Occurrence of magnetite in the sand fraction of an Oxisol in the Brazilian savanna ecosystem, developed from a magnetite-free lithology

It is relatively well established from many pedogenetic studies that, in deeply weathered Oxisols from Central Brazil, magnetite or maghemite are either inherited or transformed from magnetite of the mafic parent material. However, no similar pedogenetic pathways have been reported in the literature for other lithologies, such as limestone and pelitic rocks (shales and slates) of the Bambuí Group in Brazil. In these sedimentary, non-mafic lithologies, magnetic minerals are not likely to occur. Despite that, magnetic nodules were identified in a representative Oxisol pedon developed on this pedodomain, under savanna ( Cerrado ). Magnetic and non-magnetic fractions of nodules were separated with a hand magnet. Chemical and mineralogical compositions of these nodules were determined by conventional chemical methods, powder X-ray diffractometry (XRD), and 298 K Mössbauer spectroscopy. For the magnetic fraction, containing up to 84 dag/kg of Fe 2 O 3 but also relatively rich in Al, Ti, Cr, and Si, Mössbauer measurements were also made at 4.2 K, without and with an externally applied magnetic field of 8 Tesla, and at 100 K. Mössbauer results and structural Rietveld refinement of the XRD data consistently suggest that the iron oxide mineralogy corresponds to approximately equivalent proportions of hematite and a partially oxidised magnetite, containing 3 dag/kg of iron as FeO. Laboratory tests were conducted in an attempt to produce magnetic material by heating this non-magnetic fraction. The sample was wrapped in filter paper and heated at 300°C for 30 min, and the results were compared with the naturally occurring magnetic nodules. The saturation magnetisation value of the thermally treated sample was found to be σ = 7 J/T.kg, well below σ = 16 J/T.kg of the magnetic soil nodules. Mössbauer and XRD results indicate that the iron oxide mineralogy of this laboratory-produced magnetic sample also corresponds to a mixture of partially oxidised magnetite and hematite. Two other parts of the same non-magnetic, naturally hematite-rich precursor were mixed with charcoal, to act as reducing agent, and oven-heated at 450°C and 600°C, respectively, for 1 h, producing increasing reduction of the hematite to magnetite. These laboratory simulations support the model in which magnetite in these hematite-rich nodules was formed under the influence of seasonal burning regimes of the covering vegetation, followed by partial re-conversion of the magnetite particles to hematite un