Introduction

Silicon lies vertically below carbon in Group 14. Like carbon, silicon can also bond with other groups tetravalently, and it exhibits chirality when the four substituents are different. Organosilanes with an enantioenriched stereogenic silicon center1,2,3,4,5,6,7,8,9 are attracting growing interests not only for understanding chirality and chemistry beyond carbon, but also for preparing advanced materials (I10 and II11), bioactive molecules (III12), probes in mechanistic studies (IV13) as well as chiral auxiliaries (V14), ligands (VI15,16,17) and others types of chiral reagents (VII18,19) in asymmetric transformations (Fig. 1a). While stereogenic carbon centers are ubiquitous in nature and can be accessed by an abundance of synthetic methods, the stereogenic silicon center is unnatural and much more synthetically challenging.

Fig. 1: Importance and synthesis of organosilanes bearing an enantioenriched stereogenic silicon center.
figure 1

a Function organosilanes bearing enantioenriched stereogenic silicon centers. b Typical strategies for asymmetric construction of stereogenic silicon via resolution or desymmetrization. c Proposed mechanistic profile of enantioconvergent conversion of racemic organosilanes. d This work: Lewis base (3e)-catalyzed dynamic kinetic silyletherification of racemic chlorosilanes 1 with (S)-lactates 2 to give silicon-stereogenic silylethers 4. dr: diastereomeric ratio.

Full size image

In fact, only two strategies have so far been widely used to asymmetrically construct enantioenriched stereogenic silicon centers (Fig. 1b). The first strategy relies on the resolution of racemic organosilanes either through diastereomer formation and separation20,21, or occasionally through kinetic resolution22. This strategy suffers from a major practical limitation that the enantioenriched organosilanes cannot be obtained in more than 50% yield. A general protocol is the use of (-)-menthol to resolve racemic chlorosilanes21. In this approach, two diastereomeric silyl ethers are formed in an essentially equimolar ratio, and must be separated by fractional crystallization or repeated flash chromatography. The second strategy is the desymmetrization of prochiral organosilanes bearing two enantiotropic Si–H23,24,25,26,27,28,29,30, Si–O31, Si–C32,33,34,35,36,37,38,39,40,41 or Si–Cl15 bonds. Most reactions are catalyzed by expensive palladium or rhodium catalysts, therefore limiting the scalability and practicality of the strategy. A third strategy: enantioconvergent transformation of racemic organosilanes, has rarely been considered, except for recent progress in a Rh-catalyzed dynamic kinetic asymmetric intramolecular hydrosilylation of alkynes with racemic hydrosilanes42, and dynamic kinetic asymmetric transformation of racemic tetraorgano allyl silanes into silylethers using imidodiphosphorimidate (IDPi) catalysts43.

We began by exploiting the fact that silicon centers, particularly those with strong Lewis acidity, are prone to Lewis base-assisted racemization involving highly reactive silicon species with unstable chirality at the silicon, such as pentacoordinate silicates44,45,46. As depicted in Fig. 1c, we hypothesized that if we could establish a rapid interconversion between two enantiomeric silicon species (SiabcX) through the activation of a Lewis basic catalyst and that if one of the enantiomers reacted with the enantioenriched reactant d* much faster than the other enantiomer did, the starting racemic mixture would fully collapse to a single diastereomer bearing an enantioenriched silicon center [e.g. (S)-Siabcd]. Such enantioconvergence would be mechanistically distinct from traditional resolution or desymmetrization.

Here, we validate our hypothesis by designing and achieving a Lewis base-catalyzed dynamic kinetic47 silyletherification process of racemic chlorosilanes 1 with easily accessible (S)-lactate analogs 2 in the presence of 4-aminopyridine 3e as catalyst (Fig. 1d). This approach allowed a practical synthesis of silylethers 4 with high diastereoselectivity and in high yields on up to 10 g-scale. We were also able to transform silylethers 4 into enantioenriched hydrosilanes, deuterosilanes and tetraorganosilanes, which are difficult to synthesize using previously reported methods.

Results

Screening of reaction conditions

A variety of commercially available enantioenriched secondary alcohols were screened in CH2Cl2 at −78 °C using 1.2 equiv. of 1a21 as the model chlorosilane. No desired silylether 4a was obtained using (R)−1-phenylethan-1-ol 2a in the presence of 2.0 equiv. of NEt3 or 2,6-lutidine, suggesting that these two Lewis bases do not activate 1a (Table 1, entry 1). In contrast, 4-DMAP (3a), which functioned as both catalyst and auxiliary base, enabled facile silyletherification of either 2a or (L)-menthol (2b), providing 4a and 4b in high yields, but with a diastereomeric ratio (dr) of only 50:50 (entries 2 and 3). Using (S)−1-(pyridin-2-yl)ethan-1-ol 2c increased dr to 67:33 (entry 4), while reaction of 1a with (R)-dimethyl malate (2d), (R)-pantolactone (2e) or methyl (S)-mandelate (2f) led to the respective products 4d, 4e and 4f with respective drs of 84:16, 86:14 and 89:11 (entries 5-7). The optimal skeleton of α-carbonyl-substituted alcohols turned out to be (S)-lactates (2g-i, entries 7-9), in which 5-trifluoromethyl)furan-substituted 2i provide silylether 4i in 95% yield with dr of 92:8 (entry 10).

Table 1 Screening of reaction conditionsa
Full size table

We also screened various 4-aminopyridine catalysts. Pyridonaphthyridine 3b, the most catalytically active 4-DMAP analog reported by Steglich48, provided a slightly lower dr of 91:9 (entry 11), while the 4-primary amine-substituted pyridine 3c lowered both yield (54%) and dr (89:11) (entry 12). Using 4-aminopyridines with a 2-substitution such as 3d further reduced yield (29%) and dr (67:33) (entry 13), perhaps because the 2-substitution sterically inhibited the interaction of the nitrogen on pyridine ring with the silicon center in chlorosilane. Conversely, the 4-secondary amine-substituted pyridine 3e49 improved dr to 94:6 (entry 14). Lengthening the alkyl group on the 4-nitrogen (3f) or making it bulkier (3g) lowered either yield or diastereoselectivity (entries 15 and 16). Using 4-phenylaminopyridine 3h led to 4i in only 45% yield (entry 17), probably because the lone pair of electrons on the 4-nitrogen competitively delocalized into the phenyl ring, weakening the Lewis basicity of the pyridine nitrogen.

The combination of 0.2 equiv. of 3e as the catalyst and 2.0 equiv. of NEt3 as an auxiliary base also functioned well, giving 4i in 95% yield with dr of 93:7 (entry 18). Thus, we selected 3e as the optimal catalyst, which is an air-stable orange solid and can be prepared in two steps from tert-butyl pyridin-4-ylcarbamate on a 10-g scale in an overall yield of 85%, without the need for silica gel chromatography. Either increasing the reaction temperature to −20 °C or lowering the loading of 3e to 0.1 equiv. reduced silyletherification efficiency (entries 19 and 20). Switching the auxiliary base from NEt3 to the less basic and more sterically hindered 2,6-lutidine led to 4i in only 18% yield with a moderate dr of 87:13 (entry 21). This result implies that the auxiliary base may act as more than just an acid scavenger to influence the diastereochemical outcome.

Scope of racemic chlorosilanes 1

With the optimal reaction conditions in hand (Table 1, entry 18), the scope of racemic chlorosilanes 1 was examined using alcohol 2i (Fig. 2). Replacement of the Me group at position-a on the silicon with slightly bulkier Et, n-Pr or n-Bu groups predominantly gave (S, S)-silylethers 4j-l, albeit with lower drs. Although the drs increased with the decreasing bulkiness of alkyl substitution-a, we were unable to derive a simple rule that “smaller is better” at position-a: a sterically less demanding alkynyl group at that position (A value, 0.4 vs 1.7 for Me) led to 4m with dr of only 50:50. The alkynyl group may be too small to provide diastereomeric differentiation. Replacing the Me group with the sterically least demanding H attenuated the electrophilicity of the silicon so much that silyletherification was completely inhibited to give 4n even at room temperature. As for position-b on the silicon, replacing the t-Bu group with smaller i-Pr, n-Bu, vinyl groups or p-toluene group reduced dr to 50:50 (4o-r), while H again inhibited silyletherification (4s). These results indicate the importance of sufficient bulkiness at position-b for high diastereoselectivity.

Fig. 2: Scope of racemic chlorosilanes 1.
figure 2

aMethod A for 4j-q and 4 s: 2i (0.2 mmol), 1 (0.24 mmol), 3e (0.04 mmol), and NEt3 (0.4 mmol) in CH2Cl2 (2.0 mL) at −78 °C for 24 h. Method B for 4r and 4t-at: the corresponding hydrosilane of 1 (0.3 mmol) and BPO (0.03 mmol) in CCl4 (3 mL) at 100 °C for 24 h, followed by concentration and then reaction with 2i (0.2 mmol), 3e (20 mol %) and NEt3 (0.4 mmol) in CH2Cl2 (2.0 mL) at −78 °C for 24 h. bIsolated yield. cDetermined by 1H NMR.

Full size image

In contrast to position-a and position-b, various aryl substitutions at position-c supported good diastereoselectivity. Phenyl rings containing a variety of 4-electron-donating-substituents provided silylethers 4t-x with generally high drs. The reaction also functioned well to give silylethers 4y-ad in which the silicon was bonded to phenyl rings with 3-mono-, 3,4- or 3,5-di-, or cyclobutane-fused substituents. In contrast, the steric characteristics of the 2-substitution on the phenyl ring influenced dr: small Me (4ae) or OMe groups (4af) slightly increased dr (94:6 vs 93:7 of 4i), whereas a large phenyl group (4ag) lowered it to 72:28. Additional substitution at 3- or 5-positions disfavored formation of the (S, S)-diastereomer: dr was lower for 4ah and 4ai (88:12) than for 4ae (94:6). Chlorosilanes with electron-withdrawing substituents on the phenyl ring served as good substrates, giving 4aj-ao. The moderate drs of 4al (84:16) and 4am (82:18) imply that an electron-deficient phenyl ring probably makes the chlorosilanes more reactive, reducing diastereomeric differentiation. Naphthalene, spirobifluorene and benzofuran on the silicon were tolerated, leading to silylethers 4ap-as. The reaction also functioned well for disilyletherification using phenyl-tethered bis(chlorosilane), leading to C2-symmetric disilylether 4at in 90% yield. The observed dr of 84:16 implies that the first generated stereogenic silicon center negligibly influences the stereochemistry of the second silyletherification.

Mechanistic Studies

Analysis of reaction kinetics with different initial concentrations of components indicated first-order dependence on racemic chlorosilane 1a and 4-aminopyridine catalyst 3e, but saturation kinetics with secondary alcohol 2i and NEt3 (Fig. 3a)50,51. These findings suggest that the rate-limiting step in the overall reaction is the activation of chlorosilane 1a by 3e, after which the racemization of silicon and silylation with alcohol proceed rapidly.

Fig. 3: Analysis of reaction kinetics and 29Si NMR studies.
figure 3

a Reaction kinetics with different component concentrations. (1) First-order dependence on chlorosilane 1a. (2) Saturation kinetics with secondary alcohol 2i. (3) First-order dependence on 4-aminopyridine catalyst 3e; (4) saturation kinetics with NEt3. b 29Si NMR chemical shift of t-BuPhMeSiX 1a-X (X = Br or OTf) and the ion-paired, tetracoordinated silicon 5a-X, which was obtained by mixing 1a-X with equimolar 3e in CD2Cl2 at room temperature. c 29Si NMR spectra of 1a-Cl and 1a-Cl/3e. (1) 1a-Cl in CD2Cl2 at room temperature. (2) Equimolar mixture of 1a-Cl and 3e in CD2Cl2 at room temperature. (3) Equimolar mixture of 1a-Cl and 3e in CD2Cl2 at −78 °C.

Full size image

Halosilanes are known to undergo Lewis base-catalyzed racemization52,53,54. One of the possible pathways has been proposed to involve interconversion of the ionic tetracoordinated silicon complex55, which have been characterized or even isolated in the racemization or alcoholysis of halosilanes in the presence of pyridine56, imidazole57 or NEt358. We hypothesize that such an intermediate may also account for the racemization between SiS-1a and SiR-1a.

Consistent with this hypothesis, 29Si NMR spectra of the equimolar mixtures of 3e with either t-BuPhMeSiBr 1a-Br or t-BuPhMeSiOTf 1a-OTf at room temperature in CD2Cl2 showed new, strong 29Si resonances both appearing within ±0.3 ppm of 20.0 ppm (20.2 ppm for 1a-Br/3e and 19.7 ppm for 1a-OTf/3e) (Fig. 3b). Under the standard reaction conditions, reaction of 1a-Br with 2i provided 4i in 96% yield with 93:7 dr, while reaction of 1a-OTf with 2i provided 4i in 92% yield with 70:30 dr. These results support the existence of the ion-paired, tetracoordinated silicon species 5a-X, in which the chemical shift of the silicon lies around 20.0 ppm regardless of the counterion. Mixing t-BuPhMeSiCl 1a-Cl with 3e at room temperature did not produce a new signal around 20.0 ppm, probably for thermodynamic reasons. In contrast, mixing the two compounds at −78 °C for 10 min led to complete disappearance of the 1a-Cl signal at 25.1 ppm and appearance of a sharp peak at 19.7 ppm (Fig. 3c). These results suggest that formation of the ion-paired, tetracoordinated silicon intermediate 5a-Cl is kinetically favored at −78 °C.

Based on these experimental results, we proposed the mechanism of our dynamic kinetic silyletherification of racemic chlorosilanes in Fig. 4. To simplify the calculation, we used N-methyl-substituted catalyst 3i, which showed comparable efficiency with 3e to give 4g in 95% yield with 91:9 dr. The n–σ* interaction between neutral 3i and the racemic mixture SiS-1a and SiR-1a provides the ion-paired, tetracoordinated silicon intermediates SiS-IM1 and SiR-IM1. This process is probably irreversible at −78 °C as suggested by the 29Si NMR spectra of the equimolar mixture of 1a and 3i (Fig. 3c). Even though the concentrations of SiS-IM1 and SiR-IM1 are low, the electrophilicity of their silicon is much greater than that in the neutral chlorosilane 1a (Lewis base activation of Lewis acid59), facilitating the attack by the second 3i molecule. DFT calculations predict that SiS-IM1 and SiR-IM1 interconvert through transition states SiS-TS1 (3.4 kcal/mol), SiR-TS1 (3.3 kcal/mol) and a common symmetrical pentacoordinate silicate intermediate Si-IM2 (3.8 kcal/mol), leading to racemization of their silicon centers. Such an interconversion process is energetically very favorable, suggesting that it is rapid and reversible, consistent with the observed first-order dependence on catalyst in the overall reaction. The high electrophilicity of SiS-IM1 and SiR-IM1 also facilitates the subsequent irreversible silyletherification with (S)-lactates such as 2g. DFT calculations suggest two hexacoordinate transition states SiS/CS-TS2 and SiR/CS-TS2 accounting for the formation of SiS-4g and SiR-4g, respectively. The O atoms of the ester group in 2g coordinate to the Si atoms in the orientation trans to the catalyst, with the O…Si distances of 2.02 Å for SiS/CS-TS2 and 2.00 Å for SiR/CS-TS2. Such activation promotes the nucleophilic attack of the hydroxy group in 2g to silicon center, accompanied by deprotonation with NEt3 and releasing the catalyst. In both cases, the most sterically demanding t-Bu group forces the chiral moiety in 2g as far away from it as possible, while Me group acts as the least sterically demanding group, combining with the larger Ph group to produce the stereo-discrimination during formation of SiS-4g and SiR-4g.

Fig. 4: Elucidation of the mechanism.
figure 4

a SiS-1a reacts with catalyst 3i to give tetracoordinated silicon intermediate SiS-IM1, which was transformed into SiS-4g via the favorable transition state SiS/CS-TS2 (21.6 kcal/mol); b SiR-1a reacts with catalyst 3i to give SiR-IM1. Intermediate SiR-IM1 prefers to undergo racemization by rapid interconversion to give SiS-IM1 via pentacoordinate silicon Si-IM2, leading to SiS-4g via SiS/CS-TS2, rather than reacts with 2g to give SiR-4g via the unfavorable transition state SiR/CS-TS2 (27.0 kcal/mol); c c: Non-covalent interaction visualized by Multiwfn 3.8 (dev) software.

Full size image

DFT calculation predicts that SiS/CS-TS2 (21.6 kcal/mol) is more stable than SiR/CS-TS2 (27.0 kcal/mol) by 5.4 kcal/mol at 195 K, because the non-bonded repulsion between the Me group on the silicon and the α-Me group in 2g in the case of SiS/CS-TS2 appears to be less severe than that between the Ph group on the silicon and the α-Me group in 2g in the case of SiR/CS-TS2, as suggested by the non-covalent interaction analysis. Thus, SiR-IM1 energetically prefers racemizing to SiS-IM1, leading to the dynamic kinetic transformation of racemic 1a into SiS-4g, consistent with the observed experimental yield (97%) and dr (92:8) (Fig. 5).

Fig. 5: Diverse transformations of the carboxylic acid analog 6g.
figure 5

a 10-gram scale synthesis of 6g; b: X-ray structure of 6g, and transformations of 6g into chiral organosilanes 7, 8, 9, 10, 11, 12, and 13.

Full size image

Diverse transformations

We confirmed the synthetic usefulness of our reaction by silylating racemic chlorosilane 1a with (S)-methyl lactate 2g, a commercially available chiral feed stock, leading to (S, S)-silylether 4g on a 10-g scale, in 97% yield and dr of 92:8 (Fig. 5a). Treating 4g with aqueous LiOH followed by one recrystallization provided the corresponding carboxylic acid analog 6g as an air-stable colorless solid in 84% yield and dr ≥ 99:1. We were then able to convert 6g into various enantioenriched organosilanes.

Reaction of 6g with n-butyl-, phenyl, alkynyl or 2-furanyllithiums yielded tetraorganosilanes 7−10 in high yields, with retention of configuration60,61,62 at silicon leading to ers up to 99:1 (Fig. 5b). Synthesis of 8 would be particularly challenging using typical resolution or desymmetrization strategies because the phenyl and 4-methyl phenyl rings are sterically similar. In addition to organolithiums, the Grignard reagent allyl magnesium chloride63,64 served as a good nucleophile to afford synthetically useful chiral allylsilane 11 in 60% yield with er of 98:2. Reduction of 6g with DIBAL-H65,66 or LiAlD4 generated, respectively, hydrosilane 12 in 92% yield or deuterosilane 13 in 90% yield, with retention of configuration at silicon in both cases (er = 98:2). Both 12 and 13 are promising chiral reductants, while deuterosilane 13 may also be useful for asymmetric deuteration in drug discovery67,68,69.

Discussion

We have developed an enantioconvergent synthesis of silicon-stereogenic silylethers that involves silyletherification of racemic chlorosilanes using (S)-lactates in the presence of 4-aminopyridine catalyst. Although the racemization of silicon poses a problem for asymmetric syntheses in other contexts, we have exploited it to achieve dynamic kinetic silyletherification process of racemic chlorosilanes. chlorosilane, whereas a wide range of aryl groups is tolerated. The rate-limiting step in the reaction appears to be n-σ* interaction between the catalyst and the silicon center in chlorosilane, tetracoordinated silicon intermediate, which racemize via rapid interconversion assisted by the second catalyst molecule. The reaction provides scalable access to silylethers for subsequent preparation of diverse enantioenriched organosilanes, some of which are difficult to access by traditional synthetic methods.

Our approach may also provide the basis for Lewis base-catalyzed dynamic kinetic resolutions and dynamic kinetic asymmetric transformations involving stereogenic silicon centers. Those studies are currently underway in our group.

Methods

Method A

To a 10 mL round-bottom flask charged with 2i (48 mg, 0.2 mmol, 1.0 equiv.) in dry CH2Cl2 (2 mL) were added chlorosilanes (0.24 mmol, 1.2 equiv.), 3e (6 mg, 0.04 mmol, 20 mol%) and NEt3 (56 μL, 0.4 mmol, 2.0 equiv.) under an inert atmosphere of argon at -78 oC. The mixture was stirred for 24 h at −78 °C before warming to room temperature and removing the solvent under reduced pressure. Purification by column chromatography on silica gel (gradient eluent: Petroleum Ether to Petroleum Ether/EtOAc = 30:1) afforded the desired product 4.

Method B

Step 1: To a 10 mL round-bottom flask charged with silanes (0.3 mmol) and CCl4 (3 mL) was added benzoyl peroxide (8 mg, 0.03 mmol). The mixture was refluxed for 21 h followed by stirring for 3 h at room temperature before removing the solvent under reduced pressure. The crude product was used without purification in the next step.

Step 2: To a 10 mL round-bottom flask charged with 2i (48 mg, 0.2 mmol, 1.0 equiv.) in dry CH2Cl2 (2 mL) were added the crude chlorosilane 1 (1.2 equiv.), 3e (6 mg, 0.04 mmol, 20 mol%) and NEt3 (56 μL, 0.4 mmol, 2.0 equiv.) under an inert atmosphere of argon at -78 oC. The mixture was stirred for 24 h at −78 °C before warming to room temperature and removing the solvent under reduced pressure. Purification by column chromatography on silica gel (gradient eluent: Petroleum Ether to Petroleum Ether /EtOAc = 3:1) afforded the desired product 4.