Open Access
Add to BibSonomyAbstract
We present a method to introduce a large biaxial tensile strain in an ultra-thin germanium-on-insulator (GOI) using selective oxidation of SiGe epilayer on silicon-on-insulator (SOI) substrate. A circular patterned Si0.81Ge0.19 mesa on SOI substrate with the sidewall protected by Si3N4 or SiO2 is selectively oxidized to generate local 12 nm GOI with high crystal quality, which shows enhanced photoluminescence due to large tensile strain. Direct band photoluminescence peak significantly shifts to longer wavelength as compared to that from bulk Ge due to a combination of strain-induced band gap reduction and quantum confinement effect.
©2013 Optical Society of America
1. Introduction
Optoelectronic devices based on III-V direct-band gap semiconductors are commercial for high-performance communications systems. However, a significant challenge is that they are not compatible with Si complementary metal oxide semiconductor (CMOS) processing and require expensive substrates processed by low-throughput manufacturing. Therefore, a material with higher performance, greater energy efficiency and lower cost is crucial for the success of electronic-photonic integration on Si [1].
Germanium (Ge) has emerged as one of the most promising materials possibly to be integrated on silicon (Si) substrate for advanced metal-oxide-semiconductor field effect transistor (MOSFET) and photonic devices due to its quasi-direct band nature and much higher carrier mobility than Si. Strain in Ge will play a significant role in the enhancement of carrier mobility in advanced Ge channel MOSFET and transformation of Ge from indirect into direct band gap material. Although Ge is normally known as a poor light-emitting material since the radiative recombination through the indirect band-to-band optical transition is slow as a result of a phonon-assisted process, the Ge direct-to-indirect band separation is only 140 meV at room temperature and can be further reduced by tensile strain indicating that Ge under tensile strain attracts much attention for Si-based efficient light emission. Theoretical calculation predicted that about 1.7% in- plane biaxial tensile strain can transform Ge into direct band gap materials [2, 3].
However, it is still a big challenge to reach so large tensile strain in Ge thin films with high crystal quality using CMOS compatible technology. As is well known, Ge epilayer grown directly on Si substrate is able to obtain the maximum of 0.25% tensile strain in Ge due to thermal expansion coefficient mismatch between Ge and Si, in which light emission is achieved from the strained Ge epilayer on Si substrate utilizing an indirect valley filling effect with heavy n-type doping [4–6]. Using micromechanical technique, tensile strain as large as 1.13% in a 1.6 μm thick Ge membrane is achieved with a tungsten stressor and enhanced photoluminescence is observed with about 100 meV direct band reduction [7]. 0.4% tensile strain in Ge photonic wire (500 nm thick) grown on GaAs substrate was reported and several tens of optical gain was derived from the variable strip length method [8]. Recently, more than 2% tensile strain was demonstrated in Ge grown on lattice relaxed InGaAs/GaAs buffer layers by molecular beam epitaxy [9] and a thin Ge layer on a polyimide film using high-pressure gas [10]. However, those techniques are limited to be integrated with Si technology. GOI substrates made by layer transfer techniques or Ge condensation method presented only 0.19% maximum tensile strains [11–13].
In this work, we report a biaxial tensile strain as large as 0.67% in an ultra-thin (12 nm) GOI fabricated by locally selective oxidation of SiGe layer on SOI substrate. Film strain was evaluated by Raman spectra analysis and photoluminescence, and crystal quality of Ge layer was investigated by High resolution transmission electron microscopy (HRTEM). Room temperature photoluminescence showed a large direct band gap reduction compared with that of bulk Ge. Our work opens up the possibility of easy fabrication to ultra- thin strained GOI materials for both optical and electronic applications.
2. Material fabrication and its characterization
The substrate used in this experiment is a 4 in. silicon-on-insulator (SOI) wafer fabricated by SIMOX technology with a 38 nm top Si layer and a 380 nm BOX layer. A 60 nm Si buffer layer was grown on the substrate at 600 °C, followed by a 70 nm Si0.81Ge0.19 epilayer with a 6 nm Si cap layer at 550 °C in an ultra high vacuum chemical vapor deposition system with a base pressure of 5 x 10−8 Pa. After growth, circular SiGe mesas with diameters of 32 μm were defined by photolithography and then formed by reactive ion etching which removed all the unprotected Si and SiGe layers above the buried oxide. The etched SiGe mesa samples are divided into two groups labeled by “A” and “B”. For sample A, SiO2 film was deposited at 100 °C by plasma enhanced chemical vapor deposition to enclosure the SiGe mesas from the sidewalls. For sample B, Si3N4 film was used instead of the SiO2. After SiO2/Si3N4 deposition, all the samples were selectively oxidized and annealed at 1150 °C for Ge content less than 0.4 and then the oxide was removed to 5 nm by dilute hydrofluoric acid. Once the Ge content reached to 0.4, condensation experiments were further carried out at 900 °C for Ge fraction larger than 0.4 with several cycles to achieve the ultra-thin GOI materials. All the oxidation or annealing processes were carried out in a conventional tube furnace with pure O2 or N2 gas flow. The schematics of the ultra-thin GOI fabrication processes from sample A and B are shown in Fig. 1 . For comparison, one sample of SiGe on SOI substrate without defining patterns was also fabricated under the same condition, labeled as sample C. The thickness of the oxide layer and the SiGe layer during the Ge condensation processes were measured by ellipsometry and the Ge content in the SiGe layer was measured by Raman spectroscopy.
Fig. 1 The schematics of fabrication process of ultra-thin GOI. The mesa of SiGe on SOI were etched and protected by SiO2 (a) and Si3N4 (b), respectively, and then selectively oxidized to generate ultra-thin GOI materials. (c) Sample C without defining patterns was for comparison.
During the oxidation of Si1-xGex alloy, Ge is rejected from the oxide and piles up at the oxidizing interface and diffuses towards substrate, which is blocked by buried oxide layer to form a GOI substrate [11, 14, 15]. As a result, the Ge content in the SiGe layer increases up to the values 1. The total amounts of Ge atoms were estimated by the Ge content and SiGe thickness. The dependence of the SiGe layer thickness and the average Ge content on the oxide thickness during the Ge condensation processes is shown in Fig. 2 . As the oxide increases during the Ge condensation processes, the SiGe thickness decreases linearly since the Si is selectively oxidized to form SiO2, while Ge is ejected from the SiO2 and accumulates in the region just beneath the oxide until to form pure Ge. The values of the final SiGe thickness and the final average Ge content were found to be close to the calculated results, indicating that the total amount of Ge was conserved within the SiGe layer.
Fig. 2 Dependence of the SiGe layer thickness and the average Ge content on the oxide thickness during the Ge condensation processes. The solid lines are the theory results.
In order to evaluate the final Ge thickness more precisely and characterize the morphology of the ultra-thin GOI, HRTEM is illustrated in Fig. 3 . The condensed thin GOI layer with the thickness of 12 nm is quit uniform with perfect lattice. The thickness of 12 nm GOI is slightly thinner than the calculated value of 13 nm supposing all of the Ge atoms in 70 nm Si0.81Ge0.19 layer are condensed, implying that a small amount of Ge atoms might be incorporated into SiO2 layer or dissipation during Ge condensation processes. The HRTEM images manifested the high crystalline qualities of the GOI layer.
Fig. 3 HRTEM images of the ultra-thin GOI substrates. 12 nm GOI is uniform with a clear and sharp interface between BOX and SiO2.
3. Tensile strain in GOI and its optical properties
Raman spectroscopy was carried out to evaluate the strain in the Ge layer by the following Eq [16],
where is the Raman shift of the Ge-Ge mode in bulk Ge, which is of 300.60 cm−1 fitted by Gaussian distribution. is the Ge-Ge mode in the strained Ge and is the biaxial strain in Ge. Figure 4 shows Raman spectra for sample A, B, C and bulk Ge. The sample A, B and C are assumed to have the same thickness of 12 nm since they were oxidized with the same process. Bulk Ge of 300 μm thickness is also included in Fig. 4 for comparison. The inset is the whole Raman spectra from the sample A, in which Ge-Ge mode and Si-Si mode are located at 297.89 cm−1 and at 520.60 cm−1, respectively. The absence of Si-Ge mode means that the Si content in the GOI should be lower than the detection limit of the Raman spectrum (less than 0.5% Si atoms) [15]. Compared to bulk Ge, the peak position shifts to shorter wavenumber as tensile strain was introduced in the sample A, B and C. In this case, Ge-Ge position from sample A shifts to shorter wavenumber by about 2.71 cm−1 as compared to bulk Ge. Calculation by formula (1) shows that the Ge layer of sample A is under a tensile strain of about 0.67% ± 0.02%. For the sample B and C, the peaks of Ge-Ge mode are located at 298.41 cm−1 and 299.34 cm−1, which correspond to 0.54% ± 0.02% and 0.31% ± 0.02% biaxial tensile strains in the Ge layers, respectively.Fig. 4 Raman spectra from sample A, B, C and bulk Ge for comparison to evaluation of strain in the Ge layer. The inset is the whole Raman spectra from the sample A.
Room temperature photoluminescence of the samples was measured with an InGaAs detector array cooled by liquid nitrogen. The exciting laser is 25 mW in power and 488 nm in wavelength. As shown in Fig. 5 , the intensities of the photoluminescence from the 12-nm-thick strained GOI in sample A and sample B are comparable with that of 300 μm thick bulk Ge, but are much stronger than that of sample C by about one order of magnitude. Although the photoluminescence spectrum is cut off at 1570 nm due to the limitation of the InGaAs detector, the remarkable red shift of the photoluminescence peaks around 1600 cm−1 can be clearly observed for sample A and B as compared with the peak position for bulk Ge. With the direct band gap recombination model [17], due to the 1570 nm detection limitation of our photoluminescence system, only the shorter wavelength parts of the spectra are fitted for sample A and B as shown in Fig. 5. The photoluminescence peak shifts from 1520 nm for bulk Ge to 1636 nm and 1606 nm for sample A and B, respectively. However, the theory predicted that the direct band photoluminescence peak should be positioned at 1733 nm and 1686 nm for 0.67% and 0.54% tensile strained Ge, respectively.
Fig. 5 Room temperature direct band photoluminescence from sample A, B, C and bulk Ge for comparison.
Actually, to interpret this difference, the 12 nm Ge layer should also be considered as a quantum well as it is clipped between buried SiO2 and SiO2 top layers. We assumed the Ge/SiO2 conduction /valence band offset of 3.2 eV /4.9 eV, and the SiO2 effective masses of density of states of 0.86m0 [18] and 0.33m0 [19] for conduction and valence bands. And with the known effective masses of density of states for Ge (0.038m0, 0.043m0, and 0.284m0 for , light-hole, and heavy-hole bands, respectively [2], where m0 is the free electron mass), the calculation shows that the transition energy from the ground states of the conduction band to the heavy-hole valence band in a 12-nm-thick Ge quantum well should shift to higher energy by about 43.0 meV due to quantum confinement effect. On the other hand, the tensile strain in Ge layer caused by direct band gap reduction at the same time which give 45.6 meV, 80.3 meV and 100.2 meV energy reductions for 0.31%, 0.54% and 0.67% tensile strained Ge. Considering both effects, the final direct band transition energies are reduced by about 2.6 meV, 37.3 meV and 57.2 meV, respectively. From the band gap reduction calculation, the predicted photoluminescence peak shift is plotted in Fig. 6 as a function of the tensile strain from zero to 1.0%. The prediction is in good agreement with our experimental data.
Fig. 6 Peak energy shift as a function of strain in Ge film. The dashed line shows the theoretical results.
4. Discussion
With those experimental data, finite element calculations are used to study strain fields in the ultra-thin germanium- on- insulator. Figures 7(a) and 7(b) show the calculated stress distribution of the diameter of 32 μm Ge mesa with the sidewall surrounded by SiO2 and Si3N4, respectively. One observes from top to bottom the top oxide, the 12 nm tensile-strained Ge layer, and the BOX layer. The thermal expansion mismatch processes are simulated from 900 °C cooled down to room temperature. Comparing Fig. 7(a) with that of Fig. 7(b) shows that the strain in the Ge layer with the sidewall surrounded by SiO2 (0.83%) is larger than that surrounded by Si3N4 (0.77%). This is mostly expected from the larger difference of the expansion coefficients between Ge (8.5 x 10-6 oC−1) [20, 21] and SiO2 (5.0 x 10-7 oC−1) [22] on the sidewall than that between Ge (8.5 x 10-6 oC−1) and Si3N4 (3.3 x 10-6 oC−1) [23]. The simulations which predict SiO2 dielectric will contribute to larger tensile strain in ultra-thin GOI than that of Si3N4 dielectric are in good agreement with our experimental results. We also calculate the stress distribution of Ge layer with different Ge mesa sizes, which show that the strain will be higher if the size is reduced to several hundred nanometers.
Fig. 7 Finite element calculations of the diameter of 32 μm Ge mesa with the sidewall surrounded by SiO2 (a) and Si3N4 (b), respectively.
5. Conclusion
A method was proposed to introduce a biaxial tensile strain as large as 0.67% in an ultra-thin germanium-on-insulator (GOI) using locally selective oxidation of SiGe epilayer on silicon-on- insulator (SOI) substrate. Selective oxidation of circular patterned SiGe mesa on SOI with the sidewall protected by SiO2 or Si3N4 were carried out to achieve 12 nm high crystal quality tensile strained GOI materials due to the thermal mismatch between Ge and dielectrics. Direct band photoluminescence of GOI was significantly enhanced and the peak position shifts to longer wavelength due to the tensile strain-induced band gap reduction and quantum confinement effect. The processes used to fabricate the GOI are completely compatible with Si technology and the significant increase of tensile strain in the ultra-thin GOI is necessary for improvement of carrier mobility in advanced Ge channel MOSFETs and Si-based Ge light emission efficiency.
Acknowledgments
The authors would like to thank Dr Yuhua Zuo from Institute of Semiconductors, CAS for her help in PL measurements. This work was supported by the National Basic Research Program of China under grant No. 2012CB933503, 2013CB632103, the National Natural Science Foundation of China under grant No. 61176092, 61036003, 60837001, Ph.D. Programs Foundation of Ministry of Education of China under grant No. 20110121110025 and the Fundamental Research Funds for the Central Universities under grant No. 2010121056.
References and links
1. R. A. Soref, “The past, present, and future of silicon photonics,” IEEE J. Sel. Top. Quantum Electron. 12(6), 1678–1687 (2006). [CrossRef]
2. J. Liu, X. Sun, D. Pan, X. Wang, L. C. Kimerling, T. L. Koch, and J. Michel, “Tensile-strained, n-type Ge as a gain medium for monolithic laser integration on Si,” Opt. Express 15(18), 11272–11277 (2007). [CrossRef] [PubMed]
3. C. G. Van de Walle; “Band lineups and deformation potentials in the model-solid theory,” Phys. Rev. B Condens. Matter 39(3), 1871–1883 (1989). [CrossRef] [PubMed]
4. C. Li, Y. Chen, Z. Zhou, H. Lai, and S. Chen, “Enhanced photoluminescence of strained Ge with a δ-doping SiGe layer on silicon and silicon-on-insulator,” Appl. Phys. Lett. 95(25), 251102 (2009). [CrossRef]
5. J. Liu, X. Sun, P. Becla, L. C. Kimerling, and J. Michel, “Towards a Ge-based laser for CMOS applications,” in Proceedings of 5th IEEE International. Conference on Group IV Photonics (Institute of Electrical and Electronics Engineers, Italy, 2008), pp. 16–18.
6. S. L. Cheng, J. Lu, G. Shambat, H. Y. Yu, K. Saraswat, J. Vuckovic, and Y. Nishi, “Room temperature 1.6 microm electroluminescence from Ge light emitting diode on Si substrate,” Opt. Express 17(12), 10019–10024 (2009). [CrossRef] [PubMed]
7. D. Nam, D. Sukhdeo, A. Roy, K. Balram, S. L. Cheng, K. C. Y. Huang, Z. Yuan, M. Brongersma, Y. Nishi, D. Miller, and K. Saraswat, “Strained germanium thin film membrane on silicon substrate for optoelectronics,” Opt. Express 19(27), 25866–25872 (2011). [CrossRef] [PubMed]
8. M. de Kersauson, M. E. Kurdi, S. David, X. Checoury, G. Fishman, S. Sauvage, R. Jakomin, G. Beaudoin, I. Sagnes, and P. Boucaud, “Optical gain in single tensile-strained germanium photonic wire,” Opt. Express 19(19), 17925–17934 (2011). [CrossRef] [PubMed]
9. Y. Huo, H. Lin, Y. Rong, M. Makarova, M. Li, R. Chen, T. Kamins, J. Vuckovic, and J. Harris, “Efficient luminescence in highly tensile-strained germanium,” IEEE Int. Conf. on Group IV Photonics, 265–267 (2009).
10. C. Boztug, F. Chen, J. Sanchez-Perez, F. Sudradjat, D. Paskiewicz, R. Jacobson, M. Lagally, and R. Paiella, “Direct-bandgap germanium active layers pumped above transparency based on tensilely strained nanomembranes,” CLEO:2011, PDPA2 (2011).
11. T. Tezuka, N. Sugiyama, and S. Takagi, “Fabrication of strained Si on an ultrathin SiGe-on-insulator virtual substrate with a high-Ge fraction,” Appl. Phys. Lett. 79(12), 1798 (2001). [CrossRef]
12. J. R. Jain, D. S. Ly-Gagnon, K. C. Balram, J. S. White, M. L. Brongersma, D. A. B. Miller, and R. T. Howe, “Tensile-strained germanium-on-insulator substrate fabrication for silicon-compatible optoelectronics,” Opt. Mater. Express 1(6), 1121–1126 (2011). [CrossRef]
13. Y. Hoshi, K. Sawano, K. Hamaya, M. Miyao, and Y. Shiraki, “Formation of tensilely strained Germanium-on-Insulator,” Appl. Phys. Express 5(1), 015701 (2012). [CrossRef]
14. S. Nakaharai, T. Tezuka, N. Sugiyama, Y. Moriyama, and S. Takagi, “Characterization of 7-nm-thick strained Ge-on-insulator layer fabricated by Ge-condensation technique,” Appl. Phys. Lett. 83(17), 3516 (2003). [CrossRef]
15. S. Nakaharai, T. Tezuka, N. Hirashita, E. Toyoda, Y. Moriyama, N. Sugiyama, and S. Takagi, “Formation process of high-purity Ge-on-insulator layers by Ge-condensation technique,” J. Appl. Phys. 105(2), 024515–024518 (2009). [CrossRef]
16. P. H. Tan, K. Brunner, D. Bougeard, and G. Abstreiter, “Raman characterization of strain and composition in small-sized self-assembled Si/Ge dots,” Phys. Rev. B 68(12), 125302 (2003). [CrossRef]
17. M. El Kurdi, T. Kociniewski, T.-P. Ngo, J. Boulmer, D. Débarre, P. Boucaud, J. F. Damlencourt, O. Kermarrec, and D. Bensahel, “Enhanced photoluminescence of heavily n-doped germanium,” Appl. Phys. Lett. 94(19), 191107 (2009). [CrossRef]
18. S. Zafar, K. A. Conrad, Q. Liu, E. A. Irene, G. Hames, R. Kuehn, and J. J. Wortman, “Thickness and effective electron mass measurements for thin silicon dioxide films using tunneling current oscillations,” Appl. Phys. Lett. 67(7), 1031 (1995). [CrossRef]
19. M. I. Vexler, S. E. Tyaginov, and A. F. Shulekin, “Determination of the hole effective mass in thin silicon dioxide film by means of an analysis of characteristics of a MOS tunnel emitter transistor,” J. Phys. Condens. Matter 17(50), 8057–8068 (2005). [CrossRef]
20. G. A. Slack and S. F. Bartram, “Thermal expansion of some diamondlike crystals,” J. Appl. Phys. 46(1), 89 (1975). [CrossRef]
21. D. D. Cannon, J. Liu, Y. Ishikawa, K. Wada, D. T. Danielson, S. Jongthammanurak, J. Michel, and L. C. Kimerling, “Tensile strained epitaxial Ge films on Si (100) substrates with potential application in L-band telecommunications,” Appl. Phys. Lett. 84(6), 906–908 (2004). [CrossRef]
22. H. Tada, A. E. Kumpel, R. E. Lathrop, J. B. Slanina, P. Nieva, P. Zavracky, I. N. Miaoulis, and P. Y. Wong, “Thermal expansion coefficient of polycrystalline silicon and silicon dioxide thin films at high temperatures,” J. Appl. Phys. 87(9), 4189 (2000). [CrossRef]
23. J. Jiang, H. Lindelov, L. Gerward, K. Ståhl, J. Recio, P. Mori-Sanchez, S. Carlson, M. Mezouar, E. Dooryhee, A. Fitch, and D. Frost, “Compressibility and thermal expansion of cubic silicon nitride,” Phys. Rev. B 65(16), 161202 (2002). [CrossRef]
