- © 2014 Mineralogical Society of America
Ferrian saponite from the eastern Santa Monica Mountain, near Griffith Park (Los Angeles, California), was investigated as a mineralogical analog to smectites discovered on Mars by the CheMin X-ray diffraction instrument onboard the Mars Science Laboratory (MSL) rover. The martian clay minerals occur in sediment of basaltic composition and have 02l diffraction bands peaking at 4.59 Å, consistent with tri-octahedral smectites. The Griffith saponite occurs in basalts as pseudomorphs after olivine and mesostasis glass and as fillings of vesicles and cracks and has 02l diffraction bands at that same position. We obtained chemical compositions (by electron microprobe), X-ray diffraction patterns with a lab version of the CheMin instrument, Mössbauer spectra, and visible and near-IR reflectance (VNIR) spectra on several samples from that locality. The Griffith saponite is magnesian, Mg/(Mg+∑Fe) = 65–70%, lacks tetrahedral Fe3+ and octahedral Al3+, and has Fe3+/∑Fe from 64 to 93%. Its chemical composition is consistent with a fully tri-octahedral smectite, but the abundance of Fe3+ gives a nominal excess charge of +1 to +2 per formula unit. The excess charge is likely compensated by substitution of O2− for OH−, causing distortion of octahedral sites as inferred from Mössbauer spectra. We hypothesize that the Griffith saponite was initially deposited with all its iron as Fe2+ and was oxidized later. X-ray diffraction shows a sharp 001 peak at 15 Å, 00l peaks, and a 02l diffraction band at the same position (4.59 Å) and shape as those of the martian samples, indicating that the martian saponite is not fully oxidized. VNIR spectra of the Griffith saponite show distinct absorptions at 1.40, 1.90, 2.30–2.32, and 2.40 μm, arising from H2O and hydroxyl groups in various settings. The position of the ~2.31 μm spectral feature varies systematically with the redox state of the octahedrally coordinated Fe. This correlation may permit surface oxidation state to be inferred (in some cases) from VNIR spectra of Mars obtained from orbit, and, in any case, ferrian saponite is a viable assignment for spectral detections in the range 2.30–2.32 μm.
The core objective of the Mars Science Laboratory (MSL) spacecraft mission as implemented by the rover Curiosity is to seek evidence of past habitable environments on Mars (Grotzinger et al. 2014). Pre-mission analyses of the Gale Crater landing site from martian orbit indicated a range of sedimentary rock deposits, of proper nature and age, that might either have been deposited in potentially habitable environments, or have transported rocks from such environments (Anderson and Bell 2010; Milliken et al. 2010; Schwenzer et al. 2012; Wray 2013). Curiosity landed a few hundred meters west of a local depression in which layered and fractured rocks were visible from orbit. The science team and engineering project management decided that the rocks of the depression were an attractive science target and would allow testing and verification of engineering operations, most notably drilling a rock, sieving the drill cuttings, and delivery of them to the Chemistry and Mineralogy (CheMin) and Sample Analysis at Mars (SAM) instruments. Curiosity traversed to the depression, informally named Yellowknife Bay, and documented the presence of sandstones and mudstones and several types of diagenetic (post-depositional) alterations that could reasonably have involved water.
In Yellowknife Bay, the Curiosity rover drilled twice into a mudstone, the Sheepbed stratum, in sites named John Klein and Cumberland. The John Klein site was chosen as typical of the Sheepbed exposures and contained white-colored crosscutting veinlets. The Cumberland site was chosen for its abundance of concretionary-like structures that were relatively resistant to weathering (Grotzinger et al. 2014). Clay minerals, signs of formation or alteration in aqueous environments, are present in both drilled samples (Fig. 1), a result established by the CheMin instrument from X-ray diffraction analysis (Vaniman et al. 2014) and confirmed by the SAM instrument by analysis of evolved gases released as a function of temperature (Ming et al. 2014).
The finding of clay minerals has huge importance for the goals of the MSL mission, but (of course) the analytical instruments on Curiosity allow only a small subset of the techniques that are commonly applied on Earth to characterize clay minerals and their geological contexts. Thus, we have sought terrestrial equivalents for the occurrences of clay minerals in the Sheepbed unit. This work is a detailed characterization of one possible terrestrial equivalent, the best currently known with respect to results from CheMin, for the clay mineral detected by MSL at Gale Crater on Mars.
Mineralogy of the Sheepbed Unit
As described by Vaniman et al. (2014), drilled rock fines from the John Klein and Cumberland sites were sieved, and the <0.150 mm size fraction delivered to the CheMin instrument (Anderson et al. 2012), which produced two-dimensional X-ray diffraction patterns of the materials (Blake et al. 2012). These patterns were converted to one-dimensional patterns (Fig. 1a), and interpreted as to mineral identities and proportions using commercial software and reference diffraction patterns taken with laboratory equivalents of CheMin. Both rocks consist mostly of minerals characteristic of basalt: pyroxenes, plagioclase, and olivine, with minor alkali feldspar, pyrrhotite, and Fe-Ti oxides. They also contain amorphous material (Morris et al. 2013) and minerals characteristic of low-temperature aqueous alteration, including smectite, Ca sulfate, hematite, and akaganeite, as well as abundant magnetite/maghemite interpreted as authigenic (Vaniman et al. 2014).
The presence of smectite minerals in the John Klein and Cumberland samples is shown by the broad 001 diffractions at ~10.0 and 13.2 Å (Fig. 1a). However, the 001 diffraction is not characteristic, by itself, of exact clay mineral species because its position varies according to the speciation of the interlayer cations and their hydration state. Many standard methods for characterizing clay minerals by XRD (e.g., glycolation and cation exchange) cannot be performed by the MSL payload and, furthermore, the 2θ range of the MSL CheMin instrument (~5 to 50° 2θ, CoKα) does not include the position of the 06l diffraction band, which is normally used to differentiate between di-octahedral and tri-octahedral smectites. Instead, Vaniman et al. (2014) used the position of the 02l diffraction band to make this distinction. They report that its position for both John Klein and Cumberland (22.5° 2θ, CoKα; 4.59 Å d-spacing) implies a tri-octahedral smectite because the maximum in the 02l diffraction bands are at lower 2θ (greater d-spacing) than reference di-octahedral smectites and more similar to reference tri-octahedral smectites, although still at somewhat lower 2θ (Vaniman et al. 2014). This small difference between the martian smectites and the reference tri-octahedral smectites results from the speciation of the cations in the octahedral sites of martian vs. terrestrial reference smectites.
Here, we report on the X-ray, chemical, and spectroscopic properties of saponite from Griffith Park, California, as an analog to the Sheepbed smectites. The Griffith Park material was specifically selected for this study because it has an 02l diffraction band similar in location and shape to those of the Sheepbed smectites (Vaniman et al. 2014), it is an iron-bearing tri-octahedral smectite, and sufficient quantities are available for future studies (commercially, from museums, and from field collection). The Griffith Park saponite, originally reported by Larsen and Steiger (1917), has been studied extensively for applications in materials science and catalysis (Larsen and Steiger 1928; Rodriguez et al. 1994, 1995; Vicente-Rodrigues et al. 1996; Vicente et al. 1996, 1997, 1998; Komadel et al. 2000; Komadel 2003; Gandia et al. 2005; Komadel and Madejova 2006; Stucki 2006). However, the geological setting of its formation and its properties relevant to Mars are poorly known. Our purpose here is to characterize Griffith saponite and its formation mechanisms for comparison with the smectite minerals detected by CheMin in Gale Crater, for comparison to clay minerals detected from martian orbit by visible and near-infrared (VNIR) spectroscopy, and for comparison to Mössbauer data obtained by the Mars Exploration Rovers.
Saponite from Griffith Park: Description and geologic setting
Iron-bearing saponite was recognized first in Griffith Park, Los Angeles, California, by Larsen and Steiger (1917) as fillings of amygdules (vesicles) in Miocene basalts of the Topanga Canyon Formation in the eastern Santa Monica mountains (Critelli and Ingersoll 1995). The saponite was originally given the mineral name “griffithite,” and identified as a chlorite species because of its appearance as black lustrous crystals, up to 0.5 mm across, with strong platy cleavage. This “griffithite” loses significant water on heating to 55 °C (Larsen and Steiger 1928), consistent with it being a saponite, but was still classified as a chlorite (Neuerburg 1951) until thermal and X-ray data confirmed that it was a smectite (Faust 1955), specifically a mixed Mg-Fe3+-Fe2+ tri-octahedral smectite (Komadel 2003). The mineral name “griffithite” was then discredited (Fleischer 1955), but is occasionally used as a varietal name (e.g., Pecuil et al. 1966; Vicente et al. 1996; Changela and Bridges 2011). Here, we will call this material Griffith saponite to distinguish it from smectites and saponite from other localities.
The type locality for the Griffith saponite is Griffith Park, City of Los Angeles, California. The exact location may no longer be available, as it was at the southern end of Cahuenga pass in an area now covered by the Hollywood Freeway, Cahuenga Boulevard, and the Hollywood Bowl (Larsen and Steiger 1917; Neuerburg 1951). The type locality is in the lower member of the Topanga Canyon Formation, of Miocene age, which is a complex sequence of marine arkosic sediments, basaltic volcaniclastic sediments, basalts, and rare limestones (Neuerburg 1953; Yerkes and Campbell 1979; Critelli and Ingersoll 1995).
Large sills and irregular intrusive bodies of dark brown basalt are common in the lower member of the Topanga Formation. Many of these intrusives are amygdaloidal; the distribution of the amygdules is erratic and appears to have no relation to contacts or to inclusions. Veins of zeolites and related minerals are common in these intrusive rocks. The texture ranges from porphyritic and intergranular to intergranular; most specimens of intrusive rock are porphyritic. Phenocrysts generally comprise about 10 to 15% of the rock; the minerals are andesine-labradorite, augite, and in some specimens olivine. The groundmass consists predominantly of andesine with a little augite and magnetite; chlorite [= smectite] and chlorophaeite (?) fill spaces in the groundmass (Neuerburg 1953).
Saponite is also found in rocks of the overlying middle member of the Topanga Canyon Formation. The middle member is a complex of basaltic volcaniclastic sediments, including mudstones, conglomerates, and sandstones (Neuerburg 1953).
Small angular to sub rounded fragments of various types of basalt are set in a matrix of dark brown, opaque, claylike material that has a waxy luster (chlorophaeite?) [= saponite]. Mineral fragments consist of andesine, oligoclase, quartz, and augite, with minor amounts of myrmekite, analcite, biotite, and zircon. The boulders and cobbles consist primarily of augite basalt and olivine basalt with a few boulders and pebbles of pink granite, quartz diorite, quartzite, acidic porphyries, and arkose (Neuerburg 1953).
All of the basaltic rocks of the Topanga Canyon formation show effects of low-temperature aqueous alteration. In them, saponite is locally abundant, and zeolites of several species are locally common, as are other aqueous alteration phases including zoisite and prehnite (Neuerburg 1951).
Samples and methods
We examined saponite in three samples of altered basalt from the Griffith Park area, Los Angeles, California (Fig. 2). We have neither exact localities nor geological settings for any of these samples, except the general data given in the Introduction. One sample was obtained from the American Museum of Natural History (AMNH 89172); another set of samples was purchased from Minerals Unlimited (MU); a third sample, purchased from M & W Minerals, had been in the collection of the California Academy of Sciences (CAS). The AMNH and CAS samples are old enough that they had been curated as distinct mineral species, and so likely came from the type locality (now under the Hollywood Bowl).
A mineral separate of saponite was prepared from each sample by manually disaggregating and handpicking a portion of the sample. For the AMNH and MU samples, saponite was extracted manually from filled vesicles. From the CAS sample, we prepared three separates from vesicle fill material: fine-grained brown clay, coarse black clay, and white spherules (see below in Petrography). Portions of the mineral separates were cut to form thick sections for electron microbeam imaging and elemental analysis and portions were dry ground and dry sieved to <0.150 mm for analysis, in sequence, by backscatter Mössbauer spectroscopy (MB), X-ray diffraction (XRD), visible, near IR reflectance spectroscopy. It is likely that each handpicked separate contains a few percent of other materials.
For comparison with the Griffith saponite samples, we also analyzed by XRD a number other smectites: API-33A (Garfield nontronite), PHY07 (Pennsylvania nontronite), SWa-1 (ferruginous smectite), NAu-1 (Australian nontronite), WASCDB1 (nontronite), SWy-1 and BSDMNT1 (Na-montmorillonite), and SAz-1 and STx-1 (Ca-montmorillonite). The samples were analyzed as the <0.150 mm size fraction except for PHY07, which was analyzed as the <0.038 mm size fraction.
Portions of the saponite-bearing samples were cut to form thick sections for electron microbeam imaging and elemental analysis. The section surfaces were smoothed with abrasives down to laps with 3 μm diamond powder embedded in plastic. It was not possible to polish these samples, as the soft saponite plucked out, leaving a rougher surface than before polishing. Thus, the section surfaces are not optically flat as required for precise quantitation with EMP. In addition, the saponite-bearing samples changed significantly during the few minutes of vacuum before carbon coating. The saponite-rich areas changed from black to tan color and became visibly rougher. Thus, one cannot expect good quantification in the EMP analyses and analytical totals are significantly less than 100% even for anhydrous minerals like pyroxene and plagioclase (Tables 1 and 2); of course, analyses of saponite have even lower totals because of their structural OH (Table 3). However, we believe that element abundance ratios are not affected significantly by surface roughness, see below in Results.
Sections were imaged in backscattered electron (BSE) and secondary electron (SEI) modes and analyzed for element abundances using the CAMECA SX-100 electron microprobe at the Ares Directorate, Johnson Space Center, Houston, Texas. Analytical conditions were nominal for the instrument and laboratory. Peak intensities were measured for Kα radiation of these elements using these well-characterized standards: Si, diopside; Ti, rutile; Al, oligoclase; Cr, chromite; Fe, fayalite; Ni, NiO; Mn, rhodonite; Mg, diopside or forsterite; Ca, diopside; Na, oligoclase; K, orthoclase; and S, anhydrite. No significant quantities of S were detected, so S is not given in Table 3 The incident electron beam was at 15 kV and 10 nA, and defocused to 10 μm diameter on standards and samples. Peak X-ray intensities were counted for 30–60 s, and backgrounds were counted for the same total time. Analytical standards were run as unknowns to validate the calibrations, which were crosschecked against plagioclase and pyroxene adjacent to the smectite.
Powder X-ray diffraction
Transmission X-ray diffraction patterns for Griffith saponite samples were acquired at room temperature using the CheMin-IV instrument at the ARES Directorate, Johnson Space Center. CheMin-IV is a laboratory version of the MSL CheMin flight instrument and is used to baseline its capabilities (Blake et al. 2012). The CheMin-IV at ARES is configured with an N2 gas (derived from liquid N2) inlet tube so that measurements can be made in a dry N2 atmosphere or in (humid) lab air. The CheMin-IV, like MSL CheMin, acquires CoKα diffraction patterns in transmission geometry using radiation from powder samples that are continuously vibrated ultrasonically to agitate the material and achieve a variety of crystal orientations in the sample cell. The detector is energy sensitive, which permits post-analysis filtering to yield only diffracted (and scattered) CoKα X-rays and to exclude CoKβ X-rays and characteristic X-rays fluoresced from the target sample.
Backscatter Mössbauer (MB) spectra (Fe57: 14.4 keV) were acquired at room temperature using MIMOS-II spectrometers from SPESI. The spectrometers are laboratory equivalents of the instruments onboard the Mars Exploration Rovers (Klingelhöfer et al. 2003) with additional radiation shielding and without a reference source. The instruments were oriented vertically, so that the γ-ray beam is pointed at the ceiling and powder samples were oriented horizontally on glassine paper substrates, which are essentially transparent to the 14.4 keV γ-rays. All measurements were made in laboratory air. The source radiation was 57Co(Rh) with spectra acquired in 512 channels (folded to 256 channels). MB velocity calibration was done using the spectrum for metallic Fe foil acquired at room temperature, the MIMOS-II differential signal spectrum, and the program MERView (Agresti et al. 2006). MB parameters [isomer or center shift (CS), quadrupole splitting (QS), hyperfine field strength (Bhf), and subspectral areas of Fe-bearing phases (A)] were obtained by a least squares fitting procedure with the program MERFit (Agresti and Gerakines 2009).
VNIR reflectance spectra between 0.35 and 2.5 μm were acquired at ~25 °C with analytical spectral devices (ASD) FieldSpec3 spectrometers configured with ASD Mug Lights. One instrument was located to make measurements in (humid) lab air. A second instrument was located inside a one-atmosphere glove box configured with a transfer chamber (Plas-Lab); the glove box and transfer chamber interiors were independently and continuously purged with dry-N2 gas derived from liquid N2. Co-located in the glove box were a hot plate (Fisher Isotemp), a dewpoint meter (Vaisala DRYCAP DM70) to measure the H2O content, relative humidity (RH), and temperature of the dry N2 atmosphere (100–180 ppm by volume, <0.5%, and ~25–30 °C, respectively), and an IR thermometer (Fluke Model 66) to measure hot plate and sample surface temperatures. Spectralon (SRS-99-010; Labsphere, Inc.) was used as the reflectance standard. The spectra from the three detectors in the ASD instrument were spliced using software supplied with the instrument.
VNIR spectra were obtained in lab air for all samples. Spectral measurements for saponite sample AMNH85172_PHY were also acquired during desiccation under dry N2 at room temperature, 110 °C, and 220 °C for 436, 1152, and 212 h, respectively. Note that these spectral measurements were all made at room temperature in the dry N2 environment of the glove box; the cool-down time from 110 and 220 °C to ambient was a few minutes. The sample was remeasured in lab air after 16 h exposure to that environment.
The Griffith saponite occurs in rocks of basaltic composition, both basalts proper and clastic sediments composed primarily of basaltic detritus. The basalts consist of saponite, augite, plagioclase, titanomagnetite, and ilmenite (Figs. 3a, 3e, 3g, 3h, and 3i). The original textures of the basalts were sub-ophitic (augite partially enclosing plagioclase euhedra) to intersertal (glassy mesostasis among augite and plagioclase). From the compositions of the plagioclase and pyroxene (Figs. 4a and 4b), notably the absence of low-Ca pyroxene, the basalts were likely not tholeiitic, i.e., they were alkali olivine basalts (Hoots 1931). Relict olivine is reported is some basalts of the Topanga Canyon formation (Neuerburg 1953), but our samples contained none.
Griffith saponite was originally recognized as filling vesicles (amygdales) in the Topanga Canyon Formation basalts (Larsen and Steiger 1917; Neuerburg 1951; Critelli and Ingersoll 1995) and occurs as such in all three of our samples (Figs. 3c, 3d, 3e, 3f, and 3h). In addition, saponite is also present in two other petrographic settings. Saponite replaces olivine in the basalts, both as phenocrysts and anhedral grains among pyroxene and plagioclase (Figs. 3b, 3d, and 3i). In neither case are former olivine grains surrounded by expansion fractures, as are common when olivine is replaced by serpentine (O’Hanley 1996). Finally, saponite occurs in areas that consisted of mesostasis material, among crystals of olivine, augite, and feldspar (Figs. 3a, 3e, 3i, and 3j); this mesostasis was probably glass with quench crystals, including elongate laths of ilmenite and dendritic, hollow grains of titanomagnetite (Figs. 3f, 3g, and 3j). There is no evidence that plagioclase or augite were replaced by saponite.
The CAS sample shows evidence of several varieties of low-temperature or hydrothermal alteration, which are best exposed in the filled vesicles. Going from vesicle wall to center, the first alteration product is fine-grained, tan-colored saponite material [possibly the “chlorophaeite” of Neuerburg (1951)]; this material is also present as crack-fillings. The fine-grained saponite has a conchoidal fracture, and shows no crystal shapes, cleavages, or faces (to <5 μm as apparent in secondary electron imagery). This material is denoted as “Vf-f,” fine-grained vesicle fill, in Figures 1 to 3 and Table 3. Next inward in some vesicles is a white shell or sphere of silica, shown in X-ray diffraction to be α-quartz, i.e., agate or chalcedony (Figs. 2f and 2g), which is reported as veinlets near the type area for Griffith saponite (Neuerburg 1953). The silica contains small proportions of at least two other phases, an iron oxide and an unidentified Ca-Mg-Al silicate (Figs. 3f). Interior to both fine-grained saponite and silica (where present) is coarse-grained saponite, “Vf-c” in Figure 2 and Table 3, in grain sizes up to ~100 μm (Fig. 2d). The coarse-grained saponite fills the cores of vesicles, and occasionally the cores of silica spherules. Similar coarse-grained saponite also replaces olivine phenocrysts and crystals in the basalts, commonly with the saponite (001) crystal planes perpendicular to crystallographically oriented cracks across the original olivine (Fig. 3b). Mesostasis areas among the plagioclase and pyroxene of the basalt are also replaced by material rich in saponite, but it is difficult to classify it as fine- or coarse-grained (Figs. 3g and 3j).
The AMNH and MU samples contain only coarse-grained saponite as a common alteration material; no fine-grained saponite or silica was noted. In the AMNH and MU samples, coarse-grained saponite is present as vesicle fills, as replacements of olivine, and as replacements of mesostasis.
Properties of Griffith smectite
The chemical compositions and crystal chemistry of Griffith saponite (Table 3) are based on EMP chemical analyses and on Mössbauer spectroscopy for iron speciation and siting (Table 4). The EMP chemical analyses are of lower quality than would normally be acceptable for rock-forming minerals, in terms both of analytical totals and variability (Table 3). The low totals, averaging 85–91 wt% (Table 3), are caused by three factors: inherent H2O/OH content, polish, and porosity. The saponites must retain structural OH, even though desiccation in high vacuum; ideally, dehydrated (not dehydroxylated) saponites like these should contain ~5 wt% H2O equivalent (e.g., Larsen and Steiger 1917). Good EMP analyses rely on having an optically flat surface, and it was not possible to polish the saponite in our samples—it is soft and greasy, and plucked out on polishing so we could only flatten the surfaces to a grit size of 3 μm. Thus, this surface roughness affects analytical totals, even on the anhydrous phases like augite and plagioclase (Tables 1 and 2). Finally, the analyzed saponite was distinctly porous, as seen in secondary electron imagery (Figs. 3b and 3c). The porosity includes original intergranular space and porosity developed as the saponite dehydrated during sample preparation and analyses. For reference, the type Griffith saponite contained ~17 wt% H2O (Larsen and Steiger 1917). The variability in chemical analyses derives partially from the roughness of the sample surfaces, and the random orientations of those roughness elements with respect to X-ray detectors (and thus X-ray paths). In theory, the average of many analyses ought to even out this variability. Analyses of saponite in mesostasis areas are even more variable, because of relict unaltered phases. From the many analyses of mesostasis area, we have culled out those that clearly contain plagioclase feldspar, ilmenite, and/or titanomagnetite (Figs. 3g and 3j). However, it was not possible to distinguish small contributions from any of these phases from inherent variability of the saponite.
The chemical compositions of the Griffith saponite samples are generally similar, but with some significant differences. Among all the samples and textures, saponite shows limited variability in most cations; abundances of Al, Fe, and Mg are very similar (Table 3), and average Mg#s range only from 65 to 71% (Table 3). A puzzling difference among these samples is MnO abundance, which is an order of magnitude greater in the MU samples than in the AMNH and CAS samples. The greatest differences are in abundances of interlayer cations: Ca, Na, and K. The AMNH saponite, in all its textural settings, has significantly lower Ca than the MU and CAS samples, but an intermediate Na content.
The Griffith saponite analyzed here is similar to, but not identical with, analyses in the literature (Table 3; Fig. 5), showing that its chemical composition and Fe redox state are not narrowly fixed. The literature analyses tend to have higher Al and Si contents, slightly lower Mg#s, and Mn/FeOT ratios intermediate between MU and AMNH (and CAS). Our Griffith saponite samples include those with the lowest MnO/FeOT and lowest Fe3+/∑Fe ratios analyzed to date, excluding Larsen and Steiger (1917) who report chemical and Fe redox data but no Mössbauer, XRD, or VNIR data. There is general agreement that Griffith saponite has ~17 wt% equivalent H2O (Table 3). Our EMP analyses (in vacuum) show an average deficit of ~13 wt% in good agreement with the value of 12.3 wt% reported by Larsen and Steiger (1917) for non-hydroxyl H2O. Comparisons of these analyses are clearer in the normalized formulas, which are discussed below.
Iron mineralogy and redox state
Results of Mössbauer spectroscopy (Table 4) constrain the speciation and siting of iron in the saponite-bearing samples. Mössbauer spectra of Griffith saponite samples have three or four doublets that are assigned to Fe2+ and Fe3+ in octahedral coordination (Fig. 6; Table 4; e.g., Gates et al. 2002; Cashion et al. 2008). Samples AMNH85172_PHY and MUGPLA1_PHY are characterized by two Fe3+ doublets (doublet 3D1 with CS = 0.37 ± 0.02 mm/s and QS = 1.35 ± 0.02 mm/s and doublet 3D2 with CS = 0.38 ± 0.02 mm/s and QS = 0.80 ± 0.02 mm/s) and one Fe2+ doublet (doublet 2D1 with CS = 1.14 ± 0.02 mm/s and QS = 2.62 ± 0.02 mm/s). Doublet 3D1 has an unusually high value of QS for Fe3+, indicating that the anions coordinated to that Fe3+ are quite assymetric, i.e., the coordination is highly distorted. The other two saponite samples (CASGP1-C and -F) have these three doublets plus one additional Fe3+ doublet (3D3 with CS = 0.32 ± 0.02 mm/s and QS = 0.42 ± 0.02 mm/s). We assign this doublet to a nanophase ferric oxide (npOx) phase, which would include any combination of ferrihydrite, hisingerite, superparamagnetic hematite, and superparamagnetic goethite; none of these substances diffract coherently, and would not be apparent in XRD in the inferred proportions (Table 4). The Mössbauer spectra and parameters imply that the Fe3+ is entirely in octahedral coordination. On the basis of XRD and other data, Kohyama et al. (1973) and Kohyama and Sudo (1975) consider that the weathering sequence (under oxidizing and humid conditions) is “ferrous iron-rich saponite” (i.e., ferrosaponite) to “ferric iron-rich saponite” (i.e., ferrian saponite) to hisingerite.
Kohyama et al. (1973) also report Mössbauer data for iron-rich (and MgO-poor) saponite using a two-doublet fitting model (one doublet each for Fe2+ and Fe3+). For the ferrous doublet, they report CS ranges from 1.14 to 1.19 mm/s and QS ranges from 2.52 to 2.86 mm/s, with QS decreasing gradually with increasing extent of oxidation. For the ferric doublet, they report CS ~0.36 mm/s and QS ranges from 0.86 to 0.96 mm/s. These data are in good agreement with our measurements. For the ferrous doublet, the CS values are comparable and our value of QS (2.62 mm/s) is intermediate. For the ferric doublet, direct comparison is not possible as we used a two ferric-doublet fitting model. However, the values of CS are all comparable (and therefore insensitive to the fitting model), and the values of QS for the one-doublet fit are intermediate to those for the two-doublet fit (0.80 and 1.35 mm/s), a result expected using one-doublet vs. two-doublet ferric fitting models. Considering the difference in fitting models and bulk composition (MgO-poor vs. MgO-rich for Griffith saponite), the Mössbauer data are in good agreement.
We calculate the Fe3+/∑Fe ratio for the Griffith saponite using doublets 3D1, 3D2, and 2D1. Our Griffith saponite samples have variable proportions of Fe3+ and Fe2+ in octahedral sites with Fe3+/∑Fe ranging from 0.64 for the most reduced sample (AMNH89172_PHY) to 0.85–0.92 for the most oxidized (CASGP1-C, CASGP1-F, and MUGPLA1_PHY).
Mössbauer, chemical, and other data for Griffith saponite are discussed by Vicente et al. (1998) and Komadel et al. (2000), although they do not report Mössbauer CS and QS values. Their samples have Fe3+/∑Fe ~ 0.90, which is equivalent to values for our samples originally sourced from the California Academy of Sciences (CASGP1-C and -F) and Minerals Unlimited (MUGPLA1_PHY). Comparison of MnO concentrations (Table 3) reveals that the Vicente et al. (1998) and Komadel et al. (2000) samples of Griffith saponite correspond to our MUGPLA1 sample. Perhaps not coincidentally, their samples were also obtained from Minerals Unlimited. The Griffith saponite analyzed by Larsen and Steiger (1917, 1928) is more like our AMNH sample with respect to Fe3+/∑Fe (0.46) and MnO concentration (Tables 1 and 2; Fig. 5).
The fine-grained saponite sample (CASGP1-F) also shows a small sextet that is assigned to crystalline hematite (Table 4) and is presumably present as a matrix contamination; this sample also contains some quartz (Fig. 7). The Mössbauer spectrum of the basalt separate (AMNH89172_BAS) shows the saponite doublets (3D1, 3D2, and 2D1), an additional Fe2+ doublet (2D2) presumably from pyroxene, a hematite sextet (S1), and the magnetite pair of sextets (S2A and S2B) (Table 4). The pyroxene and magnetite correspond to igneous pyroxene and titanomagnetite as discussed above (Figs. 3c and 3e).
The crystal chemistry of the Griffith saponite samples is given in the bottom half of Table 3. It is based on EMP chemical analyses, Mössbauer Fe speciation and site assignment, and the assumption of a normal 2:1 phyllosilicate formula: I0–2M4–6T8O20(OH)4·nH2O, where I, M and T refer respectively to cations in interlayer, octahedral (metal), and tetrahedral sites. Because the Griffith saponite contain insignificant ferric iron in tetrahedral coordination (as shown by the Mössbauer data), the analyses were normalized to Si+Al = 8 to fill the tetrahedral sites. For these calculations, we assume that the Fe3+ now present as npOx (CASGP1-C and -F) was originally integral to the saponite crystal structure, i.e., in the octahedral layer (Newman and Brown 1987).
By crystal chemistry, our Griffith saponite samples are fully tri-octahedral, with 6 octahedral cations per 8 tetrahedral cations (Si+Al) within the 1σ uncertainties (Table 3). The octahedral sites are filled dominantly by Mg and Fe in a ratio near 2:1, with small or insignificant proportions of other cations. Formally, these clay minerals are (magnesian) saponites or, considering their abundance of Fe3+, ferrian saponites. Their interlayer cation compositions are dominated by Ca, and are unremarkable.
The structural normalizations of Table 3 all have excess cation charge compared to a “normal” saponite formula, I0–2M4–6 T8O20(OH)4·nH2O. The AMNH and CAS samples have total apparent formula charges of ~ +1 (within 1σ uncertainties), while the MU samples have a total apparent formula charge of ~ +2. These apparent charge excesses are not artifacts of analysis, as they are greater than what might arise from analytical uncertainties (Table 3). The apparent excess charges must be balanced somehow, and two mechanisms seem possible: unanalyzed interlayer cations and deprotonation of structural OH. Additional interlayer cations, contributing 1 or 2 charges per formula unit (Table 3), could balance the excess charge in the analyzed formulas. Possible unanalyzed interlayer cations include Li, Sr, and/or Ba. However, we think deprotonation is a more likely explanation: that one or two of the structural OH− groups in the formula is replaced by O2−. This is a known oxidation mechanism for phyllosilicates (Farmer et al. 1971; Borggaard et al. 1982; Lear and Stucki 1985; Rancourt et al. 1993), and is consistent with the inferred distortion in the octahedral sites for the Mössbauer 3D1 doublet (above).
X-ray diffraction patterns for our Griffith saponite samples (Fig. 7) are consistent with diffraction patterns in the literature (Vicente et al. 1996, 1997; Komadel et al. 2000). All the patterns show a strong 001 diffraction near 15 Å, a series of 00l diffractions, and a distinct 02l diffraction band near 4.6 Å. The strength and sharpness of the 00l diffractions imply that the Griffith saponite is highly crystalline, with little layer disordering, intercalations, or interstratifications; this is consistent with the appearance of the saponite as glossy plates. Because of the sharpness of the 001 peak, the Lorentz-polarization correction (Reynolds 1986; Moore and Reynolds 1997) resulted in negligible change in position except for sample AMHN 89172_PHY equilibrated in dry N2 in the CheMin-IV (11.8 vs. 11.9 Å after the correction). The positions of the 00l diffractions vary strongly and reversibly with hydration state (Fig. 8; Table 5). Griffith saponite (AMNH89172_PHY and CASGP1-C) dehydrated under dry N2 gas collapses to an 001 spacing of ~12 Å, implying a change from ~2 layers of H2O interlayer to <1 (Bish et al. 2003). On heating to ~220 °C in dry N2, Griffith saponite (AMNH89172_PHY) collapses further to a 001 spacing of ~10 Å, i.e., nearly no interlayer water (spectrum not shown).
It is noteworthy that the intensities of the 004 and 005 diffractions of the Griffith saponite vary with hydration state (Fig. 8). For the dehydrated saponite, with no water or hydroxyl layer between adjacent T-O-T layer packages, the 004 diffraction (at ~3 Å) corresponds to the distance between adjacent layers of O atoms (or of cations) in that package. For the hydrated saponite, the 005 diffraction at ~3 Å corresponds again to the distance between adjacent layers of O atoms in the saponite if it contains a single layer of O atoms (water or hydroxyl) between the T-O-T layer packages.
The 02l diffraction band is important as a measure of the unit-cell “b” dimension of the octahedral layer of the smectite structure, and thereby of the nature of the cations in the octahedral layer. Di-octahedral smectites have smaller 02l distances (i.e., greater 2θ values) than do tri-octahedral smectites, as shown in Figure 9 by a comparison of the montmorillonite and nontronite 02l bands with that of the tri-octahedral saponite SapCa-1. The Griffith saponite samples all have 02l spacings significantly greater than any di-octahedral smectite, and even greater than the magnesian saponite SapCa-1 (Fig. 9). This comparison suggests that the Griffith smectite is tri-octahedral (Vaniman et al. 2014). The Griffith saponite samples analyzed here show a range for the maximum in the 02l band, from a minimum near that of SapCa-1 at 4.55 Å to a maximum of 4.59 Å, and a positive correlation between the 02l maximum (in degrees 2θ) and Fe3+/∑Fe. The positive correlation extrapolates to the ferrosaponite studied by Chukanov et al. (2003) (02l = 4.72 Å and Fe3+/∑Fe = 0.27). That maximum 02l value for Griffith (ferrian) saponite is identical to the one reported by Vaniman et al. (2014) for the smectites in the martian samples John Klein and Cumberland. If the correlation between 02l maximum and oxidation state is valid, iron in the martian Sheepbed saponites is incompletely oxidized.
VNIR reflectance spectroscopy
Reflectance spectra for the four separates of Griffith saponite are shown in Figure 10a. All are characterized by overlapping Fe2+ and Fe3+ electronic and Fe2+-Fe3+ charge transfer absorptions between ~0.35 and ~1.3 μm. Qualitatively, this region of the smectite VNIR spectrum resembles that for chlorite, which also has Fe2+ and Fe3+ (King and Clark 1989). The reflectance of AMNH89172_PHY is considerably lower than those of the other three samples over this region (0.22 vs. ~0.40 at 1.0 μm), presumably because it has the lowest Fe3+/∑Fe ratio (Table 4). We have labeled the absorptions centered near 1.4, 1.9, and 2.1 to 2.5 μm as “OH,” “HOH,” and “MOH”. Spectral features in the OH region result from stretching and stretching plus bending vibrations of the H2O molecule (2ν1, 2ν3, ν1+ν3, 2ν2+v1, and 2ν2+ν3, where ν1 and ν3 are the stretching fundamental vibrations and ν2 is bending fundamental vibration of the H2O molecule) and O-H stretching vibrations of the M-OH group (where M = Fe2+, Fe3+, and Mg in saponite; Table 3). Spectral features in the 1.9 μm region result from a combination of H2O bending and stretching fundamental vibrations (ν2+v1 and ν2+ν3). Note that if the H2O molecule is absent, spectral features near 1.9 μm will not be present. Spectral features in the MOH region result from combinations of the OH stretching and bending fundamentals of the MOH group. For tri-octahedral smectites, MOH spectral features are expected from cations in octahedral sites [normally (Al,Mg,Fe2+,Fe3+)3OH], OH spectral features are expected from (Al,Mg,Fe2+,Fe3+)3OH and interlayer H2O, and HOH spectral features are expected from interlayer H2O. Adsorbed but not interlayer H2O would also contribute to OH and HOH spectral features.
The positions of the OH, HOH, and MOH spectral features for the Griffith saponite samples are compiled in Table 5. As expected for saponite equilibrated in lab air, and directly shown by the XRD data with 001 peaks at ~15 Å (Table 5), absorption features from interlayer water, near 1.4 and 1.9 μm, are prominent. Continuum-normalized spectra for the MOH region are plotted Figure 10b, and show two well-defined minima. The position of the more intense band is variable, ranging from 2.300 to 2.315 μm. The position of the less intense band ranges from 2.395 to 2.401 μm, and there is a shoulder at longer wavelengths indicating the presence of unresolved bands there.
The position of the ~2.31 μm spectral feature can be interpreted in terms of the redox state of the octahedrally coordinated Fe2+ and Fe3+. Referring to Table 5, the saponite (AMNH89172_PHY) with the lowest value of Fe3+/∑Fe (= 0.64) has the longest wavelength for the MOH band (2.315 μm). Conversely, the two saponites (CASGP1-C and -F) with the largest values of Fe3+/∑Fe (0.89 and 0.90) have the shortest wavelengths for their MOH bands (2.300 and 2.301 μm). The MU saponite (MUGPLA1) has intermediate values of Fe3+/∑Fe (0.85) and of MOH position (2.308 μm). This spectral analysis is consistent with various laboratory studies which report that the OH stretching vibration (2.6 to 2.9 μm region) shifts to longer wavelengths with possible concurrent formation of tri-octahedral domains upon chemical reduction of structural Fe3+ in nontronite, a di-octahedral Fe3+ smectite, and Griffith saponite to Fe2+ (Stucki and Roth 1976; Komadel et al. 1995, 2000; Manceau et al. 2000; Fialips et al. 2002; Komadel 2003). Recently, Chemtob et al. (2014) report a shift to shorter wavelengths for the MOH spectral region upon oxidation of Fe2+ to Fe3+ in an iron-rich, synthetic saponite.
Our crystal-chemical analysis of AMNH89172_PHY (Table 3) gives Mg/(Fe2++Fe3+) = 2 for the octahedral sites. This ratio, plus the value of Fe3+/∑Fe (0.64; Table 5), are consistent with assignment of the MOH spectral feature at 2.315 μm to (Mg2Fe3+0.64Fe2+0.36)-OH. Similarly, we can assign the 2.308 μm spectral feature for MUGPLA1B to (Mg2Fe3+0.85Fe2+0.15)-OH. If we assume that the Mössbauer doublet 3D3 (Table 4) for samples CASGP1-C and -F is associated with saponite and not npOx, then the MOH band near 2.300 μm in those samples can be associated with (Mg2Fe3+0.92Fe2+0.08)-OH.
As shown in Figure 11 for AMNH89172_PHY, desiccation greatly reduces the intensity of the spectral feature at ~1.9 μm, as interlayer H2O is removed (Fig. 11b), because that feature requires the presence of the H2O molecule. In comparison, the depth of the spectral features at 2.30–2.32 μm are relatively constant because they represent (Mg,Fe2+,Fe3+)3-OH and not interlayer H2O. The positions of these two spectral features are also invariant with respect to desiccation. For the spectral feature near 1.4 μm, desiccation both reduces its intensity (from loss of interlayer H2O) and changes its position from 1.413 to 1.404 μm (from enhanced contribution of hydroxyl bound to octahedrally coordinated cations relative to hydroxyl bound to H) (Fig. 11c). Note that changes in intensity of the OH and HOH spectral features are essentially reversible upon re-exposure of the sample to lab air.
The chemical and physical properties of the Griffith saponite samples are all similar, although there are some distinct differences. The chemical compositions of all Griffith samples examined are consistent with fully tri-octahedral saponite, with the octahedral sites completely filled (dominantly with Fe and Mg) and with no evidence for octahedral Al3+ or tetrahedral Fe3+. Most of the iron in Griffith saponite is Fe3+, ranging from 64% of all iron to 93%. With so much ferric iron, the nominal formula for a saponite does not charge-balance, with an overall excess charge of +1 to +2, well outside of nominal uncertainties. This apparent charge excess is likely compensated by substitution of OH− in the typical saponite structure by O2−—a common mechanism for iron oxidation in phyllosilicates. This suggests that iron in the original saponite was likely all (or mostly) ferrous, and that the current range of Fe3+ abundances arose by later oxidation. Interlayer cations in Griffith saponite are dominated by Ca. The VNIR spectra are typical for smectite minerals, and the position of the absorption near 2.3 μm varies with the Fe oxidation state.
Formation of the Griffith smectite
Chemical and textural evidence suggests that the saponite in Griffith Park samples was deposited from aqueous solution associated with dissolution of olivine and glass from the host basalts. The saponite forms pseudomorphs after olivine (Fig. 3c), with the saponite (001) cleavages oriented perpendicular to void spaces occurring as crystallographically oriented cracks in the olivine crystal (Fig. 3b). This observation suggests that orientations of the saponite crystals were controlled by that of the host olivine (Eggleton 1984; Delvigne 1998, p. 129; Wilson 2004). Saponite pseudomorphs after olivine are not surrounded by expansion cracks (Figs. 3b, 3f, and 3j), suggesting that replacement of olivine by saponite was isovolumetric. Because saponite contains substantial H2O and olivine does not, the replacement could not have been isochemical—a significant proportion of Mg, Fe, and Si from the olivine must have been removed into solution. Similarly, the chemistry of the saponite also shows that it did not form isochemically: saponite replacing olivine contains substantial Al, which must have come from outside the olivine, perhaps from the mesostasis glass. And, saponite replacing mesostasis glass contains substantial Mg, which would not have been abundant in the mesostasis; perhaps the Mg came from the olivine. In fact, the saponite has essentially the same composition throughout a sample, independent of its physical setting, which implies that all were deposited from an aqueous solution that allowed free chemical exchange across each rock. The mass lost from replacement of olivine and mesostasis with saponite may have been deposited in vesicles and cracks as more saponite.
It is puzzling that the CAS sample contains smectite in two textural varieties, coarse- and fine-grained, with essentially the same composition (Table 3) and separated in some areas by a quartz deposit. If this sequence of deposits represents three events or episodes of aqueous alteration, then the two varieties of smectite likely would have different chemical compositions, which they do not (Table 3). A possible solution to this difficulty is if the coarse-grained smectite was deposited from fluid into void spaces (as in the AMNH and MU samples), but that the fine-grained smectite replaced a pre-existing material that lined vesicles and cracks. It is possible, though perhaps impossible to prove, that the pre-existing material was glass from late-stage magma, mobilized within the cooling basalt to fill cracks and line vesicles. After this material solidified, quartz was deposited from solution to partially fill vesicles. Only then was the rock extensively altered to form smectite, both as deposits from solution and as replacements of olivine and glass.
Mössbauer spectroscopy shows that most of the iron in the tri-octahedral Griffith saponite samples is Fe3+. Most Fe3+-rich and Mg-poor smectites are di-octahedral and fall in the montmorillonite-nontronite series [i.e., I0–2(Al3+,Fe3+)4T8O20(OH)4·nH2O]. Thus, it seems likely that the iron in the Griffith saponite was not Fe3+ as originally deposited, but was Fe2+ and was subsequently oxidized. This inference is consistent with evidence that ferroan saponites oxidize rapidly in nature and in laboratory settings under atmospheric O2 (e.g., Badaut et al. 1985; Chuckanov et al. 2003; Neumann et al. 2011; Chemtob et al. 2014).
The formation of the Griffith saponite, as a low-temperature alteration product of basalt (and other mafic rock), is typical for iron-bearing saponite on Earth and elsewhere: “Ferromagnesian tri-octahedral smectites (saponite/iron saponite series) … are widespread in nature, especially as the main authigenic clay mineral produced by the alteration of oceanic and continental basalts and other basic volcanic material” (Güven 1988). The textural settings of the Griffith saponite are also typical for altered basalts, e.g., “…Fe-rich smectites are the dominant alteration products, occurring either as pseudomorphs or infilling veins and vesicles” (Walters and Ineson 1983). Among the many other reports of basaltic material altered to ferroan saponite or ferrosaponite, one can cite Kodama et al. (1988), April and Keller (1992), Robert and Goffé (1993), Köster et al. (1999), Parsatharathy et al. (2003), Mas et al. (2008), and Haggerty and Newsom (2003). Ferromagnesian tri-octahedral smectite is also found in deep-sea deposits (Bischoff 1972; Scheidegger and Stakes 1977; Badaut et al. 1985; Parra et al. 1985), which are, in a broad sense, also related to the alteration of basalts.
Ferromagnesian smectites are known in some martian rocks, specifically the nakhlite martian meteorites (Treiman 2005). The nakhlites contain ferromagnesian smectites that formed, on Mars, by aqueous alteration of olivine, siderite, and silicic glass (Treiman et al. 1993; Treiman 2005; Changela and Bridges 2011). The bulk chemical compositions of these alteration products are more ferroan and less aluminous than in Griffith saponite, and they do not normalize well into the formula for smectite: for Si+Al = 8, the sum of nominally octahedral cations far exceeds the limit of 6 for a tri-octahedral smectite. This discrepancy could suggest that the nakhlite saponites contain significant Mg in their interlayers, in addition to in their octahedral sites.
Ferromagnesian smectites are also found in other planetary settings, most notably in chondritic meteorites and interplanetary dust particles (IDPs). Chemical compositions of chondrite and IDP smectites vary widely, and include the Mg# of the Griffith saponite (i.e., its octahedral cations), but they typically have lower Al2O3 contents and higher Cr2O3 (e.g., Hutchison and Alexander 1987; Alexander et al. 1989; Rietmeijer 1991; Sakamoto et al. 2010; Tomeoka and Onishi 2011). Unlike the Griffith saponite, chondritic smectites generally have Na as their dominant interlayer cations.
Ferromagnesian tri-octahedral smectites can also form in other environments, including as direct precipitates and diagenetic products in alkaline and evaporitic settings (e.g., Jones and Weir 1983; Hover et al. 1999; Chukanov et al. 2003; Bristow et al. 2009). These formation mechanisms are not relevant to the formation of the Griffith saponite (Neuerburg 1951, 1953; McCulloh et al. 2002), but may be significant for Mars in general (Bristow and Milliken 2011) and the Sheepbed mudstone in particular (Grotzinger et al. 2014).
Implications for clay minerals in Yellowknife Bay, Gale Crater, Mars
Data developed here confirm that the clay minerals detected by CheMin in the Sheepbed mudstone are tri-octahedral smectites, likely saponites (Vaniman et al. 2014), and provide constraints on their crystal chemistry, oxidation state, and possible mode of origin. The smectite minerals in the Sheepbed mudstone are known primarily by their X-ray diffraction properties (Vaniman et al. 2014), and the Griffith saponite is a partial analog to the Sheepbed smectite minerals. The octahedral layers of the Griffith saponite and the Sheepbed smectites are similar in size, and thus in content of octahedrally coordinated cations, as shown by the similarities of their 02l diffraction bands. The location and shape of a smectite’s 02l diffraction bands vary according to whether the octahedral cation sites are all filled or not (tri-octahedral vs. di-octahedral), and the specific identities of those cations (Fig. 9). The locations of the Griffith saponite 02l diffraction bands, consistent with fully tri-octahedral smectites rich in iron, are consistent with their chemical compositions. Smectites in the Sheepbed mudstone have 02l diffraction bands of similar shapes and locations to those of the Griffith saponite; by analogy it seems likely that the Sheepbed smectites are ferrian saponites.
The XRD behavior Griffith saponite under humid and dry environmental conditions provides clues for the differences in locations and widths of the Sheepbed 001 diffraction peaks. In general, differences among the 001 diffractions arise primarily from differences in interlayer material (e.g., H2O, cations, and molecular species), The Griffith saponite samples all have sharp 001 diffraction peaks; after correction for Lorentz polarization, these peaks lie at ~15 Å in humid air, at ~12.8 Å after desiccation in dry N2, and at ~10 Å after desiccation on mild heating in dry N2, a variation consistent with progressive loss of interlayer H2O, with no change in interlayer cation. The Sheepbed saponite clay minerals have broad 001 diffraction peaks at ~10 and ~13.2 Å (corrected for Lorenz polarization) for John Klein and Cumberland, respectively. The same explanation can be used for the Sheepbed saponites, but it seems unlikely that the only difference between John Klein and Cumberland is the relative abundance of interlayer H2O. Alternate, more viable explanations include differences in interlayer cations (e.g., Mg vs. Ca) and pillaring. Interlayer material in smectites can reflect the smectite’s formation conditions and can be modified extensively by chemical exchange and reactions long after the smectites formed. Differences in formation and/or post-formation environments are in fact implied by variability in the amount of Ca-sulfate veins associated with the two martian saponites. Thus, the interlayer material in Sheepbed saponites retains clues to its formation and subsequent chemical processing, which may be understood through laboratory experiments on ferrian saponites.
The geological setting of Griffith saponite formation is also similar, but not identical, to that of the Sheepbed smectites in Yellowknife Bay. In both locations, the saponites formed in, and from, rocks of basaltic composition; and in both, olivine was replaced by the smectite (Vaniman et al. 2014). However, there are some differences. First, the host rocks for the saponites are not identical; both are of basaltic composition, but the Griffith saponite described here is from massive basalt while the Sheepbed smectites are in basaltic sediment. However, Griffith saponite is also reported in basaltic sediments of the Topanga Canyon Formation (Neuerburg 1953); these occurrences are under investigation. The difference between massive basalt and basaltic sediment may be unimportant chemically, but may be significant in terms of reaction rates (i.e., porosity, permeability, and specific surface areas). Second, the Sheepbed smectites are inferred to have formed (and/or been deposited) in diagenesis of lacustrine deposits (Grotzinger et al. 2014), while the Griffith saponite formed (apparently) during diagenesis in a marine environment (McCulloh et al. 2002). Differences in diagenetic environment could well account for some of the inferred differences in smectite interlayer compositions. Third, some Griffith saponite formed by replacement of glassy mesostasis material, but there is no evidence that smectite formation in Sheepbed mudstone consumed amorphous material. The amorphous material in the Sheepbed is poorly characterized, and its abundances are difficult to constrain through models of CheMin data because of the width of the amorphous material’s XRD peaks (Vaniman et al. 2014; Morris et al. 2013), so the significance of this difference is not clear. Finally, diagenesis of the Sheepbed mudstone apparently produced a significant proportion of magnetite (Fe3O4), although an igneous origin cannot be excluded. Magnetite is detected for the AMNH 89172 sample, but the oxide is associated with the basaltic separate and not the Griffith saponite separate.
Our XRD and VNIR measurements on the Griffith saponite provide a link between the XRD results for the smectite analyzed at Yellowknife Bay and the Fe-Mg smectites detectable from martian orbit by VNIR hyperspectral imaging instruments, i.e., the Compact Imaging Spectrometer for Mars (CRISM) instrument on the NASA Mars Reconnaissance Orbiter (Murchie et al. 2007) and the Observatoire pour la Mineralogie, l’Eau, les Glaces et l’Activité (OMEGA) instrument on the Mars Express orbiter (Bibring et al. 2005). Note, however, that the Sheepbed saponites do not constitute ground truth for orbital observations because clay minerals are not detected at the Sheepbed location (e.g., Milliken et al. 2010).
In CRISM spectra, Fe-Mg smectites are identified by the presence of spectral features centered near 1.9 μm (molecular H2O), near 2.3 μm (2.29–2.32 μm), and near 2.4 μm (e.g., Ehlmann et al. 2009). The spectral band near 1.4 μm should be, but is not always, detected (Ehlmann et al. 2009). For the ~2.3 μm band, a center near 2.29 μm is inferred to represent (Fe3+)2-OH in di-octahedral smectite (nontronite), and a band centered near 2.31–2.32 μm is inferred to represent (Fe3+,Fe2+,Mg)3-OH in low-Fe tri-octahedral saponite (SapCa-1) and hectorite (a Li-bearing smectite). A band centered near 2.30 μm is assigned to Fe-Mg smectite without specification of either the Fe redox state or the number of cation sites filled in the octahedral layers (e.g., Ehlmann et al. 2009). Our results for Griffith saponite show that the smectite detected at Yellowknife Bay by XRD and the smectite detections by CRISM and OMEGA that have band centers in the range 2.30 to 2.32 μm are all candidates for assignment to ferrian saponite with a range of Fe3+/∑Fe values. VNIR detections having band centers near 2.30 and 2.32 μm are the most oxidized and reduced, respectively.
We are grateful to George Harlow and the American Museum of Natural History (New York) for a sample of the Griffith saponite. Our study was assisted by D.K. Ross (EMP analyses) of Jacobs Engineering, at the Johnson Space Center. We are grateful to the ARES Division, Johnson Space Center for access to the SX-100 microprobe under a cooperative agreement with the LPI. Reviews by M.D. Dyar and J. Bridges were constructive and helpful. This work was supported by NASA grants through the Mars Science Laboratory Mission. LPI Contribution no. 1784.
↵† Special collection papers can be found on GSW at http://ammin.geoscienceworld.org/site/misc/specialissuelist.xhtml.
- Manuscript Received October 1, 2013.
- Manuscript Accepted April 29, 2014.