Regioselective Fluorohydrin Synthesis from Allylsilanes and Evidence for a Silicon–Fluorine Gauche Effect

Allylsilanes can be regioselectively transformed into the corresponding 3-silylfluorohydrin in good yield using a sequence of epoxidation followed by treatment with HF·Et3N with or without isolation of the intermediate epoxide. Various silicon-substitutions are tolerated, resulting in a range of 2-fluoro-3-silylpropan-1-ol products from this method. Whereas other fluorohydrin syntheses by epoxide opening using HF·Et3N generally require more forcing conditions (e.g., higher reaction temperature), opening of allylsilane-derived epoxides with this reagent occurs at room temperature. We attribute this rate acceleration along with the observed regioselectivity to a β-silyl effect that stabilizes a proposed cationic intermediate. The use of enantioenriched epoxides indicates that both SN1- and SN2-type mechanisms may be operable depending on substitution at silicon. Conformational analysis by NMR and theory along with a crystal structure obtained by X-ray diffraction points to a preference for silicon and fluorine to be proximal to one another in the products, perhaps favored due to electrostatic interactions.


■ INTRODUCTION
−7 New methods continue to emerge for the preparation of organofluorine compounds, including both catalytic and enantioselective systems. 8The ongoing need of drug discovery programs for fluorine-containing building blocks makes important the development of reliable, scalable, and selective strategies to generate organofluorine chemicals capable of further functionalization. 9,10Herein, we report the synthesis of 2-fluoro-3-silylpropan-1-ols from allylsilanes by a sequence of epoxidation and epoxide opening with HF•Et 3 N. Epoxide opening occurs with complete regioselectivity and appears to proceed via a blend of S N 1 and S N 2 mechanisms depending on substitution at silicon.Analysis of the 2-fluoro-3silylpropan-1-ol products revealed a conformational preference for silicon and fluorine to be in close proximity.Whereas preferred conformations of other fluorine-containing molecules is generally considered to be the result of hyperconjugation, 11−13 we hypothesize that electrostatic interactions contribute to the observed conformational bias of 2-fluoro-3silylpropan-1-ol systems.

■ RESULTS AND DISCUSSION
As part of an ongoing program investigating new reactions of allylsilanes, 14−16 we observed that epoxysilanes, prepared by epoxidation of the corresponding allylsilane, are cleanly converted to the corresponding fluorohydrin upon treatment with triethylamine trihydrofluoride (HF•Et 3 N; Table 1). 17ther HF sources such as Olah's reagent (HF•Py) resulted in significant decomposition.However, the addition of commercially available HF•Et 3 N (ca.37% HF) to a solution of the epoxysilane in dichloromethane (DCM) at room temperature produced the 2-fluoro-3-silylpropan-1-ols in uniformly high yield and with complete regioselectivity.In a typical experiment, the allylsilane was epoxidized using in situ-prepared dimethyldioxirane. 18This generally gave a sufficiently pure epoxide that could be taken directly into the epoxide opening with HF•Et 3 N ("conditions B", Table 1).An exception was allyltrimethylsilane, where the resulting epoxide proved somewhat volatile, making its isolation challenging.Instead, a one-pot epoxidation/epoxide opening adapted from the procedure of Sedgwick et al. was performed by the treatment of allyltrimethylsilane with mCPBA and HF•Et 3 N in DCM ("conditions A", Table 1). 19Under these conditions, the corresponding fluorohydrin 1 was isolated in 65% yield (entry 1) containing small amounts (∼10%) of 1-hydroxy-3-(trimethylsilyl)propan-2-yl 3-chlorobenzoate, resulting from opening of the epoxide by mCPBA-derived 3-chlorobenzoic acid. 20,21The reaction was notably slower with phenylsubstituted epoxysilanes (entries 2, 4−7), where greater phenyl substitution resulted in longer required reaction times to achieve good yields (e.g., 72 h. for Ph 3 (2, entry 2), 10 h. for (allyl)Ph 2 (5, entry 6), and 4 h.for Me 2 Ph (4, entry 4); see the Experimental Section), which may reflect a change in the mechanism (vide infra).The highest yield was obtained from allyltriisopropylsilane (entry 3), which gave the corresponding fluorohydrin 3 in 92% yield using the two-step procedure.Yields for the two-step and one-pot procedures were generally comparable (e.g., entries 4 and 5).Low isolated yields of fluorohydrin products 5 and 6 from diallylsilanes (entries 6−8) were the result of reactions (e.g., epoxidation and/or opening/ elimination) occurring at the other allyl group.The two-step procedure (conditions "A") proved optimal for producing fluorohydrin 7 from allyl(bromomethyl)dimethylsilane (60% yield, entry 9), where multiple byproducts were formed from the one-pot process.
There are a few noteworthy aspects of this transformation.First, at the outset, we were concerned about competing formation of allyl alcohol and a corresponding fluorosilane, driven by the formation of a stable Si−F bond (Scheme 1). 23y 1 H NMR analysis of the crude product mixtures, however, very little allyl alcohol was produced from any of the silanes contained in Table 1.Other minor byproducts observed were small amounts of the corresponding diol and aldehyde, the latter presumably via a Meinwald-type rearrangement. 24,25econd, this fluoride opening of epoxysilanes occurs at room temperature.Other reports of fluorohydrin synthesis by epoxide opening with HF•Et 3 N generally require heating in order to achieve high conversion. 26,27For instance, conversion of cyclohexene oxide (8) to the corresponding fluorohydrin with HF•Et 3 N required 155 °C for 5 h. 28Similarly, Adam and co-workers reported that the opening of glycidyl ether epoxide 9 with HF•Et 3 N needed a high temperature (110 °C), which gave a 53% yield of the corresponding fluorohydrins 10 as a 4:1 mixture of regioisomers. 29e attribute the rate acceleration for epoxysilane openings with HF•Et 3 N, along with complete regioselectivity, to the βsilicon effect, 30 a well-established phenomenon underpinning rate enhancements observed for other reactions involving cationic intermediates with a silicon group at the β-position. 31o benchmark the β-silicon effect in these reactions, a ∼1:1:1 mixture of allyltrimethylsilane, 1-hexene, and styrene was treated with HF•Et 3 N and mCPBA in CDCl 3 , and progress was monitored by 1 H NMR. As shown in Figure 1, after 1 h at room temperature, allyltrimethylsilane has been essentially completely consumed, converted to fluorohydrin 1 and epoxide precursor.After 2 h, no signals belonging to allyltrimethylsilane are detectable by NMR, yet significant quantities of unreacted 1-hexene and styrene remain.
In the interest of expanding to enantioenriched products by Sharpless epoxidation, 32 we also prepared and tested conversion of allylic alcohols 11 and 12 33 (Scheme 2).Use of either of these compounds resulted in low isolated yields of the corresponding fluorohydrins 13 and 14 whether by a twostep or a one-pot procedure (max.27 and 20%, respectively) due to competing elimination and formation of an Si−F species (evidenced by a singlet at ∼170 ppm in the 19 F NMR spectrum). 34It is worth noting that the d.r. of the products did not match the E/Z ratio of the starting allylsilanes, which has mechanistic implications (vide infra).Substitution of the other alkene carbon (β) or the allylic (α) position was similarly detrimental to fluorohydrin formation.The reaction of methallyltrimethylsilane (15) gave mostly unreacted starting material along with smaller amounts of unidentified byproducts.α-Hydroxy allylsilanes 16 35 and 17 36 as well as allyltrimethoxy-and allyltriethoxysilane produced exclusively elimination products.While the exact reason for the failure of these substrates is not yet known, it could be that additional substitution sterically hinders epoxide opening, thereby directing fluoride instead to attack silicon (potentially forming the fluorosilicate complex 37,38 ) and promoting elimination.Along these lines, the use of more electrophilic-at-silicon (δ + ) allylalkoxysilanes might similarly favor Si−F rather than C−F bond formation, leading to elimination over fluorohydrin formation.The Journal of Organic Chemistry To better understand the mechanism of this transformation, we attempted to prepare enantioenriched epoxysilanes 18−20 featuring differing silicon substitution by Shi epoxidation 39 of the corresponding allylsilanes (Scheme 3).Like we observed for the racemic sequence, when using allyltriphenylsilane, the Shi epoxidation was slower than the other differently substituted allylsilanes.Nonetheless, good yields of triphenylsilyl epoxide 18 could be achieved using a slightly more concentrated reaction mixture and extended reaction times.A comparison of the measured optical activity for triisopropyl silyl epoxide 19 with that previously reported 39 indicated that 18 was obtained as a 61:39 mixture of enantiomers (22% ee), in line with previously obtained values for the same transformation. 39Treatment of 18−20 with HF•Et 3 N followed by esterification with (S)-methoxy-α-(trifluoromethyl)phenylacetic acid (21) allowed for an assessment of fluorohydrin enantiopurity by 1 H NMR analysis.The enantiopurity of the resulting triisipropylsilyl fluorohydrin was determined to be 1.5:1, consistent with the ee value measured for the starting epoxide and epoxide opening via an S N 2-type process.Enantiopurities of the triphenylsilyl and dimethylphenylsilyl fluorohydrins, however, were found to be different (1.6:1 and 1.1:1).Unfortunately, allyltriphenyl-and allyldimethylphenylsilane were not included in Shi's report, 39 nor was optical rotation data available elsewhere from which the ee of the starting epoxides could be determined.Assuming a similarly low ee for 18 and 20 as that obtained for 19 (22%) by Shi's method, we were concerned that detecting minor differences between the compounds in their conversion to the corresponding fluorohydrins could be challenging.We there-

Scheme 3. Preparation of Enantioenriched Epoxysilanes and Their Corresponding Fluorohydrin with Assessment of Fluorohydrin Enantiopurity by Conversion to the Mosher Ester Derivative
The Journal of Organic Chemistry fore set out to examine alternative methods for the preparation of enantioenriched epoxysilanes to be used for understanding the mechanism of fluorohydrin synthesis.
After screening several methods (e.g., Jacobsen resolution 40 and Sharpless dihydroxylation 41 ), ultimately, we settled on Taber's alkene bromomandelation chemistry 42 for generating epoxysilanes 18−20 with an appreciable amount of enantioenrichment (Scheme 4).This protocol involves formation of diastereomeric bromomandelate adducts (22−24) that are (partially) separable by chromatography on silica.The diastereomeric purity (d.r.) of isolated fractions could be determined by NMR, which translated directly to the enantiopurity of the resulting epoxysilanes 18−20 formed upon treatment with potassium carbonate in methanol. 42omparing the enantiopurity of the starting epoxysilane 18− 20 (via the d.r. of the corresponding bromomandelate starting material 22−24 from NMR) to the enantiopurity of the corresponding fluorohydrin 2−4 (via the d.r. of the Mosher ester derivative by NMR), again differences were observed depending on substitution at silicon.
According to our analysis, the ee of the triisopropylsilyl epoxide 19 was retained in the fluorohydrin product, consistent with our previous results using 19 prepared by Shi's method 39 and suggestive of an S N 2-type epoxide opening.However, an erosion of ee was observed for the other two epoxysilanes tested (from 59 to 43% ee for Ph 3 and 64% to 38% for Me 2 Ph).The fact that complete loss of enantiopurity, which has been reported for similar transformations of styrenyl systems, 19 did not occur indicates some S N 2-type behavior.However, loss in ee suggests contribution of an S N 1 mechanism for epoxide opening involving a silyl-stabilized cation, which is consistent with the incomplete stereospecificity observed when using predominantly trans Me 2 Ph and Ph 2 allylsilanes 11 and 12 (ref.Scheme 2).−48 The fluorohydrin products obtained from this sequence are unique in that the presence of silicon adds an additional possible element of conformational control to the constraints provided by the fluorine gauche effect 49 associated with the fluorohydrin segment. 50Hyperconjugation considerations would suggest a preferred antiperiplanar C−Si and C−F arrangement, with stabilization afforded by overlap between the high energy σ C−Si and the low energy σ* C − F . 51nterestingly, NMR analysis of the different 2-fluoro-3silylpropan-1-ol products suggested a conserved preference for a gauche C−Si/C−F arrangement (Figure 2).For instance, 3-silyl fluorohydrins 1 and 2 each displayed one large (33−35 Hz) and one small (13.5−13.6Hz) 3 J HF coupling constant, consistent with a gauche rather than anti Si−F conformation.
The triphenylsilyl fluorohydrin 2 proved to be crystalline, and suitable crystals were able to be grown for analysis by Xray diffraction.Two independent structures were observed, both triple-disordered referring to uncertainty in the x,y,z planes as to where the crystal resides within the unit cell (Figure 3).In one of the solved structures, the F and OH groups are oriented gauche (dihedral angle (Ø F−OH ) = 18°), consistent with other fluorohydrin compounds. 50The other, however, shows these two groups oriented anti (Ø F−OH = 179°), perhaps influenced by the sterically large SiPh 3 .Both structures display proximity between Si and F (Ø Si−F = 6 and 38°).The dominant conformational bias in this system, therefore, appears to be a silicon−fluorine "gauche" effect.
To better understand this apparent silicon−fluorine gauche effect and its connection to the F−OH orientation in these molecules, conformational analysis by density functional theory (DFT) was performed (Figure 4).The Journal of Organic Chemistry 300°).The absolute energy minimum had both Si−F and F− OH groups gauche.This was also true for the triphenylsilyl fluorohydrin 2 (calculated energy minimum at Ø Si−F ∼ 50°for Ø F−OH = 60°), which is in fairly good agreement with results from X-ray analysis (e.g., Ø Si−F = 38°) when accounting for potential crystal forces 52,53 and the small energy differences calculated between the different conformations (ΔE ∼ 2 kJ/ mol for Ø Si−F 50°vs Ø Si−F 60°).Our working hypothesis is that the gauche−gauche conformation has the lowest energy due to a combination of hyperconjugation (e.g., σ C−H → σ C−F *) 54 and electrostatics (e.g., Si δ+ → F δ− ). 55iven not only the value of fluorine-containing compounds for drug discovery 1−7 but also an emerging interest in organosilanes for this purpose, 56 silicon-substituted fluorohydrins could present a novel platform for designing conformationally restricted biologically active structures.Alternatively, oxidative desilylation would generate 2-fluoro-1,3-propanediols, which have proven to be useful for studying enzymatic reactions involving glycerol 57,58 as well as starting points to access fluorinated carbohydrate analogues of medicinal value. 59o that end, Tamao−Fleming oxidation 60 of the 3-silyl fluorohydrin products was investigated.Ultimately it was found that after acylation of dimethylphenylsilyl fluorihydrin 4 with pivaloyl chloride (PvCl) and treatment of the resulting pivaloyl ester (4-piv) with peracetic acid (AcOOH) in the presence of sodium acetate (NaOAc) and potassium bromide, 61

■ CONCLUSIONS
In summary, various allylsilanes can be converted to the corresponding 2-fluoro-3-silylpropan-1-ols in good yield and excellent regioselectivity by epoxidation followed by epoxide opening with HF•Et 3 N.Compared with other fluorohydrin syntheses by epoxide opening with HF•Et 3 N, formation of these silicon-substituted fluorohydrins occurs more readily (e.g., at room temperature) and with higher regioselectivity that we attribute to a β-silyl effect.Reactions tended to be slower with phenyl-substituted silanes, which could be due to differences in the mechanism, which is supported by data from reactions using enantioenriched epoxysilanes.The volatility of some intermediate epoxysilanes prompted us to investigate a one-pot epoxidation/epoxide opening reaction using a combination of mCPBA and HF•Et 3 N.Yields for this onepot procedure were generally in the same range as the overall yield from a two-step process involving epoxidation with in situ-generated oxone followed by treatment with HF•Et 3 N.However, the use of mCPBA generally gave small amounts of the 3-chlorobenzoate adduct resulting from 3-chlorobenzoic acid epoxide opening.For this reason, our preferred method for substrates with suitably low volatility remains the two-step sequence.Analysis of the 3-silylfluorohydrin products by NMR, X-ray diffraction, and theory points to a preferred conformation wherein Si and F are proximal.Contrary to other fluorine gauche effects based on hyperconjugative interactions, we hypothesize that the conformational bias of 3-silylfluorohydrins is driven by an electrostatic attraction between Si δ+ and F δ− .Efforts are currently focused on further trans- General Experimental Procedures.General Procedure A1: Oxone Epoxidation.Adapted from Frohn et al.: 18 To a vigorously stirred mixture of allylsilane (1.0 mmol) and tert-butyl ammonium hydrogen sulfate (0.014 g, 0.04 mmol) in acetonitrile−dimethoxymethane (2:1, 8 mL), acetone (2.2 mL, 30 mmol), and aq K 2 CO 3 (0.1 M, 2 mL) were added oxone (3 mmol, in 8 mL of 4 × 10 −4 M EDTA solution) and aq K 2 CO 3 (1.66M, 8 mL) simultaneously via a syringe pump over the indicated time.The reaction was extracted with hexanes (3 × 20 mL), and the combined extracts were washed with brine and dried over MgSO 4 before removing the solvent on a rotary evaporator.The crude epoxide was then used directly in the next reaction without further purification.
General Procedure A2: HF•Et 3 N Epoxide Opening.A Teflon vial was charged with epoxysilane (1 equiv) and DCM (to make a 0.05− 0.1 M solution).The solution was stirred, and HF•Et 3 N (5 equiv) was added dropwise via a syringe.The reaction vessel was sealed with a Teflon screw cap, and the mixture was stirred over the indicated time at room temperature.The reaction was then quenched by pouring into a beaker containing satd.aq NaHCO 3 (75 mL) and allowed to stir until no evolution of CO 2 was observed.The mixture was transferred to a separatory funnel and extracted with DCM (3 × 25 mL).The combined organic extracts were dried over MgSO 4 , filtered, and concentrated on a rotary evaporator.The crude product was then purified by column chromatography on silica.
General Procedure B: One-Pot Epoxidation/Epoxide Opening.Adapted from Sedgwick et al.: 19 A Teflon vial was charged with mCPBA (1.3 equiv) and DCM (to make a 0.0625−0.1 M solution) and stirred until the mCPBA had dissolved.With stirring, HF•Et 3 N (5−7 equiv) was then added followed immediately by the allylsilane (1 equiv) via a syringe.The reaction vessel was sealed with a Teflon screw cap, and the mixture was stirred for the indicated time at room temperature.The reaction was quenched by pouring into a beaker containing satd.aq NaHCO 3 (75 mL) and allowed to stir until no evolution of CO 2 was observed.The mixture was transferred to a separatory funnel and extracted with CH 2 Cl 2 (3 × 25 mL).The combined organic extracts were dried over MgSO 4 , filtered, and concentrated on a rotary evaporator.The crude product was then purified by column chromatography on silica.
General Procedure C: Shi Epoxidation/Epoxide Opening.Adapted from Wang et al.: 39 To a mixture of the allylsilane (1.0 mmol) and tert-butyl ammonium hydrogen sulfate (0.014 g, 0.04 mmol) in acetonitrile (15 mL), the Shi catalyst (78 mg, 0.3 mmol) was added as a buffered solution (10 mL, 0.05 M Na 2 B 4 O 7 •10H 2 O in 4 × 10 −4 M aq Na 2 (EDTA)).With vigorous stirring, a solution of oxone (6.5 mL, 0.2 M in 4 × 10 −4 M Na 2 (EDTA)) and aq K 2 CO 3 (6.5 mL, 0.9 M) were added simultaneously via syringe pump over the indicated time.Upon completion, the reaction mixture was extracted with hexanes (3 × 20 mL).The combined organic extracts were washed with brine, dried over MgSO 4 , and filtered before removing the solvent on a rotary evaporator.The resulting epoxide product was converted to the corresponding fluorohydrin without further purification using procedure A2.
General Procedure D: Bromomandelation/Epoxide Formation/ Epoxide Opening.Adapted from Taber and Liang: 42 To a solution of (S)-mandelic acid (2.3 equiv) and 2,6-lutidine (2.6 equiv) in dry DCM (to make a 0.25 M solution) under N 2 , allylsilane was added and the flask was placed in a room temperature water bath before adding NBS (1.5 equiv).The mixture was stirred for 4−18 h before being quenched with sat.NaHCO 3 (15 mL) and extracted with DCM (2 × 15 mL).The combined organic extracts were dried over MgSO 4 , filtered, and concentrated on a rotary evaporator.The crude product was purified by chromatography on silica (10:1 to 4:1 to 1:1 Hex/ EtoAc) to yield diastereomerically enriched fractions of the bromomandelate adduct that were treated separately with K 2 CO 3 (5.0equiv) in MeOH (to make a 0.1 M solution).The reaction was stirred until completion by TLC (∼20−30 min).MeOH was then removed using a rotary evaporator, and the resulting residue was dissolved in MTBE (25 mL) and washed with aq NH 4 Cl (15 mL) and brine (15 mL).The organic phase was dried over MgSO 4 , filtered, and concentrated on a rotary evaporator.The resulting epoxide product was converted to the corresponding fluorohydrin without further purification using procedure A2.

Figure 1 .
Figure 1.Results from the treatment of a ∼1:1:1 mixture of styrene, 1-hexene, and allyltrimethylsilane (top spectrum) with HF•Et 3 N and mCPBA in CDCl 3 .By 1 H NMR, allyltrimethylsilane is clearly more reactive to these conditions, being nearly completely consumed after 1 h, whereas signals belonging to styrene (H A−C ) and 1-hexene (H D,E ) persist.

Figure 3 .
Figure 3. Crystal structure ORTEP images of triphenylsilyl fluorohydrin 2 with thermal ellipsoids at the 50% probability level.Both structures detected showed proximity between Si and F; however, the OH group was found to be either gauche (left structure; Ø F−OH = 18°) or anti (right structure; Ø F−OH = 179°).
the differentiated 2-fluoro-1,3-propanediol 25 could be obtained (Scheme 5).The low yield in this case (32%), and observed in other attempted Tamao−Fleming oxidations of 2fluoro-3-silylpropan-1-ols, was caused by competing elimina-tion, presumably via a mechanism involving intermediate Si−I.If electrostatics (i.e., attraction between Si δ+ and F δ− ) is what controls the observed Si−F gauche effect in the neutral compound, upon activation of silicon to form the corresponding silicate (Si δ− ) during Tamao−Fleming oxidation, this attraction would then become a repulsion.As a result, Si−I would adopt an antiperiplanar Si−F conformation, facilitating elimination and causing low yields of the oxidatively cleaved products.While extensive optimization of this reaction has yet to be performed, the value of 2-fluoro-3-silylpropan-1-ols as synthetic intermediates might therefore be maximized by retaining silicon and targeting functional organosilanes.

Table 1 .
Fluorohydrin Synthesis from AllylsilanesRefers only to time of the HF•Et 3 N step for conditions A. c