Hf^W, Sm^Nd, and Rb^Sr isotopic evidence of late impactfractionation and mixing of silicates on iron meteorite parent

bodies

Gregory A. Snyder a;*, Der-Chuen Lee b, Alex M. Ruzicka a, Martin Prinz c,Lawrence A. Taylor a, Alex N. Halliday b;d

a Planetary Geosciences Institute, University of Tennessee, Knoxville, TN 37996, USAb Department of Geological Sciences, University of Michigan, Ann Arbor, MI 48109, USA

c Department of Earth and Planetary Sciences, American Museum of Natural History, New York, NY 10024, USAd Institut fu«r Isotopengeologie und Mineralische Rohsto¡e, Departement Erdwissenschaften NO, ETH Zentrum, CH-8092 Zurich,

Switzerland

Received 1 July 1999; accepted 1 December 2000

Abstract

We report the first Sm^Nd and Rb^Sr isotopic analyses of silicate inclusions in four IIE iron meteorites: Miles,Weekeroo Station A and B, and Watson. We also report the Hf^W isotopic composition of a silicate inclusion fromWatson and 182W/184W of the host FeNi metal in all four IIEs. The host metal in Watson has a negative OW value(32.21 þ 0.24), similar to or higher than other iron meteorites [1,35] and consistent with segregation of metal fromsilicate early in solar system history. However, the large silicate inclusion in the Watson IIE iron yielded a chondritic OW

value (30.50 þ 0.55), thus indicating a lack of equilibration with the FeNi host within the practical lifetime of activity ofthe parent 182Hf (V50 Ma). One of the silicate inclusions in Miles is roughly chondritic in major-element composition,has a present-day ONd of +10.3, relatively non-radiogenic 87Sr/86Sr (0.714177 þ 13), and a TCHUR age of 4270 Ma. Twosilicate inclusions from Weekeroo Station and one from Watson exhibit fractionated Sm/Nd and Rb/Sr ratios, andmore radiogenic 87Sr/86Sr (0.731639 þ 12 to 0.791852 þ 11) and non-radiogenic ONd values (35.9 to 313.4). The silicateinclusion in Watson has a TCHUR age of 3040 Ma, in agreement with previously determined 4He and 40Ar gas retentionages, indicative of a late thermal event. A later event is implied for the two silicate inclusions in Weekeroo Station,which yield indistinguishable TCHUR ages of 698 and 705 Ma. Silicate inclusions in IIE iron meteorites formed over aperiod of 3 billion yr by impacts, involving an H-chondrite parent body and an FeNi metal parent body. The LILE-enriched nature of some of these silicates suggests several stages of melting, mixing, and processing. However, there islittle evidence to suggest that the silicates in the IIE irons were ever in equilibrium with the host FeNi metal. ß 2001Elsevier Science B.V. All rights reserved.

Keywords: iron meteorites; isotopes; FeNi metal; chronology; impact; silicate inclusion

1. Introduction

It was long thought that iron meteorites repre-

0012-821X / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 2 - 8 2 1 X ( 0 1 ) 0 0 2 2 2 - 9

* Corresponding author. Tel. : +1-724-775-5991;E-mail: gasnyder_tesm@yahoo.com

EPSL 5739 14-3-01

Earth and Planetary Science Letters 186 (2001) 311^324

www.elsevier.com/locate/epsl

sent the cores of asteroids that have undergoneextensive melting and separation from silicates.These irons were often split into groups basedupon whether they were thought to have under-gone fractional crystallization (magmatic irons)versus those which showed no such evidence(non-magmatic) [45,49]. However, more recentwork has allowed a di¡erent or modi¢ed interpre-tation for several groups of iron meteorites, espe-cially those that contain enclosed silicates (e.g.,[29]).

The age of the silicates contained in some ironshas been vigorously pursued. Over 30 yr ago, Bo-gard et al. [2], Burnett and Wasserburg [3,4], andSanz et al. [5] used inclusions of silicate materialin iron meteorites to date, presumably, the age ofthe iron. Although some of the silicate inclusionsin certain irons yielded Rb^Sr ages indistinguish-able from earliest solar system materials (e.g., Co-lomera, 4.61 þ 0.04 Ga), others indicated forma-tion events distinctly younger (e.g., Kodaikanal,3.8 þ 0.1 Ga). These younger ages were taken torepresent not only the age of the silicate inclusion,but the age of the host FeNi metal [3]. However,no independent test has been undertaken to simul-taneously date the silicate inclusion and the hostFeNi metal in an iron meteorite. Recent advancesin analysis of the short-lived nuclide 182Hf allowjust such an age comparison to be made.

Although other groups of irons (namely IABand IIICD; [49]) contain silicate inclusions, theIIE group of iron meteorites routinely containsabundant silicate inclusions, often making upover 20% of the sample [6]. Based upon trace-element abundances, the IIE irons have been sub-divided into two groups ^ those that exhibit ma-jor- and trace-element abundances and patternssimilar to H-group chondrites, termed the unfrac-tionated group, and those that exhibit abundancepatterns that are fractionated relative to H-groupchondrites. Three models have been put forth forthe origin of silicate inclusions in IIE irons: (1) anear-surface model involving mixing of IIE metalwith H-chondrite silicate melts [7,8] ; (2) mixing ofmetallic and silicate melts at the core^mantleboundary in a chondritic parent body [9], and(3) a combination of impact melting and plane-tary di¡erentiation [6].

To test these hypotheses, we have performed aradiogenic isotopic study of four silicate inclu-sions in three di¡erent IIE iron meteorites ^ twoin Weekeroo Station, and one each in Watsonand Miles. Furthermore, to constrain the evolu-tion of these silicate inclusions, as well as theirhost FeNi metals, we have examined both short-lived (Hf^W) and long-lived (Rb^Sr, Sm^Nd) iso-topic systems. Both W and Hf are refractory andfound in chondritic proportions in planetoids.However, W is moderately siderophile and Hf isstrongly lithophile, thus, metal^silicate fractiona-tion leads to partitioning of W into FeNi metaland Hf into silicate. If this metal^silicate fraction-ation occurs within 50 Ma of planetary evolution,then OW anomalies (positive in the silicate, nega-tive in the FeNi metal) should result [1,46]. In-deed, carbonaceous chondrites have OW values( = deviation of 182W from chondritic values atany given time) that are roughly zero, HEDs (ho-wardites, eucrites, and diogenites, thought tocome from the eucrite parent body; i.e., asteroid4-Vesta) and some SNC (shergottites, nakhlites,and chassignites ; thought to come from Mars)meteorites have OW values that are distinctly pos-itive, and metals from both iron meteorites andordinary chondrites have OW values that are dis-tinctly negative [1,10].

2. Previous isotopic studies of metal and silicate iniron meteorites

Stewart et al. [11] were the ¢rst to analyze Sm^Nd isotopes in a silicate inclusion in an iron me-teorite (Caddo County, IAB), but no silicate in-clusions in IIE iron meteorites have been studied.The Caddo Co. silicate exhibited a roughly chon-dritic initial 143Nd/144Nd isotopic composition(O143 =30.1 þ 0.2) and a well-de¢ned 147Sm/144Ndage of 4.53 þ 0.02 Ga. A plagioclase separate fromthis sample yielded the lowest measured 142Nd/144Nd to date (O142 =32.4 þ 0.4), consistent withearly silicate fractionation in the IAB parentbody [11]. Based on cooling rates and their Sm^Nd studies, Stewart et al. [11] preferred a modelof formation for the IAB Caddo Co. silicates byincorporation within the FeNi metal host at

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324312

depths s 2 km, early (within ¢rst 30 Myr) duringparent body di¡erentiation.

2.1. Isotopic studies and ages for IIE silicateinclusions

Previous isotopic studies of silicates in IIE ironshave indicated a diversity of ages which led Olsenet al. [12] to subdivide the IIEs into two groups. Ayoung group (3.7 þ 0.2 Ga) contains the ironsWatson, Kodaikanal, and Netschaevo, whereasan older group (4.5 þ 0.1 Ga) is composed of Co-lomera, Miles, Techado, and Weekeroo Station.This subdivision crosses previously determinedmineral^chemical boundaries, as both age groupsinclude samples from the fractionated and unfrac-tionated groups. In the discussion below, we haverecalculated the Rb^Sr ages and 87Sr/86Sr initialratios for analyzed silicate inclusions in Kodaika-nal [3] and Colomera [5] using a 87Rb decay con-stant of 1.402U10311 yr31 [13] and the programISOPLOT [14]. This decay constant for 87Rb isbased on the convergence of ages for Rb^Srwhole-rock data and U^Th^Pb data from chon-dritic meteorites [13].

2.1.1. Young groupThe fractionated silicates in the Kodaikanal IIE

iron have been studied intensively. Bogard et al.[15] determined K^Ar ages of 3.3 þ 0.1 and3.5 þ 0.1 Ga (K^Ar) for silicates in Kodaikanal.Burnett and Wasserburg [4] divided the silicatefrom Kodaikanal into 11 density and mineral sep-arates. All 11 separates from Kodaikanal yield awell-de¢ned line and an age of 3743 þ 44 Ma(MSWD = 1.58) with a poorly de¢ned initial87Sr/86Sr (due to large errors in the analyses) of0.719 þ 0.011. Gopel et al. [16] also obtained aconcordant U^Pb age (3676 þ 3 Ma) for Kodaika-nal which is similar to this Rb^Sr age. Most sig-ni¢cantly, silicate inclusions from Kodaikanalpoint to a relatively young age (3.67^3.75 Ga),possibly up to 800 Myr after formation of thehost FeNi metal.

The unfractionated Watson meteorite hasyielded a K^Ar age of 3.5 þ 0.2 Ga [12] and, re-cently, a more precise 40Ar^39Ar age of 3.656 þ

0.005 Ga [17]. An 40Ar^39Ar analysis of the sili-cate from the unfractionated group meteoriteNetschaevo is the only other IIE that hasyielded a similarly young age (3.79 þ 0.03 Ga;[18]).

2.1.2. Old groupThe unfractionated Techado IIE iron has been

dated by both K^Ar and 40Ar^39Ar methods. TheK^Ar determination yielded only a minimum ageof 4.2 Ga [8], but the 40Ar^39Ar method de¢nedan age of 4.482 þ 0.025 Ga [17]. Garrison andBogard [17] have also determined an 40Ar^39Arage for the fractionated Miles meteorite of4.408 þ 0.011 Ga.

Sanz et al. [5] separated silicate inclusions fromthe fractionated Colomera IIE iron into 14 min-eral separates, but did not include the three py-roxene points or a single anorthite point in theirregression. Once the most radiogenic `outer' sep-arates are included with the previous data forColomera, making a total of 20 separates, theage is increased to 4590 þ 127 Ma (MSWD =65.5), albeit with the same initial 87Sr/86Sr(0.6994 þ 0.0004). However, the high MSWD val-ues for both Colomera silicate ages indicate thatthe scatter is not solely due to analytical error.Regardless of possible problems with the Rb^Sranalyses for Colomera, there is still a suggestionof a silicate age which may be commensurate withthe age of the FeNi metal.

Previous isotopic studies of silicate inclusions inthe Weekeroo Station IIE iron included Rb^Srand K^Ar determinations. Burnett and Wasser-burg [4] analyzed size and density separatesfrom a silicate inclusion which yielded a Rb^Sr`age' of 4.33+0.23/30.12 Ga and an initial 87Sr/86Sr of 0.703. Evensen et al. [19] analyzed fourseparate inclusions which yielded a well-de¢nedRb^Sr whole-rock isochron of 4.39 þ 0.07 Gaand an initial 87Sr/86Sr of 0.7013. Bogard et al.[2] also analyzed a group of density and size sep-arates of a silicate inclusion from Weekeroo Sta-tion for K^Ar. An average of four analysesyielded an age of 4.49 þ 0.10 Ga. All of theseages overlap, within analytical uncertainty. Fur-thermore, it would appear that the silicate inclu-

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324 313

sions in Weekeroo Station, although likely coeval[29], may exhibit a range in initial 87Sr/86Sr ratios(0.7013^0.703).

2.2. Re^Os isotopic studies

With the advent of the 187Re^187Os isotopicsystem and its widespread use in planetary geo-chemistry, several groups have analyzed the FeNimetal in several classes of iron meteorites to de-termine their ages. Earlier studies assumed thatiron meteorites were coeval with chondrites,used their colinear relationship on a 187Re/186Osversus 187Os/186Os diagram, and assumed an ageof 4550 Ma to calculate the decay constant for187Re (1.53 þ 0.08U10311 yr31 ; although origi-nally it was reported as 1.62U10311 yr31)[20,21]. However, Lindner et al. [22] determineda much larger decay constant (shorter half-life)for 187Re (1.639 þ 0.050U10311 yr31), leading tothe suggestion that iron meteorites might be dis-tinctly younger than chondrites. Subsequently,Morgan et al. [23], Horan et al. [24], and Morganet al. [25] analyzed several samples from groupsIA, IIA, IIAB, IIB, IIIA, and IIIAB andcon¢rmed that iron meteorites are generally sim-ilar in age to chondrites, yielding isochrons of4.58 þ 0.04 Ga (IIAB) and 4.55 þ 0.11 Ga (IIIA).Smoliar et al. [47] did the most precise andcomplete study to that time, calculating ages forthe IIA, IIIA, IVA, and IVB irons of 4.537 þ 8,4.558 þ 12, 4.464 þ 26, and 4.527 þ 29 Ga, respec-tively. They also rede¢ned the 187Re decay con-stant to 1.666U10311 yr31 based on convergenceof the Pb^Pb ages for angrites with the IIIA Re^Os isochron [47]. Shen et al. [26] analyzed fourIIAB, four IVA, three IVB, three IAB, and twoIIIAB iron meteorites for Re^Os isotopes andgenerated a well-de¢ned isochron which de¢nesa possible `mixed' age of 4.62 þ 0.02 Ga. Sam-ples within particular groups gave similar ages,although there was a suggestion of a 60 þ 45 Matime di¡erence between `older' IVA irons (4.67 þ0.04 Ga) and `younger' IIAB irons (4.61 þ 0.01Ga). However, Smoliar et al. [27,47] have indi-cated somewhat younger ages for groups IIA(4.537 þ 0.008 Ga) and IIB (4.43 þ 0.05 Ga). Birckand Allegre [28] determined a well-de¢ned iso-

chron of 4.624 þ 0.017 Ga for a group of ironmeteorites, including those from groups IA, IIA,IVB, and IIE. They found that several FeNi metalsamples in Kodaikanal (IIE), the iron whichyielded U^Pb and Rb^Sr ages of 3.67^3.75 Gain the silicate (see above), were concordant withthe other old FeNi metal ages [28].

Re^Os isotopic systematics of certain iron me-teorites also have indicated possible late distur-bances on the parent bodies and/or later mag-matic events. A group of IIB irons (the Navajosubgroup) contained excess 187Os that was inter-preted to be due to either (1) crystallization froman evolved portion of the asteroid that had re-mained molten for some exceedingly lengthytime after most IIABs, or (2) impact remeltingof the core hundreds of millions of years afterits initial crystallization, or (3) Os di¡usion intothe Navajo group irons several hundred millionyears after crystallization (at 3.1^3.5 Ga). Shenet al. [26] analyzed schreibersites from groupIAB and IIAB irons which also contained excessradiogenic 187Os. They attributed these excesses toopen-system behavior of the schreibersites up to1 Ga after formation of the FeNi metal.

The present status of isotopic studies on ironmeteorites and their enclosed silicate inclusionsleaves several unresolved questions. Is the forma-tion of asteroidal cores, as represented by themagmatic iron meteorites, restricted to the ¢rst100 Myr, possibly the ¢rst 10^40 Myr [47] ofthe solar system [35]? If so, then how does oneexplain the young (3.8 Ga) Rb^Sr age for thesilicate inclusions in the IIE iron Kodaikanal?Were the FeNi metal and silicate formed at di¡er-ent times in IIE irons? If so, what mechanismscan be proposed to `inject' these silicates into pre-viously crystallized FeNi metal? If silicate wasinjected into previously crystallized metal, whate¡ect did this process have on the metal? Wasthe metal molten or solid at the moment of met-al^silicate mixing?

3. Silicate petrography and mineral chemistry

In depth petrographic and mineral^chemicalanalyses of silicate inclusions in Weekeroo Station

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324314

have been published in a companion paper [29].Weekeroo Station samples A and B were exca-vated from the same slab of sample AMNH(American Museum of Natural History) 2620[29]. Weekeroo Station A consists predominantlyof 30% augite phenocrysts in a cryptocrystalline(glass; 59%) mesostasis which was formerly feld-spar and silica. Weekeroo Station A also containssmall proportions of orthopyroxene (5%) and pi-geonite (3%) and trace amounts of (in decreasingabundance) phosphate, chromite, troilite, metal,and ilmenite. The silicate inclusion known asWeekeroo Station B consists of 56% feldspar,13% augite with coarse exsolution of orthopyrox-ene, orthopyroxene grains (15%), and pigeonite(10%), in a matrix of silica (3%). Minor amountsof (in decreasing order) phosphate, troilite, metal,K-feldspar, and chromite are also present. Thematrix in Weekeroo Station B is demonstrablycoarser than in Weekeroo Station A.

A split of the silicate inclusion in Watson thatwe studied contains on average 57% olivine (10^100 Wm) poikilitically enclosed by large (400^1000Wm) orthopyroxenes (23%), 12% feldspar, 5% au-gite (200^400 Wm), and (in decreasing order) mi-nor phosphate (mostly whitlockite), troilite, euhe-dral to anhedral chromite (200^350 Wm), andFeNi metal [12]. Unlike many other IIE silicates,chondrules, corona structures, and ilmenite areabsent.

Ikeda and Prinz [6] and Ikeda et al. [30] havestudied a total of 49 separate inclusions from theMiles IIE iron, and they have subdivided the sam-ples texturally into gabbroic and cryptocrystallinevarieties. The gabbroic silicates generally have ir-regular contacts with the FeNi metal, whereascryptocrystalline samples exhibit ellipsoidal, lo-bate, and rounded contacts with the enclosingFeNi metal. The contacts between the silicate in-clusions and the FeNi metal are often de¢ned bythe presence of schreibersite. The single inclusionthat we studied contained approximately 70%feldspar and 30% augite.

4. Analytical methods and data presentation

Silicate samples were crushed in an aluminum

oxide mortar under a £ow of better than class 100air and then split into two portions, a smaller,Sm^Nd^Rb^Sr split, and a larger Hf^W split.The smaller split (roughly 50 mg) was then dis-solved in HF, HNO3, and HCl, and isotope dilu-tion measurements made on a 10^15% split of thissolution with 87Rb^84Sr and 149Sm^150Nd mixedspikes. Total-process blanks for chemical proce-dures were always less than 10 pg Rb, 120 pgSr, 10 pg Sm, and 50 pg Nd. Sr and Nd isotopicdata were obtained by multidynamic analysis on aVG Sector multicollector mass spectrometer. AllSr and Nd isotopic analyses are normalized to86Sr/88Sr = 0.1194 and 146Nd/144Nd = 0.7219, re-spectively. Analyses of SRM 987 Sr and La JollaNd standards were performed throughout thisstudy and gave weighted averages (at the 95%con¢dence limit, external precision) of 87Sr/86Sr = 0.710250 þ 0.000011, and 143Nd/144Nd =0.511854 þ 0.000011, respectively. Internal, with-in-run, statistics are almost always of higher pre-cision than the external errors (see Table 2 forwithin-run statistics of samples, which are compa-rable to that of the standards). All isotope dilu-tion measurements utilized static mode multicol-lector analyses.

By convention, the Nd isotopic data are alsopresented in Tables 1 and 2 in epsilon units, de-viation relative to a chondritic uniform reservoir,CHUR (DePaolo, 1976):

ONd � ��143Nd=144Ndsample3143Nd=144NdCHUR�=

143Nd=144NdCHUR�U104

Model ages (or single-stage evolution ages) havebeen calculated for Nd isotopes:

TCHUR � 1=VUln��143Nd=144Nd30:512638�=

�147Sm=144Nd30:1966� � 1�

The TCHUR model age is determined relative to apresent-day CHUR with 143Nd/144Nd = 0.512638and 147Sm/144Nd = 0.1966. Errors in the modelages are estimated from consideration of errorsin the parent^daughter ratios and measured iso-topic ratios. Errors in isochron ages are 2-c [31]

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324 315

of the scatter as calculated in the ISOPLOT pro-gram of Ludwig [14].

The larger of the two silicate splits was used forHf and W abundance measurements and W iso-topic composition analyses [10,32]. This split wasleached with 1 M HCl to remove saw marks andfusion crust material. The samples were then di-gested sequentially with concentrated HF, 8 MHNO3, and 6 M HCl. Roughly 10% of the solu-tion was then separated and spiked with 178Hfand 186W and the remaining 90% of the solutionwas dried and redissolved in 8 ml of 4 M HF. Thechemical separations were similar to those of Salt-ers and Hart [33], although on a smaller scale,using columns ¢lled with 3.5 ml of Bio-RadAG1x8 (200^400 mesh) anion resin. The Hf andW were eluted and collected with a mixed solutionof 6 M HCl+1 M HF. The same chemical sepa-ration procedure was used for the spiked solu-tions, except the column volume was 1 ml, andHf and W were collected together. The total Wprocedural blank is V0.4 ng, which is negligiblefor all but the Watson silicate sample. The Wisotopic compositions were measured on a VGPlasma-54 multiple-collector inductively coupledplasma mass spectrometer (MC-ICPMS) [34].Standards (NIST-3163) were routinely run be-

tween samples to monitor performance and Wmemory e¡ects. Tungsten isotopic measurementswere normalized to 186W/184W = 0.927633. Theconcentrations of Hf and W were determined byisotope dilution on the MC-ICPMS. The devia-tion of 182W/184 from chondritic is determinedby the notation:

OW � f��182W=184W�sample=

�182W=184W�NISTÿ3163�31gU104 �1�

5. Hf^W isotopic compositions of IIE irons

The OW values for the FeNi metal samples inthe IIEs Weekeroo Station (two splits) and Milesare about 33, at the high end of the spectrum ofvalues obtained from iron meteorites [1,35]. Withthe exception of Goose Lake, the IIEs yield theleast negative OW values among all iron meteorites(Fig. 3).

5.1. Isotopic composition of the silicate inclusionand host FeNi metal in Watson

In comparison with other iron meteorites [1,35],

Table 2Rb^Sr, Sm^Nd isotopic composition of silicate inclusions in IIE iron meteorites

Sample Rb Sr 87Rb/86Sr 87Sr/86Sr Sm Nd 147Sm/144Nd 143Nd/144Nd ONd�o� TCHUR

(ppm) (ppm) (ppm) (ppm) (Ma)

Miles-4 4.61 66.2 0.2005 0.714177 þ 13 0.731 2.05 0.2154 0.513168 þ 19 +10.3 4270Watson 9.07 17.8 1.482 0.791852 þ 11 0.235 0.782 0.1815 0.512333 þ 23 35.9 3040Weekeroo-A 9.09 37.5 0.7004 0.752649 þ 11 0.649 8.45 0.04646 0.511950 þ 8 313.4 698Weekeroo-B 8.29 40.8 0.5861 0.731639 þ 12 0.737 5.96 0.07481 0.512075 þ 10 311.0 705

Table 1Hf^W isotopic composition of metals and a silicate inclusion in IIE iron meteorites

Sample AMNH# Hf W 180Hf/184W 182W/184W OW

(ppb) (ppb)

Watson 4762FeNi 0 1010 0 0.864809 þ 21 32.21 þ 0.24Silicate 154.2 59.8 3.040 0.864957 þ 48 30.50 þ 0.55Miles (a) 4866 0.864764 þ 19 32.73 þ 0.22Miles (b) 4866 0.864755 þ 31 32.83 þ 0.36Weekeroo (a) 2620 0.864737 þ 17 33.04 þ 0.20Weekeroo (b) 2620 0.864727 þ 17 33.16 þ 0.20

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324316

the Watson metal has a less negative OW anomaly(32.21 þ 0.24), indicating later segregation orequilibration with silicate early in solar systemevolution (Table 1). A s 1 g sample of a largesilicate inclusion from the Watson IIE iron exhib-its a chondritic OW value (30.50 þ 0.55), despite ahigh, non-chondritic 180Hf/184W (3.040; comparedto chondritic, which is 1.57 [1]), and indicatinglate derivation from chondritic materials and/orreequilibration with chondritic materials, in con-

trast to the W isotopic composition of the hostFeNi metal.

6. Rb^Sr, Sm^Nd isotopic compositions of silicateinclusions

All four of the silicate inclusions analyzed arenotably enriched in Rb, although to varying de-grees (Table 2). The silicate analyzed in Miles(designated Miles-4) has the lowest Rb/Sr (Fig.1A) and is otherwise relatively depleted in theLILE, exhibiting a Sm/Nd ratio above chondriticand a positive ONd (+10.3; Table 2; Fig. 2). Thissupra-chondritic Sm/Nd ratio (i.e., LREE deple-tion) is consistent with ion probe trace-elementdata on augite (containing the highest REE abun-dances among the major minerals, thus control-ling the bulk REE pattern) from IIE silicate in-clusions in Colomera [36] and Weekeroo Station[29]. The old TCHUR age (4270 Ma) suggests thatthis silicate has undergone little, if any, post-crys-tallization modi¢cation. However, the 147Sm/144Nd ratios for both Miles-4 and Watson di¡erfrom chondritic values by only about 9% and 8%,respectively. Thus, calculations of model TCHUR

values are relatively imprecise for these samples.The two inclusions studied from Weekeroo Sta-

tion (A and B) yield similar results, are both ex-tremely LREE-enriched (i.e., very low Sm/Nd),and exhibit negative ONd values (311.0 to313.4). Assuming derivation from a chondriticsource, the event which caused this enrichmentmust have occurred during the last 1 Gyr (Fig.2), as TCHUR ages are indistinguishable at 698^705 Ma. Rb^Sr isotopic determinations for thesetwo samples were similarly enriched in Rb and areas radiogenic as those reported by Burnett andWasserburg [3,4], although our samples do notlie along their 4.37 Ga (recalculated to 4.33 Gausing VRb = 1.402U10311 yr31) isochron (Fig. 1B).

The silicate in Watson is among the most Rb-enriched and most radiogenic of bulk silicate in-clusions measured in IIE irons (Fig. 1A). Burnettand Wasserburg [3,4] did ¢nd inclusions (D5 andD6) in the Colomera and Kodaikanal IIE ironswhich have higher Rb/Sr and, consequently, aremore radiogenic (87Sr/86Sr up to 2.052). Due to

Fig. 1. Measured 87Rb/86Sr versus 87Sr/86Sr for IIE silicates.(A) Plot of bulk analyses of silicates in IIE irons. A regres-sion of Watson and Weekeroo Station B yields an `age' of4.64 þ 0.04 Ga. Reference isochrons at 4.33 Ga and 4.39 Gaare also shown which encompass all of the data, except thosefor Weekeroo Station A and B and Watson. (B) Plot of bulksilicate and silicate separates from Weekeroo Station (dia-monds = [3,4]; stars = [19] ^ estimated from a ¢gure in theirabstract; and this study, Table 2). A regression of the Even-sen et al. [19] points yields an age of 4.35 þ 0.07 Ga and thatfor the Burnett and Wasserburg [3,4] data gives an age of4.33 þ 0.23 Ga (using VRb = 1.402U10311).

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324 317

their elevated Rb/Sr, mineral and density sepa-rates in Kodaikanal exhibited even more radio-genic isotopic compositions; up to 87Sr/86Sr val-ues of 8.847 for `feldspar glass' [3,4].

The Sm^Nd systematics of the Watson silicate(ONd =35.9) appear to be intermediate betweenthat of Miles-4 and Weekeroo Station, which isinconsistent with its relatively elevated Rb andradiogenic 87Sr/86Sr. The Watson silicate yields arelatively imprecise TCHUR of 3040 Ma (Fig. 2).

7. Discussion

7.1. Hf^W in silicate inclusions, FeNi metal, andlack of equilibration

Lee and Halliday [1] found Arispe to be theleast radiogenic iron meteorite and concludedthat Arispe was the ¢rst of the irons to undergosilicate^metal fractionation. The ages of silicate^metal fractionation were then referenced to thisearliest fractionation in Arispe. Using this logic,the IIE irons would appear to have formed laterthan most (if not all) other iron meteorites, ors 8 Ma after Arispe (Fig. 3). Conversely, the

more radiogenic character of the IIE irons couldbe due to an admixture of material that did notexperience early metal^silicate fractionation, suchas bulk chondrite material, or an admixture ofmaterial that did fractionate, but had a positiveOW.

If the silicate inclusion in Watson was cogeneticwith the metal in Watson, as has been postulatedfor other magmatic iron meteorites [35], then itshould exhibit a complementary positive OW iso-topic anomaly. Positive OW anomalies are ob-served for many of the silicates derived from aste-roids and planets, such as some SNC and HEDachondrites (Fig. 4) [10]. Such positive OW valuesare consistent with core formation early in solarsystem history creating a high Hf/W in the result-ing mantle. However, a positive OW anomaly isnot present in the Watson silicate; therefore, thesilicate and metal do not appear to be cogenetic.Metal^silicate mixing for Watson must have oc-curred later, after 182Hf had largely decayed. Thedata can be explained if metal in Watson wasderived from a source region that experiencedearly metal^silicate fractionation, and if silicatein Watson was derived from a di¡erent sourceregion that did not undergo metal^silicate frac-tionation. One reasonable explanation for this isthat the metal and silicate in Watson were derivedfrom distinct parent bodies that collided.

Fig. 2. Nd isotopic evolution diagram showing thesingle-stage evolution of radiogenic Nd (ONd = [(143Nd/144Ndsample3143Nd/144NdCHUR)/143Nd/144NdCHUR]U104) in IIEiron meteorite silicate inclusions over time. Note that thetwo silicates from Weekeroo Station converge at the CHURat 700 Ma. The silicate inclusion in Miles crosses the CHURevolution line at 4270 Ma, and that for Watson crossesCHUR at 3040 Ma.

Fig. 3. Histogram of OW values for iron meteorites (datafrom [1] and this work). Note that, with the exception ofGoose Lake which includes a large error, the IIE irons havethe least negative OW values.

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324318

7.2. Petrogenesis of silicate inclusions

A closer look at the Rb^Sr systematics versusthe Sm^Nd systematics for these silicate inclu-sions yields some additional observations. First,there appears to be a relationship between thedegree of LREE depletion in these inclusionsand their Sm^Nd model ages (Fig. 5A). Speci¢-cally, as the TCHUR ages of the IIE silicate inclu-sions become older, there is a corresponding in-crease in the 147Sm/144Nd (i.e., increasing LREEdepletion) of the samples. However, such a corre-sponding relationship is not apparent for 87Rb/86Sr in these inclusions (Figs. 5B and 6), suggest-ing either that the Rb^Sr systematics are de-coupled from the Sm^Nd systematics, or thatthe 87Rb/86Sr ratios have been altered, possiblyduring later impact metamorphism of the parentbody.

7.2.1. Weekeroo StationWe have analyzed two separate inclusions from

the Weekeroo Station IIE iron for Rb^Sr andSm^Nd. The two inclusions indicate a complex

petrogenetic history. Neither of the inclusionslies along the two lines de¢ned by previous Rb^Sr isotopic data (Fig. 1A,B). If the ages of thesetwo inclusions were considered to be similar tothe previous analyses (e.g., [3,4]), then they wouldrepresent reservoirs with vastly di¡erent initial87Sr/86Sr. In fact, silicate inclusion A, if 4.3^4.4Ga old, would project back to an initial 87Sr/86Sr of 0.708, and inclusion B would projectback to an initial 87Sr/86Sr which would approx-imate that of BABI (basalt achondrite best initial ;0.69899 þ 5; [48]) at this time.

Sm^Nd isotopic systematics of the WeekerooStation silicate inclusions point to a complexand long-lived scenario. The simplest two-stagemodel for Weekeroo Station silicate inclusions Aand B involves evolution in an H-chondrite par-ent body followed by LILE (i.e., Rb and Nd)enrichment at 700 Ma (Fig. 2). Such an enrich-ment event is also suggested by reconnaissance

Fig. 4. Histogram of OW values for metals in iron meteorites,SNC meteorites (example of silicates which may have beenin equilibrium with early FeNi metal fractionation), and theFeNi metal and silicate from the Watson IIE iron (datafrom [1,10], and this work).

Fig. 5. Plots of parent^daughter ratios versus model TCHUR

ages for the four IIE silicate inclusions analyzed in thisstudy. (A) 147Sm/144Nd versus TCHUR ages showing a positivecorrelation. (B) 87Rb/86Sr versus TCHUR ages showing no cor-relation.

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324 319

Re^Os isotopic work [37]. Although other IIEirons yielded Re^Os model ages of 4.6 Ga,Weekeroo Station was the sole iron meteoritewhich gave an unrealistic Re^Os model age of7.1 Ga. Such an impossibly old age could bedue to enrichment in radiogenic Os or a decreasein the Re/Os ratio at some later time, possibly dueto melting and loss of a sul¢de and/or metalphase. This enrichment depletion event was likelydue to impact on the parent body [29]. The e¡ectof such an impact is seen most clearly by theoccurrence of glass and cryptocrystalline silicatesin the inclusions, probably caused by remeltingand subsequent quenching of the inclusions (e.g.,[6,9,29,38]). The impact also caused dynamic mix-ing of phases, as evidenced by the disequilibriumbetween augite and glass and the preferentialalignment of large pyroxene grains [29]. Ruzickaet al. [29] discuss the possibility that igneous dif-ferentiation was impact-induced and concludethat the evidence for this is equivocal. The Sm^Nd isotopic systematics support the possibility ofimpact-induced di¡erentiation by showing thatLILE enrichment for Weekeroo Station inclusionsA and B could have occurred as recently as 700Ma ago, most likely during an impact meltingepisode.

7.2.2. WatsonAmong the samples studied, the 147Sm/144Nd

(0.1815) and present-day ONd (35.9) for Watson

is the closest to that of chondrites (0.1967 and 0,respectively). Sm^Nd isotopic studies of the sili-cate in Watson indicate a LILE enrichment event,although of a lesser magnitude than that forWeekeroo Station, and much earlier in parentbody evolution (at approximately 3040 Ma, Fig.2). The 4He^U^Th gas retention age for a split ofthe same large silicate inclusion from Watson(3.1 þ 0.3 Ga; [12]) is indistinguishable from theSm^Nd model age of the silicate we analyzed(3.04 Ga). Olsen et al. [12] also analyzed K^Arin their silicate, which yielded a slightly olderage of 3.5 þ 0.2 Ga. Independently, all three ofthese isotopic systems indicate a late-fractionationevent which occurred between 3 and 3.5 Ga ago.

The two Weekeroo Station samples are greatlyLREE-enriched (147Sm/144Nd = 0.04646^0.07481;present-day ONd =311.0 to 313.4) relative tochondrites. The silicate in Miles is also fraction-ated relative to chondrites, but in an oppositesense to both Watson and Weekeroo Station(147Sm/144Nd = 0.2154; ONd = +10.3). The recentwork of Ebihara et al. [39] on Miles supportsthis conclusion. Based upon comparisons of theREE patterns and abundances in six gabbroic andthree cryptocrystalline inclusions in Miles withthose from other IIEs, they concluded that theparental liquid for Watson was the least di¡eren-tiated and that for Weekeroo Station was themost evolved, with Miles having an intermediatecomposition.

Rb^Sr isotopes in the Watson silicate inclusionare opposite of what one might expect from theSm^Nd isotopic studies. Miles, being the mostLREE-depleted sample, also has the lowest Rb/Sr ratio, as would be expected. However, thetwo Weekeroo Station inclusions, being the mostLREE-enriched, should also have the highest Rb/Sr. Clearly, their Rb/Sr ratios are higher thanMiles, but Watson has a much higher Rb/Sr bya factor of two or more.

7.2.3. MilesIn contrast to that in Watson, the silicate in

Miles points to an even earlier LILE depletionevent in parent body evolution at 4270 Ma (Fig.2). This rather imprecise Sm^Nd CHUR modelage is, however, similar to the 39Ar^40Ar age of

Fig. 6. 87Rb/86Sr versus the proportion of plagioclase (vol%)in the four IIE silicates studied.

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324320

4408 þ 11 Ma determined by Garrison and Bo-gard [15]. Although the Hf^W isotopic studiesof Watson showed that the silicate inclusion wasnot in equilibrium with the enclosing metal, thismay not be true for Miles. Similar siderophile-element patterns between the silicate inclusionsand the host FeNi metal, as well as a relationshipbetween the abundance of metal and the sidero-phile-element abundances in the silicate inclu-sions, led Ebihara et al. [39] to suggest that `thetwo (Miles and Watson) appear to be geneticallyrelated'. Thus, we consider it likely that a majorLILE depletion event occurred on the Miles IIEparent body between 4.3 and 4.4 Ga. This eventappears to have either rehomogenized the silicateand the host FeNi metal, or caused the originalseparation of these phases. This fractionationevent could have been caused by impact but,just as plausibly, could have been the result ofigneous fractionation.

7.3. Rb^Sr isochrons: true ages or mixing ofimpact components?

Regardless of the complicated scenario one pre-fers for the origin of IIE iron meteorite silicateinclusions, two contrasting facts must be ex-plained: (1) the Rb^Sr systematics from manyother studies suggest a consistently very old age(4.3^4.5 Ga), whereas (2) the Sm^Nd systematics,with the exception of Miles-4, indicate progres-sively younger ages (Fig. 5A). The answer maylie in the impact processing which undoubtedlya¡ected these meteorites. Shock-recovery experi-ments of basaltic materials have shown that the¢rst phase to melt and form a glass is plagioclasefeldspar, followed by pyroxene, and then olivine[40,41]. This feldspar-enriched melt would likelyseparate and coalesce with other feldspar-enricheddroplets. Thus, progressive impact events can en-rich a sample in the feldspar component, as hasbeen shown for igneous-textured clasts in the Ju-lesberg ordinary chondrite [42]. Such progressiveenrichment in a feldspathic melt componentwould have important implications for the Rb^Sr and Sm^Nd systematics for silicate inclusionsderived in a like manner.

Plagioclase contains vanishingly small abun-

dances of Rb and will not generate much radio-genic Sr over an extended time period. Thus, re-setting of the Sr isotopic systematics in theplagioclase at various times in the history of aperiodically impacted sample would not changeits placement on a Rb^Sr diagram, like that inFig. 3. The `ages' generated from a sample thathas been multiply impacted and has remelted pla-gioclase several times in its history would not bechanged, but would likely retain its original crys-tallization age (i.e., the plagioclase plots very nearor practically on the y-axis and cannot be easilymoved from this axis). Thus, the placement ofsamples on Fig. 1 simply could re£ect the miner-alogy of the samples, and the derivative line couldbe a mixing line between ma¢c minerals and re-melted plagioclase and glass.

The Weekeroo Station samples exhibit petro-graphic evidence which is consistent with thismodel of preferential feldspar melting. Coarse or-thopyroxene exsolution lamellae in Weekeroo Sta-tion B pyroxenes are suggestive of slow-cooling,possibly as a result of deep burial in the parentbody. In contrast, the cryptocrystalline feldspar+silica glass in Weekeroo Station A is indicative ofrapid, likely near-surface, cooling. The preserva-tion of this coarse exsolution lamellae in Weeke-roo Station B suggests that the pyroxenes werenot remelted during the event (impact?) that cre-ated the glass in Weekeroo Station A.

Kodaikanal is the only IIE iron to be analyzedthat yielded a signi¢cantly younger Rb^Sr age(e.g., 3743 þ 44 Ma) [3,4]. If our scenario of pref-erential plagioclase melting is correct, then thissample would suggest that a shock event musthave been intense enough that other phases werealso melted, possibly totally resetting all of thephases. Just such an intense event is recorded inthe FeNi metal of Kodaikanal. In fact, Buchwald[43] stated that `few iron meteorites display suchdistortions (of silicate^metal, troilite^metal, andschreibersite^metal interfaces) as those present inKodaikanal'. He goes on to state that these dis-tortions and deformation features `require shockwith attenuated peak temperatures for their for-mation' [43].

In ma¢c rocks, olivine and pyroxenes typicallyexhibit the highest Sm/Nd of any of the major

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324 321

phases and plagioclase the lowest Sm/Nd. Withthe exception of augite, plagioclase also exhibitsthe highest abundance of Nd among the majorphases. However, the Sm/Nd of plagioclase isnot vanishingly small, and radiogenic Nd wouldbe expected to build up, given a certain period oftime. Furthermore, augite is a signi¢cant, high-Nd, high-Sm/Nd component, which also canmelt with plagioclase during an intense impact.Thus, determining the Sm^Nd distribution duringimpact melting is exceedingly complicated, and itis much more likely the Sm^Nd systematics wouldbe at least partially reset with each successiveevent. Thus, whereas the Rb^Sr systematics wouldnot yield any information other than the originaligneous fractionation event, the Sm^Nd system-atics could re£ect the impact history of the sam-ple. Supporting evidence for this is found in theSm^Nd model (TCHUR) ages, which increase withincreasing LREE depletion of the sample (Table2). Such a relationship might be expected, if suc-cessive impact events create more and more feld-spathic glass, with increasingly enriched LREE.

8. Models for IIE silicate formation

The lack of equilibration between silicate andFeNi metal in the Watson IIE iron meteoritestrongly suggests that the enclosed silicate formedmuch later than the FeNi host. Thus, it appearsthat the parent body for the IIE FeNi metal frac-tionated silicate from metal early (to achieve theW isotopic fractionation seen in the FeNi metal),but that this silicate is no longer present. Thesilicate that is enclosed within the FeNi metalwas introduced later, in some cases up to 1 billionyr later.

The silicates enclosed by the FeNi metal haveundergone variably complex histories. If these sil-icate inclusions were all derived from H-groupordinary chondrites, as is currently thought,then these silicates suggest fractionation eventswhich span over 3 billion yr. In some cases, asin Miles, the silicates represent the LREE-de-pleted crystallization products of a melt generatedwithin a few hundred million years (4.3^4.4 Gaago) of formation of the FeNi metal. Silicate in-

clusions in Watson probably are imperfectly sep-arated residues of partial melting of a chondriticprecursor 3^3.5 Ga ago. Silicate inclusions inWeekeroo Station point to a recent event, only700 Ma ago, which generated very evolved melts.All of these fractionation events could have beeninitiated by near-surface processes, presumablyimpacts and/or collisions and consequent mixing,as is suggested through textural and mineral^chemical studies [6,7,12,29,30,44].

Ruzicka et al. [29] presented two models for theformation of IIE meteorites. Both models includecollision and mixing between a metal impactorand silicate target, but the di¡erence is whetherthe target was represented by previously fraction-ated or primitive H-chondrite silicate. Due to thelarge Rb/Sr fractionations seen for IIE silicates,and especially in the Watson and Weekeroo Sta-tion silicates, we consider it unlikely that the pre-cursor was undi¡erentiated. Instead, it is muchmore feasible to fractionate an H-chondrite pre-cursor in several consecutive stages.

The metals in the IIE parent body were frac-tionated from a silicate mass early in its history,however, this silicate is either not present in theIIE iron meteorites (and by conjecture not in theparent body) or has been signi¢cantly repro-cessed. More importantly, the isotopic systematicsof the IIE irons suggest a long history of process-ing of the silicate portion of the parent body,possibly continuing up until 700 million yr ago.

Acknowledgements

This study was supported by NASA GrantsNAGW 3543 (L.A.T.), NAG 5-4745 (M.P.), and(A.N.H.).[CL]

References

[1] D.-C. Lee, A.N. Halliday, Hf^W isotopic evidence forrapid accretion and di¡erentiation in the early solar sys-tem, Science 274 (1996) 1876^1879.

[2] D. Bogard, D. Burnett, P. Eberhardt, G.J. Wasserburg,40Ar^40K ages of silicate inclusions in iron meteorites,Earth Planet. Sci. Lett. 3 (1967) 275^283.

[3] D.S. Burnett, G.J. Wasserburg, Evidence for the forma-

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324322

tion of an iron meteorite at 3.8U109 years, Earth Planet.Sci. Lett. 2 (1967) 137^147.

[4] D.S. Burnett, G.J. Wasserburg, 87Rb^87Sr ages of silicateinclusions in iron meteorites, Earth Planet. Sci. Lett. 2(1967) 397^408.

[5] H.G. Sanz, D.S. Burnett, G.J. Wasserburg, A precise87Rb/86Sr age and initial 87Sr/86Sr for the Colomera ironmeteorite, Geochim. Cosmochim. Acta 34 (1970) 1227^1239.

[6] Y. Ikeda, M. Prinz, Petrology of silicate inclusions in theMiles IIE iron, Proc. NIPR Symp. Ant. Met. 9 (1996)143^173.

[7] J.T. Wasson, J. Wang, A nonmagmatic origin of group-IIE iron meteorites, Geochim. Cosmochim. Acta 50(1986) 725^732.

[8] I. Casanova, T. Graf, K. Marti, Discovery of an unmeltedH-chondrite inclusion in an iron meteorite, Science 268(1995) 540^542.

[9] T.J. McCoy, Silicate-bearing IIE irons: Early mixing anddi¡erentiation in a core-mantle environment and shockresetting of ages, Meteoritics 30 (Suppl.) (1995) 542^543.

[10] D.-C. Lee, A.N. Halliday, Core formation on Mars anddi¡erentiated asteroids, Nature 388 (1997) 854^857.

[11] B. Stewart, D.A. Papanastassiou, G.J. Wasserburg, Sm^Nd systematics of a silicate inclusion in the Caddo IABiron meteorite, Earth Planet. Sci. Lett. 143 (1996) 1^12.

[12] E. Olsen, A. Davis, R.S. Clarke, L. Schutz, H.W. Weber,R. Clayton, T. Mayeda, E. Jarosewich, P. Sylvester, L.Grossman, M.-S. Wang, M.E. Lipschutz, I.M. Steele, J.Schwade, Watson: A new link in the IIE iron chain, Me-teoritics 29 (1994) 200^213.

[13] J.F. Minster, J.-L. Birck, C.J. Allegre, Absolute age offormation of chondrites studied by the 87Rb^87Sr method,Nature 300 (1982) 414^419.

[14] K.L. Ludwig, ISOPLOT: A plotting and regression pro-gram for radiogenic-isotope data for IBM-PC compatiblecomputers, version 2.01, USGS Open-¢le report 88-557,1990.

[15] D.D. Bogard, D.S. Burnett, G.J. Wasserburg, Cosmogen-ic rare gases and the 40K^40Ar age of the Kodaikanal ironmeteorite, Earth Planet. Sci. Lett. 5 (1969) 273^281.

[16] C. Gopel, G. Manhes, C.J. Allegre, Concordant 3675 MyrU^Pb formation age for the Kodaikanal iron meteorite,Nature 317 (1985) 341^344.

[17] D.H. Garrison, D.D. Bogard, Argon-39/Argon-40 ages ofsilicates from IIE iron meteorites, Meteoritics 30 (1995)508.

[18] S. Niemeyer, I^Xe and 40Ar^39Ar dating of silicate fromWeekeroo Station and Netschaevo IIE iron meteorites,Geochim. Cosmochim. Acta 44 (1980) 33^44.

[19] N.M. Evensen, P.J. Hamilton, G.E. Harlow, R. Klimen-tidis, R.K. O'Nions, M. Prinz, Silicate inclusions inWeekeroo Station: Planetary di¡erentiates in an iron me-teorite, Lunar Planet. Sci. X (1979) 376^378.

[20] J.-M. Luck, J.-L. Birck, C.-J. Allegre, 187Re^187Os system-atics in meteorites early chronology of the solar systemand age of the galaxy, Nature 283 (1980) 256^259.

[21] J.-M. Luck, C.J. Allegre, 187Re^187Os systematics in me-teorites and cosmochemical consequences, Nature 302(1983) 130^132.

[22] M. Lindner, R.J. Leich, G.P. Russ, J.M. Bazan, R.J.Borg, Direct determination of the half-life of 187Re, Geo-chim. Cosmochim. Acta 53 (1989) 1597^1606.

[23] J.W. Morgan, R.J. Walker, J.N. Grossman, Rhenium^os-mium isotope systematics in meteorites I: Magmatic ironmeteorite groups IIAB and IIIAB, Earth Planet. Sci. Lett.108 (1992) 191^202.

[24] M.F. Horan, J.W. Morgan, R.J. Walker, J.N. Grossman,Rhenium^osmium isotope constraints on the age of ironmeteorites, Science 255 (1992) 1118^1121.

[25] J.W. Morgan, M.F. Horan, R.J. Walker, J.N. Grossman,Rhenium^osmium concentration and isotope systematicsin group IIAB iron meteorites, Geochim. Cosmochim.Acta 59 (1995) 2331^2344.

[26] J.J. Shen, D.A. Papanastassiou, G.J. Wasserburg, PreciseRe^Os determinations and systematics of iron meteorites,Geochim. Cosmochim. Acta 60 (1996) 2887^2900.

[27] M.I. Smoliar, R.J. Walker, Re^Os isotopic constraints onthe crystallization history of IIAB iron meteorites, LunarPlanet. Sci. Conf. XXVIII (1997) 1341^1342.

[28] J.L. Birck, C.J. Allegre, Rhenium-187^osmium-187 iniron meteorites and the strange origin of the Kodaikanalmeteorite, Meteorit. Planet. Sci. 33 (1998) 647^653.

[29] A.M. Ruzicka, G.W. Fowler, G.A. Snyder, M. Prinz, J.J.Papike, L.A. Taylor, Petrogenesis of silicate inclusions inthe Weekeroo Station IIE iron meteorite: Di¡erentiation,remelting, and dynamic mixing, Geochim. Cosmochim.Acta 63 (1999) 2123^2143.

[30] Y. Ikeda, M. Ebihara, M. Prinz, Petrology and chemistryof the Miles IIE iron. I: Description and petrology oftwenty new silicate inclusions, Antarct. Meteor. Res. 10(1997) 355^372.

[31] D. York, Least-squares ¢tting of a straight line withcorrelated errors, Earth Planet. Sci. Lett. 5 (1969) 320^324.

[32] D.C. Lee, A.N. Halliday, Precise determinations of theisotopic compositions and atomic weights of molybde-num, tellurium, tin, and tungsten using ICP magnetic sec-tor multiple collector mass spectrometry, Int. J. MassSpectrom. Ion Process. 146/147 (1995) 35^46.

[33] V.J.M. Salters, S.R. Hart, The mantle sources of oceanridges, islands and arcs: The Hf isotope connection, EarthPlanet. Sci. Lett. 104 (1994) 364^380.

[34] A.N. Halliday, D.-C. Lee, J.N. Christensen, A.J. Walder,P.A. Freedman, C.E. Jones, C.M. Hall, W. Yi, D. Teagle,Recent developments in inductively coupled plasma mag-netic sector multiple collector mass spectrometry, Int. J.Mass Spectrom. Ion Process. 146/147 (1995) 21^33.

[35] M.F. Horan, M.I. Smoliar, R.J. Walker, 182W and 187Re^187Os systematics of iron meteorites: Chronology formelting, di¡erentiation, and crystallization in asteroids,Geochim. Cosmochim. Acta 62 (1998) 545^554.

[36] W. Hsu, H. Takeda, G.R. Huss, G.J. Wasserburg, Min-eralogy and chemical compositions of Colomera (IIE) sil-

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324 323

icate inclusions, Meteorit. Planet. Sci. 32 (Suppl.) (1997)A61^A62.

[37] S. Niemeyer, B.K. Esser, Re^Os chronology of IAB, IIE,and IIIAB iron meteorites, Meteoritics 26 (1991) 378^379.

[38] J.T. Armstrong, A.K. Kennedy, P.K. Carpenter, A.L. Al-bee, Petrography and trace element chemistry of Colo-mera (IIE) silicate inclusions; rhyolitic plums in the pud-ding, Lunar Planet. Sci. XXI (1990) 22^23.

[39] M. Ebihara, Y. Ikeda, M. Prinz, Petrology and chemistryof the Miles IIE iron. II: Chemical characteristics of theMiles silicate inclusions, Antarct. Meteor. Res. 10 (1997)373^388.

[40] R.B. Schaal, F. Ho«rz, Proc. Lunar Sci. Conf. 8 (1977)1697^1729.

[41] R.B. Schaal, F. Ho«rz, T.D. Thompson, J.F. Bauer, Shockmetamorphism of granulated lunar basalt, Proc. LunarPlanet. Sci. Conf. 10 (1979) 2547^2571.

[42] A.M. Ruzicka, G.A. Snyder, L.A. Taylor, Mega-chon-drules and large, igneous-textured clasts in Julesberg(L3) and other ordinary chondrites: Vapor-fractionation,shock-melting, and chondrule formation, Geochim. Cos-mochim. Acta 62 (1998) 1419^1442.

[43] V.F. Buchwald, Handbook of Iron Meteorites: Their His-

tory, Distribution, Composition, and Structure, Univer-sity of California Press, Berkeley, CA, 1975, 1375 pp.

[44] A.E. Rubin, E.A. Jerde, P. Zong, J.T. Wasson, J.W.Westcott, T.K. Mayeda, R.N. Clayton, Properties of theGuin ungrouped iron meteorite: the origin of Guin andgroup II-E irons, Earth Planet. Sci. Lett. 76 (1985/1986)209^226.

[45] J.T. Wasson, Meteorites: Their Record of early Solar-System History, W.H. Freeman and Co., New York,1985, 267 pp.

[46] C.H. Harper, S.B. Jacobsen, Evidence for 182Hf in theearly solar system and constraints on the timescale ofterrestrial accretion and core formation, Geochim. Cos-mochim. Acta 60 (1996) 1131^1153.

[47] M.I. Smoliar, R.J. Walker, J.W. Morgan, Re^Os ages ofgroup IIA, IIIA, IVA, and IVB iron meteorites, Science271 (1996) 1099^1102.

[48] D.A. Papanastassiou, G.J. Wasserburg, Initial strontiumisotopic abundances and the resolution of small time dif-ferences in the formation of planetary objects, Earth Plan-et. Sci. Lett. 5 (1969) 361^376.

[49] B.G. Choi, X. Ouyang, J.T. Wasson, Classi¢cation andorigin of IAB and IIICD iron meteorites, Geochim. Cos-mochim. Acta 59 (1995) 593^612.

EPSL 5739 14-3-01

G.A. Snyder et al. / Earth and Planetary Science Letters 186 (2001) 311^324324