Introduction

There are two basic kinds of meteorite: chondrites and differentiated, or non-chondritic, meteorites (Hutchison, 2004; Krot, Keil, Scott, Goodrich, & Weisberg, 2014; Norton, 2002; also see “Meteorites”). Chondrites have bulk chemical compositions like that of the Sun, neglecting the highly volatile elements like hydrogen, carbon, and nitrogen. They formed in the protoplanetary disk, or solar nebula, by accretion of grains of metallic iron-nickel, sulfide, silicate, once molten, millimeter-sized spherules called chondrules, and refractory materials, including calcium- and aluminum-rich inclusions (CAIs). Differentiated meteorites formed from chondritic planetesimals that melted, enabling sulfur-rich metal to sink forming metallic cores surrounded by silicate mantles and crusts. Iron meteorites are thought to be mostly derived from cores of differentiated planetesimals: a few likely formed from smaller metallic pools in the mantle or crust.

Chondrites are divided into three major classes called enstatite, ordinary, and carbonaceous, which are subdivided into 13 groups. The CI group of carbonaceous chondrites provides the best match for solar abundances (Lodders, 2020). There are also around 13 iron meteorite groups, but with some minor possible exceptions, they are not closely related to the chondrite groups nor to the meteorites derived from silicate mantles and crusts of once-molten bodies, which are called achondrites.

Iron meteorites consist almost entirely of siderophile (iron-loving) elements like nickel and cobalt, which readily dissolve in metallic iron, and chalcophile (sulfur-loving) elements like copper and silver. Most irons contain ~90% iron, 5–10% nickel, 0.4–0.6% cobalt, 0.1–0.5% phosphorus, 0.1–2 wt.% sulfur, and trace amounts of numerous other siderophile and chalcophile elements. Table 1 gives typical values and concentration ranges for 29 elements in irons (Buchwald, 1975; Scott, 1972). These were calculated after excluding silicate and silica inclusions, which are present in ~5% of irons, mostly in groups IAB and IIE (Buchwald, 1975; Ruzicka, 2014).

Iron meteorites account for only 4–5% of meteorite falls (Buchwald, 1975; Krot et al., 2014) but are more abundant among finds in certain areas, for example, around terrestrial impact craters <1 km in diameter (Buchwald, 1975) and on Mars (e.g., Ashley et al., 2012). The 12 largest individual meteorites each weigh 10 to 60 tons and are all irons (Meteoritical Bulletin Database). Figure 1 shows the Bacubirito iron, which weighs 19 tons. The largest stony meteorite weighs only 3 tons as stones are weaker than irons and break up more easily during atmospheric entry. However, several of the largest irons also broke up on entry, including the one that created Meteor Crater in Arizona, also called Barringer Crater (Figure 2).

Open in new tab

Figure 1. The Bacubirito iron meteorite was found in Mexico, weighs 19 tons, and is 4.1 m long (Terán-Bobadilla et al., 2017). It is an ungrouped iron. Like all meteorites, it is named after the location where it was found or fell. Meteorites from densely populated collection areas are also numbered.

Image from Buchwald (1975).

Open in new tab

Figure 2. Meteor Crater, Arizona, also called Barringer Meteorite Crater, has a diameter of 1,200 m; the floor is ~130 m below the surrounding plain and the rim is 40–65 m above the plain. The crater formed 50,000 years ago when a metallic body ~50–60 m in diameter hit the Earth (Artemieva & Pierazzo, 2011). The rectangular shape reflects preexisting faulting (Poelchau, Kenkmann, & Kring, 2009). Around 30 tons of meteorites weighing ~50 g to 639 kg were recovered up to 8 km away, and many thousands of tons of metal were dispersed as molten droplets (Buchwald, 1975). The irons known as Canyon Diablo contain 7–8.2 wt.% nickel and belong to group IAB, the second largest group.

Image from NASA Goddard Spaceflight Center.

Iron meteorites have been studied by a wide variety of scientists, including chemists, physicists, metallurgists, material scientists, geophysicists, astronomers, and archeologists. Planetary scientists who study iron meteorites typically focus on understanding how metallic cores and pools formed in differentiated asteroids (Chabot & Haack, 2006). However, recent studies of the isotopic compositions of iron meteorites have led to a greater appreciation for their antiquity and the clues they hold about the formation of our solar system (e.g., Kruijer, Kleine, and Borg (2019). This article reviews what studies of iron meteorites have revealed about asteroids, planets, and the solar system.

Classification

Chemical Composition

The earliest analyses of iron meteorites using neutron action analysis by Brown and Goldberg (1949), Goldberg, Uchiyama, and Brown (1951), and Lovering, Nichiporuk, Chodos, and Brown (1957) showed that the concentrations of gallium and germanium clustered into four so-called Ga-Ge groups, labeled I to IV in decreasing order of their elemental abundances. Subsequent work by Wasson (1967, 1969, 1970) and Wasson and Kimberlin (1967), using analyses of nickel and iridium in addition to gallium and germanium, defined more groups that were labeled with the letters A, B, and so on attached to the Roman numerals I–IV. Additional analytical data showed that some groups were related and four composite groups were created: IAB, IIAB, IIIAB, and IIICD (Scott, 1972; Scott & Wasson, 1975). Note, however, that groups IVA and IVB are not genetically related. Table 2 shows 14 groups that were defined by Scott and Wasson (1975), Wasson (2017), Wasson and Choe (2009), and Wasson, Choi, Jerde, and Ulff-Møller (1998).

Table 2. Iron Meteorite Groups and Their Properties*

Open in new tab

Group	Number	Ni (wt.%)	Structural Class	Example	NC/CC
IAB§	92	6–8.5	Og–Om	Campo del Cielo	NC
IC	11	6–7	Ogg, Og	Bendego	NC
IIAB	78	5.3–6.5	H, Ogg	Coahuila	NC
IIC	8	9.3–11.5	Opl	Ballinoo	CC
IID	21	9.6–11.1	Om, Of	Carbo	CC
IIE§	17	7.2–9.5	Og–Off	Weekeroo Station	NC
IIF	6	11–14	Opl, D	Corowa	CC
IIG	6	4–5	H	Bellsbank	NC?
IIIAB	~220	7.1–10.6	Om	Cape York	NC
IIICD§	12	12–23	Of–D	Tazewell	NC
IIIE	14	8.1–9.6	Og	Rhine Villa	NC
IIIF	8	6.8–8.5	Og, Om	Clark County	CC
IVA	~60	7.5–12	Of	Gibeon	NC
IVB	14	16–18	D	Hoba	CC
Ungrouped	~120	6–60	All	Tishomingo	65% CC

Open in new tab

* Notes. Numbers are largely from Goldstein et al. (2009a). Numbers in the Meteoritical Bulletin Database for major groups are ~30% higher but include paired specimens. Symbols for structural classes are defined in Table 3. NC and CC: non-carbonaceous and carbonaceous meteorites are defined chiefly on the basis of their Mo isotopic compositions (Kruijer, Burkhardt, Budde, & Kleine, 2017; Spitzer, Burkhardt, Budde, Kruijer, & Kleine, 2019; Worsham et al., 2019). Included in group IAB are those called main group IAB (MG) and subgroup sLL by Wasson and Kallemeyn (2002). Their subgroups, sLM and sLH, have the same Mo isotopic compositions as main group IAB and sLL irons (Worsham, Bermingham, & Walker, 2017) and together they are almost synonymous with group IIICD. Subgroups sHL and sHH have isotopically distinct Mo compositions from IAB irons (Worsham et al., 2017). Wasson and Kallemeyn (2002) refer to all these and other irons as the IAB complex.

^§ The so-called non-magmatic groups, IAB, IIE, and IIICD, are silicate-rich. The remaining groups, termed magmatic, have chemical trends consistent with fractional crystallization.

Iron meteorites were initially analyzed by Wasson (1967) for the three elements, gallium, germanium, and iridium, using radiochemical neutron activation analysis plus nickel by atomic absorption. As gamma ray detectors and counting equipment improved, these techniques were replaced by instrumental neutron activation analysis, which is non-destructive. This allowed many more elements to be determined: chromium, cobalt, copper, arsenic, antimony, tungsten, rhenium, platinum, and gold (Wasson, 2011; Wasson et al., 1998). Laser ablation inductively coupled plasma mass spectrometry and isotope dilution analysis have also been used to analyze group IVB irons (Campbell & Humayun, 2005; Walker et al., 2008). The most useful element for classifying iron meteorites is germanium as concentrations in irons vary by a factor of ~400,000 (0.005 to 1,900 ppm by weight, or μ‎g/g), whereas the range in each group is typically a factor 2 or less (Figure 3a). By contrast, for iridium, the range in some groups is almost as large as the total range in irons, which is a factor of ~10,000 (Figure 3b). For cobalt, the total range in all irons is only a factor of 3, and each group shows variations of only ±15% from the mean. Plots of germanium vs. nickel and iridium vs. nickel for ~700 irons are shown in Figure 3. Figure 4 shows a plot of cobalt vs. gold for most groups of irons.

Open in new tab

Figure 3. Compositions of grouped and ungrouped iron meteorites: (a) germanium vs. nickel; (b) iridium vs. nickel. Germanium shows a very large total range because it is one of the most volatile elements in iron meteorites and a very small range in each group as it is not fractionated significantly between solid and liquid iron-nickel. By contrast, iridium is highly refractory so that bulk compositions of groups lie close to the CI chondrite Ir/Ni ratio line. The very large variations of iridium within groups are due to the strong preference of iridium for solid-to-liquid iron-nickel. Concentration units for Ge and Ir: microgram/gram (= parts per million by weight).

Figures from Nancy Chabot adapted from Goldstein et al. (2009a).

Open in new tab

Figure 4. Logarithmic plot of cobalt vs. gold showing the bulk compositions of irons in 11 groups of irons. Although cobalt and gold analyses were not used to classify the iron meteorites, most groups are well resolved on this plot, and the two elements are similarly correlated in nearly all groups. This testifies to the power of the chemical classification to reveal new insights into the origins of iron meteorites. Concentration units for Co are milligram/gram (parts per thousand by weight).

Figure from John Wasson adapted from Wasson (2011).

Compositional plots for 18 elements in iron meteorites, like those in Figures 3 and 4, show several interesting features that provide important clues to the origins of iron meteorites (Scott, 1972; Scott & Wasson, 1975). First, chemical variations within groups, with three exceptions that will be discussed, are rather similar. For example, cobalt and gold are positively correlated and show nearly parallel trends within each group on a logarithmic plot (Figure 4). Iridium and nickel are negatively correlated in each group and show steep trends on the iridium vs. nickel plot (Figure 3b). The chemical variations within groups for nearly all elements are consistent with the trends expected during fractional crystallization of a large volume of molten metal such as an asteroidal core that is thoroughly mixed by convection and large enough to prevent homogenization of the solid by diffusion (Chabot & Haack, 2006; Wasson, 1985). Gold, cobalt, and nickel are preferentially concentrated in liquid iron, whereas iridium shows a strong preference for solid iron.

The three exceptional groups, IAB, IIE, and IIICD, show chemical and mineralogical characteristics that differ from those in the fractionally crystallized groups (Scott & Wasson, 1975; Wasson, 2017; Wasson & Kallemeyn, 2002; Worsham, Bermingham, & Walker, 2016). Typically, the ranges are much smaller and inter-element correlations are weaker than in the fractionally crystallized groups. For example, group IAB irons have a smaller range of gold contents (Figure 4). Note that Kallemeyn and Wasson (2002) defined a main group of IAB irons and various subgroups of what they called the “group IAB complex” (see Table 1 footnotes). Silicates are common in group IAB and related irons (Figures 7a and 7b) and in group IIE irons but rare in the fractionally crystallized groups. The three groups were once thought to have formed by sintering of solid grains of silicate and metallic iron-nickel that formed in the solar nebula and were therefore called non-magmatic irons (e.g., Matsuda, Namba, Maruoka, Matsumoto, & Kurat, 2005; Wasson, 1985). However, the large size of their taenite crystals in silicate-free regions (up to >50 cm; Buchwald, 1975, p. 391) and the presence of chemical trends consistent with chemical fractionation between solid and liquid (but not fractional crystallization) suggest that these irons formed after impact mixing of silicates and molten metal during early planetesimal collisions (Benedix, McCoy, Keil, & Love, 2000; Ruzicka, 2014; Wasson, 2017; Wasson & Kallemeyn, 2002). Given the inappropriate nature of the term “non-magmatic,” a better term for the three groups would be “silicate-rich,” even though silicates are found in other groups and IIICD is not especially silicate-rich (Goldstein, Scott, & Chabot, 2009a). However, the terms “magmatic” and “non-magmatic groups” are still in common usage.

Comparison of group bulk compositions shows that the group poorest in gallium and germanium, namely, IVB, and the ungrouped irons with similarly low levels of these elements are also poor in other volatile elements such as phosphorus, arsenic, gold, and antimony (Kelly & Larimer, 1977; Scott, 1978c; Wai & Wasson, 1979; Wasson, 1985). These six elements are more volatile than the other siderophile elements in the solar nebula, and their concentrations in the low-germanium irons are much lower than in chondrite metal. Figure 3a shows that Ge/Ni ratios in IVB are 10^3–4 times lower than in CI chondrites, which are closest to the solar values (e.g., Wasson, 1985). The volatile elements were clearly lost prior to crystallization of the molten metal. The high concentrations of nickel, iridium and similarly highly siderophile elements in group IVB and some other irons partly reflect formation in an oxidized parent body (Walker et al., 2008).

Iron meteorites that do not belong to the groups listed in Table 2 were once called “chemically anomalous irons” (Buchwald, 1975). However, this term went out of favor as they appear to have similar histories to the grouped irons, and they are now called “ungrouped.” Scott (1975) inferred that the 69 ungrouped irons known at that time were probably derived from around 50 additional bodies. Some like the South Byron trio probably come from asteroidal cores (McCoy et al., 2019). Others, like those related to group IAB irons, may come from metallic pools rather than cores (Wasson & Kallemeyn, 2002). In addition, a small number are probably formed by impact melting of known chondritic materials (See the section “Shock, Deformation, Impact Heating and Melting”). The total number of ungrouped irons grows as new irons are found and shrinks when five closely related irons achieve the status of a new group—by convention, five closely related irons are needed to define a new group (Wasson, 1985). As a result, the percentage of irons that is ungrouped has remained roughly constant at ~15%, although it appears to be higher among the irons from Antarctica and northwest Africa, which are smaller (Wasson, 1990, 2011). The ungrouped irons probably come from >50 different parent bodies (Chabot & Haack, 2006; Ruzicka, Haack, Chabot, & Scott, 2017; Wasson, 1985). Even though iron meteorites represent only 4–5% of all meteorite falls (Buchwald, 1975; Krot et al., 2014), they come from more parent bodies than any other meteorite type.

Isotopic Composition

There are two kinds of isotopic variations in meteorites (Dauphas & Schauble, 2016). Mass-dependent variations scale with the difference between the masses of the isotopes involved and result from all kinds of geological processes. For example, the oxygen isotopic compositions of Earth rocks spread along the terrestrial mass fractionation line shown in Figure 5. Mass-independent variations result chiefly from three processes: (a) the decay of radioactive nuclides, which scale with the parent/daughter ratio, (b) irradiation by cosmic rays prior to Earth impact, and (c) from isotopically diverse materials that existed in the solar nebula. The third kind of isotopic variations, also called isotopic anomalies, can in most cases be traced back to nucleosynthetic processes in evolved stars. The solids that condensed around evolved stars and survived in the interstellar medium were not fully homogenized when the solar system formed (Dauphas & Schauble, 2016; Yokoyama & Walker, 2016). Isotopic anomalies in chromium, titanium, molybdenum, oxygen, and other elements provide useful clues to the classification of meteorites and constraints on genetic relationships between iron meteorites and other meteorite types.

Open in new tab

Figure 5. Oxygen isotope plot of ${\text{δ}}^{18}\text{O}$ vs. ${\text{δ}}^1\text{O}$, which are the per mil deviations of the ${}^{18}\text{O}{/}^{16}\text{O}$ and ${}^{17}\text{O}{/}^{16}\text{O}$ ratios from terrestrial standard values, for silicates in IAB and IIICD iron meteorites and various achondrites and chondrites. Group IAB and IIICD iron meteorites plot close to the line defined by the winonaites—primitive achondrites that are chemically similar to the IAB silicates—suggesting they could be derived from the same parent body. These meteorites are clearly unrelated to acapulcoites and lodranites, which are also primitive achondrites, and CR chondrites. The line marked TFL is the terrestrial mass fractionation line defined by terrestrial silicate rocks, which has a slope of ~0.52 and is parallel to the winonaite line. Note that CCAM and Y&R are reference lines with slopes of 0.94 and 1.00, respectively, which both pass through the CAI composition. CCAM stands for carbonaceous chondrite anhydrous minerals. The Y&R line was defined by Young and Russell (1998).

Figure from Richard Greenwood, adapted from Greenwood et al. (2017).

Oxygen Isotopic Variations

Oxygen isotopic analysis is an essential tool in classifying unusual silicate-rich meteorites and evaluating possible genetic relationships (e.g., Clayton & Mayeda, 1996; Greenwood, Burbine, Miller, & Franchi, 2017). Most irons lack sufficient oxygen for isotopic analysis; the exceptions are iron meteorites with large inclusions of silicates, chromites, or phosphates.

There are three stable isotopes of oxygen: ${}^{16}\text{O}$, ${}^{17}\text{O}$, and ${}^{18}\text{O}$, which are formed in different stellar environments. Roughly one part in 500 is ${}^{18}\text{O}$, one part in 2,600 is ${}^{17}\text{O}$, and the rest is ${}^{16}\text{O}$. Oxygen isotopic ratios are expressed using the delta notation:

$$\begin{array}{l}{\text{δ}}^{18}\text{O}\mathrm{=}\left({}^{18}{\text{R}}_{\text{sample}}{-}^{18}{\text{R}}_{\text{standard}}\right)/{}^{18}{\text{R}}_{\text{standard}}{\text{where}\text{R}}^{18}\text{=}{}^{18}\text{O}/{}^{16}\text{O}\hfill \\ {\text{δ}}^{17}\text{O}\text{=}\left({}^{17}{\text{R}}_{\text{sample}}{-}^{17}{\text{R}}_{\text{standard}}\right)/{}^{17}{\text{R}}_{\text{standard}}{\text{where}\text{R}}^{17}\text{=}{}^{17}\text{O}/{}^{16}\text{O}\hfill \end{array}$$

Because delta values are very small, they are usually expressed in parts per thousand (‰, “per mil”). When the oxygen isotopic compositions of terrestrial silicate rocks are plotted on the standard oxygen isotope diagram with ${\text{δ}}^{18}\text{O}$ on the abscissa and ${\text{δ}}^{17}\text{O}$ on the ordinate, the data define a line of slope ~0.52 called the terrestrial fractionation line (TFL; Figure 5). Variations in the ${}^{17}\text{O}{/}^{16}\text{O}$ ratio are roughly half those in the ${}^{18}\text{O}{/}^{16}\text{O}$ as the mass difference between ${}^{17}\text{O}$ and ${}^{16}\text{O}$ is nearly half that between ${}^{18}\text{O}$ and ${}^{16}\text{O}$. Note that a linear relationship between ${\text{δ}}^{18}\text{O}$ and ${\text{δ}}^{17}\text{O}$ does not hold for much larger isotopic variations (see Clayton & Mayeda, 1996). Virtually all meteorites plot off the terrestrial fractionation line. The vertical deviations from the line ${\text{Δ}}^{17}\text{O}$ can be expressed as ${\text{Δ}}^{17}\text{O}={\text{δ}}^{17}\text{O}–0.52{\text{δ}}^{18}\text{O}$ (Clayton & Mayeda, 1996; more precise definitions of ${\text{Δ}}^{17}\text{O}$ are given by Greenwood et al., 2017, and Dauphas & Schauble, 2016). Values of ${\text{Δ}}^{17}\text{O}$ for rocks from the Moon, Mars, and the asteroid Vesta are 0.00, 0.32, and −0.24‰, respectively. These small ${\text{Δ}}^{17}\text{O}$ variations and much more extreme ${\text{Δ}}^{17}\text{O}$ values in CAIs are not thought to reflect nucleosynthetic processes in stars but are probably due to chemical processes in the protoplanetary disk or in the molecular cloud from which the Sun and solar system formed (e.g., Krot, 2019).

The oxygen isotopic compositions of silicates in IAB irons are plotted on the standard oxygen isotope plot in Figure 5. The data for IAB silicates lie on the mass fractionation line defined by the winonaite meteorites. These are primitive achondrites, which are strongly metamorphosed and recrystallized chondrites. Some contain a few relict chondrules but most experienced partial melting of silicate and metal-troilite. Since mineral compositions and textures are also very similar to winonaites and IAB silicates, it is highly likely that they come from the same parent body (Clayton & Mayeda, 1996; Benedix et al., 2000). Note, however, that ${\text{ε}}^{54}\text{Cr}$ analyses by Dey, Yin, Sanborn, Ziegler, and McCoy (2019) confirm that IAB metal is closely related to winonaites, but suggest that IAB silicates are not.

Group IIIAB irons contain very minor amounts of chromites, phosphates, and silicates. Oxygen isotopic analyses of these phases (with one exception) support a common parent body for IIIAB irons and main group pallasites (Greenwood et al., 2017), although Yang, Goldstein, and Scott (2010b) find problems with this. Two other groups of irons have been linked to chondrites by oxygen isotopic analyses. Silicates in the two IVA irons, Steinbach and São João Nepomuceno, and silica in Gibeon and Bishop Canyon gave slightly different ${\text{Δ}}^{17}\text{O}$ values of 1.23‰ and 1.08‰, respectively, which both overlap values for L and LL chondrites (Clayton & Mayeda, 1996; Wang, Rumble, & McCoy, 2004). Silicates in IIE irons, which are rather diverse in mineralogy, have oxygen isotopic compositions that overlap with those of unweathered H4-6 chondrites (McDermott, Greenwood, Scott, Franchi, & Anand, 2016), suggesting a possible genetic link. However, Wasson (2017) argues that the chondritic precursor for IIE irons was more reduced and resembled the chondritic clasts in the Netschaëvo IIE meteorite.

Nucleosynthetic Isotopic Variations

Nucleosynthetic isotopic variations were first identified in bulk meteorites for ⁵⁰Ti and have since been recognized for over 10 elements in all kinds of meteorites. Four elements show nucleosynthetic isotopic variations in iron meteorites: nickel, ruthenium, tungsten, and molybdenum (Dauphas & Schauble, 2016; Kruijer et al., 2017; Regelous, Elliot, & Coath, 2006; Yokoyama & Walker, 2016). Since magmatic iron meteorites are thought to come from well-mixed asteroidal cores, irons in a given group should have the same nucleosynthetic isotopic anomalies after corrections have been made for effects due to galactic cosmic rays during transit to Earth. One iron meteorite has been reclassified as a result of isotopic analysis. Wiley was originally classified as an anomalous Ni-rich member of group IIC (Wasson, 1969), but it has the most anomalous molybdenum isotopic composition of all meteorites and cannot be derived from the same source as the IIC irons (Figure 6).

Open in new tab

Figure 6. Plot of ${\text{ε}}^{95}\text{Mo}$ vs. ${\text{ε}}^{94}\text{Mo}$ (deviations of the ⁹⁵Mo/⁹⁶Mo and ⁹⁴Mo/⁹⁶Mo ratios from the terrestrial standard values in parts per 10⁴) showing that the iron meteorite groups define two populations. Five groups shown as solid blue symbols lie within error of the dashed blue line marked CC, which is defined by carbonaceous chondrites. Six iron meteorite groups shown as solid red symbols lie on the lower dashed red line marked NC, which is defined by the non-carbonaceous chondrites (ordinary and enstatite chondrites; OC and EC). Jupiter plausibly separated the two populations for several million years during its growth.

Data and interpretation from Kruijer et al. (2017, 2019), Poole, Rehkämper, Coles, Goldberg, and Smith (2017), Worsham et al. (2017, 2019), Bermingham et al. (2018) and Budde et al. (2019). Figure from Worsham et al. (2019).

The most remarkable discovery from measurements of nucleosynthetic isotopic anomalies in iron meteorites is that they can be divided into two populations. Figure 6 is a plot showing deviations of the ⁹⁵Mo/⁹⁶Mo and ⁹⁴Mo/⁹⁶Mo ratios in parts per 10⁴ from the terrestrial standard which are called ${\text{ε}}^{95}\text{Mo}$ and ${\text{ε}}^{94}\text{Mo}$. This shows that the iron meteorites fall on two parallel lines. Five groups lie on a line defined by carbonaceous chondrites and are called the CC population: IIC, IID, IIF, IIIF, and IVB. The remaining groups of irons plot on the line defined by the non-carbonaceous chondrites (enstatite and ordinary chondrites) and most achondrites; they are called the NC population (Kruijer et al., 2017). Ungrouped irons, including Wiley, are found predominantly in the CC population, the remainder are NC (see Spitzer et al., 2019; Worsham et al., 2019). The most likely explanation is that the two populations formed in different locations in the solar system. Given that carbonaceous chondrites are rich in water, the CC population probably formed further from the Sun than the NC population. This is discussed further in the section “Where Did the Iron Meteorites Form?”.

A new classification system should help explain existing compositional data (e.g., Scott, 1972) and the NC/CC division passes this test. Prombo and Clayton (1993) found a wide range of ¹⁵N/¹⁴N ratios in iron meteorites (–90 to +150‰) with small ranges in each group, but they found no correlation between the ¹⁵N/¹⁴N ratio and other parameters. However, CC irons are enriched in ¹⁵N with ${\text{δ}}^{15}\text{N}$ between +3 and +150‰ while the NC irons are depleted in ¹⁵N with ${\text{δ}}^{15}\text{N}$ between –90 and –3‰ (Worsham et al., 2019). The inferred locations of the NC and CC irons are consistent with the observation that enrichments in ¹⁵N are associated with the presence of ice and organics and that ${\text{δ}}^{15}\text{N}$ increases with increasing distance from the Sun (Füri & Marty, 2015). Even though nitrogen has only two stable isotopes, making it difficult to distinguish mass-dependent and mass-independent isotopic variations, the NC–CC division provides useful insights into the nitrogen isotopic composition of iron meteorites.

Minerals and Structure

Widmanstätten Pattern

The Widmanstätten pattern was discovered independently by William Thompson in 1804 and by Alois von Widmanstätten in 1808 (Clarke & Goldstein, 1978). For many years the pattern was thought to result from crystallization of iron-nickel from a slowly cooling melt. However, as a result of improved versions of the iron-nickel phase diagram, X-ray crystallographic studies, better chemical analyses and microscopic techniques, and laboratory experiments with iron-nickel alloys, the Widmanstätten pattern is now recognized to be a product of solid-state processes (Buchwald, 1975, p. 115; Burke, 1986, Perry, 1944). Until the chemical classification was developed, the Widmanstätten pattern provided the primary means for classifying iron meteorites.

The Widmanstätten pattern, which is visible in polished and etched slices of most iron meteorites, results from the growth on cooling of body-centered cubic (bcc) iron, called kamacite, from face-centered cubic (fcc) iron, called taenite. Note that, in metallurgy, these two phases are called ferrite ($\text{α}$) and austenite ($\text{γ}$), respectively. In pure iron, the fcc-to-bcc transformation occurs at 910°C, but addition of nickel lowers this temperature. Plates or lamellae of kamacite are oriented parallel to the octahedral or {111} planes of the taenite crystals, which can be meters or more in size (Figure 7). Irons showing this pattern are called octahedrites and they are classified according to the average width of the largest kamacite lamellae (Table 3). Kamacite lamellae are narrower in irons with relatively high bulk Ni concentrations or fast cooling rates. Irons with kamacite lamellae that are too small to be observed without a microscope are called ataxites. Fine intergrowths of kamacite and taenite between the major kamacite lamellae in octahedrites are called plessite (Goldstein & Michel, 2006). Irons with very low nickel contents (less than ~5.8 wt.%) generally form kamacite crystals up to 25 cm or more in size via a massive transformation (Yang & Goldstein, 2005). They lack taenite and are called hexahedrites. Table 3 summarizes the structural types of iron meteorites and the chemical groups in which they occur (Buchwald, 1975). Most coarse octahedrites are group IAB irons; medium octahedrites are very largely group IIIAB irons; most fine octahedrites are IVA irons; and nearly all hexahedrites are low-nickel members of group IIAB.

Table 3. Structural Divisions Among Iron Meteorites*

Open in new tab

Structural Class	Symbol	Kamacite Width (mm)	Chemical Groups
Hexahedrite	H	–	Low-Ni IIAB
Coarsest Octahedrite	Ogg†	>3.3	High-Ni IIAB
Coarse Octahedrite	Og†	1.5–3.3	IAB, IIIE
Medium Octahedrite	Om	0.5–1.3	IIIAB, IID
Fine Octahedrite	Of	0.2–0.5	IVA
Finest Octahedrite	Off	<0.2	–
Plessitic Octahedrite	Opl	<0.2 (spindles)	IIC
Ataxite	D	–	IVB

Open in new tab

* Notes. A few irons have anomalous structures and are mostly chemically ungrouped also.

† The g in Og and Ogg is the first letter of grob, German for coarse. The symbol D is for dicht, meaning compact or densely structured (Burke, 1986).

Open in new tab

Figure 7. Polished and etched slices of iron meteorites with diverse structures and inclusions. (a) Rifle IAB coarse octahedrite with many cm-sized graphite-troilite nodules. On the right side, the kamacite lamellae contain elongated cohenite grains. Courtesy of Smithsonian Institution. (b) Udei Station, a group IAB-related medium octahedrite, which has abundant angular black silicate inclusions like ~5–10% of IAB irons and a higher metallic nickel concentration of 9.5 wt.% (Wasson & Kallemeyn, 2002). Maximum width of slice is 9 cm.

Photo credit: Paul Swartz, Tucson Meteorites. (c) Cape York IIIAB medium octahedrite with a 9 cm long troilite nodule. Courtesy of Smithsonian Institution. (d) 37 cm wide slice of the Gibeon IVA fine octahedrite. Sparsely distributed dark blebs are millimeter-sized troilite nodules. Over 20 tons of Gibeon irons have been found over a large elliptical area in the Republic of Namibia measuring 275 × 100 km (Buchwald, 1975). Image from Gary Huss, University of Hawaii.

Silicates

The presence of silicates in iron meteorites, which are thought to have been once molten, has been a long-standing enigma (Burke, 1986). Given the large density difference between silicates and molten metal, silicates should have quickly floated to the surface unless trapped by crystallizing solid metal or by violent motions of the melt (e.g., Anders, 1964). Silicates are common in group IAB (Figure 7b) and group IIE irons and are very abundant in two IVA irons. Silicates are also abundant in several ungrouped irons, including Tucson, Sombrerete, and Guin (Ruzicka, 2014). Common silicate minerals in irons are listed in Table 4. In most IAB and a few IIE irons, silicates occur in clasts in roughly chondritic proportions (roughly equal proportions of olivine and pyroxene with smaller abundances of feldspar). Indeed, relict chondrules have been observed in silicate clasts in the IAB iron, Campo del Cielo, and the IIE irons, Netschaëvo and Mont Dieu (Olsen & Jarosewich, 1971; Van Roosbroek et al., 2015; Schrader, McCoy, & Gardner-Vandy, 2017). Silicate inclusions with non-chondritic mineralogy occur in the IAB irons, Caddo County and Ocotillo (Benedix et al., 2000), and in many IIE irons, including Colomera, Elga, and Weekeroo Station (Ruzicka, 2014). They include clasts of gabbro, basalt, and andesite, which are rich in feldspar, and harzburgite and lherzolite, which are depleted in feldspar and enriched in olivine and pyroxene (see also “Non-Magmatic Iron Meteorites”).

The two silicate-rich iron meteorites in group IVA, Steinbach and São João Nepomuceno, contain so much silicate that they could have been classified as stony-iron meteorites (Figure 8). However, the chemical composition and texture of their metallic portions and their metallographic cooling rates closely match the group IVA trends, so their classification is not in doubt (Haack, Scott, Love, Brearley, & McCoy, 1996). The silicate inclusions in these two irons are unique mineralogically and texturally, and their origin is especially enigmatic (Dos Santos, Scorzelli, & Varela, 2018; Ruzicka, 2014; Ruzicka & Hutson, 2006; Ulff-Møller et al., 1995). Although the oxygen isotopic composition of the silicates resembles that in L or LL chondrites (Wang et al., 2004), the mineralogy of the silicate inclusions in the two IVA irons is totally non-chondritic: coarse-grained low-Ca pyroxene, which is poor in rare-earth and other incompatible elements, and silica, plus minor chromite, metal, and troilite. Two low-Ca pyroxenes are present: orthobronzite and clinobronzite. Their compositions and the structure of the clinobronzite show that the phases equilibrated at ~1,200°C and were then quenched at ~100°C/hour (Haack et al., 1996; Reid, Williams, & Takeda, 1974). Large metallic regions show a fine octahedral structure as in IVA irons; silicate-rich regions contain smaller metal grains (Figure 8). Metallographic cooling rates at ~500°C were much lower: 1,000–3,000°C/Myr for São João Nepomuceno and ~150°C/Myr for Steinbach (Haack et al., 1996; Rasmussen, Ulff-Møller, & Haack, 1995; Yang, Goldstein, & Scott, 2008). Mössbauer studies on orthopyroxene in São João Nepomuceno gave a cooling rate at ~500°C of 10⁴ °C/Myr, within error of the metallographic data (Dos Santos et al., 2018).

Open in new tab

Figure 8. Polished and etched slices of the two IVA irons that contain abundant silicates as well as metallic iron-nickel that chemically and structurally matches the group IVA trends. (a) Steinbach contains 6 vol.% troilite, 32 vol.% metallic iron-nickel. The uniformly oriented Widmanstätten pattern shows that the metal grains are part of a large single crystal (Ruzicka & Hutson, 2006). (b) São João Nepomuceno contains a large metallic region (left side) as well as the Steinbach-like metal-silicate intergrowth. Width of slices: (a) 4.8 cm; (b) 2.5 cm.

Photo credits: (a) Mirko Graul (Bernau/Germany), Collection: Oliver Sachs (Nördlinger Ries, Germany); (b) Alex Ruzicka adapted from Ruzicka (2014).

Two other IVA irons contain platelets of tridymite, Gibeon and Bishop Canyon (see Ruzicka, 2014), and a third, Muonionalusta, contains the high-pressure mineral, stishovite, which was once probably tridymite (Holtstam, Broman, Söderhielm, & Zetterqvist, 2003). Implications of the remarkable thermal history recorded by the silicates for the origin of IVA irons are discussed in the section “Group IVA.” Note that the silicate minerals in Steinbach and São João Nepomuceno may be explained if silica had been added to convert olivine into pyroxene. This would help explain their unusual metal-silicate texture, which resembles that of the Brenham pallasite (see Figure 8a; Scott, Haack, & McCoy, 1996).

Thermal Histories Inferred From Microstructure

Studies of kamacite growth in iron meteorites provide information about their cooling rates through the temperature range of ~700 to ~300°C (e.g., Goldstein et al., 2009a). The growth of kamacite from taenite is controlled by solid-state diffusion and the equilibrium compositions of kamacite and taenite, which both increase in nickel content with falling temperature (Figure 9). Computer simulation of kamacite growth provides an estimate of the cooling rate (Goldstein & Ogilvie, 1965; Wood, 1964). Very approximate cooling rates can be inferred from kamacite plate width and bulk nickel content provided that the kamacite lamellae are well separated (Short & Goldstein, 1967). However, kamacite lamellae are commonly closely spaced so that central nickel concentrations in taenite lamellae decrease with increasing width. Figure 10 shows how nickel concentrations measured with an electron probe analyzer vary across a taenite lamella sandwiched between two kamacite lamellae in a group IVA iron meteorite. During kamacite growth, nickel atoms diffused into the interior of both phases from their interface. Taenite is strongly zoned because diffusion rates decrease exponentially with falling temperature. As a result, the nickel concentration gradient steepens toward the interface. Kamacite is more homogeneous than taenite as diffusion is faster in kamacite than in taenite. Taenite shows an M-shaped nickel profile, which largely reflects kamacite growth between 700°C and 500°C, whereas kamacite shows a small depletion close to taenite due to the reversal in the slope of the kamacite/(kamacite + taenite) phase boundary below 400°C. Figure 10 also shows a computer-generated nickel profile for a cooling rate of 200°C/Myr that matches the measured profile (Goldstein et al., 2009a).

Open in new tab

Figure 9. Fe-Ni phase diagram showing the stability fields of kamacite, the Ni-poor bcc phase, which is called α‎ by metallurgists, and taenite, the Ni-rich fcc alloy known as γ‎, and various metastable reactions that may occur (Yang, Williams, & Goldstein, 1996). The phase diagram above 450°C effectively controls the growth of kamacite. Below ~450°C, the phase diagram is more complex as the paramagnetic phase γ₁‎ may coexist with a higher-Ni, ferromagnetic phase γ₂‎. The hatched lines labeled Sp show the compositional range within which taenite starts to spontaneously decompose into magnetically ordered and disordered phases. Below ~350°C, the stable phases are kamacite and awaruite, ordered Ni₃Fe, also called ${\text{γ}}^{\prime }$, which has not been reported in iron meteorites. Instead, tetrataenite, which is ordered FeNi, ${\text{γ}}^{″}$, forms below the ordering temperature, ${\text{T}}_{\text{C}}{\text{γ}}^{″}$, of ~320°C. Tetrataenite forms a thin layer between kamacite and taenite and is the high-Ni phase that forms on cooling in the spinodal intergrowth. Taenite that is unable to nucleate kamacite on cooling will eventually cross the lines marked M_s and M_f, which show the starting and finishing temperatures for the formation of martensite, strained bcc, via a diffusionless reaction.

Figure from Goldstein et al. (2017).

Open in new tab

Figure 10. Data points show nickel concentrations measured using an electron microprobe across taenite and adjacent kamacite lamellae in the Duchesne IVA iron meteorite. The solid line shows the theoretical M-shaped nickel profile calculated for a cooling rate of 200°C/Myr. The mean cooling rate determined from 15 taenite lamellae is 100 °C/Myr (Yang et al., 2008).

Figure adapted from Goldstein et al. (2009a).

Estimates of cooling rates in irons have become more accurate since the 1970s because of improvements in the knowledge of diffusion coefficients, the effects of phosphorus on the nucleation and growth of kamacite, and the Fe-Ni-P phase diagram (Goldstein et al., 2009a; Yang & Goldstein, 2006). The importance of phosphorus can be demonstrated with laboratory experiments which show that the only way for kamacite to form in pure Fe-Ni alloys is by nucleation at grain boundaries. However, addition of only 0.1 wt.% P is enough to allow micrometer-sized oriented lamellae to form within taenite crystals. Yang and Goldstein (2005) used the Fe-Ni-P phase diagram, diffusion coefficients, and a computer model that successfully reproduced the zoning profiles in laboratory-cooled alloys to show that in most irons, formation of kamacite lamellae required prior formation of phosphide. In phosphorus-poor irons like those in group IVA, kamacite formed when taenite began to transform to martensite, a strained bcc phase, which forms without any diffusion. Figure 9 shows that in a IVA iron with 8% nickel, kamacite only formed after martensite appeared at ~600°C, around 100°C below the equilibrium temperature for a binary Fe-Ni alloy. As a result of a better understanding of kamacite nucleation mechanisms and more accurate phase diagram data and diffusion rates, estimated cooling rates for IVA irons, for example, have changed by a factor of 10–50 (Goldstein et al., 2009a).

The cooling rate of an iron meteorite at ~500°C is best determined by matching at least three nickel profiles across taenite lamellae (Figure 10) or by comparing measured and computed values for the central nickel contents of 10 or more taenite lamellae with diverse widths (Wood, 1964). Central nickel contents for group IVA irons, for example, lie in the range of 15 to 30 wt.% for lamellae with corresponding widths of 50 to 5 μ‎m (Goldstein et al., 2009a). In both methods, the orientations of the taenite lamellae relative to the polished surface are determined to eliminate geometrical errors (Yang & Goldstein, 2006). In addition, bulk nickel and phosphorus concentrations in the iron meteorites are needed. Cooling rates for irons in four groups are as follows: IAB-IIICD: 10–20°C/Myr; IIIAB, 50–300°C/Myr; IVA, 100–5,000°C/Myr; and IVB, 500–5,000°C/Myr (Goldstein, Yang, Scott, 2014; Yang et al., 2008; Yang & Goldstein, 2006; Yang, Goldstein, Michael, Kotula, & Scott, 2010a). In most cases the relative accuracy is a factor of ~2–3; the absolute errors are at least ~2–3× greater (Yang & Goldstein, 2006). Cooling rates in the IVB irons, which are ataxites with 16–18% nickel, were determined by measuring nickel profiles ≤3 μ‎m wide next to kamacite platelets using an analytical electron microscope.

For a metallic core enveloped by a silicate mantle, the cooling rate is determined almost entirely by the thickness of the silicate mantle and the regolith layer on the surface, as the thermal conductivity of metal is about 30× that of silicate rock (Chabot & Haack, 2006). The radii of parent bodies estimated from the mean metallographic cooling rate in each group range from ~3 km for the fast-cooled group IVB to ~100 km for the slowest cooled group IIAB. Note that cooling rates for group IIAB irons were determined by modeling phosphide growth (Randich & Goldstein, 1978), and need updating. The large range of cooling rates within groups IIIAB, IVA, and IVB is much larger than expected as cores within a silicate mantle would have cooled almost isothermally.

Cooling rates have also been estimated for stony irons with silicate-free metal grains that are large enough to develop Widmanstätten patterns. Cooling rate estimates for main group pallasites, which are olivine-metal mixtures, are ~2–20°C/Myr and for mesosiderites, which are metal-basalt mixtures, 0.2–5°C/Myr (Goldstein et al., 2014; Yang et al., 2010b).

The very diverse and puzzling cooling rates in groups thought to have formed in cores, which correlate with bulk nickel contents in each group, and the extraordinarily low cooling rates of mesosiderites have generated considerable debate about the veracity of the metallographic cooling rates. Concerns have been raised that cooling rate estimates were influenced by systematic errors in diffusion rates and phase equilibria, for example, or unrecognized effects of impact-induced heating or kamacite nucleation (see Wasson & Richardson, 2001; Wasson and Hoppe, 2012). For this reason and because coarse octahedrites have curved not straight kamacite-taenite interfaces that are not amenable to kamacite growth modeling, various other techniques have been developed to derive cooling rates for iron meteorites. These include measurements of the interface concentrations of nickel (and cobalt) in kamacite and taenite at their interface, which are a function of the concentration gradients in the phases and the resolution of the analytical technique (Wasson & Hoppe, 2012). Nickel concentrations at the interface correlate with the metallographic cooling rate determined from taenite zoning, as expected (Goldstein et al., 2014; Short & Goldstein, 1967). However, the technique of measuring Ni/Co ratios at the interface using an ion probe (Wasson & Hoppe, 2012) does not appear to be sufficiently precise (Goldstein et al., 2014).

Other techniques for determining cooling rates of iron meteorites are based on studies of the microstructures that form in the taenite fields between the kamacite lamellae below 400°C. Reflected light microscopy of etched sections of octahedrites reveals a fine intergrowth of kamacite and taenite called plessite in the central part of large taenite fields (Figure 11). Enclosing the plessite is a dark-etching zone called cloudy taenite, which is separated from kamacite by a thin layer of tetrataenite, tetragonal FeNi, which is just visible in Figure 11. In taenite, iron and nickel atoms are randomly distributed in the fcc structure, but in tetrataenite they are ordered on alternate (002) planes. Cloudy taenite is a nanoscale intergrowth that forms below 320°C in slowly cooled taenite with ~30–40 wt.% nickel as a result of magnetic ordering (Figure 12). The transformation occurred spinodally, that is, without any nucleation barrier. Transmission electron microscopy of cloudy taenite shows that it consists of a low-nickel matrix that appears to enclose equant grains of tetrataenite. The identity of the low-nickel matrix phase in the cloudy zone is somewhat uncertain and may depend on the cooling rate of the meteorite and the technique used for specimen preparation. Martensite, ordered Fe₃Ni, and a paramagnetic form of taenite called antitaenite have all been identified as the low-Ni phase (Blukis, Rüffer, Chumakov, & Harrison, 2017; Goldstein et al., 2009a; Scorzelli & Dos Santos, 2019). The scale of the cloudy taenite intergrowth decreases with distance from the interface with tetrataenite and increases with slower cooling (Maurel, Weiss, & Bryson, 2019; Nichols et al., 2020; Yang, Williams, & Goldstein, 1997). Cloudy taenite is not detectable in fast cooled irons (>10⁵ °C/Myr; Yang et al., 2014). Tetrataenite cannot be formed in the laboratory without enhancing diffusion rates by irradiation, and it is disordered by heating above 320°C. The tetrataenite boundary layer is widest in mesosiderites—stony-irons which have the lowest metallographic cooling rates (Goldstein et al., 2014). Note that the largest known volume of tetrataenite is found in the 1.8 kg ungrouped iron meteorite, NWA 6257, which contains 95 vol.% of tetrataenite with 43% nickel and lacks kamacite (Poirier et al., 2015).

Open in new tab

Figure 11. Reflected light photomicrograph of an etched section of Goose Lake IAB iron (medium octahedrite with 8.3% nickel) showing a rectangular field of black plessite (Pl), which is rimmed by residual taenite and a cloudy taenite rim. Between the kamacite and cloudy taenite there is a thin clear rim of tetrataenite, Fe,Ni, which is visible in the upper right. In the kamacite there are faint diagonal Neumann bands, which are deformation twin lamellae, and tiny euhedral schreibersite grains called rhabdites.

Adapted from Buchwald (1975, p. 96) by permission of ASU Center for Meteorite Studies. Scale bar is 200 μ‎m.

Open in new tab

Figure 12. Nickel X-ray map showing a 1 μ‎m wide region in the Chinautla IVA iron meteorite extending from the cloudy taenite zone on the left (CZ) though the uniform rim of tetrataenite (Tt: ordered FeNi) to kamacite (K) on the right. The cloudy taenite consists of equant islands of tetrataenite in a low-Ni matrix and the scale of the intergrowth decreases as the local bulk nickel content decreases from ~40 to 30 wt.%. The widths of the tetrataenite islands at the high-nickel edge of the cloudy zone and the thicknesses of the tetrataenite rim in iron meteorites are inversely correlated with the cooling rate.

Image from Goldstein et al. (2014).

In four groups of irons, IAB, IIIAB, IVA, and IVB, the sizes of the tetrataenite islands in cloudy taenite and the widths of the tetrataenite boundary layers are inversely correlated with the metallographic cooling rate, confirming the non-uniform cooling rates in these groups (Goldstein, Yang, Kotula, Michael, & Scott, 2009b; Goldstein et al., 2014; Yang et al., 2010a, 2010b). In moderately or well-shocked IVA irons, the cloudy taenite intergrowth is absent and the nickel profile in taenite is altered but only on a sub-micrometer scale. This shows that the wide variation of cooling rates in IVA irons is real and not an artefact of impact heating. The correlations between tetrataenite island size, the width of the tetrataenite layer, and the metallographic cooling rate for a whole range of iron and stony-iron meteorites provide two empirical methods for determining cooling rates for irons and other metal-rich meteorites lacking well-developed Widmanstätten patterns (Goldstein et al., 2009b, 2014; Nichols et al., 2020). These diverse techniques and electron probe measurements of nickel in kamacite and taenite at their interface all confirm that the mesosiderites cooled at uniquely slow rates and that main-group pallasites cooled slower than IIIAB and other iron meteorites (Goldstein et al., 2014; Yang et al., 2010b).

To explain the wide variations in the cooling rates of IVA irons—5,000°C/Myr for the early formed, low-Ni irons and 100°C/Myr for the late formed, high-Ni irons—Yang, Goldstein, and Scott (2007, 2008) inferred that the IVA irons essentially crystallized in a metallic body without a silicate mantle. They used thermal and fractional crystallization modeling (see the section “Crystallization and Chemical Trends Within Groups”) to suggest that group IVA irons crystallized from a molten metallic body that was 150 ± 50 km in radius and had less than 1 km of silicate insulation. The metallic body may have formed when a metallic core was separated from its mantle in a hit-and-run impact (see the section “Destruction of Iron Meteorite Parent Bodies”). This conclusion was also favored by Ruzicka and Hutson (2006) and McCoy et al. (2011) from their studies of IVA irons.

Group IVB irons also showed a wide range of cooling rates (500–5,000°C/Myr), but unlike the IVA irons, the high-Ni irons in group IVB cooled faster than the low-Ni ones. Yang et al. (2010a) suggested that the IVB metallic body was 65 ± 15 km in radius when it cooled without a mantle. The concept of mantle-stripped, planetesimal cores was a major and unexpected result of the 50-year effort to improve cooling rates for iron meteorites.

Shock, Deformation, Impact Heating and Melting

Shock and impact features in iron meteorites are of interest for three reasons. First, they can illuminate how and when the parent bodies of irons were broken up by hypervelocity impacts, typically at velocities of ~5 km/s, the current mean impact velocity of asteroids, and how pieces were delivered to Earth. Second, these features may record how impacts affected planetesimals during accretion when impact velocities were much lower, around the escape velocity. The third reason for studying impact features in iron meteorites is to be able to definitively distinguish primary structural features produced during formation and initial cooling from subsequent secondary impact effects (Buchwald, 1975; Stöffler, Bischoff, Buchwald, & Rubin, 1988).

Reviews on shock effects in meteorites and meteoritic breccias typically focus on stony meteorites and scarcely mention iron meteorites (see, e.g., Sharp & de Carli, 2005). This is not because shocked iron meteorites are uncommon but because they show many different characteristics from shocked silicate meteorites (Rubin & Ma, 2017). Thus, numerous high-pressure silicate minerals have been found in shock melt veins in chondrites, achondrites, and Martian meteorites (e.g., Bischoff et al., 2006; Tomioka & Miyahara, 2017), but high-pressure phases in shocked irons are rare. Group IAB irons, which are rich in silicates, are in general not significantly shocked. Some group IIE irons, like Elga, contain silicate inclusions and are heavily shocked and deformed around localized impact melts (Olsen et al., 1994; Litasov, Teplyakova, Shatskiy, & Kuper, 2019). Three high-pressure phases have been reported in irons: diamond in three IAB irons (Buchwald, 1977, pp. 384–385; Clarke, Appleman, & Ross, 1981; Clarke, Jarosewich, Ross, Wasson, & English, 1994), stishovite in a IVA iron (Holtstam et al., 2003) and an ungrouped iron, Tishomingo (see Yang et al., 2014), and allabogdanite, (Fe,Ni)₂P, in three meteorites (Britvin et al., 2019). Note that lonsdaleite, hexagonal diamond, was also reported in the three IAB irons, but is now known to be twinned diamond (Németh, Garvie, Aoki, Dubrovinskaia, & Buseck, 2014). These three high-pressure minerals are rare in irons as the only group with abundant graphite, IAB, has few shocked members, and SiO₂ and (Fe,Ni)₂P are uncommon in irons.

Fragmental and regolith breccias, which are common among stony meteorites, are non-existent among iron meteorites. Silicate rocks are brittle and readily fragmented, whereas metal deforms plastically so fragments may be rare on metallic asteroids. But even if metal fragments do accumulate on the surface, they cannot readily be welded back into coherent masses that will survive the journey to Earth’s surface as there is no equivalent in irons of the feldspathic melt that glues stony meteorite fragments together to form breccias (Bischoff & Stöffler, 1992). Iron meteorites like Udei Station (Figure 7b) that contain silicate fragments and metal grains with different orientations are not fragmental breccias, despite some superficial similarities. They were formed by mixing of fragmented solid silicate with liquid metal, not solid metal, as was once believed (see Benedix et al., 2000). Abundant silicates prevented large taenite crystals from growing in the melt.

Overviews of shock and reheating effects in iron meteorites can be found in Volume 1, Chapter 11, of Buchwald (1975) and in Rubin and Ma (2017). More information on these features and their interpretation can be found in the comprehensive descriptions of shocked irons in Volumes 2 and 3 of Buchwald (1975). In Appendix 1, Buchwald (1975) tabulates shock and reheating effects in kamacite and troilite based on his expert interpretations of the microstructures. They show clearly that some groups like IIIAB and IVA were shocked to much higher levels than groups IAB and IIAB, which are largely unshocked.

The most common shock feature in irons is found in kamacite, which transforms to a dense, hexagonal close-packed phase, $\text{ε}$-iron, under shock pressures of >13 GPa (Buchwald, 1975). On pressure release, it reverts to bcc kamacite. Since the resulting lattice is distorted, on etching it shows a characteristic hatched appearance in reflected light microscopy. This shock feature is especially common in groups IIIAB and IVA (Buchwald, 1975). Significant reheating of shocked kamacite causes recrystallization—the nucleation and growth of strain-free crystals in distorted iron-nickel. A few percent of irons that contain recrystallized kamacite are identified by Buchwald (1975).

The second most common shock feature is a fine-grained intergrowth of troilite and metal containing daubréelite grains that Buchwald (1975) and others identified as shock-melted troilite (Figure 13). In ordinary chondrites, similar occurrences of shock-melted troilite grains lack daubréelite (Scott, 1982). Some authors have doubted the interpretation of these features as shock melts and argued for other processes such as low-temperature intrusion of S-rich vapor and metamorphism after impacts (see Yang et al., 2014). However, detailed studies using scanning and transmission electron microscopy and electron probe analysis of troilite-metal intergrowths in the ungrouped irons Tishomingo (Yang et al., 2014) and Willamette (Rubin et al., 2015), and in group IIIE irons (Breen, Rubin, & Wasson, 2016) strongly support the interpretation of these intergrowths as impact-shock features. Small inclusions of troilite and daubréelite, which occur in essentially unshocked group IIIE irons, were converted to finely dispersed metal-sulfide intergrowths in severely shocked IIIE irons (Breen et al., 2016). Figures 13b and 13c show that at distances of ~200 μ‎m from the shock-melted intergrowth, kamacite was scarcely modified. This evidence for steep thermal gradients is supported by the occurrence of localized impact melting around otherwise unmelted troilite nodules in irons (e.g., Buchwald, 1975). The near absence of dendritic structure in the fine-grained, shock-melted intergrowths in iron meteorites, which are present in quenched Fe-FeS melts (Figure 14), and the morphology of spidery troilite-rich filaments in adjacent kamacite (Figure 13) imply mobilization of a complex solid-melt mixture. Numerical modeling of shock in ordinary chondrites supports the development of steep thermal gradients at iron–troilite boundaries with enhanced heating due to friction (Moreau, Kohout, & Wünnemann, 2018).

Open in new tab

Figure 13. Backscattered electron images of three troilite–daubréelite–phosphide inclusions in two group IIIE iron meteorites showing the effect of shock melting. (a) Pristine grain of troilite (tr) containing exsolved daubréelite lamellae (db), which is rimmed by schreibersite (sch) in kamacite in the weakly-shocked Rhine Villa meteorite. (b, c) In the severely shocked Aliskerovo meteorite, shock-melted troilite has been dispersed into the kamacite matrix forming extremely fine-grained, spidery troilite-rich filaments. Daubréelite forms numerous angular and subhedral equant grains. (b) Close to the troilite-melted intergrowth, kamacite was hot enough to recrystallize; near the edge, kamacite was cooler and shows decorated Neumann twin lamellae. (c) Note the 10 μ‎m wide grain of tetrataenite (tt, white) which formed after impact melting of troilite.

Images from Breen et al. (2016).

Open in new tab

Figure 14. Reflected-light photomicrograph of Mount Howe 88403, which is a 2.5 kg ungrouped iron meteorite containing abundant rounded grains of troilite (Tr) in metallic iron-nickel (white). Inset shows a 5 cm-wide polished section with troilite nodules that are evenly dispersed characteristic of Fe-S melts that cooled rapidly in hours (Scott, 1982). This iron formed by impact melting on a chondritic surface, probably that of the IIE parent body (Schrader et al., 2010; Wasson, 2017).

Image is 3 mm wide: courtesy of NASA Johnson Space Center Astromaterials Curation Office.

The fine-grained, shock-melted intergrowths of troilite and metal contain phases that were not present in the original unshocked iron meteorites. For example, some kamacite grains in the intergrowth are cobalt-rich and nickel-poor: ~2–3 wt.% cobalt and nickel, cf. ~0.6–0.8 wt.% cobalt and 5–6 wt.% nickel in the original kamacite (Breen et al. 2016; Yang et al., 2014). In addition, there are 1–10 μ‎m grains of tetrataenite with 56% nickel in the intergrowth, even though tetrataenite grains in the original meteorite were <100 nm in size. These phases may have formed after mobilization of sulfur and post-shock annealing (Breen et al., 2016; Yang et al., 2014). Joegoldsteinite, MnCr₂S₄, which was found in a shocked sulfide intergrowth in a IVA iron, Social Circle, may also have formed from shock melt (Isa, Ma, & Rubin, 2016).

There is no comprehensive scheme for characterizing progressive stages of shock metamorphism for irons as there is for chondrites (e.g., Bischoff & Stöffler, 1992; Krot et al., 2014). This may be due in part to the highly localized shock melting at metal-troilite interfaces, steep post-shock thermal gradients, and very heterogeneous shock effects in iron meteorites. For example, shock-melted troilites and cloudy taenite intergrowth can be found in the same meteorite (e.g., Colomera). Three different schemes have been developed based on (a) Canyon Diablo specimens that experienced different levels of shock and deformation on impact with Earth (Buchwald, 1975), (b) diverse shock and heating effects in IVA and some ungrouped irons (Yang et al., 2011), and (c) shocked and unshocked group IIIE irons (Breen et al., 2016). The study by Yang et al. (2011) included heating effects on cloudy taenite, steep Ni profiles at taenite rims, and cases where the Widmanstätten pattern had been largely, or completely, obliterated by heating to 700–900°C.

Several unusually sulfur-rich iron meteorites are probably impact melts as they have dendritic or cellular metal-troilite textures indicative of rapid cooling during crystallization near the surface of their parent body (see Scott, 1982). Sahara 03505 is a small (65 g) iron that probably formed by impact melting on an ordinary chondrite body (Orazio, Folco, Chaussidon, & Rochette, 2009). Mundrabilla and Georgetown are larger sulfur-rich irons with coarser dendritic structures that probably formed as impact melts on the group IAB parent body, or a related body (Wasson & Kallemeyn, 2002). Mount Howe 88403 (Figure 14), Prospector Pool, and a few other sulfur-rich irons contain aligned rounded troilite nodules indicative of rapid cooling, probably on the group IIE parent body (Schrader et al., 2010; Wasson, 2017). In addition, Nedagolla is an ungrouped sulfur-poor impact melt (Buchwald, 1975, pp. 880–882). Tucson, which is the type meteorite for brezinaite, Cr₃S₄, is a remarkably unusual iron meteorite and probably also an impact melt (Buchwald, 1975; Nehru, Prinz, & Delaney, 1982). It is one of the few irons that lacks both troilite and daubréelite and contains ~8 vol.% of highly reduced silicates and glass (Kurat, Varela, Zinner, & Brandstätter, 2010). Tucson metal, like the silicates, is highly depleted in volatile elements and contains 0.9% Si, unlike most iron meteorites (Table 1).

Note that Lovina, which is listed in the Meteoritical Bulletin Database as a Ni-rich ungrouped iron meteorite and has a beautiful dendritic structure (Teplyakova, 2011), is unlikely to be a meteorite as nuclides produced by cosmic ray irradiation are present at very low levels (Nishiizumi & Caffee, 2011).

There are few shock and impact-heating events recorded by iron meteorites that can be dated with confidence. One probable exception is the intense heating event suffered by many of the IIE iron meteorites. Three IIE irons have Ar-Ar, Pb-Pb, and Rb-Sr ages of 3.6–3.7 Gyr probably due to a major impact on the parent body (Bogard, 2011; Ruzicka, 2014). The group IIIAB and IVA irons also have a significant number of shocked meteorites and both show major peaks in their cosmic-ray exposure ages (Wasson, 1985). Keil, Haack, and Scott (1995) suggested that these peaks at ~650 Myr for IIIAB and ~400 Myr for IVA record the impacts that shocked the irons. Yang et al. (2011) discussed several possible counterarguments and favored major shock damage before these events, but the issue deserves more study.

Studies of possible impact melting and heating in irons should also consider two other potential heat sources: solar heating due to close approach to the Sun during transit to Earth (Welten et al., 2014; Wittmann et al., 2011), and blacksmiths. Buchwald (1975, p. 41) discovered that no less than 18% of all iron meteorites had been reheated by blacksmiths. Such terrestrial mistreatment can be recognized by inspection of meteorite exteriors and the study of oxidized regions where metal and oxide have reacted. Since the IIE iron Netschaëvo shows these features (Buchwald, 1975, pp. 891–894), the silicate clasts in Netschaëvo that Van Roosbroek, Pittarello, Greshake, Debaille, and Claeys (2016) inferred were impact-melted also deserve further study. The supposedly impact-melted sulfides and phosphides in the Netschaëvo clasts are unlike those in heavily shocked IIE irons (Buchwald & Clarke, 1987; Olsen et al., 1994).

Crystallization and Chemical Trends Within Groups

To test whether the chemical trends in groups of iron meteorites were produced during fractional crystallization, equilibrium solid metal/liquid metal partition coefficients need to be determined experimentally for each element using iron alloys of the appropriate composition. If the distribution coefficients are constant during fractional crystallization, the compositions of solid samples will define a straight line on a logarithmic elemental plot with a gradient given by (k_A – 1)/(k_B – 1), where k_A and k_B are the partition coefficients for the two elements (Scott, 1972). However, partition coefficients are sensitive to the concentration of sulfur, which increases during crystallization as sulfur is almost insoluble in solid iron-nickel metal (Figure 15; see also Chabot, 2004; Chabot & Jones, 2003; Chabot, Wollack, McDonough, Ash, & Saslow, 2017). For example, the partition coefficient for iridium in an Fe-S melt with 31 wt.% sulfur (the Fe-FeS eutectic composition at 988°C) is a factor of 1,000 higher than it is for Fe at its melting point of 1,538°C (Jones & Drake, 1983). By parameterizing the experimentally determined partition coefficients as a function of the sulfur concentration in the liquid, the behavior of various elements during fractional crystallization can be modeled (Chabot et al., 2017; Ruzicka et al., 2017).

Open in new tab

Figure 15. Fe-FeS phase diagram showing complete solubility of S in the liquid and virtually no solubility between the two solid phases, pure iron and troilite (FeS) (Scheinberg, Elkins-Tanton, Schubert, & Bercovici, 2016). The group IIIAB core, for example, assuming it crystallized fractionally from a liquid with 12 wt.% sulfur (Figures 16e and 16f), would form almost S-free metal until at 988°C, about 40 wt.% of the original liquid with almost all the sulfur would be left to crystallize. Fractionally crystallized irons with more than 1 wt.% sulfur are not known.

Figure 16 shows how the iridium vs. gold and the germanium vs. gold variations in three groups of irons can be modeled using fractional crystallization. The first two plots (Figures 16a and 16b) show how various initial sulfur contents between 0 and 18 wt.% affect the calculated compositional trajectories of fractionally crystallized solids. Below these plots are the results of matching these curved paths to the data for groups IVB, IIIAB, and IIAB by adjusting the initial sulfur, gold, and germanium contents of the liquid. Note that changing the initial gold and germanium contents merely shifts a curve on the logarithmic plots without affecting its shape or orientation. The data for group IVB plot define straight lines and are well matched by fractional crystallization with sulfur contents between 0 and 2 wt.% (Campbell & Humayun, 2005; Walker et al., 2008). Analytical data for irons in groups IIIAB and IIAB are best matched by assuming initial sulfur contents of 12 and 18 wt.%, respectively. These sulfur contents provide a fair match for the 1,000-fold range of iridium contents in these groups as well as the reversal in the gradient on the Ge vs. Au plots. For comparison, H and L chondrites, if melted, would form Fe-Ni-S melts with 8.5 and 15 wt.% S, respectively, so these calculated sulfur contents appear plausible. For group IVB, the sulfur-free match for group IVB is consistent with the very low observed sulfur content in this group (0.04 wt.%; Buchwald, 1975, p. 83) and the very low concentrations of volatile siderophile elements (see the section “Chemical Composition”).

Open in new tab

Figure 16. (a–h). Logarithmic plots of Ir vs. Au (left) and Ge vs. Au (right). Figures (a) and (b) show the trajectories predicted by simple fractional crystallization models, using experimentally determined partition coefficients with bulk sulfur contents of 0–18 wt.%. Figures (c) and (d) show that a melt with no sulfur can successfully match the linear trends in group IVB. To match the trends in groups IIIAB (e and f) and group IIAB (g and h) requires initial sulfur contents of 12 and 18 wt.%, respectively.

Figure by Nancy Chabot; from Ruzicka et al. (2017).

Ideal fractional crystallization should produce a much narrower array of data for group IIIAB on the Ir-Ni plot, like that for group IIAB (Figures 16e and 16g). Trapping of melt may have been responsible (Wasson, 1999). Ideal fractional crystallization does not match the chemical trends in group IVA satisfactorily as different sulfur contents are required for the iridium vs. gold and germanium vs. gold plots, namely, 3 and 9 wt.%, respectively (Chabot, 2004; Goldstein et al., 2009a). Albarède, Bouchet, and Blichert-Toft (2013) found that modeling group IVA trends with fractional crystallization requires a “negative sulfur content” and they favored a partial melting model instead. However, non-ideal fractional crystallization involving, for example, liquid trapping or liquid immiscibility may be responsible (Ruzicka et al., 2017). Another approach to explain mismatches between calculated and observed trends is to allow the dependence of the partition coefficients with sulfur concentration to vary within a range consistent with the laboratory studies instead of choosing the best fit to the experimental data (e.g., Wasson, 2016). In this case, lower bulk sulfur contents can be used to match the iron meteorite data, for example, 2 wt.% instead of 12 wt.% for group IIIAB.

Fractional crystallization of molten metal should produce isotopic as well as chemical fractionation (Dauphas & Schauble, 2016). Hopp, Fischer-Gödde, and Kleine (2018) analyzed the ruthenium isotopic composition of irons in five groups of irons and found systematic mass-dependent fractionation in groups IIAB, IIIAB, and IVB, consistent with extraction of isotopically lighter ruthenium into solid iron-nickel. However, the isotopic systematics for groups IVA and IID were more complex. Hopp et al. (2018) inferred that early formed solids may have mixed with late-stage melts in both groups.

Two elements that appear to contradict the fractional crystallization model are the chalcophilic elements, copper and chromium (Chabot et al., 2009). Chromium appears to behave like iridium, although it should theoretically be concentrated in the liquid (Scott, 1972). The contradictory behavior of chromium may reflect the crystallization of chromite from molten metal (see the section “Massive Minerals That Crystallized From Molten Metal”; Chabot et al., 2009; Wasson et al., 2007).

A major unresolved issue with the fractional crystallization model for irons is the fate of the sulfur (Benedix et al., 2014; Chabot & Haack, 2006). Consider group IIIAB irons, for example. If these irons crystallized fractionally in a metallic liquid core which initially contained 12 wt.% sulfur (Figures 16e and 16f), almost sulfur-poor metal should have crystallized until at 988°C, about 40 wt.% of the original liquid with almost all the sulfur would have finally crystallized (Figure 15). Note that the relatively minor effects of nickel are ignored in this calculation. The mean sulfur content of group IIIAB irons from planimetry of large slices is ~0.8 wt.% (Buchwald, 1975). The complete absence of sulfur-rich irons in groups IIAB, IIIAB, and other magmatic groups is a major puzzle. One possible explanation is that residual sulfur-rich liquid was expelled from the core through impact-generated fractures by what Johnson, Sori, and Evans (2019) and Abrahams and Nimmo (2019) called ferrovolcanism. These authors did not discuss the absence of sulfur-rich, fractionally crystallized irons, but ferrovolcanism may be a plausible explanation.

The generally close match between the chemical compositional trends in magmatic irons and fractional crystallization models shows that the magmatic irons crystallized from large convecting bodies of molten metal (Figure 16). This match and the general absence of silicates in these irons provide the best evidence that these irons were probably derived from planetesimal cores. However, to fully understand the crystallization history of irons, physical models for crystallization of metallic cores are needed. Core crystallization is largely controlled by the liquidus temperature at which crystallization begins and the actual temperature gradient, which for an ideally convecting molten core is given by the adiabatic temperature gradient. High pressures depress both the liquidus temperature and the adiabatic temperature. In the Earth’s core, the adiabatic temperature gradient is shallower than the liquidus gradient, and solidification starts at the center. However, for asteroid-sized bodies, the reverse situation applies and core solidification probably starts at the core–mantle boundary (Chabot & Haack, 2006; Haack & Scott, 1992; Neufeld, Bryson, & Nimmo, 2019; Scheinberg et al., 2016; Williams, 2009). However, the change in the liquidus temperature across a pure iron core of radius 150 km, for example, is only ~2 degrees so there is more uncertainty about the direction and manner of solidification. Kinetic factors and initial conditions, for example, may have been more important than equilibrium relationships (Yang et al., 2010b). If the liquid is undercooled, solid metal may nucleate within the molten core, possibly on chromite grains (Wasson, 1993).

For fractional crystallization to operate, excess sulfur must be removed from the solidification front so that the liquid is stirred well. In an outward crystallizing planetesimal core, less dense sulfur-rich liquid from the solidification front would rise, generating compositional convection. However, in an inward crystallizing planetesimal core, sulfur build-up would probably have caused dendritic rather than concentric crystallization (Haack & Scott, 1992). Chemical variations in the multi-ton Cape York iron meteorites—two-fold variations in iridium and gold, for example—support dendrite formation in the IIIAB core (Esbensen, Buchwald, Malvin, & Wasson, 1982). Dendrites that grew inward may also have been detached and fallen inward, aiding convection and generating a solid inner core (Scheinberg et al., 2016; Wasson, 2016).

Why Did Some Asteroids Melt to Form Metallic Bodies?

Urey (1955) was the first to suggest that small bodies in the solar system may have been melted by the radionuclide ²⁶Al, which has a half-life of ~0.72 Myr. Although other potential heat sources have been explored, the case for heating by ²⁶Al has become increasingly robust (Hevey & Sanders, 2006; Kleine et al., 2009; McSween, Ghosh, Grimm, Wilson, & Young, 2002; Scott et al., 2015). Heating due to ⁶⁰Fe was negligible (Trappitsch et al., 2018), and impacts did not play a major role in heating asteroids (e.g., Wilson, Bland, Buczkowski, Keil, & Krot, 2015). However, impacts can cause localized melting, especially in porous bodies (Davison, Collins, & Ciesla, 2010). Short-lived radioisotopes like ²⁶Al can be made in supernova explosions and in the Wolf-Rayet winds that preceded them, which may have triggered the collapse of the parent molecular cloud to form the Sun and its disk (Davis & McKeegan, 2014; Dwarkadas, Dauphas, Meyer, Boyajian, & Bojazi, 2017; Fujimoto, Krumholz, & Tachibono, 2018).

The initial ratio of ²⁶Al to the stable isotope ²⁷Al in the protoplanetary disk is taken to be 5.2 × 10⁻⁵ (Davis & McKeegan, 2014; Krot, 2019). This value was derived from isotopic analysis of CAIs, which have a Pb-Pb age of 4,567.3 ± 0.2 Myr and are the oldest objects that formed in the disk (Connelly et al., 2012). The assumption that ²⁶Al was homogeneously distributed in the disk has been questioned by Larsen et al. (2011), who argued on the basis of Mg isotopic heterogeneity that the ²⁶Al/²⁷Al ratio was as low as 1 × 10⁻⁵ outside the CAI-forming region. Kruijer, Kleine, Fischer-Gödde, Burkhardt, and Wieler (2014b) and Kleine and Wadhwa (2017), among others, provided persuasive arguments favoring ²⁶Al homogeneity, but the issue is not resolved (e.g., Connelly, Schiller, & Bizzarro, 2019). Assuming ²⁶Al homogeneity in the disk, material with the chemical composition of dry chondrites would have initially contained about four times the amount of ²⁶Al needed for complete melting (Figure 17). Bodies that accreted 1.4 Myr later (two half-lives of ²⁶Al) were still able to melt entirely and form metallic cores. However, 2 Myr after CAI formation (three half-lives later), there was insufficient heat for melting. Note that heat loss by conduction on these timescales would be precluded by >10 km of insulating silicate, so these estimates apply at depths of >10 km.

Open in new tab

Figure 17. Upper diagram shows the exponential decay with time of the thermal energy available in dry pristine chondritic dust from the decay of ²⁶Al, which has a half-life of 0.7 Myr. The four bold vertical lines show the heat available after successive half-lives. The lower schematic plot shows how the thermal evolution of bodies depended on their time of accretion. The vertical axis, which has an inverted logarithmic scale, shows the energy available from ²⁶Al at the time of accretion. Bodies labeled A and B, which accreted at time zero when CAIs formed and one-half life later, would have had fully molten, convecting silicate magma oceans and metallic cores at 0.3 Myr and 1.5 Myr, respectively. Body C, which accreted two half-lives after CAIs, would have contained just enough ²⁶Al to provide 1.6 kJ/g of thermal energy and cause complete melting at ~5 Myr. Body D, which accreted three half-lives (2.2 Myr) after CAI formation, would have remained solid. Formation ages for iron meteorites and chondrites are consistent with the thermal model shown here.

Figure from Sanders & Scott (2019).

When Did Asteroids Form Metallic Cores?

The best constraints on the time of core formation in asteroids come from Hf-W isotopic systematics for the magmatic irons (Kleine, Mezger, Palme, Scherer, & Münker, 2005; Kleine et al., 2009). ¹⁸²Hf decays to ¹⁸²W with a half-life of 8.9 ± 0.1 Myr. Processes that fractionate Hf from W when ¹⁸²Hf is still present will therefore produce products with diverse relative abundances of ¹⁸²W. Note that these are commonly expressed in ${\text{ε}}^{182}\text{W}$ units, which are defined as deviations in the ¹⁸²W/¹⁸⁴W ratio from the terrestrial standard in parts per 10⁴. Since both Hf and W are refractory elements, the Hf/W ratio in disk solids and in solid planetesimals is rather uniform. However, when planetesimals melted, molten metal drained inward from the silicate mantle to form a metallic core. Hafnium was retained in the silicate mantle, whereas tungsten, which is generally siderophile, entered the core. The ¹⁸²W/¹⁸⁴W ratio in the metallic core was then frozen when it formed. The time interval between CAI and core formation, which is assumed to be instantaneous, can then be estimated using the initial values for ¹⁸²W/¹⁸⁴W and ¹⁸²Hf/¹⁸⁴Hf obtained from CAIs (Kruijer et al., 2014b; see also Table 5).

When model Hf-W ages were first determined, some iron meteorites were found to have ${\text{ε}}^{182}\text{W}$ values that were lower than the CAI value, suggesting, incorrectly, that they formed before CAIs. However, irons with the lowest ${\text{ε}}^{182}\text{W}$ values in a group were found to have the longest cosmic-ray exposure ages, suggesting that ${\text{ε}}^{182}\text{W}$ values were lowered by irradiation with galactic cosmic rays which generate neutrons (Kleine et al., 2009). The effects of cosmic rays on ${\text{ε}}^{182}\text{W}$ values can be eliminated by determining the platinum or osmium isotopic composition in the same sample that was analyzed for W isotopes (Kruijer et al., 2013; Wittig, Humayun, Brandon, Huang, & Leya, 2013; Worsham et al., 2017). A plot of ${\text{ε}}^{182}\text{W}$ vs. ${\text{ε}}^{196}\text{Pt}$ for irons in the same group can be extrapolated to give a zero-exposure (or pre-exposure) ${\text{ε}}^{182}\text{W}$ value for the group (Kruijer et al., 2014a, 2017). Core formation times for 11 magmatic groups of iron meteorites determined from zero-exposure ${\text{ε}}^{182}\text{W}$ values based on this simple model are shown in Table 5 and Figure 18 (Hilton et al., 2019; Kruijer et al., 2017). Cores in the NC reservoir formed between 0.3 and 1.8 Myr after CAI formation. In the CC reservoir, they formed a little later at 2.1 to 2.8 Myr. (Non-magmatic groups of iron meteorites have younger Hf-W ages and are discussed in the section “Non-Magmatic Iron Meteorites.”)

Open in new tab

Figure 18. ${\text{ε}}^{182}\text{W}$ values and Hf-W model ages for magmatic and non-magmatic iron meteorites belonging to carbonaceous (CC) and noncarbonaceous (NC) reservoirs (Kruijer & Kleine, 2019). (1 $\text{ε}$ unit represents a 0.01% deviation in the isotopic ratio of a sample relative to the terrestrial standard.) These model ages are corrected for cosmic-ray irradiation. Magmatic irons have Hf-W model ages of 0.3–2.8 Myr after CAI, which date metal-silicate separation at core formation. The Hf-W model ages for IAB irons and four IIE irons are 3–5 Myr. Younger ages of 10–30 Myr for about half the IIE irons point to a more extended impact history for this group. Note that subgroups sHL and sHH of Wasson and Kallemeyn (2002) are isotopically distinct from IAB irons.

Data from Kruijer and Kleine (2019), Kruijer et al. (2017), Worsham et al. (2017), and Hilton et al. (2019). Figure from Kruijer and Kleine (2019).

Concerned that some iron meteorites had zero-exposure ${\text{ε}}^{182}\text{W}$ values comparable to the initial values inferred for CAIs, Humayun, Simon, and Grossman (2007) questioned whether the initial Hf and W isotopic compositions of the solar system inferred from CAIs were correct. They argued that the isotopic data for CAIs in the Allende chondrite had been affected by parent body metamorphism. Kruijer et al. (2014b) investigated these claims carefully as hydrothermal alteration effects are ubiquitous in Allende. Since the bulk compositions of CAIs plotted coherently on the Hf-W isotopic diagram used to define the initial CAI isotopic compositions, Kruijer et al. (2014b) concluded that hydrothermal alteration had not affected the tungsten isotopic compositions of CAIs. Sanders and Scott (2019) disagreed, suggesting that the tungsten isotopic composition of fine-grained CAIs had been modified during alteration by the addition of a nucleosynthetic tungsten isotopic component from the matrix. Nevertheless, they inferred that the procedure Kruijer et al. (2014b) used to correct for nucleosynthetic tungsten isotopic effects appeared to have eliminated possible errors due to alteration.

Support for the antiquity of IVA iron meteorites is provided by the uranium-isotope corrected ²⁰⁶Pb-²⁰⁷Pb ages. These ages are based on the long-lived isotopes ²³⁵U and ²³⁸U and do not require any assumption about initial isotopic homogeneity in the disk. The U-isotope corrected Pb-Pb age for orthopyroxene in the Steinbach IVA iron meteorite is 4,565.5 ± 0.3 Myr (Connelly et al., 2019). This age shows that the orthopyroxene cooled below the closure temperature for lead isotopic diffusion 1.8 ± 0.3 Myr after CAI formation. An uncorrected Pb-Pb age of 4,565.1 ± 0.1 Myr for another IVA iron meteorite, Muonionalusta, was reported by Blichert-Toft, Moynier, Lee, Telouk, and Albarède (2010). This age was revised downward by 3–7 Myr by Brennecka, Amelin, and Kleine (2018), who found a range of U isotopic compositions in troilite nodules in Muonionalusta. However, Connelly et al. (2019) attributed this range of U isotopic compositions to an analytical artefact and recalculated an age for Muonionalusta of 4,564.1±2.6 Myr (or 3.2±2.6 Myr after CAI), using their U isotopic measurements for Steinbach and other meteorites. Thus, the Pb-Pb data for IVA irons support the early formation of a metallic core in their parent body 1.5±0.6 Myr after CAI formation (Table 5).

To appreciate the considerable recent improvements in the formation ages of iron meteorites provided by the Hf-W and Pb-Pb isotopic data, note that the solidification ages of iron meteorites were previously determined using the ¹⁸⁷Re-¹⁸⁷Os isotope system (Shen, Papanastassiou, & Wasserburg, 1996; Smoliar, Walker, & Morgan, 1996). The Re-Os data suggested, for example, that IVA irons were older than IIAB irons by 60 ± 45 Myr (Shen et al., 1996). Given the large (±3%) uncertainty in the decay constant for ¹⁸⁷Re, and the greater precision and accuracy of the Hf-W and Pb-Pb ages, Re-Os ages for iron meteorites are less useful for dating solidification.

Silicates in iron meteorites have been dated by a variety of other radiometric techniques, including the short-lived I-Xe (e.g., Pravdivtseva, Meshik, Hohenburg, & Kurat, 2013) and Hf-W chronometers, and long-lived K-Ar (Bogard, 2011), Rb-Sr, and Sm-Nd chronometers. Phosphates and metal have also been dated using the short-lived Mn-Cr (Sugiura & Hoshino, 2003) and Pd-Ag chronometers (Theis et al., 2013), respectively. See Goldstein et al. (2009a), Ruzicka (2014), and Kleine & Wadhwa (2017) for overviews.

Accretion of Iron Meteorite Parent Bodies

The time of accretion for the parent body of an iron meteorite group can be estimated from the core formation age using thermal models for bodies heated by ²⁶Al. The assumptions include those used to derive Figure 17 and the model Hf-W core formation ages discussed in the section “When Did Asteroids Form Metallic Cores?” (see Hilton et al., 2019; Kruijer et al., 2017). Accretion times for iron meteorites from the NC reservoir are earlier than for irons from the CC reservoir: ~0.5 Myr vs. ~1 Myr after CAIs (Table 5).

For many years, it was thought that chondrites formed before the parent bodies of differentiated meteorites as they are clearly more primitive. However, chondrule formation ages inferred from Al-Mg isotope systematics for chondrules are mostly 2–4 Myr after CAIs (Kruijer et al., 2019; Krot, 2019), supporting the prediction from thermal models that chondrites accreted >1.5 Myr after CAIs (Figure 17). At first glance, this conclusion appears inconsistent with the statement that differentiated meteorites formed by melting of chondritic material. However, chondritic materials are merely disk solids that have a bulk chemical composition close to that of the Sun (minus the highly volatile elements).

Were chondrules present in the chondritic material that accreted to form the parent bodies of the magmatic iron meteorites? This seems plausible as a few chondrules have Pb-Pb ages that are almost as old as CAIs (see Connelly et al., 2019) and some CAIs were melted in the disk soon after they formed (Krot, 2019). Chondrules may also have been essential for planetesimal accretion (Johansen, Mac Low, Lacerda, & Bizzarro, 2015). However, most chondrules formed 2–4 Myr after CAIs (Krot, 2019) when the iron meteorite parent bodies had already melted. Thus, abundant chondrules may not have formed until the earliest planetesimals had melted. Shock waves created by large planetesimals may have melted dust in the disk (Morris, Boley, Desch, & Athanassiadou, 2012). Alternatively, collisions between molten bodies may have sprayed out silicate melt droplets (e.g., Asphaug, Jutzi, & Movshovitz, 2011; Sanders & Scott, 2017).

Where Did the Iron Meteorites Form?

Meteorites and their parent asteroids are generally thought to have formed in the asteroid belt (e.g., Anders, 1964). However, Bottke, Nesvorný, Grimm, Morbidelli, and O’Brien (2006) argued that since the iron meteorite parent bodies accreted before the chondrites, they probably formed in the terrestrial planet region where dust densities were higher and orbital periods were shorter. They suggested that protoplanets in the terrestrial planet region caused collisional destruction of the differentiated planetesimals and scattered their fragments into the belt. This would help explain the dominance of chondritic bodies in the belt and the lack of intact differentiated bodies (besides Vesta). Diverse cooling rates and the early removal of mantles from several meteorite parent bodies appeared to be consistent with this model (Scott et al., 2015). However, recent isotopic studies suggest that only a few differentiated asteroids come from the terrestrial planet region.

Warren (2011a, 2011b) first promoted the idea that meteorites can be divided into two populations based on mass-independent isotopic analyses (see the section “Nucleosynthetic Isotopic Variations”). He demonstrated that variations in ${\text{Δ}}^{17}\text{O}$, ${\text{ε}}^{50}\text{Ti}$, ${\text{ε}}^{54}\text{Cr}$, and ${\text{ε}}^{62}\text{Ni}$ among chondrites and achondrites define two distinct populations: carbonaceous chondrites and a couple of differentiated meteorites in one (CC), and non-carbonaceous chondrites and differentiated meteorites in the other (NC). Figure 19 shows an updated version of Warren’s ${\text{Δ}}^{17}\text{O}$ vs. ${\text{ε}}^{54}\text{Cr}$ plot. Warren (2011a, 2011b) tentatively suggested that carbonaceous chondrites, which are richest in water and volatiles, may have formed in the outer solar system beyond Jupiter, and that the other chondrites and nearly all achondrites formed in the inner solar system. A possible mechanism for mixing the CC and NC populations was provided by the Grand Tack model of Walsh, Morbidelli, Raymond, O’Brien, and Mandell (2011), who envisaged that Jupiter migrated through the belt scattering its contents until Saturn caught up causing the two planets to be locked in a periodic resonance. Both planets then reversed their migration directions so that Jupiter repopulated the belt with a mixture of CC and NC bodies. An alternative mechanism for mixing the CC and NC bodies together is scattering of planetesimals by Jupiter and Saturn when they formed (Raymond & Izidoro, 2017).

Open in new tab

Figure 19. Plot of ${\text{Δ}}^{17}\text{O}$ vs. ${\text{ε}}^{54}\text{Cr}$ for chondrites, achondrites, differentiated meteorites including two groups of irons, two groups of pallasites, and planets showing a very clear dichotomy between the carbonaceous and non-carbonaceous populations. Note that ${\text{Δ}}^{17}\text{O}$ is the vertical displacement form the terrestrial fractionation line on the oxygen three-isotope plot. One $\text{ε}$ unit represents 0.01% deviation from the terrestrial standard. Carbonaceous chondrites and a few differentiated meteorites plot on the right; the other population on the left includes all other chondrites and differentiated meteorites, the Earth, Moon, and Mars. The two populations are thought to have formed on either side of Jupiter.

For sources of data, see Scott et al. (2018).

The surprising discovery that iron meteorites could also be divided into two isotopic populations (CC and NC) greatly strengthened the case that Jupiter may have separated the two populations. Five groups of irons, IIC, IID, IIF, IIIF, and IVB, were found to have isotopic compositions that matched those of the carbonaceous chondrite reservoir. The remaining groups of irons plotted with the non-carbonaceous meteorites (Budde et al., 2016; Kruijer et al., 2017). The evidence was largely derived from the ${\text{ε}}^{95}\text{Mo}$ vs. ${\text{ε}}^{94}\text{Mo}$ plot (see the section “Nucleosynthetic Isotopic Variations” and Figure 6), but also from ${\text{ε}}^{183}\text{W}$, ${\text{ε}}^{100}\text{Ru}$, and ${\text{μ}}^{58}\text{Ni}$ data (Hilton et al., 2019; Nanne, Nimmo, Cuzzi, & Kleine, 2019; Poole et al., 2017; Yokoyama Nagai, Fukai, & Hirata, 2019). Grouped and ungrouped irons in the CC population tend to be richer in Ni and refractory siderophiles than the NC irons, but there is overlap between their compositions (Spitzer et al., 2019; Rubin, 2018). Interestingly, the four largest groups of irons all belong to the NC population, but four of the ten smaller groups, and most of the ungrouped irons, belong to the CC population (Table 2). Pallasites are also divided by isotopic analyses into carbonaceous (Eagle Station types and the ungrouped Milton) and non-carbonaceous (main group) and follow the same trends (Figure 19). Conceivably, CC parent bodies were generally smaller than NC bodies when they were finally emplaced in the asteroid belt.

If Jupiter did separate the two isotopic populations, the five groups of irons and the ungrouped irons in the CC population formed beyond Jupiter, while the NC irons formed inside Jupiter’s orbit. Assuming that S-type asteroids formed in the asteroid belt (Walsh et al., 2011), it is plausible that groups IIE and IVA did also, given their proximity in isotopic space to ordinary chondrites, which are linked to S-types. Group IAB irons have Mo isotopic compositions that are identical to Earth (Worsham et al., 2017), so they may have been scattered into the belt from the innermost solar system, as Bottke et al. (2006) argued for all irons. The highly reduced iron, Horse Creek, is chemically related to the enstatite achondrite and chondrites (e.g., Rubin, 2015), which are also isotopically related to the Earth (see Carlson, Brasser, Yin, Fischer-Gödde, & Qin, 2018), so it may also have formed in the innermost solar system.

Since the mechanisms for generating or preserving nucleosynthetic and ${\text{Δ}}^{17}\text{O}$ variations in meteorites are not well understood, conclusions drawn from isotopic data about formation locations of meteorites are tentative. Nevertheless, they do provide a good working hypothesis for helping understand meteorite and asteroid properties, which can be readily tested with more data.

Evidence From Asteroids

Before discussing how specific groups of irons formed, the article reviews what can be inferred about the parent bodies of iron meteorites and other differentiated meteorites from studies of asteroids from Earth and spacecraft and theoretical studies of asteroid disruption.

The asteroid belt is dominated by dark C-type asteroids, which are related spectrally to carbonaceous chondrites, and bright S-types, which are linked to ordinary chondrites (Burbine, 2014). The only asteroid thought to have a metallic core and a basaltic surface is Vesta, the second most massive asteroid, which is the parent body of howardite, eucrite, and diogenite achondrites (Righter & Drake, 1996; Fu et al., 2012; Mittlefehldt, 2015; Raymond, Russell, & McSween, 2017). The asteroid long considered as the typical differentiated asteroid is therefore unique! What happened to the parent bodies of all the other differentiated meteorites? Three types of asteroids are probably composed of differentiated materials.

Many small basaltic asteroids called V-types have been identified that are unrelated to Vesta (and its family) and are <10 km in diameter (Burbine, 2014). A-type asteroids are very largely composed of olivine (>80%) and are thought to be mostly derived from differentiated bodies. However, A-types are typically small and are scattered across the asteroid main belt. Altogether A-types account for <0.2% of the total mass of asteroids >2 km in diameter (DeMeo et al., 2019). The most likely sources for iron meteorites are the M-type asteroids, which have largely featureless spectra (e.g., Burbine, 2014; Scott et al., 2015). However, some M-types show spectral signatures of hydrated silicates so they probably have diverse compositions (Rivkin, Howell, Lebovsky, Clark, & Britt, 2000). Some larger M-types have high radar albedos, confirming that they are metal-rich, but many of these also show weak silicate features (Ockert-Bell et al., 2010; Shepard et al., 2015). Metal-rich M-type asteroids may be coated with a fine silicate regolith or silicate glass produced by impacting silicate-rich projectiles (Libourel et al., 2019).

The largest M-type asteroid, 16 Psyche, which is about 280 × 232 × 190 km in size, is the target of a planned NASA Discovery mission (Elkins-Tanton et al., 2017; Shepard et al., 2017). Psyche was initially thought to be a bare planetesimal core (Figure 20). However, Psyche, like some other M-types, has fine silicate regolith (Landsman et al., 2018) and its density of 4.0 ± 0.3 g cm⁻³ is about half that of solid metal (Viikinkoski et al., 2018). Thus, Psyche may be a porous pile of metal fragments from a core like that shown in Figure 20, or it may be the parent body of meteorites such as mesosiderites.

Open in new tab

Figure 20. Artist rendition of the metal-rich asteroid, 16 Psyche, which is the target of an upcoming NASA mission. Psyche, like other large M-type asteroids, is not likely to be a mantle-stripped metallic body as envisaged here as its density is too low (Viikinkoski et al., 2018). However, it may be highly porous piles of core fragments. The mission to Psyche should help us understand the nature of iron meteorite parent bodies and metal-rich asteroids.

Image: Arizona State University; Peter Rubin.

Despite considerable efforts to identify the parent asteroids of iron meteorites, we cannot make definitive statements about their nature. The only known metallic asteroidal core is deeply buried inside Vesta and is not the source of any iron meteorites. Given what is known about the olivine-rich A-types and the basaltic V-type asteroids, it is plausible that the iron meteorites come from numerous relatively small M-type asteroids, which are scattered throughout the main belt (DeMeo & Carry, 2014). Unless they are disguised with chondritic surfaces, which seems rather unlikely (see the section “Did Iron Meteorite Parent Bodies Have Chondritic Surfaces?”), asteroids composed of differentiated materials are much rarer than chondritic asteroids. Aside from Vesta, the non-chondritic asteroids are probably miscellaneous small fragments from a much larger original population of differentiated planetesimals.

Formation of Iron Meteorites