Table of Contents

Cleaving Oxide Films Using Hydrogen Implantation S. Thevuthasan, W. Jiang,(a,b) and W. J. Weber(b) Supported by EMSP, BES, EMSL Operations. (a) Postdoctoral Research Fellow. (b) Environmental and Health Sciences Division. Combined materials with different structures can be efficiently used in several applications related to power devices including semiconductor components, laser diodes, and solar cells. Traditionally wafer-bonding techniques have been used to join these materials together. The Smart-Cut process using hydrogen or helium implantation and subsequent heating is an emerging technology for transferring single-crystal semiconductor device layers from one substrate to another. It has been reported that this technique can be applied in fabricating silicon on insulator (SOI) layers and other electronic devices (Bruel 1995; Aspar et al. 1997, 1999; Agarwal et al. 1998). The Smart-Cut process was first reported by Bruel (1995) as a process in which heat treatment induces an in-depth microslicing of one of the two bonded wafers previously implanted with hydrogen. Subsequently, the mechanisms related to this process were investigated in the case of Si materials, and it was found that the process is controlled by hydrogen diffusion at high temperatures. On the other hand, the process is controlled by dissociation of Si-H bonds and atomic hydrogen diffusion at low temperatures. Recently, this process has been successfully applied to form silicon carbide on insulator structures (Di Cioccio et al. 1997) using polycrystalline SiC materials. Although this process has been successfully demonstrated on polycrystalline SiC substrates, the details of the mechanisms are not well established, and could be similar to those reported for Si materials. Since the Smart-Cut technology is in its early stage of development, application of this technology to other materials except Si and SiC is not known. The technological importance of oxide materials is extensive in many different areas including catalysis, gas sensors and high-Tc superconductors. Application of the Smart-Cut technology in oxide materials is an attractive process to generate thin oxide films with precise thickness on suitable substrates. In this study (Thevuthasan et al. in preparation), the first attempt was made to understand this process using SrTiO3(100) substrate as a model oxide material. A brief summary is reported in this extended abstract. Hydrogen implantation at 120 K to fluences of 5.0x1016 H+/cm2 was carried out at Implant Science Corporation. The surface orientation was (100), and 40 KeV H+ ions were used to implant the samples at a direction of 7° relative to the surface normal. There may have been a small fluctuation in the sample temperature due to the sample heating during the implantation. The damage recovery experiments were carried out in the accelerator facility at the EMSL at PNNL. The accelerator facility and the end stations are described in detail elsewhere (Thevuthasan et al. 1999). Isochronal annealing for 20-minute periods was performed at temperatures from 370 K to 570 K at 100 K intervals under high vacuum conditions with ±  5 K uncertainty in the temperature. After each annealing step, the sample was cooled down, and the RBS/C and hydrogen NRA measurements were carried out at a sample temperature of 300 K using 2.0 MeV He+ and 6.40-8.04 MeV F3+ ions. The hydrogen NRA measurements were performed using the resonant 1H(19F,ag)16O reaction, and the sample was in random orientation during the hydrogen profile measurements. At the end of 870 K annealing, the samples were removed from the ion scattering chamber and SEM experiments were performed. The channeling spectra from the irradiated and virgin regions are presented as a function of annealing temperature, along with the random spectrum, in Figure 2.8 for an ion fluence of 5.0x1016 H+/cm2. Only the RBS/C spectra from the irradiated region for 300 K, 470 K, and 570 K annealing are shown to minimize the interference between different spectra. As expected, the damage produced on both the Ti and Sr sublattices is rather dilute and may overlap somewhat, perhaps due to diffusion at the implantation temperature (300 K RBS/C spectrum). The hydrogen concentration profile measurements using the hydrogen NRA after annealing at 300 K, 470 K, and 570 K are presented in Figure 2.9. The integrated hydrogen content is 4.9x1016 H/cm2, which is consistent with the ion fluence. According to TRIM calculations, the damage peak and the peak of the hydrogen profile are expected at 275 nm and 300 nm from the surface, respectively. The experimental hydrogen profile is somewhat broad due to diffusion, and a small peak is visible around 300 nm from the surface. Annealing the sample to 370 K resulted in a slight increase in the backscattering yield from the Sr and Ti sublattices (not shown) suggesting that there is some increased disordering during annealing due to the possible interaction of H with the structure or the formation of defect clusters or bubble nuclei. The hydrogen profile after the 370 K heating appears to be peaked at the damaged peak (not shown). Figure 2.8. The channeling spectra from the virgin and 40 keV H+ (fluence ~ 5.0x1016 H+/cm2) irradiated region from SrTiO3 after 300, 470, and 570 K isochronal annealing along with the random spectrum are presented. Figure 2.9. The hydrogen profiles measured by the resonant 1H(19F,a g)16O reaction for the ion fluence of 5.0x1016 H+/cm2 are presented. The profiles were taken after 300 K, 470 K, and 570 K isochronal annealing. The damage generated in the Sr and Ti sublattices was significantly increased near the damaged peak position at 470 K annealing (Figure 2.8). The hydrogen profile data shows some hydrogen diffusion towards the surface with an enhanced peak around 275 nm from the surface. One possibility for the unusual increase in the Sr and Ti backscattering yield is the formation of hydrogen-defect clusters or H2 bubble nuclei at this annealing temperature, which could introduce sufficient strain and new surfaces to increase the backscattering for Sr and Ti sublattices near the damage peak. This interpretation is consistent with the observed shift in the hydrogen profile to coincide with the defect (vacancy) distribution. The annealing at 570 K shows significant increase in the surface peak area and a slight increase in the backscattered He+ signal from the region between the surface peak and the damage peak region. The hydrogen profile after this annealing shows a significant loss of hydrogen from the implanted region. The hydrogen NRA signal is almost in the background throughout the analyzed region of 400 nm depth from the surface. No significant difference in the aligned spectrum was observed after annealing the sample from 670 K-870 K. Figure 2.10 shows the SEM micrograph of the 5.0x1016 H+/cm2 irradiated sample after annealing at 870 K. This micrograph shows that a major portion of the surface layer has been removed from the sample. The remaining surface layer piece is approximately 300-nm thick and this agrees with the implantation depth from the surface. The cleaved surface appears to be slightly rough compared to the original polished surface. This is confirmed by the surface peak area increase in the aligned spectrum after annealing at 570 K (Figure 2.8). Since there were no significant differences between the aligned spectra after annealing at 570 K and 870 K, it appears that the cleavage of surface layer occurred after the 570 K annealing. The loss of hydrogen at 570 K annealing also confirms the surface layer cleavage after 570 K annealing. Figure 2.10. SEM micrograph from an irradiated (fluence ~ 5.0x1016 H+/cm2) and annealed sample after 870 K annealing is presented. For the first time, we have demonstrated that the oxide films can be cleaved using Smart-Cut technology. Rutherford backscattering spectrometry and channeling techniques were used to monitor the damage accumulation and thermal recovery as a function of annealing temperature. Scanning electron microscopy micrographs show that the entire surface layer has been removed after annealing at 870 K. Although the SEM micrographs were collected after annealing the samples at 870 K, it appears that the cleavage of surface layer occurred after annealing at 570 K. References Agarwal, A., T. E. Haynes, V. C. Venezia, O. W. Holland, and D. J. Eaglesham, Appl. Phys. Lett. 72, 1086 (1998). Aspar, B., E. Jalaguier, A. Mas, C. Locatelli, O. Rayssac, H. Moriceau, S. Pocas, A. M. Papon, J. F. Michaud, and M. Bruel, Electron. Lett. 35, 1024 (1999). Aspar, B., M. Bruel, H. Moriceau, C. Maleville, T. Poumeyrol, A. M. Papon, A. Claverie, G. Benassayag, A. J. Auberton-Herve, and T. Barge, Microelectron. Eng. 36, 233 (1997). Bruel, M., Electron. Lett. 31, 1201 (1995). Di Cioccio, L., F. Letertre, Y. Le Tiec, A. M. Papon, C. Jaussaud, and M. Bruel, Mater. Sci. and Eng. B 46, 349 (1997). Thevuthasan, S., C.H.F. Peden, M. H. Engelhard, D. R. Baer, G. S. Herman, W. Jiang, Y. Liang, and W. J. Weber, Nucl. Instr. Meth. A 420, 81 (1999). Thevuthasan, S., W. Jiang, and W. J. Weber, in preparation. William R. Wiley Environmental Molecular Sciences Laboratory Feedback: webmaster@emsl.pnl.gov Revised: June 12, 2001 Security & Privacy PNNL-13147