CO2 to CO Electroreduction, Electrocatalytic H2 Evolution, and Catalytic Degradation of Organic Dyes Using a Co(II) meso-Tetraarylporphyrin

The meso-tetrakis(4-(trifluoromethyl)phenyl)porphyrinato cobalt(II) complex [Co(TMFPP)] was synthesised in 93% yield. The compound was studied by 1H NMR, UV-visible absorption, and photoluminescence spectroscopy. The optical band gap Eg was calculated to 2.15 eV using the Tauc plot method and a semiconducting character is suggested. Cyclic voltammetry showed two fully reversible reduction waves at E1/2 = −0.91 V and E1/2 = −2.05 V vs. SCE and reversible oxidations at 0.30 V and 0.98 V representing both metal-centred (Co(0)/Co(I)/Co(II)/Co(III)) and porphyrin-centred (Por2−/Por−) processes. [Co(TMFPP)] is a very active catalyst for the electrochemical formation of H2 from DMF/acetic acid, with a Faradaic Efficiency (FE) of 85%, and also catalysed the reduction of CO2 to CO with a FE of 90%. Moreover, the two triarylmethane dyes crystal violet and malachite green were decomposed using H2O2 and [Co(TMFPP)] as catalyst with an efficiency of more than 85% in one batch.

The potential of the CO 2 reduction is strongly connected to the presence of protons [7,40]. However, fine-tuning of the reduction potentials of Co porphyrins is easily possible

Photophysical Properties
The UV-vis absorption spectrum of [Co(TMFPP)] in CH2Cl2 solution showed a Soret band at 437 nm and a Q band at 554 nm in line with similar Co(II) porphyrins ( Figure 1A) [51,53,[79][80][81][82]. The optical band gap (Eg), which is the energy difference between the HOMO and LUMO levels, was determined from the UV-vis absorption spectrum. The values of the optical band gap (Eg) of [Co(TMFPP)] were determined using the Tauc relation ( Figure 1B). The Eg value was 2.15 eV, which is in the normal range for Co metalloporphyrins [53,80].

Photophysical Properties
The UV-vis absorption spectrum of [Co(TMFPP)] in CH2Cl2 solution showed a Soret band at 437 nm and a Q band at 554 nm in line with similar Co(II) porphyrins ( Figure 1A) [51,53,[79][80][81][82]. The optical band gap (Eg), which is the energy difference between the HOMO and LUMO levels, was determined from the UV-vis absorption spectrum. The values of the optical band gap (Eg) of [Co(TMFPP)] were determined using the Tauc relation ( Figure 1B). The Eg value was 2.15 eV, which is in the normal range for Co metalloporphyrins [53,80].

Photophysical Properties
The UV-vis absorption spectrum of [Co(TMFPP)] in CH 2 Cl 2 solution showed a Soret band at 437 nm and a Q band at 554 nm in line with similar Co(II) porphyrins ( Figure 1A) [51,53,[79][80][81][82]. The optical band gap (Eg), which is the energy difference between the HOMO and LUMO levels, was determined from the UV-vis absorption spectrum. The values of the optical band gap (Eg) of [Co(TMFPP)] were determined using the Tauc relation ( Figure 1B). The Eg value was 2.15 eV, which is in the normal range for Co metalloporphyrins [53,80].
A first one-electron oxidation at 0.30 V that can be assigned to the Co(II)/Co(III) redox couple is broadened, but reversible. The broadening is due to the coordination of DMF after oxidation, as mentioned above. This wave is followed by a slightly larger reversible wave at 0.98 V, which is assigned to a porphyrin-centred process (Por 2− /Por 1− ) in line with previous reports [21,24,61,62,[85][86][87]. For the 4-MeO substituted derivative, a third oxidation at 1.09 V following the second oxidation at 0.92 V was reported [24,62], and for the 2,5-MeO substituted complex, second and third oxidation waves were observed at 0.62 and 1.15 V [73]. This lets us assume that, for [Co(TMFPP)], both these two porphyrin-centred processes are merged into one (larger) wave.

Electrochemical Characterisation
Cyclic voltammograms of [Co(TMFPP)] were recorded in DMF, which is a potential donor ligand and is thus prone to coordinate to the Co centre after oxidation, as has been found for most square planar coordinated M(II) porphyrins [58,59]. Two reversible one-electron reductions were found for [Co(TMFPP)] at E 1/2 = −0.91 V and E 1/2 = −2.05 V ( Figure 3). While it is generally accepted to assign the first wave to the Co(II)/Co(I) redox couple [24,61,62,86,87], the second was earlier discussed as porphyrincentred (Por 2− /Por 3− ), in line with reports on the unsubstituted [Co(TPP)] [24,61,62,86]. Alternatively, a Co(0) species after a Co(I)/Co(0) reduction has been discussed [20,88,89]. The latter description is supported by UV-vis absorption spectroscopy [89].  For [Co(TPP)] in DMF potentials of −1.88, −0.77, 0.30, and 1.05 V were previously reported for the same processes [24], while the 4-MeO substituted derivative showed −0.98 V, 0.38 V, and 0.92 V for the first reduction and the two oxidations. This means that the introduction of the four CF3 groups does not markedly affect the metal-centred first oxidation and reduction.
For a fully homogeneous diffusion-controlled electrochemical process, the peak current (Ip) for a Faradaic electron transfer varies linearly with the square root of scan rate (ν 1/2 ). From the slope of the Ip vs. ν 1/2 plot, the diffusion coefficient (D) can be determined using Randles-Sevcik equation (Equation (1)): A first one-electron oxidation at 0.30 V that can be assigned to the Co(II)/Co(III) redox couple is broadened, but reversible. The broadening is due to the coordination of DMF after oxidation, as mentioned above. This wave is followed by a slightly larger reversible wave at 0.98 V, which is assigned to a porphyrin-centred process (Por 2− /Por 1− ) in line with previous reports [21,24,61,62,[85][86][87]. For the 4-MeO substituted derivative, a third oxidation at 1.09 V following the second oxidation at 0.92 V was reported [24,62], and for the 2,5-MeO substituted complex, second and third oxidation waves were observed at 0.62 and 1.15 V [73]. This lets us assume that, for [Co(TMFPP)], both these two porphyrin-centred processes are merged into one (larger) wave.
For [Co(TPP)] in DMF potentials of −1.88, −0.77, 0.30, and 1.05 V were previously reported for the same processes [24], while the 4-MeO substituted derivative showed −0.98 V, 0.38 V, and 0.92 V for the first reduction and the two oxidations. This means that the introduction of the four CF 3 groups does not markedly affect the metal-centred first oxidation and reduction.
For a fully homogeneous diffusion-controlled electrochemical process, the peak current (I p ) for a Faradaic electron transfer varies linearly with the square root of scan rate (ν 1/2 ). From the slope of the I p vs. ν 1/2 plot, the diffusion coefficient (D) can be determined using Randles-Sevcik equation (Equation (1)): where I p is the peak current, F is the Faraday constant (F = 96485 C mol −1 ), R is the universal gas constant (R = 8.314 J K −1 mol −1 ), T = 298 K, n p is the number of electrons transferred (here, n p = 1), A is the active surface area of the electrode (0.00785 cm 2 ). Note that our plots are reported as a function of the current density, bypassing the need of the area value in Equation (1). D is the diffusion coefficient for the complex, [C 0 ] is the concentration of the catalyst (here [C 0 ] = 1 mM), and ν is the scan rate in V s −1 . The diffusion coefficient (D) was calculated from the slope of I p vs. ν 1/2 (Figure 4, right). The diffusion coefficient D for the Co(II)/Co(I) reduction is 1.98 × 10 −7 cm S −1 , while the value for the Co(II)/Co(III) oxidation is slightly larger with 9.5 × 10 −7 cm S −1 , in keeping with the assumed additional DMF ligand for the oxidised complex [Co(TMFPP)(DMF)] + . The diffusion coefficient D for the second reduction process with 1.1 × 10 −8 cm S −1 is smaller than the value for the first reduction, which is due to the increased negative charge, but still quite large.  For [Co(TPP)] in DMF potentials of −1.88, −0.77, 0.30, and 1.05 V were previously reported for the same processes [24], while the 4-MeO substituted derivative showed −0.98 V, 0.38 V, and 0.92 V for the first reduction and the two oxidations. This means that the introduction of the four CF3 groups does not markedly affect the metal-centred first oxidation and reduction.
For a fully homogeneous diffusion-controlled electrochemical process, the peak current (Ip) for a Faradaic electron transfer varies linearly with the square root of scan rate (ν 1/2 ). From the slope of the Ip vs. ν 1/2 plot, the diffusion coefficient (D) can be determined using Randles-Sevcik equation (Equation (1)): where Ip is the peak current, F is the Faraday constant (F = 96485 C mol −1 ), R is the universal gas constant (R = 8.314 J K −1 mol −1 ), T = 298 K, np is the number of electrons transferred (here, np = 1), A is the active surface area of the electrode (0.00785 cm 2 ). Note that our plots are reported as a function of the current density, bypassing the need of the area value in Equation (1). D is the diffusion coefficient for the complex, [C0] is the concentration of the catalyst (here [C0] = 1 mM), and ν is the scan rate in V s −1 . The diffusion coefficient (D) was calculated from the slope of Ip vs. ν 1/2 (Figure 4, right). The diffusion coefficient D for the Co(II)/Co(I) reduction is 1.98 × 10 −7 cm S −1 , while the value for the Co(II)/Co(III) oxidation is slightly larger with 9.5 × 10 −7 cm S −1 , in keeping with the assumed additional DMF ligand for the oxidised complex [Co(TMFPP)(DMF)] + . The diffusion coefficient D for the second reduction process with 1.1 × 10 −8 cm S −1 is smaller than the value for the first reduction, which is due to the increased negative charge, but still quite large.

Electrocatalytic H 2 Production in the Presence of Acetic Acid (AcOH)
We studied the electrocatalytic activity of [Co(TMFPP)] in DMF/acetic acid due to better solubility in this mixture than in water ( Figure 5).
On first view, our cyclic voltammetric plots show that, upon addition of acid, a catalytic current appears at the second reduction wave of [Co
Mechanistically speaking, it seems that the first reduced species which we formally describe as [Co(I)(Por 2− ] − is not catalytically very active, which would stand in contrast to previous mechanistic studies [18,70] which proposed that the reduced Co(I) species reacts with protons forming Co(II) and Co(III) species (Equations (2) to (6)) [70].
Co(I) +H +  Co(II) + ½ H2 (2) Co(III)H − +H +  H2 + Co(III) Co(III)H − +H +  ½ H2 + Co(II) Our experiments indicate that only after the second reduction, the resulting [Co(0)(Por 2− )] 2− species is active in reducing protons. In the abovementioned study on the complexes [Co(TMAP)](ClO4)2, [Co(TMPyP)](ClO4)2, and [Co(TpyP)] catalytic currents representing the proton reduction were observed at −0.95 V [70]. In a recent study, very different behaviour was found for the catalytic proton reduction using [Co(TXPP)] (H2TXPP = meso-tetra-para-X-phenylporphin) catalysts [18]. For X = Cl = catalytic waves were observed at around −2 V comparable to our findings, while for X = OMe, the H2 evolution was observed already at around −1 V. Thus, we can conclude that the substitution pattern of the meso-tetraarylporphin ligands has a strong impact on the observed Mechanistically speaking, it seems that the first reduced species which we formally describe as [Co(I)(Por 2− ] − is not catalytically very active, which would stand in contrast to previous mechanistic studies [18,70] which proposed that the reduced Co(I) species reacts with protons forming Co(II) and Co(III) species (Equations (2) to (6)) [70].
Our experiments indicate that only after the second reduction, the resulting [Co(0)(Por 2− )] 2− species is active in reducing protons. In the abovementioned study on the complexes [Co(TMAP)](ClO 4 ) 2 , [Co(TMPyP)](ClO 4 ) 2 , and [Co(TpyP)] catalytic currents representing the proton reduction were observed at −0.95 V [70]. In a recent study, very different behaviour was found for the catalytic proton reduction using [Co(TXPP)] (H 2 TXPP = mesotetra-para-X-phenylporphin) catalysts [18]. For X = Cl = catalytic waves were observed at around −2 V comparable to our findings, while for X = OMe, the H 2 evolution was observed already at around −1 V. Thus, we can conclude that the substitution pattern of the meso-tetraarylporphin ligands has a strong impact on the observed catalytic potential. Depending on these patterns, the Co(TPP) derivatives might be an active catalyst in the Co(I) oxidation state or alternatively might need to reach the Co(0) state after the second reduction for efficient proton reduction.

Electroreduction CO 2 to CO
[Co(TMFPP)] was further tested for the electrocatalytic CO 2 reduction, in CO 2 -saturated DMF, with water added as a proton source. Cyclic voltammograms of [Co(TMFPP)] in the presence of CO 2 showed marked catalytic currents at potentials at around −2 V, with the presence of water being beneficial ( Figure 6A, green and red trace). No activity was found in the absence of the catalyst ( Figure 6B).
Controlled potential electrolysis at −2.25 V for 2 h in aqueous DMF under a CO 2 atmosphere gave a FE of 90%; GC confirmed the production of CO and only traces of H 2 . Remarkably, not even traces of the very common products formate and methanol were found.

Electroreduction CO2 to CO
[Co(TMFPP)] was further tested for the electrocatalytic CO2 reduction, in CO2-saturated DMF, with water added as a proton source. Cyclic voltammograms of [Co(TMFPP)] in the presence of CO2 showed marked catalytic currents at potentials at around −2 V, with the presence of water being beneficial ( Figure 6A, green and red trace). No activity was found in the absence of the catalyst ( Figure 6B). Controlled potential electrolysis at −2.25 V for 2 h in aqueous DMF under a CO2 atmosphere gave a FE of 90%; GC confirmed the production of CO and only traces of H2. Remarkably, not even traces of the very common products formate and methanol were found.
Thus, our unsupported [Co(TMFPP)] is markedly superior in terms of efficiency and selectivity to the standard [Co(TPP)] and support with an electron-conducting material might pave the way to operate the [Co(TMFPP)] at less negative potentials. Importantly, also here, only the second reduction wave produces the catalytically active species, which we describe as Co(0) complex.  [Co(TPP)] was reported with an FE of 50% for CO alongside with traces of H 2 (FE = 2%), formate (4%), acetate (2%) and oxalate (0.4%) on electrolysis at −1.95 or −2.05 V vs. SEC in DMF [88]. When immobilised on carbon nanotubes, [Co(TPP)] allowed an efficiency of 83% at −1.15 V or 93% at −1.35 V [88] and the authors could circumvent the rapid catalyst decomposition monitored by UV-vis absorption spectroscopy.

Catalytic Oxidative Degradation of Dyes
Thus, our unsupported [Co(TMFPP)] is markedly superior in terms of efficiency and selectivity to the standard [Co(TPP)] and support with an electron-conducting material might pave the way to operate the [Co(TMFPP)] at less negative potentials. Importantly, also here, only the second reduction wave produces the catalytically active species, which we describe as Co(0) complex.
To further study the reaction kinetics of the degradation, the C t /C 0 ratios were varied and a pseudo-first order rate constant k was calculated using Equation (7) (Langmuir-Hinshelwood): C t and C 0 are the dye concentrations at times t and 0, k is the first-order rate constant. The fit of the pseudo kinetic model is shown in Figure 8, and the rate constants of degradation (k) were calculated to 0.023 and 0.034 min −1 for CV and MG, respectively. To further study the reaction kinetics of the degradation, the Ct/C0 ratios were varied and a pseudo-first order rate constant k was calculated using Equation (7) (Langmuir-Hinshelwood): Ct and C0 are the dye concentrations at times t and 0, k is the first-order rate constant. The fit of the pseudo kinetic model is shown in Figure 8, and the rate constants of degradation (k) were calculated to 0.023 and 0.034 min −1 for CV and MG, respectively. The two triarylmethane dyes crystal violet and malachite green were decomposed using H2O2 and [Co(TMFPP)] as catalyst with an efficiency of more than 85% (C0 = 25 mg/L, pH = 8, H2O2 concentration = 3 mL/L, T = 25 °C). Two 4-cyanopyridine complexes of the type [Co(II)(Por)(4-CNpy)] (Por = meso-tetrakis(para-methoxyphenyl)porphyrinato and meso-tetra(para-chlorophenyl)porphyrin) as catalysts in the degradation of organic dyes using H2O2 gave a degradation efficiency of more than 78% [10,11,26,73], while recently reported Zn(II) triazole-substituted meso-arylsubstituted porphyrin complexes gave efficiencies of up to 50% [12,15,71,72]. This shows that Co(II) porphyrins are gener- and meso-tetra(para-chlorophenyl)porphyrin) as catalysts in the degradation of organic dyes using H 2 O 2 gave a degradation efficiency of more than 78% [10,11,26,73], while recently reported Zn(II) triazole-substituted meso-arylsubstituted porphyrin complexes gave efficiencies of up to 50% [12,15,71,72]. This shows that Co(II) porphyrins are generally superior to Zn(II) derivatives in line with the assumption that the first reduction is cobalt-centred (Co(II)(Co(I)).

Materials
All reagents and solvents were purchased from ACROS ORGANICS (Geel, Belgium) or Sigma Aldrich (St. Louis, MO, USA). Solvents were purified using literature methods [90]. Silica gel 150 (35-70 µm particle size, Davisil) was used for final purification of the products. Double-distilled water was used in the experiments.

Synthesis of the meso-Tetrakis(4-(trifluoromethyl)phenyl)porphyrin (H 2 TMFPP)
Note that 3.65 g of 4-(trifluoromethyl)benzaldehyde (21 mmol) were dissolved in propionic acid (100 mL) in air and heated to 120 • C. In addition, 1.4 g of pyrrole (1.35 mL, 21 mmol) were added dropwise to the reaction and the mixture was kept at 120 • C for a further 45 min. The resulting solution was allowed to cool and the tarry mixture was filtered to give a black solid, which was rinsed with water (5 × 100 mL) and (5 × 100 mL) n-hexane and finally dried under vacuum with a yield of 1.

Instrumentation
UV-vis absorption spectra were recorded on a WinASPECT PLUS (validation for SPECORD PLUS version 4.2, WinASPECT, Jena, Germany) scanning spectrophotometer using 10 mm path length cuvettes. 1 H NMR spectra were measured on Bruker DPX 500 spectrometers (Bruker, Rheinhausen, Germany) in CDCl 3 with the solvent peak as an internal standard. FT-IR spectra were measured on a Perkin Elmer Spectrum Two FT-IR spectrometer (Perkin Elmer, Darmstadt, Germany). Elemental analysis and mass spectrometry were carried out in the nanobio chemistry platform of the ICMG, Grenoble, France. A Fluoromax-4 spectrofluorometer (Horiba Scientific, 59120 Loos, France) was used to record photoluminescence (PL) spectra at room temperature in CH 2 Cl 2 . PL quantum yield (Φ PL ) was determined using the optical method [12] with [Zn(TPP)] as standard (Φ PL = 0.031). The luminescence lifetime detection was performed upon irradiation at λ = 405 nm. The luminescence decay was analysed using the PicoQuant FLUOFIT software (PicoQuant, Berlin, Germany) [15].

Electrochemistry
Cyclic voltammetry experiments were performed using a CH-660B potentiostat at room temperature. All measurements were performed in DMF with a solute concentration of approximately 10 −3 M and nBu 4 NBF 4 (0.1 M) as supporting electrolyte. A three-electrode cell was set up with a glassy carbon working electrode, a Pt wire as counter electrode, and an Ag/AgNO 3 reference electrode. Potentials were converted into values for the saturated calomel electrode (SCE) by applying Equation (8) [11,63,91,92]:

Electrocatalytic CO 2 Reduction
The experiments were performed at room temperature under a CO 2 atmosphere in a conventional three-electrode cell sealed with Apiezon M vacuum grease. A glassy carbon electrode plate (2 cm 2 , 0.25 mm thickness) was used as the working electrode in the cathodic compartment. A 0.5 mm diameter platinum wire (10 cm length) was used as the counter electrode in the anodic compartment. The cell was purged with Ar or CO 2 for a minimum of 15 min before controlled potential electrolysis was carried out. Constant magnetic stirring was applied during electrolysis.

Gas Detection
Gas analyses were performed using a GC/MS gas chromatography (Perkin Elmer Clarus 560) instrument with a thermal conductivity detector fitted with RT-QPlot pre column + molecular sieve 5Å column. The temperature was held at 150 • C for the detector and 80 • C for the oven. The carrier gas was helium. Manual injections of 100 µL were performed during the experiment via a gas-tight Hamilton microsyringe. The total volume of the cell was 173 mL.

Faradaic Efficiency Calculation
The Faradaic Efficiency (FE) of the CO 2 reduction or hydrogen evolution reaction (HER) was calculated using Equation (9): where Z is the amount of product in mol, n is the number of the electrons (2 for both CO and H 2 ), F is the Faraday constant, and Q is the number of electrons (or charge) passed through the solution during electrolysis (I t).

Catalytic Dye Degradation
In a typical investigation, to a 10 mL aqueous solution of the dyes crystal violet (CV) and malachite Green (MG) (20 mg L −1 ), 3 mL/L of H 2 O 2 (30 wt %) were added. Next, 5 mg of the catalyst were added to this mixture at a stirring speed of 250 rpm. The reaction solution was pipetted into a quartz cell and UV-vis absorption spectra were recorded at different reaction times. Blank experiments were carried out to confirm that the reactions did not take place without catalyst in the presence of H 2 O 2 .

Conclusions
In this work, the meso-tetrakis(4-(trifluoromethyl)phenyl)porphyrinato cobalt(II) complex [Co(TMFPP)] was synthesised via modified literature methods in an excellent yield of 93% from the free-base porphyrin meso-tetrakis(4-(trifluoromethyl)phenyl)porphyrin (H 2 TMFPP). Elemental analysis, FT-IR, and 1 H NMR spectroscopy confirmed the molecular entities. UV-vis absorption spectroscopy localised the Soret band at 437 nm and the Q band at 554 nm. The optical band gap Eg was calculated using the Tauc plot method to 2.15 eV. The (αhυ) 2 over E plot suggests a semiconducting behaviour of the material. Cyclic voltammetry of the title compound showed two fully reversible reduction waves. Both the first wave, observed at E 1/2 = −0.91 V vs. SCE, and the second at E 1/2 = −2.05 V are ascribed to cobalt-centred processes Co(II)/Co(I) and Co(I)/Co(0), respectively. The first oxidation wave at around 0.3 V is metal-centred Co(II)/Co(III) and broadened through the interaction of the DMF solvent with the oxidised complex. A second oxidation is following at 0.98 V, which is presumably due to the redox couple Por 2− /Por − . [Co(TMFPP)] is a very active catalyst for the electrochemical formation of H 2 from DMF/acetic acid, with a Faradaic Efficiency (EF) of 85% at a working potential of −2.3 V. This is in line with [Co(0)(Por 2− )] 2− being the active species. The complex also catalysed the reduction of CO 2 to CO in aqueous DMF under CO 2 atmosphere with a high EF of 90% and only traces of H 2 by-product, making our derivative superior to the standard [Co(TPP)]. Also here, catalytic currents are only observed at potentials coinciding with the second reduction potential in the voltammograms, thus the same [Co(0)(Por 2− )] 2− species seems to be active as for the proton reduction. Moreover, the two triarylmethane dyes crystal violet and malachite green were decomposed in aqueous solution using H 2 O 2 and [Co(TMFPP)] as catalyst with an efficiency of more than 85% in one batch. Given the high stability of the complex and the relatively easy preparation with excellent yields, this makes [Co(TMFPP)] a versatile catalyst for important electrocatalytic reductions and oxidations. The performance on the cathodic side might be improved with the goal of less negative working potentials in future work by blending the complex with electroactive materials such as carbon nanotubes, or by immobilising the complex directly on electrodes.
Supplementary Materials: The following information is available online. Figure S1: FT-IR spectrum of [Co(TMFPP)]. Figure S2