The Provenances of Asteroids, and Their Contributions to the Volatile Inventories of the Terrestrial Planets

See allHide authors and affiliations

Science  10 Aug 2012:
Vol. 337, Issue 6095, pp. 721-723
DOI: 10.1126/science.1223474

Constraining the Birthplace of Asteroids

Many primitive meteorites originating from the asteroid belt once contained abundant water that is now stored as OH in hydrated minerals. Alexander et al. (p. 721, published online 12 July) estimated the hydrogen isotopic compositions in 86 samples of primitive meteorites that fell in Antarctica and compared the results to those of comets and Saturn's moon, Enceladus. Water in primitive meteorites was less deuteriumrich than that in comets and Enceladus, implying that, in contradiction to recent models of the dynamical evolution of the solar system, the parent bodies of primitive meteorites cannot have formed in the same region as comets. The results also suggest that comets were not the principal source of Earth's water.


Determining the source(s) of hydrogen, carbon, and nitrogen accreted by Earth is important for understanding the origins of water and life and for constraining dynamical processes that operated during planet formation. Chondritic meteorites are asteroidal fragments that retain records of the first few million years of solar system history. The deuterium/hydrogen (D/H) values of water in carbonaceous chondrites are distinct from those in comets and Saturn’s moon Enceladus, implying that they formed in a different region of the solar system, contrary to predictions of recent dynamical models. The D/H values of water in carbonaceous chondrites also argue against an influx of water ice from the outer solar system, which has been invoked to explain the nonsolar oxygen isotopic composition of the inner solar system. The bulk hydrogen and nitrogen isotopic compositions of CI chondrites suggest that they were the principal source of Earth’s volatiles.

Two recent dynamical models of the orbital evolution of the outer planets (1, 2), designed to explain the small mass of Mars and the late heavy bombardment, predict that the objects in the asteroid belt have two sources: S-complex asteroids formed in situ, whereas D-type, P-type, and C-complex asteroids formed either between the giant planets or in the trans-Neptunian region. Chondrites, the most primitive meteorites, are fragments of main-belt asteroids and are divided into several classes (ordinary, OC; Rumuruti, RC; enstatite, EC; and carbonaceous, CC) that are subdivided into numerous groups. On the basis of their reflectance spectra and other evidence, the OCs and CCs have been linked to the S- and C-complex asteroids, respectively. It was suggested (2) that D- and P-type asteroids are the sources of primitive micrometeorites, rather than meteorites. However, most micrometeorites appear to be related to CI and CM CCs (3), and at least one ungrouped CC, Tagish Lake, has been spectroscopically linked to D-type asteroids (4). At present, there is no evidence that a major type of primitive asteroid is unrepresented in our meteorite collections (3). There is isotopic evidence for a difference in the provenances of the OC-EC-RCs and the CCs (5), but there are no reliable indicators of where in the solar nebula they formed.

The proposed formation locations for the D-type, P-type, and C-complex asteroids are potentially the same as those of comets—Oort cloud comets are thought to have formed in the giant planet region and the trans-Neptunian region, while Jupiter-family comets (JFCs) probably formed in the trans-Neptunian region (6). Although chondrites share many similarities with comets (3), they do not now contain ice. The CCs, as well as the OCs and RCs, accreted water, presumably as ice, but the record of this is now preserved only in hydrated silicates (7, 8). Perhaps the CC parent bodies were larger than typical comets, so that internal heating caused much of the water to be consumed by alteration of silicates, Fe-metal, and sulfides and what remained sublimed away in the ~4 to 4.5 billion years that they have resided in the inner solar system. Nevertheless, if CCs have a cometary heritage, a record of it should be preserved in the hydrogen isotopes of their hydrated silicates.

In a static solar nebula, radial temperature gradients, combined with equilibrium and kinetic factors, mean that water D/H values should have increased with increasing formation distance from the Sun (9, 10). Given its dynamic evolution, the spatial and temporal variations in ice compositions in the solar nebula are likely to have been more complex. Nevertheless, objects that formed in the same source regions and at similar times should have accreted ice with similar hydrogen isotopic compositions. Therefore, a comparison of water D/H values in comets and and other icy outer solar system bodies with CCs is potentially a direct test of the predictions of the dynamical models.

Hydrogen isotopic compositions are available for the water in some comets and Saturn’s icy moon Enceladus. The D/H values of the OH in hydrated silicates in chondrites are likely to have been modified during alteration from the water compositions at the time of accretion. The chondrites also accreted organic matter (11), and the hydrogen in it further complicates the determination of the original isotopic compositions of their water (3). Here, we report the bulk hydrogen, carbon, and nitrogen elemental and isotopic compositions of 86 samples of chondrites (table S1) and adopt two approaches to use them for estimating or placing limits on the original hydrogen isotopic compositions of their water (3).

In the first approach, we analyzed CCs from the CM and CR groups that experienced different degrees of aqueous alteration. If the bulk hydrogen isotopic compositions are simply the products of mixing between hydrated silicates and organic matter, then, in a plot of δD versus C/H, the bulk compositions should form a line with the hydrogen isotope intercept giving the isotopic composition of the water. In Fig. 1, the data for the CMs and CRs do fall on lines, and the implied δD values for their water are Embedded Image ‰ and Embedded Image‰, respectively [δD = 1000 × (Rsample/Rstandard – 1), where Rstandard is the D/H value of standard mean ocean water]. The subparallel CM and CR trends converge on the region where the most primitive insoluble organic material (IOM) plots (Fig. 1B). This implies that the CMs and CRs shared a common primitive organic component. The IOM in most chondrites is less deuterium-rich than that in the most primitive meteorites (Fig. 1B), suggesting that there has been hydrogen isotope exchange between the water and the organics in essentially closed systems.

Fig. 1

(A) The bulk compositions of the carbonaceous chondrites analyzed here. The lines are fits to the CR and CM data. Open red squares are CMs that were not included in the CM fit. (B) An expanded scale to show that the CR and CM lines pass through the compositions of the most primitive chondritic insoluble organic matter (IOM) (11, 14). The bulk composition of the OC Semarkona is also included.

The CIs and the most primitive COs, CVs, and OCs are not common enough to use the same approach. The two CIs (Orgueil and Ivuna) and the CO (ALH 77307) analyzed here fall on the CM trend, suggesting that they all probably had similar initial water compositions. Three lithologies from Tagish Lake exhibit different degrees of alteration (12). Their bulk compositions show limited variations in δD and C/H and suggest that the water in Tagish Lake had a slightly lower D/H value than in the CMs. The bulk hydrogen isotopic composition of the OC Semarkona (Fig. 1B) has a much higher D/H value than any of the CCs.

Because all chondrites, not just the CMs and CRs, probably accreted a common organic component (11), the second approach for estimating the water compositions is by subtraction of the organic component. The hydrogen isotopic compositions of the water in CMs and CRs can be estimated, albeit with larger errors (3), by assuming that (i) the bulk carbon is in material with IOM-like C/H ratios and (ii) it has an isotopic composition like that of the most primitive IOM (δD ≈ 3500‰). This approach has been used (3) to estimate the ranges of likely water compositions in the CIs, COs, CVs, Tagish Lake, and the OCs (Fig. 2 and table S2). The ranges reflect whether the measured IOM C/H values are used, or a more primitive ratio based on the CR IOM that may better represent the IOM composition at the onset of alteration and metamorphism (11).

Fig. 2

Comparison of the estimated hydrogen isotopic compositions of water in various chondrite groups with those measured in Oort cloud and Jupiter family comets (JFC), and Saturn’s icy moon Enceladus. See (3) for details and sources.

The six measured Oort cloud comets and Enceladus have water D/H values that are enriched in deuterium by roughly a factor of 2 compared to Earth and more than an order of magnitude relative to the initial solar composition (Fig. 2). The one measured JFC has a water hydrogen isotopic composition that is similar to Earth’s (13). This suggests that there may not be a monotonic rise in the D/H values of icy bodies with formation distance from the Sun. Nevertheless, the D/H value of this JFC is still enriched in deuterium by almost an order of magnitude compared to the initial solar composition. Consequently, one would still expect that water accreted by asteroids that formed in the asteroid belt (e.g., S-complex asteroids) would have lower D/H values than objects that formed in the giant planet region and beyond (e.g., C-complex asteroids).

The OC water hydrogen isotopic composition is similar to that of Oort cloud comets [the RCs are more deuterium enriched (8)], whereas the CR water composition is like that of Hartley-2 (Fig. 2). The water compositions of the CIs, CMs, CO, CV, and Tagish Lake are less deuterium-rich than any measured comets, implying formation closer to the Sun. The lower deuterium enrichments in CCs than in OCs and RCs also run counter to the typically assumed order of formation distances for the chondrites (3). However, the hydrogen isotopic compositions in Fig. 2 and table S2 should be regarded as upper limits because isotopic fractionation associated with the observed oxidation of Fe by water during alteration and metamorphism is likely to have enriched the residual water in deuterium (14).

There has been speculation that some CCs are cometary in origin. On the basis of estimates of their preatmospheric orbits, it has been suggested that Orgueil (CI) is a fragment of a JFC (15) and that the CM Maribo is related to comet P2/Encke (16). So-called main-belt comets with typical asteroidal orbits have also recently been discovered (17). Hence, the suggestion that many outer main-belt asteroids (and therefore CCs) formed in the same regions as conventional comets is attractive, but the subterrestrial D/H values of the water in most CCs do not support this. However, most Oort cloud comets may have formed in the trans-Neptunian region, not the interplanetary region, in which case, the Oort comet-like D/H value of Enceladus provides the only constraint for the possible interplanetary origin (1) for the CCs. Enceladus appears to have accreted from material that formed in the solar disk (18), rather than in the dense subnebula around Saturn when it was growing rapidly. This suggests that the sources of the CC parent bodies were sunward of Saturn’s orbit (3 to 7 astronomical units) as it approached its final mass (1), including in the asteroid belt.

All analyzed bodies in the terrestrial planet region (within the orbit of Jupiter), including the Earth and chondrites, have oxygen that is enriched in 17O and 18O, compared to the bulk solar system, represented by the solar wind (19). The favored explanation is an early, massive influx of 16O-poor water ice that formed either in the outer solar system or the presolar molecular cloud (20, 21). The bulk and internal oxygen isotopic systematics of OCs and CCs are at least consistent with accretion of 16O-poor ices (22, 23). However, our results are in conflict with the ice influx explanation because outer solar system ice should have D/H values like those of cometary ices and molecular cloud ices are even more deuterium-rich (24, 25).

The range of D/H values in chondritic water could be explained in the context of the ice influx model if variable amounts of the accreted water had reequilibrated at high temperatures with nebular H2, thereby acquiring roughly solar hydrogen isotopic compositions. One obvious local heat source is chondrule formation, in which case one might expect an inverse correlation between chondrule abundances in chondrites (OC ≈ RC > CV ≈ CO ≈ CR > CM > Tagish Lake > CI) and their initial water D/H values given in table S2 (RC > OC > CR > CV ≈ CO ≈ CM ≈ Tagish Lake ≈ CI), but this is not the case. Alternatively, the isotopically more solar-like water reequilibrated with nebular H2 sunward of the snow line and subsequently migrated out to the chondrite-formation regions beyond the snow line—for instance, via the cold-finger effect (26). In this scenario, the water in the OCs and RCs would then be essentially pure outer solar system ice. Such an explanation is hard to understand if, as is generally assumed, the OCs and RCs formed closer to the Sun than the CCs and may have formed earlier (3). Also, there is no evidence from the internal oxygen isotope systematics of the chondrites that the water in the OCs was much more 16O-depleted than in the CCs (3), as this explanation would predict. Thus, at present there is an inconsistency between the ice influx model and our results, unless, perhaps, the major ice influx occurred long before chondrite formation and the ice that chondrites accreted formed more locally (with smaller isotopic anomalies) because transport in the disk had become less efficient.

The possibility that comets may have supplied much of Earth’s volatiles has been revived by the discovery that a JFC has water with an Earth-like hydrogen isotopic composition (13). However, Earth would not have accreted only cometary ice, but entire comets, including much organic material. From our knowledge of chondrites (11, 14) and interplanetary dust particles (27) that may come from comets, this organic material is deuterium-rich. Hence, the bulk hydrogen isotopic composition of a comet is likely to be more deuterium-rich than its water (3), in which case even comets like Hartley-2 may not be viable sources of Earth’s volatiles.

If comets were not the sources of Earth’s water, asteroids, including main-belt comets, become the most likely candidates (28), along with some nebular gas (29). Marty (29) has argued that the relative abundances of hydrogen, carbon, and the noble gases in Earth are roughly chondritic. Nitrogen is depleted relative to carbon and hydrogen, perhaps because it is trapped in the core (i.e., the volatiles were not accreted in a late veneer) (29). If Earth’s volatiles are in roughly chondritic abundances, isotopes might indicate which, if any, of the known chondrites were their major sources. The range of bulk carbon isotopic compositions of chondrites is small (3), but the hydrogen and nitrogen isotopes vary greatly (Fig. 3). No single chondrite group is identical in composition to that of the bulk Earth. The primitive CO ALH 77307 has bulk hydrogen and nitrogen isotopic compositions like Earth’s, but other isotopic evidence probably rules out COs as the major source of Earth’s volatiles (5). The CIs plot on the extension of the line connecting solar and bulk Earth. It is possible that a number of chondrite groups contributed to Earth’s volatile budget, but perhaps the simplest explanation is that most of the hydrogen and nitrogen (as well other volatiles) was accreted in CI-like material, along with ~10% contributions to both elements from material with a solar isotopic, but elementally fractionated, composition (3, 29).

Fig. 3

The bulk hydrogen and nitrogen isotopic compositions of chondrites (TL, Tagish Lake). The line connects the solar and terrestrial isotopic compositions. Bodies similar to chondrites are potential sources of Earth’s volatiles. For reasons discussed in the text, CI-like material with ~10% contributions of material with isotopically solar compositions, but a roughly chondritic H/N value, can most simply explain Earth’s bulk hydrogen and nitrogen isotopic compositions (3).

Note added in proof: Some of the discussion regarding comparison between the meteorites and the Oort cloud comets and Enceladus has been clarified from that in the version published in Science Express.

Supplementary Materials

Materials and Methods

Supplementary Text

Author Contributions

Figs. S1 to S3

Tables S1 to S5

References (30113)

References and Notes

  1. Supplementary materials are available on Science Online.
  2. Acknowledgments: All the data reported in this paper are presented in the supplementary materials. This work was partially funded by NASA Cosmochemistry grant NNX11AG67G (C.M.O’D.A.), the NASA Astrobiology Institute (C.M.O’D.A., R.B., M.L.F., L.R.N.), Carnegie Canada (C.M.O’D.A., C.D.K.H., L.R.N.), the Natural Sciences and Engineering Research Council of Canada (C.D.K.H.), the W.M. Keck Foundation (M.L.F., R.B.), and the UK Cosmochemical Analysis Network (K.T.H.). For supplying the many samples that were necessary for this work, we thank the members of the Meteorite Working Group, C. Satterwhite and K. Righter (NASA, Johnson Space Center), T. McCoy and L. Welzenbach (Smithsonian Museum for Natural History), L. Garvie (Arizona State University), S. Russell, C. Smith, and D. Cassey (Natural History Museum, London).

Stay Connected to Science

Navigate This Article