Silicon nitride

Silicon nitride is hard, strong, chemically inert and — unusually for a ceramic — good at surviving sudden temperature changes. Its thermal expansion is low and its conductivity moderate, and together those mean a piece heated or cooled quickly develops much less internal stress than an equivalent alumina one.

It keeps most of its strength to 1000 °C and above, does not creep, and is about 40 per cent the density of steel. In a rolling bearing, that lower mass means lower centrifugal load at speed, which is why hybrid bearings with silicon nitride balls run faster and cooler than all-steel ones and need less lubrication.

It is also expensive and difficult to make dense, which is the constraint on all of it.

It does not sinter easily, because the atoms barely diffuse. So it is made by reaction bonding — nitriding a silicon compact, which is cheap and leaves porosity — or by pressing with sintering aids such as yttria and alumina, which form a glassy phase at the grain boundaries and let densification happen. The glassy phase is then the weak point at high temperature, and controlling it is much of the engineering.

Hot isostatic pressing gives the best properties and the highest cost, and is what bearing balls are made by.

Rolling bearing balls, particularly in machine tool spindles, turbochargers and where electrical insulation between the races prevents bearing currents. Cutting tool inserts for cast iron and superalloys. Welding nozzles and jigs. Molten-metal handling, where nothing wets it. Turbocharger rotors. Glow plugs.

The famous use is the one that did not happen: through the 1970s and 1980s an all-ceramic gas turbine engine was pursued heavily in the United States and Japan, on the argument that a ceramic could run hotter and need no cooling. It was largely abandoned, because a component that fails without warning is unacceptable in a rotating machine and because superalloys and coatings improved faster than expected.

Known since the nineteenth century and developed as an engineering material from the 1950s. The ceramic engine programmes are its formative episode: they failed at their stated goal and produced most of the processing science that the successful applications rest on, which is a common and under-acknowledged pattern in materials development.