Rare Gas Systematics in Popping Rock: Isotopic and Elemental Compositions in the Upper Mantle

See allHide authors and affiliations

Science  20 Feb 1998:
Vol. 279, Issue 5354, pp. 1178-1181
DOI: 10.1126/science.279.5354.1178


New experimental data on the isotopic variations of neon, argon, and xenon in a popping rock imply that the40Ar/36Ar ratio of the upper mantle is less than 44,000 and that the 129Xe/130Xe ratio is less than 8.2. The elemental abundance pattern of rare gases is chondritic-like and is quite distinct from the solar pattern. These data imply that Earth accreted from planetesimals that probably underwent a transformation of their rare gas budget from solar- to chondritic-like, leaving the isotopic composition unchanged from the solar pattern.

The rare gas content of material from Earth's upper mantle is extremely low and often reflects atmospheric contamination. The method for analyzing submarine basalts is to use glassy margins of lava flows erupted at great depth under the sea (1). Even in this case, the rare gas data mostly reflect contamination. However, a few samples permit more precise measurements because they have an exceptionally high gas concentration. One of these samples is the “popping rock” 2ΠD43. This sample has a high vesicularity (16 to 18%) and high gas concentrations (2-5). It was dredged in the North Atlantic at 13°46′N at a depth of 3510 m by the research vessel Akademik Boris Petrov in June 1985. Studies of rare gases in this popping rock have shown that the He isotopic ratio is similar to the mean mid-oceanic ridge basalt (MORB)4He/3He ratio of 88,000 and that the40Ar/36Ar is high, up to 28,000, which is one of the highest values measured for MORBs (4).Thus, the isotopic compositions show that this sample reflects best the upper mantle rare gas compositions (6). It also has a 13C/12C isotopic ratio (δ13C) of up to –3.7 per mil, which indicates that the gases in this sample are rather weakly fractionated and therefore may represent the degassed mantle (5). We made two independent analyses of all rare gas isotopes in this sample by stepwise crushing (7) (Table1).

Table 1

He, Ne, Ar, Kr, and Xe abundances (in cubic centimeters per gram at standard temperature and pressure) and isotopic ratios in the 2ΠD43 popping rock samples.38Ar/36Ar ratios have considerable uncertainty and are indistinguishable (2σ level) from the atmospheric ratio (0.188). The times sign refers to the cumulative number of strokes. Numbers in parenthesis indicate a 1σ error in the last digits. Errors in abundance are 5% (1σ). NA, not analyzed.

View this table:

Because the He content in the atmosphere is very low, He is an excellent tracer for mantle sources and yields evidence for two distinct mantle reservoirs. The reservoir source of MORB has a rather constant 4He/3He ratio close to 88,000 and is typically identified as the upper mantle (8). The radiogenic signature of the MORB reservoir results both from outgassing (6) and the decay of U and Th, which increases the4He/3He ratio. The source of most OIBs (oceanic island basalts) has a much lower 4He/3He (down to 25,000) (9) and indicates the existence of a less degassed, therefore 3He-rich, reservoir thought to be located in the lower mantle.

Contamination of MORBs by atmospheric Ne, Ar, Kr, and Xe can be extremely extensive (up to 100%), and for a long time it prevented the discovery of 40Ar and 129Xe/136Xe anomalies in MORBs. Neon can serve as the Rosetta stone for solving this problem because it has three isotopes, and because of its low abundance, only the 21Ne content can be substantially changed in mantle rocks by nucleogenic production (10). Sarda and collaborators (2) showed that the 20Ne/22Ne ratio of the upper mantle is different from the atmospheric ratio (which has a “planetary” ratio of 9.8) and appears to be solar-like (13.8) (11). In a three-isotope Ne diagram, all the MORB samples fall on a straight line that is interpreted as reflecting mixing between the MORB reservoir and air. Honda and collaborators (12) found that glasses from Loihi seamount have a20Ne/22Ne ratio of up to 11.4 and a three-isotope correlation line that is steeper than the MORB line. Following the same approach as for He, they concluded that the MORB mantle is more nucleogenic than the lower mantle because it is more degassed [higher (U+Th)/Ne ratio]. In other locations, the20Ne/22Ne ratio also approaches the solar value (20Ne/22Ne ratio up to 13.2) and supports this interpretation (7, 13-16).These studies have shown that Ne is an excellent tracer of atmospheric contamination in basalts because the 20Ne/22Ne ratios of the two mixing end members are quite well constrained. The correlation of the 20Ne/22Ne ratios with21Ne anomalies directly quantifies the contamination of the nucleogenic component. However, this nucleogenic Ne component should correlate with radiogenic and fissiogenic components of the other rare gases. Thus, the contamination by atmospheric Ar and Xe can be also quantified by investigation of their correlations with Ne isotopic composition.

A solar 20Ne/22Ne ratio (13.8) is a reasonable upper limit for the mantle, but the actual mantle value could well be between a solar value and the highest MORB value measured so far. Our study and all MORB data obtained by stepwise heating (17) yield a maximum 20Ne/22Ne ratio of ∼12.5. The data include analyses from all final steps of crushing of the sample 2ΠD43 (Fig. 1 and Table 1). This ratio may indicate that the MORB mantle has a20Ne/22Ne ratio close to 12.5, as also proposed by Niedermann et al. (18). But because the origin of this value is not yet clear [whether there is nucleogenic production of 22Ne (10,18) or injection of atmospheric noble gases in the mantle], we use a solar 20Ne/22Ne ratio (13.8) to constrain the maximum 40Ar/36Ar and the129Xe/130Xe ratios in the upper mantle.

Figure 1

Three-isotope Ne-Ne diagram showing the Ne results from the 2ΠD43 samples (circles). The black squares represent all the MORB samples (17). The Loihi line is from Honda et al. (12) and Valbracht et al. (15). The MORB line is from Sarda et al. (2). mfl indicates the mass fractionation line (dashed line). In all figures, error bars indicate 1σ uncertainty.

Argon has three isotopes that could enable a deconvolution of mantle and atmospheric components similar to that used for Ne. However, it has not been demonstrated yet without ambiguity that the mantle38Ar/36Ar signature is different from that of air (19). Thus, 40Ar/36Ar ratios and—for a similar reason—the Xe isotopic ratios (129 136Xe/130Xe) measured in MORBs are just lower limits, and estimates for the upper mantle composition have used the maximum measured values of 28,000 for the 40Ar/36Ar ratio and 7.5 for the129Xe/130Xe ratio (4).Burnard et al. (20) expanded the range of40Ar/36Ar ratios up to 40,000 by analyzing Ar from single vesicles of the same sample that we used for our present study. However, the authors proposed that all of the 36Ar found still represents atmospheric contamination and suggested that the true mantle 40Ar/36Ar ratio could be up to 400,000.

Our new data on 2ΠD43 demonstrate a well-defined correlation between 40Ar/36Ar and20Ne/22Ne (Fig. 2), which we interpret as resulting from mixing of the MORB component with an atmospheric component. In such a diagram, mixing is not represented by a linear relation but by hyperbola. The best fit for this hyperbola is given byr = [22Ne/36Ar]UM/[22Ne/36Ar]airof 1.6 ± 0.1 (where r is the hyperbola parameter and UM indicates the upper mantle), and the maximum permitted40Ar/36Ar ratio for the upper mantle is 44,000 if 20Ne/22Ne = 13.8 (Fig.2). For the129Xe/130Xe-20Ne/22Ne correlation, we obtain the result that129Xe/130Xe in the upper mantle is at most ∼8.2 (Fig. 3).

Figure 2

40Ar/36Ar ratio versus20Ne/22Ne ratio from our MORB sample 2ΠD43. The best hyperbolic fit for a solar 20Ne/22Ne ratio is achieved for r = 1.6 ± 0.1 and40Ar/36Ar = 44,000 in the upper mantle. The gray dots are not considered for the fit because of a problem of Ar blanks for these two analyses.

Figure 3

129Xe/130Xe ratio versus20Ne/22Ne ratio from our MORB sample 2ΠD43. Heavy line, linear best fit; dashed line, maximum129Xe/130Xe permitted in the upper mantle.

Our upper mantle 40Ar/36Ar ratio derived from Ne-Ar systematics is close to the highest measured ratios for single vesicles (20). However, this value may still be too high if we suppose that the upper mantle20Ne/22Ne ratio can be lower than the solar value (Table 2), if some injection in the upper mantle of atmospheric Ne is possible, or if there is a nuclogenic production of 22Ne (18).

Table 2

40Ar/36Ar,129Xe/130Xe, 3He/22Ne,3He/36Ar, 3He/84Kr, and3He/130Xe ratios in the upper mantle, depending on its 20Ne/22Ne ratio (left column).

View this table:

The Ne systematics help also to estimate the rare gas elemental abundances of the upper mantle, because the correlations of20Ne/22Ne and therefrom deduced40Ar/36Ar and129Xe/130Xe ratios can be compared with the elemental ratios 3He/22Ne,3He/36Ar, and 3He/130Xe (Fig. 4). For the3He/84Kr ratio, we used a84Kr/3He-36Ar/3He correlation (Fig. 4C), because the mantle appears to have the same isotopic composition of Kr as the atmosphere. Using linear fits and assuming a solar 20Ne/22Ne, we obtain3He/22Ne = 7.3,3He/36Ar = 0.7,3He/84Kr = 21.3, and3He/130Xe = 1180 in the upper mantle. A lower-than-solar mantle 20Ne/22Ne ratio also gives lower elemental ratios (Table 2).

Figure 4

(A)20Ne/22Ne ratio versus the elemental ratio 3He/22Ne from our MORB sample 2ΠD43. (B) 40Ar/36Ar ratio versus the elemental ratio 3He/36Ar. (C)36Ar/3He ratio versus84Kr/3He ratio. (D)129Xe/130Xe ratio versus the3He/130Xe ratio. Two-component mixing is represented by a straight line because the same normalization isotope is used for the two axes in each figure.

The 3He/22Ne ratio determined by our measurements is close to calculated ratios, assuming either a steady-state model for the upper mantle or that the upper mantle is in a closed system (21, 22). This ratio indicates that this sample has unfractionated elemental ratios, as previously indicated by stable isotopes (5).

In a model of a degassed upper mantle and a lower, less degassed, mantle, the elemental ratios (3He/22Ne,3He/36Ar, and so on) should be equal in the two mantle layers as long as no elemental fractionation occured during outgassing of the upper mantle. This assumption has been verified by measurements of the 3He/22Ne ratio (16). To constrain the3He/36Ar ratio in the lower mantle, a reliable estimate of 40Ar/36Ar in the lower mantle is needed. Suggestions for this ratio range from atmospheric values (or close to them) (6) up to a value as high as >15,000 (13, 22, 23). The40Ar/36Ar ratios measured in Loihi seamount basalts, thought to best represent the lower mantle, have always been low (<5000). To quantify the atmospheric contamination in a magma chamber or during submarine eruption, Valbracht et al. (15) used the Ne-Ar systematics and restricted the40Ar/36Ar ratio to between 2000 and 5000. This case implies that the lower mantle 3He/36Ar ratio is between 0.4 and 1.1, which is close to the upper mantle value that we estimate above (0.7). Values for lower mantle Xe cannot be derived from a similar approach, because no clear anomalies of129Xe/130Xe have been observed in Loihi basalts (15). We will suppose here that the Xe isotopic ratios of the lower mantle are atmospheric, and that the3He/130Xe ratio is the same as in the upper mantle. Thus, we have experimental evidence (except for Xe) that the abundance patterns of the lower and upper mantles are similar and that we can extend the results for the upper mantle to the whole mantle.

Recently, Burnard et al. (20) suggested than the upper mantle rare gases have solar abundance patterns. This suggestion is essentially based on the assumption that the3He/36Ar ratio is solar-like [they calculated a ratio of 1.4, whereas the solar ratio is 13; and they assumed that the 4He/40Ar* ratio is 3, which has not been proved: it seems to be 1.5 (Table 1) (40Ar* is radiogenic40Ar, corrected for air contamination)]. If this model were correct, accretion would have been homogeneous with a complete loss of planetary rare gases from the planetesimals because of the high energies involved in the accretion process. The (proto-)Earth would then have acquired its mantle gas budget directly from the solar-like nebular gases, probably by solution in a magma ocean in equilibrium with a dense solar-like atmosphere that was blown off and lost quite early (24). In this scenario, an atmosphere such as we have now on Earth would have to be formed from a late volatile-rich veneer (25). Degassing of the Earth's interior should be negligible, except for the isotopes4He and 40Ar, which are produced by radioactive decay in considerable quantities, and perhaps also the Ne isotopes, whose concentration in the atmosphere is likely to be a mixture of planetary and solar Ne values (26).

However, the normalization to 36Ar and solar abundances (Fig. 5) demonstrates that the rare gas abundances in the upper mantle are similar to the planetary pattern of primitive chondrites (C1) (27) and distinct from the solar pattern. This observation justifies the application of models that explain the formation of Earth's atmosphere by mantle degassing (6, 27, 28).

Figure 5

Abundance of rare gases relative to 36Ar, normalized to solar abundance [(i/36Ar)/(i/36Ar)solar] for the upper mantle (UM), atmosphere (air), chondrite C1 (C1), and solar-composition matter (solar) [from Table 2 and (27)].

On the other hand, it has been well established that a solar-like Ne isotopy is present in the mantle (2,12-15). Even if the He/Ne ratio appears also to be solar-like (as indicated by an almost flat pattern in Fig. 5), because the Ne/Ar ratio is planetary-like (Fig. 5), the solar-like He/Ne ratio simply reflects an excess of 3He. Thus, we suggest that the materials that accreted to form Earth underwent a transformation of their rare gas abundances from solar- to planetary-like values. This elemental fractionation can be more easily achieved than an isotopic fractionation. For this reason, the isotopic ratios in Earth's interior remained solar-like. The debated existence of solar 38Ar/36Ar in the mantle (15) and the isotopic compositions of stable isotopes of Kr and Xe would support this suggestion (29). The excess of He in the mantle (compared to planetary values) may be a hint about the processes involved in the transformation from solar-like to planetary-like abundances, which should be linked to the different physicochemical behavior of He and the other rare gases. There are three important processes that may play a role: The light species, especially He, are more soluble in magmas and are more mobile by diffusion than the heavier ones, whereas they are less affected by adsorption (30, 31).

The present Earth's atmosphere is a further-evolved rare gas reservoir. It resulted from solid Earth degassing and was probably subject to additional changes by trapping [sedimentary adsorption (32)], loss into space [hydrodynamic escape (33)], and remixing with late veener components [comets, for example (26)].


View Abstract

Stay Connected to Science

Navigate This Article