ICMCTF2012 Session B7-1: Computational Design and Experimental Development of Functional Thin Films

In this study we employ an ultra-fast plasma based film deposition process termed High Power Impulse Magnetron Sputtering (HiPIMS) to control the kinetics during growth of thin silver (Ag) films. In HiPIMS, power is applied to a conventional magnetron sputtering source in short unipolar pulses of several tens to several hundreds of µs with energies ranging from several tens to several hundreds of mJ at frequencies smaller than 2 kHz. We use the time-dependent character of the HiPIMS process to modulate the flux of the film forming species seeking to control the adatom diffusion times and the nucleation process. In practice this is achieved by applying power to a Ag sputtering cathode in three different ways:

(i) In pulses of constant energy and width at various frequencies ranging from 50 Hz to 1000 Hz.

(ii) In pulses of constant width, at a certain pulsing frequency varying the pulse energy from 20 mJ to 400 mJ.

(iii) In pulses of constant energy and at a constant pulsing frequency by varying the pulse width from 50 to 500 µs.

Time- and energy-resolved mass spectrometry is employed to study the effect of the process parameters on the temporal profile of the deposition flux. Films are deposited on Si (100) substrates covered by ~2 nm native SiO₂ layer. In-situ measurements of the evolution of the residual stresses in the films using a multiple-beam optical stress sensor are combined with ex-situ imaging techniques, such as Atomic Force Microscopy, to study the nucleation characteristics and determine the film thickness at which a continuous film is formed (percolation thickness). It is shown that by increasing the pulsing frequency from 50 Hz to 1000 Hz the percolation thickness decreases from 206 Å to 146 Å. Depositions are also performed using the continuous DC Magnetron Sputtering (DCMS) process at same range of growth rates (0.1 to 10 Å/s) as those used in HiPIMS. Analysis of the DCMS grown films shows that the percolation thickness of ~160 Å does not undergo significant changes. The latter indicates that control over the nucleation and growth is a unique feature of the time domain of the pulsed process.

In this study the critical, maximum and optimum velocity of a single cold sprayed (CS) particle is estimated using the smoothed particle hydrodynamics method (SPH) by evaluating the impact shape, the coefficient of restitution and the rebound and deposit energy ratio. The contact surfaces of the particle and substrate is modeled as intersurface forces using the Dugdale-Barenblatt cohesive zone model. The application of the SPH method permits simulation of the CS process without the use of mesh and thus avoids the disadvantages of traditional numerical method in handling large deformations and tracing moving interfaces. The impact of cold sprayed particles is simulated using various metals and powder particle sizes. The materials are classified into four impact cases (soft/soft, hard/hard, soft/hard, and hard/soft), according to their physical and mechanical properties. The influence of powder and substrate material and particle size on the particle velocity range for deposition and the optimum velocity is discussed. Furthermore the numerically estimated critical, maximum and optimum velocity of the soft/soft impact case agreed well with previously obtained experimental values.

When a solid Cu substrate gets in contact with liquid Sn at 250°C for 5 seconds, two intermetallic layers with the sequence Cu₃Sn ( phase) and Cu₆Sn₅ ( phase) are formed at the substrate covered by a remaining thin layer of Sn. The typical thickness of these layers are 80 nm ( phase) and 1000 nm ( phase), building an intermetallics film on the Cu-substrate. The thermodynamic extremal principle [1] (in Ziegler's formulation) is used for the treatment of the evolution of the Cu-Sn system with the assumption of no sources and sinks for vacancies in the bulk, see e.g. [3], or non-ideal sources and sinks for vacancies in the bulk, see [4]. The interfaces between the individual phases are assumed either as ideal sources and sinks or no sources and sinks for vacancies. If these models are used for the simulation of the kinetics, no quantitative agreement is obtained for the current diffusion couple (the simulated kinetics is more than one order of magnitude slower).

The microscopic observations indicate that the newly formed intermetallic layers have a sub-micron grain structure and, thus, grain boundary diffusion can significantly contribute to the system kinetics. That is why the recent thermodynamic concept is extended so that it incorporates also the grain boundary and interface diffusion in both intermetallic phases and, thus, each intermetallic layer may develop by an interaction of bulk and interface/grain boundary diffusion. This occurs in a pronounced way in the phase, which forms a scalloped morphology, see e.g. [4]. Finally a thin film with a strongly varying microstructure develops, which is captured by the current model. An effective tracer diffusion coefficient for each component in each phase can be derived by means of the thermodynamic extremal principle [1]. Also the branching (polyfurcation) of the initial contact (Kirkendall) plane, i.e. the original solid Cu, liquid Sn contact plane, is shown, which may be exploited for the determination of the effective tracer diffusion coefficients of Cu and Sn in the and phase. A comparison of the modelling results with a series of experimental results is provided.

[1] J. Svoboda, J., I. Turek, I., F.D. Fischer, Phil. Mag., 85, 3699-3707, 2005

[2] J. Svoboda, E. Gamsjäger, F.D. Fischer, E. Kozeschnik, J. Phase Equilibria Diff., 27, 622-628, 2006

[3] J. Svoboda, F.D. Fischer, R. Abart, Acta mater., 58, 2905-2911, 2010

[4] M.S. Park, R. Arróyave, Acta mater., 58, 4900-4910, 2010

Pseudobinary transition metal nitride alloys, such as V-W-N and V-Mo-N, exhibit an electronic structure consisting of alternating high and low electron density regions, according to DFT calculations [1], leading to a simultaneous enhancement in hardness and ductility. Thus, such alloy systems are promising candidates as coatings with enhanced toughness. Recently, we demonstrated the growth of epitaxial V_xW_1-xN thin films on MgO(001) by reactive magnetron sputtering [2]. Here, we explore the effect of growth variables on the mechanical properties of V_xMo_1-xN on MgO(001).

V_xMo_1-xN thin films with 0.3≤x≤1, as determined by EDX and RBS, were deposited on MgO(001) substrates by dual reactive magnetron sputter epitaxy (MSE) in a 5 mTorr Ar+N₂ gas mixture. The substrate temperature was varied from 100 to 700 °C. For the entire composition range, at T_s = 700 °C, we obtain single-phase cubic-B1-structure V_xMo_1-xN as determined by XRD. Cross-sectional HR-TEM and SAED confirms the cubic V_xMo_1-xN solid solution viewed along both <100> and <110> zone axes. The lattice parameter varies from a = 4.149 Å for x = 0.8 to a = 4.167 Å for x = 0.4, increasing with increasing Mo content. High- resolution XRD reciprocal space mapping of a V_0.4Mo_0.6N film grown at 700 ºC shows that the film is relaxed with a = c = 4.167 Å. At temperatures between 500 and 300 °C, with x = 0.5, we obtained single phase B1 V_xMo_1-xN(001) films.

The nanoindentation hardness H and elastic modulus E of the V_xMo_1-xN alloy films are H = 14.3 GPa and E = 286 GPa for x = 0.6, and H = 12.3 GPa and E = 243 GPa for x = 0.4.

[1] D.G. Sangiovanni, L. Hultman, V. Chirita, Acta Mater 59 (2011) 2121-2134

[2] H. Kindlund et al., ICMCTF 2011, Abstract #266

We present the results of Kinetic Monte Carlo (KMC) studies on dislocation and crack formation in strained nanometre thick coatings. The Kinetic Monte Carlo (KMC) computer program developed (NUKIMOCS) enables atomistic simulation of the mechanical processes which occur at the nanoscale in thin films. The code performs off-lattice KMC calculations to determine the strain in a material at the atomic scale and how this changes with time. Off lattice KMC allows atoms to occupy any position in space as in a real atomic lattice with interaction between atoms defined by interatomic potentials. Therefore, both the elastic and kinetic properties of an off-lattice KMC model are entirely defined by the interatomic potential. The underlying physics of strain-induced microstructural evolution can therefore be presented accurately and on meaningful time and length scales.

The parameters of the interatomic potentials have been determined by fitting the energy surface obtained from first principles calculations, the elastic constants and the coefficient of thermal expansion to the unit cell. Different initial and boundary conditions were applied in order to study the behaviour of defects such as vacancies and dislocations. The local relaxation process was performed after every 10 simulation steps within a radius of three interatomic distances from the location of the event. The energy barrier between two states - the activation energy - was calculated by the ‘frozen crystal approximation’. For energy minimisation we employed a limited-memory BFGS method.

Early transition metal nitrides (TMN) and their alloys with Al are widely used in various protective coatings due to their outstanding mechanical and thermal stability. The increasing demand on coating performance from the application-side often requires sophisticated designs of the protective thin films. The major routes in a knowledge-based materials selection and design are by architecture (e.g. use of multilayer or patterning) by alloying, and by the combination of these two.

In the present paper we use Density Functional Theory (DFT) calculations to address a topic related mostly to alloying, i.e. to assess the extend of Vegard's-like linear behaviour of ternary TM-Al-N and TM-TM-N, and quaternary TM-TM-Al-N (TM=Sc, Ti, Zr, Nb, Ta, Hf) alloys. In particular, we will discuss the compositional dependence of lattice parameters, energies of formation, bulk moduli and elastic properties.

As an example we will show that although some systems for various properties exhibit reasonably linear dependence on composition (e.g. lattice parameter of cubic Ti_1–xAl_xN), in general it is dangerous to blindly use the so-called „Vegard's rule“. A consequence of the non-linear dependence of the lattice constants on e.g., the driving force for decomposing the metastable c-TM-Al-N phases will be discussed. On another front, detailed calculations of single crystal elastic constants of ZrN-AlN system reveal that the elastic responce of the cubic ternary system is very similar to ZrN up to AlN mole fraction ~40%, only after which the elastic properties change drastically towards the AlN elastic behaviour.

Enhanced toughness in hard and superhard thin films is a primary requirement for present day ceramic hard coatings, known to be prone to brittle failure during in-use conditions, in modern applications. Based on the successful approach and results obtained for TiN- and VN-based ternary thin films [1,2], we expand our Density Functional Theory (DFT) investigations to TiAlN-based quarternary thin films. (TiAl)_1-xM_xN thin films in the B1 structure, with 0.06 < x < 0.75, are obtained by alloying with M = V, Nb, Ta, Mo and W, and results show significant ductility enhancements, hence increased toughness, in these compounds. Importantly, these thin films are also predicted to be hard/superhard, with similar and/or increased hardness values, compared to TiAlN. For (TiAl)_1-xW_xN these results have experimentally been confirmed recently [3]. The general, electronic mechanism responsible for the ductility increase is rooted in the enhanced occupancy of d-t_2g metallic states, induced by the valence electrons of substitutional elements (V, Nb, Ta, Mo, W). This effect is more pronounced with increasing valence electron concentration (VEC), and, upon shearing, leads to the formation of a layered electronic structure, consisting of alternating layers of high and low charge density in the metallic sublattice. This unique electronic structure allows a selective response to tetragonal and trigonal deformation: if compressive/tensile stresses are applied, the structure responds in a “hard” manner by resisting deformation, while upon the application of shear stresses, the layered electronic arrangement is formed, bonding is changed accordingly, and the structure responds in a “ductile/tough” manner as dislocation glide along the {110}<1-10> slip system becomes energetically favored [2]. The findings presented herein open new avenues for the synthesis of hard, yet tough, ceramic coatings, by tuning the VEC of alloying elements to optimize the hardness/toughness ratio in relevant applications.

[1] D. G. Sangiovanni et. Al. Phys. Rev. B 81 (2010) 104107.

[2] D. G. Sangiovanni et. Al. Acta Mater. 59 (2011) 2121.

[3] T. Reeswinkel et. Al. Surf. Coat. Technol. 205 (2011) 4821.

New III-Nitride technology involves 1 -, 2- and 3-dimensional (nanowires, quantum wells, quantum dots) nanostructures as the building blocks of emerging novel photonic and electronic device applications. This technology, although one of the most “environment-friendly” available in the market is still far from being mature and hence devices are far from their intrinsic limits. Much more research efforts are needed to address materials related issues which are the bottleneck against the rapid advances of these emerging fields. State-of-the-art transmission electron microscopy (TEM) along with the associated spectroscopies comprise the key techniques for the structural characterisation of these heterostructured materials systems down to the atomic scale and it should be interactively combined with computational design and modeling of structures, defects and properties.

Materials issues encountered by TEM involve: a) An atomic- scale investigation of interfacial and defect structures, b) Understanding of defect introduction mechanisms and related phenomena. C) Local strain field and chemistry d) Electronic structure of defects and interfaces.

In this presentation, examples will be presented in which an hierarchical integrated multiscale framework is employed comprising high resolution TEM (HRTEM), quantitative HRTEM (qHRTEM), analytical methods provided in the scanning TEM (STEM) such as energy dispersive X-ray spectroscopy (EDX) and high-angle-annular-dark-field (HAADF) or Z-contrast imaging combined with computational modeling. Results of empirical interatomic potential simulations and density functional theory (DFT) calculations will illustrate modeling of the energetically favorable defect/interface structures and electronic properties. Image simulations using the resulting models for correlation with the corresponding experimental HRTEM images will be also shown.