AVS1996 Session SS+NS-WeM: Submonolayer Growth Kinetics

Wednesday, October 16, 1996 8:20 AM in Room 204C

Wednesday Morning

Start	Invited?	Item
8:20 AM	Invited	SS+NS-WeM-1 Low Energy Electron Microscopy Studies of Si Surface Dynamics R. Tromp, W. Theis (IBM T.J. Watson Research Center); N. Bartelt (Sandia National Laboratories) The Low Energy Electron Microscope (LEEM) is ideally suited for studying surface dynamic phenomena at the 10 nm lengthscale and up, at video time resolution. In this talk I will present recent results on the homoepitaxial growth of Si, as well as on the spatiotemporal dynamics of the second-order disordering phase transition of Si(113)-(3x1). Epitaxial growth is generally believed to be a far-out-of-equilibrium process. However, we find that growth of Si on Si, at temperatures above 500 C, occurs very close to thermodynamic equilibrium, resulting in a very large critical nucleus size of 650 dimers at a growth temperature of 650 C. Classical homogeneous nucleation theory, incorporating the results of previous LEEM experiments on equilibrium step dynamics in this temperature range, provides a quantitative description of the nucleation process. Second order phase transitions are traditionally studied by diffraction methods, averaging over the size of the probing beam, and over time. With LEEM such phase transitions can be imaged with good spatial and temporal resolution. We directly observe the temperature dependence of the correlation length in the critical fluctuations near T\sub c\, as well as critical slowing of the fluctuations as T\sub c\ is approached. From these data we obtain a value for the dynamical critical exponent z=1.9 \+-\ 0.2, in agreement with theoretical predications for non-order-parameter-conserving phase transitions.
9:00 AM		SS+NS-WeM-3 LEEM Investigation of Si/Si(111) Step Flow Growth W. Chung (Hong Kong University of Science and Technology); M. Altman (Hong Kong University of Science and Technology, Hong Kong) Step flow is one of the basic mechanisms of layer-by-layer crystal growth. According to Burton, Cabrera and Frank (BCF), the steady-state, non- equilibrium adatom concentration on a terrace during step flow growth is related to the even solution of a wave-like equation governing the supersaturation [1], grad\sup 2\\psi\ \alpha\ \psi\. This solution is a symmetric function of position with respect to its peak at the middle of the terrace. The peak value increases as a function of terrace width until island nucleation is driven on terraces which are wider than some critical terrace width, \lambda\\sub c\. Using low energy electron microscopy (LEEM), we have measured \lambda\\sub c\ as a function of temperature for Si/Si(111) step flow growth. An Ahrennius behaviour is observed with activation energy of 2.05 eV. We have also measured the distribution of nucleation position at the critical terrace width. This distribution is a probe of the adatom concentration during growth. An unexpected shift of the concentration towards the up-step side of a terrace was clearly seen. This shift is accounted for analytically by including the previously neglected, odd solution to the BCF supersaturation wave equation. An asymmetric adatom concentration may result from nonequal rates of step attachment (detachment) of an atom from (to) terraces on the up and down sides of a step. [1] W.K. Burton, N. Cabrera and F.C. Frank, Phil. Trans. Roy. Soc. London A 243, 299 (1951).
9:20 AM		SS+NS-WeM-4 Si Monomer Trapping at Islands and Steps on Si(001) Measured with STM B. Swartzentruber (Sandia National Laboratories) The growth of epitaxial films takes place through the incorporation of deposited atoms into substrate lattice positions. For the case of silicon growth on the Si(001) surface, which reconstructs with a 2x1 periodicity, the lattice is propagated through incorporation of 4-atom units at steps. The details of the growth kinetics are determined by the processes through which these 4-atom units are formed, e.g., the formation and lifetime of metastable dimers and their subsequent capture of an additional 2 atoms. These processes ultimately depend on the sticking and residence time of monomers at the steps and islands. In this work, the kinetics of Si monomers, at submonolayer coverage, are measured using STM. Si monomers are observed in empty-state images acquired between room temperature and 115 C. The monomers become trapped at the ends of rebonded-SB type dimer rows, while substrate defects and SA type steps act as reflection barriers. At elevated temperatures, monomers thermally escape from the traps and diffuse one-dimensionally along the substrate dimer row until they find another unoccupied trap or return to their original trap. From the rate at which the monomers escape from the traps, the binding activation barrier at isolated traps is estimated to be ~1.0 eV. A slightly lower barrier exists for monomers to hop between traps located at the ends of neighboring dimer rows -- a process facilitating diffusion along segments of the SB type steps. This new observation of monomer kinetics yields insight into the atom-scale processes at work during growth. This work performed at Sandia National Labs is supported by the U.S. DOE under contract DE-AC04-94AL85000.
9:40 AM		SS+NS-WeM-5 LEEM Observations of Growth and Annealing of Ge on Si(001) D. Savage, J. Maxson, M. Lagally (University of Wisconsin, Madison); R. Tromp, M. Reuter (IBM T.J. Watson Research Center) The evolution of thin-film structure and morphology as a function of temperature and dose is intimately tied to both thermodynamic and kinetic properties of the substrate and the deposited material. Typically, static measurements of morphology are taken from a growth front that has been quenched-in to deduce these properties (akin to an archeological study). Ideally, dynamical measurements will more directly reflect actual physical processes occurring during growth. We have used low-energy electron microscopy (LEEM) to follow dynamically the structural and morphological evolution of ultra-thin Ge films deposited using digermane gas source MBE. We observe different kinetic growth modes, step-flow and layer-by-layer (island nucleation), at different temperatures. Annealing studies of deposited films, while imaging the surface in a dark-field mode, show direct evidence for a thermal roughening transition that is strongly coverage dependent, where the roughening transition is defined as the temperature at which the free energy for step formation vanishes. For two-monolayer coverage, the roughening transition is estimated to be ~ 850 C as evidenced by a loss of contrast between alternate 2x1 and 1x2 domains. As the roughening temperature is approached from below, these opposing domains appear and disappear in adjacent atomically flat terraces. For one monolayer, no roughening is observed up to 1000 C. The transition temperature, therefore, lies between that of pure Si(001) and pure Ge(001), as expected in a system exhibiting a Stranski-Krastanov growth mode. The coverage dependence is consistent with recent STM results that show that step energies decrease with increasing Ge layer thickness. Results will be discussed in terms of theoretical models of surface roughening. *Research supported in part by NSF and ONR
10:00 AM		SS+NS-WeM-6 The Formation Mechanism of Step Bunching on Vicinal GaAs(001) Annealed in AsH\sub 3\ and H\sub 2\ Ambient K. Hata (Univeristy of Tsukuba, Japan); H. Shigekawa (University of Tsukuba, Japan); T. Okano (University of Tokyo, Japan) Annealing vicinal GaAs(001) in AsH\sub 3\ and H\sub 2\ ambient is well known to result in step bunching [1]. The time evolution of this step bunching showed that after it evolves into a particular size, it reaches a stationary state with its ambient. In this study we will show how the annealing conditions, such as partial pressure of AsH\sub 3\\, annealing temperature, and the miscut direction of substrates influence the stationary step bunching. For all of the experimental conditions, and on all of the substrates studied, we always observed step bunching, no regular monostep array, therefore we conclude that vicinal GaAs(001) surface annealed in AsH\sub 3\/H\sub 2\ ambient has a general tendency to form step bunching. Next, with the aim to clarify which of As or hydrogen is the direct cause, we studied the surface of vicinal GaAs(001) annealed in hydrogen, AsH\sub 3\/N\sub 2\, and nitrogen. The surface annealed in hydrogen showed a small irregular step bunching while the others did not. By consulting these experimental results, we sought the origin of this step bunching by estimating the appropriateness of the known mechanisms. We will show that the well known thermodynamic instability caused by coexistence of two phases [2] is inadequate to explain our results. Instead we reached to a conclusion that H attached to the step edge modifies the energy barrier of the step, so that the net flow among steps acts as an attractive interaction. In that case, we will show that the size of the stationary step bunching is proportional to the coverage of H at the step-edge, and the results projected by this model is exactly consistent with the experiments. [1] Appl. Phys. Lett. 63, 1625(1993), J. Appl. Phys. 76, 5601(1994) [2] Science. 251, 393(1991)
10:20 AM		SS+NS-WeM-7 Monte Carlo Study of Si(100)-2x1 Monoatomic Step Dynamics J. S\aa a\nchez, C. Aldao (Universidad Nacional de Mar del Plata-CONICET, Argentina) Motivated by recent atomic resolution STM results, we have undertaken Monte Carlo simulations of step dynamics in which sets of two dimers in the Si(100)-2x1 surface are detached from a B-type step according to their nearest-neighbors interaction energies and are attached following a simple relaxation mechanism. In particular, roughness, average kink length, and correlation length from experiments and modeling are presented in order to check the quality of our results. From simulations, attachment and detachment rates are determined and match those observed experimentally in the range 500-750 K. These results are important because they make it possible to interpret the effective activation energies for kink creation, annihilation, and diffusion as a direct consequence of the interactions among sets of two dimers and the proposed relaxation process. The characteristics of the resulting step profile are also analyzed as a function of the step width.
10:40 AM		SS+NS-WeM-8 Real-Time Imaging of the Diffusion, Nucleation and Growth of Pt on Pt(110) T. Linderoth, S. Horch, L. Pedersen, E. Laegsgaard, F. Besenbacher (University of Aarhus, Denmark) Using variable-temperature STM, Pt adatoms on the Pt(110)-(1x2) surface are studied in the sub-monolayer coverage regime. We find the adatom hopping rate to be of the order of minutes at room temperature. This system therefore provides a fascinating new opportunity to follow diffusion, nucleation and growth in real time at the atomic level. The adatoms are confined to the troughs of the missing-row reconstruction, causing the diffusion to be one-dimensional. In the temperature range from 10 \super o\C (where diffusion sets in) to 50 \super o\C, the migration of single adatoms, as well as their nucleation and the growth and stability of the resulting 1D islands is monitored using STM-movies. Dimer break-up is frequently observed, whereas larger islands are predominantly stable. From the temperature dependence of the adatom migration, a diffusion barrier is extracted. The issue of tip-influence on the diffusivity is addressed. For deposition temperatures from 80 \super o\C to 130 \super o\C, conventional "quench-and-look" growth experiments are performed, and the resulting island size distributions are analysed. Above 100 \super o\C, nucleation frequently occurs on top of the islands, even at very low coverage, indicating that upwards diffusion has become feasible. Kinetic Monte-Carlo simulations are used to determine a barrier for this and other key processes.
11:00 AM		SS+NS-WeM-9 Step Flow Instabilities in the Growth of Cu on Cu (115) and Cu (11 17) L. Schwenger, R. Folkerts, H. Ernst (CEA Saclay, France) Faceting of surfaces during growth in the step flow mode has been frequently observed and is a well known phenomenon, however, its detailed underlying origin remains in most cases unclear. One of the decisive parameters that control the stability of the flowing step train is the adatom attachment kinetics to ascending and descending steps. The investigation of growth in an unreconstructed, homoepitaxial system offers the possibility to address this kinetic aspect in its pure form. To this end, we have studied the growth of Cu on Cu(115) and Cu(11 17)in the step flow mode over a wide range of temperatures using Helium Atom Beam Scattering as a probe. We find that on both templates growth below room temperature is stable what the average distance between steps concerns, however, a strong meandering of advancing (115) steps, and, in the case of Cu(11 17), a pronounced 'fingering' of advancing (11 17) steps is observed. We interpret this observation as a manifestation of the Bales-Zangwill instability, which indicates that Cu adatoms attach preferentially to ascending steps in this temperature range. Growth above room temperature on the (115) surface leads to a destabilization of the step train, and facets are produced. According to Schwoebel and Shipsey, this requires preferential attachment of adatoms to descending steps. Thus, the adatom 'uphill current' at low temperatures due to the presence of a Schwoebel-Ehrlich barrier is overcompensated by a 'downhill current' at only high enough temperatures. Within a simple Arrhenius-type parametrization for adatom capture rate constants to ascending and descending steps this result implies that this system is not only characterized by a larger energy barrier, but also by a larger preexponential factor for migration of adatoms over descending steps with respect to diffusion on terraces. The growth on Cu (11 17) at high temperatures is currently under investigation.