and Planetary Science Letters 203 (2002) 431^444 www.elsevier.com/locate/epsl

Aqueous alteration in the Northwest Africa 817 (NWA 817)

Ph. Gillet a;, J.A. Barrat b, E. Deloule c, M. Wadhwa d, A. Jambon e, V. Sautter f , B. Devouard g, D. Neuville h, K. Benzerara i, M. Lesourd j a Laboratoire des Sciences de la Terre (UMR CNRS 5570), Ecole Normale Supe¤rieure de Lyon, 46 alle¤e d’Italie, 69346 Lyon, France b Laboratoire de Ge¤odynamique et Plane¤tologie (UMR CNRS 6112) and Universite¤ d’Angers, 2 boulevard Lavoisier, 49045 Angers Cedex, France c Centre de Recherches Pe¤trologiques et Ge¤ochimiques (UPR CNRS 2300), P.O. Box 20, 54501 Vandoeuvre les NancyCedex, France d Department of Geology, The Field Museum, Chicago, IL 60605, USA e Laboratoire MAGIE, Universite¤ de Paris VI, Paris, France f Laboratoire de Mine¤ralogie, Muse¤um National d’Histoire Naturelle, Paris, France g Laboratoire Magmas et Volcans (UMR CNRS 6524), Universite¤ Blaise Pascal, Clermont Ferrand, France h De¤partement des Ge¤omate¤riaux, ESA CNRS 7046, IPG Paris, 4 place Jussieu, 75252 Paris, France i Laboratoire de Mine¤ralogie et Cristallographie, Universite¤ de Paris VI, Paris, France j CNRS et SCIAM, Laboratoire d’Histologie, UFR de Me¤decine, Universite¤ d’Angers, 1 rue Haute de Recule¤e, 49045 Angers Cedex, France

Received 4 April 2002; accepted 11 July 2002

Abstract

Samples of a new of the family (NWA 817) contain traces of an iron-rich alteration product. Textural arguments indicate that this alteration product has been formed on the of the meteorite (). The chemical composition and structural data (X-ray diffraction, transmission electron microscopy and vibrational spectroscopy) show that the alteration mineral is a hydrous phase from the smectite family. Major elements and rare earth elements suggest that the formation of the alteration phase is related to the circulation of an aqueous fluid which composition is controlled by the dissolution of feldspars to account for a positive Eu anomaly, olivine and possibly apatite. Hydrogen isotope data display negative ND values ranging from 360 to 3280x in olivine and pyroxenes and from 3140 to 3181x in the alteration phase. The values of ND for the alteration product show a small scatter with a mean value of 3170 þ 14x. These values are lower than those previously obtained on other Martian , which give mainly positive ND values. These positive values have been interpreted as resulting from the interaction of the Martian meteorites with water from the Martian atmosphere. Ruling out the effect of terrestrial alteration, it is suggested that alteration in the NWA 817 meteorite was likely produced on Mars by the circulation of an aqueous fluid originating from a chemical reservoir, such as the Martian mantle, which has not equilibrated with a fractionated Martian atmosphere. ß 2002 Elsevier Science B.V. All rights reserved.

Keywords: Martian meteorites; nakhlite; alteration; D/H; geochemistry

* Corresponding author. E-mail address: [email protected] (P. Gillet).

0012-821X / 02 / $ ^ see front matter ß 2002 Elsevier Science B.V. All rights reserved. PII: S0012-821X(02)00835-X

EPSL 6354 24-9-02 432 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444

1. Introduction composed of ¢ne-grained feldspars, acicular pyroxenes, a silica phase (cristobalite), minute The presence in the past of large quantities of amounts of sul¢de droplets, and Ti-magnetite in water on the surface of Mars has been inferred a glassy matrix [7]. NWA 817, like the three other from the study of valley networks and other mor- known , shows no obvious shock fea- phological features. The identi¢cation and charac- tures like melt pockets, high-pressure minerals terization of water reservoirs (internal and exter- or planar deformation features [8]. The NWA nal) on Mars can also be done indirectly through 817 meteorite is an unbrecciated, olivine-bearing petrographic and chemical studies of pre-terres- clinopyroxenite with a cumulate texture [7,8].Mo- trial alteration products in the SNC meteorites dal analyses indicate the following mineral pro- which are thought to be pieces of Mars delivered portions: 69 vol% pyroxene, 15% olivine, 15% to the Earth’s surface [1]. Several such studies mesostasis and 1% Fe^Ti oxides. Key element ra- have been performed so far on the three known tios such as FeO/MnO ( = 37), Na/Al ( = 0.40), K/ meteorites of the nakhlite subclass (Nakhla, La- La ( = 449), Ga/La ( = 3.9U1034) and oxygen iso- fayette and Governador Valadares). These mete- topic composition (v17O = +0.37x) are clear evi- orites represent shallow crustal rocks showing dence for including this meteorite in the SNC clear evidence of pre-terrestrial low-temperature group [7]. Rare gas studies also indicate a martian aqueous alteration [2^5]. The predominant alter- origin for this meteorite [9]. ation products of these meteorites are: a material loosely termed iddingsite that ¢lls fractures cross- cutting magmatic minerals and mesostasis, sider- 3. The alteration products: textural observations ite and evaporitic minerals including halite, gyp- sum and anhydrite. Hydrogen isotope systematics Alteration in NWA 817 is readily apparent in indicate that these rocks have been altered by a hand samples and in thin sections as reddish £uid that equilibrated with the fractionated Mar- patches or veins (Fig. 1A). The reddish alteration tian atmosphere [6]. A fourth nakhlite, Northwest product is clearly post-magmatic as it often cross- Africa 817 (NWA 817), has recently been discov- cut olivines, the mesostasis and, more rarely, py- ered in the Sahara [7]. roxenes. In the mesostasis it occurs as patches The present study provides the description and rather than as well-de¢ned veinlets (Fig. 1B). In interpretation of the alteration products observed olivines, it occurs mostly as veinlets with, in some in this meteorite. They have been characterized cases, irregular and saw-tooth margins very sim- with the following analytical techniques: optical ilar to those described in the Lafayette meteorite microscopy, scanning electron microscopy (SEM), [3,4] (Fig. 1C). The veinlets appear to be mostly electron microprobe, secondary ion mass spec- homogeneous, with the exception of some compo- trometry (SIMS), high-resolution transmission sitional zoning mostly near their margins (Fig. electron microscopy (TEM), X-ray di¡raction, 1D). These zoning features as well as the texture IR and Raman spectroscopy (see Appendix for suggest that the veins may be hydrothermal in the description of these techniques). origin, with multiple episodes of deposition of material. Unlike the ‘iddingsite-like’ alteration materials reported from Nakhla and Lafayette 2. NWA 817: a Martian meteorite [2], optical and SEM observations show no evi- dence for the presence of a complex polycrystal- NWA 817 is a 104 g stone found in the Moroc- line and polymineralic mixture in this alteration can desert. The mineralogy consists of euhedral product. No siderite or other carbonates were ob- subcalcic augite (Wo38^40, En38^27, Fs24^34) served in association with the orange material. olivine (extensively zoned from Fa-57 in the core The extraterrestrial origin of this alteration can to Fa-86 in the rim) with crystallized magmatic only be demonstrated indirectly since we were not inclusions and a multi-component mesostasis able to observe relationships between the fusion

EPSL 6354 24-9-02 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444 433 and the alteration veinlets in the samples we with the X-ray data. Additional spectroscopic have used for making thin sections and polished data further con¢rm the mineralogical nature of sections. However, other arguments strongly sug- the alteration phase. Raman spectra exhibit broad gest that the orange material was formed during peaks at low wavenumbers (216 cm31, 278 cm31, pre-terrestrial alteration of the meteorite. A gen- 598 cm31 and 760 cm31), con¢rming its ill-orga- eral inspection of the main mass of the meteorite nized nature (Fig. 2C). The spectrum bears strong using a binocular microscope did not reveal or- similarities with those of smectites and kaolinites ange veinlets crosscutting the fusion crust or or- [13]. Infrared spectra show a very intense and ange weathering products at the surface of the large absorption band near 3600 cm31, suggestive fusion crust. The alteration product occurs in tex- of the presence of OH or H2O molecules (Fig. tural settings similar to those observed for the 2D). pre-terrestrial ‘iddingsite-like’ material in the three other nakhlites. Finally, textural evidence indi- cates that NWA 817 underwent only very mini- 5. The alteration product: chemical mal, if any, alteration in the terrestrial environ- characterization ment, thereby also lending support to a pre- terrestrial origin of the alteration of the magmatic The major element composition of the altera- minerals. Most meteorites recovered from the Sa- tion phase occurring in olivines and in the meso- haran desert have su¡ered di¡erent degrees of al- stasis is given in Table 1 and Fig. 3A. Fig. 3A teration. We have studied several of these mete- shows that the composition does not overlap orites, including three new shergottites [10^12]. with the compositional ¢elds de¢ned by the major Typically, terrestrial alteration is marked by the minerals. The low analytical totals (86^88 wt%) deposition of calcite from terrestrial £uids circu- and the absence of signi¢cant amounts of Cl, F lating in the cracks and fractures in these rocks. and C suggest the presence of water, which is Such calcite, however, is present in very minor consistent with the infrared data. The major ele- amounts in NWA 817. Additionally, the rare oc- ment composition, which is dominated by Si, Mg currences of sul¢des (which are known to be and Fe, resembles that of a ferromagnesian smec- highly susceptible to terrestrial alteration) in tite. In the absence of a more detailed crystal NWA 817 appear to be very fresh. Finally, even chemical study, including a structure re¢nement, the most altered olivine and pyroxenes of Saharan it is for the moment di⁄cult to provide a struc- meteorites never display such alteration textures. tural formula for the alteration phase and a de- Trace element composition of the alteration prod- tailed interlayer cation content of this smectite- uct provides further evidence for a pre-terrestrial like mineral. formation (see Section 4). The alteration phase (in olivines as well as in the mesostasis) is enriched in Fe relative to the olivines (Fig. 3A). In particular, it should be 4. The alteration product: structural data noted that the Fe/Mg ratio of the alteration phase is not controlled by the composition of the adja- A typical X-ray spectrum of an olivine contain- cent olivines (Fig. 3B). There is a clear correlation ing the orange alteration product is shown in Fig. between Si and Fe contents of the alteration 2A. Besides the strong olivine di¡raction peaks, phase. However, in the plot of FeO/MgO vs. additional peaks occur between 9 and 1.2 nm, SiO2/MgO, the compositions of the alteration with some peaks occurring between 1.4 and 1.5 phase in the mesostasis and in the olivine de¢ne nm. These peaks are in agreement with the d- distinct trends (Fig. 3A). This trend cannot be spacings of phyllosilicates of the smectite family. explained in the absence of a more comprehensive High-resolution TEM data reveal a poorly crys- crystal chemical characterization of this clay min- tallized material with lattice fringes varying be- eral. The plot of CaO/MgO vs. SiO2/MgO indi- tween 0.9 and 1.1 nm (Fig. 2B), in agreement cates that most of the Ca in the alteration phase is

EPSL 6354 24-9-02 434 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444

Fig. 1.

EPSL 6354 24-9-02 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444 435

Fig. 1 (Continued). Microphotographs of the alteration product in NWA 817. (A) and (B) are optical photographs taken under polarized light. (A) The alteration product forming veinlets in an olivine crystal. (B) A patch of alteration product in the meso- stasis. (C) Backscattered electron image of veinlets of alteration phase in an olivine crystal. Notice the saw-tooth margin forms of some veinlets. (D) High resolution backscattered electron image of the alteration product. Darker veins are regions less-rich in Fe. AP = alteration product; Px = pyroxene; Ol = olivine; Ms = mesostasis; Ti-m = titano-magnetite. not derived from the high-Ca phases (i.e. the py- that transported elements such as Si, Mg, Fe and roxenes or the mesostasis) (Fig. 3A). Al is abun- Ca into the rock. Nevertheless, the distinct com- dant in the alteration phase in the mesostasis positional trends observed when located in the (Al2O3 up to V4 wt%) but is only present in mesostasis or in the olivines suggest that minor minor amounts ( 6 1 wt% Al2O3) when located amounts of some of these elements (particularly, in the olivines (Table 1). These data indicate Si and Fe) were contributed by these phases. Fi- that the smectite-related phase formed in an nally, the distribution of some of the less abun- open system by the circulation of an aqueous £uid dant elements, such as Al, may be largely con-

EPSL 6354 24-9-02 436 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444

Fig. 2. Structural data on the alteration product in NWA 817. (A) X-ray pattern obtained on a single crystal of olivine contain- ing the alteration phase. The strong di¡raction lines below 0.6 nm correspond to olivine. Weak broad peaks are observed be- tween 0.9 and 1.2 nm compatible with di¡raction planes of phyllosilicates of the smectite family. (B) High-resolution transmission electron photomicrograph of alteration phase. The image shows an ill-organized crystal with di¡erent lattice spacings between 0.9 and 1.1 nm, in agreement with the X-ray data. (C) Raman spectrum obtained on the alteration phase. The spectrum is close to that of a smectite. The broad bands at 216 cm31, 278 cm31, 598 cm31 and 760 cm31 are characteristic of a poorly crystalline material. (D) Infrared spectrum of the alteration phase. The broad band near 3600 cm31 corresponds to absorption by OH or H2O molecules in the mineral.

EPSL 6354 24-9-02 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444 437

Table 1 Major element composition of the alteration products in NWA 817 and other nakhlites (in wt%)

SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO CaO Na2OK2O Total NWA 817 ^ alteration products in the mesostasis #1 47.35 0.01 2.46 0.00 27.15 0.23 8.66 0.18 0.06 0.47 86.57 #2 45.01 0.09 3.09 0.00 30.56 0.36 5.92 0.16 0.03 0.42 85.64 Average, n = 16 46.51 0.03 2.26 0.01 28.42 0.28 7.56 0.14 0.06 0.42 85.68 NWA 817 ^ alteration products in olivine crystals #3 43.31 0.06 0.17 0.00 35.87 0.43 6.40 0.19 0.13 0.47 87.03 #4 42.82 0.04 0.05 0.02 37.34 0.58 5.10 0.12 0.21 0.45 86.72 Average, n = 25 42.82 0.06 0.21 0.03 36.45 0.55 5.69 0.25 0.18 0.41 86.65 Nakhla ‘rust’ [3] 40.20 0.02 0.74 0.03 34.10 0.63 6.82 1.14 1.16 0.60 85.44 Lafayette iddingsite (total arbitrary ¢xed to 86%) [4] 42.79 0.01 4.16 0.00 25.28 0.42 12.11 0.61 0.26 0.36 86.00 trolled by contributions from the minerals in LREE enrichment in the alteration phase (La which the alteration phase was deposited. V6^8UCI). Furthermore, the LREE enrichment Trace element abundances in alteration prod- in the alteration phase is positively correlated with ucts can provide constraints on the origin of £uids the size of the Eu anomaly (Fig. 3D). Saharan responsible for their formation [5,14]. Table 2 and weathering processes are unlikely to have resulted Fig. 3C give the range of rare earth element in such a relationship. Finally, no Ce anomaly has (REE) concentrations measured in the orange been found in the REE patterns of the alteration phase located in olivine by in situ ion microprobe phase (Fig. 3C). Therefore, the REE abundances analyses [15,16]. The alteration phase is character- in the alteration product are unlikely to have been ized by a striking V-shaped REE pattern with a the result of terrestrial weathering. Additional marked positive Eu anomaly (Fig. 3B). REE work, including analyses of light elements (Li, abundances in Saharan meteorites can be strongly Be, BT) should further con¢rm this result. modi¢ed by surface processes [12,17,18]. There- fore, the e¡ects of terrestrial weathering need to Table 2 be evaluated. It has previously been noted that Range of REE abundances (in ppm) in the alteration phase. the REE patterns of terrestrially weathered min- AP 3^3.4 is the average of two analyses from the same area erals usually show light-REE (LREE) enrichment, AP 2^2 AP 3^3.4 and often have a positive or negative Ce anomaly. Y 1.81 þ 4 3.17 þ 5 Unlike the Libyan shergottites from the Dar al La 1.46 þ 5 1.94 þ 5 Gani area, NWA 817 is a relatively fresh rock Ce 2.60 þ 7 4.11 þ 8 and contains only very minor amounts of desert Pr 0.32 þ 2 0.48 þ 2 alteration products (such as calcite), mostly lo- Nd 0.95 þ 3 1.78 þ 4 cated near the surface or in some fractures. Mag- Sm 0.15 þ 1 0.41 þ 2 Eu 0.20 þ 2 0.17 þ 4 matic phases, with the exception of olivine, do not Gd 0.08 þ 3 0.18 þ 4 show LREE enrichment [19]. Olivine has the low- Tb0.019 þ 5 0.054 þ 6 est REE abundances of all the magmatic phases Dy 0.17 þ 1 0.41 þ 2 and is thus extremely sensitive to the addition of Ho 0.051 þ 5 0.115 þ 8 even small amounts of LREE during terrestrial Er 0.21 þ 1 0.46 þ 2 Tm 0.036 þ 4 0.090 þ 6 weathering. Therefore, only very small amounts Yb0.32 þ 2 0.74 þ 3 of LREE are likely to have been added by terres- Lu 0.067 þ 6 0.152 þ 9 trial contamination, which resulted in moderate (La/Sm)n 6.20 2.95 enrichment of these elements in the olivine (La (Dy/Yb)n 0.36 0.36 V0.5UCI) but cannot account for the extreme (Eu/Sm)n 3.54 1.09

EPSL 6354 24-9-02 438 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444

The heavy-REE (HREE) pattern in the altera- tion phase is similar to that in olivine (- normalized Yb/Dy V3) although average concen- trations are somewhat higher in the alteration phase (Yb V2^5UCI versus V0.7^1.1UCI in ol- ivine). As noted earlier, the degree of LREE en- richment in the alteration phase is variable (chon- drite-normalized La/Sm V3^6) and correlates with the size of the Eu anomaly (chondrite-nor- malized Eu/Sm V1.0^3.5) (Fig. 3C). These REE distributions could be explained assuming that the £uids that deposited the alteration product perco- lated through a nakhlitic sequence and dissolved various magmatic phases to di¡erent degrees. Dis- solution of olivine, which has a steep HREE-en- riched pattern, is likely to be the major source of the HREE; this would explain the similarity of the HREE pattern of the olivine and the altera- tion product. The dissolution of two other miner- als, rich in REE and present in the nakhlites, must be involved to explain the LREE patterns and Eu anomalies in the alteration product. Chlorapatite, which is the principal carrier of REE in the nakh- lites, and plagioclase both have LREE-enriched patterns (chondrite-normalized La/Sm V3 and V6, respectively). Dissolution of plagioclase is required to explain the positive Eu anomaly ob- served in the alteration product. Therefore, vari- able contributions from these two minerals can account for the range of LREE enrichments and the sizes of Eu anomalies in the alteration phase. It is worth noting that on Earth, hydrothermal £uids in mid-oceanic ridge basalts also have a positive Eu anomaly, resulting from dissolution of plagioclase in the oceanic crust by oceanic water [20]. In the absence of a more precise crystal Fig. 3. (A) Comparison of the major element compositions of the alteration phase with those of the magmatic phases in chemical study, including a structure re¢nement, NWA 817. The compositions of the alteration phase show it is impossible to provide a structural formula for correlations between the Si and Fe contents (uppermost and the alteration phase and a detailed interlayer cat- lowermost diagrams). (B) Chemical Fe/Mg pro¢le measured ion content of this smectite-like mineral. It is also on a veinlet of the alteration phase occurring in an olivine di⁄cult to propose at the present stage a compre- crystal (see also Fig. 1D). (C) Comparison of the REE com- positions of the alteration phase with those of the magmatic hensive model for the formation of the alteration phases in NWA 817. The NWA 817 whole-rock data are phase from a £uid of yet unknown composition. from Sautter et al. [7]. REE abundances are normalized to Nevertheless, the chemical data presented in this the values of Evensen et al. [40]. (D) (Eu/Sm)n vs. (La/Sm)n study provide a general outline of the alteration in the alteration phase. Same abbreviations as in Fig. 1. process which took place in this martian mag- matic rock. The composition in major elements of the alteration phase could be roughly satis¢ed

EPSL 6354 24-9-02 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444 439

Fig. 3 (Continued).

by a mix of olivine and mesostasis [3,4]. However, 3170 þ 14x, respectively). We note that the the trace element composition of the alteration large spread in the ND values in the magmatic phase cannot be accounted for by such a simple minerals (of V3140x) cannot be attributed to mixture. a statistical error in the measurements since this error is in the range of 30^40x for low hydrogen (water) content. 6. Hydrogen isotopes in NWA 817 There may be two possible explanations for these hydrogen isotopic systematics, both of Hydrogen isotope measurements have been which require that the ND value recorded in the made in situ with an ion microprobe on magmatic alteration phase re£ects that of the £uid from minerals and the alteration phase (Fig. 4) [21]. which it was formed. The spread in the ND values Details of the analytical procedure are given in in the magmatic minerals could result from het- the Appendix. The measured ND values for olivine erogeneous isotopic fractionation between these and clinopyroxene range from 3100 to 3240x, minerals and the £uid that deposited the altera- and in the alteration phase from 3147 to tion phase. Isotopic fractionation between £uids 3181x. Although the range of the ND values and silicates is generally too small to account for in the magmatic minerals is signi¢cantly larger the observed variation of ND in the magmatic than in the alteration phase, the mean values minerals, but redox reactions (accompanied by for both are similar (i.e. 3182 þ 32x and liberation of H2) of the £uid with the silicates

EPSL 6354 24-9-02 440 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444

phase formed on Earth due to the circulation of terrestrial water in the meteorite. ND values of rainwater and groundwater in Northwest Africa have been found to have minimum values down to only V380x in very exceptional cases and locations [23,24]. Let us use this value as a limit- ing one. Sheppard and Gilg [25] have reviewed the water^clay D/H fractionation. For smectites, hav- ing similar iron content as the alteration product of NWA 817, the low temperature D/H fraction- ation factor with water can reach values down to 360x. Therefore, the alteration phase, if formed on Earth, could reach ND values down to Fig. 4. Hydrogen isotope data obtained by ion microprobe 3 x on the magmatic minerals (olivines and augites) and on the 140 if it is precipitated from water with a alteration product of NWA 817. The data are expressed in ND=380x. This value is signi¢cantly higher ND units. than the one measured in the alteration phase. Moreover, the texture and setting of the alteration phase within magmatic minerals rule out the cir- culation of large amounts of water in the meteor- under reducing conditions can lead to much high- ite and thus the permanent connexion with an er fractionation (up to a few 100x) [22] and may external water reservoir. Therefore, the progres- leave a heterogeneous isotopic signature in the sive growth of the alteration phase from water magmatic minerals that previously had uniform within the meteorite will lead to an increase to- ND values. Another possibility is that the ND for ward higher values of the D/H ratio of both water the two types of minerals were both set at the and alteration phase. This phase in the D/H ratio time of deposition of the alteration phase. Subse- is not observed in the alteration phase, which in quently, interaction and incomplete equilibration turn exhibits rather constant values. with other £uids (on Mars and/or possibly on Contamination by terrestrial H (adsorbed on Earth) with somewhat di¡erent ND compositions mineral surfaces or resulting from water^mineral could result in a larger spread in the ND values in exchange) after the arrival on Earth must also be the magmatic minerals. This is because these discussed. The few negative ND values previously (nominally anhydrous) minerals have a signi¢- reported in Martian meteorites have been ex- cantly lower hydrogen content than the alteration plained in terms of contamination by terrestrial phase and are, therefore, more susceptible to hav- adsorbed water [26]. However, water adsorbed ing their hydrogen isotopic systematics altered by during sample preparation, estimated by monitor- N þ þ £uids with di¡erent D compositions. ing the H2 /H ratio as described by Deloule et al. How reasonable is it to assume that the mean [21], was minimal in our analysis of magmatic ND value of 3170 þ 14x recorded in the altera- minerals and alteration phase, con¢rming that tion phase indeed re£ects the hydrogen isotopic the measured ND values are representative of hy- composition of £uid from which it was formed drogen located in the bulk of the minerals. Never- on Mars and not that of terrestrial water? Even theless, it may still be argued that the hydrogen if the textural arguments rule out the formation of located in the bulk of the minerals in NWA 817 the alteration phase on Earth we discuss in detail has exchanged with terrestrial water and that the the potential e¡ects of the interaction with terres- negative ND values measured in this rock were trial water on the pristine D/H ratio of the mete- produced during extensive circulation of water orite, even if the textural arguments rule out the in this meteorite during its residence time in the formation of the alteration phase on Earth. Saharan desert. However, this scenario is unlikely First of all let us assume that the alteration and several lines of evidence argue against a sig-

EPSL 6354 24-9-02 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444 441 ni¢cant alteration of the ND values in the alter- 7. Conclusion ation phase during terrestrial weathering of NWA 817. As noted earlier, textural and chemical evi- The formation of the alteration phase, a ferro- dence indicates very limited circulation of terres- magnesian silicate of the smectite family, in NWA trial £uids in NWA 817. Thus, although water- 817 indicates a post-magmatic alteration process bearing clays in meteorites can exchange H with occurring under low-temperature conditions (less percolating water under desert conditions [27],it than 500‡C) [32]. Similar processes are known on seems extremely di⁄cult to obtain the most neg- Earth. Smectite formation has been observed as a ative ND values observed in both the alteration result of the interaction between basalts and other phase and magmatic minerals of NWA 817 by volcanic rocks with juvenile water (from the man- this means. In fact, heated desert water percolat- tle) or seawater [32]. If the low mean ND value in ing through the meteorite will be subjected to the alteration phase is indeed pristine and unaf- evaporation, with the consequence of a shift of fected by the residence of NWA 817 on Earth, its ND toward higher values [28]. Changes in ND then it represents a water reservoir on Mars toward lower values induced by Saharan altera- with a ND value distinct from that of the present tion have been described in carbonaceous chon- Martian atmosphere. We note that this low ND drites [29]. However, in this case the depletion in value is close to the mean value of carbonaceous D was apparently related to the hydrolysis of or- [33] and may re£ect a ‘primitive’ un- ganic matter. The absence of a signi¢cant organic fractionated water reservoir on Mars. The striking component in NWA 817 and the fact that the ND similarity of NWA 817 with the other three nakh- measurements were conducted on individual, well- lites strongly suggests that it also has a similar characterized phases rule out this possibility. crystallization age (of V1.3 Ga) [1,34]. Thus, Therefore, only the highest ND values observed the time of formation of the alteration phase in the magmatic minerals may be reasonably at- was most likely 6 1.3 Ga. By this time, the tributed to a Saharan alteration process. Conse- Mars atmosphere was already highly fractionated quently, it is highly unlikely that such alteration in its hydrogen isotopic composition. For a water can produce the low ND values measured in the reservoir to remain unfractionated on Mars 6 1.3 main water-bearing phase in NWA 817 (alteration Ga ago, it must be isolated from the Martian at- phase) and it is reasonable to assume that the mosphere and groundwater (which likely equili- mean ND value in this phase (3170 þ 14x) rep- brated with the atmosphere). Such a water reser- resents the hydrogen isotopic composition of its voir is most likely to be in the Martian mantle. parent £uid on Mars. Several previous studies have suggested that Mars This ND value is signi¢cantly lower than those has not sustained global plate tectonics [35^37] measured in other nakhlites and most other Mar- and therefore its mantle could e¡ectively remain tian meteorites (by either step heating experiments isolated from crustal and atmospheric water res- on bulk samples [26] or by ion microprobe anal- ervoirs. Therefore, the alteration products in yses on individual H-rich minerals) [6,26]. ND val- NWA 817 likely represent the ¢rst evidence of ues reported so far in Martian meteorites are alteration of the Martian crust by a ‘primitive’ mostly positive and range between +500 and aqueous £uid (i.e. with an unfractionated hydro- +4300x [6,26,30]. For the magmatic minerals, gen isotopic composition). A further implication these elevated and variable values have been in- of a low D/H ratio for primitive is terpreted in terms of interaction to di¡erent de- that at a signi¢cant proportion of the Martian grees with Martian groundwater that had equili- surface hydrogen (or water) must have been lost brated with the highly fractionated D-enriched through time. Assuming a NDof3170x for the Martian atmosphere (with a present ND value of primitive water and the current ND of the atmo- V+4000x) [31]. Clearly, the negative ND values sphere of +4200x, the surface hydrogen (water) in the alteration phase and magmatic minerals of reservoir on Mars needs to be depleted by at least NWA 17 cannot be similarly interpreted. V84%.

EPSL 6354 24-9-02 442 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444

Acknowledgements Jobin-Yvon0 spectrometer (located at the Institut de Physique du Globe de Paris) equipped with This collaborative work has been initiated confocal optics and a nitrogen-cooled CCD detec- through the French Martian sample program tor. A microscope is used to focus the excitation from the CNES and the French program PNP laser beam (514 nm lines of a Spectra Physics0 from the INSU. All the authors are extremely Arþ laser and 647 nm lines of a Coherent0 Kr indebted to Bruno Fectay and C. Bidaut who laser) to a 2 Wm spot and to collect the Raman provided this extraordinary meteorite. M. Me- signal in the backscattered direction. Accumula- zouard (ESRF Grenoble) kindly performed the tions lasting from 30 min to 1 h have been made. di¡raction experiments. B. Champagnon gave us The laser power at the sample was 2^50 mW to access to the IR spectrometer. Access to the avoid deterioration of the sample. Infrared spec- Washington University ion microprobe was cour- tra were obtained with a Perkin-Elmer Spectrum tesy of G. Crozaz. We are grateful to F. Robert, GX spectrometer equipped with a microscope C. Le¤cuyer, M. Javoy, F. Albare'de and G. Crozaz stage. Transmission spectra were recorded directly for helpful discussions during this work.[AC] on the thin section on an area 30U30 mm in size. Accumulation of 1000 scans gave good spectra. Chemical compositions in major elements were Appendix. Samples and analytical techniques obtained with a Cameca SX50 electron micro- probe (located at the Universite¤ de Paris VI). Op- Three types of samples were used to character- erating conditions for the microprobe analyses ize the post-magmatic alteration products of were 15 kV accelerating voltage with a beam cur- NWA 817: two centimeter-sized chips embedded rent of 12 nA. REEs were measured in situ with a in epoxy and polished on one side; one thin sec- modi¢ed Cameca IMS-3f ion microprobe at the tion, single crystals of olivine hand-picked from University of Washington, St. Louis, MO, USA. powders and presenting reddish alteration fea- The analytical methods are similar to those de- tures. All these samples were ¢rst studied by op- scribed by Zinner and Crozaz [16] and Lundberg tical microscopy in both transmitted and re£ected et al. [15]. Secondary ions were sputtered from the light. Backscattered electron images were taken sample surface using an O3 primary beam of with a JEOL JSM6301-F scanning electron micro- 5 nA, resulting in spot sizes about 20 Wm in diam- scope equipped with an energy dispersive spec- eter. Six large areas of the orange alteration prod- trometer. High-magni¢cation images were re- uct located in two olivine crystals were analyzed. corded in the FEG-SEM mode in order to Hydrogen isotopic compositions were measured detect subtle structural and chemical details in in situ on polished sections with the Cameca IMS the alteration products. 1270 ion microprobe located at the CRPG- X-ray di¡raction measurements were performed CNRS, Nancy. An O3 primary beam of 2^5 nA on olivine single crystals at the European Syn- were accelerated at 13 kV and focused on the chrotron Radiation Facility (Grenoble, France). sample to a beam of 30 Wm in diameter. The sec- A monochromatic beam with a wavelength of ondary beam mass resolution was set at 2000, 0.3738 Aî was used. The di¡racted X-rays were with an energy window of 55 eV, without any þ þ þ recorded on a CCD camera. The sample was energy o¡set. Secondary ions H ,D and H2 held in a silica capillary and rotated by þ 30‡. were measured during 20 min each by ion count- The beam size was around 30 Wm. TEM observa- ing and peak switching. Statistical precision rang- tions were carried out on £akes of the alteration ing from þ 30x for samples with low water con- product separated by hand-picking from crushed tent to less than 10x for samples with high crystals of olivine. The £akes have been included water content were attained. Amphibole and py- in epoxy and thin slices were cut by ultra-micro- roxene standards were used to correct for the in- tomy. strumental isotopic fractionation, calculated as Raman spectra were recorded with a T64000 a function of the Mg-number of the analyzed

EPSL 6354 24-9-02 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444 443 phases [21,38]. Results are given in NDSMOW units [8] H. Leroux, Microstructural shock signatures of major minerals in meteorites, Eur. J. Mineral. 13 (2001) 253^ (NDsample = [(D/H)sample/(D/H)SMOW31]U1000) x 272. with a precision of 40^20 , including the instru- [9] B. Marty, K. Marti, Signatures of early di¡erentiation of mental calibration uncertainty. Samples were Mars, Earth Planet. Sci. Lett. (2002) in press. carefully degassed by heating under high vacuum [10] A. Jambon, J.A. Barrat, P. Gillet, C. Go«pel, M. Javoy, before introduction into the analytical chamber. J.L. Joron, V. Sautter, One more shergottite from North The instrumental background was measured on Western Africa, Meteorit. Planet. Sci. 36 (2001) A90. [11] J.A. Barrat, P. Gillet, V. Sautter, A. Jambon, M. Javoy, international standards [39]. It is lower than 10 C. Go«pel, M. Lesourd, F. Keller, E. Petit, Petrology and ppm of water with counting rates 10^100 times geochemistry of the basaltic shergottite North West Afri- lower than the lowest counting rate observed for ca 480, Meteorit. Planet. Sci. 37 (2002) 487^499. the minerals of NWA 817. The instrumental mois- [12] J.A. Barrat, J. Blichert-Toft, R.W. Nesbitt, F. Keller, ture displays a D/H ratio in the range 50^ Bulk chemistry of Saharan shergottite Dar al Gani 476, U 36 N Meteorit. Planet. Sci. 34 (2001) 91^97. 60 10 , corresponding to D values of less [13] P. Gillet, Application of vibrational spectroscopy to geol- than 3500x. Instrumental contamination dur- ogy, in: J.M. Chalmers, P.R. Gri⁄ths (Eds.), Handbook ing the analysis of minerals containing very low of Vibrational Spectroscopy, Vol. 4, John Wiley and amounts of hydrogen like olivine and pyroxene is Sons, (2002) pp. 1^23. ruled out since no correlation was observed be- [14] A.H. Treiman, D.J. Lindstrom, Trace element geochem- þ istry of martian iddingsite in the Lafayette meteorite, tween the H counting rate and the D/H ratio. J. Geophys. Res. 102 (1997) 9153^9163. During analysis of the hydrous alteration phase [15] L.L. Lundberg, G. Crozaz, G. McKay, E. Zinner, Rare 3 the vacuum remained below 10 8 Torr, indicating earth element carriers in the and im- that the interstitial water has been correctly re- plications for its chronology, Geochim. Cosmoschim. moved from the sample surface. Furthermore, in- Acta 52 (1988) 2147^2163. [16] E. Zinner, G. Crozaz, A method for the quantitative mea- terstitial water is not included in ion probe mea- surement of rare earth elements in the ion-microprobe, surements but pumped out from the sample Intl. J. Mass Spectrom. Ion Proc. 69 (1986) 17^38. during measurement and only structural water [17] J.A. Barrat, P. Gillet, M. Lesourd, J. Blichert-Toft, G.R. of the alteration phase is analyzed.[AC] Poupeau, The Tatahouine mineralogical and chemical e¡ects of 63 of terrestrial residence, Mete- orit. Planet. Sci. 34 (1999) 91^97. [18] G. Crozaz, M. Wadhwa, The terrestrial alteration of Sa- References haran shergottites Dar al Gani 476 and 489: a case study of weathering in a hot desert, Geochim. Cosmoschim. [1] H.Y.J. McSween, What we learned about Mars from Acta 65 (2001) 971^978. SNC meteorites, 29 (1994) 757^779. [19] M. Wadhwa, J.A. Barrat, G. Crozaz, Petrogenesis of a [2] A.M. Reid, T.E. Bunch, The nakhlites part II: Where, new nakhlite from rare earth and trace elements distribu- when and how, Meteoritics 10 (1975) 317^324. tion, Meteorit. Planet. Sci. 36 (2001) A216^A217. [3] J.L. Gooding, S.J. Wentworth, M.E. Zolensky, Aqueous [20] A. Michard, F. Albare'de, G. Michard, J.F. Minster, J.L. alteration in the , Meteoritics 26 (1991) Charlou, Rare-earth elements and uranium in high-tem- 135^143. perature solutions from East Paci¢c Rise hydrothermal [4] A.H. Treiman, R.A. Barrett, J. Godding, Preterrestrial vent ¢eld (13‡N), Nature 303 (1983) 795^797. alteration of the Lafayette (SNC) meteorite, Meteoritics [21] E. Deloule, C. France-Lanord, F. Albare'de, D/H analysis 28 (1993) 86^97. of minerals by ion microprobe, in: H.P. Taylor, J.R. [5] J.C. Bridges, M.M. Grady, Evaporite mineral assemblages O’Neil, I.R. Kaplan (Eds.), Stable Isotope Geochemistry: in the nakhlite (martian) meteorites, Earth Planet. Sci. a Tribute to Samuel Epstein, Vol. 3, Geochemical Society, Lett. 176 (2000) 267^279. Houston, TX, 1991, pp. 53^64. [6] L.L. Watson, I.D. Hutcheon, S. Epstein, E.M. Stolper, [22] P. Richet, Y. Bottinga, M. Javoy, A review of hydrogen, Water on Mars clues from deuterium/hydrogen and water carbon, nitrogen, oxygen and chlorine stable isotope frac- contents of hydrous phases in SNC meteorites, Science tionation among gaseous molecules, Annu. Rev. Earth 265 (1994) 86^90. Planet. Sci. 5 (1977) 65^110. [7] V. Sautter, J.A. Barrat, A. Jambon, M. Javoy, J.P. Lor- [23] J.R. Gat, A. Issar, Desert isotope hydrology: water sour- and, P. Gillet, J.L. Joron, M. Lesourd, A new martian ces of the Sinai desert, Geochim. Cosmochim. Acta 38 meteorite from Marocco: the nakhlite North West Africa (1974) 1117^1131. 817, Earth Planet. Sci. Lett. 195 (2002) 223^338. [24] R. Gon¢antini, T. Dincer, A.M. Derekoy, Environmental

EPSL 6354 24-9-02 444 P. Gillet et al. / Earth and PlanetaryScience Letters 203 (2002) 431^444

isotope hydrology in the Hodna region Algeria, in: Iso- Society of America, Washington, DC, 1988, pp. 497^ tope Techniques in Groundwater Hydrology, IAEA, 522. 1974, pp. 293^316. [33] E. Deloule, F. Robert, J.C. Doukhan, Interstellar hydrox- [25] S.M.F. Sheppard, A. Gilg, Stable isotope geochemistry of yl in meteoritic : implications for the origin of clay minerals, Clay Miner. 31 (1996) 1^24. water in the inner , Geochim. Cosmochim. [26] L.A. Leshin, S. Epstein, E.M. Stolper, Hydrogen isotope Acta 62 (1998) 3367^3378. geochemistry of the SNC meteorites, Geochim. Cosmos- [34] B. Marti, K. Marty, K.V. Mathew, Meteorit. Planet. Sci. chim. Acta 60 (1996) 2636^2650. 36 (2001) A122. [27] S.M. Savin, S. Epstein, The oxygen and hydrogen isotope [35] C.L.J. Harper, L.E. Nyquist, B.M. Bansal, H. Wiesmann, geochemistry of clay minerals, Geochim. Cosmoschim. C.Y. Shih, Rapid accretion and early di¡erentiation of Acta 34 (1970) 25^42. Mars indicated by 142Nd/144Nd in SNC meteorites, Sci- [28] J.R. Gat, C. Browser, The heavy isotope enrichment of ence 267 (1995) 213^216. water in coupled evaporative systems, in: H.P. Taylor, [36] D.C. Lee, A.N. Halliday, Core formation on Mars and J.R. O’Neil, I.R. Kaplan (Eds.), Stable Isotope Geochem- di¡erenciated , Nature 388 (1997) 854^857. istry: a Tribute to Samuel Epstein, Vol. 3, Geochemical [37] M. Wadhwa, Redox state of Mars’ upper mantle and Society, Houston, TX, 1991, pp. 159^168. crust from Eu anomalies in shergottite pyroxenes, Science [29] R.D. Ash, T. Pillinger, Carbon, nitrogen and hydrogen in 291 (2001) 1527^1530. saharan chondrites: the importance of weathering, Mete- [38] C. Engrand, E. Deloule, F. Robert, M. Maurette, G. oritics 30 (1995) 85^92. Kurat, Extraterrestrial water in and cos- [30] L.A. Leshin, Insigths into martian water reservoirs from mic spherules from Antarctica: an ion microprobe study, analyses of martian meteorite QUE94201, Geophys. Res. Meteorit. Planet. Sci. 34 (1999) 773^786. Lett. 27 (2000) 2017^2020. [39] D.R. Bell, P.D. Ihringer, G.R. Rossman, Quantitative [31] G.L. Bjoraker, M.J. Mumma, H.P. Larson, Isotope abun- analysis of trace OH in garnet and pyroxenes, Am. Min- dance ratios for hydrogen and oxygen in the martian at- eral. 80 (1995) 465^474. mosphere, Bull. Am. Astron. Soc. 21 (1989) 991^996. [40] N.M. Evensen, P.J. Hamilton, R.K. O’Nions, Rare earth [32] N. Gu«ven, Smectites, in: S.W. Bayley (Ed.), Hydrous abundances in chondritic meteorites, Geochim. Cosmo- Phyllosilicates, Reviews in Mineralogy 19, Mineralogical chim. Acta 42 (1978) 1199^1212.

EPSL 6354 24-9-02