grain growth
Watch a tray of soap foam settle: the little bubbles slowly vanish and the big ones swell, until a few large bubbles fill the space that dozens of small ones used to. A polycrystalline ceramic does exactly this in the fire. Grain growth is the increase in average grain size during firing, as grain boundaries migrate and the bigger grains eat their smaller neighbours. A structure full of tiny grains has an enormous amount of high-energy boundary area, and nature would rather have less of it — so boundaries sweep through, small grains shrink and disappear, and the average grain gets coarser over time.
The engine is boundary curvature. A grain boundary curved like the wall of a bubble feels a pressure that pushes it toward its centre of curvature; a boundary always tends to migrate toward its concave side, which means a small grain (whose boundaries bow outward from the small grain's view) shrinks while its larger neighbour grows. The average grain size d then coarsens over firing time t in a power law, d^n minus d0^n equals K times t, where d0 is the starting size, K rises steeply with temperature (following an Arrhenius law, K proportional to exp(minus Q over R T)), and the exponent n is about 2 for ideal, clean grain growth and larger (3 or more) when pores, solutes or second phases drag on the boundaries.
Grain growth matters because it is the constant rival of densification. Both are driven by the same wish to lower energy, and both speed up when you fire hotter — but densification wants boundaries to sweep pores out to the surface, while grain growth wants boundaries to sweep past the pores and leave them behind, trapped inside grains where they can no longer be removed. Coarser grains also usually mean lower strength (a bigger grain is a bigger potential flaw). So the whole art of firing is to give the boundaries just enough mobility to close the pores, but not so much that grain growth runs away first. Sometimes coarse grains are wanted — for translucency, or low-loss magnetics — and then grain growth is encouraged, not fought.
An alumina fired at 1500 degrees C for one hour reaches 2 micron grains and 99 percent density. Held instead for eight hours, its grains coarsen to 8 microns while the density barely improves — and a few pores that were on their way out get overtaken by growing grains and sealed inside, so the longer firing actually leaves the part weaker.
More time in the furnace is not always better: past a point, grain growth wins and drags strength down.
Grain growth is not the same as sintering. Sintering removes pores and raises density; grain growth just rearranges the grains and adds no density. A part can grain-grow enormously while barely densifying — the worst of both worlds.