Organic, Organometallic and Bioorganic Catalysts for Electrochemical Reduction of CO2

Abstract A broad review of homogeneous and heterogeneous catalytic approaches toward CO2 reduction using organic, organometallic, and bioorganic systems is provided. Electrochemical, bioelectrochemical and photoelectrochemical approaches are discussed in terms of their faradaic efficiencies, overpotentials and reaction mechanisms. Organometallic complexes as well as semiconductors and their homogeneous and heterogeneous catalytic activities are compared to enzymes. In both cases, their immobilization on electrodes is discussed and compared to homogeneous catalysts in solution.


Introduction
The increased presenceo fg reenhouse gases in the atmosphere is amajor problem that needs to be addressed for asustainable future.O ft he greenhouse gases,c arbon dioxide is the principal concern due to its residence time in the atmosphere, which is estimated to be years, whereas water vapor hasaresidence time of days. [1] Humans use carbon in al inear way, transforming fossil fuels into atmosphericC O 2 .W ithout additional support for naturalp hotosynthetic systemst of ix atmospheric CO 2 ,w ithin ac ouple of decades, human civilization will have returnedt ot he atmosphere what natural photosynthesis had fixed over millions of years. As such, humans need ac yclic way of using CO 2 to have sustainable future.
Twom ain approaches have been suggested for addressing this issue-carbon capturea nd sequestration (CCS) [2] and carbon capturea nd utilization (CCU). [3] CCS describes the capture of CO 2 at its human-made origin (e.g.,f actories and power plants) and its sequestration in underground (e.g.,i no il wells and under ocean and underground bedrocks), without utilizing CO 2 as such. [4] Thism ethodi se xpensive and will not result in ac yclic use of carbon.
In contrast, the CCU approach coversabroad number of processes that can be appliedt oa ddress the issue in which CO 2 is not only captured but also used as af eedstock for various chemical products such as formica cid, carbonm onoxide, methanol,a nd methane. [5][6][7][8][9][10] Using this method, ac arbon-neutral fuel cycle mightb er ealized. For example, as ynthetic fuel, which has been created by recycling the CO 2 from the atmosphere with the helpo fr enewable energies, would release exactly the same amount of carbon to the atmosphere as used in its manufacture in the first instance. Therefore, the recycling of CO 2 into synthetic fuels is ac arbon-neutral energy vector provided that only renewable energies such as solar and wind power are used to input energy into the recycling process. This Reviews ummarizes the recent efforts to realize the photoelectro-and electrocatalytic conversion of CO 2 into synthetic fuels using organometallic, organic and bioorganic catalysts. We intentionally excluded photocatalytic approaches, because they go beyond the space and scope of such ar eview. However,i nterested readers are advisedt or efer to Ref. [11] and related books and papers therein. This article is divided into two main chapterst hat describe the homogenous (catalyst and CO 2 are in the same phase) and heterogeneous (the catalystm aterial is in the solid phase whereas CO 2 is dissolved in the electrolyte solution) electrocatalytic, photoelectrocatalytic and bioelectrocatalytic methodst oward converting CO 2 .I n each chapter there are subsections summarizing what has been done in the field, including our contributions.
Throughoutt his Review,t he applied potentials and/ort he potentialr anges are reported versus reference electrodes, such as the normalh ydrogen electrode (NHE), saturated calomel electrode (SCE), silver-silver chloride electrode (Ag/AgCl), and so forth, as they are reportedi nt he originalpapers. For ac omparisono ft hese reference electrodes, readers can refer to the conversion bar shown in Figure 1.
Assessing the catalystp erformance is of high importance for comparing different catalyst materials. There are severalf igures of merit given through this paper,n amely,f aradaic efficiency, catalytic rate constant,o verpotential, and turnover number. Faradaic efficiency defines the selectivity of ac atalyst towards ap articularp roduct and can be calculated as (moles of product/moles of electrons passed) (number of electrons needed for conversion).
The catalytic rate constant k for at ypical reaction of the following type[ Eq. (1)]: can be defineda sr ate = k[A] a [B] b ,w here a and b are usually (but not always) integers that are independent of the coeffi-Ab road review of homogeneous and heterogeneousc atalytic approaches towardC O 2 reduction using organic,o rganometallic, and bioorganic systemsi sp rovided.E lectrochemical, bioelectrochemical and photoelectrochemical approaches are discussed in terms of their faradaic efficiencies, overpotentials and reactionm echanisms. Organometallic complexes as well as semiconductors and their homogeneous and heterogeneous catalytic activities are comparedt oe nzymes. In both cases, their immobilization on electrodes is discussed and compared to homogeneous catalysts in solution. Figure 1. Reference electrodepotentials versus the normal hydrogen electrode( NHE). Ag/AgCle lectrodepotentialvalue is given for a3m KCl solution. The vacuum level for the determination of energy bands are setto À4.75 eV for NHE. [12] cients n A and n B . The dimensions of k dependo nt he exponentialt erms in the rate law.I fw ed efine the sum of the exponential terms of the concentration as p in the rate law (p = a + b + …) then k will have the dimensions of concentration 1Àp per unit time. Overpotential = applied potentialÀthermodynamic (or formal)potential for conversion.
Turnover number (TON) = moles of desired product/number of catalytically active sites (or moles of catalyst).
In this Review,w eaim to give ab road overview of the field for researchers who have been working on the topic for many years as well as researchers who are starting out and would like to pursuethis type of research.

Homogeneous Electrocatalysis for CO 2 Reduction
This section of the Review coverst he use of differentc atalysts-organometallic complexes, purely organic compounds, or bioactive materials-that are used electrochemically/photoelectrochemically.T hese catalysts are used homogeneously, which meanst hey are in the same phase as the CO 2 to be reduced.

Rhenium-and Manganese-Containing Organometallic Complexes
Organometallic complexes are one of the mostp opularc lasses of materials in the field of CO 2 reduction. Although there are many reportede xamples havingv aried molecular structure, polypyridine ligandsa re extensively used by many scientists in the field. Coveringa ll of the reported polypyridine complexes would exceed the space and scope of this Review,t herefore we focus on rhenium-and manganese-containingc omplexes. However,w ee ncourage interestedr eaders to refer to one of the latest review articles summarizing the polypyridinel igands used for CO 2 reduction. [13] Amongt he polypyridinec omplexes, Re-containing complexes are of wide interest.
The first of the Re-containing complexes was reported by Hawecker,L ehn and Ziessel in 1984. [14] In their paper,t he authors describet heir findings on [Re(bpy)(CO) 3 Cl] (bpy = 2,2'-bipyridine), whichh ad been introduced as ah omogeneous photocatalyst by the same group previously [15] for the electrochemicalr eduction of carbon dioxide to carbon monoxide. Hawecker et al. showedt hat [Re(bpy)(CO) 3 Cl] ( Figure 2) can produce 32 mL of CO if held at ap otential of À1250 mV (vs. NHE) for 14 hw ithoutd egradation, giving ar emarkablef aradaic efficiency of 98 %a nd aT ON of 300. The authorsn ote that the complex gives the highest efficiency if amixture of DMF/H 2 O( 9:1) is used together with 0.1 m Et 4 NCl as the supporting electrolyte. If no water was added,C Oproduction was observed to be much slower,l eveling off after af ew hours. [14] This study was am ilestone in the field of carbon dioxide reduction and inspired many subsequentinvestigations. Although the study of Hawecker and co-workerss et am ilestone in the field, in the initial paper,t he mechanism behind the process wasn ot elaborated in detail.H owever,t he study did have an important comparative experimental aspecti n which electrolyte solutions with and withoutw ater were used. This was an importanth int for the subsequents tudies. Sullivan et al. performed ad etailed study on the [Re(bpy)(CO) 3 Cl] complex to clarify the mechanism. [16] Their report describes the electrochemical behavior of the complex as well as the related derivatives, which led the authors toward two independent pathways for the electrochemical reduction of carbon dioxide. The derivativesu sed in the study were represented with the general formula [Re(bpy)(CO) 3 L] n + ,w here L = 4-ethylpyridine (n = 1), Cl À (n = 0) or hydride( n = 0). The authors showed that the variation of the ligand did not affect the redox potentialo f the first reversible reduction peak, whichw as observed at approximately À1120 mV (vs. NHE), and concluded that this peak originates from the reduction of bipyridine (bpy). The second (irreversible) reduction peak potential varied with the changing ligandsLsuggesting that this process is metal based. Coulometry and bulk electrolytic reduction at À1100 to À1200mV( vs. NHE) demonstrated that the first reduction is ao ne-electron process. However,t his process is coupled to the formation of the sparingly soluble green dimer [Re(bpy)(CO) 3 ] 2 .T he authors characterized this dimer by UV/Vis,I Ra nd 1 HNMR spectroscopy as well as elemental analysis. They also produced this dimer by chemicals ynthesis to further support their spectroscopic conclusions. The mechanism for the formation of the Re-Re dimer proposed by Sullivan et al. [16] can be summarized as follows [Eqs. (2)-(5)]: ½ReðbpyÞðCOÞ 3 Cl þ e À Ð½ ReðbpyC À ÞðCOÞ 3 Cl ð2Þ ½ReðbpyC À ÞðCOÞ 3 Cl fast ! ½Re 0 ðbpyÞðCOÞ 3 Cl À ð3Þ ½Re 0 ðbpyÞðCOÞ 3 Cl À slow ! ½ReðbpyÞðCOÞ 3 þ Cl À ð4Þ The lack of redox activity,i ndicating the formation of dimer speciesi nt he cyclic voltammogram of the complex, was explained by the slow rate of the process. The authors suggest that loss of Cl À might be preceded by intramolecular electron transfer to am etal-based ds*o rbital[ Eq. (3)],w hich, as ar esult, might facilitate metal-metal bond formation.T he authors conducted ac onstant-potential experiment at À1500 mV (vs. NHE), which consumed two electrons per Re atom to give ar ed-purples olution. This is believed to be the anionic form ([Re(bpy)(CO) 3 ] À )o ft he complex. If CO 2 was introduced into the electrolyte the first reversible peak at À1120 mV did not show ac atalytic enhancement, suggesting that bpy does not take parti nt he catalytic reduction of carbond ioxide if ac urrent enhancement in the second reduction peak of the complex is observed (Figure 3).
The authors summarize their findings in at wo-wayr eaction scheme (Scheme 1) and state that the first path starts with ar adicalf orm of [Re(bpy)(CO) 3 ]o ri ts solvatedf orm [Re(bpy)-(CO) 3 ]·MeCN. They also conclude that dimer formation occurs only if CO 2 is not present in the reaction medium, suggesting that carbon dioxide intercepts dimer formation. The second path (a two-electron pathway) involves the anion [Re(bpy)-(CO) 3 ] À and results in the production of CO with high current efficiency.
In 1986, Hawecker,L ehn andZ iessel publishedr esults that included the extended studies on mechanism and the origin of formed products. [17] The authorsa rgued that irradiation of the complex under optimum conditions led to arhenium-to-bipyridine charge-transfer excited state, which then was reductively quenched by at ertiarya mine to give [Re(bpy)(CO) 3 Cl] À .T hey proposed that the same should be possible in an electrochemical system in which the Re complex is reduced at an electrode. They also showedt hat the maximum efficiency was reached if the reactionm ediumc ontained 10 %H 2 O. Aw ater content exceeding 10 %c aused ad ecrease in the efficiency,a nd formation of ag reen precipitate was observed. Their resultsa re in agreement with those presented by Sullivan et al., [16] and they identified the green precipitate as the dimer of the complex. The authors also concludedt heir studies by suggesting ar eaction mechanism for the formation of CO in the presence of water (Scheme 2). [17] These studies shed light onto the possible mechanistic pathways leading to electrochemical reduction of carbon dioxide to carbon monoxide. However,t here was still ad ebate on whether a1 e À or 2e À reduction pathway is preferred or if there are circumstances under which one of those two is preferred. Spectroelectrochemical IR studies [18][19][20][21] helped to clarify the mechanismf urther. Johnson and co-workerss tudied the catalytic activityo fb oth the radical [Re(bpy)(CO) 3 Cl]C and the anion [Re(bpy)(CO) 3 ] À separately.U sing ligands other than halides, they eliminated the potential problem of both mechanisms operating simultaneously duet ot he solvente ffect. The radical[ Re(bpy)(CO) 3 Cl]C À could be stabilized at lower temperatures, [22] however,a tr oom temperature it transforms into [Re(bpy)(CO) 3 ]C due to loss of chloride. In thea bsence of strongly coordinating solvents or CO 2 this species dimerizes into [Re(bpy)(CO) 3 ] 2 ,a sc onfirmed by Johnson and co-workers. [21] In contrastt oaprevious study by Sullivan et al. [16] ,J ohnson and co-workers discovered that the radical[ Re(bpy)-(CO) 3 (CH 3 CN)]C is stable in acetonitrile and is involved in at woelectron pathway for the electrochemical reduction of CO 2 .The authors provide spectroscopic evidences howingf or the first time that CO 2 does not interactd irectly with the radical anion [Re(bpy)(CO) 3 Cl]C À ,w hich can only be regardeda st he catalyst precursor,t hus ruling out the one-electron pathway.N onetheless, ao ne-electron mechanism cannot alwaysb ee xcluded. This pathway prevails in solvents such as THF or DMF,w hich cannots tabilize 18-electron radicals [Re(bpy)(CO) 3 (solvent)]C due to their weaker coordination abilities compared to acetonitrile.T his might therefore subsequently lead to dimer formation.
Later efforts have involved modification of the chemical structure of [Re(bpy)(CO) 3 Cl] in order to improve the catalytic activity.I n2 010, Smieja and Kubiak reportedt heir results for Scheme1.Reactionpathways suggested by Sullivanand co-workers. [16] Reproduced with permission from Ref. [15].
The effect of the ligand on reductiono ft he complexes varied from À696 to À1241 mV (vs. NHE) for the first reduction and from À1481 to À1616 mV (vs. NHE) for the second reduction peak in complexes 1-5.T his difference in reduction peak potentials arises from the substituents in the 4a nd 4' positions having diverse electron-donating/withdrawing abilities. The authors argued that the pK a values of the parent pyridines determine the reduction potential and they reportedt he pK a values as an increasing trend with electron-donor character (4-carboxypyridine, pK a = 3.10;p yridine, pK a = 5.17;4 -methylpyridine, pK a = 5.94;4 -tert-butylpyridine, pK a = 5.99;4 -methoxypyridine, pK a = 6.62). [24] If the solution was saturated with CO 2 the complexes showed different increases in catalysis in their second reduction potential. Whereasc omplex 2 showed little to no current enhancement, 1 showed a3 .4-fold increase. The authors compared 1 and 4,a st hese complexes outperformed the others, and observed that compound 4 reached an 18.4fold current increase, surpassing the activity of 1 (Lehn's cata-lyst) by 3.5 times. The report also argued that catalytic activity towardsCO 2 reduction not only depends on the electrocatalyst reduction potentialb ut also on many others. Although complex 5 has the most-negative reduction potential, suggesting  . Chemical structureso fv arious rhenium complexes investigated by Smieja and Kubiak. [23] that it should possess am ore nucleophilic Re center, it showed no catalytic activity towards CO 2 .T his, according to Kubiak and co-workers, suggestsapossible p-donore ffect over a s-donor effect. Finally,t he authors explained the high and longer electrocatalytic activity of 4 over 1 by the tendency of 4 to form as table Re 0 radicala nd undergo less dimerization compared to 1;they supported this with IR spectroscopy data.
In 2012, Portenkirchnere tal. investigated [25] the effect of substituents at the 5a nd 5' positions (compound 6,F igure 5) in ac omparison to the work by Kubiak and co-workers. [23] They investigated the effect of extending the p-conjugation on catalytic activity.T he effect of extended conjugation on photophysical properties as well as the synthesis of 6 is published elsewhere. [26] The electrochemical characteristics of 6 were investigated with cyclic voltammetry and the resultsa re shown in Figure 6. Compound 6 did not show ac learp eak separation between the first and the second reduction peak. However,i ty ielded am ore positiver eduction wave at around À750 mV (vs. NHE), which is 330 mV more positivec ompared to the first reduction peak of Lehn's catalyst (1). This might be explained by the increased p-conjugation. If CO 2 was introduced to as olution of 1,t he cyclic voltammogram showeda4.5-fold increase of the second reduction peak at À1750 mV (vs. NHE), whereas for 6, a6 .5-fold increase at the same potential was observed. Absolute current density was higherf or 1 with av alue of À3.47 mA cm À2 ,w hereas it was À2.56 mA cm À2 for 6.T he authors also compared the catalytic rate constant k for these compounds. Lehn's catalyst 1 showed ac atalytic rate constant k of 60 m À1 s À1 ,w hereas compound 6 yieldedak value of 220 m À1 s À1 .F inally,t he authors comparedt he electrocatalytic and photocatalytic performance of the two compounds. Experiments were conducted in electrochemical cells having an Hg eometry.T he amount of CO, as the expected product, was quantified by gas chromatography and FTIR transmission techniques. The amount of dissolved CO in the electrolyte solution was also estimated using Henry'sl aw with aH enry constant k H of 2507 atm mol À1 solvent mol À1 CO. [27] Compound 6 showed af aradaic efficiency of 45 %, whereas Lehn's catalyst showed an efficiency of 50 %i ft he potential was held constanta t À1950 mV (vs. NHE). [25] In af urther study,P ortenkirchner et al. compared different molecular structures having Re as the metal centerf or electrocatalytic CO 2 reduction as wella sp hotocatalytic reduction. [28] They used Lehn's catalyst 1 as the benchmark compound, adopted the 4,4'-dicarboxy-substituted bpy from the study of Kubiak et al., [23] and introducedt he compound [{5,5'-bis[(2,6bis-octyloxy-4-formyl)phenylethinyl]-2,2'-bipyridyl}Re(CO) 3 Cl]. The authors counter argue the previous study [23] by showing that the carboxy-substituted complex can display ac atalytic current enhancement if CO 2 is introduced into the medium. However,t ot he authors' surprise the compound did not yield products and lost its catalytic activityw ithin minutes upon applying ab ias;t he authors attributed this to instability of the compound. [{5,5'-Bis[(2,6-bis-octyloxy-4-formyl)phenylethinyl]-2,2'-bipyridyl}Re(CO) 3 Cl] did not show as ignificant current enhancement upon contact with CO 2 .H owever,n oe xplanation was given for the lack of catalytic activity of this molecule.
Substitution at the 4-and 4'-positions of the Re-complexed bipyridyl ligandsw as shown to be effective by Kubiak and coworkers. [23] In that study,K ubiak and his team investigatedl igands with different substituents at those positions. Portenkirchner et al. made ac omparatives tudy and investigated the differences in catalytic activity arising from the same group substituted at different positions (compounds 6 and 7, Figure7). [29] The authors once again showed the effect of extended p-conjugation on the absorption characteristics ( Figure 8) noting that this can have an impact on photocatalytic properties. Rotating-disce lectrode measurements in this study showed that the diffusion coefficient of 7 was 2.5 10 À6 cm 2 s À1 ,w hich was in good agreementw ith earlierl iterature values. [23,30] If the electrolyte was saturated with CO 2 ,c ompound 7 showeda n1 1-fold increase in current at the fourth irreversible reduction at À1600 mV (vs. NHE), which the authors assigned to the metal-centered reduction. With the information on catalytic current, the authors calculated the catalytic rate constant k to be 450 m À1 s À1 .U sing the same method, compounds 1 and 6 yielded k values of 60 and2 20 m À1 s À1 ,r espectively.D espite its higher catalytic rate, compound 7 showedafaradaic efficiency of 12 %a fter 5h of electrolysis. The authors explained this phenomenon by the inhibition of catalystm aterialt hrough side reactions such as dimerization and/or H 2 evolution. Their report also emphasizes that the natureo ft he working electrode plays an important role. The authors used two different electrodes-glassy carbon and Ptnotingt hat if the workinge lectrode wasg lassy carbon the cat-   [25] Reproduced with permission from Ref. [24]. Reviews alytic rate dropped drastically to 30 m À1 s À1 .T his difference was attributed to the availability of activated protons on the Pt electrode. [29] Althoughv ariousm olecular catalysts (porphyrins, corroles, cyclams,n aphthyridines, and so forth) with metal centers such as Pd, Ru, Fe, Co, and Ni were investigated, [31][32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49] in the years after the discovery of Lehn'sc atalyst, most research has been focusedo nR e-containing complexes. With an estimated average concentration of 1ppb, Re, together with other noble metals,i so ne of the rarest elements in Earth's crust. Knowing this fact, Hawecker, Lehn and Ziessel suggested in the outlook of their inspiringw ork [14] that the metal center should be substitutedw ith more-abundant metals such as Mn, Fe, Co, or W. However,u pt on ow there are very few studies in which the Re in bipyridine ligandsiss ubstituted.
In 2011, Deronzier and co-workersi ntroduced an ew set of catalysts namely,[ Mn(L)(CO) 3 Br] where Li s2 ,2'-bipyridine (1)o r 4,4'-dimethyl-2,2'-bipyridine. [50] The authors observed two irreversible reduction peaks at around À1600 and À1700 mV (vs. Ag/AgCl), and attributedt hesep eaks to the formation of the dimer [Mn(L)(CO) 3 ] 2 and the mononuclear anion [Mn(L)(CO) 3 ] À , respectively.T hey also attributed the oxidation peak around À500 mV to breaking of dimers to give [Mn(L)(CO) 3 There was no significant current enhancement if CO 2 was introduced into solution.H owever,i ft here was 5% H 2 Oi nt he environment,anew peak appeared at ap otential that was 150 mV more positive than the first reduction peak and an enhancement of the currentw as observed. The authors explainedt his behavior with an analogyt or henium, in which aw eak Brønsted acid such as water wasi ntroduced. This helps to stabilize the rhenium-carbon dioxide intermediate and facilitate the cleavage of one of the CÀOb onds in CO 2 to yield CO. [35] After 22 ho fe lectrolysis authors achieved af aradaic efficiency of 85 %a nd at urnover number of 34. [50] In 2013, Kubiak and co-workers published ad etailed followup investigation on the effect of Brønsted acids, including IR spectroscopy results that refine the mechanism further. [51] They optimized the experimental conditions and obtainedf aradaic efficiencies approaching unity using Mn complexes.R ecently, they improved their results even furtherb yi ntroducing bulky bipyridine ligandst oe liminate dimerization reactions. They achieved faradaic efficiencies of 96 %w ithout observing formation of H 2 as as ide product. [52] These types of catalysts are of great interestn ot only for carbon monoxide formation but also due to the use of earth-abundant metals, which might pave the way towards industrial-scale applications.

OrganicCompoundsa sH omogeneousCatalysts
This parto ft he Review focuseso nt he metal free catalysts for electrochemical or photoelectrochemical reduction of CO 2 .T o avoid possible confusion,w ew ould like to emphasize that the electrochemical systems described in this part might involve metals as electrodes. However,t he focus is on the chemical entity that acts as acatalystwithin that specific system.
One of the earliest studies that used am etal-free catalyst was reported by Seshadri, Lin and Bocarsly in 1994. [53] In this study,t he authors utilized as imple molecule, namely pyridine, to perform the electrocatalytic reduction of CO 2 to methanol. The ionic form of pyridine, the pyridinium cation, acts as the catalysti nt his electrochemical system.T he authors came to this conclusion by adjusting the pH of the electrolyte to below and above 7. If the pH was greater than 7, no electrochemical features associated with pyridine were observed, indicating that the active species is the pyridiniumc ation. The optimum pH was found to be 5.4. The authors report that under aqueous conditions pyridinium reduction is coupledv ia an electrocatalytic electrochemical-chemical mechanism,l eading to the reduction of protons to H 2 (Scheme 3). Although Pd was used as the working electrode, the authors note that this mechanism is valid for other metals using aqueous, nonaqueouse lectrolytes or mixtures of both. [54]

Reviews
With this reaction mechanism in mind, the authors claimed that the observed current enhancement shown in Figure 9w as indicating an electrocatalytic electrochemical-chemical (EC) mechanism in which dissolved CO 2 or as pecies in equilibrium was reduced by the pyridinium ion, delivering pyridine and ar educed CO 2 species. The authors also performed mass spectroscopyo nc oulometrically reduced solutionsc ontaining CO 2 and pyridinium, as well as gas chromatography,d emonstrating that methanolh ad formed. Faradaic efficiencies for the methanol production varied between 22-30 %a taconstant current of 40 mAcm À2 over the courseo f1 9h on hydrogenated Pd electrodes. The authors also detected formaldehyde as as ide product.
Later in 2010, Bocarsly and his team gave mechanistic insights into how the electrochemical reduction of CO 2 to methanol proceeds. [55] They also clarified that formic acid and formaldehyde are the intermediate or side products formed during the electrochemical reduction of CO 2 to methanol. That reaction requires ap roton-assisted six-electron reduction with af ormal potential of À380 mV (vs. NHE). [55] According to their findings the mechanism involves multiple single-electron transfers. Specifically,i tw as determined to involve two-electronand four-electron-reducedi ntermediates, which are formic acid and formaldehyde, respectively.T he authors elaborated their mechanism by showing all the intermediates and side products as well as the electron flow that leads to methanol formation (Scheme 4).
Authors summarized their findings defining pyridinium radical as the one-electron charge-transfer mediator that is responsible for bringing six electrons together to drive electrochemical reduction of CO 2 to methanol. It wasn otedt hat their study is in stark contrastw ith the common idea of the need of am etal-based multi-electron transfer to achieve highly reduced species.
Bocarsly et al. expanded their studies to shed furtherl ight on the mechanism. In this study [56] they concluded that pyridinium is reduced on aP te lectrode, including ap yridiniumboundp rotont of orm as urface hydride which was further supported by the work of Batista. [57] Indeed, the study from Bocarsly and co-workers [55] created more controversy in the field and drove several studies. [58][59][60][61] Another study was conducted by SavØant and co-workers arguing the plausibility of this catalytic process. [62] The authors argue that the process apparently works only on Pt and/or Pd, metalst hat are known to reduce hydrated protons. [63] Their counterargument further continues in two main aspects:c hallenging the fact that the faradaic efficiency was found to be 20 %, [53,55] while1 00 %o ft he electrochemicalc haracteristics from cyclic voltammetryw ere attributed to the process. Second, the authors find confusingt he lack of at ypical catalytic current enhancement for whicht he replacemento ft he peak by ap lateau is accompaniedb yalarge increase. To support their argument and for comparison reasons they used acetic acid alongside pyridine. They started off investigating the ion in discussion-pyridinium (PyH + )-which is responsible for the catalytic activity.T he pH wasa djusted to approximately2pH units below the pK a of PyH + to ensure that the pyridine is in cationic form. The authors observed an increasei nt he current upon decreasing the pH and they assign thisb ehavior to the reduction of hydrated protons, supportedb ys imulated CVs. The authorss uggested that the PyHC radicali sn ot created in the pH range of interest but the electrochemical activity arises from the reduction of hydrated protons generated by the rapid dissociation of PyH + ions. The same phenomenonw as observed if pyridinew as replaced by acetic acid. Finally, SavØant andc o-workersc ompared the cyclic voltammograms of pyridine and acetic acid in presence of CO 2 at pH values adjusted according to their pK a values. At ap Ha tw hich only the "CE" (chemical reactionf ollowed by an electrochemical one) pathway is dominant, the authors observeda ne lectrochemical behaviora rising from the superposi-Scheme3.Electrochemical/chemical reactionso fpyridine under acidic conditions. tion of two acids (PyH + or AcOH and CO 2 )w hich led them to concludet hat PyHC is not formed. They observedt he same behavior with acetic acid (Figure10).
The authorss ummarized their findings by stating that they did not observe methanolo rf ormic acid formation. In addi-tion, they emphasized that the electrochemical process in question arises solely from the reduction of hydratedp rotons. Moreover,t he formation of aP yHC radical was not observed. At lower pH values the direct reduction of hydrated protons dominates,w hereas upon increasing the pH this process is realized through rapid dissociation of acids (PyH + and AcOH). [62] Portenkirchner et al. extended the study of pyridinium reductionb yi ntroducing pyridazine as ah omogeneous catalyst. [64] An earlierc omparison between pyridine and imidazole had also been published. [65] Despite having as imilar chemical structure, the two compounds have quite different pK a values: 5.14 for pyridine and 2.10 forp yridazine. [24] The authors prepared solutions containing5 0mm catalystm ateriala nd adjusted the pH to 5.3 for pyridine and 4.7 for pyridazine to obtain their protonated forms.T his pH adjustment yielded0 .25 %p yridazinium, which might support the findings of Saveant et al., in which the reduction of parentm olecule wasn ot observed and the reductive wave corresponded only to the reduction of hydrated protons created by the rapid dissociation of the parentm olecule. The electrochemical behavior of pyridinium and pyridazinium under N 2 -a nd CO 2 -saturated conditions is showni nF igure 11.Acurrent enhancement of 1.3-fold wasobservedf or pyridinium if the solution was saturated with CO 2 , while the enhancement was 5.0-fold in case of pyridazine. An ongoing debate is the effect of the working electrode on the catalytic process. This was addressed in the study by comparing glassy carbon,g old, coppera nd platinum as the working electrode and only catalytic activity was observed in case of Pt. This wasa lso supported by the work of Musgrave et al. with quantum calculations suggesting that pyridiniumw as bound to Pt surfaces with an adsorption energy of 1.0 eV per molecule. [66] The effect of catalyst concentration on the catalytic activity wasa lso studied and the difference in concentration versusc urrent enhancement was compared forp yridinium and Scheme4.Overall mechanism for the formation of formaldehyde, formic acida nd methanol. Reproduced with permission from Ref. [54]. pyridazinium.W ith increasing concentration (from 5t o 100 mm), pyridinium showed a9 .2-fold increase, whereas the increasew ith pyridazinium was1 .6-fold. The authors conducted controlled-potential electrolysis with solutionsc ontaining 50 mm catalyst materialf or over 30 h. Aliquots of electrolyte solution were analyzed by liquid injection gas chromatography equipped with af lame ionization detector.M ethanol and H 2 were detected as products and the faradaic efficiencies for methanol formation were 14 %a nd 3.6 %f or pyridinium and pyridazinium,r espectively.I ti sa lso argued in the paper that electrochemical characteristics responsible for the currente nhancement might represent as ynergetic effect arisingf rom both the catalytic activity and the effects of acids present in the solution,n amely pyridinium/pyridazinium and CO 2 .S olutions containing acetic acid did not yield methanol after 30 h of exhaustive electrolysis.
This study adds further evidences upporting the production of methanol using the small molecule pyridine as the electron shuttle to lead asix-electron reduction of CO 2 to MeOH.

Enzymes for HomogeneousBioelectrocatalysis
In considering organic and metal-organic compounds as catalysts, biocatalysts or enzymes also have to be taken into account. Enzymesa re proteins with an active site, responsible for catalyzing diverse reactions. In the case of CO 2 reduction, dehydrogenase enzymes have especiallyg ained interesta st hey are capable of reducing CO 2 to formic acid, formaldehyde, methanol, or carbon monoxide.T he use of enzymest od rive electrocatalytic reduction of CO 2 has advantages. Firstly, due to electrostatic interactions enzymes tend to retain the most favorable structure, exposing am aximum amount of catalytically active sites, which in turn increases the probability of enzymesubstrate interactions (CO 2 is the substrate in this case). Secondly,byd efinition, enzymes have high selectivity towards certain products,and side reactions are suppressed.
The report from Hçpnera nd co-workers [67] is considered to be one of the earliests tudies on enzymatic CO 2 reduction after the discovery by Stephenson and Stickland showingt he synthesis of formic acid from H 2 and CO 2 using E. coli cells. [68] The authors used nicotinamide adenine nucleotide (NADH) as the reducing agent for catalyzing the reduction of CO 2 with formate dehydrogenase (FDH) to yield HCO 2 À ,w hich is considered to be one of the safest methods to store hydrogen. [69,70] They also showedt hat the source of carbon in formate was indeedC O 2 using experiments with 14 C. In 1982, Klibanova nd co-workersp ublished as tudy showinge nzymatic production of formic acid using H 2 and CO 2 with different electron carriers. [69] The authorsa lso introduced am ethodf or utilizing formica cidt oproduce H 2 gas.
Although these early papers did not address the electrochemicalr eduction of CO 2 ,t hey were the pioneering studies that planted the idea of using enzymes as catalysts. Although enzyme-catalyzed CO 2 reductionsp rovide sustainable pathways with high selectivities and yields of generated products, the application of biocatalysts is limitedt ol aboratory-scale experiments. Due to the necessity of sacrificial co-factors such as NADH, those processes are limited, as synthesis and regeneration of co-factors is expensive. Substitutiono fc o-factors as electron donors by photochemical, photoelectrochemicalo r electrochemical strategiest hereforeb ecame highly attractive. [71] Parkinson and Weaver reported the electrochemistry of an enzymeu sing am ediatorf or the first time in 1984. [72] The authors described the use of as emiconductor,p -type indium phosphide (p-InP),t ogether with FDH to yield formic acid using CO 2 .T his was ap hotoelectrochemical system using the p-InP electrode as the source of photogenerated electrons and methyl viologen (MV 2 + )s erved as the electron shuttle for the FDH. The mechanism for the CO 2 reduction is shown in Scheme 5. The potentials given in the scheme represent the formal redox potentials of MV and FDH, respectively.U pon irradiation, an electron is excitedtothe conductionb and, whereas the hole in the valence band is extracted by the electrode and the electron in the conduction band is transferred to MV 2 + , converting it to MV + .T he authors obtained af aradaic efficiency of 89 %a nd aT ON of 21 000 using their system. [69] This Kuwabata, Ts uda andY oneyamal ater used FDH and methanol dehydrogenase (MDH) to convert CO 2 to formic acid and methanol,r espectively. [73] They used MV 2 + or pyrroloquinoline quinone (PQQ) as an electron mediatori nb uffered solutions applying potentials rangingf rom À700 to À900 mV (vs. SCE) and achieved faradaic efficiencies of 90 %. Small amountso f formaldehyde were also observed. The authors found out that methanolp roduction begins only if the formaldehyde concentration reaches1mm in the electrolyte solution.T he authors concluded that upon changing the concentrationo fb oth MDH and the mediator MV 2 + ,t he formationo fm ethanolp roceeds through formate and formaldehyde consecutively.T he faradaic efficiency for methanol was calculated as 45 %. If the electron mediatorw as changed from MV 2 + to PQQ there was ar emarkable change in the faradaic efficiencyf or methanol production.T he authors did not observe the presence of formaldehydei nt he electrochemical cell, whereas the amount of methanol reached 1.4 mmol after ac ertain initial period, yielding af aradaic efficiency of 89 %. [73] This suggests ad ifferent mechanism for the formation of methanolt hat does not involve formaldehydea sa ni ntermediate. However, the authors did not comment on the mechanism in their study.
Kim and co-workers used carbon monoxide dehydrogenase (CODH) to drivea ne lectroenzymatic reductiono fC O 2 . [74] In their report,t he authors note that MV 2 + was also required to address the enzymee lectrochemically,w hereby they used ag lassy carbon disk as the workinge lectrode. They achieved af aradaic efficiency reaching unity at thermodynamic potentials (À570 mV vs. NHE) with aT ON of 700. The high faradaic efficiency was attributed to the highly selectiveb inding of CO 2 to the enzyme active center whichc ontains Ni and Fe atoms. The electrochemical behavior of MV 2 + and CODH in solution was investigated using cyclic voltammetry ( Figure 12).
The authors argued that the electrochemical wave observed upon saturating the solutionw ith CO 2 ( Figure 12, curve b) is an indication of the electrochemical reduction of CO 2 to CO by CODH. They explaint he lack of catalytic enhancement in the currentw itht he high formal reduction potential of direct CO 2 reduction.H owever,t hey do not show any spectroscopic and/ or electrochemical data to confirm the function of the enzyme in the solution using MV 2 + .
Another study was conducted using Candida boidinii formate dehydrogenase (cbsFDH), which is an enzyme requiring electrons,p rotons and NADH to drive CO 2 reduction to formic acid. MV 2 + was used as electron shuttley ielding2 4% faradaic efficiency. [75] NADH is an expensivem aterialt ou se as the electron source and needs to be regenerated.F or that reason,a uthors introduced an Rh complex, [Cp*Rh(bpy)Cl] + .U sing cbsFDH, the authors increased selectivity towards formate withouthaving HCO 3 À as as ide product. Similar studies can be found in the literaturei nw hich the effects of different enzymet ypes having differentm etal active centers and the contribution of those metalst ot he formation of CO and formate were investigated. [76,77] Nørskov and coworkers reported ag ood overiewo ft he design parameters for enzymatic catalysts using DFT calculations. One of the important points noted in their paper is the ability of the metal center to bind CO and HCOO À while stabilizing oxygen in the latter.T hey also suggestt hat readers use noble metal centers in enzymes as such metals have rather high overpotentials for H 2 formation. [76] In their review,M ondal et al. give mechanistic insights into CO 2 reduction and in particular,t he mechanistic aspects of its reduction to CO compared to formate ( Figure 13). [71]

Heterogeneous Electrocatalysis for CO 2 Reduction
Homogeneous approaches for CO 2 reduction have been widely used throughout the history of the field. However,asdescribed in the previouss ection, electrochemical addressing of the catalyst material,w hich is distantf rom the electrochemical double Scheme5.The photoelectrochemical production of CO 2 with the enzyme formate dehydrogenase (FDH) as the catalystand methyl viologen (MV 2 + )as the mediator.p -InP: p-type indium phosphide;V B: valanceb and;CB: conductionb and. Reproduced with permissionfrom Ref. [71]. layer might be an issue. Also, recovery of the catalyst or separation of the product from the homogeneous mixture is atechnical problem.D ifferent mechanismsc an lead to degradation, inhibition, and eventual decreasei no verall efficiency.B earing these drawbacks in mind,r esearchers have focused on the direct use of catalysts using the idea of heterogeneousc atalysis. The following sections focus on heterogeneousa pproaches towards electrochemical and photoelectrochemical CO 2 reduction.

Catalyst-Functionalized Metal/Metal-like Electrodes
As tudy by Lieber and Lewis waso ne of the earliest that investigated ah eterogeneousa pproach. [78] Pyrolytic graphite or carbon clothw ere used as electrodes and they were modified with cobalt phthalocyanine, Co(Pc),v ia drop-casting or adsorption from solutions of Co(Pc) in THF.T he authors conducted cyclic voltammetry measurements in aqueous citrate buffer (pH 5) with and without aC O 2 atmosphere.I nterestingly,t here was no change in thev oltammetric behavior of Co(Pc)s urfaces in aC O 2 atmosphere.T he authors argued that the initial step did not involve binding of CO 2 on the reduced species Co(Pc) À .H owever,t hey did not excludet he possibility of very weak binding. From the resultso fe xperiments at different pH values, the authors concluded that the mechanism involves an initial protonation of Co(Pc), followed by bindingo fC O 2 .T he authors reportedt he observed products as CO (major product), H 2 and formate/oxalate (trace amounts). They achieved faradaic efficiencieso fu pt o6 0% for CO and 35 %f or H 2 .A TON of 370 000 was achieved, which represented an improvement over the previouslyr eported result by three orders of magnitude. [31] In 1989, Furuya and Matsui reported their findings on the efficiency of 16 different metal phthalocyanines (Figure 14) in the electrochemical reduction of CO 2 . [79] They focused on different transition metal groups, namely groups VIII, IIIB and IVB.
Phthalocyanines were immobilized on gas-diffusion electrodes, which were prepared using hydrophobic carbon black, hydrophilic carbon black, and polytetrafluoroethylene. The electrolyte solution was saturated with CO 2 and the cathode (gas diffusion membrane-phthalocyanine) was purged constantly with CO 2 and H 2 from the back side of the electrode. Electrolyses were conducted under galvanostatic conditions with 0.5 m KHCO 3 serving as the electrolyte. Table 1s ummarizest he metals used, the main products formed and correspondingf aradaic efficiencies achieved.
The authors also providei nsighti nto the probablem echanisms leadingt owards the products. For CO formation,t he authors suggest coordinationo fC O 2 to the metal center to be the first step. Addition of two hydrogen atoms to the metalcoordinated CO 2 was proposed to be the initial steps of formic acid formation.F inally,m ethane was proposed to form by elimination of an oxygen atom after formationo fC O. Mechanistic detailsa nd as tep-by-step breakdown of the reactions are reported in the paper. [79] Another approachi nw hich macrocycles wereu sed as catalysts came from researchers at Hokkaido University in 1991. Enyo and co-workersi mmobilized cobalt(II) tetraphenylporphyrin( CoTPP) using pyridine as the chemical anchoro n ag lassy carbon surface. [80] Preparation of modified glassy carbon electrodes is shown in Scheme 6. CO was detected as the only product after constant-potential electrolysis at potentials from À1000 to À1300 mV (vs. SCE). The authors observed the optimal potentialatÀ1100 mV and achieved afaradaic efficiency of 92 and ar emarkable TON of 10 7 .T oa ssess the effect of fixation on the stability, the authors conducted constant-po- Figure 13. Overview of the bio-inspired mechanism for the two-electronr eductiono fCO 2 to CO and formate (PCET = proton-coupled electrontransfer). Reproduced with permission from Ref. [70].  tential electrolysis experiments under homogenous conditions in which they dissolved CoTPP and pyridine in as olutiono f tetra-n-butylammonium fluoride in DMF.C atalytic activity degraded rapidly when the potential was applieda nd ab lack precipitate was observed at the bottom of the electrolysis cell, whereas the modified electrodes showedc onstantc atalytic activity foru pt o5days. The authors explained the stability of the electrodes and hence the stabilityo fc atalytically active CoTPP with the presence of the pyridine ligand, which introduces isolation of CoTPP centers from each other.T he second improvement from the pyridine ligand is the introduction of vacant sites for CO 2 due to the direction effect by the ligand. [80] Yoshida and co-workersc ontributed study of immobilized catalysts by introducing catalysts that were incorporated into coated Nafion membranes. Coated Nafion served to create ah ydrophobic environment aroundt he catalyst to suppress hydrogen evolution. [81] The catalysts used were [Re(bpy)-(CO) 3 Br] and [Re(terpy)(CO) 3 Br] (terpy = 2,2':6',2"-terpyridine). Basal-plane pyrolytic graphite (BPG) was used as the electrode, which was coated with Nafion containing 2.6 mmol catalyst material. Constant-potential electrolysis resulted in av ariety of products (HCOOH, CO and H 2 ), formic acid being the main one. Am aximum faradaic efficiency of 48 %w as achieved for formic acid with aT ON of 98 with [Re(bpy)(CO) 3 Br] as the catalyst in the Nafion matrix. Higher TONs were achieved for CO production,the highest was 198. [81] The same group also reported the immobilization of aC oPc-poly(4-vinylpyridine) matrix on ap yrolytic graphite to achieve CO 2 reduction with CO as the main product. The faradaic efficiencies varied between 31 %a nd 43 %, with remarkable TONs of around1 0 5 . [82] Interestingly,i nt he years that followed,there have not been many studies on catalystimmobilization on metal or metal-like surfaces.
Copper has alwaysb een the choice of metal if severalp roducts and higher hydrocarbonss uch as methanol, methane, propanol, or formic acid were desired. Readers are advisedt o read the detailed work of Hori et al. on the electrochemical reductiono fC O 2 using various metals. [83] Flake et al. reported their findings on copper oxide as ac atalyst for electrochemical CO 2 reduction. [84] They used copper foils as the electrodes, which formed cuprous oxide (Cu 2 O) on the surface. Thee lectrodes were tested in 0.5 m KHCO 3 electrolyte for their performance as CO 2 reduction catalysts. Methanol was the major product after 30 min electrolysis at À1100 mV (vs. SCE) with af aradaic efficiency of 38 %. If the potentials exceeded À1550 mV (vs. SCE), there wasarapid decrease in methanol production and H 2 evolution prevailed. Also, electrolysis durations of more than 30 min yielded methanea st he main product. This behavior can be explained by the reduction of Cu I species over time to Cu 0 to give methane. Another important point demonstrated by the authors was that the mechanism goes throughH 3 CO À species (Figure 15). [84] Other studies using Cu 2 Of ilms as well as catalystsh aving core-shell structures have been reported for similar materials with comparable efficiencies. [85,86] Recently,r esearchers from the University of Liverpool adapted the study of Yoshida and co-workers to use [Mn(bpy)-(CO) 3 Br].T hey immobilized the catalyst in am atrix covering ag lassy carbon electrode. [87] The electrochemical characteristics of the compound were in agreement with previous stud-Scheme6.Procedure for the preparation of CoTPP-pyridyl-glassy carbon electrode. Reproduced with permission from Ref. [79]. Reviews ies. [50,51] Using this compound, the authors obtained CO and H 2 as products.F ormationo ft he primary product was dependent on the potential applied. For example, if À1500 mV (vs. Ag/ AgCl) was applied, CO was the main product with afaradaic efficiency of 51 %. If the potential was switched to À1600 mV (vs. Ag/AgCl) H 2 evolution dominated with 81 %f aradaic efficiency.I tw as noted that use of glassy carbon, which has as mooth texture, leads to lower concentrationso ft he catalyst materiali nt he Nafion matrix. To fix this issue, the authors included multiwalled carbon nanotubes (MWCNTs) for increased surfacea rea and therefore an increased catalyst concentration. A1 0-fold current increase was observed under CO 2 -saturated conditions if MWCNTs werei ntroduced; however,H 2 was the main product instead of CO. [87] For somet ime, Nafion was the materialo fc hoice for fixing electrocatalysts onto electrodes. The idea is to have ah ydrophobic environmenta roundt he catalysta nd thusm inimize H 2 evolution to ac ertain extent. However,t his prevented the access of the electrolyte solution into the matrix in the case of an aqueous environment.
Hupp and co-workers formulated the idea of incorporating ak nown catalyst, that is, iron tetracarboxyphenylporphyrin (Fe-TCPP), into am etal-organic framework (MOF). [88] The authors note that the choice of MOF ( Figure 16) facilitated the access of solvent, reactant and electrolyte solution further into electroactive sites. Furthermore, the metalloporphyrin linkers within the MOF served as both electrocatalysts and as redoxhopping moietiesf or the delivery of reducing equivalents to the catalytic sites. The authors achieved faradaic efficiencies of up to approximately 60 %f or CO production.
The majority of studies in which porphyrins and phthalocyanines wereu sed as catalysts did not include detailed mechanistic accounts. Koper andc o-workers utilized in situ measurement techniques such as online electrochemical mass spectroscopy and online HPLC to address this issue. [89] In their study, the authors used cobalt protoporphyrin ast he electrochemical catalyst, which they immobilized on ap yrolytic graphite electrode. CO and CH 4 were the two main products.M ethanep ro-ductionw as achieved by reducing CO with HCHO as an intermediate. The mechanism proposed by the authors is given in Figure 17. [89]

Catalyst-Functionalized Semiconductor Electrodes
This part of the Review is focused on the direct use of semiconductor electrodes for reduction of CO 2 as well as catalystfunctionalized semiconductor electrodes.
One of the early studies from Halmann used p-type gallium phosphide (GaP) as the photoelectrode for driving the reduction of CO 2 . [90] GaP was immersed in an electrolyte solution together with agraphite rod as the counter electrode and asaturated calomel electrode as ar eference electrode. The choice of counter electrode wasastrategic decision, as it was reported that carbon oxidizes neither formic acid nor carbohydrates to carbon dioxide. [91] The GaP electrode was illuminated using aH ga rc lamp and was biased with À1000mV( vs. SCE). Analysis of the electrolyte revealed the presence of formica cid, formaldehyde andm ethanola tc oncentrationso f1 .2 10 À2 , 3.2 10 À2 and 8.1 10 À4 m,r espectively,a fter 18 ho fi rradiation. [90] In 1983, another study in which GaP was used for photoelectrochemical reduction of CO 2 reported faradaic yields of 15.2 %f or HCOOH and the influence of the pressureo nt he re-ductionofC O 2 wasinvestigated. [92] The use of semiconducting electrodes for reduction of CO 2 was favored in the years that followed. Canfield and Frese published their findings on the use of GaAs and InP as the catalytic materials. They used n-type GaAs to drive the electrocatalytic reduction of CO 2 to methanolw ith af aradaic efficiency of 89 %, whereas p-type GaAs and InP were used as photoelectrocatalysts to yield methanolw ith faradaic efficiencies of 52 % and 80 %, respectively. [93] Building on their own studies, [53] Bocarsly and co-workers reportedt he use of ap-type GaP semiconductor as the electrode and pyridinium as the catalyst. The conditions they used were almosti dentical for pyridinium except the electrode wasG aP insteadofhydrogenated Pd.They achieved faradaic efficiencies up to 96 %. [94] The use of semiconductors as electrodes and catalysts in solution was investigated by severalg roups.T hese studies yielded ar ange of products, from CO and H 2 [95][96][97] to methanol. [98] Several other studies focusedo ni mmobilization of catalytically active materials on semiconductor surfaces. Ghosh and Spiro reported their results on covalenti mmobilization of [Ru(bpy) 3 ] 2 + on SnO 2 surface ( Figure 18). [99] Although they did not investigate the catalytic activity of the Ru complex,t his was an important study in terms of immobilization of such compounds on semiconductor surfaces and in terms of their electrochemistry and photoelectrochemistry.

À1
.I nt he following year,t he same group reportedt heir findings on immobilized Mn complexes using the same ligand mentioned above to Figure 16. Illustration of aportion of the crystal structure of MOF-525inporphyrin free-base form, including the chemical structure of the TCPP linker and the Zr 6 -basedn ode. Reproduced with permission from Ref. [87].  3 Br] (MnP). [101] MnP was immobilizedo nm esoporous TiO 2 by dropcasting and the amount of catalystw as calculated as 34 nmol cm À2 .T he authors claim that the phosphonic acid groups act as the anchoring moieties. They achieved aT ON of 112atmoderately low overpotentials of 420 mV,with afaradaic efficiency of 67 %. The authors proposedm echanism for the process via immobilizedM nP is shown in Scheme7,a nd they noted that use of such immobilization techniques will help to improvel ong-term stability and conductivityu nder reducing conditions. In addition, the presence of a3 Ds tructure would increaset he catalyst loading as well as facilitate intermolecular interactions. [101]

Catalyst-Functionalized Organic Semiconductor Electrodes
This sectioni sf ocusedo nc atalyst-functionalized electrodes for the heterogeneous electrochemical/photoelectrochemical reductiono fC O 2 .T he term "organics emiconductor electrodes" denotes that the electrons are transferred via the organic semi-conductor to the catalyst material or to CO 2 ,a lthough it does not mean that the electrode itself is af ree-standing organic semiconductor structure.
Wrighton and co-workers reported the immobilization of Pd in ab ipyridine-based polymer,( PQ) 2 + ,a nd its catalytic activity for reduction of HCO 3 À to HCO 2 À in the presence of H 2.
[102] The authors first polymerized the bipyridine monomer ( Figure 19) on tungsten wire and then impregnatedt he polymer matrix with Pd, whichw as achieved by consecutive dipping of the electrode, first into K 2 PdCl 4 and then into 0.1 m KCl solution, and final electrochemical treatment to yield metallic Pd.
The authors conducted controlled-potential electrolysis in carbonate-containing de-oxygenated solutions using the prepared electrodes and analyzed the products. To avoid the possibilityo fo btaining HCO 2 À as the product as ar esult of degradation of the polymer electrode they also conducted experiments with 13 C-enriched carbonate solutions and confirmed the formation of formate using NMR spectroscopy and/or enzymea ssays. Af aradaic efficiency of 80 %w as achieved and the losses were attributed to the formation of H 2 or formation of palladium hydride through the reactionP d+ x H + + x e À ! Figure 17. H + and H 2 Oa re the hydrogens ources for the hydrogen evolution reaction at pH 1a nd 3, respectively.CO 2 C À is the initial intermediate for the reductiono fCO 2 to CO. CO can be further reduced to methane with HCHO as an intermediate. The catalytically inactive "resting" state of the Co is assumed to be 2 + .T he reduction of Co 2 + to Co + is supposed to trigger boththe H 2 evolution and CO 2 reduction pathways. Reproducedw ith permissionfrom Ref. [88]. PdH x as wella st he reduction of the polymer itself. [102] The study by Wrighton et al. was one of the earliests howing that heterogeneous catalysis can be achieved for CO 2 reduction and organic semiconductors can be utilized fore lectron-transfer purposes.
Another study in which MV 2 + was used as af unctional group attached to ap olymer was reported by Sariciftcia nd coworkers. [103] The authors obtained results on methylviologenfunctionalized 3-alkylpolythiophenes. The study showedc haracterization of the polymer PTV 2 + using in situ IR spectroelectrochemistry,U V/Vis ande lectron spin resonance (ESR). The authors named this new type of polymers as the third generation of conducting polymers ( Figure 20) where as olution-processable and functionalized polymer structure is achieved. ESR studies showedt hat the electrochemical addressing of viologen moiety is possible. This study did not addressi ts catalytic properties,h owever, other studies were inspired in which functionalized conducting polymers were used [103] .
Scheme7.Illustrated proposed mechanism for CO 2 reduction by TiO 2 -MnP (X = Br À in the isolated compound). Reproduced with permission from Ref. [100].  In terms of polymers that contain Re complexes as the active catalyst in their structure,astudy from Meyer and coworkers [104] was ad oor opener. The authors reported the growth of Re-containingp olymers using 4-vinyl-4'-methyl-2,2'bipyridine as both the ligand for Re and the monomer.T he authors notet hat the polymer that formed on the Pt electrode might be result of dimerizationo ri tc ould be the MeCN-containing complex as reported previously. [16] The observed green color of the freshly electropolymerized film suggested the dimerization pathway as the dominant one. However,t his color later disappeared. The authors performed controlled potentialelectrolysis in the presence of CO 2 -saturateds olutionsa t À1550 mV (vs. SCE) to drive electrocatalytic reduction of CO 2 to CO and reachedafaradaic efficiency of 92.3 %. Based on the estimated amount of the catalysto nt he polymer it was concluded that the TON was5 16. Another important point observed by authors that no formation of CO 3 2À speciesw as observed. This result was in contrast with the resultso fh omogeneous catalysis for which CO and CO 3 2À were detected together.T his study was ak ey success factor for the Re-containing polymere lectrodes in the field and was followed by as tudy from Cosniera nd co-workers in 1986. [105] In this work, [Re-(bpy)(CO) 3 Cl] was attached to pyrrole at the nitrogen atom and gave ap olypyrrole backbone from which the catalyst was ap endant group ( Figure 21). If controlled-potential electrolysis was performed in the presence of CO 2 ,t he authors found results that contradicted those of their previous study in which they observedf ormation of CO 3 2À along with oxalatea nd CO. The faradaic efficiency for CO was 78 %w hile authors also reported aT ON of 236 in this case. [105] The authors did not explain their choice of polypyrrole as the polymer backbone to drive ar eduction reaction although polypyrrole is p-type in nature;t his however might explain the decrease of catalytic activity over time.
Further studies followed investigations of the mechanism behindi mmobilized Re catalyst in polymerm atrices together with different polymers [106,107] and different metals such as Co, Fe, Ni, Os, and Ru. [108,109] The use of metal-freec arbon nanofibers for the electrocatalytic reduction of CO 2 was also reported. [110,111] Another study in which Re complex was incorporated in the main chain of ap olymer was reported by Portenkirchner and co-workers. [112] The authors used ap reviously reported modified Re complexa nd electrochemically polymerized ( Figure 22) it onto aP te lectrode potentiodynamicallyb yc ycling between À1600 and 200 mV (vs. NHE).
Structural characterization of the film duringp olymerization was performed using in situ IR spectroelectrochemistry in af low cell equipped with an Ag/AgCl reference electrode and aP tp late counter electrode. For this purpose, the film was polymerized onto aZ nSe ATRr eflection element. SEM and AFM measurements, as well as spectroscopic characterization, were performed to further investigate the Re-containing polymer films. Finally,t he film was dipped into as olution that was saturated with CO 2 .C yclic voltammetry revealed a2 0-foldc urrent enhancement in the presence of CO 2 .I fh eld at ac onstant potentialofÀ1600 mV (vs.NHE)over60min, CO was observed as the main product and 33 %f aradaic efficiency were achieved. The TON for the Re-containing polymer film was calculated as 1400.
Thiophene, aw ell-known and heavily investigated conjugated polymer buildingb lock was also used for immobilization of Re complexes. Nervi andc o-workers investigated three thiophened erivatives ( Figure 23) with Re pendant groupsa sc on-  jugated polymer building blocks for electrocatalytic reduction of CO 2 . [113] The authors investigated the catalytic properties of monomers 2 and 3 as well as their corresponding polymers. The monomers showed adecrease in the catalytic activity after 60 min. However,i ft he workinge lectrode (glassy carbon) was sonicated for 10 min the catalytic activity was restored. Electrochemically grown polymers were also tested for CO 2 reduction in CO 2 -saturated solutions. Af aradaic efficiency of 84 %w as achieved with ap olymer of 2 and3 4% with ap olymer of 3 at ap otential of À2100 mV (vs. Fc/Fc + ). [113] Apaydin et al. investigated polythiophene structures with pendantc atalyst groups for the photoelectrochemicalr eduction of CO 2 . [114] This is of particulari nterest because previous studies addressing the use of conjugated polymers lacked the information on how ap -type materialc an drive electrons through its backbone unlesst unneling wast he main driving force. [Re(bpy)(CO) 3 Cl] was attached to thiophenea tt he 3-position through an alkyl chain [3HRe(bpy)(CO) 3 Cl-Th] and then electrochemically polymerized in the presenceo fb oron trifluorided iethyle therate to yield poly[3HRe(bpy)(CO) 3 Cl-Th] ( Figure 24). After polymerization, the polymer-modified electrode wasd ipped into fresh electrolyte solution (0.1 m TBAPF 6 in MeCN) and scanned either under dark or under illumination in the presence of N 2 .The first and second peaks of the characteristics of Re complex were shiftedt om ore positivep otentials by 100 mV when the complex was illuminated, indicating light-assisted electron transport. The polymer showedafourfold increase in current upon saturation of the solutionw ith CO 2 under illuminated conditions. The authors explained the process behindt he electron transfer to CO 2 with the initial formation of an excitonu pon illumination followed by electron injectiont ot he valance band of the semiconducting polymer and then by an electron transfer from the polymer to the Re complext od rive CO 2 reduction ( Figure 25). Finally,t he authors conducted constant-potential electrolysis in CO 2 -saturated solutions at À1500 mV (vs. NHE) to obtain CO as the main product. When 10 %w ater was added, as suggested previously by Lehn and co-workers, [14] CO production dropped significantly and H 2 evolution prevailed. The authors reachedaf aradaic efficiency of 2.5 %a nd aT ON of 20 which was calculated from the estimated active sites on the surface. [114] It is explained in the study that the low faradaic efficiency can be attributed to surface limited reactivity and the choice of ah ole extractinge lectrode.H owever, this is one of the very few studies where the use of an organic semiconducting polymer and its light absorbing properties to drive photoelectrochemical reduction of CO 2 is shown.

Immobilized-Enzyme-Functionalized Electrodes
The use of enzymes for electrocatalytic reduction of CO 2 was discussed in section 2.3;h owever, am ediator( electron shuttle) was alwaysr equired. Recent studies have focused on the im-   mobilization of enzymes for direct electrochemical addressing by using ac onductive platform as well as for protecting them from environmental effects. [115] Hirst andc o-workersu sed at ungsten-containing formate dehydrogenase enzyme (FDH1), whichw as adsorbed on an electrode surface to catalyze CO 2 reduction to formate ( Figure 26). [116] The authors achieved faradaic efficiencies of around9 7-98 %a nd the enzyme retainedi ts catalytic activity over ap Hr ange of 4-8. Applied potentials to drive the reduction varied between À410 to À810 mV (vs. Ag/AgCl), which are much lower than the potentials required to use organometallic catalysts. FDH1 was the first example of this kind that can catalyzeaCO 2 reduction process with high selectivity and reversibility. [116] The study from Hirst and co-workersf ueled other heterogeneous electrocatalytic carbon dioxide works in the field. Amao and Shuto reported their findings on thec atalytic performance of viologen-immobilized FDH on an indium tin oxide (ITO) electrode. One end of the viologen was functionalized with al ong alkyl chain with ac arboxylic acida st he end cap. Carboxylic acid groups served as anchoring groups on as ol-gelprepared ITO layer,w hereas the other end of the viologen was functionalized with FDH by dippingt he electrode into an FDHcontaining solution.I ft he electrode was biased in aC O 2 -saturated pyrophosphate buffer solution at À550 mV (vs. Ag/AgCl), it yielded 23 mmol of formic acid after 3h.T he effect of the alkyl chain length was also investigated in the study and the authorsr eported at rend of increased rate of formic acid productionw ith increasing number of carbon atoms in the alkyl chain, reaching ar ate of 7.6 mmol h À1 for nine carbon atoms in the chain.
Schlager et al. demonstrated the immobilization of alcohol dehydrogenase on highlyp orousc arbon felt electrodes using aa lginate-silicate hybrid gel as the immobilization matrix. [117] Alcohol dehydrogenasec atalyzed the reduction of butyraldehyde to butanol at ap otential of À600 mV (vs. Ag/AgCl) with afaradaic efficiencyof4 0%.T he authorsalso performed acontrol experiment in which they immobilizedt he enzyme in the same matrix together with NADH as an electron donor to check the activity of the enzymes over time. They achieved a9 6% conversion if NADH was the electron donor. [117] Formation of methanol using the same methodw ith af aradaic efficiency of 10 %w as also demonstrated at ac onstantp otential of À1200 mV (vs. Ag/AgCl) [118] (Figure 27).
By expanding the idea of biocatalytic electrochemistry to include multiple enzymes immobilized on an electrode, Schlager et al. showedt hat CO 2 can be reduced to methanoli natriple cascade. [118] This studys howed the successful immobilization and electrochemical utilization of three enzymest oa chieve bioelectrocatalytic reduction of CO 2 .N ADH can be replaced in the reactionc ascade with direct electron injection to the enzymes (Scheme8). [118] Biocatalytic systems sucha se nzymes and bacteria work in the mild conditions of room temperature and atmosphericp ressure, and are superior to all other catalysts in terms of selectivity.

Summary and Outlook
We reviewed previousw ork in the field of CO 2 reduction that used organometallic, organica nd bioorganic catalysts to initiate electrocatalytic, photoelectrocatalytic and bioelectrocatalytic reactions.
Reduction of CO 2 using ah eterogeneous catalytic approach has the following advantages:i mmobilization of the catalyst materialo nt he working electrode can allow the direct use of the catalyst, bypassing the diffusion step. Another advantage of course is the reusability of catalystm aterialw ithoutt he need to recover it from the mixed chemical mediumo ft he products and catalystt ogether.T he economicso fe lectrocatalytic CO 2 reduction is still under discussion as to whether it will be feasible or not in the near future. Recent work from Kenis Figure 26. The electrocatalytic interconversion of CO 2 and formate by af ormated ehydrogenaseadsorbed onto an electrodesurface. Twoe lectronsa re transferred from the electrode to the active site (buried inside the insulating protein interior) by iron-sulfur clusters,t or educe CO 2 to formate, forming aC ÀHb ond. Conversely,ifformate is oxidized, the two electrons are transferredf rom the actives ite to the electrode. The structureo fF DH1 (which contains at least nine iron-sulfur clusters) is not known, so the structure showni st hat of the tungsten-containing formate dehydrogenasef rom Desulfovibrio gigas (PDB ID:1H0H). Reproduced with permissionf rom Ref. [115].
Scheme8.Mechanisms for CO 2 reduction catalyzedbyd ehydrogenases. Three-stepr eduction of CO 2 to methanol using NADH as as acrificial coenzyme( top) and through ad irect electron transfer to the enzyme without coenzyme(bottom). Reproducedw ith permission from Ref. [118].  [119] demonstrated the use of ag ross margin model for calculating the economicso fC O 2 reduction. The gross margin model is defined as the difference between the revenuea nd the cost of goods, divided by the revenue.T his shows that CO and HCOOHa re economically the most feasible products to pursue in terms of required potential and faradaic efficiency. However,i mprovements in catalyst durability and energy efficiency are still needed.H owever,t he activation of CO 2 can be energy demandinga nd we should certainly make use of renewable energies.O rganicp -type semiconductors can be used as electron-transfer media to provide photogenerated electrons to catalyst materials that are immobilized on the surfaces of semiconductors. Hybrids ystems in which the catalysti so f biological origin and the electrons ource is ideally ap hotoelectroactivec ompound can pave the way toward energy efficient and selectivec onversion of CO 2 .B iocatalytic systems work at room temperature and atmospheric pressure andh ave superior selectivity.These can be major factorsincalculating the economicsofl arge-scale CO 2 reduction processes.
The cyclic use of carbon in the ways described in this Review will create ac arbon-neutral energy vector,w hich is important for transforming our energy-producing sectors.