The Role of Redox Potential and Molecular Structure of Co(II)-Polypyridine Complexes on the Molecular Catalysis of CO 2 Reduction

: In this work, we report the electrochemical response of a family of Co(II) complexes, [Co II (L) 3 ] 2+ and [Co II (L’) 2 ] 2+ (L = 2,2’-bipyridine, 1,10-phenanthroline, 3,4,7,8-tetramethyl-1,10-phenanthroline, 5,6-dimethyl-1,10-phenanthroline, and 4,7-diphenyl-1,10-phenanthroline; L’ = terpyridine and 4-chloro-terpyridine), in the presence and absence of CO 2 in order to understand the role of the redox potential and molecular structure on the molecular catalysis of CO 2 reduction. The tris chelate complexes exhibited three electron transfer processes [Co II (L) 3 ] 2+ (cid:29) [Co III (L) 3 ] 3+ + 1e − , [Co II (L) 3 ] 2+ +1e − (cid:29) [Co I (L) 3 ] + , and [Co I (L) 3 ] + + 2e - (cid:29) [Co I (L)(L − ) 2 ] − . In the case of complexes with 1,10-phen and 2,2-bipy, the third redox process showed a coupled chemical reaction [Co I (L)(L − ) 2 ] − → [Co I (L − ) 2 ] − + L. For bis chelate complexes, three electron transfer processes associated with the redox couples [Co II (L) 2 ]/[Co III (L) 2 ] 3+ , [Co II (L) 2 ] 2+ /[Co I (L) 2 ] + , and [Co I (L) 2 ] + /[Co I (L)(L − )] were registered, including a coupled chemical reaction only for the complex containing the ligand 4-chloro-terpyridine. Foot to the wave analysis (FOWA) obtained from cyclic voltammetry experiments allowed us to calculate the catalytic rate constant (k) for the molecular catalysis of CO 2 reduction. The complex [Co(3,4,7,8-tm-1,10-phen) 3 ] 2+ presented a high k value; moreover, the complex [Co(4-Cl-terpy) 3 ] 2+ did not show catalytic activity, indicating that the more negative redox potential and the absence of the coupled chemical reaction increased the molecular catalysis. Density functional theory (DFT) calculations for compounds and CO 2 were obtained to rationalize the effect of electronic structure on the catalytic rate constant (k) of CO 2 reduction. a comparative study of the electrochemical response of the [Co II (bipy) 3 ](BF 4 ) 2 in MeCN solution in the presence and absence of CO 2 . As can be observed, signal Ic is not altered; however, the IIc signal (associated with the bi-electronic reduction) presents an increase in current due to molecular catalysis processes. The disappearance of Ia and IIa is also associated with the catalytic process that implies the consumption of the cobalt redox mediator. The appearance of a new signal I’ at − 0.772 V vs. Fc/Fc + is related to the oxidation of a new electrogenerated species. It should be noted that the reduction of CO 2 is activated at − 1.7 V vs. Fc/Fc+, where an increase in the current of process II is observed. 3 ] 2+ , it can be inferred that the electronic delocalization effect of three conjugated aromatic rings of the electrogenerated species in process II increased the catalytic constant.


Introduction
The transformation of carbon dioxide into value-added products has stimulated the creativity and curiosity of many generations of scientists [1]. This molecule represents an alternative carbon raw material that has the advantages of being renewable, cheap, safe, and nontoxic. Carbon dioxide can be transformed by biological, chemical, photochemical, reforming, inorganic, and electrochemical methods, with a wide variety of products [2].
From an electrochemical point of view, it is possible to carry out CO 2 reduction. In the absence of proton donors, two consecutive reduction processes have been proposed, with typical thermodynamic redox potentials: It has been established that in the first reduction, a change in the geometry of the CO 2 takes place, from linear to bent-type CO • − 2 [3], making slow electrode kinetics with a high overpotential [4,5]. Hence, experimentally, the reduction of CO 2 on a Pt electrode requires much more negative potentials for the formation of the radical anion CO 2 ·than −1.90 V vs. NHE (normal hydrogen electrode) in water and −1.97 V vs. NHE in DMF [6,7].
In the presence of proton donors, the electrochemical reduction of carbon dioxide can give different reaction products, including carbon monoxide, formic acid, methanol, and methane, as shown below [6]:  (7) Indeed, the processes are carried out at a very negative potential with a slow electron transfer, which is expected for small molecules such as CO 2 with high solvation and reorganization energies. On the other hand, CO 2 can react with its radical anion and by the addition of one electron to produce carbon monoxide and carbonate in a (ECE-DISP) mechanism. Another possible reaction is the dimerization between two radical anions, which leads to the formation of the divalent oxalate anion. As in the case of CO 2 , there are many electrochemical reactions that require a significant overpotential to be appreciated. Hence, different electrocatalytic materials can be used for this purpose [8][9][10]. However, the control of crystallinity and the number of defects make it difficult to control the electrocatalytic processes.
Another approach to catalyzing electrochemical reactions is the use of molecules as catalysts, which can be viewed as a mediated process or molecular catalysis. In electrochemistry, molecular catalysis can be defined as the use of molecules, whether in solution or immobilized on the electrode, susceptible to being reduced to transfer electrons to a target molecule with slow electrode kinetics such as CO 2 , in a concerted or sequential manner [8,10,11].
Coordination compounds with polypyridine ligands represent one of the most studied classes of molecular catalysts for CO 2 reduction. As they stabilize various oxidation states for metals and can store electrons in π* orbitals as a function of their substituents, these ligands present a great variety in the type of denticity, which includes bipyridines, terpyridines, tetrapyridines, phenanthroline, and their derivatives [12].
First, studies found in the literature correspond to the use of the compound [Re(bipy) (CO) 3 Cl] as a catalyst in the electrochemical reduction of carbon dioxide, where the only product obtained was carbon monoxide, with a faradaic efficiency of 98% [13].
On the other hand, the use of metal complexes with earth-abundant elements and polypyridine ligands for the electrochemical reduction of CO 2 has attracted the attention again in recent years [12]. The use of several Mn-bipy complexes has been reported as electrocatalysts in the reduction of CO 2 to CO [15][16][17][18].
A series of Ni(II) compounds with N-heterocyclic ligands with carbenes and pyridine have the ability to carry out the electrochemical conversion of CO 2 to CO [19,20]. The Ni complex [Ni(bipy) 3 ] 2+ presents a bi-electronic reduction at a potential of −1.58 V vs. Fc + /Fc. The dissociation of a bipyridine has been proposed by generating [Ni(bipy) 2 ] 0 , which, in the presence of CO 2 , increases the catalytic current associated with its reduction, giving CO and CO 3 − as products [21]. In Fe(II) coordination compounds with polypyridine ligands, vacant coordination sites are not necessary for the molecular catalysis of CO 2 . The mechanism involves the outer sphere homogeneous electron transfer between carbon dioxide and one of the reduced forms of the complexes [22,23].
The use of cobalt complexes with pyridylimine ligands for the electrochemical reduction of CO 2 has been informed from the literature [19], reporting the production of carbon monoxide at a potential very close to the redox potential of the Co I/0 couple (E 1/2 = −1.88 V vs. Fc + /Fc) [19]. Cobalt compounds with phenanthrolines have also been studied [20], for example, [Co(phen) 3 ] 2+ , which presents an increase in the catalytic current in the presence of CO 2 , associated with the reduction of the latter to formate [19].
Despite the wide variety of earth-abundant metal complexes reported for CO 2 reduction, there are no systematic studies that indicate the effect of redox potential and molecular structure (i.e., the metal center and the nature of the ligands) in the molecular catalysis of carbon dioxide. Therefore, in this work, we studied the electrochemical response of a family of Co(II) complexes, [Co II (L) 3 ] 2+ and [Co II (L ) 2 ] 2+ (L = 2,2'-bipyridine, 1,10phenanthroline, 3,4,7,8-tetramethyl-1,10-phenanthroline, 5,6-dimethyl-1,10-phenanthroline, and 4,7-diphenyl-1,10-phenanthroline; L'=terpyridine and 4-chloro-terpyridine), in the presence and absence of CO 2 in order to understand the role of redox potential and the electronic structure of metal complexes on the molecular catalysis of CO 2 . Foot to the wave analysis (FOWA) obtained from cyclic voltammetry experiments allowed us to calculate the catalytic rate constant (k) for mediated CO 2 reduction. DFT calculations for compounds and CO 2 were carried out to rationalize the effect of electronic structure on the catalytic rate constant (k) for CO 2 reduction.

Electrochemical Response of Cobalt Coordination Compounds
Cyclic voltammetry in its normalized current representation i/v 1/2 of Co(II) complexes at different scan rates are presented in the Supporting Information. For the compound [Co II (bipy) 3 ](BF 4 ) 2 , a reversible electron transfer I and a quasi-reversible electron transfer III, associated with the redox couples [Co II (bipy) 3 ] 2+ /[Co I (bipy) 3 ] + and [Co II (bipy) 3 ] 2+ / [Co III (bipy) 3 ] 3+ , were observed (see Figure S1) [24]. Furthermore, a bi-electronic quasireversible reduction II with a coupled chemical reaction was detected [24]. According to the literature, the dissociation of a bipy ligand and the formation of a product with a square geometry [Co I (bipy − ) 2 ] − have been proposed, which is stabilized by the electronic configuration d 8 in the electrogenerated Co(I) species [25]. This dissociation has been reported in the literature for Co, Fe, Ni, and Ru complexes with polypyridine ligands [26][27][28][29][30][31]. For [Co II (phen) 3 ](BF 4 ) 2 , a similar electrochemical response has been observed (see Figure S2). For processes, I, II, and III, the redox potentials were calculated with the equation E • = (E pa + E pc )/2 (see Table 1). According to the evidence presented, the following mechanism is confirmed:  2 ] − + L L = phen or bipy (11)  3 ](BF 4 ) 2 , mono electronic electron transfer processes were observed in I and III, while for process II, a bi-electronic quasi-reversible electron transfer was registered (see Figures S3-S5). The redox potential values are presented in Table 1. The cyclic voltammetry response of [Co II (terpy) 2 ](BF 4 ) 2 presented three reversible mono-electronic transfers. According to the above and considering the π acceptor character of the polypyridine ligands, the following mechanism is presented (see Figure S6) [ For the compound [Co II (4-Cl-terpy) 2 ](BF 4 ) 2 , three reduction signals Ic, IIc, and IIIc and two oxidation signals Ia and IIIa can be observed (see Figure S6). Processes I and III correspond to the redox couples Co(II)/Co(I) and Co(II)/Co(III), respectively. The fact that the normalized current i/v 1/2 for process IIc decreases with scan rate and the oxidation signal IIa is absent indicate a ECE mechanism [24]. The change in the mechanism associated with the presence of the chlorine atom in the terpyridine complexes can be related to the weakening of the Co-N bond, which promotes the ligand dissociation. This fact has been reported in the literature for Co, Fe, Ni, and Ru complexes with polypyridine ligands, already mentioned previously [26][27][28][29][30][31]. The following mechanism is proposed: [Co II (4-Cl-terpy) 2  Once the electrochemical behavior of the cobalt (II) coordination compounds was presented, the molecular catalysis for the reduction of CO 2 was studied. Figure 1 presents a comparative study of the electrochemical response of the [Co II (bipy) 3 ](BF 4 ) 2 in MeCN solution in the presence and absence of CO 2 . As can be observed, signal Ic is not altered; however, the IIc signal (associated with the bi-electronic reduction) presents an increase in current due to molecular catalysis processes. The disappearance of Ia and IIa is also associated with the catalytic process that implies the consumption of the cobalt redox mediator. The appearance of a new signal I' at −0.772 V vs. Fc/Fc + is related to the oxidation of a new electrogenerated species. It should be noted that the reduction of CO 2 is activated at −1.7 V vs. Fc/Fc+, where an increase in the current of process II is observed.
Once the electrochemical behavior of the cobalt (II) coordination compounds presented, the molecular catalysis for the reduction of CO2 was studied. Figure 1 prese a comparative study of the electrochemical response of the [Co II (bipy)3](BF4)2 in Me solution in the presence and absence of CO2. As can be observed, signal Ic is not alte however, the IIc signal (associated with the bi-electronic reduction) presents an incre in current due to molecular catalysis processes. The disappearance of Ia and IIa is associated with the catalytic process that implies the consumption of the cobalt redox diator. The appearance of a new signal I' at −0.772 V vs. Fc/Fc + is related to the oxida of a new electrogenerated species. It should be noted that the reduction of CO2 is activa at −1.7 V vs. Fc/Fc+, where an increase in the current of process II is observed. Subsequently, cyclic voltammetry experiments were carried out at different s rates. Figure 2 shows its normalized current representation (i/v 1/2 ), in which all the m tioned signals converge to the same potentials. Subsequently, cyclic voltammetry experiments were carried out at different scan rates. Figure 2 shows its normalized current representation (i/v 1/2 ), in which all the mentioned signals converge to the same potentials.
From cyclic voltammetry experiments of the complex [Co II (bipy) 3 ](BF 4 ) 2 in the absence and presence of CO 2 , at different scan rates, it is possible to calculate the catalytic constant using FOWA in process IIc, according to the following expression [32,33]: From cyclic voltammetry experiments of the complex [Co II (bipy)3](BF4)2 in the sence and presence of CO2, at different scan rates, it is possible to calculate the catal constant using FOWA in process IIc, according to the following expression [32,33]: The value of the slope of a graph ik/ip° vs.
allows the calculation of k, which, in this case, corresponds to a value of 9.24 M −1 s −1 Figure 3).   The value of the slope of a graph i k /i p • vs.
allows the calculation of k, which, in this case, corresponds to a value of 9.24 M −1 s −1 (see Figure 3). From cyclic voltammetry experiments of the complex [Co II (bipy)3](BF4)2 in the absence and presence of CO2, at different scan rates, it is possible to calculate the catalytic constant using FOWA in process IIc, according to the following expression [32,33]: The value of the slope of a graph ik/ip° vs.
allows the calculation of k, which, in this case, corresponds to a value of 9.24 M −1 s −1 (see Figure 3).    Similar to what was previously discussed, Figure 4 presents a comparative study of the electrochemical response of the complex [Co II (phen) 3 ](BF 4 ) 2 in a MeCN solution in the presence and absence of carbon dioxide. As can be observed, process III and the Ic signal are not altered; however, the IIc signal presents an increase in current related to molecular catalysis. The disappearance of the IIa and Ia signals and the appearance of a new broad signal (I'a) in −0.650 V vs. Fc/Fc + are also associated with the mediated CO 2 reduction.
presence and absence of carbon dioxide. As can be observed, process III and the Ic signa are not altered; however, the IIc signal presents an increase in current related to molecula catalysis. The disappearance of the IIa and Ia signals and the appearance of a new broad signal (I'a) in −0.650 V vs. Fc/Fc + are also associated with the mediated CO2 reduction. Cyclic voltammetry experiments were carried out at different scan rates, and thei normalized current representations (i/v 1/2 ) are shown in Figure 5. The catalytic constan was calculated using FOWA, and the respective plots for the [Co II (phen)3](BF4)2 complex are shown in Figure 6, where a value of k = 160 M −1 s −1 was obtained. Cyclic voltammetry experiments were carried out at different scan rates, and their normalized current representations (i/v 1/2 ) are shown in Figure 5. The catalytic constant was calculated using FOWA, and the respective plots for the [Co II (phen) 3 ](BF 4 ) 2 complex are shown in Figure 6, where a value of k = 160 M −1 s −1 was obtained. On the other hand, in the electrochemical response of the compound [Co II (terpy)2](BF4)2 in the presence of CO2, changes in current for signals IIc and IIa were observed, as well as the absence of some other species associated with an intermediate (see Figure 7). This fact suggests that this compound presents an efficient catalyst regeneration compared to the previously presented compounds. Cyclic voltammetry experiments were performed at different scan rates, and, from these experiments, the catalytic constant was calculated using FOWA (see Figures 8 and 9), obtaining a value of 57.8 M −1 s −1 . For the complex [Co II (4-Cl-terpy)2](BF4)2, no molecular catalysis was observed in the presence of CO2 associated with the fact that the rate constant of the coupled reaction in IIc deactivates the catalyst (see Figures 10 and 11). The same analysis was performed for the remaining compounds, and the catalytic constant k values are reported in Table 1 and Figures S8-S13. On the other hand, in the electrochemical response of the compound [Co II (terpy) 2 ](BF 4 ) 2 in the presence of CO 2 , changes in current for signals IIc and IIa were observed, as well as the absence of some other species associated with an intermediate (see Figure 7). This fact suggests that this compound presents an efficient catalyst regeneration compared to the previously presented compounds. Cyclic voltammetry experiments were performed at different scan rates, and, from these experiments, the catalytic constant was calculated using FOWA (see Figures 8 and 9), obtaining a value of 57.8 M −1 s −1 . For the complex [Co II (4-Cl-terpy) 2 ](BF 4 ) 2 , no molecular catalysis was observed in the presence of CO 2 associated with the fact that the rate constant of the coupled reaction in IIc deactivates the catalyst (see Figures 10 and 11). The same analysis was performed for the remaining compounds, and the catalytic constant k values are reported in Table 1 and Figures S8-S13.           An inspection of the obtained values of the catalytic constant for CO 2 reduction (see Table 1and Figure 12) indicates that the more negative the redox potential for process II with the absence of a coupled reaction, the higher the value of k. In this sense, the compound [Co(3,4,7,8-tm-phen) 3 ] 2+ presented the highest value of k, contrary to that observed in the compound [Co(4-Cl-terpy) 3 ] 2+ that presented a coupled reaction with a rate constant higher than that of the time window used in the voltammetry experiments (see Figure 11). In the case of the compound [Co(4,7-dphen-phen) 3 ] 2+ , the presence of the aromatic rings stabilized the charge on them, making it a better molecular catalyst despite having a less negative redox potential value than its analog with the 5,6-dmethyl-1,10-phenanathroline. In the case of the compounds [Co(bipy) 3 ] 2+ , [Co(phen) 3 ] 2+ , and [Co(terpy) 3 ] 2+ , it can be inferred that the electronic delocalization effect of three conjugated aromatic rings of the electrogenerated species in process II increased the catalytic constant. An inspection of the obtained values of the catalytic constant for CO2 reduction (see Table 1and Figure 12) indicates that the more negative the redox potential for process II with the absence of a coupled reaction, the higher the value of k. In this sense, the compound [Co(3,4,7,8-tm-phen)3] 2+ presented the highest value of k, contrary to that observed in the compound [Co(4-Cl-terpy)3] 2+ that presented a coupled reaction with a rate constant higher than that of the time window used in the voltammetry experiments (see Figure 11 S715). In the case of the compound [Co(4,7-dphen-phen)3] 2+ , the presence of the aromatic rings stabilized the charge on them, making it a better molecular catalyst despite having a less negative redox potential value than its analog with the 5,6-dmethyl-1,10-phenanathroline. In the case of the compounds [Co(bipy)3] 2+ , [Co(phen)3] 2+ , and [Co(terpy)3] 2+ , it can be inferred that the electronic delocalization effect of three conjugated aromatic rings of the electrogenerated species in process II increased the catalytic constant.

Molecular Orbitals Description
To relate the molecular and electronic structure with the molecular catalysis activity of CO2 reduction, DFT calculations of the species involved in the electron transfer processes were carried out, using the appropriate multiplicity according to the experimental

Molecular Orbitals Description
To relate the molecular and electronic structure with the molecular catalysis activity of CO 2 reduction, DFT calculations of the species involved in the electron transfer processes were carried out, using the appropriate multiplicity according to the experimental evidence and the convergence criteria with the level of calculation used. For the [Co II (L) 3 ] 2+ and [Co II (L ) 2 ] 2+ cations, calculations were performed considering a total spin quantum number (S) S = 3/2, according to the reported magnetic moments values [34]. For the species involved in the first electron transfer I, the electrogeneration of the species [Co I (L) 3 ] + and [Co I (L ) 2 ] + , a value of S = 1 was considered. Figure 13a,b present the semi-occupied molecular orbitals (SOMOs) of the species [Co II (bipy) 3 ] 2+ and [Co I (bipy) 3 ] + , in which the electronic density is observed to be over the metallic center and the aromatic rings of the ligand. Figure 13c shows the SOMO for the species proposed in process II [Co I (bipy)(bipy − ) 2 ] − (S = 1), where the electron density is appreciated exclusively on the aromatic rings, as has been proposed in the literature for the electrochemical reduction of coordinated diiminic ligands [26][27][28][29][30][31]. For [Co II (phen) 3 ] 2+ , DFT calculations indicate that SOMO has a higher contribution of the ligands with respect to its analog with bipy (see Figure 14). In the case of the rest of the compound with phenanthroline ligands [Co II (L) 3 ] 2+ (see Figures S14-S16), the same behavior is observed. For the reduced species with substituted phenanthroline ligands [Co I (L) 3 ] + (S = 1) and [Co I (L)(L − ) 2 ] − (S = 2), it is observed that the electron density is on the metal center for the first case and exclusively on the ligand in the second case (see Figures S14-S16). This fact can be attributed to the conjugation effect of the three fused aromatic rings in the phenanthroline ligands. For the [Co II (terpy) 3 ] 2+ and [Co II (4-Cl-terpy) 3 ] 2+ , the electron density of the SOMOs is over the metallic center and the aromatic rings of the ligand (see Figure 15 and Figure S17). For the species [Co I (terpy) 3 ] + and [Co I (4-Cl-terpy) 3 ] + , the electronic density is located in the metallic centers. On the other hand, for the species proposed in process II [Co I (terpy)(terpy − )] (S = 5/2) and [Co I (4-Cl-terpy)(4-Cl-terpy − )] (S = 3/2), it is observed that the electronic density is located on the ligands.     In Table 2, the SOMO energy values, along with the catalytic constant values for the CO 2 -mediated electrochemical reduction k, are presented. It can be seen that with increase in the number of electrons in the species in question, its energy becomes more positive. The compound with most positive energy and the highest accumulation of electrons would be expected to have the highest capacity to electrochemically catalyze CO 2 , measured through its catalytic constant k. Considering this idea, the complex with the ligands bipy, phen, 3,4,7,8-tm-phen, and 5,6-dm-phen should present very similar catalytic constants k. However, the presence of the coupled reaction decreases the k values. For the case of the compound [Co(4,7-dphen-phen) 3 ] 2+ , the fact that its species with charge accumulation [Co I (4,7-dphen-phen)(4,7-dphen-phen − ) 2 ] − presents a highly extended electron density distribution over all its aromatic rings, inferred from SOMO, increases the local reactivity sites with CO 2 . The coupled reaction for the [Co I (4-Cl-terpy)(4-Cl-terpy − )] species does not allow its SOMO energy value to be considered to establish any tendency.  Another factor that should be considered to understand the relationship between SOMO energy and catalytic processes is the faradaic efficiency, which is calculated by means of bulk electrolysis experiments. Nevertheless, the difference in time scale between cyclic voltammetry and bulk electrolysis experiments could give different mechanisms and products. Besides, sophisticated analytical techniques to identify and quantify the product are needed. Hence, a simple strategy with similar information can be proposed. Considering that CO 2 is not reduced directly to the electrode but in a catalytic pathway by the electrolyzed complex, we propose the use of the charges for the reduction of complexes in the absence (Q complex ) and presence of CO 2 (Q CO2 complex ), obtained from cyclic voltammetry experiments. The following ratio can be used as a measurement of the efficiency of the CO 2 transformation: This ratio can be used as an approach for the ability to regenerate the reduced complex by its reaction with CO 2 in a cyclic voltammetry time scale, which includes several processes such as the deactivation of the catalyst. Table 2 shows the calculated ratios for Co(II) complexes. The low values for the complex with the ligands bipy, phen, and 5,6-dm-phen in comparison with the complex containing the ligand 3,4,7,8-tm−phenanthroline is related with the diminution of the catalytic constant k for similar SOMO values.
With the same level of theory, the boundary orbital energies of CO 2 and related species in its reduction were calculated (see Table 3). As an approximation, a charge transfer of the species with charge accumulation in the cobalt complexes through their SOMOs with the LUMOs of the CO 2 , CO 2 , CO 2 •− , and CO 2 2− can be considered for the molecular catalysis of CO 2 . The SOMO values for the [Co I (L)(L − ) 2 ] − species were about 65 kcal/mol, with these being very similar to the LUMO energy values of CO 2 , confirming that the first step involves the reduction of CO 2 directly. The LUMO energy values for the CO 2 •− and CO 2 2− species suggest that their reduction requires their coordination to lower their activation energy. This interaction can be through either their oxygen or carbon atoms, as shown in Figures S18 and S19, depending on the symmetry and energy of the available orbitals of the metallic center.

Identification of Products
To identify the products from the molecular catalysis of CO 2 reduction with cobalt complexes, macro-electrolysis experiments were carried out in a two-compartment cell, applying a potential value of -2.2 V vs. Fc/Fc + . A color change was observed in the solution, from pale yellow to intense red.
We explored the presence of methanol, ethanol, and carbon monoxide by means of UPLC (ultra-performance liquid chromatography) and gas-chromatography (see experimental section for conditions). However, none of these products were detected. From the solution obtained in macro-electrolysis experiments, a white precipitate was obtained and analyzed with Fourier Transform infrared (FT-IR) spectroscopy in Attenuated Total reflection (ATR) mode. A broad signal around 1630 cm −1 was detected. The anhydrous conditions and the FT-IR response, similar to Na 2 C 2 O 4 spectra, suggest that oxalate is the main product of CO 2 reduction with Co(II) complexes

Electrochemical Characterization
The electrochemical studies were carried out using a potentiostat/galvanostat Biologic SP-300 (Lyon-France), using 1 mM solutions of each coordination compound in ultra-dry acetonitrile (MeCN), with 0.1 M of tetrabutylammonium hexafluorophosphate (TBAPF 6 ) as a supporting electrolyte. Before preparing the solutions, all complexes were dried at 100 • C in a vacuum oven for 24 h. A conventional cell with a three-electrode arrangement was used. Glassy carbon (Φ = 3 mm) was used as the working electrode (WE). Platinum wire was used as a counter electrode (CE) and silver wire as a pseudo-reference electrode (RE). The reference electrode was prepared with a silver wire immersed in 0.1 M of silver nitrate solution (AgNO 3 ) in MeCN within a separate compartment, connected through a Vycor membrane. All potentials are referenced to the Fc/Fc + couple, as recommended by the IUPAC [35].
Prior to each measurement, the solutions were bubbled with nitrogen (N 2 ) or carbon dioxide (CO 2 ) for 5 min. The cleaning of the working electrode was carried out by means of manual polishing with 0.1 µm diamond powder, and then rinsed with water and placed in an ultrasonic bath for one minute, to be rinsed again with water and dried with acetone. Voltamperograms were obtained at different scan rates from the open-circuit potential (E oc ) to the cathodic direction. Ohmic drop compensation (iR) was performed by the positive feedback method. The calculation of the solution resistance (Ru) was determined by the impedance measurement (ZIR). The determination of the catalytic constant of the system was evaluated using the foot of the wave analysis (FOWA) [36].

DFT Calculations
For redox mediators, density functional theory [37][38][39] calculations were implemented using Gaussian 09 [40]. Full geometry optimization without symmetry constraints was performed using the B3LYP [41][42][43] density functional, and Lanl2mb [44,45] basis sets were used. Optimized geometries of local minima were verified by the number of imaginary frequencies (which should be zero). Previous studies have indicated that DFT results are accurate to describe stabilities and equilibrium geometries for B3LYP density functional and Lanl2mb basis set combinations [46].

Product Identification
Here, 5 mL of CO 2 -saturated acetonitrile solution containing 1 mM of complex and 0.1 M of TBAPF 6 in a two-compartment cell with Toray carbon paper electrodes with an area of 2 cm 2 was used for macro-electrolysis experiments. The potentiostat/galvanostat Biologic SP-300 was used, applying a potential at −2.2 V vs. Fc/Fc + V for 10 min. A color change was observed in the solution, from pale yellow to intense orange.
Methanol and ethanol analyses were carried out in an ultra-performance liquid chromatography system (UPLC) model Acquity H-Class (Waters ® , Mildford, MA, USA). The UPLC system had a cooling autosampler, a quaternary solvent manager, a 2414 Model Refractive Index (RI) detector, and an oven for the analytical column. The chromatographic determination was made in a Biorad Aminex HPX-87H cation-exchange column (300 mm × 7.8 mm), using a mobile phase of 0.008 N of sulfuric acid at a flow rate (isocratic mode) of 0.4 mL/min, an oven temperature of 30 • C, a temperature cell RI of 35 • C, and an injection volume of 10 mL. Peak heights were measured using Empower3 chromatography software (Waters ® , Mildford, MA, USA). Ethanol or methanol:H 2 O solutions containing 0.7:0.7, 1.5:1.5, 3.1:3.1, 6.2:6.2, 12.5:12.5, 25:25, and 50:50% were used as reference solutions for the calibration curve (three replicates). Ethanol or methanol were identified by comparison of the retention times. All reference solutions and samples were filtered using a 0.2 mm nylon membrane.
For carbon monoxide identification, a CG Thermo Scientific Trace 1300 model (Thermo Scientific, Madison, USA) was used. For separating the target gas component, a Traceplot TG-BOND Msieve 5A, 0.32 mm × 30 cm −1 × 30 m, was employed. The detection was performed by using a thermal conductivity detector (TCD) and the output signal was monitored using Chromeleon Software (Version 7.0). For measurements, a 50 µL gas sample was injected using a gas-tight syringe Hamilton (Reno-USA) via the injection port. The injector and detector temperatures were 120 • C and 100 • C, respectively. For GC separation, the sample was injected via the sample loop into the separation column, which was heated at 50 • C, with a split ratio of 30. Ultrapure He at a flow rate of 5.0 mL min −1 was used as a carrier gas. The data were estimated by automated integration of the area under the resolved chromatographic profile using Chromeleon Software (Version 7.0). FT-IR measurements were carried out in a Shimadzu IRAffinity-1S spectrophotometer (Kyoto-Japan) with an ATR module and a range from 4500 to 650 cm −1 .
Electrogenerated cobalt species with a high accumulation of electrons and more positive SOMO energy showed a high catalytic constant k. For complexes with the same SOMO energies, the presence of coupled chemical reactions diminished their values of catalytic constant k.
The ratio Q CO2 complex Q complex was proposed as the measurement of the efficiency of the CO 2 transformation. Low values of this ratio are related with a diminution of catalytic constant k.
The anhydrous conditions in electrochemical experiments and the evidence obtained from analytical techniques allowed us to suggest that oxalate is the main product of CO 2 reduction with Co(II) complexes.