Cap Annealing Silicon Nitride Oxide Out Diffusion Suppress

Push an anneal hot enough to fully activate a shallow, high-dose implant, and a second problem shows up: the bare silicon surface starts losing dopant too — boron and phosphorus both evaporate measurably off an exposed surface at activation temperatures, and silicon itself can decompose under the wrong ambient. Capped annealing solves this by depositing a thin nitride or oxide layer over the wafer before the high-temperature step, sealing the surface shut until the anneal is done.

Boron is the dopant this matters most for. Boron has a notably high vapor pressure and a tendency to segregate toward and out of an exposed silicon surface at the temperatures needed for full activation — an uncapped high-dose boron implant can lose a measurable fraction of its as-implanted dose before the anneal even finishes, directly degrading the final sheet resistance and junction depth the process was designed to hit. A nitride or oxide cap blocks that escape path entirely, so the dose that goes in is the dose that stays in.

The cap choice itself is a trade-off. Silicon nitride is a better diffusion barrier but can introduce film stress and is harder to strip cleanly without attacking the underlying silicon; a thermal or deposited oxide cap is gentler to remove but a somewhat weaker barrier at the highest anneal temperatures. Process engineers pick cap material and thickness against the specific anneal temperature and dopant species being protected — there's no single universal answer, because the cap's only requirement is to survive exactly the one anneal step it's protecting, nothing more.