The Chauvin Mechanism: The Metallacyclobutane at the Heart of Olefin Metathesis

The four-membered ring nobody could see

Olefin metathesis exchanges the alkylidene fragments of two alkenes: R¹CH=CHR¹ + R²CH=CHR² ⇌ 2 R¹CH=CHR². Chemists at DuPont, Phillips, and Goodyear had studied the reaction since the late 1950s and run it industrially from the mid-1960s — the Phillips Triolefin Process converted propene to ethylene plus but-2-ene over WO₃/SiO₂ — long before anyone knew how the C=C bonds were being cut and re-stitched. The central puzzle was how a metal center could break both double bonds of a partner pair and rejoin the pieces in a scrambled order.

The intuitive guess, championed by Bradshaw, Calderon, and others, was pairwise: two alkenes bind side-by-side on the metal, and a transient quasi-cyclobutane forms in which both new C–C bonds are made at once, then cleaved on the perpendicular diagonal. It is a satisfying, symmetric picture — and it is wrong. In 1971 Chauvin and Hérisson (Makromol. Chem. 141, 161) proposed instead that the true chain carrier is a metal carbene, M=CHR, which reacts with one olefin at a time through a metallacyclobutane: a four-membered ring containing the metal and three carbons.

The genius of the proposal was that it demoted the cyclobutane-like species from an all-carbon ring (which would have to fragment on a specific diagonal) to a metallacycle whose two possible cleavage modes are chemically distinct: retro-[2+2] can either regenerate the starting carbene and olefin (nonproductive) or release a new carbene and a new olefin (productive). Metathesis is thus a chain of [2+2] cycloadditions and retro-[2+2] cycloreversions, with the alkylidene shuttling between partners like a relay baton.

Anatomy of the catalytic cycle

The Chauvin cycle for a generic metal alkylidene [M]=CHR runs in four repeating events:

• Olefin coordination: an alkene R′CH=CH₂ binds to the metal, typically after a labile ligand dissociates. In the Grubbs system this is the rate-limiting loss of a phosphine from (PCy₃)₂Cl₂Ru=CHPh to open a 14-electron site.
• [2+2] cycloaddition: the coordinated olefin and the M=C bond couple to form the metallacyclobutane. In frontier-orbital terms the metal-carbene π and π* interact with the olefin π and π*; the reaction is a symmetry-allowed, metal-assisted [2+2] process (the metal–carbene π/π* engaging the olefin π/π*), sidestepping the Woodward–Hoffmann prohibition on all-carbon thermal [2+2].
• Retro-[2+2] (cycloreversion): the metallacycle cleaves along the opposite diagonal, generating a new alkylidene [M]=CHR′ and expelling a new olefin RCH=CH₂.
• Turnover: the new carbene enters the next cycle. The alkylidene identity therefore changes each productive turn — this exchange of the =CHR unit is the very definition of metathesis.

Two features deserve emphasis. First, the metal's oxidation state and electron count do not change across the cycle: [2+2] and retro-[2+2] are redox-neutral, unlike oxidative addition or migratory insertion. A Ru(II) d⁶ Grubbs center stays Ru(II) throughout; a Schrock Mo(VI) d⁰ imido-alkylidene stays Mo(VI) d⁰. Second, whether a given metallacyclobutane is productive or nonproductive depends only on which diagonal cleaves. The nonproductive mode simply spits back the reactants, so most metallacyclobutane formation events do no net chemistry — a fact that dominates the observed kinetics.

The regiochemistry of ring formation sets the outcome. A monosubstituted metallacyclobutane can place its substituents α,α (1,2) or α,β (1,3) to the metal; steric preference for the less-crowded, all-substituents-away arrangement is what gives modern catalysts their E-selectivity and, in the Z-selective ruthenium variants, their Z-selectivity enforced by a cyclometalated N-adamantyl NHC chelate (a Ru–C bond formed by C–H activation at the adamantyl group).

The experiments that killed the pairwise model

Chauvin's hypothesis would have remained one idea among several without decisive kinetic tests, and the crossover experiments of 1975 supplied them. The cleanest is the double cross-metathesis: react a fully protiated symmetric olefin (1,7-octadiene, d₀) with the 1,1,8,8-tetradeutero analog (d₄, terminal =CD₂ groups) and analyze the ethylene released by mass spectrometry. The pairwise mechanism, which pairs two alkenes at once, predicts that at low conversion the ethylene should be overwhelmingly d₀ and d₄ (from same-labeled pairs), with the mixed d₂ product appearing only later as scrambling propagates.

The Chauvin mechanism predicts the opposite: because a single carbene attacks one olefin at a time, the very first productive turnovers already generate the statistical 1:2:1 ratio of d₀ : d₂ : d₄ ethylene, invariant with conversion. Experiments by Grubbs (using ¹³C and ²H labels, 1975–76), by Katz, and by Casey found the statistical distribution from the earliest measurable conversion. The pairwise model was falsified; the metallacyclobutane won.

The structural clincher came from isolating the intermediates the mechanism demanded. Schrock prepared stable, X-ray-characterized high-oxidation-state alkylidenes such as Ta(=CHtBu)(CH₂tBu)₃ and later the workhorse Mo(=CHCMe₂Ph)(=NAr)(OR)₂ imido catalysts. Tebbe's reagent Cp₂TiCH₂·ClAlMe₂ delivered a titanium methylidene equivalent, and Grubbs isolated and crystallographically characterized titanacyclobutanes (Cp₂Ti–CH₂–CHR–CH₂) that undergo clean retro-[2+2]. A four-membered ring proposed on paper in 1971 was, within a decade, sitting on a diffractometer.

A worked example: statistics, thermodynamics, and turnover

Consider the textbook demonstration Chauvin himself invoked: cross-metathesis of but-2-ene (CH₃CH=CHCH₃) with hex-3-ene (CH₃CH₂CH=CHCH₂CH₃). The methylidene-swap logic predicts one cross product, pent-2-ene (CH₃CH=CHCH₂CH₃), formed by exchanging an ethylidene (=CHCH₃) for a propylidene (=CHCH₂CH₃). At thermodynamic equilibrium the alkylidene fragments distribute binomially: with an equal starting mix, the =CHCH₃ and =CHCH₂CH₃ carbenes are equiprobable, so the alkene population settles to but-2-ene : pent-2-ene : hex-3-ene ≈ 1 : 2 : 1. That 1:2:1 signature is the fingerprint of a single carbene relay and is impossible to reconcile with a pairwise diagonal cut.

The thermodynamics are shallow but real. For simple internal alkenes, ΔH° of metathesis is near zero (a C=C is traded for a C=C), so the reaction is essentially entropy-controlled. In ring-opening metathesis polymerization (ROMP) of a strained monomer like norbornene, the ~27 kcal/mol of ring strain provides the enthalpic driving force. In ring-closing metathesis (RCM) and cross-metathesis, the trick is to make the reaction irreversible by venting the ethylene byproduct — Le Chatelier's principle applied to a gas — pulling the equilibrium toward the desired macrocycle or acyclic cross product.

Kinetically, modern catalysts are extraordinary. Grubbs second-generation catalyst (bearing an N-heterocyclic carbene, H₂IMes) reaches turnover numbers in the thousands to tens of thousands; the rate-determining step is phosphine dissociation (k₁ ~ several s⁻¹ at 25 °C for Grubbs-I; ~9.6 s⁻¹ at 35 °C), and the NHC's strong σ-donation accelerates the subsequent olefin binding and metallacycle formation. The Chauvin cycle explains why ligand dissociation gates the rate: you cannot form the four-membered ring until the olefin has a coordination site.

Limits, subtleties, and honest caveats

The Chauvin mechanism is settled for the productive chemistry, but several finer points remain live or nuanced:

• Metallacyclobutane: intermediate or transition state? For d⁰ Schrock systems the metallacyclobutane is a genuine, sometimes isolable, energy minimum. For d⁶ Ru systems, computation (Cavallo, Eisenstein, Truhlar) indicates it is a shallow minimum flanked by the true rate-limiting olefin-coordination and cycloreversion barriers; its lifetime can be very short. Calling it 'the intermediate' is correct but its depth is metal-dependent.
• Bottom-bound vs side-bound olefin. In Ru catalysts the olefin can approach trans to the NHC (bottom-bound) or cis (side-bound); DFT favors the bottom-bound pathway, and this geometry — not the pairwise question — governs stereoselectivity. This was not part of Chauvin's original picture and is still refined by computation.
• Nonproductive events dominate. Because retro-[2+2] often just regenerates the reactants, most metallacyclobutanes are catalytically 'silent.' Degenerate metathesis is real and measurable, but it does no net work — a subtlety that complicates extracting elementary rate constants from bulk kinetics.
• Catalyst decomposition. The methylidene [Ru]=CH₂ is the least stable carbene in the cycle and a major decomposition channel (via bimolecular coupling to ethylene, or β-hydride pathways). Catalyst death, not the mechanism, usually caps turnover in practice.

It is also worth stating what the mechanism is not: it is not an oxidative-addition/reductive-elimination cycle, and it is not a radical chain. The [2+2]/retro-[2+2] steps are pericyclic-like, concerted, and redox-neutral at the metal — which is precisely why the alkylidene, and not the metal's oxidation state, is the information-carrying species.

Why it mattered: from a scrambled propene stream to a Nobel Prize

Getting the mechanism right converted metathesis from an industrial curiosity into a design tool. Once chemists accepted that a well-defined metal carbene is the active species, the search shifted from ill-defined heterogeneous mixtures (WCl₆/EtAlCl₂ 'black boxes') to single-component, functional-group-tolerant catalysts you could weigh into a flask. Schrock's Mo/W imido-alkylidene alkoxides delivered high activity and, with chiral alkoxides, enantioselective metathesis. Grubbs's ruthenium benzylidenes — Grubbs-I (PCy₃)₂Cl₂Ru=CHPh and the NHC-bearing Grubbs-II, plus the Hoveyda–Grubbs chelated variants — tolerate alcohols, water, and air well enough to be used on the bench by non-specialists.

The payoff spans the sciences. Ring-closing metathesis builds macrocycles that were previously nightmarish — the C–C ring-forming step in Janssen's (Medivir-discovered) HCV protease inhibitor simeprevir is an RCM run on multi-hundred-kilogram scale. ROMP gives polymers with exact molecular weights and living-polymerization control; cross-metathesis stitches together fragments in total synthesis; and metathesis of plant-oil feedstocks (Elevance) turns fatty esters into specialty chemicals, a green-chemistry win because the byproduct is ethylene or a light olefin rather than salt waste.

The 2005 Nobel Prize in Chemistry to Yves Chauvin, Robert H. Grubbs, and Richard R. Schrock honored exactly this arc: Chauvin for the mechanistic insight that a metallacyclobutane and a shuttling carbene explain the whole reaction, and Grubbs and Schrock for turning that insight into catalysts that reshaped synthetic chemistry. The Nobel committee called metathesis a 'change-your-partners dance' — an apt image for a four-membered ring that forms, cleaves, and hands off its carbene to the next olefin in line.

Chauvin's carbene chain mechanism versus the pairwise (concerted) mechanism it displaced.

Feature	Pairwise mechanism	Chauvin (carbene) mechanism
Chain carrier	Two coordinated olefins on one metal ('quasi-cyclobutane')	Single metal alkylidene M=CHR
C–C bonds exchanged	Both new bonds form simultaneously between two olefins	One olefin at a time via metallacyclobutane
Early product distribution	Predicts a large excess of symmetric 'primary' products	Predicts the statistical (binomial) mixture from the start
¹³C double-labeling test	Predicts non-statistical crossover in early conversion	Predicts statistical scrambling — matches Grubbs (1975) & Katz experiments
Isolable intermediate	None ever characterized	Tebbe reagent, Schrock alkylidenes, and metallacyclobutanes all isolated & structurally proven