Surface ripples, crosshatch pattern, and dislocation formation: Cooperating mechanisms in lattice mismatch relaxation

Abstract

We study the interplay of elastic and plastic strain relaxation of SiGe/Si(001). We show that the formation of crosshatch patterns is the result of a strain relaxation process that essentially consists of four subsequent stages: (i) elastic strain relaxation by surface ripple formation; (ii) nucleation of dislocations at the rim of the substrate followed by dislocation glide and deposition of a misfit dislocation at the interface; (iii) a locally enhanced growth rate at the strain relaxed surface above the misfit dislocations that results in ridge formation. These ridges then form a crosshatch pattern that relax strain elastically. (iv) Preferred nucleation and multiplication of dislocations in the troughs of the crosshatch pattern due to strain concentration. The preferred formation of dislocations again results in locally enhanced growth rates in the trough and thus leads to smoothing of the growth surface.

Threading and misfit-dislocation motion in molecular-beam epitaxy-grown HgCdTe epilayers

“…[6][7][8][9][10] However, in the HgCdTe/CdZnTe epitaxial system (as well as other semiconductor systems), some of the strain energy is known to be released by other mechanisms, such as the formation of surface relief or crosshatch. [15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30] Consequently, the spacing of misfit dislocations at the interface may actually be significantly larger than the equilibrium spacing predicted by Eq. 2, assuming 100% of the elastic strain is released via the formation of misfit dislocations.…”

“…In some epitaxial systems, the stress concentration located in the valley regions of the crosshatch pattern can nucleate dislocation loops that grow to form misfit segments; however, other systems have been shown to form crosshatch patterns at the exclusion of misfit dislocations, while other systems have shown the opposite tendency, where misfit formation seems to be dominant over crosshatch formation. [19][20][21][22][23][24][25] In fact, the two strain-relief mechanisms may actually be competing mechanisms; both seeking to relieve the misfit strain in epitaxial growth of HgCdTe. Many excellent papers have been published on the topic of surface relief and crosshatch formation, and the details can be found there.…”

“…Many excellent papers have been published on the topic of surface relief and crosshatch formation, and the details can be found there. [15][16][17][18][19][20][21][22][23][24][25] Both mechanisms of strain release (crosshatchsurface relief and misfit-dislocation formation) are typically kinetically driven processes and, thus, depend on the thermal history of the sample and the growth conditions under which the epilayer was grown. Consequently, the extent of the strain energy stored in the as-grown layer can conceivably vary from layer to layer, depending on the growth parameters.…”

“…The relatively low temperatures used during the MBE growth of HgCdTe allow for epitaxial layers to grow in a metastable regime exceeding the critical thickness without the formation of misfit dislocations, resulting in a final epitaxial layer that has not fully relaxed the misfit strain by the formation of misfit dislocations or surface relief (crosshatch) and exists in a metastable state of high strain. [15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30] Both the crosshatch formation and misfit-dislocation formation during growth occur along certain crystallographic directions consistent with operating-slip systems. For (111)B-oriented growth, which is typical for the liquid-phase epitaxy growth of HgCdTe, the crosshatch pattern and the misfitdislocation network that forms are parallel to the reduce the ultimate performance of the array, understanding the dislocation formation and motion mechanisms involved is crucial when attempting to improve the ultimate performance of LWIR and VLWIR devices.…”

In situ detection of an initial elastic relaxation stage during growth of In0.2Ga0.8As on GaAs(0 0 1)

“…Relaxation of low-strained layers ( < 2 %) usually takes place through misfit dislocations, and it is well known that the surface of these layers develops an undesired crosshatched morphology once the mechanisms for plastic relaxation start [3][4][5]. Furthermore, some studies on the low-mismatched SiGe/Si [6,7] and In x Ga 1-x As/InP (x  0.53) [8] systems have shown that the surface can also roughen prior to the generation of misfit dislocations and the subsequent crosshatch development. This surface roughening is associated with an elastic relaxation process and depends strongly on the growth kinetics.…”