Superelectrophilic Intermediates in Nitrogen-Directed Aromatic Borylation

Abstract

The first examples of borylation under conditions of borenium ion generation from hydrogen-bridged boron cations are described. The observable H-bridged cations are generated by hydride abstraction from N,N-dimethylamine boranes Ar(CH₂)_nNMe₂BH₃ using Ph₃C⁺ (C₆F₅)₄B⁻ (TrTPFPB) as the hydride acceptor. In the presence of excess TrTPFPB, the hydrogen-bridged cations undergo internal borylation to afford cyclic amine borane derivatives with n = 1-3. The products are formed as the corresponding cyclic borenium ions according to reductive quenching experiments and ¹¹B and ¹H NMR spectroscopy in the case with Ar = C₆H₅ and n = 1. The same cyclic borenium cation is also formed from the substrate with Ar = o-C₆H₄SiMe₃ via desilylation, but the analogous system with Ar = o-C₆H₄CMe₃ affords a unique cyclization product that retains the tert-butyl substituent. An ortho-deuterated substrate undergoes cyclization with a product-determining isotope effect of k_H/k_D 2.8. Potential cationic intermediates have been evaluated using B3LYP/6-31G* methods. The computations indicate that internal borylation from 14a occurs via a C–H insertion transition state that is accessible from either the borenium π complex or from a Wheland intermediate having nearly identical energy. The Ar = o-C₆H₄SiMe₃ example strongly favors formation of the Wheland intermediate, and desilylation occurs via internal SiMe₃ migration from carbon to one of the hydrides attached to boron.

Introduction

The structural chemistry of cationic, trivalent boron environments (“borenium” ions according to the Nöth terminology)¹ has attracted interest over many years because of the similarity with the isoelectronic carbenium ions in terms of orbital occupancy, electron count, and net charge. In an early investigation, Ryschkewitsch and Miller reported NMR evidence that the cation 2 is in equilibrium with the picoline-BCl₃ complex 1 in the presence of excess aluminum chloride (Fig. 1).^2a More recently, Fujio et al. found that the pyridine-diphenylchloroborane adduct 3 is converted into 4 using SbCl₅ as the chloride abstracting agent.^2b According to the analogies initially recognized by Olah et al.³ and also noted by Nöth in his excellent review,¹ salt 4 is isoelectronic with trityl cation, while 2 is analogous to dichlorobenzyl cation. By the same analogy, borenium ions have been included in Olah's classification of superelectrophiles, along with several other mono-cationic species that have a positively charged heteroatom adjacent to an unoccupied p-orbital.³ Several O- and/or N-substituted borenium species have been detected in structural studies using spectroscopic^1,2,4 and crystallographic techniques,^5,6 as summarized in the review literature.^1,7

Fig. 1.

Generation of borenium ions in solution

According to prior work, borenium ions are potent electrophiles that may approach the more familiar carbenium⁸ or silylium⁹ cations in terms of reactivity. Indeed, the challenge of obtaining X-ray quality crystals of a structure related to 4 was met only recently,⁶ and required highly hindered B-aryl groups to prevent boron coordination by external electron donors. Borenium species containing nitrogen or oxygen electron pair donors are more stable because delocalization partially satisfies electron demand at boron, but they retain substantial Lewis acidity and act as catalysts in several important applications. The best known example is the enantioselective Corey-Bakshi-Shibata reduction of ketones via the intermediate 5, generated in situ from an oxazaborolidine and a borane source (Fig. 2).¹⁰ Although 5 does not carry a net positive charge, a borenium subunit can be recognized along the N–B–O segment due to the formally positive nitrogen. Structurally similar, but far more potent borenium electrophiles 9 can be generated from neutral precursors 7 by protonation at nitrogen.¹¹ In this case, 9 does carry a net positive charge and serves as a highly reactive Lewis acid catalyst despite the moderating influence of oxygen electron pairs and the tendency of triflate to form a covalent adduct at boron (8). In other applications, related O- or N-substituted (stabilized) borenium ion intermediates may be involved in the epimerization at boron in several families of chiral heterocycles,¹² while non-stabilized borenium species may play a role in C-F bond cleavage reactions,¹³ the abstraction of hydride from amine boranes using trityl or diarylmethyl cations,¹⁴ and perhaps also in some of the hydroboration chemistry of iodoborane complexes.¹⁵ Given the complexity and debatable information content of formal charges in structures related to species such as 5, 6, or 9, we omit the charges at individual atoms in most of the subsequent drawings to allow focus on the far more important net charge.

Fig. 2.

Oxazaborolidines as borenium ion precursors

In the absence of stabilizing heteroatom electron pairs or hindered aryl substituents at boron, borenium ions should be exceptionally reactive electrophiles. Furthermore, the structural analogy with carbenium ions resulting from the net positive charge and vacant p-orbital at boron suggests potentially important applications for the formation of C-B bonds. Isolated examples of relevant electrophilic borylation chemistry have been encountered over the years, usually under relatively drastic conditions (boron halide/aluminum trichloride),¹⁶ but the possible involvement of borenium species has been largely overlooked.¹⁷

Given the intensive current interest in transition metal-catalyzed applications of aryl and alkyl boranes and boronic acids,^18,19 it is time to revisit mechanistic options available to electrophilic boron for C-B bond formation. Little is known regarding the reactivity of borenium cations with carbon nucleophiles. We were especially interested to learn whether the tendency of trivalent boron to form three-center two-electron (3c2e) bonds would enable or impede the Friedel-Crafts electrophilic substitution pathway in an intramolecular context, but the mechanistic analogy with carbenium ion chemistry was a larger consideration. Therefore, our work began with the investigation of a borenium ion analogy for the classical Friedel-Crafts cyclization using benzylic amine boranes as the substrates. The study detailed below has demonstrated a series of relevant cyclizations, and has encountered evidence for a mechanism that has implications for electrophilic borylation chemistry beyond the nitrogen-directed examples described herein.

Methods and Results

A prior study in our laboratory generated the non-stabilized borenium ion 11 from triethylamine borane 10 by hydride abstraction with trityl cation and found that 11 is trapped efficiently by the starting complex 10 to form the hydride-bridged cation 12 (Fig. 3).²⁰ The 3c2e bond in 12 increases electron density at boron compared to the borenium ion 11, but 12 is a capable electrophile nevertheless, and undergoes bonding interactions with weak nucleophiles including triflate and bistriflimidate anions, trialkylsilanes, and dichloromethane.

Fig. 3.

Hydride-bridged borenium species

The above observations indicate that 12 acts as an in situ source of borenium species equivalent to 11. We therefore performed the analogous activation of N,N-dimethylbenzylamine borane 13a while monitoring intermediates by NMR spectroscopy (Scheme 1). Treatment of 13a with trityl tetrakis(pentafluorophenyl)borate (TrTPFPB) at −78 °C in CD₂Cl₂ and assay by ¹H NMR at −20 °C showed the expected conversion of 13a to 15a, according to an upfield peak at δ ¹H = −1.9 ppm and a ¹¹B chemical shift at δ 0.0 ppm. No signals for trivalent boron species such as 14a were detected. However, a highly deshielded peak did appear in the range expected for trivalent boron (δ ¹¹B = 59 ppm) when a similar experiment was performed in the more robust solvent C₆D₅Br at room temperature. The new boron signal was not consistent with the 1:2:1 triplet expected for 14a, but could be interpreted as a barely resolved, broad doublet (J ca. 150 Hz) by comparing proton-coupled and -decoupled spectra. This magnitude of B–H splitting would be consistent with the sp² environment in a free borenium ion, but the multiplicity requires a single proton at boron. Capture of 14a by an external or an internal nucleophile followed by a second hydride abstraction would satisfy the multiplicity requirement, and suggested several possible structures, but the question of cation identity was quickly resolved when addition of Bu₄NBH₄ to quench the reaction mixture produced the known benzazaborolidine 17a²¹ (72% isolated). The δ ¹¹B = 59 ppm signal must therefore be due to the trivalent boron cation 16a, stabilized by “bora-benzylic” delocalization between the formally unoccupied boron p-orbital and the aromatic π-electrons. Structure 16a also helps to understand an unusually broad, strongly deshielded signal at δ ¹H = 5.9 ppm that is coupled to boron and integrates to 1H, and can therefore be assigned as the B–H proton.

Scheme 1.

An earlier encounter with cation 16a has been reported from our laboratory, starting from the benzazaborolidine 17a.¹³ Hydride abstraction using trityl tetrafluoroborate in the presence of pyridine readily afforded the pyridine adduct 18 (as the tetrafluoroborate salt), but attempts to detect intermediates by NMR initially gave complex results. Using TrTPFPB as the hydride acceptor at −78 °C converted 17a into species having unidentified broad NMR signals, including a transient signal at δ ¹¹B = 38.7 ppm that disappeared upon adding pyridine at −50 °C or warming to room temperature. The δ 38.7 ppm signal was tentatively attributed to 16a, but anomalies were noted that could not be explained, including partial recovery of 17a after the pyridine quench. In the current study, the δ 38.7 ppm signal was detected at δ 39 ppm as a minor peak in experiments starting from either 13a or 17a (conditions designed to minimize contamination by water) while the major signal was observed at δ ¹¹B = 59 ppm. However, the δ 39 ppm signal (broad singlet; no proton coupling) became major if one equivalent of water was added to the solution obtained from 17a with TrTPFPB in CD₂Cl₂ at room temperature, and was therefore assigned as the hydroxyborenium ion 19. In support of this assignment, addition of a second equivalent of water produced the protonated boronic acid 20 (δ ¹¹B = 29 ppm; δ ¹H = 5.29 (2H, br s, OH), 4.28 (2H, d, benzylic CH₂), 2.89 (6H, d, NMe₂) ppm), identical to the salt formed by protonation of known boronic acid 21 followed by anion metathesis and extraction into CD₂Cl₂.

Having clarified the identity of the signal at δ ¹¹B = 39 ppm, we returned briefly to the NMR experiment from 17a using TrTPFPB activation, but under conditions expected to favor the formation of hydride-bridged cations (50 mol% TrTPFPB). The dominant species formed in C₆D₅Br was assigned as the hydride-bridged structure 22, based on δ ¹¹B = 11 ppm. This signal is downfield compared to neutral 17a, but far upfield from the signals of B-hydroxyborenium cation 19 or the borenium ion 16a. In the cleanest experiments, the only other ¹¹B NMR signal detected was that due to the TPFPB anion. However, distinct maxima for 16a or 17a were present in addition to the 11 ppm signal of 22 if the amount of TrTPFPB used for activation of 17a was adjusted to 83 mol% or 33 mol%, respectively. The proton chemical shifts were less characteristic, but the shift values varied as the mol% of TrTPFPB used for cation generation was changed. This behavior is consistent with an equilibrium between 22 and 17a + 16a that is fast on the ¹H NMR timescale, but slow on the ¹¹B NMR timescale. Quenching cation 22 with pyridine generates the adduct 18 previously isolated as well as recovered 17a. Thus, the hydride bridged cation 22 apparently was present as a latent source of 16a in the original experiment starting from 17a,¹³ while a different species (the hydroxyborenium ion 19) was responsible for the trivalent boron chemical shift observed (δ ¹¹B 39 ppm) in the complex NMR spectra resulting from water contamination. We note that structure 16a as re-defined in the current study remains as the only borenium ion detected to date that contains a B–H bond, but it is now clear that 16a has the δ ¹¹B = 59 ppm chemical shift.

To establish the scope of conversion from substituted benzylamine boranes into cyclic amine boranes, several experiments were conducted with modified substrates (Table 1). A comparison of solvents for the cyclizations showed that bromobenzene (or other halobenzenes) gives higher conversion and better isolated yield of 17a (72%) compared to toluene (48%) or dichloromethane (27%), so the conditions developed for the bromobenzene NMR experiments were used for the other entries of Table 1 without optimization of individual examples,²² followed by Bu₄NBH₄ reductive workup. A slurry of NaBH₄ in diglyme also gave an acceptable yield in the case of 17a (63%). Reductively quenched reaction mixtures were loaded directly onto silica gel for FC purification even though the bromobenzene caused cracking of the silica column. This resulted in poor separation for some cases, but the simple technique allowed convenient solvent removal and reasonable recoveries of 17 in addition to 10-20% unreacted 13 in typical experiments. Overall, the conversions were modest for some examples, but cyclized products were easily obtained over a range of aromatic substitutuents. Halogen substituents required longer reaction times for conversion to 17 (entries 5,6,8), especially in the case of the ortho-halogen derivatives (entries 10-12) by comparison with the para-isomers, suggesting the possibility of non-productive formation of a B–X bond between electrophilic boron and the ortho-halogen. Longer tethers were also tolerated (entries 13, 14), although activation of 25 resulted in slower cyclization (30-40% 25 recovered after 16 h), and formed 26 along with degradation products that could not be separated from 26.

Table 1.

Nitrogen-Directed Borylation^a

entry	substrate	R	n	time	Product (%)b
1	13a	H	1	4 h	17a (72%)
2	13b	p-Me	1	4 h	17b (41%)
3	13c	m-Me	1	4 h	17c (79%)c
4	13d	o-Me	1	4 h	17d (76%)
5	13e	p-Br	1	8 h	17e (53%)
6	13f	p-Cl	1	8 h	17f (73%)
7	13g	m-Cl	1	4 h	17g (67%)d
8	13h	p-F	1	16 h	17h (59%)
9	13i	m-F	1	4 h	17i (67%)e
10	13j	o-Br	1	16 h	17j (55%)
11	13k	o-Cl	1	16 h	17k (55%)
12	13l	o-F	1	16 h	17l (39%)
13	23	H	2	4 h	24 (74%)
14	25	H	3	16 h	26 f

^aReactions at room temperature in C₆H₅Br, 90 mol% TrTPFPB, followed by quench with Bu₄NBH₄.

^bYields based on TrTPFPB as limiting reagent.

^c3:1 mixture of inseparable regioisomers.

^d1:1.3 mixture of regioisomers.

^e4:1 mixture of regioisomers.

^fca. 30% (major product) + ca. 10% contaminants

To gain further insight into events leading to 16, the product-determining step was probed by a deuterium labeling study. The monodeuterated substrates 13a-D₁ and 13b-D₁ were prepared by D₂O quench of the corresponding ortho-lithiated N,N-dimethylbenzylamines 27²³ and borane complexation, giving 95% deuterium incorporation (eq. 1). The directed borylation of these substrates can occur either with loss or retention of deuterium in the product, so the ratio of 17-D₁ to 17 corresponds to k_H/k_D. Substrate 13b-D₁ proved more suitable for this study compared to 13a-D₁ because the ¹H NMR signal for the ortho-C–H of 17a-D₁ overlaps with another aromatic proton signal while all aromatic proton signals for 13b-D₁ are fully resolved in the 500 MHz ¹H NMR spectrum. Accordingly, 13b-D₁ was treated with TrTPFPB as usual, and the product 17b (41% obtained after Bu₄NBH₄ quench) was assayed using ¹H NMR. Based on the ratio of deuterium-free vs. deuterated products, the magnitude of the kinetic isotope effect (KIE) was found to be k_H/k_D = 2.8. This result indicates that the C–H(D) bond at which boron substitution occurs is broken during or before the regioselectivity-determining step (see Discussion).

(1)

Given that proton removal plays a unique role in conversion to products, we investigated additional substrates designed to provide insight regarding mechanistic details of the electrophilic borylations. In the first series, the aryllithium reagent 27a was trapped with TMSCl, followed by conversion into the amine borane 28 as usual. Treatment of 28 with TrTPFPB at room temperature resulted in efficient conversion to desilylated cyclization products, and reductive quenching with Bu₄NBH₄ gave 17a in 96% isolated yield. Even 10 mol% of the trityl salt was sufficient for 91% conversion of 28 to cyclization products within 1h at room temperature, in striking contrast to the behavior of 13a. In the latter case, the only product observed under similar conditions was the hydrogen-bridged dimer 15a, and no cyclization occurred at room temperature unless >50 mol% of TrTPFPB was used for the activation as in Table 1 (ca. 40 mol% TrTPFPB is present in addition to the amount needed to generate 15). Evidently, the TrTPFPB activates 15a for cyclization by promoting the release of the more reactive borenium ion 14a. However, no hydrogen-bridged intermediate analogous to 15a could be detected starting from 28 and TrTPFPB, and additional TrTPFPB was not required to effect the cyclization. According to this evidence, the hydrogen-bridged structure 29 would have to be considerably more reactive than 15 due to the presence of silicon, sufficient to undergo spontaneous cyclization via 30.

When an experiment from 28 and equimolar TrTFPB was monitored by ¹¹B NMR spectroscopy, we were surprised to find that the chemical shift for the cationic cyclization product does not match the δ 59 ppm value found for 16a. Instead, the observed value was δ ¹¹B = 42 ppm. We attribute the chemical shift difference to an equilibrium involving the 3c2e hydrogen-bridged silane adduct 31, formally corresponding to the interaction of 16a with Me₃SiH formed during the electrophilic borylation. Control experiments in bromobenzene-D₅ support this premise, and indicate that the chemical shift of the equilibrating mixture of 16a + 31 moves upfield as the proportion of added silane increases.²⁴ Both 16a and 31 are converted into 17a by the borohydride quench according to this interpretation.

The last series of experiments compared the above cyclizations with the analogous reaction starting from the o-tert-butyl substrate 34, available from the benzyne-derived o-tert-butylbenzaldehyde 32 via reductive amination as shown in Scheme 3. The standard activation procedure from 34 was performed in bromobenzene at room temperature, and reductive workup provided the cyclization product 35 in 66% yield. Loss of the tert-butyl group had been anticipated as a possible outcome in this reaction given the behavior of the silicon analog 28, but no such products were detected (<2% of 17a).

Scheme 3.

Discussion; Evaluation of Potential Reactive Intermediates

Most electrophilic aromatic substitution reactions proceed with no KIE,^25,26 although many exceptions are known.^25c,27 Negligible KIE has been taken as evidence not only that a σ-bonded cationic (Wheland) intermediate is involved in the reaction, but also that its formation is rate-limiting. On the other hand, the interpretation of a significant KIE can be more challenging. In a recent example, the acylation of toluene using the mixed anhydride PhCO₂Tf was characterized by k_H/k_D = 1.14 in the presence of the hindered base 2,4,6-tri-t-butylpyridine, but a substantially larger value of 1.85 was observed in the presence of TfOH.^27h Under the conditions with base added, k₂[B] is much greater than k₋₁ and k₁ is the rate-determining step (eq. 2). However, with TfOH added, k₂[B] decreases relative to k₋₁ and deprotonation becomes rate limiting.

(2)

For the electrophilic borylations, a small modification of the above argument would explain the value of KIE = 2.8 observed for conversion from 13a to 17a. In contrast to typical electrophilic aromatic substitutions, this process is conducted under exceptionally non-basic conditions because the conjugate acid of the TPFPB anion would correspond to a superacid that is known only as the etherate complex.²⁸ In this situation, the bromobenzene solvent is one candidate for the “strongest” external base, while various boron bonds, internal as well as external, constitute the alternative choices. Hydridic B–H bonds in amine boranes do have basic properties, although direct protonation of the B–N σ-bond is competitive under some conditions.²⁹ In any event, C–H bond breaking could well be rate limiting in the absence of adequate external base, resulting in k₂[B] << k₋₁ in Scheme 4. The sequence of events might then proceed from the observable H-bridged 15a via transient intermediates including the borenium ion 14a and the Wheland intermediate 36, followed by slow proton transfer and aromatization.³⁰

Scheme 4.

If no external base “B” is capable of removing the proton from 36, then k₂[B] will be too small to account for facile product formation at room temperature. In this scenario, conversion from 36 directly to the observed intermediate 16a may occur by evolution of hydrogen as the slow step, and without the intermediacy of 17a (Scheme 5; eq. 3). Another possible explanation for the observed KIE is that 36 is not on the pathway leading to products (i.e., negligible contributions from k₂[B], Scheme 4, and eq. 3, Scheme 5). Instead, conversion from 14a to the initial product 16a might take place by a C–H insertion process involving the 3c2e interactions represented by structures 37 and 38 (eq. 3b). In some respects, this sequence is reminiscent of mechanisms proposed for transition metal C–H insertions involving aromatic substrates,³¹ but the 3c2e bonding interaction between the mostly vacant borenium p-orbital of 14a and the arene σ C–H bond leads to a cationic borenium-hydrogen complex 38. Related hypervalent species may be involved in high temperature hydrogen transfer reactions catalyzed by trialkyl boranes,³² the recent hydrogen activation experiments of Stephan et al,³³ and gas phase equilibria involving BH₂(+), H₂, and BH₄(+).³⁴ The relationship between 38 and 16a has a close parallel in the cationic ammonia borane derivatives [H₃N•BH₄]⁺ (borenium ion hydrogen adduct) and [H₃N•BH₂]⁺ (borenium ion), structures that have been evaluated computationally.³⁵

Scheme 5.

(3)

The contrasting behavior of the trimethylsilyl (28) and tert-butyl (34) substrates is especially interesting in the mechanistic context. The simplest interpretation in the silicon case invokes formation of a transient hydrogen-bridged cation 29 followed by spontaneous cyclization to the Wheland intermediate 30 and desilylation (Scheme 6). Facile conversion to 30 is due to stabilization by the well known beta effect of silicon in the ipso substitution.³⁶ According to the extensive studies of Lambert et al. and Reed et al.,9a,b the naked cation Me₃Si(+) cannot simply “fall off”, but the solvents used in our study (bromobenzene, toluene) would be sufficiently nucleophilic to assist in the desilylation step from 30 by coordination at silicon.^27g Alternatively, desilylation from 30 might occur via bonding between silicon and an adjacent (nucleophilic) H–B bond to give 31 followed by formation of 16a upon loss of Me₃SiH. This alternative pathway reverses the order of events after generation of the Wheland intermediate 30, but does not change the overall result, namely the facile conversion to 16a using stoichiometric trityl activation, or to 17a using 10 mol% of TrTPFPB. No silicon-containing products were detected that might have been formed via proton removal from the isomeric Wheland intermediate 39. This observation is consistent with exclusive formation of the more stabilized 30 in the cyclization step, or reversible formation of both 30 and 39, followed by product determining desilylation.

Scheme 6.

For somewhat different reasons, an equally simple scenario might have been expected in the tert-butyl case. Activation of 34 with TrTPFPB would generate 40 as usual, and conversion to Wheland intermediates 41 and 42 is feasible in principle. However, in contrast to the silicon analogy, loss of the cation Me₃C(+) from 41 should not require nucleophilic assistance by solvent. Protonated tert-butylbenzene 43 is known to fragment to benzene and Me₃C(+) in superacid solution at temperatures well below −30 °C.³⁷ These conditions rule out assistance by nucleophiles or by base, and the analogy argues that 41 would undergo unassisted fragmentation to the amine borane 17a and Me₃C(+). Because this was not observed, we conclude that 41 was never formed. Without the stabilizing beta effect of silicon, it is easy to believe that 41 would not be formed, but similar logic suggests that 42 may also not be formed if an alternative, lower energy pathway is available to explain the conversion from 34 into 35. We have therefore explored the possibility that a C–H insertion pathway may be a viable alternative for this conversion as well as for the related events shown in Scheme 5.

The geometry and energy of potential cationic intermediates, starting with the presumed borenium ion 14a, was evaluated using B3LYP/6-31G* calculations performed using the Gaussian 03 program package.³⁸ An energy minimum (E_rel = 0.0 kcal/mol) was found for a π-complex (Min_A, Fig. 4), showing interaction between the benzene ring and the empty p orbital at boron with two short r_CB contacts at 2.38 Å for C(1) and 2.55 Å for C(2), respectively. A second local minimum (E_rel = 0.43 kcal, corrected for zero-point energy) was assigned as the Wheland intermediate 36 = Min_B based on the shorter r_CB contacts compared to Min_A, 2.31 Å for C(1) and 2.01 Å for C(2). The transition structure TS_AB (E_rel = 0.27 kcal/mol) was also identified along the relatively flat energy surface from Min_A to Min_B, similar structures that lead to TS_BC (E_rel = 18 kcal/mol; Fig. 4). Subsequent exchange between products and H₂ can occur through an isomeric transition state TS_Cexch (E_rel = 22 kcal/mol). MP2 calculations with the cc-pVTZ basis confirm the results obtained with B3LYP to within a few kcal/mol, suggesting that the broad features of the potential energy surfaces are correct.³⁹

Fig. 4.

B3LYP/6 31G* energies for cationic structures from 14a to 16a

In the context of Scheme 5, TS_BC corresponds to the 3c2e transition structure 37 for the C–H insertion pathway and the energy barrier relative to the π-complex (Min_A) is consistent with cyclization at room temperature. For the reverse reaction from 16a + H₂, the relatively high enthalpic barrier (ca. 27 kcal/mol, corresponding to somewhat higher ΔG^‡) suggests that the exchange reaction between 16a and molecular hydrogen should be too slow to be detected, but the possibility was explored experimentally (Eq. 4). Thus, 16a was generated independently by reaction of 17a with TrTPFPB, and the suspension in benzene was stirred under D₂ (ca. 2-3 atm). After 15-30 days, the heterogeneous mixture was quenched with Bu₄NBH₄ and ca. 20% of 17a was recovered with variable deuterium incorporation (MS assay). However, extensive decomposition of 16a was evident, and resulted in the formation of borane complex 46 (Eq. 4; ca. 20%) as one component of precipitated material that could not be dispersed reproducibly during the deuterium incorporation attempts. Formation of 46 was confirmed in control experiments in the absence of deuterium (46% isolated after 24 h reflux in benzene; 10 % after 12 d at rt). Faster decomposition was observed in bromobenzene (>90% conversion within 24 h at rt), but other significant decomposition products were formed in addition to 46. In the best deuterium experiment, ca. 40% D₁ incorporation was measured in both 46 and recovered 17a. These results exceed reaction rate expectations based on the calculated value of TS_BC, depending on the precision of the computations, but cautious interpretation is appropriate in any event, given the uncertainties regarding mechanism under the highly dilute, heterogeneous reaction conditions used for the deuterium incorporation studies. Among other concerns, decomposition from 16a to 46 presumably also generates (C₆F₅)₃B, a potent electrophile that may play a role in the deuterium exchange in view of its known interaction with molecular hydrogen.³³

(4)

Byproduct 46 was not formed in substantial amounts in the stoichiometric cyclization experiments from 13a and TrTPFPB under the usual conditions (4 h, rt), although traces of 46 were observed by NMR assay after 9h. On the other hand, no sign of 46 nor other decomposition products was detected when 13a was activated using 5% TrTPFPB in early attempts to achieve catalytic conversion to 17a, even at toluene reflux temperatures. These experiments encouraged the investigation of more forcing conditions. Remarkably, heating 13a with 5% TrTPFPB in toluene (160 °C bath, sealed tube) followed by quenching with Bu₄NBH₄ resulted in efficient conversion to 17a (90% isolated). Evidently, the presence of excess amine borane ensures hydride transfer to the cation 16a, and suppresses formation of the byproduct 46 in the catalytic procedure.

Attention was now turned to the cationic trimethylsilyl-containing structures generated starting from 28 (Fig. 5). Attempted optimization of a presumed π-complex analogous to Min_A (Fig. 4) having the BH₂ subunit placed near the Me₃SiC(2)–C(1) segment resulted instead in the Wheland intermediate Min_D = 30 (E_rel = 0.0 kcal/mol, Fig. 5). The structure is clear from the almost fully formed B–C(2) bond (1.67 Å) with boron nearly in the plane of the aromatic ring, and a somewhat elongated C(2)–Si bond (2.19 Å) at an angle of 105° relative to the ring. A low-lying TS_DE was found just 5.9 kcal/mol above Min_D that leads to formal migration of Me₃Si(+) from carbon to the adjacent boron-bound hydride. The result is conversion to Min_E = 31 (E_rel = −14.2 kcal/mol) with a hydride bridge linking boron (r_BH = 1.41 Å) and silicon (r_SiH = 1.61 Å), followed by B–H–Si dissociation (Min_F). On the other hand, when the initial optimization was performed with the BH₂ subunit rotated to be near the benzene C(1)–C(6) segment, a local minimum corresponding to a π-complex Min_G was found (E_rel = 6.3 kcal/mol). The higher energy path for loss of hydrogen from Min_G via TS_GH (E_rel = 23.5 kcal/mol) is analogous to the sequence of Fig. 4, but it does not compete with the more facile Si migration pathway from Min_D to Min_E. These results reflect substantial cation stabilization by the β-silyl group in Min_D compared to the regioisomeric Wheland intermediate derived from borylation at benzene C(6). Furthermore, the computations reveal a low-energy mechanism for aromatization from Min_D via TS_DE that does not require participation by an external nucleophile or base.

Fig. 5.

B3LYP/6 31G* energies for cationic structures from 28 to 16a

The remaining stages of the conversion from Min_E to 16a (stoichiometric) or 17a (catalytic) are not fully depicted in Fig. 5 because dissociation of a B–H bond into Me₃SiH and 16a (Min_F, E_rel = −6.0 kcal/mol) or dissociation of a Si–H bond into Me₃Si(+) and 17a (E_rel = + 32 kcal/mol) would be followed by adduct formation involving the high energy borenium or silylium cations and various external electron donors. The specific details would include interactions between 16a and solvent (stoichiometric conditions via Min_F), or between Me₃Si(+) and potentially bridging B–H bonds from unreacted 28 or with solvent (catalytic conditions). No attempt was made to evaluate the relevant energy profiles, but the latter pathway formally involving Si–H dissociation followed by intermolecular hydride transfer from 28 to Me₃Si(+) would nicely explain the facile catalytic conversion from 28 to 17a using 10% TrTPFPB. For similar reasons, the details of the reverse reaction from Me₃SiH and 16a (Min_F) were also not evaluated in detail. However, a transition state was found (E_rel = −2.3 kcal/mol; TS_EentE, not illustrated) for the reversible migration of Me₃Si between the two B–H hydrogens in Min_E (31). The activation barrier from Min_E is quite small (11.9 kcal/mol) and meets one of the requirements for facile isotopic exchange with an external Si–D bond. Accordingly, independently generated 16a was stirred with excess Et₃SiD in benzene at room temperature. After 10 min, 16a was isolated by precipitation with hexane, and was assayed by ¹H NMR spectroscopy. Only 8% of residual B–H signal intensity was found in recovered 16a, as expected for the reverse reaction from Min_F (16a + R₃SiH) followed by fast H/D exchange²⁰ at the stage of Min_E. This decisive experiment provides qualitative confirmation for the relative energies deduced by the computational method. It is also consistent with the observed change in the ¹¹B chemical shift of 16a upon addition of R₃SiH as mentioned earlier.

Finally, the computations were used to evaluate cationic intermediates in the tert-butyl series starting from 34. Most features of the pathway leading to 35 were found to be analogous to those shown in Fig. 4 from cation 14a to 16a. However, attempted optimization of a π-complex with boron interacting with C(2) resulted in a local energy minimum corresponding to the less hindered π-complex rotamer where boron interacts with the unsubstituted C(6). Constrained optimization indicates that the more hindered rotamer would be ca. 10 kcal/mol less stable, suggesting that simple steric repulsions are the reason why intermediate 41 is never formed. Subsequent events from the π-complex to structures 44 and 45 resemble those of Fig. 4, but the local minimum corresponding to the Wheland intermediate 42 was not found (see Supporting Information for details).

Summary

Our findings support the involvement of borenium species or hydrogen-bridged cations such as 15 and 29 in the borylations, and they raise the intriguing prospect of C–H insertion mechanisms via borenium ion intermediates at room temperature.^40,41 According to the B3LYP/6-31G* computations, the rate determining step can be described as a C–H insertion at the stage of the intermediate borenium π-complex or the corresponding Wheland intermediate. In the case of 14a, the two cations (π-complex or Wheland) have essentially identical energies and similar geometries. We did not attempt to evaluate transition states for potentially competing intermolecular proton removal at the stage of the Wheland intermediates, partly because external bases were not added in these experiments, and partly because simpler benzenium ions are remarkably stable if the counterion is non-interactive (neither basic nor nucleophilic).^30b In the present study, the tetrakis(pentafluorophenyl)borate (TPFPB) anion is sufficiently nucleophilic to react with sextet boron in the dearylation process shown in Eq. 4,⁴² but there is no evidence to suggest that TPFPB functions as a base.

The detection of hydrogen bridged cationic intermediates related to 15a with representative substrates not containing ortho-silicon indicates that hydrogen bridging is a stabilizing factor that somewhat impedes intramolecular borylation. Although we have not revisited intermolecular or intramolecular electrophylic borylations described in earlier literature reports,^17,43 we note that borenium intermediates would explain the isolated reports of surprisingly facile oxygen-directed aromatic borylations of biaryl phenols and related structures because these reactions are conducted in the presence of the oxophilic Lewis acid AlCl₃.⁴⁴ It may be rewarding to re-examine this chemistry under reaction conditions chosen to promote more specific generation of borenium intermediates.⁴⁵ If borenium electrophiles can be accessed in the absence of potentially bridging (and, therefore, stabilizing) ligands at boron, reactions should be faster and it may be possible to develop new methodology for low temperature borylation and C–H insertion chemistry.

Scheme 2.