Dislocation formation mechanism in strained InxGa1−xAs islands grown on GaAs(001) substrates

Abstract

The formation mechanism of misfit dislocations in lattice-mismatched InxGa1−xAs epilayers (0.2≤x≤1) grown on GaAs substrates has been investigated experimentally. The results suggest that 1/3〈111〉 Frank partial dislocations are grown-in at island edges in highly lattice-mismatched epilayers (x≥0.4). Then after further island growth 90° Shockley partial dislocations are nucleated to remove the stacking faults, reacting with the Frank partials to form complete 90° dislocations. An atomic model is proposed to explain the formation mechanism of the Frank partial dislocation. This model could explain the observed change in the dominant type of dislocation from the 60° at small mismatches to 90° edge dislocations at large lattice mismatches.

Alternative mechanism for misfit dislocation generation during high-temperature Ge(Si)/Si (001) island growth

“…We may further speculate that, with this model, some of the 60°perfect misfit dislocations observed could be formed by the subsequent nucleation of a 90°partial dislocation at the stacking fault surface, 12 followed by glide until the 90°p artial dislocation combines with the 30°partial dislocation to form a 60°perfect dislocation.…”

“…It has been reported that perfect dislocations can be generated by forming curved dislocation loops around the island edge, 11 by glide along ͕110͖ planes to form straight 90°sessile dislocations, 17 or by the combination of a Frank partial and a 90°partial to form a straight 90°sessile dislocation. 12 However, none of these mechanisms can apply to the Shockley partial dislocations, which could only be formed either as a grown-in dislocation or by glide on the ͕111͖ plane. For the former case, the partial dislocation and its associated stacking fault would be generated during growth at the island edge ͑the position of highest strain͒, and with further growth, the partial would be buried in the island.…”

Study of planar defect filtering in InP grown on Si by epitaxial lateral overgrowth

“…It is not clear exactly how stacking faults would form during ELOG, but incorrect depositions on strain-distorted bonds and dislocation half loops nucleating at the surface have been suggested as potential causes as mentioned earlier [17,18]. It does not appear that steric hindrance and high growth rates, as has been suggested previously [2], play a major role in the stacking fault formation since in that case stacking faults would form in the case of homoepitaxial ELOG as well.…”

“…These observations indicate that random deposition errors on facets with low stacking fault energy is an unlikely cause for the observed stacking faults, as these would be equally likely to take place in the case of ELOG on InP substrate as in ELOG on InP/Si substrate. Possibly, the formation of SFs is facilitated by residual strain in the seed layer causing distortion of bonds in the early ELOG islands, thus making incorrect depositions more likely, as has been previously observed in heteroepitaxial island growth of InGaAs on GaAs [17].…”

“…Even with this method, TDs may form during coalescence of merging growth fronts, but the same does not appear to be true for SFs [14]. Several theories aimed at explaining the formation of SFs in heteroepitaxy have been put forth [15][16][17][18], but as of yet, no study has been able to conclusively explain the presence of these defects in InP and other III-Vs on Si in ELOG layers.…”