The McMurry Coupling

What the McMurry coupling does

Most reactions that build a carbon–carbon double bond start from at least one carbon that is not a carbonyl — the Wittig needs a phosphorus ylide, the Peterson needs a silyl carbanion, an elimination needs a leaving group. The McMurry is different and slightly outrageous: it takes two ordinary carbonyl groups — two ketones, two aldehydes, or one of each — and joins them directly at the carbonyl carbons, spitting out both oxygens as titanium dioxide.

    2  R₂C=O   ──[low-valent Ti]──→   R₂C=CR₂   +   "TiO₂"

    e.g.  2  Ph₂C=O  ──TiCl₃ / Zn──→  Ph₂C=CPh₂  (tetraphenylethylene)

The net transformation is a reductive dimerization with deoxygenation. Four electrons are delivered by the titanium in total — enough to reduce two carbonyls and cleave two C–O bonds. What makes it possible is titanium's ferocious oxophilicity: forming Ti–O bonds and ultimately the TiO₂ lattice releases so much energy that stripping two oxygens off carbon becomes strongly downhill.

The reaction has two guises. Intermolecular, it couples two separate carbonyl molecules — cleanest when they are identical, because two different ketones give a statistical mix. Intramolecular, when both carbonyls live in one chain, the titanium zips the ends together to close a ring. That ring-closing version, tolerant of rings from four-membered up to 20-plus-membered macrocycles, is what earned the reaction a permanent seat in total synthesis.

The mechanism, one electron at a time

The McMurry is a surface reaction — the chemistry happens on and around the freshly reduced titanium metal, not on a dissolved single molecule — but the electron-flow logic is clean:

1. Single-electron transfer (SET). The low-valent titanium surface donates one electron into the carbonyl's π* antibonding orbital. The C=O π bond becomes a one-electron bond: a ketyl radical anion, with the radical on carbon and the negative charge (as an alkoxide) on oxygen, the oxygen now bound to titanium.
2. Radical–radical dimerization. Two ketyl radicals, both pinned close together on the metal surface, couple carbon-to-carbon. Their two radical electrons pair into a brand-new C–C σ bond. The product is a 1,2-diolate (pinacolate) — the same skeleton you would get from a pinacol coupling — held on titanium through both oxygens.
3. Deoxygenation. This is the step that makes it a McMurry rather than a pinacol coupling. Both C–O bonds break heterolytically as the electrons flow onto oxygen and into titanium. The two carbons, now each short one bond, form the second bond of a C=C double bond. The oxygens depart on the metal, ultimately as TiO₂.

  SET:       R₂C=O  +  e⁻(Ti)   →   R₂C•–O⁻(Ti)        [ketyl radical]

  couple:    2 × R₂C•–O⁻(Ti)    →   (Ti)O–CR₂–CR₂–O(Ti) [pinacolate diolate]

  deoxy:     (Ti)O–CR₂–CR₂–O(Ti) → R₂C=CR₂  +  2 "Ti–O"  →  TiO₂

Stop after step 2 and warm gently, and you can isolate the 1,2-diol (pinacol) — the McMurry and the pinacol coupling share the first half of the pathway and diverge only at deoxygenation. At the higher temperatures and with the more reducing titanium of a true McMurry, the diolate never survives: it collapses straight to the alkene. The temperature and the strength of the titanium are the dials that decide whether you stop at the diol or push through to the olefin.

Reagents, recipes, and real conditions

You cannot buy "low-valent titanium." You make it fresh by reducing a titanium(III) or titanium(IV) chloride with a stronger reductant, in an oxygen- and water-free flask, and the exact recipe changes the reactivity. The common ones:

Recipe (Ti source + reductant)	Notes
TiCl₃ + Zn–Cu couple	McMurry's 1978 improvement over the original LiAlH₄ recipe; mild, workhorse for diaryl ketones.
TiCl₃ + K (or Li, Na)	Very reducing, effective for hindered ketones; alkali metals are hazardous.
TiCl₃ + LiAlH₄	McMurry–Fleming, powerful but can over-reduce sensitive groups.
TiCl₄ + Zn	Mukaiyama's variant; convenient, generates Ti in situ from Ti(IV).
TiCl₃(DME)₁.₅ + Zn(Cu)	Fürstner–Bogdanović "instant" reagent; reproducible, scalable, macrocyclization-friendly.

• Solvent. Anhydrous THF or DME (1,2-dimethoxyethane), refluxed under argon or nitrogen. Ethers coordinate and disperse the titanium without quenching it.
• Temperature. Typically reflux, roughly 65–85 °C, for a few hours; the diolate must be pushed over the deoxygenation barrier.
• Stoichiometry. Titanium is used in large excess — often 2–4 equivalents of TiCl₃ per carbonyl — because it is a sacrificial reductant and oxygen sink, not a catalyst. It ends up as TiO₂; there is no turnover.
• Dilution matters for rings. Intramolecular macrocyclizations are run under high dilution (slow addition into a large volume of hot titanium slurry) so each molecule finds its own two ends rather than a neighbor's.
• Workup. Quench cautiously — the titanium slurry is pyrophoric-adjacent — filter off the black/gray titanium oxides, and purify the hydrocarbon product.

Scope, selectivity, and stereochemistry

The McMurry's sweet spot is exactly where other olefination methods struggle:

• Tetrasubstituted alkenes. Coupling two diaryl or dialkyl ketones gives fully substituted C=C bonds — tetraphenylethylene from benzophenone is the textbook case. Wittig chemistry chokes on tetrasubstituted alkenes because the ylide and carbonyl are both crowded; the McMurry does not care.
• Symmetric alkenes. When you couple two identical carbonyls, there is no E/Z ambiguity to worry about, and no cross-coupling mixture — the ideal use case.
• Rings, including strained and large ones. Intramolecular coupling closes carbocycles from four-membered up through macrocyclic. It is one of the few reliable ways to make certain cyclic and bridgehead alkenes.

The counterweight is stereochemistry. Because the alkene forms from a pinacolate lying on a metal surface, the reaction gives poor E/Z selectivity for unsymmetrical acyclic alkenes — frequently a near-statistical mixture. If you need a single clean geometry of a simple disubstituted alkene, reach for a Wittig, Horner–Wadsworth–Emmons, or Julia olefination instead. The McMurry earns its keep when geometry is either irrelevant (symmetric or ring-fixed) or when nothing else will build that particular crowded or macrocyclic olefin.

McMurry vs Wittig vs pinacol coupling

	McMurry coupling	Wittig reaction	Pinacol coupling
Starting carbons	Two carbonyls	One carbonyl + one ylide	Two carbonyls
Product	Alkene R₂C=CR₂	Alkene R₂C=CR'₂	1,2-diol (pinacol)
Reagent / driver	Low-valent Ti (oxophilic)	Ph₃P=CR'₂ ylide	1 e⁻ reductant (Mg, SmI₂, Ti)
Key intermediate	Ketyl → pinacolate	Oxaphosphetane	Ketyl → diolate (stops here)
Oxygen fate	Removed as TiO₂	Removed as Ph₃P=O	Retained (both O's stay)
E/Z control	Poor (surface reaction)	Good (ylide type sets E or Z)	N/A (no alkene)
Tetrasubstituted alkenes	Excellent	Difficult	N/A
Ring closure / macrocycles	Excellent (high dilution)	Moderate	Good for cyclic diols
Cross-coupling two different C=O	Statistical mixture (~1:2:1)	Fully controlled	Statistical mixture

Named application: the tamoxifen skeleton

The commercial and pedagogical poster child for the McMurry is tamoxifen, the selective estrogen-receptor modulator used to treat and prevent hormone-responsive breast cancer. Tamoxifen is a triarylethylene — three aryl groups and an ethyl substituent hung around a single tetrasubstituted C=C bond. That is precisely the kind of crowded, fully substituted alkene the Wittig hates and the McMurry loves.

   Ar–C(=O)–Et  +  Ar'–C(=O)–Ar''   ──low-valent Ti──→   Ar(Et)C=C(Ar')(Ar'')
   (a propiophenone)   (a benzophenone)                    (tamoxifen core, as E/Z mix)

A McMurry cross-coupling between a propiophenone derivative and a suitably substituted benzophenone builds the tetrasubstituted olefin core in a single step. Because the coupling is unsymmetrical, it delivers a mixture of the desired Z-tamoxifen and its inactive E-isomer that must then be separated or equilibrated — a real-world illustration of the reaction's stereochemical Achilles' heel. The tradeoff is worth it: no other one-step method assembles that specific crowded triaryl alkene so directly. The same strategy builds related triarylethylene drugs and countless dyes and fluorophores whose color comes from an extended, fully substituted olefin.

In total synthesis, the intramolecular McMurry has closed rings in targets such as the sarpagine and taxane frameworks and in porphyrin-like macrocycles, where two ketones tethered across a long chain are welded shut to form a ring-embedded double bond that no other olefination reaches cleanly.

Limitations and side reactions

• Poor cross-coupling. Two different ketones give A=A, A=B, and B=B in roughly 1:2:1. The mixed product tops out near 50% unless you bias with a large excess of one partner, exploit differing reactivity, or tether the two carbonyls together.
• Weak E/Z control. Unsymmetrical acyclic alkenes emerge as isomer mixtures — sometimes tunable by titanium recipe and temperature, but rarely to a single geometry.
• Pinacol as the arrested product. If the titanium is too mild or the temperature too low, the reaction stalls at the 1,2-diol. Under-reduced systems, or substrates that resist deoxygenation, give pinacol instead of alkene.
• Over-reduction and functional-group intolerance. The reagent is a powerful single-electron reductant. Nitro groups, some halides, epoxides, and easily reduced heteroaromatics can be attacked. Esters and amides usually survive, but sensitive reducible groups do not.
• Chemoselectivity between aldehyde and ketone. Aldehydes couple faster than ketones; in a molecule bearing both, the aldehyde tends to react first, complicating selective couplings.
• Reagent handling. The titanium slurries are air- and moisture-sensitive, sometimes pyrophoric, and the alkali-metal recipes are outright hazardous. Reproducibility historically varied batch to batch — a problem the Fürstner–Bogdanović "instant" reagent was designed to solve.

Who discovered it, and when

The reaction is named for John E. McMurry, who reported the titanium-induced carbonyl coupling with his student M. P. Fleming in J. Am. Chem. Soc. in 1974, using TiCl₃ reduced by LiAlH₄. The idea was very much in the air at the time: Toru Mukaiyama (TiCl₄/Zn) and Stanisław Tyrlik independently reported closely related titanium-based couplings in the same period, around 1973–1974. McMurry's name stuck because he systematically developed the reaction, mapped its scope over the following years, and made it a general synthetic tool.

A crucial later contribution came from Alois Fürstner and Borislav Bogdanović, who in the 1990s developed the graphite- and DME-supported "instant" low-valent titanium reagents that turned a temperamental, batch-variable slurry into a reproducible, scalable reagent — the version most often used in modern process and total-synthesis work.

Practical and safety notes

• Inert atmosphere is non-negotiable. Low-valent titanium reacts violently with air and water. Prepare and use it under argon or nitrogen, with dried solvents and glassware.
• Pyrophoric residues. The spent titanium filter cake can ignite in air while still wet with solvent. Quench and dispose of it deliberately; never let it dry in the open.
• Alkali-metal recipes demand caution. TiCl₃/K and TiCl₃/Li reductions involve reactive metals that react with water to release hydrogen and heat. Where scale or safety matters, the Zn(Cu) or Fürstner–Bogdanović recipes are far more forgiving.
• Not a catalytic reaction. Because titanium is consumed as the oxygen sink, the reaction is stoichiometric-to-excess in titanium and produces molar quantities of titanium oxide waste — a genuine drawback for large-scale green-chemistry metrics, and a reason catalytic olefination alternatives are pursued for bulk production.
• Choose the right tool. If you need one clean E- or Z-alkene of a simple substitution pattern, a Wittig or HWE is the better call. Deploy the McMurry when the target is symmetric, tetrasubstituted, or a ring that only a carbonyl-to-carbonyl weld can close.