Carbodicarbene Bismaalkene Cations: Unravelling the Complexities of Carbene versus Carbone in Heavy Pnictogen Chemistry

Synthesis of NHC-supported Bismuthenium Complexes

We hypothesized that a strong monodentate neutral donor ligand would aid in the preparation of low-coordinate bismuth cations. Our group recently reported 1 (Scheme 1), which is supported by two NHC ligands (4,5-dimethyl-1,3-diisopropylimidazolin-2-ylidene).²⁷ In an effort to synthesize an NHC supported monocation, we reacted [(NHC)BiPh₂(μ-Cl)]₂ with silver hexafluoroantimonate (AgSbF₆). Instead of the expected halide abstraction, the NHC transferred from Bi to Ag to give compound 2. This represents an unusual case of reverse-transmetalation,²⁸ where the carbene transfers to the silver-containing halide abstraction reagent. The formation of compound 2 was ascertained by single-crystal X-ray diffraction (Figure S40). Interestingly, recrystallization of [(NHC)BiPh₂(μ-Cl)]₂ from a THF/hexane mixture led to the formation of our target bismuthenium ion, [(NHC)BiPh₂]⁺, which contains a dichlorodiphenylbismuthate counteranion, [Ph₂BiCl₂]⁻ (3). Therefore, THF allows for the rearrangement of 1 to 3 by promoting NHC ligand dissociation and chloride abstraction to form the [Ph₂BiCl₂]⁻ anion. The direct reaction of more sterically crowded carbene ligands, 1,3-bis(2,6-diisopropylphenyl)-4,5-dihydroimidazole-2-ylidene (SIPr) and (2,6-diisopropylphenyl)-4,4-diethyl-2,2-dimethyl-pyrrolidin-5-ylidene (CAAC), with [Ph₂Bi(μ-Cl)]₂ led to cationic complexes (4–5) isostructural to that of compound 3. Indeed, no reaction occurs when [(SIPr)BiPh₂][Ph₂BiCl₂] (4) or [(CAAC)BiPh₂][Ph₂BiCl₂] (5) is reacted with an additional equivalent of carbene. The rearrangement chemistry and solvent-/carbene-promoted cationization observed is consistent with relatively weak carbene-Bi interactions that can be difficult to control. X-ray crystal structures for compounds 3–5 are shown in Figure S41.

We hypothesized that the stronger electron donor properties of CDC compared to NHC would promote significant Bi−Cl bond elongation, allowing for facile halide abstraction. To draw comparisons between the two ligands and to test our hypothesis, we synthesized an NHC complex of phenylbismuth dichloride, [(NHC)Bi(Ph)Cl₂]₂, as well as the CDC analogue, (CDC)Bi(Ph)Cl₂. NHC was allowed to react with (THF)Bi(Ph)Cl₂ for 1 hour at room temperature in benzene (Scheme 2). After purification, compound 6 was obtained as a white solid in 75 % yield. The ¹H NMR spectrum revealed a well-defined heptet at 4.95 ppm (J=7.0 Hz), representing the methine environment of a coordinated NHC ligand. Similarly, CDC was reacted with (THF)Bi(Ph)Cl₂ in THF at room temperature, which forms a deep yellow precipitate upon addition. After filtration and drying under vacuum, (CDC)Bi(Ph)Cl₂ (7) is obtained in 69 % yield. A broad peak at 5.22 ppm was observed in the ¹H NMR spectrum for 7 in CD₂Cl₂, representing the methine proton of a new CDC coordination environment. In the ¹³C NMR, a sharp singlet at 189.3 ppm is representative of the carbone carbon atom, this is only slightly downfield of the carbene carbon in compound 6 (185.6 ppm).

Colorless single-crystals of compound 6 were obtained from a saturated benzene solution at room temperature, while yellow crystals of compound 7 were obtained from a THF/hexane (1:2) layered mixture at room temperature (Figure 3). There are two notable differences between the solid-state structures of 6 and 7. (1) The NHC-supported bismuth complex 6 is dimeric and the CDC coordinated compound 7 is monomeric, suggesting that the strong electron donor properties of CDC negates the necessity for dimerization due to an electronically saturated bismuth atom. (2) The ^carboneC−Bi bond length in 7 [2.249(6) Å] is significantly shorter than the ^carbeneC−Bi bond length in 6 [2.346(2) Å]. This is a result of a partial π-donation from CDC to Bi in the former. As predicted, the average Bi−Cl bond lengths for 7 (2.734 Å) are longer than those in 6 (2.694 Å). Due to the observed differences in the metrical data we reasoned that CDC should lead to bismuth compounds with multiple bond character and enhanced stability. Thus, CDC-Bi cations were targeted as a platform to access bismaalkene-type species, which unlike the lighter pnictogen congers are extremely rare.

Synthesis of CDC-stabilized Mono-, Di-, and Tri-positive Bismuthenium Ions

In order to test our hypothesis that CDC would enable the formation of electropositive bismuth cations with increasing double bond character, we probed the reaction of 7 with one equivalent of AgSbF₆ in dichloromethane (DCM) (Scheme 3). Orange needle-like X-ray quality crystals of the monocation (8) were obtained in 78 % yield from a THF/hexane (1:1) layered mixture at room temperature. The ¹H NMR spectrum of 8 in [D₈]THF revealed a broad resonance at 5.03 ppm for the methine proton of the coordinated CDC ligand which is slightly downfield compared to its starting material 7. Due to the more deshielding electrophilic bismuth center, the ¹³C NMR spectrum reveals a singlet downfield from that in 7 at 191.6 ppm attributed to the carbone carbon of 8. In pursuit of a dicationic species via a second halide abstraction, we added in an additional equivalent of AgSbF₆ to 8 in DCM. Dark red X-ray quality crystals of the bismuthenium complex 9 were obtained in 43 % yield from a DCM/hexane (1:1) layered mixture at room temperature. A broad resonance attributed to the methine proton of coordinated CDC ligand in 9 is observed at 4.57 ppm in CD₂Cl₂, which is consistent with the trend observed for complexes 7 and 8. In this respect, the ¹³C NMR shows a peak at 209.7 ppm for the carbone carbon of 9, which is downfield from that of both 7 and 8, supporting substantial electron donation from CDC to bismuth.

The solid-state structures of cations 8 and 9 are shown in Figure 4. Both cations contain coordinating [SbF₆]⁻ counter anions, with Bi—F interaction distances of 2.904(2) Å (8) and 2.603(8) −2.740(7) Å (9). The CDC-bismuth bond length decreases in the order of 7 [2.249(6) Å] >8 [2.226(3) Å] >9 [2.157(11) Å], and these data support an increase in π-donor character from CDC as the bismuth center becomes increasingly more electrophilic. Notably, the C1−Bi1 bonds are significantly shorter than those in their NHC-bismuthenium congeners. Since carbones are π-basic rather than π-acidic, and the bismuth center is electrophilic, the C−Bi bond lengths are shortened due to substantial π-donation from the ^carboneC to Bi. The ^carboneC−Bi bond length in 9 is shorter than the ^CAACC−Bi bond in the subvalent carbene-bismuthindene [2.199(2) Å] and even rivals that of the bismuthio ylide (2.16 Å),^25b and the aromatic C-Bi bonds of a bismabenzene (2.154–2.160(4) Å).^25a Thus, 9 may be regarded a stabilized dicationic bismaalkene, which is unprecedented. It is noteworthy that while phosphaalkene species have been widely studied,^{22b, 22c, 29} structurally characterized examples of heavy bismuth analogues are virtually unexplored.

In order to synthesize bismuthenium ions without weakly coordinating anion contacts, we reacted CDC with bismuth tribromide in THF overnight (Scheme 4). After workup, compound 10 was obtained as a red solid in 86 % yield. The ¹H NMR spectrum of 10 shows a well-defined heptet (J=6.7 Hz) at 5.01 ppm representative of a sterically unrestricted CDC coordination environment; this is in contrast to the broad signals observed for 7, 8, and 9. Red single crystals of 10 suitable for X-ray diffraction analysis were obtained from a THF/hexane (2:1) layered mixture at room temperature.

The solid-state structure of compound 10 shows a dimeric [(CDC)BiBr₃]₂ complex with bridging bromides (Figure 5). Each bismuth atom has adopted a distorted square pyramidal geometry when considering the bromide contacts. The CDC-bismuth bond [C1−Bi1: 2.292(9) Å] is shorter than that in compound 7 [2.249(6) Å]. The bismuth-bromide bond in the apical position of 10, [Bi1−Br2: 2.6629(16) Å], is shorter than the sum of ionic radii for Bi and Br (2.68 Å).³⁰ The other two bismuth-bromide bonds are substantially longer [Bi1−Br1: 2.9196(12) Å and Bi1−Br3: 2.8390(12) Å], which support weaker Bi-Br interactions.

While halide abstraction from compound 7 will only yield mono- and di-positive bismuthenium ions (8 and 9), compound 10 is a suitable starting material for synthesizing a tri-positive bismuthenium ion. Our initial synthetic approach involved the use of THF, which would provide additional support to the bismuth center by readily filling the coordination sphere upon halide abstraction. Interestingly, when compound 10 is reacted with AgSbF₆, THF solvent polymerized into a tough insoluble solid material uncharacterizable by traditional spectroscopic methods. In an effort to circumvent this polymerization process, but still allow for THF coordination, we performed the reaction of 10 with two equivalents of AgSbF₆ in DCM and added an aliquot of THF in situ (Scheme 5). Thus, the monocation 11 was isolated in 67 % yield. Unlike the monocation 8, compound 11 does not contain any anion interactions due to the coordination of THF solvent molecules to Bi. Attempts to form the dicationic product from the reaction of 10 with four equivalents of AgSbF₆ and a variety of other halide abstraction reagents proved to be unsuccessful. This is in accordance with the observed difference in Bi−Br bond distances, which suggest that one Br atom is more tightly bound to Bi. However, facile halide abstraction was observed to generate the dication 12 when four equivalents of silver (bis-trifluoromethylsulfonyl)imide (AgNTf₂) was reacted with 10. Similar to 11, compound 12 does not contain any anion contacts to the bismuth center. The ¹H NMR integrations suggested that a third equivalent of THF coordinated to bismuth with respect to the monocation 11, to give a 1:3 THF to CDC ratio. Several attempts were made to synthesize a trication using this THF addition method, but all proved to be unsuccessful. We therefore hypothesized that the coordination of a second CDC ligand would render the last Bi−Br bond more labile due to the strong donor effect, thus weakening the Bi−Br bond. As such, the reaction 10 and two equivalents of CDC with six equivalents of AgNTf₂ successfully gave the bis(CDC)bismuthenium trication 13 in 73 % yield. The shift pattern in the ¹H NMR spectrum of 13 closely resembles that of a bis(CDC) dicationic(hydrido)boron complex,^21f further supporting that two CDC ligands have coordinated to the Bi center.

Red needle-like crystals of 11 suitable for X-ray diffraction were obtained from a THF/hexane (1:1) layered mixture at room temperature. The crystal structure reveals a monocationic [(CDC)BiBr₂(THF)₂]⁺ core, and the bismuth center has adopted a square pyramidal geometry. The ^carboneC−bismuth bond of 11 [C1-Bi1: 2.226(12) Å] is slightly longer than that of 8, which is expected for a higher coordination number. Red plate-like crystals of compound 12 were obtained from a DCM/ether (2:1) mixture at room temperature. Similar to 11, the structure reveals a square pyramidal dicationic [(CDC)BiBr(THF)₃]²⁺ core, with an additional THF occupying a coordination site. As a result of three coordinating THF molecules, the ^carboneC−Bi bond of 12 [C1−Bi1: 2.199(5) Å] is longer than 9 [C1−Bi1: 2.157(11) Å]. Dark blue plate-like crystals of compound 13 were obtained from a chlorobenzene/hexanes (10:1) mixture. The crystal structure of 13 revealed two CDC ligands coordinated to a Bi^III center, with three [NTf₂]⁻ counter anions. The ^carboneC−bismuth bonds [Bi1−C24: 2.166(2) and Bi1−C1: 2.197(2) Å] are only slightly shorter than that in 12. The average allenic bond length has increased from 1.335 Å in the free ligand to 1.437 Å suggesting substantial electron donation from the carbone carbon to Bi^III, and less back donation to the NHC moiety.

Bonding Analysis of the Bismuth and Bismuthenium Complexes

CDC-stabilized bismuthenium complexes 7–13 were calculated with quantum chemical methods using density functional theory (DFT) at the BP86-D3(BJ)/def2-TZVPP level.³¹ The nature of the chemical bonds between the CDC ligand and the remaining bismuthenium fragments [Bi] was investigated with energy decomposition analysis (EDA)³² in conjunction with natural orbitals for chemical valence (NOCV).³³ The EDA-NOCV method^{32, 33} has been proven to give deep insights into the nature of the chemical bond in a variety of main-group compounds and transition metal complexes.³⁴

The calculated equilibrium structures of 7–13 are shown in Figure S42 of Supporting Information. The calculated and experimental geometries are in agreement except for some Bi—F distances for the weakly coordinating anions in 8 and 9, which are considerably shorter than the experimental distances. This is also observed when the M06-2X functional is used for the geometry optimization.³⁵ These differences are most likely due to intermolecular forces in the solid state. Therefore, we carried out the bonding analysis with the EDA-NOCV method using the calculated and the experimental geometries (Table S3–S5). Figure S42 displays the computed bond dissociation free energy (ΔG²⁹⁸) of the CDC−Bi bond at 298 K which ranges from 27.2–62.3 kcal mol⁻¹, which indicates that the molecules are thermodynamically quite stable. We optimized the geometry of the compound 13 (CDC)₂Bi[NTf₂]₃ and the free trication [(CDC)₂Bi]³⁺. The agreement of the theoretical values of (CDC)₂Bi[NTf₂]₃ with the experimental data is reasonably good, the deviation between the calculated and X-ray values is likely to be caused by solid-state forces. The calculated geometry of the free trication [(CDC)₂Bi]³⁺ is only slightly different from the geometry in the presence of the counter ions in (CDC)₂Bi[NTf₂]₃. The bond strength of the CDC ligands to bismuth in (CDC)₂Bi[NTf₂]₃ has been estimated by calculating the bond dissociation energy for the reaction (CDC)₂Bi[NTf₂]₃ → 2 CDC + Bi[NTf₂]₃. The computed value of D₀ (ΔG²⁹⁸)=128.6 (88.2) kcal mol⁻¹ suggests that the bonds are quite strong.

Table S5 shows the numerical EDA-NOCV results of 7–13 using the calculated geometries (experimental geometries are shown in Figure S42). The calculated intrinsic interaction energy (ΔE_int) values are similar except for structures 8, 9 and 12, where the energy values using the experimental geometries are somewhat larger. However, the percentage contributions of the four components to the CDC-[Bi] interaction, i.e., Pauli repulsion ΔE_Pauli, dispersion forces ΔE_disp, electrostatic attraction ΔE_elstat and orbital (covalent) interaction ΔE_orb are very similar. This indicates that the bonding analysis using the calculated structures may safely be used for the experimentally observed species. The breakdown of the ΔE_orb term into pairwise interactions ΔE_orb(1)−ΔE_orb(3) shows that the major contribution ΔE_orb(1) comes from the σ donation of the CDC ligand [CDC]→[Bi]. The donation ΔE_orb(2) of the π lone pair of the CDC ligand¹⁴ [CDC]→[Bi] is significantly weaker (Table S5). It becomes stronger when the experimental geometries are used, because the measured carbon-bismuth distances are shorter than the calculated values (Figure S42). The largest contribution of the [CDC]→[Bi] π donation is found for 9 where it amounts to 19 % of the total orbital interactions when the experimental geometry is used (Table S3). The third pairwise orbital contribution ΔE_orb(3) is due to weak [CDC]←[Bi] σ backdonation. This is the complementary part of the overall σ bond. For the trication 13, the lowest values were provided by using the triplet states of [(CDC)₂]²⁺ and Bi⁺ with the electron configuration (6s²6p_y¹6p_z¹) for the atomic ion. The numerical results of the EDA-NOCV calculations are given in Table S5, which show also the xyz directions of the p AOs. The calculated results using other fragments are given in Table S7 of Supporting Information. We also calculated the other systems 7–12 using triplet fragments. The EDA-NOCV results in Table S8 show that the ΔE_orb values are always much higher than provided by the singlet fragments shown in Table S5.

The results of the bonding analysis suggest that the NHC^Ph substituents of the CDC ligand serve as a sink for the electronic charge, because they are rotated out of the molecular frame so that the vacant π orbitals of NHC^Ph can interact with the occupied orbitals of the [Bi] donor. The related molecular orbitals representing double dative interaction is provided in Figure S43. Note that in moving from 7→8→9 or 10→11→12, with the increase in electrophilic character of Bi, both the absolute values of [CDC]→[Bi] σ and π donation gradually increases (from 10→11 the σ donation remains essentially the same), which agrees with the increasingly shorter ^carboneC−Bi bond distances.

The pairwise orbital interactions ΔE_orb(1)−ΔE_orb(3) can be assigned to specific donations with the help of the associated deformation densities Δρ. Figure S45 shows the shape of Δρ₍₁₎−Δρ₍₃₎ of the pairwise interactions ΔE_orb(1)−ΔE_orb(3) in 9 (for the remaining systems, see Figure S45). The color of the charge flow is red→blue. It becomes obvious that, besides the σ donation and π backdonation, there is also polarization within the fragments.

We have further compared the bonding interactions in 7–13 with that in carbene-stabilized bismuthenium 3–5 and bismuth CAAC-Bi(Ph)Cl₂ complexes.⁷ The optimized geometries of the four carbene adducts are shown in Figure S46. The numerical results of the EDA-NOCV calculations are given in Table S6. A comparison of the data in Tables S5 and S6 shows that the intrinsic interaction in the CDC complexes is larger than that in the carbene adducts. The most important information comes from the breakdown of ΔE_orb into the pairwise orbital interactions between the ligands and the metal fragments. There is only one dominating σ donation ΔE_orb(1) in the carbene complexes [carbene]→[Bi] besides the π backdonation [carbene]←[Bi] ΔE_orb(2) (For associated deformation densities see Figure S47). In contrast, the CDC complexes exhibit two notable donor components from the ligand to the metal fragment. The remaining orbital interactions ΔE_rest consist of a large number of weak polarization terms within the fragments. Thus, the EDA-NOCV calculations signal a clear difference in the donor-acceptor interactions between the carbone and carbene ligands. This is best exemplified by comparing the results of the carbone complex CDC-Bi(Ph)Cl₂ (7) with the analogous carbene adduct CAAC-Bi(Ph)Cl₂. The total interaction energy ΔE_int of 7 is only slightly higher (−81.5 kcal mol⁻¹) than that of the carbene complex (−78.4 kcal mol⁻¹) but the orbital interactions ΔE_orb of the former carbone complex are much larger (−108.0 kcal mol⁻¹) than for the latter carbene species (−98.0 kcal mol⁻¹), which is mainly due to the additional contribution of the π bond (−7.3 kcal mol⁻¹). The further strengthening of the orbital interactions is partly compensated by weaker electrostatic attraction in the carbone complex. It requires the analysis of the electronic structure in order to find that 7 and the other carbone complexes possess a double dative C Bi bond, which consists of a strong σ and a significantly weaker π component, whereas the carbene complexes exhibit only a σ bond C→Bi.