the pore-boundary interaction
In the late stages of firing, two things are happening at once: grain boundaries are sweeping through the material as grains grow, and isolated pores are sitting on those boundaries waiting to be emptied. The pore-boundary interaction is the tug-of-war between them — the way a moving grain boundary and the pores in its path hold onto or let go of one another. It sounds like a footnote, but it decides whether a ceramic reaches full density or freezes short of it, so it sits at the very heart of the final stage.
Why do pores cling to boundaries at all? A grain boundary has energy, so a boundary wants to be as small in area as it can. A pore sitting on a boundary punches a hole in it, locally erasing that patch of boundary area — so the pore actually lowers the boundary's energy by being there, and the boundary is reluctant to leave the pore behind (moving away would mean re-creating the erased area). The pore therefore exerts a drag on the boundary, pinning it, much as a second-phase particle pins a boundary in Zener's picture. This attachment is doubly good for densification: not only does the pore slow runaway grain growth, but as long as the pore stays on the boundary it has a fast short-circuit — grain-boundary diffusion — for its vacancies to escape to, so it can shrink and vanish efficiently. A pore riding a boundary is a pore on its way out.
The interaction is a balance of two speeds: how fast the boundary wants to move (driven by grain growth) versus how fast the pore can keep up (limited by whatever transport shrinks and drags it). If the boundary moves gently, it and the pore migrate together and the pore is removed. If the boundary is pushed too hard — high temperature, a large grain eating small ones, abnormal grain growth — it can tear free and leave the pore stranded inside a grain, where only slow lattice diffusion serves it and densification stalls. The whole craft of final-stage sintering is keeping this interaction on the favourable side: use fine uniform powder and grain-growth inhibitors (the textbook case is a trace of MgO in alumina) so the boundaries stay slow enough never to abandon their pores until the pores are gone.
In alumina doped with a few hundred ppm of MgO, the boundaries move slowly enough to stay tethered to their pores all the way through the final stage; the pores ride the boundaries to their end and the body reaches full, translucent density — the interaction kept on the winning side.
A pore on a boundary can be emptied fast; a boundary that outruns its pores strands them and stalls densification.
A pore lowers a boundary's energy by sitting on it, so pores pin boundaries — but only while the boundary moves slowly enough to keep up. Push grain growth too hard and the boundary wins the tug-of-war, breaks away, and dooms the trapped pore.