Modeling of diffusion and ion implantation in mercury cadmium telluride: A comparison to transient enhanced diffusion in silicon

Process simulators can significantly reduce the cost and time involved in developing new semiconductor devices. They have become essential tools for process development, and their value will only increase as device architectures become more complex and device dimensions shrink. The effectiveness of these simulators as predictive tools relies on accurate modeling of atomic mechanisms underlying the critical processing steps. One of the most important aspects of semiconductor processing is defect and dopant diffusion. Diffusion in most semiconductors is controlled by point defect based mechanisms. In silicon, two point defects are present, the silicon interstitial and vacancy. In compound semiconductors, the number of point defects and associated defect chemistries are potentially much more complex. Fortunately, defects from one sublattice usually dominate. In the case of HgCdTe, the cation defects, specifically those associated with the Hg atom (Hg interstitials, Hg vacancies, and Te antisites), exist in much higher concentrations than their anion counterparts. This simplifies the modeling of diffusion in HgCdTe and allows the use of algorithms similar to these for point defect mediated diffusion in silicon. HgCdTe annealed under certain conditions can have a concentration of Hg vacancies as high as 10(18) cm(-3) These vacancies, which are doubly ionized accepters, intrinsically dope the HgCdTe. A junction can be created in this vacancy rich, p-type material through ion implantation of an electrically inactive species. During the post-implant anneal, interstitials released from the implant damage region annihilate cation vacancies, revealing a low concentration, grown-in donor. An abrupt n on p junction is formed between the revealed donors and the vacancy doped region. This is similar to the mechanism underlying enhanced diffusion of implanted B and P dopants in silicon. In this case, injected interstitials from the implant damage region displace substitutional dopant atoms, creating mobile dopant interstitials, and enhancing the dopant diffusion. The effect of the excess interstitials in both materials is transient, decaying exponentially with time. In HgCdTe, the final junction depth is dependent on the concentration of trapped interstitials in the implant damage region. The time to reach this depth is a function of the trapped interstitial release rate. Once the trapped interstitials have been depleted, the junction may smear and become unstable if the vacancies are mobile. In this paper, the similarities between transient enhanced diffusion of implanted dopants in silicon is compared to junction formation in implanted HgCdTe.