Session H42: Metals on Metals

Effects of shadowing and steering in oblique-incidence metal growth

The effects of oblique incidence on the surface morphology in metal (100) epitaxial growth are studied using a simplified model which includes shadowing but not the effects of short-range and long-range attraction of depositing atoms to the surface. Surprisingly, we find that many of the qualitative features observed in oblique incidence Cu(100) growth, including the existence of anisotropy in the submonolayer regime, as well as of a transition from anisotropic mounds to ripples perpendicular to the beam with increasing deposition angle, can be explained primarily by geometrical effects. We also find that the formation of (111) facets is crucial to the development of well-ordered ripples at large angles of incidence. A second transition from ripples oriented perpendicular to the beam to `rods' with (111) facets oriented parallel to the beam is also found at very high deposition angles and film thicknesses. When the effects of short- and long-range interactions are included in our simulations, we find two main effects. In the submonolayer regime, attraction tends to weaken the effects of shadowing and reduce the submonolayer anisotropy. However, in the multilayer regime `flux-focusing' due to long-range attraction tends to enhance the anisotropy and reduce the critical thickness/angle for the ripple transition. Near the transition from ripples to rods, sideways attraction also tends to stabilize the isotropic phase as is observed experimentally.   

Multiscale simulations of oblique-incidence Cu/Cu(100) epitaxial growth

We present the results of multiscale simulations of oblique-incidence Cu/Cu(100) epitaxial multilayer growth carried out in order to explain the experimentally observed ripple formation at large deposition angles with respect to the substrate normal. Our method combines a kinetic Monte Carlo (KMC) simulation for the thermal surface diffusion with a small-scale one-atom molecular dynamics (MD) simulation of every deposition event, in order to account correctly for shadowing as well as the short-range and long-range attraction of depositing atoms to the surface. These simulations are particularly challenging both because of the existence of significant finite-size effects at large deposition angles, as well as because of the computational time required to simulate the deposition process, and therefore an efficient algorithm for carrying out parallel simulations of deposition will be described. The relative importance of shadowing and long-range attraction, as well as the dependence of the surface morphology on such parameters as the Ehrlich-Schwoebel barrier, edge- and corner-diffusion, and deposition flux will be discussed. Preliminary results in which both the substrate atoms and the depositing atom undergo molecular dynamics in the final stages of deposition will also be presented for comparison.   

Molecular Dynamics Simulations of Upward Diffusion of Adatoms and Clusters on Facetted fcc Metal (110) Surfaces

We study upward self-diffusion of adatoms and small clusters across the outer edges of the mounds formed in fcc metal(110) homoepitaxy using molecular dynamics simulations with interatomic potentials described by the embedded-atom method. Our results show that both single adatoms and small clusters on the (111) and (100) facets of Al and Cu(110) can readily diffuse upwards and cross the outer edge of the mounds, but with different atomistic mechanisms. An adatom crosses the outer edge via a simple place exchange or indirect exchange mechanism. In contrast, the upward diffusion and outer-edge crossing of small clusters is realized by their dissociation at the edge of the mound after one or two cluster atoms are incorporated into the edge. Our simulations reveal that there truly exists efficient upward mass transport in homoepitaxy on facetted fcc metal (110) surfaces.   

Ab initio calculation of atomic interactions on Al(110): implications for epitaxial growth

Using first-principles calculations based on density-functional theory, we resolved atomic interactions between adsorbed Al atoms on Al(110). Relevant pair and trio interactions were quantified. We find that pair interactions extend to the third in-channel and second cross-channel neighbor on the anisotropic (110) surface. Beyond these distances, pair interactions are negligible. The nearest-neighbor interaction in the in-channel direction is attractive, but nearest-neighbor cross-channel interaction is repulsive. While nearest-neighbor, cross-channel repulsion does not support the experimental observation of 3D hut formation in Al/Al(110) homoepitaxial growth [1], we find that trio interactions can be significant and attractive and they support cross-channel bonding. The pair and trio interactions have direct and indirect components. We have quantified the electronic and elastic components of the indirect, substrate-mediated interactions. We also probe the influence of these interactions on the energy barriers for adatom hopping. [1] F. Buatier de Mongeot, W. Zhu, A. Molle, R. Buzio, C. Boragno, U. Valbusa, E. Wang, and Z. Zhang, Phys. Rev. Lett. 91, 016102 (2003).   

Multi-scale simulation of quantum dot formation in Al/Al (110) homoepitaxy

In experimental studies of Al(110) homoepitaxy, it is observed that over a certain temperature window (330-500K), 3D huts, up to 50 nm high with well defined and smooth (111) and (100) facets, form and self-organize over the micron scale [1]. The factors leading to this kinetic self-organization are currently unclear. To understand how these structures form and evolve, we simulated multi-layer, homoepitaxial growth on Al(110) using ab initio kinetic Monte Carlo (KMC). At the high temperatures, where nano-huts form, the KMC simulations are slow. To tackle this problem, we use a technique developed by Devita \& Sander [2], in which isolated adatoms make multiple moves in one step. We achieve high efficiency with this algorithm and we explore very high temperatures on large simulation lattices. We uncover a variety of interesting morphologies (Ripples, mounds, smooth surface, huts) that depend on the growth temperature. By varying the barriers for various rate processes, we discern the factors that determine hut sizes, aspect ratios, and self-organization. [1] F. Buatier de Mongeot, W. Zhu, A. Molle, R. Buzio, C. Boragno, U. Valbusa, E. Wang, and Z. Zhang, Phys. Rev. Lett. 91, 016102 (2003). [2] J.P. Devita \& L.M. Sander, Phys. Rev. B 72, 205421 (2005).   

Positron trapping at quantum-dot like Cu nano-particles embedded in Fe and submonolayer films of Au and Pd deposited on Cu(100) surface.

Recently clear evidence has been provided that positron spectroscopy can be used to characterize the properties of quantum-dot-like nano-particles embedded in host material even at dilute levels as a result of the preferential trapping of positrons in the nano-particles. The results of studies of sputtered surfaces of the Fe-Cu alloy with quantum-dot like Cu nano-particles embedded in the top atomic layers of Fe and submonolayer films of Au and Pd deposited on Cu(100) using Positron-Annihilation-Induced Auger-Electron Spectroscopy are analyzed by performing calculations of positron surface states and annihilation characteristics. Estimates of the positron binding energy, work function and annihilation characteristics performed for studied surfaces reveal their sensitivity to nano-particle size and coverage. Trapping of positrons at nano-particles on studied surfaces is determined from calculated positron surface state wave functions and comparison of theoretical core annihilation probabilities with experimental ones estimated from the measured Auger peak intensities.   

Adatom Diffusion on Ag(100) and Cu (100) Surface with Steps: insights from \textit{ab initio} electronic structure calculations

We present results of calculations of the activation barriers for an adatom (Ag or Cu) diffusing on terraces of Cu(100) and Ag(100), with steps on them. Our results based on the density functional theory (DFT) with the generalized gradient approximation, present systematic differences with those obtained using semi-empirical potentials. For the Ag systems, the latter are always higher than the former by about 40 meV, except for the case of hopping over the step where it is higher by 130 meV. The opposite is the case for Cu for which lower values are obtained by semi-empirical potentials than by DFT, except for the case of hopping over the step where this value is overestimated by the semi-empirical potentials. In examining the subtleties in the differences between the Ag and Cu systems, analysis of the d-band of the adatom in the hollow and bridge site show a systematic shift of the band towards higher energies when moving from hollow to saddle points. Moreover, Ag systems show a large narrowing of the d-band as opposed to the Cu cases where it is not noticeable. For adatom diffusion on the terraces (without steps) of Ag(100) and Cu(100), our calculations are in agreement with previous results.   

Quantum Size Effect on Adatom Surface Diffusion

Using scanning tunnelling microscopy, we demonstrate the nucleation density of Fe islands on surface of nanoscale Pb films oscillates with film thickness, providing a direct manifestation of quantum size effect on surface diffusion. The Fe adatom diffusion barriers were derived to be 204 and 187 meV on a 21 and 26 monolayer (ML) Pb film, respectively, by matching the kinetic Monte Carlo simulations to the experimental island densities. The effect is further illustrated by growth on wedged Pb films, where the Fe island density is consistently higher on the odd-layer films than on the even-layer films in the thickness range of 11 to 15 ML.   

Multisite Interactions in Lattice-Gas Models of Adsorbates: Reconciling Adatom Relaxations at Steps

In a lattice-gas (LG) framework for (111) cubic surfaces, pair interactions cannot distinguish A and B steps, but an orientation-dependent three adatom (trio) interaction can, as we verify with VASP for Cu(111).\footnote{TJS et al., PRB a){\bf 70}('04)245404, b){\bf 71}('05)245414, c){\bf 73}('06)115426.}$^b$ However, on Pt(111), small clusters considerably underestimate the difference.\footnote{Feibelman, Surf.\ Sci.\ {\bf 463}('00)L661; Michely et al., ibid.\ {\bf 256}('91)217.} For a sequence of overlayer configurations, we explore the role of lateral relaxations and how they complicate LG analysis. On Cu(100) our prior VASP calculations of a particular trio interaction energy (E$_{\rm d}$) gave a large positive value.$^{3c}$ This nearly cancels the attractive second-neighbor interaction energy (E$_2$), leading to a discrepancy between theory and experiment of step stiffness anisotropy.$^{3a}$ Relaxations at step edges greatly reduce this repulsion. Since position-dependent interactions are improper in LG models, we show how to deal with this phenomenon using a quarto interactions. We comment on extensions to (110) faces and analytic expressions for step stiffness.   

Heterogeneity in ultrathin films simulated by Monte Carlo method

The 3D composition profile of ultra-thin Pd films on Cu(001) has been experimentally determined using low energy electron microscopy (LEEM).$^{[1]}$ Quantitative measurements of the alloy concentration profile near steps show that the Pd distribution in the 3$^{rd}$ layer is heterogeneous due to step overgrowth during Pd deposition. Interestingly, the Pd distribution in the 2$^{nd}$ layer is also heterogeneous, and appears to be correlated with the distribution in the 1$^{st}$ layer. We describe Monte Carlo simulations that show that correlation is due to Cu-Pd attraction, and that the 2$^{nd}$ layer Pd is, in fact, laterally equilibrated. By comparing measured and simulated concentration profiles, we can estimate this attraction within a simple bond counting model. [1] J. B. Hannon, J. Sun, K. Pohl, G. L. Kellogg, \textit{Phys. Rev. Lett.} \textbf{96}, 246103 (2006)   

Diffuse x-ray scattering study of vacancy nanoclusters incorporated into thin Ag(001) homoepitaxial films

In situ diffuse x-ray scattering was used to investigate vacancies that are incorporated into Ag films on Ag(001). Knowledge of the kinetic pathways through which these vacancies are incorporated is fundamental to our understanding of how films and nanostructures grow at the atomic scale. Unexpectedly large vacancy clusters having a volume of 750 ang$^3$ were observed in 100 ML films grown at 150K. Vacancy clusters are also found to incorporate at 300K where the films grow layer-by-layer. Studies of thinner films (5, 10, 20 ML grown at 150K) indicate that the vacancy clusters begin their formation immediately upon deposition and that the vacancies, once initiated, are stable to both their enlargement and subsequent burial. This has important implications for the mechanisms that control the growth and organization of nanostructures.   

Modeling of the Kinetics of Metal Film Growth on 5-Fold Surfaces of Icosohedral Quasicrystals

During submonolayer deposition of metals on 5-f icosohedral Al- Pd-Mn and Al-Cu-Fe surfaces, experimental evidence for several system points to heterogeneous nucleation of islands at specific ``dark star'' trap sites. We model this phenomenon using a mean-field rate equation formulation for Ag on Al-Pd-Mn, where data is available for both the flux and temperature dependence of the island density. We also utilize a more sophisticated kinetic Monte Carlo simulation approach to analyze an atomistic lattice-gas model (for an appropriate ``disordered-bond-network'' of nearest-neighbor adsorption sites) describing nucleation of starfish islands observed by STM for Al on Al-Cu-Fe. Finally, we briefly describe multilayer growth morphologies (which can display kinetic roughening or quantum size effects), but which also generally reflect the submonolayer island distribution. B. Unal et al. PRB 75 (2007); C. Ghosh et al. Phil. Mag. 86 (2006) 831; Surf. Sci. 600 (2006) 1110; V. Fournee et al. PRL 95 (2005) 155504.   

Surface Self-Assembly Driven by Dislocation Annihilation and Glide

The process of growing nanometer sized ordered arrays of clusters on the misfit dislocation networks of strained metallic thin films [1, 2] requires a detailed understanding of the nucleation and film-adsorbate interaction at the atomic level. In the case of sulfur adsorption on submonolayer silver films on 0001 surface of ruthenium, the Ag's short herring bone rectangular misfit dislocation unit cell of 19(18)x16(15) Ag atoms reconstructs into a well-ordered triangular array of S filled Ag vacancy islands 18 Ag atoms apart. Atomically and time resolved variable temperature STM measurements correlated with 2D Frenkel-Kontorova models based on first principles interaction parameters show how a sequential process of Ag vacancy island formation, followed by annihilation of adjacent and opposite pairs of threading dislocation and glide of Shockley partial dislocations generates the uniform triangular array of Ag vacancy islands. We conclude that the strain in the Ag film is the driving force responsible for the surface self-assembly process. 1. Pohl \textit{et al.,} Nature 397, 238 (1999) 2. K. Th\"{u}rmer \textit{et al.}, Science 311, 1272 (2006)