Study of density effect of large gas cluster impact by molecular dynamics simulations

Large gas cluster impacts cause unique surface modification effects because a large number of target atoms are moved simultaneously due to high-density particle collisions between cluster and surface atoms. Molecular dynamics (MD) simulations of large gas cluster impacts on solid targets were carried out in order to investigate the effect of high-density irradiation with a cluster ion beam from the viewpoint of crater formation and sputtering. An Ar cluster with the size of 2000 was accelerated with 20 keV (10 eV for each constituent atom) and irradiated on a Si(1 0 0) solid target consisting of 2 000 000 atoms. The radius of the Ar cluster was scaled by ranging from 2.3 nm (corresponding to the solid state of Ar) to 9.2 nm (64× lower density than solid state). When the Ar cluster was as dense as solid state, the incident cluster penetrated the target surface and generated crater-like damage. On the other hand, as the cluster radius increased and the irradiation particle density decreased, the depth of crater caused by cluster impact was reduced. MD results also revealed that crater depth was mainly dominated by the horizontal scaling rather than vertical scaling. A high sputtering yield of more than several tens of Si atoms per impact was observed with clusters of 4-20× lower volume density than solid state.     

Computer simulation of sputtering at the low index (1 0 0), (1 1 0) and (1 1 1) surfaces of Ni3Al in a STEM

The present study is relevant to the preferential Al sputtering and/or enhancement of the Ni/Al ratio in Ni₃Al observed by the scanning transmission electron microscopy fitted with a field emission gun (FEG STEM). Atomic recoil events at the low index (1 0 0), (1 1 0) and (1 1 1) surfaces of Ni₃Al through elastic collisions between electrons and atoms are simulated using molecular dynamics (MD) methods. The threshold energy for sputtering, E_sp, and adatom creation, E_ad, are determined as a function of recoil direction. Based on the MD determined E_sp, the sputtering cross-sections for Ni and Al atoms in these surfaces are calculated with the previous proposed model. It is found that the sputtering cross-section for Al atoms is about 7-8 times higher than that for Ni, indicating the preferential sputtering of Al in Ni₃Al, in good agreement with experiments. It is also found that the sputtering cross-sections for Ni atoms are almost the same in these three surfaces, suggesting that they are independent of surface orientation. Thus, the sputtering process is almost independent of the surface orientation in Ni₃Al, as it is controlled by the sputtering of Ni atoms with a lower sputtering rate.     

Large-scale molecular dynamics simulation of sputtering process with glancing-angle Ar cluster impact on 4H-SiC

Large-scale molecular dynamics simulations with two Ar₆₈₈ cluster impacts on a 4H-SiC surface are performed to investigate the mechanism of lateral sputtering caused by two clusters collisions. The two Ar clusters are composed of 688 atoms each (referred as Ar₆₈₈) which are described by a simple Lennard-Jones potential. The initial velocities of both clusters are 2.55 × 10⁴ m/s when the acceleration voltage is 100 keV. The computational volume is 30 nm × 30 nm × 16 nm, which is constructed by 1444608 4H-SiC atoms. At 0.8 ps after the impact from the first Ar cluster on the 4H-SiC surface, a second argon cluster with predetermined incident-angle collides with 4H-SiC surface at a distance of one “diameter” away from the center of the first impact where the term “diameter” refers to the diameter of the footprint of the first impact on 4H-SiC. The incident-angle of the second argon cluster was set at 0°, 60°, or 80° for three different trials. Consequently, in each case the crater formed by the first cluster showed signs of being smeared out by the impact of the second cluster. Especially at the incident-angle of 80° the effects of surface modification were clearly noticeable.     

Molecular simulations of cluster ejection by CO2 cluster impact on carbon-based surfaces

Single (CO₂)_N (N = 1-20) cluster impact on three different carbon-based surfaces of fullerite (1 1 1), graphite and diamond (1 0 0) has been investigated by MD simulations with the cluster collision energy from 5 to 14 keV/cluster as a first step toward the general modeling of the reactive sputtering by cluster impact of a solid surface. A crater permanently remained on the fullerite and graphite surfaces while it was quickly replenished with fluidized carbon material on the diamond surface. In spite of the smaller crater size as well as the crater recovery resulting in the reduction of the surface area, the sputtering yields were the highest on diamond. The effective energy deposition near the surface contributes to the temperature rise and consequent sputtering seemed highly reduced due to the collision cascades especially on the fullerite target.     

An attempt to solve Andersen’s puzzle in surface segregation during alloy sputtering

The paper addresses CuPt alloy sputtering by Ar ions and discusses the well-known experiment performed by Andersen et al. 25 years ago, but not yet properly explained. The atomistic (binary-encounter) simulation has been applied to extract the concentrations of surface Cu and Pt atoms from the experimental data. The results of simulations favor segregation of Cu at all bombarding energies studied experimentally (1.25-320 keV). It has been shown that some mysterious results of the experiment can be explained by a reconstruction of the surface undergoing sputtering. For forecasting purposes, the sputtering of CuPt alloy with 0.25-1 keV Ar ions is also considered.     

Dynamics of cluster induced sputtering in gold

Impacts of 0.13-1.4 MeV Au₁₃ clusters onto Au(1 1 1) target are investigated in molecular dynamics simulations. The evolution of sputtered Au atoms and clusters are simulated up to 10 ns. The total sputtering yield, angular and velocity distributions of the sputtered material, as well as dimensions of impact induced craters are compared to recent experimental results. It is shown that the experimental observations can be explained by a flow of atoms from the craters. Secondary cluster ejection from crowns formed around the craters is found to be one of the main mechanisms of sputtering. The results are summed up in an empirical model.     

Modelling of deposition processes on the TiO2 rutile (1 1 0) surface

Deposition of Ti_xO_y clusters onto the rutile TiO₂ (1 1 0) surface has been modelled using empirical potential based molecular dynamics. Deposition energies in the range 10-40 eV have been considered so as to model typical deposition energies of magnetron sputtering. Defects formed as a function of both the deposition energy and deposition species have been studied.The results show that in the majority of cases Ti interstitial atoms are formed, irrespective of whether Ti was contained within the deposited cluster. Furthermore that the majority of these interstitials are formed by displacing a surface Ti atom into the interstitial site. O surface atoms are also relatively common, with Ti and TiO₂ surface units often occurring when the deposited cluster contains Ti but becoming less frequent as the deposition energy is increased. Structures that would give rise to the growth of further layers of rutile are not observed and in the majority of the simulations the energy barriers for diffusion of the end-products is high.     

Molecular dynamics simulations of large fluorine cluster impact on silicon with supersonic velocity

The etching process by very large reactive gas cluster impact was investigated by molecular dynamics (MD) simulations. Fluorine-molecule clusters with the size up to 100,000 atoms (50,000 F₂ molecules) were irradiated on silicon (1 0 0) targets at supersonic velocity regime (0.1-1 eV/atom, 1.0-3.2 km/s). The MD simulations revealed that the existence of threshold energy-per-atom around 0.3 eV/atom (1.75 km/s) to cause surface deformation and enhancement of Si desorption. When the incident energy-per-atom is less than the threshold, the incident cluster breaks up itself on the target without surface deformation. The fluorine molecules in the cluster spread in the lateral direction along the target surface, and some part of them decompose and adsorbs on the target to form silicon fluoride composites. On the other hand, the clusters penetrate the surface of silicon target when the energy-per-atom is larger than 0.3 eV/atom. In these collisional processes, the target surface is deformed to create shallow crater shape. The incident fluorine molecules are preferentially concentrated at the bottom of the crater, which resulted in high desorption yield of silicon as in the form of SiF₂, SiF₃ and SiF₄.     

Optical emission from Be, Cu and CuBe targets during ion beam sputtering

A spectral structure of the radiation (190-590 nm) emitted during sputtering of polycrystalline Cu, Be and CuBe targets by Kr⁺ ions with 5 keV have been presented. Evolution of surface composition during ion beam sputtering is investigated. Several time scales are distinguished, corresponding to different processes: the elimination of surface contaminants, the removal of the corroded layer. The implications for the use of ion beam optical spectroscopy in surface analysis are discussed. In the case of Be and Cu98 Be2, a molecular structure appears between 492 nm and 502 nm. It is similar for both samples and is ascribed to de-excitation of BeH.     

Effect of impact angle and projectile size on sputtering efficiency of solid benzene investigated by molecular dynamics simulations

Molecular dynamics computer simulations have been used to investigate the effect of the cluster size on the sputtering yield dependence on the impact angle. Ar₃₆₆ and Ar₂₉₅₃ cluster projectiles with 14.75 keV of incident energy are directed at the surface of a solid benzene crystal described by a coarse-grained representation at angles between 0° and 70°. It is observed that the shape of the angular dependence of sputtering efficiency is strongly affected by the cluster size. For the Ar₃₆₆ cluster, the sputtering yield only slightly increases with the impact angle, has a broad maximum around 40°, and decreases at larger angles. For the Ar₂₉₅₃ cluster, the yield strongly increases with the impact angle, has a maximum around 45° followed by a steep decrease at larger angles. For both investigated cluster projectiles the primary energy is deposited so close to the surface so that the sputtering efficiency only weekly benefits from the shift of the deposited energy profile toward the surface which occurs at larger impact angles. In this study, molecular dynamics computer simulations are used to probe the effect of the impact angle on the efficiency of ejection molecules emitted from solid benzene by 14.75 keV Ar₃₆₆ and Ar₂₉₅₃ clusters.     

Sputtering of clusters from copper-gold alloys

Polycrystalline Cu, Cu₂₀Au₈₀, Cu₄₀Au₆₀, Cu₈₀Au₂₀ and Au samples were bombarded with 15 keV Ar⁺, and the resulting secondary neutral yield distribution was studied by non-resonant laser post-ionisation mass spectrometry. Neutral clusters containing up to 15 atoms were observed for the targets. The yield of neutral clusters, Cu_mAu_n−m, containing n atoms, Y_n, was found to follow a power in n, i.e. Y_n∝n^-δ, where the exponent δ varied from 5.2 to 10.1. For a fixed n, the cluster yields showed a variation with number of copper atoms, m, much greater than expected for a binomial distribution suggesting that the clusters are not formed randomly above the surface and a component of preformed cluster emission occurs. In addition, the cluster compositions from the sputtered alloys were indicative of sputtering from a copper rich surface.     

Fundamentals of surfactant sputtering

We introduce a new sputter technique, utilizing the steady-state coverage of a substrate surface with up to 10¹⁶ cm⁻² of foreign atoms simultaneously during sputter erosion by combined ion irradiation and atom deposition. These atoms strongly modify the substrate sputter yield on atomic to macroscopic length scales and therefore act as surfactant atoms (a blend of “surface active agent”). Depending on the surfactant-substrate combination, the novel technique allows enhanced surface smoothing, generation of novel surface patterns, shaping of surfaces and formation of ultra-thin films. Sputter yield attenuation is demonstrated for sputtering of Si and Fe substrates and different surfactant species using 5 keV Xe ions at different incidence angles and fluences up to 10¹⁷ cm⁻². Analytical approaches and Monte Carlo simulations are used to predict the sputtering yield attenuation as function of surfactant coverage. For sputtering of Si with Au surfactants we observe high sputter yields despite a steady-state surfactant coverage, which can be explained by strong ion-induced interdiffusion of substrate and surfactant atoms and the formation of a buried Au_xSi surfactant layer in dynamic equilibrium.     

Growth of silicon nitride films by bombarding amorphous silicon with N ions: MD simulation

In this study, the molecular dynamics simulation method was employed to investigate the growth of silicon nitride films by using N⁺ ions, with energies of 50, 100, 150 and 200 eV, to bombard an amorphous silicon surface at 300 K. After an initial period of N⁺ bombardment, saturation of the number of N atoms deposited on the surface is observed, which is in agreement with experiments. During subsequent steady state deposition, a balance between uptake of N by the surface and sputtering of previously deposited N is established. The Si(N_x) (x = 1-4) and N(Si_y) (y = 1-3) bond configurations in the grown films are analyzed.     

Proton sputtering from polycrystalline Al surface interacting with highly charged ions at grazing incidence angle

Sputtering processes of protons from a polycrystalline Al surface interacting with Ar^q+ (q = 3-14) ions at a grazing incidence angle (∼0.5°) were investigated. The intensity of protons (I_H) detected in coincidence with scattered Ar atoms was measured as a function of q. I_H saturated at q ? 10, although it increased rapidly with q at 3 ? q ? 8. The angular distribution of protons with low kinetic energy (?2 eV) began to deviate from the cosine distribution and assumed a rather flat equidistribution as q increased. To analyze the sputtering processes of protons at the grazing incidence angle, a modified model of the “above-surface potential sputtering model” was proposed by considering image acceleration of projectile ions.     

Experiment and simulation of cluster emission from 5 keV Ar → Cu

The abundance distribution of neutral Cu_n clusters sputtered by 5 keV Ar impact from a polycrystalline Cu surface is measured using single-photon laser post-ionization. Molecular dynamics computer simulation is used to gain insight into the cluster sputtering process. Three different codes and two potentials are used to check the sensitivity of the results on numerics and physical input. Differences in the results obtained by the various codes and the different potentials used are discussed. While the total sputter yield is consistent with experiment, the fraction of atoms bound in clusters, and in particular the dimer fraction, are overestimated by at least a factor of 4. This is also true for a many-body potential which has been fitted to describe both bulk Cu and dimers. In detail, the simulation shows that larger clusters are emitted at later times from the target. Clusters originate mainly from regions of the surface, which are around the melting temperature of bulk Cu. Large clusters are emitted preferably from ion impacts with a high individual sputter yield. Finally, we simulate sputtering from a model Cu material with an artificially decreased cohesive energy. Here, drastic high-yield events (up to Y=78) can be observed, which produce clusters abundantly.     

Sputtering induced by cluster impact on metal targets: Influenceof electronic stopping

The effect of electronic stopping on the sputtering of metals by cluster impact is discussed. We focus on the specific case of Au₁₃ impact on a Au surface. Using molecular-dynamics simulation, we study several strategies to include electronic stopping. Electronic stopping influences both the magnitude of the sputter yield and the duration of the sputter process. In the usual procedure, electronic stopping only affects sufficiently fast atoms with kinetic energies above a threshold energy, which is of the order of the target cohesive energy. When assuming that electronic stopping holds down to thermal energies <1 eV, or even to 0 eV, the collision spike is rapidly quenched and the sputter yields become unrealistically small. Furthermore, we implement a scheme to include electronic stopping based on local (electron) density information readily available in a simulation.     

Incidence-angle dependent Cs incorporation in Si during low-energy bombardment: A dynamic computer simulation study

The implantation of Cs atoms in silicon was investigated by dynamic computer simulations using the Monte-Carlo code T-DYN that takes into account the gradual change of the target composition due to the Cs irradiation. The incorporation of Cs atoms was studied for incidence angles ranging from 0° to 85° and for four impact energies (0.2, 0.5, 1 and 3 keV). The total implantation fluences were (1-2) × 10¹⁷ Cs/cm², well above the values required to reach a stationary state. The steady-state Cs surface concentrations exhibit a pronounced dependence on impact angle and energy. At normal incidence, they vary between ∼0.57 (at 0.2 keV) and ∼0.18 (3 keV), but decrease with increasing incidence angle. Under equilibrium, the partial sputtering yield of Si exhibits the typical dependence on incidence angle, first increasing up to a maximum value (at ∼70°-75°) and declining sharply for larger angles. For all irradiation conditions a strongly preferential sputtering of Cs as compared to Si atoms is found, increasing with decreasing irradiation energy (from 4.6 at 3 keV to 7.2 at 0.2 keV) and for nearer-normal incidence.     

Study of crater formation and sputtering process with large gas cluster impact by molecular dynamics simulations

Molecular dynamics (MD) simulations of large argon clusters impacting on silicon solid targets were performed in order to study the transient process of crater formation and sputtering. The MD simulations demonstrate that the initial momentum of incident cluster is transferred to target surface atoms through multiple collision mechanism, where the initial momentum, which is along to the surface normal before impact, is deflected to lateral direction. This momentum transfer process was analyzed by the calculation of the velocity at the crater edge (the interface between cluster and target). In the case of Ar₁₀₀₀ cluster impact on Si(1 0 0) target at low energy per atom less than 40 eV/atom, the maximum value of lateral velocity of the crater edge increases in proportional to the velocity of incident cluster atoms. On the other hand, the crater edge velocity saturates over 40 eV/atom of incident energy per atom. In this case, the whole of constituent cluster atoms are implanted into the target and expand in both lateral and reflective directions at the subsurface region of the target. These MD simulations demonstrated that this collisional process result in the high yield sputtering of the target atoms.     

Modelling the erosion of beryllium carbide surfaces

Redeposition of beryllium eroded from main chamber plasma facing components of ITER onto the divertor material carbon creates a mixed material, beryllium carbide Be₂C, whose interaction with the plasma is not well known. In this study, we have investigated the erosion of Be₂C by deuterium using molecular dynamics simulations and ERO impurity modelling. We found that beryllium sputters preferentially over carbon and identified the sputtering mechanism in the ion energy range 10-100 eV to be both physical and swift chemical sputtering. In addition to single atoms, different types of small molecules/clusters were sputtered, the most frequently occurring molecules being BeD, Be₂D, and CD. The sputtering threshold was found to lie between 10 and 15 eV. The MD sputtering yields were used in plasma impurity simulations, serving as a replacement for input data obtained with TRIM. This changes the accumulation rate of impurity Be in the divertor region compared to previous estimates.     

Electronic sputtering of Gd3Ga5O12 and Y3Fe5O12 garnets: Yield, stoichiometry and comparison to track formation

The results of present paper have shown that sputtering of yttrium iron garnet (Y₃Fe₅O₁₂) under swift heavy ions in the electronic energy loss regime is non-stoichiometric. Here we are presenting additional experimental results for gadolinium gallium garnet (Gd₃Ga₅O₁₂) as target. The irradiations were performed with different ions (⁵⁰Cr (589 MeV), ⁸⁶Kr (195 MeV) and ¹⁸¹Ta (400 MeV)) impinging perpendicularly to the surface. As earlier, the sputtering yield was determined by collecting the emitted gadolinium and gallium atoms on a thin aluminium foil, placed upstream above the target and analyzing the Al catcher by Rutherford backscattering. Also for Gd₃Ga₅O_12, the emission of Gd and Ga is non-stoichiometric. Sputtering appears above a critical electronic stopping power of S_th = 11.6 ± 1.5 keV/nm, which is larger than the threshold for track formation, in agreement with other amorphisable materials. In addition, the angular distribution of the sputtered species was measured for Y₃Fe₅O₁₂ and Gd₃Ga₅O₁₂ using 200 MeV Au ions impinging the surface at 20° relatively to the surface. For the two garnets the ratio of Y/Fe (and Gd/Ga) varies with the angle of emitted species and the stoichiometry seems to be preserved only for an emission perpendicular to the surface.