Post-Implantation Annealing of Platinum in Silicon

Platinum is used to tailor the electrical behavior of silicon power devices, for example of silicon power diodes. Such diodes are typically used in applications in which inductive loads are switched at high power and high frequencies. The platinum can be introduced through diffusion from a Pt-silicide layer, or by ion implantation. Ion implantation is a major technique used for introducing dopants in silicon wafers. For platinum, it is an economically competitive technique which provides an additional design parameter, better reproducibility, and superior dose control compared to diffusion from a Pt-silicide layer. However, a thorough understanding of platinum diffusion during post-implantation annealing is required to fully utilize the capabilities of this technology. Modeling and simulation of devices are standard tools used in development to predict and optimize the device performance. To complete the simulation chain, a model which describes platinum diffusion during post-implantation annealing is needed. Although a previous model exists, it cannot reproduce some of the experimental results presented in this thesis. Accordingly, a new model for post-implantation annealing of platinum in silicon was developed in this work. The development of the new model started by revisiting old experiments of platinum diffusion from a Pt-silicide layer reported in the literature. Even though platinum diffusion from a Pt-silicide layer is considered to be a fairly well-understood process, there are several inconsistencies among the reported data and a quantitative model that can describe all available data does not exist. The mentioned studies have been used in this work, accompanied by new experiments, to calibrate the model parameters. The new experiments were especially designed to estimate suitable initial conditions of the intrinsic point defects in silicon wafers grown with the Czochralski or the float-zone method, and to investigate the influence of preannealing, the processing step typically performed before to eliminate grown-in voids remaining from crystal growth. These investigations resulted in new expressions for the solubility concentration of substitutional platinum, the transport capacity of interstitial platinum, a new upper limit for the equilibrium concentration of self-interstitials, and a parameter set for self-interstitials and vacancies. The calibrated model parameters are applicable for all considered processing steps, including preannealing of Czochralski-grown silicon wafers, platinum diffusion, and phosphorusdiffusion gettering of platinum in silicon. The results also indicate that the type of substrate material can have a significant influence on the resulting platinum depth profiles. To what extent depends on the preannealing process as well as the temperature and duration used for the platinum-diffusion annealing. Finally, the newly calibrated model for platinum diffusion from a Pt-silicide layer was extended to include platinum precipitation. This effect is used to explain the incomplete activation of the implanted platinum found after post-implantation annealing. The new model was calibrated considering data existing in the literature as well as new post-implantation annealing experiments conducted in this work. The developed model can accurately describe the measured depth profiles of the electrically active platinum concentration after annealing in a wide temperature range from 770 °C to 900 °C for implanted doses in the range from 5E11 cm-2 to 1E13 cm-2. These processing conditions were selected to be compatible with industrial manufacturing conditions. The new model is already used in industry, integrated into a state-of-the-art simulation chain for the development of platinum-diffused silicon power diodes.