Metal‐Free C−H Borylation of N‐Heteroarenes by Boron Trifluoride

Abstract

Organoboron compounds are essential reagents in modern C−C coupling reactions. Their synthesis via catalytic C−H borylation by main group elements is emerging as a powerful tool alternative to transition metal based catalysis. Herein, a straightforward metal‐free synthesis of aryldifluoroboranes from BF₃ and heteroarenes is reported. The reaction is assisted by sterically hindered amines and catalytic amounts of thioureas. According to computational studies the reaction proceeds via frustrated Lewis pair (FLP) mechanism. The obtained aryldifluoroboranes are further stabilized against destructive protodeborylation by converting them to the corresponding air stable tetramethylammonium organotrifluoroborates.

C−H activation: Indoles, pyrroles and indolenines undergo C−H borylation by boron trifluoride with assistance of base and thiourea additives.

Introduction

Efficient C−C and C−X couplings in contemporary organic synthesis including Suzuki–Miyaura coupling, Petasis reaction, Chan–Lam coupling etc., are performed by utilizing various organoboron compounds. ^[1] Consequently, a ready access to initial organoboron reagents is prerequisite in such cross‐coupling strategies. Among various organoboron compounds, organotrifluoroborates are widely used due to their increased stability under air and, especially, against protodeboronation. Generally, they are synthesized from boronic acids or their esters via fluorination with KHF₂. ^[2] In turn, synthesis of organoboronic acids and organoboronates is widely studied including both transition metal ^[3] and main group catalysis. ^[4] However, direct synthesis of organotrifluoroborates by C−H borylation is still underdeveloped.

Borylation of C(sp²)−H bonds was traditionally performed by transition metal catalysts ^[5] until it was recently shown that geometrically arranged ansa‐aminoboranes, ^[6] usually used for hydrogen activation and reductions, are also capable of activating certain electron‐rich heteroarenes. ^[7] After that, substrate scope was notably expanded ^[6c] and several reports were published concerning the structure–activity correlation of such ansa‐systems. ^[8] Beside ansa‐aminoboranes, it is also possible to activate C(sp²)−H bonds by strong Lewis acids such as B(C₆F₅)₃ ^[9] or BX₃ (X=Cl, Br).[ ^4c , ¹⁰ ] Depending on the conditions, reactions proceed either by direct electrophilic attack of the boron halide ^[10e] or by in situ generated borenium cations derived from BCl₃,[ ^4c , ^10b , ¹¹ ] BBr₃,[ ^10a , ^10d , ^10e , ¹¹ ] or catecholborane. ^[12] Regardless of the mechanisms, such borylation reactions with BF₃, the weakest among boron halides Lewis acid, has not been reported to the best of our knowledge.

Recently, we reported that combination of BF₃⋅SMe₂ with 1,2,2,6,6‐pentamethylpiperidine (PMP, 4 a) borylates terminal alkynes forming tri‐ and tetraalkynyl borates. ^[13] Such approach avoids usage of precious metals and utilizes one of inexpensive boron reagents along with the recoverable base. Also, we found that BF₃⋅SMe₂ adduct is very labile and can serve as the convenient alternative to reactive gaseous BF₃. In continuation of our efforts, we focused on extending developed methodology to the borylation of more challenging C(sp²)−H bonds.

Herein, we present the direct C−H borylation of electron rich N‐heterocycles such as indoles, pyrroles, and indolenines by BF₃⋅SMe₂. These reactions are assisted by sterically hindered amines and promoted by catalytic amounts of various thioureas (Scheme 1). Borylation products, aryldifluoroboranes, are converted to organotrifluoroborates via complexation with pyridine and further fluorination by tetramethylammonium fluoride. Such procedure overcomes the problem of unwanted protodeboronation of organofluoroboranes and makes them practical starting materials for the C−C coupling reactions.

Scheme 1.

General pattern of N‐heteroarenes borylation by boron trifluoride.

Results and Discussion

In our previous studies we showed that BF₃ together with sterically hindered base, 4 a, can borylate phenylacetylene quantitatively within 15 minutes. ^[13] The BF₃⋅SMe₂/4 a pair was also proven efficient, yet full conversions were reached after 2 hours. When these Lewis pairs were applied to our model substrate, N‐methylindole 1 a, we achieved 62–65 % in situ yields of aryldifluoroborane 2 a and minor amounts of bis(indolyl) product 3 a (Scheme 2). Because of possibility of indole borylation to both 2nd and 3rd positions, regioselectivity of borylation was confirmed by X‐ray diffraction and 2D NMR spectra of the corresponding trifluoroborate (see Supporting Information for details).

Scheme 2.

N‐Methylindole 1 a borylation with BF₃⋅L and sterically hindered amines 4 a–d gives 3‐difluoroborane 2 a and minor amounts of 3 a.

A mixture of BF₃⋅OEt₂ and PMP was shown to borylate terminal acetylenes to a remarkable degree, ^[13] whereas only traces of 2 a were observed with this reagent combination (entry 3). Addition of SMe₂ to the reaction mixture didn't enhance the reactivity of the BF₃⋅OEt₂/PMP system (entry 4). When the reaction temperature was raised to 120 °C, the in situ yield of 2 a raised to 30 % (entry 5). According to the previous studies, a higher energetic penalty required for dissociation of BF₃⋅OEt₂ results in the poor reactivity of this adduct that sharply contrasts with labile and reactive BF₃⋅SMe₂ and BF₃⋅PMP (Table 1, column 4). ^[13]

Table 1.

In situ yields of the borylation product 2 a produced in the reactions of 1 a, BF₃⋅L, and sterically hindered amines 4 a–d ^[a]

Entry	Base	L	Strength of amine‐BF3 adduct, kcal mol−1[b]	1 a, %	2 a, %[c,d]
1	4 a	SMe2	−1.9	28	62
2	4 a	4 a	−1.9	20	65
3	4 a	OEt2	−1.9	94	5
4[e]	4 a	OEt2	−1.9	83	3
5[f]	4 a	OEt2	−1.9	40	30
6	4 b	SMe2	−7.8	55	22
7	4 b	4 b	−7.8	100	0
8	4 c	SMe2	–[g]	55	35
9	4 d	SMe2	–[g]	43	47

[a] Reactions were performed in the gas‐tight NMR tubes for 24 h at 60 °C in the argon atmosphere; Yields were determined by the ¹H NMR spectroscopy against 1‐bromo‐3,5‐difluorobenzene as the internal standard. [b] Defined as ΔG of amine+BF₃→amine‐BF₃ as obtained from ref. [13]. [c] Characteristic signal of 2 a in ¹¹B and ¹⁹F NMR spectra was observed at +24 ppm and −94 ppm, respectively. [d] In addition to 1 a and 2 a, some amount of unidentified product with indole pattern was observed in ¹H NMR spectra (see Supporting Information for details). [e] 1 equiv. of SMe₂ was added to the reaction. [f] Reaction temperature 120 °C. [g] No adduct with gaseous BF₃ was detected.

We also explored the borylation of 1 a using other sterically hindered amines (Scheme 2, Table 1). Although 4 a is a sterically very demanding amine, it is capable of forming a weak Lewis adduct with BF₃ (Entry 1). When sterically less hindered and hence more strongly interacting with BF₃ N,N‐diisopropylethylamine (DIPEA, 4 b) was applied in combination with BF₃⋅SMe₂, the yield of 2 a reached only 22 % (entry 6). Moreover, when all the BF₃ in the solution was in the form of adduct with 4 b, not even traces of borylation products were found (entry 7). Sterically more hindered than 4 a bases 4 c and 4 d did not form stable adducts even with gaseous BF₃ (entries 8–9). However, borylation with 4 c–d was less effective as compared to 4 a, yet all the bases 4 a–d have similar Brønsted basicities. We hypothesized that 4 c and 4 d had stronger steric repulsive component in the borylation transitions states (vide infra) that conveyed into higher kinetic barriers in comparison to 4 a.

To gain mechanistic insight, computational studies of the borylation reaction of 1 a with BF₃⋅SMe₂/PMP pair were carried out. The mechanistic scenarios depicted in Scheme 3 were investigated via Density Functional Theory (DFT) calculations. ^[14] Despite the high nucleophilicity of indole, the 1 a‐BF₃ adduct could not be identified computationally. ^[15] Therefore, the electrophilic substitution (S_EAr) pathway can be readily excluded as a viable mechanism. For the borenium mediated pathway, the formation of the borenium/BF₄ ⁻ species (4 a‐BF₂ ⁺/BF₄ ⁻ ion pair) is thermodynamically uphill, but the relatively high barrier (34.7 kcal mol⁻¹) represented by the subsequent BF₂ ⁺ transfer transition state (TS_bor) renders this pathway unlikely as well.

Scheme 3.

Computationally examined mechanistic pathways of 1 a borylation by a PMP/BF₃ Lewis pair. Free energy data (in kcal mol⁻¹) given in parentheses are with respect to the 1 a+4 a+BF₃ reactant state.

Hence our attention turned towards the FLP‐type C−H activation mechanism, which has been described previously for ansa‐aminoboranes.[ ⁷ , ⁸ , ⁹ ] This mechanism assumes the cooperative action of Lewis acid/base centers, which was confirmed computationally (see Figures 1 and 2). ^[16] The reactants (1 a, 4 a and BF₃) form a transient weakly bound complex (4 a⋅1 a⋅BF₃), from which the C−H activation takes place concertedly via TS_FLP. The obtained free energy barrier (29.6 kcal mol⁻¹) is consistent with the reaction rate measured at elevated temperature. The concerted C−H activation leads to intermediate int, which is stabilized by N−H⋅⋅⋅F hydrogen bonding interactions. This species can favorably interact with an additional BF₃ molecule, and the (int‐BF₃) adduct can easily furnish the borylated product 2 a via a fluoride transfer (TS_Ftrans). The dissociation of the weakly bound 2 a⋅4 aH⁺/BF₄ ⁻ species yield a more stable product state (2 a+4 aH⁺/BF₄ ⁻).

Figure 1.

Free energy profile computed for the FLP‐type C−H activation pathway for 1 a borylation by a PMP/BF₃ Lewis pair. Free energy data are given relative to the 1 a+4 a+BF₃ reactant state.

Figure 2.

Structures of reaction intermediates and transition states identified computationally for the FLP‐type C−H activation pathway for 1 a borylation by PMP⋅BF₃. H atoms (except that involved in the reaction) and the four methyl groups of PMP are omitted for clarity. Forming and breaking bonds are shown as black dotted lines in TS structures; H‐bonds are highlighted by blue dotted lines. Free energy data are given relative to the 1 a+4 a+BF₃ reactant state (in kcal mol⁻¹).

We carried out kinetic studies of the 1 a borylation in the presence of varying amounts of BF₃⋅SMe₂ and 4 a. Analysis of initial rates revealed that the reaction is first order in 1 a and a positive order in BF₃⋅SMe₂. At the same time, addition of overstoichiometric 0.5 equivalents of 4 a resulted in a pronounced inhibiting effect (Figure 3). This effect was attributed to the zero kinetic order of the reaction in 4 a along with the positive order in free BF₃. Additional 4 a decreased the concentration of BF₃ and hence the reaction rate. Notwithstanding this explanation contrasted with the computationally found endergonicity of 4 a‐BF₃ dissociation (Figure 1) it was known from our previous work ^[13] that this Lewis adduct was better stabilized in a more polar solvent, DCM. In the present work, the free energies were computed for benzene that was used as a solvent for borylation reaction. However, this might not be a completely adequate description of the real reaction mixture provided the presence of polar reactants (Me₂S‐BF₃, 4 a‐BF₃) and products ([4 a‐H⁺][BF₄ ^‐]) in high concentrations.

Figure 3.

Kinetic profiles of 1 a borylation with BF₃⋅SMe₂, 4 a/4 d and various amounts of 5 a.

The ultimate picture of the kinetic processes seemed rather complex: apparently 4 a‐BF₃ is generated in situ upon mixing of the reagents that discharged some amount of free Me₂S. In addition to that, more Me₂S was generated during the borylation, suppressing dissociation of the starting BF₃ adducts. In order to derive the quantitative model of the observed kinetics a numerical simulation was carried out. Two most likely models were compared, one that followed the k[1 a][BF₃] rate law (model A) and the model B with the k[1 a][BF₃]² law. Both of them described high percentage of the total observed variability but statistically were quite comparable (see Supporting Information for details). We concluded that there is more than one mechanism in action including also the FLP mechanism that is expected to follow the k[1 a][BF₃][PMP] kinetic law. Yet such a kinetic model alone would poorly agree with the collected kinetic data, it might in combination with the models A or B.

Although 4 a together with BF₃ gives the highest in situ yield in the borylation of 1 a among the bases 4 a–d, parallel decomposition of amine by Lewis acidic BF₃ is an unwanted side reaction. ^[13] This drawback can be overcome by using the most sterically hindered base 4 d, where formation of its BF₃ adduct is sterically unfavorable. However, borylation with 4 d proceeds much slower (Figure 3) and gives the lower yield of 2 a (Table 1). This inspired us to look for additives, which could promote the borylation reaction further. Considering that formation of borocations in benzene cannot be completely excluded (Scheme 3), we turned out attention to tetramethylthiourea (TMTU) that proved assisting the autoionization of BF₃. ^[17] Accordingly, we found that the addition of small (5 mol %) amounts TMTU (5 a) to both BF₃⋅SMe₂/4 a and BF₃⋅SMe₂/4 d pairs increases the rate of 2 a formation (Figure 3). When stoichiometric amounts of 5 a were used in the absence of 4 a or 4 d, only 1 a, BF₃⋅5 a adduct, some unidentified products but no traces of 2 a were detected by ¹H or ¹¹B NMR analysis.

Our computational analysis indicates that the formation of the borenium/BF₄ ⁻ species with TMTU (5 a‐BF₂ ⁺/BF₄ ⁻ ion pair) as well the subsequent BF₂ ⁺ transfer process is notably more favored as compared to the analogous borenium formation process with PMP (see Scheme 4). ^[18] The barrier computed for the formation of borenium 1 a‐BF₂ ⁺ is 31.0 kcal mol⁻¹, which is comparable to that of the FLP pathway with PMP (29.6 kcal mol⁻¹). Furthermore, the deprotonation of 1 a‐BF₂ ⁺ by TMTU takes place via a similar barrier (29.9 kcal mol⁻¹) and the proton is then delivered to the more basic PMP. Based on these results and considering the uncertainty in the computed barriers (a few kcal mol⁻¹ at least), one possible explanation for the beneficial effect of TMTU is that it opens an alternative borylation pathway to the PMP assisted FLP mechanism. In this latter mechanism, TMTU acts as catalyst facilitating the formation and the deprotonation of the 1 a‐BF₂ ⁺ borenium intermediate. We have also explored the FLP pathway with TMTU acting as a base and we obtained a barrier of 31.7 kcal mol⁻¹, which is slightly higher than that of the TMTU promoted borenium pathway (31.0 kcal mol⁻¹), but still comparable to the most favored PMP assisted FLP pathway (29.6 kcal mol⁻¹). So an alternative explanation for the improved yields with TMTU is that it acts as a base in the FLP mechanism, which becomes important and beneficial at higher conversion rates, when most of the PMP is consumed.

Scheme 4.

TMTU assisted borenium pathway examined computationally. Free energy data (in kcal mol⁻¹) given in parentheses are with respect to the 1 a+5 a+BF₃ reactant state.

To study the role of TMTU in more details, we introduced borylation of 1 a with other thioureas 5 b–e (Scheme 5). Among the series, cyclic thioureas 5 b and 5 d give the highest in situ yield of 2 a. Interestingly, in case of 5 d minor amounts of diarylfluoroborane 3 a were also found (Table 2).

Scheme 5.

Screening of various thioureas as catalysts for borylation of 1 a.

Table 2.

In situ yields of products from the borylation of 1 a in the presence of thioureas 5 a–5 e.^[a]

Thiourea	1 a	2 a	3 a
–	43	47	–
5 a [b]	30	69	–
5 b [b]	22	77	–
5 c [b]	34	65	–
5 d [b]	19	79	2
5 e [b]	32	68	–

[a] Reactions were performed in the gas‐tight NMR tubes for 24 h at 60 °C in the argon atmosphere; Yields were determined by the ¹H NMR spectroscopy against 1‐bromo‐3,5‐difluorobenzene as the internal standard; see the Supporting Information for details; [b] 5 mol % of thiourea 5 a–e was used.

We investigated the substrate scope of borylation with 5 d. In general, electron rich N‐alkylindoles and ‐pyrroles were borylated with high to moderate yields (Scheme 6, examples 2 a–d). Moreover, N‐allyl group remained untouched in the reaction conditions (entry 2 e). Incorporation of halogens atoms as electron withdrawing groups into the indole ring decreases the conversion (examples 2 f–h). Interestingly, terminal double bond in indolenine ring (entry 2 i,j) can also be borylated with a high yield. In general, electron‐rich alkyl‐ and halosubstituted N‐heteroarenes we found to be the best substrates in the developed reaction conditions. Attempted borylation of other substrates was less successful: 5‐Methoxy substituted indole 1 k produced complex mixture with no trace of borylation product. Set of signals around +15 ppm in the ¹¹B NMR spectrum indicated possible ether cleavage. Attempt to use conventional enamines 1 l–m led only to their BF₃ adducts whereas 2‐methylbenzothiophene (1 n) remained completely intact. Interestingly, 3‐ethynylthiophene 1 o gave a complex product mixture, however, characteristic for hetarene‐BF₂ peaks at 22.95 ppm (¹¹B NMR) and −84.96 (¹⁹F NMR) were detected. Importantly, no evidence of terminal alkyne CH bond activation was found even though this substrate is known to form trialkynylborane SMe₂ adduct upon treatment with BF₃⋅SMe₂/4 a system. ^[19] Allyltrimethylsilane 1 p gave traces of borylation products according to ¹¹B and ¹⁹F spectra. At the same time, the characteristic peak for trimethylsilylfluoride multiplet found in the ¹⁹F NMR spectrum at −157 ppm indicated TMS group cleavage.

Scheme 6.

Substrate scope of the reaction and NMR yields of the N‐heteroarenes borylation products.

Aryldifluoroboranes 2 a–j were generated and partially characterized by the NMR spectroscopy in solution, however, attempted isolation of these compounds appeared to be unsuccessful due to high reactivity and reversibility of the borylation reaction (Figure 1). In line with the recent reports ^[15] we found protodeboronation to be the major decomposition pathway. Although only few 3‐BF₃‐substituted indoles, all stabilized by the strongly electron withdrawing groups on nitrogen, were known in the literature, ^[1g] we attempted conversion of in situ produced 2 a to the corresponding indolyltrifluoroborate (6 a) (Scheme 7).

Scheme 7.

Aryldifluoroboranes are accessible with high reactivity and protodeboronation of 3‐subsituted indole 2 a takes place in the presence of TMAF. However, 2 a forms pyridine (7) and N,N‐dimethylaniline (8) adducts. The former can be further converted to trifluoroborate 11 a with TMAF.

We reported previously that trialkynylboranes are converted to trialkynylfluoroborates by the treatment with tetramethylammonium fluoride (TMAF). ^[13] However, all attempts to perform direct fluorination of 2 a to 6 a with TMAF led to the complete protodeborylation, and starting material 1 a was observed instead. The reason for instability of 2 a in the presence of TMAF is not clear, but to overcome this issue, we stabilized 2 a in the form of its pyridine or N,N‐dimethylaniline adducts. Addition of any of these amines to the 2 a containing reaction mixture resulted in instantaneous formation of the corresponding adducts 7 and 8. All further attempts to isolate them led to the decomposition to 1 a, but, to our delight, 7 produced trifluoroborate 6 a without noticeable protodeboronation upon treatment with TMAF. In contrast to 7, reaction of adduct 8 with TMAF led to a complex mixture of unidentified products. To demonstrate generality of the developed two‐step method, we applied it to other substrates and isolated the corresponding trifluoroborates with the overall yields equal or slightly lower than those measured in situ during the previous borylation experiments (Scheme 8). Although 5 d was found to be the most active additive for borylation in general, we found that commercially available 5 a gives similar yields with electron rich arenes (Scheme 8, substrates 6 a–c). Borylation of less active 1‐methyl‐2‐phenylindole was successfully performed with bulky thiourea 5d, giving corresponding trifluoroborate 6 d in good yield.

Scheme 8.

A one‐pot thiourea‐assisted synthesis of aryltrifluoroborates from heteroarenes. [a] compound 11 a was isolated along with 15 % of tetramethyl‐ammonium diindolyldifluoroborate (see Supporting Information for details).

Conclusions

In conclusion, we developed a straightforward method of N‐heteroarenes C−H borylation by boron trifluoride and sterically hindered amines. Addition of thioureas as additives had a beneficial effect on the yield of borylation products. Indoles were selectively borylated into the 3‐position and indolenines to the terminal position of the double bond. Borylation intermediates, aryldifluoroboranes, are very reactive and prone to protodeboronation in presence of Me₄NF. This problem was overcame by intermediate formation of their pyridine adducts prior to treatment with Me₄NF that furnished 3‐substituted indolyl‐ and pyrrolyltrifluoroborates with high overall yields. Expansion of the reactivity of boron trifluoride towards other C−H bonds as well as studies towards avoiding protodeboronation in other organoboron compounds are subjects of ongoing studies in our groups.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

V. Iashin, D. Berta, K. Chernichenko, M. Nieger, K. Moslova, I. Pápai, T. Repo, Chem. Eur. J. 2020, 26, 13873.