Crystal Chemistry and Statistical Analysis of Iron

With an emphasis on assessment of reproducibility and precision of results, the technique of Mossbauer spectroscopy is applied to three groups of materials with increasing structural complexity: simple mineral standards, multivalent and multisite ferruginous micas, and multisite silicate glasses. In order to study the underlying accuracy of Mossbauer measurements of minerals, two mineral standards (a grunerite and an almandine/andradite garnet mix) were selected. Precision of the technique was measured through five different sets of experiments seeking to analyze the reproducibility of measurements on a single sample mount, on several identical mounts of the same sample, and on a set of mounts with different sample concentrations, run times, and background counts. The two mineral standards were analyzed by other scientists at six different laboratories; their data were also fit by the M.I.T. curve-fitting program. The standard deviation of multiple measurements on the M.I.T. apparatus is better than 0.016 mm/sec for isomer shift, 0.060 mm/sec or better for quadrupole splitting, and 1.02% on individual peak area data. The standard deviation of interlaboratory measurements on the same minerals is slightly better because only ideal run conditions were used: 0.006 mm/sec for isomer shift, 0.023 mm/sec for quadrupole splitting, and 1.44% on individual peak area data. Probable errors on different aliquots of the same sample are approximately +0.02 mm/sec for isomer shift and quadrupole splitting, and +1.5% on area data for well-resolved peaks. In contrast to the simplicity of the mineral standards, the multisite, multivalent structure of the trioctahedral micas presents a formidable challenge to the application of techniques for obtaining reproducible results. To establish a firm basis for evaluation of new results, over fifty studies on Mossbauer spectroscopy of trioctahedral micas are summarized and reviewed. New measurements were made on a suite of seventeen ferruginous samples from a range of compositions and parageneses. Characteristic spectra include doublets with the following parameters: Fe2+ cis-M2, isomer shift (6) = 1.13, quadrupole splitting (A) = 2.58 mm/sec; FeL + trans-M1, 6 = 1.12 and 1.16, A = 2.05 and 2.75 mm/sec ; Fe3 + M2, 6 = 0.40, = 0.55 mm/sec; Fe Ml, 6 = 0.40, A = 1.00 mm/sec; and Fetet 3 +, 6 = 0.20, A = 0.50 mm/sec. The effects of substitution are assessed and found to be minimal, although OH, F, and Cl (which are not easily analyzed or frequently reported) substitutions may have more of an effect. The Mossbauer and compositional data considered together show that the FeMg_ 1 and Tschermak substitutions are most common, that tetrahedral AlFe_ 1 + may be controlled by octahedral cation size, and that M2/M1 cation ordering is ubiquitous. Effects of sample preparation on Mossbauer spectra are also discussed. Finally, the technique of Mossbauer spectroscopy is applied to the structures of complicated silicate glasses. An overview of past work on iron-bearing glasses is presented, covering both geological and ceramic research on Fe-bearing borate, phosphate, and silicate glasses. Studies of multi-component synthetic compositions are reviewed in terms of the effects of each separate cation on Fe3+/Fe 2+ ratios and iron site occupancies. In the context of past work, new results on synthetic lunar and natural terrestrial glass compositions are presented. Compositions analogous to lunar green, orange, and brown glasses were synthesized under consistent conditions, then quenched into a variety of different media when the samples were removed from the furnace. Iron valence and coordination are a direct function of quench media used, spanning the range from brine/ice (most effective quench), water, butyl phthalate, silicone oil, liquid nitrogen, highly reducing CO-CO2 as, to air (least effective quench). In the green and brown glasses, Fej+ in fourand six-fold coordination is observed in the slowest-quenched samples; Fe2+ coordination varies directly with quench efficiency. Different quench rates are found to yield glass structures that are markedly different with respect to iron atom site occupation and charge. The effect of quenching method on Fe3+/Fe 2+ ratios in three silicate glasses was also studied using natural rock compositions (U.S.G.S. rock standards basalt BCR-1 and rhyolite RGM-1 and an andesite K-2B from Dennison Volcano, Alaska). Samples were equilibrated at high temperatures and then quenched into an air jet, an H2-Ar jet, or a brine/ice bath. The glasses were analyzed by both Mossbauer spectroscopy and a micro-colorimetric technique to compare the results from these commonly-used methods for ferrous-ferric iron determination. Both of these methods obtain Fe3+ estimates by difference (for low Fe3+ samples), resulting in relatively large uncertainties (3-5% for Mossbauer spectroscopy and 8% for colorimetry). Within these uncertainties, estimates of Fe3+/Fe 2+ ratios determined by Mossbauer spectroscopy and colorimetry are the same. The technique of Mossbauer spectroscopy can be successfully applied to the range of materials studied. However, the underlying, limiting precision of the technique must always be considered when drawing conclusions from Mossbauer results on the structures of materials. Thesis Supervisor: Roger G. Burns Title: Professor of Mineralogy and Geochemistry