the properties of thin films from external conditions, including substrate temperature, deposition rate, angular distribution of atoms, and incident kinetic energy. The interplay between some deposition parameters often renders the task to obtain the desired film morphology challenging. On the other hand, the control of the desired morphology is essential for the applications searched for. Predictions from computational approaches are therefore very helpful in this sense.

We have developed a 3D kinetic Monte Carlo (kMC) atomistic code to help understand the role of the elementary atomistic diffusion mechanisms and impact of deposited energy on the resulting thin film growth morphology. The simulation model aims at mimicking the full sputter-deposition process, from ejection of atoms from the target, their transport in the gas phase, and film formation. First-principle calculations are also implemented to determine the potential energy landscape for preferable adsorption sites, or energy barrier for surface diffusion.

In the work presented here, the code has been applied to simulate the growth of TiN and Cu thin films. Examples will be provided for the growth at either normal or oblique angle incidence. The code is capable of reproducing the development of TiN columnar morphology, with column tilt angle in good agreement with experimental findings. For the case of Cu, the growth proceeds in a 3D mode, with nucleation of isolated islands, percolation and formation of continuous layer. The influence of kinetic energy on film morphology and defect incorporation will be also discussed.

TiN-based materials are widely established as protective coatings for cutting tools. Grain interiors (single crystal regions) could serve as reservoirs for functional species, e.g. Al or V, which provide effective lubrication and wear protection at high temperatures by diffusing to the coating surface where they form a protective oxide scale (Al) or a lubricious oxide to reduce friction (V). The microstructure of real coatings has a polycrystalline or nanocomposite nature. Here, the intergranular region consists of quasi-amorphous (amorphous materials interfaced with TiXN, X=Al, V) tissue which serves as a diffusion highway for transporting species from bulk to the coating surface.

While approaches exist for estimating diffusivity in crystalline materials, the situation is much less explored in the case of (quasi-)amorphous materials. In this contribution, we will report on our recent progress in addressing issues related to the atomistic modeling of mass transport. To do so we employ complementary computational investigations to determine elementary point-defect migration mechanisms in crystalline as well as amorphous materials and subsequently their relative rates. Molecular dynamics (MD), which traces the motion of each particle contained in the system by numerically solving Newton’s equations, coupled to the accuracy of density functional theory (DFT) to describe interatomic forces, is the most reliable computational tool to calculate atomic jump rates as a function of temperature. We will present the DFT-based “5-frequency model” allowing us to calculate the diffusion coefficient in the crystalline fcc material. Furthermore, we also demonstrate that pressure has a notable impact on the diffusivity of V, Al, and Ti in TiN. Next, we show how to approach both crystalline and amorphous materials by employing the Green-Kubo relation together with a detailed analysis of diffusion pathways as provided by MD.

Density functional theory calculations using the all electron Wien2k code with the core-hole approach were used to calculate the N 1s and Ti 2p core electron binding energy shifts in TiN_x and TiO_xN_y. Considered point defects include N and Ti vacancies, different N and Ti interstitials and Frenkel pairs. It was shown that the majority of point defects have a significant influence on the binding energy of its first neighbours. For example, a single Ti vacancy decreases the binding energy of its nearest nitrogen neighbours by 0.6 eV. The aim of the ab initio calculations was to create a fingerprint database which could be used when analyzing experimental data.

The ab initio calculations were later utilized when analyzing XPS data from TiN and TiON thin films deposited by reactive DC and HPPMS magnetron sputtering. Thereby we were able to identify point defect-introduced features in the XPS spectra and, combined with lattice parameter and composition data, propose a quantification method. This work represents an important step on the road towards a more precise point defect identification and quantification in these materials.

Molybdenum disulfide, the most studied member of the transition metal dichalcogenides family, has been used as solid lubricant for several decades, showing extremely low friction coefficients[1] and stability to high temperature. Its lubricating properties are ascribed to the weak van der Waals interactions between sulfur atoms in the crystalline layered structure. Moreover MoS₂, even when prepared in the amorphous state or made of randomly oriented domains, can undergo shear induced structural transitions to the more ordered layered state affecting its tribological properties[2].

Exploiting a reactive classical force field[3] able to treat explicitly formation and breaking of bonds, we investigated by means of molecular dynamics simulations, the shear-induced structural changes and the possible layer formation in amorphous molybdenum disulfide. The ordering process is studied in details, with particular regard to the estimation of the thermodynamic properties that govern the process itself. A connection with crystallization theories is finally found, conferring a predictive power to the achieved results.

Overall, this study aims at gaining an atomic level understanding of the dynamics of layer formation process in MoS₂, thus controlling and possibly improving its tribological properties.

References:

[1] J.M. Martin et al., Phys. Rev. B, 48, 10583(R) (1993).

[2] J. Moser, F. Lévy, Thin Solid Films, 228, 257 (1993).

[3] T. Liang et al., Phys. Rev. B, 79, 245110 (2009).