Light-induced charge separation in molecular assemblies has been widely investigated in the context of artificial photosynthesis. Important progress has been made in the fundamental understanding of electron and energy transfer and in stabilizing charge separation by multi-step electron transfer. In the Swedish Consortium for Artificial Photosynthesis, we build on principles from the natural enzyme photosystem II and Fe-hydrogenases. An important theme in this biomimetic effort is that of coupled electron-transfer reactions, which have so far received only little attention. (i) Each absorbed photon leads to charge separation on a single-electron level only, while catalytic water splitting and hydrogen production are multi-electron processes; thus there is the need for controlling accumulative electron transfer on molecular components. (ii) Water splitting and proton reduction at the potential catalysts necessarily require the management of proton release and/or uptake. Far from being just a stoichiometric requirement, this controls the electron transfer processes by proton-coupled electron transfer (PCET). (iii) Redox-active links between the photosensitizers and the catalysts are required to rectify the accumulative electron-transfer reactions, and will often be the starting points of PCET.
Photosystem II (PSII) is a key component of the most successful solar energy-converting process on our planet: oxygenic photosynthesis. It uses light energy to accomplish the reduction of plastoquinone and the oxidation of water (Barber 2002; Diner & Rappaport 2002; Goussias et al. 2002):(1.1)Electronic excitation of the primary chlorophyll donor P680 by energy transfer from antenna pigments initiates a chain of electron transfer reactions, which separates oxidative and reductive equivalents over PSII. The four-electron water-oxidation reaction is catalysed by a Mn4Ca cluster (Ferreira et al. 2004; Haumann et al. 2005; Loll et al. 2005; Yano et al. 2006) and gives the organism an abundant source of electrons for the reduction of carbon dioxide to carbohydrates. The principles and many details of the light-harvesting and charge-separation processes are known. In contrast, the structure of the cluster, and the mechanism by which it catalyses water oxidation, are controversial and not yet fully understood.
The reactions and mechanistic principles of PSII provide important inspiration and information for biomimetic chemistry, which attempts to mimic the fundamental photosynthetic processes in synthetic molecular systems (Wasielewski 1992; Ruttinger & Dismukes 1997; Gust et al. 2001; Sun et al. 2001; Hammarström 2003; Mukhopadhyay et al. 2004; Alstrum-Acevedo et al. 2005). The goals are to improve and generalize our understanding and to ultimately pave the way for a molecular solar-energy-conversion technology by artificial photosynthesis. Light-induced charge separation on a single-electron level has been demonstrated in numerous synthetic systems, including multi-unit ‘triads’, ‘tetrads’, etc. (Wasielewski 1992; Gust et al. 2001; Baranoff et al. 2004; Alstrum-Acevedo et al. 2005). Each absorbed photon leads to the transfer of only one electron, however, while full reduction of the ultimate PSII acceptor, QB, requires two electrons before it exits to the plastoquinone pool. Water oxidation at the Mn4Ca cluster is even more demanding as four electrons have to be removed to complete the catalytic cycle. Thus, several sequences of light-induced charge separation on the single-electron level have to be coupled to achieve accumulation of charge, or rather redox equivalents, on a single molecular unit or metal cluster. This we denote accumulative electron transfer, to distinguish it from both multi-step electron transfer that is a sequence of electron transfer steps on the one-photon/one-electron level and multiple electron transfer that is the transfer of two or more electrons in a single reaction step.
Accumulation of oxidizing equivalents on the Mn4Ca cluster is coupled to the release of protons (Junge et al. 2002; Dau & Haumann 2006). Therefore, the increase in oxidation state does not lead to an increase in charge, except probably at the transition from state S1 to S2 (figure 1). Thanks to this charge-compensation mechanism, the redox potentials of the different oxidation steps are rather similar. This allows the TyzZ radical, which is the intermediate electron donor between P680 and the Mn4Ca cluster, to oxidize the cluster in every step of the cycle without wasting an unnecessary amount of free energy in the early steps.
The coupling of electron and proton transfer on the donor side of PSII begins with the oxidation of the tyrosine TyrZ, which is the electron donor that regenerates the P680 chlorophylls after oxidation. The phenolic group of TyrZ becomes very acidic upon oxidation and looses its proton to a nearby base, most probably the His190 that is in hydrogen-bonding contact (Diner & Rappaport 2002; Ferreira et al. 2004; Loll et al. 2005). The proton-coupled electron transfer (PCET) from TyrZ has been much studied and debated regarding the nature of the electron–proton coupling and its influence on the reaction rate and mechanism under different conditions (Tommos & Babcock 2000; Diner & Rappaport 2002; Diner et al. 2004; Renger 2004). The following reduction of the TyrZ radical by the Mn4Ca cluster is also a PCET reaction, and thus the PCET of TyrZ may have direct mechanistic implications for the water oxidation.
From this section, it is clear that the concept of coupled electron-transfer reactions is vital for the function of PSII. While there have been numerous synthetic molecular systems shown to undergo light-induced electron transfer on the single-electron level, very few have displayed accumulative electron transfer in a molecular unit or complex (Molnar et al. 1994; Heyduk & Nocera 2001; Huang et al. 2002, 2004; Konduri et al. 2002). Likewise, comparatively few synthetic molecular systems have shown light-induced PCET. Within the Swedish Consortium for Artificial Photosynthesis (CAP), we have focused on synthetic systems mimicking the donor side of PSII and made the first studies of light-induced electron transfer from tyrosine and synthetic manganese complexes to appended Ru(II)polypyridine complexes as photo-active units (Sun et al. 2001; Hammarström 2003; Lomoth et al. 2006). More recently, we have also made Fe2 and linked Ru–Fe2 complexes based on the active site structure of natural Fe-hydrogenases, achieving biomimetic hydrogen production (Ott et al. 2004a,b). In the present paper we briefly review some of our work relevant to the theme of coupled electron transfers in PSII.
2. Results and discussion
(a) Accumulative electron transfer from manganese complexes
At least two problems arise in accumulative electron transfer from a molecular unit or complex, which are not relevant for charge separation on the single-electron level. One is thermodynamic and is due to the build-up of charge on the complex. This makes it increasingly difficult to remove the next electron, unless each electron transfer is coupled to a charge-compensation mechanism. The other problem is kinetic and is due to the partly oxidized complex being able to act as both an electron donor and an acceptor to the excited state of the photo-active unit. As an illustration of this point, the Mn4Ca cluster of PSII in the states S1 to S3 is thermodynamically very able to accept an electron from the excited *P680, which would lead to an electron flow in a counterproductive direction. To avoid this reaction, the electron transfer steps in the productive direction must be more rapid to compete kinetically. The importance of both these points is illustrated below.
(i) Charge compensation in accumulative electron transfer
We have previously shown the light-induced accumulative oxidation of a manganese dimer linked to a photo-active Ru(II)(bpy)3 complex (bpy is 2,2′-bipyridine, figure 2; Huang et al. 2002, 2004). Upon successive laser flash excitations, electrons were transferred from the excited Ru unit to the external, irreversible acceptor [Co(NH3)Cl]2+, and the photo-oxidized Ru(III) oxidized the manganese dimer from the Mn2(II,II) to the Mn2(III,IV) state. This was initially surprising, as only two oxidation steps of the manganese complex were found electrochemically: Mn2(II,II)→Mn2(II,III)→Mn2(III,III), both occurring below the Ru3+/2+ redox potential. We could later show, however, that the presence of water in the photoreactions led to a charge-compensating ligand exchange of the acetates for oxo ligands. A thorough study by Fourier-transform infra-red (FTIR) spectroelectrochemistry and electrospray ionisation- (ESI-) mass spectrometry (Eilers et al. 2005) and extended X-ray absorption fine structure (EXAFS; Magnuson et al. 2006) led us to the following, somewhat simplified scheme (Lomoth et al. 2006): at low water concentration (less than or equal to 10%) and on short time scales (less than 30 s), ligand exchange occurs predominantly in the Mn2(III, III) state. In 90% water instead, essentially all acetates have already dissociated in the Mn2(II,II) state.
The ligand-exchanged species in the Mn2(II,II) and Mn2(II,III) state inferred from the FTIR and mass spectrometry data (Eilers et al. 2005) have water and hydroxo ligands, and they do not carry a lower charge than the corresponding states with acetate ligands. This may seem surprising, as charge compensation was evoked to explain the facilitated oxidation. However, the exchange for water ligands opens the possibility for a proton-coupled oxidation, in which the oxidized Mn2(III,III) and Mn2(III,IV) states produced are stabilized by a charge-compensating proton release. Indeed, the data suggest the release of about one, one and (at least) two equivalents of protons upon oxidation of the Mn2(II,II), Mn2(II,III) and Mn2(III,III) states, respectively. Also, the Mn2(III,IV) complex detected by ESI-mass spectrometry appeared with two fully deprotonated, water-derived oxo ligands and has thus the same charge as the bis-acetato Mn2(II,III) complex. Interestingly, our data indicate that the electrochemical potential required to produce the di-oxo Mn2(III,IV) state in the presence of 10% water is just above that required to generate the bis-acetato Mn2(II,III) state in neat acetonitrile. Although we could not determine the formal potentials, this would imply a compression of three oxidation steps within a narrow potential range of probably less than 0.15 V, thanks to a charge-compensating ligand exchange and proton-coupled oxidation. These reactions mimic the important stepwise oxidation of the manganese cluster of PSII, which is coupled to a charge-compensating deprotonation, presumably of water-derived ligands (Junge et al. 2002; Dau & Haumann 2006).
The FTIR spectroelectrochemistry data in dry acetonitrile (Eilers et al. 2005), when the acetates remain coordinated, may serve as references for IR-spectral changes expected for oxidation of carboxylate-bridged di-manganese units of the Mn4Ca cluster of PSII. For the Mn2(bpmp)(Ac)2 complex (i.e. the same complex as in figure 2 but not linked to a Ru complex), the asymmetric stretch wavenumber changes as little as from 1594 cm−1 (Mn2II,II) to 1592 cm−1 (Mn2II,III) to 1586 cm−1 (Mn2III,III) upon oxidation. The wavenumbers for the symmetric stretch changed somewhat more, from 1422 cm−1 (Mn2II,II) to 1390 cm−1 (Mn2II,III) to 1384 cm−1 (Mn2III,III), but were broader and overlapped more with other absorption peaks. Thus, each peak shifts 6 cm−1 or less in most cases, which would be very difficult to detect in FTIR difference spectra of PSII. More comparisons with IR data from other model Mn–carboxylate complexes in different oxidation states up to MnIV would be necessary to say whether these small changes are typical or not and in order to guide the interpretation of FTIR data from PSII.
(ii) Competing kinetics in accumulative electron transfer
The excited state of the Ru unit of the Ru–Mn2(II,II) complex in figure 2 has a lifetime of 110 ns (Sun et al. 2000). This allows efficient electron transfer to an acceptor to generate the Ru3+ state, and subsequent oxidation of the manganese dimer. When the manganese is already oxidized, however, the Mn2(II,III) quenches the excited state so that its lifetime is much shorter (G. Eilers, L. Hammarström and R. Lomoth 2006, unpublished data). Although we have not elucidated the quenching mechanisms, the effect is that the shorter lifetime reduces the yield of electron transfer to the acceptor and is presumably one reason for the relatively low yield of manganese oxidation per flash (Huang et al. 2002).
In the corresponding Ru–Ru2 complex, where the Ru dimer is isostructural with the Mn dimer above, we examined the quenching reactions in some detail in different oxidation states of the appended Ru dimer: Ru-Ru2(II,II), Ru-Ru2(II,III) and Ru-Ru2(III,III) (Xu et al. 2005). In all states the excited state lifetime was very short, on the sub-nano-second time scale, owing to quenching reactions. For Ru-Ru2(II,II), we concluded that the main quenching mechanism was exchange energy transfer, while in the higher oxidation states we could not discriminate between exchange energy transfer and electron transfer from the excited state to the Ru dimer. Nevertheless, irrespective of which mechanism is responsible for quenching, the net result is not productive for photo-oxidation of the Ru dimer, and the short excited state lifetime makes it difficult to obtain efficient electron transfer to an external acceptor. Instead we managed to increase the excited state lifetime by one order of magnitude by synthetically manipulating the remote ligands of the photo-active Ru(II) unit to localize the excited state (a metal-to-ligand charge-transfer state) away from the appended dimer and thus decrease the quenching rate. Using this approach, we have previously decreased the quenching in a Ru–Mn complex up to a factor of 600 (Abrahamsson et al. 2002). Importantly, the rate of the desired electron transfer to the photo-oxidized Ru unit in the subsequent reactions was not decreased, because the electron is transferred to a metal-based orbital coupled via the same linking ligand.
These results show that unwanted quenching reactions of the photosensitizer might be very rapid and thus a real problem for accumulative electron transfer, also when the units are at approximately 15 Å distance. Nevertheless, the approach to localize the excited state away from the quenching Mn2 or Ru2 unit, while the ‘hole’ on the oxidized photosensitizer is still on the Ru metal, shows some promise. Note that it may also be analogous to the situation in PSII, for which a different distribution of the *P680 and P680+ states over the four central chlorophylls is discussed (Dekker & van Grondelle 2000; Renger & Holzwarth 2005).
(b) Long-lived charge separation in a manganese-based triad
By attaching two naphthalene diimide (NDI) electron-acceptor units to the Ru–Mn2 complex above, we obtained the first manganese-based charge separation triad (figure 3; Borgström et al. 2005). Excitation of the Ru unit lead to rapid (τ=40 ns) electron transfer to the acceptor and oxidation of Mn2(II,II). The charge-separated state had an average lifetime of 600 μs at room temperature, which is already unusually good for triad systems. By performing the experiments at 140 K instead (in fluid butyronitrile), the charge separation lifetime was dramatically increased to approximately 0.5 s, which is on the same time scale as charge recombination in photosynthetic reaction centres. We could detect both the oxidized donor and the reduced acceptor, and follow their recombination in a one-to-one ratio. The reaction was also reversible and we could run the triad through at least five charge-separation cycles without any detectable change in reactivity. This shows that it is a genuine charge-separated state of the triad.
The strong temperature dependence of the charge recombination process was analysed by Marcus theory (Marcus & Sutin 1985)(2.1)where kET is the observed electron transfer rate constant; λ is the reorganization energy for electron transfer; A is a pre-exponential factor; and the other symbols have their conventional meanings. From the slope of the plot of ln(kETT1/2) versus 1/T, and the value of ΔG0=−1.07 eV, we derived a reorganization energy λ=2.0 eV. This is twice as high as typically observed for electron transfer in polar media and implies a large inner reorganization energy of the manganese complex. By analysing the crystal structures of the Mn2(II,II) and Mn2(II,III) dimers, with reasonable assumptions for the bond force constants, we found that an inner reorganization energy in the order of 1.0 eV is reasonable. The bond-length changes are mainly a shortening of all the Mn–ligand bonds (also along the Jahn–Teller axis of Mn(III)) of the manganese that is oxidized. This supports the experimental value, which includes an additional approximately 1.0 eV of solvent reorganization energy.
Conventional wisdom concerning long-lived charge separation is that the recombination reaction should be in the Marcus inverted region, where the driving force (−ΔG0) is larger than the reorganization energy. This combines a reasonable energy content with a somewhat activated, and thus slow, recombination (figure 3). Interestingly, our results show long-lived charge separation in a different regime, in which the energy content is also high (approx. 1.0 eV) but the activated, slow recombination lies in the Marcus normal region. Because electron transfer from manganese is often accompanied by a large reorganization energy, it is a slow donor (Abrahamsson et al. 2002), but once it is oxidized this intrinsic property helps to maintain a long-lived charge separation. A long-lived charge-separated state may allow for further light-induced charge separation and accumulative electron transfer, which we are currently exploring in this type of triad system.
(c) Proton-coupled electron transfer from tyrosine and tryptophan
As a model for PCET from tyrosine, we studied the intramolecular PCET of the Ru(bpy)3–tyrosine complex of figure 4 (Sjödin et al. 2000, 2002, 2004). The Ru unit was photo-oxidized to Ru(III) by a laser flash in the presence of an external electron acceptor (methylviologen, [Ru(NH3)6]3+ or [Co(NH5)Cl]2+). The subsequent PCET from the tyrosine is bidirectional, meaning that the electron goes to Ru(III) and the proton goes to the external aqueous solution. Bidirectional PCET reactions are often found in radical enzyme systems such as PSII, but more rarely in model systems (Chang et al. 2004; Mayer & Rhile 2004). The figure shows the pH dependence of the PCET rate constant. At pH>10 the tyrosine was already deprotonated and a rapid electron transfer was observed. At pH<10 a pH-dependent rate was observed, a pH dependence of a kind that had not been identified before.
The stepwise PCET mechanisms—proton transfer followed by electron transfer (PTET) and electron transfer followed by proton transfer (ETPT)—would not show this pH dependence. Based on well-established models and considerations, the limiting rate constant for proton transfer to H2O, OH− or the base form of the buffer are all too slow to explain the observed rates by any PTET mechanism (Sjödin et al. 2004, 2005). They would also give a slope of either 0 or 1 in figure 4, which is different from the slope of approximately 0.4 we observed. Instead we assigned this to a concerted electron–proton transfer (CEP) mechanism, in which the reaction coordinates for electron and proton are evolved to a common transition state. The assignment to a CEP mechanism was supported by a significant kinetic deuterium isotope effect and an activation energy that is larger than expected for a pure electron transfer.
As the tyrosine changes pKa from 10 to −2 upon oxidation, the driving force for the overall PCET reaction increases with 59 meV per pH unit. We found that the rate followed the same dependence on pH, and thus on driving force, as that expected from the Marcus theory for pure electron transfer. This is in contrast to the previously accepted idea that a CEP with proton release to the aqueous solution should not show pH dependence (Krishtalik 2003). Our finding is thus new, but a physical model that connects the microscopic PCET events to the macroscopic free energy dependence on pH, due to dilution of the released proton, remains to be developed to give a complete explanation of our results.
A bidirectional CEP reaction may be expected to have a larger reorganization energy than a pure electron transfer, owing to the additional solvent repolarization due to the proton release (Hammes-Schiffer & Iordanova 2004) and the internal bond-length changes in the phenolic group (Sjödin et al. 2005). Nevertheless, it can outcompete an ETPT mechanism because it uses all available free energy in a single reaction step. With the larger reorganization energy, the dependence of the rate on driving force in the Marcus normal region is less steep. Thus, by increasing the oxidant strength of the Ru(III) unit with ethyl ester substituents, we could favour ETPT and switch the dominating mechanism from CEP to the pH-independent ETPT (figure 5). The CEP mechanism could only compete at high pH values. In the Ru(bpy)3–tryptophan complex, we changed instead the pKa of the oxidized amino acid from −2 to approximately 4.7. This increased the driving force for the initial electron transfer step of ETPT, while the driving force for CEP was approximately the same as in the original Ru(bpy)3-complex. Thus, we obtained a similar result as for the ester-substituted Ru(bpy)3tyrosine: a pH-independent rate at low to neutral pH, followed by a pH-dependent rate at higher pH. As the deprotonation of the tryptophan radical is much slower than for the tyrosine radical, we could obtain direct spectroscopic evidence for a switch from a stepwise ETPT at neutral pH to a CEP reaction at high pH (figure 5). In the former case, the TrpH+ intermediate deprotonated with a time constant of 130 ns. This is expected for an Eigen acid with a pKa of 4–5, and close to the value of 300 ns reported for TrpH+ deprotonation in DNA photolyase (Aubert et al. 2000). The competition between the ETPT and CEP mechanisms are governed by the higher driving force and higher reorganization energy for CEP. While ETPT is favoured by high oxidant strengths, CEP is energy conservative in that it uses all available free energy in a single reaction step. As low overall driving forces are typical for biological PCET reactions, one may predict that they most often follow a CEP rather than an ETPT mechanism.
Our original results in figure 4 showed strong kinetic similarities with data for TyrZ oxidation in Mn-depleted PSII, and we suggested that at pH<7, the latter also followed a CEP mechanism with proton release to the bulk. At higher pH, the TyrZ oxidation is faster and pH independent, however, owing to an internal hydrogen bond to His190. The kinetic difference compared with the low pH region—rate increase, activation energy and kinetic isotope effect—is less dramatic with this hydrogen bond than for complete deprotonation of the tyrosine as in our synthetic complex. We mimicked this situation in bimolecular oxidation of substituted phenols with internal hydrogen bonds to carboxylate groups on the phenol (Sjödin et al. 2006). The PCET still followed a CEP mechanism, but was now independent of pH because the proton was transferred to the carboxylate base. The driving force at neutral pH was smaller owing to the low pKa values of the carboxylates. Still the rate was higher without the hydrogen bond, and the kinetic isotope effect was smaller. This may tentatively be ascribed to the combined effect of a smaller reorganization energy and better proton vibrational wave function overlap due to the stronger hydrogen bond (Hammes-Schiffer & Iordanova 2004; Sjödin et al. 2006). A phenol with a stronger hydrogen bond showed a larger rate increase and a smaller kinetic isotope effect than a phenol with a weaker hydrogen bond. This is an analogy to the case in PSII, where the TyrZ oxidation rate increases due to a hydrogen bond to His190 in the Mn-depleted system. In native PSII, this hydrogen bond is believed to be stronger and the rate then is even higher.
The paper has briefly summarized aspects of coupled electron transfer in the recent biomimetic efforts of the Swedish CAP. Just as extensive studies of single-electron transfer in model systems were pivotal for our understanding of single-electron transfer in biology, we now need model studies of different types of coupled electron-transfer reactions to understand the natural systems. Moreover, we need to master these reactions on a fundamental level in order to develop molecular systems for solar energy conversion by artificial photosynthesis.
H. Dau (Freie University, Berlin). You suggested that in synthetic complexes a Mn-oxo group may be sufficient to form O2 with an outer-sphere water molecule. I think that it also will be important to consider that groups of the Mn complex may be needed to accept the protons from the outer-sphere water. Can you comment on that?
L. Hammarström. Yes, proton-accepting bases may facilitate the deprotonation of water, and thus the water oxidation. We are working to introduce such groups in synthetic complexes.
A. Aukauloo (University of Paris-Sud, France). You proposed the formation of a Mn(V)=oxo species with the bpmp ligand, but this ligand bears a phenol group. Do you think that this group stays in its normal oxidation state or gets oxidized to a phenoyl radical?
L. Hammarström. This part of our investigation is still at an early stage. As we presently see the same reaction signature with a range of complexes as reported for Mn2(tpy)2(H2O)2, i.e. formation of 16O18O with oxone in H218O, I suggested that the proposed mechanism for water oxidation involving a Mn(V)=O is more general to a variety of Mn-complex structures than presently believed, or that the mechanism is different from what has been proposed. Thus, Mn(V) may well never be involved in the bpmp complex.
V. Pecoraro (University of Michigan, USA). It is likely that the tBuOOH reactions generating O2 are going through tBuOOOOtBu giving tBuO+1O2.
L. Hammarström. We did not look for 1O2 as our isotope results already showed that this oxidant was unsuitable for our purpose. The oxygen seems to come entirely from manganese-catalysed degradation of tBuOOH.
The contributions of all present and past members of CAP, many of which are found in the list of references, are gratefully acknowledged. This work has been supported financially by the Swedish Energy Agency, the Knut and Alice Wallenberg Foundation, the Swedish Research Council, the Swedish Foundation for Strategic Research and the European Commission (‘SOLAR-H’ NEST-STRP 516510). L.H. is a Research Fellow of the Royal Swedish Academy of Sciences.
One contribution of 20 to a Discussion Meeting Issue ‘Revealing how nature uses sunlight to split water’.
- © 2007 The Royal Society