Electrocatalytic and Solar‐Driven CO2 Reduction to CO with a Molecular Manganese Catalyst Immobilized on Mesoporous TiO2

Abstract Electrocatalytic CO2 reduction to CO was achieved with a novel Mn complex, fac‐[MnBr(4,4′‐bis(phosphonic acid)‐2,2′‐bipyridine)(CO)3] (MnP), immobilized on a mesoporous TiO2 electrode. A benchmark turnover number of 112±17 was attained with these TiO2|MnP electrodes after 2 h electrolysis. Post‐catalysis IR spectroscopy demonstrated that the molecular structure of the MnP catalyst was retained. UV/vis spectroscopy confirmed that an active Mn–Mn dimer was formed during catalysis on the TiO2 electrode, showing the dynamic formation of a catalytically active dimer on an electrode surface. Finally, we combined the light‐protected TiO2|MnP cathode with a CdS‐sensitized photoanode to enable solar‐light‐driven CO2 reduction with the light‐sensitive MnP catalyst.

The reduction of CO 2 to CO is viewed as ap otentially lucrative and renewable source of ak ey chemical feedstock, as well as astrategy to reduce rising atmospheric CO 2 levels. Electrocatalysis by molecular transition-metal complexes is av iable means of achieving this transformation, typically offering excellent tunability [1] and selectivity [2] as well as providing opportunities to study the catalytic mechanism. [3] Alternatives based on inexpensive solid-state materials usually offer less well-defined catalytic centers that prevent adetailed understanding of the catalytic mechanism. [4] Immobilization of such molecular catalysts on electrode surfaces makes efficient use of the active metal centers and therefore enables at rue appraisal of properties,s uch as the turnover number (TON). [5] However,i nm ost cases reported to date,molecular catalysts were deposited on carbon [5c, 6] and Pt-based [7] electrodes.T hese offer low transparencytovisible light, and only in very few cases have the surface-bound catalytic intermediates been characterized spectroscopically in situ. [2c, 8] Bimolecular reaction mechanisms,inwhich active dimers form during catalysis,h ave not been observed on electrode surfaces,a nd it has been thought that such mechanisms would be impeded by immobilization of amonomeric pre-catalyst. [5b, 9] First-row transition-metal complexes based on [MnBr-(CO) 3 (L)] (L = bipyridine and derivatives) have emerged in recent years as promising electrocatalysts for CO 2 reduction owing to their high selectivity and low overpotential for catalysis. [10] They also contain only Earth-abundant elements, which is as ignificant advantage over analogous Re-based catalysts. [7b,8,11] Thelow overpotential is adirect consequence of the bimolecular reaction mechanism, whereby aMn 0 -Mn 0 dimer is formed after the first reduction of the homogeneous molecular catalyst, which then reduces CO 2 to CO (L = 4,4'dimethyl-2,2'-bipyridine). [10a] However,t he maximum TONs achieved by this class of complex for electrocatalytic CO production are 34 after 18 h, [10a] and 36 after 6h. [12] Mn catalysts have been integrated onto electrodes in polymer films,such as Nafion, where they reached aTON of 14 based on the total amount of catalyst used. [13] From electrochemical measurements it was proposed that the Mn 0 -Mn 0 dimer forms in the polymer matrix, although this was not spectroscopically verified. Preliminary studies of an electro-polymerized pyrrole-based Mn catalyst deposited on silicon nanowires have also suggested photoelectrochemical (PEC) CO 2 reduction, based on cyclic voltammetry (CV) results. [14] Herein, we present an ovel Mn I CO 2 reduction electrocatalyst with ap hosphonate functionality (MnP,S cheme 1) that allows anchoring and direct wiring between the catalytic center and ametal oxide surface, [15] as has been achieved for an analogous phosphonate-modified Re complex. [16] We employ am esoporous TiO 2 electrode,b ecause it offers 1) long-term stability and conductivity under reducing conditions, [17] 2) at hree-dimensional morphology for high cata-lyst loading and to facilitate close inter-molecular interactions,and 3) transparencyfor spectroelectrochemical characterization of catalytic intermediates. [18] Thee lectrochemical investigations establish the heterogenized MnP as the bestperforming Mn electrocatalyst to date,which was enabled by ad ynamic TiO 2 j MnP interface and dimerization of the immobilized Mn catalyst. Finally,wepresent the first example of CO 2 reduction by aMncatalyst driven by full UV/Vis solarspectrum irradiation, circumventing the typical photo-instability [13b,19] of these compounds by combining the TiO 2 j MnP hybrid cathode in the dark with aCdS-sensitized photoanode.
MnP (Scheme 1) was synthesized by coordination of 4,4'bis(phosphonic acid)-2,2'-bipyridine to pentacarbonyl manganese(I) bromide in ethanol under N 2 while protected from light. Thep roduct was isolated as an orange solid in 63 %y ield and characterized by CHNP microanalysis, 31 P-NMR spectroscopy,h igh-resolution mass spectrometry,a nd infrared (IR) spectroscopy (n CO = 2030CO = , 1946CO = , and 1930 cm À1 , Figure 1a), which confirmed a fac-Mn tricarbonyl species. [19] Full synthetic and characterization details can be found in the Supporting Information. MnP was insoluble in CH 3 CN and therefore characterized by CV in DMF ( Figure S1 in the Supporting Information). Ac atalytic wave at E onset = À1.8 V versus Fc + /Fc (Fc = [(h-C 5 H 5 ) 2 Fe]) was observed when H 2 O was added and the cell was purged with CO 2 .The presence of water in the electrolyte solution is known to significantly increase electrocatalytic CO 2 reduction activity,b ya llowing the Mn-Mn dimer to directly react with CO 2 . [10a] Mesoporous TiO 2 electrodes were prepared by ad octorblading procedure,a pplying as uspension of commercial P25 TiO 2 nanoparticles (anatase/rutile (8/2) mixture,average particle size 21 nm) to af luorine-doped tin oxide (FTO) coated glass electrode,a nd further experimental details can be found in the Supporting Information. Scanning electron microscopy (SEM) on the resultant electrode revealed am esoporous film with at hickness of approximately 6 mm ( Figure S2 a). Loading of the catalyst onto the TiO 2 electrode was achieved by drop-casting am ethanol solution of MnP, resulting in 34 nmol Mn per cm 2 of geometrical surface area. Thepresence of IR bands at n CO = 2032 and 1928 cm À1 in the attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectrum confirmed the presence of MnP on the electrode (TiO 2 j MnP;F igure 1a). Immobilization and electronic communication of the MnP with ametal oxide was confirmed by adsorbing MnP on conducting and mesoporous tin-doped indium oxide (ITO) electrodes instead (film thickness approximately 7 mm, see Figure S2 ba nd Supporting Information for experimental details). CV with ITO j MnP in anhydrous CH 3 CN (1.0 m Bu 4 NBF 4 )d isplayed ar eversible wave at E = À1.6 Vv ersus Fc + /Fc, assigned to the reduction of Mn I to Mn 0 . Thep eak current was linearly dependent on the scan rate,i ndicative of an immobilized species in good electronic communication with the electrode ( Figure S3).
TiO 2 becomes conductive at potentials more negative than the conduction band (CB), thus the CV of TiO 2 j MnP can be employed to study electrocatalytic CO 2 reduction. The CV scan of ab are (Mn-free) TiO 2 electrode in CH 3 CN/H 2 O (19/1, 0.1m Bu 4 NBF 4 )s hows the filling and emptying of the conduction band of TiO 2 (Figure 1b), as confirmed by the increase in absorbance in the l = 600-850 nm region of the electronic spectrum at an applied potential, E appl ,o fÀ1.8 V versus Fc + /Fc ( Figure S4). [17b,20] Comparable CV features are observed with abare TiO 2 electrode under CO 2 or TiO 2 j MnP under N 2 .H owever,T iO 2 j MnP purged with CO 2 showed an increased current with an onset of E = À1.6 Vversus Fc + /Fc, indicative of electrocatalytic CO 2 reduction by the heterogenized MnP catalyst ( Figure 1b). Furthermore,t he ratio of cathodic to anodic charge in the forward and reverse CV scans increased from approximately 1:1t o4 :1 by changing TiO 2 to TiO 2 j MnP under CO 2 ,s uggesting that conductionband electrons of TiO 2 are consumed by the Mn catalyst on the CV timescale and are therefore unavailable for discharging during the anodic scan.
Theincreased current arising from TiO 2 j MnP under CO 2 was confirmed as being the result of the reduction of CO 2 to CO by controlled-potential electrolysis (CPE). Figure 2a shows the gaseous products formed when TiO 2 j MnP electrodes were held at E appl = À1.7 Vv ersus Fc + /Fc in the dark under CO 2 ,a nd monitored by gas chromatography (GC). After 2h,a na verage of 1.10 AE 0.25 Cw as passed, with the production of 3.75 AE 0.56 mmol CO,c orresponding to aF aradaic efficiency( FE) of 67 AE 5%.T he FE for H 2 production was 12.4 AE 1.4 %, and the formation of formate was not detectable by ion chromatography.T he TON CO of 112 AE 17 was calculated based on the amount of MnP drop-cast onto the electrode,a nd is thus al ower limit since it assumes all MnP remains bound and active throughout CPE. This is the highest TON CO based on the total amount of catalyst used for aM nc atalyst in CO production, and was achieved at al ow overpotential (h)o fa pproximately 0.42 V, calculated using astandard potential for CO 2 reduction to CO (E 0' (CO 2 /CO)) of À1.28 Vversus Fc + /Fc in these conditions. [21] This is one of the lowest overpotentials observed for at ransition-metalbased catalyst in non-aqueous solution, [1a, 2a, 22] matched only by am odified Fe-porphyrin in homogeneous DMF solution (h = 0.41 V) [21a] and aMncatalyst that achieved aTON co of 36 after 6h(h = 0.35 V). [12] Figure 1. a) Solution FTIR of MnP and ex situ ATR-FTIR spectra of TiO 2 j MnP before and after controlled-potential electrolysis(CPE) for 20 min at E appl = À1.7 Vversus Fc + /Fc. b) CV scans of TiO 2 and TiO 2 j MnP (geometrical surface area = 1.0 cm 2 )u nder N 2 and CO 2 . Conditions:C H 3 CN/H 2 O(19/1), 0.1 m Bu 4 NBF 4 , n = 100 mVs À1 ,A g/ AgCl reference electrode (RE), Pt counter electrode (CE), room temperature.

Communications
TiO 2 j MnP exhibited good CO selectivity,w ith aC O:H 2 ratio of approximately 12:1 after 1hCPE, although this ratio was reduced to 5.4:1 after 2h,p resumably ar esult of desorption or degradation of the Mn catalyst during the second hour of electrolysis.I nt he absence of either CO 2 or the Mn catalyst ( Figures S5 aand S5 b), no CO was produced. H 2 production by bare TiO 2 was 1.91 AE 0.31 mmol after 2h, compared to 1.43 AE 0.22 mmol for TiO 2 j MnP with as urface coverage of 22 nmol cm À2 and 0.69 AE 0.08 mmol with ac overage of 34 nmol cm À2 (see Figure 2a, Figure S5, and Table S1). Increasing amounts of MnP on TiO 2 therefore suppress H 2 in favor of CO production, suggesting that H 2 production by TiO 2 j MnP may originate from unmodified areas of the TiO 2 rather than the catalyst itself.
IR and UV/Vis spectroscopies confirmed the molecular nature of MnP during catalysis on TiO 2 .F igure 1a shows an ATR-FTIR spectrum of TiO 2 j MnP taken after CPE for 20 min (Q = 0.37 C, approximate TON CO = 34), revealing peaks at n CO = 2042 and 1943 cm À1 .T hese vibrational CO stretches closely match the spectrum of the as-prepared electrode,w ith as light shift explained by exchange of coordinated Br À for as olvent molecule,a nd therefore demonstrate that the molecular structure of the catalyst remains largely unchanged during catalytic turnover. Deactivation of the Mn catalyst to am aterial that is no longer molecular would be unlikely to give high CO selectivity, corroborating Figure 2a.
TheU V/Vis spectra of TiO 2 j MnP before,d uring,a nd after 20 min CPE with E appl = À1.7 Vversus Fc + /Fc are shown in Figure 2b.D uring CPE, bands at 630 and 820 nm were observed, which are assigned to the formation of an Mn-Mn dimer by comparison to similar peaks formed during homogeneous CPE of the unmodified [MnBr(bpy)(CO) 3 ] (Table S2 for assignment). [1a, 2b] We excluded the formation of the mononuclear doubly reduced MnP anion, analogues of which are also known to reduce CO 2 when dimer formation is impeded, [1b,2c] due to the lack of as trong peak at approximately 548 nm as found in an analogous Mn compound in THF [3a] (difference spectrum in Figure S6). After CPE for 20 min, the TiO 2 j MnP was left under CO 2 without an applied potential, and the peaks resulting from the dimer were lost (Figure 2b). This was corroborated by the IR spectrum in Figure 1a,which indicated mainly the presence of the Mn I monomer,b ut with as mall peak at 1865 cm À1 and ab roadening of the peak at 1943 cm À1 ,a ssigned to as mall amount of remaining dimer. [3b] These data are consistent with the mechanism shown in Scheme 1, with the formation of as teady-state concentration of the catalytically active Mn-Mn dimer.T his intermediate then reacts with CO 2 before it can be identified ex situ, reforming the Mn I monomer as detected in the IR spectrum.
Immobilization of MnP on mesoporous TiO 2 creates ah igh local concentration of Mn 0 under reducing conditions at the electrode surface.Phosphonic acid modified molecules, such as MnP,display some lability when bound to TiO 2 , [23] and phosphate buffer has been used to displace anchored catalysts from TiO 2 particles,d emonstrating ad ynamic interaction. [24] We propose that the high activity and low overpotential of this system is due to either temporary desorption of the catalyst, followed by dimerization and re-anchoring within mesoporous TiO 2 ,o rt he high local concentration of MnP placing the metal centers in an environment where they are predisposed to dimerization upon reduction.
Manganese carbonyl compounds,s uch as MnP,s how instability under illumination, [19] and tend to undergo photolysis and release CO ligands. [25] Consequently,the few reports of Mn-based CO 2 reduction photocatalysis use monochromatic or narrowly filtered light to prevent decomposition of the catalyst. [14, 25a, 26] This photo-instability was observed for TiO 2 j MnP,which displayed asignificantly lower CO production of 0.39 AE 0.16 mmol (12 AE 3% FE) when CPE was performed under UV-filtered 1sun illumination (l > 420 nm to avoid TiO 2 band-gap excitation in this experiment) at À1.7 Vv ersus Fc + /Fc for 2h ( Figure S7). Thes ignificant H 2 production (1.74 AE 0.6 mmol, 59 AE 8% FE) is consistent with degradation of MnP and possibly the formation of acatalytically active Mn deposit. Therefore,TiO 2 j MnP cannot be used directly in aC O 2 reducing photocathode that efficiently absorbs sunlight and exposes the catalyst to irradiation.
An alternative strategy to drive CO 2 reduction using full solar-spectrum irradiation was implemented, integrating MnP into aphotoelectrochemical circuit with aphotoanode,wired to TiO 2 j MnP,w hich was kept in the dark. CdS-sensitized ZnO nanosheet electrodes were prepared following ar eported procedure (SEM in Figure S8 a), [27] which absorb ab road spectrum of light below 530 nm according to the electronic spectrum shown in Figure S8b.T hese ZnO j CdS electrodes gave an anodic photocurrent in the presence of triethanolamine (TEOA) as ahole scavenger with an onset of À1.65 Vversus Fc + /Fc,apotential at which TiO 2 j MnP gives ac athodic current from CO 2 reduction (Figure 3a). The linear-sweep voltammetry (LSV) scan of at wo-electrode, two-compartment PEC cell comprising aC dS j ZnO photoanode and aT iO 2 j MnP cathode (kept in the dark) in Figure 3b shows as mall photocurrent at zero bias,w hich

Angewandte Chemie
Communications increased as ab ias potential (U appl )w as applied. To confirm that CO was produced, we performed CPE in atwo-electrode configuration in CH 3 CN/H 2 Oelectrolyte solution (19/1, 0.1m Bu 4 NBF 4 ,0 .1m TEOA, purged with CO 2 ). An applied potential of 0.6 Vf or 1h passed ac harge of 0.26 C, and 0.36 AE 0.07 mmol of CO (26 %F E, 2.6:1 CO:H 2 ratio, TON CO = 11, Figure S9) was measured. Thel ower CO production performance compared to the three-electrode electrocatalytic system could be due to the potentially disruptive presence of TEOAi nt he electrolyte solution, the lower charge passed and the different potential at the cathode.N evertheless,t his is the first example of full spectrum solar-light driven CO 2 reduction with aMncatalyst.
In conclusion, we have presented MnP as an ovel Mnbased CO 2 reduction catalyst that allows immobilization onto am esoporous TiO 2 electrode with its phosphonic acid anchoring groups.T he TiO 2 j MnP cathode achieved efficient CO 2 reduction to CO,r eaching an unprecedented TON CO of 112 AE 17 at an overpotential of 0.42 Vafter 2hCPE. During electrocatalytic CO 2 reduction, aMn-Mn dimer was formed, which is an important catalytic intermediate in homogeneous solution. This is,toour knowledge,the first observation of the dynamic formation of active catalytic dimers on as urface, providing as trategy for retaining homogeneous reaction mechanisms whilst also gaining the advantages of heterogeneous catalysis.Finally,weutilized the CO 2 reduction activity of TiO 2 j MnP at al ow overpotential to assemble aP EC cell with aC dS-sensitized photoanode,d emonstrating that Mn catalysts can be used in solar-driven CO 2 reduction in spite of their photo-instability.T his work represents an advance in moving molecular CO 2 reduction electrocatalysis towards af ull artificial photosynthetic system. This was achieved through the immobilization of the catalyst, attainment of ah igh TONa tl ow overpotential, and implementation of aPEC cell.