Fig. 6.1 presents the RBS spectra of the 6.0 ∙ 1016 𝑐𝑚−2 sample, including the as- implanted curve (green), the RUMP simulation curve (black), the PLM/channelled curve (blue) and the PLM/random curve (purple). The simulation curve (to best fit the random as-implanted case) shows that after implantation the total amorphous thickness is ~360 𝑛𝑚 and the Sn concentration is ~8.5 𝑎𝑡. %. After PLM, the RBS/random spectrum shows a small change in the Sn profile with the Sn distribution spread more uniformly throughout the depth of the Ge-Sn layer. The Sn concentration at this stage is estimated to be ~7.5 𝑎𝑡. % and no significant surface peak, i.e. no surface segregation, is noticeable in the spectrum. By using both the RBS/random and the RBS/channelled spectra, from chapter 2 the substitutional fraction of the implanted Sn ions is calculated to be ~82%. In the RBS/channelled spectrum, a noteworthy feature is the peak below
Ion beam synthesis and photoluminescence study of strained-relaxed Ge-Sn alloys
the surface in the Ge part of the spectrum. While this peak does suggest a region of defects below the sample surface, it will be shown in later sections that the subsurface defective regions in this sample are not associated with the large amorphous blobs illustrated in section 5.2 due to oxygen contamination. It is, rather, the mechanism for strain relaxation in the material.
Fig. 6.1: RBS spectra of the Ge-Sn samples implanted at 350 𝑘𝑒𝑉 and the dose of 6.0 ∙ 1016 𝑐𝑚−2 (8.5 at. % Sn).
Ion beam synthesis and photoluminescence study of strained-relaxed Ge-Sn alloys
Fig. 6.2: XRD-𝜔/2𝜃 scan on the (224) plane (a) and reciprocal space mapping on the (224) planes (b) of the 6.0 ∙ 1016 𝑐𝑚−2 sample.
Ion beam synthesis and photoluminescence study of strained-relaxed Ge-Sn alloys
The XRD-𝜔/2𝜃 scan on (224) planes of the 6 ∙ 1016 𝑐𝑚−2 sample is shown in Fig. 6.2(a). The Ge substrate peak is at 0 𝑠 for reference. Due to the lattice expansion, a single XRD peak of the Ge-Sn layer is at a lower Bragg angle, i.e. to the left of the substrate peak. To characterise simultaneously the in-plane and out-of-plane strain in the Ge-Sn alloys, XRD/reciprocal space mapping is employed and shown in Fig. 6.2(b). The most noteworthy feature in this figure is that the Ge-Sn peak is located along a diagonal axis relative to the Ge peak of the substrates. The full relaxation axis is determined by way of the in-plane and out-of-plane lattice constants (a and c respectively) having the same values. The RSM result therefore indicates that the Ge-Sn layer is fully relaxed because its peak is situated on the axis of full relaxation. By applying Eq. 9 in section 2.2.1.2, the lattice constant of the Ge substrate is ~5.65 Å, consistent with existing data in the literature [118], whereas, the lattice constant of the Ge-Sn layer is calculated to be ~5.7 Å. This gives the total lattice expansion of the Ge- Sn layer as compared to the Ge substrates of ~0.9%. This strain relaxation is encouraging because according to Ref. [23] fully relaxed alloys with a Sn concentration > 6.5 𝑎𝑡. % (7.5 𝑎𝑡. % from RBS/C in this sample) would be sufficient for the direct bandgap transition.
The Raman spectroscopy data are shown in Fig. 6.3. Due to the introduction of Sn into Ge lattice the 1st order Ge-Ge phonon mode shown in this figure shifts towards lower wavenumber, similar to what was found in section 5.2. The peak positions of the pristine Ge/5/6/6.8 ∙ 1016 𝑐𝑚−2 are 300.32/295.54/294.22/292.9 𝑐𝑚−1, respectively. Therefore, the peak shifts of each sample as compared to pristine Ge are 4.78 𝑐𝑚−1 for the 5 ∙ 1016 𝑐𝑚−2 sample, 6.1 𝑐𝑚−1 for the 6 ∙ 1016 𝑐𝑚−2 and 7.42 𝑐𝑚−1 for the 6.8 ∙ 1016 𝑐𝑚−2 sample.
As mentioned previously in section 5.2, peak shift in Raman spectroscopy combines an impurity component and a strain component as follows:
∆𝜔𝐺𝑒−𝑆𝑛 = 𝜔𝐺𝑒− 𝜔𝐺𝑒−𝑆𝑛 = 𝑎 ∙ ∁𝑆𝑛+ 𝑏 ∙ 𝜀𝑖𝑛−𝑝𝑙𝑎𝑛𝑒;
hence for a strain-free Ge-Sn alloy (𝜀𝑖𝑛−𝑝𝑙𝑎𝑛𝑒 = 0), it is possible to calculate the substitutional Sn concentration from the shift of the Raman peak when the correlation between the shift and the Sn concentration is known. Such a correlation has previously been investigated, such as in Ref. [112] or in Ref. [87]. The value 𝑎 of the above equation is found to be −(82 ± 4) 𝑐𝑚−1 in Ref. [112] and −(75.4 ± 4.5) 𝑐𝑚−1 in Ref. [87]. An averaged parameter calculated from these two studies would be −(78.7 ±
Ion beam synthesis and photoluminescence study of strained-relaxed Ge-Sn alloys
4.25) 𝑐𝑚−1. From the obtained peak shifts from the Raman spectroscopy data, the substitutional Sn concentration is estimated to be (6.1 ± 0.37) 𝑎𝑡. % for the 5 ∙ 1016 𝑐𝑚−2 sample, (7.75 ± 0.46) 𝑎𝑡. % for the 6 ∙ 1016 𝑐𝑚−2 sample and (9.4 ± 0.5) 𝑎𝑡. % for the 6.8 ∙ 1016 𝑐𝑚−2 sample. These calculated values from Raman spectroscopy are in good agreement with the Sn concentration obtained from the RBS data and might be sufficient for the realisation of a direct bandgap material [23].
Further analysis using a curve fitting procedure indicates that the pristine Ge and the 5/6/6.8 ∙ 1016 𝑐𝑚−2 samples exhibit a FWHM of 3.4, 5.3, 6.15 and 6.9, respectively. In section 5.2, the FWHM of the pristine sample and the PLM sample are 2.7 𝑐𝑚−1 and > 7 𝑐𝑚−1, respectively. Thus, as compared to the set of samples in section 5.2, the microscopic structure of the 350 𝑘𝑒𝑉 Sn implanted set of samples appears to be slightly improved.
Ion beam synthesis and photoluminescence study of strained-relaxed Ge-Sn alloys