The [2,3]-Wittig Rearrangement

What the reaction does

Start with an ordinary allyl ether, but one where the carbon on the other side of the oxygen carries a removable proton — an allylic ether of the type R-O-CH₂-CH=CH₂ where R also bears an α-C-H. Rip that α-proton off with a strong base and you get a carbanion sitting right next to the ether oxygen. That carbanion is unhappy: it is a high-energy anion parked beside an electron-rich oxygen. It solves its problem in one clean, concerted move.

The carbanion swings around and bonds to the far carbon of the allyl group. As that new carbon-carbon bond forms, the original carbon-oxygen bond breaks, and the electron pair that used to hold the allyl group to oxygen collapses onto the oxygen as an alkoxide. The allyl double bond shifts one position over (allylic transposition). After aqueous workup you have a homoallylic alcohol — an alcohol with a C=C two carbons away — that has a brand-new C-C bond and, very often, one or two new stereocenters set with high selectivity.

In one sentence: a base-generated carbanion converts an allyl ether into a homoallylic alcohol by a concerted [2,3]-sigmatropic shift, trading a C-O bond for a C-C bond. The reaction is prized because it forms C-C bonds and quaternary/tertiary carbons under mild, low-temperature, thermally quiet conditions with excellent stereocontrol.

The step-by-step mechanism

The whole event is two operations: make the carbanion, then let it rearrange.

1. Deprotonation (or transmetalation) to form the carbanion. A strong, non-nucleophilic base removes the proton α to oxygen. The α-C-H of a simple ether has a pKa around 40-45, so you need n-BuLi or s-BuLi (and often an α-stabilizing group like phenyl, vinyl, or a carbonyl). In the Still-Wittig variant you instead install an α-tributylstannyl group and do a fast tin-lithium exchange with n-BuLi at -78 °C, which cleanly delivers the exact carbanion you want. Either way the product is a lithiated carbanion α to the ether oxygen.
2. The [2,3]-sigmatropic shift. The carbanion's lone pair (the HOMO) overlaps with the terminal (C-3) carbon of the allyl group. Three things happen in one concerted, five-atom cyclic transition state: (a) the carbanion carbon forms a bond to allyl C-3, (b) the allyl π bond migrates from C2=C3 to C1=C2, and (c) the O-C1 σ bond breaks heterolytically, dumping the electron pair onto oxygen. Because it is a [2,3]-shift, you number outward from the two ends of the breaking O-C1 σ bond: on the alkoxide side you count 2 atoms (O and the carbanion carbon Cα), and on the allyl side you count 3 atoms (C1-C2-C3, forming the new bond at C3). Those 2 + 3 atoms close into a five-membered envelope transition state.
3. Protonation. The resulting lithium alkoxide is quenched on aqueous workup to give the neutral homoallylic alcohol.

    Allyl ether           Carbanion             [2,3]-TS            Homoallylic
                                             (5-membered)           alkoxide

    R-CH₂-O-CH₂          R-CH⁻-O-CH₂           R-CH···CH₂           R-CH-CH₂-CH=CH₂
            \                    \              :  ⌒  ‖                    |
             CH=CH₂    ──base──►  CH=CH₂  ───►  O   CH   ───────►         O⁻  (→ OH)
                                              (C1) (C3)
              1    2   3

  step 1:  base pulls the α-C-H  → carbanion α to O
  step 2:  carbanion C bonds to allyl C-3; O-C1 breaks; π shifts C2=C3 → C1=C2
  step 3:  workup protonates the alkoxide → homoallylic alcohol

The key electron-arrow logic uses three curved arrows: one from the carbanion lone pair to allyl C-3, one from the C2=C3 π bond to C1-C2, and one from the O-C1 σ bond onto oxygen. All three arrows move electron pairs around the cyclic array — the lone pair, the migrating π bond, and the breaking σ bond — so this is a six-electron, suprafacial-suprafacial process through a Hückel-aromatic (4n+2, n=1) transition state. The Woodward-Hoffmann rules classify it as a thermally allowed [2,3]-sigmatropic shift, which is why it runs cleanly and concertedly at low temperature while the stepwise [1,2] radical alternative needs more heat.

Reagents, base, and conditions

• Base. The proton is only weakly acidic, so the base must be strong and non-nucleophilic. Standard choices: n-butyllithium or sec-butyllithium (1.0-1.1 equiv) in THF; LDA or LiHMDS for ester/carbonyl-stabilized substrates; KH for potassium-templated variants. For the Still-Wittig, n-BuLi does a tin-lithium exchange on an α-stannyl ether rather than a deprotonation.
• Temperature. Very low — typically -85 °C to -60 °C (a dry-ice/acetone or liquid-N₂/pentane bath). Cold temperatures suppress the competing [1,2]-Wittig radical pathway and prevent the carbanion from decomposing.
• Solvent. THF is standard; THF/HMPA or added TMEDA can be used to modulate aggregation state and selectivity. Ethers only — never protic solvents, which would quench the carbanion instantly.
• α-Stabilizing group. A phenyl, vinyl, alkynyl, or carbanion-stabilizing group α to oxygen lowers the pKa and makes deprotonation feasible with LDA. Non-stabilized carbanions generally require the Still-Wittig tin route.
• Workup. Quench with saturated NH₄Cl(aq) or dilute acid to protonate the alkoxide, then extract and purify.

A representative recipe: dissolve the α-stannyl allyl ether in dry THF, cool to -78 °C, add 1.05 equiv n-BuLi dropwise, stir 10-30 min (the tin-lithium exchange and rearrangement are fast at this temperature), then quench with NH₄Cl. Yields of 70-95% are typical for well-designed substrates.

Scope, selectivity, and stereochemistry

The [2,3]-Wittig is valued far more for its stereocontrol than for the raw bond formation. Three levels of selectivity fall out of the folded five-membered transition state:

• Alkene geometry (E/Z). The new double bond forms from the internal allyl carbons. The reaction shows a strong general preference for the (E)-alkene, because the envelope transition state places the internal substituent pseudo-equatorial (exo) to avoid allylic A^1,3 strain. Selectivities of 90:10 or better are common.
• Diastereoselectivity (syn/anti). When both fragments carry substituents, the reaction sets two stereocenters at once. Their syn/anti relationship is coupled to the alkene geometry through the same transition state: (E)-ethers channel to one diastereomer, (Z)-ethers to the other. This "geometry-locks-diastereomer" behavior is the reaction's signature.
• Enantioselectivity. Because the shift is suprafacial and intramolecular, chirality transfers — a defined stereocenter in the starting material dictates the configuration of the new one. Chiral-auxiliary and asymmetric-base (chiral lithium amide) versions give the homoallylic alcohol with high ee.

Scope: allyl, crotyl, prenyl, propargyl (giving allenes), and benzyl-vinyl ethers all work. The α-carbon can bear stabilizing groups (aryl, vinyl, carbonyl, cyano) or, via the Still route, be non-stabilized. Ring-contraction and ring-expansion variants let you use the rearrangement to forge medium-ring and macrocyclic systems.

How it compares to related sigmatropic shifts

	[2,3]-Wittig	[1,2]-Wittig	Claisen ([3,3])	Anionic oxy-Cope ([3,3])
Order of shift	[2,3]-sigmatropic	[1,2]-shift	[3,3]-sigmatropic	[3,3]-sigmatropic
Mechanism	Concerted, suprafacial	Stepwise radical pair	Concerted, chair-like	Concerted, charge-accelerated
Charge	Anionic (carbanion)	Anionic (carbanion)	Neutral	Anionic (alkoxide)
Bonds traded	Break C-O, form C-C	Break C-O, form C-C	Break C-O, form C-C	Break C-C, form C-C
Needs base?	Yes (strong)	Yes (strong)	No	Yes (make alkoxide)
Temperature	-85 to -60 °C	Higher (competes)	150-250 °C thermal	0-25 °C (huge rate boost)
Substrate	Allyl ether + α-C-H	Any ether + α-C-H	Allyl vinyl ether	1,5-hexadien-3-ol
Stereocontrol	Excellent (linked E/Z + dr)	Poor (racemizes)	Excellent (chair)	Good
Product	Homoallylic alcohol	Alcohol (scrambled)	γ,δ-unsat. carbonyl	δ,ε-unsat. carbonyl

Worked example: crotyl benzyl ether to a homoallylic alcohol

Take (E)-crotyl benzyl ether, PhCH₂-O-CH₂-CH=CH-CH₃. The benzylic position is the acidic site (a phenyl group stabilizes the α-carbanion), and the crotyl group is the migrating allyl unit.

    Ph-CH₂-O-CH₂-CH=CH-CH₃   ──n-BuLi, THF, -78 °C──►   Ph-CH(OH)-CH(CH₃)-CH=CH₂

    (E)-crotyl benzyl ether                              1-phenyl-2-methyl-but-3-en-1-ol
                                                         (homoallylic alcohol, new C-C bond,
                                                          two new stereocenters; the new alkene
                                                          is a terminal vinyl, and the (E)-crotyl
                                                          geometry sets the syn/anti diastereomer)

• Deprotonation. n-BuLi (1.05 equiv) removes the benzylic proton at -78 °C, giving a phenyl-stabilized carbanion α to oxygen.
• [2,3]-shift. The benzylic carbanion bonds to the crotyl terminal carbon (C-3, the =CH-CH₃ end becomes the new methyl-bearing tertiary stereocenter), the O-CH₂ bond breaks, and the double bond migrates to become a terminal vinyl group.
• Product. After NH₄Cl workup: 1-phenyl-2-methylbut-3-en-1-ol — a homoallylic alcohol with two adjacent stereocenters whose syn/anti relationship is set by the (E)-geometry of the crotyl group. Typical yields 75-90%, dr often > 85:15.

The same substrate at higher temperature, or without an allyl group, would drift into the [1,2]-Wittig radical pathway and give scrambled, racemized products — which is exactly why chemists keep it cold and keep an allyl group in place.

A real named application: total synthesis

The [2,3]-Wittig — especially the Still-Wittig tin variant — is a standard chirality-transfer tool in complex-molecule synthesis:

• Macrocyclic ring formation. Still used ring-forming [2,3]-Wittig rearrangements of cyclic allylic ethers to make medium and large carbocyclic rings with defined stereochemistry — a strategy applied to macrolide and cembranoid frameworks. The intramolecular version stitches a ring while transposing the alkene in one step.
• Polypropionate and polyketide fragments. The linked E/Z and syn/anti control makes the reaction a clean way to install the alternating methyl/hydroxyl stereocenters of polypropionate natural products (portions of erythromycin-type and other macrolide chains).
• Nakai and Marshall's work. Takeshi Nakai and James Marshall developed the acyclic diastereo- and enantioselective versions in the 1980s that turned the reaction from a curiosity into a predictable stereocontrolled C-C bond-forming method used across natural-product synthesis.
• Allene synthesis. Propargylic ethers rearrange to give allenes — the [2,3]-shift across a triple bond delivers substituted allenic alcohols that are hard to make otherwise.

Limitations and side reactions

• Competing [1,2]-Wittig. The biggest rival. Homolysis of the C-O bond gives a radical pair that recombines with loss of stereochemistry. It is favored by higher temperature, stabilized radicals (benzylic on both sides), and the absence of a good allyl acceptor. Stay cold and use an allylic ether to keep the [2,3] path dominant.
• α-Elimination / carbene pathways. Some carbanions can undergo β-elimination or, if a leaving group is present, α-elimination to a carbene rather than rearranging.
• Proton-source sensitivity. Any protic impurity quenches the carbanion before it can rearrange, killing the yield. Rigorously dry, anhydrous, and often degassed conditions are mandatory.
• Weakly acidic substrates. If the α-position has no stabilizing group, direct deprotonation fails or is unselective; you must pre-install a tin (Still) or silicon handle for a metalation-triggered version.
• Retro-[2,3] and regiochemical scrambling. With certain symmetric or highly stabilized systems the shift is reversible or can pick the wrong terminus, eroding selectivity.

Historical discovery

Georg Wittig (yes, the same Wittig of the Wittig olefination and the 1979 Nobel Prize) and Löhmann first reported the base-induced rearrangement of ethers in 1942 (Justus Liebigs Ann. Chem. 550, 260), observing that α-metalated ethers rearranged to alcohols. That original chemistry was largely the messy [1,2] pathway. The concerted, stereodefined [2,3] variant was recognized and rationalized later, once the sigmatropic framework of R. B. Woodward and Roald Hoffmann (1965-1969) gave chemists the orbital-symmetry language to distinguish a concerted suprafacial [2,3]-shift from a radical [1,2]-shift.

The reaction's modern power dates from two developments in the late 1970s and 1980s: W. Clark Still's α-alkoxystannane / tin-lithium exchange protocol (the Still-Wittig rearrangement, 1978), which made non-stabilized carbanions accessible and selective, and the systematic stereochemical studies of Takeshi Nakai and James A. Marshall, who mapped the acyclic diastereo- and enantiocontrol that made the reaction a reliable synthetic tool. Today it sits alongside the Claisen and Cope rearrangements as one of the workhorse sigmatropic C-C bond-forming reactions.

Practical and safety notes

• Pyrophoric bases. n-BuLi and s-BuLi ignite on contact with air and react violently with water. Handle under inert atmosphere with proper cannula/syringe technique; titrate before use.
• Cryogenic baths. -78 °C dry-ice/acetone and colder liquid-N₂ slush baths demand cryo gloves and eye protection; acetone baths are flammable.
• Organotin toxicity. The Still-Wittig tributylstannyl reagents and the Bu₃SnH/Bu₃Sn byproducts are toxic and environmentally persistent; use closed handling and dispose of tin waste properly.
• HMPA hazard. HMPA, a common additive for tuning selectivity, is a suspected carcinogen — substitute DMPU where possible.