Sharpless Asymmetric Dihydroxylation

What the Sharpless AD does

A carbon-carbon double bond is flat. Both faces of an alkene are, on their own, equivalent-looking planes — attack one and you get one enantiomer of the diol, attack the other and you get its mirror image. Plain osmium tetroxide (OsO₄) will happily dihydroxylate an alkene, but it has no way to tell the two faces apart, so on a prochiral alkene it gives a 50:50 racemate. The Sharpless reaction bolts a chiral cinchona-alkaloid ligand onto the osmium so that only one face fits into the binding pocket. The result: two new C–O bonds, on the same face, with the absolute configuration you chose in advance.

The whole thing runs at 0 °C to room temperature in a tert-butanol/water mixture with commercially premixed reagents called AD-mix. There are two of them:

• AD-mix-β — contains the ligand (DHQD)₂PHAL (from dihydroquinidine). Delivers the diol from the "top" face in the Sharpless mnemonic.
• AD-mix-α — contains the pseudoenantiomeric ligand (DHQ)₂PHAL (from dihydroquinine). Delivers the diol from the "bottom" face — the opposite enantiomer.

Pick the mix and you pick the enantiomer. Same flask, same substrate, same conditions — just swap α for β and the product's optical rotation flips sign.

The mechanism, arrow by arrow

The reaction is a catalytic cycle with a stoichiometric co-oxidant regenerating the metal. Follow one alkene through it:

1. Ligand binding. The bis-cinchona ligand coordinates to osmium(VIII) through one of its quinuclidine nitrogen lone pairs. This is not a spectator — binding accelerates the reaction (ligand-accelerated catalysis) and builds the chiral wall that discriminates the two alkene faces.
2. [3+2] cycloaddition. The alkene's π electrons and two of the osmium's oxo (Os=O) oxygens close a five-membered ring in a single concerted, suprafacial step. Both new C–O bonds form at once, on the same face, giving an osmium(VI) glycolate (a cyclic osmate ester). Osmium drops from +8 to +6; the alkene's π bond is gone. (The alternative "[2+2] then rearrange" pathway was debated for years; kinetic isotope-effect work by Sharpless and by Houk's calculations settled it in favor of a concerted [3+2].)
3. Hydrolysis. Water (aided by the base and, for most substrates, added methanesulfonamide) cleaves the osmate ester. Out comes the free syn-1,2-diol; osmium is released as an Os(VI) species. Because there was never a free carbon intermediate, the two hydroxyls stay cis — the syn relationship is mechanically guaranteed.
4. Reoxidation. The stoichiometric oxidant — potassium ferricyanide K₃Fe(CN)₆ under basic conditions (or NMO in the older Upjohn protocol) — pushes Os(VI) back up to Os(VIII). The regenerated OsO₄ ligates a fresh cinchona alkaloid and the cycle turns again.

        R                 R   OH
         \                 \ /
          C=C   ──AD-mix──→  C
         /    \             / \
        R'     H           C   H     (syn-1,2-diol, single enantiomer)
                          / \
                        R'   OH

  cycle:  Os(VIII)O4 · L*  +  alkene  →  Os(VI) glycolate       ([3+2], syn addition)
          Os(VI) glycolate  +  H2O     →  syn-diol  +  Os(VI)    (hydrolysis)
          Os(VI)  +  2 Fe(CN)6^3-  +  2 OH-  →  Os(VIII)  +  ...  (reoxidation)

The two-phase tert-BuOH/water system matters: the osmylation and reoxidation happen in different phases, which keeps the concentration of Os(VIII) low. Low [Os(VIII)] suppresses the racemic, ligand-free "second cycle" pathway that would erode the ee — a subtle but crucial design choice that distinguishes the high-ee ferricyanide conditions from the older NMO conditions.

Reagents, catalyst, and conditions

• Osmium source. Not volatile OsO₄ itself but the non-volatile, easier-to-handle salt potassium osmate, K₂OsO₂(OH)₄, at 0.2–1 mol%. It is oxidized in situ to the active OsO₄.
• Chiral ligand. A dimeric cinchona alkaloid, most commonly (DHQD)₂PHAL or (DHQ)₂PHAL, at ~1 mol%. PHAL = phthalazine, the spacer that links the two alkaloid units into a binding cleft.
• Stoichiometric oxidant. 3 equiv K₃Fe(CN)₆ with 3 equiv K₂CO₃ (modern, high-ee) or 1.1–1.5 equiv NMO (older Upjohn conditions, cheaper but usually lower ee).
• Additive. 1 equiv methanesulfonamide (CH₃SO₂NH₂) for all internal alkenes — it accelerates hydrolysis of the osmate ester roughly 30-fold and keeps turnover high. Omit it for terminal alkenes, where it can actually slow things down.
• Solvent / temperature. tert-BuOH : H₂O (1:1), 0 °C to room temperature, typically 6–24 h.
• Workup. Quench with sodium sulfite (Na₂SO₃) to reduce residual osmium, extract, then scavenge trace Os for pharma-grade material.

A convenient practical fact: one bottle of "AD-mix-β" already contains the osmate, the ligand, the ferricyanide, and the carbonate in the right ratio. You add ~1.4 g of AD-mix per mmol of alkene, plus the methanesulfonamide, and stir.

Selectivity and stereochemistry: the mnemonic

Sharpless distilled the facial selectivity into a picture that predicts the outcome for almost any alkene. Draw the alkene in a plane with its largest substituent (R_L) in the lower-left, the smallest group toward the lower-right, and the remaining substituents on top. Then:

• AD-mix-β (DHQD) attacks from above the plane (β face).
• AD-mix-α (DHQ) attacks from below the plane (α face).

The mnemonic works because the alkene has to slide edge-on into a chiral U-shaped groove formed by the two quinoline rings sitting over the phthalazine floor. The largest substituent points out of the cleft; the enzyme-like pocket only accommodates one facial approach. It's essentially a small-molecule active site.

Two stereochemical guarantees fall out of the mechanism:

• syn addition, always. The concerted [3+2] delivers both oxygens to the same face, so you get the cis (syn) diol, never the trans. Compare epoxidation-then-hydrolysis, which gives the anti diol.
• Predictable absolute configuration. With the mnemonic, you know which enantiomer you'll get before you run the reaction — the reason AD is a planning tool in total synthesis, not just a functional-group interconversion.

Sharpless AD vs related alkene oxidations

	Sharpless AD	OsO₄ / NMO (Upjohn)	Sharpless epoxidation	Epoxidation + hydrolysis
Product	syn-1,2-diol	syn-1,2-diol	2,3-epoxy alcohol	anti-1,2-diol
Diastereo­selectivity	syn (cis)	syn (cis)	epoxide	anti (trans)
Enantioselective?	Yes — chiral ligand	No — racemic on prochiral alkenes	Yes — (+)/(–)-DET	No (unless epoxide was chiral)
Substrate requirement	Any alkene (broad)	Any alkene	Allylic alcohol only	Any alkene
Metal	Os (0.2–1 mol%)	Os (1–5 mol%)	Ti(OiPr)₄	none (peracid) or none
Chiral controller	(DHQD)₂PHAL / (DHQ)₂PHAL	—	(+)- or (–)-diethyl tartrate	—
Typical ee	90–99.8%	0% (racemic)	90–98%	—
How you pick the enantiomer	AD-mix-α vs -β	can't	(+) vs (–)-tartrate	can't

The clean division of labor is worth remembering: Sharpless epoxidation handles allylic alcohols and makes chiral epoxides; Sharpless dihydroxylation handles essentially any alkene and makes chiral syn-diols. Together with the later Sharpless aminohydroxylation (which installs one OH and one NH₂), they form a toolkit for enantioselective alkene functionalization.

Worked example: (E)-stilbene → (R,R)-hydrobenzoin

The textbook demonstration. trans-Stilbene (PhCH=CHPh) is a symmetric, trans-disubstituted alkene — the class the AD handles best.

    PhCH=CHPh  ──AD-mix-β, CH3SO2NH2──→  (R,R)-PhCH(OH)-CH(OH)Ph
               t-BuOH/H2O (1:1), 0 °C            (R,R)-hydrobenzoin

• Reagents. 1.0 mmol trans-stilbene, 1.4 g AD-mix-β, 1.0 mmol methanesulfonamide.
• Conditions. 5 mL tert-BuOH + 5 mL water, stir at 0 °C for ~6 h (monitor by TLC), warm to RT to finish.
• Workup. Add solid Na₂SO₃, stir 30–60 min, extract with ethyl acetate, wash, dry, recrystallize.
• Result. (R,R)-1,2-diphenyl-1,2-ethanediol (hydrobenzoin) in ~90% yield and ~99% ee. Switching to AD-mix-α gives the (S,S) enantiomer under otherwise identical conditions.

The near-perfect ee here (stilbene routinely gives 99.8% ee) is why this substrate is the standard proof-of-concept for the reaction and a common undergraduate lab experiment demonstrating catalytic asymmetric synthesis.

Real-world applications

• Anti-HIV drugs. The AD reaction installs the chiral diol handles used to build the aminoindanol and related stereocenters in HIV protease inhibitors; it was an early showcase of catalytic asymmetric methods in process chemistry.
• Diltiazem and taxol side chains. AD provides enantiopure syn-diol and, via the aminohydroxylation cousin, amino-alcohol fragments used in cardiovascular drugs and in the C-13 side chain of Taxol (paclitaxel).
• Chiral building blocks. A 1,2-diol is a versatile synthon — selectively protect one OH, convert to an epoxide, a cyclic sulfate, or an amino alcohol. Enantiopure diols from AD feed into terpenes, sugars, prostaglandins, and macrolide syntheses.
• Sugar synthesis. Because AD sets two adjacent stereocenters with predictable configuration, it is a workhorse for building the polyol backbones of carbohydrates and iminosugars.
• Fine-chemical process. The sub-mol% osmium loading, non-volatile osmate salt, and premixed AD-mix reagents make the reaction robust enough for kilo- to multi-kilo-scale asymmetric manufacture with osmium scavenged to low ppm in the product.

Limitations and side reactions

• Terminal and cis-alkenes are hard. Monosubstituted (terminal) alkenes and cis-1,2-disubstituted alkenes present two similar prochiral faces, so ee frequently sags to 70–90%. Specialized ligand backbones (PYR for terminal, indoline-based for others) recover some of it, but these remain the weak substrates.
• Osmium toxicity and cost. Even at ≤1 mol%, osmium is toxic and must be scavenged from the product. OsO₄ vapor is dangerous to eyes and lungs — a key reason the non-volatile osmate salt and closed AD-mix format are used.
• Over-oxidation. Under the older NMO conditions, a "second catalytic cycle" through an Os(VIII) trioxoglycolate can cleave the diol (to carbonyls) and erode ee. The ferricyanide two-phase system was designed specifically to shut this down.
• Electron-poor alkenes are sluggish. Very electron-deficient olefins (e.g. some α,β-unsaturated carbonyls) react slowly; more forcing conditions or tuned ligands are needed.
• Chelating groups can misdirect. Substrates with basic nitrogens or strong metal-coordinating groups can bind osmium or the ligand and scramble the facial selectivity.
• Hydrolysis-limited turnover. Without methanesulfonamide, slow osmate-ester hydrolysis bottlenecks the cycle for hindered internal alkenes; the additive is essentially mandatory there.

Historical discovery

Stoichiometric osmylation of alkenes to cis-diols dates to the early 20th century, and the catalytic Upjohn dihydroxylation (OsO₄ + NMO co-oxidant) was reported by Van Rheenen and coworkers in 1976. The asymmetric breakthrough came from K. Barry Sharpless at MIT and later Scripps. In 1988 his group reported that adding a chiral cinchona alkaloid to a catalytic osmylation gave enantioenriched diols; the crucial insight was ligand-accelerated catalysis — the chiral ligand doesn't just select a face, it speeds the reaction, so the selective pathway is also the dominant one.

Over 1989–1992 the group engineered the dimeric PHAL-linked ligands, the ferricyanide two-phase conditions, and the methanesulfonamide accelerant, arriving at the robust "AD-mix" formulation that made the reaction practical for any lab. Sharpless shared the 2001 Nobel Prize in Chemistry (with William Knowles and Ryoji Noyori for asymmetric hydrogenation) "for his work on chirally catalysed oxidation reactions" — the epoxidation and the dihydroxylation together. He went on to win a second Nobel in 2022 for click chemistry, one of only a handful of people to win two Nobel Prizes.

Safety and handling notes

• Use the osmate salt, not OsO₄. K₂OsO₂(OH)₄ is a non-volatile solid; it avoids the acute inhalation hazard of volatile osmium tetroxide, whose vapor severely irritates and can permanently damage eyes and respiratory tissue.
• Quench reductively. Destroy residual osmium at workup with sodium sulfite, sodium bisulfite, or thiourea before opening the flask to air.
• Contain the waste. Osmium-containing aqueous waste is collected separately and treated; do not pour Os-laden solutions down the drain.
• Ferricyanide + base. The K₃Fe(CN)₆/K₂CO₃ oxidant is benign as handled, but keep it out of contact with strong acid (to avoid HCN evolution) during disposal.
• Product purity. For pharmaceutical intermediates, residual osmium must be reduced to low ppm — resin scavengers and repeated aqueous washes are standard.