Meteorite Kr in Earth’s Mantle Suggests a Late Accretionary Source for the Atmosphere

See allHide authors and affiliations

Science  11 Dec 2009:
Vol. 326, Issue 5959, pp. 1522-1525
DOI: 10.1126/science.1179518


Noble gas isotopes are key tracers of both the origin of volatiles found within planets and the processes that control their eventual distribution between planetary interiors and atmospheres. Here, we report the discovery of primordial Kr in samples derived from Earth’s mantle and show it to be consistent with a meteorite or fractionated solar nebula source. The high-precision Kr and Xe isotope data together suggest that Earth’s interior acquired its volatiles from accretionary material similar to average carbonaceous chondrites and that the noble gases in Earth’s atmosphere and oceans are dominantly derived from later volatile capture rather than impact degassing or outgassing of the solid Earth during its main accretionary stage.

There are many possible origins for planetary gases, but there is little evidence to identify whether or not the volatiles within our planet have the same accretionary origin as those that now form our oceans and atmosphere. After collapse of the solar nebula, planetesimals were formed from the dust within the first 0.1 to 1 million years (1). Although it took a further 100 million years of violent impacts between these planetary embryos before the terrestrial planets we see today remained, 60% of Earth’s mass accumulated within the first 10 to 20 My (1, 2). Solar nebula gas may have been present for as long as 10 My after nebula collapse was initiated (3) and contemporaneous with the terrestrial planet precursors. Indeed, gravitational capture of solar nebula gases by the early Earth (4) is used as a starting point for many noble gas and volatile accretionary models of both the terrestrial atmosphere and the interior (5, 6). Alternatively, volatiles may have been carried to the early Earth by accreting dust, planetesimals (7), and cometary material (8). These sources would be composed of gases from implanted solar wind (911), adsorbed nebula gases (12), and minor amounts of presolar material. Although accretionary models of the inner planets suggest that water and other volatiles may have been acquired from cometary and other icy body material (13), others contend that impact degassing of accretionary material or early planetary degassing played a major role in atmosphere formation (14).

Within Earth’s mantle, primordial 20Ne/22Ne ratios show a similarity between the terrestrial mantle value and that of solar wind that has been implanted into meteorite surfaces and the lunar regolith (9, 10). Although the source of Earth’s primordial light noble gases (He and Ne) is most likely from addition of solar wind–irradiated material to the surfaces of accreting material, it is not possible to predict from this result whether heavy noble gases (Ar, Kr, and Xe) trapped within the primordial Earth have a similar implanted solar wind composition. This is because, when compared to solar elemental ratios, enrichment of Kr and Xe in chondrites by ~104 and 105, respectively, relative to Ne require only minor input from chondritic material for it to dominate the noble gases signature (15). Identifying the original isotopic composition of the heavy noble gases within Earth is nevertheless critical in developing our understanding of the relationship and timing of volatile interaction between the deep Earth and the atmosphere (16, 17).

Magmatic CO2 natural gases have proved to be an invaluable resource for studying mantle gases (10, 1619). Here, we report Kr and Xe isotopic measurements of CO2 gases to determine the isotopic composition of heavy noble gases in Earth’s interior. Our samples were collected from the Bravo Dome gas field in New Mexico, United States, which produces almost pure (99%) CO2 from the Tubb sandstone at depths of 600 to 700 m. Magmatic gas in this field has been in contact with groundwater to varying degrees, providing a spatially coherent mixture of gas with two different origins. End-member compositions are represented by near-pure magmatic gas in the western portions of the field and a mixture of dissolved air-derived gas and crustal radiogenic noble gases that have accumulated in the groundwater to the east (10, 16). To date, Kr and Xe measurements have been limited by the precision of single-collector instruments (16, 18, 19). To overcome these limitations, we used multicollector noble gas mass spectrometry (20).

Isotope data of 86Kr/82Kr versus 84Kr/82Kr show a linear trend from atmospheric values to compositions with nonair ratios (Fig. 1A and table S1). Samples showing the greatest deviation from air also possess the lowest Kr concentration, as would be expected from samples with the least air contamination (fig. S1). The largest deviations from air of Kr isotopes also correlate with the highest mantle-derived 3He/4He, 20Ne/22Ne, and 124-128Xe/130Xe isotopic compositions observed in previous work (10, 16). Unfractionated 38Ar/36Ar data (17) and the high precision 124-128Xe/130Xe data presented here exclude mass fractionation during transport or storage as the cause of the isotopic deviations from air observed in Kr and nonradiogenic Xe. We interpret the Kr isotopic deviations from air as a resolvable primordial mantle Kr signal. Samples were corrected for 86Kr and 84Kr produced from U fission in the crust (Fig. 1B) to give a Kr isotopic composition for each sample that must lie on the present day mantle-air mixing line. This mixing line serves as a reference line for comparison of terrestrial data with possible primordial or accretionary volatile sources. The mixing line shown represents the minimum fission correction because correcting for fission U and Pu-derived Kr within the mantle may decrease the gradient of this mixing line further depending on the U/Pu of the mantle source (20) (fig. S4).

Fig. 1

Kr isotope measurements of well gases show a clear, nonsolar primordial component. (A) Individual analyses of all sample aliquots (n = 125) are plotted. Uncertainties in well gas data and primordial end-members are 1σ. Data are compared with those from solar-Kr, AVCC-Kr, Kr-Q, and Air-Kr (31). The dashed line represents the solar mass fractionation line. (B) Averages of each sample are plotted both as uncorrected data (black) and as crustal fission corrected (blue) (see fig. S4 for corrections involving both U and Pu fission). The bold line is a York fit (32) freely fitted through the data. Thin black lines are 1σ error envelopes to the data. The dashed line represents the solar mass fractionation line. Fission correction is based on crustal U fission determined from 136Xe excesses over air and assumes no Kr/Xe fractionation. Uncorrected data do not pass through air because of a crustal U fission component to Kr as observed previously in Xe isotopes (16); when corrected for fission, extrapolation of the best fit line passes through atmosphere. A summary of primordial end-members is presented in table S4.

The air-mantle mixing line is consistent with a fractionated solar or average carbonaceous chondrite (AVCC) source for the primordial mantle Kr. The maximum Pu/U fission correction (fig. S4) produces an air-mantle mixing line that remains within the AVCC error envelope but no longer within error of the air–fractionated solar mixing line. Importantly, our data exclude unfractionated solar nebula gas (solar) as a primordial or accretionary component now preserved in the mantle sampled by these gases. AVCC is the product of tightly bound Ar, Kr, and Xe found in carbonaceous residues of meteorites, referred to as the Q component, and other minor noble gases trapped in different phases. The Kr isotope system has the ability to distinguish between AVCC and Kr-Q. Because the air-mantle mixing line excludes Kr-Q as the sole source of Kr, the mechanism of accretion does not distinguish between the different locations of noble gases within meteorites—an important consideration for non-noble gas volatile elements.

Xe isotope data provide further perspective on the Kr observations. Samples from Bravo Dome have 124-136Xe/130Xe in excess of air values (16, 19). The light Xe isotopes are unaffected by radiogenic addition, and these deviations from air represent a resolvable primordial component in the mantle Xe record. The trend described by our 124Xe/130Xe and 126Xe/130Xe data (Fig. 2) is indistinguishable from an AVCC (or Q) component and is consistent with published well gas and mid-ocean ridge basalt (MORB) reservoir data (table S4). The Xe isotope results are also consistent with the Kr observation that the primitive heavy noble gases within Earth are sourced by a chondritic reservoir with a composition similar to that of carbonaceous chondrites or a fractionated solar nebula source. At present, solar nebula Xe cannot be resolved from AVCC without further improvement in the precision of the solar and AVCC Xe end-member values. We focus our discussion on 124Xe/130Xe and 126Xe/130Xe ratios because of possible 128Xe production by neutron capture in iodine that may increase uncertainty in end-member 128Xe/130Xe. Additional 124-128Xe/130Xe data are plotted in fig. S5.

Fig. 2

Nonradiogenic Xe isotope measurements of well gases show a primordial component that is completely consistent with Kr data. (A) Reduced scale view to show Xe nonradiogenic isotope data, plotted as 124Xe/130Xe versus 128Xe/130Xe with errors at 1σ uncertainty. The bold line is a York fit (32), freely fitted through the well gas data presented in table S2. Thin black lines are 1σ error envelopes. Open squares are Harding County and Caroline well gases (19). Open diamonds are MORB averages computed from (21). The open circle is modern atmosphere (atm). (B) Expanded scale view to incorporate possible end-members, plotted as 124Xe/130Xe versus 126Xe/130Xe. The bold line is the error-weighted best fit, freely fitted through the data. Thin black lines are error envelopes of 1σ. Uncertainties in the primordial end-members are 1σ. Isotopic compositions of solar Xe, AVCC-Xe, Xe-Q, and Air-Xe are tabulated in (31) and table S4.

To examine whether the well gases, which sample the subcontinental lithospheric mantle, are representative of the convecting mantle, we turn to previous observations. These show that the 3He/4He, 20Ne/22Ne, 40Ar/36Ar, 124,126, 128,129Xe/130Xe, and 3He/20Ne/36Ar/84Kr/130Xe of the well gas mantle source is almost indistinguishable from that of MORB, represented by the 2ΠD43 popping rock (10, 16, 21, 22). In addition, both reservoirs have an identical 36Ar/84Kr/130Xe abundance composition that is indistinguishable from seawater and a small (10%) sedimentary contribution, itself unique in isotope character and composition in the solar system. This is interpreted as evidence for seawater recycling into both mantle reservoirs and argues against isolation of these reservoirs from each other over geological time (16). The only significant difference between the MORB and well gas mantle reservoirs is in their respective Pu-derived Xe content, which has been interpreted as reflecting a different closure age and isolation of the two reservoirs for over 4 Gy (21). Many mantle models balance an accretionary gas-rich reservoir flux into the convecting mantle with radiogenic in-growth and a mantle degassing with possible atmosphere recycling to achieve the present-day convecting mantle noble gas abundance and isotopic composition [e.g., variants on (23)]. In particular, we argue that there is no a priori reason why the nonradiogenic primordial Kr and Xe components should differ between reservoirs when both mantle reservoirs preserve indistinguishable nonradiogenic primordial He, Ne, and Xe. The overwhelming similarities between the reservoirs lead us to conclude that the extension of the mantle sampled by the well gases to the rest of the mantle is indeed valid.

Taken together, the Kr and Xe data are consistent with the early mantle acquiring a composition that is mass-fractionated and depleted in the lighter isotopes relative to the solar nebula (Fig. 1). A source that provides both this signature and a mechanism of delivery to the early Earth is accretion of meteoritic material similar to AVCC (Figs. 1 and 2). A similar explanation has been argued to preserve and deliver a solar wind–implanted Ne isotope signature to Earth’s early mantle (9, 10). We nevertheless cannot preclude the possibility of other more-complex processes that mass-fractionate solar nebula gases and introduce these into the early Earth’s interior.

Planetary evolution models often argue that the unique noble gas elemental and isotopic character of the terrestrial atmosphere can be explained by a combination of fractionation in the atmosphere and deep planetary outgassing of noble gases with isotopic compositions similar to the solar nebula (5, 24, 25). These models are attractive because they invoke preferential loss of the light elements and isotopes during hydrodynamic escape (i.e., mass-fractionating gas loss induced by upward drag from escaping hydrogen). The energy for this process may originate from either impacts or enhanced ultraviolet radiation from the early Sun. This light isotope depletion is broadly seen in Earth’s atmosphere relative to possible solar system reservoirs. The heavy noble gases in Earth’s atmosphere are assumed to have evolved together through mass fractionation during early atmosphere loss (5). In such scenarios, extensive fractionation of Xe isotopes required to reduce 124-128Xe/130Xe ratios from solarlike values to those observed in the atmosphere would also have produced fractionation in Kr isotopes to a much greater extent than that observed in the atmosphere today. This type of model requires subsequent input of solar Kr, but negligible additional solar Xe, to buffer these ratios back to modern-day air (5). Earth’s mantle has been assumed to provide the buffering source (5).

Our results, however, show that Earth’s mantle does not presently contain significant unfractionated solar nebula Kr. If a solar nebula Kr component was ever present in the early Earth’s mantle, evidence for it has been erased by degassing and overprinting with Kr that is isotopically heavier than air (Fig. 1). This “heavy” Kr cannot generate modern atmosphere through degassing followed by mass fractionation during atmosphere loss. We are therefore led to the conclusion that neither early outgassing nor impact-driven degassing of the early mantle represented by these gases are responsible for generating the noble gases in the present-day atmosphere. The absence of resolvable solar nebula Kr in the mantle precludes any role of this source in the buffering of heavily fractionated Kr (necessitated by Xe fractionation) back to atmospheric Kr.

We suggest that, although the Kr and Xe trapped in the mantle are primary signatures of the accreting Earth, they play only a minor role in determining the noble gas composition of the modern atmosphere. It is possible that a fractionated solar atmosphere may have been preserved as a remnant of major impact events, perhaps as late as 60 My after the start of the solar system (26), if atmosphere loss during Moon formation is substantial (27, 28). However, to achieve the approximately solar Kr/Xe ratios and isotopic compositions in the present atmosphere, an additional high Kr/Xe source is still required. Experiments on amorphous water ice suggest cometary ices could provide such a reservoir (29, 30). Earth’s modern atmosphere is then a mixture of a relatively late fractionated solar gas and an even later cometary source. A similar conceptual model has been proposed as an explanation for why Xe is unexpectedly depleted in Earth’s atmosphere relative to Kr (30).

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Tables S1 to S4


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank Oxy and Amerada Hess corporations for permission to sample the Bravo Dome Field. We thank D. Blagburn and B. Clementson for laboratory support and R. Pepin for discussions on the manuscript. This work was funded by UK Natural Environment Research Council.
View Abstract

Navigate This Article