Fluence loss due to Rutherford backscattering for very low energy ion implantation in silicon

Abstract

Modern semiconductor device manufacturing process requires ion implantation as low as 0.5 keV. These are source/drain extension implants with high fluences. It is conceivable that the implanted ions cannot be totally retained due to surface sputtering and backscattering. So far the industry has taken a pragmatic route, i.e. implant with enough dopant so that the device reaches certain performance characteristics. Rutherford binary collision model says the stopping power cross-section is inversely proportional to the square of ions energy. This would suggest that the fluence loss from backscattering can be severe when the implant energy is very low. This work is our attempt to verify the theory's validity with such implants. In this work we experiment with boron ions of 250 eV to 5 keV and arsenic ions of 500 eV to 10 keV. With each energy the incident angle of ions to silicon wafers is changed from 0° to 30° and to 60°. A low fluence of 5 × 10 13/cm 2 is used for all the wafers in hope that backscattering effect can be studied without much influence from surface sputtering. Secondary ion mass spectrometry (SIMS) is used to probe for the retained fluence with each implant condition. This paper discusses the retained fluence as a function of species, energy and incident angle of ultra-low energy ions into silicon surface. It is observed that the fluence loss increases at larger tilt angles and the fluence loss due to scattering does not follow the energy dependence as depicted by the conventional Rutherford backscattering theory. At ultra-low energies the surface conditions and surface deposition mechanisms become more important and may compete with the scattering and sputtering processes. We also discuss the difficulties in dealing with such ultra-low energy implants such as the lack of established stopping power models, the issues with ion implantation and SIMS artifacts.