Electrocatalytic Co 2 Reduction with a Membrane Supported Manganese Catalyst in Aqueous Solution †

[Mn(bpy)(CO) 3 Br] cast in a Nafion membrane is an active heterogeneous electrocatalyst with good selectivity for CO 2 reduction to CO in neutral aqueous electrolyte. Addition of multi-walled carbon nanotubes (MWCNT) leads to a B10 fold current enhancement and stable CO : H 2 yields (1 : 2) at À1.4 V vs. Ag/AgCl at pH 7. The electrochemical reduction of CO 2 to products such as methanol, formic acid and CO offers an environmentally benign route to the production of high value fuels from a waste feedstock. The production of syngas (CO : H 2 B 0.5) from CO 2 and water is of particular interest as it can be used to make a wide range of hydrocarbon products through Fischer–Tropsch chemistry. As a result of this, significant efforts towards the development of both homogenous and supported (heterogeneous) metal complexes for CO 2 electro-catalysis are underway. One of the most commonly studied homogenous CO 2 catalysts is [Re I (bpy)(CO) 3 X] (bpy = 2,2 0-bipyridine and X = Cl À , Br À or solvent) which has received extensive interest both as a photocatalyst and electrocatalyst due to its activity and selectivity for CO 2 reduction, often to CO. 4–6 A limitation of [Re(bpy)(CO) 3 X] is the high cost of the metal centre and the relatively high overpotential for CO production. 4 Several recent reports have demonstrated that [Mn

[Mn(bpy)(CO) 3 Br] was synthesized according to literature methods. 1 The synthesis was conducted completely in the absence of light by covering all reflux apparatus in aluminium Electronic Supplementary Material (ESI) for ChemComm.This journal is © The Royal Society of Chemistry 2014 foil prior to addition of reagents and under a N 2 atmosphere.Mn(CO) 5 Br (200 mg, 0.72 mmol) and 2,2'-bipyridine (115 mg, 0.74 mmol) were dissolved in 30 ml diethyl ether and refluxed for 3 hours at 40°C.The mixture was left to cool overnight, resulting in the formation of an orange powder precipitate.The product was removed from the mother liquor by cooling in an acetone-dry ice bath for 2 hours.The orange powder was filtered off and rinsed with cold diethyl ether until the solvent ran colourless.The product was left on a vacuum pump for 48 hours to dry.Yield = 61%;

Film preparation:
The membrane was prepared from [Mn(bpy(CO) 3 Br] (typically 5.5 mg) dissolved in 0.5 ml of acetonitrile, mixed with 0.5 ml a Nafion/alcohol solution (5.0 % w/w) in a 1:1 volume ratio in the dark.In a typical experiment 10 μl (corresponding to 140 nmol of [Mn(bpy(CO) 3 Br] in 2.5 % Nafion/MeCN 1:1) of this yellow solution was transferred onto a polished glassy carbon electrode using an autopipette and left to dry in air in the dark for several hours, yielding a pale yellow polymer film that completely coated the working area.Prolonged CPE experiments (Fig. S5) used only 5 μl of the casting solution, corresponding to 70 nmol of catalyst deposited.Casting solutions for MWCNT/Nafion/[Mn(bpy)(CO) 3 Br] films were prepared as above with the addition of MWCNT (5.5 mg) to the casting solution, that was then sonicated in the dark for 15 minutes.5 μl of this darkly coloured casting solution was then transferred onto a polished glassy carbon electrode using an autopipette and left to dry in air in the dark for several hours 1.3 Apparatus: All electrochemistry was performed with a PalmSens 3 potentiostat.Gas chromatography was performed using an Agilent 6890N employing N6 helium as the carrier gas (5 ml.min -1 ).A 5 Å molecular sieve column (ValcoPLOT, 30 m length, 0.53 mm ID) and a pulsed discharge detector (D-3-I-HP, Valco Vici) were employed.CO and H 2 peak areas were quantified with multiple calibrant gas injections and were re-calibrated daily.NMR spectra were recorded on a Bruker Advance 400 NMR spectrometer with CD 3 CN as the solvent and an amber quartz NMR tube.ESI MS and elemental analyses were performed by the University of Liverpool analytical services.FTIR spectra were recorded using a JASCO 4000 FTIR.Scanning electron microscopy of films cast on glass slides was recorded using a Jeol 6610 SEM operating at an accelerating voltage of 20 kV.Profilometry was measured using an Ambios Technology XP200.

Electrocatalysis:
The electrolyte was a phosphate buffer (30 mM Na 2 HPO 4 + 30 mM NaH 2 PO 4 in Milli-Q water).The electrolyte was pH = 7.0 when air saturated and pH = 6.2 when CO 2 saturated.Cyclic voltammograms of the catalyst films were performed in a pearshaped flask with a glassy carbon disc working electrode (BASi, geometric surface area = 0.0717 cm 2 ), a Pt basket counter electrode and a Ag/AgCl (3 M NaCl) reference electrode (BASi).All CV and CPE experiments were performed in the dark as the Mn complex is photosensitive.The flask was purged with either argon or CO 2 for 20 minutes prior to scanning.Controlled potential electrolysis (CPE) was performed in a custom made pyrex Hcell with the two compartments separated by a fine glass frit.The same electrodes as above were used but with an extra double junction (porous glass frit) covering the Pt counter electrode to minimize re-oxidation of CO or other reaction products.Both compartments were magnetically stirred throughout the CPE reaction.Results presented are based on an average of at least two independent measurements, with typical variability of ~ ± 10 % in product yields with ~3.5 x 10 -10 mol of electroactive catalyst.It was noted that in the selectivity and activity was highly dependent upon the coverage of the GCE by the Nafion film and the electroactive concentration, making it important to ensure that the entire GCE surface is covered and that no cracks in the membrane occur upon drying.The faradaic efficiencies (FE) achieved were calculated by taking the measured product yields (GC) and charge passed (Q) and accounting for the requirement of 2 electrons to produce one CO molecule. ] The overall FE for all gas phase products measured (H 2 + CO) was 75 % and 69 % in the - The highest concentration of a liquid phase CO 2 reduction product would be via a 2 electron pathway (e.g.HCO 2 H formation). 15.75 mC would lead to a maximum product yield of 82 nmol.The total electrolyte volume of the cell is 23 ml, leading to a maximum concentration of ~3 M for liquid phase CO 2 reduction products.), 4 suggesting an electron density at the Mn centre in the presence of MWCNT that is significantly higher than in the absence of MWCNT, but still well below that present in the doubly reduced complex.A linear dependence of peak current for the reduction of [Mn(bpy)(CO) 3 X] n+ at -1.15 V to the scan rate is measured indicating that a surface immobilised redox process is occurring, in line with expectations for a membrane supported electro-active species.Using the relationship 5 :

Catalyst and Membrane
Where n is the number of electrons transferred, F is Faraday's constant,  is the scan rate in V s -1 , A the electrode surface area (cm 2 ),  the surface coverage (mol cm -2 ), R the gas constant (J mol -1 K -1 ) and T the temperature (K), it was possible to calculate a typical surface electroactive coverage of catalyst of ~3.5 x10 -10 mol in the experiments where 140 nmol of catalyst were dissolved in the casting solution.The concentration of electroactive material in the MWCNT sample was assessed through analysis of the dimer re-oxidation, at ca. -0.5 V as the 2 successive reductions under argon are found to be heavily overlapped, using the relationship between surface coverage and charge: The measured electroactive concentration, the confirmation of surface confined electrochemistry and the film morphology studies are of relevance when considering the mechanism of dimer formation within a Nafion film.Several mechanisms can be proposed (1)   diffusion of the Mn species through the Nafion channels, (2) diffusion of the Mn complex into solution following leaching and (3) the presence of closely packed Mn complexes in localized clusters that do not require diffusion for dimerization.We favour the 3rd hypothesis for the following reasons: (i) Both ( 1) and ( 2) require the movement of electroactive Mn(bpy)(CO) 3 X] n .The linear scan rate dependence of the peak current for both the reduction of [Mn(bpy)(CO) 3 X] n (Fig. S8) and the re-oxidation of the dimer (Fig. S9) indicate that a surface immobilised redox process is occurring this allows us to rule out rapid free diffusion (on the CV timescales studied) of [Mn(bpy)(CO) 3 X] through either the Nafion channels (~2.4 nm) 6 or solution.
(ii) Figures S14-S17 clearly show that no [Mn(bpy)(CO) 3 X] n species are found in the post electrolysis electrolyte, in-line with expectations for a catalyst insoluble in aqueous solutions, ruling out the mechanism ( 2) and the possibility of the Mn complexes being dissolved, transported or leached back out of the membrane.
(iii) The formation of clusters of [Mn(bpy)(CO) 3 X] n , required for the 3rd mechanism, is probable given that the when the volatile organic solvents (CH 3 CN, propanol) are removed through drying during film preparation, the [Mn(bpy)(CO) 3 Br] will precipitate out leading to small deposits within the Nafion membrane.Catalyst aggregates have also been previously reported for a related [Re(bpy)(CO) 3 Br] system. 7Optical images (Fig. S10(b)) shows that yellow clusters appear to be localised at the film edges, and profilometric data (Fig. S9) shows enhanced roughness upon the inclusion of [Mn(bpy)(CO) 3 Br].SEM images (Fig. S11) of the MWCNT/[Mn(bpy)(CO) 3 Br] membrane also show the presence of large clusters at the film edge which may be either MWCNT bundles or the [Mn(bpy)(CO) 3 X] n+ .
(iv) In the absence of fast catalyst diffusion (mechanism (3)) we would anticipate that only Mn clusters close to the GCE interface within the Nafion in the absence of MWCNT would be electrochemically active, in line with our reported very low measured concentrations of electrochemically active catalyst (Figs.S7, S8).This is due to Nafion being a poor conductor of electrons.A mobile catalyst within the Nafion would be expected to lead to higher concentrations of electrochemically active species.The inclusion of MWCNT effectively increases the electrode surface area making it more probable that a Mn cluster is close to the electro-active surface leading to higher electroactive concentrations and current densities.

Catalyst benchmarking:
The activity of the Nafion/[Mn(bpy)(CO) 3 Br] catalysts reported can be directly compared to those previously reported for the related Nafion/[Re(bpy)(CO) 3 Br] system once normalised for surface area of GCE, Table 1 main text. 7Compared to the rhenium analogue we find higher TON and improved selectivities are achieved with Mn at -1.4 to -1.5 V vs Ag/AgCl in addition to the presence of catalytic activity at lower applied potentials.Direct benchmarking between the activity of the homogenous [Mn(bpy)(CO) 3 Br] system 1 and the heterogeneous Nafion immobilised catalyst by comparison of overall TON in a defined time period is however not appropriate.It has recently been highlighted by Savéant and co-workers 8,9 that a fair activity comparison between heterogeneous and homogenous systems requires the consideration of only the catalyst concentration present in a reaction layer close to the electrode surface in the homogenous experiment.If this is not done the TOF reported for the homogenous systems will be greatly disadvantaged due to the presence of the bulk catalyst and furthermore TOF obtained will be highly dependent upon the volume-surface ratio of the cell used.For example it was reported that bulk electrolysis over 22 hours of [Mn(bpy)(CO) 3 Br] (2.5 x10 -5 mol of catalyst) at ca.-1.35 V vs Ag/AgCl led to Q = 130 C being passed with a FE of 85% for CO, 1 which we calculate to be a TON ~ 23, giving a bulk TOF ~ 0.0003 s -1 which is well below the real TOF = 0.45 s -1 (log TOF = -0.35s -1 at  = 0.51 V) calculated by Savéant et al. using a foot-of-the-wave analysis of CV data. 9gure S10 shows the TOF for our heterogeneous system calculated from the CPE data reported in the main text.As can be seen from figure S10, we report significantly lower TOF for the Nafion immobilised system compared to the homogenous solution at a similar overpotential ( ~ 0.6 V).As noted in the main text the diffusion of CO 2 is likely to be a major limiting factor in these polymer immobilised materials.It is also highly desirable to be able to compare the extrapolated TOF at  = 0 V between two systems; however the TOF 0 in Nafion (log TOF 0 ~ -3.0 s -1 cf.-8.7 s -1 for [Mn(bpy)(CO) 3 Br] in CH 3 CN + 5% H 2 O) is believed to be artificially high as unlike the foot-of the-wave method we are unable to account for how the concentration of CO 2 within Nafion varies with potential and time during our CPE measurement.

Fig
Fig. S3: UV/Vis spectra of [Mn(bpy)(CO) 3 Br] in MeCN (blue) or MeCN/Nafion solution (green, red).Spectra show that ligand exchange in the dark appears to be promoted in the presence of Nafion.

Fig. S13 :
Fig. S13: Amperometric detection (black) of a [Mn(bpy)(CO) 3 Br] (140 nmol)/Nafion film in 30 mM Na 2 HPO 4 + 30 mM NaH 2 PO 4 buffer at pH ~ 7.0 held at -1.5 V vs Ag/AgCl.The experiment is initially carried out under an Argon atmosphere before CO 2 is briefly introduced (10 s purge).GC headspace analysis (red) of the electrochemical cell at the reaction start, immediately prior to CO 2 introduction and at the reaction end clearly showing a large increase in the rate of CO production upon introduction of the CO 2 .

Fig. S14 :
Fig. S14: CV of post CPE electrolyte (from [Mn(bpy)(CO) 3 Br]/Nafion membrane held at -1.5 V for four hours) with a clean glassy carbon electrode under Argon and CO 2 , confirming that no electroactive materials are leached out of the membrane during experiments.

Fig. S15 :
Fig. S15: CV of post CPE electrolyte (from [Mn(bpy)(CO) 3 Br]/MWCNT/Nafion membrane held at -1.25 V for four hours) with a clean glassy carbon electrode under Argon and CO 2 , confirming that no electroactive materials are leached out of the membrane during experiments.

Fig. S21 .
Fig. S21.The overall TON for CO is ca. 5 (70 nmol of [Mn(bpy)(CO) 3 Br] deposited onto a GCE (0.07 cm 2 ) in a 5 l solution (1:1 Nafion (2.5%wt):CH 3 CN)), during CPE at -1.5 V vs. Ag/AgCl in a CO 2 purged 60 mM phosphate buffer (pH~7).The TON is well in excess of that achievable by catalyst decomposition.The sample was run for a period of 880 minutes, then briefly repurged with CO 2 (grey dashed line) prior to the experiment resuming.a TON are based on the total amount of [Mn(bpy)(CO) 3 Br] deposited and not the electroactive content of the film.

Fig. S23 :
Fig. S23: Measured dependence of TOF (from preparative scale electrolysis) with overpotential for Nafion/[Mn(bpy)(CO) 3 Br] in pH 7 electrolyte.The experimental data presented is likely to be strongly affected by variations in C CO2 within the Nafion membrane.