Effect of the 2-R-Allyl and Chloride Ligands on the Cathodic Paths of [Mo(η3-2-R-allyl)(α-diimine)(CO)2Cl] (R = H, CH3; α-diimine = 6,6′-Dimethyl-2,2′-bipyridine, Bis(p-tolylimino)acenaphthene)

The new, formally Mo(II) complexes [Mo(η3-2-R-allyl)(6,6′-dmbipy)(CO)2Cl] (6,6′-dmbipy = 6,6′-dimethyl-2,2′-bipyridine; 2-R-allyl = allyl for R = H, 2-methallyl for R = CH3) and [Mo(η3-2-methallyl)(pTol-bian)(CO)2Cl] (pTol-bian = bis(p-tolylimino)acenaphthene) share, in this rare case, the same structural type. The effect of the anionic π-donor ligand X (Cl– vs NCS–) and the 2-R-allyl substituents on the cathodic behavior was explored. Both ligands play a significant role at all stages of the reduction path. While 2e–-reduced [Mo(η3-allyl)(6,6′-dmbipy)(CO)2]− is inert when it is ECE-generated from [Mo(η3-allyl)(6,6′-dmbipy)(CO)2(NCS)], the Cl– ligand promotes Mo–Mo dimerization by facilitating the nucleophilic attack of [Mo(η3-allyl)(6,6′-dmbipy)(CO)2]− at the parent complex at ambient temperature. The replacement of the allyl ligand by 2-methallyl has a similar effect. The Cl–/2-methallyl ligand assembly destabilizes even primary radical anions of the complex containing the strongly π-accepting pTol-Bian ligand. Under argon, the cathodic paths of [Mo(η3-2-R-allyl)(6,6′-dmbipy)(CO)2Cl] terminate at ambient temperature with 5-coordinate [Mo(6,6′-dmbipy)(CO)3]2– instead of [Mo(η3-2-R-allyl)(6,6′-dmbipy)(CO)2]−, which is stabilized in chilled electrolyte. [Mo(η3-allyl)(6,6′-dmbipy)(CO)2]− catalyzes CO2 reduction only when it is generated at the second cathodic wave of the parent complex, while [Mo(η3-2-methallyl)(6,6′-dmbipy)(CO)2]− is already moderately active at the first cathodic wave. This behavior is fully consistent with absent dimerization under argon on the cyclic voltammetric time scale. The electrocatalytic generation of CO and formate is hampered by the irreversible formation of anionic tricarbonyl complexes replacing reactive [Mo(η3-2-methallyl)(6,6′-dmbipy)(CO)2]2 along the cathodic route.


■ INTRODUCTION
There is a strong interest in the electrocatalytic reduction of CO 2 that offers a sustainable route to a variety of valuable chemical feedstocks for organic synthesis or chemical fuel. Transition-metal complexes have been identified as highly effective catalysts for the 2e − reduction of CO 2 , allowing one to take advantage of energy-saving proton-coupled pathways. 1,2 The original reports have mostly focused on complexes based on rare and precious metals, such as rhenium in [Re(bipy)-(CO) 3 Cl] (bipy = 2,2′-bipyridine), where the active catalyst is the 2e − -reduced 5-coordinate anion [Re(bipy)(CO) 3 ] − . 3−7 The costs associated with such materials directed current research efforts toward Earth-abundant metals, such as Mn.
[Mn(bipy)(CO) 3 ] − has only recently been identified as a catalyst in the presence of small amounts of Brønsted acids. 8−11 Although catalysts with impressive performance based on Earth-abundant first-row transition metals, such as Fe, Co and Ni, are now widely known, 12 much less attention has been paid to the Group 6 metals (Cr, Mo, W).
Thus, the ultimate goals of the study were to probe (i) the steric and electronic consequences of allylic methyl substitution on the cathodic path, (ii) the effect of the Cl − ligand in comparison to SCN − on the structures and reactivity of the reduced complexes, and (iii) the effect of the alternative coordination sphere including a stronger π-acceptor redoxactive ligand. At the same time, the peculiar secondary reactivity accompanying the dimerization step along the cathodic path at ambient temperature was further explored to assign the ultimate reduction products.

■ EXPERIMENTAL SECTION
Materials and Methods. All synthetic and electrochemical operations were carried out under an atmosphere of dry argon gas using standard Schlenk techniques. Tetrahydrofuran (THF) was freshly distilled under dry argon from ketyl radicals derived from the reaction of metallic Na and benzophenone, butyronitrile (PrCN) and dichloromethane (DCM) were distilled from CaH 2 , and acetonitrile (MeCN) was distilled from P 2 O 5 . The supporting electrolyte, Bu 4 NPF 6 (Acros Organics), was recrystallized twice from ethanol and dried under vacuum at 373 K for 5 h. Just prior to the experiment, the electrolyte was dried again overnight at 373 K. The precursor complexes, [Mo(η 3 -2-R-allyl)(MeCN) 2 (CO) 2 Cl] (R = H, CH 3 ), were prepared in good yields by a slight modification of the literature procedures. 27 The ligand pTol-Bian was prepared according to a literature procedure involving the condensation reaction of acenaphthenequinone and 2,6-dimethylaniline. 28 All other compounds were purchased from Sigma-Aldrich and used without further purification. The target complexes were prepared by facile thermal substitution of the labile MeCN ligands in the precursor complexes. The purity and identity of the final products were confirmed by infrared and NMR spectroscopy and single-crystal X-ray diffraction. 1 H NMR spectra were recorded on a 400 MHz Bruker NanoBay spectrometer. Elemental analyses were carried out by Medac Ltd. 2 Cl] (0.62 mmol, 0.2 g) in dry DCM (10 mL) was mixed under dry argon with a solution of the appropriate α-diimine ligand (typically 0.65 mmol) in dry DCM (10 mL). The mixture was stirred for 4 h, and then the volume was reduced by half. The crude product was precipitated by slow addition of hexane (10 × 5 mL). Roughly 100 mg of the precipitate was recovered by inert filtration and washed with cold hexane (2 × 10 mL). Spectroscopically pure samples were prepared by column chromatography on silica, using either DCM/hexane (9/1, v/v) or DCM/diethyl ether (9/1, v/v) as the eluent, where necessary. Following the purification, yields ranged between 15 and 50%. Crystals for X-ray analysis were grown by slow evaporation of DCM.
Cyclic Voltammetry. Cyclic voltammograms of complexes 1−3 were recorded with a Metrohm Autolab PGSTAT302N potentiostat operated with the NOVA 2.14 software. The airtight singlecompartment electrochemical cell housed a Pt-microdisk working electrode (active area of 0.4 mm 2 ) polished with 0.25 μm diamond paste (Kemet), a coiled-Pt-wire counter electrode, and a coiled-Agwire pseudoreference electrode. All values are reported vs the ferrocene/ferrocenium (Fc/Fc + ) redox couple, which served as the internal standard for most measurements and was added just before the final potential sweep. Where necessary, decamethylferrocene (Fc*/Fc* + ) served this purpose in order to avoid overlap with the nearby Mo(II)/Mo(III) oxidation. In THF, the value of E 1/2 (Fc*/ Fc* + ) = −0.48 V vs Fc/Fc + has been determined for this work. Solutions contained 10 −1 M Bu 4 NPF 6 and 10 −3 M analyte.
IR Spectroelectrochemistry. IR spectroelectrochemical experiments were performed using a Bruker Vertex 70v FT-IR spectrometer. An internal DLaTGS detector and an external Bio-RAD FTS 60 MCT detector (linked to the spectrometer and housing the cryostat) served for measurements at T = 298 and 223 K, respectively. The in situ electrolyses at ambient temperature were conducted using an airtight OTTLE cell. 32 The cell was equipped with Pt-minigrid (32 wires/cm) working and auxiliary electrodes, an Ag-microwire pseudoreference electrode, and optically transparent CaF 2 windows. The course of the spectroelectrochemical experiment was monitored by thin-layer cyclic voltammetry. The electrode potential control during the thin-layer CV was achieved using a PalmSens EmStat3 potentiostat, operated with PSTrace5 software. Low-temperature spectroelectrochemical measurements were carried out with a cryostatted OTTLE cell of a similar design. 33 Solutions contained 3 × 10 −1 M Bu 4 NPF 6 and 3 × 10 −3 M analyte.
Organometallics pubs.acs.org/Organometallics Article Computational Studies. Density functional theory (DFT) calculations 34 were performed using the Amsterdam Density Functional (ADF) program. 35−37 Geometries were optimized without symmetry constraints using the local density approximation (LDA) of the correlation energy (Vosko−Wilk−Nusair) 38 and the generalized gradient approximation (Becke's 39 exchange and Perdew's 40,41 correlation functionals) with gradient correction. Unrestricted calculations were performed for open-shell complexes. Solvent (THF) was considered in all geometry optimizations and singlepoint calculations, using the COSMO approach implemented in ADF. Relativistic effects were treated with the ZORA approximation. 42 Triple-ζ Slater-type orbitals (STOs) were used to describe all of the valence electrons of H, O, C, N, Cl, and Mo. A set of two polarization functions was added to H (single ζ 2s, 2p), O, C, N, and Cl (single ζ, 3d, 4f), and Mo (5d, 4f). Frequency calculations were performed to obtain the vibrational spectra and to check that intermediates were minima in the potential-energy surface. Three-dimensional representations of the structures and molecular orbitals were obtained with Chemcraft. 43 ■ RESULTS AND DISCUSSION Characterization and Crystal Structure Analysis. In THF, the IR spectra of complexes 1−3 exhibit two ν(CO) bands. For complexes 1 and 2, these absorption bands are almost identical in terms of both the intensity pattern and wavenumbers: viz., 1945 and 1861 cm −1 . In comparison with the reference, [Mo(η 3 -allyl)(6,6′-dmbipy)(CO) 2 (NCS)] (1948, 1866 cm −1 ), the absorption bands are slightly shifted to smaller wavenumbers, which reflects the increased π-backdonation experienced by the CO ligands upon the replacement of NCS − by the stronger π-donor Cl − . Finally, in complex 3, the ν(CO) bands are observed at 1956 and 1886 cm −1 , reflecting the increased acceptance from the pTol-Bian ligand in comparison with 6,6′-dmbipy in complex 2.
The structures of 1·CH 2 Cl 2 , 2, and 3 are presented in Figure  1. Crystallographic data and selected bond lengths are summarized in Tables S1 and S2 in the Supporting Information. All three complexes adopt the type A pseudooctahedral (equatorial) structure, which has been observed for the [Mo(η 3 -allyl)(x,x′-dmbipy)(CO) 2 (NCS)] (x = 4−6) series, 21 with both donor nitrogen atoms of the chelating ligand N∩N in positions trans to the CO ligands. While most of the bipy complexes known exhibit this arrangement, it is worth noting that this structure is observed here for the first time in a complex of an N-aryl-Bian ligand. Indeed, all the analogous complexes with the large and strong π-acceptor, Naryl-Bian, always adopt the less symmetrical type B (axial) structure. 20,44−48 As has been widely observed previously, in all three complexes, the open face of the allyl ligand lies over the CO ligands (i.e., the endo isomer is preferred).
The Mo−Cl bond lengths in the three complexes show a trend reflecting subtle variations of the chloride environment: 2 (2.5145(15) Å) > 1·CH 2 Cl 2 (2.4914(8) Å) > 3 (2.4873(7) Å). In all three complexes, the central allyl carbon (meso-C) is closer to the metal center than are the terminal C atoms. For instance, in 1·CH 2 Cl 2 , this distance is 2. DFT calculations, 34 using the ADF program, 35−37 were performed on the parent structures 1−3 and all their possible derivatives described below. They have revealed that the equatorial (type A) isomer is indeed always preferred over the axial (type B) isomer. The energy difference (in kcal mol −1 ) increases on going from 1 (5.23) to 2 (7.92), reflecting the extra stabilization of the equatorial isomer induced by the influence of the extra methyl group in 2-methallyl (Table S3 in the Supporting Information).
The energy difference between the two isomers of 3 is only 1.72 kcal mol −1 . This is the first known complex of an N-aryl-Bian ligand that does not prefer the axial isomer and is also the first example of a 2-methallyl complex with a N-aryl-Bian-type ligand. This unusual arrangement reflects the large repulsion between the methyl substituents of pTol-Bian and the 2methyl substituent of the allylic ligand, which overcomes the tendency to avoid cis orientation between the CO and Ndonor atoms and the steric effects occurring in the equatorial isomers.
The structural parameters (Table S2 in the Supporting Information) are well reproduced by DFT calculations (Table  S4 in the Supporting Information).
Cyclic Voltammetry. Cyclic voltammetry of 1−3 was conducted in argon-saturated THF/Bu 4 NPF 6 ( Figures 2 and 3 and Figure S12 in the Supporting Information) and PrCN/ Bu 4 NPF 6 (Figures S1−S3 and S13 in the Supporting Information) at 298 or 195 K on a Pt-microdisk electrode. The redox potentials determined for 1−3 are summarized in Table 1.
In the negative potential region, there is a reversible 6,6′dmbipy-based reduction (R1) at E 1/2 = −2.04 V (THF) or −2.03 V (PrCN), producing the radical anion [Mo(η 3allyl)(6,6′-dmbipy)(CO) 2 Cl] •− ([1] •− ). As is the case for the NCS − progenitor, there is no evidence for the formation of the 5-coordinate anion [Mo(η 3 -allyl)(6,6′-dmbipy)(CO) 2 is then seen on the reverse anodic scan as a new anodic wave O1′ at E p,a = −1.74 V. The final detectable cathodic process, R2′, at E p,c = −2.82 V (THF) or −2.79 V (PrCN), corresponds to the partly reversible reduction of the 5coordinate anion to the 5-coordinate dianion [1-A] 2− . The cathodic behavior hardly changes at low temperature (195 K), although the R2′ wave becomes fully reversible and slightly shifts to In the positive potential region, 2 also undergoes a reversible metal-based oxidation to [2] + at E 1/2 = 0.06 V, which is less positively shifted than for 1 due to the stronger electron donation from the 2-methallyl group stabilizing 2 + . The cathodic behavior of 2 strongly differs from that of 1, as the initial reduction in THF is a totally irreversible 2e − (ECE) process occurring at E p,c = −2.25 V (THF) or −2.14 V (PrCN). In comparison to 1 in THF, this corresponds to a ca. 150 mV negative shift of the parent reduction potential, as the replacement of allyl with 2-methallyl increases the LUMO energy. A similar negative potential shift was observed following the replacement of the bipy ligand with 4,4′dmbipy. 21,49 The anodic wave O1′, assigned to the oxidation of the 5coordinate anion [2-A] − , is observed at E p,a = −1.83 V (THF) or −1.71 V (PrCN) on the reverse scan that started directly beyond R1. There is only one other detectable cathodic wave, R2′, which is also shifted to the more negative potential of E p,c = −2.98 V (THF) or −2.89 V (PrCN) and corresponds to the 1e − reduction of [2-A] − formed at R2. This behavior resembles that of [Mo(η 3 -allyl)(CO) 2 (4,4′-dmbipy)(NCS)]; 21 however, in this case there is no follow-up reduction of the dimer species [2-D] on the (sub)second CV time scale (i.e., no R(D) wave is detected). This means either that the dimer is reduced at the same electrode potential as for the parent complex (R1) (cf. [Mn(iPr-dab)(CO) 3 Br] 24 ) or that the ultimate dimerization reaction (ECEC) is inhibited or too slow on the CV time scale to be observed. The first option is highly unlikely, given the large (almost 500 mV) separation between R1 and R(D) determined for the closely related complexes with the 4,4′ and 5,5′-dmbipy ligands. 21 At T = 195 K, the initial R1 wave of 2 becomes fully reversible, with E 1/2 = −2.02 V (THF) or −2.07 V (PrCN). The subsequent wave, R2, at E p,c = −2.60 V (THF) and −2.66 V (PrCN), corresponds to the irreversible reduction of stable [2] •− , yielding the 5-coordinate anion [2-A] − . The latter reduces again to the corresponding 5-coordinate dianion at R2′ with E 1/2 = −2.82 V (THF) and −2.90 V (PrCN).
Finally, complex 3 also undergoes a reversible metalcentered oxidation to [3] + at E 1/2 = 0.05 V vs Fc/Fc + , testifying to the donor power of the 2-methallyl and Cl − ligands, capable of stabilizing the formal Mo(III) oxidation state, despite the significantly increased π-acceptor capacity of the pTol-Bian ligand in comparison to 6,6′-dmbipy. This anodic behavior is quite remarkable when it is compared to closely related reference systems, such as [Mo(η 3 -allyl)(2,6xylyl-Bian)(CO) 2 (NCS)] that becomes irreversibly oxidized at ca. 0.6 V vs Fc/Fc + , a positive potential shift of more than 500 mV. 20 Then, 3 is reversibly reduced to [3] •− at much less negative potentials in comparison to 1 or 2: viz., E 1/2 = −1.34 V (THF) and −1.32 V (PrCN). This marked stabilization of the LUMO of 3 is fully consistent with the increased πacceptor capacity of the pTol-Bian ligand. However, this reduction potential is still more negative than the value determined for the above Mo(2,6-xylyl-Bian)(NCS) reference  At T = 195 K, the CV response of 3 at negative potentials closely resembles the courses recorded for 1 and 2. In comparison to the scans at room temperature, R1 shows a totally reversible shape comparable with that of the internal ferrocene standard. The irreversible wave R2 due to [3] •− reduction shifts slightly negatively to −2.03 V in THF and becomes well developed in both THF and PrCN. This temperature-dependent behavior indicates some reorientation of [3] •− at the cathodic surface at ambient temperature. This cathodic step generates the genuine 5-coordinate anion, [3-A] − , which is oxidized on the reverse anodic scan at O1′, E p,a = −0.99 V (THF) or −1.03 V (PrCN), and reduced at R2′ to the corresponding stable dianion. The much larger separation between R2 and R2′ for 3 in comparison to 1 and 2 (Table 1)  Computational Studies. DFT calculations were performed to determine the ground-state geometries, electronic structures and energies, and to reproduce the vibrational spectra of complexes 1−3 and their oxidized and reduced forms introduced in the preceding CV section. The geometryoptimized structures are depicted in Figure 4 for 2 and in Figures S4 and S5 in the Supporting Information for 1 and 3, respectively. The equatorial isomer is the most stable one for all of the neutral parent complexes, as discussed above.
The calculated IR ν(CO) wavenumbers are practically identical (Table S5 in the Supporting Information) for 1 and 2, with the symmetric ν(CO) modes at 1878 and 1879 cm −1 , respectively, and the antisymmetric modes at 1797 cm −1 in both cases. The experimental wavenumbers 1945 and 1861 cm −1 (in THF) for 1 are converted into 1886 and 1805 cm −1 by the application of a 0.97 scaling factor, in good agreement with the calculated values. For 3, the calculated wavenumbers are somewhat larger, with the symmetric mode at 1891 cm −1 and the antisymmetric mode at 1821 cm −1 . It is important to apply the scaling factor to calculated ν(CO) values for identification purposes of all studied 6-coordinate complexes ( Table 2). It is redundant for the strongly π-delocalized 5coordinate anions, [X-A] − .
As depicted in Figure 5 for 2 and Figures S6 and S7 in the Supporting Information for 1 and 3, respectively, the HOMOs of 1−3 have a strong contribution from the metal, being bonding between the metal and the π-acceptor carbonyls, but π-antibonding between the Mo center and the π-donor chloride ligand. Hence, the 1e − oxidation reaction, described in the preceding CV section, converts formally Mo(II) to Mo(III). The HOMO-1 and HOMO-2 do not differ significantly. Conversely, the LUMO, LUMO+1, and LUMO +2 are almost completely localized on the 6,6′-dmbipy ligand in 1 and 2 but only partially localized on the pTol-Bian ligand in 3, indicating that the initial reduction step influences these ligands.
The localization of the initial reduction at the α-diimine ligand is reflected by the shortening of the 6,6′-dmbipy interring bond, for example, from 1.481 to 1.433 Å in [1] •− and from 1.481 to 1.442 Å in [2] •− (see Table S4). On the other hand, the oxidation process does not affect this bond, which remains at 1.480 Å in both [1]  The loss of the chloride ligand from the primary 6coordinate radical anion affords the 5-coordinate radical, [X-R], which in principle may adopt either a square-planar geometry (SP), derived from the equatorial isomer, or a trigonal-bipyramidal geometry (TBP), derived from the axial isomer. 21 In these species, the SP geometry minimizes steric constraints between substituents and is preferred for [1-R], [2-R], and [3-R]. Only for [1-R] is a TBP geometry less stable by 1.56 kcal mol −1 also possible, suggesting that the 2-methyl group in the allyl plays a relevant role in the control of the metal coordination environment.
The direct reduction of the 5-coordinate radicals affords the active 2e − catalysts for these systems, the 5-coordinate anions [X-A] − . The latter may, in principle, exist in either a closedshell singlet (diamagnetic) or an open-shell triplet (paramagnetic) state. As in the previous cases, 21 the former state is more stable by a high margin: 16.48 kcal mol −1 (1), 19.17 kcal mol −1 (2), and 11.77 kcal mol −1 (3). The three 5-coordinate anions adopt an SP geometry.
These SP 5-coordinate anions may react with coordinating solvents such as PrCN to form new 6-coordinate anionic complexes. The equatorial isomer forms easily from the SP precursor, by accepting the ligand electrons in the well oriented LUMO+1 ( Figure S9 in the Supporting Information). However, this geometry is only found for [1-PrCN] − . Both [2-PrCN] − and [3-PrCN] − adopt the axial isomer geometry (it is shown in Figure S5 for the latter). These derivatives are not very stable, probably due to the negative charge in the acceptor fragment. In particular, the large reorganization of the Mo(η 3 -2-methallyl)(6,6′-dmbipy)(CO) 2 fragment required to form [2-PrCN] − makes the formation of this species very unlikely. The steric constraints imposed by the pTol-Bian and 2methallyl ligands seem to be more important and the fragments barely reorganize when the sixth ligand (Cl or PrCN) adds, therefore allowing for the formation of the solvent complex.
For the 6,6′-dmbipy complexes 1 and 2, the loss of the chloride ligand from the radical anions, forming the 5-   Table S5 in the Supporting Information). The origin of this phenomenon has already been discussed. 21    The two complexes of 6,6′-dmbipy, 1 and 2, formed dimers with a long and weak Mo−Mo bond (3.886 Å in [1-D] and 3.955 Å in [2-D]), as shown in Figure 4 and Figure S4 in the Supporting Information. Their IR spectra are characterized by three strong ν(CO) bands appearing at 1855, 1847, and 1782 cm −1 for both Mo−Mo-bound dimers. No dimer of this type could be obtained from calculations for 3.
IR Spectroelectrochemistry at Low Temperature. IR spectroelectrochemistry has been proven to be an invaluable tool for unraveling the mechanistic details of different cathodic paths. The data presented in this section support the major insights gained from the cyclic voltammograms and DFT calculations in the preceding sections. The IR ν(CO) absorption data recorded for parent 1−3, their oxidized and reduced products, and key reference compounds are summarized in Table 2 (and complemented with relevant DFT data taken from Table S5 in the Supporting Information). It is convenient to begin the discussion with the cathodic paths of 1−3, determined at low temperature (223 K), as these results are the most straightforward to assign.
Reducing     [1] •− is unstable on the SEC time scale even at low temperature, despite the fully reversible cathodic wave R1 in the cyclic voltammogram ( Figure 2). This is a consequence of the strong π-donation from the Cl − ligand, which is less tunable than that of the isothiocyanate ligand via In contrast to 1, reducing 2-methallyl complex 2 in PrCN at the cathodic wave R1 under the same low-temperature conditions (Figure 8)  This red ν(CO) shift reflects the increased electron density at the CO ligands imposed by the 2-methallyl ligand, which has a stronger effect in [2-A] − than in parent 2 due to the widely delocalized nature of the π-bonding in the 5-coordinate anion (see the preceding DFT section). These IR SEC results are consistent with the different CV behaviors of 1 and 2 ( Figures  2 and 3, respectively), clearly confirming that the methylated allyl group significantly destabilizes the 1e − -reduced intermediate [2] •− . At low temperature, this results in the rapid formation of stable [2-A] − already at R1 via [2-R] (Scheme 1); the dimerization is inhibited. At ambient temperature, however, the cathodic course in the thin-layer cell becomes more complex, as described in the next section.
Perhaps most surprising in the studied series is the lowtemperature cathodic behavior of 3 (ν(CO): 1951, 1876 cm −1 ). On the basis of the recorded CV responses and the strongly π-accepting nature of the N-aryl-Bian ligand, one would expect the corresponding radical anion, [3] •− , to persist in the electrolyte. However, the initial reduction of 3 at R1 generated a mixture of two species absorbing in the ν(CO) region (Figure 9), akin to the case for 1. On comparison with the reference complexes (Table 2) (Figure 3d), and the CV responses of 3 in PrCN do not show any substantial difference from this behavior ( Figure S3 in the Supporting Information). Obviously, the strong coordinating ability of the PrCN solvent needs to be considered. The solvento anion  − is formed already at R1 (Figure 9), most likely from an equilibrium between [3] •− and [3-R] reducible to [3-A] − that coordinates a donor solvent molecule. Alternatively, [3-R] coordinates PrCN prior to the ultimate reduction. Such a cathodic behavior has been well documented: for example, for [Re(bipy)(CO) 3 Cl] in PrCN. 4 IR Spectroelectrochemistry at Ambient Temperature. In line with the ordinary reversible anodic cyclic voltammetric scans, both studied Mo−2-methallyl complexes 2 ( Figure  S10b, Supporting Information) and 3 ( Figure 10) are oxidized on the SEC time scale to the corresponding stable, formally Mo(III) cationic products. On the other hand, [1] + is unstable at room temperature ( Figure S10a in the Supporting Information) and slowly decomposes (decarbonylates) during the electrolysis. The accompanying blue shifts of the ν(CO) bands (summarized in Table 2 and reflected in the DFTcalculated values) to larger wavenumbers are significant. They comply with the depopulation of the largely Cl−Mo-based HOMO of the parent complexes ( Figure 5 and Figures S6 and  S7 in the Supporting Information), having the expected large effect on the degree of CO π-back-donation that decreases in the formally Mo(III) products. The reversible oxidation of complex 3 is truly remarkable. The Mo−bipy bond lengths barely change upon oxidation, while the internal bonds in pTol-Bian display larger changes. Therefore, the stability of   Conducting IR SEC in the negative potential region at ambient temperature in THF/Bu 4 NPF 6 reveals additional complexity in the cathodic paths of 1 and 2 in comparison to the straightforward cathodic behavior seen at 223 K (see the preceding section). The reduction of 1 at R1 (Figure 11) leads to a mixture of products, a very minor component of which is the radical anion [1] •− . Initially, the mixture contains two major secondary products that can be identified from their ν(CO) stretching wavenumbers.
Electrochemical reduction of 3 in THF at ambient temperature ( Figure 13) exhibits a cathodic behavior similar to that encountered for this complex in PrCN at low temperature. The initial reduction at R1 produces once more the unstable radical anion [3] •− , transforming to the solvated 6-coordinate anion [3-THF] − (ν(CO): 1897, 1800 cm −1 ). Dimer [3-D] was neither observed on the SEC time scale nor could be calculated using approaches that led to dimers [X-D] for 1 and 2. This is likely a result of the steric hindrance from the bulky pTol-Bian ligand destabilizing the dimer conformation.  The transient appearance of dimer [1-D] along the cathodic path of 1 might be considered highly surprising, as results (CV, SEC, DFT) from the previous series [Mo(η 3 -allyl)(x,x′dmbipy)(CO) 2 (NCS)] (x,x′ = 4−6) indicated that the most sterically demanding 6,6′-dmbipy ligand stabilized the 5coordinate anion, [1-A] − , toward dimerization. This study, however, reveals that the true story is more complicated. The dimer is expected to form in a zero-electron coupling reaction between the 5-coordinate anion and the yet nonreduced parent complex (Scheme 1). Thus, the proclivity of the dimer formation is dependent on several factors. The first is the inertness of the parent complex itself. If the Mo−X (X = Cl, NCS) bond in the parent complex is weaker, then it is obviously more susceptible to this form of nucleophilic attack on the appropriate time scale, and the dimer therefore has a higher chance to form. We conclude firmly that the Mo−Cl bonds in 1 and 2 are weaker than the Mo−N(CS) bond in the reference complex [Mo(η 3 -allyl)(6,6′-dmbipy)(CO) 2 (NCS)]. Second, the stability of the 1e − -reduced intermediate, i.e. the radical anions [1] •− and [2] •− , also plays a role. The more reactive Mo−Cl bond facilitates a greater amount of the 5coordinate anions being available to react with the parent during the initial cathodic step, driving the reduction mechanism more along the pathway involving the dimer. This conclusion underlines the need to determine the exact mechanism of the rapid concomitant conversion of [1-D] to [Mo(6,6′-dmbipy)(CO) 3 Y] − on the time scale of IR spectroelectrochemistry. The cathodic pathways described in this study have a strong effect on the results gathered during electrochemical reduction under a CO 2 atmosphere, which are presented in the next section.
Cyclic Voltammetry and IR Spectroelectrochemistry under a CO 2 Atmosphere. The CV studies of 1 and 2 in THF were repeated under an atmosphere of CO 2 ( Figure 14) in order to probe for any catalytic activity of 5-coordinate anions [X-A] − (X = 1, 2) along the cathodic paths toward the 2e − catalytic reduction of CO 2 . For complex 1, the 1e − cathodic wave R1 (cf. Figure 2) remains unchanged, producing stable [1] •− . However, catalytic current enhancement is observed at the R2 wave, where [1-A] − is produced via the     (Figure 7). For the CV of complex 2, the behavior is different. Interestingly, a modest increase in the cathodic current is observed already at R1, which most likely corresponds to the catalytic reduction of CO 2 by [2-A] − that has already been identified as the dominant product at this wave on the CV time scale (Figure 3a). Correspondingly, the anodic counter wave O1′ is absent on the reverse anodic scan starting directly beyond R1. However, the bulk of the catalytic current enhancement is not seen until slightly more negative potentials are reached, where also a new quasi-reversible wave is detected at ca. −2.7 V. The latter may correspond to reduction of an unreactive intermediate adduct of [ 25,54 Under the same conditions, 3 was not catalytically active toward the CO 2 substrate along the cathodic CV scan, which is consistent with previous observations on the poor catalytic performance of a closely related Mo−allyl complex with 2,6dimethylphenyl-Bian. 20 Indeed, IR spectroelectrochemistry in the preceding section has provided no evidence for the cathodic generation of 5-coordinate [3-A] − undergoing an electrophilic attack by CO 2 .

■ CONCLUSIONS
This work strongly supports our ongoing efforts to characterize the fascinating redox reactivity of the formally Mo(II) complexes [Mo(η 3 -2-R-allyl)(α-diimine)(CO) 2 X] (X = halide, pseudohalide). This study based on [Mo(η 3 -2-R-allyl)(6,6′dmbipy)(CO) 2 Cl] (R = H, CH 3 ) has resulted in several important discoveries. First, the interplay of steric and electronic effects between the various ligands (X = halide, pseudohalide; α-diimine; R-allyl) is more complex than was originally anticipated; it is also important to consider the effects of different time scales, in order to fully appreciate the whole situation. For instance, the replacement of the NCS − ligand with Cl − initially (when analyzing the CV scans) does not seem to affect the cathodic path strongly. On the other hand, IR SEC has revealed that there is actually a strong effect on the stability of the primary radical anions at ambient temperature and the reactivity of the ECE-generated, 2e −reduced 5-coordinate anions toward the parent complexes, resulting in Mo−Mo dimerization. In contrast to the dimethylbipy substitution in the 6,6′-position, the substitution at the meso-carbon of the allyl ligand results in a strongly decreased stability of the radical anions toward the cleavage of the Mo− Cl bond. The new Cl − and 2-methallyl ligand assembly studied in this work also eliminates the usually stabilizing influence of the π-acceptor pTol-Bian ligand on the singly reduced species, resulting not only in a different parent molecular structure (Atype) in comparison to other Mo−N-aryl-Bian complexes but also in the increased reactivity of the radical anion (even at low temperature).
■ ASSOCIATED CONTENT * sı Supporting Information