Imidazolylidene Cu(II) Complexes: Synthesis Using Imidazolium Carboxylate Precursors and Structure Rearrangement Pathways.

Copper(II) complexes of type (NHC)CuX2 (X = OAc, Cl, Br, BF4, and NO3) bearing monodentate N-heterocyclic carbenes (NHCs) were prepared by in situ decarboxylation of imidazolium carboxylates as a new synthetic methodology for Cu(II)-NHC complexes. In contrast to the classical deprotonation method, the decarboxylation protocol does not require anaerobic conditions and provides access to complexes with NHCs that are unstable as free carbenes such as N,N'-diisopropyl-imidazolylidene and N,N'-dimethyl-imidazolylidene. Spectroscopic evidence of the formation of the Cu-CNHC bond is provided by UV-vis and EPR, in particular by the 44 MHz carbene hyperfine coupling constant using a 13C-labeled imidazolylidene ligand. A variation of the nature of the carbene N-substituents and the anions bound to the Cu(II) center is possible with this methodology. These variations strongly influence the stability of the complexes. Structural rearrangement and ligand reorganization was observed during recrystallization, which are comprised of heterolytic Cu-CNHC bond dissociation for unstable NHC ligands as well as homolytic Cu-X bond cleavage and disproportionation reactions depending on the nature of the anion X in the copper complex.


Introduction
N-Heterocyclic carbenes (NHCs) 1 have emerged as ligands for transition metals to perform a multitude of catalytic transformations. 2 In particular, Cu-containing NHC complexes have been used as catalysts for a broad range of reactions such as the Huisgen cycloaddition, hydrosilylation, carboxylation, olefination, C-C bond cleavage, C-C and C-N coupling reactions, in which the active species is Cu(I). 3 While soft-acid soft-base Cu(I)-C interactions lead to highly stable Cu(I)-NHC complexes, less favourable intermediate/hard-soft Cu(II)-C interactions make the isolation of Cu(II)-NHC complexes difficult. Consequently and in contrast to low valent copper(I) NHC complexes, which are widely known, [3][4][5]

higher valent
Cu(II)-NHC analogues are very rare. Most of the few examples reported so far employ tethered-NHC ligands to favour coordination to the Cu(II) center (for example complexes A-D, [6][7][8][9] Fig.   1). These Cu(II)-NHC complexes are typically prepared by transmetalation using an NHC transfer reagent or by direct complexation to the isolated free carbene. [6][7][8][9][10][11][12][13][14][15][16] Complexation of Cu(II) by unsupported monodentate NHC ligands has been explored little with only (1)Cu(OAc)2 being authenticated by X-ray crystallography (E, Fig. 1). 12 The use of carboxylate precursors of NHCs has been demonstrated for the preparation of NHC metal complexes of Ir, Rh, Ru, Pd, Pt, and Cu, and showed a high tolerance to water. [17][18][19][20] An advantage of this method is the generation of the desired complex through an in situ decarboxylation process induced by the metal precursor, which avoids the handling of the free carbene. Inspired by the early work of Saegusa and Cahiez who demonstrated that Cu(II) is promoting decarboxylative Cu-C bond formation with, e.g., benzoate, 21,22 we envisaged the preparation of a variety of Cu(II)-NHC complexes using different imidazolium carboxylate precursors and diverse Cu(II) salts. Herein we describe a series of compounds using this approach and their characteristic EPR and UV/vis spectroscopic properties in solution. In particular, we demonstrate that their preparation and stability is distinctly influenced by the nature of the N-wingtip groups of the NHC ligands and by the anion of the copper precursor.

Results and Discussion
New synthetic route using an imidazolium carboxylate as NHC precursor. We have applied the decarboxylation strategy for the synthesis of the known Cu(II)-NHC complex (1)Cu(OAc)2 12 (Scheme 1). When starting from the carboxylate precursor 1-CO2, the desired complex (1)Cu(OAc)2 was obtained in similar yields as via the deprotonation route (63%), yet in only one single step and in an aerobic atmosphere. 23 This is in contrast to the published route that requires the deprotonation of the imidazolium salt 1H.Cl and isolation of the air-sensitive free carbene 1, followed by its reaction with Cu(OAc)2 under inert and dry conditions. 12 Scheme 1. Synthesis of complex (1)Cu(OAc)2 following either the deprotonation (top) or the decarboxylation (bottom) routes starting from ligands 1H.Cl and 1-CO2, respectively. Support for the formation of (1)Cu(OAc)2 was obtained by infrared spectroscopic analysis (Fig.   2, see Figure S2 for full spectra). The vibration bands at 1674 and 1305 cm -1 (assigned to the carboxylate functional group for the ligand precursor 1-CO2) are absent in the spectrum of (1)Cu(OAc)2, suggesting that the CO2 functional group is no longer pendant to the ligand framework. Moreover, the acetate bands in the 1600 cm -1 area are sharper and more clearly defined than the broad features in the precursor Cu(OAc)2 salt, indicative of a better defined ligand arrangement in the complex due to NHC coordination to the Cu(II) center. The identity of complex (1)Cu(OAc)2 prepared via the decarboxylation route was verified by comparison with an authentic sample of (1)Cu(OAc)2 prepared by the deprotonation route as reported in the literature. 12 Both samples were recrystallized from a saturated CH2Cl2 solution by pentane diffusion and the crystals were analyzed by powder X-ray diffraction. The powder X-ray diffractograms of both samples of (1)Cu(OAc)2 showed well resolved crystalline intensities, in a monoclinic crystal pattern with C2/c space group ( Figure 3). The identical intensities in the 8-30° 2θ range for both samples indicate that the structure of both crystalline species is the same. The diffractogram features a prominent maximum at 2 = 8.835° with corresponding d spacing value of 10.001 Ǻ, and strong intensities at 11.090 and 11.277°. Significantly, the diffractogram of (1)Cu(OAc)2 calculated from the reported single crystal X-ray diffraction data 12 shows an essentially identical pattern especially in the low theta range with strong intensities at 8.843, 11.077 and 11.272° ( Figure 3; Tables S1, S2).   (1) were water-free according to elemental analysis, indicating that the use of reagent-grade moist solvents for the decarboxylation route did not induce coordination of water to the copper center.
In contrast, reaction with Cu(OTf)2 afforded a sticky solid that did not provide any evidence for Cu-NHC complex formation. dissolved in CH2Cl2 were studied by absorption spectroscopy. All complexes exhibit a very strong absorbance in the UV region, attributed to ligand π-π* transitions, and a weak band in the vis-NIR domain which was assigned to Cu(II) d-d transitions ( Figure 4, Table 1). The d-d transition bands display a shoulder at lower energy which is characteristic for a squarepyramidal metal coordination geometry at copper. 32     The UV-vis absorption characteristics of the Cu(II) complexes (1)CuX2 are strongly dependent on the nature of the anion X -. While all complexes display intense ligand π-π* transitions in the UV region, complex (1)CuCl2 features additionally a substantial band at 471 nm and a weak absorption in the NIR range around 880 nm, which was attributed to Cu(II) d-d transitions ( Figure 4, Table 1). Complex (1)CuBr2 has a much stronger absorption at 649 nm with a shoulder at lower energy (855 nm) that is similar to the absorption properties of complex (1)CuCl2 (Table 1) (featureless spectrum), which were measured as controls. The spectrum of Cu(OAc)2 displays a resolved but rather weak g ﬩ signal around 2.07, while the spectrum of the mixture of Cu(OAc)2 and 1H.Cl was featureless and no signal was resolved. The spectrum of (1)Cu(OAc)2 is identical irrespective of the applied synthetic methodology (deprotonation vs decarboxylation) and therefore supports the suitability of decarboxylation to form a Cu(II)-CNHC bond (vide infra). The signals were notably broadened, which was attributed to a mixture of two species in solution, possibly due to variations in the coordination modes of the acetate ligands. 34 The splitting of the signal at low field (2750 G) has been attributed to partially separated 63 Cu and 65 Cu absorptions from a Cu(OAc)2 spectrum. 35 The spectrum was simulated with EasySpin, using two isomers with slightly deviating parameter sets, {g1 = 2.290, g2 = g3 = 2.055, A1( 63 Cu) = 160 × 10 -4 cm -1 } and {g1 = 2.280, g2 = g3 = 2.055, A1( 63 Cu) = 185 × 10 -4 cm -1 } ( Figure S17). The parameters determined for (1)Cu(OAc)2 are different from those previously reported for Cu(II) complexes with chelating NHC ligands (Table 2). For example, Meyer's tris-NHC Cu(II) complex D (cf Fig. 1) has a tetrahedral geometry and displays a rhombic signal with similar g values and a lower hyperfine coupling constant. 8 Bis-alkoxy-NHC Cu(II) complexes akin to complex C (cf Fig. 1   To establish the carbene hyperfine coupling constant in Cu(II)-NHC complexes, the 13 Clabelled ligand 13 C-1H.Cl was synthesized 36 and complex ( 13 C-1)Cu(OAc)2 was isolated upon reaction with tBuOK and Cu(OAc)2. While elemental analysis, mass spectrometry and powder X-Ray diffraction analyses of crystals of ( 13 C-1)Cu(OAc)2 are identical with those of the nonlabelled complex (1)Cu(OAc)2 (cf experimental part and Fig. S1), EPR spectroscopy shows distinct differences ( Figure 5). The hyperfine signals splitting is especially clear in the g// region, with a simulated coupling constants a( 13 C)1 = 44 × 10 -4 cm -1 ( Figure S18). Moreover, this large coupling constant unambiguously confirms the formation of a CNHC-Cu bond. In addition, the EPR data provide support for the integrity of (1)Cu(OAc)2 in solutions used for EPR spectroscopy (CH2Cl2/toluene and DMF/toluene, Fig. S19).

NMR spectroscopy and other solution state considerations of complexes (1-4)Cu(OAc)2.
Finally, our investigations also included 1 H NMR spectroscopy of a solution of (1-4)Cu(OAc)2 in CD2Cl2. As expected, broad signals characteristic of paramagnetic effects were observed in the 0-12 ppm range ( Figure S27). While allocation of signals was not trivial, the larger than expected number of signals suggests the co-existence of different isomers or aggregates, in agreement with the EPR measurements. Diffusion ordered spectroscopy (DOSY) experiments were performed to distinguish between different aggregate sizes. Actually, the diffusion constants of the various species appeared essentially identical and did not lead to a clear separation of species ( Figure S28, Table S3). This similarity suggests only minor changes, e.g., imparted by mono-vs bidentate bonding of the acetate ligand. The 13 C{ 1 H} NMR spectra did not show any resonance that is attributable to the CO2 group (C = 152 for ligand precursor 1-CO2) and hence support successful decarboxylation en route to the formation of the (NHC)Cu(OAc)2 complex ( Figure S29).

Similarly, NMR spectroscopic analyses of complexes (1)CuCl2 and (1)CuBr2 dissolved in
CD2Cl2 revealed only broad signals in the 1 H NMR spectra characteristic for paramagnetic species (Figures S30-31). Therefore, UV-vis and NMR analyses of (1)CuBr2 strongly suggest the formation of a Cu(II)-NHC complex, even though this complex is EPR silent as a solid or in frozen solution at 77 K. In contrast, sharp signals were noted in the NMR spectra of  The rearrangement of complex (4)Cu(OAc)2 is very similar, however the copper acetate complex anion is composed of alternating paddlewheels Cu2(OAc)4 cluster and monomeric Cu(OAc)2 units, both linked with a bridging  2 , 2 -bound acetate ligand, indicating versatility of the Cu(OAc)n synthon (Figure 9). Again, significant hydrogen bonding is present between the protons of the imidazolium cation 4H + and the acetate oxygen atoms ( Figure S34). Similar copper carboxylate polymers were previously described, though with different linkers such as bridging 4-4'-bipyridine, 39,40 or alternate with Cu(II) complexes such as (1,10phenanthroline)Cu motifs. 41 Also tetrametallic systems were reported that contain one paddlewheel cluster linked to two Cu(carboxylate)2 motifs similar to the complex anion resulting from rearrangement of (4)Cu(OAc)2, yet terminated by water molecules. 42  43 The [(2)2Cu] + copper(I) ion was also identified by mass spectrometry with a signal at m/z 671.3188 (calc. 671.3175) from a MeCN solution of the isolated powder of (2)Cu(OAc)2 ( Fig. 10-11). Of note, a frozen solution of the crystals was EPR silent, which is expected for a Cu(I) salt and Cu(II) acetate clusters, 44,45 yet in marked contrast to the behaviour of the complex before crystallization (cf Figure 6). Crystallization of (1)CuBr2, afforded co-crystals of imidazolium dibromocuprate in which the imidazolium cation is either brominated at C2 position or protonated, i.e. 1Br.CuBr2 and 1H.CuBr2. The refinement converged at a 87:13 ratio of the two cations, and also the CuBr2anion is disordered in the same proportion, indicating that each cation has its specific anion ( Figure 11). The formation of 1Br.CuBr2 is further supported by a mass spectrometry signal in positive ionization mode at m/z 467.2051 (calc. for 1Br + 467.2062) from a MeCN solution of the powder of (1)CuBr2 (Fig. S15). Figure 11. Re-assembly of (1)CuBr2 into 1Br.CuBr2 and 1H.CuBr2 and ORTEP representations of the two co-crystallized salts with a refined weighting of 87% 1Br.CuBr2 and 13% 1H.CuBr2 (50% probability ellipsoids; hydrogen atoms omitted for clarity). In the presence of a bromide or chloride anions rather than acetate, both pathways are observed.
NHC-Cu dissociation affords the imidazolium cations 1H + , while disproportionation is evidenced by the formation of the imidazolium cations 1Cl + and 1Br + . Disproportionation as discussed for (2)Cu(OAc)2 leads to a mixture of (1)CuBr and (1)CuBr3 when starting from (1)CuBr2. While reductive NHC-OAc elimination is prevented, reductive elimination from (1)CuBr3 is well conceivable and produces the brominated imidazolium cation 1Br + together with the dibromo cuprate anion as 1Br.CuBr2. This reaction pathway is supported by the fact that the cation 1Br + was previously synthesized from the Cu(I) complex (1)CuBr in the presence of an oxidant through the formation of a putative Cu(III)-NHC tris(halide) complex. 48 The similarity of product formation for the bromide and chloride complexes suggests that the disproportionation and reductive elimination is a rather general pathway for copper re-assembly with halides and likely with other anions that are susceptible to reductive C-X bond formation. procedures. 17,24,36,51,52 The labelled imidazolium salt 13 C-1H.Cl was prepared using the same procedure than for 1H.Cl with 99% labelled 13 C-paraformaldehyde (Aldrich). 36 Complexes syntheses. The synthesis of complexes (NHC)CuX2 was accomplished using one of the following methods. Crystals were grown from slow diffusion of Et2O into a solution of the complex in CH2Cl2.
Route a: Under nitrogen atmosphere, the imidazolium salt (0.20 mmol) was reacted with tBuOK The solution was filtered and pentane was added until a solid precipitated. The solid was collected by filtration, washed with Et2O (2 x 2mL) and dried under vacuum.