Iron meteorites Crystallization, thermal history, parent ...escott/Goldstein ea chem review.pdf ·...

ARTICLE IN PRESS

0009-2819/$ - se

doi:10.1016/j.ch

�CorrespondE-mail addr

Chemie der Erde 69 (2009) 293–325www.elsevier.de/chemer

INVITED REVIEW

Iron meteorites: Crystallization, thermal history, parent bodies, and origin

J.I. Goldsteina,�, E.R.D. Scottb, N.L. Chabotc

aDepartment of Mechanical and Industrial Engineering, 313 Engineering Lab, University of Massachusetts,

160 Governors Drive, Amherst, MA 01003, USAbHawaii Institute of Geophysics and Planetology, University of Hawaii, Honolulu, HI 96822, USAcThe Johns Hopkins University Applied Physics Laboratory, 11100 Johns Hopkins Road, Laurel, MD 20723, USA

Received 9 August 2008; accepted 6 January 2009

Abstract

We review the crystallization of the iron meteorite chemical groups, the thermal history of the irons as revealed bythe metallographic cooling rates, the ages of the iron meteorites and their relationships with other meteorite types, andthe formation of the iron meteorite parent bodies. Within most iron meteorite groups, chemical trends are broadlyconsistent with fractional crystallization, implying that each group formed from a single molten metallic pool or core.However, these pools or cores differed considerably in their S concentrations, which affect partition coefficients andcrystallization conditions significantly. The silicate-bearing iron meteorite groups, IAB and IIE, have textures andpoorly defined elemental trends suggesting that impacts mixed molten metal and silicates and that neither groupformed from a single isolated metallic melt. Advances in the understanding of the generation of the Widmanstattenpattern, and especially the importance of P during the nucleation and growth of kamacite, have led to improvedmeasurements of the cooling rates of iron meteorites. Typical cooling rates from fractionally crystallized iron meteoritegroups at 500–700 1C are about 100–10,000 1C/Myr, with total cooling times of 10Myr or less. The measured coolingrates vary from 60 to 300 1C/Myr for the IIIAB group and 100–6600 1C/Myr for the IVA group. The wide range ofcooling rates for IVA irons and their inverse correlation with bulk Ni concentration show that they crystallized andcooled not in a mantled core but in a large metallic body of radius 150750 km with scarcely any silicate insulation.This body may have formed in a grazing protoplanetary impact. The fractionally crystallized groups, according toHf–W isotopic systematics, are derived originally from bodies that accreted and melted to form cores early in thehistory of the solar system, o1Myr after CAI formation. The ungrouped irons likely come from at least 50 distinctparent bodies that formed in analogous ways to the fractionally crystallized groups. Contrary to traditional viewsabout their origin, iron meteorites may have been derived originally from bodies as large as 1000 km or more in size.Most iron meteorites come directly or indirectly from bodies that accreted before the chondrites, possibly at 1–2AUrather than in the asteroid belt. Many of these bodies may have been disrupted by impacts soon after they formed andtheir fragments were scattered into the asteroid belt by protoplanets.r 2009 Elsevier GmbH. All rights reserved.

Keywords: Widmanstatten pattern; Crystallization; Cooling rates; Chemical groups; Parent bodies; Iron meteorites; Fractional

crystallization; Magmatic; Parent bodies

e front matter r 2009 Elsevier GmbH. All rights reserved.

emer.2009.01.002

ing author. Tel.: +1413 545 2165; fax: +1 413 545 1027.

ess: [email protected] (J.I. Goldstein).

www.elsevier.de/chemer

dx.doi.org/10.1016/j.chemer.2009.01.002

mailto:[email protected]

ARTICLE IN PRESSJ.I. Goldstein et al. / Chemie der Erde 69 (2009) 293–325294

1. Introduction

Recent studies have argued that the iron meteorites,like stony-irons and achondrites, come from solarsystem bodies that melted allowing mm-sized andsmaller metal and silicate grains in chondrite materialto segregate into much larger domains (e.g., Krot et al.,2008; Weisberg et al., 2006). According to the textbooks,iron meteorites are derived from over 50 bodies thatwere 5–200 km in size, most of which melted to formmetallic cores and silicate mantles. These bodies arethought to have accreted in the asteroid belt after thechondrites and to have been broken open by impactslong after the bodies had cooled slowly.

Other studies suggest that almost all of thesestatements may be incorrect. Iron meteorites may havebeen derived originally from bodies as large as 1000 kmor more in size that melted. There is a growing consensusthat most iron meteorites come from bodies thataccreted early – even before the parent bodies of thechondrites – and that 26Al, which has a half-life of0.7Myr, is the major heat source that melted them.Evidence from Hf–W radiometric dating of ironsand chondrites shows that most irons come frombodies in which metallic cores formed o1Myr afterthe growth of the oldest objects, Ca–Al-rich inclusionsin chondrites (Kleine et al., 2005; Markowski et al.,2006a, b; Qin et al., 2008; Burkhardt et al., 2008).The early accretion of igneously differentiatedasteroids is also inferred from consideration ofthe heating effects of 26Al in planetesimals. Homo-geneity of Mg isotopic compositions of diversemeteorite parent bodies suggests that 26Al was homo-geneously distributed in the solar system (Thraneet al., 2006). Therefore there would have been sufficientthermal energy from 26Al to melt cold planetesimalsthat accreted within 1.5Myr of CAI formation andwere large enough (420 km radius) so that little heatwas lost for several half-lives of 26Al (Hevey andSanders, 2006). In addition, chondrule ages determinedby Al–Mg and Pb–Pb dating are 1.5–5Myr afterCAI formation indicating that chondrites accretedafter differentiated asteroids when 26Al concentrationswere no longer adequate to melt asteroids (see Scott,2007).

The requirement that iron meteorite parent bodiesaccreted before those of the chondrites is one ofseveral arguments advanced by Bottke et al. (2006) fortheir claim that these bodies accreted not in theasteroid belt but closer to the Sun at 1–2AUwhere planetesimals accreted faster (see Section 7.7).Thus, the parent bodies of the iron meteorites couldhave been much more diverse than those of thechondrites and the irons may tell us more about thebodies that accreted to form the terrestrial planets thanthe chondrites.

In this paper we discuss how this new view of theformation of the iron meteorites was developed. Wereview the general structure and classification of ironmeteorites and outline the crystallization and formationof the iron meteorite chemical groups and their thermalhistory as revealed by metallographic cooling rates. Wediscuss other evidence, chiefly isotopic, that elucidatesthe thermal and igneous histories of irons, and finishwith conclusions about the possible origins and forma-tion of their parent bodies.

2. Composition, structure, and chemical groups

of the iron meteorites

2.1. Composition and structure

Iron meteorites are Fe–Ni alloys containing minoramounts of Co, P, S, and C. The Ni content varies froma minimum of 5.1 up to 60wt% although the vastmajority of irons have between 5 and 12wt%. The 10largest irons, which each weigh more than 10 tons (e.g.,Buchwald, 1975), are meter sized and most were largesingle crystals of taenite (fcc Fe–Ni) after solidificationand at high temperatures in their parent bodies. TheWidmanstatten pattern of the irons was revealedindependently by Thomson and by von Widmanstattenin 1804 and 1808, respectively (see review by Clarke andGoldstein, 1978) when polished sample surfaces wereetched by various chemicals (Fig. 1). This structure canbe used to determine the cooling rate of each meteorite,as discussed in Section 4.

The Widmanstatten pattern develops as a two-phaseintergrowth of kamacite (a-bcc, ferrite) and taenite(g-fcc, austenite), and forms by nucleation and growthof kamacite from taenite during slow cooling of theparent body (Owen and Burns, 1939). The conventionalexplanation of Widmanstatten pattern formation, whichis only partly correct but will suffice for this introduc-tion, is based on the binary Fe–Ni equilibrium phasediagram (Yang et al., 1996). A meteorite of a givenFe–Ni content cools from the one-phase taenite (g)region into the two-phase a+g region, where kamacite(a) nucleates and grows as the meteorite continues tocool. Kamacite nucleates on the close packed octahedral{1 1 1} planes of taenite, forming a Widmanstattenpattern (Fig. 1). In three dimensions, kamacite growsas two-dimensional plates into the surrounding taenite.As cooling continues, kamacite grows at the expense oftaenite and the Ni content of both kamacite and taeniteincreases.

The nature and scale of kamacite has led to astructural classification of iron meteorites. Hexahedrites(H) are one-phase kamacite with Ni of 5–6.5wt% Ni.Octahedrites (O) have visible (to the eye) Widmanstatten

ARTICLE IN PRESS

Fig. 1. Polished and etched slices of iron meteorites showing Widmanstatten patterns. (a) Carlton – group IIICD, Of. Kamacite

plates (blue) formed on the close-packed planes of the parent taenite phase. Plessite, a fine mixture of kamacite and tetrataenite,

formed in the prior taenite regions between the kamacite plates. Schreibersite precipitates are observed in the centers of some of the

kamacite plates. Scale – 1 cm along the bottom. (b) Canyon Diablo – group IAB, Ogg. Note cohenite precipitates in the centers of

several kamacite plates. A large rounded sulfide occurs in the right-hand bottom corner. Scale – 10.5 cm along the bottom. (c) Mt.

Edith – group IIIAB, Om. Narrow bands are kamacite, gray angular areas enclosed by these are plessite. Short angular bands

surrounded by kamacite are schreibersite (Fe–Ni)3P; rounded black inclusions are troilite, FeS. Scale – 19 cm along the bottom. (d)

Tawallah Valley IVB, D. Kamacite plates form on the close packed planes of the parent taenite phase. The matrix is plessite, a fine

mixture of kamacite and tetrataenite. Numerous schreibersite precipitates are observed in or near kamacite plates. (b and c:

Courtesy of Smithsonian Institution). (For abbreviations, Of, Om, etc., see Table 1 footnotes.)

J.I. Goldstein et al. / Chemie der Erde 69 (2009) 293–325 295

patterns with kamacite widths of 0.2–3mm and Niconcentrations of 6 to �12wt% that are generallyinversely related to kamacite width. Ataxites (D) havemicroscopic Widmanstatten patterns where kamacite issimilarly oriented and o0.2mm in width (Fig. 1d) andNi contents of �10 to 420wt%.

The presence of P, S, and C in iron meteorites leads tothe formation of precipitates of schreibersite (FeNi)3P(Figs. 1a, c, d), troilite (FeS) (Figs. 1b, c), cohenite(FeNi)3C (Fig. 1b) and other Fe–Ni carbides. The bulkconcentrations of these and other elements are compiledby Buchwald (1975). P and C are very soluble in theliquid phase and at high temperatures, fcc taenite canaccommodate �1wt%. In most cases, phosphides andcarbides exsolve in the solid state during cooling. S is

less soluble in taenite and sulfides usually form as theliquid metal solidifies. As discussed in Section 3, theconcentration of these elements in the liquid, particu-larly S, has a major effect on the solidification behaviorand the eventual distribution of major, minor, and traceelements in the meteorite. These elements tend tosegregate to the liquid phase during solidification,lowering the melting point of the liquid and allowingfor the formation of eutectic material at the lowesttemperatures. As discussed in Section 4, the amount of Ppresent in the metal greatly influences the nucleationtemperature, the reaction process, and the diffusion rateof Ni as the Widmanstatten pattern develops. Silicatesare present in some iron meteorites, mainly groups IAB,IIICD, and IIE (Mittlefehldt et al., 1998) and their


composition and significance are discussed in Sections 3and 5.

2.2. Trace elements and chemical groups

Early measurements of Ga and Ge in iron meteoritesshowed that their concentrations fall into four distinctgroups, labeled I–IV, in order of decreasing abundances(Goldberg et al., 1951; Lovering et al., 1957). Subse-quently, Wasson (1967) and Wasson et al. (1998)analyzed �700 iron meteorites, initially for Ni, Ga,Ge, and Ir and later for Cr, Co, Cu, As, Sb, W, Re, Pt,and Au, using instrumental neutron activation analysis.The original four Ga–Ge groups were resolved into 14clusters labeled with additional letters A–G attached tothe Roman numerals (Table 1). About 15% of the ironsdo not fit these groups and are labeled as ungrouped. Gaand Ge are the most useful elements to classify irons,because the range within most groups is only a factor ofless than 2.5, whereas the total range between all groupsvaries by factors of 103–4. The power of this classifica-tion to reveal correlations between numerous diverseproperties including mineralogical, chemical and iso-topic parameters shows that the members of each groupare closely related and formed together in one parentbody (Buchwald, 1975; Scott and Wasson, 1975; Haackand McCoy, 2004). Figs. 2a,b show the variation of Gevs. Ni and Ir vs. Ni for the groups and ungrouped irons.The abundances of these elements in CI chondritesnormalized to Ni are also shown.

Table 1. Properties of 14 groups of iron meteorites and the ungro

Group Number Ni (wt%) Structureb

IABc�110 6–60 Og-D

IC 11 6–7 Ogg, Og

IIAB 78 5.3–6.5 H, Ogg

IIC 8 9.3–11.5 Opl

IID 21 9.6–11.1 Om, Of

IIE 17 7.2–9.5 Og-Off

IIF 6 11–14 Opl, D

IIG 6 4.1–4.9 H

IIIAB �220 7.1–10.6 Om

IIICDc 12 12–23d Of-D

IIIE 14 8.1–9.6 Og

IIIF 8 6.8–8.5 Og, Om

IVA 61 7.5–12 Of

IVB 14 16–18 D

Ungroupedc �110 6–35 Ogg-D

aBased largely on the listed references and the Meteoritical Bulletin Databa

specimens and irons with tentative classifications.bStructure: H, hexahedrite; Ogg, Og, Om, Of, Off – coarsest, coarse, medium

Buchwald (1975, Table 26a) for definitions.cWasson and Kallemeyn (2002) define 120 irons from groups IAB and IIIC

irons are included in their scheme as sLM and sLH subgroups, respectively.

crystallization.dIrons with lower Ni concentrations are excluded.

Gallium and Ge vary between groups because theyare the most volatile siderophile elements and the rangeincreases with increasing volatility. In group IAB,element/Ni ratios for siderophiles are broadly compar-able to CI chondritic values, but in groups IVA, IVBand the ungrouped irons, with similarly low concentra-tions of Ga and Ge, element/Ni ratios are depletedrelative to CI chondrites and decrease in order ofincreasing volatility: Au, P and As, Cu and Ga, and Ge(Wasson, 1985; Scott, 1977a). Two possible explana-tions for the depletion of volatile elements have beenproposed: in the solar nebula before accretion, or duringmajor planetary impacts prior to metal solidification.Ga and Ge show small variations within groupsbecause, unlike Ir (Fig. 2), they are not fractionatedsignificantly between solid and liquid metal. Compar-isons of the chemical trends within groups and theirmineralogy suggests that there are two very differenttypes of groups: (1) groups IIAB, IID, IIIAB, IVA, IVB,and possibly the smaller groups such as IC and IIIF,which are very largely free of silicates, have composi-tional trends that can be explained by chemicalfractionation during solidification of molten iron; (2)groups IAB, IIICD, and IIE, with more abundantsilicates, the chemical trends are very different andcommonly much weaker (Scott, 1972). In the first type,chemical variations are largely consistent with fractionalcrystallization modeling using experimentally deter-mined solid metal/liquid metal partition coefficients,whereas in the second, the trace element distribution

uped irons.a

Example Reference

Canyon Diablo Choi et al. (1995)

Bendego Scott and Wasson (1976)

Coahuila Wasson et al. (2007)

Ballinoo Wasson (1969)

Carbo Wasson and Huber (2006)

Weekeroo Sta. Wasson and Wang (1986)

Corowa Kracher et al. (1980)

Bellsbank Malvin et al. (1984)

Cape York Wasson (1999)

Tazewell Choi et al. (1995)

Kokstad Malvin et al. (1984)

Clark Co. Kracher et al. (1980)

Gibeon Wasson and Richardson (2001)

Hoba Walker et al. (2008)

Butler Scott (1979), Wasson (1990)

se: http://tin.er.usgs.gov/meteor/metbull.php. Numbers exclude paired

, fine, and finest octahedrites; Opl, plessitic octahedrite, D, ataxite. See

D and �35 ungrouped irons as a ‘‘group IAB complex’’. IIIC and IIID

The silicate-bearing groups IAB and IIICD did not form by fractional

http://tin.er.usgs.gov/meteor/metbull.php

ARTICLE IN PRESS

Fig. 2. Logarithmic plots of (a) Ge vs. Ni and (b) Ir vs. Ni of bulk compositions of all iron meteorites. The chemical groups are

shown in distinct colors and symbols. Iron meteorite data provided by J.T. Wasson; references are given in Table 1.


cannot be modeled by fractional crystallization alone(Chabot and Jones, 2003; Chabot and Haack, 2006).These two types of iron meteorite groups were called‘‘magmatic’’ and ‘‘non-magmatic’’ (Wasson, 1985).However, it is likely that metal in groups IAB, IIICD,and IIE was once molten, so this terminology, althoughwidespread, is rather misleading. We refer to groupsIAB, IIICD, and IIE as the ‘‘silicate-bearing groups’’,and the ‘‘magmatic’’ groups are called the ‘‘fractionallycrystallized groups’’. Note, however, that there are twosilicate-rich IVA irons and one silicate-bearing IIIABiron. Fractionally crystallized iron meteorite groups arewidely thought to be derived from the cores of asteroidsthat melted, whereas the silicate-bearing groups maycome from bodies that were not heated sufficiently formetallic cores to have formed (Haack and McCoy,2004). In this case, a group of iron meteorites from thecore of a differentiated body should have cooled atvirtually identical rates because of the high thermalconductivity of metal compared with mantle and crustmaterials. However, there is evidence in several frac-tionally crystallized groups for diverse cooling rates sothese irons could not have cooled in an insulatedmetallic core. At least one group, IVA, appears to haveformed in a metallic body that crystallized and cooledwith virtually no silicate mantle.

The classification and number of parent bodiesrepresented by the silicate-bearing groups is moreuncertain as their chemical trends are ill-defined.Groups IAB and IIICD contain assemblages ofgraphite, troilite and silicates like those in chondritesand might have formed in the same body (McCoy et al.,

1993; Benedix et al., 2000), whereas group IIE ironsmostly contain differentiated silicates and lack graphite(Mittlefehldt et al., 1998). Wasson and Kallemeyn(2002) defined a ‘‘group IAB complex’’ consisting of156 meteorites previously classified in groups IAB andIIICD plus a number of ungrouped irons. These werereorganized into a main group of 70 irons with6.4–7.5wt% Ni (previously called group IA), fivesubgroups (two being IIIC and IIID), 2 grouplets and27 more distantly related irons. Although Wasson andKallemeyn (2002) recognized that these irons couldcome from several bodies (as suggested by O isotopes),they argued that they formed in similar ways. We followKrot et al. (2008) and Weisberg et al. (2006) and use theearlier definitions for groups IAB and IIICD.

3. Crystallization of iron meteorite groups

3.1. Introduction

By studying the chemical variations within a group ofrelated irons, we can gain insight into the bulkcomposition of the metallic liquid from which theysolidified since the relative concentrations of siderophileelements in the core reflects the bulk composition ofthe parent asteroid. In addition we can learn about theprocess by which the irons crystallized, and theenvironment on the parent asteroid in which theyformed. The groups show surprisingly large differencesin bulk compositions and their internal chemicalvariations, so that it is not possible to identify a typical


group of iron meteorites. Consequently, we explore thesolidification of iron meteorites using a group-by-groupapproach. The bulk composition of a metallic liquid hasa crucial influence on the trace element chemistry duringcrystallization and the evolution of the metallic melt.Thus, before examining individual iron meteoritegroups, we briefly discuss the importance of bulkcompositions as related to the evolution of ironmeteorites.

3.2. Effects of bulk composition

during crystallization

Experiments to measure how diverse elements parti-tion between coexisting solid metal and liquid metalshow that the partition coefficient (D) is greatly affectedby the concentration of S, P, or C in the metallic liquid(Willis and Goldstein, 1982; Jones and Drake, 1983).Additionally, some elements are especially sensitive, e.g.,the partition of Ni varies little with changing composi-tion of the metallic liquid, while that for Ir can changeby orders of magnitude during crystallization in theFe–Ni–S system. During fractional crystallization of themetallic melt, the non-metals are largely excluded fromthe crystallizing solid metal and consequently enrichedin the metallic liquid. Thus, during crystallization, themetallic liquid composition is continuing to evolve tohigher non-metal contents, which in turn influences thechemical partitioning of elements between the crystal-

Fig. 3. Experimental determinations of solid metal/liquid metal par

Fe–Ni–S, (b) Fe–Ni–P, and (c) Fe–Ni–C systems. References for th

non-metals to the metallic liquid can cause large changes in the par

variations in iron meteorite groups (Corrigan et al., 2009).

lizing solid metal and the residual liquid metal. Thischanging metallic liquid composition must be taken intoaccount when attempting to understand the behavior ofelements during the crystallization of iron meteorites.

Fig. 3 shows experimentally determined solid metal/liquid metal partition coefficients for Au, Ge, Ir, and Niin the Fe–Ni–S (Chabot et al., 2003; Chabot and Jones,2003), Fe–Ni–P (Corrigan et al., 2006), and Fe–Ni–C(Chabot et al., 2006) systems. The non-metals S, P, andC can affect partitioning behavior during the crystal-lization of the metal. However, S can have a much largerinfluence than either P or C if all three elements arepresent in significant amounts (Fig. 3). For example,D(Ge) increases by two orders of magnitude from the S-free system to the Fe–FeS eutectic composition of31wt% S. In contrast, D(Ge) only increases by a factorof 3 from a P-free to a Fe–Fe3P eutectic composition(10wt% P) and about a factor of 5 from the C-free tothe Fe–C eutectic composition of 4.3 wt% C.

Unlike the binary Fe–S and Fe–P systems, the Fe–S–Psystem shows a large liquid immiscibility field (Ragha-van, 1988). Thus, as fractional crystallization proceedsand the liquid is enriched in non-metals, the bulkcomposition may enter the immiscibility field, causingtwo immiscible molten phases to form, one S-rich,P-poor and one S-poor, and P-rich. A similarliquid immiscibility field exists in the Fe–S–C system(Raghavan, 1988). Chabot and Drake (2000) conductedexperiments that suggested the liquid immiscibility fieldin the part of the Fe–S–P system that is relevant to the

tition coefficients (D) for Au, Ge, Ir, and Ni are shown in (a)

e data are in the sources shown on the figure. The addition of

tition coefficients, which are crucial to understanding chemical

ARTICLE IN PRESS

Fig. 4. Logarithmic plots of Ir vs. Au. (a) The solid composition during simple fractional crystallization of Fe–Ni liquids with

3–18wt% S; (b–f) data for groups IIAB, IIIAB, IVA, IID, and IVB. All graphs are scaled with a factor of 20 on the Au x-axis and a

factor of 10,000 on the Ir y-axis, to enable comparisons between the calculated curves and the different iron meteorite groups.

Simple fractional crystallization models were run until the Fe–FeS eutectic composition of 31wt% S was reached. On (d–f), model

curves have tick marks (x-marks) that correspond to every 10% of crystallization. Sources for the iron meteorite data: IIAB

(Wasson et al., 2007), IIIAB (Wasson, 1999), IVA (Wasson and Richardson, 2001), IID (Wasson and Huber, 2006), and IVB

(Walker et al., 2008).


crystallization of iron meteorites is actually smaller thanshown by Raghavan (1988). Yet Chabot and Drake(2000) still concluded that some iron meteorite groups,despite the smaller liquid immiscibility field, might stillhave encountered liquid immiscibility in the Fe–S–Psystem. Thus, it is important to understand the bulknon-metal composition of the iron meteorite parentmelts and to assess if immiscible liquids may haveformed, especially during the later stages of fractionalcrystallization.

The bulk S content of the parent melt of an ironmeteorite group cannot be inferred from measurementsof the bulk S content of iron meteorites themselves,Sulfur is virtually insoluble in solid metal, and thus isfound only in inhomogeneously distributed troilitenodules that likely represent melt trapped during thecrystallization process. Instead, the S concentrations ofthe melt have to be inferred by studying distributions ofother elements most sensitive to the amount of S presentin the liquid during crystallization.

Fig. 4a shows calculated fractional crystallizationtrends for Ir vs. Au using the partitioning values ofFig. 3 and for different bulk S concentrations. Tradi-tionally, element trends for iron meteorites were plotted

vs. Ni content, due in part to the much higher Niconcentrations than other elements, making Ni contentseasier to measure. However, D(Ni) has a value close tounity at all S concentrations (Fig. 3), resulting in limitedNi variations during crystallization of iron meteoritegroups. Therefore, Ni is a poor choice for assessing thesuccess of crystallization models and Au is commonlyused instead (Haack and Scott, 1993; Wasson, 1999;Wasson and Richardson, 2001).

For higher bulk S concentrations, the Ir vs. Au trendbecomes steeper and shows a larger overall fractionationof Ir (Fig. 4). The Ir vs. Au trends for five iron meteoritegroups are also shown in Fig. 4, plotted on the sameoverall scale as the models in Fig. 4a. Fig. 5 showssimilar model and iron meteorite plots for Ge vs. Au. Asshown by the models in Figs. 4a and 5a, simplefractional crystallization can produce large fractiona-tions in Ir and distinctly non-linear Ge trends, both ofwhich are sensitive to the bulk S concentration. Ir andGe are thus good choices for deducing constraints onthe crystallization history of iron meteorites. Thechemical trends for the five groups (Figs. 4 and 5)suggest a wide range of bulk S compositions fromessentially S-free to about 18wt% S. The estimated S

ARTICLE IN PRESS

Fig. 5. Logarithmic plots of Ge vs. Au. (a) solid composition during simple fractional crystallization of Fe–Ni liquids with

3–18wt% S; (b–f) data for groups IIAB, IIIAB, IVA, IID, and IVB. All graphs are scaled with a factor of 20 on the Au x-axis and a

factor of 6 on the Ge y-axis. Data sources are listed in the caption for Fig. 4. Simple fractional crystallization models were run until

the Fe–FeS eutectic composition of 31wt% S was reached. On (d–f), model curves are marked with tick marks (x-marks) that

correspond to every 10% of crystallization.

J.I. Goldstein et al. / Chemie der Erde 69 (2009) 293–325300

values are dependent on the crystallization modelutilized.

3.3. Crystallization of the chemical groups

Different types of crystallization models have beenused to gain insight into the evolution of iron meteoritegroups. Some of the earliest attempts to model thecrystallization of iron meteorites used best-fit curves toexperimentally determined solid metal/liquid metalpartition coefficients in simple fractional crystallizationcalculations (e.g., Jones and Drake, 1983). Subsequentwork, noting the shortcomings of simple fractionalcrystallization, added complexities to model additionaleffects that might be caused by assimilation ofadditional metal to the core during crystallization(Malvin, 1988), dendritic crystallization of the core(Haack and Scott, 1993), the onset of liquid immisci-bility (Ulff-Møller, 1998), and incomplete mixing in themolten core (Chabot and Drake, 1999). Mathematically,these models resembled the simple fractional crystal-lization calculations, using fits to experimental partitioncoefficients, but with appropriate modifications. Morerecently, crystallization models have been developedthat are mixtures of fractionally crystallizing solid and

liquid trapped during the crystallization process (e.g.,Wasson, 1999; Wasson and Richardson, 2001; Wassonand Huber, 2006; Wasson et al., 2007). The trapped meltmodel offers an attractive explanation for the scatterwithin iron meteorite groups. However, in contrast toprevious models, the partition coefficients used in thesetrapped melt models are not set by best-fits to theexperimental values but, rather, are determined usingestimates of the amount of trapped melt in individualiron meteorites; the partitioning values determined bythis method do not always agree with the experimentalvalues (for reviews, see Chabot and Haack, 2006).

3.3.1. Crystallization of the IIIAB group

The crystallization history of the IIIAB group hasbeen investigated by many workers because this is thelargest iron meteorite group, its elemental trends arewell defined, and the crystallization trend appears to bewell sampled (e.g., Wasson, 1999). Using a simplefractional crystallization model, based on parameteriza-tions of experimentally determined solid metal/liquidmetal partition coefficients, the IIIAB Ir vs. Au and Gevs. Au trends are fairly well matched by a bulk S contentfor the IIIAB parent melt of about 12wt%. This modelcan explain the three orders of magnitude variation in Ir

ARTICLE IN PRESSJ.I. Goldstein et al. / Chemie der Erde 69 (2009) 293–325 301

in IIIAB (Fig. 4c), and the curved Ge trend (Fig. 5c),which arises due to changes in D(Ge) with changing Scontent during fractional crystallization. Thus, to firstorder, the IIIAB group seems to represent a metallicmelt that began with about 12wt% S and experiencedsimple fractional crystallization as the melt solidified.

More complex models have addressed the short-comings of the simple model, which misses the high-Au,low-Ir, later crystallizing portion of the IIIAB trend(Fig. 4c). Models involving liquid immiscibility in theFe–S–P system (Ulff-Møller, 1998) or incompletemixing in an inwardly crystallizing core (Chabot andDrake, 1999) matched the low-Ir portion of the IIIABtrend. There is also a considerable amount of scatteraround the general IIIAB trend, especially for Ir(Fig. 4c), which is reflected in the two-fold variationsof Ir in the Cape York shower (Esbensen et al., 1982).These local variations may result from inward dendriticcrystallization of the core (Haack and Scott, 1993) ortrapped melt (Wasson, 1999), the latter using assumedpartition coefficients, suggesting a much lower initial Scontent for the IIIAB bulk composition of 2.4wt% S.

3.3.2. Crystallization of the IVB group

Though there are only 12 well-analyzed members ofthe IVB group, their chemical data suggest a simplecrystallization history (Rasmussen et al., 1984; Camp-bell and Humayun, 2005; Walker et al., 2008). The Ir vs.Au (Fig. 4f) and Ge vs. Au (Fig. 5f) IVB trends are bothwell matched by simple fractional crystallization in a S-free system. Campbell and Humayun (2005) fit the IVBtrends for a number of elements and calculated partitioncoefficients that are in good agreement with theexperimentally determined ones. Small troilite nodulesare present, but their abundance suggests that theseirons have the lowest S concentrations of any group(0.02–0.05wt% S; Buchwald, 1975, Table 30). Chabot(2004) demonstrated that crystallization trends werevery similar for S contents from 0 to 2wt% and that theIVB Ge vs. Au and Ir vs. Au trends were equally well fitwith a S content of 2wt% as in a S-free system. Overall,the crystallization of the IVB group appears to beconsistent in every aspect with simple fractional crystal-lization of a nearly S-free metallic melt. The well-behaved nature of the group may suggest that thecrystallization process is simpler in general for a systemessentially free of non-metals such as S. However,the sub-chondritic Re/Os and Pt/Os inferred fromfractional crystallization models remain a mystery(Walker et al., 2008).

3.3.3. Crystallization of the IIAB group

Figs. 4b and 5b show that the IIAB Ir and Ge trends arewell defined (Wasson et al., 2007) and generally consistentwith simple fractional crystallization from a metallic meltwith an initial S content of about 18wt%. Given this high

S content, it seems likely that liquid immiscibility in theFe–S–P system would have been encountered during thefractional crystallization (Chabot and Drake, 1999).Having the metallic liquid separate into two immiscibleliquids, one S-rich and one P-rich, would affect theelemental crystallization trends (Ulff-Møller, 1998). How-ever, the first-order simple fractional crystallization modeldoes quite well at fitting the IIAB trends and nodiscontinuity in either the Ge or Ir trend is observed(Figs. 4b and 5b). Additionally, the IIAB trends actuallyshow considerably less scatter than observed for the IIIABgroup. Having a bulk S concentration of 18wt% alsoraises the question of where is that S now, since less than45% of the melt will crystallize before the Fe–FeS eutecticcomposition of 31wt% S is reached by the liquid. The lackof S-rich meteorites may reflect the friable nature of suchmaterials, causing them to not survive the journey toEarth’s surface (Kracher and Wasson, 1982). The trappedmelt model of Wasson et al. (2007), using assumed ratherthan experimentally determined partition coefficients,suggests a much lower bulk S concentration of 7.5wt%and is able to explain the low-Ir members not reproducedby simple fractional crystallization on Fig. 4b. However,even with an initial S content of 7.5wt%, Wasson et al.(2007) concluded that the IIAB melt experienced liquidimmiscibility during crystallization.

3.3.4. Crystallization of the IVA group

The elemental trends in group IVA (e.g., Wasson andRichardson, 2001) are less consistent with simplefractional crystallization than trends in groups IIIAB,IVB, and IIAB. The Ir vs. Au trend appears to beformed by fractional crystallization of a metallic meltwith a bulk S content of 3wt% (Fig. 4d). However, theGe vs. Au trend is not matched by fractional crystal-lization with 3wt% S but, rather, by a significantlyhigher S content of 9wt% (Fig. 5d). If the Ge and Irtrends were established during the simple fractionalcrystallization process, both elements should exhibitpartitioning behavior consistent with the single Scontent of the bulk metallic liquid. The discrepancybetween the Ge–Au and Ir–Au trends suggests thatsimple fractional crystallization was modified signifi-cantly by some other process. Using an initial S contentof 0.4wt%, the trapped melt model of Wasson andRichardson (2001) had success at matching the Ir–AuIVA trend but has not been applied to the Ge–Au trend.

3.3.5. Crystallization of the IID group

Although a small group with 21 members, group IIDshows well defined and interesting chemical trends(Wasson and Huber, 2006). As in the IVA group, theIr (Fig. 4e) and Ge (Fig. 5e) trends suggest different bulkS contents when fit by simple fractional crystallization.As with the IVA group, the inferred initial bulk Scontent consistent with the Ge trend (12wt%) is about


6wt% higher than the S content consistent with the Irtrend (6wt%). However, due to the small number of IIDmembers, only two meteorites are inconsistent with theGe trend as modeled with 6wt% S. Overall, the IIDgroup, like the IVA group, appears to have solidified ina process that was more complicated than simplefractional crystallization. The trapped melt model ofWasson and Huber (2006) suggests an initial S contentof 0.7wt% and formation by crystallization of a P-richlower liquid in a stratified, two-layer core. The trappedmelt model nicely reproduces the Ir–Au IID trend buthas not been used to model the IID trends of Ge–Au.

3.3.6. Crystallization of the IAB group

The IAB group is a ‘‘silicate-bearing group’’ ratherthan a ‘‘fractionally crystallized group.’’ Fig. 6 plots theIr–Au and Ge–Au trends, and also included are IIICDand other irons that were classified as belonging to thelarger ‘‘IAB complex’’ of Wasson and Kallemeyn(2002). By comparing these trends to those of Figs. 4and 5, it is clear that the IAB group shows considerablescatter in the compositional values of its members incomparison to the previously discussed groups. Despitethese compositional variations, it is still interesting toexamine if any of the trends are consistent with simplefractional crystallization. Fig. 6 shows that many of theIAB group irons, but not the lower Ge members, aregenerally consistent with simple fractional crystalliza-tion of a metallic melt with a very high bulk S content.

Fig. 6. Comparison of Ir–Au and Ge-Au trends in the IAB group (b

models for 0–27wt% S (a and c). All graphs are scaled to have factor

Ge y-axis, to enable comparisons between the models and the IAB

Wasson and Kallemeyn (2002). Simple fractional crystallization mod

was reached. On (b and d), model curves are marked with tick mar

The formation and evolution of the IAB groupcontinues to be debated. Wasson and Kallemeyn (2002)argued that the main group was not formed by fractionalcrystallization but, rather, involved crystal segregationwith solid and melt essentially in equilibrium andmultiple impact induced melting events to create thedifferent subgroups. Others have suggested fractionalcrystallization of melts in parent bodies that weredisrupted during the crystallization process, incompletedifferentiation of the parent asteroid, partial meltingfollowed by mixing events, the crystallization of S-richcores, or a combination of multiple processes (Kracher,1985; McCoy et al., 1993; Benedix et al., 2000; Takeda etal., 2000). The common presence of silicate inclusions isan important constraint on the history of these meteor-ites. Overall, there is agreement that the IAB ironmeteorites experienced a crystallization history andenvironment substantially different from the ‘‘fraction-ally crystallized’’ groups and from just simple fractionalcrystallization of an undisturbed metallic melt.

4. Cooling rates

4.1. Formation of the Widmanstatten pattern

4.1.1. Binary Fe–Ni

The conventional explanation for the formation of theWidmanstatten pattern is based on the binary Fe–Ni

and d) with those calculated for simple fractional crystallization

s of 10 on the Au x-axis, 2000 on the Ir y-axis, and 1000 on the

iron meteorite trends. Data for the iron meteorites are from

els were run until the Fe–FeS eutectic composition of 31wt% S

ks that correspond to every 10% of crystallization.

ARTICLE IN PRESS

900

800

700

600

500

400

300

200

1000 10 20 30 40 50 60 70

Ni CONTENT (wt. %)

TEM

PE

RAT

UR

E (°

C)

α

Ms

γ

γ1γ 2

(γ1)(γ 2) γ’

γ "Tc

γ"

Tcγ

Fig. 7. Fe–Ni binary phase diagram (Yang et al., 1996). a is a

low-Ni bcc phase, g a high-Ni fcc phase, g1 a low-Ni

paramagnetic fcc phase, g2 a high-Ni ferromagnetic fcc phase,

g0 is ordered Ni3Fe, g00 is ordered FeNi–tetrataenite, and Ms is

the martensite starting temperature. Tcg is the Curie tempera-

ture of the g phase. Tcg00 is the ordering temperature of FeNi,

g00.4

5

6

7

8

9

10

11

12

0Distance (μm)

Ni (

wt.%

)

TAENITE TAENITE

KAMACITE

Ni PEAK

10 μm

T

T

K

5 10 15 20 25

Fig. 8. (a) Reflected light micrograph of a Fe–9.0wt% Ni

alloy cooled from 700 1C through the kamacite+taenite phase

field at a rate of 4 1C/h to 500 1C and then removed from the

furnace. A kamacite precipitate (K) formed at a taenite/taenite

grain boundary (taenite is indicated by a T) Arrows mark the

position of the taenite/taenite boundary. (b) Ni traverse

measured with the electron microprobe across the kamacite

precipitate (Reisener and Goldstein, 2003).


equilibrium phase diagram (Fig. 7, Yang et al., 1996) inwhich a meteorite cools from the one-phase taenite (g)region of the diagram into the two-phase a+g regionwhere kamacite (a) nucleates and grows. However, theWidmanstatten pattern does not form by this process.Experimental evidence shows that for Fe–Ni alloys, theg-solid solution transforms to a supersaturated a-solidsolution, a2, martensite during cooling (Allen andEarley, 1950). No composition change takes place. Onlyduring reheating from low temperatures does thesupersaturated solid solution a2 decompose into a+g(kamacite+taenite), with equilibrium compositionsgiven by the a+g tie line at the reheating temperature.Reisener and Goldstein (2003) slow cooled a number ofbinary Fe–Ni alloys from taenite into the two-phasea+g fields and found that kamacite formed along theoriginal taenite grain boundaries and at taenite triplejunctions (Fig. 8). However, no intragranular (matrix)kamacite precipitates, such as Widmanstatten kamaciteplates, were observed. Narayan and Goldstein (1984a)analyzed a number of binary Fe–Ni experimental alloysand also found an absence of intragranular kamacite.Therefore, there is strong experimental evidence that thereaction sequence g-a+g is not effective for thenucleation of the Widmanstatten pattern in single crystalregions of iron meteorites.

4.1.2. Effect of P and the Fe–Ni–P phase diagram

In a study of the effect of P on the formation of theWidmanstatten pattern in iron meteorites, Goldsteinand Doan (1972) produced Widmanstatten kamacite inthe intergranular or matrix region of the original taenitewith an Fe–Ni alloy containing as little as 0.1wt% P.Fig. 9 shows the microstructure of a 9.8wt% Ni,0.3wt% P alloy slowly cooled from the all taenite fieldto 650 1C. A Widmanstatten pattern of kamacite platesis observed in the intragranular (matrix), which was fcctaenite at higher temperatures. In this alloy kamaciteformed by the reaction path g-g+Ph-a+g+Ph,where Ph is schreibersite, (Fe–Ni)3P. As suggested byNarayan and Goldstein (1984a), the nucleation ofintragranular kamacite in low P alloys does not begin

ARTICLE IN PRESS

10 μm

Fig. 9. Reflected light micrograph of the microstructure of a

9.8wt% Ni–0.3wt% P alloy slowly cooled to 650 1C. The

microstructure is a Widmanstatten pattern of kamacite in a

taenite matrix. A few phosphides are present as seen appearing

above the plane of polish (note arrow) (Goldstein and Doan,

1972).

400

500

600

700

800

900

210P (wt.%)

T (°

C)

3 phase corner

8.5 wt.% Ni

Ms

c b

α+γ+Ph.

γ+Ph.

γ

α+γ

α+Ph.

a

Fig. 10. Iso-Ni concentration section at 8.5wt% Ni from the

Fe–Ni–P phase diagram. Ms represents the martensite start

temperature. Vertical dashed lines a–c represent alloys with

different P contents cooling from high temperature to low

temperature. Alloy ‘‘a’’ forms a Widmanstatten pattern by


when the alloy enters the two- phase a+g field, butstarts only when the alloy enters the three-phasea+g+Ph field. It is clear that P has a very importantrole in the nucleation of the Widmanstatten pattern iniron meteorites and also in metal particles of stony-ironscontaining P (e.g., pallasites, mesosiderites).

mechanism II, g-g+Ph-a+g+Ph. Alloy ‘‘b’’ forms a

Widmanstatten pattern by mechanism III, g-(a+g)-a+g+Ph. Alloy ‘‘c’’ forms a Widmanstatten pattern by

mechanism V, g-a2+g-a+g (Yang and Goldstein, 2005).

The g corner of the three-phase a+g+Ph phase field is shown

by the filled circle. Open circles are experimental data from

Doan and Goldstein (1970).

4.1.3. Nucleation of the Widmanstatten pattern in high

and low P meteorites

The ternary Fe–Ni–P phase diagram has beendetermined experimentally to temperatures below500 1C (Doan and Goldstein, 1970; Romig and Gold-stein, 1980). The formation of the Widmanstattenpattern is determined by the bulk Ni and P content ofthe meteorite and by its path through the ternaryFe–Ni–P phase diagram as the meteorite cools from theall taenite field at high temperatures. Depending on theNi and P content, the kamacite nucleation temperaturecan be determined from either the (g+Ph)/(a+g+Ph)phase boundary, the (a+g)/(a+g+Ph) phase bound-ary, or the martensite start temperature, Ms, where a2 –martensite forms (Yang and Goldstein, 2005). As anexample, Fig. 10 shows an iso-Ni concentration sectionat 8.5wt% Ni from the Fe–Ni–P diagram (Doan andGoldstein, 1970). Alloy ‘‘a’’ is a high P meteoriteas it passes first through the g+Ph phase field beforea-kamacite nucleates. The mechanism for Widmanstat-ten pattern formation is: g-g+Ph-a+g+Ph (Me-chanism II, Yang and Goldstein, 2005). Fig. 9 illustratesa Widmanstatten pattern, which formed by MechanismII in an experimental alloy.

Narayan and Goldstein (1984a) determined experi-mentally that the a phase does not nucleate when the Pcontent is too low to enter the three-phase a+g+Phfield via the g+Ph phase field. The a phase nucleatesonly when g is saturated in P and enters the a+g+Phfield upon cooling. The reaction path for low Pmeteorites is shown for alloys ‘‘b’’ and ‘‘c’’ in Fig. 10.For a meteorite which cools below the (a+g)/(a+g+Ph) boundary before the Ms temperature isreached (alloy ‘‘b’’ in Fig. 10), the reaction for formingthe Widmanstatten pattern is g-(a+g)-a+g+Ph.The designation (a+g) indicates that the alloy passesthrough the a+g two-phase field but kamacite nuclea-tion is suppressed. This reaction scheme is calledMechanism III (Yang and Goldstein, 2005). For ameteorite which cools below Ms before it reaches the(a+g)/(a+g+Ph) phase boundary (alloy ‘‘c’’ in

ARTICLE IN PRESS

0.01

0.1

1

5Ni Content (wt.%)

P C

onte

nt (w

t.%)

IIIABIVA

Mechanism III

Mechanism V

Massive transformation

Mechanism II

7 9 11 13 15 17 19

Fig. 11. Dependence of the formation of the Widmanstattenstructure on the Ni and P content of groups IIIAB and IVA

(Yang and Goldstein, 2005). Mechanism II represents the

reaction g-g+Ph-a+g+Ph. Mechanism III represents the

reaction g-(a+g)-a+g+Ph, and Mechanism V represents

the reaction g-a2+g-a+g.

K

16151413121110987654

WE

IGH

T %

Ni

DISTANCE

1000 Å

� �

Fig. 12. Experimental evidence that computer simulation

programs for the growth of kamacite are applicable for the

growth of the Widmanstatten pattern in iron meteorites. (a)

TEM photomicrograph showing an a-kamacite crystal (K)

that nucleated and grew in a matrix of taenite. The alloy

contained 6.88wt% Ni and 0.49wt% P and was cooled from

790 to 650 1C at a rate of 5 1C/day (Narayan and Goldstein

(1984a). The taenite matrix has transformed to martensite

below 500 1C during cooling. (b) Ni concentration profile

across a kamacite/taenite interface in the Fe–Ni–P alloy from

(a). The TEM thin section was analyzed with an analytical

TEM with X-ray analysis capability. The solid line is the

profile predicted by the numerical model (Narayan and

Goldstein, 1984b).


Fig. 10), the reaction for forming the Widmanstattenpattern is g-a2+g-a+g, Mechanism V (Yang andGoldstein, 2005). The a2 martensite starts to form belowMs, the martensite start temperature (Fig. 7). As coolingproceeds, newly formed martensite begins to decomposeinto a+g. Some of the high-temperature g phase mayalso be retained.

The bulk Ni and P contents of the meteoritedetermine which of the three reactions control theformation of the Widmanstatten pattern. Fig. 11shows the Ni and P contents of chemical groups IIIABand IVA as well as the Ni–P boundaries for theregion of applicability of mechanism II for high P ironsand of mechanisms III and V for low P irons using theFe–Ni–P phase diagram. This diagram shows that theWidmanstatten structure in the high P members ofgroup IIIAB forms by mechanism II, g-g+Ph-a+g,while in the low P members of groups IIIAB theWidmanstatten structure forms by either mechanism III,g-(a+g)-a+g+Ph or mechanism V, g-a2+g-a+gU The Widmanstatten structure of all themeteorites in low P group IVA forms by mechanismV. Therefore, for the most populous group, IIIAB, allthree mechanisms are applicable while for chemicalgroup IVA, only mechanism V is applicable. Todetermine the cooling rate of an individual meteoritein any of the chemical groups (Table 1), the appropriateformation mechanism for the Widmanstatten patternmust be established first, using the bulk Ni and Pcontent, so that an appropriate cooling rate simulationcan be applied.

4.1.4. Growth of the Widmanstatten pattern

The formation of the Widmanstatten pattern in ironmeteorites is controlled initially by nucleation asdiscussed in Section 4.1.3 and then by kamacite growth


in the surrounding taenite as temperature decreases.Experiments have been performed to nucleate and growintragranular a – kamacite as a function of cooling timeand temperature in Fe–Ni–P alloys (Narayan andGoldstein, 1984a, b). Fig. 12a shows an a-kamacitecrystal that nucleated and grew in taenite and Fig. 12bshows the Ni concentration profile measured across thea-kamacite/g-taenite interface of the kamacite crystal.Narayan and Goldstein (1984b) observed that growthkinetics are dictated by the bulk diffusion of Ni intaenite and equilibrium partitioning of Ni and Pbetween kamacite and taenite takes place at the a/ginterface. A numerical model to simulate growth of thiskamacite crystal was developed based on these observa-tions. The solid line on Fig. 12b is the profile predictedby the numerical model and forms an excellent fit withthe measured data. The numerical model used tosimulate Ni redistribution and kamacite growth of theexperimental alloys is the same model used to simulatethe formation of the Widmanstatten pattern in ironmeteorites. Although the numerical model used forWidmanstatten pattern growth extrapolates Ni distribu-tions for a process that takes millions of years, it hasvalidity based on its successful application to experi-mental alloys.

4.2. Metallographic cooling rates

4.2.1. Cooling rate model

The metallographic cooling rate model simulatesgrowth of the Widmanstatten pattern and distributionof Ni content in kamacite and taenite phases duringgrowth. The model considers five major factors: (1)mechanism for Widmanstatten pattern formation, (2)kamacite nucleation temperature, (3) effect of impinge-ment, (4) Fe–Ni and Fe–Ni–P phase diagrams, and (5)interdiffusion coefficients.

Depending on the bulk Ni and P content of ameteorite, the reaction that controlled Widmanstattenformation and the appropriate nucleation temperaturefor the Widmanstatten pattern can be determined(Fig. 11). Impingement is caused by overlapping Nigradients from the growth of adjacent kamacite platesspaced apart by a distance L. The effect of impingementis to restrict the growth of kamacite and increase theamount of Ni in taenite during formation of theWidmanstatten pattern (Saikumar and Goldstein,1988). The Fe–Ni and Fe–Ni–P phase diagrams can beused to obtain the equilibrium phase compositions ofkamacite and taenite for the P-bearing metal phasesduring the cooling process. The interdiffusion coeffi-cients of Ni in kamacite and taenite control the rate atwhich Ni can be transferred between the two phases. Phas the effect of increasing the diffusion rate of Ni in

taenite significantly (Dean and Goldstein, 1986; Yangand Goldstein, 2006).

Numerical models have been developed which simu-late diffusion controlled kamacite growth in taenite(Widmanstatten pattern development) in the Fe–Ni–Pphase system. A constant cooling rate is usuallyassumed for the temperature range in which theWidmanstatten pattern forms. The model includes theapplicable nucleation mechanisms (II, III, and V) forWidmanstatten pattern formation and the kamacitenucleation temperature for the bulk Ni and P content ofa specific iron meteorite. The model also includes theequilibrium tie lines in the Fe–Ni and Fe–Ni–P phasediagrams and the binary and ternary diffusion coeffi-cients as a function of temperature and composition.Impingement is considered by assuming or measuringthe distance (L) between adjacent kamacite plates. As L

decreases, overlapping Ni gradients in the taenitebetween two kamacite plates are calculated. The Niprofile in kamacite and taenite for a specific cooling rateand bulk meteorite Ni and P content is the output of thecomputer model. Fig. 17 shows calculated Ni profiles forIVA irons Bishop Canyon and Duchesne, each with adifferent cooling rate and impingement distance. [Fordetailed discussions, see Hopfe and Goldstein (2001),and Yang and Goldstein (2006)].

4.2.2. Compositional measurements

Ni, Fe, Co, and P compositions are measured acrosskamacite–taenite–kamacite areas using the electronprobe microanalyzer (EPMA) on properly preparediron meteorite samples (Fig. 13). A typical M shaped Niprofile is obtained across each taenite band. Measure-ments of the central taenite Ni content and thecorresponding taenite half-width are also made(Fig. 13). The half-width of the taenite band can onlybe measured after the kamacite/taenite band orientationwith respect to the analyzed surface is obtained (Yangand Goldstein, 2006).

4.2.3. Methods for the measurement of metallographic

cooling rates

A variety of metallographic methods have been usedto determine the cooling rates of the Widmanstattenpattern in iron meteorites. The taenite profile matchingand the taenite central Ni content methods are the mostaccurate. In the taenite profile-matching method, the Nicontent vs. distance profile in taenite is computed forseveral cooling rates from the cooling rate model as afunction of bulk Ni and P content and taenite half-width. The calculated Ni composition profile in taenite(Ni vs. distance) is then plotted and compared with Nicomposition profiles measured for a given meteoritewith the EPMA (Goldstein and Ogilvie, 1965). For eachtaenite band the orientation of the kamactie plates mustbe measured in order to obtain accurate distances

ARTICLE IN PRESS

10

15

20

25

30

35

40

1Taenite Half Width (micron)

Ni C

onte

nt (w

t.%)

Meas.

Calc.

Bishop Canyon

2000°C/My3000

10

15

20

25

30

35

40

1Taenite Half Width (micron)

Ni C

onte

nt (w

t.%)

Meas.

Calc.

50°C/Myr100

Duchesne

200

10 100

10 100

a

b

Fig. 14. Central Ni content vs. taenite half-width for Bishop

Canyon and Duchesne IVA irons. The taenite half-width

measurements were corrected for effects of orientation.

Calculated cooling rate curves are also plotted for each

meteorite.

0

10

20

30

40

50

60

70

-20Distance (micron)

Ni C

onte

nt (

wt.%

)

Meas.Calc.

Bishop Canyon

3000 °C/Myr

0

10

20

30

40

50

60

70

-30Distance (micron)

Ni C

onte

nt (

wt.%

)

Meas.Calc.

Duchesne

200 °C/Myr

-10 0 10 20

-20 -10 0 10 20 30

a

b

Fig. 13. Comparison between measured and calculated Ni

profiles across the taenite phase in the Bishop Canyon and

Duchesne IVA irons. The taenite half-width is 3.5 and 8.0 mmfor Bishop Canyon and Duchesne, respectively.


traversed by the EPMA. Fig. 13 shows measured andcalculated Ni composition profiles for the BishopCanyon and Duchesne IVA irons that have differentcompositions and cooling rates. The measured coolingrate of Bishop Canyon is 3000 1C/Myr, while that ofDuchesne is 200 1C/Myr. To obtain a statisticallyaccurate cooling rate for a meteorite, measured andcalculated Ni profiles from a minimum of threekamacite bands need to be compared.

In the taenite central Ni content method, the Nicontent in the center of taenite is computed for severalcooling rates from the cooling rate model as a functionof bulk Ni and P content and taenite half-width. Thecalculated central Ni content is then plotted vs. the half-width of the taenite for several cooling rates (Wood,1964). Ni contents in the center of taenite phases ofvarious half-widths are measured for a given meteoritewith the EPMA. These data are plotted on the samegraph as the computer simulated iso-cooling curves ofcentral Ni content vs. taenite half-width. The measured

data should fall along one of the computed iso-coolingrate curves (Wood, 1964). A variation of this coolingrate method was developed by Rasmussen (1981), whomeasured the local bulk Ni and bulk P contents for eachtaenite lamella. Fig. 14 shows computed iso-cooling ratecurves and measured taenite central Ni contents for theBishop Canyon and Duchesne IVA irons. The datafollow iso-cooling rate curves from which a cooling ratecan be measured. A cooling rate of 2500 1C/Myr with a2s uncertainty range of 1.3, 1920–3250 1C/Myr, forBishop Canyon and a cooling rate of 100 1C/Myr with a2s uncertainty range of 2.9, 30–290 1C/Myr, forDuchesne were measured (Yang et al., 2007). Thesevalues compare well with those obtained for thesemeteorites from the profile-matching method (Fig. 13).


Central Ni content measurements of about 10 taenitebands are needed to obtain a statistically accuratecooling rate. For each band the orientation of thekamactie plates must be measured to determine accuratetaenite half-widths. The measured data for the taenitecentral Ni content method is usually easier to obtainthan for the taenite profile-matching method. Therefore,the taenite central Ni content cooling rate method isusually employed.

Other metallographic cooling rate methods (kamacitebandwidth, taenite maximum Ni, and kamacite centralNi content) have been developed, although they havelimited applicability. The kamacite bandwidth methodof Short and Goldstein (1967) relates the width of theWidmanstatten kamacite to the cooling rate. Themethod cannot be employed since the effects ofimpingement are not considered and a constant amountof undercooling before nucleation of the Widmanstattenpattern is assumed (Saikumar and Goldstein, 1988). Thetaenite maximum Ni method of Short and Goldstein(1967) relates the maximum Ni content in the taenitenext to the kamacite/taenite border measured with theEPMA with the cooling rate. The method has fallen outof favor because it is sensitive to impingement effectsand to the exact placement of the electron beam close tothe kamacite/taenite interface. A correlation betweenthe measured and calculated Ni content in the center ofa kamacite band for a given cooling rate has beenobserved (Powell, 1969; Haack et al., 1996a). Unfortu-nately the uncertainty in the detailed shape of thea/(a+g) solvus line for the Fe–Ni–P phase diagramleads to errors in the predicted Ni variation in thekamacite phase as a function of cooling rate. Thismethod is also relatively insensitive to cooling ratevariations (Hopfe and Goldstein, 2001).

The hexahedrites, which have Ni contents of5–6.5wt%, do not form a Widmanstatten pattern oncooling. These meteorites contain sufficiently high P sothat it is possible to obtain cooling rates by measuringNi gradients across kamacite–phosphide–kamaciteregions with the EPMA. A comparison is made usingthese data with computer simulated Ni profiles ofphosphide growth in the Fe–Ni–P system as a functionof cooling rate (Randich and Goldstein, 1978). For eachphosphide, the orientation of the plates must bemeasured to order to obtain accurate distances to usein comparing measured and simulated Ni growthprofiles.

4.3. Measured cooling rates

Since the early studies of Wood (1964) and Goldsteinand Ogilvie (1965), metallographic cooling rates havebeen measured for a large number of irons. Thecomputer simulation methods have become more

accurate as the major effect of P on the nucleation andgrowth of the Widmanstatten pattern has been recog-nized. For example, the measured cooling rates of IVAirons Bishop Canyon and Duchesne have increased overthe last 40 years by a factor of 10–50 (Goldstein andShort, 1967; Yang et al., 2008a) as the accuracy of themethod has improved. The latest cooling rate measure-ments for the IIIAB and IVA chemical groups (Yangand Goldstein, 2006; Yang et al., 2007, 2008a) arethe most accurate to date, as all five major factors forthe measurement of cooling rates are considered in thesimulation model and kamacite/taenite orientations aremeasured. Potential inaccuracies in each of these factorsare discussed for the IIIAB irons by Yang and Goldstein(2006).

To establish whether there is any variation of coolingrate within a given chemical group, it is important tominimize the uncertainty in the cooling rate measure-ment of each meteorite and to evaluate their accuracy.The major uncertainty is due to measured deviationsfrom the calculated iso-cooling curve for the taenitecentral Ni content method. For example, each datapoint for Bishop Canyon or Duchesne in Fig. 14represents one measurement from a single taenite band.The deviation from the average iso-cooling curve, thecooling rate of the meteorite, gives the uncertainty in thecooling rate. Recent measurements list the cooling ratemeasurement along with the 2s uncertainty in thecooling rate determination, which range from a factor of1.3 for Bishop Canyon to 3.0 for Duchesne, with anaverage of 2.0 for 24 IIIAB and IVA irons (Yang andGoldstein, 2006; Yang et al., 2007, 2008a). Clearly, toestablish cooling rate trends in chemical groups, carefulattention must be paid to determining errors in themeasurement of each individual cooling rate. At thistime it is not possible to determine if cooling rates varywithin a chemical group, unless the individual coolingrates vary by more than a factor of 2.

Measured variations in cooling rates for a specificchemical group that exceed the statistical uncertainty inthe cooling rate of 2.0 provide strong evidence that theparent asteroid does not have a conventional metal coresurrounded by a silicate mantle. The highest qualitycooling rate measurements have been obtained forgroups IIIAB and IVA (Fig. 15), particularly studiesin which the orientation of kamacite/taenite interfaceshave been measured (Yang and Goldstein, 2006;Yang et al., 2007, 2008a). Cooling rates vary from 56to 338 1C/Myr for the IIIA irons and from 100 to6600 1C/Myr for the IVA irons, variations of 6 and 66,respectively. In both cases, the cooling rate rangesexceed those expected for samples from a core enclosedby a silicate mantle, as such samples should haveindistinguishable cooling rates.

Table 2 summarizes the measured cooling ratedata for chemical groups IIIAB, IVA, IAB, IIICD

ARTICLE IN PRESS

IIIAB

IVA

10

100

1,000

10,000

100,000

7Ni content (wt.%)

Coo

ling

rate

(°C

/Myr

)C

oolin

g ra

te (°

C/M

yr)

1000

100

10

16 7 8 9 10 11

Bulk Ni (wt. %)

8 9 10 11

Fig. 15. Cooling rate measurements for iron meteorites in chemical groups IIIAB and IVA (Yang and Goldstein, 2006; Yang et al.,

2007). Variations of a factor of 6 and 66 are observed in Groups IIIAB and IVAB, respectively. The error bar for each meteorite

represents the 2s uncertainty range.

Table 2. Cooling rate variations in iron meteorite chemical groups.

Group Cooling rate variation (1C/Myr) Authors Method

IAB 2–3 Goldstein and Short (1967) 1

63–980 Rasmussen (1989) 1

25–70 Herpfer et al. (1994) 2

IIICD 87–480 Rasmussen (1989) 1

IIAB 0.8–10 Randich and Goldstein (1978) 3

IIIAB 1.0–10 Goldstein and Short (1967) 1

21–185 Rasmussen (1989) 2

56–338 Yang and Goldstein (2006) 2

IVA 7–90 Goldstein and Short (1967) 1

2–96 Rasmussen (1982) 2

19–3400 Rasmussen et al. (1995) 2

100–6600 Yang et al. (2008a) 2

IVB 2–25 Goldstein and Short (1967) 1

110–450 Rasmussen et al. (1984) 1

1400–17,000 Rasmussen (1989) 1

1. Kamacite bandwidth method (Short and Goldstein, 1967).

2. Taenite central Ni content method (includes effects of P on phase diagram and on diffusion coefficients for Ni in taenite).

3. Phosphide growth simulation.


(a subgroup of the group IAB complex, Wasson andKallemeyn, 2002), IIAB, and IVB. Much of the oldercooling rate data (chemical groups IAB, IIICD, IVB)

were obtained using the kamacite bandwidth method(Short and Goldstein, 1967). The kamacite bandwidthmethod cannot be used as discussed in the previous


section, since the effects of impingement are notconsidered and a constant amount of undercoolingbefore nucleation of the Widmanstatten pattern isassumed. It is possible that the IVB cooling ratesobtained by the kamacite bandwidth method may beuseful, but only if the orientation of the bands withrespect to the polished surface are measured so thataccurate bandwidths are obtained. From the best-quality data given in Table 2, it appears that there aresignificant, 2s42, variations in cooling rates not onlyfor IIIAB and IVA irons but possibly for IIAB andIVB irons.

Possible cooling rate variations for chemical groupsIIAB and IVB (Table 2) need to be re-investigated inorder to learn more about the size, silicate mantle, etc. oftheir parent asteroidal bodies. The number of measuredcooling rates for Group IAB (Herpfer et al., 1994) islimited and the cooling rates for subgroup IIICD of thegroup IAB complex (Table 2) need to be remeasured.The effect of C content on the IAB and IIICD irons mayalso have an effect on phase boundaries and interdiffu-sion coefficients. It is important to obtain cooling ratesfor the IAB irons as this is the second largest chemicalgroup and is silicate rich. This chemical group experi-enced a crystallization history different from the‘‘fractionally crystallized’’ groups such as IIIAB, IVB,and IVA. Measured cooling rates will help in under-standing how silicates were incorporated in the parentalmelt, the nature of the mantle, and the nature of thesolidification process in the parent asteroidal body.

Fig. 16. (a) Reflected light micrograph of the microstructure of

decomposed taenite near the kamacite–taenite boundary of the

Dayton IIICD iron meteorite (Yang et al., 1997a). (b) TEM

bright field image of a kamacite/decomposed taenite region in

the Tazewell, IIICD iron meteorite (Reuter et al., 1988). K is

kamacite, CT1 is the outer taenite rim, CZ is the cloudy zone

region, CT2 is an unetched part of the cloudy zone called clear

taenite 2, and M is finely decomposed martensite (high-Ni

plessite).

4.4. Cooling rates obtained from the cloudy

zone structure

Although most studies of the thermal history ofmetallic Fe–Ni grains have focused on the metallo-graphic cooling rates applicable at �650–400 1C, relativecooling rates at lower temperatures can also be obtainedby analyzing the microstructure of taenite rims. Duringcooling to low temperatures (o400 1C), several phasetransformations take place within the Ni gradient in thetaenite phase (Fig. 13). Fig. 16a shows an opticalmicrograph of the microstructure that formed in thetaenite Ni gradient close to the kamacite–taenite inter-face. A clear taenite region (CT1) is observed next to thekamacite–taenite interface followed by a heavily etchedregion called the cloudy zone (CZ) (Scott, 1973). As theNi content decreases away from the kamacite–taeniteborder, there is an unetched cloudy zone (CT2) and thena decomposed martensite region (M). Fig. 16b displays abright field transmission electron microscope image ofthe CT1/CZ microstructure in a thin section of theTazewell IIICD iron at much higher magnification. Theclear taenite zone, CT1, contains FeNi, g2, which issometimes ordered as tetrataenite. The cloudy zone, CZ,

is two phase with a high-Ni fcc FeNi island phasesurrounded with a low-Ni honeycomb phase, which isusually bcc martensite (Reuter et al., 1988). Fig. 17ashows Fe and Ni scanning X-ray maps of a 2� 2 mmkamacite-clear taenite–cloudy zone interface region in athin section of the Carlton IIICD iron meteorite. A two-phase structure of the cloudy zone and the chemicalcomposition of the island and honeycomb phases areobserved. Fig. 17b shows the Ni and Fe variation alonga line scan across the kamacite-clear taenite–cloudyzone. The Ni content of the clear taenite regionvaries from 450wt% at the kamacite–taenite interfaceto �45wt% at the boundary with the cloudy zone.

ARTICLE IN PRESS

Fe Wt.% Ni Wt.%2μm field of view

-500 0 500 1000 1500-500 0 500 1000 1500nmnm

5 101520253035404550554550556065707580859095

K CT CZ

-400

-200 0

200

400

600

800

1000

1200

1400

1600

0

10

20

30

40

50

60

70

80

90

100

Wt.

%

Distance [nm] from Kamacite

KCT

CZNi

Fig. 17. (a) Ni and Fe distribution across the kamacite–clear

taenite–cloudy zone region in the Carlton iron meteorite,

IIICD. The size of the scan is 2� 2 mm. (b) Ni and Fe

concentration (wt%) across the same 2 mm region (blue is Fe

and green is Ni).


The cloudy zone region has �42wt% Ni at theboundary with tetrataenite and grades to �30wt% atits boundary with the decomposed martensite. Thestructure of the cloudy zone develops by a spinodalreaction in which a high-Ni phase (g2) (island phase)forms in a matrix of low-Ni kamacite or taenite (g1)(honeycomb phase) (Yang et al., 1997a). The spinodalreaction occurs at low temperatures o350 1C (Fig. 7).Both the formation of tetrataenite and the spinodalreaction are cooling rate dependent. Tetrataeniteincreases in width, and the size of the high-Ni islandphase in the cloudy zone increases with decreasingcooling rate; both these microstructures can givecooling rate information at low temperatures o350 1C(Yang et al., 1997b). At Ni contents below �30wt%,plessite, a fine mixture of kamacite and tetrataeniteforms at low temperature and is observed in the centersof taenite regions between the plates (Fig. 1). The

structure of plessite and the mechanism of formation arediscussed in detail by Buchwald (1975), Zhang et al.(1993), Yang et al. (1997a), and Goldstein and Michael(2006).

Yang et al (1997b) proposed an empirical cooling rateindicator for meteoritic metal based on the size of thehigh-Ni particles, island phase, in the cloudy zone,which together with the surrounding low-Ni phase forma ‘‘honeycomb’’ structure (Figs. 16 and 17). Theymeasured the size of the high-Ni taenite particles, islandphase, in the cloudy zone of various iron, stony-iron,and stony meteorites by high-resolution scanningelectron microscopy (SEM) at the outer edge of thecloudy zone adjacent to the tetratenite rim and showedthat the size increased with decreasing cooling rate. Theyfound that the size of the high-Ni particles ranges from400 to 450 nm in mesosiderites, which cooled veryslowly, to �20–40 nm in the rapidly cooled IVA ironmeteorites. Since the inverse relationship between high-Ni particle size (island phase size) and metallographiccooling rate holds not only for iron meteorites but alsofor metal in stony-irons, and stony meteorites, the scaleof the cloudy taenite microstructure provides a valuableguide to relative cooling rates of metal-bearing meteor-ites at 350–200 1C.

Measurements of the high-Ni particle size are moredifficult to obtain with the SEM for fast-cooledmeteorites. As the island phase size approaches theresolution of the SEM (p10 nm), specimen preparationbecomes more critical. The amount of chemical attackor etching used to develop the cloudy zone microstruc-ture can affect the apparent size and also the visibility ofthe microstructure as observed even in a high-resolutionSEM. In addition, the finest cloudy zone particles areespecially sensitive to shock heating, which can cause thecloudy zone to quickly coalesce or even disappear. High-Ni particle sizes o10 nm are measured by makingelectron transparent thin sections and analyzing thecloudy zone region using the TEM (Figs. 16 and 17).

Since Yang et al. (1997b), more accurate metallo-graphic cooling rates have been measured (Hopfe andGoldstein, 2001; Yang and Goldstein, 2006; Yang et al.,2007, 2008a) and more high-Ni particle sizes have beenmeasured (Yang et al., 2007; Goldstein et al., 2008). Thedirect relation between decreasing island phase size andincreasing cooling rate in the IVA irons (Fig. 18) is thesame as that observed by Yang et al. (1997b) for theoverall trend for iron, stony-iron, and stony meteorites.This variation takes place over a cooling rate range of 4orders of magnitude (0.2–6600 1C/Myr). As recognizedby Wasson and Richardson (2001), a straight linerelationship fits the data in Fig. 18. Cooling ratescannot be derived directly from cloudy taenite particlesizes because there is no effective model currentlyavailable for spinodal growth in the Fe–Ni system.However, the relative cooling rates of two meteorites,


(CR1) and (CR2), can be estimated from the ratio oftheir respective island phase or high-Ni particle sizes(IPS2/IPS1), namely: (CR1/CR2) ¼ (IPS2/IPS1)

n. Theparameter ‘‘n’’ obtained from Fig. 18 equals 2.470.4.

1

10

100

1000

0.1Cooling rate (K/Myr)

Hig

h N

i par

ticle

siz

e (n

m)

MesosideriteIIIABIVAPallasite

1 10 100 1000 10000

Fig. 18. Variation of high-Ni particle size vs. metallographic

cooling rate in mesosiderites, pallasites, IIIAB and IVA irons.

Data sources: (1) High-Ni particle size: mesosiderites and

pallasites (Yang et al., 1997b), IIIAB irons Cumpus and

Spearman using SEM measurements (Yang et al., 1997b).

Urachic (46 nm), Point of Rocks (44 nm), Chupaderos (58 nm),

Casas Grandes (42 nm), and Bella Roca (53 nm) using SEM

measurements, this study, IVA irons using TEM measure-

ments (Goldstein et al., 2008). (2) Metallographic cooling rates

– mesosiderites (Hopfe and Goldstein, 2001), pallasites (Yang

et al., 1997b), IIIAB irons (Yang and Goldstein, 2006), and

IVA irons (Yang et al., 2008a).

Table 3. Examples of iron meteorite ages determined by diverse is

Isotopic

system

Process dated Group

182Hf–182W Core formation IIAB, IID, IIIAB

IVB187Re–187Os Crystallization of molten

Fe–Ni

IIAB

IIIAB

IVA

IVB147Sm–143Nd Silicate closure at �1000K IAB (CC)53Mn–53Cr Phosphate closure at

�1000K

IIIAB

129I–129Xe Silicate closure at �1100 1C IAB (CC)40K–40Ar Silicate closure at

�575–700K

IAB (CC)

IAB40K–40Ar Impact reheating IIE (Watson)

Meteorite abbreviation: CC, Caddo County.aAbsolute ages for Hf–W and Al–Mg chronometers assume CAI formatiobUsing the angrite LEW 86010 to anchor the Mn–Cr chronometer (see Kita

initial 53Mn/55Mn is used as a proxy for the CAI initial (Shukolyukov and LcAge increases by 1Myr if the Shallowater age of Gilmour et al. (2006) isdAge increased by 21Myr using new 40K decay constant for consistency w

This equation can be used to obtain relative coolingrates, for example, between meteorites in the samechemical group. For the IVA irons, the differences in thehigh-Ni particle sizes vary by 2.970.5, indicating thatcooling rates in the temperature range where theWidmanstatten pattern forms vary by about a factorof 15 (Yang et al., 2007). For the IIIAB irons, thedifferences in the high-Ni particle sizes vary by1.7570.5, indicating that the cooling rates in thetemperature range where the Widmanstatten patternforms vary by about a factor of 5.

In TEM studies of IVA irons, Goldstein et al. (2008)observed that the widths of the tetrataenite, CT1,regions (Figs. 16 and 17) correlate directly with theisland phase size in the cloudy zone and with the coolingrate of the meteorite. The measurement of tetrataenitewidths can be easily corrected for orientation effects,since the electron transparent thin section obtainedby focused ion beam (FIB) techniques is cut normal tothe kamacite/taenite interface. Measurement of thewidths of tetrataenite regions can provide anotherindependent measurement of relative cooling rates atlow temperatures.

5. Iron meteorite ages

Radioactive isotopes with diverse chemical propertiesand half-lives of 4Myr to 40Gyr have decayed in ironmeteorites or their progenitor materials, providingconstraints on many different stages in their formation(e.g., Chabot and Haack, 2006; Wadhwa, 2007). The

otopic systems.

Age (Myr)a Age since

CAI (Myr)

References

, 456771.2 �0.271.2 Markowski et al.

(2006a, b)

4530750 37750 Cook et al. (2004)

4517732 50732 Cook et al. (2004)

4456725 110725 Smoliar et al. (1996)

4527729 40730 Smoliar et al. (1996)

4530720 40720 Stewart et al. (1996)

456371b 4.170.5b Sugiura and Hoshino

(2003)

4557.970.1c 9 Bogard et al. (2005)

4527711d 40711 Bogard et al. (2005)

4360–4560 7–200 Vogel and Renne (2008)

367677 890 Bogard et al. (2000)

n at 4567.170.16Myr (Amelin et al., 2002, 2006).

et al., 2005). The age decreases by 1Myr if the carbonaceous chondrite

ugmair, 2006).

used.

ith Vogel and Renne (2008).


ages in Table 3, are representative of those providing thebest constraints on their formation. Absolute ages arelisted as well as ages relative to CAIs, which formed4567Myr ago according to U–Pb dating (Amelin et al.,2002, 2006).

5.1. Hf–W metal–silicate fractionation ages

182Hf decays into 182W with a half-life of 8.9Myr.Since Hf is lithophile and W is siderophile, the ratio of182W in irons to the stable reference isotopes 183W and184W may constrain the time at which metal and silicatewere separated so that they were no longer able toequilibrate. For irons derived from bodies that weremelted sufficiently to form cores (probably all majorgroups except IAB and IIE), metal–silicate equilibriumprobably ceased when metal droplets settled throughsilicate melt to form a core, as in the terrestrial magmaocean (Rubie et al., 2003). After allowance for cosmic-ray effects, the 182W/184W ratio in these irons isremarkably similar to the inferred initial ratio ofCa–Al-rich inclusions (CAIs), which are the first objectsformed in the solar system (Markowski et al., 2006a, b;Kleine et al., 2005; Schersten et al., 2006; Qin et al.,2008; Burkhardt et al., 2008). For Negrillos (IIA) andGibeon (IVA), where the cosmic-ray effects are bestconstrained, the model ages for core formation in theirparent bodies are �1.071.3Myr after crystallization ofCAIs (Markowski et al., 2006b; Burkhardt et al., 2008).(A negative age would imply that core formationpreceded CAI formation!) For eight fractionally crystal-lized groups, Burkhardt et al. (2008) infer that coreformation took place o1Myr after CAIs. Irons insilicate-bearing groups IIE and IAB tend to have lower182W/184W ratios, implying more recent metal–silicateexchange (Markowski et al., 2006a). This might reflectyounger formation by impact melting of chondriticmaterial after cores formed in other bodies, or moreplausibly, partial metal–silicate exchange during cooling(Markowski et al., 2006a). [Note that the three ironslisted by Markowski et al. (2006a, b) as IIICD (See Choiet al., 1995) are now included in the main IAB group(Wasson and Kallemeyn, 2002).]

Some irons have especially low 182W/184W values,suggesting they may be older than CAIs. However, suchlow abundances of 182W are almost certainly a result ofcosmic-ray effects during long space exposure, notfrom early silicate–metal separation (Markowski et al.,2006b; Qin et al., 2008). Nevertheless, Humayun et al.(2007) have raised concerns about the accuracy of theCAI initial 182W/184W ratio obtained from AllendeCAIs, as aqueous fluids have caused alteration inAllende. The W contents of some CAIs in Allende havedefinitely been disturbed by alteration, however, the182W/184W value for CAIs derived by Kleine et al. (2005)

was obtained from both separated minerals andbulk CAIs and has been confirmed by Burkhardt et al.(2008).

Studies of Hf–W systematics of CAIs in less alteredchondrites and 182W/184W ratios in irons with shortcosmic-ray exposure ages would clearly be invaluable toconstrain more precisely when cores formed in asteroids.However, except possibly for IAB and IIE members, it isreasonably certain that iron meteorites come from thefirst bodies that accreted. Asteroids larger than �30 kmacross that accreted o1Myr after CAIs would havebeen heated sufficiently by 26Al to melt and form cores(Sanders and Taylor, 2005; Schersten et al., 2006).Chondrites probably come from bodies that accreted41.5Myr after CAIs formed when there was insuffi-cient 26Al to melt asteroids as chondrule formation agesare 2–5Myr after CAIs (Fig. 19). In addition, there is aninverse relationship between peak metamorphic tem-perature in a chondrite group and the chondruleformation age, supporting the concept that differen-tiated bodies accreted before the chondrite parent bodies(Scott, 2007).

5.2. Re–Os metal solidification ages

Re and Os are highly siderophile elements but smalldifferences in their preference for solid and liquid Fe–Nicause the Re/Os ratio to rise during crystallization.Strong correlations between the abundance of 187Os,which is the decay product of 187Re, and the Re/Os ratioamong the bulk compositions of irons in groups allowtheir crystallization times to be inferred (Smoliar et al.,1996). Since the decay constant for 187Re is poorlyknown, absolute ages have been derived by pegging theRe–Os age of IIIAB irons to the Pb–Pb crystallizationage of coarse-grained angrites. The inferred Re–Os agesof 4530750 and 4517732Myr for IIAB and IIIABirons, respectively in Table 3 from Cook et al. (2004) areconsistent with earlier studies by Smoliar et al. (1996)and others. These data do not reveal any significantdifferences between the low- and high-Ni ends ofindividual groups nor between groups, but they do limitthe duration of core crystallization to a period of lessthan 30Myr.

5.3. Mn–Cr cooling ages

Mn-bearing phosphates in IIIAB irons show acorrelation between the 53Cr/52Cr ratio and the Mn/Crratio, implying that 53Mn (half-life 3.7Myr), decayed inthe phospates to 53Cr (Sugiura and Hoshino, 2003). Theslopes of the isochrons for the phosphate sarcopside infour IIIAB irons (all with bulk Ni of �9–10%) givethe 53Mn/55Mn ratio when the mineral cooled throughthe isotopic closure temperature of �1000K. Using the

ARTICLE IN PRESS

Fig. 19. Ages of iron meteorites, chondrules and CAIs (major

components of chondrites), and angrites (basaltic meteorites)

inferred from four different radiometric techniques:182Hf–182W, 53Mn–53Cr, 26Al–26Mg, and 207Pb–206Pb. The W

isotopic compositions of ‘‘magmatic’’ which are those irons

that were fractionally crystallized, suggest that they come from

bodies in which cores formed o1Myr after CAIs. Chondrule

ages are 2–4Myr after CAIs, implying that chondrite parent

bodies accreted after those that melted to form the fractionally

crystallized irons. Group IAB and IIE irons (the silicate-

bearing irons) have higher 182W/184W ratios and younger

metal–silicate equilibration ages than the other groups,

reflecting later formation or metal-silicate exchange. Mn–Cr

isotopic data for phosphates in IIIAB irons suggest they

formed 4–5Myr after CAIs, around the time that the older

angrite basalts crystallized. Vertical lines show how relative

chronologies inferred from short-lived isotopes (182Hf, 53Mn,26Al) are linked to absolute ages inferred from U–Pb dating.

Error bars are 2s, but those derived using short-lived

radioisotopes do not include errors in the anchors’ ages.

Chondrule ages are identified by the group (CR, CO, CV, etc.)

and angrites by an abbreviated name (LEW 86010, SAH

99555, Angra dos Reis, etc.). Adapted from Scott and Sanders

(2009), who give the sources of the age data for chondrules,

CAIs, and angrite basaltic meteorites. For iron meteorite data

sources, see Table 3.


LEW 86010 angrite to anchor the Mn–Cr relative agesto the Pb–Pb absolute ages (Fig. 19) gives a time of4Myr after CAIs for the IIIAB phosphates to havecooled through �1000K. This Mn–Cr age wouldprobably require a metallographic cooling rate of�500C/Myr (cooling from �1500 to �700 1C in2Myr assuming that cooling started 2Myr after CAIformation). This rate is rather higher than the rate of�100C/Myr calculated by Yang and Goldstein (2006),but given the uncertainties in the Mn–Cr anchor age andthe metallographic cooling rate, the Mn–Cr cooling timeis not implausible.

5.4. Ar–Ar cooling and impact reheating ages

The decay of 40K to 40Ar in K-bearing silicateminerals in irons provides constraints on the time ofisotopic closure. Because the closure temperature infeldspar of �600–700K is lower than for most otherchronometers, this chronometer can be readily reset byimpact heating (e.g., Bogard et al., 2000, 2005). Ar–Arages of around 4.5Gyr for IAB silicates are thought toreflect primary cooling at this time while the 3.7Gyr ageof the Watson IIE iron probably reflects late impactreheating (Table 3).

6. Relationships with other meteorites

Possible links between iron meteorite groups andvarious stony and stony-iron meteorites have beenproposed primarily on the basis of O isotopic composi-tions of silicates, phosphates, and chromites (Franchi,2008). However, many differentiated meteorites haverather similar O isotopic compositions so that improve-ments in the precision of the O isotopic analyses havecaused some interpretations to be revised. For example,Clayton and Mayeda (1996) found that IIIAB irons hadO isotopic compositions indistinguishable from those ofeucrites, mesosiderites, and main-group pallasites fromwhich they inferred that the four types could all bederived from a single body. However, more precisemeasurements using laser fluorination showed thatmain-group pallasites and eucrites had distinctly differ-ent O isotopic compositions from eucrites and mesosi-derites and could not be from the same differentiatedbody (Greenwood et al., 2006). We conclude that exceptfor the link between group IAB irons and winonaites,the iron meteorite groups are probably not derived frombodies that supply us with other kinds of meteorites.

6.1. Group IAB irons and winonaites

Silicates in certain group IAB irons are mostcommonly angular inclusions with approximately chon-dritic mineralogy, but also occur as grains in sulfide andgraphite nodules and non-chondritic inclusions such asgabbroic clasts (Benedix et al., 2000; Takeda et al.,2000). The chondritic silicate inclusions are closelysimilar in mineralogy and O isotopic composition towinonaites, which are strongly metamorphosed andpartly melted, metal-bearing meteorites, some of whichappear to contain relict chondrules (Clayton andMayeda, 1996; Benedix et al., 1998). There is additionalevidence for a strong link between these two groups asthe Mount Morris (Wisconsin) meteorite, classified as awinonaite, may actually be a large silicate clast from thePine River IAB iron that was found nearby (Bevan and


Grady, 1988). In addition, the Winona meteorite itselfmay be a silicate inclusion from the Canyon Diablo IABiron (J.T. Wasson, private communication).

6.2. IIE irons and H chondrites

Similarities in the O isotopic and mineral composi-tions of silicate inclusions in IIE irons and H chondrites(Fig. 20) have persuaded many authors that the IIEirons formed on the H chondrite body by impactmelting (e.g., Casanova et al., 1995; Clayton andMayeda, 1996; Ikeda and Prinz, 1996; Gaffey andGilbert, 1998). However, Bogard et al. (2000) inferredfrom differences in exposure ages and mineralogy thatIIE irons and H chondrites came from separate bodies.High-precision analyses of H chondrite falls by Folcoet al. (2004) suggest that IIE irons and H chondriteshave different O isotopic compositions (Fig. 20).Although Folco et al. (2004) favored a common parentbody in view of some similar degassing ages of�3.7Gyr, Franchi (2008) concluded that this is unlikelyunless the internal structure and isotopic variation in theH chondrite body is poorly understood. We need high-precision analyses of IIE silicates to allow us moreinsight into the potential relationship between IIE ironsand H chondrites.

Fig. 20. Oxygen isotopic compositions of silicates in group IIE

irons and H4-6 chondrites (adapted from Franchi, 2008). IIE

silicates overlap with the shaded field of H4-6 silicates from

Clayton et al. (1991). However, more precise data for H

chondrite falls show only marginal overlap in D17O with IIE

irons (Folco et al., 2004), suggesting that IIE irons and H

chondrites may come from separate bodies. D17O is the vertical

displacement from the terrestrial fractionation line on a plot of

d18O vs. d17O, where d18O and d17O are the deviations in per

mil of the 18O/16O and 17O/16O ratios in the meteorite sample

from the standard values. Geological processes may shift

samples horizontally on this diagram but not vertically.

6.3. Group IIIAB irons and main-group pallasites

The composition of metallic Fe–Ni in main-grouppallasites is similar to that of Ni-rich IIIAB irons andcomparable to the calculated liquid composition aftermost of the IIIAB core had crystallized and the liquidwas near eutectic composition (Scott, 1977b; Wassonand Choi, 2003). The major discrepancies are somehigh-Ir pallasites and the relatively high concentrationsof Ga and Ge in pallasite metal. A common origin forIIIAB irons and main-group pallasites has also beeninferred from the O isotopic compositions of chromitesand phosphates in IIIAB irons (Clayton and Mayeda,1996; Fig. 21). However, laser fluorination analyses areneeded to test this. Strong evidence against formation ofIIIAB irons in the main-group pallasite body comesfrom studies of the cloudy zone microstructure showingthat the pallasites cooled more slowly than the IIIABirons (Yang et al., 2008b; Fig. 18).

6.4. Group IVA irons and L chondrites

The O isotopic compositions of silica and orthopyr-oxene inclusions in IVA irons are surprisingly hetero-geneous and overlap those of L and LL chondrites,suggesting that the irons may have formed from

Fig. 21. Oxygen isotopic compositions of silicates and oxides

in IIIAB irons, main-group pallasites, mesosiderites, and HED

achondrites (adapted from Franchi, 2008). Less accurate

Group IIIAB data are from Clayton and Mayeda (1996);

others from Greenwood et al. (2005, 2006). Large spread in the

D17O for IIIAB oxides, which probably reflects analytical

error, overlaps those of silicates in main-group pallasites,

mesosiderites and HED achondrites, suggesting possible links

with one or more of these groups. Chemical compositions of

metal in IIIAB irons and main-group pallasites support a

common source, but their cooling rates appear incompatible.


L/LL-like material (Clayton and Mayeda, 1996; Wanget al., 2004; Franchi, 2008). However, it seems unlikelythat these irons and chondrites come from the samebody as the former are derived from a body that wasalmost entirely molten, whereas the latter were notmelted significantly. Group IVA irons were not asso-ciated with L or LL chondrites when they cooled as thethermal model of Yang et al. (2007, 2008a) precludes thepresence of significant amounts of silicate aroundthe source of the IVA irons when the Widmanstattenpattern formed. The heterogeneity among the IVAoxide grains was attributed by Wang et al. (2004) toincomplete homogenization during parent body accre-tion, heating, and differentiation.

7. Formation of iron meteorite groups

Here we integrate chemical and isotopic constraintson the origins of groups IAB, IIE, IIIAB, IVA, and IVB,and the ungrouped irons. The first two groups are the‘‘silicate-bearing’’ groups, whose origin has not beenreviewed recently; the last three are the best-studied‘‘fractionally crystallized’’ groups.

7.1. Group IAB irons

Several unique features of IAB irons are difficult toreconcile with formation in a single metallic core: thehigh abundance of chondritic silicates, the lack of clearfractional crystallization trends like those in othergroups (Fig. 6), and the close association with thewinonaites, which were strongly metamorphosed butnot extensively melted (Benedix et al., 1998). Kracher(1985) argues that IAB irons might come from a singleS-rich core, but most authors favor an origin in metallicpools. These are thought to have formed either byimpact melting of cold chondritic material (Choi et al.,1995; Wasson and Kallemeyn, 2002) or by catastrophicimpact fragmentation and reassembly of a body thatwas already hot and partly molten (Benedix et al., 2000,2005).

Cooling rates of IAB irons during kamacite growthare too poorly constrained to make definitive statementsabout burial depths and body sizes (Table 2). Radio-metric ages of silicates are somewhat inconsistent, butBogard et al. (2005) infer that partial melting and impactdisruption and reassembly of the IAB body occurred at4.56Gyr and that metamorphism lasted until 4.52Gyr.However, they also speculate that the IAB body mayhave experienced a more complex history with milderimpact heating around 4.52Gyr. Vogel and Renne(2008) Ar–Ar dated plagioclase grains from four IABirons and found that some contained plagioclase grainswith a range of ages that were uncorrelated with grain

size. For example in Caddo County, they found a rangefrom 4.4770.02 to 4.5670.02Gyr. They inferred thatthe silicate grains in a single iron cooled at differentrates to below 350 1C (the approximate blockingtemperature for K–Ar) at different locations prior toan impact that mixed metal and silicate. However, thisseems unlikely as it would appear to require steepthermal gradients in the meteorites when they cooledthrough 350 1C after the Widmanstatten pattern formed.Group IAB irons are largely unshocked, so localizedimpact heating also appears unlikely and the explana-tion for the very diverse Ar–Ar ages is unclear.

7.2. Group IIE irons

IIE irons contain silicate inclusions of chondritic anddifferentiated composition (Wasson and Wang, 1986;Mittlefehldt et al., 1998; Bogard et al., 2000) as well asGe isotopic compositions uniquely light and variable(Luais, 2007). Because of their complexity, we list theIIE irons in Table 4 together with some of theircharacteristic properties. Four have heterogeneouslydistributed angular silicate inclusions 1–10 cm in sizebroadly chondritic in composition – mostly olivine andorthopyroxene with minor plagioclase and phosphates(Buchwald, 1975; Olsen et al., 1994; Casanova et al.,1995). Another five have 5–10 vol% globular silicateinclusions up to 1 cm in size that are gabbroic incomposition containing plagioclase (or glass), K-feld-spar, pyroxene with only trace amounts, if any, ofolivine (Wasserburg et al., 1968; Ikeda and Prinz, 1996;Ruzicka et al., 1999). In addition, crystals of nearly purepotassium feldspar up to 11 cm in length were found onthe exterior of Colomera (Wasserburg et al., 1968).

Mittlefehldt et al. (1998) suggest that IIE inclusionsshould be viewed as a sequence extending from primitiveto highly differentiated: (1) chondritic inclusions likethose in Netschaevo that contain chondrules, (2) clastslike Techado that lack chondrules and appear unmelted,(3) totally melted, silicate clasts like the one in Watsonthat lack metal but are roughly chondritic in composi-tion, (4) plagioclase–orthopyroxene–clinopyroxeneclasts that could represent basaltic partial melts asin Miles and Weekeroo Station, and (5) plagioclase–clinopyroxene clasts like those in Colomera, Kodaikanaland Elga.

Group IIE irons can also be divided into an old groupwith formation ages of 4.570.1Gyr and cosmic-rayexposure ages of 50–600Myr, and a young group(Netchaevo, Watson, Kodaikanal) with ages of3.770.1Gyr resulting from late impact heating andcosmic-ray exposure ages of 3–15Myr (Bogard et al.,2000; Table 3). These subgroups can also be distin-guished on the basis of their N and Ge isotopiccompositions (Mathew, 2000; Luais, 2007). However,

ARTICLE IN PRESS

Table 4. Group IIE irons and their properties.

Meteorite Ni (wt%)a Silicatesb Ages (Gyr)c Techniquec

Arlington 8.42 Absent

Barranca Blanca 8.07 Absent

Colomera 7.86 Differentiated 4.5 Ar–Ar, Rb–Sr

Elga 8.25 Differentiated

Garhi Yasin 8.3 Chondriticb

Kodaikanal 8.71 Differentiated 3.68 Pb–Pb, Rb–Sr

Leshan 9.5 Absent

Miles Differentiated 4.41 Ar–Ar

Netschaevo 8.6 Chondritic 3.74 Ar–Ar

Techado 8.88 Chondritic 4.49 Ar–Ar

Tobychan 7.82 Absent

Verkhne Dnieprovsk 8.6 Absent

Watson 001 Chondritic 3.68 Ar–Ar

Weekeroo Sta. 7.51 Differentiated 4.4–4.5 Ar–Ar, I–Xe, Rb–Sr

aWasson and Wang (1986).bBogard et al. (2000), Buchwald (1975).cBogard et al. (2000) for references.


there is no simple relationship between the nature of asilicate clast, its age, and the composition of theassociated metal. Oxygen isotopic compositions of thesilicate inclusions suggest that all silicate-bearing IIEirons could come from a single body, but high-precisionlaser fluorination analyses are needed to check this(Fig. 20).

Given the heterogeneous distribution of silicates inmany IIE irons, it is not surprising that some lacksilicates (Table 4). However, the composition of themetal in IIE irons is not especially unusual (there areseveral ungrouped irons that are compositionallysimilar) so the classification of these silicate-free IIEirons is less secure.

Although there are many radiometric ages for IIEirons, there are no modern determinations of metallo-graphic cooling rates to constrain parent body size orburial depths. However, kamacite bandwidth variationssuggest that IIE irons cooled at diverse rates and burialdepths, not in a core. The uniquely high concentrationof IIE members impact heated during the Late HeavyBombardment at �3.8Gyr suggests that the IIE parentbody was large at that time, conceivably as large asVesta (Bogard, 1995).

Group IIE was excluded in the discussion in Section 3of chemical variations due to crystallization because theIIE irons show complex patterns and their analyticaldata are somewhat incomplete. Although As, Au,and Ni vary systematically among IIE irons in asimilar fashion to their behavior in IIIAB irons, Wassonand Wang (1986) argued against a fractional crystal-lization origin for IIE as the variations of W and Irwith Ni are much smaller than those in IIIAB andresemble those in IAB irons. They concluded that IIEirons formed in individual pools of impact-produced

melt, consistent with the diverse cooling rates discussedabove.

The remarkable diversity of properties has lead to awide range of models for the formation of IIE irons, andtwo ungrouped irons, Guin and Sombrerete, which havesimilar silicate inclusions. These include (1) an inter-mediate stage in the gravitational separation of metaland silicates before the core had formed (Wasserburg etal., 1968), (2) impact mixing of materials from a singlebody that was undergoing igneous differentiation(Bogard et al., 2000; Ruzicka et al., 2006), (3) near-surface impact melting of a chondritic body (Wassonand Wang, 1986; Rubin et al., 1986), and (4) impactmixtures from collisions between metallic projectileswith a target of H chondrite affinity (Ruzicka et al.,1999). Given the increasing evidence for impacts duringdifferentiation caused by 26Al heating, early formationof IIE irons by impact mixing of materials from a singlepartly melted body appears most plausible. However, itis possible that impacts mixed materials from severalasteroid-sized bodies or that protoplanetary collisionswere involved.

7.3. Group IIIAB irons

This group has chemical trends consistent with aS-rich molten metallic core in a differentiated body thatwas undisturbed when it fractionally crystallizedand cooled. However, the range of cooling rates of�60–300 1C/Myr in IIIAB irons exceeds the error limitsof 2s ¼ 2.0 (Fig. 15a). After allowance for the change inkamacite nucleation mechanism that causes a step in thecooling rate vs. bulk Ni curve, the IIIAB irons appear toshow an inverse correlation between bulk Ni and

ARTICLE IN PRESS

1,800

1,600

1,400

1,200

1,000

800

600

400

20010 100 1,000 10,000 100,000

Cooling rate (K/Myr)

Tem

pera

ture

(K)

Center

0.4R0.7R

0.8R 0.9R0.97R

Twp

Tcz

100000

10000

1000

100

107 8 9 10 11

Ni content (wt. %)

Coo

ling

rate

(°C

/Myr

)

9.0 wt% S

3.0 wt% S

Fig. 22. Models for crystallization and cooling of a 150 km


cooling rate (Yang and Goldstein, 2006), similar to theIVA group though with a smaller range. The most likelyexplanation is that the body in which the IIIAB ironscooled had a small silicate mantle that was thicker thanthe IVA mantle (o1 km), but much thinner than thetens of kilometers expected for an undisturbed differ-entiated body. Detailed thermal models are needed toestablish the dimensions of the core and silicate mantle.These data will help to constrain the nature of theimpact – whether asteroidal or protoplanetary – andwhether the core crystallized before or after the impact.The antiquity of the phosphate in IIIAB irons – only4–5Myr younger than CAIs (Fig. 19) – and its lowtemperature of isotopic closure (�750 1C) imply that thecore crystallized only a few Myr after CAIs formed.Such rapid core crystallization would be aided by animpact that removed all but a few kilometers of silicatefrom a molten metallic body, perhaps several tens ofkilometers in radius.

One IIIAB iron, Puente del Zacate, contains a single7mm wide silicate clast in a troilite nodule. Mostsurprisingly, these silicate minerals resemble those ingroup IAB irons in that they are a chondritic mixture ofolivine, pyroxene, and plagioclase mixed with graphite(Olsen et al., 1996), and their O isotopic compositionclosely matches that of the phosphates and chromites inIIIAB (Fig. 21). How a small chondritic fragment couldbe added to core material during an impact is notknown. However, its presence in sulfide suggests that itwas trapped in residual melt and was therefore added tothe liquid prior to rapid crystallization of the IIIABmetal, as inferred above.

radius metallic body to explain the thermal histories and bulk

Ni of IVA irons (Yang et al., 2007, 2008a). (a) Calculated

cooling rates as a function of temperature for various radial

depths (R is radius) provide a good match for the observed

range of 100–6600 1C/Myr during the growth of the Widman-

statten pattern at TWP as well as the 15-fold cooling rate range

during the formation of the cloudy taenite intergrowth at TCZ.

(b) Comparison between calculated cooling rates for IVA irons

(Fig. 20b) with two theoretical curves derived from the thermal

model in (a) and a fractional crystallization model for the Ni

distribution assuming an initial bulk S content for the IVA

body of 3 and 9wt%.

7.4. Group IVA irons

According to conventional models, the average cool-ing rate for IVA irons of 200–1500 1C/Myr implies thatthey are derived from the core of a differentiated bodyof radius 10–20 km (Chabot and Haack, 2006). How-ever, Yang et al. (2007, 2008a) rejected this model as thecooling rates of 13 IVA irons at 650–500 1C decreasedsystematically from 6600 to 100 1C/Myr with increasingbulk Ni (Fig. 15) and measurements of the size of thehigh-Ni particle size in the cloudy zone showed thatcooling rates at 300 1C decreased with bulk Ni by afactor of �15. In addition, a mantled core that cooled at3000 1C/Myr (a lower limit for the fast cooled, low-NiIVA irons) would require a differentiated body only4 km in radius, yet a body of this size would nothave been melted sufficiently by 26Al to form a core(Yang et al., 2007).

The only plausible explanation for the IVA coolingrates is that the irons did not cool in a mantled core or adisrupted and reaccreated body (Haack et al., 1996b),but in a large metallic body with scarcely any silicate

insulation. Yang et al. (2007, 2008a) calculated that ametallic body of 150750 km radius would accommo-date the ranges of cooling rates they inferred for IVAirons from both kamacite growth modeling and cloudytaenite particle sizes (Fig. 22a) and that it wassurrounded by o1 km of silicate material. Since impactscapable of removing virtually all of the mantle from asolid core would disrupt and disperse core material andpreclude any correlation between cooling rate andcomposition, Yang et al. (2007) proposed that the150 km radius body was molten following an impact.


A possible mechanism for making such a body wasproposed by Asphaug et al. (2006), who suggested thatthe parent bodies of iron and stony-iron meteorites wereformed in grazing impacts between protoplanets. Theirnumerical models suggested that such impacts couldconvert the central part of the projectile into a string ofmetal-rich bodies. Inwards fractional crystallization ofone of these molten metallic bodies would ensure thatthe early-formed low-Ni IVA irons would cool quicklynear the surface and that bulk Ni would be correlatedinversely with cooling rate.

Yang et al. (2008a) tested this model further by usingthe fractional crystallization model for IVA irons tocalculate how bulk Ni would vary with burial depth in a150 km radius metallic body that crystallized inwards.Since no single S concentration for group IVA satisfiesthe simple fractional crystallization model, values of 3and 9wt% were used (Figs. 4d and 5d). The variation ofNi with burial depth was then combined with thethermal model in Fig. 22a to infer how cooling ratevaried with bulk Ni. Fig. 22b shows that this modelingfor a 150 km radius metallic body provides a very goodmatch for the IVA cooling rates.

Two silicate-rich irons, Steinbach and Sao Joao Nepo-muceno, have metal compositions and cooling rates indi-cating that they come from the IVA body. How thesesilicate-rich irons could have formed has been a majormystery because the silicates, silica and pyroxene, shouldhave quickly floated up to the core–mantle boundary of adifferentiated body. A protoplanetary impact of the kindenvisaged by Asphaug et al. (2006) may provide an environ-ment for trapping silicates under an advancing solidificationfront, as well as producing the silicates from magmas withmore chondritic compositions (Ruzicka and Hutson, 2006).

A very different impact model for formation of IVAirons and stony-irons infers that they were produced byimpact melting of L/LL chondrite material near thesurface of an asteroid that generated an asteroidal core(Wasson et al., 2006). They argued that the inferredcooling rates were incorrect and that the irons cooled atthe same rate, contrary to Fig. 15b.

Group IVA, like IVB irons, have extraordinarily lowconcentrations of moderately volatile siderophile ele-ments. Fig. 2a shows that Ge/Ni ratios in these irons(and some ungrouped irons) are more than a factor of1000 lower than in CI chondrites. Given the evidence forimpacts during the formation of IVA irons, loss ofvolatiles like Ga, Ge, and S from molten iron may haveoccurred during asteroidal or protoplanetary impacts(Wasson et al., 2006; Yang et al., 2007).

7.5. Group IVB irons

The remarkable depletion of Ga and Ge in IVB irons(Fig. 2) is clearly a result of high-temperature metal–

vapor equilibration (Kelly and Larimer, 1977; Campbelland Humayun, 2005; Walker et al., 2008). Most authorshave invoked condensation and accretion at highnebular temperatures, rather than impact volatilization,as the mean group IVB composition matches thatpredicted by equilibrium condensation calculations.Depletion of W and some other less siderophile elementsmay reflect oxidation in the IVB body. Chemical trendssuggest that these irons come from a fractionallycrystallized core with o2wt% S. However, twocharacteristics of the cooling rate data are difficultto reconcile with crystallization and cooling in the coreof a differentiated body: the high cooling rates of103–4 1C/Myr (Rasmussen, 1989), which require adifferentiated body that was too small to be melted by26Al heating, and the diverse cooling rates that areincompatible with an insulated core. However, unlikegroup IVA, the IVB irons may show a positivecorrelation between cooling rate and bulk Ni. A pro-posed link between IVB irons and angrites (Campbelland Humayun, 2005) needs further testing.

7.6. Ungrouped irons

There does not appear to be a fundamental differencebetween the ungrouped irons and those in groups. It isunlikely, for example, that a significant number arechemically reprocessed members of other groups. Somesmall irons weighing o20 g could be impact-producedmetallic slugs from chondritic asteroids (Wasson, 2000).However, most of the larger ungrouped irons appear tohave experienced chemical fractionation processes likethose that produced the groups (Scott, 1979). Abun-dances of volatile siderophile elements like Ga, Ge, Cu,and As among both ungrouped and grouped irons tendto be correlated. Deviations of the ungrouped ironsfrom the overall As–Ga trend are inversely related totheir Ir concentrations, consistent with the inverse As–Irtrends and small Ga variations within fractionallycrystallized groups. This suggests that most ungroupedirons were fractionally crystallized, like most groupedirons.

The ungrouped irons contain sets of two to fourrelated irons that may well achieve group status as moreirons are discovered. The arbitrary minimum numberfor a group is set at five. In addition, some of theungrouped irons that Wasson and Kallemeyn (2002)include in their IAB complex may come from the samebody. The total number of parent bodies currentlysampled by the ungrouped irons is probably somewhathigher than earlier estimates of 40–50 (Scott, 1979;Wasson, 1990). Interestingly, the fraction of ungroupedirons recovered from Antarctica (�0.35) is much higherthan for non-Antarctic irons, even after excludingthe smallest samples (Wasson, 1990, 2000). The high


proportion of ungrouped Antarctic irons is pro-bably related to their much smaller mean mass(�100� smaller than for non-Antarctic ones). Smallermeteoroids may sample more diverse locations in theasteroid belt, because they are ejected with higherimpact velocities (Wasson, 1990) or because they driftslower into resonances.

7.7. Source of iron meteorites

Since the ungrouped and grouped irons probablycome from over 60 different asteroids, about two-thirdsof all the asteroids sampled by meteorites were meltedsufficiently to allow metal–silicate segregation. How-ever, this high proportion is not reflected in the spectralproperties of large intact asteroids or their fragments(Bottke et al., 2006). Types C and S asteroids, generallyconsidered to be chondritic, dominate over likelyigneous types such as A and V types, which arecomposed of olivine and basalt, respectively. (Note thatnearly all V types probably come from Vesta.) Thevirtual absence of stony and stony-iron meteorites thatare related to the irons likely reflects the much greaterstrength of metallic Fe–Ni than rock and its longerimpact lifetime (Burbine et al., 1996).

A possible explanation for the abundance of igneousmeteorites, the scarcity of igneous asteroids, and theearly formation of iron meteorite parent bodies is thatmelted planetesimals were formed at 1–2 AU from theSun and their fragments were scattered into the asteroidbelt by protoplanets (Bottke et al., 2006). Since mostirons cooled in 10Myr or less (rather than 100Myr ormore, as inferred �30 years ago), it becomes moreplausible that the iron meteorite parent bodies weredisrupted in the first few Myr inside 2 AU (Bottke et al.,2006). Disruption by grazing protoplanetary impactsrather than bombardment by smaller projectiles mighthelp to explain how many differentiated bodies weredisrupted while Vesta’s crust was preserved (Asphauget al., 2006). Note that disruption inside 2 AU may haveproduced bodies that were �1–10 km in size; disruptionduring the past 1Gyr into meter-sized fragmentsthat were exposed to cosmic rays occurred in theasteroid belt.

8. Summary

Contrary to traditional views about their origin,iron meteorites may have been derived originally frombodies as large as 1000 km or more in size. Most ironmeteorites come directly or indirectly from bodies thataccreted before the chondrites, possibly at 1–2 AUrather than in the asteroid belt. Many of these bodiesmay have been disrupted by impacts before they cooledslowly.

Chemical variations within the larger chemicalgroups, excluding IAB and IIE, are broadly consistentwith fractional crystallization implying that each che-mical group comes from a single molten metallic pool orcore. However, the groups differ considerably in bulkcomposition and crystallization conditions. Only thenearly S-free group IVB is well modeled by simplefractional crystallization of an isolated metallic meltusing experimentally determined partition coefficients.The presence of S complicates the crystallization historyof the other groups and there are unresolved issuesabout the scatter in chemical trends (IIIAB), liquidimmiscibility (IIAB), trapping of liquid (IIAB, IIIAB,IVA, IID), and formation of late stage members (IIIAB,IIAB). The IVA and IID groups have well-definedchemical trends but no crystallization model hasexplained both the Ir and Ge trends with a single bulkS content.

The irons in groups IAB and IIE, many of whichcontain silicates, show considerable scatter on mostbinary elemental plots; neither formed from a singleisolated metallic melt. Impacts were involved in theirformation but whether they generated molten metalpools or merely mixed preexisting molten metal withsilicate is not certain. The IAB silicate inclusions arecogenetic with the winonaites, which are stronglymetamorphosed chondrites. No other group of silicateor metal–silicate meteorites is strongly linked to anygroup of irons.

The cooling rates of irons determined from kamacitegrowth models have increased over time (in one case bya factor of 50) as models have become ever moresophisticated. Typical cooling rates for irons in theirparent asteroidal bodies at �500–700 1C are100–10,000 1C/Myr with cooling times of 10Myr orless. This increase in measured cooling rate is due largelyto the recognition of the importance of P in thenucleation and growth of the Widmanstatten pattern.

The highest quality cooling rates are those for groupsIIIAB, �60–300 1C/Myr, and IVA, �100–6600 1C/Myr.These variations greatly exceed a factor of 2, which isthe uncertainty of the cooling rate method. The widerange of cooling rates for IVA irons shows that thesemeteorites cooled not in a mantled core but a largemetallic body of radius 150750 km with scarcely anysilicate insulation. The correlation between bulk Ni andcooling rate requires inwards crystallization of a moltenmetallic body, which may have formed by a grazingimpact between protoplanets. Protoplanetary impactsmay have also been involved in the formation of IIIABirons; thermal histories of groups IIAB and IVB, need tobe re-investigated to see if they formed in a similar way.Impact volatilization and high-temperature nebularcondensation have been invoked to explain the deple-tions of Ga, Ge and other volatile siderophiles in groupsIVA and IVB, respectively.


After excluding ungrouped irons weighing o20 g thatmay be simple impact melts, silicate-bearing irons, andothers loosely linked to group IAB, we infer that theremaining ungrouped irons probably come from at least50 other bodies that formed in analogous ways to thefractionally crystallized groups. They have 182W/184Wratios showing they were derived originally from bodiesthat were melted by 26Al to form cores o1.5Myr afterCAI formation and before the formation of chondrites.These constraints, the evidence for rapid cooling ofmany irons in a few Myr, the lack of other kinds ofmeteorites related to the irons, and the small fraction ofdifferentiated bodies in the asteroid belt all suggest thatirons could be derived from fragments of planetesimalsor protoplanets that formed and broke up at 1–2 AUand were subsequently scattered into the asteroid belt byprotoplanets.

Acknowledgments

This work was supported by NASA GrantsNNG05GK84G and NNX08AG53G (J.I. Goldstein,P. I.), NNX08AE08G (K. Keil, P. I.) and NNG0-6G113G (N. L. Chabot, P. I.). We thank J.T. Wasson(UCLA) for kindly supplying the data for Fig. 2.Helpful reviews of this paper were given by J. Yang(UMass), A. Kracher (Iowa State), and H. Haack (Nat.Hist. Museum of Denmark). We thank Associate EditorKlaus Keil for soliciting this Invited Review.

References

Allen, N.P., Earley, C.C., 1950. The transformations a-g and

g-a in iron-rich binary iron-nickel alloys. J. Iron Steel

Inst. 166, 281–288.

Amelin, Y., Krot, A.N., Hutcheon, I.D., Ulyanov, A.A., 2002.

Lead isotopic ages of chondrules and calcium–aluminum-

rich inclusions. Science 297, 1678–1683.

Amelin, Y., Wadhwa, M., Lugmair, G., 2006. Pb-isotopic

dating of meteorites using 202Pb–205Pb double-spike:

comparison with other high-resolution chronometers.

Lunar Planet. Sci. 37, 1970.

Asphaug, E., Agnor, C.B., Williams, Q., 2006. Hit-and-run

planetary collisions. Nature 439, 155–160.

Benedix, G.K., McCoy, T.J., Keil, K., Bogard, D.D.,

Garrison, D.H., 1998. A petrologic and isotopic study of

winonaites: evidence for early partial melting, brecciation,

and metamorphism. Geochim. Cosmochim. Acta 62,

2535–2553.

Benedix, G.K., McCoy, T.J., Keil, K., Love, S.G., 2000. A

petrologic study of the IAB iron meteorites: constraints

on the formation of the IAB–Winonaite parent body.

Meteorit. Planet. Sci. 35, 1127–1141.

Benedix, G.K., Lauretta, D.S., McCoy, T.J., 2005. Thermo-

dynamic constraints on the formation conditions of

winonaites and silicate-bearing IAB irons. Geochim.

Cosmochim. Acta 69, 5123–5131.

Bevan, A.W.R., Grady, M.M., 1988. Mount Morris

(Wisconsin) – a fragment of the IAB iron Pine River?

Meteoritics 23, 349–352.

Bogard, D.D., 1995. Impact ages of meteorites: a synthesis.

Meteoritics 30, 244–268.

Bogard, D.D., Garrison, D.H., McCoy, T.J., 2000. Chronol-

ogy and petrology of silicates from IIE iron meteorites:

evidence of a complex parent body evolution. Geochim.


Bogard, D.D., Garrison, D.H., Takeda, H., 2005. Ar–Ar and

I–Xe ages and the thermal history of IAB meteorites.


Bottke, W.F., Nesvorny, D., Grimm, R.E., Morbidelli, A.,

O’Brien, D.P., 2006. Iron meteorites as remnants of

planetesimals formed in the terrestrial planet region.

Nature 439, 821–824.

Buchwald, V.F., 1975. Handbook of Iron Meteorites. Uni-

versity of California Press, Berkeley, CA.

Burbine, T.H., Meibom, A., Binzel, R.P., 1996. Mantle material

in the main belt: battered to bits? Meteoritics 31, 607–620.

Burkhardt, C., Kleine, T., Bourdon, B., Palme, H., Zipfel, J.,

Friedrich, J., Ebel, D., 2009. Hf–W mineral isochron for

Ca, Al rich inclusions: age of the solar system and the

timing of core formation in planetesimals. Geochim.


Campbell, A.J., Humayun, M., 2005. Compositions of group

IVB iron meteorites and their parent melt. Geochim.


Casanova, I., Graf, T., Marti, K., 1995. Discovery of an

unmelted H-chondrite inclusion in an iron meteorite.

Science 268, 540–542.

Chabot, N.L., 2004. Sulfur contents of the parental metallic

cores of magmatic iron meteorites. Geochim. Cosmochin.

Acta 68, 3607–3618.

Chabot, N.L., Drake, M.J., 1999. Crystallization of magmatic

iron meteorites: the role of mixing in the molten core.


Chabot, N.L., Drake, M.J., 2000. Crystallization of magmatic

iron meteorites: the effects of phosphorus and liquid

immiscibility. Meteorit. Planet. Sci. 35, 807–816.

Chabot, N.W., Haack, H., 2006. Evolution of asteroidal cores.

In: Lauretta, D.S., McSeeen, H.Y. (Eds.), Meteorites

and the Early Solar System II. University Arizona Press,

pp. 747–771.

Chabot, N.L., Jones, J.H., 2003. The parameterization of solid

metal–liquid metal partitioning of siderophile elements.


Chabot, N.L., Campbell, A.J., Jones, J.H., Humayun, M.,

Agee, C.B., 2003. An experimental test of Henry’s Law in

solid metal–liquid metal systems with implications for iron

meteorites. Meteorit. Planet. Sci. 38, 181–196.

Chabot, N.L., Campbell, A.J., Jones, J.H., Humayun, M.,

Lauer, H.V., 2006. The influence of carbon on partitioning

behavior during planetary evolution. Geochim. Cosmo-

chim. Acta 70, 1322–1335.

Choi, B.-G., Ouyang, X., Wasson, J.T., 1995. Classification

and origin of IAB and IIICD iron meteorites. Geochim.



Clarke Jr., R.S., Goldstein, J.I., 1978. Schreibersite growth

and its influence on the metallography of coarse-structured

iron meteorites. Smithsonian Contributions to the Earth

Sciences, vol. 21. Smithsonian Institution Press, Washing-

ton, DC.

Clayton, R.N., Mayeda, T.K., Olsen, E.J., Goswami, J.N.,

1991. Oxygen isotope studies of ordinary chondrites.

Geochim. Cosmochim. Acta 55, 2317–2337.

Clayton, R.N., Mayeda, T.K., 1996. Oxygen isotope studies of

achondrites. Geochim. Cosmochim. Acta 60, 1999–2017.

Cook, D.L., Walker, R.J., Horan, M.F., Wasson, J.T.,

Morgan, J.W., 2004. Pt–Re–Os systematics of group IIAB

and IIIAB iron meteorites. Geochim. Cosmochim. Acta 68,

1413–1431.

Corrigan, C.M., McCoy, T.J., Chabot, N.L., McDonough,

W., 2006. Trace element partitioning in the Fe–Ni–P

system: applications to P-rich iron meteorites. Lunar

Planet. Sci. 37 (Abstract no. 2314).

Corrigan, C.M., Chabot, N.L., McCoy, T.J., McDonough,

W.F., Watson, H.C., Saslow, S.A., Ash, R.D., 2009. The

iron-nickel-phosphorus system: effects on the distribution

of trace elements during the evolution of iron meteorites.

Geochim. Cosmochim. Acta, in press.

Dean, D.C., Goldstein, J.I., 1986. Determination of the

interdiffusion coefficients in the Fe–Ni and Fe–Ni–P

systems below 900 1C. Metall. Trans. 17A, 1131–1138.

Doan, A.S., Goldstein, J.I., 1970. The ternary phase diagram,

Fe–Ni–P. Metall. Trans. 1, 1759–1767.

Esbensen, K.H., Buchwald, V.F., Malvin, D.J., Wasson, J.T.,

1982. Systematic compositional variations in the Cape York

iron meteorite. Geochim. Cosmochim. Acta 46, 1913–1920.

Folco, L., Bland, P.A., D’Orazio, M., Franchi, I.A., Kelley,

S.P., Rocchi, S., 2004. Extensive impact melting on the H-

chondrite parent asteroid during the cataclysmic bombard-

ment of the early solar system: evidence from the

achondritic meteorite Dar al Gani 896. Geochim. Cosmo-

chim. Acta 68, 2379–2397.

Franchi, I., 2008. Oxygen isotopes in asteroidal materials. Rev.

Miner. 68, 345–397.

Gaffey, M.J., Gilbert, S.L., 1998. Asteroid 6 Hebe: The

probable parent body of the H-type ordinary chondrites

and the IIE iron meteorites. Meteorit. Planet. Sci. 33,

1281–1295.

Gilmour, J.D., Pravdivtseva, O.V., Busfield, A., Hohenberg,

C.M., 2006. The I-Xe chronometer and the early solar

system. Meteoritics 41, 19–31.

Goldberg, E., Uchiyama, A., Brown, H., 1951. The distribu-

tion of nickel, cobalt, gallium, palladium, and gold in iron

meteorites. Geochim. Cosmochim. Acta 2, 1–25.

Goldstein, J.I., Ogilvie, R.E., 1965. The growth of the

Widmanstatten pattern in metallic meteorites. Geochim.


Goldstein, J.I., Short, J.M., 1967. The iron meteorites, their

thermal history and parent bodies. Geochim. Cosmochim.

Acta 31, 1733–1770.

Goldstein, J.I., Doan, A.S., 1972. The effect of phosphorus on

the formation of the Widmanstatten pattern in iron


Goldstein, J.I., Michael, J.R., 2006. The formation of plessite

in meteoritic metal. Meteorit. Planet. Sci. 41, 553–570.

Goldstein, J.I., Yang, J., Kotula, P., Michael, J.R., Scott,

E.R.D., 2009. Thermal histories of fast-cooled iron

meteorites from zoned taenite microstructures using trans-

mission electron microscopy. Meteorit. Planet. Sci., in

press.

Greenwood, R.C., Franchi, I.A., Jambon, A., Buchanan, P.C.,

2005. Widespread magma oceans on asteroidal bodies in

the early Solar System. Nature 435, 916–918.

Greenwood, R.C., Franchi, I.A., Jambon, A., Barrat, J.A.,

Burbine, T.H., 2006. Oxygen isotope variation in stony-

iron meteorites. Science 313, 1763–1765.

Haack, H., McCoy, T.J., 2004. Iron and stony-iron meteorites.

In: Davis, A.M., Holland, H.D., Turekian, K.I. (Eds.),

Meteorites, Comets, and Planets, Vol. 1. Treatise on

Geochemistry. Elsevier-Pergamon, Oxford, pp. 325–345.

Haack, H., Scott, E.R.D., 1993. Chemical fractionations in

Group IIIAB iron meteorites: origin by dendritic crystal-

lization of an asteroidal core. Geochim. Cosmochim. Acta

57, 3457–3472.

Haack, H., Scott, E.R.D., Rasmussen, K.L., 1996a. Thermal

and shock history of mesosiderites and their large parent

asteroid. Geochim. Cosmochim. Acta 60, 2609–2619.

Haack, H., Scott, E.R.D., Love, S.G., Brearley, A.J., McCoy,

T.J., 1996b. Thermal histories of IVA stony-iron and iron

meteorites: evidence for asteroid fragmentation and reac-

cretion. Geochim. Cosmochim. Acta 60, 3103–3113.

Herpfer, M.A., Larimer, J.W., Goldstein, J.I., 1994. A

comparison of metallographic cooling rate methods used

in meteorites. Geochim. Cosmochim. Acta 58, 1353–1365.

Hevey, P.J., Sanders, I.S., 2006. A model for planetesimal

meltdown by 26Al and its implications for meteorite parent

bodies. Meteorit. Planet. Sci. 41, 95–106.

Hopfe, W.D., Goldstein, J.I., 2001. The metallographic

cooling rate method revised: application to iron meteorites

and mesosiderites. Meteorit. Planet. Sci. 36, 135–154.

Humayun, M., Simon, S.B., Grossman, L., 2007. Tungsten

and hafnium distribution in calcium-aluminum inclusions

(CAIs) from Allende and Efremovka. Geochim. Cosmo-

chim. Acta 71, 4609–4627.

Ikeda, Y., Prinz, M., 1996. Petrology of silicate inclusions in

the Miles IIE iron. Proc. NIPR Symp. Antarctic Meteorites

9, 143–173.

Jones, J.H., Drake, M.J., 1983. Experimental investigations of

trace element fractionation in iron meteorites, II: The

influence of sulfur. Geochim. Cosmochim. Acta 47,

1199–1209.

Kelly, W.R., Larimer, J.W., 1977. Chemical fractionations in

meteorites. VIII – Iron meteorites and the cosmochemical

history of the metal phase. Geochim. Cosmochim. Acta 41,

93–111.

Kita, N.T., Huss, G.R., Tachibana, S., Amelin, Y., Nyquist,

L.E., Hutcheon, I.D., 2005. Constraints on the origin of

chondrules and CAIs from short-lived and long-lived

radionuclides. In: Krot, A.N., Scott, E.R.D., Reipurth, B.

(Eds.), Chondrites and the Protoplanetary Disk, ASP

Conference Series, vol. 341. Astronomical Society of the

Pacific, San Francisco, pp. 558–587.

Kleine, T., Mezger, K., Palme, H., Scherer, E., Munker, C.,2005. Early core formation in asteroids and late accretion

of chondrite parent bodies: evidence from 182Hf–182W in


CAIs, metal-rich chondrites, and iron meteorites. Geochim.


Kracher, A., 1985. The evolution of partially differentiated

planetesimals: evidence from iron meteorite groups IAB

and IIICD. J. Geophys. Res. 90 (suppl.), C689–C698.

Kracher, A., Wasson, J.T., 1982. The role of S in the evolution

of the parental cores of the iron meteorites. Geochim.


Kracher, A., Willis, J., Wasson, J.T., 1980. Chemical

classification of iron meteorites—IX. A new group (IIF),

revision of IAB and IIICD, and data on 57 additional irons.


Krot, A.N., Keil, K., Scott, E.R.D., Goodrich, C.A., Weis-

berg, M.K., 2008. Classification of meteorites. In: Davis,

A.M. (Ed.), Meteorites, Comets and Planets, Treatise on

Geochemistry Update 1. Elsevier, Amsterdam, pp. 1–52.

Lovering, J.F., Nichiporuk, W., Chodos, A., Brown, H., 1957.

The distribution of gallium, germanium, cobalt, chromium,

and copper in iron and stony-iron meteorites in relation to

nickel content and structure. Geochim. Cosmochim. Acta

11, 263–278.

Luais, B., 2007. Isotopic fractionation of germanium in iron

meteorites: significance for nebular condensation, core

formation and impact processes. Earth Planet. Sci. Lett.

262, 21–36.

Malvin, D.J., 1988. Assimilation-fractional crystallization

modeling of magmatic iron meteorites (abstract). In: Lunar

and Planetary Science XIX. Lunar and Planetary Institute,

Houston, pp. 720–721.

Malvin, D.J., Wang, D., Wasson, J.T., 1984. Chemical

classification of iron meteorites, X—Multielement studies

of 43 irons, resolution of group IIIE from IIIAB, and

evaluation of Cu as a taxonomic parameter. Geochim.


Markowski, A., Quitte, G., Halliday, A.N., Kleine, T., 2006a.

Tungsten isotopic compositions of iron meteorites: chron-

ological constraints vs. cosmogenic effects. Earth Planet.

Sci. Lett. 242, 1–15.

Markowski, A., Leya, I., Quitte, G., Ammon, K., Halliday,

A.N., Wieler, R., 2006b. Correlated helium-3 and tungsten

isotopes in iron meteorites: quantitative cosmogenic correc-

tions and planetesimal formation times. Earth Planet. Sci.

Lett. 250, 104–115.

Mathew, K.J., 2000. Isotopic signatures and origin of nitrogen

in IIE and IVA iron meteorites. Geochim. Cosmochim.

Acta 64, 545–557.

McCoy, T.J., Keil, K., Scott, E.R.D., Haack, H., 1993.

Genesis of the IIICD iron meteorites: evidence from

silicate-bearing inclusions. Meteoritics 28, 552–560.

Mittlefehldt, D.W., McCoy, T.J., Goodrich, C.A., Kracher,

A., 1998. Non-chondritic meteorites from asteroidal bodies.

Rev. Miner. 36, 4-1–4-195.

Narayan, C., Goldstein, J.I., 1984a. Nucleation of intragra-

nular ferrite in Fe–Ni–P alloys. Metall. Trans. 15A,

861–865.

Narayan, C., Goldstein, J.I., 1984b. Growth of intragranular

ferrite in Fe–Ni–P alloys. Metall. Trans. 15A, 867–874.

Olsen, E., Davis, A., Clarke Jr., R.S., Schultz, L., Weber,

H.W., 9 other coauthors, 1994. Watson: a new link in the

IIE chain. Meteoritics 29, 200–213.

Olsen, E.J., Davis, A.M., Clayton, R.N., Mayeda, T.K.,

Moore, C.B., Steele, I.M., 1996. A silicate inclusion in the

Puente del Zacate IIIA iron meteorite. Science 273,

1365–1367.

Owen, E.A., Burns, B.D., 1939. X-ray study of some meteoritic

irons. Philos. Mag. 28, 497–512.

Powell, B.N., 1969. Petrology and chemistry of mesosider-

ites—I. Textures and composition of nickel-iron. Geochim.


Qin, L., Dauphas, N., Wadhwa, M., Masarik, J., Janney, P.E.,

2008. Rapid accretion and differentiation of iron meteorite

parent bodies inferred from 182Hf–182W chronometry and

thermal modeling. Earth Planet. Sci. Lett. 273, 94–104.

Raghavan, V., 1988. Phase Diagrams of Ternary Iron Alloys,

vol. 2. Indian National Scientific Documentation Centre,

New Delhi, India, 360pp.

Randich, E., Goldstein, J.I., 1978. Cooling rates of seven

hexahedrites. Geochim. Cosmochim. Acta 42, 221–233.

Rasmussen, K.L., 1981. The cooling rates of iron meteorites –

a new approach. Icarus 45, 564–576.

Rasmussen, K.L., 1982. Determination of the cooling rates

and nucleation histories of eight group IVA iron meteorites

using local bulk Ni and P variation. Icarus 52, 444–453.

Rasmussen, K.L., 1989. Cooling rates and parent bodies of

iron meteorites from Group IIICD, IAB, and IVB. Phys.

Scripta 39, 410–416.

Rasmussen, K.L., Malvin, D.J., Buchwald, V.F., Wasson,

J.T., 1984. Compositional trends and cooling rates of group

IVB iron meteorites. Geochim. Cosmochim. Acta 48,

805–813.

Rasmussen, K.L., Ulff-Moller, F., Haack, H., 1995. The

thermal evolution of IVA iron meteorites: evidence from

metallographic cooling rates. Geochim. Cosmochim. Acta

59, 3049–3059.

Reisener, R.J., Goldstein, J.I., 2003. Ordinary chondrite

metallography: Part 1. Fe–Ni taenite cooling experiments.


Reuter, K.B., Williams, D.B., Goldstein, J.I., 1988. Low

temperature phase transformations in the metallic phases of

iron and stony-iron meteorites. Geochim. Cosmochim.

Acta 52, 617–626.

Romig, A.D., Goldstein, J.I., 1980. Determination of the

Fe–Ni and Fe–Ni–P phase diagrams at low temperatures

(700–300 1C). Metall. Trans. 11A, 1151–1159.

Rubie, D.C., Melosh, H.J., Reid, J.E., Liebske, C., Righter,

K., 2003. Mechanisms of metal-silicate equilibration in the

terrestrial magma ocean. Earth Planet. Sci. Lett. 205,

239–255.

Rubin, A.E., Jerde, E.A., Zong, P., Wasson, J.T., Westcott,

J.W., Mayeda, T.K., Clayton, R.N., 1986. Properties

of the Guin ungrouped iron meteorite – the origin of Guin

and of group-IIE irons. Earth Planet. Sci. Lett. 76,

209–226.

Ruzicka, A., Fowler, G.W., Snyder, G.A., Prinz, M., Papike,

J.J., Taylor, L.A., 1999. Petrogenesis of silicate inclusions

in the Weekeroo Station IIE iron meteorite: differentiation,

remelting, and dynamic mixing. Geochim. Cosmochim.

Acta 63, 2123–2143.

Ruzicka, A., Hutson, M., 2006. Differentiation and evolution

of the IVA iron meteorite parent body: clues from


pyroxene-geochemistry in the Steinbach stony-iron

meteorite. Meteorit. Planet. Sci. 41, 1959–1987.

Ruzicka, A., Hutson, M., Floss, C., 2006. Petrology of silicate

inclusions in the Sombrerete ungrouped iron meteorite:

implications for the origin of IIE-type silicate bearing irons.


Saikumar, V., Goldstein, J.I., 1988. An evaluation of the

methods to determine the cooling rates of iron meteorites.

Geochim. Cosmochim. Acta 52, 7l5–726.

Sanders, I.S., Taylor, G.J., 2005. Implications of 26Al in

nebular dust: formation of chondrules by disruption of

molten planetesimals. In: Krot, A.N., Scott, E.R.D.,

Reipurth, B. (Eds.), Chondrites and the Protoplanetary

Disk, ASP Conference Series, vol. 341. Astronomical

Society of the Pacific, San Francisco, pp. 915–932.

Schersten, A., Elliott, T., Hawkesworth, C., Russell, S.,

Masarik, J., 2006. Hf–W evidence for rapid differentiation

of iron meteorite parent bodies. Earth Planet. Sci. Lett. 241,

530–542.

Scott, E.R.D., 1972. Chemical fractionation in iron meteorites

and its interpretation. Geochim. Cosmochim. Acta 36,

1205–1236.

Scott, E.R.D., 1973. The nature of dark-etching rims in

meteoritic taenite. Geochim. Cosmochim. Acta 37,

2283–2294.

Scott, E.R.D., 1977a. Composition, mineralogy, and origin of

group IC iron meteorites. Earth Planet. Sci. Lett. 37,

273–284.

Scott, E.R.D., 1977b. Geochemical relationships between

some pallasites and iron meteorites. Miner. Mag. 41,

265–272.

Scott, E.R.D., 1979. Origin of anomalous iron meteorites.

Miner. Mag. 43, 415–421.

Scott, E.R.D., 2007. Chondrites and the protoplanetary disk.

Ann. Rev. Earth Planet. Sci. 35, 577–620.

Scott, E.R.D., Sanders, I.S., 2009. Implications of the

carbonaceous chondrite Mn–Cr isochron for the formation

of early refractory planetisimals and chondrules. Geochim.

Cosmochim. Acta, in press.

Scott, E.R.D., Wasson, J.T., 1975. Classification and proper-

ties of iron meteorites. Rev. Geophys. Space Phys. 13,

527–546.

Scott, E.R.D., Wasson, J.T., 1976. Chemical classification of

iron meteorites—VIII. Groups IC, IIE, IIF and 97 other

irons. Geochim. Cosmochim. Acta 40, 103–115.

Short, J.J., Goldstein, J.I., 1967. Rapid methods of determin-

ing cooling rates of iron and stony iron meteorites. Science

156, 59–61.

Shukolyukov, A., Lugmair, G.W., 2006. Manganese–chro-

mium isotope systematics of carbonaceous chondrites.

Earth Planet. Sci. Lett. 250, 200–213.

Smoliar, M.I., Walker, R.J., Morgan, J.W., 1996. Re–Os ages

of group IIA, IIIA, IVA, and IVB iron meteorites. Science

271, 1099–1102.

Sugiura, N., Hoshino, H., 2003. Mn–Cr chronology of

five IIIAB iron meteorites. Meteorit. Planet. Sci. 38,

117–143.

Stewart, B., Papanastassiou, D.A., Wasserburg, G.W., 1996.

Sm-Nd systematics of a silicate inclusion in the Caddo IAB

iron meteorite. Earth Planet Sci. Lett. 143, 1–12.

Takeda, H., Bogard, D.D., Mittlefehldt, D.W., Garrison,

D.H., 2000. Mineralogy, petrology, chemistry, and39Ar–40Ar and exposure ages of the Caddo County IAB

iron: evidence for early partial melt segregation of a gabbro

area rich in plagioclase–diopside. Geochim. Cosmochim.

Acta 64, 1311–1327.

Thrane, K., Bizzarro, M., Baker, J.A., 2006. Extremely brief

formation interval for refractory inclusions and uniform

distribution of 26Al in the early solar system. Astrophys.

J. 646, L159–L162.

Ulff-Møller, F., 1998. Effects of liquid immiscibility on trace

element fractionation in magmatic iron meteorites: a case

study of group IIIAB. Meteorit. Planet. Sci. 33, 207–220.

Vogel, N., Renne, P.R., 2008. 40Ar–39Ar dating of plagioclase

grain size separates from silicate inclusions in IAB iron

meteorites and implications for the thermochronological

evolution of the IAB parent body. Geochim. Cosmochim.

Acta 72, 1231–1255.

Wadhwa, M., 2007. Long-lived chronometers. In: Davis, A.M.

(Ed.), Meteorites, Comets and Planets, Treatise on Geo-

chemistry Update 1. Elsevier, Amsterdam, pp. 1–25.

Walker, R.J., McDonough, W.F., Honesto, J., Chabot, N.L.,

McCoy, T.J., Ash, R.D., Bellucci, J.J., 2008. Modeling

fractional crystallization of group IVB iron meteorites.


Wang, P.-L., Rumble, D., McCoy, T.J., 2004. Oxygen isotopic

composition of IVA iron meteorites: implications for the

thermal evolution derived from in situ ultraviolet laser

microprobe analysis. Geochim. Cosmochim. Acta 68,

1159–1171.

Wasserburg, G.J., Sanz, H.G., Bence, A.E., 1968. Potassium

feldspar phenocrysts in the surface of the Colomera, an iron

meteorite. Science 161, 684–687.

Wasson, J.T., 1967. The chemical classification of iron

meteorites, I. A study of iron meteorites with low

concentrations of gallium and germanium. Geochim.


Wasson, J.T., 1969. The chemical classification of iron

meteorites—III. Hexahedrites and other irons with germa-

nium concentrations between 80 and 200 ppm. Geochim.


Wasson, J.T., 1985. Meteorites: Their Record of Early Solar-

System History. W.H. Freeman, 267pp.

Wasson, J.T., 1990. Ungrouped iron meteorites in Antarctica:

origin of anomalously high abundance. Science 249,

900–902.

Wasson, J.T., 1999. Trapped melt in IIIAB irons; solid/liquid

elemental partitioning during the fractionation of the IIIAB

magma. Geochim. Cosmochim. Acta 63, 2875–2889.

Wasson, J.T., 2000. Iron meteorites from Antarctica: more

specimens, still 40% ungrouped. Workshop on Desert

Meteorites. Lunar & Planetary Science Institute, Contribu-

tion, No. 997, pp. 76–80.

Wasson, J.T., Choi, B.-G., 2003. Main-group pallasites-

chemical composition, relationship to IIIAB irons, and

origin. Geochim. Cosmochim. Acta 67, 3079–3096.

Wasson, J.T., Huber, H., 2006. Compositional trends among

IID irons; their possible formation from the P-rich lower

magma in a two-layer core. Geochim. Cosmochim. Acta 70,

6153–6167.


Wasson, J.T., Kallemeyn, G.W., 2002. The IAB iron–meteor-

ite complex: a group, five subgroups, numerous grouplets,

closely related, mainly formed by crystal segregation in

rapidly cooling melts. Geochim. Cosmochim. Acta 66,

2445–2473.

Wasson, J.T., Richardson, J.W., 2001. Fractionation trends

among IVA iron meteorites: contrasts with IIIAB trends.


Wasson, J.T., Wang, J., 1986. A nonmagmatic origin of

group-IIE iron meteorites. Geochim. Cosmochim. Acta 50,

725–732.

Wasson, J.T., Choi, B.-G., Jerde, E.A., Ulff-Moller, F., 1998.

Chemical classification of iron meteorites XII. New

members of the magmatic groups. Geochim. Cosmochim.

Acta 62, 715–724.

Wasson, J.T., Huber, H., Malvin, D.J., 2007. Formation of

IIAB iron meteorites. Geochim. Cosmochim. Acta 71,

760–781.

Wasson, J.T., Matsunami, Y., Rubin, A.E., 2006. Silica

and pyroxene in IVA irons; possible formation of

the IVA magma by impact melting and reduction of

L–LL-chondrite materials followed by crystallization and

cooling. Geochim. Cosmochim. Acta 70, 3149–3172.

Weisberg, M.K., McCoy, T.J., Krot, A.N., 2006. Systematics

and evaluation of meteorite classification. In: Lauretta,

D.S., McSween, Jr., H.Y. (Eds.), Meteorites and the Early

Solar System II. University of Arizona Press, pp. 19–52.

Willis, J., Goldstein, J.I., 1982. The effects of C, P, and S

on trace element partitioning during solidification in

Fe–Ni alloys. Proc. Lunar Planet. Sci. Conf. 13th. Part I.

J. Geophys. Res. 87 (suppl.), A435–A445.

Wood, J.A., 1964. The cooling rates and parent planets of

several iron meteorites. Icarus 3, 429–459.

Yang, C.-W., Williams, D.B., Goldstein, J.I., 1996. A revision

of the Fe–Ni phase diagram at low temperature. J. Phase

Equil. 17, 522–531.

Yang, C.-W., Williams, D.B., Goldstein, J.I., 1997a. Low-

temperature phase decomposition in metal from iron,

stony-iron, and stony meteorites. Geochim. Cosmochim.

Acta 61, 2943–2956.

Yang, C.-W., Williams, D.B., Goldstein, J.I., 1997b. A new

empirical cooling rate indicator for meteorites based on the

size of the cloudy zone of the metallic phases. Meteoritics

32, 423–429.

Yang, J., Goldstein, J.I., 2005. The formation of the

Widmanstatten structure in meteorites. Meteorit. Planet.

Sci. 40, 239–253.

Yang, J., Goldstein, J.I., 2006. Metallographic cooling rates of

the IIIAB iron meteorites. Geochim. Cosmochim. Acta 70,

3197–3215.

Yang, J., Goldstein, J.I., Scott, E.R.D., 2007. Iron meteorite

evidence for early formation and catastrophic disruption of

protoplanets. Nature 446, 888–891.

Yang, J., Goldstein, J.I., Scott, E.R.D., 2008a. Metallographic

cooling rates of IVA iron meteorites. Geochim. Cosmo-

chim. Acta 72, 3043–3061.

Yang, J., Goldstein, J.I., Scott, E.R.D., 2008b. Thermal

history and origin of the main group pallasites. Lunar

Planet. Sci. 39 (Abstract no.1938).

Zhang, J., Williams, D.B., Goldstein, J.I., 1993. The micro-

structure and formation of duplex and black plessite in iron


Iron meteorites Crystallization, thermal history, parent ...escott/Goldstein ea chem review.pdf ·...

Documents

Transcript of Iron meteorites Crystallization, thermal history, parent ...escott/Goldstein ea chem review.pdf ·...