| Summary: | This thesis deals with the mechanisms involved in reactively sputtering a metal target in an Inert/reactive gas glow discharge and with the electrical transport and optical properties of A1/A1N granular metal (or cermet) films produced by this technique. Experiments are described in which an A1 target is sputtered in Ar/N₂ and Ar/O₂ atmospheres. The relationships between chemical processes occurring on the target surface, substrate surface, and in the glow discharge of a dc planar magnetron sputtering system are studied for the purpose of controlling film composition. The positive feedback mechanisms which lead to the well known transitions between bare and covered target surfaces are correlated with glow discharge characteristics. These data are shown to be in agreement with a model which assumes two distinct mechanisms for target coverage: (1) chemisorption of neutral reactive gas species from the sputtering gas; and (2) ion plating of reactive gas species from the sputtering current. This model allows estimation of the stability of the glow discharge against the positive feedback mechanisms and indicates under what circumstances voltage control of the glow discharge will permit sustained operation at all degrees of target surface coverage. With the voltage control method, a one to one correspondence between target voltage and film composition is established. In addition, a method is presented for calculating the film composition from only the glow discharge characteristics. My experiments show that voltage controlled, reactive dc, planar magnetron sputtering is ideally suited to the deposition of A1/A1N cermets of controlled composition. X-ray diffraction, transmission electron microscope (TEM), Hall, and resistivity vs. temperature data for these A1/A1N cermets are presented as a function of metal volume fraction (Xv) and correlated with the glow discharge characteristics of the deposition process. Metal precipitates are seen to form and, thereby, the film properties are interpreted in terms of granular composites of A1 and A1N crystallites when the A1/N ratio becomes greater than one. A percolation threshold is observed in the conductivity at a critical volume fraction (Xvc) of 0.72 ± .02. The conductivity, σ, exhibits power law behavior both above and below Xvc. Above Xvc, σ ~ (Xv - Xvc)[sup t], with t = 1.75 ± .1, in excellent agreement with the theoretical prediction of 1.7 for a mixture of two "normal" conductors (i.e. metallic or semiconductor conduction, but not hopping or tunneling). Below Xvc, conduction is via hopping and σ ~ (Xvc -Xv)[sup –s], with s = 4.3 ± .1. For a mixture of normal conductors below Xvc, s is predicted to be 0.7, while there is no theoretical prediction for s when conduction is via hopping. This power law behavior of hopping conductivity warrants further theoretical, as well as experimental, investigation. Further, the temperature behavior of the conductivity is consistent with the view that hopping is from defect to defect within the A1N grains as opposed to direct metal grain to metal grain hopping. The temperature behavior of the condcutivity also indicates that electron localization effects become important for Xvc < Xv < 0.8. In spite of the obvious granular nature of these films, neither the effective medium or Maxwell-Garnett theories for granular materials appears adequate in describing their optical properties. Observable structure in the UV optical absorption and IR reflectivity seem to be properties of A1N and not due to the microgranular structure of the A1/A1N composite. That optical properties predicted in the granular theories are not observed, even though the films are granular, is attributed to the effect of a large number of single and multiple atom A1 inclusions, with other than bulk optical properties, that are not taken into account in these theories. === Science, Faculty of === Physics and Astronomy, Department of === Graduate
|
|---|
