K.; Bhaduri, S.; Armstrong, W. H. Chem. Rev. 2004, 104, 3981. (328) (a) Sarneski, J. E.; Didiuk, M.; Thorp, H. H.; Crabtree, R. H.; Brudvig, G. W.; Faller, J. W.; Schulte, G. K. Inorg. Chem. 1991, 30, 2833. (b) Blackman, A. G.; Huffman, J. C.; Lobkovsky, E. B.; Christou, G. J. Chem. Soc., Chem. Commun. 1991, 989. (c) Vincent, J. B.; Folting, K.; Huffman, J. C.; Christou, G. Biochem. Soc. Trans. 1988, 16, 822. (d) Mukhopadhyay, S.; Staples, R. J.; Armstrong, W. H. Chem.Commun. 2002, 864. (e) Ménage, S.; Girerd, J. J.; Gleizes, A. J. Chem. Soc. Chem. Commun. 1988, 431. (329) (a) Collomb, M.-N.; Deronzier, A.; Richardot, A.; Pécaut, J. New J. Chem. 1999, 23, 351. (b) Baffert, C.; Collomb, M.-N.; Deronzier, A.; Pécaut, J.; Limburg, J.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chem. 2002, 41, 1404. (330) (a) Stebler, M.; Ludi, A.; Bürgi, H. B. Inorg. Chem. 1986, 25, 4743. (b) Ruiz, R.; Sangregorio, C.; Caneschi, A.; Rossi, P.; Gaspar, A. B.; Real, J. A.; Munõz, M. C. Inorg. Chem. Commun. 2000, 3, 361. (331) (a) Goodson, P. A.; Hodgson, D. J.; Michelsen, K. Inorg. Chim. Acta 1990, 172, 49. (b) Goodson, P. A.; Hodgson, D. J.; Glerup, J.; Michelsen, K.; Weihe, H. Inorg. Chim. Acta 1992, 197, 141. 11995

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

C. W.; Shinopoulos, K. E.; Crabtree, R. H.; Brudvig, G. W. Dalton Trans. 2010, 39, 3985. (344) Chen, H.; Tagore, R.; Das, S.; Incarvito, C.; Faller, J. W.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chem. 2005, 44, 7661. (345) Yamazaki, H.; Igarashi, S.; Nagata, T.; Yagi, M. Inorg. Chem. 2012, 51, 1530. (346) A recent mechanistic computational study has also been published, see: Hatakeyama, M.; Nakata, H.; Wakabayashi, M.; Yokojima, S.; Nakamura, S. J. Phys. Chem. A 2012, 116, 7089. (347) Chen, H.; Tagore, R.; Olack, G.; Vrettos, J. S.; Weng, T.-C.; Penner-Hahn, J.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chem. 2007, 46, 34. (348) Tagore, R.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chem. 2008, 47, 1815. (349) Limburg, J.; Brudvig, G. W.; Crabtree, R. H. J. Am. Chem. Soc. 1997, 119, 2761. (350) Tagore, R.; Chen, H.; Zhang, H.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chim. Acta 2007, 360, 2983. (351) Kurz, P.; Berggren, G.; Anderlund, M. F.; Styring, S. Dalton Trans. 2007, 4258. (352) Baffert, C.; Romain, S.; Richardot, A.; Lepretre, J.-C.; Lefebvre, B.; Deronzier, A.; Collomb, M.-N. J. Am. Chem. Soc. 2005, 127, 13694. (353) Yagi, M.; Narita, K. J. Am. Chem. Soc. 2004, 126, 8084. (354) Chen, H.; Faller, J. W.; Crabtree, R. H.; Brudvig, G. W. J. Am. Chem. Soc. 2004, 126, 7345. (355) Chen, H.; Collomb, M.-N.; Duboc, C.; Blondin, G.; Rivière, E.; Faller, J. W.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chem. 2005, 44, 9567. (356) Gao, Y.; Crabtree, R. H.; Brudvig, G. W. Inorg. Chem. 2012, 51, 4043. (357) Narita, K.; Kuwabara, T.; Sone, K.; Shimizu, K.-i.; Yagi, M. J. Phys. Chem. B 2006, 110, 23107. (358) Yagi, M.; Toda, M.; Yamada, S.; Yamazaki, H. Chem. Commun. 2010, 46, 8594. (359) Hirahara, M.; Yamazaki, H.; Yamada, S.; Matsubara, K.; Saito, K.; Yui, T.; Yagi, M. Catal. Sci. Technol. 2013, 3, 1776. (360) Yamazaki, H.; Nagata, T.; Yagi, M. Photochem. Photobiol. Sci. 2009, 8, 204. (361) Yagi, M.; Narita, K.; Maruyama, S.; Sone, K.; Kuwabara, T.; Shimizu, K. Biochim. Biophys. Acta, Bioenerg. 2007, 1767, 660. (362) Li, G.; Sproviero, E. M.; Snoeberger, R. C., III; Iguchi, N.; Blakemore, J. D.; Crabtree, R. H.; Brudvig, G. W.; Batista, V. S. Energy Environ. Sci. 2009, 2, 230. (363) Li, G.; Sproviero, E. M.; McNamara, W. R.; Snoeberger, R. C., III; Crabtree, R. H.; Brudvig, G. W.; Batista, V. S. J. Phys. Chem. B 2010, 114, 14214. (364) Nepal, B.; Das, S. Angew. Chem., Int. Ed. 2013, 52, 7224. (365) Najafpour, M. M.; Moghaddam, A. N.; Dau, H.; Zaharieva, I. J. Am. Chem. Soc. 2014, 136, 7245. (366) Baffert, C.; Collomb, M.-N.; Deronzier, A.; KjaergaardKnudsen, S.; Latour, J.-M.; Lund, K. H.; McKenzie, C. J.; Mortensen, M.; Preuss Nielsen, L.; Thorup, N. Dalton Trans. 2003, 1765. (367) Poulsen, A. K.; Rompel, A.; McKenzie, C. J. Angew. Chem., Int. Ed. 2005, 44, 6916. (368) Seidler-Egdal, R. K.; Nielsen, A.; Bond, A. D.; Bjerrum, M. J.; McKenzie, C. J. Dalton Trans. 2011, 40, 3849. (369) Sameera, W. M. C.; McKenzie, C. J.; McGrady, J. E. Dalton Trans. 2011, 40, 3859. (370) Rardin, R. L.; Tolman, W. B.; Lippard, S. J. New J. Chem. 1991, 15, 417. (371) Berggren, G.; Thapper, A.; Huang, P.; Kurz, P.; Eriksson, L.; Styring, S.; Anderlund, M. F. Dalton Trans. 2009, 10044. (372) Berggren, G.; Thapper, A.; Huang, P.; Eriksson, L.; Styring, S.; Anderlund, M. F. Inorg. Chem. 2011, 50, 3425. (373) Beckmann, K.; Uchtenhagen, H.; Berggren, G.; Anderlund, M. F.; Thapper, A.; Messinger, J.; Styring, S.; Kurz, P. Energy Environ. Sci. 2008, 1, 668. (374) Kurz, P. Dalton Trans. 2009, 6103.

(c) Arulsamy, N.; Glerup, J.; Hazell, A.; Hodgson, D. J.; McKenzie, C. J.; Toftlund, H. Inorg. Chem. 1994, 33, 3023. (d) Gohdes, J. W.; Armstrong, W. H. Inorg. Chem. 1992, 31, 368. (e) Pal, S.; Olmstead, M. M.; Armstrong, W. H. Inorg. Chem. 1995, 34, 4708. (f) Bossek, U.; Weyhermüller, T.; Wieghardt, K.; Nuber, B.; Weiss, J. J. Am. Chem. Soc. 1990, 112, 6387. (g) Canãda-Vilalta, C.; Huffman, J. C.; Streib, W. E.; Davidson, E. R.; Christou, G. Polyhedron 2001, 20, 1375. (h) Wieghardt, K.; Bossek, U.; Ventur, D.; Weiss, J. J. Chem. Soc. Chem. Commun. 1985, 347. (i) Wieghardt, K.; Bossek, U.; Nuber, B.; Weiss, J.; Bonvoisin, J.; Corbella, M.; Vitols, S. E.; Girerd, J. J. J. Am. Chem. Soc. 1988, 110, 7398. (332) (a) Bhula, R.; Gainsford, G. J.; Weatherburn, D. C. J. Am. Chem. Soc. 1988, 110, 7550. (b) Kitajima, N.; Osawa, M.; Imai, S.; Fujisawa, K.; Moro-oka, Y.; Heerwegh, K.; Reed, C. A.; Boyd, P. D. W. Inorg. Chem. 1994, 33, 4613. (c) Yano, S.; Doi, M.; Tamakoshi, S.; Mori, W.; Mikuriya, M.; Ichimura, A.; Kinoshita, I.; Yamamoto, Y.; Tanase, T. Chem. Commun. 1997, 997. (d) Yoshino, A.; Miyagi, T.; Asato, E.; Mikuriya, M.; Sakata, Y.; Sugiura, K.-I.; Iwasaki, K.; Hino, S. Chem. Commun. 2000, 1475. (e) Alexiou, M.; Dendrinou-Samara, C.; Karagianni, A.; Biswas, S.; Zaleski, C. M.; Kampf, J.; Yoder, D.; PennerHahn, J. E.; Pecoraro, V. L.; Kessissoglou, D. P. Inorg. Chem. 2003, 42, 2185. (f) Auger, N.; Girerd, J.-J.; Corbella, M.; Gleizes, A.; Zimmermann, J. L. J. Am. Chem. Soc. 1990, 112, 448. (g) Sarneski, J. E.; Thorp, H. H.; Brudvig, G. W.; Crabtree, R. H.; Schulte, G. K. J. Am. Chem. Soc. 1990, 112, 7255. (333) (a) Vincent, J. B.; Christmas, C.; Chang, H. R.; Li, Q.; Boyd, P. D. W.; Huffman, J. C.; Hendrickson, D. N.; Christou, G. J. Am. Chem. Soc. 1989, 111, 2086. (b) Cinco, R. M.; Rompel, A.; Visser, H.; Aromi, G.; Christou, G.; Sauer, K.; Klein, M. P.; Yachandra, V. K. Inorg. Chem. 1999, 38, 5988. (c) Hagen, K. S.; Westmoreland, T. D.; Scott, M. J.; Armstrong, W. H. J. Am. Chem. Soc. 1989, 111, 1907. (d) Dubé, C. E.; Wright, D. W.; Bonitatebus, P. J., Jr.; Pal, S.; Armstrong, W. H. J. Am. Chem. Soc. 1998, 120, 3704. (e) Blondin, G.; Davydov, R.; Philouze, C.; Charlot, M.-F.; Styring, T.; Åkermark, B.; Girerd, J.-J.; Boussac, A. J. Chem. Soc., Dalton Trans. 1997, 4069. (f) Stibrany, R. T.; Gorun, S. M. Angew. Chem., Int. Ed. Engl. 1990, 29, 1156. (g) Grillo, V. A.; Knapp, M. J.; Bollinger, J. C.; Hendrickson, D. N.; Christou, G. Angew. Chem., Int. Ed. Engl. 1996, 35, 1818. (h) Chan, M. K.; Armstrong, W. H. J. Am. Chem. Soc. 1990, 112, 4985. (i) Bouwman, E.; Bolcar, M. A.; Libby, E.; Huffman, J. C.; Folting, K.; Christou, G. Inorg. Chem. 1992, 31, 5185. (j) Afrati, T.; Dendrinou-Samara, C.; Raptopoulou, C. P.; Terzis, A.; Tangoulis, V.; Kessissoglou, D. P. Angew. Chem., Int. Ed. 2002, 41, 2148. (334) Herrero, C.; Quaranta, A.; Protti, S.; Leibl, W.; Rutherford, A. W.; Fallahpour, R.; Charlot, M.-F.; Aukauloo, A. Chem.Asian J. 2011, 6, 1335. (335) Castillo, C. E.; Romain, S.; Retegan, M.; Leprêtre, J.-C.; Chauvin, J.; Duboc, C.; Fortage, J.; Deronzier, A.; Collomb, M.-N. Eur. J. Inorg. Chem. 2012, 5485. (336) Sun, L.; Berglund, H.; Davydov, R.; Norrby, T.; Hammarström, L.; Korall, P.; Börje, A.; Philouze, C.; Berg, K.; Tran, A.; Andersson, M.; Stenhagen, G.; Martensson, J.; Almgren, M.; Styring, S.; Åkermark, B. J. Am. Chem. Soc. 1997, 119, 6996. (337) Sun, L.; Hammarström, L.; Norrby, T.; Berglund, H.; Davydov, R.; Andersson, M.; Börje, A.; Korall, P.; Philouze, C.; Almgren, M.; Styring, S.; Åkermark, B. Chem. Commun. 1997, 607. (338) Huang, P.; Magnuson, A.; Lomoth, R.; Abrahamsson, M.; Tamm, M.; Sun, L.; van Rotterdam, B.; Park, J.; Hammarström, L.; Åkermark, B.; Styring, S. J. Inorg. Biochem. 2002, 91, 159. (339) Naruta, Y.; Sasayama, M.; Sasaki, T. Angew. Chem., Int. Ed. Engl. 1994, 33, 1839. (340) Shimasaki, Y.; Nagano, T.; Takesue, H.; Ye, B.-H.; Tani, F.; Naruta, Y. Angew. Chem., Int. Ed. 2004, 43, 98. (341) Limburg, J.; Vrettos, J. S.; Liable-Sands, L. M.; Rheingold, A. L.; Crabtree, R. H.; Brudvig, G. W. Science 1999, 283, 1524. (342) Limburg, J.; Vrettos, J. S.; Chen, H. Y.; de Paula, J. C.; Crabtree, R. H.; Brudvig, G. W. J. Am. Chem. Soc. 2001, 123, 423. (343) The [(H2O)(tpy)Mn(μ-O)2Mn(tpy)(OH2)]3+ complex may be capable of mediating electrochemical oxidation of H2O, see: Cady, 11996

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(375) Berends, H.-M.; Manke, A.-M.; Näther, C.; Tuczek, F.; Kurz, P. Dalton Trans. 2012, 41, 6215. (376) Young, K. J.; Takase, M. K.; Brudvig, G. W. Inorg. Chem. 2013, 52, 7615. (377) Gao, Y.; Liu, J.; Wang, M.; Na, Y.; Åkermark, B.; Sun, L. Tetrahedron 2007, 63, 1987. (378) Privalov, T.; Sun, L.; Åkermark, B.; Liu, J.; Gao, Y.; Wang, M. Inorg. Chem. 2007, 46, 7075. (379) Gao, Y.; Åkermark, T.; Liu, J.; Sun, L.; Åkermark, B. J. Am. Chem. Soc. 2009, 131, 8726. (380) Karlsson, E. A.; Lee, B.-L.; Åkermark, T.; Johnston, E. V.; Kärkäs, M. D.; Sun, J.; Hansson, Ö .; Bäckvall, J.-E.; Åkermark, B. Angew. Chem., Int. Ed. 2011, 50, 11715. (381) Karlsson, E. A.; Lee, B.-L.; Liao, R.-Z.; Åkermark, T.; Kärkäs, M. D.; Saavedra Becerril, V.; Siegbahn, P. E. M.; Zou, X.; Abrahamsson, M.; Åkermark, B. ChemPlusChem 2014, 79, 936. (382) Arafa, W. A. A.; Kärkäs, M. D.; Lee, B.-L.; Åkermark, T.; Liao, R.-Z.; Berends, H.-M.; Messinger, J.; Siegbahn, P. E. M.; Åkermark, B. Phys. Chem. Chem. Phys. 2014, 16, 11950. (383) Lundberg, M.; Blomberg, M. R. A.; Siegbahn, P. E. M. Inorg. Chem. 2004, 43, 264. (384) Anderlund, M. F.; Högblom, J.; Shi, W.; Huang, P.; Eriksson, L.; Weihe, H.; Styring, S.; Åkermark, B.; Lomoth, R.; Magnuson, A. Eur. J. Inorg. Chem. 2006, 5033. (385) Plietker, B. Iron Catalysis in Organic Chemistry; Wiley-WCH Verlag GmbH & Co. KGaA: Weinheim, 2008. (386) Gopalaiah, K. Chem. Rev. 2013, 113, 3248. (387) Bolm, C.; Legros, J.; Le Paih, J.; Zani, L. Chem. Rev. 2004, 104, 6217. (388) Junge, K.; Schröder, K.; Beller, M. Chem. Commun. 2011, 47, 4849. (389) Liu, L.-X. Curr. Org. Chem. 2010, 14, 1099. (390) For reviews on high-valent iron complexes, see: (a) Nam, W. Acc. Chem. Res. 2007, 40, 522. (b) Hohenberger, J.; Ray, K.; Meyer, K. Nat. Commun. 2012, 3, 1. (391) For selected recent examples on development and studies on high-valent iron complexes, see: (a) Fukuzumi, S.; Morimoto, Y.; Kotani, H.; Naumov, P.; Lee, Y.-M.; Nam, W. Nat. Chem. 2010, 2, 756. (b) Morimoto, Y.; Kotani, H.; Park, J.; Lee, Y.-M.; Nam, W.; Fukuzumi, S. J. Am. Chem. Soc. 2011, 133, 403. (c) Afanasiev, P.; Kudrik, E. V.; Albrieux, F.; Briois, V.; Koifman, O. I.; Sorokin, A. B. Chem. Commun. 2012, 48, 6088. (d) Wang, Y.; Janardanan, D.; Usharani, D.; Han, K.; Que, L., Jr.; Shaik, S. ACS Catal. 2013, 3, 1334. (392) Ellis, W.; McDaniel, M. D.; Bernhard, S.; Collins, T. J. J. Am. Chem. Soc. 2010, 132, 10990. (393) Ertem, M. Z.; Gagliardi, L.; Cramer, C. J. Chem. Sci. 2012, 3, 1293. (394) Liao, R.-Z.; Li, X.-C.; Siegbahn, P. E. M. Eur. J. Inorg. Chem. 2014, 728. (395) Demeter, E. L.; Hilburg, S. L.; Washburn, N. R.; Collins, T. J.; Kitchin, J. R. J. Am. Chem. Soc. 2014, 136, 5603. (396) Fillol, J. L.; Cordolà, Z.; Garcia-Bosch, I.; Gómez, L.; Pla, J. J.; Costas, M. Nat. Chem. 2011, 3, 807. (397) Codolà, Z.; Garcia-Bosch, I.; Acuña-Parés, F.; Prat, I.; Luis, J. M.; Costas, M.; Lloret-Fillol, J. Chem.Eur. J. 2013, 19, 8042. (398) Kasapbasi, E. E.; Whangbo, M.-H. Inorg. Chem. 2012, 51, 10850. (399) Acuña-Parés, F.; Codolá, Z.; Costas, M.; Luis, J. M.; LloretFillol, J. Chem.Eur. J. 2014, 20, 5696. (400) Acuña-Parés, F.; Costas, M.; Luis, J. M.; Lloret-Fillol, J. Inorg. Chem. 2014, 53, 5474. (401) Mitra, M.; Lloret-Fillol, J.; Haukka, M.; Costas, M.; Nordlander, E. Chem. Commun. 2014, 50, 1408. (402) Chen, G.; Chen, L.; Ng, S.-M.; Man, W.-L.; Lau, T.-C. Angew. Chem., Int. Ed. 2013, 52, 1789. (403) A recent paper also demonstrated that the catalytic conditions have a significant influence on the homogeneity and heterogeneity of the catalysis of iron complexes: Hong, D.; Mandal, S.; Yamada, Y.; Lee, Y.-M.; Nam, W.; Llobet, A.; Fukuzumi, S. Inorg. Chem. 2013, 52, 9522.

(404) Coggins, M. K.; Zhang, M.-T.; Vannucci, A. K.; Dares, C. J.; Meyer, T. J. J. Am. Chem. Soc. 2014, 136, 5531. (405) Hoffert, W. A.; Mock, M. T.; Appel, A. M.; Yang, J. Y. Eur. J. Inorg. Chem. 2013, 3846. (406) Ghosh, A.; de Oliveira, F. T.; Yano, T.; Nishioka, T.; Beach, E. S.; Kinoshita, I.; Münck, E.; Ryabov, A. D.; Horwitz, C. P.; Collins, T. J. J. Am. Chem. Soc. 2005, 127, 2505. (407) de Oliveira, F. T.; Chanda, A.; Banerjee, D.; Shan, X.; Mondal, S.; Que, L., Jr.; Bominaar, E. L.; Münck, E.; Collins, T. J. Science 2007, 315, 835. (408) Chanda, A.; Shan, X.; Chakrabarti, M.; Ellis, W. C.; Popescu, D. L.; de Oliveira, F. T.; Wang, D.; Que, L., Jr.; Collins, T. J.; Münck, E.; Bominaar, E. L. Inorg. Chem. 2008, 47, 3669. (409) Kundu, S.; Annavajhala, M.; Kurnikov, I. V.; Ryabov, A. D.; Collins, T. J. Chem.Eur. J. 2012, 18, 10244. (410) Barnett, S. M.; Goldberg, K. I.; Mayer, J. M. Nat. Chem. 2012, 4, 498. (411) Chen, Z.; Meyer, T. J. Angew. Chem., Int. Ed. 2013, 52, 700. (412) Zhang, M.-T.; Chen, Z.; Kang, P.; Meyer, T. J. J. Am. Chem. Soc. 2013, 135, 2048. (413) Zhang, T.; Wang, C.; Liu, S.; Wang, J.-L.; Lin, W. J. Am. Chem. Soc. 2014, 136, 273. (414) Shafirovich, V. Ya.; Khannanov, N. K.; Strelets, V. V. Nouv. J. Chim. 1980, 4, 81. (415) Elizarova, G. L.; Matvienko, L. G.; Lozhkina, N. V.; Parmon, V. N.; Zamaraev, K. I. React. Kinet. Catal. Lett. 1981, 16, 191. (416) Harriman, A.; Porter, G.; Walters, P. J. Chem. Soc., Faraday Trans. 2 1981, 77, 2373. (417) Brunschwig, B. S.; Chou, M. H.; Creutz, C.; Ghosh, P.; Sutin, N. J. Am. Chem. Soc. 1983, 105, 4832. (418) Elizarova, G. L.; Matvienko, L. G.; Lozhkina, N. V.; Parmon, V. N. React. Kinet. Catal. Lett. 1984, 26, 67. (419) Wasylenko, D. J.; Ganesamoorthy, C.; Borau-Garcia, J.; Berlinguette, C. P. Chem. Commun. 2011, 47, 4249. (420) Sun, Y.; Bigi, J. P.; Piro, N. A.; Tang, M. L.; Long, J. R.; Chang, C. J. J. Am. Chem. Soc. 2011, 133, 9212. (421) Stracke, J. J.; Finke, R. G. J. Am. Chem. Soc. 2011, 133, 14872. (422) Wasylenko, D. J.; Palmer, R. D.; Schott, E.; Berlinguette, C. P. Chem. Commun. 2012, 48, 2107. (423) McGuire, R., Jr.; Dogutan, D. K.; Teets, T. S.; Suntivich, J.; Shao-Horn, Y.; Nocera, D. G. Chem. Sci. 2010, 1, 411. (424) Dogutan, D. K.; McGuire, R., Jr.; Nocera, D. G. J. Am. Chem. Soc. 2011, 133, 9178. (425) Ertem, M. Z.; Cramer, C. J. Dalton Trans. 2012, 41, 12213. (426) Lai, W.; Cao, R.; Dong, G.; Shaik, S.; Yao, J.; Chen, H. J. Phys. Chem. Lett. 2012, 3, 2315. (427) Lei, H.; Han, A.; Li, F.; Zhang, M.; Han, Y.; Du, P.; Lai, W.; Cao, R. Phys. Chem. Chem. Phys. 2014, 16, 1883. (428) Wang, D.; Groves, J. T. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 15579. (429) Nakazono, T.; Parent, A. R.; Sakai, K. Chem. Commun. 2013, 49, 6325. (430) Rigsby, M. L.; Mandal, S.; Nam, W.; Spencer, L. C.; Llobet, A.; Stahl, S. S. Chem. Sci. 2012, 3, 3058. (431) Wang, L.-P.; Van Voorhis, T. J. Phys. Chem. Lett. 2011, 2, 2200. (432) Leung, C.-F.; Ng, S.-M.; Ko, C.-C.; Man, W.-L.; Wu, J.; Chen, L.; Lau, T.-C. Energy Environ. Sci. 2012, 5, 7903. (433) Yin, Q.; Tan, J. M.; Besson, C.; Geletti, Y. V.; Musaev, D. G.; Kuznetsov, A. E.; Luo, Z.; Hardcastle, K. I.; Hill, C. L. Science 2010, 328, 342. (434) Lowry, M. S.; Goldsmith, J. I.; Slinker, J. D.; Rohl, R.; Pascal, R. A.; Malliaras, G. G.; Bernhard, S. Chem. Mater. 2005, 17, 5712. (435) Pizzolato, E.; Natali, M.; Posocco, B.; Montellano López, A.; Bazzan, I.; Di Valentin, M.; Galloni, P.; Conte, V.; Bonchio, M.; Scandola, F.; Sartorel, A. Chem. Commun. 2013, 49, 9941. (436) Montellano López, A.; Natali, M.; Pizzolato, E.; Chiorboli, C.; Bonchio, M.; Sartorel, A.; Scandola, F. Phys. Chem. Chem. Phys. 2014, 16, 12000. 11997

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(437) Fu, S.; Liu, Y.; Ding, Y.; Du, X.; Song, F.; Xiang, R.; Ma, B. Chem. Commun. 2014, 50, 2167. (438) For reviews on polyoxometalates, see: (a) Hill, C. L.; ProsserMcCartha, C. M. Coord. Chem. Rev. 1995, 143, 407. (b) Hill, C. L. Chem. Rev. 1998, 98, 1. (c) Long, D.-L.; Burkholder, E.; Cronin, L. Chem. Soc. Rev. 2007, 36, 105. (d) Long, D.-L.; Tsunashima, R.; Cronin, L. Angew. Chem., Int. Ed. 2010, 49, 1736. (e) Song, Y.-F.; Tsunashima, R. Chem. Soc. Rev. 2012, 41, 7384. (f) Streb, C. Dalton Trans. 2012, 41, 1651. (g) Lv, H.; Geletii, Y. V.; Zhao, C.; Vickers, J. W.; Zhu, G.; Luo, Z.; Song, J.; Lian, T.; Musaev, D. G.; Hill, C. L. Chem. Soc. Rev. 2012, 41, 7572. (h) Sumliner, J. M.; Lv, H.; Fielden, J.; Geletii, Y. V.; Hill, C. L. Eur. J. Inorg. Chem. 2014, 635. (439) Howells, A. R.; Sankarraj, A.; Shannon, C. J. Am. Chem. Soc. 2004, 126, 12258. (440) Geletii, Y. V.; Botar, B.; Kögerler, P.; Hillesheim, D. A.; Musaev, D. G.; Hill, C. L. Angew. Chem., Int. Ed. 2008, 47, 3896. (441) Sartorel, A.; Carraro, M.; Scorrano, G.; De Zorzi, R.; Geremia, S.; McDaniel, N. D.; Bernhard, S.; Bonchio, M. J. Am. Chem. Soc. 2008, 130, 5006. (442) Geletii, Y. V.; Huang, Z.; Hou, Y.; Musaev, D. G.; Lian, T.; Hill, C. L. J. Am. Chem. Soc. 2009, 131, 7522. (443) Puntoriero, F.; La Ganga, G.; Sartorel, A.; Carraro, M.; Scorrano, G.; Bonchio, M.; Campagna, S. Chem. Commun. 2010, 46, 4725. (444) Orlandi, M.; Argazzi, R.; Sartorel, A.; Carraro, M.; Scorrano, G.; Bonchio, M.; Scandola, F. Chem. Commun. 2010, 46, 3152. (445) Natali, M.; Orlandi, M.; Berardi, S.; Campagna, S.; Bonchio, M.; Sartorel, A.; Scandola, F. Inorg. Chem. 2012, 51, 7324. (446) Sartorel, A.; Miró, P.; Salvadori, E.; Romain, S.; Carraro, M.; Scorrano, G.; Di Valentin, M.; Llobet, A.; Bo, C.; Bonchio, M. J. Am. Chem. Soc. 2009, 131, 16051. (447) Geletii, Y. V.; Besson, C.; Hou, Y.; Yin, Q.; Musaev, D. G.; Quiñonero, D.; Cao, R.; Hardcastle, K. I.; Proust, A.; Kögerler, P.; Hill, C. L. J. Am. Chem. Soc. 2009, 131, 17360. (448) Liu, Y.; Guo, S.-X.; Bond, A. M.; Zhang, J.; Geletii, Y. V.; Hill, C. L. Inorg. Chem. 2013, 52, 11986. (449) Quiñonero, D.; Kaledin, A. L.; Kuznetsov, A. E.; Geletii, Y. V.; Besson, C.; Hill, C. L.; Musaev, D. G. J. Phys. Chem. A 2010, 114, 535. (450) Lee, C.-Y.; Guo, S.-X.; Murphy, A. F.; McCormac, T.; Zhang, J.; Bond, A. M.; Zhu, G.; Hill, C. L.; Geletii, Y. V. Inorg. Chem. 2012, 51, 11521. (451) Sartorel, A.; Truccolo, M.; Berardi, S.; Gardan, M.; Carraro, M.; Toma, F. M.; Scorrano, G.; Prato, M.; Bonchio, M. Chem. Commun. 2011, 47, 1716. (452) Toma, F. M.; Sartorel, A.; Iurlo, M.; Carraro, M.; Parisse, P.; Maccato, C.; Rapino, S.; Rodriguez Gonzalez, B.; Amenitsch, H.; Da Ros, T.; Casalis, L.; Goldoni, A.; Marcaccio, M.; Scorrano, G.; Scoles, G.; Paolucci, F.; Prato, M.; Bonchio, M. Nat. Chem. 2010, 2, 826. (453) Quintana, M.; Montellano López, A.; Rapino, S.; Toma, F. M.; Iurlo, M.; Carraro, M.; Sartorel, A.; Maccato, C.; Ke, X.; Bittencourt, C.; Da Ros, T.; Van Tendeloo, G.; Marcaccio, M.; Paolucci, F.; Prato, M.; Bonchio, M. ACS Nano 2013, 7, 811. (454) Guo, S.-X.; Liu, Y.; Lee, C.-Y.; Bond, A. M.; Zhang, J.; Geletii, Y. V.; Hill, C. L. Energy Environ. Sci. 2013, 6, 2654. (455) Piccinin, S.; Sartorel, A.; Aquilanti, G.; Goldoni, A.; Bonchio, M.; Fabris, S. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 4917. (456) Kuznetsov, A. E.; Geletii, Y. V.; Hill, C. L.; Morokuma, K.; Musaev, D. G. J. Am. Chem. Soc. 2009, 131, 6844. (457) Besson, C.; Huang, Z.; Geletii, Y. V.; Lense, S.; Hardcastle, K. I.; Musaev, D. G.; Lian, T.; Proust, A.; Hill, C. L. Chem. Commun. 2010, 46, 2784. (458) Murakami, M.; Hong, D.; Suenobu, T.; Yamaguchi, S.; Ogura, T.; Fukuzumi, S. J. Am. Chem. Soc. 2011, 133, 11605. (459) Weakley, T. J. R.; Evans, H. T., Jr.; Showell, J. S.; Tourné, G. F.; Tourné, C. M. J. Chem. Soc., Chem. Commun. 1973, 139. (460) Huang, Z.; Luo, Z.; Geletii, Y. V.; Vickers, J. W.; Yin, Q.; Wu, D.; Hou, Y.; Ding, Y.; Song, J.; Musaev, D. G.; Hill, C. L.; Lian, T. J. Am. Chem. Soc. 2011, 133, 2068.

(461) Lv, H.; Rudd, J. A.; Zhuk, P. F.; Lee, J. Y.; Constable, E. C.; Housecroft, C. E.; Hill, C. L.; Musaev, D. G.; Geletii, Y. V. RSC Adv. 2013, 3, 20647. (462) Stracke, J. J.; Finke, R. G. ACS Catal. 2013, 3, 1209. (463) Vickers, J. W.; Lv, H.; Sumliner, J. M.; Zhu, G.; Luo, Z.; Musaev, D. G.; Geletii, Y. V.; Hill, C. L. J. Am. Chem. Soc. 2013, 135, 14110. (464) Natali, M.; Berardi, S.; Sartorel, A.; Bonchio, M.; Campagns, S.; Scandola, F. Chem. Commun. 2012, 48, 8808. (465) Finke, R. G.; Droege, M.; Hutchinson, J. R.; Gansow, O. J. Am. Chem. Soc. 1981, 103, 1587. (466) For an example where distinct differences have been observed in the electrochemical behavior between freely diffusing and electroflocculated films of iridium nanoparticles, see: Gambardella, A. A.; Bjorge, N. S.; Alspaugh, V. K.; Murray, R. W. J. Phys. Chem. C 2011, 115, 21659. (467) Lieb, D.; Zahl, A.; Wilson, E. F.; Streb, C.; Nye, L. C.; Meyer, K.; Ivanović-Burmazović, I. Inorg. Chem. 2011, 50, 9053. (468) Brasch, N. E.; Buckingham, D. A.; Clark, C. R.; Rogers, A. J. Inorg. Chem. 1998, 37, 4865. (469) Ohlin, C. A.; Harley, S. J.; McAlpin, J. G.; Hocking, R. K.; Mercado, B. Q.; Johnson, R. L.; Villa, E. M.; Fidler, M. K.; Olmstead, M. M.; Spiccia, L.; Britt, R. D.; Casey, W. H. Chem.Eur. J. 2011, 17, 4408. (470) Wu, J.; Liao, L.; Yan, W.; Xue, Y.; Sun, Y.; Yan, X.; Chen, Y.; Xie, Y. ChemSusChem 2012, 5, 1207. (471) Zhu, G.; Geletii, Y. V.; Kögerler, P.; Schilder, H.; Song, J.; Lense, S.; Zhao, C.; Hardcastle, K. I.; Musaev, D. G.; Hill, C. L. Dalton Trans. 2012, 41, 2084. (472) Tanaka, S.; Annaka, M.; Sakai, K. Chem. Commun. 2012, 48, 1653. (473) Song, F.; Ding, Y.; Ma, B.; Wang, C.; Wang, Q.; Du, X.; Fu, S.; Song, J. Energy Environ. Sci. 2013, 6, 1170. (474) Goberna-Ferrón, S.; Vigara, L.; Soriano-López, J.; GalánMascarós, J. R. Inorg. Chem. 2012, 51, 11707. (475) Soriano-López, J.; Goberna-Ferrón, S.; Vigara, L.; Carbó, J. J.; Poblet, J. M.; Galán-Mascarós, J. R. Inorg. Chem. 2013, 52, 4753. (476) Han, X.-B.; Zhang, Z.-M.; Zhang, T.; Li, Y.-G.; Lin, W.; You, W.; Su, Z.-M.; Wang, E.-B. J. Am. Chem. Soc. 2014, 136, 5359. (477) Evangelisti, F.; Car, P.-E.; Blacque, O.; Patzke, G. R. Catal. Sci. Technol. 2013, 3, 3117. (478) Bassil, B. S.; Ibrahim, M.; Al-Oweini, R.; Asano, M.; Wang, Z.; van Tol, J.; Dalal, N. S.; Choi, K.-Y.; Ngo Biboum, R.; Keita, B.; Nadjo, L.; Kortz, U. Angew. Chem., Int. Ed. 2011, 50, 5961. (479) Al-Oweini, R.; Bassil, B. S.; Palden, T.; Keita, B.; Lan, Y.; Powell, A. K.; Kortz, U. Polyhedron 2013, 52, 461. (480) Gao, J.; Cao, S.; Tay, Q.; Liu, Y.; Yu, L.; Ye, K.; Mun, P. C. S.; Li, Y.; Rakesh, G.; Loo, S. C. J.; Chen, Z.; Zhao, Y.; Xue, C.; Zhang, Q. Sci. Rep. 2013, 3, 1853. (481) Cao, R.; Ma, H.; Geletii, Y. V.; Hardcastle, K. I.; Hill, C. L. Inorg. Chem. 2009, 48, 5596. (482) Zhu, G.; Glass, E. N.; Zhao, C.; Lv, H.; Vickers, J. W.; Geletii, Y. V.; Musaev, D. G.; Song, J.; Hill, C. L. Dalton Trans. 2012, 41, 13043. (483) Heussner, K.; Peuntinger, K.; Rockstroh, N.; Nye, L. C.; Ivanovic-Burmazovic, I.; Rau, S.; Streb, C. Chem. Commun. 2011, 47, 6852. (484) Chen, G.; Chen, L.; Ng, S.-M.; Lau, T.-C. ChemSusChem 2014, 7, 127. (485) Mills, A. Chem. Soc. Rev. 1989, 18, 285. (486) Kiwi, J.; Kalyanasundaram, K.; Grätzel, M. Solar Energy Materials: Structure and Bonding, 49th ed.; Springer Verlag: Berlin, 1982; pp 37−125. (487) For early examples of using metal oxides as water oxidation catalysts, see, for example: (a) Kiwi, J.; Grätzel, M. Angew. Chem., Int. Ed. Engl. 1979, 18, 624. (b) Kiwi, J.; Borgarello, E.; Pelizzetti, E.; Visca, M.; Grätzel, M. Angew. Chem., Int. Ed. Engl. 1980, 19, 646. (c) Harriman, A.; Richoux, M.-C.; Christensen, P. A.; Mosseri, S.; Neta, P. J. Chem. Soc., Faraday Trans. 1 1987, 83, 3001. (d) Nagoshi, 11998

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(516) Ullman, A. M.; Nocera, D. G. J. Am. Chem. Soc. 2013, 135, 15053. (517) Bediako, D. K.; Surendranath, Y.; Nocera, D. G. J. Am. Chem. Soc. 2013, 135, 3662. (518) Dincă, M.; Surendranath, Y.; Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 10337. (519) Young, E. R.; Nocera, D. G.; Bulović, V. Energy Environ. Sci. 2010, 3, 1726. (520) Lai, Y.-H.; Lin, C.-Y.; Lv, Y.; King, T. C.; Steiner, A.; Muresan, N. M.; Gan, L.; Wright, D. S.; Reisner, E. Chem. Commun. 2013, 49, 4331. (521) Zhong, D. K.; Sun, J.; Inumaru, H.; Gamelin, D. R. J. Am. Chem. Soc. 2009, 131, 6086. (522) Zhong, D. K.; Gamelin, D. R. J. Am. Chem. Soc. 2010, 132, 4202. (523) Zhong, D. K.; Cornuz, M.; Sivula, K.; Grätzel, M.; Gamelin, D. R. Energy Environ. Sci. 2011, 4, 1759. (524) Steinmiller, E. M. P.; Choi, K.-S. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 20633. (525) Pijpers, J. J. H.; Winkler, M. T.; Surendranath, Y.; Buonassisi, T.; Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 10056. (526) Young, E. R.; Costi, R.; Paydavosi, S.; Nocera, D. G.; Bulović, V. Energy Environ. Sci. 2011, 4, 2058. (527) Reece, S. Y.; Hamel, J. A.; Sung, K.; Jarvi, T. D.; Esswein, A. J.; Pijpers, J. J. H.; Nocera, D. G. Science 2011, 334, 645. (528) Surendranath, Y.; Lutterman, D. A.; Liu, Y.; Nocera, D. G. J. Am. Chem. Soc. 2012, 134, 6326. (529) Chakrabarty, R.; Bora, S. J.; Das, B. K. Inorg. Chem. 2007, 46, 9450. (530) Dimitrou, K.; Folting, K.; Streib, W. E.; Christou, G. J. Am. Chem. Soc. 1993, 115, 6432. (531) Dimitrou, K.; Brown, A. D.; Concolino, T. E.; Rheingold, A. L.; Christou, G. Chem. Commun. 2001, 1284. (532) McCool, N. S.; Robinson, D. M.; Sheats, J. E.; Dismukes, G. C. J. Am. Chem. Soc. 2011, 133, 11446. (533) La Ganga, G.; Puntoriero, F.; Campagna, S.; Bazzan, I.; Berardi, S.; Bonchio, M.; Sartorel, A.; Natalic, M.; Scandola, F. Faraday Discuss. 2012, 155, 177. (534) Berardi, S.; La Ganga, G.; Natali, M.; Bazzan, I.; Puntoriero, F.; Sartorel, A.; Scandola, F.; Campagna, S.; Bonchio, M. J. Am. Chem. Soc. 2012, 134, 11104. (535) McAlpin, J. G.; Stich, T. A.; Ohlin, C. A.; Surendranath, Y.; Nocera, D. G.; Casey, W. H.; Britt, R. D. J. Am. Chem. Soc. 2011, 133, 15444. (536) Symes, M. D.; Surendranath, Y.; Lutterman, D. A.; Nocera, D. G. J. Am. Chem. Soc. 2011, 133, 5174. (537) Singh, R. N.; Mishra, D.; Anindita; Sinha, A. S. K.; Singh, A. Electrochem. Commun. 2007, 9, 1369. (538) Ahn, H. S.; Yano, J.; Tilley, T. D. Energy Environ. Sci. 2013, 6, 3080. (539) Jiao, F.; Frei, H. Angew. Chem., Int. Ed. 2009, 48, 1841. (540) Yusuf, S.; Jiao, F. ACS Catal. 2012, 2, 2753. (541) Rosen, J.; Hutchings, G. S.; Jiao, F. J. Am. Chem. Soc. 2013, 135, 4516. (542) Gardner, G. P.; Go, Y. B.; Robinson, D. M.; Smith, P. F.; Hadermann, J.; Abakumov, A.; Greenblatt, M.; Dismukes, G. C. Angew. Chem., Int. Ed. 2012, 51, 1616. (543) Zidki, T.; Zhang, L.; Shafirovich, V.; Lymar, S. V. J. Am. Chem. Soc. 2012, 134, 14275. (544) Armandi, M.; Hernandez, S.; Vankova, S.; Zanarini, S.; Bonelli, B.; Garrone, E. ACS Catal. 2013, 3, 1272. (545) Shevchenko, S.; Anderlund, M. F.; Thapper, A.; Styring, S. Energy Environ. Sci. 2011, 4, 1284. (546) Wee, T.-L.; Sherman, B. D.; Gust, D.; Moore, A. L.; Moore, T. A.; Liu, Y.; Scaiano, J. C. J. Am. Chem. Soc. 2011, 133, 16742. (547) Grzelczak, M.; Zhang, J.; Pfrommer, J.; Hartmann, J.; Driess, M.; Antonietti, M.; Wang, X. ACS Catal. 2013, 3, 383. (548) The organic ligand framework in cobaltocene (CoCp2) has previously been shown to undergo decomposition, thus resulting in

K.; Yamashita, S.; Yagi, M.; Kaneko, M. J. Mol. Catal. A: Chem. 1999, 144, 71. (488) For early examples where iridum oxides were used as H2O oxidation catalysts, see, for example: (a) Kiwi, J.; Grätzel, M. Angew. Chem., Int. Ed. Engl. 1978, 17, 860. (b) Beni, G.; Schiavone, L. M.; Shay, J. L.; Dautremont-Smith, W. C.; Schneider, B. S. Nature 1979, 282, 281. (489) For early examples where ruthenium oxides were used as H2O oxidation catalysts, see, for example: (a) Kalyanasundaram, K.; Grätzel, M. Angew. Chem., Int. Ed. Engl. 1979, 18, 701. (b) Kalyanasundaram, K.; Mikik, O.; Pramauro, E.; Grätzel, M. Helv. Chim. Acta 1979, 62, 2432. (c) Duonghong, D.; Borgarello, E.; Grätzel, M. J. Am. Chem. Soc. 1981, 103, 4685. (490) Mills, A.; Duckmanton, P. A.; Reglinski, J. Chem. Commun. 2010, 46, 2397. (491) Zhang, Y.; Ren, T. Chem. Commun. 2012, 48, 11005. (492) Zhang, Y.; Judkins, E. C.; McMillin, D. R.; Mehta, D.; Ren, T. ACS Catal. 2013, 3, 2474. (493) Sun, K.; Pang, X.; Shen, S.; Qian, X.; Cheung, J. S.; Wang, D. Nano Lett. 2013, 13, 2064. (494) Youngblood, W. J.; Lee, S.-H. A.; Kobayashi, Y.; HernandezPagan, E. A.; Hoertz, P. G.; Moore, T. A.; Moore, A. L.; Gust, D.; Mallouk, T. E. J. Am. Chem. Soc. 2009, 131, 926. (495) Lee, S.-H. A.; Zhao, Y.; Hernandez-Pagan, E. A.; Blasdel, L.; Youngblood, W. J.; Mallouk, T. E. Faraday Discuss. 2012, 155, 165. (496) Hoertz, P. G.; Kim, Y.-I.; Youngblood, W. J.; Mallouk, T. E. J. Phys. Chem. B 2007, 111, 6845. (497) Yagi, M.; Tomita, E.; Sakita, S.; Kuwabara, T.; Nagai, K. J. Phys. Chem. B 2005, 109, 21489. (498) Zhao, Y.; Swierk, J. R.; Megiatto, J. D., Jr.; Sherman, B.; Youngblood, W. J.; Qin, D.; Lentz, D. M.; Moore, A. L.; Moore, T. A.; Gust, D.; Mallouk, T. E. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 15612. (499) Frame, F. A.; Townsend, T. K.; Chamousis, R. L.; Sabio, E. M.; Dittrich, T.; Browning, N. D.; Osterloh, F. E. J. Am. Chem. Soc. 2011, 133, 7264. (500) El Wakkad, S. E. S.; Hickling, A. Trans. Faraday Soc. 1950, 46, 820. (501) Benson, P.; Briggs, G. W. D.; Wynne-Jones, W. F. K. Electrochim. Acta 1964, 9, 275. (502) Benson, P.; Briggs, G. W. D.; Wynne-Jones, W. F. K. Electrochim. Acta 1964, 9, 281. (503) Kanan, M. W.; Nocera, D. G. Science 2008, 321, 1072. (504) Surendranath, Y.; Dincǎ, M.; Nocera, D. G. J. Am. Chem. Soc. 2009, 131, 2615. (505) Esswein, A. J.; Surendranath, Y.; Reece, S. Y.; Nocera, D. G. Energy Environ. Sci. 2011, 4, 499. (506) Lutterman, D. A.; Surendranath, Y.; Nocera, D. G. J. Am. Chem. Soc. 2009, 131, 3838. (507) Risch, M.; Khare, V.; Zaharieva, I.; Gerencser, L.; Chernev, P.; Dau, H. J. Am. Chem. Soc. 2009, 131, 6936. (508) Kanan, M. W.; Yano, J.; Surendranath, Y.; Dincǎ, M.; Yachandra, V. K.; Nocera, D. G. J. Am. Chem. Soc. 2010, 132, 13692. (509) Du, P.; Kokhan, O.; Chapman, K. W.; Chupas, P. J.; Tiede, D. M. J. Am. Chem. Soc. 2012, 134, 11096. (510) Farrow, C. L.; Bediako, D. K.; Surendranath, Y.; Nocera, D. G.; Billinge, S. J. L. J. Am. Chem. Soc. 2013, 135, 6403. (511) Hu, X. L.; Piccinin, S.; Laio, A.; Fabris, S. ACS Nano 2012, 6, 10497. (512) McAlpin, J. G.; Surendranath, Y.; Dincǎ, M.; Stitch, T. A.; Stoian, S. A.; Casey, W. H.; Nocera, D. G.; Britt, R. D. J. Am. Chem. Soc. 2010, 132, 6882. (513) Gerken, J. B.; McAlpin, J. G.; Chen, J. Y. C.; Rigsby, M. L.; Casey, W. H.; Britt, R. D.; Stahl, S. S. J. Am. Chem. Soc. 2011, 133, 14431. (514) Surendranath, Y.; Kanan, M. W.; Nocera, D. G. J. Am. Chem. Soc. 2010, 132, 16501. (515) Bediako, D. K.; Costentin, C.; Jones, E. C.; Nocera, D. G.; Savéant, J.-M. J. Am. Chem. Soc. 2013, 135, 10492. 11999

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(577) Boppana, V. B. R.; Jiao, F. Chem. Commun. 2011, 47, 8973. (578) If the Mn2O3 impurity was not completely removed, the amount of this impurity was below the limit of detection. (579) Zaharieva, I.; Chernev, P.; Risch, M.; Klingan, K.; Kohlhoff, M.; Fischer, A.; Dau, H. Energy Environ. Sci. 2012, 5, 7081. (580) Takashima, T.; Hashimoto, K.; Nakamura, R. J. Am. Chem. Soc. 2012, 134, 1519. (581) Morita, M.; Iwakura, C.; Tamura, H. Electrochim. Acta 1979, 24, 357. (582) Trasatti, S. Electrochim. Acta 1984, 29, 1503. (583) Pyrophosphate is a redox-inactive molecule that can be used for the uptake of MnIII atoms from the surface of heterogeneous manganese oxides through chelation. (584) Takashima, T.; Hashimoto, K.; Nakamura, R. J. Am. Chem. Soc. 2012, 134, 18153. (585) Gorlin, Y.; Jaramillo, T. F. J. Am. Chem. Soc. 2010, 132, 13612. (586) Gorlin, Y.; Lassalle-Kaiser, B.; Benck, J. D.; Gul, S.; Webb, S. M.; Yachandra, V. K.; Yano, J.; Jaramillo, T. F. J. Am. Chem. Soc. 2013, 135, 8525. (587) Ghanotakis, D. F.; Babcock, G. T.; Yocum, C. F. FEBS Lett. 1984, 167, 127. (588) Miyao, M.; Murata, N. FEBS Lett. 1984, 168, 118. (589) Yocum, C. F. Coord. Chem. Rev. 2008, 252, 296. (590) Yachandra, V. K.; Yano, J. J. Photochem. Photobiol., B 2011, 104, 51. (591) Najafpour, M. M.; Ehrenberg, T.; Wiechen, M.; Kurz, P. Angew. Chem., Int. Ed. 2010, 49, 2233. (592) Najafpour, M. M. Dalton Trans. 2011, 40, 3793. (593) Najafpour, M. M. Dalton Trans. 2011, 40, 3805. (594) Najafpour, M. M.; Nayeri, S.; Pashaei, B. Dalton Trans. 2011, 40, 9374. (595) Najafpour, M. M.; Pashaei, B.; Nayeri, S. Dalton Trans. 2012, 41, 4799. (596) Birkner, N.; Nayeri, S.; Pashaei, B.; Najafpour, M. M.; Casey, W. H.; Navrotsky, A. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 8801. (597) Esswein, A. J.; McMurdo, M. J.; Ross, P. N.; Bell, A. T.; Tilley, T. D. J. Phys. Chem. C 2009, 113, 15068. (598) Zaharieva, I.; Najafpour, M. M.; Wiechen, M.; Haumann, M.; Kurz, P.; Dau, H. Energy Environ. Sci. 2011, 4, 2400. (599) Shevela, D.; Koroidov, S.; Najafpour, M. M.; Messinger, J.; Kurz, P. Chem.Eur. J. 2011, 17, 5415. (600) Mishra, A.; Wernsdorfer, W.; Abboud, K. A.; Christou, G. Chem. Commun. 2005, 54. (601) Hewitt, I. J.; Tang, J.-K.; Madhu, N. T.; Clérac, R.; Buth, G.; Anson, C. E.; Powell, A. K. Chem. Commun. 2006, 2650. (602) Mishra, A.; Yano, J.; Pushkar, Y.; Yachandra, V. K.; Abboud, K. A.; Christou, G. Chem. Commun. 2007, 1538. (603) Nayak, S.; Nayek, H. P.; Dehnen, S.; Powell, A. K.; Reedijk, J. Dalton Trans. 2011, 40, 2699. (604) Kotzabasaki, V.; Siczek, M.; Lis, T.; Milios, C. J. Inorg. Chem. Commun. 2011, 14, 213. (605) Mukherjee, S.; Stull, J. A.; Yano, J.; Stamatatos, T. C.; Pringouri, K.; Stich, T. A.; Abboud, K. A.; Britt, R. D.; Yachandra, V. K.; Christou, G. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 2257. (606) Kanady, J. S.; Tsui, E. Y.; Day, M. W.; Agapie, T. Science 2011, 333, 733. (607) Tsui, E. Y.; Tran, R.; Yano, J.; Agapie, T. Nat. Chem. 2013, 5, 293. (608) Tsui, E. Y.; Agapie, T. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 10084. (609) Kanady, J. S.; Mendoza-Cortes, J. L.; Tsui, E. Y.; Nielsen, R. J.; Goddard, W. A., III; Agapie, T. J. Am. Chem. Soc. 2013, 135, 1073. (610) There exists an example where the redox-inactive metal has been exchanged in layered manganese oxides, where it was revealed that the presence of Ca and Sr resulted in optimal catalytically active H2O oxidation materials, similar to the OEC, see: Wiechen, M.; Zaharieva, I.; Dau, H.; Kurz, P. Chem. Sci. 2012, 3, 2330.

the formation of catalytically active cobalt nanoparticles, see: Hong, D.; Jung, J.; Park, J.; Yamada, Y.; Suenobu, T.; Lee, Y.-M.; Nam, W.; Fukuzumi, S. Energy Environ. Sci. 2012, 5, 7606. (549) Ruettinger, W.; Yagi, M.; Wolf, K.; Bernasek, S.; Dismukes, G. C. J. Am. Chem. Soc. 2000, 122, 10353. (550) Wu, J.-Z.; De Angelis, F.; Carrell, T. G.; Yap, G. P. A.; Sheats, J.; Car, R.; Dismukes, G. C. Inorg. Chem. 2006, 45, 189. (551) Cady, C. W.; Crabtree, R. H.; Brudvig, G. W. Coord. Chem. Rev. 2008, 252, 444. (552) Kuznetsov, A. E.; Geletii, Y. V.; Hill, C. L.; Musaev, D. G. J. Phys. Chem. A 2010, 114, 11417. (553) Brimblecombe, R.; Swiegers, G. F.; Dismukes, G. C.; Spiccia, L. Angew. Chem., Int. Ed. 2008, 47, 7335. (554) Brimblecombe, R.; Koo, A.; Dismukes, G. C.; Swiegers, G. F.; Spiccia, L. J. Am. Chem. Soc. 2010, 132, 2892. (555) Brimblecombe, R.; Koo, A.; Dismukes, G. C.; Swiegers, G. F.; Spiccia, L. ChemSusChem 2010, 3, 1146. (556) Dismukes, G. C.; Brimblecombe, R.; Felton, G. A. N.; Pryadun, R. S.; Sheats, J. E.; Spiccia, L.; Swiegers, G. F. Acc. Chem. Res. 2009, 42, 1935. (557) Brimblecombe, R.; Kolling, D. R. J.; Bond, A. M.; Dismukes, G. C.; Swiegers, G. F.; Spiccia, L. Inorg. Chem. 2009, 48, 7269. (558) Brimblecombe, R.; Bond, A. M.; Dismukes, G. C.; Swiegers, G. F.; Spiccia, L. Phys. Chem. Chem. Phys. 2009, 11, 6441. (559) Hocking, R. K.; Brimblecombe, R.; Chang, L.-Y.; Singh, A.; Cheah, M. H.; Glover, C.; Casey, W. H.; Spiccia, L. Nat. Chem. 2011, 3, 461. (560) For examples of decomposition of molecular manganese complexes in solution or by thermolysis that produces active nanostructured manganese oxide materials, see: (a) Hill, J. P.; Palza, H.; Alam, S.; Ariga, K.; Schumacher, A. L.; D’Souza, F.; Anson, C. E.; Powell, A. K. Inorg. Chem. 2008, 47, 8306. (b) Hill, J. P.; Alam, S.; Ariga, K.; Anson, C. E.; Powell, A. K. Chem. Commun. 2008, 383. (c) Nayak, S.; Malik, S.; Indris, S.; Reedijk, J.; Powell, A. K. Chem. Eur. J. 2010, 16, 1158. (561) For eaxmples of decomposition of molecular manganese complexes that generates catalytically active nanostructured manganese oxide materials for O2 evolution, see: (a) Tanase, T.; Lippard, S. J. Inorg. Chem. 1995, 34, 4682. (b) Najafpour, M. M.; Moghaddam, A. N. Dalton Trans. 2012, 41, 10292. (c) Singh, A.; Hocking, R. K.; Chang, S. L.-Y.; George, B. M.; Fehr, M.; Lips, K.; Schnegg, A.; Spiccia, L. Chem. Mater. 2013, 25, 1098. (562) Najafpour, M. M.; Sedigh, D. J.; Pashaei, B.; Nayeri, S. New J. Chem. 2013, 37, 2448. (563) Chen, Z.; Jiao, Z.; Pan, D.; Li, Z.; Wu, M.; Shek, C.-H.; Wu, C. M. L.; Lai, J. K. L. Chem. Rev. 2012, 112, 3833. (564) Najafpour, M. M.; Rahimi, F.; Amini, M.; Nayeri, S.; Bagherzadeh, M. Dalton Trans. 2012, 41, 11026. (565) Najafpour, M. M. Chem. Commun. 2011, 47, 11724. (566) Najafpour, M. M.; Sedigh, D. J. Dalton Trans. 2013, 42, 12173. (567) Zhou, F.; Izgorodin, A.; Hocking, R. K.; Armel, V.; Spiccia, L.; MacFarlane, D. R. ChemSusChem 2013, 6, 643. (568) Najafpour, M. M.; Rahimi, F.; Aro, E.-M.; Lee, C.-H.; Allakhverdiev, S. I. J. R. Soc. Interface 2012, 9, 2383. (569) Jin, K.; Park, J.; Lee, J.; Yang, K. D.; Pradhan, G. K.; Sim, U.; Jeong, D.; Jang, H. L.; Park, S.; Kim, D.; Sung, N.-E.; Kim, S. H.; Han, S.; Nam, K. T. J. Am. Chem. Soc. 2014, 136, 7435. (570) Najafpour, M. M.; Moghaddam, A. N. New J. Chem. 2012, 36, 2514. (571) Najafpour, M. M.; Moghaddam, A. N.; Sakha, Y. Dalton Trans. 2013, 42, 11012. (572) Najafpour, M. M.; Pashaei, B. Dalton Trans. 2012, 41, 10156. (573) Jiao, F.; Frei, H. Chem. Commun. 2010, 46, 2920. (574) Post, J. E. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 3447. (575) Robinson, D. M.; Go, Y. B.; Greenblatt, M.; Dismukes, G. C. J. Am. Chem. Soc. 2010, 132, 11467. (576) Robinson, D. M.; Go, Y. B.; Mui, M.; Gardner, G.; Zhang, Z.; Mastrogiovanni, D.; Garfunkel, E.; Li, J.; Greenblatt, M.; Dismukes, G. C. J. Am. Chem. Soc. 2013, 135, 3494. 12000

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001

Chemical Reviews

Review

(611) Indra, A.; Menezes, P. W.; Zaharieva, I.; Baktash, E.; Pfrommer, J.; Schwarze, M.; Dau, H.; Driess, M. Angew. Chem., Int. Ed. 2013, 52, 13206. (612) Wiechen, M.; Spiccia, L. ChemCatChem 2014, 6, 439. (613) Smith, R. D. L.; Prévot, M. S.; Fagan, R. D.; Zhang, Z.; Sedach, P. A.; Siu, M. K. J.; Trudel, S.; Berlinguette, C. P. Science 2013, 340, 60. (614) Smith, R. D. L.; Prévot, M. S.; Fagan, R. D.; Trudel, S.; Berlinguette, C. P. J. Am. Chem. Soc. 2013, 135, 11580. (615) Smith, R. D. L.; Sporinova, B.; Fagan, R. D.; Trudel, S.; Berlinguette, C. P. Chem. Mater. 2014, 26, 1654. (616) Conrad, F.; Bauer, M.; Sheptyakov, D.; Weyeneth, S.; Jaeger, D.; Hametner, K.; Car, P.-E.; Patscheider, J.; Günther, D.; Patzke, G. R. RSC Adv. 2012, 2, 3076. (617) Gao, M.-R.; Xu, Y.-F.; Jiang, J.; Zheng, Y.-R.; Yu, S.-H. J. Am. Chem. Soc. 2012, 134, 2930. (618) Pintado, S.; Goberna-Ferrón, S.; Escudero-Adán, E. C.; GalánMascarós, J. R. J. Am. Chem. Soc. 2013, 135, 13270. (619) Bensaid, S.; Centi, G.; Garrone, E.; Perathoner, S.; Saracco, G. ChemSusChem 2012, 5, 500.

12001

dx.doi.org/10.1021/cr400572f | Chem. Rev. 2014, 114, 11863−12001