Geochrmrca n Cosmorhlmrca Acta Vol. 50. pp. 909-926 0 Pergamon Journals Ltd.1986. Printed in U.S.A.
Chemical systematics of the Shergotty meteorite and the composition of its parent body (Mars)* J. C. LAUL and M. R. SMITH Chemical Technology Department, Battelle, Pacific Northwest Laboratories. Richland. WA 99352. U.S.A.
and H. WANKE, E. JAGOUTZ, G. DREIBUS.H. PALME, B. SPE~EL and A. BURGHELE Max-Planck-lnstitut fur Chemie. Saarstrasse 23, D-6500 Mainz. Federal Republic Germany and M. E. LIPSCHUTZand R. M. VERKOUTEREN Chemistry Department, Purdue University, West Lafayette. IN 47907. U.S.A. (Received June 28. 1985; accepted in revisedjbrm January 30, 1986) Abstract-We report chemical data for 60 elements by INAA and RNAA in two bulk samples, for 30 elements in various mineral separates of Shergotty, and results of leaching experiments with IM HCI on powdered ahquots of Shergotty and EETA 7900 1, lithologies A and B. Shergotty is homogeneous in major element composition but heterogeneous with respect to LIL trace elements (-20%). The heterogeneity is even greater for volatile and siderophile trace elements. The mineral data, including three clinopyroxene fractions with variable Fe0 contents, maskelynite and minor phases (Ti-magnetite, ilmenite, quartz, K-rich phase), show that major minerals do not account for the rare earth elements (REE) in the bulk meteorite. Instead, the REE are to a large extent concentrated in accessory whitlockite and apatite (shown by leaching with 1M HCI): together with the majority of REE (La. 96%, Yb 70%) Cl and Br are quantitatively dissolved by leaching. The REE patterns of the leachate of Shergotty and EETA 79001 are different. The Shergotty leachate may consist oftwo components. Component 1 is similar to that of EETA 79001 leachate (whitlockite), component 2 is enriched in light REE and may be responsible for the higher LREE contents of Shergotty in comparison to the other shergottites. There is some evidence that Shergotty was an open system and component 2 was introduced after crystallization. The REE patterns of the residues of Shergotty and EETA 79001 are identical indicating that the parent magmas ofboth meteorites are compositionally similar. Based on cpx separates with the lowest REE content, the REE pattern in the Shergotty parent magma was calculated. It is enriched in LREE and has a subchondritic Nd/Sm ratio. The negative Eu anomaly in the phosphates indicates that at least some plagioclase crystallized before phosphate. Based on several element correlations in SNC meteorites, it was suggested (DREIBUS and W~~NKE, 1984) that both the Shergotty parent body (SPB, very probably Mars), and the Earth accreted from the same two chemically different components: component A. highly reduced and devoid of volatile elements and an oxidized component B containing also volatile elements. The SPB (Mars) mantle is 2-4 times richer in volatile and moderately siderophile elements than the Earth, indicating a higher portion of component B in the SPB. The concentrations of chalcophile elements in the SPB mantle are low, reflecting equilibration with a sulfide phase and subsequent segregation of sulfide into the core. Unlike the Earth (WXNKE,1981). the SPB (Mars) may therefore have accreted almost homogeneously. INTRODUCTION
THERE ISINCREASINGgeochemical evidence that three groups of rare achondrites, shergottites, nakhlites, and Chassigny (SNC meteorites) come from the planet Mars (NYQUIST et al.. 1979; STOLPERand MCSWEEN, 1979; WOOD and ASHWAL, 1981; SHIH et al., 1982; DREIBUSet al., 1982; SMITH et al., 1984; B~GARD et al., 1984; DREIBUS and W.&NKE, 1984). The recent discovery of four lunar meteorites (ALHA 8 1005, YA 791197, YA 82192, and YA 82193) has added credibility to the Martian hypothesis (Geophys. Res. Lett. 10, 1983; Sym. Anturc. Meteor. Japan, 1985). It is ob-
viously possible to eject meteorites from a body like the Moon. The SNC meteorites (4 shergottites, 3
* Presented at the Shergotty-Nakhla-Chassigny Symposia, 16th Lunar and Planetary Science Conference, 1985.
nakhlites. and Chassigny) have a number of properties in common. They show igneous unbrecciated textures with mineralogies indicating a high oxygen fugacity environment. They exhibit a distinctive ‘80/‘60 ratio (CLAYTON and MAYEDA, 1982, 1983), complex REE patterns and they have similar abundance ratios for FeO/MnO, K/La, K/U, W/La, Ga/La, and Na/Al. Compared to other achondrites they have high volatile element contents (WOOD and ASHWAL,198 1; DREIBUS et al.. 1982; DREIBUS and WANKE, 1984). All these chemical and isotopic similarities suggest a common parent body. However, the most striking feature among SNC meteorites is their common young crystallization age of less than 1.3 AE, far younger than any other known meteorite. A recent crystallization age of 360 ?I I6 m.y. for Shergotty by the Sm-Nd method (JAGOUTZ and WANKE, 1986) makes Shergotty the youngest known meteorite. A detailed chemical study is therefore of great interest.
J. <‘. Laul er al
Among the SNC meteorites, there are however differences in terms of exposure ages, shock histories. mineralogies and trace element signatures. The group of shergottites (Shergotty, Zagami, EETA 7900 1 and ALHA 77005) have a number of properties in common. They have REE patterns with a smooth rise (no Eu anomaly) towards heavy REE and a variable hump between Gd-Ho. Shergottites have been appreciably shocked, as evidenced by the plagioclase conversion to maskelynite (STOLPER and MCSWEEN, 1979). Shergotty and EETA 79001 provide direct evidence in favor of Martian origin. Shergotty’s major element composition (e.g. Fe0 content) is similar to the Martian soil composition measured by Viking spacecraft (CLARK et al., 1982; MCSWEENand STOLPER, 1980). However, the strongest evidence for a Martian origin comes from trapped noble gases in glasses ofEETA 79001 (BOGARDand JOHNSON, 1983) which have similar elemental and isotopic ratios as those measured in the Martian atmosphere (OWEN et al., 1977). In addition to the rare gas isotopes, the 15N/14Nand 14N/40Arratios in the same shock-altered phases (glasses) of EETA 79001 agree weIl with the Martian atmospheric measurements (BECKERand PEPIN, 1984; WIENS el al., 1986). Shergotty fell in I865 in India. Through the courtesy of the Geological Survey of India, Calcutta, 30 grams of Shergotty were made available for multidisciplinary studies by various groups in a consortium. We, in a miniconsortium of Battelle-Northwest, Mainz and Purdue groups, focussed on the chemical aspects of Shergotty. Bulk chemical data on Shergotty have been reported in the literature by DUKE (1968), MCCARTHY et al. ( 1974), SMITH and HERVIG (1979) and STOLPERand MCSWEEN(1979), and trace elements such as REE by SCHNETZLERand PHILPOTTS( 1969), SHM et al. ( 1982) and DREIBUS el al. (1982). However, Shergotty is a coarse-grained rock and there are variations in trace element contents due to sampling. Here we report chemical data on 60 elements (major, minor and trace) in the same aliquots by instrumental and radiochemical neutron activation analysis (INAA and RNAA) and XRF. Two fragments of Shergotty, B and C (- 7 cm apart), were analyzed to examine for chemical variations. In addition, we report chemical data (30 elements) in Shergotty mineral separates (four clinopyroxene separates of varying Fe content, maskelynite. quartz, a K-rich phase, ilmenite, Ti-magnetite) and 1M HCl leachates and residues from bulk samples of Shergotty. In our petrogenetic modeling of Shergotty and discussions pertinent to other shergottites, we also include chemical data on the bulk, clinopyroxenes, phosphate-rich phase, and 1M HCl leachate and residue from EETA 79001 (lithologies A and B). EXPERIMENTAL OurmainfocuswasonfragmentsB(12gm)andC(l3gm) which were taken -7 cm apart and represented opposite sides of the main Shergotty meteorite. Various scientific studies
and the sample distribution plans from fragments B and C’ are outlined in detail by LAUL ( 1986). In brief, a split from fragment B, which weighed 3.5 gm, was broken into small chips (>5 mm) and labelled as SH23. In the Mainz Laboratop, 1.3 gm of SH23 was pulverized in an agate ball mill and labelled as SH23 I. Different splits of SH23 1 powder were allocated to each of the three laboratories for analyses. From fragment C (SH I), different subaliquots were assigned to each of the three laboratories by the Johnson Space Center (JSC) Various minerals were separated from fragment C tSH 10.6.3 gm) by E. Jagoutz, Mainz. Ten-minute leaching experiments with lM HCI were performed on fragment B powder. Anslytical procedures were as follows: Bulk samples. Bulk samples were analyzed lor 60 elements (major, minor and trace) by three laboratories (BattelleNorthwest; Max-Planck-Institut. Mainz; and Purdue University) using mainly INAA and RNAA. The elements Si, Ti. Al, Mg, Ca. and P in fragment B were measured by XRF. S using a Leybold Heraeus analyzer (CSA 2002). and C and F by a specific ion electrode method employed at the Maim Laboratory. Detailed descriptions of the INAA and RNAA methods applied by the Mainz group are outlined by WKNKI et al. (1972. 1973) and that of Battelle-Northwest by LAUL (1979) and LAUL et 01. (1982). The contents of W, Cu, As, Au and MO were determined by a metal extraction technique (RAMMENSEEand PALME, 1982). The Purdue group determined the concentrations of a suite of I5 trace siderophile, chalcophile. and volatile elements (Ag, As, Au, Bi, Cd, Co. Cs, Ga, In, Rb, Sb. Se, Te, Tl, Zn and U) by RNAA using radiochemical separation techniques outlined by VERKO~JTEREN and LIPSCHUTZ(1983). Analytical results of all these investigations are compiled in Table I. Mineral separules. The procedure for mmeral separation is described by JAGOUTZand W~;NKE(1986). Two to four mg each of clinopyroxene separate (Fe-rich to Fe-poor), maskelynite, quartz, ilmenite, and Ti-magnetite were tifit analyzed by INAA using normal Ge(Li) and coincidence/noncoincidence Ge(Li)-NaI(Tl) counting for the suite of 25-30 elements at Battelle-Northwest. The coincidence/noncoincidence system permits greater sensitivity in the determination of many elements, especially for small samples. Since measurements for REE were important in order to obtain well defined REE
patterns for the mineral fractions,the sampleswere reimuiiited and processed by RNAA using the REE separation scheme of LAUL et al. (1982).
Leaching with1M HCI. The bulk powder was first inadiated with thermal neutrons and then treated with I M HCI for 10 minutes and filtered through a fine filter paper. Solution and residue were counted by INAA for REti, Cl, Br, As, K, SC, Co, Cr. Na,O. FeO. CaO. and P,Os (B-countinel Since not all the REE-in the residue could be well determinYkdby INAA, the sample was therefore processed by RNAA using the REE separation scheme OfLAUL ef al. (I 982). Leaching experiments were performed at Mainz and Battelle-Northwest CHEMICAL
Table 1 shows the chemical data for 60 elements (major, minor and trace) in Shergotty bulk fragments B and C, and also the literature values. The precision of the data is mostly due to counting statistics, which in most cases is about -t 1 to 10%. At sub ppb level the precision is +-5- 15%. The agreement among the results from the three laboratories is excellent, indicating that the accuracy is in the range of precision. The REE data of the Mainz group are obtained by INAA, whereas those of Battelle-Northwest by RNAA. For the elements La, Sm, Eu, Tb, Yb and Lu, which are equally well determined by INAA and RNAA, the agreement between the two methods is excellent, whereas the concentrations of Ce. Nd. Gd. Dy. and Ho determined by INAA have larger un
Composition of Shergotty and Mars
Table 1. Chemical Data on Shergotty (Fragments B and C) by INAA and RNAA) . Benelle
8 SH231 94
0.8 7.14 19.6 9.5 9.52 1.40 0.19
TiOZ *W3 Fe0 MgD cao Na20 '(20
9.72 1.30 0.15
cr203 Sum Li ppm
54.0 260 37.5 70
SC V co Ni
56.5 265 38.5 JO
0 60 5.7.
7.27 54 67
10.5 0.044 0.025 <20 19
Cd PP~ In ppm 8b ppb Te Ppb
32.0 2.44 5.8 0.88 4.2 1.60 0.65 2.6 0.52 3.7 0.73 0 30 1.80 0.26 2.15 0.29
8a La Ca pr Nd Sm Eu Gd Tb ov HO Trll Yb LU Hf la
4.3 1 46
0.56 0.38 1.19.
2.0. 1.97 0.27. 0.25 0.48
at al. fl982).
et al. (1982)
et at. data
et al. (1982)
12. 15, 16
Bulk data are taken
TI Th ppm U
4.7 0.70 3.3 1.30 0.55 2.1 0.41 3.0 0.60 0.25 1.50 0.22 1.50 0.18
W jr ppb AU
1 ppm CS
Mars Soil lb)
and Hervig Schnatzler
et al. (19841 subtracted
certainties (2%15%) relative to the RNAA values. Three analyses of fkgment B powder (SH23 I) by RNAA (normalized to Yb) are shown in Table 2. The absolute concentrations of
the REE vary somewhat, probably because of varying amounts of accessory phosphate phases (discussed later). However, when the data are normalized to Yb in each run, the precision of
J. C. Laul et al.
912 Table 2.
of REE Data in Shergotty
!A11 vdlues in pp~l
This Work RNAA'.
Wt. 'mg! *
1 1.80 = :.RO
*The data is from Jagoutz and W&ke based on counting
jhih et al (1982)
: .58 1.46 _ _ _~___ ____-_(1986). statistics
are mostly *l-5%.
I* each RNAA run 1s normalized
the triplicate analyses is very good (~5%). The RNAA data for Nd and Sm for fragment B powder are in excellent agreement with the isotope dilution data of JAGOUTZand W~;NKE ( 1986). The REE data reported by SHIH ef a/. ( 1982) are h> isotope dilution (I.D.). Although their absolute REE concentrations are about 60% lower than ours, their relative elemental ratios such as Nd/Sm of 2.57 is in excellent agreement with our Nd/Sm ratios of 2.62 determined by RNAA (Table 2). Bulk composition. Fragments B and C are essentially identical in major and minor elements, and arc characterized b) 50% SiOz, 0.87% TiOz, 7.0% A&03, 19.7% FeO. 9.5% MgO. 9.7% CaO, 0.19% KzO, 0.52% MnO and 0.18% CrzOx (Table I ). Fragment A, not listed in Table I, was also analyzed for a few major and trace elements. and contains 19.2% FeO. 10.0% CaO, 0.15% K20. 1.4%Na*O, and 0.20% CrzOz. These concentrations are identical to those of fragments B and C. These values agree well with the literature values reported b) DUKE (1968) MCCARTHY et al. (1974), SMITH and HERVIG ( 1979) and STOLPER and MCSWEEN(1979) indicating that Shergotty is homogeneous in bulk composition. Its composition is similar to the Martian soil composition (Table Ii reported by CLARK er al. ( 1982). Large-ron-lirhuphile (LIL) elements. Fragments B and C’ differ in many LIL trace elements (K, Ba, Sr, REE. Zr. Hf Ta and Th). The RNAA data for REE of fragments B and C‘ are normalized to C I abundances (PALME et al.. 198 I) and the patterns are shown in Fig. 1. The REE patterns are about 10X Cl-chondrite with a slight decline from La-Nd and increase from Nd-Sm and show a smooth rise towards heavy REE with a maximum (small hump) at Dy (14X) for sample B. The difference in the REE contents between samples B and C IS around 20%. If we include the REE data of SHIH d al. ( 1982). the variation spans 60 to 70%. Our limited REE data for fragment A are identical to values of SHIH et al. However. the SHIHet a/. REE pattern is very similar to our REE patterns for fragments B and C. The Cl chondrite normalized Sm/Nd ratio of 1.20 by SHIH et al. (1982) and JAGO~JTZand W,&NKF ( 1986) obtained by isotope dilution method is the same as ours by RNAA (Table 2, Fig. I). The REE abundances reported by DREIBUSe/ al. ( 1982) lie between those of fragments
to Yb = 1.80 ppn.
B and C with the maximum shifted towards ‘l’b. l‘hese KEI-. data were obtained by INAA and Dy and Ho may have larger errors. Nevertheless. a small hump in the region of Dy appears real. and is consistent with isotope dilution data of SHIn i‘r
Shergottv Sulk . Fragment B ( RNAn Cl Fragment c 1 * Fragment A A Dreibus et al (19821 0 Sh,h et al 119821 X Jagout and W&he
t ID t
,I /I ,,*l/ , Eu Gd Tb DY Ho Tm Lu Ei Yb REE ,on,c Radu I Sm
FIG. I. Cl normalized REE patterns and Sc in various fragments of Shergotty bulk. The REE contents including the literature values vary by a factor of about 2. All samples show a hump between Cd-Ho. C I chondrite values (ppm) are used for normalization in all figures. These values are: K 517, La 0.245, Ce 0.638. Pr 0.096, Nd 0.474. Sm 0.154, Eu 0.058, Cd 0.204. Tb 0.037. Dy 0.254. Ho 0.057, Er 0.166. Tm 0.026. Yb 0.165, Lu 0.025 and SC 5.9 (PAI.MEet 01 198 I)
Composition of Shergotty and Mars
(Fe0 = I?‘&%) with Fe/(Fe + Mg) atom ratios (Fe’) of 0.57, 0.41 and 0.36 respectively (Table 3). The Fe0 contents are within the range of microprobe data of 1f-35% Fe0 (STOLPERand MCSWEEN, 1979). A CaO content of 7.6% suggests that FeO-rich cpx is pigeonite, whereas a CaO content of 15.3% indicates that the FeOpoor cpx is primarily augite. In our cpx separate with 2 1.4% Fe0 and 10. I % CaO, both an augite and a pigeonite component must be present, while the FeOpoor cpx fractions represent the cores of the pyroxenes. Figure 2 displays the Cl normalized abundances of K, REE, SC, V, Cr for the Fe-rich “rim” cpx to Fepoor “core” cpx. A split from the Fe-rich cpx was remeasured for REE by RNAA and its REE concentrations turned out to be lower. Obviously, this sample is highly inhomogeneous. None of the other mineral separates analyzed by JAGOUT~ and WXNKE (1986) had Nd or Sm concentrations as high as our Fe-rich cpx sample. All these other separates were prepared in the same manner as our sample by handpicking from the same magnetic fraction. The sample with the highest Nd and Sm contents analyzed by JAGOUTZ and W~;NKE had a Nd content of 2.27 ppm, compared to 9.0 ppm Nd of the Fe-rich cpx in this work. As discussed below, it is very likely that the high LREE concentrations in the Fe-rich cpx separate are due to the admixture of trace amounts of phosphates which may contain up to 0.1% La (SIMON et al., 1985).
al. ( 1982). Although the absolute abundances of LIL elements vary considerably, their normalized patterns are similar. The REE are mostly concentrated in accessory phases such as phosphate(discussed later}. Variations in the absolute contents of REE may reflect the modal abundances of phosphates present in the samples. Volatileand siderophileelements. The heterogeneity is even greater for voiatile elements such as Cd, Te, TI, and Bi and the sideropbiles Au and Ag (TabIe 1). Sampie SH23 1(fragment B) was ground in an agate ball mill to a grain size of less than IO pm. It is therefore conceivable that the high concentrations of some of these elements were introduced during processing, although data reported in the literature are similarly high or
are even higher (see Table I ). Mineral chemistry
Chemical data on 30 elements are listed in Table 3 for Shergotty minerals: three clinopyroxene fractions with different Fe0 contents, maskelynite, quartz, a Krich phase, Ti-magnetite, and ilmenite. The REE were measured by RNAA. Petrographic studies indicate that Shergotty contains 70% pyroxene, 25% maskelynite, 2% magnetite, 0.3% pyrrhotite, 0.2% ilmenite, and 2% whitlockite (STOLPERand MCSWEEN, 1979; SIMON et al., 1985). Pyroxenes consist of augite and pigeonite in equal proportion (STOLPERand MCSWEEN, 1979). Pyroxene. Our three clinopyroxene samples show noticeable differences in chemical composition ranging from Fe-rich (Fe0 = 30.9%) to Fe-poor dinopyroxene
Table 3. Chemical Data in Shergotty Minerals by INAA and RNAA”
CPX Fe-Rich (%I
no, [email protected]
Fe0 Mg’J cso F&,0 KiO Ml0 Ca
0.401.3 30.9 13.0 76 0.30 0.020 0.76 0.174
-zz <0.5 1.3 21.4 17.4
1.4 17.4 17.4
10.1 0.15 0.018 0.63 0.371
15.3 0.19 0.010 0.54 0.550
60 520 46.0 120 50 <40 0.27 0.81 <2 0.32 0.14 0.68 0.19 <3 0.26 0.11 0.67 0.10
<0.5 10.1 5.40 0.32
co.5 1.1 1.30 0.21
Q&p;82 20.3 2.0 58.5
8.6 6.7 36.7
2.20 13.6 12.2
3.8 2.75 1.29
<2 2.0 0.17 0.145
5.2 7.2 1.30 0.185
3.0 7.0 t .30 1.70
4.5 2.5 (4001
2070 290 1410
1020 49.4 3450
400 36.0 _
10 200 0.27
20 40 0.23
40 100 1.3
460 <50 2.30
200 <50 3.40
130 330 25.5
0.67 <2 0.15 0.10 co.4 0.051 0.4 LO.181 IO.0551 0.23 0.032 3 1 0.20 0.10
3.3 2.5 0.80 0.46 0.20 1.6 0 30 0.93 0.14 17.4 t.3 0.6
5.6 <6 1.30 0.41 1.6 0.44 3.4 0.62 0.25 1.36 0.19 11.0 2.82 0.56
54 30 6.70 1.90 7.0 1.05 6.3 1.20 0.60 3.40 0.52 4.7 0.80 6.0
PPm SC V CO Ni Zn Sr La CO Nd Sm Eu Cd 2 Ha Tm Yb l.U HI T* Th
78.0 290 60.0 400 90 <25 10.1 27 9.0 0.50 0.20 1 .o 0.25 <3 042 0.19 1.1 0.17 0.50
3.31 8.6 2.8 0.41 0.16 11.41 0.23 _ 0.35 0.15 1.0 0.16
130 67 c30 0.31 1.0
0.81 0.12 0.10
lRNAA was used for REE: Ia) spKt from Fe-M + t-5%
0.020 <3 0.023 10.04) 0.063 0.011 0080
o- -0, Split from ‘Rim’ CPX
= 21.4% 0 41
Fe0 = 17.4% Fe’ = 0.36
/ I , 1
Tb Dy Ho
FIG. 2. Cl normalized Shergotty Fe-rich “nm” cpx to Fe-poor “core” cpx LIL patterns. Data for tive c’p~ separates are from JACOUTZ and WANKE (IY86). The large differences in LREE contents between Fe-nch and Fe-poor cpx fractions are remarkable. Fe-rich pyroxene probably contain LREE-rich phosphates (set’ text for details).
The Sm and Nd concentrations of some other pyroxene separates which were analyzed by JAGOUTZ and W~NKE (1986) are also included in Fig. 2. The enormous Nd-Sm fractionation in our Fe-rich cpx sample is striking. Crystallization of phosphate cannot produce this fractionation, since phosphate does not discriminate between Sm and Nd. Interestingly, the light REE (LREE) are considerably enriched relative to the heavy REE (HREE) in the Fe-rich cpx, while the LREE are depleted in the Fe-poor “core” cpx. The other cpx samples lie in between. The highly fractionated LREE (e.g., La/Sm = 20.5) observed in the Fe-rich pyroxenes could reflect the fractionation of REE during closed system crystallization, as cpx preferentially incorporates HREE. Detailed calculations, however. show that the observed high enrichment of LREE cannot be modeled, even assuming extreme values for cpx/melt partition coefficients. It is therefore unlikely that the LREE are sited in the cpx. As phosphates are the main carrier of REE in Shergotty (see next section), we suspected the presence of phosphates in the pyroxene fraction. We therefore performed a leaching experiment on the residual fraction of the Fe-rich cpx-separate (first column, Table 3). An aliquot of this separate had been analyzed by RNAA, as discussed before. The 2.64 mg
treated with separately. The results of this experiment are given in Table 4. Since the LREE (e.g. La) content of the 2.08 mg fraction was significantly below that of the original 5.6’3 mg fraction. we expected high LREE concentration in the residual cpx-fraction. Surprisingly the LREE content of this fraction is even below that of the 2.08 mg fraction. A LREE-enriched phase must have been lost during handling. The results of the leaching experiment. nevertheless, show the following interesting trends. The LREE are almost completely removed from tfrc cpx-fraction, while the HREE are only slightly affected. The leachate has a Cl-normalized La/Sm ratio of2.85. while the residue has a La/Sm ratio lower than 0.2. Similar large differences between leach and residue ol pyroxene separates were found by JAGOUIZ anti WXNKE ( 1986) for the Nd/Sm ratio. We therefore believe that the high enrichments in light REE observed in the Fe-rich cpx-separates are due to the presence of trace amounts of a phase or phases soluble in diluted HCl. Phosphates are possible candidates as phosphates sometimes intersect cpx in the form of needles (JAGOUTZ and WANKE. 1986). The fraction of phosphate cpx
separate was, after neutron irradiation,
I M HCI: residue and leachate were then counted
Composition Table 4.
Chemical data of leachate with 1M HCl (10 min) and residue of Shergatty Fe-rich cpx (fraction of Fe-rich cpx of Table 3) by INAA, (P - by B-counting)
Shergotty Fe-rich cpx (2.64 ms)
Fe0 cao p205
(ppm) Na K
0.13 <0.6 0.31
70 1160 5700 73 575 ___
(SMITH and maskelynite La at 1.1 X anomaly at
HE~VIG, 1979). The REE pattern of our (Fig. 3) is reasonable for plagioclase with Cl, Sm at 0.46 X Cl and a positive Eu 12.4 X C 1. Minor phases. The composition of quartz. a K-rich phase, Ti-magnetite and ilmenite are reported for completeness. The quartz separate contains - 19% normative plagioclase (An3,AbszOr,). The K-rich phase is probably associated with intercumulus phases and is somewhat similar to the “tan” two-phase assemblage reported by STOLPERand MCSWEEN(1979). The Timagnetite separate is different from the titano-magnetite composition measured by STOLPER and MCSWEEN (1979). Our sample is richer in Ti, suggesting that it contains some ilmenite. On the other hand, the “ilmenite” separate is far too low in Ti. with .41,Ca. Mg, K. and Na abundances indicating the presence of other minerals. Trace element abundances of minerals generally portray their purity, and K, REE and SC patterns of
in the cpx-separate can be estimated from the concen_-0, 500 -0-____o-* -TX0 tration of P in the leach. A content of 1350 ppm P 'd' IPhosphate-Phase) " I (P205 = 0.3 1%) was obtained by instrumental /Y1 counting, corresponding to a phosphate content of 0.77%. Because of the intimate intergrowth between phosphate and clinopyroxene, the REE contents in the 100 SHERGOTTY cpx-fraction listed in Table 3 do not reflect the REE I I contents of pure pyroxene crystals. This explains the apparent discrepancy between the lower contents of REE in the bulk leach residue compared to the concentration in pyroxene. Since maskelynite also appears to be occasionally intergrown with phosphate, similar arguments apply to this mineral phase also. However, these phosphates adjacent to or intergrown with the clinopyroxene are different from the major portion of phosphates, large whitlockite and apatite crystals, since they have a C 1-normalized La/Sm ratio of 1.0. This is even more evident from the isotope studies by JAG~UTZ and WANKE (1986) who found that all cpx-separate leached and unleached sample as t well as leach and residue plot along a single isochron, while the data point for the bulk leachate which con1 tains more than 90% of the Nd, is clearly off this isochron. 0.5 1 Other trace elements SC, V, and Cr show an anticorrelation with cpx Fe’ values, reflecting that the pyroxene distribution coefficients (D’s) for these elements are greater than I, thus depleting the residual parent magma as crystallization proceeds. Potassium, as ex‘t 11,111 III 9 pected, is less partitioned in Fe-poor cpx than in Fe0.1 1 ’ ’ ’ ’ Ho Tm Lu t K La Ce Pr Nd Sm GdTb rich cpx. Yb SC EU GY Muskelynite. Norm calculations for the maskelynite REE Ionic Radii separate (Table 3) indicate relatively pure plagioclase FIG. 3. Cl normalized LIL (K, REE, SC)patterns in Sher(94%) composed of An49Ab.,90rz. This plagioclase gotty bulk, major minerals (pyroxene and maskelynite), minor composition is in agreement with results from microminerals (ilmenite, Ti-magnetite, quartz, and a K-rich phase) and IM HCI leachate. probe analyses of maskelynite reported for Shergotty !
J. c‘. Laul el ul
our mineral separates are shown in Fig. 3. llmenite and Ti-rich magnetite separates have REE patterns including a Dy hump similar to Shergotty whole rock. The Ti-magnetite separate shows the most pronounced Dy hump of all samples. The high REE levels in ilmenite and Ti-magnetite may be due to phosphate impurities. The quartz separate is low in REE with La at 1 X Cl, and has a positive Eu anomaly at 1.7 X C I indicative of plagiociase or K-feldspar impurities. The potassium-rich phase has K at 20 X C 1, flat trivalent REE at 5 X C 1 and a positive Eu anomaly at 8 x C 1. As mentioned before, the REE contents of major minerals, pyroxene and maskelynite, do not account for the total REE in the bulk rock. The missing REE are concentrated in the whitlockite and apatite accessory phases that are clearly visible in thin sections. STOLPER and MCSWEEN(1979), SMITH and HERVIG (1979), and SIMON et al. (1985) reported -2% whitlockite and La at the 0.0 1 to 0.1% level. This finding is further supported by the leaching study discussed below. Leachate and residue of bulk sample
Chemical data of 1M HCI leachates and residues for Shergotty and EETA 79001 are listed in Table 5. The REE concentrations in the Shergotty ieachate are very high, thus the errors associated with INAA measurements are low (+l to 5%). From the PzOs content of apatite and whitlockite (microprobe analyses) and from the PzO~ content of the bulk rock, the Cl content of
apatite (2.05% Cl from microprobe analysis) and the amount of Cl (98 ppm) measured in the leachate. WI’ estimate an apatite content of 0.44% and a whitlockltr content of 1.26% resulting in a total phosphate Conteni of 1.70%. Based on the Ca content of 0.66%. and rhc Cl content in the leach, it appears that phosphate 15 completely dissolved in the I M HCI leach. Figure 4 shows the percent of elements leached from a bulk Shergotty sample by leaching. Chlonne and Br are quantitatively removed together with the ma,ior fractions of La (96%). The heav) REE are :,umewhar less affected (Yb 66%). Nine percent of the (la is dissolved, corresponding to a phosphate content
Chemical Data of Leachate with 1 M HCI (IO min) and Residue of Shergotty and EETA 79001 (A and B) by INAA and RNAA EETA Shergotty
A 166 myi
6 175 mg, __.._Hesldur
0 56 13
0 97 0 33
0 25 : 0 O!> 0 0 7 ‘1
0 21 1 20
0 16 41
2 70 0 51
0 67 0 10 1700
0 51 !I 080
0 04 14
0 80 16
120 1 32
1 20 0 16 14
0 46 0 077
0 91 19 7
0 20 %i Iii
0 060 0 40
I1 13 mgi
are by INAA.
for REE are by RNAA
1 48 16 5 97
Composition of Shergotty and Mars A strong positive Eu anomaly in the that maskelynite is not dissolved. If plagioclase contribution (24%) from corrected LIL pattern in the residue roxene.
EETA 79001 (LITHOLOCIES I : t a
FIG. 4. Percent element leached in I M HCl(l0 min.) relative to the bulk in Shergotty. Cl and Br are quantitatively leached. Soluble REE range from 96% La to 66% Yb. Fe, Na, Co, SC, K and Cr are leached below the 3% level.
12% higher than those of LREE (530 X Cl). The remaining HREE are uniformly decreasing in abundance to 380 X Cl for Lu. A corresponding increase from Gd to Lu is observed in the REE pattern of the residue which is largely governed by the pyroxene. The combination of both fractions (leachate + residue) yields a REE pattern identical to that of bulk Shergotty sample B.
Wt. (mgl (‘%J TiOz
MgD cao NaZO fWJ MnO CrzD3 SC V
27 4 1.70
0.042 0.47 0.61
La Nd Sm
0.10 0.006 14
0.34 0.50 0 22
Yb LU Hf
1.25 0 18 0.94
4.87 6.9 2.5 12.7 2.00
Errors an RNAA
et al. 119841 are
9.0 8.6 0.26
0.27 0.0060 0.50 0.64
70 180 46 180 0.42
40 250 0.19
0.72 3.2 0.71 0.98 0.40 2.30
2.5 1 65
CPX Most Magnetic
50 206 30 71
10 100 7.5
ELI Gd Tb HO Tm
(B1 58.0 1.1
45 70 0.41
A AND B)
Table 6 lists chemical data for 30 elements in the bulk and in phosphate-rich phases of lithologies A and B. and bulk and cpx separates (least and most magnetic) of lithology B. Table 5 shows the chemical data of 1M HCl leachates and residues of lithologies A and B. Although bulk and mineral data were reported by SMITH et al.( 1984). these samples were re-irradiated for determination of the REE by RNAA to characterize the REE patterns better. Our chemical data for the bulk are in good agreement with those of BURCHELE et al. (1983). The Cl chondrite normalized LIL (K. REE, SC) patterns for the bulk, minerals and leachates are displayed in Fig. 5. Unlike Shergotty, bulk samples of EETA 79001 show a depletion in the LREE, a smooth increase towards HREE (no Eu anomaly), and a hump at Dy. Lithologies A and B have identical REE patterns except that B has twice the REE content of A, corresponding to phosphate contents of 1.53% for A and 2.96% for B. Leaching experiments with IM HCI on lithologies A and B yield REE patterns that are identical to the REE patterns of the phosphate-rich phase (0.12 mg) from lithology A. Differences in the REE contents of
Table 6. Chemical Data of EETA 79001* (Lithology A and B) Bulk and Mineral Separates by INAA and RNAA Lithology
residue indicates we subtract the the residue, the is typical of py-
1.3 <2 0.77 0.29 16+02 0.36 0.54 024+0.2 1.57 0.24 1.4
for REE by RNAA
J. C. Laul et al
63% clinopyroxene (MCSWEEN and JAROSEWICH. 1983). As in Shergotty, the least magnetic cpx of hthoiogy B has an Fe0 content of 15.4% (Fe’ = 0.34) whereas the most magnetic cpx has Fe0 of 3 1.O% (Fe’ ==0.66) Their REE patterns are typical of pyroxene (Fig. 5t and closely resemble the whole rock REE patterns. The most magnetic cpx has about 2.2 times higher LB content than the least magnetic cpx. The HREE of the least magnetic cpx are similar to the HREE of residue B. As in Shergotty, the REE in EETA 7900 1 lithologies A and B are largely residing in phosphates.
COMPARISON OF RBE PATTERNS BETWEEN SHERGOTTY AND EETA 79001. [email protected]
REE patterns of residue, bulk, and leachate of Shergotty and EETA 79001 are shown in Fig. 6. ‘The sim-
EETA 79001 A4
La Cs Pr
Dv Tb .Ho
REE Ionic Radii FIG. 5. Cl normalized LIL (K, REE and Sc) patterns in EETA 79001 (lithologies A and B) bulk, pyroxene (least magnetic and most magnetic), phosphate “rich” phase and leachates. The REE patterns of a phosphate rich separate and leachates are identical.
\ 1’ &V \p ’ , \e’
” the leachates correspond to the different phosphate contents in lithologies A and B (MCSWEEN and JAROSEWICH, 1983). Normalized to the phosphates, the REE concentrations are indistinguishable. As in Shergotty, the REE in the EETA 79001 are highly soluble in 1M HCI (La 938, Yb 72%). The major mineral indicator elements SC, Co, Cr. and Fe are again dissolved below the 3% level. Leachates and the phosphate-rich separate of EETA 79001 show considerable depletion in LREE, a negative Eu anomaly and a hump around Dy. These patterns are different from the Shergotty leachate pattern. The reason for their difference is discussed later. The REE patterns of lithologies A and B residues are shown in Fig 6. Like the Shergotty residue, these two samples have low concentrations of LREE and are typical of pyroxene. They also show a positive Eu anomaly. The residue from lithology B has up to 30% higher REE content, except Eu (60% enrichment), than the corresponding residue of sample A. This is due to mineralogical differences. Lithology A has 18% plagioclase, 69% pyroxene (66% cpx, 3% opx), and 10% olivine, whereas lithology B has 30% plagioclase. and
Shsrgpttv ,.--.+_. l. -*, ,,’ -.____.--l / ,x-x. Leachats 1M I-ICI (10 min)
0 9 * .a
,,I,IIli,i Sm Eu
Tm Lu Er Yb
REE lon~c Radii FIG. 6. Cl nomx&ed REE patterns of residue, bulk and leachate of Shergotty and EETA 7900 1. The residues have
identical REE patterns but the leachates show different patterns. The Shergotty leachate shows evidence of two corn ponents.
Composition of Shergotty and Mars
ilarity of REE patterns including the positive Eu anomaly between the residues of Shergotty and EETA 79001 is striking. A significant positive Eu anomaly in the residue and a negative Eu anomaly in the leachate suggests that at least some plagioctase crystallized before the phosphates. The similar REE patterns of the residues suggest that Shergotty and EETA 79001 crystallized from magmas with similar REE compositions. Since bulk Shergotty and EETA have high REE contents relative to residues (pyroxene) and show the characteristic hump at Dy, it is evident that the bulk REE pattern is governed by accessory phosphate phases (whitlockite and apatite). The EETA 79001 leachate pattern of phosphate readily explains the observed trend. However, the Shergotty leachate is nearly tlat at LREE (530 X Cl) and shows no hump at Dy, yet the bulk and pyroxene show a hump similar to the phosphate of EETA 7900 1 (Figs. 2 and 6). This observation can be explained if we assume that there are two accessory components high in REE. Component 1 may be a phosphate phase similar to the phosphate phases of EETA 79001. Component 2 is enriched in LREE and is also soluble in 1M HCI. Thus, the combination of both components results in the observed REE pattern of the leachate with a nearly tlat LREE pattern 530 X Cl (La), and no hump at Dy. The presence of component 2 would also explain the approximately flat LREE pattern of bulk Shergotty relative to the LREE depleted pattern of EETA 79001 lithologies A and 3. The phosphate phase in EETA 79001 is almost completely whitlockite (MCSWEEN and JAROSEWICH, 1983), whereas the phosphate phases in Shergotty include both whitlockite ( 1.26%)and apatite (0.44%). It is therefore possible that component 2 is apatite. It is, however, not possible to characterize the apatite REE pattern directly, because in the HCl leach both apatite and whitlockite with different Mg/Fe + Mg and probably very different REE patterns and abundances are dissolved. Since the REE contents of Shergotty and EETA 79001 augites are similar, it is reasonable to assume that the’REE contents in whitlockites from both meteorites are also similar. In this case, the REE pattern due to 0.44% apatite in Shergotty can be calculated. The results of this calculation are shown in Fig 6. The LREE in the calculated apatite are enriched relative to the HREE. The apatite (component 2) may have been introduced postmagmatically and may be of hydrothermal origin. TREIMAN(1985) reported amphibole in pigeonite in Shergotty, thus indicating the presence of a hydrous melt. In terrestrial hydrothermal environment, LREE fractionate significantly from HREE because of different solubilities of the hydroxides (SILLENand MARTELL, 1964), consistent with our calculated apatite. On the other hand, merrillite (whitlockite) and fluorapatite in eucrites are also enriched in LREE relative to HREE (Yb), and the LREE in merrillite are enriched 30 to 100 times over apatite (DELANEY et cd., 1984). Our leaching study shows arsenic as being quanti~tively soluble in the Ieachate, evidence in favor of hydrothermal origin. On the other hand,
arsenic can be freely substituted for P in phosphates. It will be interesting to examine other volatiles in the leachate. At this stage it is mere speculation that component 2 is introduced by a hydrothermal fluid, although there is some evidence for metasomatism with a votatile-rich (e.g. Cl, Br) fluid. As mentioned before, it is possible that Shergotty and EETA 79001 crystallized from compositionally similar parent magmas. After crystallization Shergotty may have been an open system, and Cl, Br, As, and REE with LREE enrichment (component 2) may have been introduced at this sta8e. This is supported by the considerably lower concentrations of Cl, Br, As, and La in EETA 79001 lithologies A and B, compared to Shergotty, despite similar P contents. Shergotty and EETA 7900 1 not only have different Nd/Sm and Rb/ Sr ratios but also differ in their Nd isotopic composition (NYQUIST et al., 1979; SHIH et al., 1982; WOODEN et al.. 1982). This indicates crystallization from two separate, although com~sitionally similar, magmas. Shergotty’s parent magma The concentrations of K, REE and SC are used to model the parent magma of Shergotty. According to petrographic studies, Shergotty has 70% pyroxene in about equal proportions of augite and pigeonite from which - 30% is suggested to be cumulate phenocrysts (STOLPERand MCSWEEN, 1979). Since our magnesian pyroxene compares well with their p~rne~~st composition, we can apply appropriate partition coefficients (D values) of pyroxene to infer the trace element composition of the parent magma in equilibrium with these pyroxenes. MCKAY et al. ( 1986) experimentally measured I) values for La, Ce, Nd, Sm, Eu, Gd, Yb, and Lu elements for pyroxenes under SNC compositions and oxygen fugacities, and showed that D values for REE varied over an order of magnitude between pyroxenes with Wo 10 (wollastonite) and Wo40. The corresponding D values for Wo28 and Wo32 (a&e) are shown in Table 7. The values of Tb, Dy and Ho are interpolated based on known terrestrial augite patterns. Among the cpx mineral fractions analyzed (Fig. 2) our Fe-poor cpx (Fe’ = 0.36) has 28% normative wollastonite, Sm at 2. I X Cl and a Cl-normalized Sm/ Nd ratio of i .3, On the other hand, the augite (102)R reported by JAGOUTZand W~~NKE(1986) has the lowest Sm content at 1.34 X Cl and a chondritic normalized Sm/Nd ratio of 1.9, similar to the ratio observed in the bulk residue. The augite sample with the lowest Sm content may be the closest approximation for pure core material. Pyroxenes of this composition should be among the first minerals to crystallize. Since JAGOUTZ and W~~NKE(1986) determined only the concentrations of Sm and Nd, we assumed a pattern for the other REE similar to that of the pyroxenes analyzed in this work (Fig. 2). This REE pattern is designated as “corrected Fe-poor cpx” in Fig. 7. Using D values of Wo28, the equilibrium parent magmas for uncorrected and corrected cpx are shown in Fig. 7. In both cases, the parent magma is enriched in LREE
+ REE data for Augite (UO2R) and (YO32) are from McKay et al. (1986); for Garnet from Philpatts et al. (1972). Values
are by interpolation.
(La-Nd) and shows an overall S-shaped pattern. The uncorrected magma has higher absolute concentrations and has a Nd/Sm ratio of I .3, whereas the corrected magma has a subchondritic Nd/Sm ratio of 0.87. The first crystallizing cpx may have had still higher contents of CaO (and lower FeO). Using D values for cpx with Wo32 (MCKAY ef al.. 1986), the calculated REE pattern ofthe magma has an identical shape but somewhat lower absolute concentrations (La-Nd) than the magma calculated from pyroxenes with Wo28. It is important to note that the corrected parent magma has a subchondritic Nd/Sm ratio, which is also consistent with the estimates of SHIH et al. ( 1982) and MCKAY (‘I a/ (1986) using different approaches. However, the Sm/ Nd systematics of Shergotty require a source material enriched in LREE with a C 1 normalized Nd/Sm ratio. slightly above one on a time integrated basis (SHIH CI al., 1982). The subchondritic Nd/Sm ratio would require at least a two-stage evolution of Shergotty. The enriched HREE pattern of the parent magma would argue against any significant involvement of garnet in the residual source. The D values of garnet for REE (Lu/La = 250, Table 7) would obviously deplete HREE in the equilibrium partial melts. GROSS
cosmic abundances. The refectory lithophlle elements such as Ba, La (REE), SC. Zr. Hf. Th. Ti. Ta. and II are enriched by factors of 4 to 10 indicating that these elements are enriched in the crust of the Shergotty parent body (SPB) (see also TREIMAN cv ul.. 1986j. The elements Ca, Sr, V. W. and Al are enriched by factors of 2 to 3.5. The relatively high abundance of W IS surprising, since W is a moderately siderophile element and should have been. at least partly, extracted. together with other siderophile elements. into the core of the SPB. Since W is one of the few slderophile elements. which is not chalcophile. this indicates segregation of a sulfide phase rather than a metal phase during core formation on the SPB (DREIBUS and WAN& 1984). Manganese is present at the Cl abundance level (Fig. 8 ). The siderophile (and also chalcophile) elements Ir Ni, and Au are strongly depleted by 3 X IO -’ with a Ni/Au ratio of I, reflecting the fact that these elements are concentrated in the core. Similarly, chalcophilc elements Te, Se. Bi, As, and S are depleted by factors of 2 X 1O-3 to 1 X IO-*. Other elements with chalcophile tendencies Co, Ag, Zn, Cd, TI. In. Mo, Cu, and Sb are depleted between 0.02 to 0.10. They may have largely partitioned into the sulfide phase. For the first time it was possible to obtain data on all halogens, including iodine (I), in an SNC meteorite.
The superior data for 60 elements from three laboratories on Shergotty powdered aliquot B (Table I ) are normalized to Si and Cl abundances and the pattern is shown in Fig. 8. Following a suggestion of GAIUAPATHY et al. (1970), the elements are arranged by group number in the periodic table with symbol sizes proportional to period number. Based on gross chemical elemental comparison, certain chemical fractionation trends on the Shergotty parent body (SPB) can be inferred. Alkalis Li, Na and Cs are at the cosmic (C 1) level. while K and Rb are enriched at about 1.5 times the
11Ilii Sm EuGdTb REE ionicRadii
), ,,, Tm Lu Vb
,! ,& SC
FIG. 7. Estimates of the REE contents ofthe Shergotty parbased on Fe-poor cpx. The pattern designated “uncorrected” is calculated from data of the Fe-poor cpx fraction (Table 3). The pattern designated “corrected” is calculated from Sm-Nd data on cpx-fractions analyzed by JAGOUTZ and WANKE (19861. See text for details
Composition of Shergotty and Mars Shergony
vs. Cosmic Abundances
’ ’ 8 Gmupl
FIG. 8. Shergotty abundances for 60 elements are normalized to Si and Cl. Elements are arranged by group number in the periodic table, with symbol size proportional to period number. The depletion of volatile, siderophile and chalcophile elements and the enrichment of refractory LIL elements indicate igneous processes in the SPB.
High Br/I and Cl/I ratios (normalized to C 1) indicate depletion of I relative to Brand Cl. A similar depletion of I is observed in a large number of terrestrial basalts. However, the bulk Earth Br/I ratio may still be chondritic, since the major fraction of terrestrial I is contained in sediments, while the ocean water is the main reservoir for Br and Cl (DREIBUSet al., 1983). Similar processes on the SPB may have established the observed I depletion in Shergotty. It can, of course, not be excluded that a high Br/I ratio is characteristic for the whole parent planet. Shergottites have high concentrations of phosphorus, when compared to terrestrial, eucritic and lunar basalts. This reflects, in part, the high contents of moderately volatile elements on the Shergotty parent body (SPB). The normalized abundance of P is in the same range as that of other incompatible elements of similar volatiiity, e.g., K, Rb, Cs (Fig. 8).
LITHOPHILE, SIDEROPHILE AND VOLATILE TRACE ELEMENT PATI-ERNS IN SHERGO-ITITES The abundances of lithophile, siderophile, and volatile trace elements (Co, Au, As, Sb, Ga, Se, Te, Bi, Ag, In, Tl, Zn, Cd, Rb, Cs, and U) in Shergotty, Zagami, EETA 79001 (hthologies A and B), and ALHA 77005 are normalized to Si and then to C 1 chondrites. These patterns, shown in Fig. 9, display many similar features which are characteristic of the shergottites.
Lithophile elements like Rb, Cs, and U are enriched. Siderophile elements Co, Au, As, and Sb are depleted by a factor of 10, and the chalcophiles Se, Te, Bi, and Cd are depleted by factors up to 104. Despite the general similarity in abundance patterns, each shergottite shows its own characteristics. In terms of genetic relationship, there is no systematic trend among shergottites. Element correlations DREIBUS et al. (1982), BURGHELE et al. (1983) DREIBUS and WANKE (1984), TREIMAN and DRAKE (1984) and TREIMAN et al. (1986) have noted several element correlations in SNC meteorites. Potassium correlates strongly with Rb and Cs. The K/Rb ratio is 282 + 39 and that of Rb/Cs is 15. Aluminum, a refractory element, shows a strong correlation with the moderately volatile elements Na and Ga. Figure 10 shows the excellent correlation of K vs. the refractory element La in all SNC meteorites and that of the moderately siderophile element W vs. La in shergottites. The K/La ratio is noted to be 655 and that of W/La is 0.21. Since we have demonstrated in the past that these ratios do not change under magmatic differentiation processes, we assume that they are characteristic for bulk planets such as Earth and Moon (WKNKE et al., 1974). In applying the same rationale to the Shergotty parent body (SPB), we conclude that both ratios
J. (‘. Laul A ul.
Zagaln EET A79001,
As Sb Go Se Te BI
_____~ Zn Cd Rb Cs
FIG. 9. Lithophile, siderophile and volattle trace elements in sheraottites are normahzed to SI and i i abundances. These patterns show identical features which are characte&ic of shergomtes (sources. BURCHEIA cl al.. 1983: SMITHcl al., 1984).
K/La and W/La are higher in the SPB than tn Earth and Moon (WAN= et al.. 1974 and NEWSOM and PALME, 1984). From the Br-La correlation in the shergottites, DREIBUS and WANKE (1985) have estimated the halogen content in the SPB mantle. Since there is also a good correlation of Tl with Br in shergottites (Fig. I l), we can estimate the Tl abundance in the SPB
mantle. The original mantle abundance ot II could have been higher, since Tl may have been partly removed into a sulfur-rich core. Figure 12 shows the FeO/MnO correlation for SNC meteorites. Earth and Moon. The absolute concentrations of MnO in SNC meteorites vary between 0.45 and 0.55’%, though the FeO/MnO ratio is 40.5 A 3 i for all shergottites and Nakhla. The MnO content o!
EETA 79001 B /
01 FIG. 10. Correlation of the moderately volatile element K and the moderately siderophile refractory element W with the refractory element La for all SNC meteorites (DREIBUSand WANKE. 1984).
FIG. I 1. Bromine versus ‘ITfor shergottites and terrestnal rocks. Data sources: Shergottites; Br data from Maim lahoratory. Tl data from SMITH etal.(1984). Earth: Br and ‘V data from HERTOCENef al. (1980).
Composition of Shergotty and Mars Table
MnO - Fe0 Correlation 0.80 0.70 -
Rb Zn Ga
77.8 7.6 0.36 14.24
/ / /
FIG. 12. MnO versus Fe0 correlation in SNC, Earth and Moon. The FeO/MnO ratio of shergottites and Nakhla is 40.5 rf: 3.3. Fe-rich to Fe-poor cpx show a somewhat linear fractionation trend.
the pyroxene separates varies systematically with Mg/ Fe from 0.75 to 0.54%. In a system with approximately chondritic composition, where 01, opx, and cpx are the major MnO containing phases, the bulk solid-liquid partition coefficients for Fe0 and MnO are only slightly below one. Therefore the Fe0 and MnO contents of partial melts reflect the concentrations of these elements in the respective source regions. Hence, the fact that Mn is present in Cl abundance in shergottites indicates the Cl abundance for the whole parent planet. In comparing the Cl FeO/MnO ratio (101) with that of shergottites and Nakhla (40.5), we deduce a C 1 and Si normalized Fe0 abundance of 0.40 for the mantle of the SPB. The higher FeO/MnO ratio of 5 1 for Chassigny (an olivine cumulate) can be explained by the preferential partitioning of Fe0 into olivine. COMPARISON
the sum of MgO + Fe0 in all shergottites and Nakhla (Fig. 13). From this correlation the SPB mantle should contain 68 ppm Co. Given a Co D+,, = 68.3 (PALME et al., 1979), removal of 16.7% FeS is required to account for this concentration of Co on a planet with initial C 1 abundances. In order to estimate the composition and mass of the core, it was assumed that the whole planet contains Fe and Ni in Cl abundances relative to Si, and S in an abundance of 0.35 (mean abundance of elements, derived from component B; see below). The fraction of Fe not contained as oxide in the mantle, as well as all Ni and S, were assumed to be present in the elemental state. This would lead to a core of 2 I .7% of the planet’s mass. The mass of the core obtained in this way is only slightly higher than the amount of (Fe, Ni)S which was found to be required for the depletion of Co and other chalcophile elements (DREIBUS and WiiNKE, 1984). To explain the elemental abundance pattern of the Earth’s mantle, accretion of the Earth from two chemically different components has been proposed by
OF THE EARTH AND MARS
The chemical composition of Mars as SPB derived from analytical data of SNC meteorites (BURGHELE et al., 1983; DREIBUS and W;~NKE, 1984), is given in Table 8. The estimate of the bulk composition is based on the assumptions that in the SPB mantle, Si, Mg, and Cr and all refractory elements (Al, Ca, Ti, REE, U, etc.) are present in C 1 abundances. The normalized Fe0 abundance of 0.40 (see above) corresponds to an absolute concentration of 17.9% FeO. From the element correlations (Al/Na/Ga; K/La/W; K/Rb/Cs; Br/ Tl, etc.) observed for the SNC meteorites, we can estimate the trace element abundances in the SPB mantle. For example, from the abundance of the refractory element Al and the observed AI/Na/Ga ratios, the abundances of Na and Ga are calculated. As in terrestrial and lunar basalts, Co concentration is proportional to the abundance of the mafic portion expressed by
FIG. 13. Cobalt versusMgO + Fe0 in basalts and mantle nodules from the Earth and in SNC meteorites. All data from the Maim laboratory.
J. C. Laul et al
RINGWOOD(1979) and W.&NKE(198 1). In the model of W,&NKE(198 l), component A is highly reduced, i.c it contains Fe and all siderophile elements as metals. Si partly as metal and Cr, Mn, and V as metals or sulfides. It is devoid of all elements more volatile than Na, but contains all other elements in C 1-abundance ratios. Component B is highly oxidized (Fe. Co, Ni. W, Mn, Cr, V as well as all other siderophile and oxyphile elements as oxides), containing all elements including moderately volatile and at least some of the volatile elements (In, etc.) in C 1 abundances. Accretion started with component A only. Due to high temperatures reached during accretion, a segregation of metal. i.e. core formation, is thought to have occurred contemporaneously with accretion. Component B was added in increasing proportions after the Earth reached about 60% of its present mass. Hence. according to this inhomogeneous accretion model for the Earth, most of the matter supplied by component B never came into contact with metallic iron of component A. In this way, the high abundance of Ni, etc., in the Earth’s mantle is explained. According to this model, the Earth consists of about 85% of component A and 15% of component ?. The mantle abundances normalized to Si and Ci for the Earth and SPB = Mars are shown in Fig. 14. The abundances of moderately volatile and some moderately siderophile elements (Fe, Ga, Na, K. P. F. Rb, Zn, Cl, and Br) derived from component B are about 2 to 4 times higher on Mars than on the Earth. This suggests that Mars is richer in volatiles than the Earth (DREJBUSand WXNKE, 1985). CHEN and WASSERBURG (1986) reported high non-radiogenic Pb (volatile) in Shergotty, consistent with our observations. The similar abundances of elements with highly different geochemical behaviors (e.g. W, K, Ga and P) in both planets are striking and underlines the validity of the two component model for the formation of inner planets (RINGWOOD, 1979; W~~NKE,198 1). Using the abundances of moderately volatile and moderate11 siderophile elements, DREIBUSand W.&NKE( 1985) es-
timated for Mars a mixing ratio of60% of component A and 40% of component B. All chalcophile elements are considerabi! depleted relative to their concentranons in the Earth, how ever. as noted above, the W abundance is 111gh. Since W is moderately siderophile but only slIghtI> chalc+ phile, we suggest that the chalcophile elements wcr( extracted from the SPB mantle by a sulfur-rtch FeNi alloy. This leads to the conclusion that contr’ar? to rht Earth, Mars accreted almost homogeneousl! ( DREIB~IS and WANKE, 1984, 1985). Sulfur. mainlh supphed h> component B. and FeNi from component A x:’ thought to be responsible for the formation ota sulfurrich FeNi alloy. During core formation. the Fe-hi-S alloy scavenged chalcophile elements. depending on their sulfide-silicate partition coefficients. Sulhde-silicate partition coefficients are low for W and Ga and high for Co and Cu. This explains why a large fraction of W supplied by component A remained in the mantle, while the highly chalcophile elements Co and Cu supplied by component B were effectively extracted into the core. CONCLUSlONS
1. Shergotty is homogeneous on a cm scale in major element composition but heterogeneous with respect to LIL trace elements. The heterogeneity is greater fol volatile and siderophile trace elements. 2. The REE are concentrated in accessory phosphate phases (whitlockite, apatite). These phases are soluble in 1M HCl and they govern the shape of REE patterns in bulk samples of shergottites. 3. Shergotty and EETA 79001 residues from leaching with 1M HCI are identical in LIL patterns, indicating compositionally similar parent magmas. However, their leachate (phosphate) patterns are different. suggesting two components for the Shergotty leachate. Component 1 is similar to the EETA 79001 leachate. where component 2 is enriched in LREE and probabl? rich in volatiles. 4. Shergotty Fe-rich “rim” cpx is LREE enriched and Fe-poor “core” cpx is LREE depleted. The high enrichment of LREE in the Fe-rich cpx is at least part11 3ue to the intergrowth of phosphate with cpx. These phosphates must have different REE patterns than the large phosphate grains comprising the major portion of phosphates in Shergotty. 5. A negative Eu anomaly in the leachates and positive Eu anomaly in the residues of Shergotty and EETA 79001 suggest that some plagioclase crystallized before phosphates. 6. Heavy REE enrichment estimated for the parent magma of Shergotty implies that garnet is not required in the source region. 7. Estimates of the bulk composition of the SPB FIG. 14. Estimated abundances of moderately volatile and (Mars) derived from various element correlations in moderately siderophile elements in the Mars mantle (SPB) and in the Earth’smantle, following the approach of DREIBUS SNC meteorites suggest that Mars and Earth accreted and W~~NKE(1984). Thallium abundance for the Earth’s from the same two chemically different components. mantle was estimated from data by HERTOGENet al. C1980). .A and B (DREIBUS and WANKE. 1984).
Composition of Shergotty and Mars
8. The SPB (Mars) mantle is about 2 to 4 times richer in volatile and moderately siderophile elements including Fe0 than the Earth, its higher portion of the oxidized, volatiles containing component B is obvious. Tungsten abundance is high, but the abundances of all chalcophile elements are low, indicating chemical equilibrium between the silicate and sulfide phases during core formation. Unlike the Earth, Mars probably accreted homogeneously (WANKE, 198 1: DREIBUS and WANKE, 1984). Acknowledgements--We are very grateful to the Geological Survey of India, Calcutta, India, for the availability of Shergotty sample for International Consortium studies. Without their generosity, this consortium would not have been possible. We thank J. S. Schmitt, Battelle, Pacific Northwest Laboratories, for his assistance in the experiments. Samples analyzed by INAA in Maim were irradiated in the TRIGA-reactor of the lnstitut fur anorganische Chemie and Kemchemie der Universimt Mainz. We thank the reviewers, J. L. Gooding, G. A. McKay, and A. L. Treiman for their valuable comments that greatly improved the manuscript. The support by NASA grants NAS 9-15357 (J. C. Laul) and NAG 9-48 (M. E. Lipschutz) and by Deutsche Forschungsgemeinschaft (H. Wanke), is gratefully acknowledged. Editorial handling: M. J. Drake
REFERENCES BECKER R.
H. and PEPINR. 0. (1984) The case for a Martian origin of the shergottites: nitrogen and noble gases in EETA 79001. Earth Planet. Sci. Lett. 69, 225-242. BOGARD D. D. and JOHNSONP. (1983) Martian gases in an Antarctic meteorite. Science 221, 65 l-654. BOGARDD. D., NYQUISTL. E. and JOHNSONP. (1984) Noble gas contents of shergottites and implications for the Martian origin of SNC meteorites. Geochim. Cosmochim. Acta 48, 1723-1739. BURGHELEA., DREIBUSG., PALME H., RAMMENSEEW.. SPEITEL B., WECKWERTH G. and W~CNKEH. (1983) Chemistry of shergottites and the Shergotty parent body (SPB): further evidence for the two component model of planet formation. (abstr.). Lunar Planet. Sci. XIV, 80-8 1. CHEN J. H. and WASSERBURGG. J. (1986) Formation ages and evolution of Shergotty and its parent planet from PbU-Th systematics. Geochim. Cosmochim. Acta (this issue). CLARK B. C., BAIRDA. K., WELDONR. J.. TSUSAKID. M., SCHNABELL. and CANDELARIAM. P. (1982) Chemical composition of Martian fines. J. Geophys. Res. 87, B12, 10059-10067. CLAYTONR. N. and MAYEDAT. K. (1982) Oxygen isotopes in carbonaceous chondrites and in achondrites. (abstr.). Lunar Planet. Sci. XIII. 117- 118. CLAYTONR. N. and MAYEDAT. K. (1983) Oxygen isotopes in eucrites, shergottites, nakhlites and chassignites. Earth Planet. Sci. Lett. 62, I-6. DELANEYJ. S., O’NEILLC. and PRINZ M. (1984) Phosphate minerals in eucrites (abstr.) Lunar Planet. Sci. Xl: 208209. DREIBUSG. and W~NKE H. (1984) Accretion of the earth and the inner planets. Proc. ofthe 27th Int. Geological Congress 11, I-20, VNU Science Press. DREIBUSG. and WKNKEH. ( 1985) Mars, a volatile-rich planet. Meteoritics 20, 367-382. DREIBUS,G., JAGOUTZE. and W.&NKEH. (1983) Distribution of Cl, Br, and iodine in mantle and crust. Terra Cognita 3, 122.
DREIBUS Ct.. PALME H., RAMYENSEE W.. SPETTEL B.. WECKWERTHG. and WXNKE, H. (1982) Composition of Shergotty parent body, further evidence of a two component model for planet formation. (abstr.). Lunar Planet. Sci. XIII. 186-187. DUKE M. B. (1968) The Shergotty meteorite: Magmatic and shock metamorphic features. In Shock Meramorphism al Natural Materials (eds. B. M. FRENCHand N. M. SHORT). pp. 6 13-62 1. The Mono Book Corporation. GANAPATHYR., KEAYSR. R.. LAUL J. C. and ANDERSE. (1970) Trace elements in Apollo I 1 lunar rocks: Implications for meteorite influx and origin of Moon. Pror,..-lpol/o 11 Lunar Sci. Con/: 2, 1117-I 142. Geophvs. Research Lett. 10 (1983) Special volume on ALHA 8 1005 meteorite. HERTOCENJ.. JANSSENSM.-J. and PALMEH. (1980) Trace elements in ocean ridge basalt glasses: Implications for fractionations during mantle evolution and petrogenesis. Geochim. Cosmochim. Acta 442 125-2 143. JAGOUTZE. and W.XNKEH. (1986) Sr and Nd isotopic systematics of Shergotty meteorite Geochim. Cosmochim. Acra (this issue). LAUL J. C.. LEPEL E. A.. WEIMER W. C. and WOGMAN N. A. (1982) Precise trace rare earth analysis by radiochemical neutron activation. J. Radioanal. Chem. 69, 18 I- 196. LAULJ. C. (1979) Neutron activation of geological materials. Atomic Energy Review IAEA 17, 603-695. LAUL J. C. (1986) The Shergotty consortium and SNC meteorites: An overview. Geochim. Cosmochim. .&a (this issue). MCCARTHYT. S., ERLANKA. J.. WILLISJ. P. and AHRENS L. H. ( 1974) New chemical analyses of six achondrites and one chondrite. Meteoritics 9, 2 15-22 I. MCKAY G. A., WAGSTAFFJ. and YANG S. R. (1986) Clinopyroxene REE distribution coefficients for shergottites: The REE content of the Shergotty melt. Geochtm COW mochim. Acra (this issue). MCSWEEN H. Y., JR. and STOLPERE. M. (1980) Basaltic meteorite. Sci. Amer. 242, No. 6. 54. MC?NEEN H. Y., JR. and JAROSEWICH E. (1983) Petrogenesis of the Elephant Moraine A79001 meteorite: multiple magma pulses on the shergottite parent body. Geochim. Cosmochim. Acta 47, 1501-1514. NEWSOMH. E. and PALME H. (1984) The depletion of siderophile elements in the Earth’s mantle: New evidence from molybdenum and tungsten. Earth Planet. Sci. Lett. 69,354364. NYQUISTL. E., WOODJ., BANSALB., WIE~MANNH.. MCKAY G. and BOGARDD. D. (1979) Rb-Sr age of the Shergotty achondrite and implications for metamorphic resetting of isochron ages. Geochim. Cosmochim. Acta 43, 1057- 1074. OWEN T., DIEMANNK., RUSHNECKD. R.. BILLERJ. E., HoWARTHD. W. and LAFLEURA. L. (1977) The composition of the atmosphere at the surface of Mars. J. Geophvs. Res 82,4635-4639. PALMEH., GOBEL E. and GRIEVE R. A. F. (1979) The distribution of volatile and siderophile elements in the impact melt of East Cleatwater (Quebec). Proc. Lunar Planet. Scr Conf: 10th. 2465-2492. PALMEH., SUESSH. and ZEH H. D. (1981) Abundances of the elements in the solar system. In Landolt-Bornstem, VoI. 2a (Astronomy and Astrophysics), pp. 257-272. SpringerVerlag, Berlin. PHILPOTTSJ. A., SCHNETZLERC. C. and THOMAS H. H. (1972) Petrogenetic implications of some new geochemical data on ecologitic and ultrabasic inclusions. Geochim. Cosmochim. Acta 36, 1131-I 166. RAMMENSEEW. and PALMEH. (1982) Metal-silicate extraction technique for the analysis of geological and meteoritic samples. J. Radioanal. Chem. 71, 40 l-4 18. RINGWOOD A. E. (1979) Origin of rhe Earth and Moon. Springer-Verlag, Berlin. SCHNETZLERC. C. and PHILFQTTSJ. A. (1969) Genesis of
J. C. Laul et al
the calcium-rich achondrites in light of rare-earth and barium concentrations. In MeteoriteResearch (ed. P. M. MILLMAN), pp. 206-2 16. D. Reidel, Dordrecht. SHIH C. Y., NYQUIST L. E., B~GARD D. D., MCKAY G. A.. WOODENJ. L., BANSALB. M. and WIESMANNH. (1982) Chronology of petrogenesis of young achondrites Shergotty, Zagami, ALHA 77005: late magmatism on a geologically active nlanet. Geochim. Cosmochim. Acta 46.2323-2344. SILLBNL. G. and MARTELLA. E. (1964) Stabiiity constants of metal-ion complexes. Chem. Sot. London, Spec. Publ. 17,754. Also in Rare Earth Element Geochemistry (1984). ed. P. HENDERSON,Chapter 9 by S. E. Humphris on the Mobility of the REE in the Crust. SIMONS. B., PAPIKEJ. J. and BEAUCHAMPR. H. (1985) Petrography and silicate mineral chemistry of the Shergotty meteorite. (abstr.). Lunar Planet. Sci. XVI. Suv~l. . . A, 2324. ~ ’ SMITHJ. V. and HERVIG R. L. (1979) Shergotty meteorite: mineralogy, petrography, and minor elements. Meteoritic.7 14, 121-142. SMITH M. R., LAUL J. C., MA M.-S, HUSTONT., VERKOUTEREN R. M., LIPSCHUTZM. E. and SCHMITTR. A. ( 1984) Petrogeneais of the SNC (shergottites, nakbhtes, chassignites) meteorites; implications for their origin from a large dynamic planet, possibly Mars. Proc. Lunar Planet. Sci. [email protected]
l&h, B6 12-B630. STOLPERE. M. and MCSWEENH. Y., JR. (1979) Petrology and origin of the shergottite meteorites. Geochim. Cosmochim. Acta 43, 1475-1498. Symposium (10th) on Antarctic Meteorites (1985) National Institute of Polar Research, Tokyo, Japan. TREIMANA. H. (1985) Amphibole and hercynite spine1 in Shergotty and Zagami: Magmatic water, depth of crystallization, and metasomatism. Meteoritics 20, 227-242. TREIMAN A. H., DRAKE M. J., HERTOGENJ., JAN.SSENS M. J., WOLF R. and EBIHARAM. (1986) Core formation in the Earth and Shergottite Parent Body (SPB): Chemical evidence from basalts Geochim. Cosmochim. Acta (this issue).
TREIMANA. H. and DRAKE M. J. (1984) Core lormalion m the shergottite parent body (SPB) (abstr.). Meteoriticc 19, 324-325. VERKOUTERENR. M. and LIFXHUTZ M. E. (1983) Cumberland falls chondritic inclusions-II. Trace element cons tents of forstetite chondrites and meteorites of similar redox state. Geochim. Cosmochim. Acta 47, 1625-I 637 WKNKE H. (1981) Constitution of terrestnal planets. I%u Trans. Roy. Sot. London A303,287-302. W~~NKJZ H., BADDENHAUSEN H., BALACIXU A., I‘EXHU F.. SPEI-~ELB., DREIBUSG.. PALME H., [email protected]
M., KRUSE H., WLOTZKAF. and BEGEMANNF. (1972) Multielement analyses of lunar samples and some imphcations of the results. Proc Lunar Sci. Coqi 3rd. 125! 1268. WANKE H.. BADDENHAUSENH., DREIBUS G., JAUXJTZ t.. KRUSE H., PALME H., SPETTEL B. and TFXHKE F. (1973~
Multielement analyses of Apollo 15, 16, and 17 samples and the bulk composition of the Moon. Proc I.unar Sci Coqf 4th, 1461-1481. WXNKE H., PALME H., BADDENHAUSENH.. DREIBUS Ci ~ JAC~UTZ E., KRUSE H., SPETEL B.. TESCHKE F. and THACKERR. ( 1974)Chemistry ApolIo 16 and I7 samples:
Bulk composition, late stage accumulation and early differentiation of the moon. Proc. Lunar Sci. C’o~f 5th. ! 307. 1335. WIENS R. C., BECKERR. H. and PEPIN R. 0. (1986) The case for a Martian origin of the shergottites 11: Trapped and indigenous gas components in EETA 79001 glass. Earth Planet. Sci. Left. (in press). WOOD C. A. and ASHWALL. D. (198 1) SNC meteorites: ~gneous rocks from Mars? Proc. Lunar Planet. Sri Conf 1?th 1359-1375. WIDEN J., SHIHC.-Y., NYQUISTL., BANSALB., WIESMAN~X H. and MCKAY G. (1982) Rb-Sr and Sm-Nd isotopic constraints on the origin of EETA 79001: a second Antarctic shergottite. Lunar Planet. Sri. XIII, 879-880.