The Lunar Impact Vapor Paradox

NLSI Lunar Science Conference (2008) 2123.pdf THE LUNAR IMPACT VAPOR PARADOX. Paul H. Warren1, Edward D. Young1,2 and William I. Newman2,3,4, 1 Institute of Geophysics and Planetary Physics (pwarren@ucla.edu), Departments of 2Earth and Space Sciences, 3 Physics and Astronomy, and 4Mathematics, University of California, Los Angeles, CA 90095-1567, USA Introduction A large fraction of the Moon’s crater-pocked crust, about half of the surface rock material, consists of impact-melt breccia. Physical modeling [e.g., 2,3] implies that major impacts generally produce vapor along with melt, in a roughly 1:9 mass ratio. Yet on the Moon, notwithstanding the recent discovery of impact-vapor condensates in Apollo-14 regolith breccia 14076 [1], instead of being roughly 1/9 as abundant as impact-melt breccia, condensates appear to be less than 0.001 times as abundant. The order of magnitude of the impact-melt inventory is scarcely in doubt. Why is the abundance of vapor condensate in the sampled lunar megaregolith so disproportionately low? The 14076 vapor condensates Caveat: 14076 is relevant as the only significant sampling of lunar vapor condensates; and because from their distinctive range of compositions we infer that lunar condensate should in general be easy to identify. However, the 14076 condensates are not necessarily representative, e.g., in terms of spheroid sizes, of all lunar condensates. The anorthositic-regolith half of 14076 (total mass 2.0 g) is uniquely endowed with high-alumina, silicapoor (HASP) material of impact evaporation-residue origin [4,5]. Complementary vapor condensates, dubbed [1] GASP (gas-associated spheroidal precipitate), occur both as spheroids, up to 5 m across, and as clasts up to 200 m. All GASP is distinctively depleted in the same refractory major oxides that are characteristically enriched in HASP: Al2O3 and CaO. Among the clasts, excluding two instances of obvious contamination by Na-rich substrate-derived melt, bulk Al2O3 averages 0.3 wt%. Pyroxene compositions are also weird; e.g., En82Wo0.45 with 0.07 wt% Al2O3. These materials manifestly condensed from impact vapor associated with the 14076 HASP. Bulk GASP spheroids range in mg from 7 to 84 mol%, and in FeO/SiO2 (wt.) from 0.002 to 0.67 [1]. Bulk compositions and some textures (e.g., lobate boundaries between silicic and mafic domains) suggest liquid immiscibility at ~ 1680°C. Compositional data show remarkably little mixing between clasts (former mush-puddles) of GASP and HASP. Apparently condensation did not commence until after an intermediate stage of neither net evaporation nor net condensation, during which expansion of the vapor cloud carried the eventual GASP matter well apart from the HASP. Expected total yield of lunar impact vapor At least 10 lunar impact basins have major ring diameter between 700 and 2500 km [6]. Basins form by enlargement of an original “transient” crater with diameter DTC. Both DTC and the volume of impact melt, Vmelt, are commonly modeled as simple functions of impact energy. Several different proposed approaches for scaling between DTC and final D for complex craters give results consistent to within ~10% relative [7]. For scaling between DTC and Vmelt, high-resolution hydrocode models [8], corrected to vertical impact, confirm the a power law relationship proposed by [9]: Vmelt = c DTC3.85 (1) where c is a constant related to impactor density and velocity. For nonterrestrial application, results from (1) must be scaled to g0.83 [2]. Combined with the relationship between DTC and final D [7], (1) implies that melt+vapor yield is proportional to D3.41. The size distribution of lunar craters scales as D–b, where b is ~2.4 [e.g., 2]. Comparison between these two expoments (3.4 and 2.4) implies that a handful of large basins must account for the vast majority of all the melt+vapor ever produced by lunar impacts. Translation from Vmelt into mass of impact vapor mvap depends on impact velocity vi, but typical vi during the late heavy bombardment (LHB) were presumably at least as high as the modern average asteroidal-lunar vi of 16 km/s [10]. One LHB model [11] invokes highinclination impactors, of which 84% hit at ≥20 km/s. The widely cited LHB model of [12] implies vi averaged ~23 km/s. Hydrocode experiments [e.g., 3] indicate that at 15 km/s, mvap/mmelt ≈ 0.10. Assuming average mvap/mmelt ~1/9, the total mass of vapor from the 10 largest known lunar impact basins [6; discounting Procellarum] is 4×1019 kg. If efficiently condensed into rock, this would amount to 0.9 vol% of the Moon’s crust. Another way of stating the problem: Roughly half of the lunar megaregolith, by mass, consists of impact-melt breccia, within which (on average) slightly more than half represents quenched impact melt. The corresponding proportion of impact-vapor condensate, if mvap/mmelt was ~1/9 and condensation was efficient (and burial was no more efficient than in the case of impact melt), would be 3 wt% of the megaregolith. The paradox Yet sample obseravtions suggest that condensates constitute less than 0.01% of the megaregolith. In 14076, clasts of GASP are up to 200 m across; HASP up to 340 m. The total number of comparable-sized lunar highland clasts and regolith fragments that have been scrutinized in thin-section petrologic surveys without ever, apart from 14076, detecting GASP (or HASP >>30 m across) is conservatively 500,000 [e.g., a total of 200,000 from just three sur- NLSI Lunar Science Conference (2008) 2123.pdf veys: 13-15]. Even assuming a very low efficiency of detection, say 10%, it seems clear that the total proportion of GASP-like condensate in the megaregolith is <<0.01% and probably <0.001%. Among sub- m sized regolith spheroids, ~15% have GASP-like “VRAP” (volatile-rich alumina-poor) compositions [16]. But the total sub- m component of an average surface soil is only 1 wt%. Many soil grains have thin (<200 nm) rims that may be partly vapor condensate [17]. However, most soil-grain rim matter probably forms by solarwind sputtering [18], or from vapors produced by micrometeorite impacts into regolith [19]. Assuming an average soil grain has the equivalent of a pure condensate rim 10 nm thick, and factoring in the average grain-size distribution of lunar soil, rim condensates constitute ~0.2 wt% of the total regolith (sensu stricto, an ~10 meter thick layer). The combined proportion of VRAP and grain-rim condensate in the megaregolith (several km thick) is probably <<0.01 wt%. In summary, the observed abundance of lunar condensate appears low by at least 2 orders of magnitude. Condensates, quo vadis? We [20] envisage two possible mechanisms for engendering the minimal presence of condensates in the Apollo+Luna samples: (A) The process of condensate formation and deposition might in principle be highly inefficient, relative to escape of vapor and entrained condensates from the Moon’s gravity. An estimate from Melosh [2] (unfortunately without any explanation of derivation) is that the fraction of an impact vapor cloud that never condenses may be “up to 50 percent.” Applying the kinetic theory of gases, and assuming that the vapor becomes saturated at T ~ 4000 K and P ~ 2 MPa [21], typical molecules will be undergoing roughly 1010 collisions per second — ample for efficient condensation. Condensates formed from vapor expanding at a velocity vexp in excess of 2.38 km/s (the lunar vesc) will generally never land on the Moon. But the fraction of the gas expanding at vexp < vesc is far in excess of the ~0.01 needed to explain the condensate paradox. (B) Condensates may have formed and landed in abundance, but some quirk(s) of the depositional process led to their destruction and/or burial. In particular, we postulate that condensates are undersampled because in large impacts (responsible for most of the vapor) they tend to land so close to ground zero that they mostly end up as a trace, cryptic component assimilated into melt formed by the same impact; or else end up assimilated into, or simply covered by, mare basalt. This model assumes that the largest lunar impacts engender very large proportions of melt, equivalent to 1/4 to 1/2 of the volume of the transient crater [e.g., 2]. We further assume that, by a combination of direct ejection, sloshing, and buoyant rise, most of the surface of the final, collapsed basin (with D ~ 2DTC) was soon covered by hot impact melt. Provided most of the condensates landed no further than the final basin rim, the biggest batches would settle while still hot onto (or very close to) still-turbulent impact melts, where they would be efficiently assimilated into the nascent matrices of impactite breccias; and thus never survive as discrete, recognizable materials. The keys to efficiency for mechanism B are twofold. First, compared to classical ideal-gas models [e.g., 22,23], expansion of the vapor must be several times slower, as suggested by [24]. This has a dampening effect on mechanism A, but (A) seems completely inadequate, anyway. Reasons cited [24] for slow expansion are the reduction in sound speed that results from initial condensation, and the far from spherical geometry of the initial vapor distribution. In a realworld expansion, two phases are probably present even before the onset of condensation, because at typical vi (~20 km/s) in a basin-scale event, most vapor probably forms beneath many km of matter that never undergoes more than 50% vaporization. The other key premise for efficiency of (B) is that expansion flow of the relevant (vexp < vesc) vapor is efficiently channeled toward nearvertical trajectories by the wall of ejecting impact melt that surrounds the nascent crater [e.g., 25,26]. Mechanism B is most efficient if vi is not drastically higher than the modern asteroidal-lunar average of ~16 km/s. References: [1] Warren P. H. (2008) GCA, in press. [2] Melosh H. J. (1989) Impact Cratering. [3] Pierazzo E. et al. (2001) In Impact Markers in the Stratigraphic Record (6th ESF Impact Workshop). [4] Vaniman D. T. & Bish D. L. (1990) Am. Min. 75, 676. [5] Papike J. J. et al. (1997) Am. Min. 82, 630. [6] Wilhelms D. E. (1987) The Geologic History of the Moon. [7] McKinnon W. B. et al. (2003) In Impact Cratering: Bridging the Gap Between Modeling and Observations, LPI abst. 8015. [8] Ivanov B. A. & Artemieva N. A. (2002) In Catastrophic Events and Mass Extinctions: Impacts and Beyond (GSA SP 356). [9] Grieve R. A. F. & Cintala M. J. (1992) Meteoritics 27, 526. [10] Ivanov B. A. (2001) Space Sci. Rev. 96, 87. [11] Morbidelli A. et al. (2001) MPS 36, 371. [12] Gomes R. et al. (2005) Nature 435, 466. [13] Vaniman D. T. et al. (1976) PLSC 7, 199. [14] Simon S. B. et al. (1978) PLPSC 9, 1813. [15] Simon S. B. et al. (1982) PLPSC 13, A232. [16] Keller L. P. & McKay D. S. (1992) PLPSC 22, 137. [17] Keller L. P. & McKay D. S. (1997) GCA 61, 2331. [18] Bernatowicz T. J. et al. (1994) Science 264, 1779. [19] Taylor L. A. et al. (2001) MPS 36, 285. [20] Warren P. H. et al. (2008) Icarus, submitted. [21] de Niem D. (2002) In Catastrophic Events and Mass Extinctions: Impacts and Beyond (GSA SP 356). [22] Raizer Y. P. (1960) Sov. J. Exp. Theor. Phys. 37, 1229. [23] Moses J. I. et al. (1999) Icarus 137, 197. [24] Melosh H. J. & Pierazzo E. (1997) LPS 28, abst. 1369. [25] Ivanov B. A. (2005) Sol. Sys. Res. 39, 381. [26] de Niem D. et al. (2007) Plan. Space Sci. 55, 900.