Ab initio electronic structure calculations are reported which directly relate local chemical and distortion environments of oxygen-coordinated octahedral Fe2+ to corresponding hyperfine parameter distributions in Mossbauer spectroscopy. Changes in the quadrupole splitting (QS) and other properties of the electric field gradient (EFG) with various distortions were investigated on model clusters including the bare octahedra FeO 610- and Fe(OH)64-, and various seven-octahedra sections of an octahedral sheet, through self-consistent charge (SCC) Xalpha ab initio calculations. The percent change in the EFG over the range of distortion parameters found in trioctahedral micas is greatest with flattening for the clusters compared, suggesting that flattening is the most important structural distortion in determining the EFG. The independent parameters of the EFG tensor---particularly the principal value VZZ, the asymmetry parameter eta, and the direction of the principal axis with regards to the octahedral sheet---are examined in detail as functions of octahedral flattening for each of the thirteen possible configurations of Mg2+ and Al3+ cations in the first nearest neighbour octahedra of a Fe2+-centred seven-octahedra cluster. It is demonstrated that the argument that the EFG tensor at a particular site is largely determined by the point group symmetry of the corresponding crystallographic site is not correct. Averages and distributions of eta, principal axis angles and quadrupole splittings are simulated by taking EFG tensor results from all seven-octahedra clusters, and using probabilities for the occurrence of each possible cation configuration in a chemically disordered octahedral sheet of a given bulk composition. Simulations of all hyperfine parameter distributions, including the isomer shift (IS) and the hyperfine magnetic field, were performed using ensembles of Fe(OH)6 4- octahedra generated by applying random distortions to the average structure. Distributions of the QS at 0 and 300 K agree well with experimental results, possessing a low QS tail and a sharp edge at high QS values. This high edge is due to a universal upper bound on the QS, corresponding to the maximum of the QS versus flattening curve. The width-to-mean ratio of the distribution of the nuclear electron spin polarization, the dominant contribution to the hyperfine magnetic field, is large compared to that of the other parameter distributions. The variation of the spin polarization due to local structural variation alone is found to be sufficient to account for the widths and intensities of lines in magnetic hyperfine spectra. The widths of most distributions increase monotonically with the static disorder parameter, and are universal functions of one another, providing a useful constraint in fitting Mossbauer spectra. The correlations between hyperfine parameters are non-linear and considerably more complicated than those supposed in current spectral fitting models. The correlation between QS and isomer shift increases with the amount of disorder in the material. Isomer shift and spin polarization have a strong non-linear correlation. The complexity of the correlations between hyperfine parameters, together with the fact that optimal fits of Mossbauer spectra are not unique, strongly implies that fits to spectra require supporting electronic structure calculations and crystal chemical models of the measured materials for a correct interpretation. It is also shown that published mathematical models of electric field gradient tensor distributions in amorphous materials, derived assuming an isotropic solid, do not describe the distributions in [6]Fe2+ sites in disordered materials, except under special symmetry conditions.
| Identifer | oai:union.ndltd.org:uottawa.ca/oai:ruor.uottawa.ca:10393/29399 |
| Date | January 2006 |
| Creators | Evans, R. James |
| Publisher | University of Ottawa (Canada) |
| Source Sets | Université d’Ottawa |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | 212 p. |
Page generated in 0.0017 seconds