Redox-Switchable Behavior of Transition-Metal Complexes Supported by Amino-Decorated N-Heterocyclic Carbenes

The coordination chemistry of the N-heterocyclic carbene ligand IMes(NMe2)2, derived from the well-known IMes ligand by substitution of the carbenic heterocycle with two dimethylamino groups, was investigated with d6 [Mn(I), Fe(II)], d8 [Rh(I)], and d10 [Cu(I)] transition-metal centers. The redox behavior of the resulting organometallic complexes was studied through a combined experimental/theoretical study, involving electrochemistry, EPR spectroscopy, and DFT calculations. While the complexes [CuCl(IMes(NMe2)2)], [RhCl(COD)(IMes(NMe2)2)], and [FeCp(CO)2 (IMes(NMe2)2)](BF4) exhibit two oxidation waves, the first oxidation wave is fully reversible but only for the first complex the second oxidation wave is reversible. The mono-oxidation event for these complexes occurs on the NHC ligand, with a spin density mainly located on the diaminoethylene NHC-backbone, and has a dramatic effect on the donating properties of the NHC ligand. Conversely, as the Mn(I) center in the complex [MnCp(CO)2 ((IMes(NMe2)2)] is easily oxidizable, the latter complex is first oxidized on the metal center to form the corresponding cationic Mn(II) complex, and the NHC ligand is oxidized in a second reversible oxidation wave.

We entered this area of chemistry in 2018, by reporting preliminary studies on the redox behavior of the thiourea A NMe2 and A (NMe2)2 , derived from the mono-substituted 4-(dimethylamino)imidazol-2-ylidene IMes NMe2 and the di-substituted 4,5-bis(dimethylamino) imidazol-2-ylidene IMes (NMe2)2 , respectively ( Figure 1) [35]. While we studied the latter amino-derived NHCs mainly in the direction of organometallic catalysis with remarkable results [36][37][38][39][40], we actually initiated this program by considering the analogy between the NHC backbones and the electron-rich tri-and tetra-aminoethylene moieties, which are well-known electron-rich reducing agents [41,42]. We indeed show that both thioureas are redox-active, due to the presence of the dimethylamino groups onto their backbone. While A NMe2 can be oxidized mono-electronically to the corresponding persistent radical-cation [A NMe2 ] •+ , A (NMe2)2 exhibits two oxidized forms, namely, the radical cation [A (NMe2)2 ] •+ and the diamagnetic dication [A (NMe2)2 ] 2+ and both are remarkably stable. Thus, we decided to extend this study to the case of the transition-metal complexes of the diamino-substituted IMes (NMe2)2 ligand (1), and we report herein that this NHC ligand exhibits redox-active behavior with a variety of metal centers. of redox-switchable catalysts by a strong modification of the electronic properties through a redox stimulus [7,34], or permitted the access to multiple oxidation states of the complexes. We entered this area of chemistry in 2018, by reporting preliminary studies on the redox behavior of the thiourea A NMe 2 and A (NMe 2 ) 2, derived from the mono-substituted 4-(dimethylamino)imidazol-2-ylidene IMes NMe 2 and the di-substituted 4,5bis(dimethylamino)imidazol-2-ylidene IMes (NMe 2 ) 2, respectively ( Figure 1) [35]. While we studied the latter amino-derived NHCs mainly in the direction of organometallic catalysis with remarkable results [36][37][38][39][40], we actually initiated this program by considering the analogy between the NHC backbones and the electron-rich tri-and tetra-aminoethylene moieties, which are well-known electron-rich reducing agents [41,42]. We indeed show that both thioureas are redox-active, due to the presence of the dimethylamino groups onto their backbone. While A NMe 2 can be oxidized mono-electronically to the corresponding persistent radical-cation [A NMe 2] •+ , A (NMe 2 ) 2 exhibits two oxidized forms, namely, the radical cation [A (NMe 2 ) 2] •+ and the diamagnetic dication [A (NMe 2 ) 2] 2+ and both are remarkably stable. Thus, we decided to extend this study to the case of the transitionmetal complexes of the diamino-substituted IMes (NMe 2 ) 2 ligand (1), and we report herein that this NHC ligand exhibits redox-active behavior with a variety of metal centers.  2 and A (NMe 2 ) 2 derived from the amino-decorated IMes NMe 2 and IMes (NMe 2 ) 2, respectively, and of the transition-metal complexes of IMes (NMe 2 ) 2, subject of the present study.

Redox Properties of IMes (NMe 2 ) 2 Copper(I) Complex 2
We started our investigation by studying the behavior of the copper(I) complex [CuCl(1)] (2), which is synthesized by reacting the free stable NHC 1, generated from 1·HOTf by deprotonation with one equivalent of potassium bis(trimethylsilyl)amide (KHMDS), with an excess of CuCl and NaCl at room temperature ( Figure 2a). These unusual conditions are necessary to avoid the competitive formation of the cationic homoleptic complex [Cu(1)2] + under standard conditions (1 equivalent of CuCl) [43]. The 1 H NMR spectrum of 2 is particularly simple, as only four singlets are observed, due to the C2v symmetry of the molecule, and to the equivalence of the twelve protons of the NMe2 substituents, as already observed in the previous complexes of ligand 1 [36,39,40]. The redox behavior of complex 2 was first examined by cyclic voltammetry (CV) measurements (Figure 2b), and compared to the parent unsubstituted complex Figure 1. Redox behaviors of the thiourea A NMe2 and A (NMe2)2 derived from the amino-decorated IMes NMe2 and IMes (NMe2)2 , respectively, and of the transition-metal complexes of IMes (NMe2)2 , subject of the present study.

Redox Properties of IMes (NMe2)2 Copper(I) Complex 2
We started our investigation by studying the behavior of the copper(I) complex [CuCl(1)] (2), which is synthesized by reacting the free stable NHC 1, generated from 1·HOTf by deprotonation with one equivalent of potassium bis(trimethylsilyl)amide (KHMDS), with an excess of CuCl and NaCl at room temperature ( Figure 2a). These unusual conditions are necessary to avoid the competitive formation of the cationic homoleptic complex [Cu(1) 2 ] + under standard conditions (1 equivalent of CuCl) [43]. The 1 H NMR spectrum of 2 is particularly simple, as only four singlets are observed, due to the C 2v symmetry of the molecule, and to the equivalence of the twelve protons of the NMe 2 substituents, as already observed in the previous complexes of ligand 1 [36,39,40]. The redox behavior of complex 2 was first examined by cyclic voltammetry (CV) measurements (Figure 2b), and compared to the parent unsubstituted complex [CuCl(IMes)] ( Figure S1). While complex [CuCl(IMes)] displays no clear oxidation wave, the bis-amino-substituted complex 2 features two fully reversible one-electron transfer processes at E 1/2 = +0.71 V, and +1.11 V vs. SCE. This CV pattern is highly similar to that observed for the corresponding thiourea A (NMe2)2 [35], with the exception that both oxidation processes in 2 occur at higher potentials. This suggests that the redox events in complex 2 might also be located on the NHC ligand, and prompted us to explore in more detail the mono-and di-oxidized complexes [2]  [CuCl(IMes)] ( Figure S1). While complex [CuCl(IMes)] displays no clear oxidation wave, the bis-amino-substituted complex 2 features two fully reversible one-electron transfer processes at E1/2 = +0.71 V, and +1.11 V vs. SCE. This CV pattern is highly similar to that observed for the corresponding thiourea A (NMe 2 ) 2 [35], with the exception that both oxidation processes in 2 occur at higher potentials. This suggests that the redox events in complex 2 might also be located on the NHC ligand, and prompted us to explore in more detail the mono-and di-oxidized complexes [2] + and [2] 2+ . Complex 2 was chemically oxidized using an equimolar amount of 1,1′diacetylferrocenium tetrafluoroborate ([Fc (Ac) 2](BF4), E1/2 = +0.95 V [44]) in CH2Cl2 at −40 °C, to give a deep blue solution of the one-electron oxidized species [2] + , which was characterized immediately by EPR spectroscopy at room temperature in CH2Cl2/toluene (1:1) solution ( Figure 2c). The X-band EPR spectrum of [2] + is consistent with a delocalized organic radical (Landé factor of giso = 2.00338), and displays a symmetrical, hyperfine Complex 2 was chemically oxidized using an equimolar amount of 1,1 -diacetylferrocenium tetrafluoroborate ([Fc (Ac)2 ](BF 4 ), E 1/2 = +0.95 V [44]) in CH 2 Cl 2 at −40 • C, to give a deep blue solution of the one-electron oxidized species [2] + , which was characterized immediately by EPR spectroscopy at room temperature in CH 2 Cl 2 /toluene (1:1) solution ( Figure 2c). The X-band EPR spectrum of [2] + is consistent with a delocalized organic radical (Landé factor of g iso = 2.00338), and displays a symmetrical, hyperfine splitting pattern. The simulated spectrum fits perfectly with the experimental spectrum, and is composed mainly of hyperfine interactions between the radical and twelve equivalent protons (a H = 18.6 MHz), with the two exocyclic nitrogen atoms (a N = 16.1 MHz) and with the copper center (a Cu = 36 MHz), and, to a lower extent, by a hyperfine coupling constant with the two nitrogen atoms of the diaminocarbene unit (a N = 3.8 MHz) (Figure 2d). The optimized geometry computed by DFT for [2] •+ confirms that the copper center remains linear and dicoordinate, which is in agreement with a Cu(I) center, and shows, in particular, a planarization of the two exocyclic nitrogen atoms (Σ N = 359.5 • and 359.6 • ), with a twist of the NMe 2 plane relative to the carbenic heterocycle of ca. 34 • , most probably due to steric congestion, as already encountered in the related thiourea derivative [A (NMe2)2 ] •+ . Similarly, the SOMO orbital of [2] + is mainly localized on the bis(dimethylamino)ethylene backbone, involving C−N π-antibonding and C−C π-bonding characters, with small contributions of the carbene and copper centers (Figure 2e). The DFTcomputed hyperfine coupling (hfc) constants are consistent with the hfc values obtained from the simulation of the EPR spectrum. The DFT calculation distinguishes two groups in particular for the hydrogen atoms from the NMe 2 groups, depending on whether they are localized on the inner or outer methyl groups, and gives two distinct hfc coupling constants of a H = 22.3 MHz and a H = 18.9 MHz. Eventually, the mono-oxidized complex [2] + is shown to be stable in solution for 2-3 h at room temperature.
We then tried to generate the di-oxidized complex [2] 2+ . Due to the high potential observed for the second oxidation wave in CV (E 1/2 = +1.11 V vs. SCE), strong oxidizing agents were employed. Thus, complex 2 was reacted at a low temperature (−50 • C) with 2.5 equivalents of [N(p-BrPh) 3 ](SbF 6 ) (Magic Blue, E 1/2 = +1.16 V vs. SCE) or [NO](SbF 6 ) (E 1/2 = +1.46 V vs. SCE), but only mixtures of unidentified products were detected by NMR, and even if the reaction mixture was indeed diamagnetic, no definite conclusion could be drawn on the nature of [2] 2+ .
Overall, this first study confirms that NHC ligand 1 behaves as a redox-active ligand in organometallic complexes, and that the mono-oxidized complex [2] + is generated and characterized by EPR spectroscopy.

Redox Properties of IMes (NMe2)2 Dicarbonyl Rhodium(I) Complex 3
We then turned our attention to the rhodium carbonyl complex [RhCl(CO) 2 (1)] (3) that we had previously synthetized [36], with the aim of evaluating the overall donor properties of the NHC by means of IR spectroscopy [11].
The cyclic voltammogram of complex 3 displays a first reversible oxidation wave at E 1/2 = +0.71 V vs. SCE, and a second irreversible oxidation process at E pa = +1.26 V vs. SCE (Figure 3b), whose reversibility may be restored at higher scan rate (10 V · s −1 ). Interestingly, the first oxidation wave of 3 occurs at the same potential as for complex 2, and, hence, corresponds to an oxidation of the NHC backbone. To confirm the nature of the mono-oxidized state [3] + , we then carried out the chemical oxidation of 3 using one equivalent of 1,1 -diacetylferrocenium tetrafluoroborate in CH 2 Cl 2 at −20 • C (Figure 3a), and analyzed [3] + by EPR spectroscopy in CH 2 Cl 2 /toluene solution (1:1) at room temperature. Unlike the Cu(I) compound [2] + , the Rh(I) complex [3] + is stable under these conditions only for several minutes. Analogously to [2] + , the EPR spectrum of [3] + features a symmetrical hyperfine coupling pattern (Figure 3c), which is perfectly simulated, taking into account hyperfine interactions between the single electron and the twelve hydrogen atoms of the NMe 2 groups (a H = 18.8 MHz) and with the two exocyclic nitrogen atoms (a N = 15.6 MHz). These values are very close to those calculated for [2] + , and are corroborated by DFT calculations, which give similar hfc values for those interactions. In addition, the DFT calculations indicate two much smaller hyperfine couplings with the two endocyclic nitrogen atoms and with the 103 Rh center (I = 1 2 ), which were not resolved in the experimental spectrum. The experimental and simulated EPR spectra and the computed SOMO orbital in [3] + (Figure 3d) are, thus, in agreement with an organic radical delocalized on the bis(dimethylamino)ethylene backbone of the NHC ligand IMes (NMe2)2 , leaving the Rh(I) center untouched, as already observed for the thiourea A (NMe2)2 and the Cu(I) complex 2. Therefore, it is possible to quantify the effect of the mono-oxidation of the IMes (NMe2)2 ligand on its electronic properties by recording the IR spectra of complexes 3 and [3] + , as the average frequency of the two ν CO bands in the complexes [RhCl(CO) 2 (NHC)] in CH 2 Cl 2 is linearly correlated to the Tolman electronic parameter (TEP) value of the NHC ligand [45,46]. While the carbonyl stretching in 3  (Figure 3d) are, thus, in agreement with an organic radical delocalized on the bis(dimethylamino)ethylene backbone of the NHC ligand IMes (NMe 2 ) 2, leaving the Rh(I) center untouched, as already observed for the thiourea A (NMe 2 ) 2 and the Cu(I) complex 2. Therefore, it is possible to quantify the effect of the mono-oxidation of the IMes (NMe 2 ) 2 ligand on its electronic properties by recording the IR spectra of complexes 3 and [3] + , as the average frequency of the two νCO bands in the complexes [RhCl(CO)2(NHC)] in CH2Cl2 is linearly correlated to the Tolman electronic parameter (TEP) value of the NHC ligand [45,46]. While the carbonyl stretching in 3 are measured at 1992 and 2075 cm −1 (TEP = 2047 cm −1 ), the two bands shift to higher frequencies upon oxidation to [3] + and are observed at 2010 and 2090 cm −1 (TEP = 2060 cm −1 ). A scale of TEP values of selected, relevant NHC ligands is depicted in Figure 4. While the ligand IMes (NMe 2 ) 2 is a better donor than the regular IMes, the mono-oxidized cationic A scale of TEP values of selected, relevant NHC ligands is depicted in Figure 4. While the ligand IMes (NMe2)2 is a better donor than the regular IMes, the mono-oxidized cationic radical [IMes (NMe2)2 ] •+ is a quite weak electron-donor, and is an even weaker donor than the naphthoquinone-annelated IMes NQ (TEP = 2056.6 cm −1 ) [22,47], the (amino)(amido)carbene SIMes =O (TEP = 2058 cm −1 ) [48], and the cationic IMes (NMe3)+ (TEP = 2055.7 cm −1 ) [49]. However, it remains a greater donor than the least electron-donating NHC ligands represented by the diamidocarbene SIMes (=O)2 (TEP = 2068 cm −1 ) [50], the cationic NHC B (TEP = 2067 cm −1 ) [51], or the dicyano-substituted NHC C (TEP = 2067 cm −1 ) [52]. Interestingly, the difference in TEP values of 13 cm −1 between the two redox states of IMes (NMe2)2 Molecules 2022, 27, 3776 6 of 17 indicates that the latter constitutes a new class of redox-switchable NHC ligand, with the previous examples relying mainly on the incorporation of redox-active ferrocene-derived moiety, or of a naphthoquinone, into the NHC skeleton [7,53]. Moreover, the relatively high shift between the two oxidation states IMes (NMe2)2 / [IMes (NMe2)2 ] •+ can be explained by the fact that the redox event in IMes (NMe2)2 occurs directly onto the carbenic heterocycle and, thus, has a strong effect on the electron-donating properties of the carbenic center.

Redox Properties of Half-Sandwich IMes (NMe 2 ) 2 Fe(II) and Mn(I) Complexes [4](BF4) and 5
The study continued with the two isoelectronic d 6 , 18e-valence electron halfsandwich complexes [CpFe(CO)2 (1) [55], and was isolated in very good yield. Both complexes were fully characterized by spectroscopic and analytical methods, and their molecular structures firmly established by single-crystal X-ray diffraction analysis ( Figure 5). Beyond the expected pianostool coordination geometries around the metal and the metal−C NHC bond lengths, which are within the classical range for such complexes [54,55], the C4−C5 bond lengths are characteristic of a double C=C bond (1.35 Å) [56], and the pyramidalization of the two exocyclic nitrogen atoms (Σ N~3 50 • ) indicates that the nitrogen lone pairs do not interact with the π-system of the carbenic heterocycle, as previously observed in other complexes [36,39,40]. their molecular structures firmly established by single-crystal X-ray diffraction analysis ( Figure 5). Beyond the expected piano-stool coordination geometries around the metal and the metal−CNHC bond lengths, which are within the classical range for such complexes [54,55], the C4−C5 bond lengths are characteristic of a double C=C bond (1.35 Å) [56], and the pyramidalization of the two exocyclic nitrogen atoms (ΣN ~ 350°) indicates that the nitrogen lone pairs do not interact with the π-system of the carbenic heterocycle, as previously observed in other complexes [36,39,40]. While complexes [4](BF4) and 5 are isoelectronic, their cyclic voltammograms differ substantially ( Figure 6). The iron complex [4](BF4) shows a similar electrochemical behavior to those of previously studied derivatives 2 and 3, with a reversible oxidation wave at E1/2 = +0.84 V vs. SCE, and an irreversible second oxidation process at Epa = +1.36 V vs. SCE, which indicates that the NHC ligand 1 is still the location where the redox process takes place. Conversely, the much more electron-rich Mn(I) complex 5 exhibits three oxidation processes at E1/2 = +0.08 V, +0.75 V, and +1.33 V vs. SCE, with the two first to those of previously studied derivatives 2 and 3, with a reversible oxidation wave at E 1/2 = +0.84 V vs. SCE, and an irreversible second oxidation process at E pa = +1.36 V vs. SCE, which indicates that the NHC ligand 1 is still the location where the redox process takes place. Conversely, the much more electron-rich Mn(I) complex 5 exhibits three oxidation processes at E 1/2 = +0.08 V, +0.75 V, and +1.33 V vs. SCE, with the two first being fully reversible and the latter only partially. The potential at +0.08 V vs. SCE of the first additional oxidation wave lies in the typical range of potentials for the oxidation of Mn(I) to Mn(II) centers in a series of [CpMn(CO) 2 (NHC)] reported by us [57,58] and others [59], and, thus, corresponds to a metal-centered oxidation process. The lower potential of this process for 5 (E 1/2 = +0.08 V vs. SCE), compared to the process for the parent [CpMn(CO) 2 (IMes)] (E 1/2 = +0.16 V vs. SCE), is also consistent with the more electron-donating character of ligand 1, relative to the IMes ligand. These first assumptions are confirmed by the study of the mono-oxidized species [4] 2+ and [5] + . The Fe-complex [4] 2+ is generated chemically by oxidation with 1,1′diacetylferrocenium tetrafluoroborate, and the resulting red/purple solution analyzed by X-band EPR and IR spectroscopies at room temperature ( Figure 6). The same hyperfine coupling pattern is observed as encountered for the previous complexes [2] + and [3] + , and

These first assumptions are confirmed by the study of the mono-oxidized species [4] 2+ and [5] + . The Fe-complex [4] 2+
is generated chemically by oxidation with 1,1diacetylferrocenium tetrafluoroborate, and the resulting red/purple solution analyzed by X-band EPR and IR spectroscopies at room temperature ( Figure 6). The same hyperfine coupling pattern is observed as encountered for the previous complexes  [57].
The full reversibility of the second oxidation wave in the CV of 5 then prompted us to explore the oxidation of [5](SbF 6 ). The latter is cleanly oxidized by using one more equivalent of AgSbF 6 in CH 2 Cl 2 to give a dark blue solution, from which the dicationic complex [5](SbF 6 ) 2 is isolated as a black powder in 80% yield. In contrast to the first metal-centered oxidation of complex 5, leading to ca. 120 cm −1 high frequency shift of ν CO bands, the average stretching CO frequencies are much less affected upon the [5] + → [5] 2+ transformation (+7.5 cm −1 ), being consistent with the occurrence of electron transfer at the NHC ligand. The intense absorption at 578 nm observed in the UV-vis spectrum of [5](SbF 6 ) 2 ( Figure S3) is similar to that observed for the radical cation of [A (NMe2)2 ] •+ (590 nm), previously attributed to the electronic transition involving SOMO orbital by TD-DFT calculations [32]. Important changes are worth mentioning in the molecular structure of

This does not allow us to conclude what the spin ground state of [5] 2+ is (S = 1 or S = 0)
, and what is expected to be observed by EPR.
As it turns out, no signal is detected by recording the EPR spectra of complex [5](SbF 6 ) 2 at various temperatures (from 4.2 K to room temperature). DFT only predicted that the two single electrons are expected to be weakly magnetically coupled. The fact that no EPR signal is observed could be due to a relatively high zero-field splitting constant of the EPR spectrum of the S = 1 (typically D > 1 cm −1 ), possibly associated with very short relaxation times. These features would lead either to low-field (X-band) EPR silent S = 1 system (D > 1cm −1 ) [62], or to a very broad EPR signal (very short relaxation times) [63], which is lower in intensity than the detection limit of our EPR spectrometer.

Materials and Methods
All manipulations were performed under an inert atmosphere of dry nitrogen by using standard vacuum line and Schlenk tube techniques. Glassware was dried at 120 • C in an oven, for at least three hours. THF, diethyl ether, pentane, toluene, and CH 2 Cl 2 were dried using an innovative technology solvent purification system. Imidazolium triflate IMes (NMe2)2 ·HOTf (1·HOTf) [36], [RhCl(CO) 2 (IMes (NMe2)2 )] [36], and [CpFe(CO) 2 I] [64] were prepared according to procedures in the literature. Following one such procedure, 1,1 -diacetylferrocenium tetrafluoroborate was prepared through the oxidation of 1,1 -diacetylferrocene by benzoquinone/HBF 4 ·OEt 2 [65]. CuCl was purified by dissolution in the minimum of concentrated HCl, followed by precipitation by adding distilled water, and filtration through a fritted funnel. The off-white powder was dried carefully and stored in a glovebox. Technical quality [CpMn(CO) 3 ] was purified by the crystallization from hexane, as described earlier [66]. The 0.5 M solution of KHMDS in toluene was prepared by weighing 2.0 g of solid KHMDS (stored in a glovebox) in a volumetric flask, and adding toluene to complete to 20 mL. AgSbF 6 (Aldrich) was stored in the glovebox. All other reagents were commercially available, and used as received.
NMR spectra were recorded on Bruker AV300 or AV400 spectrometers. Chemical shifts are reported in ppm (δ) compared to TMS ( 1 H and 13 C), using the residual peak of deuterated solvent as internal standard [67]. Solution IR spectra were recorded in 0.1 mm CaF 2 cells using a PerkinElmer Frontier FT-IR spectrometer, and given in cm −1 with relative intensities in parentheses. UV-vis spectra were recorded on a Perkin Elmer Lambda 950 UV/Vis/NIR spectrometer in 1 cm quartz cell under nitrogen atmosphere (CH 2 Cl 2 , 10 −4 M sample concentration). Elemental analyses were performed by the microanalytical service of the LCC, and mass spectra (ESI mode) by the mass spectrometry service of the "Institut de Chimie de Toulouse". X-band EPR data were recorded using Elexsys ESP 500 and EMX Brüker EPR spectrometers. Typical settings: 2 mW microwave power and 0.1 mT modulation amplitude. For all species, spectra recorded with lower modulation amplitudes do not reveal more lines or different patterns (data not shown). Landé factors were measured by comparison with a reference DPPH sample (g = 2.0036). The numerical simulation of EPR spectra was performed using EasySpin software [68].

η 5 -Cyclopentadienyl)-Dicarbonyl-(1,3-Dimesityl-4,5-Bis(Dimethylamino)Imidazol-2 -ylidene)iron (II) Tetrafluoroborate [4](BF 4 ).
A solution of KHMDS (0.5 M in toluene, 380 µL, 0.19 mmol, 1.02 equiv.) was added dropwise to a solution of 1·HOTf (100 mg, 0.185 mmol) in toluene (5 mL), at 0 • C. After 30 min, the toluene was evaporated, in order to remove HMDS, and the crude product was dissolved anew in 5 mL toluene, and filtered over Celite under nitrogen. The toluene was evaporated, and the carbene was dissolved in dry THF (5 mL). CpFe(CO) 2 I (61.8 mg, 0.2 mmol, 1.08 equiv.) was added as a solid in the dark (under visible light irradiation, a carbonyl ligand can be removed and replaced by the iodine counter-anion). The solution was stirred for 1 h at room temperature, and THF was evaporated (always in the absence of light). The solid was taken up in dichloromethane (5 mL), degassed water (5 mL) and NaBF 4 (61.5 mg, 0.56 mmol, 3 equiv.) were added, and the reaction mixture was stirred for 30 min. The organic phase was isolated, dried over Na 2 SO 4 , and evaporated. The yellow solid was eventually washed with ether, and crystallized by the slow diffusion of a dichloromethane solution in diethyl ether, to afford [4](BF 4 ) (67 mg, 54%) as yellow crystals, some of which were suitable for X-ray diffraction. 1  To a suspension of 1·HOTf (216 mg, 0.4 mmol) in toluene (30 mL), a solution of KHMDS (0.9 mL of 0.5 M, 0.45 mmol, 1.1 equiv.) was added, and the resulting, slightly yellow solution was sonicated for 5 min, and then stirred at RT for 15 min. CpMn(CO) 3 (82 mg, 0.4 mmol, 1 equiv.) was added, and the reaction mixture was irradiated with UV light (370-380 nm), using a custom LED-based reactor (ca. 15W power), until ν CO bands of the product at the measurements of 1904 and 1831 cm -1 ceased to increase (ca. 1h). The toluene was evaporated under vacuum, and the residue was purified by column chromatography on silica (1 × 6 cm). A slightly yellow band containing traces of cymantrene was eluted with 10:1 hexane/toluene mixture, followed by yellow-orange band of the target product. The solution was evaporated under vacuum, and the crude product was crystallized from ether/hexane mixture at -20 • C, to afford 5 (195 mg, 86%) as yellow crystals. Single crystals, suitable for X-ray diffraction, were obtained from concentrated solution in toluene at room temperature. 1  To a stirred yellow solution of complex [5](SbF 6 ) (105 mg, 0.13 mmol) in CH 2 Cl 2 (5 mL), solid AgSbF 6 (45 mg, 0.13 mmol, 1 equiv.) was added, in one portion, at −20 • C. The color of the reaction mixture immediately changed to deep blue, and black precipitate appeared. The reaction mixture was slowly warmed to room temperature, stirred for additional 30 min, and the volatiles were removed under vacuum. The residue was repeatedly extracted with CH 2 Cl 2 under sonication (6 × 5 mL), combined dark blue solutions were filtered through Celite, and Et 2 O (25 mL) was added dropwise under vigorous stirring, to induce the precipitation of the product. The supernatant was removed by decantation, and the precipitate was washed with ether (10 mL) and dried under vacuum to give [5](SbF 6 ) 2 (108 mg, 80%) as deep violet powder. Single crystals, suitable for X-ray diffraction, were obtained by the layering of toluene on the solution of  Electrochemical studies Cyclic voltametric measurements were carried out with a Autolab PGSTAT100 potentiostat, controlled by GPES 4.09 software. Experiments were performed under argon at room temperature, in a homemade airtight three-electrode cell. The reference electrode consisted of a saturated calomel electrode (SCE) separated from the solution by a glass frit. The counter electrode was a platinum wire of ca. 1 cm 2 apparent surface. The working electrode was a Pt microdisk (0.5 mm diameter). The supporting electrolyte, [nBu 4 N](PF 6 ) (99% electrochemical grade), was dried at 120 • C and stored under Ar. The CH 2 Cl 2 solutions of 2, 3, [4](BF 4 ), and 5 used for the electrochemical studies were typically 10 −3 M in complex, and 0.1 M in supporting electrolyte. Before each measurement, the solutions were degassed by bubbling Ar, and the working electrode was polished with a polishing machine. Potentials are referenced to the SCE, and were calibrated using the Fc/Fc + couple, by adding ferrocene (10 −3 M) at the end of the experiments. Under the experimental conditions employed in this work, the half-wave potential (E 1/2 ) of the Fc + /Fc couple in CH 2 Cl 2 occurs at E 1/2 = +0.46 V vs. SCE. Cyclic voltametric studies include the study of the compounds in oxidation, and in reduction at various scan rates (from 50 mV/s to 10 V/s).
X-ray diffraction analyses. X-ray diffraction data were collected on a Bruker SMART / APEX II diffractometer ( [4](BF 4 )), an Oxford diffraction Gemini diffractometer (5), an Xcalibur Gemini Ultra diffractometer ( [5](SbF 6 )), or on a Bruker D8/APEX II/Incoatec Mo IµS Microsource diffractometer ([5](SbF 6 ) 2 ) using, in most cases, MoKα radiation (λ = 0.71073 Å, graphite monochromator), besides complex [5](SbF 6 ), for which CuKα radiation (λ = 1.54184 Å, graphite monochromator) was used. Crystal and structure refinement details are given in the Table S1. All calculations were performed on a PC compatible computer using the WinGX system [69]. The structures were solved using the SIR92 program [70], which revealed in each instance the position of most of the non-hydrogen atoms. All the remaining non-hydrogen atoms were located by the usual combination of full matrix least-squares refinement, and difference electron density syntheses using the SHELX program [71]. Atomic scattering factors were taken from the usual tabulations. Anomalous dispersion terms for the Mn and Fe atoms were included in Fc. All non-hydrogen atoms were allowed to vibrate anisotropically. The hydrogen atoms were set in idealized positions (R 3 CH, C-H = 0.96 Å; R 2 CH 2 , C-H = 0.97 Å; RCH 3 , C-H = 0.98 Å; C(sp 2 )-H = 0.93 Å; U iso 1.2 or 1.5 times greater than the Ueq of the carbon atom to which the hydrogen atom is attached), and their positions refined as "riding" atoms. After completing the initial structure solution of complexes [5](SbF 6 ) and [5](SbF 6 ) 2 , it was found that ca. 10% and 37% of the total cell volume was filled with disordered solvent, accounting for 84 and 386 electrons per unit cell, respectively. These solvent molecules, likely to be Et 2 O in partial occupation in case of [5](SbF 6 ) and CH 2 Cl 2 /toluene for [5](SbF 6 ) 2 , could not be well-modelled in terms of atomic sites, and the SQUEEZE procedure [72] was, therefore, applied to the data to remove this contribution. CCDC 2170556-2170559 contains the supplementary crystallographic data for the structures unveiled in this paper. These data can be obtained, free of charge, from the Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Conclusions
Following a first study on the thiourea derivatives, we then conducted a complete exploration of the redox behavior of a series of transition-metal complexes supported by the NHC ligand IMes (NMe2)2 (1), whose carbenic heterocycle is substituted by two dimethylamino groups. The ability of ligand 1 to act as a redox-active ligand is shown through different spectroscopic, and even crystallographic, techniques on the oxidized forms of the complexes. In a similar behavior to the thiourea derivative, the two dimethylamino groups in 1 play a critical role in the stabilization of the mono-oxidized ligand [IMes (NMe2)2 ] •+ (1 •+ ) in the complexes, by allowing an efficient delocalization of the spin density over the diaminoethene backbone. Conversely, the bis-oxidized state of the ligand is not generated in these complexes. The oxidation process of the NHC ligand is shown to have a profound effect on its overall electronic donation, leading to a new redox-switchable NHC. An obvious limitation of this ligand is a relatively high oxidation potential, which would limit its implementation in organometallic catalysis. We are, therefore, considering further developments, aimed at improving both the accessibility and stability of the oxidized forms of the NHC ligands, by changing the nature and location of the redox-active moiety, the NHC platform, and/or the denticity of the NHC ligand. Investigations along those lines are currently being performed in our laboratories.