Goethite is an important iron-bearing mineral found in nearly all types of soils and sediments that typically occurs as particles of nanoscale crystallites. Here we show that poorly crystalline nanophase goethite can undergo dehydroxylation and alteration to produce strongly magnetic Fe-oxide nanoparticles. After moderate reductive heating, synthetic oriented aggregates of nanogoethite were converted to stoichiometric sub-micron magnetite. In contrast, oxidative heating produced only nanocrystalline hematite. We present a novel characterization of the products of nanogoethite alteration, which represents an important pathway for the production of both superparamagnetic and stable single-domain magnetic particles in a wide variety of sedimentary environments, including soils and lacustrine and marine settings in which goethite is commonly present. In nature, the mechanism of magnetite formation observed in our experiments could contribute to the widespread phenomenon of magnetic enhancement in many soil types, and could be responsible for secondary magnetizations often observed to accompany diagenesis in sedimentary rocks. Thus, nanogoethite may be an important precursor of fine magnetic particles and magnetic remanence carriers in the environment, particularly in soils and catchment areas affected by wildfire as well as in sediments subjected to deep burial diagenesis and low-grade metamorphism.
Widespread occurrence of goethite has been observed in nearly all sediment types on Earth, albeit typically at low concentrations and in nanocrystalline form that may hinder its detection. Nanogoethite is likely the most stable naturally occurring Fe-oxyhydroxide, and can form rapidly from ferrihydrite at ambient soil temperatures (Yee et al., 2006). Hansel et al. (2004) suggested that goethite may be the dominant phase involved in Fe-redox reactions in many sediment systems. The phenomenon of magnetic enhancement in many soil types has been linked to transformation of weakly magnetic Fe-rich phases into strongly magnetic Fe-oxide nanoparticles, and goethite has been identified in some sites as a likely precursor candidate (Guyodo et al., 2006). However, few goethite alteration processes capable of producing magnetic nanoparticles have been documented (Usman et al., 2013). Biologically induced remineralization of goethite by dissimilatory iron-reducing bacteria is very limited, in contrast to more reactive Fe-oxyhydroxides such as lepidocrocite (e.g., Ona-Nguema et al., 2002) and ferrihydrite (e.g., Glasauer et al., 2003). Formation of insoluble Fe2+ complexes at the crystal surfaces during bioreduction may passivate goethite, preventing further reduction and transformation, and may inhibit formation of secondary phases (Hansel et al., 2004). Abiotic reductive dissolution is also much slower in goethite than in ferrihydrite and lepidocrocite (Postma, 1993). Yet, nanophase goethite has been found to be the main reactive Fe-bearing phase in a wide range of marine and lake sediments (van der Zee et al., 2003). Here we describe the formation and properties of nanoscale magnetite particles produced by laboratory nanogoethite alteration, and discuss the wider implications for the occurrence of magnetic nanoparticles in the environment and the evolution of magnetic signatures in goethite-bearing sediments.
Goethite was synthesized following the protocol of Schwertmann and Cornell (1991) using an FeCl2 · 4H2O precursor to which NaHCO3 was added, and was oxidized gradually by air bubbling over 48 h. Goethite purity was verified with energy dispersive spectroscopy on a high-resolution transmission electron microscope (HRTEM) and X-ray diffraction (XRD). HRTEM was performed on a JEOL 2100F microscope with a field emission gun. Powder XRD patterns were collected using Co(Kα) radiation on a Panalytical XPert diffractometer with a sealed cell that maintained anoxic measurement conditions. Magnetic characterization included low-temperature (T < 300 K) frequency-dependent susceptibility and hysteresis loops measured at 10 K in a maximum field of 2.5 T on a Magnetic Properties Measurement System (MPMS; www.qdusa.com/products/mpms3.html), as well as room-temperature hysteresis measured on a vibrating sample magnetometer. Relative room-temperature frequency dependence of susceptibility, χfd, was calculated as
Rietveld refinement of powder diffraction patterns was performed with the XND code (Berar and Baldinozzi, 1998), which accounts for anisotropic line-shape broadening. Starting crystallographic parameters were taken from Forsyth et al. (1968), Blake et al. (1966), and Hill et al. (1979) for goethite (space group Pbnm), hematite, and magnetite, respectively. Scale factors, unit-cell parameters, specimen displacement, and line-shape parameters were refined. Size-broadening effects were estimated from Lorentzian line widths varying as a function of 1/cos. Anisotropic Lorenztian line-width parameters were refined for goethite and hematite, and the isotropic line-shape parameter along  was refined for magnetite. Mean coherent domain (MCD) sizes along various crystallographic directions were then calculated from line-width parameters using Scherrer’s formula (Table 1).
Alteration experiments were performed in a gas-mixing furnace contained inside an Ar-filled glove box. Sample powders were heated either in a reducing atmosphere of a flowing CO-CO2 mixture (20% / 80%) or oxidized in flowing air. Samples were heated to between 210 and 270 °C for 40–155 min (Table 1). Subsamples from reduction experiments were prepared for characterization in an oxygen-free glove box, and anoxic sample conditions were maintained during XRD and magnetic measurements.
Rietveld refinements of XRD patterns for the synthetic nanogoethite indicate an MCD size in the range 5.9–8.6 nm (Table 1). In HRTEM, the nanogoethite consists of highly oriented elongated aggregates of crystallites, with a particle size near 10 nm by 50 nm (Fig. 1A). Simulated HRTEM electron diffraction patterns for individual particles are nearly single crystalline, reflecting the high degree of crystallite orientation (Fig. 1B). Crystallite sizes determined from HRTEM appear somewhat smaller than those indicated by XRD, however the similar orientations of neighboring crystallites make the size difficult to determine visually. Magnetic susceptibility of the nanogoethite (Fig. 2A) decreases on warming from 10 K, with little frequency dependence. Hysteresis loops measured at 10 K after zero-field cooling have an antiferromagnetic character with no magnetic remanence or coercivity (Fig. 2B).
Upon heating nanogoethite in air at 250 °C for 2.5 h (sample G01), a pronounced color change from yellow-ochre to red occurs, consistent with the well-characterized dehydroxylation reaction that produces a topotactic transformation from goethite to hematite (Naono et al., 1987). In HRTEM, the reaction product consists of elongated particles of aggregated nanohematite crystallites with the same general morphology as the nanogoethite but containing nanopores (Fig. 1C). Crystallite size between the goethite and the hematite is indistinguishable in HRTEM, however XRD indicates a smaller MCD size of 3.2 by 5.8 nm for the hematite. XRD patterns for sample G01 have correspondingly broad hematite peaks with no goethite peaks detected (Fig. 3). The low-temperature susceptibility curve for sample G01 is distinct from that of the nanogoethite and lacks a clear Morin transition, which is suppressed below the critical hematite grain size of 20 nm (Özdemir et al., 2008). Sample G01 has a decreased high-field slope in low-temperature hysteresis relative to the goethite, and is also dominantly antiferromagnetic (closed) in shape (Fig. 2B).
In contrast, heating nanogoethite in a CO-CO2 atmosphere produced sample colors ranging from dark red to pure black, varying as a function of heating time and temperature. Samples with a distinctly black color (G02, G05) are marked by a strong Verwey transition near 120 K (Fig. 2A), identifying the primary alteration product as stoichiometric magnetite. XRD patterns (Fig. 3) and Rietveld refinements (Table 1) indicate that sample G05 is composed of pure magnetite and that sample G02 is a mixture of 50% magnetite and 50% hematite (± 7%), which is roughly consistent with the respective saturation magnetization, MS, values of 86 and 66 Am2/kg at 10 K. In HRTEM, G02 and G05 appear as rounded, equant to slightly elongated, highly crystalline grains (Fig. 1F). Lower heating temperatures (samples G03, G04) produce weaker but discernable Verwey transitions (Fig. 2A), marking the onset of magnetite formation. An MS value of 5.9 Am2/kg for sample G03 is consistent with a magnetite concentration of 4% (± 2%) estimated from XRD, while the magnetite fraction in sample G04 cannot be unambiguously quantified with XRD due to coinciding magnetite and hematite peaks, despite an MS value only slightly lower than that of sample G03. Sample G06 (210 °C, 40 min) is the most weakly reacted sample, however the hysteresis loop shape indicates the presence of a ferromagnetic phase (Fig. 2B).
No goethite peaks are detected in XRD patterns of heated samples, suggesting that dehydroxylation of goethite to hematite occurs very rapidly in the experiments. Hematite XRD peaks occur in all reduced samples except G05, which appears to be entirely converted to magnetite. Rietveld refinements of XRD data (Table 1) indicate that hematite crystallites in reduced samples are larger than in sample G01, with MCD lengths along  ranging from 4.2 to 7.6 nm. Hematite crystallite size increases and the fraction of hematite decreases with longer heating (G06 versus G04) and increased temperature (G04, G03, G02). Some differences in the morphology of nanohematite particles can be discerned in HRTEM images (Figs. 1D and 1E), although magnetite was not identified with HRTEM in 210 °C samples due to the low concentrations. The evolution of nanohematite to larger crystallite sizes and smaller mass fractions corresponds to increasing magnetite concentrations, as indicated by MS values in low-temperature hysteresis loops (Fig. 2B; see the GSA Data Repository1). The formation of magnetite in the experiments thus appears to occur by a two-step reaction process:
Room-temperature χfd is remarkably high in G03, G04, and G06, with values of 22%, 30%, and 12%, respectively, reflecting a significant portion of superparamagnetic (SP) magnetite particles, which typically occur at sizes below 20 nm at room temperature (Till et al., 2011). Values of χfd above 10% typically only occur in systems of non-interacting SP particles (Till et al., 2011) because magnetostatic interactions greatly reduce both bulk susceptibility and χfd (Muxworthy, 2001). The initial stages of magnetite formation during goethite alteration thus must proceed from initially isolated nucleation of SP-sized magnetite crystallites in a matrix of goethite and/or hematite. As the reaction proceeds, magnetite particles grow larger and the spacing between them decreases, with increasing magnetostatic interaction. The magnetic particle assemblage transitions from one dominated by weakly interacting SP magnetite to one of strongly interacting single-domain (SD) magnetite as alteration progresses.
DISCUSSION AND CONCLUSIONS
Recrystallization of nanohematite as evidenced by crystallite growth appears to accompany the process of reduction to magnetite. The high crystallinity and larger grain size of the magnetite contrast strongly with the pseudomorphed poorly crystalline nanohematite produced at the same heating time and temperature (sample G01 versus G02). While a topotactic relationship exists between hematite and magnetite, and pseudomorphs of the two phases can occur, here we see evidence of extensive and rapid recrystallization of intermediate hematite during moderate reductive heating. We hypothesize that the extremely fine-grained, poorly crystalline nature of hematite formed by nanogoethite dehydroxylation results in high reactivity under reducing conditions that allows nanohematite to easily recrystallize and achieve transformation to magnetite at moderate temperatures.
Alteration of goethite to a strongly magnetic Fe-oxide such as magnetite or maghemite due to fire has been documented in soils containing organic matter that acts as a reductant. Although experimental burning of grassland soils indicates that fire is an unlikely source of significant magnetic enhancement in prairie soils (Roman et al., 2013), other soil types (Le Borgne, 1960; Blake et al., 2006) and lake sediments (Rummery, 1983; Gedye et al., 2000) in previously burned areas nevertheless commonly exhibit strongly enhanced magnetic susceptibility. Several studies of fire-enhanced soil magnetic susceptibility have identified goethite as the Fe-bearing parent phase whose transformation produces strongly magnetic Fe-oxide nanoparticles (Anand and Gilkes, 1987; Ketterings et al., 2000; Clement et al., 2011). Our results demonstrate that temperatures as low as 210 °C are effective in producing magnetic enhancement from pure nanocrystalline goethite, whereas the previously estimated minimum fire temperature needed to produce fine magnetic particles was 400 °C (Rummery, 1983). Alteration of aluminous goethite may require slightly higher temperatures, as Al tends to stabilize goethite against dehydroxylation (Ruan and Gilkes, 1995). The role of fire-induced mineral alteration in producing fine magnetic particles may be particularly important in lacustrine and riverine sediments because catchment areas affected by wildfires commonly experience increased surface runoff and sediment delivery (Smith et al., 2013).
In addition to fire-induced magnetic enhancement by reductive alteration of goethite at elevated temperatures, it may be possible for the same reaction chain to occur at lower temperatures over longer time scales. Langmuir (1971) predicted nanogoethite to have no stability relative to hematite on geological time scales. On laboratory time scales, goethite is reportedly stable against dehydroxylation up to 100 °C (Koch et al., 1986); however, ambient-temperature transformation of goethite to magnetite can occur via more complex abiotic reactions catalyzed by aqueous Fe2+ (Usman et al., 2013) without hematite present. Such reactions are likely to be more important than alteration by dehydroxylation at in situ conditions in soils and shallow sediments, although it is unclear whether such chemical processes occur in nature or are geologically significant.
The results described here also have implications for the evolution of Fe-mineral assemblages in marine metasediments, which commonly undergo diagenetic alteration. Diagenetic hematite growth from a goethite precursor has been previously recognized in pelagic limestones (Channell et al., 1982). Given the common occurrence of nanogoethite in a wide range of marine settings (van der Zee et al., 2003), alteration of goethite in anoxic or suboxic deep-sea sediments may also be a likely mechanism for the neoformation of fine magnetite and the increased magnetization observed to occur with deep-burial diagenesis or low-grade metamorphism (Aubourg et al., 2012). In an intra-basin comparison of claystones, Abdelmalak et al. (2012) identified goethite as a major phase in sediments that had undergone only minor diagenesis, while more deeply buried sediments were dominated by stoichiometric magnetite. Simulated burial experiments have also documented neoformation of nanophase magnetite in organic-rich claystones at temperatures up to 130 °C for several days (Kars et al., 2012), consistent with the fine SP oxides produced in our weakly reacted reduced samples.
The stable SD magnetite produced by more strongly reacted nanogoethite is capable of carrying a strong and stable magnetic remanence. Therefore, sub–Curie temperature growth of magnetite in goethite-bearing metasediments will cause profound changes in the paleomagnetic information recorded by such rocks, giving rise to a chemical remanent magnetization, and thereby altering the paleomagnetic signal originally recorded by the sediments at the time of deposition. In light of our results, the origin of nanoscale authigenic magnetite responsible for secondary magnetizations in many carbonates (e.g., Jackson and Swanson-Hysell, 2012) bears reexamination. This work highlights the previously overlooked but potentially important role of goethite as a precursor of magnetic remanence carriers and fine magnetic particles in a vast range of sediments and environmental settings.
The authors thank Nicolas Menguy for assistance with TEM imaging. We thank Ramon Egli and an anonymous reviewer for helpful reviews of this work. This work was supported by the Agence National de Recherche of France under project 2010-BLAN-604-01. This is IPGP contribution 3584.