From porous Si to patterned Si substrate: Can misfit strain energy in a continuous heteroepitaxial film be reduced?

The use of patterned Si substrates (specifically porous Si) for the reduction of heteroepitaxial film strain energy [S. Luryi and E. Suhir, Appl. Phys. Lett. 49, 140 (1986)] has been studied recently by several research groups. We report experimental studies and discuss the validity of the original concept. We studied molecular beam epitaxial growth of continuous GexSi1−x film on 〈100〉 porous Si substrates using Rutherford backscattering spectrometry, transmission electron microscopy, and x-ray rocking curves. The results show predominantly 60° dislocations with long misfit segments. There is no reduction in either strain in the films or dislocation density compared to the samples grown on regular Si substrates. In order to reduce defect densities, the stress fields from different growth areas of patterned substrate must not interfere. Arguments based on crystallographic considerations are presented which demonstrates that, first, the porous Si substrate is not suitable for such applications because of the nature of interconnected growth areas, and second, any kind of decaying stress field requires a certain degree of wafer warpage which does not always present in a realistic MBE-grown heterostructure. It is the conclusion of this discussion that the strain energy in a continuous heteroepitaxial film grown on patterned substrates does not reach a limiting value, but continues to increase with the film thickness.