Development of α-selective silylation of alkenyl boronates. The reaction conditions for the synthesis of geminal borosilanes was first examined with trans-1-pentenylboronic acid pinacol ester as the model substrate (Supplementary Tables 1–3). The use of 1.2 equivalent of phenylsilane, 0.5 mol% photocatalyst 4-CzIPN (1,2,3,5-tetrakis-(carbazol-yl)-4,6-dicyanobenzene), 5 mol% HAT catalyst ethyl thioglycolate, and 5 mol% base DIPEA (N,N-diisopropylethylamine) in methyl tert-butyl ether (0.1 M) under blue light irradiation was found optimal, affording the corresponding geminal borosilane in 92% yield and 14:1 regioselectivity. The choice of thiols has little effect on the regioselectivity (see Supplementary Table 1). Control experiments showed that the photocatalyst, light, and thiol are all necessary components for an effective transformation (Supplementary Table 3). The reaction yield decreased from 92–71% in the absence of DIPEA.
With the optimal conditions in hand, the substrate scope was evaluated (Table 1). A broad range of β-alkyl-substituted alkenyl boronates were incorporated to give geminal borosilanes 11–27 in 45–96% yields with high regioselectivity. Identical results were obtained regardless of whether trans or cis alkenyl boronates were used (11). Different functional groups including silyl ether (16), alkyl chloride (17), tetrahydropyranyl ether (21), and epoxide (22) were well-tolerated. Heterocycles such as proline (24), indole (25) and isoxazole (26) were accommodated as well. For unsubstituted vinyl boronate, a 1 to 1.7 ratio of α-silylation and β-silylation products (28) was noted, possibly due to the steric effect during the radical addition. The reaction of isopropenylboronic acid pinacol ester delivered no hydrosilylation product (29) (Supplementary Table 5), demonstrating that β-silylation is indeed unfavorable under our reaction conditions. We continue to evaluate different substituents at the β position of alkenyl boronates such as an electron-withdrawing ester group (30), an electron-donating ethoxy group (31) and an electron-neutral boronic acid MIDA (N-methyliminodiacetic acid) ester group (32)39. The corresponding geminal borosilanes were all synthesized in good yields with high α-selectivity. Interestingly, (2-pinacolethenyl)boronate MIDA ester which contains sp2- and sp3-hybridized boron substituents on the olefin underwent selective silylation α to the Bpin moiety (32, 16:1). β-Aryl-substituted alkenyl boronates were then investigated, and the hydrosilylation products were formed with exclusive α-selectivity (33–46, 41–81% yields), regardless of electron-withdrawing or electron-donating groups on the aryl substituent. Some sensitive functionalities in transition-metal catalyzed hydrosilylation or hydroborylation, including aryl chlorides (41) and aryl bromides (44) were well-tolerated with our protocol.
The incorporation of different patterns of silanes such as primary, secondary and tertiary silanes in hydrosilylation could lead to silicon products with distinct reactivities40. However, transition-metal catalyzed methods for borosilane synthesis are generally limited to one type of silanes (e.g. secondary silanes)7–14. It was found that primary, secondary and tertiary silanes were all competent substrates in our transformation, delivering structurally diverse geminal borosilanes effectively (47–52). A siloxane substrate was converted to 52 in 51% yield, indicating the potential utility of this method in silicone polymer chemistry41. Moreover, diverse boryl alkenes reacted smoothly under the optimal reaction conditions, embracing different boronates (53–56), a free boronic acid (57), and a boronic acid MIDA ester (58). The good efficiency, regioselectivity and functional-group tolerance prompted us to evaluate the potential of this method for derivatization of complex drug-like molecules, and the reactions proceeded well with carprofen, ibuprofen, gemfibrozil and indomethacin derivatives (59–62).
Development of γ-selective silylation of allyl boronates. Having established a general method to access geminal borosilanes, we next turned our attention to vicinal borosilanes. The extremely low reactivity of isopropenylboronic acid pinacol ester (29) suggests that a completely different strategy is required to synthesize vicinal borosilanes. We were inspired by the studies on 1,2-boron radical migration42–46 and found that the switch of alkene substrates to α-substituted allyl boronates readily delivered vicinal borosilanes (Table 2). This strategy worked well for α,α-dialkyl-substituted allyl boronates (63–69, 52–84% yields), and the presence of a methyl group at the γ position was tolerated (69). α-Aryl-substituted allyl boronates were also competent substrates (70, 74%). Changing the α-substituent of an allyl boronate to an isopropyl group furnished the migrated product 71 and non-migrated product 71' in 1:5 ratio, while α-methyl allyl boronate only gave non-migrated product 72. The trends are in good accordance with the decreased stability of the radical intermediate derived from 1,2-boron radical migration42–46. Likewise, a wide range of primary, secondary and tertiary silanes participated in the silylation/migration cascade smoothly (73–81, 65–88% yields). The formed vicinal borosilanes 78–81 contain additional handles such as silane, boronate and ester groups for further molecular diversification.
Further synthetic utilizations. The synthetic utility of the divergent photo-HAT hydrosilylation is further illustrated in Fig. 2. The selective synthesis of geminal borosilane 82 was achieved by two metal-free transformations from the herbicide clodinafop-propargyl in one pot (65% yield, Fig. 2a). The highly selective monofunctionalization of trihydrosilanes and dihydrosilanes through photo-HAT catalysis offers opportunities for stepwise decoration of silicon atoms to access structurally diverse multi-borosilanes. By selectively sequencing the α-selective silylation of alkenyl boronate or γ-selective silylation of allyl boronate processes, different types of multi-borosilanes (83–86) could be obtained (Fig. 2b). The boryl or silyl groups in the geminal and vicinal borosilanes are poised for conversion to a range of valuable products47,48. For example, the silyl group in the products could be converted through Si-H bond functionalization to produce silicon-containing compounds such as silanols (87, 88), silylether (89), silylchloride (90), as shown in Fig. 2c. The boryl group represents an extremely versatile synthetic handle in organic synthesis, which are known to undergo amination, oxidation and arylation to assemble C-N, C-O and C-C bonds, respectively. To showcase this flexibility, the boryl group in the products was converted to amines, alcohols, and arenes with the silyl group intact (91–93) (Fig. 2c). By tuning the oxidative conditions, both the silyl and boryl group underwent oxidation, giving rise to 1,2-diol product 94. Furthermore, gram-scale reactions were demonstrated in a batch reactor (47, 78% yield, 1.19 g) as well as an operationally simple continuous-flow reactor (47, 85%, 7.44 g per day production) (Fig. 2d).
Mechanistic considerations. To shed some light on the reaction mechanism and the origin of regioselectivity, a series of control experiments as well as density functional theory (DFT) calculations were conducted. The radical nature of these transformations was confirmed by radical scavenger and radical clock experiments (see Supplementary Discussion). 1H NMR spectroscopy showed the formation of complex C (Fig. 3) upon mixing ethyl thioglycolate (A) with equimolecular DIPEA (B) at room temperature (Supplementary Fig. 7). Stern–Volmer fluorescence quenching and cyclic voltammetry studies indicated that the excited photocatalyst was reductively quenched by the thiol-DIPEA complex C (Supplementary Figs. 6 and 8), giving rise to thiyl radical I. Deuterium-labeling experiments (Supplementary Figs. 9–14) support the HAT pathways in which thiyl radical I abstracts a H atom from silane to produce the silyl radical, and thiol N donates a H atom to the radical adduct to give the products22,23. The conformation of thiyl radical I and thiol N was suggested by DFT calculations (Supplementary Figs. 20 and 21). Both light on/off experiments and the calculated quantum yields (Ф = 0.109) supported photocatalytic cycles instead of radical chain propagation49. Accordingly, a photocatalytic mechanism was proposed (Fig. 3). The excited 4CzIPN (E1/2 (PC*/PC•−) = + 1.35 V vs SCE)50 oxidizes the thiol-DIPEA complex C (E1/2 ox = + 0.92 V vs SCE, Supplementary Fig. 8) to generated the thiyl radical I. The thiyl radical I abstracts a hydrogen atom from silane to afford the silyl radical and thiol N. Subsequently, the silyl radical adds to the α-position of the alkenyl boronate to deliver an alkyl radical intermediate which undergoes polarity-matched HAT process with thiol N to give geminal borosilane. The generated thiyl radical I finally accepts one electron from 4CzIPN•− to close both catalytic cycles. This photocatalytic cycle is thermodynamically feasible according to DFT calculations (Supplementary Fig. 20). Interestingly, computational studies also revealed the important role of thiol-DIPEA complexation, which not only lowers the energy barrier for single electron oxidation by excited photocatalyst, but also stabilizes the formed radical cation intermediate I (Supplementary Fig. 21). Moreover, the energy barrier for hydrogen atom abstraction from phenylsilane by I is much lower due to complexation (Supplementary Fig. 20). The SET event between reduced photocatalyst and thiyl radical I for photocatalyst regeneration also benefits from the complexation. These results indicate that the thiol-DIPEA complexation promotes both SET and HAT processes, which provides rational direction for other thiol-mediated HAT reactions51.
We next sought to elucidate the origin of α-selectivity in the hydrosilylation of alkenyl boronates. It is noted that α-selectivity was also observed in the hydrosilylation of alkenyl boronates using engineered carbon nitrides as heterogeneous photocatalysts20. However, the reaction scope is very limited and the reason for α-selectivity in this heterogeneous process remains unknown. Here, the silyl radical addition and subsequent hydrogen atom transfer with thiols were analyzed by calculations with (E)-1-pentenylboronic acid pinacol ester (R1) and phenylsilane as the model substates (Fig. 4 and Supplementary Fig. 22). The calculated energy diagram illustrates that the addition of silyl radical to alkenyl boronic esters determines the regioselectivity because the transition states (S1 or S2) have the highest energy in the reaction pathways52–54. This also explains why similar regioselectivity was observed with different thiols (Supplementary Table 1). The energy barrier of silyl radical adding to α-position of R1 is 1.64 kcal·mol− 1 lower than that to β-position (S1 vs S2), which means the α-addition rate is approximately 16 times faster than β-addition. This is very close to the observed selectivity in the crude reaction mixture (α/β = 14:1). Despite higher stability of the generated intermediate T2 after β-addition28–34, there are two reasons for the kinetic-controlled α-selectivity. The radical addition processes are nearly irreversible at room temperature, thus the equilibrium between α- and β-addition products cannot be reached. Besides, HAT from thiol N to the radical intermediate T1 is both kinetically and thermodymically favored (ΔG≠ = 11.71 kcal·mol− 1, ΔG = -7.89 kcal·mol− 1) due to polarity-match24,55. The higher HAT rate of T1 compared to T2 further reduces the concentration of the radical T1. Overall, the kinetically favored radical addition and energetically favored back HAT process contribute to the α-selective silylation of alkenyl boronates. Similar elucidation is also found for cis-alkenyl boronates (Supplementary Fig. 22).
As for the allyl boronates, a silyl radical could selectively add to the sterically more accessible γ-position of the allyl boronic ester to give a β-boryl carbon-centered radical intermediate. At this stage, a 1,2-boron migration process influenced by the α-substituents on the allyl boronates took place42–45. The migration was steered by thermodynamic effects to generate a more stable carbon radical. DFT calculations indicate the migration energy barrier for α,α-dimethyl allyl boronate is low (ΔG≠ = 9.23 kcal·mol− 1) and the rearranged radical intermediate T7 is more stable than the non-migrated radical T5 (ΔG = -1.69 kcal·mol− 1) (Supplementary Fig. 23). Moreover, the HAT reaction rate of rearranged radical T7 with thiol is much faster compared to T5, thereby allowing selective synthesis of vicinal borosilanes. The plausible mechanism for the synthesis of vicinal borosilanes was then proposed in light of all the experimental data (Supplementary Fig. 19).