Is lunar magma ocean (LMO) gone with the wind?

球是地球唯一的天然卫星，几千年来一直为神话、占卜、天文、文学和艺术所景仰。人类对月球形成的描述有好几百年了，但我们对月球的真正认识还是从上个世纪美国NASA的载人登月阿波罗计划11-17 (1969-1972) 开始。这些登月计划带来了珍贵的月球样品，这些样品和之后无人太空船的遥感数据为我们揭开了月球奥秘的层层面纱。
 

有关月球的成因曾经有几种假说，但目前广为认可的是“巨大陨石撞击说”。这个假说认为，地球形成不久有一个火星大小的天体与地球相撞，飞出许多碎片进入其轨道，汇聚形成月球。这一碰撞会释放巨大能量,足以融化月球外层几百公里深, 形成臆想的“全月球岩浆洋” (Lunar magma ocean，LMO) 假说。

 

这一假说实际上依赖其它假说，包括 (1)“月球高地”月壳是由低密度的斜长石岩 (CaAl2Si2O8) 漂浮在“岩浆洋”聚集而成；(2)月壳斜长石岩普遍存在的证据是月岩样品有高铕 (Eu) 元素的异常，因为和岩浆比较，斜长石岩喜欢Eu这个元素；(3) “月球低地” (月海)玄武质岩具有低铕 (Eu) 元素的异常，被认为是“岩浆洋”底部铁镁矿物堆晶再熔融的产物，与月壳斜长石岩构成铕 (Eu)元素的互补关系；(4)月岩中缺少含水矿物一直被认为是月球“撞击说”和“月球岩浆洋”说最有力的证据。
 
尽管人们暂时不情愿放弃这些“假说-支撑-假说”的循环论证，近年来的一系列研究有理、有力、充分地证明，这些普遍认可的假说必须放弃。

The Moon is the only natural satellite of the Earth, and has been the subject of numerous works of arts, literature, mythology, astrology and astronomy for millennia. The scientific speculation of its origin has also been in the literature for centuries. But it is the manned landing of Apollo missions 11-17 (1969-1972) with returned lunar rocks and soils and the unmanned post-Apollo spacecraft with remote-sensing data that allow the development of geological models on the origin of the Moon, its internal structure and its subsequent histories [1].
Several models have been proposed for the origin of the Moon, but the single generally accepted model today is the 'giant impact' hypothesis [1]. It assumes that the Earth-Moon system formed as the result of a giant impact by a Mars-sized body that collided with the proto-Earth, blasting material into its orbit to form the Moon. This Moon formation mechanism would help explain the high angular momentum of the Earth-Moon system and the small size of the iron lunar core.
The energy released during such a giant impact would have been sufficient to melt the outer few hundreds of kilometers (<1000 km?) of the Moon, forming the postulated global lunar magma ocean (LMO; Fig. 1a). The evidence for this 'LMO hypothesis' came from the highly anorthositic compositions of the lunar highland crust. The 'anorthositic compositions' refer to rocks dominated by CaO-rich plagioclase (CaAl 2 Si 2 O 8 ), which is less dense than the magma and would float to form the lunar crust and highlands (Fig. 1b) [2]. This is the 'plagioclase floatation hypothesis'. As noted [3], the 'LMO hypothesis' was initially based on a few anorthositic particles of Apollo 11 soil [2]. These two interdependent hypotheses are actually based on a single observationthe Moon's anorthositic upper crust has a large positive Europium (Eu) anomaly. Eu is a rare earth element and a positive Eu anomaly means it is relatively more abundant than its neighboring rare earth elements Sm and Gd, i.e. Eu/Eu * (=Eu N /[Sm N × Gd N ] 0.5 ) > 1. This is because plagioclase prefers Eu over Sm and Gd during its crystallization from magma. Studies of the Apollo samples also show that basaltic rocks from the lunar Maria have a negative Eu anomaly, i.e. Eu/Eu * < 1. This observation has been used to further support the 'LMO hypothesis'. Complementary to the plagioclase flotation, the dense mafic minerals olivine and pyroxene crystallized from the LMO would sink to form the lunar mantle with Eu/Eu * < 1 (Fig. 1b), whose subsequent melting would produce Maria basalts with inherited Eu/Eu * < 1. This forms the 'cumulate origin hypothesis' for the lunar mantle. Additional evidence for the 'LMO hypothesis' came from the incompatible element enriched KREEP material (enriched in potassium [K], rare earth elements [REE], phosphorous [P] etc.), which was interpreted to represent the very last drops of residual melt of the LMO solidification. An entirely independent line of evidence in support of the 'LMO hypothesis' is the lack of waterbearing minerals, which was used to suggest the Moon to be extremely dry, resulting in the general acceptance of the 'dry Moon hypothesis'. This is logical because water and other volatiles would have been entirely evaporated during the giant impact and LMO processes as also manifested by the depletion of moderately volatile elements like Sodium (Na) in lunar rocks.
The Eu-anomaly-based 'plagioclase flotation hypothesis' for lunar crust and 'cumulate hypothesis' for lunar mantle are foundation to the 'LMO hypothesis'. The dry 'Moon hypothesis' is considered as verification of the 'LMO hypothesis'. In this short communication, we briefly summarize and further elaborate some key elements of the recent work by O'Hara and Niu [4] on the 'LMO hypothesis', and challenge that this single generally accepted 'hypothesis should be entirely reconsidered because its foundation is shaken and because the 'dry Moon hypothesis' has been proven to be problematic as emphatically pointed out from the very beginning of the Apollo program by the second author [5,6] and his 107-page meticulous review in 2000 [7].
The 'wet' vs. 'dry' Moon. As one of the NASA's Principal Investigators of experimental petrology on returned lunar samples of Apollo missions, O'Hara and co-authors predicted in early 1970s that the Moon may have been a water-rich planet and may still be so in its interior [7] because water was required to explain the observed phase equilibria of lunar rocks. The best explanation was that partial melting of amphibole-bearing lunar mantle produced wet basaltic magmas. Such wet magmas underwent low-pressure gabbroic (plagioclase + clinopyroxene ± olivine ± titanium-iron-rich oxides) fractionation, effectively giving rise to the various rock types and lithologies of the Moon. The lack of hydrous minerals and depletion of volatile elements (including Na and varying species of S, C, F, Cl, etc.) in lunar samples simply resulted from volatilization loss during magma emplacement and volcanic eruption into the hard vacuum like 'atmosphere' of the Moon. Furthermore, the interpreted pyroclastic deposits on the Moon would offer a convincing line of evidence for wet magmas and hence wet lunar interior (abundant water and other volatiles). This 'wet Moon' view with experimental demonstrations has been in the literature for ∼45 years [7], but it has been entirely overlooked because this was a minority view that has been readily and quickly buried in the monopoly of the 'LMO hypothesis' [7,8]. The ability of analyzing abundant water and other volatiles in lunar glasses, minerals and melt inclusions in recent years [9][10][11][12] offers convincing evidence in support of the very old 'wet' Moon hypothesis, and effectively put the 'dry Moon' assumption to an end. For example, it has been demonstrated explicitly that the H 2 O, F and S concentrations in the primitive lunar mantle source to be at least 110, 5.3 and 70 ppm, respectively, which are similar to those in terrestrial MORB mantle [12]. This casts doubt on the 'giant impact hypothesis' as all the volatiles would have been totally evaporated and severely questions the validity of the 'LMO hypothesis'.
The Eu anomaly. The 'plagioclase flotation hypothesis' would be reasonable only if the lunar crust in general and the lunar highlands in particular all have Eu/Eu * > 1. However, this is not true [4,7]. If anything, there is a weak negative (Eu/Eu * < 1), not positive PERSPECTIVES (Eu/Eu * > 1), Eu anomaly for the lunar crust. The significant inverse correlations of Eu/Eu * with the abundances of incompatible elements such as Thorium (Th) of all the analyzed lunar samples (rocks and soils) suggest the likelihood that most lunar crustal materials have Eu/Eu * < 1 (Fig. 2a). Indeed, the gamma-ray spectrometer data [13] by the Lunar Prospector demonstrate that no more than ∼4% of the global lunar crustal materials (rocks and soils) have Eu/Eu * > 1 (Fig. 2b). This straightforward observation denies the 'plagioclase flotation hypothesis' for the lunar crust. The weak negative (Eu/Eu * < 1) Eu anomaly in the Maria basalts is a straightforward consequence of cotectic crystallization of plagioclase + clinopyroxene + olivine rather than inherited from the assumed lunar mantle [4,7], which argues against the 'cumulate hypothesis' for the lunar mantle. All these further question the validity of the 'LMO hypothesis'. The experimental studies also show the physical difficulty to separate plagioclase (to float) from the co-precipitating clinopyroxene + olivine (to sink) [7], making the 'LMO hypothesis' all the more feeble. It is possible that olivine and pyroxene might begin to crystallize to form much of the lunar mantle prior to plagioclase crystallization. This might partially reconcile the physical separation difficulty, but if so, there is no reason that the lunar mantle would have a negative Eu anomaly required by the 'LMO hypothesis'.
A recent work claims the globally widespread pure anorthosite on the Moon [14], apparently in support of the 'plagioclase flotation hypothesis'. However, this claim is inconsistent with the bulk lunar crust major element composition: 42.33 wt% SiO 2   Significant inverse correlations of Eu/Eu * with the abundances of incompatible elements such as Th in lunar materials (rocks and soils) have allowed calculations of Eu anomalies on lunar materials without REE data, an approach widely used in the literature (see [7]). (a) Modified from [4] to show such inverse correlations defined by Apollo and Luna samples [1] represented by solid circles fit with curve and equation in blue and by highland rock chips represented by the thick curve and equation in red derived from the data in [19]. These observations state that lunar materials with Th < ∼0.62 ppm have a positive Eu anomaly with Eu/Eu * > 1, whereas lunar materials with Th > ∼0.62 ppm have a negative Eu anomaly with Eu/Eu * < 1. (b) Modified from [4] to show the distribution frequency of Th concentrations obtained by gamma ray spectrometry (GRS) for the whole lunar surface (see [4]). With the total of 259 200 0.5 • -by-0.5 • bin averages, the Th concentration peaks at ∼1.5 ppm, and is variably higher for the far majority of the lunar surface materials. If Th = 0.8 ppm is conservatively chosen as the division, no more than 4% of the lunar surface would have a positive Eu anomaly with Eu/Eu * > 1. That is, the lunar crust has on average a negative, 'not positive', Eu anomaly and >96% lunar surface/crustal materials do not require the big plagioclase flotation hypothesis that is purely based on the positive-Eu-anomaly argument for the lunar petrogenesis. We refer interested readers to the high quality global Thorium data given in [13,20], especially the global average Th = 1.7 ppm (see fig. 28 of [13]), which is equivalent to a large global 'negative' Eu anomaly with Eu/Eu * = ∼0.6. Hy (orthopyroxene), 18.39 wt% Ol (olivine) and 2.30 wt % Il (ilmenite) [4]. This bulk lunar crustal composition is NOT anorthositic, but feldspathic basalt or feldspathic gabbroic composition only slightly more feldspathic than the oceanic lower crust bulk composition (despite the lost volatile elements like Na) [15]. To defend the 'plagioclase flotation hypothesis', one could argue that the globally high Th content [13] and thus the low Eu/Eu * (see Fig. 2) of the lunar crust may have resulted from compositional contamination by the lunar-wide basin formation ejecta [16], but this argument has no significance because the 'plagioclase floatation hypothesis' postulated since the Apollo program was based on lunar surface rocks already so contaminated in the early history of the Moon. Nevertheless, this contamination can help explain the observed 'mechanical mixing' of lunar compositional variation recognized by O'Hara and Niu [4].
While debate may continue for some time, we suggest that the 'LMO hypothesis' be abandoned because its foundation hypotheses (plagioclase flotation for lunar crust and mafic mineral sinking for lunar mantle) are tested to be false, and because its verifying 'dry Moon hypothesis' is also proven to be wrong. It follows that the 'giant impact hypothesis' for the origin of the Moon is severely questionable because abundant water and other volatiles remain well preserved in lunar rocks, supporting a wet, not a dry, Moon as advocated by O'Hara over 45 years ago. We emphasize that all the foregoing interdependent hypotheses need entire reconsideration in order to genuinely understand the origin and evolution of the Moon and its petrogenesis. It is beyond the scope of this short communication, but we suggest that more effort should be devoted to the effects of low pressure (due to the low gravity and thus relatively high mantle dT/dP) and hard vacuum 'atmosphere' on the lunar petrogenesis. As for the origin of the Moon, it is possible that the present-day angular momentum of the Earth-Moon system may not be used as a constraint and alternative possibilities may be explored [17]. An elegant review on the origin of the Moon from physical, chemical and isotopic perspectives is given by Halliday [18], and isotopic similarities of the Moon and the Earth offer constraints on physical models of Moon formation.