Azeotropes

Why can no amount of careful distilling push ethanol past about ninety-six percent purity?

Distilling looks like the sort of thing that should get better the harder you try. Ethanol boils near 78 degrees Celsius and water at 100, so heat the mixture, catch what leaves first, and repeat. Repeat enough times, or build a tall enough column, and you should approach pure ethanol as closely as your patience allows.

You cannot. At ordinary atmospheric pressure the process stalls at about 95.6 per cent ethanol by mass — a figure I am recalling as the standard value rather than deriving — and a column of a thousand plates does no better than a column of forty. Something here is not merely slow. Something is closed.

Ask first why distillation works at all, since that answer contains the answer to why it stops. Boil a mixture and the vapour above it is not a sample of the liquid: it is richer in whatever escapes more readily. Condense it and you have a liquid of new composition. Boil again, enrich again. Every stage of a column is one such step, and the column's power is the number of steps stacked.

Notice what that depends on. Each step gains something only because the vapour composition differs from the liquid it came from. That difference is the whole engine — and it is not a constant. It varies with composition, and no law says it must stay on the same side of zero.

So put the question sharply: could there be a composition at which the vapour is identical to the liquid? If so, boiling changes nothing, condensing gives back what you started with, and every subsequent plate does the same. The column has run out of anything to do — not because it is inefficient, but because there is nothing left for efficiency to act on.

That is an azeotrope, and it exists because mixtures need not behave as the naive picture assumes. If ethanol and water simply diluted each other, each component's tendency to escape would fall in proportion to its share of the liquid, the more volatile one would always be enriched, and no azeotrope could exist. But unlike molecules do not interact as like ones do. Here the mixed state is less favourable than that simple picture allows, so both components escape more readily than their shares suggest and the mixture boils below either pure liquid — about 78.2 degrees Celsius, slightly under ethanol's own. That enhancement is not shared evenly: when ethanol is dilute its escaping tendency is boosted greatly, and as ethanol becomes the bulk of the liquid the boost transfers to the water. Somewhere between, the two cross, and there vapour and liquid have identical composition. The usual molecular story — water's hydrogen-bonded network paying an entropic price to accommodate ethanol's hydrocarbon end — is standard but more argued over than the thermodynamic fact itself.

Now separate the two kinds of obstacle, because this is where getting it wrong is most tempting. Most limits we meet are kinetic: the outcome is allowed and merely slow, so more time, surface or energy eventually gets there. This one is not. At the azeotropic composition the two phases are at equilibrium with each other, and no quantity of plates, reflux, patience or heat changes an equilibrium. It is a genuine thermodynamic constraint on the process as posed — which is exactly why the industrial answers all work by changing the question rather than trying harder.

And that gives the test. If the barrier really is a property of the equilibrium between two phases, and not of some ethanol-water compound forming at that ratio, it must move when the equilibrium moves. Pressure moves it: the azeotropic composition shifts with pressure and, below roughly a tenth of an atmosphere, the azeotrope stops existing altogether, so a vacuum column walks straight past 96 per cent. Sources quote that vanishing pressure a little differently, so treat the value as approximate; the direction and the disappearance are the robust part. A third component moves it too — an entrainer such as cyclohexane forms a new azeotrope that carries the water out overhead. Neither trick makes distillation better. Each replaces the phase equilibrium with a different one.

What would refute all this? If the 96 per cent mixture were a chemical compound — an ethanol hydrate, as was once proposed — its composition would be a fixed stoichiometric ratio, unmoved by pressure. It is neither: 95.6 per cent by mass is no simple whole-number molar ratio, and the composition demonstrably slides as pressure changes. Observe an azeotropic composition that stayed put under pressure and matched a tidy ratio, and the compound reading would win instead.

Three clarifications, the first correcting the commonest misreading.

An azeotrope is not an inseparable mixture. It is a mixture that this particular method cannot separate, because the method is built on a difference that has locally gone to zero. Adsorb the water onto a molecular sieve that admits water and excludes ethanol, or drive it through a selective membrane, and the separation proceeds without difficulty. Both are used industrially to make anhydrous ethanol. Nothing about the mixture forbids separation.

Nor is the azeotrope a place the mixture is trying to reach. It is a fixed point the process converges on from either side: start below it and the distillate climbs toward 96 per cent while the residue tends toward water; start above it and the residue tends toward pure ethanol instead. Which product you can obtain pure depends on which side you begin.

Contrast this with the isotope case elsewhere in this collection, where repeated condensation fractionates rainfall progressively as air moves inland. That works precisely because its fractionation never reaches unity — the tilt between phases survives every step, so the process merely depletes its reservoir and never stalls. An azeotrope is the other outcome: the tilt itself runs out.

Every term the collection defines is gathered in the glossary.