Experimental Investigation and Modeling of the Effects of High-Dose Ion Implantation Damage on Boron Diffusion in Silicon.

As integrated circuit device dimensions shrink, it becomes imperative to have a predictive capability for impurity redistribution in high-dose source/drain implanted regions during subsequent thermal anneals. The final doping profile depends on high concentration diffusion, implant damage, dopant solubility, amorphous layer regrowth, extended defect formation, and dopant segregation. The purpose of this work is to experimentally characterize the effects of high-dose implantation damage on dopant diffusion and to implement physically-based models for these effects into SUPREM-IV, Stanford's two-dimensional process simulator. Epitaxial silicon is used to grow a boron buried marker layer to carefully study the effect of implant damage alone on dopant diffusion. Silicon implants are used to generate surface damage, and the motion of the boron layer is monitored during furnace annealing in inert or oxidizing ambients. In inert ambients, boron diffusion is enhanced compared to equilibrium conditions due to point defects generated by the implant. The enhanced diffusion is characterized as a function of implant dose and energy, and anneal time and temperature. For short-time, low-temperature anneals, enhanced diffusion is independent of silicon implant dose, even above the amorphization level. For long times, enhanced diffusion of the marker layer is less for implant doses above the amorphization level. Extended defects form anytime amorphous layers are generated and are responsible for the smaller enhancements observed for long times. This hypothesis is confirmed by the anneals in oxidizing ambients. Oxidation -enhanced diffusion of boron is reduced in regions containing extended defects compared to regions without them. Some of the interstitials injected during the oxidation which normally enhance the diffusion of boron are absorbed by the defects. Transmission electron microscopy shows the growth of these defects during the thermal oxidations. A new equilibrium pair diffusion model has been implemented into SUPREM-IV. In addition, a one-dimensional model for extended defect growth has been implemented into SUPREM-IV and extended to account for dissolution of interstitial -type clusters. Agreement between experimental and simulated results of implant damage effects on dopant diffusion has been improved by the implementation of these new models.