Valence losses at interfaces in aluminium alloys

The power and potential of electron energy loss spectroscopy (EELS) as an analytical technique in electron microscopy is undisputed. Much of the focus of current research is on the extraction of structural and bonding information from energy loss near-edge structure. However, the valence loss region contains a wealth of information including data on the optical, electronic and physical properties of a material. Despite the fact that the signal intensity is considerably greater in this region, it is often overlooked as considerable data processing is required to extract the information.

In particular, the valence loss EELS spectrum of a material can be used to calculate its complex dielectric function. However, close to an interface between two dielectric media, a peak corresponding to the interface is observed in the EELS spectrum. This peak must be removed before accurate bulk dielectric data can be determined. However, relativistic and non-relativistic equations exist to model the EELS response of two and three-layer systems. This latter case is of particular interest in the investigation of thin intergranular films within ceramic or composite systems. The thickness of such films can have a significant effect on the strength of the ceramic or composite but is relatively insensitive to material composition. Recent models have suggested that the thickness of the intergranular film is correlated with the dispersion forces acting on the interface. These forces can be expressed in terms of the Hamaker constant, which depends on the complex dielectric functions of the components. If the dielectric functions can be determined using valence loss spectroscopy it should be possible to measure the Hamaker constant and thus obtain information on the dispersion forces. However, due to the complexity of the three-layer system it is of interest initially to explore the simpler two-layer case.

The aim of this project was to investigate EELS from two-layer systems and relate the results to the existing theory. Two systems were investigated, magnesium silicide platelets within an aluminium matrix and silicon precipitates within an aluminium matrix. Both systems were prepared through thermal treatment of a 6061 Al alloy.

The majority of the data presented in this thesis was acquired using EELS. However, energy dispersive x-ray spectroscopy (EDX) and electron microscopy were also used. EELS was performed on two different electron microscopes, the VG HB5 STEM and the FEI Tecnai TF20 (S)TEM. The bulk of the results were acquired on the HB5.

To facilitate the comparison of theoretical and experimental results, the data was separated into bulk and interface components. The component amounts were then plotted against distance from the interface. Bessel functions were then fitted to this plot to give characteristic values. These values represented how well the optimal interface position had been chosen, the comparative decay of the interface plasmon on each side of the interface and the relative thickness of the bulk material.

The experimental data from most of the interfaces examined indicated significant variations in the thickness of the sample. Despite this, the experimental results were found to follow the trend suggested by the theoretical equations. Analysis of the characteristic values indicated that the data from the HB5 and Tecnai for an interface showed a strong correlation. However, comparison of the experimental values with the theoretical reference showed a deviation of ~20%.

Though the source of this deviation was not clear, a number of possible causes were investigated. Theoretical models were generated of systems with a variety of thickness profiles. In addition, systems containing steps, wide and narrow bulk plasmons and a thin interfacial layer of a third material were all considered. The deviation between the results from experiment and the simple theoretical model was believed to be consistent with the factors affecting EELS from a real interface. In particular, thickness variations and imperfections at the interface were found to be the most likely cause of the discrepancy between theory and experiment.

Finally, additional work that could be performed to extend the applicability of this thesis to three-layer systems is discussed. In particular, the determination of dielectric functions from thin interfacial phases is considered.