Modeling two-dimensional solid-phase epitaxial regrowth using level set methods

Abstract

Modeling the two-dimensional ͑2D͒ solid-phase epitaxial regrowth ͑SPER͒ of amorphized Si ͑variously referred to as solid-phase epitaxial growth, solid-phase epitaxy, solid-phase epitaxial crystallization, and solid-phase epitaxial recrystallization͒ has become important in light of recent studies which have indicated that relative differences in the velocities of regrowth fronts with different crystallographic orientations can lead to the formation of device degrading mask edge defects. Here, a 2D SPER model that uses level set techniques as implemented in the Florida object oriented process simulator to propagate regrowth fronts with variable crystallographic orientation ͑patterned material͒ is presented. Apart from the inherent orientation dependence of the SPER velocity, it is established that regrowth interface curvature significantly affects the regrowth velocity. Specifically, by modeling the local SPER velocity as being linearly dependent on the local regrowth interface curvature, data acquired from transmission electron microscopy experiments matches reasonably well with simulations, thus providing a stable model for simulating 2D regrowth and mask edge defect formation in Si.

Solid-Phase Epitaxy

“…Commercial devices are now on the nanometer scale, and processing steps may involve the amorphization of spatially confined regions usually defined by a surface mask (Figure 7.2(B)) followed by SPE anneals [21][22][23]. It can be seen that, in this case, the c-a interface is no longer planar, and this has a significant impact on the resulting crystal, with SPE proceeding in more than one direction.…”

Defective Solid-Phase Epitaxial Growth of Si

“…In fact, the templating of the growth interface and growth interface self-impingement were both shown to occur during the SPEG process in patterned material (Morarka et al, 2009). In terms of why this templating and self-impingement occurs, this is partly due to the differences in the velocities of the portions of the growth interface with [001] and [110] normal directions , but is also due to the growth interface being curved under the mask edge and being pinned at the wafer surface near the mask edge due to the presence of an oxidation layer.…”

“…However, this only explains templating along the {111} plane with normal tilted away from the mask edge. For the {111} plane with normal tilted toward the mask edge, simulations indicated the templating to be due to local variation in the growth interface velocity owing to the variable crystallographic orientation of the growth interface and the growth interface being curved (Morarka et al, 2008(Morarka et al, , 2009). However, the exact origin of this is still not entirely clear.…”

Continuum modeling of implantation and thermal processes for advanced devices formation

“…The movement of amorphous/crystalline (a/c) interfaces due to the recrystallization of amorphous Si can be described with the specific solution fields, either the distance field by the level-set method [29] [30] or the phase field by the phase field method (PFM) [31] [32]. In this paper, we briefly describe some details of the PFM as implemented in Sentaurus Process [4].…”

“…However, it is highly relevant for 2D or 3D process simulation of complete device fabrication flows, where recrystallization occurs in various directions and the amorphized region is often confined by Si/SiO 2 boundaries. The formation of nano-facets and nano-ridges with {111} surfaces [35] and its consequences on SPER near material boundaries can be modeled accurately by the level-set method [29] [30] and by lattice kinetic Monte Carlo (LKMC) simulations [36], but not yet by the continuum phase field model. Dopant redistribution using the snow plow effect might strongly depend on the crystal direction of regrowth, but unfortunately, this dependency has not been studied yet in dedicated experiments.…”