Journal of Geophysical Research: Planets

Fe oxidation processes at Meridiani Planum and implications for secondary Fe mineralogy on Mars



[1] Fe oxidation processes may have occurred during groundwater-mediated diagenesis in Meridiani Planum sediments. To address this question, melanterite oxidation experiments were conducted at epsomite saturation as a function of pH. Results show that schwertmannite is initially formed from acidic Fe oxidation and that its formation and aging to mixtures of jarosite and nanocrystalline goethite is strongly controlled by pH over the range ∼2.0–4.0. The pH is controlled in turn by Fe oxidation and Fe3+ hydrolysis. In one 77-d oxidation experiment, nanocrystalline hematite was tentatively identified by Mössbauer spectroscopy. Accordingly, aging experiments with synthetic nanocrystalline goethite were conducted (1) to further resolve the formation mechanisms of Fe-phases identified from oxidation experiments and (2) to test whether low water activity (aw) controls the thermodynamically favored goethite to hematite transition at low temperature. Mössbauer spectroscopy and total X-ray scattering show no observable changes after 4 months of aging, and instead, these results point to a jarosite precursor for the tentatively identified hematite. On the basis of these results, we suggest that the oxidation and maturation of initially formed Fe2+-bearing saline minerals may account in large part for the distribution of secondary Fe minerals at the Martian surface, contributing to the association of Fe oxides and Mg/Ca sulfates observed from orbital surface analyses. We hypothesize that oxidation of Fe2+ sulfates at low temperature could account for sustained diagenetic acidity in addition to much of the observed Fe mineralogy in Meridiani Planum outcrop rocks. The origin of the gray crystalline hematite at Meridiani, however, is deserving of further experimental work to test this mechanism.

1. Introduction

[2] Saline mineralogy at the Martian surface is distinct from saline mineralogy on Earth. One of the most important factors contributing to this difference is the abundance and mobility of Fe in surficial environments on Mars [McLennan and Grotzinger, 2008; Tosca and McLennan, 2006; Tosca et al., 2005]. Under the acidic conditions that appear to have influenced much of the Martian surface, substantial quantities of Fe can be released by the chemical weathering of mafic lithologies [Tosca et al., 2004]. In this geochemical setting, Fe becomes highly mobile regardless of its oxidation state. To date, there are several locations on the surface of Mars where Fe sulfate and Fe oxide minerals have been identified. Specifically, detailed characterization of these minerals has been carried out by the Mars Exploration Rover (MER) Spirit in soils at Gusev Crater, and by Opportunity at Meridiani Planum [Klingelhöfer et al., 2006; Morris et al., 2006a, 2006b; Squyres and Knoll, 2005]. In addition, orbital analyses show that in areas where saline minerals have been detected (i.e., Mg and Ca sulfates), the Fe component in the sediments is usually represented by Fe oxide minerals [Bibring et al., 2005; Gendrin et al., 2005]. For example, Bibring et al. [2007] have recently cited specific localities across the Martian surface where sulfate minerals are geologically coupled with ferric oxides including regions of Valles Marineris and Margaritifer Terra. In contrast, detections of jarosite and soluble Fe sulfates at the Martian surface by Mössbauer spectroscopy provide evidence that Fe has a tendency to behave as a soluble element that partitions into Fe-bearing sulfate minerals during surficial processes on Mars [McLennan and Grotzinger, 2008; Tosca et al., 2005]. The long-term stability of soluble Fe-bearing minerals under oxidizing conditions and changing climate is thus somewhat uncertain. However, the transformation of the soluble Fe component of saline assemblages to Fe oxides and Fe-hydroxysulfates is a mechanism that may be more important than originally thought in controlling the distribution of Fe minerals at the Martian surface.

[3] The partitioning of a redox-sensitive element such as Fe into secondary minerals introduces a complicated set of mineral-forming reactions that can take place during the formation and modification of chemical sediments at the Martian surface. As a result, Fe sulfates and oxides may form by a variety of mechanisms. In this manuscript, we focus on the geochemistry and mineralogy of Fe in Meridiani Planum sediments as an example of the potentially complex behavior that Fe may display in the Martian sedimentary record. In addition, we provide experimental evidence to suggest that the long-term maturation of soluble Fe components at the Martian surface is a mechanism that may reveal detailed information on pH and oxidizing conditions over Martian geologic history and may have generated much of the Fe mineralogy at Meridiani Planum.

[4] Of the chemical and mineralogical data obtained on Meridiani Planum outcrop, perhaps one of the more conspicuous pieces of evidence for a changing chemical environment is the oxidation state of Fe. The Mössbauer spectrometer aboard the Opportunity Mars Exploration Rover (MER) has returned data from almost all outcrops, and results reveal that roughly 90% of the total Fe resides in the ferric (Fe3+) state [Klingelhöfer et al., 2006, 2004; Morris et al., 2006b]. The Fe is partitioned between the various minerals that have been identified by the Mössbauer spectrometer, including hematite (in two populations: (1) spherule hematite and (2) hematite contained within the light-toned outcrop), jarosite, trace amounts of igneous silicates, and an Fe3+-bearing component (referred to as Fe3D3) which may include one or more phases that are likely to be poorly crystalline and contain octahedrally coordinated Fe3+ [Klingelhöfer et al., 2006, 2004; Morris et al., 2006b]. Such an observation is striking because as the provenance of the evaporitic sediments is certainly basaltic, it implies that almost all of the Fe involved in the system has been oxidized from Fe2+ in basalt to the predominantly Fe3+-bearing materials now identified in the outcrop. As such, the environment at Meridiani Planum has been repeatedly labeled as “oxidizing” [Klingelhöfer et al., 2004; Squyres and Knoll, 2005], but a potentially complicated history that involves Fe geochemistry in these sediments suggests that additional constraints stand to be uncovered with further study. Indeed, gaining insight into Fe oxidation processes has the potential to yield information about key chemical properties of aqueous Martian environments such as pH, SO42–, oxidation conditions, and the activity of water (which is related to salinity).

[5] Below, we describe an experimentally based investigation of Fe oxidation pathways at Meridiani Planum and the resulting effects on solution composition and mineral speciation in high ionic strength brines. The results of this investigation shed light on (1) the apparent paucity of Fe2+-bearing secondary phases at Meridiani Planum and elsewhere at the Martian surface, (2) the postalteration mineral transformations that have taken place (if any) to create jarosite, hematite and other Fe-bearing secondary phases, and (3) the timing of these reactions/transformations. We also discuss stability relations of various Fe-bearing minerals based on experimental and theoretical data, implications for general aqueous processes and Fe mineralogy at the Martian surface, and provide testable hypotheses for current and future exploration missions such as Mars Express, Mars Reconnaissance Orbiter (MRO), Phoenix and Mars Science Laboratory (MSL).

2. Fe Geochemistry and the Formation of Meridiani Planum Sediments

[6] All of the mineralogical and chemical data collected on Meridiani Planum outcrop are consistent with a detailed hypothesis of sediment formation that is outlined in several previous publications [see Clark et al., 2005; Grotzinger et al., 2005; McLennan et al., 2005; McLennan and Grotzinger, 2008; Squyres et al., 2006a; Squyres and Knoll, 2005; Squyres et al., 2006b; Tosca et al., 2005]. To accompany the following discussion, only a brief summary will be presented here.

[7] The outcrop at Meridiani Planum can be best described as reworked, impure evaporitic sediments [McLennan et al., 2005]. The sediments themselves are mixtures containing an “evaporitic component,” composed mostly of sulfate salts (Mg- and Ca-bearing sulfates from APXS bulk chemical trends and Mini-TES spectral deconvolution) [Christensen et al., 2004; Clark et al., 2005; Glotch et al., 2006; McLennan et al., 2005]. As mentioned above, the sediments also include the oxidation products jarosite, hematite, and the Fe3D3 component [Klingelhöfer et al., 2004]. The remaining fraction of the sediments can be referred to as a “terrigenous component” which contains largely altered siliciclastic material perhaps including amorphous silica, a common alteration product of basalt [Clark et al., 2005; Glotch et al., 2006; McLennan, 2003; McLennan et al., 2005; Squyres et al., 2006b]. The siliciclastic material is interpreted to be altered because chemical trends show that after subtraction of all identified or inferred mineral components, the bulk composition of the residual solids closely resembles a basalt in which a large fraction of divalent cations has been removed relative to Al3+ [Squyres et al., 2006b]. This trend is consistent with acidic alteration of basalt over time, because even though all components are dissolved and “mobile” in the strictest sense, divalent cations are lost in a greater proportion than Al3+ [Tosca et al., 2004]. This chemical trend among RATed Meridiani outcrop rocks is discussed in greater detail and also is shown graphically by Squyres et al. [2006b], providing direct chemical evidence for altered basaltic residue in the sediments.

[8] The depositional environment of the Meridiani sediments first involves chemical weathering of basaltic material and evaporative concentration at what is most likely a shallow playa lake environment [Grotzinger et al., 2005; McLennan et al., 2005]. The chemical weathering products and residual basaltic materials, along with the saline mineral assemblages that have formed from evaporation, are then subjected to aeolian transport (resulting in rounded, sand-sized grains composed of the finer-grained mixture) and deposition into dune-interdune facies [Grotzinger et al., 2005; McLennan et al., 2005]. The upper portion of the Burns Formation identified at Eagle, Endurance and Erebus Craters shows soft-sediment deformation and festoon geometry cross bedding, consistent with subaqueous deposition of the altered basalt-evaporite sand grains in what were most likely transient interdune depressions where groundwater saturated and eventually breached the surface [Grotzinger et al., 2005, 2006]. Syndepositional and postdepositional diagenetic modification of the sediments was controlled by a fluctuating groundwater table percolating through bedded sediments in more than one discrete episode [McLennan et al., 2005]. The sediments that are being characterized by the Opportunity Rover display a series of diagenetic features such as recrystallization textures, secondary porosity (moldic and sheet-like), and hematitic concretions, all of which constrain the relative timing of groundwater percolation into the sediments [McLennan et al., 2005]. It is important to note that in most places, the primary sedimentary fabric (i.e., sorted and rounded grains deposited in an aeolian setting to form coherent bedding structures) is largely preserved, despite the evidence for significant but selective diagenetic modification of some components [Grotzinger et al., 2005; McLennan et al., 2005].

[9] One hypothesis for the observation of selective dissolution of some chemical components while preserving others is Fe oxidation. McLennan et al. [2005] and Tosca et al. [2005] suggested that Fe oxidation processes occurring at high ionic strength (so as to preserve the most soluble Mg sulfate component) would provide a simple mechanism to explain the variation in Fe mineralogy as well as to selectively alter one sediment component after deposition. The hypothesis accounts for the appearance of moldic secondary porosity (indicative of mineral dissolution) by the oxidation of a soluble Fe2+-bearing component deposited with the sediments [McLennan et al., 2005; Tosca et al., 2005]. However, the behavior of Fe can be potentially complex.

[10] For the purposes of the remaining discussion, we divide the geochemical evolution of the sediments into three stages: (1) chemical weathering of basaltic sediments, (2) evaporation, and (3) diagenesis.

[11] We define chemical weathering as the modification of primary basaltic sediments by aqueous fluids resulting in mineral dissolution and, if aqueous conditions permit, the subsequent precipitation of secondary phases (e.g., clays, (hydr)oxides or hydroxysulfates). The end of chemical weathering is marked by the onset of evaporative concentration, resulting in the precipitation of evaporite phases, such as sulfate and/or chloride salts. Diagenesis refers to the chemical processes that have altered the sediments during or after deposition. We are only considering those processes that have been initiated by an aqueous phase and only briefly discuss phase transitions in response to atmospheric conditions over geologic time below.

[12] The predominantly Fe3+-bearing assemblage that has been analyzed by the Opportunity Rover may have formed by a variety of oxidation processes over the course of chemical weathering, evaporation, sedimentary transport, and diagenesis. Because of the number of possible pathways to generate the “end-product” Fe mineralogy in the context of the hypothesis described above, there are several nonunique solutions for the evolution of Fe oxidation and mineralogy at Meridiani Planum.

[13] Bulk compositional analyses show that the silici-clastic component of the Meridiani sediments is compositionally and mineralogically consistent with a substantially weathered residual basaltic material [Glotch et al., 2006; Squyres et al., 2006b] mixed with saline mineral precipitates. This analysis indicates that chemical weathering has played an important role in controlling the initial fluid chemistry that has led to the mixing of saline minerals with residual weathered material. However, the nature of chemical weathering processes that have taken place at Meridiani Planum is less constrained and is so far generally agreed to have been initiated by acid conditions [McLennan and Grotzinger, 2008; Squyres and Knoll, 2005] in which significant Fe is released from ferromagnesian mineral phases such as olivine, pyroxene and Fe-Ti oxide minerals [Burns, 1993; King and McSween, 2005; Tosca et al., 2004]. The overwhelming proportion of the Fe initially released to the fluid from chemical weathering is Fe2+ and may be oxidized by molecular oxygen according to the following reaction:

equation image

We are most interested in this process from two perspectives: (1) the rate of Fe oxidation by molecular oxygen relative to Fe oxidation under high ionic strength conditions and (2) the fate of Fe3+ after oxidation and the subsequent mineral phases which tend to control Fe and other components upon their precipitation.

[14] The rate of reaction (1) has been found to be highly dependent on pH at levels above approximately 3.5 [Singer and Stumm, 1970; Stumm and Morgan, 1996]. Above pH 3.5, the rate law for this reaction may be written as

equation image

where [O2] is the concentration of dissolved oxygen (mol kg−1), [Fe2+] is the concentration of Fe2+(aq) (mol kg−1), {OH} is the activity of OH in solution, and k is the rate constant in mol−1 kg min−1. At pH levels below approximately 3.5, the rate law is independent of {OH} (which is related to pH through the dissociation of H2O) and reduces to

equation image

Burns [1993] and King and McSween [2005] reviewed other critical factors controlling the rate of this reaction. Fe oxidation by molecular oxygen is most affected by pH and oxygen content of the atmosphere. Figure 1 depicts the relationship between Fe2+ oxidation rate and pH as a function of atmospheric oxygen content. Although the pH is constrained to be within the range of jarosite stability (pH ∼ 1.5–4.5), the presence of jarosite in the Meridiani outcrop and the excess sulfur present in bulk chemical analyses do not allow the oxygen content of the paleoatmosphere to be well constrained.

Figure 1.

Rates of Fe2+ oxidation as a function of pH for various atmospheric oxygen levels. All rates become first-order dependent on pH above pH∼3.5. At oxygen levels equal to the present Martian atmosphere, Fe2+ oxidation rates are 4 orders of magnitude lower than the same pH values under the Earth's atmosphere. Using estimates for atmospheric oxygen levels of an early Mars from Catling and Moore [2003], Fe2+ oxidation rates are an additional 7 orders of magnitude lower, or 11 orders of magnitude lower than observed at identical pH conditions on present-day Earth.

[15] One recent evaluation of the evolution of atmospheric oxygen is discussed by Catling and Moore [2003]. Catling and Moore [2003] suggested that early in Martian history, the atmosphere contained a much lower proportion of oxygen owing to a greater flux of chemically reducing volcanic emanations. Model calculations show that increased volcanic activity early in Martian history would have buffered photochemically produced oxygen, creating a more reducing atmospheric environment than the present Martian surface. Presumably oxygen content is tied to volcanic activity and the proportion of oxygen in the Martian atmosphere has risen due to atmospheric loss processes in combination with the dropoff in volcanic activity [Catling and Moore, 2003; Head et al., 2001]. Atmospheric oxygen is contained within Fe-bearing sediments found across the Martian surface, as its conspicuous orange color and the substantial proportion of oxidized Fe suggest [Kolb et al., 2006; McLennan and Grotzinger, 2008]. The question relevant to Meridiani Planum is how much oxidation and when, and how this process was influenced by the evolution of the global Martian climate.

[16] As soon as Fe2+ is oxidized, the precipitation of secondary Fe-bearing minerals is also highly dependent on pH, sulfate and Fe concentration [Bigham and Nordstrom, 2000; Jambor et al., 2000]. Fe-bearing phases that are likely to form from fluids at relatively low pH and in the presence of sulfate are jarosite, schwertmannite and goethite. Hematite may also form directly from aqueous solution, but its formation is often restricted to higher pH usually through a precursor phase such as ferrihydrite (e.g., Schwertmann et al. [1999]; also see detailed summaries by Burns [1993], Burns and Fisher [1990], Cornell and Schwertmann [2003], and King and McSween [2005]). The most important points are the general range in pH where these minerals precipitate and their stabilities relative to one another. Thus, more than one mechanism may result in the same distribution of Fe-bearing minerals.

[17] From the preceding discussion, we are left with three main possibilities for the fate of Fe at Meridiani Planum, as shown schematically in Figure 2: (1) Fe oxidation occurred largely during chemical weathering and Fe3+ is redistributed among different minerals (jarosite, Fe3D3 and hematite) during later diagenesis (path A), (2) Fe oxidation occurred throughout chemical weathering, evaporation, and diagenesis (path B), or (3) the majority of Fe oxidation occurred during groundwater-mediated diagenesis and was relatively minor during chemical weathering (path C).

Figure 2.

Schematic outlining end-member pathways for Fe oxidation during the evolution of Meridiani Planum sediments. Path A corresponds to a history where the majority of Fe was oxidized during weathering with Fe oxidation products transported and distributed with chemical sediments. Path B represents continuous Fe oxidation, occurring over chemical weathering, evaporation, and sediment diagenesis. Path C corresponds to a history where the majority of Fe was oxidized after sediment deposition, during groundwater-mediated diagenesis.

[18] Any path of chemical evolution where Fe2+ is not yet completely oxidized will result in Fe2+-bearing evaporite phases at the outset of diagenesis [Tosca et al., 2005]. From our previous discussion, it appears that Fe2+ was not likely to have been significantly oxidized under an ancient Martian atmosphere that contained a significantly lower proportion of O2 than the present Mars or Earth. This observation is consistent with the identification of siderite (a Fe2+-bearing carbonate that forms in slightly acidic waters under an atmosphere containing CO2) in several SNC meteorites [Bridges et al., 2001; Bridges and Grady, 2000; Bridges et al., 2004]. The detection of preterrestrial siderite in at least five SNC (Nakhlite) occurrences suggests that rates of Fe2+ oxidation were not so rapid as to completely oxidize Fe2+ before its formation. However, it is worth noting that the SNC meteorites are most reflective of subsurface assemblages and may not represent true thermodynamic equilibrium with an overlying atmosphere. Nevertheless, this evidence collectively increases the likelihood that Fe2+-bearing evaporite salts were deposited with soluble Mg and Ca components before diagenesis. If a soluble Fe2+-bearing component was present in the sediments, it is apparently now absent. Could such a component survive diagenesis and aging at Meridiani Planum? Additional Fe oxidation reactions must have taken place to produce the Fe3+-bearing jarosite-hematite-Fe3D3 mineralogy observed at Meridiani Planum, but several mechanisms remain.

[19] If a soluble Fe2+-bearing component was present during or after sediment deposition, what mineral phases were involved? The partitioning of Fe2+ and Fe3+ during evaporation has been investigated with geochemical modeling and more recently with experimental methods [Tosca and McLennan, 2006, 2007; Tosca et al., 2005]. The effect of the Fe2+/FeT ratio on the resulting evaporite assemblage is somewhat complicated by the array of possible mineral phases [Fernández-Remolar et al., 2005; Hudson-Edwards et al., 1999]. The general relationship that has been suggested and supported by observations in Fe-rich acid mine drainage environments is that Fe2+ (aq) is partitioned into melanterite (Fe2+SO4·7H2O) and the amount of melanterite increases with increasing Fe2+ (aq) [Fernández-Remolar et al., 2005; Hudson-Edwards et al., 1999; Tosca and McLennan, 2007; Tosca et al., 2005]. As the amount of Fe3+ increases as a result of oxidation conditions before evaporation, the amount of melanterite decreases, and mixed valence Fe-bearing salts (i.e., copiapite, römerite, bilinite, and voltaite) appear. At high amounts of Fe3+ in solution at the outset of evaporation, the evaporite minerals become a function of the evolving pH, with jarosite precipitating at pH values at less than 3 to about 1. As the pH decreases to less than 1 and approaching −1, ferricopiapite begins to dominate. However, even less acidic conditions (an initial pH of 2) upon evaporation can result in extremely acid levels once they are concentrated and at this stage, rhomboclase may dominate, precipitating at pH levels around –2 and below.

[20] In considering the evaporite component at Meridiani Planum, the two most important factors are the extent of Fe oxidation and the initial pH upon evaporation. A soluble Fe2+-bearing phase such as melanterite may have been a component of the initial saline assemblage at Meridiani Planum, but it is not likely to have survived multiple episodes of groundwater diagenesis. Accordingly, with the assumption that some amount of a Fe2+-bearing saline component was deposited in the evaporite component of the sediments, we are most interested in the reactions that would have occurred in response to groundwater-mediated diagenesis. Drawing from constraints on sediment texture and composition as well as mineral water equilibrium modeling, we begin with the constraint that any such oxidation reaction occurred at or near Mg sulfate (specifically, epsomite) saturated conditions.

3. Experimental Fe2+ Oxidation at High Ionic Strength

3.1. Experimental and Analytical Methods

[21] The primary goal of the experiments performed in this study is to investigate the melanterite oxidation process at high ionic strength and its effect on pH and mineralogy. The experiments were constructed to allow the oxidation of an Fe2+-containing brine (generated from melanterite dissolution) over time with continuous monitoring of pH and changes in mineralogy. The experiments began with the synthesis of a “matrix” fluid saturated with reagent-grade MgSO4·7H2O (epsomite), with approximately 10 g excess to maintain saturation. The matrix fluid was synthesized in either 500 mL or 1 L volumes and added to Pyrex beakers. The solutions were continuously stirred with Teflon stir bars and varying amounts of reagent-grade Fe2+SO4·7H2O (melanterite) were added and allowed to oxidize over time as a result of constant contact with ambient atmosphere (pO2 = 0.21 atm). Experiments were conducted for at least 77 d each for initial melanterite loadings of 1, 10, 200, and 400 g L–1 (e.g., equivalent to adding 400 g melanterite to 1 L of matrix solution). The beakers were covered with paraffin film with two 1-inch lengths of 2 cm OD Teflon tubing inserted through the film to accommodate a pH electrode and a sampling port. Stirring beakers were isolated from the ambient heat given off from stir plates by placing styrofoam sheets between the beakers and stir plates. The temperature of the experiments was kept at an average 21.5°C ± 1.5. Beakers were also corrected for evaporation through periodic weighing and the addition of small amounts of deionized water. The experimental approach is designed to simulate chemical processes that may occur as a result of groundwater diagenesis, or infiltration of water into sediments largely composed of Mg sulfate. During the experiments, aliquots of suspended solid were collected and centrifuged at 5000 or 14000 rpm for 45 min at 25°C. The parent brine was decanted and the supernatant was extracted by pipette, filtered and analyzed for all major solution components. Because residual amounts of brine will evaporate and precipitate unwanted salts upon drying, the moist, centrifuged sample was blotted dry and then added to 50 mL of deionized water and centrifuged again for 45 min at 25°C. Samples were then allowed to air-dry in a fume hood overnight. This preparation removed the brine without affecting the insoluble precipitates removed from solution. As a check to this process, blotted, centrifuged samples and “rinsed” samples were both analyzed by XRD, SEM-EDS and in some cases, Mössbauer, and no differences were noted between samples other than the appearance of residual epsomite and melanterite evident in XRD and SEM analyses.

[22] The two different sample preparation techniques (decanting/blotting or decanting/blotting/rinsing) allowed the presence of highly soluble salt minerals that may have precipitated during the experiment in response to oxidation to be analyzed rather than risking the dissolution of such phases during the washing and filtering process. However, the only highly soluble phases that were identified in the unwashed sample preparations were epsomite and melanterite, both of which are residual products of experiments conducted under epsomite and melanterite saturation, respectively.

[23] In all experiments, a pH electrode was immersed in solution and in some experiments (1, 10, 400 g L–1), pH was monitored with a data logging device to collect time-resolved pH data over the course of the experiments. Because the oxidation of Fe2+ and the precipitation of Fe3+-bearing phases strongly controls pH, experiments were allowed to continue until the final pH of the solution reached a value that was stable over several weeks and the precipitated phases were allowed sufficient time to attain a steady state condition in solution, where solids and solution were largely invariant.

[24] As discussed in further detail below, a supplementary investigation on the stability of nanocrystalline goethite at low water activity was also conducted. This investigation was aimed at (1) resolving the formation mechanisms of tentatively identified hematite produced from oxidation experiments and (2) testing the hypothesis of whether low water activity (aw) controls the thermodynamically favored goethite to hematite transition at low temperature. Nanocrystalline goethite with a target crystallite size of 7 nm was synthesized following the methods described by Mazeina and Navrotsky [2005]. The goethite synthesis proceeded by adding 250 mL of a 2.5 molar KOH solution to 60 mL of a 0.5 molar Fe(NO3)3X9H2O solution. The KOH was added at a rate of 10 mL min–1 with continuous stirring. The resulting mixture was aged for 100 h at 60°C. After aging, the residual electrolyte was decanted from the vessel and the goethite precipitate was centrifuged at 5000 rpm at 25°C. Residual electrolyte was again decanted and replaced with deionized water. The process was repeated until all residual electrolyte was removed and the conductivity of the supernatant approached 0.1 μS cm−1, corresponding to that of deionized water. At the final rinse stage, residual deionized water was decanted and the goethite slurry was added to electrolyte solutions of varying strength, corresponding to varying water activity values. Three aging experiments were conducted with the synthetic nanocrystalline goethite all at pH 2.0: 2.98 mol kg–1 MgSO4, 3.50 and 5.73 mol kg–1 MgCl2. Using published Pitzer ion interaction models to calculate the water activity for the three electrolytes used in our experiments [Pitzer et al., 1999; Rard and Clegg, 1999], we estimate water activity values of 0.91, 0.65 and 0.34, respectively, for the electrolyte solutions listed above.

[25] The decanted supernatant was analyzed for major elements (Mg, FeT, Ca, K, Na, and Al) with a direct current argon plasma emission spectrophotometer (DCP-AES). Solutions were diluted with 4% HNO3 and external solution standards were prepared in 4% HNO3 matrix with concentrations bracketing the unknown sample range as closely as possible. Two-point calibration curves were calculated for every two sample analyses. Total and Fe2+ determinations were made in solution using a HACH DR 2000 UV/Vis spectrophotometer and the 1,10-phenanthroline method. Sulfate analyses were conducted using a barium chloride (SulfaVer4®) colorimetric method. Eh was measured with a platinum combination oxidation-reduction potential (ORP) electrode (Orion model 9678) which was calibrated against ZoBell's solution, which is a well characterized, equimolar solution of Fe2+ and Fe3+ stabilized in 0.1 molal KCl, where Fe is stabilized by aqueous complexing with cyanide [Nordstrom, 1977].

[26] Solid analyses were conducted using powder X-ray diffraction (XRD), scanning electron microscopy with energy dispersive X-ray microanalysis (SEM-EDS) and Mössbauer spectroscopy. Powder X-ray diffraction data were collected using a Scintag PAD X powder diffractometer at 45 kV and 25 mA with Cu Kα radiation. Data were collected between 10 and 50 degrees 2Θ, with a scan step of 0.02 degrees and between 2 and 6 s of counting time per step. XRD patterns were analyzed with Crystallographica Search-Match© software which identifies diffraction peak positions and searches against Powder Diffraction File catalog patterns returning “best fit” matches given a set of chemical constraints on the system. Scanning electron microscopy was performed using a LEO 1550 SFEG SEM at 15 kV and using a 30 μm aperture. The SEM is also equipped with an EDAX energy dispersive X-ray spectrometer (EDS), capable of analyzing semiquantitatively for elements heavier than and including C.

[27] High-energy X-ray total scattering data (to permit pair distribution function (PDF) analysis and nanocrystal structure refinement) were collected on nanocrystalline goethite samples at the 1-ID (∼100 keV, ë = 0.1240(6) A) beam line at the Advanced Photon Source at Argonne National Laboratory [Michel et al., 2007a, 2007b]. A CeO2 standard (NIST diffraction intensity standard set 674a) was used to calibrate the sample-to-detector distance and the nonorthogonality of the detector relative to the incident beam path. The radiation scattered by the calibrant and samples was collected on an amorphous Si detector system manufactured by General Electric. Conversion of data from two to one dimension was done using the program Fit-2D [Hammersley et al., 1996; A. P. Hammersley, ESRF internal report ESRF98HA01T, 1998]. A polarization correction was applied during integration of the data. The total scattering structure function S(Q) and PDF G(r) were obtained using PDFgetX2 [Qiu et al., 2004] where standard corrections were applied as well as those unique to image plate geometry [Chupas et al., 2003]. The composition incorporated in the normalization was FeOOH.

[28] The model for goethite (ICSD 71808) was fitted to the experimental PDF and the structural parameters of the model were refined using the program DiffPy [Farrow et al., 2007]. The parameters varied during the refinement were included in the following order: (1) scale, resolution dampening (σQ); (2) unit cell parameters; (3) r-dependent peak-width ratio (srat); (4) atomic coordinates; (5) isotropic displacement parameters (U); and (6) correlated motion (δ). Additional global parameters related to the PDF and a “goodness-of-fit” indicator, expressed as a weighted residual value (Rw), are included in Table 1. The refined parameters pertaining to the structure are included in Table 2.

Table 1. Refined Lattice Parameters, PDF Global Sharpening and Attenuation Parameters, Residual Values, and Scale Factors for Unreacted and Aged Nanocrystalline Goethitea
Unreactedaw = 0.91aw = 0.65aw = 0.34
  • a

    Values in parentheses indicate uncertain digits.

a, Å4.603(5)4.603(5)4.602(4)4.605(4)
b, Å9.944(9)9.951(3)9.945(3)9.949(9)
c, Å3.024(3)3.024(4)3.024(1)3.025(2)
σQ, Å–10.062(4)0.065(2)0.062(5)0.061(8)
Linear correlation factor δ1.512(5)1.501(1)1.501(4)1.499(5)
Low r sigma ratio srat1.055(2)1.076(8)1.065(4)1.059(3)
Low r cutoff, Å4.
Rw, %15.9(4)14.2(6)15.1(5)14.8(1)
Scale, %124.7(4)126.0(9)125.4(7)121.6(5)
Table 2. Refined Atom Coordinates and Isotropic Displacement Parameters for Space Group Pbnm for Unreacted and Aged Nanocrystalline Goethitea
AtomWyckoff PositionxyzU, Å2
  • a

    Values in parentheses indicate uncertain digits.

Goethite (Unreacted)
Goethite (aw = 0.91)
Goethite (aw= 0.65)
Goethite (aw= 0.34)

[29] Samples were also analyzed with diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy. DRIFT measurements on Fe oxidation products were collected using a Nicolet Nexus 670 Fourier transform infrared spectrometer fitted with an Avatar DRIFT accessory and an MCTA detector. Approximately 40–50 mg of each sample were mixed with KBr and pressed into a sample holder. H2O and CO2 were purged from the DRIFT accessory using a Whatman dry gas generator. Diffuse reflectance spectra were collected from 650 to 4000 cm−1 with a resolution of 2 cm−1 and sample spectra were ratioed against a KBr powder background.

[30] Mössbauer spectra were collected on selected experiment products. Approximately 10–20 mg of each sample were gently crushed and mixed with sugar to thicken the sample and spread it over the holder. Mössbauer spectra were acquired at temperatures of 295 K using a source of ∼60 mCi 57Co in Rh on a WEB Research Co. model WT302 spectrometer (Mount Holyoke College). For each sample, the fraction of the baseline due to the Compton scattering of 122 keV gammas by electrons inside the detector was subtracted. Run times were 12–24 h, and baseline counts were ∼2 million after the Compton correction, as needed to obtain reasonable counting statistics. For samples extracted from goethite stability experiments, baseline counts reached 5.5–9.5 million after the Compton correction. Spectra were collected in 1048 channels and corrected for nonlinearity via interpolation to a linear velocity scale defined by the spectrum of the 25 μm Fe foil used for calibration. Data were then folded before fitting.

[31] To model the data, we used an in-house program from the University of Ghent (Belgium) called Mexfieldd because we expected magnetic sextets to be present. Although these spectra did not contain magnetic components, they did contain heavily overlapping Fe3+ distributions that could not be modeled with model-independent software. Mexfield was thus used to provide Lorentzian line shapes and the capability of solving the full Hamiltonian.

[32] Isomer shift, quadrupole splitting, line width, and doublet area were generally allowed to vary freely in the fits, though constraints were used rarely to ensure that line width did not fall below ∼0.22 mm s–1. Errors on isomer shifts are estimated at ±0.04 mm s–1 because of high peak overlap and low signal-to-noise ratios. Quadrupole splitting values are ± 0.05 mm s–1. The distribution of area among multiple Fe3+ doublets is probably ±5–10% absolute.

3.2. Results

3.2.1. Acid Generation

[33] When melanterite was added to all epsomite-saturated brines, pH immediately decreased from the initial value of 5.7. This decrease was observed within the first 5–10 min of each experiment and is a function of the amount of melanterite added to solution. Figure 3 shows the results of experiments that employed pH data loggers. An initial decrease in solution pH is observed for each experiment, occurring over 5–15 min. The magnitude of the pH decrease increases from 1 to 10 to 400 g L–1 melanterite added. Solution samples analyzed immediately (within the first 5 min of beginning the experiment) revealed that the amount of Fe3+ in solution at the beginning of the experiment was below the detection limit. Frau [2000] attributed this phenomenon to the hydrolysis of the Fe2+ ion as a result of melanterite dissolution (if oxidation and hydrolysis of the Fe3+ ion is insignificant):

equation image

The equilibrium constant for the hydrolysis reaction of Fe2+ can be compared to the only other major acid producing or consuming reaction in the system if the oxidation of Fe2+ is neglected for this stage of the experiment: protonation of aqueous sulfate to form the bisulfate ion, HSO4:

equation image

As noted by Frau [2000], the equilibrium constant for reaction (4) is equal to 10–9.5 and the equilibrium constant for reaction (5) is equal to 10–12.1 [Stumm and Morgan, 1996]. This difference would indicate that the aqueous Fe2+ ion behaves as a stronger acid than the SO42– ion behaves as a base [Frau, 2000], resulting in net acid production. Our results agree with melanterite dissolution experiments in pure water conducted by Frau [2000] in that an initial drop in pH was observed within minutes of melanterite addition. As the experiment proceeded, the initially produced acidity sets the pH at a level where Fe2+ oxidation (according to reaction (1)) and Fe3+ hydrolysis cause a further gradual decrease in pH over time. This effect can be seen in Figure 4, which shows the evolution of solution pH for the 400 g L–1 (melanterite saturated) experiment over time. The initial decrease in pH is shown as a sharp decrease in the first 20 min. After this interval, the pH continues to decrease at a shallower slope. For these long-duration experiments conducted at melanterite saturation, the pH reached a minimum of 2.05 after 77 d.

Figure 3.

Evolution of pH during Fe oxidation experiments conducted in epsomite-saturated solutions. Upon the addition of 1, 10, and 400 g L–1 melanterite, the pH of the solution shows a rapid decline, corresponding to Fe2+ hydrolysis. In these experiments, the magnitude of the initial pH decrease is proportional to the amount of melanterite added.

Figure 4.

Long-term pH evolution in a 400 g L−1 melanterite oxidation experiment. At 400 g L−1, melanterite is saturated in addition to epsomite in this experiment, simulating groundwater flow through porous sulfate-rich sediment. After the initial decline in pH resulting from melanterite dissolution, pH continues to decrease, resulting mainly from the hydrolysis of the Fe3+ ion upon Fe2+ oxidation.

3.2.2. Mineralogical Evolution

[34] In all of the experiments the color of the solution changed from green to yellow to brown over time. The first yellowish and cloudy colors of the solutions could be observed after several hours had elapsed (∼5–10 h). All experiments resulted in the presence of identifiable precipitates. The melanterite-saturated (400 g L−1) experiment exhibited the most variation in mineralogy.

[35] The initial precipitate in all experiments was identified by a combination of powder XRD, SEM-EDS and by Mössbauer to be schwertmannite (ideally FeO(OH)3/4(SO4)1/8). Figure 5a shows the initially extracted precipitate from a 10 g L–1 experiment. The schwertmannite generally displays fibrous morphology and occurs in “bundles” of individual fibers that are approximately 1–2 μm long and less than 100 nm wide. SEM-EDS results show that the schwertmannite was generally of consistent chemical composition – semiquantitative analyses gave Fe:S ratios of approximately 6 (Figure 5a). Powder X-ray diffraction of the initial precipitate yielded broad peaks (Figure 6), the strongest of which occurred at 35.2 degrees 2Θ (d = 2.55 Å), consistent with schwertmannite [Bigham et al., 1994, 1990]. A Mössbauer spectrum of the initial precipitate obtained from a 400 g L–1 experiment after 7 d is shown in Figure 7. The spectrum is identical to that of a natural schwertmannite standard, and is composed of three subspectra representing a distribution of varying nuclear environments for 57Fe in the schwertmannite structure. Mössbauer parameters for this sample are given in Table 3.

Figure 5.

Scanning electron micrographs of Fe oxidation products from experiments described in this study. (a) Schwertmannite precipitated from a 10 g L–1 experiment. Energy-dispersive X-ray microanalysis shows a Fe/S ratio of ∼6 for this phase, consistent with the range of SO4 reported in the literature. (b) Schwertmannite and jarosite precipitated from a 10 g L–1 experiment. EDS analyses show that the jarosite is K bearing with a composition along the K-H3O join. (c) Schwertmannite precipitated from a 400 g L–1 experiment. (d) Jarosite precipitated from a 400 g L–1 experiment.

Figure 6.

Powder X-ray diffractograms of Fe oxidation products precipitated from a 200 g L–1 experiment. After 7 d, schwertmannite can be identified by a series of weak reflections, the strongest of which occurs at 35.2 degrees 2Θ (d = 2.55 Å). After 21 d, jarosite can be identified, increasing in abundance over the duration of the experiment. At 77 d, goethite appears in addition to jarosite, with broadened peaks suggesting nanometer-sized crystallites.

Figure 7.

The 57Fe Mössbauer spectrum of Fe oxidation products formed after 7 d elapsed in a 400 g L–1 experiment. The only mineral phase identified, consistent with XRD and SEM-EDS results, is schwertmannite. The schwertmannite phase is composed of three doublets, the distribution of which reflects the varying nuclear environment of 57Fe in the schwertmannite structure. The Mössbauer parameters for this sample are listed in Table 3.

Table 3. Mössbauer Parameters of Fe Oxidation Products From a 400 g L–1 Experimenta
Sample NamePhaseδ, mm s−1Δ, mm s−1ΓArea, %
  • a

    Errors for IS (s), QS (d) and area are ±0.04, ±0.05, and ±5–10% absolute. Schw, schwertmannite; Jar, jarosite; np-Gt, nanophase goethite; (np-Hm), nanophase hematite.

400 g/L 7 daysSchw0.370.520.3033
400 g/L 14 daysSchw0.380.800.6677
400 g/L 21 daysSchw0.370.820.7066
400 g/L 77 daysSchw0.280.760.352

[36] As the experiments progressed, jarosite appeared after the formation of schwertmannite in the 10, 200, and 400 g L–1 experiments. Jarosite was evident in the 400 g L–1 experiment after 14 d and increased in abundance over time. Clear evidence for jarosite can be seen in the XRD data, where the jarosite diffraction peaks are sharply defined compared to schwertmannite, which occurs as a poorly crystalline phase. Powder XRD supports the presence of jarosite after 21 d in the 200 and 400 g L–1 experiments and after at least 77 d in the 10 g L–1 experiment (see Figure 6). In addition, SEM-EDS analyses show that the jarosite occurs as larger hexagonally shaped crystals approximately 1–2 μm in size (Figures 5b and 5d). SEM-EDS results also show that the jarosite precipitated is K and H3O-bearing. The K is a result of K contamination to the solution from the pH electrode used during the experiment. As the fill solution of the electrode is KCl, K was progressively leached into the beaker over the course of the experiment. However, as basaltic weathering solutions invariably contain several ppm K [Tosca et al., 2004], this is a relevant chemical component to include in the system, as it affects jarosite solubility. Figure 5b shows that, while individual analyses vary, the average K:S ratio in the jarosite is 0.42, which is consistent with a K-H3O solid solution composition of approximately (K0.84, H3O0.16)Fe3(SO4)2(OH)6. Finally, a jarosite doublet appears in Mössbauer spectra acquired on precipitates sampled at 14 d from the 400 g L–1 experiment (Figure 8).

Figure 8.

The 57Fe Mössbauer spectrum of Fe oxidation products from a 400 g L–1 experiment. The spectrum is dominated by schwertmannite, but jarosite and nanocrystalline goethite appear after 14 d.

[37] Goethite was also observed in all of the experiments, but at different time intervals and in varying abundances. For example, in the 1 g L−1 experiment, goethite appeared earlier and was the dominant phase as opposed to the 10, 200, and 400 g L−1 experiments, where goethite did not appear until at least 14–21 d and remained less abundant. All of the goethite identified from the experiments is nanocrystalline – the powder XRD peaks corresponding to goethite are broadened and SEM-EDS analyses of the goethite could not resolve individual particle sizes. Figure 6 shows XRD data for the 200 g L−1 experiment after 77 d. Nanocrystalline goethite can be identified in this sample from several broad diffraction peaks in addition to jarosite. In addition, Mössbauer spectra of the extracted solids from the same experiment at 14 and 21 d show a consistent goethite feature with average isomer shift and quadrupole splitting values of 0.37 and 0.50 mm s−1, respectively (Figures 8 and 9and Table 3).

Figure 9.

The 57Fe Mössbauer spectrum of Fe oxidation products from a 400 g L−1 experiment. Schwertmannite decreases in abundance at 21 d, with nanophase goethite increasing in abundance.

[38] The precipitates formed from the series of Fe oxidation experiments described above crystallize as a function of pH. Figure 10 shows powder X-ray diffractograms collected on final products from four melanterite oxidation experiments: 1, 10, 200, and 400 g L−1. For each experiment, the evolution of pH was dictated by the amount of melanterite present. For each sample, the pH value at which schwertmannite was identified is listed. At decreasing values of the pH at which schwertmannite ages, jarosite increases in abundance and nanocrystalline goethite decreases. The amount of jarosite appears to correspond to the amount of melanterite added, and thus, the pH. At lower pH values of 3.0 and 2.7 jarosite dominates the 200 and 400 g L−1 experiments. During the 1 g L−1 experiment which reached the highest pH values, jarosite was absent and only goethite with a minor amount of schwertmannite remained. It should also be noted that in forced hydrolysis reactions of Fe2(SO4)3 solutions, the concentration of Fe3+(aq) exerts control on the precipitation of goethite and jarosite, with low and high concentrations favoring these two phases, respectively, and may play a minor role in determining the nature of the solid Fe3+ precipitates identified in our experiments [Kandori et al., 2004]. From these results, we suggest that schwertmannite acts as a precursor phase to jarosite and goethite at the conditions investigated in the experiments. The transformation from schwertmannite to jarosite is seen in XRD, Mössbauer and FT-IR data sets for the 200 and 400 g L−1 experiments. FT-IR data show a splitting of the broad absorption feature centered at ∼1130 cm−1 to three distinct features at 1011, 1087, and 1213 cm−1. The broad feature present at 7 d corresponds to ν3(SO4)2− vibrations in the schwertmannite structure, whereas the three distinct absorptions at 1011, 1087 and 1213 cm−1 correspond to δOH bending and ν3(SO4)2− stretching vibrations in the jarosite structure (Figure 11) [Bishop and Murad, 1996, 2005].

Figure 10.

Powder X-ray diffractograms of final Fe oxidation products formed after 77 d in 1, 10, 200, and 400 g L−1 experiments. The pH values indicate the minimum pH at which schwertmannite has precipitated. The pH at which schwertmannite ages controls the relative proportions of jarosite and nanocrystalline goethite. At a pH of 4.4, the final products are largely goethite and a minor amount of schwertmannite. At decreasing pH (controlled by increasing amounts of melanterite addition), goethite decreases in abundance and jarosite increases in abundance, yielding final products composed largely of jarosite with minor amounts of nanocrystalline goethite (J, jarosite; G, goethite; S, schwertmannite).

Figure 11.

Diffuse reflectance (DRIFT) FT-IR spectrum of Fe oxidation products produced from a 10 g L−1 experiment. At 7 d elapsed, the broad absorption feature centered at ∼1130 cm−1 appears, arising from ν3 (SO4)2− vibrations in schwertmannite. As the experiment progresses, the broad feature at 1130 cm−1 splits to three distinct absorptions at 1011, 1087, and 1213 cm−1 (corresponding to δOH bending and ν3 (SO4)2− stretching vibrations), resulting from the increasing abundance of jarosite formed in the experiment.

[39] Finally, after 77 d, analysis of Mössbauer spectra of products formed from the 400 g L−1 experiment yielded an additional and new component. The component has distinct Mössbauer parameters from schwertmannite, jarosite and nanocrystalline goethite, all of which display consistent behavior in Mössbauer spectra. The IS and QS values of the component are 0.44 and 0.57 mm s−1, respectively. With these distinct parameters, we assign the component to nanocrystalline hematite (Figure 12). However, it is important to note that this particular feature can be nonunique to Mössbauer spectroscopy in this system because several Fe-bearing phases exhibit features corresponding to the resulting range in QS and IS. Tentatively identified hematite in this experiment could be derived from either a nanocrystalline goethite or jarosite precursor. Thus, to constrain the mechanism of hematite formation in this system, nanocrystalline goethite aging experiments were conducted and are described below.

Figure 12.

The 57Fe Mössbauer spectrum of Fe oxidation products from a 400 g L−1 experiment. The product is dominated by jarosite and nanocrystalline goethite with a trace amount of a doublet assigned to residual schwertmannite. After 77 d, a distinct phase appears in the Mössbauer spectrum, consistent with nanocrystalline hematite.

4. Discussion

4.1. Acid Generation and Sustaining Acidic Conditions at Meridiani Planum

[40] An important observation from this experimental investigation of melanterite oxidation, and one that is particularly relevant to Meridiani Planum geochemistry, is the efficiency of the acid-generating processes in which aqueous Fe is involved. In the experiments described above, large amounts of acid are liberated from the oxidation of small amounts of melanterite. In general, the acid generating reactions of the Fe oxidation process result in solutions that typically achieve “steady state” pH values of less than or approximately 2.0, which is consistent with equilibrium phase boundary predictions between jarosite-goethite and jarosite-hematite [Bigham et al., 1996b; Tosca et al., 2005]. The initial pH decrease observed within the first 15 min of all experiments is also consistent with data reported by Frau [2000], where a similar trend in pH was observed for melanterite dissolution experiments in pure water. Acid generation resulting from Fe oxidation is an important observation because it may have been crucial in fixing and maintaining an initially acidic pH throughout much of the geochemical evolution of the Meridiani Planum sediments.

[41] There are several forms of Fe-bearing sulfate minerals that incorporate Fe2+ and show similar acid-producing potential upon dissolution and further oxidation. For example, Jertz and Rimstidt [2003] conducted dissolution experiments of natural sulfate-rich samples collected from an acid mine drainage site. In their experiments, Jertz and Rimstidt [2003] monitored pH as a function of wt % of the solid sample added. Samples dominated by melanterite, fibroferrite, halotrichite and copiapite were used and their acid-generating efficiencies were monitored. For the melanterite-dominated sample, addition of greater than 0.2 wt. % fixed the pH at a value of approximately 3.0. For the copiapite and halotrichite samples, addition of 0.2 wt. % resulted in a pH of 2.5 and for the fibroferrite sample, a pH of 2.0. The acid-generating potential of the dissolution of Fe2+, Fe3+, and mixed valence Fe sulfate minerals in water is clear. A combination of hydrolysis processes rapidly fixes an acidic pH for many of the Fe sulfates, and the presence of a soluble Fe-bearing component in the Meridiani sediments after deposition could have played a major role in maintaining acidity.

[42] The question of sustaining an acidic environment in the Meridiani system has remained an open one largely because of the seemingly incompatible nature of acidic conditions with a basaltic substrate [McLennan et al., 2005; Tosca et al., 2005]. However, the variety of acid-generating processes involving Fe(aq) and their rapid and strong control on pH in the system provides a simple means to provide acid to the system. Because the amount of acid generated is a function of the amount of mineral added to water, this relationship can be translated to a water-to-rock ratio or, water-to-melanterite ratio in the sediments. The ratio of water to melanterite during diagenesis in the Meridiani Planum sediments may have been controlled by porosity, permeability and simply the mass fraction of melanterite (or another Fe-bearing sulfate) that was present in the initial evaporite assemblage.

4.2. Schwertmannite and Its Stability

[43] In acidic Fe sulfate systems, mineralogy is controlled by a number of factors. From field and laboratory studies, pH and SO4 concentration are among the most important variables controlling Fe mineral speciation in this system [Bigham and Nordstrom, 2000; Jambor et al., 2000]. The precipitation of schwertmannite in all of our experiments supports the likelihood that schwertmannite may be found at the Martian surface if (1) oxidation of Fe2+ in acid sulfate environments occurs and (2) it remains stable over geologic time. Laboratory studies invoking inorganic and microbial Fe2+ oxidation at acidic pH consistently report initial schwertmannite precipitation [Acero et al., 2006; Bigham et al., 1990, 1996a; Murad et al., 1994] as do countless field studies characterizing the mineralogy of numerous acid sulfate environments on Earth [Murad and Rojik, 2003; Nordstrom and Alpers, 1999; Regenspurg et al., 2004]. Bishop and Murad [1996] suggested schwertmannite as an important Fe oxidation product on Mars on the basis of distinct features present in reflectance spectra. In fact, schwertmannite (and other poorly crystalline Fe oxide or hydroxide phases) has recently been suggested as a component in Meridiani outcrop material from visible and near-IR multispectral outcrop analysis [Farrand et al., 2007] and from coordinated MER and OMEGA spectral analyses of the Meridiani Planum region [Bibring et al., 2007]. Schwertmannite is also consistent with the Fe3D3 component of Mössbauer spectra analyzed in situ by the Opportunity rover [Klingelhöfer et al., 2004].

[44] Perhaps more important in a geological sense is the metastability of schwertmannite with respect to goethite and jarosite. In dry conditions out of contact with aqueous solution, schwertmannite appears to be stable [Bigham et al., 1996b; Bishop and Murad, 1996]. However, schwertmannite in water may completely transform to goethite or jarosite. The rate of this transformation has been shown to be a function of pH, with the rate of transformation increasing with increasing pH [Schwertmann and Carlson, 2005]. The pH of the aging solution exerts a strong control on the relative amounts of jarosite and/or goethite produced from schwertmannite. Agreement among several laboratory and field studies also shows that schwertmannite controls Fe(III) and SO4 concentrations in the short term, even down to a pH of 1.9 [Nordstrom and Alpers, 1999]. However, on geological timescales, schwertmannite is a precursor to goethite and jarosite; under most near surface conditions, it will decompose to a mixture of these phases depending on pH. Our results suggest that over the somewhat narrow pH range of ∼2.0–4.0, the resulting assemblages change from jarosite-dominated to goethite-dominated over relatively short timescales.

[45] The clearest evidence for the strong influence of pH on schwertmannite transformation can be seen in powder XRD analyses of final Fe oxidation products from 1, 10, 200, and 400 g L−1 experiments. In Figure 10, the relative amount of jarosite to goethite increases with a decreasing pH of schwertmannite aging. This result is consistent with other studies that have investigated the evolution of Fe sulfate mineralogy under acidic conditions [Acero et al., 2006; Bigham et al., 1996a; Jönsson et al., 2005; Kawano and Tomita, 2001]. The formation of goethite in our experiments is most likely the result of a schwertmannite transformation process rather than direct hydrolysis from solution because of the significant amount of time elapsed before the occurrence of goethite in the 200 and 400 g L−1 experiments. This is consistent with previous studies investigating the aging of schwertmannite to goethite [Acero et al., 2006; Jönsson et al., 2005]. In the 1 and 10 g L−1 experiments, goethite appeared sooner and may have either been a more rapid and complete transformation process induced by the higher pH, or direct hydrolysis of Fe3+ from solution. The formation of jarosite, on the other hand, is likely to be a transformation reaction rather than a separate aqueous precipitate. Support for this hypothesis comes in part from a study performed by Acero et al. [2006], who reported the formation of both goethite and jarosite from aging a schwertmannite sample. Mössbauer spectroscopy of Fe oxidation products from a 400 g L−1 experiment support a transformation and aging process of schwertmannite to both goethite and jarosite. In Figure 13, the peak area of schwertmannite steadily decreases with a simultaneous increase in the abundances of both jarosite and goethite. This observation can be best explained by the presence of two competing reactions in this system, the transformation of schwertmannite to goethite (by dissolution-reprecipitation):

equation image

and the transformation of schwertmannite to jarosite, by the same mechanism:

equation image

Because of the relatively rapid transformation rates associated with the above phase changes, schwertmannite identification on Mars could constrain not only Fe, SO4, pH and oxidation conditions, but also the duration of liquid water in the associated area.

Figure 13.

Mössbauer spectrum peak areas (expressed in %) for the evolution of a 400 g L−1 experiment to 77 d. The peak areas are taken as estimates of mineral abundance in the samples. As the experiment progresses, the abundance of schwertmannite decreases with a concomitant increase in jarosite and nanocrystalline goethite abundance. This trend in mineral abundances suggests that schwertmannite transforms to both jarosite and goethite in this experiment.

4.3. Nanocrystalline Goethite and Its Stability

[46] All goethite identified in our experiments is nanocrystalline (i.e., possessing an ordered crystallographic domain on the order of tens of nanometers). The evidence for the nature of goethite in our samples comes directly from XRD analyses of all samples containing goethite where diffraction peaks are significantly broadened (for example, see Figure 6). In addition, individual goethite particles could not be distinguished at the resolution of the scanning electron microscope. The nanocrystalline nature of goethite is common in acid sulfate conditions [Waychunas et al., 2005]. Its widespread presence is the result of rapid nucleation of goethite particles under specific conditions in combination with slow growth kinetics resulting from the presence of the sulfate ion [Cornell and Schwertmann, 2003; Nordstrom and Alpers, 1999]. The nanocrystalline nature changes the solubility of goethite and also contributes to its persistent formation at low pH values (i.e., 2.0) from solution despite the high sulfate activity used in our experiments. In addition to being another possible contributing phase to the Fe3+-bearing Fe3D3 component in the Mössbauer spectra of Meridiani outcrop, a closer examination of nanocrystalline goethite stability shows that this phase may serve as precursor to facilitate hematite formation under low-temperature high ionic strength acid sulfate conditions.

[47] Langmuir [1971] demonstrated, on a thermodynamic basis, the instability of goethite relative to hematite as a function of grain size. The main conclusion from the effect of particle size on goethite stability is that under nearly all surficial geologic conditions on Earth, nanocrystalline goethite is unstable with respect to coarse-grained hematite. Recently, these same stability relationships were revisited by Mazeina and Navrotsky [2005], who determined the surface contribution to the enthalpy of formation of goethite through calorimetry, providing an important step in quantifying the particle size effect on goethite stability. In addition, low water activity favors the dehydration of goethite and at least thermodynamically, drives the transformation to hematite [Langmuir, 1971].

[48] A variety of observations support the effect of water activity on the formation of goethite and hematite in specific environments. For example, a study conducted on the aging of ferrihydrite [Torrent et al., 1982] shows that relative humidity (equivalent to the activity of water) controls the conversion of ferrihydrite to goethite and/or hematite. Torrent et al. [1982] showed that low relative humidity favors hematite crystallization, while high relative humidity favors goethite formation. In addition, Trolard and Tardy [1987] investigated the stability relationships of Al-bearing goethite and hematite and showed that lower water activity favors hematite crystallization over goethite at aw values less than 0.883. In combination with particle size effects on stability, the net result is that a small increase in particle size makes goethite unstable with respect to hematite at all water activities. The water activity values for the experimental solutions used in this study range from 0.90 (1 g L−1 melanterite) to 0.84 (400 g L−1 melanterite), based upon calculations performed using measured solute molalities. Thus, we can expect hematite to be thermodynamically favored in place of goethite in these experiments and possibly in many surface environments where saline waters exist in contact with goethite.

[49] To test the hypothesis that aging of nanocrystalline goethite at low water activity may in fact drive the transition to hematite, aging experiments were conducted with synthetic goethite using electrolyte solutions with varying water activities. Nanocrystalline goethite was synthesized as described above and the estimated crystallite size of the final product was approximately 25 nm, using the Scherrer equation and instrumentally corrected peak broadening from powder X-ray diffraction analysis (Figure 14) [Klug and Alexander, 1974]. Freshly synthesized goethite and goethite samples subject to four months of aging were analyzed by Mössbauer spectroscopy and by total X-ray scattering using pair distribution function (PDF) analysis [Michel et al., 2007a,2007b]. The total X-ray scattering and PDF method is especially suited for the analysis and structure refinement of nanocrystalline materials; it is a powerful technique in extracting quantitative structural information and in identifying neoformed nanocrystalline precipitates produced during the aging experiments that conventional powder XRD may not detect (see Michel et al. [2007a, 2007b] for detailed discussion). The X-ray scattering results are shown in Figure 15. From analysis and comparison of the PDFs of unaltered goethite and aged goethite under three electrolyte solutions, there is no appreciable change in the starting material and no newly formed secondary phase present above the detection limit of the method (approximately 5%). Similarly, Mössbauer spectra acquired on the same materials shows that the spectrum of synthetic goethite is essentially unchanged after 4 months of aging. Figure 16 shows the Mössbauer spectra for each material with corresponding Mössbauer parameters listed in Table 4. Three diffuse sextets are present in each spectrum, with very similar parameters and slightly different areas. The slight asymmetry and line broadening are typical for Mössbauer spectra of iron oxides, and can result from many factors, including suuperparamagnetic relaxation, different kinds of magnetic excitations, cluster ordering, and surface effects. However, we were able to fit the spectra with three hyperfine field distributions to accommodate the broadening.

Figure 14.

Powder X-ray diffractogram of unreacted, synthetic nanocrystalline goethite.

Figure 15.

Experimental PDF G(r) versus distance r for unreacted, synthetic nanocrystalline goethite. The calculated PDF for goethite is plotted to show that the synthetic product is indeed goethite as the difference between experimental data and the calculated model is minimal. The inset plots the weighted total scattering structure function Q[S(Q)-1] and depicts the location of diffraction maxima for reference.

Figure 16.

Mössbauer spectra of synthetic goethite samples after 4 months of aging in various electrolyte solutions. (a) Unaltered, synthetic goethite, (b) 2.98 mol kg−1 MgSO4 solution, (c) 3.50 mol kg−1 MgCl2 solution, and (d) 5.73 mol kg−1 MgCl2 solution.

Table 4. Mössbauer Parameters of Nanocrystalline Goethite From Aging Experiments Conducted at Varying Water Activitya
 Unreacted Goethiteaw = 0.91aw = 0.65aw = 0.34
  • a

    Errors for IS (s), QS (d), and area are ±0.04, ±0.05, and ±5–10% absolute.

Sextet 1
δ, mm s−10.370.370.370.37
Δ, mm s−1−0.27−0.27−0.26−0.27
Γ, mm s−10.360.230.290.35
Percent area33293536
Sextet 2
δ, mm s−10.370.290.380.37
Δ, mm s−1−0.20−0.28−0.20−0.20
Γ, mm s−11.010.850.880.98
Percent area39363442
Sextet 3
δ, mm s−10.370.360.370.37
Δ, mm s−1−0.28−0.29−0.28−0.28
Γ, mm s−10.370.330.310.32
Percent area28353122

[50] Interpretation of the Mössbauer parameters is complicated by their very small grain sizes. However, the samples are well constrained by thermodynamics to be either goethite or mixtures of goethite with hematite. Bulk goethite has 293 K parameters of δ = 0.37 mm s−1, Δ = −0.27 mm s−1, and BHf = 38.1 T [Morris et al., 1985]. Bulk hematite spectra have 293 K sextets with δ = 0.37–0.39 mm s−1, Δ = –0.22–0.41 mm s−1, and BHf = 46.4–52.2 T [Morris et al., 1985; Cornell and Schwertmann, 2003]. Nanocrystalline (nanophase) hematite (∼10 nm crystallite size) is a superparamegnetic doublet [Morris et al., 1989], which is not observed in these spectra. Given that the BHf values for all our data are <36 T, the presence of hematite in any of the samples is unlikely. Thus, it appears that all three sextets represent distributions of iron in goethite.

[51] Although thermodynamic analysis of nanocrystalline goethite suggests that transformation to hematite is favored under low water activity, this reaction may suffer from kinetic controls to prevent its completion at least over laboratory timescales. While the results from synthetic goethite aging experiments do not support the transformation of goethite to hematite under high ionic strength, they do suggest that if the secondary phase present in the melanterite oxidation experiments identified by Mössbauer spectroscopy (Figure 12) is indeed hematite, then another mechanism is likely to have been responsible for its formation.

[52] The nanocrystalline goethite aging experiments have essentially duplicated the chemical conditions present in the 400 g L−1 experiment and no hematite was produced over similar timescales. However, a more attractive mechanism for the formation of hematite may be the transformation of jarosite to hematite at high ionic strength. Indeed, Barrón et al. [2006] found that the conversion of jarosite to hematite occurs with increasing ionic strength at low temperature and acidic conditions. However, the ionic strength (∼12 molal) of epsomite-saturated solutions used in this study is significantly higher than the range of ionic strength investigated by Barrón et al. [2006]. The increase in ionic strength may prevent the formation of goethite and favor the formation of hematite, but the amount of time required to yield hematite observable to Mössbauer spectroscopy, for example, is unknown. Nevertheless, either the conversion of goethite or jarosite to hematite are viable mechanisms for hematite at Meridiani Planum and at other occurrences in association with sulfate minerals identified from orbit [Bibring et al., 2007].

4.4. Constraints on Fe Mineral Formation at Meridiani Planum

[53] Diagenetic processes relevant to Meridiani Planum have been investigated by several workers, but only three studies have addressed mechanisms for the distribution of Fe mineralogy throughout the sediments. Barrón et al. [2006] report the results of aging experiments conducted on jarosite at varying pH levels and ionic strength; they found that jarosite could directly convert to hematite at low ionic strength (0.25 molal and lower) slightly acidic to circumneutral pH (4–8), with phosphate and elevated temperatures (∼60°C) accelerating the process. Other products that were formed in their experiments included goethite, lepidocrocite, and ferrihydrite. However, no conversion to hematite was reported to take place at experiments conducted at pH 4–6 at high ionic strength (2.2 molal) regardless of phosphate content. The mechanism for hematite precipitation in these experiments is the dissolution/reprecipitation of jarosite to hematite under weakly acidic conditions and in the presence of phosphate. The study presents a simple mechanism for the formation of hematite under these conditions, but the question remains as to whether the same processes occur at high ionic strength values approaching those of a Mg sulfate saturated brine (>12 molal at 25°C). The results of the Barrón et al. [2006] study do imply, however, that the jarosite must have been deposited with sediments prior to diagenesis. It is not possible to constrain whether the jarosite formed during weathering, evaporation, diagenesis or a combination of these processes. Nevertheless, the main point of the model is that jarosite itself acts as a hematite precursor, suggesting that formation of hematite from jarosite at low temperature, acidic high ionic strength conditions is a likely mechanism and one that may account for almost all observed Fe mineral components identified in Meridiani Planum outcrop. As our results show, this complex mineral assemblage and the associated acidic conditions are formed solely from the oxidation of Fe2+-bearing sulfate salts such as melanterite.

[54] Another geochemical model for the distribution of Fe-bearing minerals at Meridiani Planum is one suggested by Morris et al. [2005] and more recently by Golden et al. [2007]. The model proceeds by a gradual Fe3+ hydrolysis mechanism, initiated by the chemical weathering of basalt that causes an increase in pH, precipitating aqueous Fe3+ in the form of jarosite and hematite spherules. The model is based on field observations from the Mauna Kea locality in Hawaii as well as more recent laboratory simulations of the process [Golden et al., 2007]. Field observations and supporting laboratory work involve hydrothermal conditions (e.g., ∼150°C) to drive the forced hydrolysis of Fe3+ to hematite spherules. However, in the experiments the aging solution included a significant Cl concentration (from Fe added as a FeCl3 component) in the presence of SO4. It is unclear how the reaction may proceed at higher observed SO4/Cl ratios characteristic of those measured in Meridiani Planum outcrop. This model implies that the majority of Fe3+ is oxidized before diagenesis and exists in an unspecified mineral phase or that the sediments are deposited with the evaporitic sediments and later altered during diagenesis to release Fe3+ in solution.

[55] From the experimental results discussed in this study, we suggest that the majority of the Fe-bearing mineralogy analyzed at Meridiani Planum originated from the oxidation and alteration of the soluble Fe component deposited with evaporite minerals. The mineralogical evolution of Fe2+ oxidation of saline components based on our experimental results is shown schematically in Figure 17. The overall process is acid generating. Mineral assemblages composed largely of jarosite, goethite and minor amounts of schwertmannite with relative amounts are strongly controlled by pH. The pH is controlled by sediment porosity and the amount of soluble Fe-bearing minerals in contact with the infiltrating diagenetic fluid. The transformation of jarosite or nanocrystalline goethite (which are ubiquitous in acid sulfate environments) to hematite results in a final assemblage that represents several secondary Fe mineral components identified in Meridiani outcrop. Interestingly, spectral analyses of the Meridiani hematite from orbit are consistent with its evolution from a goethite precursor [Glotch et al., 2004]. The model has the advantage that the majority of Fe oxidation occurs in response to diagenesis. Only the Fe component in the evaporite portion of the sediments is affected, providing a possible explanation for the formation of moldic and sheet-like secondary porosity as suggested by McLennan et al. [2005] and Tosca et al. [2005].

Figure 17.

Schematic outlining the general mineralogical evolution during high-ionic strength Fe oxidation as suggested by the experimental results described in this study. Schwertmannite appears as the initially precipitated Fe phase provided Fe oxidation occurs under slightly acidic conditions (pH 3–5). Schwertmannite acts as a precursor to jarosite and goethite in this system and the relative abundance of these phases is controlled by the pH at which schwertmannite ages. Nanocrystalline goethite and jarosite are thermodynamically unstable with respect to hematite under these conditions. However, the mechanism of the goethite transformation to hematite at low-temperature remains unresolved.

[56] However, whether the model can fully account for the spherical morphology of hematite concretions found in the outcrop is less clear. Aggregation-based growth mechanisms have been observed experimentally and naturally for a variety of Fe oxide and hydroxide systems [Bailey et al., 1993; Banfield et al., 2000]. The morphology of the aggregate particle may range from spherical to rod-like to a double ellipsoid depending on Fe3+ concentration, pH and temperature [Bailey et al., 1993], but aggregation of nanoparticles into larger spherical particles may act as a growth mechanism.

5. Implications for Secondary Fe Mineralogy at the Martian Surface

[57] One of the most important points to note from the preceding discussion is that all of the acid-generating and mineral formation processes discussed above stem from the instability of Fe2+-bearing sulfate salts, which, as estimates of oxygen content of the early Martian atmosphere suggest, are a natural consequence of evaporation of fluids derived from basaltic weathering under acidic conditions. Because much of the Martian surface has been influenced by acidic conditions, soluble Fe-bearing minerals would be expected to coexist with Mg and Ca sulfate minerals [Tosca and McLennan, 2006; Tosca et al., 2005]. However, it is interesting to note that for areas of the Martian surface where Mg and Ca sulfate minerals have been identified from orbit, the soluble Fe component is conspicuously absent. Instead Fe occurs as a variety of Fe oxide, hydroxide and hydroxysulfate minerals. Thus, in addition to Fe oxidation occurring in response to chemical weathering, the oxidation of Fe2+-bearing sulfate minerals appears to be a mechanism of equal importance in controlling secondary Fe mineralogy at the Martian surface.

[58] The sediments analyzed at Meridiani Planum provide a detailed set of data with which to test hypotheses for Fe oxidation in saline assemblages. The experimental results described in this study provide one example of the general sequence of Fe oxidation products that can be produced as a consequence of the maturation of soluble Fe2+ minerals over geologic time. It is equally important to recall that cycling of Fe2+(aq) and Fe3+(aq) between various minerals and in solution exerts a strong control on pH and is an important source of acidity to the Martian surface in addition to the input and recycling of volcanically derived acidic volatiles.

[59] The number of potential reactions that may arise from mineral transformations taking place during chemical weathering, saline mineral production and aging at the Martian surface may be significant. However, the important relations can be summarized by grouping the reactions into three categories that may occur between chemical weathering, evaporation and “diagenetic” processes taking place at the Martian surface: oxidation, hydrolysis, and dehydration. Oxidation and dehydration reactions are clear, but hydrolysis reactions, as used here, refer to the process of Fe3+ hydrolysis as a function of increasing pH, driving polymerization and depending on chemical species present, the precipitation of “ochreous” phases such as schwertmannite, ferrihydrite, jarosite, goethite and hematite. Figure 18 shows a flowchart that begins with the release of Fe2+ from the chemical weathering of basalt at the Martian surface. The fate of Fe2+ or Fe3+ into various mineral classes and their relationship to one another is also depicted. Many of the relationships that have been used to construct such a diagram come from studies of acid mine drainage environments on Earth [see Bigham and Nordstrom, 2000; Jambor et al., 2000], as well as recent thermodynamic constraints on Fe sulfate mineral stability [Majzlan et al., 2006; Tosca et al., 2007]. One important point to note from Figure 18 is that not all oxidation processes of Fe-bearing saline minerals directly result in the precipitation of oxide and hydroxide phases. The processes are strongly controlled by pH. Recent analyses of Paso Robles class soils at Gusev Crater are an example of where Fe oxidation processes may have occurred under extremely low pH; probably too low to allow the precipitation of phases such as jarosite, schwertmannite, goethite or hematite. Analyses of various spectral data sets collected on Paso Robles-type soils reveal that they are composed mainly of Fe and SO4 [Ming et al., 2006; Yen et al., 2007]. Among the best matches to thermal IR, visible and Mössbauer spectra are minerals such as ferricopiapite and rhomboclase [Lane et al., 2007, 2006]. These minerals constrain the pH to values of below 1 and also suggest that Fe oxidation has occurred and Fe has remained extremely soluble in these phases. As such, Fe3+ salt precipitation in Figure 18 is divided into those minerals precipitating at a pH generally less than ∼1–2 and those that precipitate as pH increases above pH ∼1–2, where Fe3+ hydrolysis becomes increasingly important.

Figure 18.

Flowchart depicting major Fe oxidation and precipitation processes in acidic (pH ≤ 4), SO4-rich systems. The reactions shown are not meant to be exhaustive but rather to show the evolution of soluble Fe-bearing sulfate minerals toward goethite, hematite, and other ochreous phases over time and oxidizing conditions. The reactions illustrated with black arrows are classified as weathering/evaporation reactions; those reactions most likely to occur during chemical weathering of basalt and/or the evaporation of a fluid derived from basaltic weathering. The reactions illustrated with red arrows represent diagenetic reactions, or those reactions occurring in response to processes occurring after deposition. Three major classes of diagenetic reactions among Fe-bearing minerals are used: oxidation, hydrolysis, and dehydration. See text for discussion.

[60] A last point to note from Figure 18 is that goethite and hematite can arise from any combination of oxidation, hydrolysis or dehydration. In reality, kinetic factors often exert the most control on the distribution of such phases, but nevertheless, it is important to note that they are the stable thermodynamic end products in almost all cases, except those of exceedingly low pH.

[61] The high-resolution mineral spectroscopic data returned from Mars Express and the Mars Reconaissance Orbiter missions may shed light on Fe oxidation processes through geologic associations between Fe oxide minerals and sulfate minerals. Indeed, results from the OMEGA spectrometer aboard the Mars Express mission show a correlation between Fe oxide minerals and sulfate deposits [Bibring et al., 2005, 2007; Gendrin et al., 2005]. Already, hematite deposits are observed in association with layered sulfates in areas of Valles Marineris [Bibring et al., 2005, 2007; Gendrin et al., 2005]. More recent analysis of results obtained with the MGS-TES instrument also show the association of two gray hematite units with light-toned sulfate-rich materials at Aureum and Iani Chaos [Glotch and Rogers, 2007]. The geologic relationships between hematite-rich and sulfate-rich units in the latter regions suggest similar aqueous formation environments [Glotch and Rogers, 2007]. Although it is not possible at this time to discern the origin of Fe oxide minerals at these locations to the point where a distinction may be made between a weathering-derived origin or an origin derived from the oxidation of Fe2+-bearing sulfates, these hypotheses may be tested in greater detail with results from ongoing orbital and landed missions as well as future missions to Mars, such as Phoenix and MSL.

[62] Nevertheless, the present evidence for the abundance and mobility of Fe(aq) in the Martian hydrosphere suggests possible widespread importance of Fe-bearing sulfates. Fe-bearing sulfates are sensitive to environmental factors such as O2 content, pH, relative humidity/activity of water and SO4 concentration. Although the number of possible reactions between Fe-bearing sulfates and oxides can be substantial, a better understanding of Fe mineral stability under Martian surface conditions will help constrain mechanisms of formation. Thus, Fe minerals identified in association with other saline precipitates may ultimately be exploited as a unique set of geochemical probes yielding information that few minerals are able to faithfully retain.


[63] This work was supported by a grant from the Mars Fundamental Research Program of NASA (NNX06AB67G) to S.M.M. The authors acknowledge additional support from NASA grants NNX06AB62G and NNG06G130G to M.D.D. This work is also funded in part by the Center for Environmental Molecular Science (CEMS), NSF awards CHE0221934, DMR-045244, and EAR-0510501, and DOE award DE-FG02-03ER46085. Data collection performed at XOR beam line 1-ID at the Advanced Photon Source, Argonne National Laboratory supported by the U.S. Department of Energy, Office of Science/Basic Energy Sciences (DE-AC02-06CH11357).