Pre-amorphization damage in ion-implanted silicon

Ion implantation in silicon with doses below the amorphization threshold can lead to the formation of dislocations after high-temperature annealing. We have studied this for implants of 0.1–1 MeV B, Si, P, Ga, As, In, and Sb ions after annealing at 900°C using cross-sectional transmission electron microscopy. Pre-amorphization damage, also called category I dislocations, is observed if the total number of silicon atoms displaced by the implant exceeds a critical value before reaching the threshold dose for amorphization. These dislocations are of interstitial type and result from agglomeration of mobile silicon interstitials. The critical number of displaced Si atoms required for pre-amorphization damage formation increases with the mass of the implanted species and was determined by Rutherford backscattering spectrometry and channeling analysis to range from 1.5 × 10¹⁶/cm² for B ions to (1.5-2) × 10¹⁷/cm² for Sb ions. This increase with mass is attributed to an increasing collision cascade density resulting in a lower fraction of the measured damage being in the form of mobile Si interstitials needed for dislocation formation. In contrast to keV implants, category I defects are observed for high-mass species at MeV energies because the critical number of mobile interstitial silicon atoms is reached prior to the amorphization threshold. The critical number can be used to manipulate secondary defect formation. First, introducing a second damage profile can influence where the secondary defects form. Results are presented for MeV B or As implants in combination with low-energy Si irradiations. Depending on the separate implant parameters the position where secondary defects form can be influenced. Second, a comparison between channeling and random implants of B or P ions in Si(100) wafers shows that higher doses can be reached without formation of secondary defects by channeling implants than by random implants due to the lower amount of damage produced by a channeled ion. In either case, secondary defect formation is observed after high-temperature (900°C) annealing only if the total number of displaced Si atoms exceeds a critical value of ∼ 1.5 × 10¹⁶/cm² and ∼ 5 × 10¹⁶/cm² for the B and P implants, respectively. Third, higher total doses can be introduced without forming secondary defects by repetitive subthreshold implants each followed by an anneal to remove the implant damage. While a single 6 × 10¹³ In/cm² implant results in a high density of dislocation loops after annealing, we demonstrate that instead using four separate 1.5 × 10¹³ In/cm² implants each followed by an anneal leads to the formation of only a few partial dislocations. Pre-amorphization damage formation and annihilation is shown to influence transient tail diffusion of B. This has been investigated as a function of B implant condition, dose, energy, time, temperature, and as a function of further Si or Ge implants. Significantly longer transient tail diffusion is observed for B implants along [100] than for random implants reflecting the differences between the random and channeling implants in the position of the damage distribution relative to the B profile. A second implant with 1 MeV ²⁹Si ions below the amorphization threshold can significantly reduce B tail diffusion if the damage for the Si dose is high enough to form pre-amorphization damage during the anneal. Lower Si doses do not influence B diffusion. Annealing of extended defects results in anomalous diffusion as well. These results demonstrate that Si interstitials cause the enhanced B diffusion. Transient B tail diffusion is completely prevented only if the B-implanted silicon is amorphized so that the B profile is completely incorporated in the amorphized region.