Weakly Stabilized Primary Borenium Cations and their Dicationic Dimers

Abstract

Hydride abstraction from monocationic hydride bridged salts [H(H₂B–L)₂]⁺ [B(C₆F₅)₄]⁻ (L = Lewis base) generates an observable primary borenium cation for L = iPr₂NEt, but with L = Me₃N, Me₂NPr, or several N-heterocyclic carbenes, highly reactive dicationic dimers are formed.

Borocations are accessible by activation of Lewis base-borane complexes with electrophiles.¹ Recent applications include aromatic or aliphatic C–H borylation and alkene hydroboration using activated reagents derived from amine boranes 1^2,3 or N-heterocyclic carbene boranes (NHC boranes; 2).^4,5 Both catalytic and stoichiometric activation methods were developed, but attempts to understand the stoichiometric activations encountered puzzling results. Thus, stoichiometric activation of typical 1 or 2 using strongly electrophilic “hydridophiles” was shown to generate transient species equivalent to the borenium cations 3 or 4, but borenium intermediates were not detected. Instead, we observed intermediate cationic hydride-bridged “dimers” 5 or 6,^6,7 apparently derived from a 3c2e interaction between the mostly empty borenium p-orbital and a nucleophilic B–H bond in the starting borane complex. In cases where nucleophilic anions were present, the tetracoordinate complexes 7 were observed instead of the borocations 5 or 6. However, it soon became clear that additional activated intermediates would have to be considered.

In our recently reported example of activation using the trityl salt 8 as hydridophile, amine borane 9 afforded a relatively stable H-bridged dimer 10 using 50 mol% of 8 (eq. 1).^2a No further reaction occurred over days at rt, but addition of a second 50 mol% portion of 8 to 10 resulted in facile (10 min, rt) conversion into the cationic borylation product 11. Clearly, the final 1:1 stoichiometry of trityl salt 8 to 9 was responsible for greatly increased reactivity, but we could only speculate about the nature of the activating event(s) pending further evidence.

Hints regarding the key activation process were obtained from several distinct directions, and the first important clue came from experiments with the NHC borane 2a. Thus, treatment of 2a with 1 equiv of AlCl₃ in CH₂Cl₂ (eq. 2) gave the expected solution intermediate 6a (¹¹B δ −24.3 ppm).⁸ When the solution containing 6a was treated with excess AlCl₃, a transient signal appeared at ¹¹B δ +11.7 ppm that was replaced by a dominant new signal at δ +47.3 ppm after 24 h at rt. Although neither the initial nor the final (relatively stable) solution species could be isolated, the dichloroborenium cation structure 12 for the final product was inferred from chemical shift comparisons with the analogous picoline-derived PicBCl₂⁺ Al₂Cl₇⁻ (δ +47.3 ppm).⁹

The details of the presumed disproportionation leading to 12 were not investigated, but we were intrigued by the identity of the +11.7 ppm species (unknown I). This chemical shift is at least ca. 15 ppm downfield compared to any known, formally tetracoordinate NHC borane complex (NHC–BH₂X, 7 with L = NHC),¹⁰ and suspicions inevitably turned to the borenium cation structure 4a. However, this structure did not fit intuitive or computational chemical shift evaluations (see Supporting Information).

Additional clues were obtained by studying amine boranes under conditions (bromobenzene solution instead of CH₂Cl₂) that afforded relatively long-lived activated intermediates. Thus, N,N-dimethylpropylamine borane 13 reacted with 50 mol% of trityl salt 8 in d₅-PhBr to give a stable solution of 14 (Scheme 2). Addition of a second 50 mol% of 8 drastically increased reactivity as usual, but also afforded a transient ¹¹B signal at δ +13.2 ppm (unknown II), similar to the NHC-derived +11.7 ppm species (unknown I) mentioned above. Within 48h at rt, the transient signals were almost entirely replaced by an H-coupled ¹¹B peak at δ +67.4 ppm. Along with observation of a very broad ¹H signal (1H) centered around 4.8 ppm, and by analogy with the cyclization of 9, this species could be assigned as the secondary borenium cation 15. The assignment is supported by reductive quenching with n-Bu₄NBH₄, which afforded ca. 80% of the borylation product 16. These observations raise the possibility that the +13.2 ppm transient (unknown II) is an intermediate in the borylation. Although the timescale for survival of unknown II was quite long (hours at rt), the unknown could not be isolated.

Scheme 2. Amine borane activation.

In an attempt to suppress internal borylation pathways and also to destabilize H-bridged intermediates, we explored activation of the hindered N,N-diisopropylethylamine borane 17. Thus, treatment with 100 mol% of the trityl salt 8 in d₅-PhBr gave two new boron species (4:1 ratio after 5 min at rt) as evidenced by ¹¹B NMR assay. While the major new cationic product can be assigned as 18 (δ −0.5 ppm), the minor product (δ +53.9 ppm) is clearly a tricoordinate boron cation, tentatively assigned the unprecedented nitrogen-bound primary borenium structure 19.¹¹ This structure is also supported by the observation of a broad ¹H resonance near δ 4.8 ppm, typical for B−H resonances in tricoordinate boron cations.¹² Further characterization or isolation of 19 was precluded by rapid decomposition and formation of complex unidentified species within minutes at rt, not to mention partial degradation of the [B(C₆F₅)₄]⁻ counterion over 2h at rt according to ¹H and ¹⁹F NMR data. Therefore, the structural assignment of 19 relies on comparisons with the known ¹¹B chemical shifts of non-stabilized, but more highly substituted borenium species including the tertiary cation 20 (δ +85.1 ppm in CD₂Cl₂)^2c and the secondary cations 11 (δ +71.1 ppm in CD₂Cl₂, +69.5 ppm in d₅-PhBr) and 15 (δ +67.4 ppm in d₅-PhBr). Neglecting the structural differences between 20, 11 and 15, replacement of a B-alkyl group by a hydrogen atom shifts the ¹¹B signal upfield by roughly 15 ppm. A comparable, additional upfield shift would be expected for the primary borenium cation 19, qualitatively consistent with the observed chemical shift of +53.9 ppm. This value is also consistent with the ¹¹B chemical shift computed using the GIAO method (+55.8 ppm in PhBr).

Similar activation of trimethylamine borane 21 using 100 mol% of 8 in d₅-PhBr gave contrasting results (Scheme 2). Although the initially formed product at rt was the previously reported hydride-bridged dimer 22,⁶ formation of a second product (¹¹B δ +12.9 ppm; unknown III) occurred more slowly. After 24 h at rt, the ratio of unknown III to 22 reached 2.5:1, and decomposition occurred more slowly than in the experiments starting from 17. Nevertheless, repeated efforts to isolate the highly sensitive unknown III by crystallization were not successful.

After a major effort in the amine borane series had encountered no example of an isolable species having the unexplained δ 11-13 ppm ¹¹B chemical shift, we opted to re-examine the NHC boranes. Activation of NHC–BH₃ complexes 2a-c with 0.5 equiv of the trityl salt 8 predictably resulted in clean formation of solutions containing the H-bridged cations 6a-c, respectively (X = B[C₆F₅]₄), with ¹¹B chemical shifts in the range of −22 to −25 ppm, ca. 10 ppm downfield compared to the starting 2a-c (Scheme 3). Additionally, salt 6c was prepared in benzene suspension, isolated as a sensitive crystalline solid in 97% yield, and fully characterized by ¹H, ¹¹B, ¹³C and ¹⁹F NMR spectroscopy.

Scheme 3. NHC borane activation.

Subsequent addition of a second 50 mol% portion of 8 to the solution of 6b resulted in another hydride abstraction event, producing additional Ph₃CH along with a new species having a ¹¹B NMR signal at δ +11.9 ppm (unknown IV), nearly identical to that (δ +11.7 ppm) observed initially for unknown I using AlCl₃ in the activation step (eq. 2). Formation of the new product was rather slow, taking several hours at rt to reach completion. In contrast, a similar activation sequence starting from the more substituted NHC borane 2d was quite fast, reaching full conversion of 6d to a new product (¹¹B δ +8.0 ppm; unknown V) within 10 min at rt. Attempts to obtain X-ray quality crystals were not successful, but the improved reactivity of 2d stimulated experiments with the analogous tetrasubstituted NHC borane 2e.

The standard activation method from 2e and trityl salt 8 afforded a finely divided solid having a solution ¹¹B chemical shift of δ +10.6 ppm, while activation with Ph₃C⁺ [HCB₁₁Cl₁₁]⁻ in PhF afforded single crystals suitable for X-ray crystallography. Two NHC subunits and two counterions were present, so the structure was clearly a dication. However, the boron atoms were disordered, and key features of the structure remained unresolved. On the other hand, the dicationic nature of the substance became obvious at this step, and was also consistent with the limited solubility of the solid.

Finally, after a lengthy optimization study, activation of 2e with Ph₃C⁺ Al₂Br₇⁻ in fluorobenzene afforded a relatively well behaved, but highly sensitive crystalline product that allowed acquisition of high quality X-ray diffraction data. Subject to the considerations discussed below, a definitive fit for the corresponding reactivity and X-ray data is the unprecedented dimeric hydride-bridged dication structure 23 (X⁻ = Al₂Br₇⁻; Fig. 1). By analogy, the standard activation procedure using 8 must afford a similar structure (23 with X^‒ = [B(C₆F₅)₄]⁻).

Figure 1.

ORTEP plot of 23 (X⁻ = Al₂Br₇⁻, the counterion has been omitted for clarity).

The X-ray structure of 23 conforms to the crystallographically imposed C_i symmetry, with the inversion center rendering both “NHC–BH₂⁺” subunits equivalent. The same symmetry considerations define the mutually parallel orientation of the mean planes of both 5-membered rings with interplane separation of 0.46Å. Another bonding parameter of interest is the C–B bond distance of 1.58Å, which is only marginally shorter than the corresponding C–B bond in the starting carbene borane 2e (1.59Å).¹³ The crucial B···B distance in 23 is 1.78Å, and is essentially identical to B···B distances measured by X-ray crystallography in neutral RBH₂ dimers (1.78-1.79Å)^14,15 as well as to the B···B distance in the parent B₂H₆ determined by electron diffraction (1.77Å)¹⁶ or X-ray methods (1.78Å).¹⁷ A substantially shorter B···B distance has been established for a structure containing a B–B bond (X-ray,1.49Å; neutron diffraction, 1.48Å).¹⁸ Hydrogens near the central boron atoms were located on difference Fourier maps and were allowed to refine isotropically as independent atoms. The exact location of the B–H hydrogens is uncertain due to their inherently low contribution to the electron density map. However, there can be little doubt that each boron retains two attached hydrogens, based on quenching studies described in the next paragraph, and no doubt that 23 has the indicated overall symmetry.

Addition of p-dimethylaminopyridine to freshly generated 23 (X = B[C₆F₅]₄) as a suspension in bromobenzene afforded a homogeneous solution of the boronium salt 24 (¹¹B δ −14.7 ppm), identical to the fully characterized material prepared independently from 6e (see Supporting Information). Furthermore, treatment of 23 with n-Bu₄NBH₄ resulted in clean conversion to the starting NHC borane 2e. Finally, addition of 2e to 23 afforded the monocationic 6e. In each of these quenching experiments, a single product was formed exclusively according to NMR assay. This evidence provides strong support for the presence of an intact “NHC–BH₂⁺” subunit in 23, while the X-ray and ¹¹B chemical shift data support the dimeric, dicationic structure (GIAO computed chemical shift δ +10.8 ppm vs. 10.6 ppm observed).

Based on the analogy of ¹¹B chemical shifts, all of the unknowns I-V can be assigned as the hydride bridged dicationic dimers with reasonable confidence (Fig. 2), even though 23 is the only structure to be fully defined. Analogous hydride-bridged dications are known in transition metal chemistry,¹⁹ but we could find no main group precedents for 23. We did find diverse dicationic and tricationic boron-based structures,²⁰ but no hydride-bridged dimers potentially capable of dissociating into a primary borenium cation. In contrast, reversible dissociation is implicit for all of the species shown in Fig. 2. Furthermore, the detection of a primary borenium monomer 19 under similar activation conditions indicates that the hydride bridged dimeric dications are thermodynamically more stable in most, but not all of the cases studied.

Figure 2.

Proposed structures of unknowns I-V.

The balance between dimer and monomer depends on the interplay between steric as well as electronic factors. The monomeric borenium salt [(Ph₃P)₂C→BH₂]⁺ [BH(C₆F₅)₃]⁻ isolated and characterized by Alcarazo et al. is stabilized by the sterically demanding Ph₃P subunits, as well as by σ and π donation from the carbon-bound ligand.^8a,21 In contrast, all of the NHC-derived (carbon-bound) borenium salts (Fig. 2) prefer the unprecedented dicationic dimer structures (IV, V, 23), presumably because of lower π donation from the NHC subunit. Analogous nitrogen-bound borocations are far more reactive due to minimal stabilizing factors and increased electron demand from the electronegative, cationic amine ligand. Thus, the dicationic dimer II was detected as a labile solution intermediate during the conversion of amine borane 13 into the cyclic borenium salt 15, while the more hindered amine borane 17 was converted into an observable, but highly reactive primary borenium salt 19. Apparently, 19 prefers the monomeric structure in part because steric repulsion between the isopropyl groups destabilizes the corresponding (undetected) dimeric dication.

Since the dimeric dication 23 is now proven to be an energy minimum, we assume that dicationic transition states are also possible. Thus, the direct interaction of 6e with the trityl salt 8 becomes a plausible activation pathway leading to 23.²² The role of dication intermediates in the C–H borylation events also remains uncertain, but the options are more clear. The dissociation of dimeric dications into monomeric monocations followed by C–H insertion provides the simplest explanation, but the dimeric dications may already be sufficiently electrophilic for direct insertion into a C–H σ-bond.

Scheme 1. Lewis base borane complex activation.