Highly Enantiomerically Enriched Secondary Alcohols via Epoxide Hydrogenolysis

In this article, we report the development of ruthenium-catalyzed hydrogenolysis of epoxides to selectively give the branched (Markovnikov) alcohol products. In contrast to previously reported catalysts, the use of Milstein’s PNN-pincer-ruthenium complex at room temperature allows the conversion of enantiomerically enriched epoxides to secondary alcohols without racemization of the product. The catalyst is effective for a range of aryl epoxides, alkyl epoxides, and glycidyl ethers and is the first homogeneous system to selectively promote hydrogenolysis of glycidol to 1,2-propanediol, without loss of enantiomeric purity. A detailed mechanistic study was conducted, including experimental observations of catalyst speciation under catalytically relevant conditions, comprehensive kinetic characterization of the catalytic reaction, and computational analysis via density functional theory. Heterolytic hydrogen cleavage is mediated by the ruthenium center and exogenous alkoxide base. Epoxide ring opening occurs through an opposite-side attack of the ruthenium hydride on the less-hindered epoxide carbon, giving the branched alcohol product selectively.

Table S1.Results of catalyst screening and optimization.Entries in bold are also included in Table 1 in the main text.The column "e.e.(%)" lists the enantiomeric excess of the 1-phenylethanol product, and "b:l" lists the ratio of the branched 1-phenylethanol product to the linear 2-phenylethanol product.

Energies Calculated by DFT
Table S3 below shows the energies calculated by DFT for all structures reported in this paper relevant to propylene oxide hydrogenolysis catalyzed by RuCl (Scheme 6, Figure 2, and Figures S15 -S18).The column E(BS2) represents the solvent-corrected electronic energy in hartrees, calculated with the B97X-D functional and the def2-TZVP basis set.The column G(corr) represents the correction to the Gibbs free energy calculated at 298.15 K after geometry optimization using the B97X-D functional and the def2-SVP basis set.The column SS G (kcal) represents the Gibbs free energy for each isolated species at 298.15 K in kcal/mol, including the addition of 1.89 kcal/mol for each molecule to convert to a 1 M standard state, except for the solvent 2-propanol, to which 3.42 kcal/mol are added to convert to the 13.08 M standard state.The column Mass Balance lists the small molecules included in the total free energy for the calculation of reaction pathways.The column G(total, kcal) is the sum of standard-state free energies of the ruthenium complex and any small molecules included for mass balance.The column G(rel) is the total free energy referenced against RuH-solv.
Table S10 (page S33) shows the energies calculated by DFT for the reversible dehydrogenation of 2-propanol to acetone, catalyzed by RuPNN HEt (Figure 6).The DFT method is the same as that described above.The column G(rel) in Table S10 is the total free energy referenced against p, the 2-propanol hydrogen-bond adduct of RuPNN HEt .

Derivation of the Rate Law for Catalytic Epoxide Hydrogenolysis
Because both RuO i Pr and RuH occupy a significant fraction of the ruthenium speciation under catalytic conditions, saturation kinetics are expected, as described below.In the following derivation, we have assumed that the hydrogenolysis product 2-tetradecanol will interact with the ruthenium species similarly to the 2-propanol solvent.Since the solvent at 13.08 M is always at a much higher concentration than the product, we have ignored potential inhibition by the 2-tetradecanol product in the kinetic analysis.
In the analysis of experimental data, as well as in the DFT calculations, we have taken the standard state of the reacting species RuH, RuO i Pr, epoxide, and H2 to be 1.0 M in solution.The solvent isopropyl alcohol has a standard state defined by its neat molarity of 13.08 M.
The two-step sequence in Scheme 7 can be written in linear form as: If RuO i Pr and RuH are the dominant ruthenium species present, we can represent the sum of their concentrations as [Ru]total: We can apply the steady-state approximation to the RuH intermediate: Substituting ([Ru]total -[RuH]) for [RuO i Pr] gives: Solving for [RuH] gives: From reaction 2 above, the rate of product formation (the negative of the rate of epoxide consumption) is: Substituting in the steady-state concentration of RuH gives:

S24
Here, we can make a simplifying assumption that k 2 [epoxide] is negligible in the denominator.If k 2 [epoxide] were significant, saturation behavior in [epoxide] would be observed at high epoxide concentrations.This assumption is also consistent with the energy barriers predicted by DFT.The barrier from RuH-solv to e-TS (corresponding to k2) is 26.4 kcal/mol, while the barrier from RuH-solv backwards to c-TS PNN (corresponding to k-1) is 18.1 kcal/mol.This assumption gives the rate law: At this point, reaction 1 is a fast pre-equilibrium and reaction 2 is rate-determining.We can replace k-1/k1 with the reciprocal of the equilibrium constant K1 for reaction 1.Since the thermodynamic equilibrium constant K1 is calculated using the standard state of 13.08 M for iPrOH (see below), the term (k-1/k1)[ i PrOH] is equal to 1/K1.

Determination of the Equilibrium Constant K1 by NMR Spectroscopy
The equilibrium between RuO i Pr and RuH was analyzed by NMR spectroscopy in the absence of epoxide.In the glovebox, a stock solution was prepared in non-deuterated isopropyl alcohol containing 0.0137 M RuCl and 0.20 M KO t Bu.After stirring for five minutes, the solution was filtered, and 0.70 mL was transferred to a J. Young NMR tube.The NMR tube was removed from the box and hydrogen gas was added, after which the NMR tube was sealed and shaken.Thirteen such samples were prepared, with hydrogen pressures ranging from 0.2 bar to 5.6 bar.For each sample, an unlocked 1 H NMR spectrum was recorded, using a long delay between scans of 20 seconds to allow for accurate integration.The relative integrations for the H2 signal at 5.03 ppm, the RuH signal at -4.90 ppm, and the RuO i Pr signal at -15.70 ppm were used to calculate the molarities of the three species.Table S4 below summarizes this data.
Table S4.NMR Measurements of the RuH/RuO i Pr equilibrium.For the equilibrium between RuO i Pr and RuH (reaction 1 on page S25), the thermodynamic equilibrium constant K1 is defined as follows:

Integrations (arbitrary units) Concentrations (M)
Assuming that the only ruthenium species present are RuH and RuO i Pr, we can solve for [RuO i Pr]: We can then substitute this expression for [RuO i Pr] into the equilibrium equation: Solving for [RuH]/[Ru]total gives: Plotting [RuH]/[Ru]total vs [H2] allows K1 to be determined through a least-squares fit, as shown in Figure 1 in the main text.K1 was determined to be 89 ± 6, corresponding to G° = -2.66 ± 0.04 kcal/mol at 298.15 K. Table S8 shows the calculation of k2 for each independent kinetic experiment in Table S5.The molarity of hydrogen is calculated using the Henry's law constant for hydrogen in isopropyl alcohol at 298.15 K, equal to 0.003483 M/bar. 7By taking the average and standard deviation, we find a global rate constant k2 of 0.0152 ± 0.0016 M -  The data shown in Figure 5 were collected using the same apparatus and procedure described on page S28 for kinetic studies.Reaction solutions, with a total volume of 10.0 mL, were prepared with 0.050 mmol RuCl or RuPNN HEt , 2.50 mmol (R)-styrene oxide (99% e.e.), 0.188 mmol KO i Pr (from a 5% solution in isopropyl alcohol), and 0.50 mmol tetradecane as internal standard.Aliquots were removed at regular intervals and analyzed by GC-FID using the method described above in Table S2.The measured concentrations, as well as the calculated yields, e.e.'s, and branched : linear ratios reported in Figure 5, are shown in Table S9 below.

Concentrations Measured in Kinetic Experiments
Table S9.Time Course Data for Hydrogenation of (R)-Styrene Oxide by RuCl vs. RuPNN HEt .
In the above table, "% Yield" is the total yield of phenylethanol isomers, "% e.e." is the enantiomeric excess of the 1-phenylethanol product, and "b : l" is the ratio of branched 1-phenylethanol to linear 2-phenylethanol.

Table S2 .
GC temperature programs and retention times for chiral epoxides and their hydrogenolysis products.
a The underlined retention times indicate the major enantiomer of the epoxide reactant and branched product listed in Table2in the main text.bThe enantiomers of 1-tetradecene oxide were not resolved with either available GC column.Because the 2-tetradecanol product after hydrogenolysis was measured to have >99% e.e., we infer that the epoxide reactant also had >99% e.e.Chiral GC Traces for Racemic and Enantiomerically Enriched Products

Ring Opening Including Explicit 2-Propanol Figure
S15 below shows the calculated pathway for epoxide ring-opening, including an explicit solvent molecule to stabilize the developing negative charge on the epoxide oxygen.The barrier of 26.7 kcal/mol is slightly higher than the corresponding barrier of 26.4 kcal/mol in the MEP.

Table S5 . Concentration data from kinetic experiments in Figure 3.
Concentrations are in moles/liter.