Sharpless Asymmetric Dihydroxylation [Os]

This process is based on the stereospecific syn-dihydroxyaltion of alkenes by osmium tetroxide first reported by Criegee in 1936 (JLAC36-522-75).


The observation that this stoichiometric reaction is accelerated by the addition of certain tertiary amines (ligand accelerated catalysis), coupled with the high affinity displayed by quinuclidine 1 for OsO4 (JSCDT77-941), led to the use as ligands of ester derivatives of the cinchona alkaloids dihydroquinine 2 (DHQ) and dihydroquinidine 3 (DHQD) in an asymmetric variation (JACS80-102-4263). Further incorporation of N-methylmorpholine-N-oxide (NMO) as stoichiometric oxidant [Os(VI) to Os(VIII)] permitted the catalytic use of expensive and highly toxic osmium tetroxide (JACS88-110-1968). The ee values obtained were, however, modest (ca. 60% ee with styrene as substrate). The enantioselectivity was improved by use of a phthalazine linker (introduced by reaction with 4) to give (DHQ)2-PHAL 5 and (DHOQ)2-PHAL 6  (JOC92-57-2768). Coupled with the discovery that K3Fe(CN)6 in the presence of K2CO3 is an effective stoichiometric oxidant for catalytic OsO4 oxidation (JOC90-55-766) resulted in the development of AD-mix-α and AD-mix-β as a convenient to use mixtures of metal, ligand, oxidant and base (JOC92-57-2768).


The standard procedure for the dihydroxylation of 1 mmol of substrate requires 1.4 g of AD-mix-α containing K3Fe(CN)6 (0.980 g, 3 mmol), K2CO3 (0.410 g, 3 mmol), (DHQ)2-PHAL (7.8 mg, 0.01 mmol) and K2OsO2(OH)4 (1.5 mg, 0.004 mmol). The latter Os(VI) salt is used as a non-volatile and safer form of osmium. To this is added 5 mL of H2O and 5 mL of t-BuOH, and after cooling to 0 oC, the substrate is added in one portion. The resulting biphasic heterogeneous slurry (i.e. triphasic overall) is stirred vigorously at 0 oC for typically 6-24 h. For non-terminal alkene substrates the addition of CH3SO2NH2 (1 eq. with respect to substrate) is recommended to reduce the reaction time (see below). On completion, the reaction is quenched with Na2SO3 (1.5 g), and after warming to room temperature, the product is extracted with EtOAc or CH2Cl2. Methanesulfonamide may be removed from the organic extracts by washing with KOH (2 M). Illustrative reactions as:


The DHQ and DHQD ligands have the opposite configuration required for enantiomers at positions 8 and 9, but the same R configuration at position 3 (and are therefore diastereoisomers). Fortunately position 3 has only a small influence on ligand selectivity such that DHQ and DHQD are often referred to as pseudo-enantiomeric. In the example above the catalyst containing the DHQD (β) derived ligand is a little more selective than the DHQ (α) derived alternative (as is the case with many substrates). The following mnemonic may be used to predict facial selectivity. The bottom-right and to a much lesser extent the top-left quadrants impose steric barriers. The top-right quadrant is open to substituents of moderate size, and the bottom-left quadrant is an attractive area with respect to aromatics, or suitable aliphatic groups when aromatic substituents are absent. This mnemonic is consistent with the high facial selectivity observed for mono-substituted, 1,1-disubstituted, (E)-disubstituted and trisubstituted alkenes. (Z)-Disubstituted alkenes result frequently in modest enantioselectivity. Some tetra-substituted alkenes have resulted in high enantioselectivity. For substrates with more than one double bond (which may be conjugated) it is usually the most electron rich which reacts first.


Under the biphasic conditions ligand coordination to OsO4 and cycloaddition occur in the organic phase, and following hydrolysis of the resulting Os(VI) glyoxylate 6, reoxidation to Os(VIII) occurs in the aqueous phase. An advantage of these conditions is the non-participation of a secondary catalytic cycle involving oxidation of the Os(VI) glyoxylate 6 to a trioxo Os(VIII) glycolate (with loss of L*) followed by a poorly selective second cycloaddition. The operation of this secondary cycle results in lower enantioselectivities where NMO is used as an oxidant in an organic solvent. Hydrolysis of the Os(VI) glyoxylate 6 is accelerated by MeSO2NH2, shortening the reaction time such that completion occurs at 0 oC within a reasonable time. This additive is not required for terminal alkenes (mono or 1,1-disubstituted). The many contributions made by Sharpless towards this practical, predictable and highly selective asymmetric dihydroxylation method contributed towards the award of the Nobel Prize in Chemistry (2001 – together with Knowles and Noyori for work in other areas of asymmetric catalysis).


Particularly good substrates for the Sharpless AD reaction are allylic alcohol derived 4-methoxybenzoates (JACS94-116-12109, JACS95-117-10805). For example, substrate 7 results in an excellent ee in contrast to the allylic alcohol 8 from which it is derived.


This example aside, most of the work on 4-methoxybenzoates reported by Corey has been with the pyridazine (PYDZ) linked DHQ and DHQD ligands, but there appears to be little difference (other than a reduced molecular weight) to the phthalazine linked ligands used in the AD-mixes. A model to account for the excellent facial selectivity has been proposed illustrated below with (DHQD)2-PYDZ and allyl 4-methoxybenzoate as substrate (98% ee). A U-shaped conformation for the ligated-OsO4 complex creates a pocket approximately 7.7 Å wide such that the substrate in the s-cis conformation is bound by π-stacking/hydrophobic interactions which for this substrate include its aryl group and the blue methoxyquinoline moiety. The proximity of the axial (Oa) and one of the equatorial (Oe) oxygens of the nitrogen-coordinated (activated) OsO4 moiety is followed by participation in a minimal motion [3 + 2] cycloaddition via the transition state shown resulting in a Os(VI) glyoxylate The involvement of the Oe/Oa oxygens (bond angle 90o) rather than Oe/Oe (120o) is reasoned to be the similarity of the former to the experimentally observed geometry of O-Os-O in the resulting adduct (JCSCC78-853).


In support of this qualitative model proposed by Corey are the near identical enantioselectivities obtained with the conformationally restricted bis-quinidine derived macrocyclic ligand shown below (JACS93-115-12579). Subsequent calculated modelling studies identified the face-to-face π-binding of the Corey model, and also the significance with some substrates of a second face-to-edge (CH-π) binding interaction with the phthalazine/pyridazine linker. For styrene, the binding of which is represented below, approximately 50% of the total stabilisation interaction calculated is with the red quinoline ring (JACS99-121-1317). For trans-stilbene, a C-H bond(s) of the additional phenyl group participates in face-to-edge interaction with the linker such that this is of approximately equal significance to face-to-face binding (JACS99-121-10186). The two quinoline groups operate as the attractive area of the bottom-left quadrant of the mnemonic, and in addition to interacting with aromatic alkene substituents, there is also the likelihood that CH-π interactions occur with suitable alkane substituents, but this appears not to have been explicitly discussed in the literature.


That enzyme-like reversible substrate binding occurs with (DHQD)2-PYDZ.OsO4 prior to rate-limiting irreversible cycloaddition was established by the determination of Michaelis-Menten parameters Km (Michaelis constant) and Vmax (maximum velocity) values for a several substrate including those shown below (JACS96-118-319). The lower Km values for styrene 9 and allyl 4-methoxybenzoate 10 illustrate the greater binding affinity of these substrates compared to 1-decene 11. The bulky  triisopropylsilyl (TIPS) group of 12 significantly inhibits binding. From these data the relative initial rate of reaction is given by VAKB/VBKA (values here given relative to 12). The experimental difference between 9 and 11 was determined as 9, close to the calculated value of about 11 (68 vs. 6).


The low binding affinity of the bulky TIPS substituted substrate 12 permits the operation of at least one non-bound competitive pathway with opposite facial selectivity, such that the overall reaction is slow and poorly selective. Facial selectivity is a result of there being a single available orientation of the substrate with Oa and O­e that also enables favourable interaction with the binding pocket. The facility of this binding is a key factor in determining the selectivity and rate of the reaction, and corresponds to examples of enzyme catalysis where substrate binding reduces the entropic barrier by an increase in the effective concentration (in this case the concentration of one of the two prochiral alkene faces). With trans-stilbene, the operation of both face-to-face and face-to-edge binding results in this being one of the best substrates with respect to rate and enantioselectivity (99% ee).

Review: CR94-94-2483