Modeling of Transition Metal Redistribution to Enable Wafer Design for Gettering

Abstract

We use standard solubility relations and diffusion‐limited rate equations to create a model for the gettering of Fe in silicon. The model employs experimentally determined values for diffusivity, ion pairing, binding potentials, and precipitate densities. The model provides a means to evaluate the relative effectiveness of solubility enhancement induced segregation gettering, internal gettering (IG), precipitation, and back‐side gettering outdiffusion. The materials variables are p‐type doping level, density of bulk IG sites, and back‐side IG site density. From this work, an understanding of the interactions of the various gettering mechanisms is developed. We follow the contaminant concentration at various positions in the wafer as a function of time and temperature. Negative temperature ramps are modeled to simulate the inevitable cooling step following high‐temperature processing. The results indicate that segregation from epitaxial layers to heavily doped substrates brings orders of magnitude improvement in Fe removal over IG alone and that segregation is effective in reducing contamination levels even when initial contamination levels are very low. The best gettering occurs when IG sites work together with segregation. A well‐designed wafer has a high density of IG sites to accelerate equilibration during cooling and to enhance mass transport from the segregation interface. Higher p‐doping levels in the substrate enhance the segregation coefficient, creating a steeper gradient of [Fe] from the front surface, and slower cooling allows for the greatest amount of equilibration to occur and therefore the most effective Fe gettering. A time‐temperature‐transformation diagram approach is introduced to provide a comprehensive description of the wafer and process design parameters for effective gettering. © 2000 The Electrochemical Society. All rights reserved.