Understanding and modeling of gettering of iron in silicon
Abstract
The first comprehensive attempt to understand and model the gettering of iron in silicon is presented here. Quantitative experimental results are obtained and used in predictive simulations of gettering. Algorithms for simulating segregation and relaxation gettering are proposed and included in a predictive finite differences gettering simulator. Such algorithms require numerous material parameters such as effective diffusivity, segregation coefficient, precipitation site density and radius. In this study, the following material parameters were obtained experimentally, the effective precipitation site densities for internal gettering of iron, the segregation coefficient for iron between silicon and aluminum, and the effective diffusivity of iron in p++ silicon. It is observed that the oxide precipitate as a whole acts as a sink for iron at low temperatures. At higher temperatures, the effectiveness of the oxide precipitates as sinks for iron decreases and an effective iron precipitate site density is introduced. The segregation coefficient of iron between silicon and aluminum was also determined to be between 105 to 10 6 in the 750--950°C range. The FeB dissociation time constant was experimentally determined to have an activation energy of 0.91 eV. This can now be used to calculate the effective diffusivity of iron in p ++ silicon which is a function of the migration enthalpy and the iron-boron pairing and dissociation time constants. Using these results, simulations of internal, aluminum backside, and heavily doped substrate gettering were performed to understand the kinetics and optimize the gettering of iron. It is shown that for internal gettering, small device regions of low precipitate site densities with optimal precipitate (OP) cooling profiles result in the best. For aluminum gettering, it is shown that for short annealing times, as during rapid thermal annealing (RTA), only iron from the region close to the aluminum layer can be gettered. Additionally, it is shown that precipitated iron can drastically extend the required gettering times for aluminum gettering. For the heavily doped substrates, published data for iron solubility was examined and the iron trap level was determined. The simulations demonstrated that the iron diffusivity in the p++ silicon is the rate limiting factor for iron gettering.
- Publication:
-
Ph.D. Thesis
- Pub Date:
- December 1998
- Bibcode:
- 1998PhDT.......357H