A Formal Rearrangement of Allylic Silanols

We show that 1M aqueous HCl/THF or NaBH4/DMF allows for demercurative ring-opening of cyclic organomercurial synthons into secondary silanol products bearing terminal alkenes. We had previously demonstrated that primary allylic silanols are readily transformed into cyclic organomercurials using Hg(OTf)2/NaHCO3 in THF. Overall, this amounts to a facile two-step protocol for the rearrangement of primary allylic silanol substrates. Computational investigations suggest that this rearrangement is under thermodynamic control and that the di-tert-butylsilanol protecting group is essential for product selectivity.


Introduction
Rearrangement reactions can be grouped based on mechanism. One category contains true pericyclic reactions, involving a concerted flow of electrons that results in the breaking of a σ bond, simultaneous rearrangement of a π system, and formation of a new σ bond [1]. These rearrangements are effected thermally or through Lewis acid catalysis, and many landmark reactions (Cope [2], Claisen [3], Ireland-Claisen [4][5][6][7], Mislow-Evans [8], etc.) fall into this category. The second category contains formal sigmatropic processes, where the rearrangement proceeds through a discrete organometallic intermediate. A prominent example of this latter process is the Overman transposition of allylic trichloroacetimidates, which is catalyzed by either mercuric or palladium (II) salts (Scheme 1) [9].

Synthetic Investigations
This reaction was discovered serendipitously. To remove unreacted after the cyclization reaction, we explored using a 1M aqueous HCl worku sternation, we recovered starting material with trace amounts of the rearr (Scheme 3). Repeating this experiment led to the same result. We thus re aqueous HCl was promoting a demercuration reaction leading to starting m eration. We wondered if we could change the product distribution to favor silanol. Scheme 3. 1M aqueous HCl promotes an allylic rearrangement reaction. Scheme 2. Few protocols exist for the transposition of free and protected allylic alcohols.

Scheme 2.
Few protocols exist for the transposition of free and protected al

Synthetic Investigations
This reaction was discovered serendipitously. To remove unreacted after the cyclization reaction, we explored using a 1M aqueous HCl worku sternation, we recovered starting material with trace amounts of the rearra (Scheme 3). Repeating this experiment led to the same result. We thus rea aqueous HCl was promoting a demercuration reaction leading to starting m eration. We wondered if we could change the product distribution to favor t silanol.
All attempts to improve the product distribution with A were unsuc ever, we observed a dramatic improvement upon switching to an organom strate with a pendant isopropyl group ( All attempts to improve the product distribution with A were unsuccessful. However, we observed a dramatic improvement upon switching to an organomercurial substrate with a pendant isopropyl group (Table 1, Entry 1). Increasing the reaction temperature gave identical results, but dropping the temperature to 0 • C led to a decrease in yield and selectivity (Table 1, Entries 2-3). Interestingly, treatment with two equivalents of NaBH 4 in either DMF or DMSO (Table 1, Entries 4-5) was equally effective in forming rearranged product. Demercuration reactions are known with NaBH 4 , but generally the products consist of mercury simply substituted with H, OH, or I [35]. There was a profound solvent dependence on outcome with markedly worse reactions observed in MeOH, DMA, and THF ( The nature of the substituent trans to the HgCl group greatly affected the ratio of the two regioisomeric products (Scheme 4). Generally, as the steric bulk of the linear alky chain increased, so too did the yield of the terminal alkene regioisomer (Scheme 4, Entries 1-3). With branching at the α carbon (Scheme 4, Entries [4][5] or at the β carbon (Scheme 4 Entry 6), the terminal alkene regioisomer predominated. When there was competition be tween formation of a terminal alkene and its tri-substituted isomer, the more substituted olefin formed exclusively (Scheme 4, Entry 7).
Our optimized protocols were compatible with a wide array of substrates (Scheme 5). While protocol A (1M aq. HCl/THF) was tested with the majority of substrates, protoco B (NaBH4/DMF) was used for those bearing acid sensitive functionality (Scheme 5, Entries 1-2). In all but one instance (Scheme 5, Entry 10), terminal alkene products were greatly favored; in many cases (Scheme 5, Entries 1-3), these were the exclusive product. A variety of functional groups, including ketals (Scheme 5, Entries 1-2), halogens (Scheme 5, Entry 3; Scheme 5, Entry 6), and alkyl ethers (Scheme 5, Entry 3; Scheme 5, Entry 8) were wel tolerated. We were pleased to successfully convert product 40 into a single diastereomer of a protected pentitol using a combination of catalytic K2OsO4•2H2O and stoichiometric NMO (Scheme 6). The nature of the substituent trans to the HgCl group greatly affected the ratio of the two regioisomeric products (Scheme 4). Generally, as the steric bulk of the linear alkyl chain increased, so too did the yield of the terminal alkene regioisomer (Scheme 4, Entries 1-3). With branching at the α carbon (Scheme 4, Entries [4][5] or at the β carbon (Scheme 4, Entry 6), the terminal alkene regioisomer predominated. When there was competition between formation of a terminal alkene and its tri-substituted isomer, the more substituted olefin formed exclusively (Scheme 4, Entry 7).
Our optimized protocols were compatible with a wide array of substrates (Scheme 5). While protocol A (1M aq. HCl/THF) was tested with the majority of substrates, protocol B (NaBH 4 /DMF) was used for those bearing acid sensitive functionality (Scheme 5, Entries 1-2). In all but one instance (Scheme 5, Entry 10), terminal alkene products were greatly favored; in many cases (Scheme 5, Entries 1-3), these were the exclusive product. A variety of functional groups, including ketals (Scheme 5, Entries 1-2), halogens (Scheme 5, Entry 3; Scheme 5, Entry 6), and alkyl ethers (Scheme 5, Entry 3; Scheme 5, Entry 8) were well tolerated. We were pleased to successfully convert product 40 into a single diastereomer of a protected pentitol using a combination of catalytic K 2 OsO 4 •2H 2 O and stoichiometric NMO (Scheme 6). Scheme 6. Product 40 serves as a convenient precursor for a protected pentitol.

Mechanistic Studies
In order to better understand the observed selectivity, we turned to DFT calculat using the ORCA software package [37,38]. All calculations were performed using B3LYP functional [39,40] with D3BJ dispersion correction [41,42] using the RIJCOSX proximation [43]. The def2-TZVP basis set [44] was used, and implicit water solvation Scheme 5. Substrate Scope. a Relative stereochemistry shown unless explicitly indicated. b The yield of the internal alkene regio-isomer is shown in parentheses. c Isolated yield unless otherwise mentioned. d Single diastereomer but relative stereochemistry unassigned. e Yield estimated using 1 H NMR integration against methyl phenyl sulfone as an internal standard. f CCDC number 2052702. g dr =~1 Scheme 6. Product 40 serves as a convenient precursor for a protected pentitol.

Mechanistic Studies
In order to better understand the observed selectivity, we turned to DFT calculation using the ORCA software package [37,38]. All calculations were performed using the B3LYP functional [39,40] with D3BJ dispersion correction [41,42] using the RIJCOSX ap proximation [43]. The def2-TZVP basis set [44] was used, and implicit water solvation wa applied using the SMD model [45]. When mercury was present, the def2-ECP [46] wa Scheme 6. Product 40 serves as a convenient precursor for a protected pentitol.

Mechanistic Studies
In order to better understand the observed selectivity, we turned to DFT calculations using the ORCA software package [37,38]. All calculations were performed using the B3LYP functional [39,40] with D3BJ dispersion correction [41,42] using the RIJCOSX ap- proximation [43]. The def2-TZVP basis set [44] was used, and implicit water solvation was applied using the SMD model [45]. When mercury was present, the def2-ECP [46] was applied automatically. When multiple conformations were possible, a systematic rotor search was performed in Avogadro [47] to identify the lowest energy conformation as a starting point. Further details and atomic coordinates are reported in the Supplementary Materials.
Noting that lower temperature led to lower selectivity for the terminal alkene isomer, we investigated the reaction thermodynamics to determine whether the product distribution was due to equilibrium or kinetics. The simplified reaction mechanism for protocol A is: organomercury + HCl → alkene + HgCl 2 For methyl substrate 1, the internal alkene 26 is calculated to be preferred by 0.9 kcal/mol ( Figure 1). Experimentally, the observed 2.95:1 ratio favoring 26 over 27 corresponds to an expected ∆G of 0.64 kcal/mol according to a Boltzmann population analysis at room temperature: Noting that lower temperature led to lower selectivity for the terminal alkene isomer we investigated the reaction thermodynamics to determine whether the product distribu tion was due to equilibrium or kinetics. The simplified reaction mechanism for protoco A is: organomercury + HCl  alkene + HgCl2 For methyl substrate 1, the internal alkene 26 is calculated to be preferred by 0.9 kcal/mol ( Figure 1). Experimentally, the observed 2.95:1 ratio favoring 26 over 27 corre sponds to an expected ΔG of 0.64 kcal/mol according to a Boltzmann population analysis at room temperature: For isopropyl substrate 4, the terminal alkene 33 is calculated to be preferred by 0.5 kcal/mol ( Figure 2). Experimentally, the observed 3.0:1 ratio favoring 33 over 32 corre sponds to an expected ΔG of 0.65 kcal/mol by the same equation above. Because these calculated values align reasonably well with experiment, it appears that the observed re action selectivity is due to equilibrium thermodynamics. For isopropyl substrate 4, the terminal alkene 33 is calculated to be preferred by 0.5 kcal/mol ( Figure 2). Experimentally, the observed 3.0:1 ratio favoring 33 over 32 corresponds to an expected ∆G of 0.65 kcal/mol by the same equation above. Because these calculated values align reasonably well with experiment, it appears that the observed reaction selectivity is due to equilibrium thermodynamics.
To better understand this thermodynamic preference, we also modeled the deprotected allylic alcohols. For methyl substrate 1, the internal alkene (E)-2-buten-1-ol was 0.02 kcal/mol higher in Gibbs Free Energy than terminal alkene 3-buten-2-ol or nearly isoenergetic. However, for isopropyl substrate 4, the internal alkene was 0.28 kcal/mol lower in Gibbs Free Energy than the terminal alkene. Importantly, the molecular dipole of the internal alkene is about 1 Debye larger than the dipole of the terminal alkene in both cases, so implicit water solvation significantly stabilizes the internal alkene with its larger molecular dipole. Because the deprotected allylic alcohols fail to account for the observed selectivity, we conclude that the pendant di-tert-butylsilanol group plays a critical role in determining the thermodynamic selectivity of the reaction.  To better understand this thermodynamic preference, we also modeled th tected allylic alcohols. For methyl substrate 1, the internal alkene (E)-2-buten-1-ol kcal/mol higher in Gibbs Free Energy than terminal alkene 3-buten-2-ol or nearly getic. However, for isopropyl substrate 4, the internal alkene was 0.28 kcal/mol Gibbs Free Energy than the terminal alkene. Importantly, the molecular dipole o ternal alkene is about 1 Debye larger than the dipole of the terminal alkene in bo so implicit water solvation significantly stabilizes the internal alkene with its lar lecular dipole. Because the deprotected allylic alcohols fail to account for the o selectivity, we conclude that the pendant di-tert-butylsilanol group plays a critica determining the thermodynamic selectivity of the reaction.
For the reaction to be under thermodynamic control, it must be reversible. T tion barriers must be low enough to be overcome rapidly at room temperat kcal/mol). We began by modeling mercuronium rearrangements based on literatu edent for similar reactions [48]. An intrinsic reaction coordinate (IRC) calculation rearrangement transition state leading to the major terminal alkene isomer 33 is i below (Figure 3). Starting from 4, the silanol oxygen is protonated to form 59, the O bond is broken with concomitant mercuronium formation. The transition st very late and product-like, and the potential energy surface in this region is very f plicating analysis. A stationary point could not be located for the discrete mercu product 61. Instead, 61 spontaneously engages in a 5-exo ring closure to reversibly an isomeric alkylmercury species. This pathway is ultimately not productive as products of this type are isolated experimentally. Most likely, mercuronium 61 short-lived and is rapidly abstracted by chloride to form HgCl2, which was not m The overall calculated reaction pathway is exothermic by 12 kcal/mol and h kcal/mol barrier from SM 59 to TS 60. It follows that the reverse reaction starti alkene 33 should have a barrier of about 20 kcal/mol, which establishes the rea feasibly reversible at room temperature. For the reaction to be under thermodynamic control, it must be reversible. The reaction barriers must be low enough to be overcome rapidly at room temperature (<20 kcal/mol). We began by modeling mercuronium rearrangements based on literature precedent for similar reactions [48]. An intrinsic reaction coordinate (IRC) calculation from the rearrangement transition state leading to the major terminal alkene isomer 33 is included below (Figure 3). Starting from 4, the silanol oxygen is protonated to form 59, then the C-O bond is broken with concomitant mercuronium formation. The transition state 60 is very late and product-like, and the potential energy surface in this region is very flat, complicating analysis. A stationary point could not be located for the discrete mercuronium product 61. Instead, 61 spontaneously engages in a 5-exo ring closure to reversibly re-form an isomeric alkylmercury species. This pathway is ultimately not productive as no side products of this type are isolated experimentally. Most likely, mercuronium 61 is very short-lived and is rapidly abstracted by chloride to form HgCl 2 , which was not modeled. The overall calculated reaction pathway is exothermic by 12 kcal/mol and has an 8 kcal/mol barrier from SM 59 to TS 60. It follows that the reverse reaction starting from alkene 33 should have a barrier of about 20 kcal/mol, which establishes the reaction as feasibly reversible at room temperature.
The same transition state analysis was performed for the pathway leading to the minor internal alkene product 32, for which a reaction barrier of only 5.1 kcal/mol was observed. Were this reaction under kinetic control, 32 would be overwhelmingly preferred over 33 with a ∆∆G ‡ of over 3 kcal/mol. This preference for a more substituted mercuronium ion is expected from Markovnikov selectivity rules due to the partial carbocation character of the mercuronium ion. Overall, for methyl organomercury substrate 1, the internal alkene 26 is favored by both thermodynamics and kinetics, whereas for isopropyl organomercury substrate 4, kinetics favors the internal alkene 32, and thermodynamics favors the terminal alkene 33.  The same transition state analysis was performed for the pathway leading to nor internal alkene product 32, for which a reaction barrier of only 5.1 kcal/mol served. Were this reaction under kinetic control, 32 would be overwhelmingly p over 33 with a ΔΔG ‡ of over 3 kcal/mol. This preference for a more substituted ronium ion is expected from Markovnikov selectivity rules due to the partial carb character of the mercuronium ion. Overall, for methyl organomercury substrate 1 ternal alkene 26 is favored by both thermodynamics and kinetics, whereas for is organomercury substrate 4, kinetics favors the internal alkene 32, and thermod favors the terminal alkene 33.

Materials and Methods
All reagents were obtained commercially unless otherwise noted. Solvents w rified by passage under 10 psi N2 through activated alumina columns. Infrared (I tra were recorded on a Thermo Scientific™ Nicolet™ iS™5 FT-IR Spectrometer (W MA, USA); data are reported in frequency of absorption (cm −1 ). NMR spectra w orded on a Bruker Avance (Billerica, MA, USA) 400 operating at 400 and 100 M NMR spectra were recorded at 400 MHz.
Data were recorded as: chemical shift in ppm referenced internally using resi vent peaks, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = mul overlap of nonequivalent resonances), integration, coupling constant (Hz). 13 C NM tra were recorded at 100 MHz. Exact mass spectra were recorded using an elec ion source (ESI) either in positive mode or negative mode and with a time-of-fligh analyzer on a Waters LCT PremierTM mass spectrometer (Milford, MA, USA) given in m/z. TLC was performed on pre-coated glass plates (Merck) and visualize with a UV lamp (254 nm) or by dipping into a solution of KMnO4-K2CO3 in water f by heating. Flash chromatography was performed on silica gel (230-400 mesh). R phase HPLC was performed on a Hamilton PRP-1.7 μm, 21.2 × 250 mm, C18 Hg(OTf)2 was purchased from either Alfa Aesar (Ward Hill, MA, USA) or Strem

Materials and Methods
All reagents were obtained commercially unless otherwise noted. Solvents were purified by passage under 10 psi N 2 through activated alumina columns. Infrared (IR) spectra were recorded on a Thermo Scientific™ Nicolet™ iS™5 FT-IR Spectrometer (Waltham, MA, USA); data are reported in frequency of absorption (cm −1 ). NMR spectra were recorded on a Bruker Avance (Billerica, MA, USA) 400 operating at 400 and 100 MHz. 1H NMR spectra were recorded at 400 MHz.
Data were recorded as: chemical shift in ppm referenced internally using residue solvent peaks, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet or overlap of nonequivalent resonances), integration, coupling constant (Hz). 13 C NMR spectra were recorded at 100 MHz. Exact mass spectra were recorded using an electrospray ion source (ESI) either in positive mode or negative mode and with a time-of-flight (TOF) analyzer on a Waters LCT PremierTM mass spectrometer (Milford, MA, USA) and are given in m/z. TLC was performed on pre-coated glass plates (Merck) and visualized either with a UV lamp (254 nm) or by dipping into a solution of KMnO 4 -K 2 CO 3 in water followed by heating. Flash chromatography was performed on silica gel (230-400 mesh). Reversed phase HPLC was performed on a Hamilton PRP-1.7 µm, 21.2 × 250 mm, C18 column. Hg(OTf) 2 was purchased from either Alfa Aesar (Ward Hill, MA, USA) or Strem Chemicals (Newburyport, MA, USA). Di-tert-butylsilyl Bis(trifluoromethanesulfonate) was purchased from either TCI America (Portland, OR, USA) or from Sigma-Aldrich (St. Louis, MO, USA).

Conclusions
In summary, we presented a two-step protocol for the rearrangement of allylic silanols. We had previously demonstrated that primary allylic silanols are readily transformed into cyclic organomercurials using Hg(OTf) 2 /NaHCO 3 in THF. Here, we show that using either 1M aqueous HCl/THF or NaBH 4 /DMF allows for demercurative ring-opening to form secondary silanol products bearing terminal alkenes. Computational investigations suggest that this rearrangement is under thermodynamic control and that the di-tert-butylsilanol protecting group is essential for product selectivity.