A combined electron microprobe (EMP) and Raman spectroscopic study of the alteration products in Martian meteorite MIL 03346



[1] We examine the secondary alteration products in MIL 03346 using Raman spectroscopic and electron microprobe traverses. Discussion focuses on the single olivine in ,177 supplemented with observations from ,168 and ,169. Traverses start at the rim and progress into the interior. Dark brown, nearly opaque, laihunite [Fe2+Fe3+2(SiO4)2] is present as overgrowths, and 20–50 µm veins of reddish-brown stilpnomelane [(K,Na,Ca)4(Ti0.1,Al2.3,Fe3+35.5,Mn0.8,Mg9.3) (Si63Al9)(O,OH)206∗n(H2O)] occur inside the olivine. Stilpnomelane crosscuts and postdates the laihunite; veins are in sharp contact with the host olivine but lined by ~5 µm films of jarosite [KFe3+3(SO4)2(OH)6] from a later generation of alteration. An interstitial laihunite also hosts stilpnomelane. The most recent secondary phases are gypsum and bassanite in our X-ray maps of ,168 and ,169. Ca-sulfates were not observed in X-ray maps of ,177 but were detected in our Raman point count. All sulfates are believed to be Martian. The groundmass of MIL indicates rapid cooling from elevated temperatures with fO2 near QFM. Reports of laihunite synthesis by olivine oxidation at elevated temperatures (100–800°C) suggest the overgrowths formed during consolidation. In terrestrial rocks, stilpnomelane is a product of late diagenesis to garnet-grade metamorphism. In MIL, stilpnomelane appears to be a secondary phase formed at the lower end of this stability range, at conditions akin to diagenesis. Raman spectra indicate that the stilpnomelane, jarosite, and Ca-sulfates are hydrated. The stilpnomelane contains Cl and was followed by jarosite, a product of acid alteration, and the deposition of Ca-sulfates and halide salts from more neutral chloride solutions.

1 Introduction and Purpose of Research

[2] Kuebler et al. [2013, this issue] studied the iddingsite alteration reactions of two terrestrial basalts and compared these to the alteration products in the Martian shergottite Allan Hills (ALHA) 77005. Alteration products in the nakhlite meteorites are also referred to as “iddingsite” and purported to contain smectites, Fe oxides, and oxyhydroxides, but the nakhlite alteration products have a different mode of occurrence and contain greater proportions of phyllosilicate [Bunch and Reid, 1975; Gooding et al., 1991; Treiman et al., 1993; Bridges et al., 2001]. The alteration products in Nakhla and Lafayette have been dated at 650 (Ar-Ar) and 680 Ma (Rb-Sr) [Shih et al., 1998; Swindle and Olson, 2002], significantly younger than the host clinopyroxenite, and presumably formed by interaction with meteoric groundwater. Nakhlite alteration products fill veins with sharp, distinct boundaries instead of topotactically replacing the olivine, implying that the olivine crystal structure is completely disrupted, whereas the Lunar Crater (LC), Mauna Kea (MK), and ALHA iddingsite inherits part of its structure from the host olivine and has the same extinction angle as its host olivine [Brown and Stephen, 1959]. We consider the nakhlite alteration products to be sufficiently different from the iddingsite samples to merit distinction.

[3] All of the nakhlites are clinopyroxenites with similar radiogenic and cosmic-ray exposure ages and are thought to derive from different depths of a single cumulate pile [McKay and Schwandt, 2005; Treiman, 2005; Mikouchi et al., 2006]. The mesostasis of MIL 03346 (MIL) hosts plagioclase, laihunite, cristobalite, and magnetite crystals whose forms indicate rapid crystallization and is thought to be one of the shallowest-formed nakhlites [McKay and Schwandt, 2005; Mikouchi et al., 2006]. The mesostasis is glassy in areas but altered in others, including jarosite and iron staining around the magnetite. Few olivines are present and are irregularly distributed throughout the meteorite. The olivine has ~Fo43 core compositions but are zoned to fayalite (Fo7–11) at their rims. Most of the secondary alteration veins in these MIL sections differ in form from those of Lafayette and contain a different secondary phase that we identify as stilpnomelane. The stilpnomelane veins have a smooth morphology (see Figures 1a–1f) while the saw-tooth olivine alteration veins of Lafayette pictured in [Treiman et al., 1993; Treiman, 2005] host secondaries with a micaceous morphology. Changela and Bridges [2010] analyzed more complex secondary alteration veins in Lafayette that contain both smooth and micaceous alteration products. Likewise, Figure 1 of Rao et al. [2005] pictures smooth vein-filling alteration products in Nakhla. The relationship between the smooth and micaceous phyllosilicates is unclear at present, but it may be that the smooth alteration products occur metastably and reorganize themselves into micaceous materials. The most recent secondary phases are the gypsum and bassanite that fill late-stage fractures and clearly postdate all other secondary phases (see Figure 11). All of the sulfates are inferred to be Martian: the jarosite because it predates the Ca-sulfates and the Ca-sulfates because they occur in both Antarctic and non-Antarctic nakhlites.

Figure 1.

(a) Transmitted light photo (PPL, 20 X) of reddish-brown alteration veins and greenish-brown jarosite in an olivine of MIL 03346,168. (b) BSE image indicating the location of traverse 3 on the single olivine in ,177. Traverse 3 crossed two alteration filled veins, two large cracks, and one smaller crack. The width of the laihunite-bearing rim is between 4 and 27 µm wide (see text). The main secondary phase filling the veins is stilpnomelane as identified by Raman spectroscopy (dark in BSE; see text for discussion regarding its identification), but both alteration veins are lined with jarosite (lighter than stilpnomelane in BSE but darker than olivine). (c) PPL photo (20×) of an olivine in ,169 with reddish-brown stilpnomelane veins transecting darker brown laihunite overgrowths and overprinted by a later generation of jarosite alteration. (d) BSE close-up of jarosite crosscutting the stilpnomelane in the first alteration vein of traverse 3. (e) PPL photo (20×) of stilpnomelane, laihunite, and jarosite-bearing olivine in ,169 adjacent to glassy (quenched) matrix materials; note that laihunite overgrowths do not form continuous rims around the olivine. (f) BSE image of incipient saw-tooth veins located near profile 1 on ,177. (g) PPL photo (10×) of an interstitial laihunite grain in ,168 with relict olivine (two transparent spots) and veins of stilpnomelane. (h) Reflected light photo (10×) of the interstitial laihunite shown in Figure 1g with the stilpnomelane veins and olivine remnants labeled. (i–l) Respectively, Mg, Si, Fe, and Mn X-ray maps of the interstitial laihunite shown in Figures 1g and 1h; maps demonstrate the composition of the laihunite and stilpnomelane relative to the adjacent clinopyroxene (laihunite contains less SiO2 than the pyroxene and is zoned with respect to FeO, MgO, and MnO; stilpnomelane has SiO2 and MnO contents similar to pyroxene but less MgO and more FeO).

[4] As in Kuebler et al. [2013], we review the phase identifications made by Raman spectroscopy, document the alteration reactions and secondary mineral occurrences using Raman spectroscopic and electron microprobe (EMP) traverses (or single-line X-ray profiles), and use these data with petrographic observations to constrain the sequence and conditions of alteration on Mars.

2 Analytical Techniques and Methods

[5] For this study, laser Raman spectroscopy and electron microprobe (EMP) images, X-ray profiles, and spot analyses were made on polished thin and thick sections. Spectra were collected on a HoloLab 5000® (Kaiser Optical Systems, Inc.) spectrometer using a partially unpolarized 532.3 nm Nd:YAG laser with a 6 µm beam as the excitation source. The spectral region is set to 0–4370 cm−1 by the HoloPlex grating and a super-notch filter used to reject the reflected laser light. Manual traverses were made across olivine alteration veins in sections ,168 and ,177 of MIL using 1–3µm steps. The absolute wavelength scale of the CCD camera was calibrated using a neon arc lamp and a cubic spline routine in the Holograms® software to smooth and fit the spectrum; the resulting calibration curve has a standard deviation of 0.003 nm/pixel. The absolute frequency of the laser can be calibrated against a silicon wafer or cyclohexane; cyclohexane is preferred because it has multiple peaks across the analyzed spectral region. The spectra in this study were collected with the laser frequency calibrated against a silicon wafer. The silicon wafer is also used to monitor the laser wavelength at the start and end of each working day and demonstrate slight deviations in the laser frequency so the olivine peak positions are corrected using the measured peak position of the silicon wafer and a standard value of 520.5 cm−1 before plotting them on the olivine calibration curve. The olivine calibration curve was created with data collected when the laser wavelength was calibrated to cyclohexane and may explain why the data points in Figure 7 fall slightly below the median calibration curve. The CCD camera has a spectral resolution (peak separation) of 3 pixels or 6.2 cm−1.

[6] The alteration products were analyzed by making manual traverses from the rim of the olivine through the veins and into the core (“manual” traverses are made without using the subroutines in the stage mapping panel of the Holograms software so that the microscope can be focused as often as desired). Three manual traverses were made on the single olivine in thick section ,177 (manual traverses 1, 3, and 4) and two additional traverses on thin section ,168, but epoxy peaks hindered the use of these spectra for detailed analysis of the alteration products. One automated traverse (traverse 2, also known as Raman point count [Haskin et al., 1997]) was made across the entire width of thick section ,177 to determine the relative mineral proportions (194 useful spectra from a total of 202 spectra); these data are presented in Table 1. Our jarosite and apatite abundances may be overestimated; these phases frequently occur with plagioclase, which may have minor peaks at similar Raman shifts—the presence of OH peaks at appropriate Raman shifts confirms their presence at some locations.

Table 1. Estimated Mineral Abundances from Traverse 2 on MIL Thick Section ,177
  • a

    cpx, clinopyroxene; ol, olivine; laih, laihunite; plag, plagioclase; mgt, magnetite; ilm, ilmenite; cris, cristobalite; Ca-sulf, gypsum and bassanite; jaro, jarosite; apat, apatite; unid., unidentified phyllosilicate; hem, hematite.

  • b

    There is only one olivine in thick section ,177 accounting for the lower abundance of olivine in our analysis relative to McKay and Schwandt [2005], Day et al. [2006], and Stopar et al. [2005].

Mineral phaseacpxollaihplagmgt + ilmstilpcrisCa-sulfjaroapatunid.hemsum
Mineral proportionsb72.770.861.206.295.430.600.993.

[7] All of the olivine traverses were acquired using a 64 s accumulation time, 20× long-working distance objective (0.4 NA), and low laser power (≤6.0 mW) because of the potential for oxidizing some alteration phases [De Faria et al., 1997; Wang et al., 1998, 2004]. A higher laser power (~9 mW) was used for the Raman point count on ,177. The olivine doublets were curve fit using the least squares fitting subroutine of the Grams AI software with a mixed Gaussian-Lorentzian peak shape and linear baseline. The constraint-free iteration option was used for fitting all parameters until convergence or a minimum was attained. Table 1 lists the samples and parameters used to collect each Raman traverse and corresponding EMP traverse.

[8] Electron microprobe (EMP) traverses and X-ray maps were collected on thin sections ,168 and ,169 of MIL using the Cameca SX 100 at the Johnson Space Center during the summer of 2006. Because samples ,168 and ,169 had been carbon coated, Raman spectra were collected on a third sample—thick section ,177. Additional profiles (single-line element maps) and point analyses were collected on thick section ,177 using the JEOL 8200 electron microprobe at Washington University in St. Louis. Samples ,168 and ,169 were made from ,166 (an exterior sample whose parent is ,3) and ,177 from potted butt ,28 which includes both interior and exterior materials. A 15 kV accelerating voltage was used with a 25 nA beam current and 5 µm spot size for olivine point analyses made near Raman traverses 1 and 3 on ,177, but most vein-filling materials (stilpnomelane and jarosite) were analyzed in spot mode (1 µm spot size). A 15 kV accelerating voltage was used with a 10 nA beam current and 5 µm spot size for microprobe traverses a–h on ,168. All quantitative analyses (13 elements on ,177; 16 elements on ,168 and ,169) used a combination of silicate, metal, and oxide standards plus anhydrite or troilite for S, apatite for P, tugtupite for Cl, and fluorite for F. X-ray matrix effects were corrected using a PAP correction routine on the Cameca SX 100 and ZAF methods incorporated into the Probe-for-Windows control software on the JEOL 8200 microprobe [Armstrong, 1988; Pouchou and Pichior, 1991]. EMP analyses and Raman spectra were collected in similar locations on MIL ,177 but not as carefully located as those of Kuebler et al. [2013].

3 Sample Analysis

3.1 Identification of Alteration Phases by Raman Spectroscopy

3.1.1 Laihunite

[9] Olivine in MIL are unusual in that they contain laihunite [Fe2+Fe3+2(SiO4)2], secondary stilpnomelane [(K,Na,Ca)4(Ti0.1,Al2.3,Fe3+35.5,Mn0.8,Mg9.3)(Si63Al9)(O,OH)206∗n(H2O)], and jarosite [KFe3+3(SO4)2(OH)6] (see reference spectra in Figure 2). The identification of these phases requires some discussion because laihunite was only tentatively identified by Rost et al. [2006], and the stilpnomelane identification differs from the usually inferred smectite/saponite/illite. We initially identified laihunite in ,177 by comparing our Raman spectra of the olivine rims with the published standard spectrum [Kuebler et al., 2006] and ratioing the signal to noise of the 570 cm−1 laihunite peak to that of the olivine doublet (see Figure 3). Laihunite is monoclinic (whereas olivine is orthorhombic) and therefore has a different spectrum; the major laihunite peaks occur at 312, 570 with a shoulder near 595 and 896 cm−1. All of these peaks are present in the spectra of the olivine rim, but the 570 cm−1 peak is strongest (most apparent) and easiest to use for establishing the presence of laihunite; the laihunite peak at 312 cm−1 overlaps the olivine peak at 300 cm−1, and the laihunite peak at 896 cm−1 occurs between the olivine doublet and the group of olivine peaks between 850 and 1000 cm−1.

Figure 2.

Reference spectra of the mineral phases identified in MIL 03346. From top to bottom, they are as follows: (a and b) Laihunite [Fe2+Fe3+2(SiO4)2] and coexisting fayalite from Lai-He, China—sample belongs to A. Wang; (c) Fo41 olivine from Rustenburg, S. Africa—sample belongs to A. Hofmeister (Harvard sample #118652); (d) stilpnomelane [(K, Na, Ca)4(Ti0.1, Al2.3, Fe3+35.5, Mn0.8, Mg9.3)(Si63Al9)(O,OH)206∗n(H2O)] from Friedericke bei Weilburg, Nassau, Germany (sample once belonged to Michigan School of Mines); (e) cristobalite [SiO2] in a porphyritic augite andesite core from Honeybee Robotics; (f) Iconofile jarosite [KFe3+3(SO4)2(OH)6] pigment sample; (g) ferrihydrite [Fe5O12H9] powder from Jownsuu, Pohjoiskarjala, Finland; this ferrihydrite was oxidized by the Raman laser even at very low laser powers.

Figure 3.

Spectra from the olivine rim at the start of traverse 3 on ,177. Laihunite, an Fe3+-bearing olivine, is indicated by the relative strength of the 570 cm−1 peak versus that of the olivine doublet. Laihunite is inferred to be present where this ratio is >0.5 and may be present in spectra whose ratio is between 0.2 and 0.5 (04160707 to 04160710). The top spectrum is the first in the traverse and the distance of each subsequent spectrum from the starting point provided in µm on the y axis. An additional peak (indicated by a gray asterisk) is present in spectrum 04160732 and occurs near 665 cm−1.

[10] Our standard Fo40 olivine spectra have a minor peak near 570 cm−1 so the presence or absence of laihunite cannot be evaluated according to the presence or absence of the peak. The peak position of this minor peak in 13 standard olivine spectra is usually between 569.0 and 571.3 cm−1 but may be as high as 582.4 and 582.8 cm−1, depending on crystal orientation, so that peak position is also an unreliable means for discriminating the presence of laihunite when the signal is weak (the laihunite peak mixes with the minor olivine peak). In the standard Fo40 spectra, the ratio of the height of this minor olivine peak to the height of the doublet is, on average, 0.04 and never exceeds 0.08, so we base our interpretation of the presence of laihunite on the relative signal-to-noise ratio of the 570 cm−1 peak versus that of the olivine doublet (using whichever peak in the doublet is taller, usually the 820 cm−1 peak). As a set, the MIL traverses have lower signal-to-noise ratios than the standard olivine spectra with background values for this ratio ranging from 0.07 to 0.11. We interpret laihunite to exist at the rims of the olivine where this ratio exceeds 0.5; areas having values between 0.2 and 0.5 are deemed transitional. It is cautioned that the jarosite lining the alteration veins also has a peak of moderate strength at this position, so that laihunite can only be interpreted as present when the other jarosite peaks are relatively weak. Laihunite is also present in the groundmass of MIL. The laihunite domains in the terrestrial standard are more oxidized than the MIL olivine overgrowths such that the laihunite peaks dominate the spectra (ratio ≫ 1) and appear to completely replace the olivine [Kuebler et al., 2011].

3.1.2 Jarosite

[11] MIL 03346 jarosite have Raman spectral patterns similar to that of a jarosite pigment sample (see Figure 2) and an altered Hawaiian tephra sample, HWMK515, identified as nearly pure natrojarosite by XRD [Ming et al., 1996]. The MIL jarosite spectra have an H2O peak centered between 3410 and 3420 cm−1 and occasionally have a shoulder near 3495 cm−1; these peak positions are similar to the jarosite in ALHA 77005. Several jarosite reference spectra are presented in Figure 13 of Kuebler et al. [2013] and the Raman spectrum of jarosite discussed in context with other Fe sulfates by Chio et al. [2005]. XRD of the pigment sample indicates hydronium jarosite [(H3O)Fe3+3(OH)6(SO4)2], but alkali jarosites are hard to distinguish from hydronium jarosite by XRD [Dutrizac and Jambor, 2000]. We identify the sulfate lining these veins as K-Mg jarosite on the basis of our EMP data. Clark et al. [2005] report the jarosite of Meridiani Planum to be Na-K jarosite according to Mössbauer data but indicate that the APXS data are not consistent with this conclusion. Ca-sulfates occur in fractures and are more sparsely distributed than the jarosite; they crosscut and appear to postdate all other forms of alteration in MIL ,168 and ,169. Gypsum and bassanite, originally identified by EMPA, were confirmed by Raman spectroscopy [Kuebler et al., 2007; reference spectra in Wang et al., 2009]. Ca-sulfates were not observed in the X-ray maps on ,177, but weak Ca-sulfate peaks are present in multiple spectra from the automated Raman point count suggesting its presence in the groundmass.

3.1.3 Stilpnomelane

[12] Spectra of the reddish-brown silicate veins in MIL were run through a spectral search/match routine (after baseline subtraction) that identified stilpnomelane as a potential match. The chemical composition of this material, as indicated by EMPA, is consistent with stilpnomelane in the WU Raman collection and also with published stilpnomelane analyses (see Tables 2 and 3) [Deer et al., 1966; Eggleton, 1972], but its optical characteristics are not consistent with canonical stilpnomelane (low relief, no cleavage traces). According to our supplier (Excalibur), the Raman standard on which our identification was made is from Friedericke bei Weilburg, Nassau, Germany (a locality mentioned in the 1868 edition of Dana) and once belonged to the mineral collection of the Michigan School of Mines. The Raman peaks of the reddish-brown silicate are broad like a glass, but the XRD pattern of the Nassau standard matches that of stilpnomelane and demonstrates its crystallinity (see Figures 4 and 5). The breadth of the Raman peaks may be due to the large size of the stilpnomelane unit cell, which has a modulated (island-like) 2:1 structure [Eggleton, 1972; Eggleton and Chappell, 1978; Guggenheim and Eggleton, 1987, 1994]. By analogy with spectra of strongly zoned minerals, we infer that minor differences in the bond lengths of the modulated crystal structure produce translational disorder and overlapping mode frequencies in the resulting Raman spectra (i.e., broad peaks) [Kuebler et al., 2006].

Table 2. Samples Used in Study, Raman and EMP Data Collection Parameters
Sample, thin SectionTraverseaDate CollectedLaser pwr (in mW)Acc. time# of Spectra in TraverseStep Size (in um)Corr. Traverse or ProfileDate CollectedQuantitative AnalysesAcc. Voltage (keV)Beam Current (nA)Spot Size (in um)
  1. a

    Raman traverses 1, 3, and 4 were manual traverses made on the single olivine in thick section ,177. Traverse 2 was an automated traverse (Raman point count) across the width of the thin section (manual traverses permit user to focus on sample as often as desired versus automated traverses where sample is only brought into focus once).

MIL 03346127 January 20076.04 s × 164201.0228 May 200819 June 200815251
,177229 January 20079.24 s × 1615399.8n/an/an/an/an/an/a
 316 & 17 April 20075.94 s × 162461.4328 May 200819 June 200815251
 416 & 17 July 20075.54 s × 162781.4128 May 200819 June 200815251
,168111 & 12 June 20106.04 s × 161761.4a04 August 200615105
 b04 August 200615105
 c04 August 200615105
 d4 & 5 August 200615105
 215 June 20106.04 s × 16812.7e11 August 200615105
 f11 August 200615105
 g11 & 12 August 200615105
 h12 August 200615105
 Unalt olivine12 August 200615105
Laihunite #73110 October 20104.210 s × 10411.4None22 October 201015251
Lai-He, China217 October 20105.110 s × 10861.4None22 October 201015251
Nassau stilpnomelaneNone08 July 20029.35 s × 10Std spectrumNone22 October 2002152010
AW #1121 stilpnomelaneNone12 July 200014.55 s × 10Std spectrumNone19 February 2002153010
Table 3. Four Best Stilpnomelane Matches Reported in Extended Volumes of Deer, Howie, and Zussman [1966], Reduced with the More Recent Structural Determination of Eggleton [1972]a
Analysis #3456
LocationZuckmantel,Zuckmantel,Anna Mine,Kangerdlussuaq,
 SilesiaSilesiaBaern,E. Greenland
Original referenceHutton [1938]Hutton [1938]Hutton [1938]Wager and Deer [1939]
  1. a

    Extended volumes reduced stilpnomelane analyses using eight oxygen per formula unit, but the average formula of 37 samples reported by Eggleton [1972] is (Ca,Na,K)4(Ti0.1Al2.3Fe35.5Mn0.8Mg9.3)[Si63Al9](O,OH)206∗nH2O. We reduced the stilpnomelane analyses using 11.25 oxygen per formula unit; excess alumina was applied to the ferromagnesium sum. Note difference in alkali contents relative to Table 4 (tr = trace).

Si + Al4.0004.0004.0004.000
Ti + Fe2+ + Fe3+ + Mn + Mg + excess Al2.6522.6302.6462.441
Ca + Na + K0.2560.2060.2870.355
Sum cations6.9086.8376.9336.796
Figure 4.

Comparison of the Raman spectra acquired on four standard stilpnomelane to that of the MIL vein-filling material. EMP data are available for only two of these and are included in Table 3. (a) #79-10-6 and (b) #1121 belong to the private collection of A. Wang, (c) is a stilpnomelane from Laytonville, CA, and (d) is a quartz-stilpnomelane schist from Nassau, Germany. (e) Spectrum of the reddish-brown vein-filling materials in vein 1 of traverse 3 on the olivine in MIL thick section ,177. The Nassau sample provides the closest match to the MIL alteration product; their Si-Ob-Si peaks are about 10 cm−1 lower than those of the two Chinese standards but have different Si-Onb peak positions; the Si-Onb peak position of the MIL stilpnomelane is similar to that of AW #79-10-6, while the Si-Onb of the Nassau sample is closer to that of AW #1121. The Laytonville spectrum is weak but suggests the presence of three component peaks in the Si-Ob-Si region: near 560, 600, and 680 cm−1. The 467 cm−1 peak in the Laytonville spectrum may be the vibration producing the ~500 cm−1 shoulders in the other three stilpnomelane or could belong to quartz which has a peak near 465 cm−1. AW # 1121 has no good H2O peak due to a high background at high wavenumbers.

Figure 5.

Comparison of the Nassau stilpnomelane XRD pattern with two stilpnomelane XRD patterns from the Jade database (PDF cards #00-045-1355 and 00-025-0174) and an XRD pattern of quartz (PDF card #00-002-0458). Lines shown in green correlate with stilpnomelane and those in red with quartz.

[13] Figure 4 is a comparison of the Raman spectra acquired from four stilpnomelane standards (the Nassau standard; a stilpnomelane from Laytonville, CA; plus two from China) and the stilpnomelane spectrum from vein 1 of traverse 3 on MIL ,177 with the fewest peaks attributable to other phases. EMP data for two of the Raman standards are included in Table 3. Spectra from the Nassau stilpnomelane most closely resemble the stilpnomelane in MIL; the main feature of its spectrum is the broad Si-Ob-Si (bridging oxygen) peak near 590 cm−1. The Si-Onb (non-bridging oxygen) peaks near 904 and 1156 cm−1 are less intense, as is typical of polymerized structures (Onb < 2) [Wang et al., 1994] and the H2O peak at 3592 cm−1 rather weak. The stilpnomelanes in MIL have Si-Ob-Si peak shifts similar to the standards (peak center ranges from 578 to 595 cm−1 and sometimes suggest the presence of two unresolved peaks, even when jarosite is not present) and are chemically similar to the standards but contain ~5 wt % less alumina, ~5 wt % more FeOtot and have lower probe totals. The stilpnomelanes in MIL have relatively strong, broad H2O peaks centered between 3568 and 3579 cm−1 that sometimes appear as a shoulder on the jarosite H2O peak. The Chinese stilpnomelane standards have spectra similar to the Nassau standard, but their Si-Ob-Si and Si-Onb peak positions are higher and the lattice vibrations (peaks below 450 cm−1) are lower. AW #79-10-6 has an H2O peak position similar to that of the Nassau sample (3590 cm−1), but the H2O peak of AW #1121 is swamped by a high background. Three distinct Si-Ob-Si peaks (560, 600, and 680 cm−1) are apparent in the Laytonville spectrum, but this is a relatively weak spectrum. These may be the component peaks of the broader Si-Ob-Si features observed in the other stilpnomelane standards. Likewise, the 467 cm−1 peak may represent the vibration responsible for the ~500 cm−1 shoulder in the spectra of the three other standards (or may belong to quartz). The 930 cm−1 Si-Onb peak position of the Laytonville stilpnomelane resembles the two Chinese stilpnomelane, but the 1154 cm−1 peak position is closer to that of the Nassau stilpnomelane. The Laytonville stilpnomelane has an H2O peak position similar to that of MIL (3567 cm−1).

[14] Spectra acquired in the stilpnomelane veins of MIL have different background shapes than the spectra comprising the rest of the traverse; there may have been some distortion of the relative peak heights during baseline subtraction, but we maintain this mineral identification on the basis of the quality of the Raman and XRD matches and the similarity in chemical composition with the WU Raman standards (see Table 3) [Kuebler et al., 2011] and published compositions despite the atypical optical characteristics. Gooding et al. [1991] note that their average Nakhla “rust” composition agrees equally well with stilpnomelane and dioctahedral smectites but indicate two individual analyses to resemble smectite and two others to have affinities to chlorite [(Mg,Fe2+,Fe3+,Mn,Al)12(Si,Al)8O20(OH)16] or vermiculite [(Mg,Ca)0.6-0.9(Mg,Fe3+,Al)6[(Si,Al)8O20](OH)4∗nH2O] [see also Gooding, 1985]. Historically, the smectite interpretation has been preferred. The major Si-Ob-Si bands of the stilpnomelane occur ~50 cm−1 lower than those of the coarse alteration products in Lafayette [see spectrum in Kuebler et al., 2004]. The stilpnomelanes in MIL are even more Al deficient than the Lafayette phyllosilicates; this concurs with reports of stilpnomelane occurring in Al-poor environments [Guggenheim and Eggleton, 1987]. The Fe-rich trioctahedral phyllosilicate of Kuebler et al. [2004] is equivalent to the fine-grained phyllosilicate of Treiman et al. [1993; see also Wang et al., 2002] and, based on morphology, may be the amorphous silica gel of Changela and Bridges [2010].

3.2 Changes Along a Typical Raman Traverse (Traverse #3)

[15] Traverse 3 started at the rim of the olivine in thick section 177 and progressed into its interior, so the first alteration vein is the one closest to the edge of the olivine in Figure 1b. A second, narrower vein occurs further into the olivine. The first four spectra in the Raman traverse have an unusually strong peak at 570 cm−1, attributed to the Fe+3-bearing laihunite (see Figures 2 and 3). The width of the laihunite-bearing rim along this traverse is between 4 and 27 µm, depending on the value of the ratio of peak signal-to-noise ratios chosen to define its presence. The ratio of the first four spectra is >0.5 and does not drop below 0.2 until ~13 µm into the traverse but varies up and down and does not drop below and stay below 0.2 until 27 µm into the traverse. By similar estimates, the width of the laihunite rim is between 10 and 20 µm wide in traverse 1 and between 25 and 40 µm wide in traverse 4.

[16] Olivine doublets are generally strong throughout traverse 3 with an average signal-to-noise ratio above 800. The olivine doublets start diminishing about 17 µm ahead of the leading edge of the first alteration vein (see Figure 6, spectrum 04160784), an indication that the olivine is altered, are weak in nine or ten spectra, and occur as remnant features in two spectra at the leading edge of the alteration vein. Very weak doublets are apparent in many of the spectra collected inside the vein and probably represent olivine at depth in the thin section. Jarosite peaks occur throughout most of the traverse but are strongest and only dominate the spectra acquired in proximity to, or just inside, the alteration veins. Jarosite peaks are weaker and more sparsely distributed in the interior of the olivine. Ten spectra inside the first alteration vein have the Raman spectral pattern of stilpnomelane (in which the ~595 cm−1 peak is stronger than the ~570 cm−1 jarosite peak with a subordinate peak near 305 cm−1). A few extra peaks are present in the stilpnomelane spectra (683, 714 cm−1) and are assumed to belong to another incipient phyllosilicate. The olivine doublet reappears quickly on the opposite side of the first alteration vein, dominating the spectra before the traverse emerges from the vein (trailing alteration front occurs at 04160801). A similar reaction was observed in the spectra approaching the second vein; olivine doublets start to diminish 10 µm ahead of the vein, and jarosite appears 4 µm ahead of the vein but only dominates the two spectra adjacent to the leading edge of the second vein. Again, weak olivine doublets and remnants occur throughout the alteration and reappear rapidly at the opposite edge of the vein (in the last two spectra acquired in the vein). The stilpnomelane pattern dominates the four spectra acquired inside the second alteration vein. Three spectra in the middle of the second vein contain hematite peaks but may belong to a phase that oxidized under the laser to produce hematite-like peaks [Wang et al., 2004]. Photos acquired during analysis suggest that some of the materials in the second vein oxidized under the laser. Oxidation might be an indication of ferrihydrite or jarosite. Ferrihydrite is reported to occur in Lafayette by Treiman et al. [1993] and the WU ferrihydrite standard oxidized under a green laser at laser powers as low as 0.4 mW (but the stilpnomelane did not, even at laser powers as high as 15 mW). Alternatively, the phase being oxidized by the laser could be jarosite, as was observed in the iddingsite alteration of ALHA 77005 [Kuebler et al., 2013].

Figure 6.

Spectral changes across the first alteration vein in traverse 3. The alteration fronts occur at 04160784 and 0801 (per reflected light microscopy). Spectra 0787, 0793, 0796, and 0799 were collected inside alteration vein 1; all other spectra were collected in the adjacent olivine on either side of the vein. Peaks belonging to olivine, jarosite, and the 570 cm−1 peak (here the 570 cm−1 peak belongs to jarosite) are identified by black, dark gray, and light gray asterisks, respectively. Spectrum 04160793 is representative of the stilpnomelane spectral pattern in MIL, but jarosite peaks and a weak olivine doublet are also present. Spectral changes across the vein are asymmetric in that the distance over which the olivine doublet diminishes upon approach to the alteration vein is wider than the distance over which it reappears on the opposite side of the vein. Jarosite is apparent inside the olivine in transmitted light and the jarosite film at the leading edge of the vein thicker than the opposite side. Distances are reported relative to the starting point of the traverse (1.414 µm steps).

3.3 Microprobe Results on Thick Section ,177 and the Laihunite Standard

3.3.1 Microprobe Results on MIL Fayalite and Laihunite Versus Standard Laihunite

[17] The single olivine crystal in thick section ,177 is zoned in both Fe and Mg with laihunite and fayalite (Fo11) occurring at the rims and Fo43 in the cores. The range in electron microprobe compositions correlates well with the compositional range implied by the olivine doublets plotted on the Raman calibration curve of Kuebler et al. [2006]; see Tables 4 and 5 and Figure 7. Both quantitative analyses and X-ray profiles (single-line X-ray maps) were acquired on MIL, but only quantitative analyses were collected on the laihunite standard (AW #73 from Lai-He, China; see Table 4). The laihunite in MIL is obvious in plane-polarized light (PPL), but the Fe-Mg zoning is too strong and no clear boundary exists between the olivine and laihunite so that its presence is not obvious in bask-scattered electron (BSE) images. The olivine and laihunite oxide totals are both high in MIL (ranging from 98.7 to 102.3 wt % with only two analyses below 100%), so we could not discern the presence of laihunite in the EMP analyses by varying the Fe2+/Fe3+ ratios to optimize the cation sums. The EMP analysis of the rim of the olivine in ,177 does, however, resemble the analysis of the standard laihunite in Table 4. Laihunite in the standard has less FeOtot and lower oxide totals than the coexisting fayalite, reflecting the presence of Fe3+. There is a clear boundary between the laihunite-bearing and host fayalite domains of the standard, so both phases can be discerned in reflected light (laihunite is brighter than fayalite in reflected light) and BSE images, as well as in plane-polarized light (PPL). EMP analyses of the standard laihunite could be reduced using the published formula [Dyar et al., 1998], indicating that the standard has the proper Fe2+/Fe3+ ratio.

Table 4. EMP Analyses of Secondary Phases Encountered in the Olivine of MIL Thick Section ,177 and Thin Section ,168a
Line #404142434445464724    
Phase Id.JarositeJarositeStilpJarositeJarositeStilpJarositePoss. chlStilpStilpStilpStilpStilp
Notespt 1pt 1pt 2pt 3pt 1pt 2pt 3pt 1 MIL ,168MIL ,168RamanRaman
 unk 27unk 28unk 29unk 30unk 31unk 32unk 33unk 34unk 13travs e–htrav dNassau stdAW#1121
 vein 1vein 2vein 2vein 2vein 3vein 3vein 3vein 4trav 1 veinave. (N = 58)ave. (N = 7)ave. (N = 6)ave. (N = 3)
Corr. Spectrum0416073504160785∼04168630416080004160860∼04160915041608680416091501270795    
Spot Size (µm)500000005551010
  • a

    All analyses from traverse 3 except line #24.

  • b

    Analyses were reduced using 11 oxygens per formula unit for jarosite, 11.25 for stilpnomelane.

  • c

    Analyses were reduced using the same jarosite cation sums as Kuebler et al. [2013]. The Fe2+/Fe3+ ratios for jarosite were adjusted to bring the B cation sums to 3.0. The A cation sums may be <1.0 if there are any divalent cations substituting for monovalent cations.

  • d

    The Fe2+/Fe3+ ratios of stilpnomelane were set to 0.00001.

  • e

    The stilpnomelane tetrahedral sum should =4, the ferromagnesian sum should =2.66, the excess Al in line #47 was applied to the ferromagnesian sum.

  • f

    n.a. = not analyzed, n.d. = not detected.

  • g

    Numbering of cracks and veins differs from text (major cracks were included in the count during EMP acquisition but only stilpnomelane veins were counted during Raman analysis).

  • h

    The point where line #24 was acquired is ∼40 µm above the spot indicated in 01270795.bmp.

New total88.8086.5280.1289.0883.8483.7090.7990.0983.1185.8483.8088.0189.65
Corrected total93.6491.3683.9193.8688.4587.4895.9093.2986.8090.0388.0091.8393.02
For stilpnomelane             
Si + Al + P + S  3.67  3.70 4.003.733.483.373.933.91
Fe2+,3++Mn + Mg + Ti + Cr + Ni + Zn + V  2.69  2.68 2.732.672.903.082.492.56
Ca + Na + K  0.01  0.01
For jarosite             
A cations1.130.78 1.220.88 0.51      
B cations3.003.00 3.003.00 3.00      
X cations2.322.04 2.162.35 2.70      
Total cations6.445.836.386.386.236.396.206.826.426.436.516.506.75
Table 5. EMP Analyses of Olivine Along Traverse 3 in Thick Section ,177a
Line #1415161718192021224142
Phase Id.ol + laiholivineolivineolivineolivineolivineolivineolivineolivinelaihunite 73fayalite 73
Distance (µm)04692138184230276322368  
Notespt 1pt 2pt 3pt 4pt 5pt 6pt 7pt 8pt 9pt 1pt 2
 unk 3unk 4unk 5unk 6unk 7unk 8unk 9unk 10unk 11unk 20unk 21
Corr. Spectrum0416070304160736041607690416080204160834041608670416090004160933041609501010102510101004
Spot Size55555555511
  • a

    The distance between the digitized MIL analyses is ~46 µm.

  • b

    Analyses were reduced using 4 oxygen per formula unit for the olivine and 8 for the laihunite in standard 73.

  • c

    The Fe2+/Fe3+ ratios were varied to bring the cation sums as close to 3.00 as possible for all MIL olivine. The Fe2+/Fe3+ ratios were varied to bring the laihunite cation sums to 5.00 and the Fa sums to 3.00 for standard 73.

  • d

    S and P were included in the tetrahedral sum so that all cations are accounted for.

  • e

    Mn and Fe3+ were included in the calculation of the Mg#.

  • f

    Laihunite is a non-stoichiometric olivine with the approximate composition Fe2+Fe3+2(SiO4)2.

  • g

    n.a. = not analyzed, n.d. = not detected.

  • h

    The step size used in the corresponding Raman traverse = 1.414 µm.

Corrected total101.28100.80101.21100.93100.39101.01101.57101.37101.96102.05100.87
For olivine           
sum oct2.
sum tet0.991.000.990.991.000.990.991.000.991.861.00
Total cations3.
Mg# of ol0.140.330.400.400.410.410.410.410.410.020.02
Figure 7.

Plot of olivine doublet peak positions from the spectra acquired along traverse 3 on MIL ,177. Plot indicates that the range of olivine compositions measured from core to rim is consistent with previously published olivine compositions for MIL 03346 (Fo5–44) [McKay and Schwandt, 2005]. Data points fall slightly below the calibration curve, probably because the Raman spectrometer was calibrated against Si instead of cyclohexane at the time of acquisition.

[18] An interstitial grain of laihunite was found during our petrographic analysis of ,168 and identified by Raman spectroscopy but not analyzed quantitatively by EMP (Figures 1g–1l). Small patches of olivine remain inside the laihunite and veins of stilpnomelane can be discerned (the stilpnomelane is a lighter brown than the host laihunite in transmitted light and darker in reflected light). X-ray maps indicate that the interstitial laihunite is chemically heterogeneous with Si contents lower than the adjacent pyroxene, zoned with respect to Mn and Fe and strongly zoned with respect to Mg. The stilpnomelane veins have less Fe, Mn, and Mg but more Si than the host laihunite (see Figure 1i–1l).

3.3.2 Microprobe Results on MIL Stilpnomelane Versus Standard Stilpnomelane

[19] Only four stilpnomelane analyses were acquired (5 µm spot size) on thick section ,177 (see Table 3). Three of these are quite similar (line #24, 42, and 45), and the fourth contains less SiO2 but has significant Al2O3 (line #47; 18.8 wt % Al2O3) and may be chlorite. The three low alumina analyses (<1 wt % Al2O3) have less alumina than the Nassau Raman standard and the 37 analyses reported in Eggleton [1972]; the fourth has significantly more alumina than any of these. CaO, Na2O, and K2O are low in all of the analyses (sum to <0.1). The stilpnomelane occurrences in the middle of the veins are wider (typically tens of µm) than the jarosite films, so overlap of the beam onto adjacent phases is not considered a problem. The 65 stilpnomelane analyses made on thin section ,168 are consistent with the first three analyses on ,177 with an average of 41.6 wt % SiO2, 37.7 wt % FeO, and less than 0.5 wt % Al2O3 (traverses d–h; see Table 3). The average of six EMP analyses made on the Nassau standard and three on Chinese standard AW #1121 is provided in Table 3 for reference. The tetrahedral cation sums of the stilpnomelane include Si and Al plus any detected S and P, and the ferromagnesian sum includes Mn, Ti, Cr, and Ni in addition to Mg and Fe (so that all analyzed elements are accounted for). The reddish-brown color of the stilpnomelane indicates that the iron in the stilpnomelane is ferric so the Fe2+/Fe3+ ratio was set to 0.0001; the resulting cation sums are satisfactory indicating that the Fe2+/Fe3+ ratios do not need to be varied. Excess alumina in the Al-rich analysis was applied to the ferromagnesian sum. The stilpnomelane in MIL has similar SiO2 contents to the phyllosilicates in Lafayette (40–45 wt %) but 10 wt % more FeO and 7 wt % less MgO; the stilpnomelane also has a lower alkali content than the Lafayette alteration products (Al2O3 + Na2O + K2O + CaO = 6.55 wt %) [Kuebler et al., 2004]. The stilpnomelane in MIL contains similar Cl concentrations as the iddingsite in ALHA 77005 (0.2 wt % Cl on average, ranging up to 0.45 wt % Cl).

3.3.3 Element Mobility Trends Across Secondary Veins

[20] Because only a few quantitative analyses were collected on ,177, we use our seven microprobe traverses from thin section ,168 to study element mobility trends in MIL (Figures 8b–8h) [Kuebler et al., 2007]—traverse a could not be used because of an error during analysis. All of the data points from the traverses were plotted (even bad data points from cracks and fractures) because most of the secondaries occur in cracks and fractures and because some secondary phases are hydrated and have low totals. All of the traverses were collected using a 5 µm spot size and 1–2 µm steps which may exaggerate the apparent mixing trends, but jarosite alteration pervades the olivine in ,177, so some SO3 may actually be present in the olivine adjacent to the cracks and fractures. We collected 10 quantitative analyses each in the unaltered cores of three olivine; these are plotted in Figure 8a. According to our X-ray maps, PPL and BSE images, traverses b, c, and f only cross Ca-sulfate-bearing fractures (gypsum and bassanite); these are plotted in Figure 8b and indicate a loss of FeO and MgO from the olivine and an increase in CaO (FeO and MgO decrease as the traverse passes out of the olivine and into the fracture). Traverses d and e (Figures 8c and 8d) both cross stilpnomelane veins and end in fayalite at or near the rim of olivine 2 in ,168 (location of e shown in BSE image of Figure 11a; other traverse locations are shown in Figures 11b–11d). Traverse d indicates the strongest olivine zoning of all the traverses. The stilpnomelane contains ~40–45 wt % SiO2, 33–38 wt % FeO, with <1 wt % CaO; the olivine and stilpnomelane form distinct clusters in these plots, reflecting the sharp contact between these phases. Traverse e crosses a Ca-sulfate-bearing fracture before passing into stilpnomelane; the presence of the sulfate is indicated by the CaO trend. There are two FeO trends in Figure 8d (one to the origin and the other to ~43 wt % FeO) indicating that jarosite is also present. The FeO and MgO trends in Figure 8c mimic those in the sulfate-bearing traverses, but there is no coincident increase in CaO; according to positional notes taken during analysis, these are bad analyses acquired where traverse d landed in a void or crack. Traverse g (Figure 8e) passes through two stilpnomelane veins in a region of jarositic alteration and roughly parallels the edge of the olivine shown in Figure 1a (olivine 1). Again, two FeO trends are apparent, reflecting the presence of jarosite. Traverse h (Figure 8f) crosses two stilpnomelane veins and passes into a laihunite overgrowth (olivine 3, not shown). The laihunite has several wt % less FeO than the coexisting fayalite and produces scatter in the cluster of olivine data points; jarosite is also present.

Figure 8.

Plot of wt% FeO, MgO, and CaO versus wt% SiO2 of data points from (a) unaltered olivine and (b–f) traverses b–h on MIL 03346,168. (a) Data points represent 10 analyses from each of three unaltered olivine cores having ~Fo40 composition. According to our PPL, BSE, and X-ray maps of ,168, traverses b, c, and f cross fractures containing only Ca-sulfate; these are plotted in Figure 8b and indicate a loss of FeO and MgO from the olivine and an increase in CaO. (c and d) Traverses d and e cross stilpnomelane veins and end at or near the fayalitic rims of olivine 2 (location of traverse e shown in Figure 11a; the olivine and stilpnomelane form distinct clusters reflecting the sharp contacts between these phases. Traverse e crosses a Ca-sulfate vein before passing into stilpnomelane; the presence of the sulfate is indicated by an increase in CaO. There are two FeO trends in traverse e (one to the origin and the other to ~43 wt % FeO) suggesting that jarosite is also present. The FeO and MgO trends of traverse d mimic those in sulfate-bearing traverses, but the lack of a coincident increase in CaO suggests that these are bad analyses. (e) Traverse g passes through two stilpnomelane veins in a region of jarositic alteration and roughly parallels the edge of the olivine pictured in Figure 1a (olivine 1). Again, two FeO trends are present indicating jarosite. (f) Traverse h crosses two stilpnomelane veins lined by jarosite before passing into the laihunite overgrowth of olivine 3 (not shown); laihunite has less FeO than the coexisting fayalite and is indicated by scatter in the olivine data.

3.3.4 X-ray Profiles

[21] X-ray profiles (single-line X-ray maps) were taken near the locations of the Raman traverses on thick section ,177. These are shown in Figure 9 where the alteration products are indicated by positive spikes of the elements composing the alteration phases (e.g., the Al, P, K, and S in the jarosite lining the alteration veins) concurrent with negative spikes and bands in the Fe and Mg of the olivine. Zoning in the olivine is apparent in the profiles as an increase in Fe and decrease in Mg at the rims of the olivine (left side of Figure 9) and in the X-ray maps of the entire olivine shown in Figure 10. Note that there are negative spikes in the Si profile where jarosite lines the alteration veins but that the stilpnomelane in the center of these veins has more Si than the host olivine. The locations of veins 1 and 2 are indicated in the Mg profile where their width is most apparent. The major spike in Al (off the scale of the profile) represents a single Al-rich stilpnomelane analysis (possibly chlorite; see below) lining the big crack interior to vein two.

Figure 9.

X-ray profiles (single-line element maps, top) were collected near the locations of the Raman traverses and represent compositional variation from the rim of the olivine (left) to the core (right). Positive peaks of Al, P, K, and S indicate the presence of jarosite in cracks and fractures concurrent with negative Fe and Mg stilpnomelane bands in the olivine (alteration veins 1 and 2 are labeled in the Mg profile). The ratio of the 570 cm−1 peak signal-to-noise ratio to that of the olivine doublet (bottom) is used to demonstrate the presence of laihunite at the rim of the olivine and the presence of jarosite and stilpnomelane in the alteration veins.

Figure 10.

X-ray maps of the olivine in thick section ,177. The BSE image, Fe and Mg maps, indicate zoning in the olivine; the Ca map represents the surrounding clinopyroxene; Al and Na highlight the glassy mesostasis; the S map indicates the distribution of jarosite along cracks and fractures and provides a visual estimate of the volume of solution present during alteration; the Si map highlights the stilpnomelane veins.

[22] The X-ray profiles suggest variable jarosite compositions (i.e., Al, P, K, and S peaks in Figure 9 have variable heights); this was confirmed by quantitative analyses made at select locations to calibrate the profiles (see Table 3). Ideal jarosite has 9.4 wt % K2O, 47.8 wt % Fe2O3, 32.0 wt % SO3, and 10.8 wt% H2O. The MIL 177 jarosite microprobe totals (before the Fe2+/Fe3+ correction by stoichiometry) range from 80.3 to 90.8 suggesting a wide range in H3O (9.2–19.7 wt% H2O, by difference). Sodium and hydronium may both substitute for K; hydronium substitution is more common in jarosite than alunite and is usually inferred where microprobe analyses suggest a deficiency in cations and an excess of H2O [Ripmeester et al., 1986]. In comparison to the ideal jarosite composition, K2O is deficient (ranging from 2.9 to 4.2 wt% K2O) with the sum of the A cations being significantly <1 in all but two analyses (this may occur when divalent cations substitute for monovalent cations because charge balance is maintained by vacancies [Scott, 1987; Papike et al., 2006]. FeO ranges from 38.3 to 52.5 wt% in the jarosite analyses, with some spots deficient in FeO and others in excess relative to the ideal composition. The Fe2+/Fe3+ ratios were varied to bring the sum of the B cations to the ideal value of 3.0 and the inferred Fe2+ applied to the A cation sums. SO3 ranges widely, from 13.2 to 30.9 wt%, so all but one of the analyses appear to be deficient in SO3 relative to the ideal formula. According to Scott [1987], SiO2 is a permissible substitution in the XO4 site. Because SiO2 is present in significant levels, ranging from 2 to 25 wt%, and because the ionic sizes of Si and P are even smaller than that of Fe2+, it is unlikely that these components substitute into the A sites. Therefore, we included Si and P in the sums of the X cations. This produces an apparent excess of X cations in all but one analysis (sum of X cations ranging from 2.0 to 2.7, ideal value = 2.0). According to the mark the electron beam left on the jarosite films in spot mode at 8000 X, we do not believe there was any overlap onto the adjacent phases during quantitative analysis. The traverses on ,168 and ,169 used a 5 µm spot size and overlapped onto the adjacent olivine or stilpnomelane so the discussion above is based on the quantitative analyses of the jarosite in ,177. We initially identified the Ca-sulfates (bassanite and gypsum) in thin sections ,168 and ,169 by the CaO content of their EMP analyses, but the FeO and MgO contents of the sulfate analyses immediately adjacent to the host olivine are high because of the broad electron beam and their SO3 contents and totals low because of the high beam current (necessary for analyzing the olivine and other secondary phases). The bassanite and gypsum identifications were confirmed using Raman spectroscopy [see reference spectra in Wang et al., 2009].

3.4 Comparison to Alteration Assemblages in Other Nakhlites

[23] Collectively, the secondary assemblages of the nakhlites are described as mixtures of smectite or illite with dispersed Fe oxides, sulfate, siderite, and halite [Chatzitheodoridis and Turner, 1990; Gooding et al., 1991; Treiman et al., 1993; Bridges and Grady, 1999, 2000]. The Fe oxides are variously identified as hematite, magnetite, maghemite, ferrihydrite [Fe5O12H9], or goethite and the smectite sometimes specified to be saponite [(½Ca,Na)0.7(Mg,Fe,Al)6 (Si,Al)8O20(OH)4∗nH2O]. Day et al. [2006] report the pyrrhotite of MIL to be altered. Lafayette, one of the deeper-formed nakhlites, contains the highest proportion of alteration with occurrences in both the olivine and mesostasis [Treiman et al., 1993]. The reddish-brown alteration products are more restricted in the shallower-formed nakhlites; forming veins in olivine with fewer occurrences in the mesostasis. The olivine alteration veins in these three MIL thin sections are in sharp contact with their host olivine (in reflected light); the alteration products have a smooth texture and, as per our Raman analysis, contain stilpnomelane. Veins in NWA 817 have saw-toothed outlines as do those in Lafayette, but the texture of the vein-filling materials are smooth like those in MIL 03346 (compare images in [Treiman et al., 1993 and Treiman, 2005] to those of [Sautter et al., 2002; Gillet et al., 2002]). Gillet et al. [2002] note the difference in the character of the NWA 817 veins, including the lack of iron oxides, and report them to have compositions similar to ferromagnesian smectite but declined to provide a structural formula without a more detailed crystal chemical study of the material. Changela and Bridges [2010] imaged alteration veins in Lafayette containing complex textures, having smooth regions with rosettes of coarse phyllosilicates and median veins of finer-grained phyllosilicates. Alteration veins in NWA 998 are reportedly narrower than in other nakhlites [Bridges et al., 2004; Treiman and Irving, 2008].

[24] Calcium sulfates (gypsum, bassanite, and anhydrite) occur in most nakhlites and Chassigny [Wentworth and Gooding, 1994], and jarosite is tentatively identified in Nakhla [Treiman and Gooding, 1991] but only demonstrated to be present in MIL 03346 and the paired Y-000593 and Y-000749 [Fries et al., 2006; Noguchi et al., 2009]. Siderite [FeCO3] is reported to occur with interstitial halite, anhydrite, and minor chlorapatite in Nakhla [Chatzitheodoridis and Turner, 1990; Bridges and Grady, 1999]; with the interstitial oxides of Governador Valadares [Bridges and Grady, 2000]; and is suggested to be the fine-grained material lining the center of the olivine alteration veins in Yamato 000593, Lafayette, and NWA 998 [Ries, 1998; Bridges and Grady, 2000; Bridges et al., 2004]. Day et al. [2006] did not observe any gypsum, anhydrite, siderite, or goethite in sections ,111 and ,118 of MIL but found one instance of Ba celestite. We did not detect any siderite in the Raman point count on MIL and siderite was not reported in NWA 817 by Sautter et al. [2002] or Gillet et al. [2002]. Gooding et al. [1991] reported a single occurrence of magnesium sulfate (epsomite or kieserite) in an interior sample of Nakhla but was not found by Chatzitheodoridis and Turner [1990]. Bridges et al. [2001] report carbonate abundances to decrease from Lafayette to Governador Valadares to Nakhla; the lack of siderite in MIL and NWA 817 suggests that these meteorites complete the sequence. Gillet et al. [2002] reports NWA 817 to be a very fresh hot desert find with only minor (terrestrial) calcite.

4 Inferences Regarding Alteration

4.1 The Terrestrial or Martian Provenance of Nakhlite Secondary Minerals

[25] The Martian origin of the phyllosilicates is demonstrated by the presence of veins truncated by fusion crust, offsets produced by micro-faulting during impact events, and Ar-Ar ages that predate their arrival on Earth [Gooding et al., 1991; Treiman et al., 1993; Shih et al., 1998; Swindle and Olson, 2002]. Jarosite has been identified in Mössbauer spectra of rocks and soils at both MER landing sites [Klingelhöfer et al., 2004; Morris et al., 2006] and is present in the shergottite ALHA 77005 [Smith and Steele, 1983, 1984; Kuebler, 2009], and its presence previously noted in MIL per the Raman spectral maps of Fries et al. [2006]. Jarosite is present in X-ray maps of the groundmass of all three MIL thin sections (see S map in Figure 10); we infer the jarosite to be Martian because it predates the Ca-sulfates, which are inferred to be Martian.

[26] The carbonate and sulfate components of the nakhlites are inferred to be Martian because they occur in both falls and finds from around the globe [Bridges and Grady, 2000; Bridges et al., 2001; Gooding et al., 1991; Gooding, 1992; Dreibus et al., 2006], but similar phases occur in Antarctic soils and terrestrial contaminants may be present [Wentworth et al., 2005]. Treiman et al. [1993] observed occurrences of Ca-sulfates in both open and closed (i.e., not exposed to the atmosphere) vugs of the Lafayette fusion crust and inferred these sulfates to have been mobilized during passage through the Earth's atmosphere. Photo-documentation of MIL 03346 demonstrates that no efflorescent salts (calcium sulfates) were present on its surface at the time of collection but were present after an accidental thawing [Satterwhite and Righter, 2006], presumably having migrated to the surface from the interior of the meteorite. Occurrences of gypsum and bassanite in the X-ray maps of ,168 and ,169 (exterior samples) are obvious late-stage deposits, appearing in a fracture crosscutting a stilpnomelane-bearing vein of one olivine and with impact melt glass in another (Figures 11a and 11f). Calcium sulfates were not detected in the X-ray maps of the olivine in ,177 (cut from a more interior piece of the meteorite), but weak 1007 cm−1 peaks are observed in multiple spectra from the Raman point count and indicate trace amounts of gypsum and jarosite. No siderite or other carbonate was detected in any of the thin sections by EMP or Raman spectroscopy.

Figure 11.

(a) BSE image of Ca-sulfate in a fracture crosscutting a vein of stilpnomelane in olivine 2 of thin section ,168. (b–f) Sulfur X-ray maps of the Ca-sulfate in thin sections ,168 and ,169; maps indicate that low levels of jarosite (gray) occur throughout the mesostasis but that Ca-sulfates (white, ~41 wt % CaO) only occur along late-stage fractures and cracks.

4.2 Inferences Regarding Conditions, Sequence of Alteration

[27] Most references report laihunite to be an oxidized variety of fayalite with the approximate composition of Fe+2Fe+32(SiO4)2, but the terms ferrifayalite and oxidized olivine are also found in the literature [Kitamura et al., 1984; Schaefer, 1985; Kan and Coey, 1985; Kondoh et al., 1985; Khisina et al., 1995; Dyar et al., 1998]. Dyar et al. [1998] report the CNMMN to have approved usage of the term laihunite to describe distorted olivine having Fe3+ in M2 sites, Fe2+ in M1 sites, and vacancies in alternate rows parallel to (001). The WU Raman standard is from the laihunite type locality, a metamorphic magnetite mine near the village of Lai-He in Liaoning Province, China, and was originally identified by XRD [A. Wang, personal communication, 2003]. Laihunite has been identified in a variety of geologic settings—iron ores, eulysite (an Fe-Mn rich metamorphic rock), banded iron formations (BIF), as well as volcanic tuffs [Ferrifayalite Research Group, 1976; Laihunite Research Group, 1976, 1982 in Schaefer, 1985; references in Kitamura et al., 1984]—but the literature is not clear regarding the mechanism of formation at these locations (oxidation, alteration, metasomatism).

[28] Laihunite has reportedly been synthesized by the oxidation of fayalite at elevated temperatures with or without an aqueous fluid (400–800°C [Kondoh et al., 1985]; 100 and 300°C [Banfield et al., 1990; Iishi et al., 1997]). That laihunite was only detected in the olivine overgrowths of ,168, ,169, and ,177 (does not form continuous rims) and not observed adjacent to any of the traversed fractures inside the olivine of ,177 suggests that it formed by the oxidation of fayalite as the magma cooled (a late primary phase or early secondary phase) and not as a reaction product of the fluids that produced the stilpnomelane or jarosite. Crosscutting relationships demonstrates the stilpnomelane to have formed after the laihunite (stilpnomelane crosscuts the laihunite at the edge of the olivine in Figure 1e and crosscuts primary zoning in the Mg X-ray map in Figure 1i), as do the young geochronologic ages reported for the vein-filling materials in Lafayette. We found stilpnomelane veins crosscut by jarosite (see Figure 1d) and one stilpnomelane vein offset by a planar fracture that is crosscut by a continuous zone of jarosite alteration. We therefore infer the jarosite to represent a third generation of alteration. Late-stage fractures hosting Ca-sulfates crosscut both the stilpnomelane and jarosite and are considered to be late third or fourth generation alteration products.

[29] Stilpnomelane occurs in a variety of geologic environments (greenschists, iron formations, in xenoliths within Skaergaard gabbros and granophyres, and as a surface weathering product of sulfide deposits [Deer et al., 1992; formula from Eggleton, 1972; Klein, 2005] and is usually indicative of low to medium-grade metamorphism (late diagenesis to garnet-grade metamorphism) [Klein, 1983]. The Nassau stilpnomelane Raman standard is a fragment of schist (banded quartz and stilpnomelane) whose metamorphic history does not extrapolate to the MIL alteration environment. Stilpnomelane is common in very low grade metamorphic mafic rocks and metasediments of appropriate composition (i.e., glauconite bearing); it has been synthesized from gels and from magnesian chlorite + albite at 630°C and 2 kbar, perhaps requiring pressures as low as 1 kbar to form [Deer et al., 1992; Winkler, 1979]. French [1973] reviews the criteria for evaluating whether a secondary phase formed under the conditions of diagenesis (1–2 kbars, 150–200°C) or low-grade metamorphism (2–3 kbars, 200–350°C) and indicates that the stability range of stilpnomelane may be large enough to encompass both diagenetic and low-grade metamorphic conditions. The vitric texture of the MIL mesostasis suggests rapid crystallization; MIL is presumed to have formed in a relatively shallow environment, so we infer the stilpnomelane in MIL to have crystallized at the lower end of its range of stability.

[30] The alteration products in the analog image of the Delvigne Atlas [1998; plates 130 and 131] more closely resemble the alteration products in Lafayette than those in MIL and are indicated to be a mixture of chlorite and smectite. Chlorite and saponite are both olivine alteration products but may also form from other minerals and may occur together or separately. Saponite is a microcrystalline to lamellar clay, whereas chlorite has a micaceous habit [Delvigne et al., 1979]. Smectites occur as hydrothermal alteration products near hot springs and geysers or in soils produced by the alteration of basic rocks [Deer et al., 1992]. Chlorite is a common hydrothermal alteration product in igneous rocks but may also occur in argillaceous sediments and low to moderate grade metamorphic rocks. On Earth, the increased temperatures and pressures associated with diagenesis are thought to facilitate the conversion of trioctahedral smectites into chlorite [Deer et al., 1992].

[31] Cristobalite, in particular, indicates that the mesostasis of MIL crystallized rapidly at elevated temperatures (β-cristobalite is the stable silica polymorph between 1470 and 1713°C) [Deer et al., 1992]. The coexistence of fayalite, magnetite, and cristobalite in the groundmass suggests that fO2 reached QFM during the late stages of crystallization, and the presence of laihunite and hematitic stains in the groundmass suggests that conditions may have become even more oxidizing during the final stages of crystallization (or post consolidation). The presence of an oxidizing environment is consistent with results from Viking experiments (surface materials inferred to contain oxidants—peroxides or perchlorates formed by exposure to short wavelength UV light) [Oyama and Berdahl, 1977; Klein, 1998; Lodders and Fegley, 1998; Hecht et al., 2009]; these atmospheric oxidants when combined with volcanogenic SO3 and HCl, plus H2O in the form of frost or dew, form acids capable of dissolving basalts [Haskin et al., 2005].

[32] Jarosite is present as thin films in cracks and veins of MIL olivine and is present in the mesostasis where it postdates the stilpnomelane. The sulfur X-ray map in Figure 10 gives a visual estimate of the volume of solution present during alteration. These films are narrower than the jarosite veins found in ALHA 77005,34 and 51, but jarosite has been identified in more thin sections of MIL than ALHA. The MIL pattern of jarosite deposition suggests infiltration. The presence of jarosite suggests an acidic environment, but gypsum and bassanite (and other previously reported secondaries: halite, siderite, anhydrite, and epsomite) suggest deposition from more neutral chloride solutions. The sulfates and salts may later precipitate from the same fluids that deposited the jarosite or represent a separate generation of alteration (compare occurrences of jarosite and Ca-sulfate in Figures 10 and 11).

4.3 Inferences Regarding the Setting of Nakhlite Olivine Alteration

[33] Raman spectra of the Lunar Crater and Mauna Kea alteration products demonstrate their iddingsite to be fine-scale intergrowths of a volumetrically minor polymerized silicate with either goethite or maghemite [Kuebler et al., 2013]. Altered olivine only occur in the oxidized surface rinds (outer 1 cm) of the Mauna Kea samples and are surface weathering products formed atop of Mauna Kea sometime after extrusion. The Lunar Crater iddingsite is attributed to surface alteration during the pluvial episodes of the Pleistocene (the Lunar Crater flow is inferred to have been more permeable before the accumulation of its aeolian mantle). ALHA 77005 iddingsite is also a fine-scale intergrowth containing akaganéite but is overprinted by a later generation of jarosite alteration; the iddingsite appears to be restricted to the light lithology (olivine poikilitically is enclosed by orthopyroxene and predates many of the shock features) and is therefore interpreted to have formed at depth as a product of deuteric alteration. The ALHA, LC, and MK alteration products are optically continuous with their host olivine and presumed to have formed by topotactic growth of the secondary phases (the secondary phases inherit part of the olivine structure).

[34] In contrast, the secondary alteration veins in the nakhlites are dominated by phyllosilicates that are not optically continuous with their host olivine, occur in veins that are in sharp contact with the host olivine (suggesting complete disruption of the olivine crystal structure), indicate different element mobility trends, and are significantly younger than their host lithology. Lafayette and the deeper-formed nakhlites host smectite with masses of goethite or ferrihydrite [see images in Treiman et al., 1993; Treiman, 2005], but analysis of the MIL alteration products indicates none of these (except for the potential ferrihydrite/jarosite laser oxidation features in vein 2).

[35] The nakhlite analog sample pictured in the Delvigne Atlas [1998] resembles the saw-tooth vein-filling materials in Lafayette and is likely to be from Volcans National Park which lies a few miles north of Kivu Lake and straddles the boundary between the Democratic Republic of the Congo and Rwanda (the Atlas gives the sample location as Northern Kivu, DRC, and states that the sample was collected from an outcrop). According to Google Earth Satellite imagery, this is a modern basalt complex hosting several calderas (e.g., Nyiragongo and Nyamulagira) and is associated with the western limb of the East African Rift. Proximity to Lake Kivu implies alteration at high water/rock ratios but SO2 fumaroles and dry CO2, CH4, and hydrocarbon gas vents (“mazuku” or “evil winds”) are also present [Smets et al., 2010; Tedesco et al., 2007]. This is a modern, equatorial volcanic complex implying that alteration occurred in a warm, humid climate shortly after extrusion (in contrast to the discrepancy between the reported nakhlite crystallization age and the age of their alteration products). Unfortunately, no Kivu samples were available for analysis so we cannot review the secondary phase identifications and element mobility trends of MIL with respect to a terrestrial analog as we did for ALHA 77005.

[36] In reference to the analog site (but without specific knowledge of where in the park the sample pictured in the Atlas was collected; i.e., its proximity to a groundwater source or fumarole), the MIL phyllosilicate and sulfate alteration products are inferred to have formed in discrete episodes; that the phyllosilicates formed in a warm, humid climate in the presence of groundwater and the sulfates deposited later by gas vents or fumaroles (or related seepage; as per the sulfur X-ray map in Figure 10). Phyllosilicate alteration of the Kivu basalts occurred in a shallow environment shortly after extrusion. However, the MIL alteration veins contain stilpnomelane, suggesting conditions akin to diagenesis, and are significantly younger (650–680 Ma) [Shih et al., 1998; Swindle and Olson, 2002] than their host clinopyroxenite (1.3 Ga) indicating a significant span of time between accumulation and alteration. It is unclear at present how to reconcile this age discrepancy relative to the terrestrial analog. Volcanism on Mars is longer lived than on Earth, and again, there is a difference in the host lithology. MIL is inferred to be a shallow-formed nakhlite but is still a cumulate lithology. We anticipate differences in the temperature, composition, and possibly the volume of the fluids effecting alteration in the Kivu basalt and MIL; confirmation of the conditions of alteration will require field work.

[37] MIL secondaries indicate the oxidation of olivine to form laihunite, the addition of SiO2 with the formation of stilpnomelane, and the loss of FeO and MgO with an influx of CaO associated with sulfate deposition. No mixing is indicated between the olivine and stilpnomelane which form discrete clusters of data points in Figures 8c and 8d (instead of the linear trends observed for the iddingsite samples), reflecting the sharp contacts between these phases. Laihunite is inferred to have formed early at an elevated temperature. An oxidation reaction for the conversion of fayalite to laihunite can be written as

display math(1)

The presence of magnetite and silica as reaction products is consistent with the inference that laihunite formed during the final stages of crystallization (magnetite and cristobalite occur in the mesostasis). We can write an approximate chemical reaction for the formation of stilpnomelane in MIL assuming a Fo50 olivine composition and the stilpnomelane formula of Eggleton [1972] (see footnote in Table 2). The cations supplied to form the stilpnomelane are assumed to have been mobilized from the surrounding augite and mesostasis. The formula used in the reaction below requires the influx of alkalis (Na, Ca, K, and Al), but our microprobe analyses indicate the stilpnomelane in MIL to be alkali poor (sum of CaO, Na2O, and K2O < 0.1 wt %; Al2O3 < 0.5 wt %):

display math(2)

This reaction assumes that the glass in the mesostasis has the composition of K-feldspar, assumes stoichiometric alteration, and that fluid was present to aid cation mobility.

[38] A few of the data points in the MIL ternary fall along the Fo–Fa tie line (olivine zoning; see Figure 12) but others scatter above or below this toward the SiO2—(FeO + Fe2O3) join. Data that trend toward the FeO + Fe2O3 apex represent jarosite and those that fall above the Fo–Fa tie line represent stilpnomelane. The reaction for MIL implies a water/rock ratio similar to the ALHA reaction [Kuebler et al., 2013]. Both jarosite and stilpnomelane are hydrated and require an influx of water (more so than hematite or goethite), but their distribution within MIL is irregular; portions of the MIL groundmass contain jarosite and are visibly altered, whereas other areas retain their glassy texture.

Figure 12.

MIL ternary diagram; data points that fall along the Fo–Fa tie line illustrate the range of zoned olivine compositions. Data trending toward the FeO + Fe2O3 apex represent jarositic alteration, while those that fall above the Fo–Fa tie line represent stilpnomelane.

[39] Deeper-formed nakhlites are reported to contain more abundant alteration products suggesting that alteration may have been longer lived at depth with higher water/rock ratios (e.g., Nakhla, Lafayette). Siderite was not observed in MIL so no direct inferences can be made regarding the relative timing of its formation versus that of the stilpnomelane. Bridges and Grady [1999] review the textural and geochemical evidence pertaining to the formation of the siderite-anhydrite-halite-chlorapatite assemblage of Nakhla, citing textural evidence in favor of a high temperature, hydrothermal origin (i.e., vein deposits) but report geochemical data that support a low temperature, low pressure origin. This conflicting evidence led Bridges and Grady [1999] to conclude that the siderite, apatite, and halite in Nakhla most likely formed during the final cooling stages but Bridges and Grady [2000] and Bridges et al. [2001] to favor deposition from a brine. Kivu Lake consists of mildly basic (pH ~9.5) and CO2 saturated meteoric waters [see Tassi et al., 2009, Table 1], for concentrations of dissolved constituents in depth profiles of the five main basins, temperatures not reported); we do not know how the local meteoric groundwaters compare to the lake water but expect the Bridges and Grady [2000] hypothesis is more accurate. The jarosite films in MIL may represent fluids associated with the venting of volcanic gases. Halite's solubility suggests that dry conditions persisted after its deposition and implies that the stilpnomelane formed prior to the siderite assemblage. While Bridges and Grady [1999] report these salts to be roughly coeval with the alteration veins in Nakhla, the gypsum and bassanite in MIL clearly postdate all other secondary phases. It may be that the Ca-sulfate efflorescents in MIL were remobilized from anhydrite.

5 Conclusions

[40] Laihunite, stilpnomelane, jarosite, gypsum, bassanite, and possible ferrihydrite have been identified in the olivine of MIL 03346 (by either Raman spectroscopy or EMPA). The laihunite is inferred to be a late primary phase or early secondary oxidation product associated with the late-stage olivine overgrowths. Reports of laihunite synthesis suggest elevated temperatures (100–800°C) and indicate the presence of an oxidizing environment, but Raman analyses suggest that the MIL overgrowths are less oxidized than the Lai-He standard (see Figures 13 and 14; Kuebler et al., 2011). By analogy with the Kivu alteration site, we assume that the MIL stilpnomelane formed in the presence of groundwater (but at greater depth, under conditions akin to diagenesis) and that the jarosite and Ca-sulfates were deposited by volcanic vents or from the brine from that deposited the siderite assemblage (third generation of alteration; the sulfur X-ray map in Figure 10 suggests the presence of a fluid phase). The siderite and anhydrite reported in deeper-seated nakhlites [Bridges and Grady, 1999] were not observed in MIL but are thought to predate the jarosite and postdate the stilpnomelane. The Ca-sulfate efflorescents may be remobilized anhydrite. All of the secondary phases are inferred to be Martian.

Figure 13.

Variation in the signal-to-noise ratio of the 570 cm−1 laihunite peak relative to that of the olivine doublet along traverses 1 and 2 on the standard; the ratios indicate that the olivine in the standard is more oxidized than the olivine overgrowths of MIL ,177. Traverse locations are indicated on the photo inset (the brighter phase is laihunite; tick marks on the cross hairs are 10 µm apart). This may explain why we could not reduce our EMP data of the MIL overgrowths using the published laihunite formula.

Figure 14.

Spectral changes across the contact between fayalite and laihunite in traverse 1 on the Lai-He standard. In this traverse, the fayalite spectrum is completely replaced by that of the laihunite, whereas the fayalite doublet persists throughout the MIL overgrowths. BSE imaging of the standard indicates a complex, fine-scale intergrowth so the replacement reaction implied by the Raman spectra may be deceptively simple. Distances are reported relative to the starting point of each traverse (1.414 µm steps).

[41] Again, there is a difference in lithology between the terrestrial analog and Martian meteorite with the Martian sample representing a deeper-seated lithology. The terrestrial analog site is an active volcanic field so its alteration products are roughly contemporaneous with their host basalt. In contrast, the alteration products of the nakhlites are significantly younger than their host clinopyroxenite [Shih et al., 1998; Swindle and Olson, 2002]. How to resolve this discrepancy is not clear at present. Confirmation of the inferred mechanisms/conditions of alteration will require field work.

[42] We recognize that the optical properties of the stilpnomelane in MIL are inconsistent (lower relief, no cleavage) with stereotypical stilpnomelane. The stilpnomelane identification is based on the similarity of the Raman spectra of the MIL vein-filling materials with four WU Raman stilpnomelane standards as presented in Figure 4 and the similarity of the Nassau stilpnomelane XRD pattern with the reference stilpnomelane XRD patterns as shown in Figure 5. This identification is also supported by the similarity in chemical composition of the MIL alteration products with two of the WU Raman standards, published analyses in Eggleton [1972] and the extended volumes of Deer et al. [1966]. The MIL alteration products have more FeO but less alumina and alkalis than the Raman standards or the Lafayette phyllosilicates (MIL stilpnomelane have, on average, <1.5 wt % Al2O3, Na2O + K2O + CaO <0.5 wt %, and ~37 wt % FeOtot, ~43 wt % SiO2). The Raman peaks of the stilpnomelane spectra are broad, and their peak positions vary somewhat between the standards and MIL but do not resemble dioctahedral phyllosilicates. The major stilpnomelane (Si-Ob-Si) peak positions are >100 cm−1 lower than biotite-group (trioctahedral) phyllosilicates and >50 cm−1 lower than the vein-filling phyllosilicates in Lafayette [Wang et al., 2002; Kuebler et al., 2004].


[43] We thank the LPI and Allan Treiman for supporting the first author's research during the summer of 2006, the Johnson Space Center for the use of their electron microprobe facilities, and Kevin Righter for access to the Antarctic meteorite collection. I would also like to thank Alian Wang for granting me access to the Raman spectrometer at Washington University. Randy Korotev, Brad Jolliff, Alian Wang, Bob Dymek, and Jill Pasteris at Washington University in St. Louis all provided constructive reviews. The reviewers from JGR, Tim Glotch and Melanie Kaliwoda, were also helpful. I would also like to extend my appreciation to everyone in the Earth and Planetary Sciences Department at Wash University and especially the Planetary Surface Materials Research Group for their guidance and support. This research was supported in part by grant NNX07AQ34G (Mars Fundamental Research Program) to Dr Brad Jolliff.