Growth dynamics of SiGe nanowires by the vapour-liquid-solid method and its impact on SiGe/Si axial heterojunction abruptness

Abstract

The vapour–liquid–solid (VLS) method is by far the most extended procedure for bottom-up nanowire growth. This method also allows for the manufacture of nanowire axial heterojunctions in a straightforward way. To do this, during the growth process, precursor gases are switched on/off to obtain the desired change in the nanowire composition. Using this technique, axially heterostructured nanowires can be grown, which are crucial for the fabrication of electronic and optoelectronic devices. SiGe/Si nanowires are compatible with complementary metal oxide semiconductor (CMOS) technology, which improves their versatility and the possibility of integration with current electronic technologies. Abrupt heterointerfaces are fundamental for the development and correct operation of electronic and optoelectronic devices. Unfortunately, the VLS growth of SiGe/Si heterojunctions does not provide abrupt transitions because of the high solubility of group IV semiconductors in Au, with the corresponding reservoir effect that precludes the growth of sharp interfaces. In this work, we studied the growth dynamics of SiGe/Si heterojunctions based on already developed models for VLS growth. A composition map of the Si–Ge–Au liquid alloy is proposed to better understand the impact of the growing conditions on the nanowire growth process and the heterojunction formation. The solution of our model provides heterojunction profiles that are in good agreement with the experimental measurements. Finally, an in-depth study of the composition map provides a practical approach to the drastic reduction of heterojunction abruptness by reducing the Si and Ge concentrations in the catalyst droplet. This converges with previous approaches, which use catalysts aiming to reduce the solubility of the atomic species. This analysis opens new paths to the reduction of heterojunction abruptness using Au catalysts, but the model can be naturally extended to other catalysts and semiconductors.