Papers

AVS2001 Session EL-WeA: Semiconductor Growth

Wednesday, October 31, 2001 2:00 PM in Room 124

Wednesday Afternoon

Start	Invited?	Item
2:00 PM		EL-WeA-1 Epitaxial Growth by Low-energy Plasma-enhanced CVD H. von Känel (ETH Zürich, Switzerland); M. Kummer, A. Dommann (Interstate University of Applied Science, Switzerland); C. Rosenblad, J. Ramm (Unaxis Semiconductors, Liechtenstein); T. Hackbarth, M. Zeuner (Daimler Chrysler Research Center, Germany) Epitaxial growth of Si and SiGe by a low-energy plasma-enhanced CVD (LEPECVD) process is described. LEPECVD is based on a low voltage DC arc discharge, leading to very efficient decomposition of the gaseous precursors SiH₄ and GeH₄. In addition, direct immersion of the substrate in the intense plasma strongly enhances surface kinetics in particular at low growth temperatures. As a result, epitaxial growth rates in LEPECVD can be as high as 10 nm/s, nearly independent of the substrate temperature in the range between 500°C and 750°C. LEPECVD is ideally suited for the growth of SiGe-MODFET and SiGe-MOSFET structures, requiring thick graded Si_1-xGe_x buffer layers with low defect densities. Concentration profiles are easy to control because Ge incorporation is entirely determined by the germane to silane flux ratio. X-ray reciprocal space mapping, defect etching and scanning force microscopy have been used to characterize buffers grown with Ge end concentrations between 10 % and 100 %. LEPECVD combined with other low temperature growth techniques, such as molecular beam epitaxy (MBE) has been shown to lead to superior performance of n-channel MODFETs.¹ Alternatively, the growth rates in LEPECVD can be lowered in order to allow for the synthesis of active channels with abrupt interfaces by this technique alone. Examples will be shown with strained Ge-rich channels up to 100 % Ge on relaxed buffers with Ge end concentrations between 50 and 70 %. Such structures have given rise to p-MOSFETs with effective hole mobilities exceeding the one of standard Si p-MOSFETs by up to a factor of four ². ¹ T. Hackbarth et al., 11^th European Workshop on Molecular Beam Epitaxy, Hinterzarten, Germany, 02/06/01. ² G. Höck et al., Appl. Phys. Lett. 76, 3920 (2000).
2:40 PM		EL-WeA-3 Self-Limited Layer-by-Layer Growth of Si by Alternated SiH₄ Supply and Ar Plasma Exposure D. Muto, T. Seino, T. Matsuura, J. Murota (Tohoku University, Japan) Self-limited layer-by-layer growth of Si without substrate heating has been investigated using ECR plasma. First, 5Å-thick epitaxial Si films on Si(100) were continuously deposited by Ar plasma with SiH₄ at the Ar pressure of 16mTorr and the SiH₄ partial pressure of 0.02mTorr with the microwave power of 800W. Then, by alternated SiH₄ supply for 2min at the pressure of 10mTorr and the Ar plasma exposure for 5 or 20 sec at the pressure of 16mTorr with the microwave power of 200W, atomic-order Si epitaxial growth was carried out. The deposited Si films thickness was determined by AFM. Surface structure and hydrogen termination on the Si surface were evaluated by RHEED and FTIR/RAS, respectively. For the alternated process on the wet-treated surface, the Si deposition was scarcely observed on the wet-treated surface by the alternated process. By the continuous deposition, the wet-treated surface changed the dimer monohydride structure. As a result, the deposition rate was 0.3Å/cycle when the Ar plasma exposure time was 5sec. In this deposition cycle, the intensity of the dimer monohydride after SiH₄ exposure was about 2/3 as that after Ar plasma exposure. When the Ar plasma exposure time increased to 20sec, the deposition rate was 0.5Å/cycle. The intensity of dimer monohydride was higher than that of 5sec Ar plasma exposure. These results mean that SiH₄ molecule is adsorbed on the dimer monohydride structure, next Si epitaxial growth with the generation of the dimer monohydride structure is performed due to decomposition of adsorbed SiH₄ by Ar plasma exposure. In conclusion, atomic-order layer-by-layer epitaxial growth of Si is achieved by SiH₄ self-limited adsorption and enhanced reaction due to Ar plasma exposure.
3:00 PM		EL-WeA-4 Detrimental Effects in using Surfactant Assisted Growth G.G. Jernigan, P.E. Thompson, M. Fatemi, M.E. Twigg (U.S. Naval Research Laboratory) Using STM, XPS, TEM, and XRD, we have investigated the use of Sb as a surfactant for the growth of SiGe quantum wells (QW) by molecular beam epitaxy. Our XPS results indicated that Ge surface segregation was inhibited from a Si capping layer after the growth of the QW with the use of an Sb surfactant, but Ge segregation during the growth of the QW was only partially prevented by the Sb surfactant. The differing effect Sb had on the start and the end of the growth of a SiGe QW led us to believe that something additional to the suppression of Ge segregation was occurring. An XRD analysis of many QW samples showed that Sb did not relieve strain in the SiGe QW but resulted in broader diffraction peaks due to compositional fluctuations within the QW. In agreement with the XRD, TEM showed that there was a greater amount of strain contrast within the QW due to compositional fluctuations. We will report our most recent results, using an in situ STM, which shows that Sb surfactant assisted growth has the detrimental effect of roughening the Si and SiGe surface while making the interface more compositionally abrupt.
3:20 PM		EL-WeA-5 Optical and Structural Characterization of GaN Films Grown by Molecular Beam Epitaxy on SiC Coated Si Substrates M. Lopez-Lopez, M. Cervantes-Contreras, M. Melendez-Lira, M. Tamura (CINVESTAV-IPN, Mexico) High-quality GaN layers are difficult to grow on Si substrates due to the large lattice mismatch (17%), and the problems associated to the growth of a polar material on a non-polar substrate. MBE growth of GaN directly on Si frequently results in films with a mixture of the stable hexagonal phase and the metastable cubic phase. Single-crystal cubic GaN films can be obtained by coating the Si substrates with a thin SiC layer.¹ Here we present a study of the effects of the orientation of SiC-coated Si substrates on the MBE growth of GaN. The GaN layers were grown in a conventional MBE system with an RF activated nitrogen plasma source. (100)- and (111) oriented Si substrates were chemical treated in a HF solution, and then annealed in the MBE preparation chamber under a C₂H₂ partial pressure. This resulted in the formation of ~2.5 nm-thick SiC epitaxial layer. Transmission electron microscopy and x-ray diffraction results showed that the growth on SiC-coated Si(100) leads to cubic GaN films, the growth on SiC-coated Si(111) resulted in predominantly hexagonal GaN. 12K-photoluminescence spectroscopy (PL) of the films on (100) substrates showed an intense emission at ~3.1 eV associated to a donor-acceptor pair transition. However, the PL spectra of films on (111) substrates showed an additional peak at ~2.4 eV, which could be associated to crystal defects. 300K-photoreflectance (PR) spectra presented transitions at ~3.2 and ~3.4 eV for GaN on (100)- and (111) Si substrates, respectively. A quantitative analysis of the PR line spectra was carried out using the third-derivative function theory. We obtained band-gap energy values lower than those generally accepted. These red shifts could be associated to residual tensile strain in the epilayers due to the lattice mismatch and the difference in thermal expansion coefficients between GaN and the Si substrate. ¹ Y. Hiroyama and M. Tamura: Jpn. J. Appl. Phys. 37 (1998) L630.
3:40 PM		EL-WeA-6 Detailed Modeling of Si Gas-source MBE: Descriptions on Growth Rate and Hydrogen Coverage T. Murata, M. Suemitsu (Tohoku University, Japan) In response to recent requirements from CMOS technologies, Si CVD is attracting renewed attentions. The attentions include needs for more complete understanding of the growth kinetics and a precise modeling of the growth. We here present the results of our growth experiments and a growth model based on the results. What is unique in our experiments is the observation of the surface hydrogen coverage θ during growth, which is doubtlessly a key parameter in the description of the growth in CVD mode and yet has been quite rarely obtained in the past. To obtain θ we employed Si gas-source molecular beam epitaxy using disilane, and have conducted temperature-programmed-desorption measurements on the surface quenched from the growth. Growth rate and θ were obtained as a function of both the growth temperature and the source-gas pressure. We then tested two growth models based on the results. While the conventional 2-site-adsorption model well described the temperature- and the pressure-dependence of the growth rate, it failed to reproduce the behavior of θ. In contrast, the 2-site/4-site adsorption model developed previously by the authors¹ showed almost complete fits to both the growth rate and the hydrogen coverage. The model assumes the 2-site and 4-site adsorptions to dominate the growth at low and high temperatures, respectively, and the present results prove its excellent ability to describe the Si growth in CVD mode. This work was supported in part by a Grant-in-aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology, Japan. ¹ Suemitsu et al., Jpn. J. Appl. Phys. 36(1997) L625. .
4:00 PM		EL-WeA-7 Growth and Doping Kinetics in Si Gas-source MBE with In-situ Phosphrous-doping M. Suemitsu, Y. Tsukidate (Tohoku University, Japan) Being free of ion damages and redistribution of dopants during post-anneals in ion implantation, in-situ doped epitaxy is attracting renewed attention in ULSI processings. Remaining barriers are found mostly in N-type doping, where addition of PH₃ exhibits notable retardation of the growth and yet difficulties in achieving demanded high doping levels. In the studies presented here, we have experimentally and theoretically investigated the growth and doping kinetics in Si gas-source MBE using SiH₄ and PH₃. Experimentally, the growth rate (R_g) and, more importantly, the hydrogen (H) and phosphorous (P) coverages on the growing surface were obtained by temperature-programmed desorption. The R_g Arrhenius plot showed a seemingly conventional separation into a high- and a low-temperature (LT) region with a low and a high activation energy, respectively, but obtained P and H coverages indicate that the domain I (5502 desorption while the domain II (T<550C) by both P₂ and H₂ desorptions. In domain I, the P coverage amounted to ~0.3 ML, some three orders of magnitude as high as in the bulk. This indicates a very rapid surface segregation of P atoms during growth, as well as its major role in impeding high doping levels. Surprisingly, mild PH₃ addition (<10ppm) in the high-temperature region were found to accelerate the growth rate. With the known site-blocking effects of P atoms[1], this suggests enhanced stickng probability of silane molecules at Si sites in the presence of surface P atoms. Based on these exprimental results, a theoretical model for growth/doping has been constructed and compared with experimental R_g and P and H coverages.
4:40 PM		EL-WeA-9 Improved Surface Preparation for High Quality Homoepitaxial Growth of SiC W.V. Lampert, C.J. Eiting, S.A. Smith, L. Grazulis, J.S. Solomon, T.W. Haas (Air Force Research Laboratory, Materials and Manufacturing Directorate) Surface quality is a key factor in determining the quality of films grown by molecular beam epitaxy (MBE). Fullerene (C ₆₀) and silicon (Si) effusion cells were used to grow superior quality homoepitaxial 6H-SiC. Prior to ex-situ processing, the wafers used for this study had the scratched surfaces and subsurface damage typical of commercial SiC wafers. Our ex-situ surface preparation includes chemical-mechanical polishing (CMP), carbon dioxide (CO ₂) cleaning, hydrogen fluoride (HF) etching, de-ionized (DI) water rinsing, and nitrogen gas drying. Various surface analysis tools, such as Auger electron spectroscopy (AES) and atomic force microscopy (AFM), have been used to characterize the effects of these ex-situ processing steps. The results show that while our CMP processing adds surface contamination, it also leaves a scratch-free and stepped surface that is much more suitable for epitaxial growth. The results further show that the HF etching process step removes the contaminants left by the CMP process step and does not adversely effect the stepped surface. Results of these measurements and their implications for successful growth are discussed.
5:00 PM		EL-WeA-10 Initial Stages of Al_xSe_y Heteroepitaxial Growth on Si(111) J.A. Adams, A.A. Bostwick (University of Washington); E. Rotenberg (Advanced Light Source, Berkeley); F.S. Ohuchi, M.A. Olmstead (University of Washington) Aluminum selenide is a largely unstudied material with interesting possibilities as a wide band gap semiconductor. Bulk Al₂Se₃ has a 3.1 eV band gap, and its hexagonal lattice constant (defected wurtzite structure) is about 1.3% larger than Si(111). However, very little is known about the properties of Al_xSe_y heteroepitaxial films. Unlike Ga_xSe_y, which is stable in both layered GaSe and defected zincblende Ga₂Se₃ structures, layered AlSe has not been reported in either bulk or thin film form. We have used an Al₂Se₃ evaporative source to deposit thin films of Al_xSe_y on Si(111) by molecular beam epitaxy. We investigated their electronic and atomic structure using angle-resolved valence band and core-level spectroscopy and diffraction, and compare these results to our previous work on Ga_xSe_y/Si(111). The initial Al_xSe_y/Si interface appears to form a bilayer structure similar to GaSe-terminated Si, although the temperatures for bilayer formation and for Se-evaporation from the film are higher for AlSe than for GaSe. Surface states (resonances) are well within the bulk Si bands, at least 1.2 eV below the Fermi level at the zone center. Despite the initial lone-pair termination, subsequent growth leads to photoelectron diffraction forward-focussing peaks along directions expected for the covalently-bonded wurtzite structure and not for layered AlSe.