Sharpless Epoxidation

A flat molecule has two faces — pick one

A carbon–carbon double bond is flat. Approach it from the top and you get one product; approach from the bottom and you get its mirror image. With an ordinary oxidant the two faces are indistinguishable, so an alkene is epoxidized to a 50:50 mixture of left- and right-handed epoxides — a racemate, useless if you need one specific hand of a chiral drug.

The Sharpless epoxidation breaks that symmetry. It takes an allylic alcohol (a C=C double bond with an –OH on the immediately adjacent carbon), parks it onto a chiral titanium catalyst, and delivers a single oxygen atom to one chosen face of the double bond. The product is a 2,3-epoxy alcohol with two new adjacent stereocenters set in a single, predictable handedness:

          OH                            O
          |                            / \
   R—CH=CH—CH2          ───────►   R—CH—CH—CH2—OH
   (allylic alcohol)               (2,3-epoxy alcohol,
                                    one enantiomer, >90% ee)

   Catalyst:  Ti(OiPr)4  +  (+)- or (−)-diethyl tartrate
   Oxidant:   t-BuOOH (TBHP)
   Solvent:   CH2Cl2, ~ −20 °C, 4 Å molecular sieves

The catalyst supplies the chirality; the substrate need not be chiral to begin with. That is the whole game of asymmetric catalysis: a small amount of one-handed catalyst stamps its handedness onto a flood of achiral starting material. K. Barry Sharpless reported the reaction in 1980 and shared the 2001 Nobel Prize in Chemistry for it.

How the titanium–tartrate complex works

Mix titanium(IV) tetraisopropoxide, Ti(OiPr)₄, with a single enantiomer of diethyl tartrate (DET) and the two assemble into a chiral complex. The best-supported active species is a dimeric Ti₂(tartrate)₂ unit in which each titanium carries two oxygens from the bridging tartrate and exchangeable alkoxide sites. Those exchangeable sites are the business end:

1. The allylic alcohol's –OH displaces an isopropoxide and binds to titanium. This is the anchoring step — it locks the substrate's distance and orientation relative to the chiral ligand.
2. tert-Butyl hydroperoxide (TBHP) also binds titanium, again by ligand exchange, as a Ti–OOtBu alkylperoxide, positioning its electrophilic oxygen right next to the alkene.
3. Oxygen transfer is intramolecular: the peroxide's proximal O — the one coordinated to titanium — is delivered across the double bond in a concerted, syn-addition as the O–O bond breaks (the distal oxygen leaves as tert-butoxide). Both new C–O bonds form on the same face, so the epoxide geometry is set in one step.

Because the substrate and oxidant are both clipped onto the same rigid chiral scaffold, only one face of the alkene can reach the active oxygen. The other face is blocked by the tartrate ester arms. That geometric discrimination — not any difference in bond energies — is the source of the enantioselectivity.

Ligand-exchange cycle (schematic):

  Ti(OiPr)4 + tartrate  ──►  [Ti2(tartrate)2(OiPr)4]   (chiral catalyst)
                                   │  + allylic alcohol  (− iPrOH)
                                   ▼
                          [Ti–O–CH2–CH=CH–R]            (substrate anchored)
                                   │  + t-BuOOH          (− iPrOH)
                                   ▼
                      [Ti–OOtBu  ···  C=C]              (oxidant aligned)
                                   │  concerted O transfer
                                   ▼
              [Ti–O–epoxy alcohol] + t-BuOH             (product + spent oxidant)
                                   │  + fresh substrate (turnover)
                                   ▼
                          back to catalyst

Predicting the product: the tartrate mnemonic

The single most useful thing to memorize is which tartrate gives which hand. Draw the allylic alcohol flat on the page with the C=C in the plane and the hydroxyl group pointing to the lower right. Then:

                (−)-DET delivers O from ABOVE
                            ↓  O
                   ════════════════════
         R   C === C   CH2—OH   (OH to lower right)
                   ════════════════════
                            ↑  O
                (+)-DET delivers O from BELOW

  (−)-DET = D-(−)-diethyl tartrate  (unnatural)  → top-face oxygen
  (+)-DET = L-(+)-diethyl tartrate  (natural)    → bottom-face oxygen

Both tartrate enantiomers are cheap, food-grade-derived, and sold by the bottle, so you select the product's absolute configuration simply by choosing which tartrate to throw in. This predictability — "draw it this way, add this tartrate, get that enantiomer" — is exactly why the reaction became a textbook standard rather than a lab curiosity.

Conditions, loadings, and why it must be bone-dry

The original 1980 procedure used stoichiometric titanium and tartrate. The 1986 breakthrough was discovering that 3–4 Å molecular sieves scavenge trace water and keep the catalyst intact, allowing genuinely catalytic loadings — often 5–10 mol% Ti and a slight excess of tartrate. Representative conditions:

Component	Typical amount	Role
Allylic alcohol	1.0 equiv	Substrate (anchors to Ti via –OH)
Ti(OiPr)₄	5 – 10 mol%	Lewis-acidic metal centre
(+)- or (−)-DET / DIPT	6 – 12 mol% (slight excess over Ti)	Chiral ligand; sets the facial selectivity
TBHP (in decane or toluene)	1.5 – 2.0 equiv	Terminal oxidant — the O source
4 Å molecular sieves	~10 wt% of substrate	Keep system anhydrous → catalytic turnover
CH₂Cl₂, −20 °C to 0 °C	solvent / temperature	Cold improves ee; dry, non-coordinating solvent

Water is the enemy: it hydrolyzes the Ti–tartrate complex back to inactive titanium oxides and erodes both yield and ee. Tartrate diisopropyl ester (DIPT) is sometimes preferred over DET because its bulkier esters give marginally higher selectivity for some substrates. Use the workup-friendly anhydrous TBHP solutions, not aqueous hydrogen peroxide.

Real numbers: selectivity, rates, and resolution

Enantioselectivity is reported as enantiomeric excess (ee), the difference between the two enantiomers as a fraction of the total:

ee (%) = (|R − S| / (R + S)) × 100

A 95:5 ratio of enantiomers  →  ee = (95 − 5)/(95 + 5) × 100 = 90%
A 98:2 ratio                 →  ee = 96%

For the textbook case of trans-2-hexen-1-ol and related primary allylic alcohols, the Sharpless epoxidation routinely delivers 90–98% ee. The two faces differ in activation free energy by only a few kJ/mol, but selectivity is exponential in that gap:

ratio = exp(−ΔΔG‡ / RT)

For 95:5 (19:1) at 253 K (−20 °C):
  ΔΔG‡ = R·T·ln(19) = 8.314 × 253 × 2.944
       ≈ 6,190 J/mol ≈ 6.2 kJ/mol

For 99:1 at 253 K:
  ΔΔG‡ = 8.314 × 253 × ln(99) ≈ 9.7 kJ/mol

A barrier difference smaller than one hydrogen bond decides the product's handedness — and running cold (−20 °C rather than 25 °C) sharpens it, because the same ΔΔG‡ buys a larger ratio at lower T. In kinetic resolution of a racemic secondary allylic alcohol, the fast and slow enantiomers can differ in rate by k_rel > 100, letting you stop near 50% conversion and recover the unreacted alcohol in >95% ee alongside a high-ee epoxide.

Sharpless vs Jacobsen–Katsuki epoxidation

Two great asymmetric epoxidations divide the alkene world between them. The choice is dictated almost entirely by whether your substrate carries an allylic –OH.

	Sharpless epoxidation	Jacobsen–Katsuki epoxidation
Best substrate	Allylic alcohols (–OH directs)	Unfunctionalized cis-alkenes, conjugated alkenes
Catalyst metal	Titanium(IV)	Manganese(III) (salen complex)
Chiral source	Diethyl/diisopropyl tartrate	Chiral salen ligand (e.g. from 1,2-diaminocyclohexane)
Oxidant	tert-Butyl hydroperoxide (TBHP)	NaOCl (bleach), PhIO, or m-CPBA
Why it's selective	Substrate –OH binds metal; rigid geometry	Alkene approaches the chiral Mn=O pocket
Choosing the enantiomer	Swap (+)-DET for (−)-DET	Swap the salen ligand antipode
Typical ee	90 – 98% for allylic alcohols	up to ~98% for favourable cis-alkenes
Year / Nobel	1980; Sharpless, Nobel 2001	1990 (Jacobsen, Katsuki)

The headline: if there is an allylic alcohol, reach for Sharpless; if it is a plain or cis-disubstituted alkene, reach for Jacobsen–Katsuki. They are complementary, not competing.

Where it shows up: the epoxide as a chiral linchpin

The 2,3-epoxy alcohol is prized because it is a convergent intermediate: an electrophilic three-membered ring flanked by a hydroxyl handle. A nucleophile opens the ring stereospecifically (see epoxide ring opening), and the original handedness is transmitted into the product. Real uses include:

• Sharpless's own total synthesis of natural products. The reaction was a workhorse in syntheses of sugars, leukotrienes, and many polyols, where multiple stereocenters had to be set in series with predictable absolute configuration.
• (+)-Disparlure, the sex pheromone of the gypsy moth, is a chiral cis-epoxide made on multigram scale by an early Sharpless route — pheromone traps need the correct enantiomer because the moth's receptors are chiral.
• Glycidol and its derivatives. Enantiopure glycidol (2,3-epoxy-1-propanol) made by Sharpless chemistry feeds into beta-blocker drugs such as propranolol, whose activity resides in one enantiomer.
• Process-scale chiral building blocks. Because both tartrate antipodes are inexpensive and the catalyst is sub-stoichiometric, the reaction scaled to industrial chiral-pool manufacturing in a way most asymmetric methods of its era could not.

Common misconceptions and pitfalls

• "It epoxidizes any alkene." No — it needs an allylic (or homoallylic, with lower selectivity) hydroxyl to anchor the substrate. Isolated alkenes give poor or zero ee. Use Jacobsen–Katsuki, Shi epoxidation, or a peracid instead.
• "The substrate provides the chirality." The opposite. The substrate is usually achiral; the tartrate dictates the product's handedness. Swap the tartrate and you flip the product enantiomer.
• "Wetter is fine." Trace water hydrolyzes the active Ti–tartrate dimer and tanks both yield and ee. Molecular sieves and anhydrous TBHP are not optional for catalytic loadings.
• "More tartrate, more selectivity." You want a modest excess of tartrate over titanium (roughly 1.1–1.2 : 1). A large excess of free tartrate can poison turnover by over-saturating the metal.
• "ee and yield are the same thing." They are independent. You can have 99% ee at 40% yield, or 80% ee at 95% yield. ee measures handedness; yield measures how much material you recovered.
• "It sets the configuration of the alcohol carbon." The new stereocenters are the two former alkene carbons (C2 and C3 of the epoxy alcohol). The carbinol carbon was already there; the epoxide is built on the double bond, not the C–OH.