|Home | About | Journals | Submit | Contact Us | Français|
The discovery two decades ago of metal-free stable carbenes, especially imidazol-2-ylidenes [N-heterocyclic carbenes (NHCs)], has led to numerous breakthroughs in organic and organometallic catalysis. More recently, a small range of complexes has been prepared in which alternative NHC isomers, namely imidazol-5-ylidenes (also termed abnormal NHCs or aNHCs, because the carbene center is no longer located between the two nitrogens), coordinate to a transition metal. Here we report the synthesis of a metal-free aNHC that is stable at room temperature, both in the solid state and in solution. Calculations show that the aNHC is more basic than its normal NHC isomer. Because the substituent at the carbon next to the carbene center is a nonbulky phenyl group, a variety of substitution patterns should be tolerated without precluding the isolation of the corresponding aNHC.
For decades, carbenes, which feature a neutral divalent carbon atom with two nonbonding electrons, were considered prototypical reactive intermediates (1). Today, thanks to the availability of stable carbenes (2, 3), these molecules, especially the so-called N-heterocyclic carbenes (NHCs) (I) (4–6) (Fig. 1, top left), are recognized as versatile ligands for transition metal–based catalysts (7–10) and as metal-free organic catalysts in their own right (11–14). As expected, NHCs I usually bind metals via the carbene center (C2) to give η1 complexes II. However, in 2001, Crabtree and co-workers discovered that 2-pyridylmethylimidazolium salts react with IrH5(PPh3)2 to give 1 with the imidazole ring bound the “wrong way,” at C5 and not at C2 (15, 16) (Fig. 1, center). Since that time, a few other complexes of type IV featuring the so-called abnormal NHCs (aNHCs) (III) (17–20) as ligands have been prepared (21–23) (Fig. 1, top right). Experimental and theoretical data suggest that aNHCs III are even stronger electron-donor ligands than are NHCs I. In line with these observations, initial catalytic screening of aNHC metal complexes IV reveals promising results for the activation of unreactive bonds such as C–H and H–H (24–26). As an example, an aNHC palladium complex has been reported to be an efficient catalyst in the Heck olefination of aryl bromides, whereas the corresponding NHC analog is virtually inactive under identical conditions (24).
Lassaletta and co-workers (27) have shown that the deprotonation of imidazo [1,5-a] pyridinium salts 2 leads to free NHC 3 that can be isolated (Fig. 1, bottom). In contrast, using C2-substituted precursors, such as 4, Lassaletta et al. did not observe the corresponding free aNHC. However, by performing the deprotonation reaction in the presence of [Rh(COD)Cl]2, they were able to isolate the corresponding aNHC complex 5. Because calculations predicted that the parent aNHC III (where R is equal to H) is only about 17 kcal mol−1 higher in energy than its NHC isomer I (18), it seemed that free aNHC derivatives were reasonable synthetic targets. We report here the isolation of a metal-free member of this class of heterocyclic compound.
By analogy with the classical synthetic route used to prepare NHCs, we chose imidazolium salt 6 as a precursor to the desired aNHC 9 (Fig. 2). The pKa (where Ka is the acid dissociation constant) for loss of the C5-bound proton in the parent imidazolium salt (~33) (28) was calculated to be nine units higher than that for loss of the C2-bound proton (29); we therefore replaced the C2 hydrogen with a phenyl group. To offer kinetic protection to the C5 position, we appended bulky 2,6-di-isopropyl-phenyl (Dip) substituents at both nitrogen atoms, as well as a second phenyl group at C4. Imidazolium salts 6 with various counterions were prepared in good yields after slight modifications to known synthetic procedures (30–32). They were fully characterized by spectroscopic methods, with a single-crystal x-ray diffraction study carried out for the bromide salt 6 (Br−) (Fig. 3, left).
All attempts to deprotonate the imidazolium tetrafluoroborate salt 6 (BF4−) failed. However, small anions such as Cl− and Br− are known to accelerate heterolytic C–H bond cleavage through hydrogen bonding, and this effect has been used with C2- and C5-unsubstituted imidazolium salts to favor metallation of C2 (with the more acidic proton) over C5 (33). We reasoned that with C2 protected in 6, small anions should promote the desired deprotonation reaction at C5. Indeed, when 6 (HCl•Cl−) was treated with two equivalents (34) of a lithium base such as n-butyllithium (nBuLi) or lithium diisopropylamide (LDA), the proton nuclear magnetic resonance (1H NMR) spectrum of the resulting product showed the disappearance of the singlet at 8.7 parts per million (ppm) arising from C5(H) of 6. In the 13C NMR spectrum, the C5 carbon gives rise to a very broad resonance at 190 ppm, which is significantly downfield of the corresponding resonance for the precursor 6 (124 ppm). Although these data indicated that a deprotonation had occurred, the shape of the 13C NMR signal, as well as the calculated chemical shift (32) for the C5 carbon of the free aNHC 9 (205 ppm), suggested that the new compound was the aNHC lithium 7. Similar complexation has previously been observed for other singlet carbenes such as small NHCs (35) and bis (diisopropylamino) cyclopropenylidene (36, 37); in both cases, coordination of the lithium cation was apparent from the broadening and the upfield shift of the carbene 13C NMR resonance. Subsequently, we sought to sequester the lithium cation through addition of excess  crown-4 to a diethylether solution of 7 (where X is Br). This treatment induced a clean rearrangement to generate 8, which was isolated in 45% yield. This product formally results from the deprotonation of an isopropyl substituent of the Dip group by the carbene center of the aNHC lithium adduct 7, followed by nucleophilic addition of the resulting benzyl anion to C2. However, calculations indicated that the rearrangement of 9 into 8 is exothermic by only 6.1 kcal mol−1 and involves an activation barrier of 20.3 kcal mol−1 for the proton transfer. Therefore, we hypothesized that the observed rearrangement was catalyzed by a component of the (crown)LiBr system, and that the formation of 8 did not imply that the free aNHC 9 was too reactive to be isolated.
Sodium and potassium bases have proven more appropriate than lithium bases in generating free carbenes (35–37), because the corresponding carbon-heavy alkali metal bonds are more labile, which favors precipitation of the salt. When the deprotonation of imidazolium 6 (HX•X−, where X is Cl or Br) was performed with two equivalents (34) of potassium hexamethyldisilazide (KHMDS) in tetrahydrofuran, a clean reaction occurred, with the 13C NMR spectrum of the resulting product showing a very sharp signal at 201.9 ppm. After the products were worked up, the free aNHC 9 was isolated as a green powder (480 mg, 68% yield), and single crystals were obtained by recrystallization from a dry hexane solution at −78°C (Fig. 3, right).
In the solid state, both the free aNHC 9 and the imidazolium salt 6 (Br−) feature a fully planar ring (maximum deviation for N1–C2–N3–C4–C5–C21–C31–C41–C53 was 1.9 and 6.3 pm for 9 and 6, respectively), confirming the delocalization of the π system. This electronic structure is corroborated by the values of the endocyclic C–N [6: 1.335 ± 5 to 1.409 ± 5; 9: 1.354 ± 2 to 1.408 ± 3 Å] and C–C bond lengths [6: 1.351 ± 5, 9: 1.385 ± 3 Å], which are halfway between those of single and double bonds. The carbene bond angle N1–C5–C4 for 9 [101.03 ± 17°] is more acute than the corresponding angle in the cationic precursor 6 [108.0 ± 3°]. This feature is consistent with increased s character of the σ lone-pair orbital on the carbene atom in 9 as compared with the C–H+ bonding orbital in 6. A similar relationship is observed in NHCs and their NHC(H+) precursors (3, 4).
Calculations predict aNHC 9 to be 14.1 kcal mol−1 less stable than its isomeric normal NHC with the phenyl group bonded to C5 instead of C2. Figure 4 shows the two highest occupied molecular orbitals (HOMOs) of 9. The HOMO (−4.403 eV) is a σ-type lone-pair orbital at C5; the HOMO-1 (−4.879 eV) is a C5–C4 π-bonding orbital, which exhibits antibonding conjugation with the π orbital of the phenyl substituent at C4. These molecular orbitals are much higher in energy than those of the isomeric NHC (−5.000 and −5.279 eV, respectively), which indicates that aNHCs are more basic than NHCs. Indeed, calculations predict that the first (287.0 kcal mol−1) and second (144.6 kcal mol−1) proton affinity of aNHC 9 are significantly higher than those of normal NHCs (229.9 to 270.6 and 38.9 to 106.5 kcal mol−1, respectively) (38).
Although abnormal NHC 9 is sensitive to air and quantitatively rearranges to 8 upon heating in benzene at 50°C for 48 hours, it is stable at room temperature for a few days both in the solid state (melting point: decomposition at 65°C) and in solution (39). The availability of stable aNHCs not only provides easy access to a variety of transition-metal complexes, but also allows for their use as organocatalysts. As a proof of concept (40–43), (aNHC) AuCl complex 10 and (aNHC)-CO2 adduct 11 have been prepared in 79 and 95% isolated yields by simply reacting 9 with chloro (dimethylsulfide) gold(I) and CO2, respectively (Fig. 2).
Because the substituent at the C4 of 9 is a nonbulky benzene ring, a variety of substitution patterns should be tolerated without precluding isolation of the corresponding aNHC. The substituent at C4 is in conjugation with the carbene center, which opens the possibility of substantially modulating the electronic character of the ring system.
We thank NIH (grant R01 GM 68825) and the Deutsche Forschungsgemeinschaft for financial support of this work and the Consejo Nacional de Ciencia y Tecnología (E.A.P.) and Alexander von Humboldt Foundation (P.P.) for postdoctoral fellowships. Metrical data for the solid-state structures of 6 (Br−), 8, 9, and 10 are available free of charge from the Cambridge Crystallographic Data Centre under reference numbers CCDC-735754, CCDC-735755, CCDC-735756, and CCDC-744123, respectively.
Supporting Online Material
Materials and Methods