V local structure in phosphate glasses

The aim of this work is to study the vanadium local environment in two V-P-Na glasses. These glass systems are important for their technological applications, like power cells and high power laser devices. The compositions of the glasses analysed are: i) VPNa10 - 10% mol. V2O5, 45% mol. P2O5 and 45% mol. Na2O and ii) VPNa80 - 80% mol. V2O5, 10% mol. P2O5 and 10% mol. Na2O. They have been chosen because only few studies on the V structural role in glasses with high V content are available in the literature. The analysis has been carried out by means of V K-edge X-ray Absorption Spectroscopy (XAS). Since XAS is a local probe and it is an element selective technique, it is a suitable tool for studying the local structure of disordered systems. To be able to determine the structural role of V in the glasses and to take into account the complexity of the data analysis, this study has been carried out by investigating all regions of the absorption spectrum using different approaches for the different parts of the absorption spectrum: i) the pre-edge peak region by using a fingerprint method that consists in comparing this region of the glasses with that of a selected number of model compounds; ii) the Extended X-ray Absorption Fine Structure (EXAFS) region by generating theoretical scattering paths and fitting the theoretical EXAFS function with the experimental one; iii) the X-ray Absorption Near Edge Spectroscopy (XANES) region by performing a fit of the theoretical spectrum calculated by the MXAN code with the experimental one. For the last point a preliminary analysis on a set of model compounds was necessary to test the theory on the different V speciation and to find the best theoretical starting condition to calculate the theoretical XANES of the glasses. The position and intensity of the pre-edge feature give important information about the V coordination geometry and oxidation state by comparing the glasses pre-edge peak to those of the model compound with known structure. The choice of the standards has been made in order to represent the most common V coordination geometries and oxidation states. Namely the standards of choice were: Goldmanite and V2O3 for [6]V 3+; Cavansite for [5]V 4+; V2O4 for [6]V 4+; Palenzonaite and Vanadinite for [4]V 5+; V2O5 for [5]V 5+. With the V XAS data obtained for the standards it has been possible to build a plot of the centroid position of the pre-edge peak vs. its intensity. The mixing lines represent the (ideal) position of a hypothetical compound with a speciation intermediate between the model compounds reported in the plot. The lines were built by linear combination of the preedge peak values of the two extremes for each valence state. The pre-edge peak analysis on the two glasses has shown: i) V in VPNa10 is predominantly 4+ with a coordination geometry about 60% square pyramidal and 40% octahedral; ii) V in VPNa80 is mainly 5+ with a coordination geometry approximately 65% square pyramidal and 35% tetrahedral. Starting with the preliminary information obtained by the pre-edge peak analysis we have built a model structure for the two glasses to generate EXAFS scattering paths to fit the experimental EXAFS. The EXAFS analysis has shown that in the case of VPNa80 the multiple scattering contribution is much higher (due to the higher V content) and we had to take it into account. A single scattering approximation was adequate for VPNa10. The EXAFS analysis confirmed the presence of mixed V coordination geometries for the two glasses. We have obtained a radial distribution function which also suggested information on the second coordination shell around V. In VPNa10 the second coordination shell was mainly composed by P atoms but a possible V contribution was not prominent. In VPNa80 the fit was compatible with a second coordination shell composed by a mix of V and P atoms. In this sample there was evidence of a strong geometry distortion that, however, was impossible to quantify due to the scalar determination of the bond distances by the EXAFS analysis. Moreover, the EXAFS fit of VPNa80 was not as accurate as that of VPNa10. For these reasons we decided to perform a structural refinement of the glass structures by means of V K-edge XANES region FMS theoretical calculations. Calculations of theoretical XANES spectra have been tested on the model compounds by means of two codes: FEFF and MXAN. Both FEFF and MXAN codes exploit FMS theory in the Muffin Tin (MT) approximation: the former one performs a Self Consistent Field calculation of the potentials while the latter one determines the physical parameters through a fit with the experimental XANES and allows a geometrical fitting of the structural parameters. The FMS theory was able to reproduce the experimental XANES spectra of the model compounds and the MT parameters (MT radii and interstitial potential) have been extracted through the MXAN fit from the model compounds analysis to be used in the study of the glass samples. In particular we have shown that the theory was more accurate for the minerals with lower V content than for the 3 oxides, especially close to the absorption edge, where the MS effects are stronger and the photoelectron mean free path is higher. The MT parameters (MT radii and interstitial potential) are correlated with the V oxidation state and coordination geometry. These values could be used as starting values in the analysis of more complex structures and thus we have built a database to be used in the glass analysis. Theoretical XANES spectra of the glasses have been calculated by means of the MXAN code. They were in very good agreement with the experimental ones, reproducing all of the features with good relative positions and intensities. The V speciation, as found by XANES refinement, was in good agreement with both the EXAFS and pre-edge peak analysis for both the glasses. The bond distances were consistent with literature data and the oxidation state extrapolated by bond valence sums agreed with the values found in the pre-edge peak analysis. The coordination found in the second shell suggests the presence of mainly bridging oxygen atoms, as already found in literature for V-P glasses. V in VPNa10 was found to be mainly 4+, with a geometry from 60% to 70% square pyramidal and from 30% to 40% octahedral. A vanadyl and a trans oxygen atoms were found in the 5-fold geometry; all the other oxygen atoms were coordinated with a P atom in the second shell. V in VPNa80 was mainly 5+, with a geometry from 65% to 70% square pyramidal and from 30% to 35% tetrahedral. We were not able to determine the second coordination shell around the 4-fold V; a vanadyl and a trans O atoms were found in the 5-fold geometry again; three of the basal O were coordinated with 2 V in the second shell, while the opposite basal O was coordinated with a P atom. Preliminary analysis at the Na K-edge XANES are presented in order to estimate the Na-V and Na-O bond distances and the Na local geometry. We have shown that FMS theory is a suitable approximation to calculate XANES spectra of disordered structures since some of the limitations of the theory (mainly inaccuracy in calculating MS near the edge where the metal-metal interaction is strong and the photoelectron mean free path is high) are less prominent in disordered structures due to the lack of long range order. By FMS it was possible to get deep insights into the local structure of V, including bond distortions and geometry of the second coordination shell. In conclusion, the analysis of the V XAS spectra of the VPNa glasses has been successfully carried out. The study allowed also to gain information on the advantages/ disadvantages of the FEFF and MXAN codes by investigating in detail the V K-edge XANES on the V-bearing standards with the two codes. This allowed to approach the study of the glasses with rigorous preliminary information.