- Split View
-
Views
-
Cite
Cite
M. J. O’HARA, Flood Basalts and Lunar Petrogenesis, Journal of Petrology, Volume 41, Issue 7, July 2000, Pages 1121–1125, https://doi.org/10.1093/petrology/41.7.1121
- Share Icon Share
Abstract
The popular interpretation of lunar maria, as sequences of primary picritic flood basalts derived by remelting a mantle that had accumulated from a global magma ocean, has many unsatisfactory aspects. The expertise of Keith Cox would have been valuable in their interpretation.
BASALT PETROGENESIS 1950–1975
Few among us will ever see more basalt at a glance than one can see any fine night with a full Moon. The rocks of the lunar maria have been likened to terrestrial continental flood basalts (CFB), Keith Cox’s chosen field of specialization. Basaltic magmas have played a very important role in planetary crust formation (see Basaltic Volcanism Study Project, 1981) and basaltic igneous activity has predominated on at least seven extra-terrestrial bodies in the Solar System (e.g. Shearer et al., 1998; Beatty et al., 1999). Samples of at least four of these bodies are available through meteorite discoveries, or through manned and unmanned mission returns, yet in the 40 years of the existence of Journal of Petrology it has carried remarkably few papers dealing with the petrology and petrogenesis of extra-terrestrial basalts. The extent of compartmentalization is illustrated by the paucity of discussion of extra-terrestrial igneous materials in standard texts such as those by Carmichael et al. (1974) and Wilson (1989) and the virtual absence of reference to the lunar meteorites in recent comprehensive accounts of lunar materials by Taylor et al. (1991) or Papike et al. (1998).
For almost 40 years Turner & Verhoogen (1951, 1960) and Carmichael et al. (1974) provided the standard texts for igneous petrologists and geochemists. Ocean island basalts (OIB), mid-ocean ridge basalts (MORB) and CFB were each advanced as primary magmas. However, petrographic, major element and experimental petrology data were challenging this simplified view. Data for Hawaiian tholeiitic OIB demonstrated that the common basalts were differentiated and that only the picrites could be primary magmas (Yoder & Tilley, 1962; O’Hara, 1965). MORB bore the petrographic imprint of partial crystallization and magma mixing (Muir & Tilley, 1964). Flood basalts displayed fractionation and modification in low-pressure environments (Cox et al., 1965, 1967; Cox & Hornung, 1966). Karoo flood basalt compositions were modified towards low-pressure plagioclase-saturated cotectic character, the characteristic imprint of modification of the liquid compositions by partial crystallization at low pressures.
The debate quickened during the run-up to the first lunar sample return with the publication of experimental studies adopting a strictly primary magma standpoint with regard to common basalts (Green & Ringwood, 1967). Other studies simultaneously showed that partial melts of a dry peridotite mantle at high pressure could only be picritic, not basaltic in composition (O’Hara & Yoder, 1967). A synthesis of all available experimental petrology data emphasized this for basalts generally (O’Hara, 1968a) and for MORB and Hawaiian OIB specifically (O’Hara, 1968b; Jamieson, 1970).
Common basalt compositions, however, carried a trace element signal that could not be reconciled with closed system perfect fractional crystallization, but was consistent with equilibrium between crystals and liquid at low but variable mass fractions of melt (Gast, 1968). Gast played a major role in the inception, design and implementation of the lunar sample science programme, being chairman of the Lunar Science Application and Planning Team from 1969, Chief of the NASA Planetary and Earth Sciences Division at the Manned Spacecraft Centre, Houston, from 1970 and an active Principal Investigator throughout.
The debate ran throughout the period of the lunar sample returns and by public acclaim effectively discounted the field, major element, petrological and experimental data for terrestrial basalts in favour of one particular interpretation of the trace element data (Carmichael et al., 1974; Taylor, 1975). This conclusion was, however, founded on two implicit but erroneous assumptions. The partial crystallization processes that could give rise to the low-pressure cotectic character of common basalts are not synonymous with closed system perfect fractional crystallization. Low mass fractions of equilibrium or aggregated fractional partial melting are not the only processes that can account for the trace element geochemistry.
LUNAR EXPLORATION PROGRAMME
Selection of Principal Investigators by NASA for the anticipated lunar sample returns took place early in 1967. K. G. Cox, discoverer of the first armalcolite-group mineral in terrestrial flood basalts (Von Knorring & Cox, 1961), was already an international figure committed to a role for crustal modification in the field of flood basalt studies (Cox et al., 1965, 1967; Cox & Hornung, 1966) but was excluded from the group chosen despite the first recovered lunar mare basalts being widely held to be analogous to terrestrial flood basalts (Taylor, 1975). Cox and co-workers published several more major papers on flood basalts during the period of the lunar landings, and more than 20 papers on flood basalt petrogenesis, detailed elsewhere in this volume, were to follow. Many of these papers have been extensively cited in the literature, but Cox was not a member of the Lunar Science Institute’s Basaltic Volcanism Study Project (1981), which later addressed a substantial effort to terrestrial CFB volcanism, and none of his contributions is referenced in two recent summaries of lunar work (Papike et al., 1998; Shearer & Papike, 1999). Cox’s potential contributions to lunar petrogenesis may prove to have been sorely missed.
The following digest of the lunar petrogenetic debate is in extended abstract form without discussion or references. The full version of this material will be published in a separate issue of this journal.
The Apollo 11 hand specimens recovered in mid-1969 were extremely titanium-rich mafic to picritic basalts in which quench textures were abundant. The views that dominated the subsequent discussions were that these specimens came from and were representative of flood basalts, that they were quenched liquid compositions, and that they were primary partial melts from the lunar interior. Samples from five later missions arrived within 24 months of the first Lunar Science meeting, allowing little time for mature reflection. Interpretations were maintained compatible with the original viewpoint. The broad outlines of the ‘established’ view of lunar petrogenesis were thus determined in a hothouse atmosphere within no more than 12 months in 1969–1970 and with an ignorance of the detailed field relationships of the samples that persists to this day. Unlike the situation with regard to terrestrial basalts, the interpretation reached in 1970 has undergone no major modification in the ensuing 30 years.
The concentrations of highly incompatible trace elements in the lunar mare basalts ranges from 10 to 100 times chondritic and is consistent with their representing 1–10% partial melts of some chondritic source region if they are primary magmas, melt fractions no higher and generally lower than are widely accepted for terrestrial MORB. Different, generally lower, values for the required mass fraction of melting arise if the source region is postulated to be a cumulate from a global magma ocean. Those mare basalt samples richer in incompatible elements display increasing negative anomalies in the concentration of Eu relative to other rare earth elements (REE), indicative of substantial separation of plagioclase feldspar from the system at low oxygen fugacity somewhere in their history.
However, the picritic mare basalt hand specimens display, on cooling from the all liquid state, the silicate crystallization sequence of olivine joined in most cases by one or both pyroxenes long before plagioclase appears, with one or more oxide phases (spinel, ilmenite, armalcolite) also commonly present from an early stage. This late appearance of plagioclase during the cooling history is enhanced at higher pressures. If the hand specimens are to represent the primary or parental magma, the negative Eu anomaly cannot be the product of plagioclase (gabbro) separation from liquids of those compositions anywhere within the Moon. Therefore, if a primary magma model was to be retained, the source region had already to contain an in-built negative Eu anomaly. A global magma ocean several hundreds of kilometres deep was postulated to have formed by impact heating during a rapid accretional phase at the birth of the Moon. Cooling was postulated to be accompanied by flotation of plagioclase crystals in the dense magma away from plagioclase-saturated ferro-magnesian minerals accumulating mainly at the floor. Europium, preferentially gleaned by the plagioclase crystals, was selectively removed from the evolving magma and from the ferro-magnesian cumulate mantle that it was precipitating, creating a plagioclase- and Eu-depleted source rock from which the basalts could later be derived by renewed melting. This model required that at the time of derivation of the <3·9-Gy-old mare basalts the average lunar highland crust contained a substantial positive Eu anomaly, existence of which has been widely publicized in a multitude of places (e.g. Taylor, 1975).
Identification of this positive anomaly rested on the correlation in returned highland samples between Th concentration, measurable over great areas by gamma-ray counting from orbit, and REE concentrations. Absolute Eu concentrations vary little in lunar samples whereas those of the other REE vary greatly. The lower the Th, the lower the total REE concentration, and the greater the likelihood of Eu being enriched relative to the other REE. However, the existence of this positive anomaly is not firmly established by the original sample dataset when combined with Th values in the equatorial belt overflown by the Apollo 15 gamma-ray spectrometer, or by a greater array of more recent data, combined with the inferences about Th contents to be drawn from the latest remote sensing of the entire lunar surface. The average lunar highland crust has a small negative Eu anomaly rather than the requisite large positive anomaly. The problem is not alleviated by the observation that the highland feldspathic basalt component rich in potassium, REE and phosphorus (KREEP) is special, is concentrated around the Imbrium basin, and carries most of the REE signal in the average highlands. KREEP was part of the average lunar crust 200 My before the mare basalts were extracted from the remaining lunar mantle and if the average crust contained no Eu anomaly at the time of basalt derivation, neither did the average mantle.
Plagioclase may have near neutral buoyancy in basic magmas yet in terrestrial magma bodies it accumulates neither by crystal flotation nor by residual phenocryst enrichment. Cumulate rocks from the disrupted Eucrite Parent Body show no evidence of plagioclase flotation. In the latter, as in terrestrial gabbro bodies, both plagioclase and pyroxene have crystallized together, probably mainly at the floor. Dense olivine, orthopyroxene and spinel are closely associated with plagioclase in lunar highland troctolites and norites. The reality of the key process of plagioclase flotation from a global magma ocean has yet to be substantiated.
The subsequent remelting of parts of the postulated cumulate pile in the lunar mantle also presents difficulties. Once the cumulate has consolidated as a result of global cooling it is unlikely to undergo renewed partial melting in the same environment. The postulated cumulate pile would be gravitationally unstable (chemically denser and cooler cumulates formed towards the top, provided plagioclase had floated) leading to the possibility of renewed partial melting of hot plumes rising towards the surface or melting of anchors of dense cumulates sinking into hotter regions. There is, however, no evidence of the crustal deformations that could be expected to accompany such events.
The premise that primary magmas were erupted in abundance was widely accepted in the 1960s and 1970s. It was particularly seductive because it immediately invested a sample with fundamental importance. It is a stepping stone to mantle source mineralogy, pressure of formation and the mantle dynamics leading to its partial melting. Breathtaking vistas in planetology and cosmology unfold before the possessor of such a talisman. But primary magmas have uncomfortable attributes. Investigation of their phase equilibria at high pressures constrains the permissible mineral assemblage and mineral chemistry (but not the mineral proportions) of the mantle residue from which they may have separated at any postulated depth.
If the residual mantle is required to be at least bimineralic (olivine, pyroxene), and the mass fractions of melting are postulated to be low, the phase equilibria for each sample define a unique depth and temperature within the lunar mantle at a specific time. The assemblage of these depth–temperature points for all the primary magmas might define an approximate palaeo-selenotherm. The possible selenotherm at 3·5 ± 0·3 Ga so defined by the putative primary magmas is grossly super-adiabatic in the Moon, yet the extensive surface tectonic deformation that might result from this situation is manifestly absent.
To account for the negative Eu anomaly in the manner postulated, the cumulate mantle at the required pressures would have to contain plagioclase-saturated, alumina-rich pyroxenes, almost certainly accompanied by traces of plagioclase crystals. The phase petrology of an assemblage of true primary magmas at higher pressures, however, also defines the permissible ‘occult’ phases with respect to which a mineral assemblage is just saturated, even if that crystalline phase is not present in significant amount. The remoteness from plagioclase saturation in liquids of hand-specimen composition at low pressure is exacerbated by increase of pressure. Liquids of the compositions of the mare basalt hand specimens are far from equilibrium with plagioclase or a plagioclase-saturated mineral assemblage at any depth and, therefore, cannot be primary magmas formed in the way required by the model.
Small mass fraction partial remelting products formed at 0·2–1·5 GPa (60–450 km depth in the Moon) from a mineral assemblage close to plagioclase saturation at those depths should yield liquids characterized by the low-pressure crystallization sequence of olivine joined next and for a considerable interval by plagioclase, to be followed then by calcium-rich pyroxene before calcium-poor pyroxene. Terrestrial MORB, which display precisely this crystallization sequence at low pressure, are thought to be formed within the above pressure range by generally greater mass fractions of partial melting of mantle mineral assemblages that may not even be plagioclase saturated, both factors that will not enhance plagioclase relative to pyroxene crystallization from the liquids at low pressure. The low-pressure phase petrology of the mare basalt hand specimens (and that of the various green and red glasses) is inconsistent with their postulated modes of origin.
The average composition of the basaltic target rock converted to regolith at each site and across the vast majority of the mare surfaces is much more feldspathic than that of the hand specimens, casting further doubt on the identification of the latter as true magma compositions. If average regolith compositions approximate to the average erupted magma composition at each site, the mare lavas were close to simultaneous saturation with plagioclase and several other silicate and oxide phases on eruption. A similar situation is widespread in terrestrial basalts and those of the Eucrite Parent Body. Gabbro fractionation from some more primitive composition within the lunar crust could then afford a satisfactory explanation for the average lava compositions and their negative Eu anomalies, obviating any requirement for a global magma ocean. As compositions modified by low-pressure processes, the basalt samples have much to say about crust-forming processes of immediate interest to the petrologist, but little to say about the lunar mantle until those near-surface processes have been unravelled and their effects stripped away.
The very flat surfaces and vast lateral extents of mare basalts imply very high eruption rates during individual eruptions. Very few flow fronts are detectable anywhere on the maria and the morphology of some mare fillings is equally suggestive of giant lava lakes. Overall eruption rates in the early and middle phases of the activity, on which we have few constraints, will have determined whether the filling was a long succession of individually cooled flow units, predominantly a succession of thick sills and intercalated lavas, or a single deep, slow-cooled lake of lava. The late surface eruptions at a single site may, however, span 250 My, two orders of magnitude greater than the time span associated with the major eruptive phase of terrestrial CFB volcanism. This is just one of several points of distinction between lunar mare volcanism and terrestrial CFB volcanism.
Terrestrial CFB may undergo extensive fractionation and crustal assimilation in deep-seated magma chambers. Contamination and assimilation, such popular processes 70 years ago, were unfashionable as factors in basalt petrogenesis at the time of the Apollo missions, but have since been extensively rehabilitated. The incompatible trace element signal in the mare basalts, around which much of the debate revolves, may have been significantly altered by assimilation of a few percent of KREEP-rich materials in the crust or incorporation of KREEP-rich regolith during their flow across the surface. These complications remain to be evaluated.
Solutions to most of the problems that then remain regarding the hand specimens and the various green and red picritic glasses must be sought in some combination of ferro-magnesian phase enrichment by quench-crystal sinking, impact remelting of cumulate horizons and selective volatilization on eruption into hard vacuum. The debate would be greatly advanced if we had access to a few continuous 100-m-long cores or sections through the basalts in the circular maria basins.
Io, the innermost moon of Jupiter and currently the body displaying the highest rate of volcanic activity in the Solar System, is similar in size and density to the Moon. The style of eruption seen later at Io was unknown when the Apollo missions returned their samples. It is extremely doubtful whether the scientific community would have so decisively rejected a role for selective volatilization in the evolution of the lunar samples had Io been visited first, yet there is almost no mention of Io in the latest reviews of lunar petrogenesis, and as little mention of the Moon in recent studies of volcanism on Io. Selective volatilization from small basaltic liquid particles during fire-fountaining into hard vacuum leads to losses of sulphur and sodium in particular. It produces low oxygen fugacities in the remaining liquid, and a bias towards normative anorthite and hypersthene in the silicate composition. These are precisely the major chemical differences between lunar and terrestrial basalts. If eruption processes on the Moon were similar to those on Io it may be excessively optimistic to seek any primary or parental liquids among the returned samples.
*e-mail: sglmjo@cardiff.ac.uk
REFERENCES
Basaltic Volcanism Study Project (
Beatty, J. K., Petersen, C. C. & Chaikin, A. (eds) (
Carmichael, I. S. E., Turner, F. J. & Verhoogen, J. (
Cox, K. G. & Hornung, G. (
Cox, K. G., Johnson, R. L., Monkman, L. J., Stillman, C. J., Vail, J. R. & Wood, D. N. (
Cox, K. G., MacDonald, R. & Hornung, G. (
Gast, P. W. (
Green, D. H. & Ringwood, A. E. (
Jamieson, B. G. (
Muir, I. D. & Tilley, C. E. (
O’Hara, M. J. (
O’Hara, M. J. (
O’Hara, M. J. & Yoder, H. S. (
Papike, J. J., Ryder, G. & Shearer, C. K. (
Shearer, C. K. & Papike, J. J. (
Shearer, C. K., Papike, J. J. & Rietmeijer, F. J. M. (
Taylor, G. J., Warren, P., Ryder, G., Delano, J., Pieters, C. & Lofgren, G. (
Turner, F. J. & Verhoogen, J. (
Von Knorring, O. & Cox, K. G. (
Wilson, M. (