the glassy grain-boundary phase
In many polycrystalline ceramics, a thin skin of glass wets every grain boundary, like a film of syrup between stacked pancakes or the mortar between bricks that never quite set to stone. This is the glassy grain-boundary phase: an amorphous (non-crystalline) layer, often only one to a few nanometres thick, that coats the boundaries and pools in the pockets where three or four grains meet (the triple junctions). It is easy to overlook because it is so thin, yet it can rule a ceramic's behaviour out of all proportion to its tiny volume.
Where does it come from? Two main sources. In traditional clay-based wares (porcelain, stoneware), the feldspar flux melts during firing into a glass that flows around the crystalline grains and freezes there on cooling — the body is glued together by glass. In advanced ceramics it usually comes from sintering aids and impurities: silicon nitride, for instance, is sintered with additives like alumina and yttria that, together with the silica skin on the powder, form a liquid at firing temperature. That liquid is what lets liquid-phase sintering work — grains dissolve into it and re-precipitate, densifying the body fast — but on cooling the liquid is trapped as a glassy film. It is essentially frozen liquid: atoms with sharp short-range order but no long-range crystalline repeat.
This film is the classic Achilles' heel of high-temperature ceramics. As temperature rises toward the glass's softening range, the film goes soft and viscous, and grains slide over one another on it — this is the dominant mechanism of high-temperature creep in silicon nitride and many others, and it can slash strength above about 1000 degrees C. The same film raises dielectric loss in electronic ceramics and offers a fast diffusion path. The engineering answer is often to crystallise the film after sintering (a controlled heat treatment that devitrifies the glass into a refractory crystal), or to choose additives that give a more refractory boundary phase — trading a little sintering ease for much better high-temperature life.
A silicon nitride turbocharger rotor sintered with a yttria-alumina liquid is strong at room temperature but starts to creep above 1000 degrees C as its nanometre-thin boundary glass softens. Re-heating it to crystallise that glass into a refractory yttrium-silicate phase pushes its useful temperature several hundred degrees higher.
A film only nanometres thick, invisible in an optical micrograph, can set the entire high-temperature limit of the part.
Do not read 'glassy' as 'weak everywhere'. At room temperature the glassy film is a rigid solid and often causes no trouble. It becomes the villain only near and above its softening range, which is why the problem is a high-temperature one.