Stressed multidirectional solid-phase epitaxial growth of Si

Abstract

The study of the solid-phase epitaxial growth (SPEG) process of Si (variously referred to as solid-phase epitaxy, solid-phase epitaxial regrowth, solid-phase epitaxial crystallization, and solid-phase epitaxial recrystallization) amorphized via ion implantation has been a topic of fundamental and technological importance for several decades. Overwhelmingly, SPEG has been studied (and viewed) as a single-directional process where an advancing growth front between amorphous and crystalline Si phases only has one specific crystallographic orientation. However, as it pertains to device processing, SPEG must actually be considered as multidirectional (or patterned) rather than bulk in nature with the evolving growth interface having multiple crystallographic orientations. Moreover, due to the increasingly ubiquitous nature of stresses presented during typical Si-based device fabrication, there is great interest in specifically studying the stressed-SPEG process. This work reviews the progress made in understanding the multidirectional SPEG and, more importantly, stressed multidirectional SPEG process. For the work reviewed herein, (001) Si wafers with ⟨110⟩-aligned, intrinsically stressed Si3N4/SiO2 patterning consisting of square and line structures were used with unmasked regions of the Si substrate amorphized via ion implantation. It is revealed that the stresses generated in the Si substrate from the patterning, both in line and square structures, alter the kinetics and geometry of the multidirectional SPEG process and can influence the formation of mask-edge defects which form during growth to different degrees as per differences in the substrate stresses generated by each type of patterning. Likewise, it is shown that application of external stress from wafer bending during SPEG in specimens with and without patterning can also influence the geometry of the evolving growth interface. Finally, the effect of the addition of SPEG-enhancing impurities during multidirectional stressed growth is observed to alter the evolution of the growth interface, thus suggesting that stress influences on growth are much less than those from dopants. Within the context of prior work, attempts are made to correlate the prior observations in single-directional stressed SPEG with the observations from patterned stressed SPEG reviewed herein. However, as is argued in this review, it ultimately appears that much of the research performed on understanding the single-directional stressed-SPEG process cannot be reasonably extended to the multidirectional stressed-SPEG process.