Open Access

Rosaria Brescia

• The present thesis is devoted to the study of the Ir/yttria-stabilized zirconia (YSZ)/Si epitaxial multilayer as a substrate for the nucleation and growth of single-crystal quality diamond films, potentially useful for optical and electronic applications, by plasma-assisted chemical vapour deposition (CVD). Particular attention is devoted to the bias-enhanced nucleation (BEN) treatment, the only method allowing the early generation of a high density of epitaxially oriented diamond grains on Ir. In the first part of this work, the microstructure of epitaxial YSZ films with cubic and tetragonal phases deposited on Si substrates has been analysed by diffraction contrast transmission electron microscopy (TEM). Two-beam bright field imaging of cubic YSZ films allowed to demonstrate the decrease of the threading dislocation density in cubic YSZ films with increasing thickness and/or upon high temperature annealing treatments. In addition to this qualitative result, a quantitative estimationThe present thesis is devoted to the study of the Ir/yttria-stabilized zirconia (YSZ)/Si epitaxial multilayer as a substrate for the nucleation and growth of single-crystal quality diamond films, potentially useful for optical and electronic applications, by plasma-assisted chemical vapour deposition (CVD). Particular attention is devoted to the bias-enhanced nucleation (BEN) treatment, the only method allowing the early generation of a high density of epitaxially oriented diamond grains on Ir. In the first part of this work, the microstructure of epitaxial YSZ films with cubic and tetragonal phases deposited on Si substrates has been analysed by diffraction contrast transmission electron microscopy (TEM). Two-beam bright field imaging of cubic YSZ films allowed to demonstrate the decrease of the threading dislocation density in cubic YSZ films with increasing thickness and/or upon high temperature annealing treatments. In addition to this qualitative result, a quantitative estimation of the dislocation density could be deduced by direct visualisation, in good agreement with former X-ray diffraction (XRD) studies performed on the same systems. These results confirm former TEM studies on the evolution of dislocations in complex oxide systems upon high temperature treatments, indicating a promising pathway to improve the structural quality of this type of films. Epitaxial films of tetragonal YSZ on Si substrates contain different texture components, characterised by the longer axis (the c-axis) of the structure lying parallel or perpendicular to the substrate surface. The combination of TEM imaging and electron diffraction analyses allowed here to identify the different variants as specific features in the films. In contrast to previous literature reports, a higher volume fraction of the material exhibits the c-axis oriented perpendicular to the substrate surface after deposition. With high temperature annealing treatments, a distribution closer to the statistical one, but still with the c-oriented component dominating over the remaining two, was observed. Moreover, in sharp contrast to former reports, no twinning relationship among the various texture components was found for these films. This means that for all orientational variants the three axes of the tetragonal cell are parallel to the base vectors of the Si substrate. This behaviour in particular makes these films more suitable for the deposition of additional epitaxial layers. In the second part of this work experimental and simulation studies have been carried out in order to derive a comprehensive picture of the mechanisms of BEN of diamond on Ir buffer layers. This process is fundamentally different from the analogous process on the more common Si and 3C-SiC substrates. In view of the exceptional epitaxial orientation obtained since the early stages of diamond growth, the mechanisms taking place during BEN on Ir need to be thoroughly clarified. For the first time, BEN has been performed in this work on the Ir/YSZ/Si system that is, as already mentioned, the most interesting from the technological application point of view. On both (001)- and (111)-oriented Ir surfaces, substantial analogies have been observed as for epitaxial Ir films on other substrate systems (Ir/SrTiO3, Ir/Al2O3, Ir/MgO): the product of BEN is an extremely thin (1-2 nm thick) carbonaceous film in which diamond features are extremely tricky to detect. The nuclei for the diamond epitaxial growth are created within localised areas, the domains, and they are too small and defective to generate coherent signals in HRTEM and RHEED. XPD instead can identify a successful BEN treatment, being able to detect a clear pattern ascribable to diamond for the C 1s signal in case the epitaxial nuclei are formed. Extremely short growth steps after BEN were applied in this work with the aim of identifying the earliest appearance of diamond features in HRTEM and RHEED. Both techniques, indeed, start to detect crystalline diamond features after only 5 s growth following the interruption of the BEN treatment. In particular, this stage is very interesting as a transition step between BEN and growth: a mixture of individual diamond grains, extended epitaxial diamond layers and residues of the TEM-amorphous film created during BEN are found. In contrast, after 5 s further growth (10 s total growth time) only individual grains (8 nm wide) are found on top of the Ir substrate. The following evolution of crystalline diamond features, with increasing growth time (from 5 s to 2 min), allows to extrapolate the distribution of diamond nuclei corresponding to the termination of the BEN treatment (0 s growth): these should be 1.6 nm thick grains about 20 nm apart. Moreover, the lateral size extrapolated for the diamond grains at the end of the BEN process would be few nm. In addition to this, the quantitative analysis of the XPD C 1s patterns would suggest the presence of 9 nm wide diamond grains at the end of the BEN treatment. However, structurally perfect diamond grains with these dimensions would be clearly visible both in HRTEM images and in RHEED patterns. These results suggest that XPD patterns are not only generated by structurally perfect single crystalline grains (acting as nuclei for the following epitaxial growth). A substantial contribution may come from a diamond matrix characterised by a highly defective nature, which is not visible by HRTEM and RHEED and which is etched off to a large extent during the first seconds of the deposition process. Attempts were made to generate artificially the damaged diamond phase produced by the BEN procedure on the Ir surface by implantation of a bulk diamond layer with high energy C+ ions. At certain damage conditions, the RHEED pattern of diamond was lost, and a feeble XPD pattern was still observed for the C 1s signal. This result is qualitatively similar to the one obtained after a successful BEN treatment. However, the weakness of the XPD pattern indicates a higher disorder in the ion-implanted films than in the product of successful BEN processes. This is also confirmed by post-implantation growth experiments on thin diamond fims, demonstrating the loss of the majority of epitaxial grains corresponding to the same damage conditions. These results overall suggest that there are profound differences between the defective crystalline structures formed by the incident ions during BEN at high temperature and the statistically distributed defects generated in the keV carbon ion implantation process at room temperature. One additional peculiar aspect of BEN on Ir surfaces consists in the spatial arrangement of the early epitaxial diamond grains, hence of the former nuclei. These are found within the domains, μm-sized patterns which are often observed as isolated regions on the Ir substrates. At the domain edges, a sharply defined boundary between nucleated and empty regions is found. Such a spatial distribution of epitaxial features within size-limited regions is hard to explain by assuming simultaneous formation of epitaxial grains, e.g. by condensation following switching off the bias voltage, as it was proposed in literature. It can also be ruled out that the nuclei are generated in independent events. According to our model a single, or few, real nucleation events take place inside each domain. Under the harsh ion bombardment during BEN, growth perpendicular to the surface is hindered. However, this restriction does not apply to lateral growth. Thus, the latter is the basic mechanism giving rise to the unique pattern formation observed exclusively for diamond nucleation on Ir. Only in one sample extended crystalline diamond regions have been found after 5 s growth. In all other samples, especially after longer growth steps, the diamond islands were observed isolated, at a typical distance of about 20 nm. As a consequence, lateral growth has to be combined with mechanisms that cause splitting into individual islands and the ordering of the islands. This ordering can be attributed to repulsive interactions between neighbouring grains via strain in the substrate, the strength of which was estimated here by finite element simulations. The outcome of the simulations qualitatively indicates that a film split into individual grains would be indeed characterised by a lower strain energy with respect to a 2D layer. This could explain the tendency of an initially extended pseudo-2D layer to split into smaller grains, under the ion-bombardment taking place during BEN. An even lower strain energy configuration is represented by grains that are few nanometers apart. These results would globally support the misfit-strain driven interaction among diamond islands on Ir as responsible of the observed arrays of early grains within the domains. In addition to this, BEN experiments on Ir show a variety of shapes for the domains, ranging from circular to straight-edge ones, to fuzzy and annular shapes. These different morphologies are obtained under 'macroscopically' identical conditions, suggesting the fundamental influence of locally variable conditions. A simple generic Monte Carlo model assuming only two basic competing processes, i.e. the lateral spread of existing diamond nuclei and their back-etching, has been shown here to reproduce the observed variety. The simulations have shown that the competition between these two counteracting processes can nicely reproduce the patterns that have been observed experimentally. Further work will be done in order to describe the underlying processes on an atomistic level. The peculiarities of BEN of diamond on Ir, i.e. the tiny amount of generated material, the exceptional degree of epitaxial alignment since the very beginning of the growth process, the regular arrangement of nuclei and their localisation within well-defined areas, suggest that diamond nucleation on Ir will remain a scientifically challenging topic in the future. In spite of an extensive search for alternative materials that could also act as template for the heteroepitaxial deposition of diamond, Ir is indeed still the only single crystalline material that combines excellent epitaxial alignment with the potential for upscaling to wafer size dimensions. The resulting high technological potential should therefore guarantee that the interest in the described layer structure will rather increase than decrease in the near future.…