Use of a Purged FOUP to Improve H-Terminated Silicon Surface Stability Prior to Epitaxial Growth

A H-terminated Si surface serves as the starting surface for hetero-epitaxial growth. The presence of oxygen on the wafer surface caused by surface re-oxidation can significantly reduce the epitaxial quality. In this paper we evaluated the capabilities of an Entegris EBM (Entegris Barrier Material) purged FOUP (Front-Opening-Unified-Pod) to reduce this re-oxidation. The surface cleanliness is one of the factors affecting (hetero-) epitaxial quality. Clean starting surfaces can be obtained by a high temperature H 2 -bake (T > 800 °C). [1] Surface reflow during these high temperature bakes can significantly impact the integrity of small structures, e.g. causing corner rounding of FINFET structures. [2] As the clean starting surface can no longer be obtained by the H 2 -bake at relative high temperatures, more stringent requirements are put on the pre-epi clean and on the impact of queue time between wet-chemical clean and epitaxial growth. We have evaluated the re-oxidation of H-terminated 300 mm Si surfaces in a non-purged FOUP compared to an EBM purged FOUP while the FOUP is open and docked to the factory interface. Re-oxidation or degradation of the H-terminated Si surface has been evaluated by capping the surface with an epitaxial-grown 80nm Si-cap/80nm Si 0.8 Ge 0.2 bilayer. The H-terminated surface was created in a separate step before the epitaxial growth by applying a thermal treatment in H 2 in a conventional ASM Epsilon3200 epi reactor. The Si wafer was unloaded in between the thermal treatment and the Si-cap/SiGe epitaxial growth step. This procedure offers the most-controllable scheme for creating a H-passivated surface but the learnings are also applicable to wet-chemical pre-epi treatments. The Si 0.8 Ge 0.2 / Si substrate interfacial cleanliness was evaluated at the center of the wafer by SIMS (Fig. 1). The EBM purged FOUP shows a lower interfacial degradation compared to the standard FOUP. The uniformity of the surface degradation has been investigated by surface light scattering (data not shown). While the light-scattering intensity and its range increased significantly after queuing for 1hr in a non-purged FOUP, this increase was suppressed when using the N 2 -purged FOUP. The light scattering measurements also show that the epitaxial quality is better in a N2-purged FOUP compared to a non-purged FOUP but degraded compared to the loadlock reference. In this paper we evaluated the impact of queue time of an H-terminated Si surface in a non-purged FOUP compared to an EBM N 2 -purged FOUP. The interfacial degradation as measured by light scattering and SIMS for 1hr of queue time is shown to be less when the EBM N 2 -purged FOUP is used. However the epitaxial quality degradation was shown not to be completely suppressed. References [1] R. Loo et al. Solid State Phenomena 145-146, (2009) 177 [2] N. Sato et al. Appl. Phys. Lett. 65 (1994) 1924 Figure 1: Degradation of the Si 0.8 Ge 0.2 -on-Si interface as measured by SIMS. The interfacial oxygen concentration is plotted as a function of the queuing condition.