Quantitative use of EELS Mo‐M2 ,3 edges for the study of molybdenum oxides: elemental quantification and determination of Mo valence state

Intro There is currently a strong revival in the study of molybdenum oxides triggered by the recent developments following their nanostructuration. 1,2,3 This should open the way for an emerging field of research aiming at the characterization and optimization of Mo‐based nanodevices. EELS performed in a TEM is an unrivaled tool for such analyses even though EELS analyses of Mo oxides can be tricky. Since Mo‐L 2,3 white lines are situated around 2500 eV, they cannot be used with confidence, such high‐energies implying excessively long dwell times and therefore unavoidable irradiation beam damages. Furthermore, these lines are located too far away from the O‐K edge to allow Mo valence determination and Mo/O elemental quantification from the same spectra. On the other hand, Mo‐M 2,3 white lines are located at lower energies and are closer to the O‐K edge. The main issue in using these edges is however the delayed maxima of the Mo‐M 4,5 edges ( Fig. 1a ) that hinders the background subtraction with the usual inverse power low function. In this contribution, we use a combination of EELS experiments, multiplet and density functional theory (DFT) calculations to establish that elemental quantification and Mo valence states can indeed be reliably derived from Mo‐M 2,3 edges. Material & Methods EELS spectra were acquired on commercial MoO 3 (Mo VI ) and MoO 2 (Mo IV ) powders using a Hitachi HF2000 TEM (100 kV) equipped with a cold FEG and a modified Gatan PEELS 666 spectrometer. The energy resolution was 1.5 eV and the energy dispersion 0.20 eV/pixel. EELS spectra were acquired at magic angle condition for the Mo‐M 2,3 edges to avoid anisotropy effects playing a role in the M 2,3 intensity ratio determination. Experiments were performed at liquid nitrogen temperature to minimize carbon contamination and irradiation beam damage. Background subtraction for the M 2,3 edges is based on the determination of post‐edge parameter ( Fig.1b )to avoid the detrimental effect of the Mo‐M 4,5 edges on the background subtraction. After removal of the multiple‐inelastic scattering effects, M 3 /M 2 intensity ratios were determined by subtracting a two steps function followed by area integration. Theoretical intensity ratios were also derived from multiplet calculations by using the CTM4XAS program, 4 the crystal field splitting parameter being determined from DFT calculations with the Wien2K code. 5 Results To determine the feasibility of elemental quantification, the k‐factors ( Fig. 2a ) and the corresponding standard errors ( Fig. 2b ) were determined as a function of the width of the energy window used for the integration. The standard errors reach a minimum close to 2% for energy windows of 15 and 20 eV for MoO 2 and MoO 3 respectively. The relative difference to the mean value presents also strong variations depending on the energy window and the best accuracy (2%) is found for a width of 10 eV. The precision and the accuracy of these results validate the method we used to subtract the background. In addition, theoretical M 3 /M 2 ratios were also determined from multiplet calculated spectra ( Fig. 3a ) and compared to experimental ratios ( Fig. 3b ). The agreement between experiences and calculations is excellent and strengthens our experimental methodology. All these results will be detailed together with the possibility to discriminate the two oxides thanks to chemical shifts and energy‐loss near‐edge structures. This work provides thus a complete picture on the ability to obtain a wealth of precise and accurate chemical information on Mo oxides from the conjugated analyzes of O‐K and Mo‐M 2,3 edges. It will also open interesting opportunities for the EELS studies of a large variety of materials as it is directly transposable to the whole family of 4d transition metal oxides. 6