Self-healing coatings

Why does a paint deliberately filled with tiny bubbles of liquid protect steel better than a tougher, more perfect film?

Take a good protective paint and stir into it great numbers of tiny liquid-filled shells. You have made the film softer, less dense, and studded with inclusions that concentrate stress — by every measure of the coating itself, worse. Panels made this way nevertheless survive salt-spray exposure that eats through the tougher, unfilled film.

So the assumption to interrogate is not what the capsules do, but the one underneath: that a coating protects in proportion to how good it is.

Ask what a barrier coating is actually doing. It stands between steel and the water, oxygen and dissolved ions steel needs in order to corrode. Now ask how such a barrier fails, and notice something odd about the arithmetic. If ninety-nine point nine percent of the film is perfect and one scratch reaches bare metal, the intact area does not supply ninety-nine point nine percent of the protection. It supplies almost none of it at that scratch.

Worse. Corrosion is an electrochemical circuit needing an anode where metal dissolves and a cathode where oxygen is reduced. The scratch is a small anode; the film's edges disbond slightly, and the large area under them can host the cathodic half-reaction. A tiny anode wired to a large cathode carries a fierce local current density, which is why a scraped painted panel corrodes in a narrow line far faster than an unpainted one corrodes all over. A barrier's performance is set by its worst point, and the rest of the film can make that point worse.

Which reframes the design problem. There are only two moves: make a defect less likely, or make a defect matter less. Toughness, thickness and adhesion pursue the first and have a ceiling — stone chips, handling, thermal cycling and ultraviolet embrittlement will eventually produce a defect, and its consequence is not bounded by how good the film was. Spending on perfection buys delay; spending on response buys tolerance.

So what would a response look like? Triggered by the damage itself, because nobody inspects the underside of a bridge girder. Local, because the reagent must be at the defect. And sitting there inert for years beforehand.

That is the shape of the microcapsule design. Shells tens to a few hundred micrometres across — recalled as a range, since it varies enormously by system — hold a liquid healing agent. A crack running through the brittle film reaches a capsule and ruptures it, and the released liquid is drawn into the crack by capillary action. Then it must set: in the best-known laboratory system the released monomer meets a catalyst dispersed through the surrounding resin and polymerises; in others the trigger is ambient moisture or oxygen reaching the freshly opened crack.

For corrosion protection specifically, the common industrial version is less romantic and more effective. The capsules or carriers hold a corrosion inhibitor, and what reaches the defect is not glue but chemistry that passivates the exposed steel. The film need not be rebuilt at all; the bare metal simply has to be made unreactive.

Here the split between what can happen and what does happen is the entire mechanism. That inhibitor could passivate that steel on the day the paint was mixed — nothing thermodynamic prevents it. What prevents it is a wall. Encapsulation is a purely kinetic device: it holds a reagent thoroughly capable of reacting in a place from which it can reach nothing, and makes damage the only key. The reagent is not made stable. It is made unreachable.

One consequence follows immediately, and it is the honest limit of the idea. The reservoir is finite. Every capsule fires once and is spent, so capsule size and loading decide how many events, and how large an event, a patch of film can answer.

Three refinements, and one of them is a test.

This is not the same thing as passivation, though the two are often filed together. A stainless steel's chromium oxide film also heals when scratched, but its reservoir is the metal itself plus the oxygen in the air, which is why it heals at the same spot indefinitely. A capsule coating heals a site a fixed number of times, and then that site is an ordinary scratch. Nor is it a sacrificial anode, which is also a finite reservoir but is consumed continuously everywhere it is connected; capsule redundancy is spent only where and when damage occurs, which is what makes so small a reservoir go so far.

The capsules must be well bonded to the resin, which sounds like a manufacturing detail and is actually a requirement. A poorly bonded capsule is a hole, and a crack in a brittle film deflects around a hole rather than through it — so the crack passes by, no agent is released, and the coating is simply a weaker film.

And the test that decides whether any of this is real: build three panels, one unfilled, one filled with the healing agent, one filled with capsules containing nothing. Scribe all three, expose them, and measure how far corrosion creeps from the scribe. If the empty-capsule panel performs as well as the filled one, nothing was healed and the capsules merely blunted cracks — the account above would be wrong. Then scribe the filled panel twice through the same point: a reservoir-limited mechanism must fail the second time at that site while a fresh site still heals. A coating that healed one spot over and over would not be this mechanism at all.

• Vascular coatings, which replace isolated capsules with refillable channels, and what that costs.

Every term the collection defines is gathered in the glossary.