Journal of Geophysical Research: Planets

Formation of silica by low-temperature acid alteration of Martian rocks: Physical-chemical constraints



[1] Theoretical geochemical modeling has been used to evaluate the formation conditions of amorphous silica during aqueous alteration of typical Martian igneous rocks at 0°C. The models show that some silica can form during low-temperature alteration of mafic to ultramafic rocks over a large range of pH and water/rock ratios. Silica-dominated deposits, like those found at the Columbia Hills in Gusev crater on Mars, could form at solution pH below ∼2 and water/rock ratios of ∼102–104. High-water/rock conditions could represent acid flow through rocks, solution discharge from an acid spring, and/or surface flow of released solutions. Low pH favors dissolution of silicates and saturation of solution with respect to relatively insoluble silica, which then precipitates. Partial evaporation or freezing of released solutions would also cause precipitation of amorphous silica. Modeling shows that Ti oxides are also present in silica-rich deposits. More soluble minerals (e.g., ferric oxides, phyllosilicates) could precipitate downstream from partially neutralized, evaporated, or frozen solutions. Temperatures above ∼0°C are not required to form abundant silica through acid alteration of Martian rocks.

1. Introduction

[2] The presence of iron oxyhydroxides, specific Fe and Mg sulfates, and Mg phyllosilicates in Martian surface materials is consistent with aqueous alteration of primary mafic and ultramafic rocks [e.g., Clark et al., 2005; Bibring et al., 2005; Ming et al., 2006]. The formation of Martian salts and iron oxides from mafic silicates implies a release of SiO2 into aqueous solution followed by precipitation of silica [McLennan, 2003] and/or silica-rich minerals. The presence of silica and/or silica-rich minerals on Mars is consistent with data obtained at Mars Pathfinder [e.g., McLennan, 2003] and Opportunity rover [Glotch et al., 2006] landing sites and with orbital thermal infrared [e.g., Kraft et al., 2007; Bandfield, 2008] and near-infrared [Milliken et al., 2008] spectral observations of surface regions. Recently, silica-rich outcrops and soil (up to ∼98 wt % SiO2) have been found at the Columbia Hills in Gusev crater [Squyres et al., 2008]. These silica-rich samples often have relatively high Ti concentrations. For example, the sample with 98 wt % SiO2 contains 1.5 wt % TiO2 [Squyres et al., 2008]. Thermal infrared spectra of these materials are consistent with opaline silica [Ruff et al., 2007; Squyres et al., 2008]. Although the silica is likely a product of aqueous processes [Squyres et al., 2008], the conditions of rock alteration and silica deposition remain to be constrained. Here we used thermochemical equilibrium and coupled kinetic-thermodynamic models to investigate conditions under which abundant silica may form by low-temperature alteration of typical Martian rocks.

2. Methods

[3] We explored the effects of solution pH, solution/rock ratio (expressed as water/rock, W/R, ratio), and primary rock composition on secondary mineralogy, solution chemistry, and timing of rock alteration at 0°C. These factors were investigated using two numerical approaches to model water-rock interactions. In both models, a series of acid H2SO4-HCl solutions with an S/Cl molal ratio of 5.2 were used, consistent with Martian soil composition [e.g., Clark et al., 1982]. Solutions with an initial pH < 5 were chosen because of chemical and mineralogical signs of low-pH alteration of Martian rocks [e.g., Clark et al., 2005; Ming et al., 2006; Hurowitz et al., 2006] and preferential deposition of silica from low-pH solutions due to its low dissolution rates and low solubility under these conditions [Dove and Rimstidt, 1994; Dove, 1995]. The low temperature was chosen because of relevance to current Martian surface conditions and the inability of metastable amorphous silica to form in high-temperature (>∼150–200°C) solution-rock systems [Dove and Rimstidt, 1994]. Although the focus was on abundant secondary silica, the formation of other minerals was also modeled because of their influence on silica precipitation under some conditions.

[4] Equilibrium mineral assemblages and solution compositions were calculated in the O-H-K-Mg-Ca-Al-C-Si-S-Na-Cl-Fe-Ti solution-rock system open with respect to current Martian atmospheric O2 and CO2. Simplified chemical compositions of the Martian basaltic meteorite Shergotty, Adirondack-type olivine basalt from Gusev crater, and the Martian dunite meteorite Chassigny were used as input compositions to represent a range of Martian rocks with different SiO2 content (Table 1). Throughout the text, the Shergotty composition will be referred to as basalt, the Adirondack-type basalt will be referred to as olivine basalt, and the Chassigny composition will be referred to as dunite. Calculations of thermochemical equilibria between aqueous solutions, solids, and dissolved gases were performed with the GEOCHEQ code [Mironenko et al., 2000] over a range of W/R ratios and solution pH. The code uses thermodynamic data for aqueous solutes [Shock et al., 1989, 1997] and minerals (Table 2). In addition, aqueous hydroxyl complexes of Ti are included from Knauss et al. [2001]. Some models were open with respect to solution pH; these models predict mineral assemblages formed with continuous input of acid solution. Other modeled systems were closed, allowing pH to change along with rock alteration.

Table 1. Compositions of Martian Rocks Used in Modelsa
 Shergottyb (Basalt)Adirondackc (Olivine Basalt)Chassignyd (Dunite)
Table 2. Major Secondary Minerals Formed in Modelsa
MineralFormulaGibbs Free Energy at 0°C (kcal/mol)Referenceb
Amorphous silicaSiO2−202.531
AnalcimeNaAlSi2O6 · H2O−736.751
GypsumCaSO4 · 2H2O−428.434
StelleriteCa2Al4Si14O36 · 14H2O−4,752.95
Ca montmorillonitecCa0.165 Mg0.33 Al1.67Si4O10(OH)2−1,270.96
Mg montmorillonitecMg0.495Al1.67 Si4O10(OH)2−1,267.16
Na montmorillonitecNa0.33Mg0.33 Al1.67 Si4O10(OH)2−1,269.66
K montmorillonitecK0.33Mg0.33Al1.67 Si4O10(OH)2−1,271.76
Ca saponitedCa0.165Mg3Al0.33Si3.67O10(OH)2−1,343.66
Mg saponitedMg3.165Al0.33Si3.67O10(OH)2−1,339.86
Na saponitedNa0.33Mg3Al0.33Si3.67O10(OH)2−1,342.46
K saponitedK0.33Mg3Al0.33Si3.67O10(OH)2−1,344.56

[5] A coupled kinetic-thermodynamic model developed for acid weathering [Zolotov and Mironenko, 2007] was also used to explore silica precipitation. This model considers experimentally determined rates of mineral dissolution, as well as solubility-controlled precipitation in the O-H-C-S-Cl-Mg-Fe-Ca-Na-Al-Si system open with respect to current Martian atmospheric O2 and CO2. The thermodynamic block of this model is based on the GEOCHEQ code and database and also uses the data presented in Table 2. In addition to the data sources for the kinetic parameters described by Zolotov and Mironenko [2007, Table 2], we used kinetic data for diopside from Brantley [2004] and dissolution rates of augite as a proxy for hypersthene. Here we explored the timing of acid weathering of 0.1 mm spherical monomineral grains with the simplified normative mineralogy (wt %) of the Adirondack-type basalt: plagioclase, 36.8; diopside, 13.2; hypersthene, 18.9; olivine, 21.7; magnetite, 5.3 [McSween et al., 2006]. This grain size is similar to that of typical fine sand particles observed by the Mars Exploration Rovers [e.g.,Fergason et al., 2006]. The grains were exposed to different amounts of H2SO4-HCl solutions with different initial pH, and temporal changes in rock mineralogy and solution composition were computed.

3. Results and Discussion

[6] Equilibrium calculations for olivine basalt systems open with respect to acid solution show that amorphous silica forms at W/R < ∼104, over a wide range of acid pH (Figure 1a). Rutile, which represents Ti oxide phases in the model, forms under all conditions in which amorphous silica forms. The very high W/R boundary of silica precipitation implies that acid groundwater systems, acid spring waters, and acid surface streams would be saturated with silica. Evaporation, freezing, or other consumption (e.g., rock hydration) of these solutions would result in silica deposition. Models indicate that less silica precipitates at higher solution pH, corresponding to the increase in silica solubility [Dove, 1995] and the decrease in solubility of secondary silicates. At very low solution pH (<∼1–2) and W/R ratios of ∼102–104, amorphous silica and small amounts of rutile are the only precipitates. At lower W/R ratios, some gypsum can form in addition to amorphous silica and rutile. Modeling suggests that deposits with 60–100 wt % amorphous silica form at solution pH ∼2–3 (Figure 1a). At higher pH, goethite and Al phyllosilicates precipitate in addition to less abundant silica. Although formation of highly soluble salts, such as Mg sulfates and chlorides, was not modeled, they could form at low-W/R conditions upon significant evaporation or freezing [e.g., Zolotov and Mironenko, 2007].

Figure 1.

(a) Precipitation conditions of secondary minerals formed through acid alteration of olivine basalt at 0°C over a range of W/R ratios and fixed pH values. The curves with arrows show conditions of first deposition for a given mineral. Note that at the kaolinite-montmorillonite line, kaolinite stops being formed in favor of montmorillonite. Some rutile forms over the entire range of conditions depicted. The gray curves delineate conditions where silica comprises a given weight percent of the mineral assemblage. Overall, conditions of silica precipitation are similar for the basalt and dunite compositions. (b) Volumes of amorphous SiO2 formed by rock alteration at W/R = 1000. The curves with arrows show conditions where a given mineral starts to precipitate.

[7] Although the conditions of silica precipitation are similar for all modeled rock compositions, the volumes of silica are different (Figure 1b). For pH < ∼3–4, the volume of deposited amorphous silica is limited by protolith bulk SiO2 content. Silica volume increases with increasing SiO2 content of the bulk rock composition (basalt > olivine basalt > dunite). At pH higher than 3–4, alteration of dunite can produce more silica than the olivine basalt. Therefore at the higher pH, the amount of silica in alteration assemblages does not correlate directly with protolith SiO2 content.

[8] Figure 2 shows an example of equilibrium mineral assemblages formed through rock alteration by solutions with a specified initial pH in a closed system. In these models, different amounts of acid solutions interact with a rock (here, olivine basalt) leading to equilibrium mineral assemblages and somewhat neutralized solutions. Calculated equilibrium mineralogy and solution pH depend on the amount of solution in the system (i.e., W/R ratio); lesser amounts of solution at lower W/R ratios result in thorough neutralization. In systems with initial solution pH < ∼2, abundant silica precipitates at W/R ≈ 102–104, corresponding to a final solution with pH < ∼3. These results are similar to the fixed pH open system models discussed above. Silica and rutile are the only modeled precipitates at elevated W/R ratios and lower pH values. However, little or no silica precipitates from the highest-W/R systems, reflecting complete dissolution of silica (and all silicates) in large volumes of solution.

Figure 2.

Equilibrium mineralogy from alteration of 1 kg of olivine basalt in a closed system by different amounts of acid solutions with initial pH (a) 1 and (b) 3 at 0°C. The bold black curve shows changes in pH with W/R ratio. Rutile is present at a nearly constant abundance (∼1 cm3) in all equilibrium assemblages. At high-W/R low-pH conditions, amorphous silica and rutile are the only minerals in equilibrium with acid solution.

[9] When acid solution reacts with a comparable mass of rock, silica is only a minor constituent of the mineral assemblage in equilibrium with neutralized solution (Figure 2a). The minerals consist of primarily clay minerals, goethite, carbonates, and gypsum. If initial solution pH is higher than ∼2, silica is less abundant and there are no conditions where it forms alone (Figure 2b). Goethite and kaolinite may form instead of silica at lower pH. If small amounts of solution react (low W/R), the equilibrium mineral assemblages do not reflect initial solution pH.

[10] Trends in calculated mineralogy with W/R ratios shown in Figure 2 can also be thought of as trends with acid weathering reaction progress in a closed system. High W/R ratios can represent early stages of weathering when pH is still low and only a small portion of rock is reacted. Low W/R ratios can represent advanced degrees of alteration when pH is elevated through neutralization of solution. This can be envisioned as time-independent titration of the solution with rock. From this perspective, amorphous silica and rutile are the first phases that precipitate during rock alteration by solutions with initial pH values below ∼2. Silica remains the most abundant precipitate until solution pH increases to ∼3–5. Decrease in silica abundance with increasing pH can be considered as dissolution of early formed silica and precipitation of secondary silicates. Although halting of acid alteration at early stages can account for precipitation of some silica (e.g., coatings on primary phases), formation of abundant silica would require another pathway, for example, the scenario described in section 5.

[11] The pattern of silica formation is similar for all considered rock types at a given initial pH (Figure 3). Higher pH in initial and partially neutralized fluids leads to smaller amounts of silica for all rock compositions. For initial solution pH < ∼2 and high W/R ratios, reacted rock is mostly dissolved, and the amount of precipitated silica correlates with rock SiO2 content. However, for initial solution pH > ∼2 and high W/R ratios, alteration of dunite and basalt leads to similar amounts of silica deposited at pH > 6. Relatively abundant silica deposited from dunite could be accounted for by lesser production of Al-rich phyllosilicates. In neutralized solutions, independent of initial pH, basalt alteration produces relatively large amounts of silica, and dunite produces slightly more silica than olivine basalt. The dunite produces slightly more silica than the olivine basalt because of the reason stated above and because of the lower SiO2 content in serpentine compared to clay minerals formed in olivine basalt alteration. Alteration of olivine-bearing rocks at neutral and alkaline conditions does not lead to abundant silica.

Figure 3.

Volume of silica formed through acid alteration of Martian rocks in a closed system at 0°C. Initial solution pH was (a) 1 and (b) 3. Gray curves show changes in pH with W/R ratio.

[12] Kinetic modeling of olivine basalt alteration in a closed system also demonstrates that silica-rich deposits can form at pH < ∼2 and high W/R ratios (∼102 to < ∼5 × 103) (Figures 4 and 5) . At these conditions, the majority of primary silicates dissolve, and silica is the only or most abundant precipitate (gypsum forms under some conditions; Ti oxides are expected but not modeled). These calculations demonstrate that thorough dissolution of silicates is the necessary step that precedes deposition of abundant silica (Figure 5a). Lower W/R ratios and/or higher initial pH leads to rapid neutralization through solution-rock interactions, the majority of primary minerals remain unaltered; secondary minerals consist of phyllosilicates, goethite, carbonates, and gypsum; and abundant silica does not form (Figure 4 and Figure 5b). In turn, very high W/R ratios (>∼104) result in complete dissolution of silicates, and SiO2 does not precipitate.

Figure 4.

Modeled volume of amorphous silica formed through acid weathering of 0.1 mm mineral grains that represent the olivine basalt. Lighter curves represent temporal changes in volumes of silica for different W/R ratios. Bold curves show corresponding changes in pH. (a) For initial solution pH = 1, silica dominates the mineral assemblage at W/R ∼102 to < ∼5 × 103. At these W/R ratios, most of the rock's minerals are dissolved when abundant silica precipitates (see Figure 5a). (b) For initial solution pH = 3, silica is not dominant at any considered W/R ratio. The majority of the primary minerals remain unaltered at times of 102–103 years (see Figure 5b).

Figure 5.

Modeled volume of primary and secondary minerals during acid weathering of 0.1 mm mineral grains of olivine basalt at W/R = 100. Bold curves show corresponding changes in pH. (a) For initial solution pH = 1, silica forms first among secondary minerals and dominates the secondary assemblage when most primary minerals dissolve. (b) For initial solution pH = 3, silica is a first-formed secondary phase but is not dominant. Silica coexists with unaltered minerals and several secondary phases deposited from neutralized solutions; pl, bytownite; kaol, kaolinite; goet, goethite; mont, Mg montmorillonite; zeol, the zeolite stellerite.

4. Comparison With Other Studies

[13] Modeled precipitation of silica in acidic low-temperature aqueous environments is consistent with other numerical models, experiments, and field observations. Equilibrium modeling of acid weathering of Kilauea basaltic materials [Schiffman et al., 2006] demonstrated that the first phase to precipitate at low pH was amorphous silica. Kinetic-thermodynamic models of Zolotov and Mironenko [2007] also revealed very early deposition of amorphous silica from acid-basalt systems followed by its dissolution into neutralized solutions. Tosca et al. [2004] and LaClair [2006] reported the formation of pervasive amorphous silica through low-temperature batch acid weathering of a variety of basaltic materials. Morris et al. [1998] observed a SiO2-enriched basaltic tephra from near the summit of Mauna Kea volcano, which had been altered by acid aerosols. Thin coatings of amorphous silica and jarosite on basaltic tephra near Kilauea caldera could also be a result of acid fog-type weathering [Schiffman et al., 2006].

[14] Silica also precipitates from low-temperature acid solutions formed through oxidation of sulfides in their parent rocks or in mine tailings. Acid mine drainages are often in equilibrium with silica [Nordstrom and Alpers, 1999]. Burns [1988] discussed subsurface silica formed below gossans together with hydrated iron oxides and sulfates. Silica rock coatings are reported in an arctic environment in Sweden as a result of pyrite oxidation and acid rock weathering [Thorn et al., 2001]. Precipitated opaline silica is also observed in acid lake/groundwater systems in Australia, where other precipitates are sulfates, halite, and iron oxides [e.g., Benison and Goldstein, 2002]. Large-scale movement of dissolved SiO2 by discharged low-pH solutions could have contributed to the formation of widespread opal-bearing silcretes in Western Australia [McArthur et al., 1991]. Interestingly, a common characteristic of silcretes is the presence of Ti in anatase [e.g., Hutton et al., 1972; Taylor and Eggleton, 2001]. At least in some cases, silcretes may indicate precipitation of silica in an environment where Al is mobilized, such as an acid environment [Taylor and Eggleton, 2001].

5. Formation of Silica-Rich Deposits at Gusev Crater and Elsewhere

[15] The modeling shows that some amorphous silica can form in secondary assemblages during low-temperature alteration of mafic/ultramafic rocks over a range of acid pH and W/R ratios. It follows that in some areas on Mars in which silica phases have been detected, acid-rock alteration could be responsible. For example, the presence of an amorphous, dominantly opaline, silica component [Glotch et al., 2006] together with jarosite [Klingelhöfer et al., 2004] in sulfate-rich rocks at Meridiani Planum is consistent with aqueous deposition of silica at low pH [Zolotov and Mironenko, 2007]. Surfaces rich in high silica phases detected together with sulfates in Hellas Basin [Bandfield, 2008] could result from acid weathering processes. An association of silica phases with jarosite at several locations near Valles Marineris is also consistent with acid-rock interactions during the history of these surface materials [Milliken et al., 2008].

[16] Very silica rich deposits, like those observed near Home Plate in Gusev crater [Squyres et al., 2008], could form at solution pH below ∼2 and W/R ratios of ∼102–104. In fact, the low solubility of silica phases in acids [Dove, 1995] implies the involvement of a large mass of solution to form silica-rich deposits. High-W/R conditions may imply, for example, solution discharge in spring environments and subsequent outflow from a spring/pool, as depicted schematically in Figure 6. Solution flow through near-surface rocks could account for the elevated W/R ratio and cause dissolution of silicates followed by silica precipitation along the solution path. After discharge of silica-saturated solutions, precipitation of more silica could be driven by freezing and/or evaporation, which would be major factors at low Martian surface temperatures and pressures. However, formation of silica-dominated deposits requires that more soluble elements are transported away, such as by solution flow. Deposited silica could originate both from SiO2 released to solution during mineral dissolution below the acid spring and from dissolution of nearby or local rocks/sediments. In each case, dissolution of a silicate mineral precedes deposition, and some aqueous transport of SiO2 is probably unavoidable.

Figure 6.

A diagram showing a possible low-temperature acid spring setting for the formation of silica-rich surface deposits. This environment could provide the high W/R ratios and low-pH conditions to dissolve silicates and Ti-bearing minerals, to release SiO2 and Ti into aqueous solution, and to saturate solutions with silica and Ti oxides. Partial evaporation and/or freezing of released solutions could cause precipitation of amorphous silica and Ti oxide(s), which are significantly less soluble than other rock-forming minerals at low pH. Other secondary minerals could form below the surface by alteration from initial acid flows and farther from the spring after solutions have been partially neutralized, evaporated, or frozen. The vertical and lateral mineralogical profiles could be similar to the sequence modeled with decreasing W/R ratio in Figure 2a and reflect changes in mineral solubility in higher-pH solutions. Layer 1, amorphous silica and some Ti oxides; layer 2, amorphous silica + goethite + gypsum + Ti oxides; layer 3, amorphous silica + goethite + phyllosilicates + gypsum + Ti oxides. The scheme may represent conditions at Home Plate in Gusev crater at the time of silica deposition.

[17] As initial surface acid flows penetrate into the ground, they would alter rocks/sediments and become partially neutralized. Silica would be formed closest to the surface (at lower pH and higher W/R ratios), followed by formation of iron oxides and oxyhydroxides and secondary silicates deeper in the sediments (Figure 6). Eventually, silica deposition from surface acid flows could seal the underlying sediment/rocks and limit neutralization of subsequent acid flows near the spring. Acid solutions would flow farther before becoming neutralized sufficiently to precipitate other minerals. This could lead to the precipitation of iron oxides/oxyhydroxides, phyllosilicates, and salts farther from acid springs (Figure 6). The observation of hematite-rich deposits, with Ca sulfates, below and near the silica-rich deposits close to Home Plate [Schröder et al., 2008] is consistent with this sequence of mineral deposition (see Figures 2 and 5). The proximity of some of the silica-rich materials to soils containing ferric sulfates [e.g., Yen et al., 2008; Squyres et al., 2008] is consistent with low-pH conditions. Therefore, we suggest that silica-rich materials observed nearby Home Plate could have formed from surface acid flows originating at a spring. This is consistent with the Home Plate depression being filled by spring water at the time of silica deposition [Ruff et al., 2007].

[18] Titanium enrichment in some of the silica-rich deposits at Home Plate [Squyres et al., 2008] is also consistent with our modeling results, in which rutile precipitates together with silica. Although Ti is normally considered immobile during weathering, solubility of TiO2 increases at lower pH [Van Baalen, 1993]. It is possible that both SiO2 and TiO2 were dissolved from parent rocks and precipitated together from discharged solutions. Precipitation could have occurred very near the primary materials, as occurs in acid bleaching, or farther from the source. SiO2 and TiO2 dissolved and precipitated in either scenario may be indistinguishable in the Martian deposits with the currently available data. Note also that Ti enrichments at Home Plate may also represent lag deposits of primary minerals that have not been dissolved by acid flows.

[19] Although Martian surface materials could have been affected by acid precipitates (aerosols, rains, and snows) resulting from volcanic emissions [e.g., Settle, 1979] or impact processes [Zolotov and Mironenko, 2007], the high-silica deposits at Home Plate may imply a localized process. Local generation of acid solutions could have occurred through rapid oxidation of H2S emitted from magma [Mironenko and Zolotov, 2007] or by aqueous oxidation of metal sulfides [Burns, 1988; Burns and Fisher, 1990], which have been tentatively detected in some rocks near Home Plate [Morris et al., 2007]. The Zn and Ni in some silica-rich deposits at Home Plate [e.g., Ming et al., 2008; Squyres et al., 2008] could have been released from sulfides during acid weathering [Taylor and Eggleton, 2001]. Elevated fluid temperatures are not necessary for near-surface generation of acids and precipitation of abundant silica. Although a higher-temperature scenario remains a possibility for Home Plate silica-rich deposits, the observed features could result from acid solution–rock interactions at temperatures as low as 0°C.


[20] We thank Douglas Ming and Peter Schiffman for helpful reviews and Everett Shock, Michael Kraft, and Elizabeth Rampe for comments. This work was supported by the ASU NASA Space Grant Program (A.M.) and NASA Mars Fundamental Research Program grants NNX07AQ70G (M.Z.) and NNX06AB25G (T.S.).