Wittig Reaction

Why the Wittig matters

• Predictable C=C bond placement. The new alkene appears exactly where the old C=O sat — no rearrangement. This is unique among alkene-forming reactions; eliminations (E1, E2) place the alkene under Saytzeff or Hofmann rules that depend on substrate, while Wittig places it where the chemist puts the carbonyl. For total synthesis, this means an alkene is just one Wittig step away from any carbonyl in your intermediate.
• Mild conditions tolerate dense functionality. Stabilized ylides like Ph₃P=CHCO₂Et react with aldehydes in CH₂Cl₂ or toluene at room temperature in 1-12 h. Free alcohols, esters, amides, halides, ethers, and protected amines all survive. The reaction is one of the gentlest C=C-forming methods in the synthetic toolkit.
• Industrial-scale vitamin synthesis. BASF's vitamin A process uses a Wittig olefination as the key step linking C15 + C5 fragments. Tonnage exceeds 3000 t/yr globally; combined with DSM and Adisseo, world vitamin A production is around 6000 t/yr, and the Wittig is the dominant route. β-carotene (for animal feed and food coloring) is made by double Wittig at similar scale (~500 t/yr).
• 1979 Nobel Prize justification. Wittig's 1954 paper transformed alkene synthesis from a haphazard elimination problem into a reliable, regiospecific construction. By 1979 the reaction was used in 30+ commercial vitamin syntheses, every advanced organic chemistry curriculum, and the synthesis of many marketed pharmaceuticals. The Nobel committee cited 'the development of phosphorus-containing compounds into important reagents in organic synthesis.'
• E/Z control via ylide tuning. Stabilized ylides give 95:5 E-selectivity; non-stabilized give 85:15 Z. Schlosser modification (n-BuLi, then PhLi, with no Li⁺ salts in the second stage) sharpens Z-selectivity to >98:2 for non-stabilized cases. This pair of regimes covers most needs without separate chiral catalysts.
• Phosphonate variant (HWE) bypasses Ph₃P=O removal. The Horner-Wadsworth-Emmons modification uses (RO)₂P(=O)-CH₂-EWG instead of triphenylphosphonium. The byproduct is water-soluble dialkylphosphate, washed away in the aqueous extraction. HWE is the dominant variant for stabilized ester/amide/nitrile alkenes and is highly E-selective (>95:5).
• Compatible with macrocyclization. Intramolecular Wittig olefinations close 14-26-membered macrocycles in many natural product syntheses (FK506, erythronolide, epothilone). Macrolactonization gives the C=O; Wittig converts it to a C=C (often E) without epimerizing nearby chirality centers. Ring-closing metathesis is the modern competitor but Wittig predates RCM by 40 years.

Common misconceptions

• "All ylides behave the same." Stabilized and non-stabilized ylides differ in reactivity by orders of magnitude. Ph₃P=CHCO₂Et is bench-stable in CHCl₃ at room temperature for weeks; Ph₃P=CH₂ ignites in air and must be made and used at -78 °C in dry THF. Treating both with the same protocol gives wildly different yields.
• "The mechanism is via a betaine." The textbook 'betaine' intermediate (a zwitterion P⁺-C-C-O⁻) was the original mechanistic proposal but was never observed. Modern NMR studies (Maryanoff, Vedejs) detect the oxaphosphetane directly and find no evidence for an open-chain betaine. The reaction is a concerted [2+2] cycloaddition with asynchronous bond formation.
• "Wittig works on every carbonyl." Hindered ketones (camphor, fenchone, di-tert-butyl ketone) react sluggishly or not at all because the [2+2] approach requires close P-O contact. Ester carbonyls do not react with normal Wittig ylides — only the more nucleophilic Tebbe or Petasis reagents olefinate esters.
• "Salt effects don't matter." LiCl, LiBr, and even NaCl can shift Z:E selectivity by 10-20 percentage points for non-stabilized ylides. The Schlosser modification deliberately exploits this — generating a betaine, equilibrating with PhLi/HCl, then re-deprotonating to flip stereochemistry from Z to E.
• "HWE is just Wittig with a smaller P group." HWE has a different mechanism — the phosphonate carbanion is much harder than a phosphonium ylide, the [2+2] is more reversible, and only stabilized variants work. HWE is roughly 1000× more E-selective for ester-stabilized cases than the analogous Wittig because the trans-oxaphosphetane is more accessible.
• "Triphenylphosphine oxide is easy to remove." Ph₃P=O has mp 156 °C and is sparingly soluble in cold pentane. On silica it co-elutes with many polar products. Aqueous extraction does not remove it. Removal usually requires silica chromatography, recrystallization from hot hexane, or a CaCl₂ adduct precipitation. This is the single biggest practical drawback of the Wittig and the reason HWE often wins in industry.

Mechanism of the Wittig reaction

The mechanism has three observable stages. Step 1: ylide generation. Triphenylphosphine attacks an alkyl halide R'CH₂X by SN2 to give a phosphonium salt [Ph₃P⁺-CH₂R'] X⁻ (mp typically 200-300 °C, isolable, stable to air). A strong base then removes the α-H. For non-stabilized ylides, the α-H pKa is ~22 and you need n-BuLi (pKa 50) at -78 °C in THF. For stabilized ylides where the α-carbon bears an ester, amide, or nitrile, the α-H pKa drops to ~9 and even Na₂CO₃ in water deprotonates within minutes. The ylide is written as either Ph₃P⁺-CR'₂⁻ (ylide) or Ph₃P=CR'₂ (ylene) — the two forms are resonance contributors with comparable weight.

Step 2: [2+2] cycloaddition to oxaphosphetane. The ylide and carbonyl approach in a roughly cis manner, with the nucleophilic carbon attacking the carbonyl carbon while phosphorus reaches for the oxygen. The transition state is a four-centered, asynchronous [2+2] that bypasses the symmetry forbidden ground-state [2+2] by virtue of phosphorus's d-orbital participation and polar character. The product is a 1,2-oxaphosphetane — a four-membered ring with P (apical, pentavalent) and O on adjacent corners. Maryanoff observed these directly by ³¹P NMR (-55 to -65 ppm) at -78 °C in 1985. With non-stabilized ylides, oxaphosphetane formation is irreversible; with stabilized ylides, it is reversible, allowing thermodynamic control.

Step 3: retro-[2+2] elimination. Above -30 °C the oxaphosphetane breaks apart in a concerted retro-[2+2]. The two new bonds are C=C (the alkene) and P=O (in Ph₃P=O). The P-C and C-O bonds of the ring break simultaneously. The strong P=O bond (~544 kJ/mol) is the thermodynamic driving force — it is one of the strongest single bonds in chemistry, and its formation is what makes the Wittig essentially irreversible. Retro-[2+2] of the cis-oxaphosphetane gives the Z-alkene; trans-oxaphosphetane gives the E-alkene. The ratio reflects the kinetic vs thermodynamic preference for the two diastereomeric oxaphosphetanes set in step 2.

Wittig vs Horner-Wadsworth-Emmons vs Julia

Reaction	Reagent	α-H pKa	Typical base	Byproduct	Selectivity	Substrate scope
Wittig (non-stabilized)	Ph₃P=CHR (R = alkyl)	~22	n-BuLi, NaHMDS	Ph₃P=O	~85:15 Z:E	Aldehydes, unhindered ketones
Wittig (stabilized)	Ph₃P=CHCO₂R	~9	Na₂CO₃, NaH	Ph₃P=O	~95:5 E:Z	Aldehydes, ketones (slow)
Schlosser-Wittig	Ph₃P=CHR + PhLi/HCl	~22	n-BuLi then PhLi	Ph₃P=O	>98:2 E:Z	Aldehydes, salt-sensitive
Horner-Wadsworth-Emmons	(EtO)₂P(O)-CH₂-EWG	~14	NaH, K₂CO₃, DBU	(EtO)₂P(O)O⁻ Na⁺ (water-soluble)	>95:5 E:Z	Aldehydes, ketones; needs EWG
Still-Gennari (Z-HWE)	(CF₃CH₂O)₂P(O)-CH₂-EWG	~12	KHMDS, 18-crown-6	Phosphonate	~95:5 Z:E	Aldehydes; Z-selective HWE
Julia-Kocienski	BT-SO₂-CH₂R (benzothiazole sulfone)	~22	KHMDS, NaHMDS	BT-SO₂Na, SO₂	>95:5 E:Z	Aldehydes; non-stabilized E
Tebbe / Petasis	Cp₂Ti=CH₂	n/a (carbene)	n/a	Cp₂Ti(O)	n/a (only =CH₂)	Esters, amides, ketones

Famous syntheses using the Wittig

• BASF Vitamin A (1971-present). The C15 phosphonium salt (β-ionylideneethyltriphenylphosphonium chloride) couples with the C5 aldehyde glyceraldehyde acetal under NaOMe in MeOH to give Vitamin A acetate after hydrolysis. The Wittig step runs at 50 °C, completes in 4 h, and is run on >1000 t/yr scale at BASF's Ludwigshafen site. The all-trans selectivity is critical for biological activity.
• Corey prostaglandin syntheses (1969 onwards). E. J. Corey's PGF₂α total synthesis uses a Wittig olefination of the Corey lactone aldehyde with Ph₃P=CHCO₂H-bearing side chain to install the cis-Δ⁵,⁶ double bond. Z-selectivity from the non-stabilized ylide is essential because the natural prostaglandin geometry is cis. The Corey synthesis became the standard route for industrial PGF₂α (Lutalyse for veterinary use, Latanoprost for glaucoma).
• Nicolaou Taxol C/D-ring (1994). A Wittig coupling at C-12 installs the exocyclic methylene that becomes part of the oxetane ring. The reaction was carried out on a complex polycyclic intermediate at -78 °C with no epimerization at the eight nearby stereocenters.
• Roche β-carotene (industrial). Two C20 phosphonium salts react with a C2 dialdehyde (glyoxal) under base to give all-trans β-carotene in a single double-Wittig step. Roche (now DSM) ran this at multi-100-tonne scale for animal feed and food-coloring applications.
• Stork prostaglandin / Lasonolide work. Gilbert Stork's enantioselective prostaglandin syntheses (1976) and his lasonolide A total synthesis (2002) both deploy intramolecular Wittig macrocyclizations to close 14- and 18-membered rings. The reaction tolerates the pre-existing tetrahydropyran and tetrahydrofuran rings without scrambling stereochemistry.
• Wittig's original 1954 paper. Wittig reacted methylenetriphenylphosphorane with benzophenone in ether at room temperature and obtained 1,1-diphenylethylene plus triphenylphosphine oxide in 84% yield. The paper concluded with a one-page table of similar examples and noted the lack of side products — the cleanness of the reaction is what convinced the community within five years.