Oxygen Isotopes and the Moon-Forming Giant Impact

See allHide authors and affiliations

Science  12 Oct 2001:
Vol. 294, Issue 5541, pp. 345-348
DOI: 10.1126/science.1063037


We have determined the abundances of 16O,17O, and 18O in 31 lunar samples from Apollo missions 11, 12, 15, 16, and 17 using a high-precision laser fluorination technique. All oxygen isotope compositions plot within ±0.016 per mil (2 standard deviations) on a single mass-dependent fractionation line that is identical to the terrestrial fractionation line within uncertainties. This observation is consistent with the Giant Impact model, provided that the proto-Earth and the smaller impactor planet (named Theia) formed from an identical mix of components. The similarity between the proto-Earth and Theia is consistent with formation at about the same heliocentric distance. The three oxygen isotopes (Δ17O) provide no evidence that isotopic heterogeneity on the Moon was created by lunar impacts.

The Moon is generally considered to have formed from the debris produced in a collision between the proto-Earth and a Mars-sized impactor (1, 2). Assuming this Giant Impact model to be correct, materials from the proto-Earth, the impactor [named Theia (3)], and any additional material added to the Moon after the impact may have introduced isotopic heterogeneity as a consequence of collision and reaccretion. To test for inherent isotopic heterogeneity in the Moon, we searched for small isotopic variations in oxygen. The existing Δ17O data of three anorthosites, one dunite, one green glass clod, three mineral analyses of a single basalt, and six soil samples from the Apollo missions (4) and nine lunar meteorites (5) range from –0.204 to +0.094 per mil (‰) (Fig. 1). This large range of 0.3‰ might be because the Moon was not homogeneous, either due to incomplete mixing of material from the proto-Earth and Theia or addition of material to the Moon after the Giant Impact. Here we report the results of new analyses of 16O, 17O and 18O for 31 lunar rocks with the use of a higher precision CO2 laser fluorination technique (610).

Figure 1

Comparison between conventional and new laser16O, 17O, and 18O measurements of lunar samples. Δ17O gives displacement from the terrestrial fractionation line experimentally defined as Δ17O = δ17O – δ18O × 0.5245. Squares, Apollo Missions (4); diamonds, lunar meteorites (5); circles, this study.

The Δ17O values of 4 ferroan anorthosites, 15 mare basalts, 2 KREEP basalts, 1 norite, 1 troctolite, 3 lunar volcanic glasses, 4 breccias, and 1 meteorite analyzed for this study yield a range of –0.015 to +0.018‰ (Table 1). Compared with the existing data for lunar meteorites (5) and rocks from the Apollo missions (4), the range of Δ17O data is reduced by a factor of 10 (Fig. 1). None of the rock, glass, and breccia samples plot more than 0.018‰ off the terrestrial fractionation line (TFL), which is within the uncertainty of the technique as calculated from 10 analyses of a terrestrial olivine standard Δ17O = –0.003 ± 0.017‰ (3σ error of the mean). Samples 74220 and 79155 display larger variations in Δ17O with the use of conventional techniques (4) than are obtained with laser fluorination. The differences in Δ17O among lunar samples obtained by conventional fluorination (4, 5) relate in part to the poorer reproducibility of conventional fluorination. Lunar impacts may also have contributed chondritic material to lunar soils and perhaps created a part of the Δ17O variation in six lunar soil samples (4). However, none of the igneous rocks studied indicates the existence of an isotopically distinct reservoir within the Moon. Nor have Δ17O variations in glass and breccia samples been identified that were created by lunar impacts.

Table 1

δ18O and δ17O analyses of lunar samples. n, number of analyses; p, powdered sample; f, large fragment; 1σ, 1 standard deviation; qtz., quartz; norm., normal; ol., olivine.

View this table:

The reproducibility of laser δ18O and δ17O measurements is typically better than ±0.1‰, as documented by duplicate analyses of lunar rocks (Table 1). The δ18O of picritic, olivine-normative, quartz-normative, and pigeonite mare basalts cannot be distinguished from fresh mid-ocean ridge basalt (MORB) glasses worldwide (11), whereas high-Ti mare basalts (TiO2 > 9%) give δ18O values lower than MORB glasses. This probably reflects the high normative ilmenite content of the high-Ti mare basalts. There is a systematic difference in δ18O between lunar basalts and anorthosites that averages 0.4‰ (Table 1). This is most likely caused by equilibrium fractionation between feldspar and silicate melt. In some cases, the apparent δ18O variation among lunar rocks may reflect sample heterogeneity because the fragments analyzed were small and may therefore have been dominated by disproportionate amounts of olivine, ilmenite, or plagioclase. Only the powdered samples that were used for W isotope measurements should be considered representative of bulk rock samples because of their larger size (>1 g). Despite these reservations, the average δ18O of the lunar samples is identical to the terrestrial mantle within uncertainty.

The Δ17O average and reproducibility of all lunar rocks is 0.003 ± 0.005‰ (3σ error of the mean). On the basis of a 99.7 % confidence interval, the lunar and terrestrial fractionation lines are identical to within 0.005‰. Earth and the Moon are distinct, however, from other large planetary bodies like Mars for which the martian meteorite Δ17O line is displaced by +0.32‰, or asteroid 4 Vesta for which the howardites, eucrites, and diogenites yield Δ17O values that lie, on average, –0.28‰ below the terrestrial fractionation line (Fig. 2). Besides their differences in Δ17O, Earth and the Moon both yield δ18O ∼ 5.5‰ and therefore are, on average, more enriched in 18O than Mars [δ18O ∼ 4.3‰ (12)] or Vesta [δ18O ∼ 3.3‰ (5)].

Figure 2

Three oxygen isotope plots of lunar rocks. Composition of Martian meteorites (12) and HED meteorites (5), supposed to be fragments of asteroid 4 Vesta, are given for comparison.

Computer simulations of the Moon-forming Giant Impact indicate that the Moon inherited a significantly higher proportion of material from Theia than the proto-Earth [e.g., see (1, 2)]. The resolution of smooth particle hydrodynamic modeling of potential Moon-forming Giant Impacts has been improved recently. These high-resolution studies show that a potential impactor must have been very close in size to Mars. A significantly larger or smaller impactor would produce a higher angular momentum than observed for the Earth-Moon system or would throw too much iron into orbit, which would have created a more iron-rich Moon (2). On the basis of these high-resolution models, it has been estimated that 70 to 90% of the Moon is derived from Theia. An identical oxygen mass fractionation line for the Moon and Earth, therefore, cannot be explained by assuming that similar proportions of material came from the silicate portions of the proto-Earth and Theia. Only if the proto-Earth and Theia Δ17O values were identical to within 0.03‰ would it be possible that the average Δ17O value of the Moon plots within 0.005‰ on the terrestrial fractionation line.

Some computer models assume a larger size for the impactor, i.e., a mass ratio of 7:3 between the proto-Earth and Theia (2, 13). All these models assume that the Earth had only achieved about two-thirds of its final mass after the Giant Impact, because a larger proto-Earth would produce greater angular momentum for the Earth-Moon system than that observed. Models assuming that the proto-Earth had reached just 66% of its mass after the Giant Impact (2, 3) and identical Δ17O of the Moon and Earth require that late incoming material came from the same reservoir as the material that made up Theia and the proto-Earth. If this was another planetesimal, it must have formed from an identical mix of components as the proto-Earth and Theia. Because this is unlikely, the oxygen isotope data are easier to reconcile with a Giant Impact model involving a Mars-sized impactor.

The three isotopes of oxygen are heterogeneously distributed in the solar system (14). The largest mass-independent oxygen isotope variations of more than 40‰ are measured on minerals from calcium-aluminum–rich inclusions (CAIs) of the Allende meteorite (15, 16). The different mass-dependent fractionation lines for asteroid 4 Vesta, Mars, and the Earth-Moon system (Fig. 2) provide evidence that the average provenance of the raw material of these objects is significantly different (5). There is, however, no obvious relation between oxygen isotopes and current heliocentric distance from the sun as found for 53Cr/52Cr (17). This might indicate that oxygen isotope compositions were not monotonically zoned through the solar system or that oxygen isotope alteration continued on icy planetesimals (18). However, computer simulations of the collisional growth stage of the inner solar system (19) demonstrate that terrestrial planets were fed from a zone with a heliocentric distance of 0.5 to 2.5 astronomical units and beyond. Regardless of how heterogeneous the early inner solar system was at the beginning, it developed toward a homogeneous composition by collisional growth. This is endorsed by the small Δ17O differences of about 0.6‰ observed for the Earth-Moon system, Mars, and Vesta compared with more than 10‰ differences among chondrites. Collisional growth will smooth out pre-existing heterogeneities but is unlikely to result in identical oxygen isotopic compositions for all planets because a correlation between final heliocentric distance and average provenance of a planet is predicted (19). The differences in Δ17O among large planetary embryos and planets depend on final heliocentric distance because oxygen isotopes were heterogeneously distributed in the early solar system. Therefore, the progenitors of the Moon and Earth formed at a similar heliocentric distance. A similar orbit of Theia compared with the proto-Earth would result in a relatively small encounter velocity. This is consistent with the assumptions of most Giant Impact simulations (20).

If the proto-Earth and Theia grew from a similar mix of components at a similar distance from the Sun, then not only the oxygen isotopes but also the chemical composition and other isotope ratios should be similar. Therefore, differences between the Moon and Earth, such as the depletion of volatiles or the high FeO content of the lunar mantle compared with the terrestrial mantle, may be of secondary origin. Such differences might be produced during accretion and differentiation of the proto-Earth, Theia, the Moon, and the present Earth (21, 22).

Significant amounts of meteoritic material may have been admixed to the Moon after it formed, perhaps the equivalent of the late veneer on Earth. Although evidence of large impacts is still visible on the surface of the Moon, we have not found any indication of meteoritic material admixed to any of our lunar samples using oxygen isotopes. This does not exclude addition of meteoritic material completely but does limit the amount to several percent. The proportions can be calculated for different meteorite classes. The high (H), low (L), and very low (LL) iron ordinary chondrites are displaced from the TFL by +0.7, +1.0, and +1.3‰ on average (5). Admixing 3% H-chondritic material, for example, would be detectable when 0.016‰ deviation is considered significant (Fig. 3). Even less material from L or LL or carbonaceous chondrites would be detectable because these groups are further displaced from the TFL. Although CI-chondrites plot on the TFL, this group is characterized by high δ18O values above 16‰. Any proportion larger than 5% would increase the δ18O value by at least 0.5‰. This is also not found for any of the analyzed rocks. Planetesimals with a similar Δ17O as Mars admixed to any of the lunar samples would need to be present in proportions of >5% to be detectable using oxygen isotopes (Fig. 3). Therefore, we can exclude the possibility that >3 and >5% of these primitive meteoritic or differentiated planetesimals, respectively, have been admixed to any of the studied lunar samples. The only meteorites that are difficult to exclude are the highly reduced enstatite chondrites and aubrites. These have Δ17O and δ18O values either identical or very similar to Earth and the Moon (23,24) and, therefore, could have contributed more than 5% to lunar rocks.

Figure 3

The Δ17O values for lunar samples plot within standard deviation (2σi) error of ± 0.016‰ (long-dashed lines) on the TFL. If the impactor had formed from the same raw material as Mars or the HED parent body, then all lunar samples must have obtained, within 2%, the same portion from the impactor and proto-Earth as obtained by Earth using the triple standard error of the mean (3σmean) as significant, shown by short-dashed lines. On average, the H-chondrites plot 0.7‰ above the TFL, allowing a maximum of 3% chondritic material mixed into any of the studied lunar samples, 2σ confidence level. Other chondrite groups like L, LL, or carbonaceous chondrites show an even larger deviation from the TFL and, therefore, even less of these primitive materials can be mixed into the lunar samples. For the mixing lines in this figure, identical oxygen abundances have been assumed for all objects.


View Abstract

Stay Connected to Science

Navigate This Article