Determination of Chemical Valence State by X-ray Emission Analysis Using Electron Beam Instruments: Pitfalls and Promises

Abstract

The use of X-ray emission spectroscopy to determine the valence states of first-row transition elements is evaluated using measurements of X-ray wavelength shifts, line shapes, and relative X-ray line intensities (Lβ/Lα and Kβ/ Kα). X-ray wavelength (or energy) centroids and line shapes are shown to vary not only with oxidation state but also with chemical bonding for the same oxidation state. In many of the cases studied, the ratio of Kβ to Ka is shown to vary depending upon the differential absorption at a similar level to what is cited in the literature as due to the valence state. Similarly, in a number of cases studied, the ratio of Lβ to Lα is shown to vary by differences in self-absorption (and at low electron beam energies, differences in overvoltage) by levels comparable to what is ascribed in the literature as due to valence state. Differences were found among compounds with different oxidation states that exceeded the magnitude predicted by absorption or overvoltage effects; however, no systematic variations were found that were dependent only on the valence state of the element (as opposed to which other elements it was bonded). The results show that oxidation state cannot be simply and unambiguously determined by X-ray emission spectroscopy at the resolution obtainable with conventional wavelength-dispersive detectors on electron microscopes or microprobes. Of the various techniques employed in X-ray emission spectroscopy, the most promising for oxidation-state determination appears to be measurement of the wavelength shift and/or satellite line shape variation, but then only when comparing compounds in which the cation has the same bonding environment. It is critically important to correct for self-absorption effects when attempting to relate variations in Lβ/Lα to differences in oxidation state or bonding environment, regardless of the electron beam energy used in analysis. Specifically, Lβ/Lα measurements cannot be used to determine the oxidation state of Cu in high-temperature superconductors.