Lindlar Catalyst: A Poison That Makes It Stop

A catalyst built to be bad at its job

Palladium is the finest hydrogenation metal there is, and that is precisely the problem: give ordinary 10% Pd/C an alkyne and hydrogen and you get the alkane. Lindlar's catalyst is that metal, systematically ruined.

The recipe, published by Herbert Lindlar (1909–2009) at Hoffmann-La Roche in Basel (Helv. Chim. Acta 1952, 35, 446), carries three deliberate handicaps. The palladium is precipitated as roughly 5 wt% Pd onto calcium carbonate rather than carbon — chalk is a low-area, mildly basic support, so it disperses the metal poorly, offers the desorbed alkene almost no pore network to linger in, and keeps acid-catalysed side reactions off the table. The fresh catalyst is treated with lead(II) acetate, leaving about 3.5 wt% Pb alloyed into the surface. Finally, about 5 mol% quinoline on substrate goes in as a second moderator, usually added by the chemist rather than the supplier.

Lead and quinoline do related jobs. Lead dilutes the contiguous Pd ensembles, breaking up the multi-atom sites that grip an alkene tightly enough to hydrogenate it, blunting the step edges and kinks that are a particle's most aggressive sites, and making it far harder for hydrogen to dissolve into the lattice as bulk β-PdHx. Quinoline's nitrogen lone pair then coordinates to whatever aggressive palladium is left. The result still splits H2 and still binds an alkyne, but has lost its appetite for an alkene.

The Horiuti–Polanyi mechanism, step by step

Everything happens on the metal surface, by the sequence Horiuti and Polanyi proposed in 1934 (Trans. Faraday Soc. 30, 1164), still the working model ninety years on.

• 1. Hydrogen chemisorbs dissociatively. H2 strikes the palladium and comes apart into two surface hydride atoms — essentially barrierless on a clean Pd(111) terrace, which is why 1 atm and room temperature suffice.
• 2. The alkyne lies down. It adsorbs side-on across two or more Pd atoms, C–C axis parallel to the terrace, in a di-σ/π mode using both orthogonal π systems — a far stronger grip than an alkene manages with one.
• 3. The first hydride inserts. One surface H migrates onto a carbon, giving a σ-bonded vinyl — the half-hydrogenated state — still anchored through the other carbon.
• 4. The second hydride follows, from the same side. It arrives from the same metal face, because that is the only place hydrogen exists. Both hydrogens therefore land on one face of the former triple bond: the addition is syn, and the substituents are forced onto the same side. That is the Z alkene, and no other geometry is available from a single surface encounter.
• 5. The alkene lets go. With one π bond and two substituents in the way it binds far more weakly than the alkyne did, and desorbs before a third hydrogen finds it.

Step 4 is not an assumption. Run the reaction with D2 and the dominant product is the cis-1,2-dideuterio alkene, both deuteriums on one face — the classic proof of suprafacial delivery. On hotter, unmoderated palladium you also see scrambled and monodeuterated products, itself evidence that step 3 is reversible.

Thermodynamics cannot stop this — kinetics can

There is no thermodynamic ledge for the alkene to sit on. For the parent system, hydrogenating acetylene to ethylene has ΔH = −175 kJ/mol, and hydrogenating ethylene onward to ethane a further −137 kJ/mol — the second step is worth another three quarters of the first. Internal alkynes behave the same way (2-butyne to cis-2-butene, about −153 kJ/mol, then ≈ −119 kJ/mol to butane).

Worse, the isomer you want is the wrong one. cis-2-Butene is 4.2 kJ/mol LESS stable than trans-2-butene, because the two methyls on the same side push against each other. Lindlar hydrogenation deliberately delivers the uphill isomer, so its selectivity is entirely kinetic: the product is whatever the surface makes fastest.

The kinetic lever is competitive adsorption. The alkyne binds palladium roughly 10–100× more strongly than the alkene it produces, so while any alkyne remains it monopolises the surface. Take Langmuir-style competition with K(alkyne)/K(alkene) = 50: at 99% conversion the alkyne is 1% of the organic material in solution, yet the ratio of occupied sites is still 50 × (1/99) ≈ 0.51 — it holds a third of the covered surface while nearly gone from the flask. When the last alkyne goes that protection vanishes and the alkane forms quickly — hence the sharp break in the H2 uptake curve, and the instruction to stop at exactly 1.00 equivalent of H2.

The poisons buy the rest. Teschner and co-workers (Science, 2008, 320, 86) showed by in-situ X-ray spectroscopy that selectivity is settled just below the surface, by two subsurface species pulling opposite ways. Hydrogen dissolved in the lattice as bulk β-PdHx is the culprit: it feeds an indiscriminate hydrogen flux up to the surface that over-reduces and isomerises. Carbon dissolved into the near-surface lattice is the cure, filling the interstitial sites hydrogen would otherwise occupy; a palladium particle only becomes selective once that carbon is in place.

Running it: hardware, stoichiometry and a worked hydrogen budget

The apparatus is deliberately unimpressive: a flask, a stir bar, a hydrogen balloon or low-pressure Parr shaker, and a gas burette if you are serious. Conditions are 1 atm H2 at 20–25 °C, catalyst at 5–10 wt% on substrate, in hexane, ethyl acetate or toluene, quinoline added before the hydrogen, complete in 1–4 h.

The number that matters is the gas: one mole of H2 at 25 °C and 1 atm occupies 24.5 L (V = RT/P = 0.08206 × 298.15 / 1 = 24.47 L). So for a real run:

• Substrate: 5.00 mmol of an internal alkyne.
• 1.00 equivalent of H2 = 5.00 × 10⁻³ mol × 24.47 L/mol = 122 mL of hydrogen at 25 °C, 1 atm.
• Stop the instant the burette reads 122 mL. Let it run to 147 mL (1.20 equiv) and you have, to first order, turned 20% of your hard-won Z-alkene into alkane — essentially inseparable from the alkene on silica.

Without a burette, monitor by TLC or GC and quench at the first disappearance of alkyne; never stir on under H2 just to be sure. Filtration is then where people set fire to things. A palladium filter cake loaded with adsorbed hydrogen and wet with volatile solvent is pyrophoric and ignites that solvent as it dries in air. Keep it wet, quench under water, and dispose of it as lead-bearing waste — lead(II) acetate is a reproductive toxin and a REACH substance of very high concern, which is why lead-free substitutes (Pd–Ag, Pd–Bi, Pd–Cu single-atom alloys) are an active target.

How you prove you actually got the Z isomer

A geometry claim needs evidence, and three measurements do almost all the work.

• Proton NMR vinyl coupling. The three-bond H–H coupling across a double bond depends on the dihedral angle: for a 1,2-disubstituted alkene, cis gives 3J ≈ 10–12 Hz and trans gives 3J ≈ 15–16 Hz. The gap is large, unambiguous, and the first thing anyone checks.
• Infrared. The =C–H out-of-plane bend is diagnostic: a strong band near 675–730 cm⁻¹ for a cis-disubstituted alkene against roughly 965 cm⁻¹ for trans. The AOCS trans-fat method (Cd 14d-99) rests on that band.
• Capillary GC. On the long, highly polar cyanopropyl-polysiloxane columns AOCS Ce 1h-05 specifies for cis/trans fatty-acid analysis, the E isomer elutes ahead of the Z and a 97:3 ratio integrates directly. Silver-ion (argentation) TLC or HPLC separates them more cleanly still, because Ag⁺ complexes the less hindered cis π bond more strongly.

The catalyst is specified too: commercial Lindlar ships against a palladium assay (ICP-OES, nominal 5 wt%) and a lead assay (≈ 3.5 wt%), with metal dispersion measured by CO chemisorption. Batch-to-batch variation in lead level is the commonest reason a published procedure fails to reproduce — an under-leaded lot over-reduces, an over-leaded one does nothing — so careful groups burn a calibration run on a model alkyne first.

Vitamin A, pheromones, and the largest Lindlar-like reactor on Earth

Lindlar was not doing a methods study for its own sake — he worked inside Roche's vitamin A and carotenoid programme in Basel, whose industrial routes ran through acetylenic intermediates that had to be half-hydrogenated to a cis double bond without touching the polyene chain already in the molecule; nothing in 1952 could do it. The reproducible preparation appeared in Organic Syntheses in 1966 (Lindlar and Dubuis, vol. 46, p. 89) and has barely changed since. It is now the default way to install a Z double bond — insect pheromones are overwhelmingly Z-alkenes and their mating-disruption lures are made this way at tonne scale, as are leukotrienes and prostaglandins.

Its largest relative is not in a flask at all. Every steam cracker on Earth makes ethylene contaminated with 0.5–2% acetylene, and polymer-grade ethylene must carry less than about 1 ppm, because acetylene poisons Ziegler–Natta and metallocene polymerisation catalysts and forms shock-sensitive metal acetylides in the cold box. The fix is an acetylene converter: a fixed bed of Pd–Ag on alumina, often only 0.01–0.05 wt% Pd, with carbon monoxide metered into the feed as the moderator in place of lead and quinoline — the same problem at millions of tonnes a year. Lose the exotherm and the bed runs away to ethane; let oligomerisation take hold and it fouls with green oil, the C4-and-heavier tar that forces a shutdown.

Failure modes, and the reactions it gets confused with

Lindlar hydrogenation fails in a small number of well-characterised ways.

• Overshoot. Past 1.00 equiv of H2 the alkane grows, and it will not chromatograph away.
• Z to E drift on a long run. A Z-alkene that re-adsorbs can take one hydrogen to give a σ-bonded alkyl, and in that half-hydrogenated alkyl the C–C bond is genuinely single and free to rotate; β-hydride elimination then releases it as the E isomer, or with the double bond walked a carbon along the chain. (The σ-vinyl from the alkyne cannot do this — its C–C bond is still double, so it cannot rotate — though it can eliminate the hydrogen it just received, which is what scrambles deuterium labels.) This is the origin of the 3% in a 97:3 ratio; it worsens with time, temperature, H2 pressure and under-poisoned catalyst, and it is the main thing quinoline suppresses.
• Hydrogenolysis, or no reaction at all. Palladium can cleave propargylic and benzylic C–O bonds; conversely, thiols, sulfides, phosphines and free amines poison Pd properly rather than partially.
• Terminal alkynes. They react fast, and the terminal alkene has no geometry to protect, so the stop is far less clean.

And two look-alikes worth keeping straight:

• Sodium in liquid ammonia at −78 °C is the complementary reaction, not a variant: with no metal surface, a solvated-electron sequence in solution ends at a vinyl anion that settles into the less crowded geometry — the E (trans) alkene, typically >98% E. Terminal alkynes merely deprotonate to the acetylide (pKa ≈ 25).
• The Rosenmund reduction also uses poisoned palladium — Pd on BaSO4 with quinoline-sulfur or thiourea — but it halts an acyl chloride at the aldehyde, in hot xylene, before the alcohol.

Ways to stop an alkyne at the alkene stage — and the one route that gives the opposite geometry

Method	Conditions	Product	Where it breaks
Lindlar catalyst	~5 wt% Pd/CaCO3 + Pb(OAc)2, ~5 mol% quinoline, 1 atm H2, 20–25 °C	Z (cis) alkene, commonly 97:3 Z:E	Past 1.00 equiv H2 the alkane grows; long runs drift to E
Unmodified 5–10% Pd/C	1 atm H2, 25 °C, no moderator	Alkane — no stop at all	Nothing blocks the alkene from re-adsorbing
Na (or Li) in liquid NH3	−78 °C, NH3, alcohol proton source, no H2 gas	E (trans) alkene, >98% E	Terminal alkynes are only deprotonated to the acetylide
P-2 nickel (nickel boride)	Ni(OAc)2 + NaBH4, ethylenediamine, 1 atm H2	Z (cis) alkene, ~95%	Boride activity is batch-sensitive; slower to quench
Zn(Cu/Ag) couple	Activated Zn dust, aqueous methanol, 25–60 °C, no H2 gas	Z (cis) alkene	Slow; chosen when Pd would attack other groups
Industrial acetylene converter	Pd–Ag on alumina, 0.01–0.05 wt% Pd, CO moderator, ~30–120 °C	Ethylene, residual acetylene below ~1 ppm	Runaway exotherm to ethane; fouling by green oil