AVS1996 Session SS+NS-ThM: Cluster Diffusion and Multilayer Film Growth

Thursday, October 17, 1996 8:20 AM in Room 204C

Thursday Morning

Start	Invited?	Item
8:20 AM		SS+NS-ThM-1 Magic Size Effects for Heteroepitaxial Island Diffusion J. Hamilton (Sandia National Laboratories) There are a number of experimental reports of rapid diffusion of heteroepitaxial islands (containing tens or hundreds of atoms) on surfaces.* This rapid diffusion of heteroepitaxial islands is explained here by a mechanism involving nucleation and motion of dislocations across the island. The activation energies for this process are calculated using a Frenkel-Kontorova model. At "magic island sizes" and misfits, the activation barriers for heteroepitaxial island diffusion are dramatically lowered. Total energy calculations and molecular dynamics simulations using the embedded atom method confirm the existence of a rapid diffusion process for two-dimensional islands. The rate of diffusion for a 61 atom silver island on the Ru(0001) surface at room temperature is found to be approximately 1/100th of the rate of diffusion for a single silver atom on the same surface! A signature of this process is a strongly temperature dependent activation energy for diffusion. *e.g. Kellogg, G. L. Surf. Sci. 187, 153-164 (1987)
8:40 AM		SS+NS-ThM-2 Influences of Cluster Mobility on Coarsening of Adatom and Vacancy Clusters D. Sholl (The Pennsylvania State University); R. Skodje (University of Colorado); K. Fichthorn (The Pennsylvania State University) A number of recent experiments have clearly demonstrated that large, compact clusters of adatoms or vacancies can diffuse on single-crystal surfaces. This fact can have significant implications for the mechanism and rate of adlayer coarsening. We will discuss the coarsening of adlayers which contain mobile clusters in two physically relevant regimes. First, we will present a detailed theory for the late-stage coarsening of adlayers when clusters diffuse in a manner which conserves cluster mass. This situation is an accurate description of vacancy clusters on a variety of metal surfaces.Secondly, we will discuss the coarsening of clusters when the evolution of clusters occurs due to mass exchange between the cluster and a two-dimensional gas phase. The traditional description of adlayer coarsening in this situation is that coarsening takes place by Ostwald ripening and that cluster mobility is irrelevant. While experiments have shown that this description is very accurate for some adlayers, such as Si clusters on Si(001) [1], it is not at all appropriate for Ag clusters on Ag(100) [2]. In the latter system, dynamic coalescence of diffusing clusters dominates the adlayer coarsening [2]. We will present a theoretical framework which allows the existence of Ostwald ripening and coarsening due to cluster mobility to be unified.[1] W. Theis, N. C. Bartlelt and R. M. Tromp, Phys. Rev. Lett. 75 (1995) 3328. [2] J.-M. Wen, J. W. Evans, M. C. Bartelt, J. W. Burnett, and P. A. Thiel, Phys. Rev. Lett 76 (1996) 652.
9:00 AM	Invited	SS+NS-ThM-3 Formation, Diffusion, & Coarsening Processes for Ag Clusters on Ag(100) J. Wen, J. Evans, M. Bartelt, C. Zhang, J. Burnett, P. Thiel (Iowa State University) Two-dimensional Ag clusters form during Ag deposition on Ag(100) at and above 300K [1]. Scanning Tunneling Microscopy (STM) is used to analyze the flux and temperature dependence of the island density immediately following deposition. These studies reveal a transition from irreversible to reversible cluster formation, as temperature increases slightly above 300K. Comparison with simulations [2] yields precise estimates for the terrace diffusion barrier, and the surface dimer bond-energy. STM studies of subsequent adlayer dynamics reveal significant diffusion of large clusters, containing 100 to 700 atoms, with a diffusion coefficient of order 10\super -17\ cm\super 2\ s\super -1\ at 300K [3]. The diffusion coefficient varies weakly, if at all, with cluster size in this range. This observation rules out periphery diffusion as the main mechanism of cluster diffusion, suggesting instead two-dimensional evaporation-condensation. These STM studies also show that diffusion of these clusters controls coarsening of the Ag films under all but extreme conditions of low coverage and large island separation, up to a coverage around 0.65 monolayers (which is just below the percolation threshold) [4]. This contrasts normal expectations of Ostwald ripening as a predominant coarsening mechanism. This work was supported by NSF Grant No. 9317660 --------- [1] P.Bedrossian et al, Surf. Sci. 334, 1 (1995); H.A. van der Vegt et al, ibid v.330, 101 (1995). [2] M.C.Bartelt and J.W.Evans, Surf. Sci. v.28, 421 (1993). [3] J.-M.Wen, S.-L.Chang, J.W.Burnett, J.W.Evans, and P.A.Thiel, PRL v.73, 2591 (1994). [4] J.-M.Wen, J.W.Evans, M.C.Bartelt, J.W.Burnett, and P.A.Thiel, PRL, v.76, 652 (1996).
9:40 AM		SS+NS-ThM-5 Selective Deposition of Al on H/Si(100) Surfaces T. Shen, C. Wang, J. Tucker (University of Illinois, Urbana) Selective metallization is one of the major challenges in the process of nanofabrication. Chemical vapor deposition (CVD) may provide a solution to selectivity, but the film morphology can depend on many factors, such as substrate structure, temperature, precursor flux, and contamination. We will present a STM study of the initial stages of the Al nucleation and growth on various Si(100) surfaces using dimethylaluminum hydride (DMAH) as the CVD precursor. It is found that the initial growth follows the Volmer-Weber mode. The majority of Al crystallites appear in two orthogonal (110)-oriented domains which agree with the epitaxial relationships of Al films grown by MBE, thermal evaporation, or ionized cluster beam method. The substrate surface, however, dictates the nucleation rate. Without using H\sub 2\ as a carrier gas, the nucleation on clean and oxidized Si(100) surfaces is much suppressed compared to H-terminated surfaces. Moreover, different reconstructed H-terminated surfaces show different nucleation rates, which indicates the importance of lattice matching between the underlying substrate and the epitaxial layer. Since the H/Si(100)-1x1 has the smallest lattice misfit from Al(110), it is the most preferred growth surface. Details of the nucleation dependence on substrate temperatures and precursor flux will be analyzed.Supported by the Office of Naval Research URI:N00014-92-J-1519
10:00 AM		SS+NS-ThM-6 A Helium Atom Scattering Study of the Structure and Dynamics of Epitaxial Thin Films of Lead Adsorbed on Cu(111) J. Braun (Max-Planck-Institut f\um u\r Str\um o\mungsforschung, Germany); J. Toennies (Max-Planck-Insitut f\um u\r Str\um o\mungsforschung, Germany) The growth and dynamics of thin lead films on Cu(111) has been investigated using high resolution helium atom scattering (HAS). Previously it had been shown that lead grows in a layer by layer mode on Cu(111) at surface temperatures lower than 200 K [1]. The phonons along the <110> and the <112> azimuthal directions were measured for 1,3,4,5,6,8 and 30 monolayers of lead using the time of flight (TOF) technique. For the first monolayer a mixing is observed between the Rayleigh wave of the adsorbate and of the substrate and another dispersive mode emerges along the <110> azimuthal direction. The development of the Rayleigh mode of the lead films can be followed for coverages of 3 monolayers and higher. A surprisingly large number of additional modes occurs for ultrathin films (up to 5 monolayers) whereas for coverages greater than 8 monolayers the behaviour of the (111) surface of bulk lead is approximated. The dispersion curves show several irregularities which we attribute to Kohn anomalies which are enhanced because of the reduced dimensionality. This will be discussed in connection with the earlier reported anomalies in the dispersion curves of bulk lead [2]. The experimentally determined dispersion curves are also compared with those of thin films of thallium adsorbed on Cu(111), which have also been measured. Additionally we present the preliminary results of a lattice dynamics analysis of thin films based on a force constant scheme. [1] B.J.Hinch, C.Koziol, J.P.Toennies and G.Zhang, Europhys. Lett. 10(1989)341. [2] B.N.Brockhouse, K.R. Rao and A.D.B. Woods, Phys. Rev. Lett. 7(1961)93.
10:20 AM		SS+NS-ThM-7 Charge Limited Cluster Growth from Deposition and Segregation D. Bonnell, X. Li (University of Pennsylvania); Y. Liang (Pacific Northwest National Laboratory) Recent STM studies of the initial stages of deposition of metal on oxide substrates show surprisingly narrow cluster size distributions for a variety of metal-oxide systems. Furthermore, tunneling spectroscopy of isolated metal clusters on oxides indicates that the interface contact potential of small clusters is size dependent. Finally, recent STM observations of cation and anion segregation to oxide surfaces appear to be limited by the development of surface charge. We attempt to bring these observations together in the development of a general model that considers the effect of local charge on surface morphological evolution.
10:40 AM		SS+NS-ThM-8 Kinetics of Mound Formation in Epitaxial Growth J. Amar, F. Family (Emory University) The results of kinetic Monte Carlo simulations of epitaxial growth on fcc(100) and bcc(100) surfaces are reported. We have used our model to simulate Fe/Fe(100) deposition at room temperature and have compared our results with recent experiments. Excellent agreement is found for the selected mound angle, coarsening exponent n, and kinetic roughening exponent \beta as well as for the mound morphology. The general dependence of the surface skewness, mound angle and coarsening kinetics on temperature, deposition rate, and strength of the step barrier to interlayer diffusion is also studied and compared with recent experiments. For the case of a very large step barrier we find n = 1/3 which is significantly larger than found in previous models but in agreement with recent experiments on Rh/Rh(111). The results of an analytic calculation of the surface current and selected mound angle as a function of the Ehrlich-Schwoebel step barrier and substrate temperature are also presented. Depending on the sign of the step barrier and the magnitude of the prefactor for diffusion over a step various scenarios are possible, including the existence of a critical temperature T\sub c\ for mound formation above which (for a positive step barrier) or below which (for a negative step barrier) quasi-layer-by-layer growth will be observed. For the case of Fe/Fe(100) deposition our calculation implies an upper bound for T\sub c\ which is consistent with experiment. The weak parameter-dependence of our estimates for the mound angle at room temperature confirms and explains the good agreement found in previous estimates assuming different values of the step barrier.
11:00 AM		SS+NS-ThM-9 STM Evidence for the Step Density Interpretation of RHEED Intensity Oscillations in GaAs(111)A and (110) Homoepitaxy T. Jones, D. Holmes, J. Sudijono, C. McConville, B. Joyce (Imperial College, United Kingdom) Scanning tunnelling microscopy (STM) has been used to study in detail the growth of the first few monolayers (MLs) of GaAs on GaAs(111)A and GaAs(110) surfaces by molecular beam epitaxy. Specific emphasis has been directed at the relationship between the variation in the surface step density, as measured by STM, and in situ RHEED intensity oscillation studies. GaAs(111)A homoepitaxy is relatively straightforward. At low depositions (0.25 ML), a high density of two dimensional (2D) islands is nucleated on the large terraces of the smooth starting surface. Coalescence of the islands occurs with increasing deposition (0.5 ML), and by 1 ML this is largely complete, confirming the layer-by-layer growth mode deduced from the RHEED intensity oscillations. The cycle continues with subsequent deposition and the growth front becomes distributed over several incremental layers, consistent with the damping and overall intensity decrease of the RHEED oscillations. GaAs(110) homoepitaxy is more complex because of the variety of growth modes that occur on this surface orientation. The monolayer growth mode is analogous to that seen on GaAs(111)A with the RHEED oscillations reflecting the temporal variation in surface step density. The bilayer growth regime is more complicated, however, since islands exhibiting both a single (2 =C5) and double layer height (4 =C5) are seen in STM= images. A superposition of the monolayer and bilayer periods is observed and the RHEED beam appears to be sensitive only to the monolayer steps. The role of the bilayer steps is to delay the progression of monolayer step completion.
11:20 AM		SS+NS-ThM-10 Giant Interfacial Stress in Metal on Metal Heteroepitaxy H. Ibach, A. Grossmann (Forschungszentrum J\um U\lich, IGV, Germany) Near equilibrium, the growth modes in epitaxial growth are determined by the specific free surface energies of the substrate \gamma\ \sub S\ , the deposited material \gamma\ \sub a\ and the specific free energy of the interface \gamma\ \sub i\. Layer by layer growth, at least for the first few layers, is expected, if \gamma\ \sub S\ \>\ \gamma\ \sub a\ \+\ \gamma\ \sub i\. For metal on metal systems, it is in general assumed that the interfacial energy \gamma\ \sub i\ is small. Hence, good candidates for layer by layer growth should be deposits of low surface energy on substrates bearing a large surface energy. In this paper we report on results which call this concept into question. By employing the cantilever method, we have measured the change in the surface \/\ interfacial stress during the deposition of Ag on Pt \(\111\)\ and also Cs on Ni \(\111\)\. For both systems we find a huge stress which increases linear with the thickness of the deposited layer up to a thickness of three layers, from where the induced stress begins to level off. The stress is an order of magnitude larger than expected from the elastic deformation of the deposited film due to the misfit of the lattice constant. It is proposed that the stress arises from the transfer of electronic charge from the adlayer into the substrate. It is furthermore shown that the huge stress makes the Ag \/\ Pt \(\111\)\ system instable and that the stress is relieved by annealing to higher temperatures via interfacial mixing.
11:40 AM		SS+NS-ThM-11 The Effect of Substrate Misorientation on Nucleation and Strain Relaxation for Epitaxially Grown CaF\sub 2\ on Si(111) Substrates B. Kim, C. Ventrice, Jr., L. Schowalter (Rensselaer Polytechnic Institute); T. Thundat (Oak Ridge National Laboratory) The effect of substrate misorientation on the surface morphology and strain relaxation mechanisms of MBE-grown CaF\sub 2\ overlayers on vicinal Si(111) surfaces has been studied with AFM and in-situ RHEED. At 1045K, CaF\sub 2\ growth is initiated by the formation of a reacted CaF layer followed by the complete overgrowth of an additional CaF\sub 2\ monolayer (ML). Growth of CaF\sub 2\ beyond these initial two MLs depends on the degree of miscut of the Si substrate. On Si substrates tilted toward the [1 1 -2] azimuth by a miscut angle \>=\ 0.5\super o\, the atomic step edges on the Si surface bunch together forming flat terraces that are ~200 nm wide. Further growth on this orientation proceeds only through the nucleation and lateral propagation of relatively thick CaF\sub 2\ islands of constant height. For CaF\sub 2\ films grown on substrates with a miscut angle < 0.5\super o\, the CaF\sub 2\ layer remains relatively uniform without the thick island nucleation/propagation mode. The onset of dislocation formation, through the formation of trigons, is observed in CaF\sub 2\ films grown on "on-axis" substrates for films thicker than 3 nm with ex-situ AFM. As observed by Tromp et al. with LEEM, the observation of dislocations emanating from the trigon corners, gives strong evidence that these dislocations are <1 -1 0> edge dislocations with their Burgers vector parallel to the interface. The CaF\sub 2\ films grown on "off-axis" substrates do not form these dislocation networks. Instead, the thick CaF\sub 2\ islands remain atomically flat. This data suggests that when heteroepitaxy proceeds by the nucleation and coalescence of thick islands, strain relief proceeds by a completely different mechanism.