End-of-range damage produced by n-type dopants is frozen-in if that damage is produced in germanium-rich regions in crystalline silicon. This is not the case with p-type dopants. High germanium doses were implanted into silicon; the wafers were annealed at 1000 C for 30 minutes under nitrogen. Phosphorus or boron then were implanted into the crystalline siliconfollowed by a 900 C, 30 minute neutral ambient anneal.

Explanation of the freeze-in phenomenon relies on results of semiempirical quantum chemical calculations that showed that group V dopants, whether interstitial or substitutional, as well as interstitial silicon, would be attracted to the vicinity of substitutional germanium sites in germaniumrich regions ofthe crystalline silicon. This general preferential diffusion behavior is not calculated to occur with the p-type dopants from group III. When substitutional, in fact, the group III impurities are predicted to be repelled from the germanium-rich regions.

TEM studies show that any residual damage with boron as the implant and germanium present is low in density and within 20 nm of the silicon interface. This is contrary to studies involving phosphorus where damage is exclusively associated with the end-of-range of the phosphorus implant. Both sets of results are in accord with the theoretical considerations.