Mossbauer spectroscopy of phyllosilicates: effects of fitting models on recoil-free fractions and redox ratios

Clay minerals are ubiquitous constituents in soils on Earth, are occasionally found in meteorites, and may also occur on planetary surfaces in the presence of water. However, little is known about the fundamental Mossbauer parameters (the intrinsic isomer shift, 61, the characteristic Mossbauer temperature, theta(M), and the recoil-free fraction, f) that are characteristic of clay minerals and critical to the correct interpretation of the Fe3+/Sigma Fe ratios as well as the mineral modes. Spectra of well characterized single mineral samples at multiple temperatures may be used for the determinations of f. Hence, measurements of five-layer silicates with a range of layer types are presented here: nontronite, Fe-smectite, glauconite, annite and biotite. The spectra were fitted using three different software packages: WMOSS from Science, Engineering & Education Co. in Minnesota; Recoil, from the University of Ottawa in Canada; and two programs used at the University of Ghent in Belgium. Four different approaches to modelling line shapes were used: (1) Lorentzian; (2) pseudo-Voigt (convolution of Lorentzian and Gaussian curves); (3) quadrupole-splitting distributions (QSD); and (4) a technique that does not assume a particular line shape (subsequently referred to as 'model-independent'). Values of delta(1), theta(M) and f were determined using the method of De Grave & Van Alboom (1991). Results show that multiple doublets are routinely required by all models to represent Fe-site occupancy, even when all the Fe atoms of the same valence are in the same site, as is the case for dioctahedral smectite, nontronite, mica and glauconite. Consistent values of centre shift (delta) and quadrupole splitting (Delta) were obtained for the two distributions of Fe-M2(3+) in the smectites. In glauconite, a single Fe2+ doublet was clearly resolved and gave systematic values for 8, A and area, but the two Fe3+ doublets were less defined. In annite, two Fe2+ and two Fe3+ doublets were modelled, while three Fe2+ and one Fe3+ doublet were used for biotite. Three different programs that use Lorentzian line shapes gave very similar results for delta, Delta and area. The two different implementations of QSD line shapes gave similar but sometimes slightly different results, and the pseudo-Voigt and model-independent fits usually fell between the ranges for Lorentzian and QSD results. The value of delta(1) is similar to 0.58 mm/s for Fe3+ and similar to 1.31 mm/s for Fe2+ across all models and line shapes, which is expected because the Fe3+ has an additional shielding 3d electron. Values for theta(M) data are nearly identical for Fe3+ in nontronite and Fe-smectite (similar to 450 K), somewhat varied for Fe3+ in glauconite and biotite (theta(M) = similar to 730 K and similar to 615 K, respectively), and relatively distinct for Fe2+ (similar to 350 K). Some values for theta(M) and f could not be determined due to the non-monotonic behaviour of the fitted values for 6 as a function of temperature. Values of f(295) were 0.821-0.917 for Fe3+ and 0.662-0.743 for Fe2+, consistent with previous studies of the recoil-free fraction in micas and other silicates. Calculated scatter in delta, Delta, area and f values as a function of different line shapes and computer software was significantly reduced at lower temperatures. Sources of error in each of the calculated parameters are discussed.