The Corey-Winter Olefination

What the Corey-Winter olefination does

You have a 1,2-diol — two hydroxyl groups on adjacent carbons — and you want a carbon-carbon double bond in their place. The obvious moves all have problems. Acid-catalyzed dehydration only removes one OH and leaves the other; it scrambles carbocations and needs harsh conditions. A double elimination via mesylates or halides demands strong base and often isomerizes the product. The Corey-Winter olefination sidesteps all of that: it deletes both oxygens at once, forms the alkene between the two former alcohol carbons, and does it under neutral, non-basic conditions.

The trick is to lash the two oxygens together into a single small ring that carries a built-in leaving-group trigger. A 1,2-diol plus a thiocarbonyl source gives a five-membered cyclic thionocarbonate (a 1,3-dioxolane-2-thione): the two former OH oxygens now flank a C=S carbonyl. A trivalent phosphorus reagent then attacks the sulfur, plucks it off, and the ring collapses — expelling carbon dioxide and leaving behind the alkene.

      OH  OH                    O   O
       |   |    (1) Im₂C=S       \\ /
    R—CH—CH—R'  ─────────→    R—CH   CH—R'   ─P(OMe)₃, Δ→   R—CH=CH—R'  + CO₂ + S=P(OMe)₃
                 (thiocarbonyl)    C                          (alkene)
                                   ‖
                                   S
                            cyclic thionocarbonate

The net result is a stereospecific syn-elimination: whatever relative configuration the two diol carbons had is written directly into the geometry of the double bond. That, plus the gentleness of the conditions, is the whole reason chemists reach for this reaction when nothing else survives.

The mechanism, arrow by arrow

Break the sequence into two operations — build the ring, then tear it down.

Step A — form the cyclic thionocarbonate. The diol reacts with a thiocarbonyl-transfer reagent. Classically this is thiophosgene (Cl₂C=S); in the modern, safer protocol it is 1,1'-thiocarbonyldiimidazole (TCDI, Im₂C=S). Each hydroxyl oxygen, acting as a nucleophile, displaces a leaving group (chloride, or imidazole) from the central thiocarbonyl carbon. Two sequential additions/eliminations stitch both oxygens onto the same carbon, closing the five-membered 1,3-dioxolane-2-thione ring. With TCDI, two equivalents of imidazole are released as the byproduct.

Step B — desulfurize and fragment. This is the step that names the reaction, and it is where the electron flow matters:

1. Phosphorus attacks sulfur. The lone pair on trivalent phosphorus (in P(OMe)₃) is a soft nucleophile; it attacks the soft, polarizable thiocarbonyl sulfur. This forms a new P–S bond and pushes the C=S π electrons onto the carbon.
2. A dioxycarbene is unmasked. As the sulfur leaves — carrying the bonding pair off toward phosphorus, ultimately as the very stable S=P(OMe)₃ (trimethyl thiophosphate) — the ring carbon is left electron-deficient with two oxygen substituents: a stabilized 1,3-dioxolan-2-ylidene carbene. The two flanking oxygens donate lone pairs into the empty carbene orbital, which is why this particular carbene is accessible under such mild conditions.
3. Cheletropic fragmentation expels CO₂. The carbene collapses: the two C–O ring bonds break, the two former diol carbons form the new π bond between them, and the central O–C(carbene)–O unit leaves as carbon dioxide. Because both C–O bonds sit on the same face of the ring, they leave together — a concerted syn elimination.

     O   O                     O   O                    R—CH
      \\ /       (MeO)₃P:         \\ /  ⊖                     ‖   +  O=C=O  +  S=P(OMe)₃
   R—CH  CH—R'  ───────→      R—CH  CH—R'   ───→        R'—CH
       C                          C:  (carbene,           alkene
       ‖                       dioxolan-2-ylidene)
       S                       — S already gone as S=P(OMe)₃

An alternative, non-carbene mechanism has been proposed for some conditions: a second equivalent of phosphite adds to the ring carbon to give a phosphorus-stabilized carbanion (an ylide-like species), which then does the C–O cleavages in a stepwise fashion before decarboxylating. Both routes deliver the same alkene with the same syn stereochemistry — the geometry is set by the ring, not by which fragmentation pathway operates.

Reagents, conditions, and the two phosphorus choices

The reaction is a two-flask sequence in practice; each flask has its standard recipe.

• Thiocarbonyl source. Thiocarbonyldiimidazole (TCDI) is the reagent of choice today — a stable, weighable solid, far safer than thiophosgene. Typical conditions: 1.1–1.5 equiv TCDI, refluxing toluene or THF, with a catalytic amount of DMAP to accelerate ring closure. Thiophosgene (the original reagent) is cheaper and very reactive but is a volatile, acutely toxic liquid requiring a base (pyridine, DMAP) to mop up HCl.
• Phosphorus reductant — the classic. Trimethyl phosphite, P(OMe)₃, often used neat as both reagent and solvent, at 120–160 °C for several hours. Triethyl phosphite works similarly. Cheap and reliable, but the high temperature is the drawback.
• Phosphorus reductant — the mild variant. 1,3-Dimethyl-2-phenyl-1,3,2-diazaphospholidine — the Corey-Hopkins reagent — is a much stronger P-nucleophile because the two ring nitrogens donate electron density onto phosphorus. It desulfurizes the thionocarbonate at or near room temperature, which rescues thermally fragile substrates that char in hot phosphite.
• What leaves. The volatile, easily removed byproducts are the appeal: CO₂ gas bubbles out, and the phosphorus departs as S=P(OMe)₃ (trimethyl thiophosphate) or the corresponding phosphine sulfide. No metal residues, no salts to filter beyond the phosphorus byproduct.

Selectivity and stereochemistry

The defining feature is stereospecificity. Because both C–O bonds are welded into one rigid five-membered ring, they can only leave syn to each other — there is no conformational freedom to eliminate anti. The consequence is a clean transfer of diol relative configuration to alkene geometry:

• A threo (anti) 1,2-diol — the two R groups on opposite faces — gives the (E)/trans alkene.
• An erythro (syn) 1,2-diol — the two R groups on the same face — gives the (Z)/cis alkene.

This makes the reaction genuinely useful as a stereodefined synthesis: pair it with a stereoselective dihydroxylation (Sharpless AD, OsO₄) that sets the diol diastereomer, and you control the alkene geometry two steps upstream. Contrast this with acid dehydration or E1, which give the thermodynamic mixture and lose all stereocontrol.

Corey-Winter vs other olefination and elimination routes

Property	Corey-Winter	Acid dehydration / E1	Barton-McCombie	McMurry coupling
Starting material	1,2-diol	mono-alcohol	mono-alcohol (via xanthate)	two carbonyls
Product	alkene (C=C from the two C–OH carbons)	alkene (loses one C–OH + β-H)	alkane (C–OH → C–H)	alkene (couples two C=O)
Conditions	neutral, non-basic; RT–160 °C	strong acid, heat	radical chain, Bu₃SnH / AIBN	low-valent Ti (TiCl₃/Zn), reflux
Stereochemistry	stereospecific syn — diol config → alkene geometry	thermodynamic (E-major), no control	not applicable (no new C=C)	poor E/Z control
Carbocation rearrangement	none — no cation intermediate	common — shifts and scrambling	none — radical, but can epimerize	none
Strained / anti-Bredt alkenes	excellent — trans-cyclooctene, cyclobutene	fails — isomerizes or polymerizes	not applicable	moderate
Functional-group tolerance	high — esters, acetals, epoxides survive	low — acid-sensitive groups die	high	low — many FGs reduced by Ti
Key byproducts	CO₂ + S=P(OMe)₃	H₂O	Bu₃SnSMe + COS	Ti oxides
Best use case	stereodefined or strained alkene from a diol	simple, robust substrates	delete a single OH	symmetric alkenes, ring closure

Worked example: trans-cyclooctene from trans-1,2-cyclooctanediol

Trans-cyclooctene is a landmark strained alkene — the smallest isolable trans-cycloalkene, so distorted that it is chiral and slowly racemizes by ring-flipping. It cannot survive strong base or acid, which makes it a textbook Corey-Winter target.

  trans-cyclooctane-1,2-diol
        │  (1) Im₂C=S (1.3 eq), DMAP, toluene, reflux 3 h
        ▼
  cyclic thionocarbonate (fused to the 8-membered ring)
        │  (2) P(OMe)₃, neat, 130 °C, 6 h
        ▼
  trans-cyclooctene  +  CO₂↑  +  S=P(OMe)₃

• Why it works. The neutral, non-basic desulfurization never generates a carbanion or carbocation that could relieve ring strain by isomerizing back to the more stable cis alkene. The syn-elimination geometry, fixed by the fused ring, delivers the strained trans double bond directly.
• Why alternatives fail. Base-mediated elimination (E2 on a dimesylate) would equilibrate the strained trans product; acid dehydration would give the thermodynamic cis-cyclooctene. Corey-Winter is one of the few ways to isolate the strained isomer cleanly.
• Practical note. Because trans-cyclooctene is reactive (it is the alkene half of many bioorthogonal "click" ligations with tetrazines), the mild workup — just evaporate the CO₂ and distill away the phosphorus byproduct — protects the fragile product.

Real-world applications

• Strained and anti-Bredt alkenes. Cyclobutene, cyclopropene, bridgehead alkenes, and trans-cyclooctene are all accessible by Corey-Winter because the reaction never routes through a species that could relieve strain. This is its signature niche.
• Bioorthogonal chemistry. Trans-cyclooctene (TCO) tags — the fastest inverse-electron-demand Diels-Alder partners for tetrazine ligation, used in live-cell imaging and pretargeted PET/radiotherapy — are frequently made or elaborated through Corey-Winter-derived strained-alkene chemistry.
• Complex-molecule total synthesis. When a diol sits inside a densely functionalized intermediate studded with acid- and base-sensitive groups (esters, acetals, epoxides, silyl ethers), Corey-Winter installs the alkene without collateral damage. It appears in steroid, terpenoid, and nucleoside-analog syntheses where a stereodefined olefin is needed late-stage.
• Nucleoside deoxygenation. Sugar-derived 1,2-diols in nucleoside chemistry are converted to unsaturated (2',3'-dideoxy-didehydro) analogs — a route relevant to antiviral nucleoside building blocks — because glycosidic bonds and base rings tolerate the neutral conditions.
• Stereocontrolled olefins from dihydroxylation. Coupled with an asymmetric dihydroxylation that sets the diol diastereomer, Corey-Winter becomes a two-step, geometry-defined alkene synthesis where the E/Z outcome is chosen at the diol stage.

Limitations and side reactions

• Requires a 1,2-diol. The whole strategy hinges on having two adjacent hydroxyls to bridge. A single alcohol cannot form the ring — for a mono-ol deoxygenation you want Barton-McCombie instead.
• Ring-closure geometry. The two OH groups must be able to reach across to a common thiocarbonyl carbon. Rigid trans-diaxial diols on large or conformationally locked rings can be slow or reluctant to cyclize into the five-membered thionocarbonate.
• Thermal load with P(OMe)₃. The classic phosphite protocol needs 120–160 °C for hours; thermally fragile substrates can decompose. The Corey-Hopkins diazaphospholidine variant fixes this by running near room temperature.
• Reagent hazards. Thiophosgene is acutely toxic and volatile; TCDI is the safer modern substitute but is moisture-sensitive. The phosphite byproducts are odorous organophosphorus sulfides.
• Not for stereorandom targets. If you don't already control the diol diastereomer, the stereospecificity you gain is moot — a stereorandom diol gives a mixture of alkene geometries, mirroring the starting mixture.

Who discovered it, and when

The reaction was reported in 1963 by Elias James Corey and Roland Arthur Edwin Winter at Harvard, in their paper on a new stereospecific olefin synthesis from 1,2-diols. Corey — who would win the 1990 Nobel Prize in Chemistry for the logic of organic synthesis and retrosynthetic analysis — recognized that binding both hydroxyls into a single ring would force a syn-elimination and give clean stereocontrol, a hallmark of his design-driven approach to synthesis.

A closely related sulfur-free transformation was independently developed by Eastwood and co-workers, which is why the reaction is sometimes called the Corey-Winter-Eastwood olefination. Eastwood's variant converts the diol into a 2-alkoxy-1,3-dioxolane (an orthoester-derived cyclic acetal) and fragments it thermally to the alkene plus a carboxylate — same stereospecific syn-elimination, but with cheaper reagents and no phosphite or thiocarbonyl. Later, Corey and Hopkins introduced the highly nucleophilic 1,3-dimethyl-2-phenyl-1,3,2-diazaphospholidine reagent, extending the thionocarbonate method to substrates too fragile for hot trimethyl phosphite and cementing Corey-Winter as the standard mild route from diols to alkenes.