Benzyne: The Ring With a Triple Bond It Cannot Hold

Eliminate first, add second: the mechanism step by step

Drop chlorobenzene into potassium amide dissolved in liquid ammonia at its boiling point of −33 °C and you get aniline. That should not happen. Benzene rings do not do SN2 — there is no backside to attack — and there is no nitro group here to stabilise an anionic adduct.

• Step 1 — deprotonation. Amide (NH₂⁻, the conjugate base of a solvent with pK_a ≈ 38) removes a hydrogen from the carbon next to the chlorine. A benzene C–H is brutally weak as an acid, pK_a ≈ 43, but the halogen's inductive pull drains its neighbour and makes that one position measurably easier to strip.
• Step 2 — loss of halide. The resulting 2-chlorophenyl carbanion expels chloride from the adjacent carbon. Deprotonation then loss of leaving group from the adjacent atom is the textbook definition of an E1cb elimination — the carbanion is a real, discrete intermediate, not a bystander, though with the heavier halides the two steps blur toward something closer to a concerted E2.
• Step 3 — addition. The product, o-benzyne, is a ferociously electrophilic neutral molecule. Ammonia or amide adds across the strained bond, and a final protonation gives aniline.

Which step is slow is the subtle part, and the evidence says it is usually not step 1. Run the reaction in deuterated ammonia and the ortho positions of a halobenzene swap H for D far faster than the ring eliminates: the carbanion forms and reverts many times before anything leaves. That is unsurprising arithmetic — deprotonating a benzene C–H (pK_a ≈ 43) with amide (ammonia pK_a ≈ 38) is uphill by about five pK units, so only a trace of carbanion exists at any instant. With deprotonation as a pre-equilibrium, the observed rate is the product of two opposing trends, and they split cleanly: base-catalysed H/D exchange, a pure acidity measure with no elimination involved, runs F > Cl > Br > I, while the amination itself runs Br > I > Cl > F. Fluorobenzene is the best-acidified substrate and the slowest one, because C–F is punishingly hard to break and that second step is what sets the pace. Change the base and the balance moves: phenyllithium or LDA deprotonate essentially irreversibly, which is why fluoroarenes are the preferred aryne precursors there — and why Wittig reached for fluorobenzene.

The triple bond is a fraud: geometry, orbitals and 1846 cm⁻¹

Every textbook draws o-benzyne (1,2-didehydrobenzene, C₆H₄) with a triple bond, then immediately apologises for it. The aromatic 6π sextet above and below the ring is untouched — benzyne is still aromatic. The new bond forms in the plane of the ring, from the two sp² hybrid orbitals that used to hold the departed hydrogens.

Those lobes sit at internal ring angles of roughly 127°, where a genuine alkyne demands 180° — the linear geometry that leaves its two in-plane orbitals exactly parallel, side by side, for maximum overlap. Bending the frame to 127° tips each lobe by about 27°, so benzyne's pair miss being parallel by 53° and barely see each other. Bond length hides all of that: gas-phase microwave spectroscopy (Brown, Godfrey and Rodler, 1986; later refined on ¹³C isotopologues by Kukolich and co-workers) put the C1–C2 distance at ~1.24 Å, and with benzene at 1.39 Å and a real alkyne at 1.20 Å, that is about four-fifths of the way to a full triple bond. Distance is the one measurement that flatters benzyne; everything that probes bond strength says it is worth about half a normal one.

The vibration is the cleanest evidence. Trapped in solid argon at ~10 K, o-benzyne's C≡C stretch appears at 1846 cm⁻¹ (Radziszewski, Hess and Zahradník, 1992), against ~2150 cm⁻¹ for an ordinary alkyne (the C≡C band sits near 2100–2140 cm⁻¹ on a terminal alkyne and climbs to about 2190–2260 cm⁻¹ on an internal one). At 1846 cm⁻¹ the bond completes one full oscillation every ~18 fs: ν = c·ν̃ = (3.00 × 10¹⁰ cm/s)(1846 cm⁻¹) = 5.53 × 10¹³ Hz, and 1/ν = 1.8 × 10⁻¹⁴ s. Treat those two carbons as an isolated oscillator with reduced mass μ = 6 amu and the force constant comes out at k = μ(2πν)² ≈ 1.2 × 10³ N/m, about 12 mdyn/Å — above a C=C double bond (~9.6) and well below a true C≡C triple bond (~15.6).

Worked numbers: the second C–H costs 35 kcal/mol less than the first

You can price benzyne's extra bond straight out of thermochemistry.

• Benzene's gas-phase heat of formation is 19.8 kcal/mol; its C–H bond dissociation energy is 112.9 kcal/mol; ΔH_f(H•) = 52.1 kcal/mol.
• So ΔH_f(phenyl radical) = 19.8 + 112.9 − 52.1 = 80.6 kcal/mol.
• Negative-ion energetics give ΔH_f(o-benzyne) = 106.6 ± 3.0 kcal/mol (446 kJ/mol).
• Pulling the second hydrogen therefore costs 106.6 + 52.1 − 80.6 ≈ 78 kcal/mol.

The second C–H bond is about 35 kcal/mol weaker than the first, and that difference is exactly what the new in-plane bond pays back. Compare an ordinary alkyne π bond — roughly 65–75 kcal/mol, depending on whether you price it by rotation barriers or by the same stepwise-BDE cycle used here — and the verdict is quantitative: benzyne's "triple bond" delivers barely half the bonding of a normal one, because the orbitals are bent away from each other.

The other number that matters is the singlet–triplet gap of 37.5 ± 0.3 kcal/mol (157 kJ/mol), measured by Wenthold, Squires and Lineberger in 1998 from the photoelectron spectrum of the o-benzyne radical anion. A gap that large means the two electrons are firmly paired in one bonding orbital: o-benzyne is a closed-shell singlet, an ordinary (if very strained) molecule, not a diradical. Its isomers are a different story — m-benzyne (ΔH_f ≈ 122 kcal/mol, gap ~21 kcal/mol) is borderline, and p-benzyne (ΔH_f ≈ 138 kcal/mol, gap ~3.8 kcal/mol) really is a singlet biradical, which is precisely why the p-benzyne made by the Bergman cyclization can abstract hydrogen atoms from DNA.

Cine substitution and the 1953 experiment that settled it

o-Benzyne is symmetric, so the incoming nucleophile cannot tell the two carbons apart. Half the time the nitrogen lands on the carbon that used to carry the chlorine; half the time it lands on the carbon that used to carry the hydrogen. Substitution that shows up one position over from the leaving group is called cine substitution, and it is the fingerprint of an aryne.

In 1953 John D. Roberts, with Simmons, Carlsmith and Vaughan, proved it at MIT with a ¹⁴C label. He made chlorobenzene carrying ¹⁴C at the carbon bearing the chlorine, treated it with potassium amide in liquid ammonia, degraded the aniline and counted where the radioactivity sat. The label came back split ~50:50 between C1 and C2. No addition–elimination mechanism can move a label; only a symmetric intermediate can.

The corollary is the sharpest diagnostic in the field: block both ortho positions and the reaction dies. 2,6-dimethylchlorobenzene has no ortho hydrogen, so potassium amide leaves it alone. With an unsymmetrical aryne the split is no longer 50:50. A 3-methoxy substituent, for example, pushes the nucleophile to the carbon that puts the developing negative charge closest to the inductively withdrawing group, giving mainly the meta product. Garg and Houk's aryne distortion model (2010) turned that into a predictive rule: compute the two internal angles of the distorted aryne and the nucleophile attacks the carbon with the larger one.

Making it on purpose: from liquid ammonia to fluoride-triggered triflates

Georg Wittig proposed a dehydrobenzene intermediate in 1942 from fluorobenzene and phenyllithium, and in 1955 he and Pohmer nailed it by trapping benzyne with furan in a Diels–Alder reaction, isolating the bicyclic oxygen-bridged adduct. Furan remains the standard trap: benzyne is a voracious enough dienophile to pull a Diels–Alder out of dienes most alkynes cannot touch. Left alone, two benzynes dimerise to biphenylene, and three can trimerise to triphenylene.

The teaching-lab route diazotises anthranilic acid to benzenediazonium-2-carboxylate, which sheds N₂ and CO₂ on warming and collapses to benzyne, trapped by tetraphenylcyclopentadienone to give 1,2,3,4-tetraphenylnaphthalene plus CO — a beautiful experiment and a genuinely dangerous one.

Modern aryne chemistry runs on the Kobayashi precursor: 2-(trimethylsilyl)phenyl trifluoromethanesulfonate, introduced by Himeshima, Sonoda and Kobayashi in 1983. Fluoride (CsF, TBAF, or KF with 18-crown-6) attacks silicon, the resulting carbanion ejects triflate, and benzyne appears in acetonitrile at room temperature. No organometallic or amide base, no liquid ammonia, no −33 °C bath, and enough functional-group tolerance that esters, ketones and amines survive — which is why arynes became a synthetic workhorse. Hoye's hexadehydro-Diels–Alder reaction (Nature, 2012) needs no reagent at all: heating a tethered triyne cyclises it to an aryne with nothing left to remove. Industrially, benzyne is encountered rather than made — the old Dow phenol process, chlorobenzene with sodium hydroxide at about 350 °C and several hundred atmospheres, scrambles isotopic labels, which is to say it partly runs through an aryne.

How you prove something exists for a flicker

Nobody has a bottle of benzyne. Everything known about it comes from four families of measurement, and they agree.

• Chemical trapping. Generate benzyne with a diene present and isolate the cycloadduct; yield ratios against competing traps give relative rates.
• Isotopic labelling. Roberts's ¹⁴C experiment, plus the deuterium-exchange work: the ortho positions of halobenzenes exchange with the solvent faster than they eliminate, which proves the carbanion is real and reversibly formed. That is also why the reaction shows no large primary kinetic isotope effect — the rate is decided downstream, when C–X breaks.
• Matrix isolation. Photolyse a precursor such as benzocyclobutenedione or phthaloyl peroxide inside solid argon at ~10 K, where nothing can diffuse and nothing can dimerise, and record the infrared spectrum. That is where the 1846 cm⁻¹ band comes from. Gas-phase microwave spectroscopy later gave the full rotational constants and therefore the bond lengths.
• Molecular imprisonment. In 1997 Ralf Warmuth, working in Donald Cram's laboratory, generated o-benzyne inside a hemicarcerand — a molecular container with no opening big enough to escape through — and recorded its ¹H and ¹³C NMR spectra at low temperature. The aryne carbons resonate far downfield of ordinary alkyne carbons.

Thermochemistry comes from a fifth technique, both halves of it run on the o-benzyne radical anion made in a flowing-afterglow mass spectrometer: energy-resolved collision-induced dissociation gives the heat of formation, and the photoelectron spectrum of the same ion gives the singlet–triplet gap.

Hazards, failure modes and the look-alikes it gets confused with

Benzenediazonium-2-carboxylate is a friction-and-shock-sensitive explosive when dry. Dried samples have detonated on scraping, so it is generated in situ and kept wet with solvent, always. Old sodium or potassium amide is the other classic hazard: fresh amide is white or grey, but on exposure to air it oxidises to a yellow-to-brown mixture containing peroxide and nitride species that can explode on disturbance. Discoloured amide is destroyed, never used.

The commonest synthetic failure is regiochemical: run a strong-base amination on a substituted aryl halide expecting one product and an aryne hands you an often inseparable mixture of ipso and cine isomers. The fix is to remove the mechanism — palladium catalysis, blocked ortho positions, or a substrate genuinely activated for SNAr.

And that is the look-alike. Nucleophilic aromatic substitution (SNAr) is addition–elimination; the aryne route is elimination–addition — the same two operations in the opposite order. SNAr needs a powerful electron-withdrawing group ortho or para to the leaving group, goes through a negatively charged Meisenheimer adduct with an sp³ carbon, and is strictly ipso. The aryne route needs no activating group at all, goes through a neutral strained molecule, and scrambles the position. The halogen ordering trips people the other way: SNAr runs F ≫ Cl ≈ Br > I, because the polarised C–F bond makes the ipso carbon electrophilic and stabilises the anionic adduct, whereas aryne amination runs Br > I > Cl > F. The discriminators that actually separate them are whether an ortho hydrogen is required and whether the label moves. Two further impostors: the radical-chain SRN1 mechanism, which also attacks unactivated aryl halides but is killed by radical scavengers and gives no cine product; and p-benzyne from the Bergman cyclization, same formula but a hydrogen-abstracting biradical, not a closed-shell electrophile.

Four ways to attach a nucleophile to a benzene ring — and why only one of them scrambles the position

Route	What the ring must have	Intermediate	Where the new bond ends up
Aryne (elimination–addition)	A leaving group <em>and</em> an ortho C–H; no activating group needed	o-Benzyne — neutral, strained, symmetric	Either of the two aryne carbons: ipso <strong>or</strong> cine (~50:50 when the ring is symmetric)
SNAr (addition–elimination)	A strong electron-withdrawing group ortho or para (classically NO₂)	Meisenheimer anion — ipso carbon now sp³, charge delocalised onto the withdrawing group	Ipso only; the position never moves
Diazonium (SN1-like)	An aryl diazonium salt from an aniline	Aryl cation, after N₂ leaves	Ipso only, but with radical side-products
Pd cross-coupling (Buchwald–Hartwig)	An aryl halide or triflate plus a Pd/ligand system	Ar–Pd(II)–Nu, via oxidative addition	Ipso only, and tolerant of almost everything else on the ring