The present-day inventory of volatile elements in inner Solar System terrestrial bodies is a function of initial delivery and subsequent volatile depletion histories. Unravelling this history can, therefore, elucidate mechanisms of planetary growth and differentiation. In the case of the Earth–Moon system, the contrasting fate of volatiles, between a relatively volatile-poor Moon and a water-rich Earth, provides important evidence for their formation and evolution. Despite their clear importance, however, the fate of the volatile elements and the origin of the volatile depletion of the Moon is not well understood1,2. An important observation of Apollo lunar samples was their strong depletion in volatile elements compared with terrestrial samples3. More recent studies of the volatile content in the Moon have shown that isotopes of moderately volatile elements, such as Cl and Zn, are highly depleted, compared with bulk silicate Earth4,5,6. In addition, some lunar rocks are enriched in the heavier isotopes of Fe compared with bulk silicate Earth7. This observation was originally interpreted in terms of Fe volatilization during the giant impact7; however, igneous processes provide a more robust interpretation for these results8,9. In contrast, studies of OH contents and D/H in apatites from mare basalts, as well as Apollo pyroclastic glass beads, have argued for elevated volatile contents in portions of the lunar interior10,11,12,13. Reconciling these seemingly contrasting observations remains critical to examining models of lunar formation (for example, ref. 14) and of lunar mantle differentiation15, since a volatile-rich Moon is not a logical consequence of large-scale collisional events, or magma ocean differentiation4. To examine the origin of lunar volatiles, we employ high-precision Zn isotopes and abundances. Zinc is a moderately volatile element with a 50% condensation temperature of 726 K under solar nebula conditions16. Isotopic fractionation during magmatic processes has a negligible effect (<0.1‰ for δ66Zn; per mil deviation of 66Zn/64Zn from the JMC-Lyon standard)17. However, Zn isotopes are highly fractionated during high-temperature volatilization in planetary rocks and variations in δ66Zn larger than 1‰ in terrestrial igneous rocks have only been observed in association with evaporation events18,19. The absence of isotopic fractionation during igneous processes makes Zn isotopes powerful tracers of high-temperature evaporation during volatile depletion episodes in planets. Recent work on lunar samples has focused largely on using Zn isotopes to study mare basalts, regolith and pyroclastic glass beads6,18,20. These studies have shown that mare basalts are consistently enriched in the heavier Zn isotopes compared with terrestrial basalts, which has been interpreted as a consequence of a whole-scale evaporation event on the Moon6, or during magma ocean differentiation4. It has also been suggested that similar isotopic fractionation could possibly have occurred during eruptive degassing of volatiles from mare basalt lavas into vacuum5.

To obtain the most comprehensive understanding of the Zn isotopic composition of lunar samples, we investigated the Zn isotopic compositions and abundances of lunar crustal rocks, which include magnesian and alkali suite samples (collectively termed here as MGS) and ferroan anorthosite (FAN) plutonic rocks. MGS are interpreted to have formed within the FAN-dominated lunar crust21 and, therefore, have not experienced degassing in a vacuum. FAN samples were formed by crystallization of a plagioclase floatation crust above a magma ocean after >80% differentiation21. FANs have highly variable volatile element abundances, with Zn ranging from 0.1 to>50 p.p.m. (for example, ref. 22), possibly consistent with transport and redistribution of Zn at the lunar surface. New data are also presented for the pyroclastic green glass 15426 to further examine whether pyroclastic glasses are enriched in the light isotopes of Zn due to recondensation of isotopically light vapour18. Additional data are also reported for three lunar regolith samples and several mare basalt samples to confirm previously observed heavy Zn isotope enrichment in these lithologies6,18,20.


Lunar basalts

New data (Supplementary Table 1) for four high-Ti basalts (δ66Zn=+1.5±0.6‰, 2 s.d.) and three low-Ti basalts (δ66Zn=+1.5±0.4‰, 2 s.d.) (Supplementary Note 1) fall within the range observed previously for mare basalts6,20, and confirm that there is a consistent enrichment in the heavier isotopes of Zn compared with terrestrial basaltic rocks (δ66Zn=+0.28±0.05‰ for the bulk silicate Earth17) (Fig. 1). Combining our new data together with literature data6,20 (Supplementary Table 2), we calculate a new average value for lunar basalts of δ66Zn=+1.4±0.5‰ (2 s.d., n=26). A few exceptions (five samples out of 30) are isotopically lighter and are excluded from this average due to secondary condensation effects (Supplementary Discussion).

Figure 1: δ66Zn values for lunar rocks.
figure 1

The grey dashed box in a represents the magnified area shown in b. Lunar mare basalts, alkali and magnesium suite samples are enriched in the heavier isotopes of Zn in comparison with the terrestrial mantle composition (orange square). Ferroan anorthosites exhibit significant isotopic variability, to light and heavy Zn isotopic compositions. Isotopic data for lunar samples are from this study and refs 6, 18, 20. The bulk silicate Earth value is from ref. 17.

Lunar crust and interior

In contrast to the relative homogeneity of mare basalts, FAN samples show large Zn isotopic variability (δ66Zn between −11.4 and +4.2‰) and the light isotope enrichment is correlated with the reciprocal Zn abundance (Fig. 2). In comparison, the two MGS rocks that we measured are both isotopically heavy (Supplementary Table 1).

Figure 2: δ66Zn versus reciprocal Zn concentration in ferroan anorthosites.
figure 2

The correlation between δ66Zn and 1/Zn suggests redistribution and mixing of Zn between a surface reservoir rich in Zn that is isotopically light and a Zn poor, isotopically heavy primary magmatic reservoir.

Pyroclastic glass and regolith

The pyroclastic green glass (15426) is enriched in Zn compared with mare basalts (50 p.p.m. versus 2 p.p.m., which is in good agreement with previous studies22) and is isotopically lighter than the mare basalts (δ66Zn=−0.98‰). This result confirms consistent light isotopic enrichment of Zn in both high- and low-Ti pyroclastic glass beads18,20, with δ66Zn for the pyroclastic glasses ranging from −1 to −4.1‰. Three regolith samples are enriched in the heavier isotopes of Zn, with δ66Zn up to +6.4‰, confirming previous results18,20 that the lunar regolith is isotopically heavier than lunar basalts through impact gardening processes (Supplementary Discussion). The Zn abundances of the different samples fall within the range of previous measurements: mare basalts: 1–5 p.p.m.; FAN: 0.6–75 p.p.m.; MGS: 3 p.p.m.; and pyroclastic green glass: 50 p.p.m. (Supplementary Table 2).


Enrichment in the lighter isotopes of Zn in Apollo 17 glasses was originally interpreted as the consequence of degassing of basaltic magma during a lava-fountaining event18. During magmatic ascent, the volatile elements are thought to have vapourized, leading to light isotope enrichment through kinetic isotopic fractionation. This vapour is then considered to have recondensed and coated the bead surfaces18. The new results on the Apollo 15 green glass (15426) offer the same interpretation for this pyroclastic deposit. The green glass 15426 is coarser (with a mean radius of the grains of 100 μm) than the two Apollo 17 glasses (mean radii of 50 μm)23. Therefore, a prediction, based on surface/volume, is for more limited light isotope enrichment of Apollo 15 glass beads due to surface/volume effects, as our results show (Apollo 15426 δ66Zn−1‰, [Zn]53 p.p.m. versus Apollo 74220 δ66Zn −3.4 to −4‰ and [Zn]230 p.p.m.).

Enrichment in the heavier isotopes of Zn in lunar mare basalts confirms that most lunar mare basalts have a homogeneous Zn isotopic composition, with 25 samples (out of 30) enriched by >1‰ (δ66Zn=+1.42±0.5‰; 2 s.d., n=26). It has previously been argued that the heavy isotope enrichment in mare basalts reflects a global phenomenon, where basalts representing four different Apollo mission sampling locations (Apollo 10, 12, 15 and 17), as well as unknown regions sampled by lunar meteorite LaPaz Icefield 02205, all show nearly identical enrichments6. The heavy Zn isotopic composition in the source of the mare basalts together, with the depletion in absolute abundances of moderately volatile elements, suggests that the lunar material suffered a whole-scale high-temperature evaporation process, with the preferential partitioning of the lighter isotopes of Zn in the vapour. This evaporation process most probably occurred during large-scale melting of the Moon, such as during the giant impact6.

Highland rocks represent the primary crust of the Moon and are the oldest known lunar samples (for example, ref. 24). The large range of Zn abundances in FAN samples suggests two possible scenarios: (1) the source of the FAN had a highly variable Zn content, which is reflected in the FAN samples, or (2) the Zn abundance represents secondary redistribution of Zn at the surface of the Moon during cataclasis of the FAN, as well as mixing with some chondritic components. Partial melting does not seem to extensively fractionate Fe/Zn due to the fact that these two elements share similar ionic radii25. The five different FAN samples analysed here have variable Fe/Zn (Supplementary Table 3), which, therefore, suggests that either the source of the anorthosites had variable Fe/Zn, that Fe and Zn were fractionated during plagioclase crystallization, or that the Zn elemental variations are a result of secondary evaporative processes. Since Zn isotopes are not fractionated by >0.2‰ during igneous processes17, the large Zn isotopic variations, which are correlated with the Zn abundance (Fig. 2), indicate that the Zn systematics of FAN are due to high-temperature evaporation effects. Thus, Zn-rich FAN contains a large fraction of an isotopically light vapour, while the most Zn-poor FAN (15415) retains primary magmatic characteristics.

Apollo sample 15415 is unbrecciated, pristine26 and is enriched in the heavier isotopes of Zn (δ66Zn=4.2‰). On the other hand, 65315, which is enriched in Zn (75 p.p.m.) is a brecciated cataclastic anorthosite with exogenously derived (meteoritic) highly siderophile element additions27. Similarly, 67955 is also a brecciated sample, formed from an impact melt, with a significant component of meteoritic siderophile elements28 and has disturbed Zn isotope systematics. These lines of evidence indicate that the brecciated samples incorporated isotopically light Zn-rich vapour during cataclasis or residence on the lunar surface. Transport of Zn at the surface of the Moon, during later impact bombardment of the crust and during the assimilation of exogenous Zn, implies that the Zn isotopic composition of cataclastic FAN samples do not represent the composition of their source reservoirs.

Significant enrichments in the heavy isotopes of Zn observed in the MGS, which solidified within the lunar crust, point towards a higher volatile depletion in their source than mare basalts. The origin of the MGS is generally interpreted as magmatic intrusions within the FAN crust (for example, ref. 21), formed by either a very high degree of partial melting29 or by assimilation of FAN crust30. Evaporation and condensation in a vacuum at the surface of the Moon has, therefore, played no role in the composition of these samples. Enriched heavier isotopes of Zn observed in the MGS points towards a source already depleted in volatile elements. In turn, this provides strong support to loss of Zn and moderately volatile elements during a whole-scale evaporation of the Moon early in its history either following the giant impact or during a lunar magma ocean phase4,6, and not locally during degassing, as has been proposed for some of the mare basalts (ref. 5). A third scenario would be that the impactor was already depleted in volatile elements before the giant impact (for example, refs 31, 32). The new data cannot constrain this scenario; however, a second process of volatile loss is required to explain the differences in Zn isotopic composition between mare basalts and pristine highland samples. A more likely scenario is, therefore, that volatile element of the Moon occurred during lunar formation and differentiation.

An outstanding question is how Zn and other moderately volatile elements were lost from the Moon, while water, which is far more volatile (for example, ref. 32), seems to have been retained in some samples (for example, refs 10, 11, 12, 13). It has been suggested that hydrodynamic escape after the giant impact was not a viable mechanism to explain the loss of water (or potassium) from the Earth–Moon system and that the volatile element content of the Moon was inherited from the precursor material33. However, this model fails to explain the extreme volatile depletion of the mare basalt source rocks, which are the most representative derivative rock type from the lunar mantle, or the later evolution of volatiles in the Moon revealed from crustal rocks. In a two-phase vapour-melt numerical model of the dynamical and thermodynamical evolution of the proto-lunar disk34, where the vapour phase is viscous due to magnetorotational instability, the volatile elements are predicted to accrete to Earth, leaving the material that formed the Moon enriched in refractory elements and correspondingly depleted in volatiles. Calculations of the chemical composition of the vapour of the lunar protoplanetary disk35 indicates that Zn enters the volatile phase of the disk at temperatures <3,000 K, leaving the Moon depleted in Zn and more volatile elements. Our new results support this mode of origin, but also suggest that continued volatile depletion in a lunar magma ocean is required4 to explain differences in δ66Zn between mare basalts and pristine lunar crustal rocks.


Sample preparation

One-gram chips of samples (Supplementary Note 1) were powdered with an agate mortar and pestle. Samples were then dissolved in a mixture of HF/HNO3 for several days in Teflon beakers at 140 °C. Zinc purification was achieved in a 0.25-ml column using AG-1X8 (200–400 mesh) anion-exchange resin. After applying the sample solution in HBr onto the resin, further HBr was added to remove elements causing matrix effects. Purified Zn was collected in a HNO3 solution following the procedure of ref. 18. The resulting Zn fraction was further purified in smaller 100-μl columns following the same ion-exchange chromatography procedure.

Sample analysis

Samples were analysed using a Thermo-Fisher Neptune Plus multicollector inductively coupled plasma mass spectrometer at the Department of Earth and Planetary Sciences, Washington University in St Louis. Masses 62, 63, 64, 65, 66, 67 and 68 were measured by Faraday cups positioned, respectively, and the intensity of 62Ni was measured to control and correct the isobaric interference of 64Ni (ref. 6). All samples fell on the mass-dependent fractionation line (Supplementary Fig. 1), and the typical analytical uncertainty (2 s.d.), is 0.04‰ for δ66Zn and 0.05‰ for δ67Zn and δ68Zn. Instrumental mass bias was corrected by bracketing of the standard and the sample and is further discussed in ref. 17. The Zn concentrations were also obtained from multicollector inductively coupled plasma mass spectrometer data by comparing the intensity of the sample signal with that of the standard, with an estimated precision of ±10%.

Additional information

How to cite this article: Kato, C. et al. Extensive volatile loss during formation and differentiation of the Moon. Nat. Commun. 6:7617 doi: 10.1038/ncomms8617 (2015).