Asymmetric retro-[1,4]-Brook rearrangement of 3-silyl allyloxysilanes via chirality transfer from silicon to carbon.

An asymmetric retro-[1,4]-Brook rearrangement of 3-silyl allyloxysilanes has been developed via Si-to-C chirality transfer. Mechanistic studies reveal that the silyl group migrates with retention of configuration. The stereochemical outcome of the newly formed stereogenic carbon center, which has remained a longstanding question, is also clarified, suggesting a diastereoselective Si to C chirality transfer without loss of enantiomeric excess. This journal is © The Royal Society of Chemistry.

Introduction

Intramolecular O-to-C silyl migration, now called retro-Brook (or West) rearrangement, was first reported by Speier and later systematically studied by West.[1] The retro-Brook rearrangement occurs only under special circumstances[2] and so has been less investigated than Brook rearrangement.[3] But, it comprises a powerful synthetic tool because diverse organosilanes could be constructed from more accessible silyl ethers by a rapid and regio- and stereoselective manner. A covalent Si–O bond is cleaved and a Si–C bond is formed via silyl migration. Thus, the stereochemical courses at the migrating silicon center and the stereochemical control at the forming carbon center comprise two important stereochemical issues. Tomooka and co-workers[4] reported the first example of practically useful level of retro-[1,4]-Brook rearrangement of allyloxysilane by use of HMPA as a co-solvent. In this work, they also showed, for the first time, that the silyl migration proceeded with retention of configuration at the silicon center (Scheme 1). In contrast, more efforts have been directed toward diastereoselective formation of the Si–C bond to generate synthetically useful chiral organosilanes. Nearly all previous studies have used stereogenic C1,[2] C2 (ref. 2) or C3 (ref. 2) centers in substrate I to control diastereoselective formation of the Si–C bond in III. When the migrating silicon is stereogenic,[5] it might be used as a stereochemical controller by Si to C chirality transfer, which was redefined by Oestreich.[6] Achieving this in practice is quite challenging. There are only two examples we know come from a preliminary study by Tomooka and co-workers.[4] The 3-Me allyloxysilane with SiOMePht-Bu as the migrating silyl group afforded a dr of 83 : 17, while the corresponding 3-SiMe3 allyloxysilane only migrated with a dr of 66 : 34. In both cases, the stereochemistry of the formed stereogenic carbon center were not determined.

Scheme 1

Diastereoselective retro-Brook rearrangement.

Oestreich rationalized the difficulties in achieving high diastereoselectivity during Si-to-C chirality transfer as follows.[6] The relatively long Si–C bond disfavors formation of a compact transition state II, which weakens diastereoselectivity. At the same time, all three substituents on the migrating silyl group can affect the stereochemical course, requiring the careful selection of three substituents that together allow efficient stereochemical control. Despite these difficulties, Oestreich described an intermolecular Pd- catalyzed asymmetric hydrosilylation using chiral silane (Scheme 2a),[7] and Leighton demonstrated an intramolecular Hosomi–Sakuraiallylation involving a chiral allylsilane intermediate (Scheme 2b).[8] In both of these cases, either cyclic silanes or acyclic silane with three distinct, sterically demanding substituents were used to achieve the high stereochemical control.

Scheme 2

Chirality Transfer from Silicon to Carbon.

Here we report an asymmetric retro-[1,4]-Brook rearrangement of 3-silyl allyloxysilanes 1via an efficient Si-to-C chirality transfer (Scheme 2c). The combination of SiMePht-Bu as the migrating silyl group and SiPh3 as the terminal silyl group proved most effective, giving geminal bis(silyl) aldehyde 3 and enol derivatives 4 in good yield with high diastereoselectivity. The overall stereochemical outcome of the migrating silicon center and the newly formed carbon center were clarified by detailed mechanistic studies.

Results and discussion

This project arose from our interest in developing chiral geminal bis(silanes) reagents and synthons.[9] These species contain two different silyl groups, making the carbon to which they are attached a stereogenic center. In previous work, we achieved asymmetric C–C or C–H bonds formation via 3,3-sigma tropic rearrangement of optically pure 3,3-bis(silyl) allylic alcohols, allowing asymmetric synthesis of crotyl geminal bis(silanes).[10] We were curious whether asymmetric C–Si bond formation could be another efficient strategy to construct chiral geminal bis(silanes). Our s-BuLi-promoted retro-[1,4]-Brook rearrangement of 3-silyl allyloxysilanes[11] appeared to be a suitable model to test this possibility. The reaction tolerates a wide range of migrating and non-migrating silyl groups, making it practical for identifying the best pair of silyl groups.

We initially fixed t-BuPhMeSi as the migrating silyl group (Table 1). Entries 1–7 showed an obvious steric bias for the non-migrating silicons (Si1) at the 3-position of 1. When Si1 was an SiMe3 group, geminal bis(silyl) aldehydes 3a were generated as a nearly 1 : 1 mixture of two diastereomers (entry 1). Even when one methyl was replaced with a phenyl group, dr did not improve for the corresponding products 3b (entry 2). These results imply that the small methyl group does not permit good diastereochemical control. Diastereoselectivity improved progressively when steric demand at Si1 increased from SiMe3 to SiEt3, Si(n-Pr)3, Si(i-Pr)3 and finally SiPh3 (entries 3–6). The largest SiPh3 group imposed the strongest stereochemical control, providing 3f at the highest dr of 90 : 10 (entry 6). Interestingly, an inverse steric bias for Si1 was observed when the migrating silicon was switched from t-BuPhMeSi to 1-NpPhMeSi. The largest SiPh3 group afforded a dr of only 65 : 35, while the smallest SiMe3 provided the best dr of 86 : 14 (entries 7–9). We also tested the silicon combination in which the t-BuPhMeSi functioned as a chiral auxiliary, while SiPh3 migrated (entry 10). The reaction gave the aldehyde 3f with a dr of 74 : 26 lower than that obtained in entry 5.

Screening of Silyl Groupsa

Entry	Sub.	Si1	Si2	Prod.	Yieldb	drc
1	1a	Me3Si	t-BuPhMeSi	3a	62%	56 : 44
2	1b	Me2PhSi	t-BuPhMeSi	3b	65%	47 : 53
3	1c	Et3Si	t-BuPhMeSi	3c	53%	67 : 33
4	1d	(n-Pr)3Si	t-BuPhMeSi	3d	50%	65 : 35
5	1e	(i-Pr)3Si	t-BuPhMeSi	3e	45%	83 : 17
6	1f	Ph3Si	t-BuPhMeSi	3f	60%	90 : 10
7	1g	Ph3Si	1-NpPhMeSi	3g	65%	65 : 35
8	1h	Et3Si	1-NpPhMeSi	3h	73%	83 : 17
9	1i	Me3Si	1-NpPhMeSi	3i	63%	86 : 14
10	1j	t-BuPhMeSi	Ph3Si	3f	55%	74 : 26

Reaction conditions: 1 (0.15 mmol), s-BuLi (0.60 mmol), HMPA (0.6 mmol), 0.5 mL of THF, at −78 °C for 0.5 h, then H2O at rt overnight.

Isolated yields.

Ratios were determined from 1H NMR analysis of crude product.

Next we examined the scope of electrophiles for quenching the lithium enolate intermediate generated from 1f. The reaction tolerated triethylsilyl chloride (entry 1), various acyl chlorides (entries 2–9) and chlorocarbonates (entries 10 and 11) to provide 3,3-bis(silyl) enol derivatives 4 in good yields with high diastereoselectivity (Table 2). The enol double bond formed exclusively with Z-selectivity. The relative stereochemistry of the products was unambiguously established based on X-ray diffraction analysis of 4d crystals.[12] Methyl iodide was also a suitable electrophile, but less reactive than acyl chloride, giving 4l in 40% yield with O-alkylation selectivity (entry 12).

Scope of Electrophilesa

Entry	Electrophiles	Product	Yieldb
1	Et3SiCl		4a (60%)
2			4b (R = H, 65%)
3			4c (R = Br, 65%)
4			4d (R = NO2, 60%)
5			4e (67%)
6			4f (50%)
7			4g (66%)
8			4h (R = Me, 50%)
9			4i (R = Ph, 55%)
10			4j (R = Me, 70%)
11			4k (R = Ph, 70%)
12	Meld		4l (40%)

Reaction conditions: 1f (0.15 mmol), s-BuLi (0.60 mmol), HMPA (0.6 mmol), 0.5 mL of THF, at −78 °C for 0.5 h, then electrophile at rt for 2 h.

Isolated yields.

Ratios were determined from crude 1H NMR analysis of product.

10.0 equiv. of MeI.

The silicon can migrate with either retention or inversion of configuration. Thus, the relative stereochemistry of 3f may not reflect the stereochemical course of the migrating silicon, or how it controls the stereochemical outcome of the resulting stereogenic carbon center. In particular, if the enantiomerically defined silyl group racemizes during migration, the carbon center can be constructed diastereoselectively, but not enantioselectively. The observation by Tomooka and co-workers that silicon migrates with retention of configuration in their simple allyloxy system[4] does not necessarily apply to our case, since the non-migrating silicon may affect the stereochemical course. To gain a definitive answer to this question, we used enantiomerically defined 10 as a stereochemical probe (Scheme 3).[13] Following the procedure developed by Oestreich,[14] a 1 : 1 mixture of 5 was separated by several cycles of silica gel chromatography, affording 6 in diastereomerically pure form. Reduction of 6 with DIBAL-H provided hydrosilane 7 in 77% yield. Subsequent chlorination of 7 with CCl4 delivered chlorosilane 8, which directly reacted with the potassium salt of 9, giving 10 in 90% yield. The high er of 10 (96 : 4) suggests that transformation from 6 to 10 proceeds in a stereospecific manner at the silicon center, and should follow the known sequence of retention-retention-inversion.[15] Thus, the absolute configuration of the silicon in 10 was assigned as R. Under the optimal retro-[1,4]-Brook rearrangement conditions, 10 was converted into aldehyde 11. The major isomer showed an er of 96 : 4, indicating that the silicon migrated stereospecifically. X-ray diffraction analysis of 11[16] unambiguously confirmed the R-configuration of the silicon, indicating that migration proceeds with retention of configuration as in Tomooka's case. The X-ray diffraction analysis of 11 also established the S-configuration of the new stereogenic carbon center. The result revealed that the migration proceeded by a diastereoselective Si to C chirality transfer without loss of enantiomeric excess.

Scheme 3

Preparation of enantiomerically defined 10 and its retro-[1,4]-Brook rearrangement to form 11.

A plausible mechanism to explain our results is proposed in Scheme 4, based on the model we proposed for the racemic version of the reaction.[11] The α-deprotonation of 10 gives the corresponding allylic anion, which adopts the endo-orientation assisted by Li–O coordination.[17] The O-to-C silyl migration takes place irreversibly via two possible pentacoordinated silicate transition states or intermediates, TS-1 and TS-2.[18] In this way, the configuration of the silicon center is retained without racemization. While TS-2 suffers a severely steric repulsion between the Ph group on Si1 and one of the Ph groups on Si2, the interaction between the Me group on Si1 and the Ph group on Si2 appears being tolerable in the case of TS-1. These considerations are supported by the preliminary results from density functional theory calculations, which showed TS-1 to be more stable than TS-2 by 6.2 kJ mol−1. Our model also explains the observed steric bias for substituents on Si1. Substituents smaller than the Ph group might not be large enough to create an appreciable difference between the non-bonded interaction with the Me group in TS-1 and with the Ph group in TS-2. As a result, 3 forms with poor diastereoselectivity (Table 1, entries 1–4).

Scheme 4

Plausible reaction mechanism.

Conclusions

In summary, Si-to-C chirality transfer has been used as an efficient strategy to achieve asymmetric retro-[1,4]-Brook rearrangement of 3-silyl allyloxysilanes. The SiMePht-Bu and SiPh3 groups, in which SiMePht-Bu migrates, function as the best combination to give geminal bis(silyl) aldehyde and enol derivatives with high diastereoselectivity. The silyl group migrates with retention of configuration. Enantioselective generation of the stereogenic carbon center suggests that Si-to-C chirality transfer is a promising method to construct optically pure chiral organosilanes. Further applications of this strategy are being explored in our group.

Conflicts of interest

There are no conflicts to declare.