Implications for early hydrothermal environments on Mars through the spectral evidence for carbonation and chloritization reactions in the Nili Fossae region

Authors


  • This article was corrected on 16 DEC 2015. See the end of full text for details.

Abstract

[1] Previous identification of serpentine and magnesium carbonate in the eastern Nili Fossae region of Mars indicates hydrothermal alteration of an olivine-rich protolith. Here we characterize Fe/Mg phyllosilicates associated with these units and present spectral evidence for the presence of a talc component, distinguishable from saponite. Locations with magnesium carbonate are exclusively associated with talc-related phyllosilicates. In the westernmost portions of the Nili Fossae region, where a mafic protolith dominates, Fe/Mg phyllosilicates display spectral evidence for a wide degree of chloritization. We propose that Noachian Fe/Mg smectites were uniformly buried by Hesperian lava flows that initiated hydrothermal alteration in the eastern Nili Fossae region. The chloritization of smectites may have produced silica-rich fluids necessary for the serpentinization of olivine; temperature and depth constraints indicated by their distribution also suggest a hydrothermal system was present. The subsequent carbonation of serpentine and/or olivine in eastern Nili Fossae, while requiring an additional CO2 source, provides an explanation for the limited occurrence of serpentine and the colocation of carbonate and talc-bearing material throughout this area. The consequence of the hypothesized carbonation reaction and the presence of serpentine provides geochemical constraints for the proportion of CO2 present in the fluids that interacted with the protolith. If this carbonation reaction was a widespread phenomenon, it may have been an important process in the ancient Martian carbon cycle and could have provided a sink for CO2 in the past.

1 Introduction

[2] Recently, alteration phases including Fe/Mg-phyllosilicates, zeolites, prehnite, chlorite, serpentine, illite (or muscovite), kaolinite, and hydrated silica have been identified on Mars in the Nili Fossae region, Tyrrhena Terra (south of Syrtis Major), and Noachis Terra (west of Hellas Basin) [Mustard et al., 2008; Ehlmann et al., 2009; Fraeman et al., 2009; Ehlmann et al., 2011]. These minerals, identified from high spatial and spectral resolution data provided by Compact Reconnaissance Imaging Spectrometer for Mars (CRISM), are interpreted to be products of low-grade thermal metamorphism or hydrothermal interaction, along with diagenesis, suggesting that past aqueous alteration of the subsurface crust occurred at elevated temperatures [Ehlmann et al., 2009, 2010].

[3] The distinction between low-temperature sedimentary processes (weathering and diagenesis) that occur during and after lithification at the near-surface environment, and low-grade metamorphism is difficult, as some of the same minerals may be generated in both environments. This distinction is particularly difficult to make using orbital detection of these mineral phases, because evaluation of rock textures and identification of the full mineral assemblage is impossible. Additionally, the temperature at which recrystallization or new mineral formation occurs is highly dependent upon the protolith. Thus, the regional context of metamorphism may be difficult to decipher. Despite these limitations, with the assumption of a basaltic protolith for Mars [McSween et al., 2009], metamorphic assemblages can be used to constrain and identify particular pressure and temperature conditions and metamorphic facies [McSween et al., 2011; Ehlmann et al., 2011]. Here we will use mineral assemblages to constrain pressure and temperature conditions experienced in the Nili Fossae region during the early history of Mars.

2 Background

2.1 Previously Identified Metamorphic Assemblages on Mars

[4] Ehlmann et al. [2011] identified several metamorphic mineral assemblages on Mars using Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) spectra. It is worth noting that an “assemblage,” as used in their work, refers to a collection of individual minerals identified in pixels that are in close proximity to each other, not necessarily all in the same CRISM pixel, and is by necessity an incomplete tally of coexiting minerals. This nontraditional definition of an assemblage is justified because visible and near-infrared spectra of minerals do not combine linearly, making spectrally nondominant minerals hard to identify in compositionally mixed pixels. The assemblages identified by Ehlmann et al. [2011], typically associated with host craters, are the following: (Assemblage 1) prehnite-chlorite-silica, (Assemblage 2) analcime-silica-Fe/Mg smectite-chlorite, and (Assemblage 3) chlorite-illite(muscovite). Additionally, (Assemblage 4) serpentine has been found both associated with craters and in bedrock units. These assemblages have been identified in the Nili Fossae region, Tyrrhena Terra, and Noachis Terra. The distribution of alteration minerals in the Nili Fossae region identified by Ehlmann et al. [2009], as well as phases identified in this study, is shown in Figure 1.

Figure 1.

Thermal Emission Imaging System daytime infrared mosaic of the Nili Fossae region, with locations of alteration minerals as identified by Ehlmann et al. [2009] and this study. Colored circles indicate the locations of minerals identified by Ehlmann et al. [2009] (open circles) and the present study (circles with cross hairs). Symbols are placed side-by-side where identifications fall within a single CRISM Full Resolution Targeted (FRT) observation. The orange triangles indicate a detection of Fe/Mg phyllosilicate with a band depth at 2.29 µm (BD2290) of less than 0.03 and are associated with chloritization. Blue triangles indicate a detection of Fe/Mg phyllosilicate (BD2290 > 0.03) associated with carbonation of serpentine). See section 2.2 for a description of this parameter. See section 5.1 and Figure 4 for the discrimination of “probable” versus “possible” talc identification. Blue shaded areas encompass regions where hypothesized carbonation of serpentine occurred. Darker blue area indicates where olivine-rich protolith is present [e.g., Mustard et al., 2007].

2.1.1 Prehnite-Chlorite-Silica and Analcime-Silica-Fe/Mg Smectite-Chlorite Assemblages

[5] Prehnite-bearing rocks (Assemblage 1) are of particular interest; prehnite forms in hydrothermal or metamorphic environments [e.g., Frey and Robinson, 1999] and has a unique OH overtone absorption feature at 1.48 µm [Clark et al., 2007], allowing for its discrimination from chlorite, which shares a ~2.35–2.36 µm absorption feature. The stability of prehnite (along with pumpellyite) defines the onset of sub-greenschist facies metamorphism, which typically occurs between 300°C and 400°C [e.g., Schiffman and Day, 1999]. Pumpellyite is spectrally indistinguishable from an Mg-rich chlorite, making prehnite the key indicator mineral for spectral studies. Analcime-bearing rocks (Assemblage 2) indicate the onset of zeolite facies metamorphism [Arkai et al., 2007], though the presence of the silica phase in this assemblage is not indicative of a particular environment, as it forms in a variety of conditions [McLennan, 2003]. Analcime shares diagnostic absorption features (at 1.42, 1.91, and 2.52 µm) with other zeolites as well as a unique, weaker feature at 1.79 µm. Ehlmann et al. [2011] suggest the most probable formation mechanism for Assemblages 1 and 2 is the interaction of hydrothermal (<400°C) fluids with a basaltic protolith.

2.1.2 Chlorite-Illite Assemblage

[6] Chlorite and illite (Assemblage 3) are typically products of hydrothermal activity or low-grade metamorphism on Earth. During burial diagenesis, dioctahedral smectite can ultimately convert to illite, and trioctohedral smectite converts to chlorite. Mixed-layered clays, such as illite/smectite (I/S) or chlorite/smectite (C/S), have phyllosilicate layers that are mixed at the mineral structural level. The proportion of different phyllosilicate layers in these mixed-layered clays changes with burial depth and thus is a useful parameter for characterizing the environment of formation. Chloritization (the conversion of smectite to chlorite) appears to be a much more common process on Mars than on Earth, where illitization dominates (the conversion of smectite to illite) [Milliken et al., 2010]. This is consistent with the low abundance of K+ on Mars required for illitization [Milliken et al., 2010].

2.1.3 Serpentine and Carbonate Assemblages

[7] The serpentine-bearing rocks (Assemblage 4) in the Nili Fossae region are indicative of hydrothermal alteration of olivine-rich rocks. Ehlmann et al. [2009] identified serpentine-bearing rocks in association with an olivine-rich draping unit overlying Fe/Mg-smectite, a stratigraphic relationship found throughout Nili Fossae [Mangold et al., 2007]. The spectral signature of magnesium carbonate, first identified by Ehlmann et al. [2008], is also associated with the underlying Fe/Mg-smectite and has been hypothesized to be a weathering product of either the olivine or serpentine in this region [Ehlmann et al., 2009, 2010; Brown et al., 2010]. The abundant and compositionally varying olivine in the Nili Fossae region [Hoefen et al., 2003; Mustard et al., 2005; Hamilton and Christensen, 2005; Koeppen and Hamilton, 2008] particularly dominates in the northeast portion of the area shown in Figure 1. The predominance of this mineral suggests an olivine-rich protolith is appropriate for this region of Nili Fossae.

[8] The following two-step reaction series describes the process of serpentine and talc/carbonate formation via the hydrothermal alteration of an Mg-rich olivine protolith [e.g., Brown et al., 2010]:

Step 1:

Serpentinization

display math(1)
Alt. Step 1a:

Serpentinization

display math(1a)
Step 2:

Carbonation

display math(2)

[9] Step 1 requires the presence of aqueous silica available in significant amounts, dissolved from mafic rocks as predicted by McLennan [2003]. The “alternate” step 1 results in the production of magnetite, though petrographic studies of hydrothermally altered peridotites indicate significant magnetite formation may not begin until more than 60% of the primary rock is serpentinized [Bach et al., 2004]. Furthermore, although magnetite has a distinctively low albedo and small abundances would certainly affect the spectral characteristics of magnetite-bearing material, the mineral has no characteristic absorption features that significantly deviate from the apparent continuum in the visible and near-infrared wavelength ranges, and would difficult to identify if present. Thus, the lack of magnetite detection is not indicative of the absence of this reaction. Equation ((1a)) also provides a condition where methane may be produced in the presence of CO2; Brown et al. [2010] and Ehlmann et al. [2010] suggested serpentinization may be one potential source of telescopic observations of methane [Mumma et al., 2009] that have been contested [Zahnle et al., 2011]. If this region were the source of methane on Mars, it would require currently active serpentinization in the subsurface [Brown et al., 2010]. Serpentine is known to be very reactive in the presence of CO2, and carbonation of serpentinite may represent a significant but neglected sink in Earth's global carbon cycle [Klein and Garrido, 2011]. Thus, the identification of talc as the phyllosilicate-bearing unit in the Nili Fossae region would be relevant to understanding carbonate formation mechanisms on Mars and potential atmospheric sources and sinks for the Martian atmospheric CO2.

2.2 Near-Infrared Identification of Fe/Mg Phyllosilicates

[10] Fe/Mg phyllosilicate can be identified through overtones and combinations of fundamental vibrational absorption features from ~1 to 2.6 µm using the CRISM data set. Phyllosilicate minerals have OH stretching vibrational absorption features at ~1.4 µm while combination tones of H2O have absorptions near 1.9 and 1.4 µm [e.g., Bishop et al., 1994]. The OH in the phyllosilicate mineral structure is bound to an octahedrally coordinated cation whose composition (typically Fe2+, Fe3+ Al3+, and Mg2+) will subtly shift the precise band center of the 1.4 µm feature [Bishop et al., 2002a, 2002b]. Variations in the shape and wavelength of the ~2.2–2.3 µm OH stretching and bending overtone absorption features are also due to the specific cation(s) bonded to the hydroxyl in the octahedral layer. By itself, the Al-OH bond typically gives rise to an absorption feature around 2.2 µm, whereas varying degrees of Fe and Mg exchange cause the absorption feature to shift to around 2.3 µm [e.g., Clark et al., 1990; Bishop et al., 2002b]. Specifically, the Fe-OH absorption feature falls near 2.28 µm whereas the Mg-OH feature falls near 2.31 µm [e.g., Clark et al., 1990]. For a detailed characterization and description of the majority of the spectral end-members present in the Nili Fossae region, see Ehlmann et al. [2008, 2009], Mustard et al. [2007], Ehlmann et al. [2011], and Brown et al. [2010].

3 Data and Methods

3.1 CRISM Data Processing

[11] The distribution and composition of phyllosilicate minerals on Mars can be mapped and measured using CRISM data, which measures reflected sunlight in the visible and near-infrared portions of the spectrum (0.362–3.92 µm) [Murchie et al., 2007a]. CRISM can acquire data in a targeted mode with a high spatial resolution of 18 or 36 m/pixel in 544 bands, or in one of two survey modes at 100 or 200 m/pixel in 72 bands. CRISM data were processed using the CRISM Analysis Toolkit interface (http://pds-geosciences.wustl.edu/missions/mro/crism.htm#Tools). Data were converted to I/F (a ratio of measured spectral radiance to the incoming solar flux) using the procedures described by Murchie et al. [2007a, 2007b]. The I/F values were divided by the cosine of the incidence angle as a simple photometric correction to provide an approximation of reflectance. The gas bands in the photometrically corrected values were removed, following the volcano scan method, scaling and dividing an atmospheric transmission spectrum derived from a ratio of spectra from the top and base of Olympus Mons to the depth of the atmospheric 2 µm CO2 feature in the target spectrum [e.g., Mustard et al., 2005]. A spectral and spatial noise removal routine [after Parente, 2008] was also applied to the data. Equations for established spectral parameters from Pelkey et al. [2007] were applied to the corrected CRISM images to map different spectral compositions. The BD2290 (band depth at 2.29 µm) and D2300 (drop at 2.3 µm) spectral parameters were used to map Fe/Mg smectites. As described in Ehlmann et al. [2009], where possible a pixel region of spectrally bland material was used to ratio against a pixel region of spectral interest in the same column of the unprojected CRISM image, to remove any residual atmospheric or instrument-related spectral features. If no spectrally bland material was present within the columns of interest, larger regions were chosen that were not necessarily from the same columns.

3.2 Excavation Depth Calculations

[12] Metamorphism through burial diagenesis commonly occurs in the formation of illite and chlorite phases (see section 2.1.2). The minimum depth of burial may be estimated utilizing equations relating crater size to the depth of excavation. Using relationships between a crater's final diameter (D) and the excavation depth (DE), the depth of excavation may be calculated for different displacement mechanisms in cratering mechanics. There are several models that relate the size of the transient crater cavity (Dt) to D on Mars; but for reasons outlined in Tornabene et al. [2008], the following equation from Grieve [1981] and Melosh [1982] can be used to provide a conservative lower limit of Dt:

display math(3)

[13] The depth of excavation of phyllosilicates exposed through crater ejecta (de) is a function of Dt and can be approximated from Melosh [1989]:

display math(4)

[14] The central peak of a complex crater exposes the deepest-seated materials sampled by a crater. The following relationship between the stratigraphic uplift (SU) in a central peak is based on data from terrestrial impact structures [Grieve and Pilkington, 1996]:

display math(5)

[15] Finally, a rough estimate of rim uplift (hR) can be approximated for complex craters by the following relationship [Garvin et al., 2003]:

display math(6)

though problems including stratigraphic overturn and subsequent erosion arise when estimating this parameter. The resulting excavation depths calculated from the above approximations are presented, and their implications are discussed in section 5.2.

4 Results

4.1 Newly Observed Locations of Metamorphic Alteration Minerals in Nili Fossae

[16] Using the criteria described by Ehlmann et al. [2009] for the spectral identification of metamorphic alteration phases, we have identified several new occurrences of such phases in the Nili Fossae region (Figure 1, colored circles that include a cross hair pattern). None of the minerals in these newly identified locations are inconsistent with the mineral assemblages previously described by Ehlmann et al. [2011]. Ehlmann et al. [2011] found that prehnite and chlorite are always identified together in mixed spectra. However, we have identified at least one location where prehnite is the dominant phase (Figure 2). This is particularly noticeable by the lack of a 1.41 µm band and the preservation of subtle prehnite absorption features longward of 2.4 µm that are not present in chlorite.

Figure 2.

Newly identified prehnite-bearing spectra (without mixing of a chlorite phase). (top) Ratioed CRISM spectra with black vertical lines at 1.48, 1.91, 2.01, 2.22, 2.28, and 2.35 µm indicating the presence of prehnite. Gray vertical lines at 1.41 and 2.24 µm indicative of chlorite and do not appear coincident with CRISM spectra, particularly example b. (bottom) Library spectrum of prehnite (GDS613.a), chlorite (chlorite_smr13.4984), clinochlore Fe:Mg = 0.57 (clinochlore_gds158.5327), and clinochlore Fe:Mg = 1.5 (clinochlore_gds157.5542) [Clark et al., 2007]; pumpellyite Ca-rich (ZE-EAC-001-c1ze01) and pumpellyite Mg-rich (ZE-EAC-002-c1ze02) [Reflectance Experiment Laboratory (RELAB)].

4.2 Mapping of Fe/Mg Phyllosilicates

[17] A map of BD2290 values from multispectral ~231 m/pixel CRISM tiled data as compared to a map of D2300 values, shows that two Fe/Mg phyllosilicate populations occupy geographically distinct regions surrounding Nili Fossae (Figure 3). Similarly, in the CRISM targeted data, the BD2290 > 0.03 Fe/Mg phyllosilicates (blue triangles) are found in the olivine-rich eastern portion of the region (blue shaded areas), whereas the BD2290 < 0.03 Fe/Mg phyllosilicates (orange triangles) are found mostly, but not exclusively, in the western portion of the region (Figure 1). It should be noted that the BD2290 and D2300 parameters are also sensitive carbonate, so care was taken to identify spectra without a 2.5 µm feature in the targeted data mapping of phyllosilicates in Figure 1, and for identifying Fe/Mg phyllosilicates throughout the region. Wherever carbonate has been identified (blue circles), only BD2290 > 0.03 Fe/Mg phyllosilicates are found. BD2290 < 0.03 Fe/Mg phyllosilicates are dominantly associated with occurrences of prehnite/chlorite phases (orange circles). There are some locations that possess both low and high BD2290 Fe/Mg phyllosilicates. The geochemical significance of the geographic associations is discussed in section 5.5.

Figure 3.

THEMIS daytime infrared mosaic of the Nili Fossae region with CRISM multispectral mapping of the (top) D2300 and (bottom) BD2290 parameters. White circles indicate where outcrops with relatively high D2300 and low BD2290 values are present. Blue arrows indicate where both parameters are relatively high. Images are centered at ~73°E, 19.5°N.

5 Discussion

[18] The olivine-rich draping unit represents a protolith from which the serpentine likely formed in the Nili Fossae region [Ehlmann et al., 2010]. The identifications of 2.31 and 2.51 µm absorption features in this region have been uniquely attributed to magnesium carbonate, which is likely a product of partial alteration of the olivine-rich unit [Ehlmann et al., 2008]. Mangold et al. [2007] and Ehlmann et al. [2009] noted that where the olivine-rich unit exhibits absorptions at 1.91 and 2.32 µm, due to H2O and Mg-OH, respectively, the 2.5 µm carbonate absorption is weak or absent. Ehlmann et al. [2008] suggested impact or volcanic heating of the olivine-bearing material led to hydrothermal alteration along the contact with the underlying water-bearing Fe/Mg-smectite unit and formed the carbonate-bearing unit. Brown et al. [2010] proposed a different spectral interpretation, suggesting that the Fe/Mg-smectite phase could actually be talc based on their assertion that the two phases are spectrally indistinguishable at the ~2.3 µm region, the phase may be hidden with the presence of carbonate, and that talc is the more likely phase given the context of an olivine-rich protolith associated with magnesium carbonate. The consequence of this new spectral interpretation was a modified hydrothermal hypothesis, in which the phyllosilicate was formed at the same time as the overlying carbonate-bearing unit by a single hydrothermal event [Brown et al., 2010]. If correct, this hypothesis suggests that stratigraphic relation of the phyllosilicate and carbonate-bearing units are simply the result of different temperatures regimes within the zone of hydrothermal alteration [Brown et al., 2010]. Upon closer inspection, we maintain that talc is in fact spectrally distinguishable from saponite, and show that the ~2.3 µm absorption features of Fe/Mg phyllosilicates in eastern Nili Fossae are more consistent with talc than saponite.

5.1 Spectral Analysis of Fe/Mg Phyllosilicates in Nili Fossae

[19] Figure 4 shows laboratory spectra of saponite, talc, and other Fe/Mg-phyllosilicates. Laboratory spectra are typically acquired from pure phases with ideal lighting conditions, high signal to noise, and limited residual artifacts in the spectra. One of the major challenges working with remotely derived spectra in a natural setting is that surfaces are likely intimate mixtures of materials rather than pure phases, have variable lighting conditions, lower signal to noise than in the laboratory setting, residual atmospheric artifacts, and may have artificially injected slopes from spectral ratioing to bland areas within scene. Despite these complications, the spectrally dominant phase of a mixture can often be identified using the band centers of diagnostic absorption features of the phase. The diagnostic band centers for both talc and saponite are 2.31 and 2.39 µm, and thus, the band centers of these features cannot be used to distinguish between the two phases. Although laboratory spectra of talc displays a narrow 1.4 µm band due to the Mg-OH bond and lacks the 1.9 µm hydration band with no interlayer H2O in the mineral structure (Figure 4), these absorptions are not particularly definitive if the talc is mixed with even small proportions of a hydrated phase. The 1.4 µm sharp feature should be present but may be somewhat asymmetric or overprinted by a broader 1.4 µm hydration feature from a mixed component. The saponite 1.4 µm feature also has such characteristics (see Figure 4); thus, both the 1.4 and 1.9 µm features are not diagnostic.

Figure 4.

Comparison of Fe/Mg phyllosilicate versus talc laboratory spectra. (left) Full laboratory spectral range shown. Both saponite and talc share similar absorption features at 2.31 and 2.39 µm [e.g., Brown et al., 2010]. (right) Same spectra shown at 2.1 to 2.5 µm for clarity. Local continuum between ~2.2 and 2.42 µm is shown by black diagonal lines. A larger gap between the local continuum and the septum at 2.35 µm (dark gray vertical line) is present in the talc spectrum. Saponite, talc, and actinolite share an absorption minimum at ~2.31 µm, while phologopite occurs at longer wavelengths (~2.32), a difference resolvable at CRISM spectral resolution. Library spectra are from Clark et al. [2007].

[20] Although the ~2.3 µm band centers are similar for talc and saponite, the gap between the local continuum and the septum at 2.35 µm (between the two absorption bands) is greater in the talc spectrum than the saponite spectrum. Laboratory spectra of a full suite of Fe/Mg phyllosilicates (shown) reveal that talc is unique within this group in having a septum with this characteristic gap and an absorption feature at 2.31 µm. Phlogopite shares the septum gap feature; however, the ~2.3 µm feature is shifted to significantly longer wavelengths (2.33 µm), a difference resolvable with CRISM spectral resolution. Some Fe/Mg phyllosilicates (e.g., chlorite) do have a reflectance at 2.35 µm that falls below the local continuum but lack the septum that is present in the talc spectrum. Spectra of the amphibole actinolite do display a septum at 2.35 µm that falls below the local continuum and the 2.31 µm absorption feature, but for reasons described below, this phase is thought less likely to be present in this region of Nili Fossae. The magnitude of the continuum-septum gap is given by the relative band depth between the local continuum defined at 2.15 and 2.42 µm and the local maximum of the septum at 2.35 µm. To distinguish between talc and other phyllosilicate phases that also display reflectances at 2.35 µm below the local continuum (e.g., chlorite), a second spectral parameter is necessary. These phases have an additional absorption feature at 2.21 µm, which talc lacks. Thus, a plot of relative band depth at 2.35 µm versus relative band depth at 2.21 µm (Figure 5) allows for the separation of talc, prehnite/chlorite, and other Fe/Mg phyllosilicates.

Figure 5.

Relative band depth at 2.35 µm verses 2.21 µm for Fe/Mg phyllosilicates in the Nili Fossae (NF) region (symbol outlined in black) as compared to library spectra of actinolite, phologopite, and talc (blue region); prehnite and chlorite (red region); Fe/Mg smectites (green region); and Al-OH minerals (purple region). CRISM spectral contrast has been multiplied by a factor of 6 for comparison with library spectra fields. Bubble diameter is proportional to 2.29 µm band depth. As shown in Figure 4, the band depth at 2.35 µm can be used to differentiate between saponite and talc. NF phyllosilicates with 2.29 µm band depth (BD2290) values of greater than 0.03 do not trend toward library chlorite phases (indicating chloritization), but rather trend toward the composition of library talc. The path of these phyllosilicates originates partway up the chloritization process and is distinct from phyllosilicates (BD2290 < 0.03) that could originate from the saponite end-member. NF phyllosilicates that fall within the blue “Talc” field are considered a “probable” talc identification in Figure 1, while those blue bubbles that fall along the trend 2 line are considered “possible.” Library spectra from Clark et al. [2007] and RELAB.

[21] CRISM data plotted in Figure 5 were derived from CRISM ratioed I/F data, where the spectral contrast has been adjusted (by a factor of 6×) to be comparable to laboratory spectra. This factor is applied equally to all wavelengths and is consistent between samples, so the relative variability between data points and calculated band depths is maintained. If the spectral contrast of the CRISM data were not altered, the spectral parameter values would fall much closer to the origin, making parameter comparisons to laboratory data much more difficult. The population of CRISM spectra from Nili Fossae falls along two distinct trendlines when plotted in this new spectral parameter space. Trendline 1 is composed of Fe/Mg phyllosilicates with BD2290 < 0.03 (orange circles) and Nili Fossae spectra dominated by prehnite/chlorite features (red circles). As shown in Figure 5, Trendline 1 falls roughly along a vector between laboratory spectra of Fe/Mg smectite and prehnite/chlorite. Points along this trendline represent increasing abundances of prehnite/chlorite toward the upper right portion of the plot (in the direction of the points representing laboratory spectra of prehnite/chlorite). Trendline 2 is composed of Fe/Mg phyllosilicates with BD2290 > 0.03 (blue triangles) and is approximately orthogonal to the Trendline 1 in the parameter space of Figure 5. Interestingly, the laboratory spectra of talc (as well as actinolite and phologopite) fall along the extension of Trendline 2. The spectral parameters used in Figure 5 were chosen to separate the talc from Fe/Mg smectite and chlorite, so points along Trendline 2 represent increasing abundances of talc toward the upper left portion of the plot (in the direction of the points representing the laboratory spectra of talc).

[22] As we previously noted, talc is spectrally similar to actinolite, though considering the olivine-rich protolith and the context of associated assemblages with the Fe/Mg phyllosilicate phase (Figure 1), we argue that talc is the most likely spectral candidate. Actinolite is typically formed by metamorphism of basaltic rocks. The spectra of the Fe/Mg phyllosilicate material have a 1 µm absorption consistent with an olivine component [Brown et al., 2010], suggesting the phyllosilicate may be mixed with remaining olivine-rich protolith (residual reactant in the carbonation process). If the spectra were actually associated with actinolite, there would be other (unobserved) mineral end-members expected, such as pyroxenes. With increasing metamorphism, the onset of actinolite occurs at the transition from the prehnite-pumpellyite facies to the greenschist facies. While this trend may be expected transitioning eastward of the prehnite-pumpellyite assemblages observed in the Nili Fossae region, the assemblages associated with actinolite typically would include other minerals, such as epidote, which have significant spectral features in the near-infrared have not yet been observed in this assemblage.

[23] Data from a mathematical linear mixture of laboratory saponite and chlorite would fall within the blue “Talc” field in Figure 5. However, there are several lines of reasoning that suggest the Fe/Mg phyllosilicates (Figure 5, blue bubbles) are not just a mixture of these two phases. First, if they were a simple linear or intimate mixture, we would expect these data to trend between the green “Fe/Mg-smectite” field and the red “Prehnite/Chlorite” field. The trend for this phase (Trendline 2) is conversely orthogonal to a vector between these fields. The BD2290 values (Figure 5, bubble size) also indicate that these phyllosilicates are unlikely to be mixtures with Al-OH members such as montmorillonite or kaolinite, as the BD2290 values for these samples are very small.

[24] It is useful to consider the paleoenvironmental conditions for the Nili Fossae region that would be consistent with the trendlines in Figure 5. A possible interpretation of the points along Trendline 2 is that they represent varying degrees of chloritization of Fe/Mg smectite. Points toward the upper right portion of the plot would represent greater degrees of chloritization and perhaps correlated with greater burial depth. The points along Trendline 2, which progress toward pure talc, may represent varying degrees of carbonation (equation (2)()). Interestingly, Trendline 2 does not appear to originate with pure Fe/Mg smectite. Rather, it originates in a region partway along Trendline 1. Chloritization normally occurs when parent rock is buried, so the fact that Trendline 2 originates partway along the chloritization trendline suggests that carbonation occurred after burial to a particular depth.

5.2 Depth Dependence and Implications of Fe/Mg Phyllosilicates in Nili Fossae

[25] Additional insight can be gained by considering the geographic and depth distribution within the Nili Fossae region of samples along the two trendlines in Figure 5. The phyllosilicates that fall along Trendline 1 are dominantly associated with the western portion of the Nili Fossae region (prehnite/chlorite and Fe/Mg phyllosilicates with BD2290 < 0.03, Figure 1). Because the points representing Fe/Mg phyllosilicates and prehnite/chlorite are distributed along Trendline 1, rather than concentrated at a single point, varying degrees of chloritization (burial diagenesis) are inferred to have occurred. Calculating the depth of excavation for exposed phyllosilicates can provide further evidence for the chloritization process. Throughout this region, Fe/Mg phyllosilicates are excavated from some initial depth (DE) below the currently exposed surface via crater impact processes and are located in crater ejecta, rims, and central peaks.

[26] Results from equations (3), (5), (6)(3)–(6) (section 3.2) are expressed as a general depth of excavation (DE) and have been calculated for all occurrences of phyllosilicates associated with Trendline 2 (Figure 5) that have a discernable contextual relationship to either a crater ejecta blanket, rim, or central peak. To illustrate the depth relationship of different phyllosilicate compositions, of DE has been plotted as a function of the spectral parameters presented in Figure 5 (Figure 6). A trend of inferred chloritization with increasing DE is observed; those phases located at the origin of Trendline 1 are excavated from shallower depths than those phases located at the end of Trendline 1 (Figure 6, gray arrow). This relationship suggests that burial diagenesis (chloritization) is in fact responsible for the compositional distribution of Fe/Mg-phyllosilicates in the western portion of the Nili Fossae region.

Figure 6.

Spectral evidence and depth relationship for Fe/Mg phyllosilicates indicative of the chloritization process. (top) CRISM ratioed reflectance spectra of several examples of Fe/Mg phyllosilicates (BD2290 < 0.03) and prehnite/chlorite from several different CRISM scenes. Spectra exhibit shift in major absorption feature from ~2.31 to 2.35 µm (position indicated by vertical lines) and appearance of a ~2.2 µm shoulder (see Figure 2 for comparison of laboratory prehnite/chlorite spectra); spectral variability is indicative of points plotted along Trendline 1 in Figure 5. (bottom) Depth of excavation (DE) versus relative band depth at 2.35 and 2.21 µm for Fe/Mg phyllosilicates in the Nili Fossae region. Phyllosilicates associated with Trendline 2 (talc-related phyllosilicates) are not shown here, as the majority of them are unassociated with apparent craters. Points are projected onto the X-Y axis (spectral parameters) via gray vertical lines. The dark gray arrow indicates Trendline 1 (from Figure 5) or the inferred trend of chloritization.

5.3 Spatial Distribution and Implications of Fe/Mg Phyllosilicates in Nili Fossae

[27] By contrast to the Fe/Mg phyllosilicates in the western portion of the Nili Fossae region, those in the eastern portion of the region do not have such marked spectral variability and typically are unassociated with craters. The fact that the low-carbonation end of Trendline 2 emanates from a localized area along Trendline 1 suggests that the ~160,000 km2 area in eastern Nili Fossae, where the BD2290 > 0.03 Fe/Mg phyllosilicates are concentrated, experienced a relatively uniform depth of burial. This is also the portion of Nili Fossae where the olivine-rich cap unit and underlying magnesite are found, as well as the two documented occurrences of serpentine in the region. The local geographic association of olivine, serpentine, magnesite, and potentially talc provides the necessary solid reactants and products for the serpentinization and carbonation processes described in equations ((1)) and (2). An example of the stratigraphic relationship of Fe/Mg-phyllosilicates from eastern Nili Fossae is shown in Figure 7. Here the talc-bearing phyllosilicates (red) overlay Fe/Mg-phyllosilicates that follow the chloritization trend (see Figure 7d). Chloritization cannot be observed in a single exposed outcrop (down section), as there needs to be kilometers of burial to significantly change the spectra. A second example of a talc-carbonate assemblage is shown in Figure 8. The decrease in carbonate signature and increase in talc signature down section is consistent with the hypothesized carbonation reaction. The green spectrum (Figure 8, right) reveals a 1.4 µm band with a sharp drop that may indicate mixing of the talc 1.4 µm band with a broader feature from an unknown hydrated mineral. Furthermore, the strength of the 2.39 µm feature (due to talc) increases while the strength of the 2.5 µm feature (due to carbonate) decreases down section.

Figure 7.

Example stratigraphic relationship of Fe/Mg phyllosilicate material exposed in an escarpment in the eastern portion of the Nili Fossae region. (a) CRISM scene FRT00009D44 spectra of talc-related phyllosilicate (BD2210 < 0; BD2355 > 0.01) indicated in red and chlorite-related phyllosilicate (BD2210 > 0; BD2355 > 0.01) indicated in green, overlain on High Resolution Imaging Science Experiment (HiRISE) scene ESP_016193_2005 and draped on Mars Orbiter Laser Altimeter (MOLA) topography for stratigraphic context; ~5× vertical exaggeration. (b) Same scene draped over MOLA topography where red = OLINDEX2 (olivine), green = LCPINDEX (low-calcium pyroxene), and blue = HCPINDEX (high-calcium pyroxene) overlain on IRA (reflectance at 1.3 µm); see Pelkey et al. [2007] for a description of these parameters. (c) Approximate contact drawn between the upper talc-related unit (red, layered) and chlorite-related unit (green, brecciated) based on morphologic differences in the same HiRISE scene. (d) End-member Fe/Mg phyllosilicate spectra from the two stratigraphic units with comparison laboratory spectra. (black) Laboratory spectra of talc and saponite [Clark et al., 2007] for comparison with the upper talc-related unit. Gray vertical lines and purple continuum lines across 2.1–2.45 µm have been added for comparison. (red) Talc-related end-member spectrum averaged from 40 pixel for the numerator s1:[416,84] and a spectrally bland denominator s2:[416,246] are plotted with the ratioed spectrum (s1/s2). Dominant absorptions are present at 1.40, 1.92, 2.31, and 2.39 µm. Note the drop in reflectance at 2.35 µm from the purple continuum line. (green) Chlorite-associated spectrum averaged from 33 pixels for the numerator s3:[262,154] and a spectrally bland denominator s4:[262,29] are plotted with the ratioed spectrum (s3/s4). Dominant absorptions are present at 1.41, 1.92, and 2.34 µm.

Figure 8.

Example stratigraphic relationship of talc-related phyllosilicates, carbonate, and olivine-rich material in the eastern portion of Nili Fossae. (left) CRISM scene FRT00003E12 RGB, where red = OLINDEX (olivine), green = D2300 (phyllosilicate), blue = BDCARB (carbonate) [Pelkey et al., 2007], overlain on Context Camera (CTX) image P03_002176_2024_XI_22N283W. Bluish-purple color (and arrow) indicative of olivine-bearing carbonate material; light yellow color (and arrow) indicative of the presence of all three phases; bright green material (eroding from walls) indicative of talc-bearing material. (right) CRISM unratioed and ratioed spectra for each of the three phases (indicated by color). Unit underlying Hesperian cap (purple spectrum) exhibits a strong bowl-shaped 1 µm feature, indicative of olivine, and absorptions at ~2.3 and 2.5 µm indicative of carbonate. A hydrated phase is also present (1.9 µm absorption). Below the olivine-carbonate unit, a talc-bearing phase (green spectrum) is identified by the appearance of a relatively strong 1.39 µm absorption feature and an increase in the band depth of the 2.39 µm feature. There is no remnant of the carbonate 2.5 µm absorption feature (as is weakened but present in the yellow spectrum, which is taken from between the two units), although a hydrated phase not attributed to talc (1.9 µm) remains. This hydrated phase may also be responsible for the broadened absorption at 1.39 µm, which appears asymmetric longward of the absorption minimum.

[28] Evidence for the serpentinization and carbonation reactions appears throughout this ~160,000 km2 area, though neither could have run to completion across the entire region because both olivine and serpentine are still observed. Complete carbonation of serpentine is achieved with low fluid to rock ratios at temperatures ≤ 200°C, but if temperatures remain this low, extended fluid interaction at low temperatures can result in the decarbonation and silicification of the altered rock [Klein and Garrido, 2011]. This scenario would not be detectable in the wavelength range used in this study if the replacement is crystalline quartz, as it has no significant spectral features. Amorphous silica (e.g., opal) is detectable with CRISM; however, none is observed within the region shaded in blue (Figure 1). If temperatures increase above 200°C, even high levels of fluid interaction will not cause the dissolution of carbonate [Klein and Garrido, 2011]. Because the magnesium carbonate in Nili Fossae persists, either the temperature of the olivine-rich protolith remained low but fluid interactions were limited or temperatures above 200°C were reached.

[29] The carbonate-bearing layer in the Nili Fossae region is thought to be mixed with other phases, such as olivine and nontronite, and is matched well with laboratory mixtures containing 80% magnesite [e.g., Ehlmann et al., 2008]. In order to assess the upper limits of the effects of the carbonation of serpentine in this region, we can calculate the total amount of CO2 and the associated atmospheric pressure trapped in the magnesite in this layer. As an upper limit, we assume the magnesite-bearing layer is uniformly thick across the entire ~160,000 km2 area where magnesite or talc-related phyllosilicate is present (Figure 1, blue shaded areas), with an abundance of 80% magnesite. The upper limit of the thickness of the carbonate-bearing layer can be approximated using the thickness of the olivine-bearing layer (meters to tens of meters thick) [Mustard et al., 2007]. Although carbonates typically occur in only a small portion of each CRISM targeted image, the rest of the carbonate and/or talc-bearing material is assumed to be buried (likely under Hesperian lava flows from Syrtis or the olivine-rich unit) or otherwise obscured. If this thickness is assumed to be 10 m for the entire ~160,000 km2 area, using the density of magnesite, we can derive the total mass of magnesite from the calculated volume of the unit. That mass is stoichiometrically equivalent to the grams of CO2 trapped assuming equation (2) is responsible for the presence of carbonate. Assuming a constant atmospheric volume to the present-day Mars, we can use the proportionality of the current mass of CO2 in the Mars atmosphere to the current atmospheric pressure to solve for an atmospheric pressure with the additional mass of CO2. If we assume this reaction took place across the entire ~160,000 km2 area, the amount of CO2 sequestered would be equivalent to ~6000 Pa (~0.06 atm) of CO2 locked in the rock. More conservatively, if this reaction only took place over the region, the olivine-rich protolith is present (Figure 1, dark blue shading), ~0.03 atm of CO2 would be sequestered. This is ~5–10 times the current Martian atmospheric pressure and is consistent with atmospheric CO2 partial pressures (~0.01–2 bars) required for hydrogeochemical modeling of the observed mineral assemblages [Van Berk and Fu, 2011]. Though the upper limit of the Nili Fossae region carbonates certainly does not account for all of the hypothesized sedimentary carbonates from ancient Mars [e.g., Kasting, 1991], it may reveal an important geologic process by which CO2 sequestration may have occurred. Furthermore, the findings here suggest the mineralogic evidence for this reaction may typically be buried in the subsurface [e.g., Michalski and Niles, 2010].

5.4 Similar Assemblages Outside of the Nili Fossae Region

[30] Evidence for subsurface carbonate is not limited to the Nili Fossae region. The presence of a Fo60-80 olivine-carbonate assemblage in the Comanche Spur by the Mars Exploration Rover at Gusev Crater is hypothesized to have formed during the Noachian under neutral pH hydrothermal conditions [Morris et al., 2010]. The assemblage has been compared to carbonates in the Nili Fossae region, and Morris et al. [2010] have indicated that the relative remoteness of carbonate outcrops may imply that such environments had multiple occurrences in the Noachian terrain across the globe. Although no significant phyllosilicate component has been observed in the olivine-carbonate assemblages, orbital observations of phyllosilicates in an outcrop near the Comanche Spur appear Mg-rich [Carter and Poulet, 2012]. Given the similarity to Nili Fossae assemblages, the Mg-rich phyllosilicates may also have a talc component. CRISM spectra of the Colombia Hills carbonate have an additional 1.9 µm feature, similar to those in Nili Fossae, indicating the presence of chemically bound water or mixture with a hydrated phase [Carter and Poulet, 2012].

[31] Michalski and Niles [2010] have identified carbonate- and phyllosilicate-bearing material exhumed and exposed by crater central uplifts (e.g., Leighton Crater). These authors prefer the hypothesis that these ancient exposed deposits were likely buried and metamorphosed sediments rather than products of hydrothermal alteration in the subsurface, based on morphological evidence of sedimentary features; but the authors acknowledge that CO2 rich fluids interacting with subsurface basaltic rocks may also explain the presence of these assemblages. The fact that the occurrences of olivine, Mg-phyllosilicate, and carbonate-bearing assemblages are not localized to one region suggests hydrothermal carbonate formation may have been an important process for CO2 sequestration during the Noachian.

5.5 Physical and Chemical Constraints of Chloritization, Serpentinization, and Carbonation

[32] On Earth, corrensite (C/S) forms in a variety of environments, including hydrothermally altered basalts, ultrabasic rocks, burial diagenetic sequences of volcanic and sedimentary rocks, and in contact metamorphic zones (as summarized by Środoń [1999]). Because of the recognized transitional sequence of S → C/S → C during burial, the appearance of corrensite can be used as a paleogeothermometer, as first identified by Hoffman and Hower [1979]. In hydrothermal and diagenetic systems, corrensite has been reported to be stable between 150°C and 300°C [Schiffman and Fridleifsson, 1991; Hoffman and Hower, 1979]. Using an end-member that has presumably undergone the least amount of chloritization in the Nili Fossae region as the end-member to represent the onset of corrensite, we bound the chloritization process from ~0.5 to ~3.0 km depth (see Figure 6, bottom), yielding a maximum paleogeothermal gradient of ~60°C/km. This gradient is equivalent to ~20× the estimated current geothermal gradient of ~3°C/km, (derived from an estimated global Martian heat flux of ~6 mW/m2 and crustal thermal conductivity of ~2.0 W/mK [e.g., Hahn et al., 2011]). The equivalent heat flux (120 mW/m2) doubles the highest estimates of crustal heat flow during the Noachian of ~50–60 mW/m2 [McGovern et al., 2004; Hahn et al., 2011] and approximates mean oceanic heat flow in young (<~10 Ma) oceanic crust on Earth [e.g., Pelayo et al., 1994; Stein and Stein, 1994]. Though chloritization appears to increase with depth in the Nili Fossae region (Figure 6, bottom), which is consistent with burial diagenesis, the maximum geothermal gradient indicated by the C/S paleogeothermometer is significantly higher than predicted for typical Noachian crustal heat flow and thus requires one of the following: (1) an elevated thermal gradient, (2) significant crustal stripping (a minimum of ~2.5 km removed since the formation of chlorite), or (3) indicates the spectral identification of C/S are not fully characterized by the parameters in Figure 5. An elevated thermal gradient in this region could have been sourced from local hydrothermal circulation related to volcanism (Nili Patera) and/or an impact-induced hydrothermal system (Isidis Basin).

[33] The chemical reactions that drive chloritization of smectite may be written in two ways: (1) to preserve the original saponite 2:1 layers remain intact and that mineralogic change only involves ionic substitution within the structure where Al is a mobile component, and (2) without conservation of the original 2:1 layers and where Al is immobile [e.g., Chang et al., 1986].

  • 1.Chloritization (Al is mobile)
    display math(7)
  • 2.Chloritization (Al is immobile)
    display math(8)

[36] Both forms of the reaction include aqueous silica as a product (although equation (7) produces silica more per mole of saponite). This is important to note, as aqueous silica is a required reactant in the serpentinization process (equation ((1)). In eastern Nili Fossae, the burial of saponite and subsequent chloritization at depth may have produced the source of silica-rich fluids necessary for serpentinization. The widespread fracture system identified in the Noachian basement rocks [e.g., Mangold et al., 2007] could have provided conduits for silica-rich fluid flow to interact with overlying stratigraphy. Since CRISM data do not indicate compositional variation in the linear fractures [Mustard et al., 2009], these erosionally resistant ridges may be composed a form of mineralized silica that is spectrally neutral in the CRISM wavelength range (e.g., quartz). While both chloritization reactions require water for the formation of chlorite, the proportion of water necessary in the second reaction (equation (8)) can easily be accounted for by dewatering of saponite interlayer water (up to ~4 mol) [Chang et al., 1986]. If chloritization proceeds via equation (7), then a supplemental source of water would have been necessary (at least ~3 additional moles) to complete the reaction.

[37] While in situ measurements of assemblages (at the mineral scale) are necessary to confidently determine the set of reactions that actually produced the observed mineralogy, we can still use remote data to constrain the environmental conditions necessarily to produce the stratigraphy and mineralogy (at the >20 m scale) in the Nili Fossae region. Figure 9 (top) shows an activity-activity diagram indicating the activities of aqueous SiO2 and CO2 that represent the pathways and stability fields for the minerals involved in the carbonation reaction. The carbonation pathways indicated by the dashed lines are the geochemical results if SiO2 is buffered by the reaction. A serpentinite assemblage containing brucite will only be stable at very low CO2 concentrations, and with increasing CO2 will be replaced by a serpentine + magnesite assemblage, then a talc + magnesite assemblage, and finally a quartz + magnesite assemblage [e.g., Klein and Garrido, 2011, and references therein]. The amount of olivine [(Fe,Mg)SiO4] relative to pyroxene [XY(Si,Al)2O6] in the starting lithology is important for determining the amount of excess silica released during water-rock interactions. Excess silica is more prevalent with a pyroxene-rich protolith and promotes the formation of talc [Bach et al., 2004] as aSiO2 increases.

Figure 9.

Geochemical constrains of the carbonation reaction. (top) Activity-activity diagram depicting phase equilibria in an MgO-SiO2-H2O-CO2 system at 100°C and 10 MPa. Phase boundaries (solid lines) represent equal activities of minerals in the fields on either side of the line. Diagram was produced from isothermal, isobaric invariant points reported in Klein and Garrido [2011]. Dashed lines represent an example carbonation pathway at differing temperatures. Invariant points (i1, i2, and i3) identify SiO2 and CO2 activities where three phases coexist in equilibrium. (bottom) T-XCO2 diagram illustrating the low temperature equilibria (at 1 kbar) calculated from Berman [2007]. Stability field of the serpentine chrysotile shown in gray.

[38] We may also use the presence of serpentine to help constrain the geochemical environment in which the carbonate assemblages formed. For example, Figure 9 (bottom) shows temperature versus the mole fraction of CO2 at 1 kbar for the reactions that constrain the stability of serpentine (specifically chrysotile) in association with talc and magnesite. As fluid temperature decreases and/or CO2 activity increases, the stability of serpentine gives way to produce a talc-magnesite assemblage. For fluids with XCO2 values greater than ~0.1 for chrysotile (or ~0.2 for antigorite), serpentine is no longer stable, and olivine reacts directly to the talc-magnesite assemblage. One interpretation of the relative paucity of serpentine in Nili Fossae and in association with other olivine-rich locations on Mars [e.g., Ehlmann et al., 2010] has been that the carbonation reaction consumed the majority of serpentine [e.g., Brown et al., 2010]. However, it may also be the case that an elevated CO2 mole fraction in the hydrothermal fluids lead to the preferential formation of a talc-magnesite assemblage, as serpentine is thermodynamically unstable under these conditions. Brucite also becomes unstable at temperatures above ~350°C independent of the XCO2 (calculated from Berman [2007] as in Figure 9, bottom). Brucite has not yet been identified in these assemblages, which may indicate the carbonate-serpentine assemblages reached temperatures above 350°C.

6 Conclusions

[39] The olivine-bearing unit that overlies the Noachian phyllosilicate-bearing unit in the Nili Fossae region may have been a target of hydrothermal alteration in the past. We put forth the following hypothesized sequence of events for the development of alteration assemblages and stratigraphic relationships found in the easternmost portion of the region (Figure 10):

  1. [40] Alteration of Noachian-aged crust to Fe/Mg smectite.

  2. [41] Uniform burial and the onset of chloritization beneath olivine-rich lava flows and/or impact ejecta. Chloritization drives the production of aqueous silica, and fractures related to the Isidis impact provide a conduit for fluid flow to the boundary between the altered Noachian crust and the olivine-rich unit.

  3. [42] The overlying olivine-rich unit provides a heat source that continues to drive a thermal gradient promoting the hydrothermal alteration and serpentinization of olivine.

  4. [43] Continued alteration via high temperature (> 200°C) or low-temperature (≤ 200°C) fluid-limited carbonation of serpentine to form magnesium carbonate and talc-bearing material mixed with the underlying Noachian mixed-layer clay. This reaction requires additional CO2 that may be derived locally (melting of buried H2O-CO2 ice, e.g., Changela and Bridges [2011]) or from depth (if another fluid source is present). It is presumably in contact with the atmospheric CO2 cycle.

  5. [44] Continued burial diagenesis of Noachian phyllosilicates with the onset of Early Hesperian volcanic activity forming the Syrtis lava cap unit. This may be the heat source for the elevated thermal gradient necessary to explain the distribution of chlorite at depth.

Figure 10.

Schematic diagram through a possible geologic time sequence of the eastern portion of the Nili Fossae region. See section 6 for a description of time steps.

[45] We also propose that the uniform distribution and consistent colocation of carbonate and talc-bearing mixed-layer clay is consistent with the Brown et al. [2010] modification of the Ehlmann et al. [2008] hydrothermal hypothesis, suggesting that these units formed through a single hydrothermal event. This hypothesis suggests that hydrothermal activity was not restricted to the Noachian era but was present and possibly prevalent through the Hesperian period via volcanic flows. As Ehlmann et al. [2011] suggests, long-term hydrothermal settings may provide habitable environments through an extended period of time in Martian history. Further characterization of Fe/Mg phyllosilicates in this region and in those associated with abundant olivine (e.g., Argyre and Terra Tyrrhena [Koeppen and Hamilton, 2008]) may indicate carbonation of serpentine was a widespread process on Mars and is the reason for the limited occurrence of serpentine on Mars as shown in Ehlmann et al. [2008]. If carbonation of serpentine was a global process, this reaction could provide a significant atmospheric sink for CO2 and may have controlled part of the Martian global carbon cycle in the past.

Acknowledgments

[46] We would like to thank Ted Labotka for his guidance on carbonate geochemistry calculations, and Devon Burr and Michael Essington for their feedback on an early version of this manuscript. We would also like to thank Bethany Ehlmann and Patrick Pinet for providing valuable comments and suggestions that significantly improved this work. This work was funded by Arizona State University subcontract #10-254 (under Jet Propulsion Laboratory contract #1224808) for Dr. Moersch's participation on the THEMIS Science Team.

Erratum

  1. In the originally published version of this article, the caption for Figure 8 contained an error. The text following “CRISM scene” was “FRT000064D9,” when it should be “FRT00003E12.” This error has since been corrected and this version may be considered the authoritative version of record.

Ancillary