ICMCTF2015 Session B7: Computational Design and Experimental Development of Functional Thin Films

First-principles calculations have proved to be a useful tool to assist knowledge-based coating design. Their major role is to provide an insight and explanation to complement experimental observations, as well as to yield material-related trends to explore new horizons for evermore demanding applications. In this presentation we will discuss adsorption phenomena of single atoms as well as simple molecules on various surfaces of a wide variety of materials, and thereby address topics of adhesion transfer, surface diffusion, and friction.

When implanting Al and Fe atoms into the TiN (001) surface, Al atoms need to overcome approximately 40% higher energy implantation barrier than Fe atoms. However, the implantation force for both species is approximately the same, though the repulsive forces act closer to the surface on Al and further from the surface on Fe atoms.

Calculations of potential energy landscapes reveal that O experiences the same diffusion barrier on both, CrN and TiN (001) surfaces. On the other hand, the same procedure suggests that H ad atoms are significantly more mobile on TiN than on CrN surfaces.

A comparison of adsorption energies and potential energy landscapes for H, O,-OH, and H2O groups on Si-doped graphene clearly show preferential attraction to Si-sites as compared to C-sites. These results combined with ab initio molecular dynamics simulations of gas interactions with the a-C surface rationalise the low-friction effect in Si containing a-C thin films experimentally observed at increased temperatures beyond 240 ∘C.

The sliding of components is known to result in third body formation, which characteristically consists of mechanical mixing and grain refinement in the near surface region as well as transfer film formation on the counterface. Understanding the behavior of these so called third bodies can help to optimize the tribological response of sliding components by selecting the right combination of materials, environmental and contact conditions (e.g. pressure, contact area). This study investigates the tribolayer properties at the interface of metal and ceramic (i.e. WC/W) sliding contact using various experimental approaches and classical atomistic simulations. Experimentally, nanoindentation and micropillar compression tests as well as adhesion mapping by means of atomic force microscopy are used to evaluate the strength of tungsten-carbon tribolayers. In order to capture the influence of environmental conditions, a detailed chemical and structural analysis is performed on the worn surfaces by means of XPS mapping and depth profiling along with transmission electron microscopy of the debris particles. Experimentally, the results indicate a decrease in hardness and modulus of the worn compared to the unworn one. Atomistic simulations of nanoindentation on deformed and undeformed specimens are used to probe the strength of the WC tribolayer and despite the fact that the simulations do not include oxygen, the simulations correlate well with the experiments on deformed and undeformed surfaces, where the difference in behavior is attributed to the bonding and structural differences of amorphous and crystalline W-C. Adhesion mapping suggests a decrease in surface adhesion, which based on chemical analysis is attributed to surface passivation.

Gold films attract great interests because of their desirable properties and are widely used in various electric and electronic devices. The stability of gold films is of great importance to the long-term performance and reliability of these devices. Passivation layer is widely used to package metallic films to obtain stable performance.

In this paper, response of gold film with and without passivation layer during thermal treatment and natural storage was studied. Substrate curvature method was employed to study the residual stress based on the calculation with Stoney equation. Surface topography and morphology of the samples were observed by atomic force microscope (AFM) and scanning electron microscope (SEM). Results show that the presence of passivation layer altered the response of gold film to thermal treatment. Influence of passivation material and layer thickness on the performance of gold films was also analyzed and discussed.

To systematically explore and optimize the design of functional thin films the availability of computational tools that allow to understand and predict the relation between atomistic growth mechanisms and the technologically desired properties of the thin film is crucial. A challenge in this respect is the complexity of realistic conditions in thin film growth: The growth occurs under conditions that are often far away from thermodynamic equilibrium and intended as well as unintended forign chemical elements are commonly present in high concentrations. As a consequence the surfaces present under actual growth conditions have little in common with the chemically clean and ideal structures that are accessible to experimental surface science studies.

Atomistic ab initio calculations that are free of any materials-specific input parameters have evolved over the last two decades into a powerful tool to study the structure, the chemistry as well as the kinetics of surfaces under realistic growth conditions. The key ideas behind this new approach and the opportunities which it opens will be discussed for a few selected examples: (i) The identification of suitable selective surfactants to control the roughness/smoothness of thin films, (ii) the design of atomistically controlled growth strategies to achieve doping levels well above/below thermodynamic equilibrium, and (iii) the application of surface phase diagrams to control adatom kinetics and spinodal decompositions.

Transition metal nitride thin films are largely used nowadays in a wide range of applications from microelectronics to mechanical machining operations. Predictive design of these materials can be achieved by using multi-scale computational approaches to mimic real deposition conditions and gain insights on the microstructural and morphological evolutions during growth. The accurate description of growth kinetics and energetics remains a challenging task, especially for magnetron sputtering for which particle bombardment may alter the resulting microstructures.

To explain experimental findings of TiN thin film growth regimes during sputter- deposition, different approaches can be used. We chose to perform kinetic Monte Carlo (kMC) simulations using a 3D rigid lattice model. At this time only one 3D lattice orientation is considered. The lattice is obtained by a superposition of two FCC sub-lattices, each of them corresponding to a kind of atom (metal sub-lattice and N sub-lattice). For both kinds of particles, two main events are taken into account, 3D diffusion and deposition event.

A 3D diffusion event can take place, obeying certain conditions that fulfil the symmetry of the system, between sites belonging to the same sub-lattice. To compute the diffusion barrier to a vacant site, different diffusion models have been used, all of them taking into account the interaction of the diffusing atom with its nearest and next nearest neighbours. Energy barriers were taken from existing DFT data of the literature. Together with a given diffusion model a deposition model has been used as well. Starting with normal incidence deposition for both kinds of particles, the deposition model has been extended to take into account the different aspects of a real sputter deposition experiment (e.g., angle distribution from SIMTRA) or GLAD geometry.

Different combinations of diffusion-deposition models will be firstly discussed. Then, for a given model, the influence of the most important experimental parameters (substrate temperature, deposition rate, angular distribution of particles arriving at the substrate, stoichiometry) on the resulting morphology of the growing 3D structure will be presented. Together with 3D diffusion some microscopic mechanisms (interaction of the incident particle with an already existing 3D object in the neighbourhood of its trajectory, reflection or diffusion into the volume) have been also included.

The Modified Embedded Atom Method (MEAM) interatomic potential within the classical Molecular Dynamics (MD) framework enables realistic, large-scale simulations of important model materials such as TiN. As a step toward understanding atomistic processes controlling the growth of TiN on a fundamental level, we perform large-scale simulations of TiN/TiN(001) deposition using a TiN MEAM parameterization which reproduces experimentally-observed surface diffusion trends, correctly accounts for Ehrlich barriers at island step edges [1], [2], and has been shown to give results in excellent qualitative and good quantitative agreement with Ab Initio MD based on Density Functional Theory (DFT) [3], [4]. Half a monolayer of TiN is deposited on 100x100 atom TiN(001) substrates at a rate of 1 Ti atom per 50 ps, resulting in simulation times of 125 ns. The TiN substrate is maintained at a typical epitaxial growth temperature, 1200 K during deposition using Ti:N flux ratios of 1:1 and 1:4 with incident atom energies of 2 and 10 eV to probe the effects of N₂ partial pressure and substrate bias on TiN(001) growth modes. We observe nucleation of Ti_xN_y molecules; N₂ desorption; the formation, growth and coalescence of mixed <100>, <110>, and <111> faceted islands; as well as intra- and interlayer mass transport mechanisms. For equal flux ratios at 2 eV incidence energy, islands begin to form atop existing islands at coverages ≳ 0.25 ML, leading to 2D multilayer growth. At 10 eV, the film growth mode shifts toward layer-by-layer growth. We discuss the implications of these results on thin film growth and process tailoring. Our classical MD predictions are supported and complemented by DFT-MD simulations.

[1] D. G. Sangiovanni, D. Edström, L. Hultman, V. Chirita, I. Petrov, and J. E. Greene, “Dynamics of Ti, N, and TiNx (x=1–3) admolecule transport on TiN(001) surfaces,” Phys. Rev. B, 86, 155443 (2012).

[2] D. Edström, D. G. Sangiovanni, L. Hultman, V. Chirita, I. Petrov, and J. E. Greene, “Ti and N adatom descent pathways to the terrace from atop two-dimensional TiN/TiN(001) islands,” Thin Solid Films, 558, 37 (2014).

[3] D. G. Sangiovanni, D. Edström, L. Hultman, I. Petrov, J. E. Greene, and V. Chirita, “Ab initio and classical molecular dynamics simulations of N2 desorption from TiN(001) surfaces,” Surf. Sci., 624, 25 (2014).

[4] D. G. Sangiovanni, D. Edström, L. Hultman, I. Petrov, J. E. Greene, and V. Chirita, “Ti adatom diffusion on TiN(001): Ab initio and classical molecular dynamics simulations,” Surf. Sci., 627, 34 (2014).

In the present work, the CVD process for MT-Ti(C,N) coating in the vertical hot-wall reactor was studied through the Computational Fluid Dynamics (CFD) based on the Finite Volume Method (FVM). By means of commercial FLUENT software, the flow characteristics and temperature gradients in the preheater was simulated and validated with the benchmark solutions from the experimental measurement. The computational model was then applied to investigate the deposition procedure of MT- Ti(C,N) coating from TiCl₄-CH₃CN-N₂-H₂ gas mixture. The thermal and hydrodynamic characteristics of the flow within the reactor was simulated. The influence of concentration of gas species on the growth rate of Ti(C,N) coating deposited on specimens, which are located at different position of the reactor, is predicted. The computational predictions of the growth rate are in agreement with the experimental measurements. The CFD approach is of general validity and applicable to optimize the process parameters and to provide theoretical guidance for improving coating uniformity in thermal CVD process.

The financial supports from the Doctoral Scientific Fund Project of the State Education Committee of China (Grant No. 20120162110051) and the National Natural Science Foundation of China (Grant nos. 51371201 and 51371199) are greatly acknowledged.

Most commonly known hard transition-metal nitrides crystallize in rocksalt structure (B1). The discovery of ultraincompressible pyrite-type PtN₂ 10 years ago has raised a question about the cause of its exceptional mechanical properties. We answer this question by a systematic computational analysis of the pyrite-type PtN₂ and other transition-metal pernitrides (MN₂) with density functional theory. Apart from PtN₂, the three hardest phases are found among them in the 3d transition-metal period. They are MnN₂, CoN₂, and NiN₂, with computed Vickers hardness (H_V) values of 19.9 GPa, 16.5 GPa, and 15.7 GPa, respectively. Harder than all of these is PtN₂, with a H_V of 23.5 GPa. We found the following trends and correlations that explain the origin of hardness in these pernitrides. (a) Charge transfer from M to N controls the length of the N-N bond, resulting in a correlation with bulk modulus, dominantly by providing Coulomb repulsion between the pairing N atoms. (b) Elastic constant C44, an indicator of mechanical stability and hardness is correlated with total density of states at EF, an indicator of metallicity. (c) Often cited monotonic variation of H_V and Pugh’s ratio with valence electron concentration found in rocksalt-type early transition-metal nitrides is not evident in this structure. (d) The change in M-M bond strength under a shearing strain indicated by crystal orbital Hamilton population is predictive of hardness. This is a direct connection between a specific bond and shear related mechanical properties. This panoptic view involving ionicity, metallicity, and covalency is essential to obtain a clear microscopic understanding of hardness. This work was funded by NSF CMMI 1234777, and has been accepted in Phys. Rev. B (2014).