Protracted core formation and rapid accretion of protoplanets

See allHide authors and affiliations

Science  06 Jun 2014:
Vol. 344, Issue 6188, pp. 1150-1154
DOI: 10.1126/science.1251766

The chronology of planetary embryos

Protoplanets, or early planetary embryos such as iron meteorite parent bodies, formed in the early protoplanetary disk from dust, debris, and planetesimals. Defining the precise chronology of accretion and differentiation—including core formation—of these planetary embryos will aid in a richer understanding of the chemical evolution of the solar system. Through high-precision tungsten isotope measurements, Kruijer et al. show that the timing of accretion and core formation for iron meteorite groups falls within 0.6 to 2 million years of the age of the solar system (see the Perspective by Elliott). Differences of timing within this group are probably a function of volatile contents of the parent bodies or spatial and chemical heterogeneity within the protoplanetary disk.

Science, this issue p. 1150; see also p. 1086


Understanding core formation in meteorite parent bodies is critical for constraining the fundamental processes of protoplanet accretion and differentiation within the solar protoplanetary disk. We report variations of 5 to 20 parts per million in 182W, resulting from the decay of now-extinct 182Hf, among five magmatic iron meteorite groups. These 182W variations indicate that core formation occurred over an interval of ~1 million years and may have involved an early segregation of Fe-FeS and a later segregation of Fe melts. Despite this protracted interval of core formation, the iron meteorite parent bodies probably accreted concurrently ~0.1 to 0.3 million years after the formation of Ca-Al–rich inclusions. Variations in volatile contents among these bodies, therefore, did not result from accretion at different times from an incompletely condensed solar nebula but must reflect local processes within the nebula.

Magmatic iron meteorites are generally considered to sample the metal cores of differentiated protoplanetary bodies that formed after the segregation and subsequent crystallization of metallic melts (1). Each of the magmatic iron meteorite groups represents metal from a distinct parent body. The groups are primarily distinguished by different contents of moderately volatile elements such as Ga and Ge, relative to Ni. The order-of-magnitude variations in volatile depletion probably arise from chemical fractionations induced by processes within the solar nebula, suggesting that the conditions of parent body accretion varied in time and/or space.

Precise determination of the timing of core formation for each magmatic iron meteorite group is critical for determining the accretion rate for each parent body, for identifying the heat sources responsible for melting and differentiation, and for assessing whether the timing of accretion played a role in establishing the different degrees of volatile depletion observed in the magmatic iron meteorite groups. The extinct 182Hf-182W chronometer [half-life (t1/2) = 8.9 million years (My)] is ideally suited to precisely constrain the time of core formation, but its application to iron meteorites has been hampered by cosmic ray–induced neutron capture effects on W isotope compositions (25). Thus, although previous studies have shown that core formation in most iron meteorite parent bodies probably occurred within ~2 My after the formation of Ca-Al–rich inclusions (CAIs), it has not yet been possible to resolve differences in the timing of accretion and core formation (6).

We used Pt isotope compositions to quantify and correct measured W isotope compositions of iron meteorites for the effects of neutron capture (7, 8) and obtained combined high-precision Pt and W isotope data for metal samples from the major iron meteorite groups (IIAB, IID, IIIAB, IVA, and IVB) (9). All iron meteorite groups exhibit well-defined empirical ε182W-ε196Pt correlations, whose intercepts provide pre-exposure ε182W (the 182W/184W unbiased by galactic cosmic rays) for each group (Fig. 1 and table S1). Our results reveal small but resolvable differences in pre-exposure ε182W among magmatic iron meteorite groups, with the IID iron meteorites having the highest pre-exposure ε182W of −3.15 ± 0.07 (±95% confidence) and the IIAB iron meteorites having the lowest value of −3.40 ± 0.03. Model ages of metal segregation in the iron meteorite parent bodies, relative to the formation of CAIs, can be calculated as the time of Hf/W fractionation from an unfractionated reservoir with chondritic 180Hf/184W of 1.28 ± 0.03 (10, 11). At face value, the distinct pre-exposure ε182W values yield resolved and very precise Hf-W ages spanning a total range of ~0.7 to ~2.9 My after CAI formation (table S6).

Fig. 1 ε182W versus εiPt for the major magmatic iron meteorite groups.

(A to F) IIAB, IIIAB, IVA, IVB, and IID iron meteorites, respectively. ε182W (6/4) and ε19iPt (8/5) are 0.01% deviations from the terrestrial 182W/184W ratios (normalized to 186W/184W, denoted 6/4) and iPt/195Pt ratios (normalized to 198Pt/195Pt, denoted 8/5). Solid lines are best-fit regressions through the data with their 95% confidence envelopes (dashed lines) and pre-exposure ε182W intersecting the ordinate at εiPt = 0. W isotope analyses were performed using multicollector inductively coupled plasma mass spectrometry (solid symbols) or thermal ionization mass spectrometry (open symbols). Error bars represent external uncertainties (2 SD for Pt and 95% confidence for W). The investigated IID iron meteorites Carbo and Rodeo have nearly identical Ir/Pt, so in this specific case, ε182W (6/4) versus ε192Pt (8/5) also show a well-defined correlation (Fig. 1F), providing an additional precise estimate of the pre-exposure ε182W (7). Small downward corrections for nucleosynthetic heterogeneity have been made to the IID and IVB data points (7, 9).

With the exception of the IID iron meteorites, the pre-exposure ε182W values exhibit inverse correlations with Ga/Ni and S contents, estimated for the bulk compositions of each core (Fig. 2). This suggests that the degree of volatile element depletion exerted some control on the timing of core formation. The inverse correlation of core formation model age with the degree of volatile depletion is remarkable because it is opposite to what might be expected for the accretion time of iron meteorite parent bodies. Volatile-poor bodies (such as IVA and IVB) would be expected to have accreted earlier, at a time when the solar nebula was yet not fully condensed, in comparison to more volatile-rich bodies (such as IIAB) (12, 13). Our data, however, suggest that core formation in the IVA and IVB iron meteorite parent bodies occurred later than in the IIAB parent body (Fig. 2 and table S1). This may indicate that the IVA and IVB, as well as the IID, parent bodies accreted later than or over a longer period of time than the IIAB parent body. However, linking the time of core formation to an age of accretion requires knowledge of the temperature at which melting and metal segregation occurred. This temperature is largely controlled by the S content of the iron meteorite parent bodies (14, 15), and thus was different for each body. The IIAB iron meteorites have the highest S content and hence the lowest liquidus temperature of ~1330°C, whereas the IVB iron meteorites exhibit the lowest S content and highest liquidus temperature of ~1600°C. The inverse ε182W-versus-S correlation observed among the magmatic iron meteorites therefore may primarily reflect different melting temperatures of the metal within their parent bodies.

Fig. 2 ε182W versus volatile element ratios and bulk S contents.

(A) Pre-exposure ε182W versus CI chondrite–normalized Ga/Ni. (B) Pre-exposure ε182W versus inferred S concentration (weight %) in the core (9). Error bars on ε182W represent 95% confidence limits of the mean. The Ge, Ga, and Ni concentrations are from (19) and references therein.

To quantify the relation between accretion age and time of melting and metal segregation, we consider two end-member models for core formation (9). After the initial accretion, the still undifferentiated iron meteorite parent bodies probably consisted of an unequilibrated mix of different components, including metallic Fe, FeS, and silicates. Upon heating, due primarily to the presence of 26Al, melting probably began at the Fe-FeS eutectic temperature (~1000°C at 1 atm) (14) (Fig. 3). The first model assumes that this Fe-FeS melt did not segregate, but that heating continued to the liquidus temperature of the core, and only then did the metal melt segregate to form the core. In this model, the observed ~1-My time difference in core formation model ages between the IIAB and IVA parent bodies is consistent with the time required to raise the temperature inside the parent body from the liquidus temperature of the IIAB to that inferred for the IVA iron meteorites (Fig. 3). Despite their different core formation ages, the IIAB, IIIAB, and IVA iron meteorite parent bodies, therefore, could have accreted within a narrow time interval, between ~0.1 and ~0.3 My after the formation of CAIs. However, for a given accretion time, this model cannot explain the higher ε182W of the IID and IVB iron meteorites, which plot to the right of the heating curves of the IIAB, IIIAB, and IVA groups (Fig. 3).

Fig. 3 Internal temperature versus time after CAI formation for iron meteorite parent bodies assuming a single event of metal segregation.

Solid symbols show the model ages of metal segregation inferred for the iron meteorite parent bodies. Also shown are model results for postaccretional internal heating by 26Al decay of a spherical protoplanet with a radius of 40 km (9). Solid curves show the temperature evolution at half the radius (i.e., at 20 km depth) in planetesimals accreted at 0.1 to 0.3 My (hashed area) after CAI formation, and for 0.6 My in the case of the IVB iron meteorites (dark blue curve). Horizontal dashed lines show the eutectic melting temperature in the Fe-FeS system and that of pure Fe at atmospheric pressure.

The second model assumes that, as a consequence of their high densities, the Fe-FeS eutectic melts rapidly segregated by permeable flow to form cores (16). The parent bodies subsequently continued to heat, eventually leading to the melting of silicates and finally of pure Fe metal at ~1600°C (14). Therefore, in this model, core formation and metal-silicate separation began with eutectic melting of Fe-FeS but probably did not resume until pure Fe metal melted at higher temperatures (Fig. 3). Because melting was a multistage process occurring over a period of time, the early-segregated Fe-FeS melts would have had less radiogenic ε182W than the later-segregated pure Fe metal melts. In addition, as the early-segregated Fe-FeS melts removed some W to the initial cores, the residual mantles would have developed suprachrondritic 180Hf/184W, potentially leading to considerably higher ε182W in the mantles over a short period of time. The final W isotope compositions of the metal cores therefore would reflect particular mixes of early- (having lower ε182W) and late- (having higher ε182W) segregated metal fractions. Consequently, for different bodies with uniform formation ages, the metal cores of S-rich bodies would be characterized by lower ε182W as compared to S-poor bodies, because they would contain a larger fraction of the early-segregated Fe-FeS melt.

We modeled the 182W evolution in the iron meteorite parent bodies, accounting for the 180Hf/184W in the mantles after a first melt extraction (Fig. 4A). For an accretion time of 0.25 My, our thermal model predicts that the first Fe-FeS melts formed at ~0.7 My (t1 in Fig. 4) and that the melting temperature of pure Fe metal was reached ~0.6 My later, at ~1.3 My (t2 in Fig. 4). For an earlier accretion time, at ~0.1 My, the difference between t1 and t2 becomes much smaller, leaving too little time for the generation of a significant 182W difference between early- and late-segregated metal. For accretion times later than ~0.3 My, the onset of melting is too late to explain the low ε182W of –3.40 ± 0.03 of the IIAB iron meteorites. The 180Hf/184W of the mantle after extraction of an Fe-FeS melt at t1 is obtained by estimating the fraction of Fe metal melted at the eutectic, using the S content of each iron meteorite group (9). The 180Hf/184W values of the mantles reached after a first melt extraction are different for each parent body, because the amount of early-segregated Fe-FeS melt decreases with decreasing S content of the bulk core. Thus, the mantles of S-rich bodies (such as IIAB and IID) initially have higher 180Hf/184W than those of S-poor bodies (such as IVA) (Fig. 4A). After extraction of a Fe-FeS melt, the mantles of the iron meteorite parent bodies evolved to ε182W values between ~ –3.3 and –3.1 until the melting temperature of pure Fe metal was reached (Fig. 4A). At that point, all Fe metal segregated to the core and was mixed with the earlier-segregated Fe-FeS melts. The IIAB iron meteorites have a very low ε182W of –3.40 ± 0.03 that is only slightly elevated as compared to the solar system initial value of –3.48 ± 0.06, indicating that melting and metal segregation in the IIAB iron meteorite parent body must have started very early. The IIAB composition, therefore, is dominated by the early-segregated Fe-FeS melt and as such represents one end member in the mixing model. The mixing of early- and late-segregated metal to form bulk iron cores illustrates that the variable ε182W, as well as the inverse ε182W–versus-S correlation of the magmatic iron meteorites, can be reproduced (Fig. 4B). The mixing model also reproduces the offset of the IID iron meteorites from the ε182W–versus-S correlation, which reflects the high 180Hf/184W of the IID mantle after extraction of an early Fe-FeS melt.

Fig. 4 History of core formation for a two-stage metal segregation process.

(A) ε182W evolution diagram illustrating the effect of a two-step metal segregation history of the IIAB, IIIAB, IVA, and IID parent bodies, shown for an accretion time of 0.25 My after CAI formation. For a given accretion time, the thermal model constrains the time interval (hashed area) of melting and core formation between the early (t1) and late (t2) metal segregation step. Dashed lines show modeled ε182W curves for different 180Hf/184W ratios of the IID, IIIAB, and IVA residual mantles after early core segregation at t1, represented by the IIAB iron meteorites. The modeled ε182W compositions of the late-segregated (low-S) core fractions at t2 are also plotted. (B) Diagram of ε182W versus S for the IIAB, IID, IIIAB, and IVA iron meteorite groups with mixing curves demonstrating that the pre-exposure ε182W of the IID, IIIAB, and IVA iron meteorites can be explained by mixing of (i) an early- (high-S) segregated melt, represented by the IIAB iron meteorites, and (ii) the modeled ε182W compositions of the residual mantles, which supplied a late- (low-S) segregated melt.

Only the IVB iron meteorites appear inconsistent with the models presented above. They are strongly depleted in volatile elements, including S (17, 18). Therefore, only a very minor early-segregated Fe-FeS metal fraction could form, and so this body evolved with chondritic 180Hf/184W until pure Fe melted at high temperature and subsequently segregated to form the core. However, for accretion times of ~0.1 to 0.3 My, our model predicts that the melting temperature of pure Fe is reached between ~0.7 and 1.3 My after CAI formation (Fig. 3); that is, earlier than suggested by the Hf-W model age of 2.9 ± 0.5 My for the IVB iron meteorites. Thus, either the IVB parent body accreted later than the other iron meteorite parent bodies (at ~0.6 My after CAI formation; Fig. 3), or the precursor material of the IVB parent body had higher-than-chondritic Hf/W. The inferred bulk composition of the IVB parental melt is strongly fractionated relative to chondrites, indicating substantial high-temperature processing of the precursor materials in the solar nebula before parent body accretion (17, 18). Relative to other refractory siderophile elements, W is depleted, reflecting either core formation under relatively oxidized conditions—where W becomes less siderophile—or subchondritic W abundances of the bulk IVB parent body with a bulk 180Hf/184W as high as ~2 (9). Using a 180Hf/184W of 2 results in a Hf-W model age of metal segregation of 1.8 ± 0.3 My after CAI formation, which is in good agreement with the timing of pure Fe metal melting inferred for the IIAB, IID, IIIAB, and IVA parent bodies (Fig. 4). The IVB parent body, therefore, may have accreted at about the same time as the other bodies.

The Hf-W results indicate that core formation in iron meteorite parent bodies occurred over at least ~1 My. Differences in the time of metal segregation reflect either distinct melting temperatures of the metal or variations in the proportions of early- and late-segregated metal fractions, which in turn are controlled by the bulk S concentrations of the parent bodies. Regardless of differences in the time of core formation, the parent bodies of the IIAB, IID, IIIAB, IVA, and IVB iron meteorites probably accreted at about the same time, between ~0.1 and ~0.3 My after CAI formation. Our data, therefore, rule out the possibility that strongly volatile-depleted parent bodies (IVA and IVB) accreted much earlier than less depleted bodies (IIAB), indicating that the variable depletions of moderately volatile elements in the iron meteorite parent bodies do not mirror the increasing condensation of the moderately volatile elements in the solar nebula over time. Rather, they seem to reflect more local processes within the nebula, resulting in spatially distinct chemical heterogeneities and thus variable volatile depletions of the dust accreting to planetesimals. This is consistent with the observation that the IVB iron meteorites, which are among the most strongly volatile-depleted meteorites, formed from material that underwent substantial high-temperature processing before parent body accretion.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S5

Tables S1 to S6

References (2039)

References and Notes

  1. Materials and methods and supplementary text are available on Science Online.
  2. Acknowledgments: We thank T. Elliott, A. Halliday, and an anonymous reviewer for their comprehensive and constructive reviews. We gratefully acknowledge the Field Museum of Natural History (Chicago), the Smithsonian Institution (Washington, DC), and the American Museum of Natural History (New York City) for providing the samples for this study. We also thank R. Wieler, J. Wasson, R. Hin, F. Nimmo, and W. van Westrenen for discussions and U. Heitmann for technical support during sample preparation. This study was supported by a Förderungsprofessur to T.K. of the Swiss National Science Foundation (grant no. PP00P2_123470) and NASA Cosmochemistry grant NNX13AF83G to R.J.W.. The data reported in this paper are tabulated in the supplementary materials (tables S1 to S6).

Stay Connected to Science

Navigate This Article