Three-way catalytic converter

Why does one exhaust brick need three different precious metals doing three different jobs?

Bolted under a petrol car is a ceramic brick, honeycombed with thousands of thin channels, coated with a rough oxide layer, and carrying a few grams of platinum, palladium and rhodium. Those grams are why the part is worth stealing.

The obvious question is why three metals. If one of them catalyses exhaust chemistry, why not use more of the best one and drop the others? And the less obvious question, which turns out to be the interesting one: why does this brick need a sensor, a feedback loop and an engine management computer working alongside it, when a catalyst is supposed to be a passive lump that simply sits there and helps?

Look first at what is coming out of the engine, because the name is a clue. There are three pollutants, not one. Carbon monoxide, from fuel that burned but not all the way. Unburnt hydrocarbons, from fuel that barely burned at all. And nitrogen oxides, which are not unburnt fuel in any sense — they are nitrogen and oxygen from the air itself, forced together by the heat of combustion.

Now ask what fixing each one requires. Carbon monoxide needs another oxygen atom to become carbon dioxide. Hydrocarbons need oxygen too, to become carbon dioxide and water. But nitric oxide needs its oxygen taken away, so the nitrogen can pair off and leave as harmless N2.

Sit with that for a moment, because it is the crux. Two of the three jobs are oxidations and the third is a reduction, and they are being asked of the same gas stream at the same instant. One wants oxygen present; the other wants it absent. This is not a case where a better catalyst would do all three — it is a case where the three jobs make contradictory demands of their shared environment.

So the specialisation is not laziness or hedging. Each metal is genuinely better at its own reaction: platinum and palladium are excellent oxidation catalysts for carbon monoxide and hydrocarbons, while rhodium is unusually good at splitting the nitrogen-oxygen bond and, importantly, at steering the product toward N2 rather than ammonia or nitrous oxide. Three reactions, three specialists.

But specialists sharing one workspace create a problem the individual specialists do not have. Ask what conditions would let all three work at once. There is one: a mixture in which the oxygen is almost exactly enough to burn the fuel and no more — neither a surplus that would starve the reduction, nor a deficit that would starve the oxidations. For petrol that is an air-to-fuel ratio near 14.7 to 1, and the window in which all three conversions stay high is startlingly narrow, on the order of a per cent either side.

Which answers the second question. The sensor and the feedback loop exist because the specialists cannot each be given their own conditions. The zirconia lambda sensor in the exhaust reports whether the mixture just burned rich or lean, and the engine management system trims the injectors continuously, hunting back and forth across the ideal point many times a second. And because that hunting still overshoots, the washcoat contains cerium oxide, which absorbs oxygen when the mixture runs lean and releases it when the mixture runs rich — a chemical flywheel that smooths the excursions the electronics cannot eliminate. The buffer is a coordination cost, and it exists purely because the jobs were divided.

The first thing to correct is the picture of a catalyst as an inert filter. Nothing is trapped. Every atom that goes in comes out; what the metals do is offer a surface on which bonds break and re-form at temperatures where they otherwise would not, so the same atoms leave in a different arrangement. That is also why a converter is not consumed, and why it can nonetheless be ruined: lead deposits on the active sites permanently, which is the real reason leaded petrol had to be abolished before this technology could work at all, and sulfur and silicone contamination degrade it more slowly.

Second, the converter is nearly useless cold. The metals need roughly 250 to 300 degrees Celsius before conversion switches on — the light-off temperature — and a disproportionate share of a modern car's total emissions is produced in the couple of minutes before that happens. Much of the engineering effort of recent decades has gone into reaching light-off faster, by moving the brick closer to the engine or heating it electrically, rather than into the catalysis itself.

Third, the three-way scheme is specific to engines that can run at that stoichiometric point. A diesel runs deliberately lean, with surplus oxygen always present, so the reduction leg simply cannot work no matter which metals are fitted. Diesels therefore need a different architecture entirely — typically a urea-dosed selective catalytic reduction system, where the reducing agent is carried on board and injected, because the exhaust will never supply it.

Finally, the split between platinum and palladium is partly chemistry and partly economics. Both catalyse the oxidations; their relative loading has shifted over the years with their relative prices and with tolerance for sulfur. Rhodium's role has no such substitute, which is a large part of why rhodium is the most expensive of the three.

• Why nitrogen oxides form at all in a hot cylinder, and how exhaust gas recirculation attacks the problem before the converter sees it.
• What ammonia slip is, and why the diesel answer to NOx brought a pollutant of its own.

Every term the collection defines is gathered in the glossary.