Dopant Precipitation Clustering High Dose Supersaturation

Push implant dose past the solid solubility limit (see our activation-limits article), and the excess dopant doesn't just sit around inertly — it actively clusters together during anneal into precipitates: small, electrically inactive aggregates of dopant atoms, sometimes bound with silicon interstitials, that form through the same nucleation-and-growth physics that governs crystallization or phase separation in any supersaturated system.

Precipitation is a direct consequence of doing the thing high-dose implants are meant to do. A heavily doped source/drain or emitter region needs concentrations well above equilibrium solubility to hit target sheet resistance — that supersaturation is the entire point. But supersaturation is thermodynamically unstable, and given enough thermal energy, the system will always relax toward a lower-energy state by segregating the excess dopant into clusters, exactly the way a supersaturated sugar solution eventually crystallizes out.

Every precipitate is a double loss, not a single one. Dopant locked inside a precipitate is electrically inactive, so it doesn't lower sheet resistance the way the design intended — but precipitates also introduce structural defects at the precipitate/lattice boundary, which can act as the same kind of deep-level trap sites covered in our deep-level-defects article, adding leakage on top of the lost activation. A process that overdoses to "guarantee" low resistance can end up with both worse resistance *and* worse leakage than a correctly dosed one.

Anneal time and temperature control how much precipitation happens, not just how much activation happens. Nucleation requires clusters to reach a critical size before they become stable and start growing — below that threshold, a cluster is more likely to dissolve back into solution than to keep growing. Spike and millisecond anneals (see our RTA article) limit precipitation for exactly the same reason they limit diffusion: less time at temperature means less opportunity for clusters to reach that critical nucleation size in the first place.