A Boronium Ion with Exceptional Electrophilicity

During our studies on aromatic borylation,^[1] we considered the combination of a highly electrophilic R₂BNTf₂ reagent with a base that would neutralize the HNTf₂ byproduct of borylation without deactivating the electrophile. In principle, these requirements might be satisfied by 1,8-bis(dimethylamino)naphthalene (1), a hindered and exceptionally basic aniline that finds numerous applications as a basic catalyst or reagent due to its legendary lack of nucleophilicity.^{[2, 3]} Strong electrophiles interact weakly, if at all, with the amine nitrogens, and very few examples are known where stable bonds to nitrogen can be formed between 1 and electrophilic groups larger than hydrogen.^{[2, 4-7]} Among these exceptional cases, cyclic boronium structures 2 and 3 are relatively stable because the subunits BH₂ and BF₂ have minimal steric requirements.^[5] However, the more hindered BMe₂ derivative 4 has not been detected and no analogous BR₂ structures are known.^{[5a, 8]} In view of this long history, we were somewhat surprised to find that an adduct is readily formed simply upon mixing 1 with the 9-BBN bistriflimide reagent 5a despite the transannular steric demands of the 9-BBN core and the need to form adjacent quaternary bonds to boron as well as nitrogen.^{[9, 10]} The remarkable structural features and unusual reactivity of this adduct are the subject of the following communication.

The previously unreported 5a was easily prepared from the commercially available 9-BBN dimer and bis(trifluoromethanesulfonyl)imide upon heating in toluene. Combination of the bulky boron reagent 5a with a stoichiometric amount of 1 in CD₂Cl₂ at room temperature formed a deep red solution that turned colorless within seconds of mixing the reagents. Analysis of the resulting solution by ¹¹B NMR spectroscopy revealed a signal at δ 16.2 ppm, suggesting that a single tetracoordinate^[11] boron atom is present in the product. The ¹⁹F NMR spectrum showed a single peak at δ -79.4 ppm, which is characteristic of bistriflimide anion,^[12] so the boron-containing fragment was thus identified to be a cation. The ¹H NMR spectrum suggested that the solution structure of the cation is highly symmetrical on the NMR timescale at room temperature. Only four groups of protons corresponding to the diamine subunit 1 were observed, including one sharp singlet for all four methyl groups, and a well-resolved (at 500 MHz) AMX system for the aromatic protons. Other peaks in the ¹H and ¹³C NMR spectra were also consistent with a symmetrical time-averaged structure for the cation (for example, a single ¹³C methyl peak at δ 57.1 ppm, and only 3 peaks for the 9-BBN cage carbons). Since the covalent adduct 6 (X = NTf₂) is ruled by observation of the bistriflimide anion,^[12b] and the tricoordinate cationic borenium structure 7 (X = NTf₂) is not consistent with the observed ¹¹B NMR chemical shift, the most plausible structure for the species formed from 1 and 5a is the exceptionally hindered boronium salt 8a. Formation of 8a may follow the logical sequence in Scheme 1, but alternative mechanisms have not been excluded.^[13] Formation of >95% 8a depends on the low nucleophilicity of the counterion, and only partial conversion to the boronium ion was observed when 5b was used instead of 5a. Thus, when equimolar 1 and the triflate reagent 5b were mixed in CD₂Cl₂, the resulting solution showed both the starting 1 and the product 8b (ca. 1.2:1 ratio by ¹H NMR assay).

Scheme 1.

The low temperature ¹H NMR behavior of 8a in CD₂Cl₂ is complex and indicates the presence of unsymmetrical species. Decreasing the temperature broadens the ¹H NMR singlet corresponding to the N-methyl groups, until it turns into a broad set of at least three maxima that are not fully resolved even at -80°C. Additional information about the solution structure is provided by the very different shielding of the two bridgehead hydrogen atoms of the 9-BBN cage observed in the low temperature ¹H NMR spectra. While the spectrum taken at room temperature shows only one peak for both bridgehead protons, two distinct resonances are observed at -80 °C. One of the bridgehead hydrogens gives rise to a peak at an unremarkable δ 1.57 ppm, while the other hydrogen appears at δ 0.33 ppm. Such prominent shielding by the aromatic ring is consistent with the 9-BBN cage being tilted toward one side of the naphthalene plane to place the shielded proton above the aromatic π-system.

Slow cooling of the solution of 8a in a mixture of CH₂Cl₂ and hexanes produced large platelike crystals, suitable for X-ray diffraction studies.^[14] Due to strain imposed by the hindered environment, the structure of 8a is non-symmetrical, the B-N bonds are very long, and both the 9-BBN cage and the bis(dimethylamino)naphthalene unit are severely twisted. While the C15-B1-C19 angle (106°) and C-B bond lengths (1.63 Å, B1-C15; 1.61 Å, B1-C19) are within the expected range for 9-BBN derivatives,^{[9a, 15]} the C20-C19-C15-C22 and C18-C19-C15-C16 dihedral angles are in excess of 10°. Furthermore, strong distortion of the diamine base is evidenced by the N2-C9-C1-N1 dihedral angle of 28.9°, compared to 20.3° in 1.^[16] On the other hand, the aniline N⋯N distance in 8a (2.65 Å) is substantially shorter than that reported for 1 (2.79 Å), and only slightly exceeds that in salts of protonated 1, such as the sulfonimide salt 1·HNMs₂ (2.60 Å).^[17]

The arrangement of the 9-BBN cage in crystals of 8a is also noteworthy. The bridgehead carbons C15 and C19 are quite distinct, and C15 is pseudo-axial with respect to the distorted half-chair boron heterocycle. This places C15-H above the aromatic π-system, consistent with the low temperature ¹H NMR result indicating substantial shielding of one of the bridgehead protons. Another prominent structural detail is the length of the B-N bonds (B1-N1 1.72 Å; B1-N2 1.73 Å), compared to values of 1.58-1.60 Å in simpler boronium cations such as [H₂B(NMe₃)(MeIm)]⁺ or [H₂B(NH₂Me)(MeIm)]⁺ (MeIm = 1-methylimidazole).^[18,19] Since the 1.72-1.73 Å distance greatly exceeds the sum of covalent radii for B and N atoms (1.55 Å),^[20] the calculated Pauling bond order for both B-N bonds in 8a is only ca. 0.55.^[21]

The unusual structural features prompted computational modeling of the cation 8a.^[22a] Gas phase geometry optimization at the M06-2X/6-31G(d,p) level produced a structure that is in close agreement with the X-ray data (for example, B-N bond lengths are within 0.01 Å of the experimental values).^[22b] NBO analysis^[22c] performed on the optimized structure indicates that the boron atom carries the bulk of the positive charge (NBO charge 0.97), and Wiberg bond orders of the two B-N bonds are 0.52 and 0.53.

It was also of interest to compare the interactions of other amines with the potent Lewis acid 5a in solution. When triethylamine was combined with 5a in CD₂Cl₂, clean formation of the borenium ion 10 was observed (Scheme 2). No evidence for a boronium structure 11 was detected, even when excess triethylamine was used. The borenium character of 10 is substantiated by a strongly deshielded ¹¹B NMR peak at δ 85.1 ppm, as well as the bistriflimide anion peak at δ -79.5 ppm in the ¹⁹F NMR spectrum,^[12a] evidence that rules out the alternative structure 9. For simplicity, 9 is tentatively shown as a precursor of 10, although direct conversion from 5a is not ruled out.

Scheme 2.

In contrast to triethylamine, 4-(dimethylamino)pyridine (DMAP) reacted with a stoichiometric amount of 5a to afford mostly the isolable boronium cation 14 according to the ¹¹B NMR shift of δ 3.0 ppm (CD₂Cl₂, rt) along with traces of the borenium cation 13 (δ 66.5 ppm).^[23] A much better way to generate 13 in situ was to protonate the amine borane complex 12 with Tf₂NH. This method confirmed the chemical shift of 13 and afforded solutions also containing relatively minor amounts of the boronium salt 14 (ca. 7-11:1 13:14). However, the more hindered 2,6-di-tert-butyl-4-methylpyridine did not interact with 5a at room temperature according to ¹H and ¹¹B NMR assay.

The most remarkable feature of the borenium salt 10 is the absence of any stabilizing π-donor or n-donor substituents at boron, in contrast to 13 and to all previously reported persistent borenium ions generated in the condensed phase.^{[24, 25]} A comparison of ¹¹B NMR shifts for 10 (δ 85.1 ppm) and 13 (δ 66.5 ppm) indicates extensive cation stabilization by π delocalization between DMAP and the boron atom. Other π-stabilized borenium cations have been observed in the ¹¹B chemical shift range of δ 58.2 to 66 ppm,^{[1, 24, 25]} suggesting that 10 may be an exceptionally electrophilic member of the borenium family of structures.

High electrophilicity of boron cations is crucial for potential applications in electrophilic aromatic borylation.^{[1, 26]} Thus, different combinations of the bistriflimide 5a with basic amines generated reagents that react with electron-rich heterocycles to provide B-heteroaryl 9-BBN derivatives along with the protonated amines. The reagent consisting of 5a and the non-complexing 2,6-di-tert-butyl-4-methylpyridine was the most reactive, and borylated N-methylindole in seconds at room temperature to afford 16a (>95% conversion by NMR spectroscopy), while the cationic reagents 8a and 10 required several hours at 50 °C for similar conversion. No added base was needed with 8a or 10 because both reagents already contain a “built-in” base (proton sponge 1 and triethylamine, respectively) to neutralize the HNTf₂ that forms during borylation. On the other hand, neither 13 nor 14 reacted with N-methylindole under these conditions. ^[27]

While the boronium salt 8a is less potent than the reagent from 5a and 2,6-di-tert-butyl-4-methylpyridine, 8a is a far more convenient borylating agent. Practical access to 8a on gram scale is possible using a one-pot procedure from 9-BBN, 1, and HNTf₂ without having to isolate the highly sensitive 5a (see Supporting Information). Crystallized 8a is much easier to handle compared to 5a, and even survives up to a month of exposure to dry air (dessicator over Drierite), in contrast to 5a or 10. Furthermore, the aromatic borylation products obtained using 8a are easy to isolate (Table 1). Crystalline products 16a-d were obtained in high purity simply by extracting the reaction mixtures with hexanes, where neither the unreacted 8a nor the byproduct 1·HNTf₂ is soluble, followed by solvent evaporation. This procedure minimizes the risk of competing protodeboronation using 8a, but it is not feasible with the reagent from 5a and 2,6-di-tert-butyl-4-methylpyridine due to differences in reagent solubility.

Table 1.

Borylation of nitrogen heterocycles using 8a.^[a]

Het-H	Product, Het-BBN	Time	Yield, %
		1.5h	96
		1.5h	98
		3.5h	97[b]
		5.5d	97[b]

^[a]1.05 equiv of 8a; CH₂Cl₂; 50 °C.

^[b]2.10 equiv of 8a

The structures of boranes 16a-d were established by multinuclear NMR spectroscopy, as well as X-ray crystallography in the case of 16a.^[28] Borylation of N-methylindole afforded exclusively the 3-substituted regioisomer, in sharp contrast to the previously reported reaction with B(C₆F₅)₃, which produces the 2-borylated N-methylindole.^[29] Pyrrole 15c gave a mixture of mono-borylated regioisomers along with some of the diborylated 16c using one equivalent of 8a (ca. 80% conversion of 15c), but two equivalents of the borylating agent cleanly produced the diborylated pyrrole 16c. In the reaction with unsubstituted indole, the known N-borylation product was produced first,^[30] followed by much slower C3-borylation to afford 16d.

Several reactions of 8a suggest that it is in equilibrium with the starting 1 and 5a. Thus, equimolar 8a and Tf₂NH produced the protonated diamine (1·HNTf₂) and released 5a (NMR assay). Furthermore, reaction of 8a with triethylammonium bistriflimide (Et₃NH⁺ Tf₂N⁻) yielded 1·HNTf₂ and the tricoordinate cation 10, representing an unusual route from boronium to borenium ions involving the formal migration of the 9-BBN fragment to a different amine. These events can be understood if dissociation of 8a to 1 + 5a is the first step. The same dissociative mechanism may also help explain the borylations of Table 1, although the identity of the key boron electrophile is not clear. Depending on the timing of bond dissociation and borylation events, a role for the tricoordinate borenium ion 7^[1] or even a dicoordinate borinium ion^[26a] cannot be ruled out at this point. The equilibrium between 8a and 1 + 5a is not directly observable by ¹H NMR spectroscopy, but the analogous process does occur in the related system 8b and 1 + 5b, containing the more nucleophilic triflate anion (vide supra).

To summarize, the covalent boron bistriflimide 5a was used to access the unusual boron salts 8a and 10 by exploiting the excellent leaving group ability of bistriflimide anion. The triethylamine-derived 10 expands the range of borenium salts observed in the condensed phase, proving that resonance delocalization of the positive charge is not required for a persistent borenium cation. The hindered boronium salt 8a cautions against interpreting the name “proton sponge” too literally: in fact, this study proves that 1 can act as a chelating ligand for species much larger than a proton. Also, the unusual structure of 8a raises a rhetorical question: is there a distinct boundary between the cationic species called “boronium” (tetra-substituted B), “borenium” (tri-substituted B), and “borinium” (di-substituted B), according to Nöth’s terminology?^[24] While 8a should be most appropriately called a boronium salt, the long B-N distances increase the “borinium-like” character, and the unusual reactivity adds a small hint of a borenium ion (7). Aside from the structural features of 8a, the chemoselectivity of its formation also deserves attention. In view of earlier reports that strong electrophiles attack the aromatic system of 1^[3] or abstract hydride from one of the N-methyl groups,^{[3a, 31]} it is quite intriguing that the reaction between 1 and 5a proceeds to form 8a, the most hindered of all plausible products. As evidenced by the electrophilic borylations (Table 1), the steric hindrance of 8a is responsible for extraordinary reactivity compared to less hindered boronium salts such as 14.^[32]

Figure 1.

ORTEP drawing of 8a (50% probability thermal ellipsoids). The counterion and hydrogen atoms are omitted for clarity.