Electron spectroscopic characterization of corrosion reactions of active metal systems.

Abstract

The corrosion chemistry of two active-metal systems has been studied primarily with X-ray Photoelectron Spectroscopy. First, the interaction of metallic lithium thin films with simple glass surfaces was investigated. Li was deposited on SiO₂, sodium silicate, potassium silicate, and B₂O₃ glasses in an ultra-high vacuum deposition and analysis chamber. The reaction of Li with SiO₂ results in substantial reduction of the glass matrix to form a thin product layer. A negatively-charged Si species was identified based on an unusually low XPS Si(2p) binding energy. The interaction of Li with alkali silicate glasses resulted in substantially less matrix breakdown than for SiO₂, but exchange of lithium for either sodium or potassium cations occurred at the Li/glass interface. The reaction between Li and B₂O₃ was limited to the top layers of the glass, as a passivating layer formed at the Li/B₂O₃ interface. Investigations into the oxidation of polycrystalline iron surfaces were initiated. Clean surfaces were exposed to controlled amounts of pure oxygen gas. The resulting oxide composition and thickness was determined using ultraviolet photoelectron spectroscopy (UPS), electron energy loss spectroscopy (EELS), and XPS. The results indicated the formation of a bilayer structure, with FeO near the oxide-metal interface, and Fe₃O₄ at the outer surface. Film growth was approximately logarithmic with time, and was strongly pressure dependent. Studies of the electronic properties of the characterized iron oxide surfaces were conducted by measuring the rate of electron transfer between the surface and redox-active species in an electrochemical cell. A strong dependence on film thickness was indicated. Photoemission of electrons from a solid is an inherently complex process; this is especially the case for XPS of clean and oxide-covered active metals. Improved theoretical models of XPS lineshapes were developed which provided new insight into the physical processes involved in photoemission. Additionally, these models provided improved qualitative and quantitative data interpretation through the use of least-squares fitting techniques.