1. Introduction

SiGe is a promising channel material due to its high carrier mobilities [1]. Strain engineering is also effective for the SiGe channel, in which. Ge or C implantation in the source and drain (S/D) region is effective to induce compressive or tensile strain, respectively. After implantation, however, the samples are necessary to be thermally annealed to remove lattice damage. Moreover, considering the existing impurity profiles such as B and P, the annealing process has to be minimum thermal budget. Thus, we focused on a laser annealing (LA) process. LA can remove lattice damage efficiently and minimize the thermal budget compare to the other thermal processes [2-3]. However, the correlation between LA conditions and strain has not been reported in detail.

In this study, we demonstrated strain evaluation in laser-annealed SiGe thin layers with and without C or Ge ion implantation by Raman spectroscopy.

2. Experiment

The 20 nm thick Si_0.55Ge_0.45 films were grown on (001) Si substrates. After P and B ion implantation and rapid thermal annealing to simulate impurity profile, C or Ge are ion implanted. Implantation conditions are multiple step with 2keV/1E15, 4keV/2E15 and 8keV/7E15 for C, and single step of 3keV/5E16 for Ge. The samples with and without C or Ge implantation were then laser-annealed with the conditions of 0.6, 1.7, 2.4 J/cm² laser power. Raman spectroscopy and X-ray photoelectron spectroscopy (XPS) were carried out for the evaluation. In the Raman measurement, the spectrometer with the focal length of 2,000 mm and excitation source of UV (l = 355 nm) laser were used. In the XPS measurement, X-ray source was Al-Ka (1486.6 eV) and Ar⁺ sputtering was performed to evaluate depth profile.

3. Results and discussion

Figure 1 shows Ge concentration profiles before and after LA measured by XPS. As shown in Fig. 1, we confirmed that Ge redistribution depending on LA power. For 0.6 J/cm², Ge were redistributed only at the sample surface. For 1.7 J/cm², Ge were redistributed not only at the surface but also at the depth of approximately 13 nm. However, for 2.4 J/cm², Ge redistributed far deep in the Si substrate beyond SiGe layer probably due to the deep melting. Then, we estimated the strain in the film by using the following equation,

(1), (Ref. [4]), where x is Ge average concentration between the surface and 20 nm depth, and is in-plane strain.

Figure 2 shows strain depending on the LA power measured by Raman spectroscopy. As shown in Fig. 2, we confirmed that all samples were compressively strained. Up to 1.7 J/cm², the compressive strain increased as laser power increased. However, for 2.4 J/cm², the compressive strain was lower than that for 0.6 J/cm² probably due to the Ge redistribution.

Figure 3 shows comparison of Raman peak shift with and without C ion implantation after LA at 1.7 J/cm². As shown in Fig. 3, we confirmed that the peak of C implanted SiGe were shifted even lower wavenumber than that without C implantation. Therefore, C implanted SiGe may be strained compressively more than that without implantation. However, high C concentration even without strain can also be a cause in the Raman shift as described for the case of Ge incorporation by equation (1), therefore, it should be carefully discussed, hopefully with more characterization data.

Acknowledgements

The authors thank to Dr. N. Sawamoto for her support in Transmission Electron Microscopy observation.

Reference

[1] Jacopo Franco. et al., IEEE Transactions on Electron Devices, 60, 1 (2013).

[2] C. W. White. et al., SCIENCE. 204, 4392 (1979).

[3] R. T. Young. et al., Appl. Phys. Lett. 32, 139 (1978).

[4] T. S. Perova. et al., J. Appl. Phys, 109, 033502 (2011).