AVS2001 Session EL-WeM: Si Surface Dynamics and Reactions

Wednesday, October 31, 2001 8:20 AM in Room 124

Wednesday Morning

Start	Invited?	Item
8:20 AM	Invited	EL-WeM-1 Step Motion and Morphology on Si(111) M.S. Altman (Hong Kong University of Science and Technology) Surface steps are of interest for their role as templates for growth. Step morphology determined by thermodynamics can be influenced through step motion kinetics. Low energy electron microscopy observations of step flow, island nucleation and growth, and spiral growth at screw dislocations are the basis for studying step motion kinetics and the influence of surfactants upon growth. Sb and In surfactants are found to have notable and opposite influence on step kinetics. These results demonstrate that surfactants can be used to controlkinetically driven morphological instabilities, i.e., bunching and meandering, in step flow motion, and support the view that surfactants function at least partially through step passivation. However, surface phase separation is also observed as a consequence of surfactant incorporation during growth, which leads to a distinct step flow instability. Self-organized periodic step bunching patterns on vicinal surfaces will also be discussed.
9:00 AM		EL-WeM-3 Step Fluctuations on a Chemically Heterogeneous Surface: Al/Si(111)-(√3x√3)¹ I. Lyubinetsky, D.B. Dougherty, E.D. Williams (University of Maryland at College Park) Analysis of equilibrium step fluctuations has been extended for an adsorbate-induced reconstruction. The lateral and temporal correlations of single-height step fluctuations has been studied with variable temperature STM at 550-750° C on a vicinal Si(111) surface, miscut by 0.5° in the [2,-1,-1] direction, after formation of the Al-induced (√3x√3) surface phase. From direct measurement of the spatial step-correlation function, the step diffusivity, b²/a, is found to exponentially increase from 0.7 Å at 550°C to 1.19 Å at 750°C. The effective kink creation energy of 0.14 eV then has been extracted from corresponding Arrhenius plot. Results of analysis of lateral step correlations at elevated temperatures will also be compared with results obtained in traditional way by quenching surface down to the room temperature. The temporal correlation functions in this temperature region are shown to scale as t^1/2, consistent with either step-edge attachment or terrace-crossing diffusion as a rate limiting process. The prefactor of the temporal correlation function also varies exponentially with temperature yielding an effective activation energy of 2.5 eV for the rate-limiting step in surface mass transport. Using both lateral and temporal correlation functions, the kinetic parameters governing mass transport have been extracted for different temperatures, and atomistic models for the step fluctuations will be discussed. ¹ Work supported by the UMD-NSF-MRSEC.
9:20 AM		EL-WeM-4 Atomic and Electronic Structure of Si Layers on CaF₂/Si(111) A.A. Bostwick, J.A. Adams (University of Washington); E. Rotenberg (Advanced Light Source, Berkeley); M.A. Olmstead (University of Washington) Many forms of nanocrystalline silicon have been observed to luminesce at room temperature under photon or electron excitation. However, the relative importance of electron confinement and interface compounds is still an open question for many of these materials. Our group recently developed a means to fabricate ultrathin, crystalline, arsenic-terminated silicon quantum wells on calcium fluoride using electron irradiation in the presence of an arsenic surfactant flux.¹ With no intrinsic oxygen and a well-defined surface and interface structure, these films make excellent candidates for probing the properties of oxygen-free silicon nanostructures. We have studied such films with angle-resolved valence band and core-level photoemission spectroscopy and diffraction and near-edge xray absorption spectroscopy. Our initial studies show no surface states in the bulk Si energy gap and a flat, weakly dispersing density of valence states. We also discuss growth mechanisms for these films, which are sensitive to the irradiation and arsenic exposure conditions. ¹ B. R. Schroeder, S. Meng and M. A. Olmstead, Appl. Phys. Lett. 77, 1289 (2000).
9:40 AM		EL-WeM-5 Surface Mass Transport and Island Nucleation during Growth of Ge on Laser Textured Si(001) T. Schwarz-Selinger (Max Planck Institut für Plasmaphysik, Germany); D.G. Cahill (University of Illinois, Urbana) Shape transitions and coarsening of coherent three-dimensional islands in Stranski-Kranstanow crystal growth have been extensively studied but quantitative descriptions of island nucleation kinetics are hindered by incomplete understanding of surface mass transport on the wetting layer. To gain new insights on wetting layer mass transport, we manipulate the spatial distribution of island nucleation by modifying the substrate morphology with laser texturing: fluid flow in the melt zone created by tightly-focused nsec pulses from a frequency doubled Nd:YAG produces shallow, micron-sized dimples on the Si substrate. We then use gas-source molecular beam epitaxy to deposit Ge and, finally, measure the distribution of three-dimensional Ge islands surrounding the laser dimples by atomic force microscopy. As expected, island nucleation starts at the vicinal surfaces near the rim of the dimple. These initial islands create a denuded zone of suppressed island nucleation on the flat regions of the substrate surrounding the dimple. Nucleation theory predicts that the extent of the denuded zone L should be comparable to the the island-island separation d but, in the limit of slow growth rates, we observe L/d>20. By comparing the denuded zone lengths produced by different growth rates (varied by a factor of 100) and substrate temperatures (500-600°C), we extract an activation energy for surface mass transport.
10:00 AM		EL-WeM-6 Non-thermal SiO₂ Film Growth on Si(100) using Laser-generated O(¹D) and O(³P) T.C. Coulter, A.C. Tuan (University of Washington); W.P. Hess, J.W. Rogers, Jr (Pacific Northwest National Laboratory); Y. Ono (Sharp Labs of America) Thermal oxidation of silicon by O₂ or H₂O at high temperature (usually 800-1000°C) is currently used to achieve device-quality films for microelectronic applications. These high temperatures can degrade other device characteristics. As device dimensions shrink, it becomes even more important to keep processing temperatures low. Thus, many alternative low-temperature oxidation methods have been explored. A successful low-temperature method must produce thin silicon oxide layers with good uniformity, abrupt interfaces, and device-quality electrical characteristics. In order to develop the best low temperature deposition strategy, it is necessary to understand the details of the oxidation mechanism. High-temperature thermal oxidation studies suggest that atomic oxygen, and not molecular O₂, may be the oxidizing species diffusing through the oxide to react at the silicon interface. Low temperature plasma oxidation proceeds faster than thermal oxidation, which is often attributed to the presence of charged and neutral atomic species. Recent studies using a modified plasma for SiO₂ growth suggest that excited oxygen atoms in the ¹D state may play an important role in oxidation.¹ To elucidate the roles of excited and ground state neutral oxygen atoms in silicon oxidation, we have used photolytically generated ground state O(³P) and excited state O(¹D) atoms to oxidize Si(100) at low temperature. In contrast to plasma oxidation, where many oxygen species of differing energy and charge are present, we can study the contributions of O(³P) and O(¹D) individually, and compare their kinetics and oxidation mechanism. The growth rate and oxidation kinetics were studied with in-situ ellipsometry, and oxide stoichiometry and interface quality were determined with XPS. ¹M. Hirayama, K. Sekine, Y. Saito, and T. Ohmi. IEEE Transactions on Electron Devices. 47(7), 1370 (2000).
10:20 AM		EL-WeM-7 On the Use of Angle-resolved XPS for Resolving Composition Structure of Ultrathin Inhomogeneous Oxide Layers T. Conard, H. De Witte, W. Vandervorst, J. Petry (IMEC, Belgium); R. White, K.S. Robinson (Thermo VGScientific, UK) With the downscaling of electronic, the industry faces a large number of challenges. Among those, layers with an EOT lower than 1nm have to be engineered for the gate oxide. The materials considered can be silicon oxynitride, ZrO₂, Al₂O₃ as a single layer or as multistack. One common characteristic of all these films is there very limited thickness (a few nm at most) and the inhomogeneous distribution of elements throughout the stack. These films challenge thus most of the analysis techniques used for determining the layer structure and composition such as SIMS. For very thin layers, XPS is a possible alternative to sputtering techniques as it allows a non-destructive analysis through the whole film. The development of angle resolved XPS instruments with the ability to acquire spectra simultaneously over multiple angles should also give the possibility to retrieve the layer composition as a function of depth. This work concentrates on the interpretation of such ARXPS data and its application as an alternative and complementary technique to SIMS profiling. Particular in our experiments is the simultaneous collection of up to 16 emission angles (using a Thermo VG Scientific ThetaProbe) which provides unparalleled capabilities to reconstruct non-homogenous depth profiles in very thin layers. We will present results showing that small differences in the nitrogen distribution (near surface, in film, interfacial) inside thin (~3.5 nm) SiON layers can be determined using simultaneous ARXPS. In addition, the technique can also be used to understand the strong modification observed in the profiling of ZrO₂/SiO₂/Si stacks using Ar+ ions at energies between 500 eV and 3keV. The depth profiles reconstructed from the ARXPS provide information on the redistribution of Zr and O in the ion bombarded film allowing to interpret these modifications in terms of oxygen preferential sputtering and sputtering induced oxygen diffusion in the ZrO₂ layer.
11:00 AM		EL-WeM-9 Silicon-Oxide Formation on Gold T. Vdovenkova, A.J. Slavin (Trent University, Canada) The Au-Si system has been well studied as a model system for metal-silicon interfaces. In contrast, the present work is one of the few studies of the growth of a silicon-oxide film on a gold substrate. The gold is the electrode, about 200-nm-thick, of a quartz-crystal microbalance (QCM) used for measuring the deposited mass. In early work on the deposition of Si onto a Au film about 10-nm thick, Si was visible in Auger spectroscopy (AES) from the start of evaporation. In contrast, in our study the evaporation of several monolayer-equivalents of Si under UHV conditions at room temperature resulted in the Si dissolving into the gold without the accumulation of any Si on the surface. Subsequent exposure to oxygen gas did not result in the segregation of Si to the surface. However, a layer of oxide from 0.5 to 1.8 nm thick was produced by a series of evaporations of the Si in the presence of O₂ gas at 1 x 10^-5 torr with the sample held at 78°C, probably aided by oxygen ions produced at the hot filament of the evaporator. The film growth has been studied by AES and EELS, using the QCM to monitor the amount of silicon oxide deposited. The average stoichiometry of the oxide was roughly SiO_1.75, based on an AES peak position of 78 eV compared to 92 for elemental bulk Si and 76 for SiO₂. One atomic layer of Si evaporated in vacuum onto the oxide film gave an AES peak at 90 eV, thought to be elemental Si; this showed that the Si oxide acted as a barrier to Si in-diffusion. This extra Si did not oxidize at 78°C under O₂ gas at 1 x 10^-5 torr with the evaporator filament off. The intensity of the 90-eV peak decreased in comparison with the peak for Si in SiO_x, under prolonged exposure to the electron beam. This research was supported by NSERC, Canada.
11:20 AM		EL-WeM-10 Kinetics of the Selective Oxidation of Si(100) versus W by H₂O Steam in Hydrogen Y. Liu, J. Hebb (Axcelis Technologies, Inc.) The selective oxidation (SELOX) of Si versus tungsten (W) is an important process to form W gate electrodes on SiO₂ dielectric in the advanced CMOS devices. The SELOX was studied using a small quartz reactor, a catalytic water vapor generator (WVG), a quadrupole mass spectrometer (QMS), an ellipsometer and a 4-point probe. New kinetics data were obtained for the wet SiO₂ growth on Si(100) at 1 atmosphere and 900-1150°C, with the steam (H₂O) in H₂ percentage being 0 to 80%. A Si(100) or a W-covered Si(100) wafer was rapidly heated to a desired temperature in an inert gas. Fast gas sequencing was carried out to expose the wafer to (1) H₂ for W passivation, (2) steam+H₂ for Si oxidation and (3) H₂ for WO_x reduction. Gases in the reactor were analyzed with the QMS. The SiO₂ thickness was measured with the ellipsometer while the W film was characterized by its sheet resistance change. Preliminary results showed that the SiO₂ thickness is proportional to the steam percentage at a given set of oxidation temperature and time, suggesting that the SiO₂ growth follows the first-order kinetics. At a 20% steam percentage and during the first 60 sec, the SiO₂ growth rates at 955°C and 1047°C have been determined to be 0.96 Å/sec and 2.89 Å/sec, respectively. This yields an activation energy of 1.67 eV in agreement with the published values using O₂ or steam+O₂. Between 60 and 180 sec, the growth rates decreased to 0.73 Å/sec and 1.81 Å/sec, respectively. Hence, the earlier oxidation stage is kinetically controlled by an interface reaction step while the later stage is limited by a different slow step. Besides the detailed kinetic studies for the wet Si oxidation in the H₂ reducing ambient, systematic data will be presented for the W oxidation and WO_x reduction to explore the selectivity window for Si SELOX.