Heterogeneous Integration of InGaAs Nanowires on Si(111) for Si Photonics

Abstract

1. Introduction In recent years, as the advancement of miniaturization of Si LSI in CMOS technology, the electrical connection inside and outside of the chip is facing technical limitations in terms of speed and miniaturization. To solve this problem, optical interconnection and Si photonics, in which various photonic devices are integrated on a Si platform, has been attracting much attention. One of their challenges is light sources, because Si is an indirect-gap semiconductor, and direct integration of highly efficient light emitters based, for instance, on InGaAs has a various problems, such as generation of misfit dislocations due to lattice mismatch between InGaAs and Si, and growth of polar materials on non-polar substrates. A small footprint of nanowires (NWs) grown by selective-area Metal-Organic Vapor-Phase Epitaxy (MOVPE) enables us to overcome these issues because it accommodates lattice mismatch. In this study, we investigated growth behavior of InGaAs NWs, which have compatible bandgap for optical communication band, on Si(111) substrates by selective-area growth. 2. Experimental Si(111) substrates partially masked with 20nm-thick SiO2 for NW growth were prepared by thermal oxidation, electron-beam lithography, and wet chemical etching. InGaAs NWs were grown by selective-area MOVPE, using trimethylgallium (TMGa), trimethylindium (TMIn), and arsine (AsH3) as source materials. To form (111)B polarity on non-polar Si(111) surfaces, special care was taken following Ref. [1]. After the removal of native oxide on Si surfaces by thermal cleaning in a hydrogen atmosphere at above 900°C, AsH3 was supplied for the replacement of outermost Si with As. Then, In order to form a high-quality interface of InGaAs and Si, InGaAs layers were grown firstly by flow-rate modulated epitaxy (FME), in which TMIn/TMGa and AsH3 was supplied alternately with hydrogen intervals. The growth temperature was 670°C. The intended composition of In in InGaAs was 47%, which had a bandgap corresponding to 1.55 μm, and had a lattice mismatch with Si by about 7.7%. 3. Results and discussion We found that the formation of InGaAs NWs critically depended on the diameter d of the NWs (or mask opening). When d was less than 200 nm, hexagonal structures surrounded by vertical {-110} facets appeared and vertical InGaAs NWs were formed in the most of the mask opening. This indicates that (111)B polarity was formed successfully on Si(111) surface and the growth direction of InGaAs was controlled into the vertical <111> direction. However, we also found no InGaAs growth took place in some openings. This is presumably due to a failure of initial nucleation process in some openings. It is noted that number of openings without growth increased as d decreased. On the other hand, when d was larger than 200nm, InGaAs in the hexagonal NW form decreased and anomalous lateral growth over the mask were observed. These results are understood as follows. When d is small, possibility of nucleation of InGaAs on Si is also small. Thus, single, or sometimes, no nuclei were formed in the opening for small d. As d becomes larger, possibility of nucleation increases, resulting in the increase of yield of NW growth. However, if d becomes too large, it is likely that multiple nuclei are formed [2]. Moreover, as the growth proceeds, those nuclei will coalesce but, because of the lattice mismatch between InGaAs and Si, the coalesced InGaAs are thought to contain threading dislocations. Finally, they are ended in laterally overgrown structures, as reported in Ref [2]. Thus, it is important to make sure that a single nucleus is formed for each opening for the growth of NWs. As also reported in Ref. [2], formation of nuclei on Si depends on In composition of InGaAs. In addition, the optimal growth procedure for the NWs on Si is different between InAs [1] and GaAs [3]. These findings suggest that optimal growth conditions, particularly at the initial stage of the growth, is composition-dependent for InGaAs, Thus, optimization of AsH3 thermal treatment and FME process is necessary to increase the yield of NW growth for optical communication band. 4. Summary We have achieved to grow InGaAs NWs vertically on Si(111) substrates with In composition around 47 %, and it was found that the nanowire facet growth changes due to influence of the nucleation and the lattice mismatch. Detailed investigation of InGaAs NWs by m-photoluminescence measurement and electrical characterization will be presented. 5.