Technical Comments

Noble Gases in Mantle Plumes

Science  23 Mar 2001:
Vol. 291, Issue 5512, pp. 2269a
DOI: 10.1126/science.291.5512.2269a

Trieloff et al. (1) provided neon isotope data to argue that mantle neon originates as a trapped component in meteorites rather than having a pristine solar composition. That finding, if true, has important consequences for understanding the mechanism of volatile input into the mantle, because it rules out models involving direct incorporation of solar gases.

The main argument that Trieloff et al. put forward was that a dunite from Hawaii and a basalt glass from Iceland show the same upper limit 20Ne/22Ne value, 12.55 ± 0.11, in the last stages of crushing. This value, although distinct from the solar wind value of 20Ne/22Ne = 13.8, is indistinguishable from the mixture of implanted neon from solar wind and solar energetic particles (20Ne/22Ne = 11.2) that is trapped in lunar and meteorite grains (2). Trieloff et al. argued that the constant 20Ne/22Ne value lower than solar wind cannot be due to the addition of an air contaminant (20Ne/22Ne = 9.8), because that would require a mechanism capable of contaminating the sample in proportion to the amount of 22Ne released in the final stages of crushing, which varied from 1.7 × 10−12 to 12.4 × 10−12 cm3 at standard temperature and pressure.

Air-like noble gas contamination of samples is not removed by exposure to vacuum together with moderate heating, and requires a high-energy environment for release. There is no evidence that the first steps of the ball mill crushing technique employed by Trieloffet al. quantitatively remove this air contamination. Air-like noble gases account for lower Ne and Ar isotope values in almost all other reported samples. In basalt glass, air contamination is proportional to vesicularity, and air is almost certainly trapped within the matrix surrounding the vesicles (3). The final stages of crushing that release the last gas trapped in inclusions may also release the last vestiges of vesicularity-related contamination in proportion to the varying amounts of Ne released. This would provide a mechanism to produce the observed 20Ne/22Ne “plateau” effect in the Iceland glass. Irrespective of whether the same mechanism applied to the dunite sample examined by Trieloffet al., any measured 20Ne/22Ne values can only be considered lower limits on the mantle composition. Harrison et al. (4), using a pressure fracturing technique rather than a high-energy ball mill, reported20Ne/22Ne = 13.75 ± 0.32—indistinguishable from the solar value—in the same Iceland sample analyzed by Trieloff et al. The lower20Ne/22Ne values reported by Trieloff et al. thus may only represent the upper limit of the ball mill crushing technique.

If the mantle were indeed dominated by the proposed trapped Ne component, no major portion of the mantle would contain20Ne/22Ne values greater than ∼12.5. The mantle that supplies mid-ocean ridge basalts (MORB) provides a useful test case. Employing the same release technique as Trieloff et al., Moreira et al. (5) also found a maximum value of 20Ne/22Ne ≈ 12.5 in the gas-rich MORB sample 2πD43, and showed a clear correlation between40Ar/36Ar and 20Ne/22Ne for different release steps. This correlation suggests that if the MORB-source mantle has 20Ne/22Ne = 12.5, then it would have 40Ar/36Ar ≈ 25,000 (Fig. 1). Fisher reports 40Ar/36Ar > 35,000 in MORB from the East Pacific Rise (6), and Burnardet al. report 40Ar/36Ar at up to 40,000 in another portion of the 2πD43 sample, with the gas released from individual vesicles using a laser (7). Combined with the Ne-Ar correlation (5), this value is consistent with solar Ne in the MORB-source (Fig. 1).

Figure 1

20Ne/22Ne correlates with 40Ar/36Ar measured in the MORB 2πD43 sample (5). Whole-rock40Ar/36Ar values exceed 35,000 (6), and individual-inclusion analysis gives values of up to 40,000 (7). These argon ratios require, under relationships described by Moreira et al. (5), that the MORB-source mantle have a 20Ne/22Ne value significantly higher than the “plateau” value of 12.5 reported by Trieloff et al. (1) for Loihi and Iceland.

Figure 11

40Ar/36Ar ratios versus 1/36Ar for MAR popping rock 2πD43 data. (A) Ball mill crushing data from Moreira et al. (5), Kunz (8), and Staudacher et al. (9). (B) Pressure fracturing data from Trieloff and Kunz (7). (C) Laser extraction data from Burnard et al. (6). Although steep trends toward a y-intercept of 295.5 (open symbols) indicate local atmospheric contamination of the sample, the horizontal trends (solid symbols; mean values with 1ς error bands) suggest values close to 30,000 as the best estimate for the upper mantle value. (D) Fit hyperbola obtained by minimizing χ2through variation of the curvature parameter c =20Ne/36Armantle/20Ne/36Aratmospheric (c = 3.4), using all available popping rock data (5,8). A mantle endmember of40Ar/36Ar = 32,000 is consistent with Ne-B–like (20Ne/22Nemantle= 12.7) composition (arrow connecting Fig. 1C to Fig. 1D), if the Ne-Ar correlation with all available popping rock data is considered. (A fit curve with the endmember composition preferred by Ballentine et al.—40Ar/36Armantle = 40,000,20Ne/22Nemantle=13.8—yields minimum χ2 at c = 2.4, but the χ2 are not lower, that is, these parameters do not provide a better fit than the curve plotted here.)

The constancy of the 20Ne/22Ne plateau and the similarity of that value with the average value found trapped in meteorites are important observations (1). Nevertheless, release techniques other than ball mill crushing show the mantle to have 20Ne/22Ne as high as 13.75 (4). The general conclusion that the mantle contains solar Ne (8) rather than the Ne trapped in meteorites cannot be overturned.


Response: Ballentine et al. propose that the solar neon component in both plume and MORB mantle reservoirs has 20Ne/22Ne = 13.8 ± 0.1 (1), higher than the “meteoritic” value of 12.52 ± 0.18 [Ne-B (2)] that we suggested (3), and that this component is preserved in an Icelandic basalt glass, as shown by one measurement with20Ne/22Ne = 13.75 ± 0.32 (4). This and all other available values greater than 12.52, however, can (within given uncertainties) be explained as representing the high-value tail of a statistical distribution, which must inevitably be present in view of limited analytical accuracy. Among all 315 Ne measurements that we compiled [supplemental figures 2, 3, and 4 of (3)], 74 20Ne/22Ne ratios were indistinguishable from the Ne-B value within their specific 1ς uncertainty. That data subset can be regarded as part of a set of superposed Gaussian distributions with center values indistinguishable from Ne-B, in which 65% of the values (that is, the 74 observed) agree within 1ς uncertainties with Ne-B. Another 15% (∼17 values calculated) would be expected to have values between 1ς and 2ς higher than Ne-B, which is in good agreement with the observed number of 15 values (3). The > 2ς tail, in which 2.3% of all values, or a total of ∼3, would be expected, is represented by only one value (13.75 ± 0.32).

The second, indirect argument of Ballentine et al. crucially depends on the MORB mantle argon composition. Data on the 2πD43 sample from three labs (5–9) display steep mixing arrays between an atmospheric and a mantle component with high40Ar/36Ar ratios (Fig. 1). Horizontal trends indicate a largely contamination-free mantle endmember, with average 40Ar/36Ar values of 23,700 ± 2300 (Fig. 1A), 27,800 ± 3000 (Fig. 1B), and 32,400 ± 4200 (Fig. 1C). These data point to a mantle source with40Ar/36Ar ≈ 32,000. A single value of 40,000 ± 4000 (i.e., with a 2ς deviation) can be expected with a certain statistical probability (2.3%). In turn, a MORB endmember with 40Ar/36Ar = 32,000 (Fig. 1, C and D) and Ne-B like composition can be reconciled with a χ2 fit curve (Fig. 1D) through all available popping rock data (5,8). It should be noted that the difference between ball mill crushing and pressure fracturing 40Ar/36Ar data is 4100 ± 3800, which would correspond to a20Ne/22Ne difference of 0.19 ± 0.17 (Fig. 1D).

Ballentine et al. suggest the Ne-B–like20Ne/22Ne plateaus are artifacts caused by simultaneous release of solar neon and an air contaminant from sites nonseparable by ball mill extraction. Such a hypothesis, however, has to explain why the major part of vesicularity-related atmospheric neon in basalt glasses (10) is indeed separable by ball mill crushing, as evidenced by varying 20Ne/22Ne ratios in the initial extractions (3, 5,8, 9). Only a minor part would partition into the sites that are nonseparable, but the required amounts are highly variable [e.g., 17% of the sample's mantle neon and 1.7% of the atmospheric neon for Dice 11 (3), and 5% of the sample's mantle neon and 0.1% of its atmospheric neon for 2πD43#2 (5)], and in every case a value of 12.5 would have to be produced. Such an effect appears highly improbable, because it has to occur in different rock types (dunites and basalt glasses) from different worldwide localities (the Mid-Atlantic Ridge, Loihi, Iceland) and different mantle environments (MORB mantle, plume reservoirs).

Present worldwide data display a significant decrease at20Ne/22Ne > 12.52 (3,11, 12), irrespective of whether ball mill crushing or other extraction techniques, such as stepwise heating, were used. If this is due to a proportional mixing effect, it must furthermore be asked why it yields a value of 12.5—a value that perfectly coincides with meteoritic solar neon—and not 11.9, 13.0, or some other value. We conclude that a definitive proof of solar neon (20Ne/22Ne = 13.8) in the mantle is not available and that Ne-B composition is a valid alternative that relates Earth's solar noble gas inventory to implantation by solar corpuscular radiation, with all of the implications stated in our original study (3).

  • * Present address: Mineralogisches Institut der Universität Heidelberg, Im Neuenheimer Feld 236, D-69120 Heidelberg, Germany; e-mail: E-mail:trieloff{at}


  1. 1-1.
  2. 1-2.
  3. 1-3.
  4. 1-4.
  5. 1-5.
  6. 1-6.
  7. 1-7.
  8. 1-8.
  9. 1-9.
  10. 1-10.
  11. 1-11.
  12. 1-12.

Related Content

Navigate This Article