43rd Lunar and Planetary Conference (2012) 1738.pdf

FORMING THE FROM TERRESTRIAL SILICATE-RICH MATERIAL – 2012 EDITION. W. van Westrenen1, R. J. de Meijer2, V.F. Anisichkin3 and D.V. Voronin3. 1VU University Amsterdam, De Boelel- aan 1085, 1081 HV Amsterdam, The Netherlands ([email protected]), 2Stichting , de Weehorst, 9321 XS Peize, The Netherlands, 3Lavrentyev Institute of Hydrodynamics of SB RAS, Novosibirsk, .

The elephant in the room: Data from new mis- required size difference between Earth and impactor sions by both traditional and new space faring nations would lead to major differences in pressure- and new measurements on -era samples and -time conditions for core formation in lunar are revolutionizing our knowledge of these bodies (which predates Moon formation in all the Moon. Although many of the ‘classic’ views of the giant impact models). The resulting differences in Si surface, interior, and of the Moon, and and Hf-W systematics would be resolvable their evolution through time have changed in light of with current analytical techniques, and are not ob- these new data, the prevailing model of the formation served. of the Moon through a giant impact [1,2] continues to Resolution of this discrepancy without changing be virtually universally adhered to. one or more of the the main premises of the giant im- This is surprising, because a wide range of recent pact model requires complete isotopic homogenisation studies shows that our best estimate of lunar bulk che- of Earth and material after the impact. Turbulent mistry is completely inconsistent with dynamical mod- exchange between partially molten and vaporised els of giant impacts that reproduce the current physical Earth and Moon shortly after the impact has been in- properties and dynamics of the Earth-Moon system. voked to explain the similarity in O [14]. The Such models predict the chemical composition of the effectiveness and dynamics of this mechanism are con- Moon to differ significantly from that of the Earth. tested [15,16]. Even if this process could explain the O From a chemical point of view, an alternative Moon isotopes similarity, it is highly unlikely that such a formation hypothesis that is much easier to defend mechanism can also fully homogenise initial differenc- would result on a Moon that is simply composed of es in the isotopic compositions of much heavier, re- terrestrial silicate material with a composition equiva- fractory elements including Si, Cr, Ti, Nd, Hf and W. lent to the Bulk Silicate Earth (BSE). Here, we outline For example, [17] suggest that this scenario would one such model. Our main aim is not to convince you lead to differences in Si isotope compositions for Earth of the validity of our alternative hypothesis (although and Moon, which are not observed [6]. Given the large admittedly that would be nice); our goal is to convince and growing uncertainty surrounding the feasibility of you that (a) the giant impact model is facing serious current versions of the giant impact model in light of problems in light of a growing body of increasingly geochemical data, alternative hypotheses for lunar sophisticated chemical analyses and dynamical simula- formation need to be explored. In our view, a reasona- tions and (b) alternative models should be developed ble starting point for alternative hypotheses should be and tested. the notion that the Moon is so similar to the BSE be- Recent chemical analyses and dynamical model- cause it was formed from the BSE. Below we outline ing: Measurements of the isotopic composition of lu- a scenario that is consistent with this explanation.. nar rocks [3-6], of their Sm-Nd [7] and Hf-W [8,9] Forming the Moon from the BSE: We revisit the systematics, and the content of the lunar hypothesis that the Moon was formed directly from [10,11], demonstrate that the bulk silicate Earth and terrestrial silicate material, as first proposed in the ‘fis- the Moon show an surprisingly high degree of similari- sion’ hypothesis by [18]. According to [18], ty. This is fully inconsistent with a giant impact model the Moon originated from a hot, fast-spinning Earth. In for the formation of the Moon. High-resolution Darwin’s model, centrifugal marginally ex- smooth-particle hydrodynamic (SPH) simulations have ceeded equatorial attraction, and the Moon was formed shown that a grazing between the proto-Earth from resonant effects of solar . In the 1960s, and a -sized (Theia) can produce Earth- Ringwood and Wise updated Darwin’s hypothesis by Moon systems with the correct , , and including models for the thermal evolution and internal . Current simulations indicate that differentiation history of the Earth. They suggested 70-80% of the of the Moon originates from that core-mantle differentiation led to a reduced mo- Theia, not from Earth [12]. This discrepancy between ment of of the Earth and hence to a larger angu- geochemical data and dynamical models can not be lar velocity. The starting point for these modified resolved by proposing that the proto-Earth and the models is a proto-Earth that is rotating rapidly (rota- impactor formed at a similar distance from the , for tion period of ~ 2.7 h) with gravitational forces at the example in one of Earth’s Lagrange points [13]. The Earth’s surface only barely exceeding centrifugal 43rd Lunar and Conference (2012) 1738.pdf

forces. In this situation, a slight increase in angular the history of the Earth-Moon system [23] does not velocity would allow part of Earth’s equatorial mass to require a giant impact either: a wide range of silicate be launched into space. The main problem with this fragments is produced in our proposed scenario (see Darwin-Ringwood-Wise (DRW) model is the fact that Figure). the angular momentum of the current Earth-Moon sys- Conclusions: Moon formation models have to be tem is only ~ 27% of that required for a 2.7 h rotation consistent with lunar chemistry. Current versions of rate of the proto-Earth. In the absence of viable models the giant impact model are not. Alternative models in for a decrease in angular momentum by a factor of which the Moon is formed from terrestrial material four during Moon formation and subsequent Earth- deserve more detailed study. We provide one (radical) Moon system evolution, the ‘fission’ hypothesis was alternative hypothesis – we hope others will follow. abandoned before the first lunar sample was collected Figure: Hydrodynamic in the mission. An alternative hypothesis: We re-examined the dy- model of the aftermath namics of the Earth-Moon system and the energetics of of a nuclear explosion at the boundary be- initial Earth-Moon separation. In contrast to previous tween metallic core ‘fission’ models, our starting assumption is that the () and silicate angular momentum of the proto-Earth before Moon mantle (dark grey) of formation must have been close to (within 10 per cent an Earth-sized planet. of) that of the present- Earth-Moon system. This Top panel: t~2700s, same assumption is made in recent three-dimensional middle panel: t~4000s, hydrodynamic simulations of a giant impact origin for bottom panel: the Moon [12]. We then calculated the amount of t~13000s after explo- energy that would be required to separate a Moon from sion. Light-grey ma- the Earth in this case, and tried to identify processes terial is plasma / ga- other than giant impacts that could deliver this amount seous. Numbers denote of energy in a very short time period. Nuclear fission sizeable silicate frag- is the only known natural process that we could identi- ments. Fragment 8 fy, and we explored if and how a nuclear explosion in reaches Earth the interior of the Earth could lead to the ejection of and has Moon-like

silicate material to form the Moon. The -poor na- mass. From [22]. ture of the Moon requires that a fission-induced explo- sion leading to the launch of the Moon would have to References: [1] W.K. and Davis D.R. have occurred outside Earth’s metallic core, with the (1975) 24, 504 [2] A.G.W. and Ward W.R. core-mantle boundary (CMB) region as the most likely (1976) LPSC VII, 120 [3] Wiechert U. et al. (2001) source region [19]. Depending on details of the trig- Science 294, 345 [4] Leya, I. et al. (2008) EPSL 266, 233 gering of a nuclear explosion the amount of energy [5] Georg, R.B. et al. (2007) 447, 1102 [6] Army- required to launch the Moon is on the order 0.5-5*1029 tage R.M.G. et al. (2012) GCA 77, 504 [7] Boyet M. J. This corresponds to a small fraction of the fissile and Carlson R.W. (2007) EPSL 262, 505 [8] Touboul content of the CMB at the time [20]. The dynamic ef- M. et al. (2007) Nature 450, 1206 [9] Münker C. fects of such an explosion have been explored by hy- (2010) GCA 74, 7340 [10] Saal A.E. et al. (2008) Na- drodynamic modelling [21,22]. The Figure shows that ture 454, 192 [11] Hauri E.H. et al. (2011) Science the rapidly expanding plasma resulting from a deep 333, 213 [12] Canup R. M. (2008) Icarus 196, 518 nuclear explosion disrupts and expels overlying man- [14] Pahlevan K. and Stevenson D. J. (2007) EPSL tle, and gaseous materials, eventually resulting in 262, 438 [15] Zindler A. and Jacobsen S. B. (2009) the formation of a Moon-sized silicate body in an orbit LPSC 40, 2542 [16] Melosh J. (2009) Ann. Meteor. around Earth. This scenario by definition results in an Soc. 72, 5104 [17] Pahlevan K. et al. (2011) EPSL identical isotopic composition of Earth and Moon for 301, 433 [18] Darwin G.H. (1879) Phil. Trans. Roy. both lighter and heavier elements as well as water, Soc. 170, 1 [19] de Meijer R. J. and van Westrenen W without relying on full post-collisional equilibration. (2008) African J Science 104, 111. [20] Carlson Note that the decay products of the explosion do not R.W. and Boyet M. (2009) EPSL 279, 147 [21] Ani- end up in the Moon in these dynamical simulations. sichkin V.F. (1997) Combustion, Explosion, and Shock Our model is also consistent with the angular momen- Waves 33, 117 [22] Voronin D.V. (2007) Int J Geol 2, tum and energy of the present-day Earth-Moon system. 16 [23] Jutzi M. and Asphaug E. (2011) Nature 476, The recently proposed presence of two early in 69.