A comprehensive electron energy-loss spectroscopy study of niobium (Nb) and stable Nb-oxide phases (NbO, NbO2, Nb2O5) was carried out. In this work (Part I), the plasmons and energy-loss near-edge structures (ELNES) of all relevant Nb edges (Nb-N2,3, Nb-M4,5, Nb-M2,3, Nb-M1, and Nb-L2,3) up to energy losses of about 2600 eV and the O-K edge are analyzed with respect to achieving characteristic fingerprints of Nb in different formal oxidation states (0 for metallic Nb, +2 for NbO, +4 for NbO2, and +5 for Nb2O5). Chemical shifts of the Nb-N2,3, Nb-M4,5, Nb-M2,3, and Nb-L2,3 edges are extracted from the spectra that amount to about 4 eV as the oxidation state increases from 0 for Nb to +5 for Nb2O5. Four different microscopes, including a 200 keV ZEISS Libra with monochromator, were used. The corresponding wide range of experimental parameters with respect to the primary electron energy, convergence, and collection semi-angles as well as energy resolution allows an assessment of the influence of the experimental setup on the ELNES of the different edges. Finally, the intensity of the Nb-L2,3 white-line edges is correlated with niobium 4d-state occupancy in the different reference materials.

A comprehensive electron energy-loss spectroscopy (EELS) study of niobium (Nb) and stable Nb-oxide phases (NbO, NbO2, Nb2O5) was carried out. Part II of this work is devoted to quantitative EELS by means of experimental k-factors derived from the intensity ratio of the O-K edge and the Nb-M4,5 or Nb-M2,3 edges for all three stable Nb-oxides. The precision and accuracy of the quantification are investigated with respect to the influence of intensity-measurement energy windows, background subtraction, and sample thickness. Integration-window widths allowing optimum accuracy are determined. Owing to background-subtraction errors, the Nb-M4,5 edges rather than Nb-M2,3 are preferred for quantification. Different approaches are applied to improve the precision with regard to thickness-related errors. Thus, a precision up to ±1.5% is achieved by averaging spectra from all three reference oxides to determine a k-factor using Nb-M4,5. Using the experimental k-factor, the determination of atomic concentration ratios CNb/CO in the range of 0.4 (Nb2O5) to 1 (NbO) was found to be possible with an accuracy of 0.6% (relative deviation between measured and nominal composition), whereas ratios of calculated partial ionization cross sections lead to less accurate results.