On the other hand, intermixing of two different-sized components blurs the heterointerface and thus is highly detrimental for heterostructures. Intriguingly, these two modes of strain relaxation may couple. This has been widely discussed in the context of “strain-enhanced diffusion”.

In this talk, we present our latest work which shows that these two modes can couple in an entirely different way, with important implications for the thermal stability of strain-relaxed template layers. Specifically, strain relaxation by dislocations can suppress intermixing between the heterolayer and the substrate. Intermixing is suppressed because, once the strain is fully relaxed by dislocations, intermixing would actually increase the strain.

We demonstrate this effect by carrying out Monte Carlo (MC) atomistic simulations for the thermodynamic equilibrium of heterolayer films. Continuum modeling for the thermodynamics and the kinetics of these systems reproduces the MC results. In this way, we can predict the extent and rate of intermixing in thin films as a function of temperature, degree of relaxation and other parameters. Our MC equilibrium results for Ge on effectively thin substrates like silicon-on-insulator (SOI) show a significant suppression of alloying in the film. The effect is even stronger for InAs on GaAs, due to the larger misfit. Continuum modeling allows us to extend these results to very thick (effectively semi-infinite) substrates, where we find a dramatic slowing-down of intermixing over long times.

The effect, which is of interest not only for planar films, but also for any structure relaxed by dislocations, including islands and other nanostructures, opens new possibilities for the control of stress and property optimization in semiconductor devices.

Theoretical studies based on first-principles calculations have provided increased understanding of mixing and decomposition thermodynamics in Ti_1-xAl_xN and related refractory metastable solid solutions. In this work we move on from studies of the bulk towards investigation of surfaces and growth processes. We present results from calculations of the effects of Al-addition on the atomistic processes on TiN and Ti_1-xAl_xN surfaces. These have impact on texture evolution and short range clustering tendencies during thin film growth.

The thin film growth mechanism of materials with a NaCl (B1) structure such as TiN and MgO is well understood. Therefore, MgO was chosen as a model material to investigate the compositional effect when Mg is replaced by other elements (M= Al, Cr, Y, Ti and Zr) to form a mixed oxide Mg(M)O thin film.

Kinetic arguments related to the mobility of the adparticles [1] explain the growth mechanism of the pure MgO thin films. The kinetic conditions also explain other observations on increasing the M content, namely the change in the preferential orientation of the MgO crystallites and the formation of a solid solution between MgO and the oxide of the metal M. However, when the M content in the thin films increases, XRD pole figures reveal a clear transition from a highly crystalline to a fully amorphous structure. These transitions were noticed for all systems studied, and its origin can not be explained from the same kinetic arguments.

The analogy between this transition and the crystalline to liquid transition noticed in the packing of hard spheres suggests a thermodynamic basis. The good correspondence between MD simulations and the experimental observations allows to check this idea. The transition can be understood from a lowering of the packing density of the filled octahedral positions in the MgO structure, similar to the hard sphere system. As the transition is a first order transition, the model predicts a mixture of crystalline and amorphous material at the phase boundary. This prediction is confirmed by TEM analysis which shows the presence of nanocrystals in an amorphous matrix.

This fundamental research can assist in tailoring the physical properties of materials as the latter depend strongly on the crystallographic properties of the thin films as indicated by hardness, surface energy and optical measurements.

[1] S. Mahieu, P. Ghekiere, D. Depla, and R. De Gryse, Thin Solid Films 515, 1229-1249 (2006).

In deposition of compounds by reactive sputtering is difficult to associate high deposition rates with a precise control of the film stoichiometry due to the formation of compound over the target surface, which leads to target poisoning and, as consequence, problems of stability. A typical feature of reactive sputtering is the hysteresis phenomenon observed from the partial pressure of the reactive gas (p) as function of the reactive gas flow rate (Q_total). In order to describe this process, a reliable model was developed, called Berg’s model, which allows the understanding of relations between the experimental parameters and the hysteresis loop. From Berg’s model is possible to predict a decrease in the hysteresis loop using small target areas, small collecting areas (small target-to-substrate distances) or non-realistic pumping speeds. On the other hand, an alternative configuration, based in conventional magnetron devices, presents itself as a way to suppress the hysteresis loop in, for example, TiN deposition. This system, called triode magnetron sputtering (TMS), consists of a grid inserted between the target and the substrate. This grid works as the discharge anode and the variation of the grid-to-target distance modifies the plasma impedance. This arrangement allows that all enclosed parameters can operate in a wider range than those used in conventional magnetron devices. In this work, an adaptation from Berg’s model was done to the TMS in order to describe the reactive sputter deposition of TiN at different effective grid areas. In the adapted model, in addition to the balance equations for the target and collecting area (substrate and chamber walls), another equation was written to the grid. As the grid is inserted between the target and collecting area, only the sputtered material, which is not collected by the grid, can be deposited over the collecting area. The results shown that increasing the effective grid area decreases the hysteresis loop until completely suppress it, which means that the role of the collecting area decreases.

There is a pressing need for a far better scientific understanding of the thermodynamics and transport properties of fluid mixtures in small systems. Small systems exist at the interfaces between solids and fluids, fluids and fluids, pores, and at critical points. In engineering, one such small system might be the multiphase formation of mixtures of polymers or monomers condensing on the surface of a substrate which lead to unpredictable morphologies. The transport properties of these complex mixtures over very long periods of time cannot be predicted using elementary approaches, for which no adequate theoretical treatment has been developed

The gas-liquid critical point, or consolute point in a fluid mixture, provides a convenient physical model whereby the size of the system (in terms of the correlation length) may be adjusted to be of macroscopic proportions and the physics may be probed using the scattering of optical photons. It is in the interplay between "smallness" and non-linearity that even very weak symmetry-breaking fields, such as gravity or the surface potential, may have enormous influence on the structure and transport properties of a condensing film.

This work explores the theoretical framework with which structure and morphology may be predicted in condensing polymeric films by considering the effects of diverging correlation lengths ξ at or near critical consulate points. Experimental methods based on elastic and inelastic light scattering technique are further presented as a means of validating the theory of small systems.

For many years, Al₂O₃ has been one of the most important materials in CVD-coatings, used in tools for metal machining applications. The first Al₂O₃ coated tungsten carbide cutting insert from Sandvik Coromant was introduced already in 1975 [1]. The room-temperature stable α-phase is the most commonly used Al₂O₃ polymorph. More than one decade ago it was found that the addition of a small amount of H₂S to the gas vastly enhanced the growth rate of α-Al₂O₃ and eliminated the dog-bone effect such that a uniform deposition of the edges and flat surfaces of the insert could be achieved [1].

There have been studies of the effect of H₂S on the macroscopic growth and gas phase reactions [1-4], but although there are indications that the H₂S interaction with the Al₂O₃ surface is strong [5, 6], there are no published results where the influence of H₂S on growth of α-Al₂O₃ has been proved on an atomic scale.

In this Paper we investigate the effect of H₂S on CVD-growth of α-Al₂O₃ on an atomistic level. We have applied a multiscale approach, where we use thermodynamic modeling (THERMOCALC and the Scientific Group Thermodata Europe (SGTE) database SSUB3) for gas phase reactions, and Density Functional Theory (DFT) [7] for surface reactions.

We use the ThermoCalc equilibrium calculations and data from non-equilibrium kinetic modeling [8], to obtain the relevant molecules used as input for our first principles calculations. The heat of formation of a number of possible relevant reactions of these molecules at three different surfaces and with different surface terminations of α-Al₂O₃ were calculated. The internal energy of the reaction was calculated using DFT, whereas the remaining part of the free energy was added from the SSUB3 database. Reaction barriers were calculated using the Nudge Elastic Band (NEB) approach. The effect of H2S compared to H₂O on the surface of Al₂O₃ was studied in detail. From our study we are able to present a possible scenario, supported by our theory, where H₂S acts as a catalyst in the CVD -growth of α-Al₂O₃.

[1] P. Mårtensson, Surface & Coatings Technology 200 (2006) 3626.

[2] K.H. Smith, J.N. Lindström, US Patent, #4 619 866 (1986).

[3] T. Oshika, A. Nishiyama, K. Nakaso, M. Shimada, K. Okuyama, J. Phys., IV France 9 (1999) 877.

[4] T. Oshika, M. Sato, A. Nishiyama, J. Phys., IV France 12, 113 (2002).

[5] O. Saur, T. Chevrau, J. Lamotte, j. Travert, J-C. Lavalley, J. Chem. Soc 77 (1981) 427.

[6] C.R. Apesteguia, J.F. Plasa de los Reyes, T.F. Garetto, J. M. Parera, Applied Catalysis 4 (1982) 5.

[7] G. Kresse and J. Furthmüller, Phys. Rev. B 54 11169 (1996).

[8] P. Tan, J. Muller, D. Neushutz, J. of Electrochemical Soc. 152 288 (2005).

In general, the substrates which require surface modification using overlay coatings are three dimensional in nature, and consequently, normal incidence of the depositing flux cannot be ensured over the entire substrate area. This off – normal incidence can affect the columnar structure development and orientation of the coatings. The column orientation subsequently becomes the function of the angle of incidence of the sputtered flux resulting in anisotropies in macroscopic properties of the functional coatings. The present study aims to understand the structure development of coatings deposited on planar substrates kept at various inclinations with respect to the depositing flux using characterising tools like scanning electron microscopy and X – ray diffraction. The following aspects were focused on,

i. to identify the columnar microstructure and relate the column orientation with the angle of incident flux

ii. to study the effect of various deposition parameters like deposition time and duration on the column orientation

iii. the structure evolution of elemental coatings at the atomistic level is also simulated based on non – lattice, ballistic growth models

iv. to simulate the effect of deposition geometry on the column orientation

Coatings of copper and titanium nitride were deposited by magnetron sputter deposition method on silicon wafers of size 1 cm x 1 cm. The former is important in micro electronics industry and the later has wide tribological applications. The substrates were mounted symmetrically below the target race – track on holders which could align them in different inclinations (0 to 90 degrees). The planar magnetron used had a target diameter of 75 mm. Elemental targets of copper and titanium were used with argon as the sputtering gas. Stoichiometric TiN was deposited in a mixture of argon and nitrogen gas introduced at suitable flow rates using flow controllers. The deposition of TiN was done for different durations and that of Cu was done for different magnetron bias conditions so as to control the sputter yield and hence the deposition rate.

Well defined columnar microstructure of the coatings was revealed in the cross sectional micrographs of both types of coatings studied. The XRD analysis also revealed strong dependence of the substrate inclination angle on the crystallite size and preferred orientation of the diffraction planes.

Details of the characterisation and simulation results regarding the effect of the various deposition parameters considered will be presented.

Alumina thin films deposited by PVD are of high interest due to their outstanding mechanical and thermal properties. The crystal structure of these films is usually referred to as γ-Al₂O₃, however the crystal structure of γ-Al₂O₃ is not well defined even not for bulk materials. Alumina thin films were deposited by reactive dual magnetron sputtering (dms) at 550 °C on cemented carbide substrates. The alumina grain size is smaller than 50 nm measured by dark field imaging in the transmission electron microscope (TEM). The crystal structure was analyzed by energy-filtered electron diffraction showing a characteristic background due to a disordered structure different from γ-Al₂O₃ and any other known alumina phase. Indicating this we call the structure of the thin films pseudo γ-Al₂O₃. Reflections belonging to γ-Al₂O₃ with d lattice spacings between 0.2 nm and 0.4 nm were not visible but an intensity distribution corresponding to an amorphous structure was detected. Texture as origin for the missing reflections could be ruled out. Diffraction analysis was performed at room temperature and -145 °C and showed the static character of the disorder. Radial intensity distribution functions were determined from diffraction patterns and were compared to x-ray diffraction data published for γ-Al₂O₃. Differences between the two crystal structures were highlighted and discussed with respect to lattice spacings and intensities of the various reflections. For analyzing the crystal structure of these materials, energy filtering is absolutely necessary to identify reflections with small intensity on top of a large background. The films contained Ar with a mole fraction of about 5.6 at% as shown by energy dispersive x-ray (EDX) spectroscopy in the TEM. The plasmon energy of the disordered phase was found to be 26.1 eV as determined by electron energy-loss spectroscopy (EELS). This value corresponds to that of bulk alpha alumina [15]. A model is presented showing Ar and Al atoms as being primari ly responsible for the disorder. Mechanical properties of the films were investigated by nanoindenting. A Young modulus of 315±8 GPa was found. For bulk sapphire the Young modulus is 381±2 GPa and for the CVD deposited α-Al₂O₃ the Young modulus is 315.8±2 GPa.