TEXTURE AND STRESS ANALYSIS OF THIN FILMS
K. Helming
FG Textur und
Anisotropie kristalliner Stoffe" TU Clausthal, Großer Bruch 23, D-38678
Clausthal-Zellerfeld, Germany
Keywords:
texture components, ODF-methods, texture modeling,
heterostructures, multi-layers, sample symmetry position
sensitive detector, adhesivity
Solid state processes in polycrystalline materials are often associated with changes of crystallite orientation [1]. A microscopical understanding requires the determination of texture which is quantitatively given by the orientation distribution function (ODF). The ODF is traditionally calculated from pole figures, measured by x-ray or neutron diffraction. For cubic or hexagonal metals 3 - 4 pole figures are required for calculating the ODF. The angular resolution of 5° (~1000 sample directions) is usually sufficient to describe the most typical textures of deformed or recrystallized metals and alloys.
In the last two decades texture analysis has broadened significantly. Due to the low crystal symmetry modern multi-phase materials usually require much larger amounts of data to calculate the complete ODF for each crystalline phase. For materials like rocks or ceramics ODF-calculations (from 20 and more pole figures) rendered more difficult because Bragg reflections often coincide, and the degree of overlapping increases with decreasing crystal symmetry. Due to the low diffracted intensities and the expected sharp textures, pole measurement on thin films or multilayers may require long times of measurement and dense grids of measurement (>30 000 sample directions) in the pole figures. Furthermore effects of absorption and overlapping of Bragg reflections must be considered. On the other hand the searched textures are mostly very simple, they may often be explained by a few sharp texture components. Therefore intelligent strategies for pole figure measurement can be developed which allow the calculation of sample directions, component orientations and volume fractions using a-priori knowledge combined with the necessary minimum of measured data input.
Diamond, GaN and SiC thin films investigated here were thin (<2 µm) and polycrystalline and sometimes contained more than one crystallographic phase. On the other hand the substrates were thick and single crystalline. Thus, applying X-ray diffractometry, film reflections tended to be rather weak and broad, whereas intense fundamental substrate reflections were accompanied by spurious substrate reflections caused by Bremsstrahlung radiation. An experimental solution to this problem is the use of a strictly monochromatic beam and a suitable beam collimation. Using synchrotron radiation the intensity of the film reflections can be increased considerably. Application of the component method [2] allows to automatically separate up to five overlapping Bragg reflections (even if they originate from different wavelengths). Even peak widths below 1 degree (FWHM) as observed e.g. with epitactic SiC and GaN films can be modelled with ease using the component method.
In general the use of components reduces the amount of data necessary for a quantitative texture description. Processes and properties correlated with texture are understood more readily due to this high compression of texture information. Orientation differences between the preferred orientations of the calculated components can give an indication as to which crystal planes or directions are invariant during a process modifying the texture. An uncommon statistical symmetry may be observed with respect to the sample coordinate system. For each of the presented examples the sample symmetry of the films was identical with the subgroup of the substrates crystal symmetry. This subgroup belongs to the crystal axis parallel to the substrate normal. Based on this finding, the symmetry of the deposition processes (e.g. surface kinetics) and correlation between substrate and layer (e.g. metric coincidences) may be concluded. For each of the presented examples the sample symmetry of the films was identical with that subgroup of the substrates crystal symmetry which belongs to the crystal axis parallel to the sample normal direction.
In the field of orthopaedics and
dentistry, hard coating for implant surfaces are used to improve
both the osteophile properties and the long-term stability of the
material. In order to obtain an adequate adhesivity and shearing
strength, it is important to determine the correlation between
the process-induced intrinsic stresses during the coating
deposition and the process parameters such as substrate
temperature and substrate bias voltage. TiN films were deposited
by physical vapour deposition technique. The coatings are
characterised by their adhesivity and compressive stress using
the scratch test method and a bending-strip technique. To get
more information about the texture, growth characteristic and the
induced stresses of the TiN films, XRD spectra were measured with
a position sensitive detector. The intention was to examine the
influence of energy and flux of the charged particles striking
the substrate during the film growth process. The energy of
incoming ions can be affected by a negative substrate voltage,
the flux density and the energy by a hf plasma between the
substrates and the process chamber wall. Obviously, both the
increase of the bias voltage and the number of charged particles
striking the substrate are leading to a preferred film growth in
the <111> direction. A quantitative description of texture
by means of a small number of components is an extraordinarily
concise method. It reveals relevant information about the texture
even in those cases (as presented here) where texture information
contained in the measured pole figures is rather limited. The
peak shift to larger 2 values with higher tilt angles of the TiN
reflections in the measured XRD spectra confirm the lateral
compression which was measured by the mechanical method.