First, it is worth noting that the deposition mechanism of Ge using GeCl4:C3H8O2 is considered to be the same with the deposition of Ge on Si substrate. From the cyclic voltammogram study, it shows two reduction processes where Ge (IV) is reduced to Ge(II) prior to the deposition of Ge on substrate as summarized by Eqs.(1) and (2).
Fig.1(a)shows the FESEM image of the patterned Ge microstrip before annealing where the morphology shows continuous and uniform strips. Based on the EDX spectra (data not shown), the
deposited Ge film was found to be highly pure without any excessive contaminants such as Pt metal. Fig. 1(b) shows the color coded normal direction (ND) EBSD images of the Ge microstrip where it can be understood from the random distribution of colors that the crystal structure is amorphous.
Fig.2 shows the Raman spectra of the as-deposited layer and the heat treated Ge microstrip. It can be seen that no significant Ge–Ge vibration mode peak was observed for the as-deposited Ge,
thus confirming that the structure was amorphous. From the Raman spectra, the main peak at 300 cm1 which corresponds to Ge–Ge vibration mode was clearly observed after the rapid annealing process, confirming the crystallization of Ge. The value of full width half maximum (FWHM) of this peak was estimated to be around 3.4cm1, which is very close to the value of the standard single crystalline bulk Ge wafer of 3.1cm1, as shown in Fig. 2.
Fig. 3(a) shows the FESEM image of heat treated Ge microstrip. Due to severe agglomeration, the strip' slength shrunk to be about 70 mm. Fig.3(b) shows the lateral strain profile evaluated by the Raman line scan of the heat treated Ge strip for the length of 60 mm, where negative values indicate compressive strains. Here, it clearly shows large compressive strain has been generated up to 0.13%.The EBSD image, as shown in Fig.3(c), indicates the single-crystalline (111)oriented Ge strip with 70 mm length and 15 mm width. It was reported that only several mm Ge grain was grown without any seeding area. Therefore, it can be concluded that MLG is playing the key role to enhance single-crystallization of (111) oriented Ge.Three locations which indicated the high compressive strain, medium strain, and low strain were chosen for the HRTEM observation.Fig.3(d) shows the HRTEM image of the location with the high compressive strain. Here, the introduction of C atoms into the grown Ge layer was clearly recognized.Fig.3(e) shows the HRTEM image of the location with the medium compressive strain. Here,it shows less introduction of C atoms in the crystalline Ge.
Fig.3(f) shows the HRTEM image of the location with the low compressive strain. No structure of graphene was observed, suggesting that the measured area does not have a coverage of graphene since the total coverage of graphene on the entire substrate was initially 95%. The interface of crystalline Ge and SiO2 was clearly observed.The results seem to suggest that the compressive
strain increases with the introduction of C atoms into the Ge layer. It was shown in Fig.3(f) that crystalline Ge was observed on SiO2 where there is no graphene layer. In general, the crystallization of large grain single-crystal Ge on bare SiO2 is not possible to be obtained. This strongly suggests that the induction of lateral growth could start from the intermixing of Ge/graphene region.The region of the mixing of C atoms from the MLG and Ge is speculated to be the possible nucleation region of Ge (111) and the lateral growth has been induced from this nucleation region probably by the spatial gradient of solidification temperature strips grown by Si-seeded RMG, where Si (100) substrates were employed as crystal seed.Similarly, high carrier mobility(about 1000cm2/V s)is also expected for this orientation-controlled single-crystal GOI strips obtained by the present study.These carrier mobilities are much higher than that for poly-Ge(about 140cm2/V s) obtained by a conventional seed-free processsuch as a solid-phase crystallization.