Some aspects of the minor element chemistry of lunar mare basalts

The principal minor element (including Ti) characteristics of mare basalts which must be explained by an acceptable theory of petrogenesis are reviewed. Thes include: (i) Theabsolute abundances of incompatible elements vary over a twentyfold range yet therelative abundances within this group rarely deviate by more than a factor of two from the chondritic relative abundances. (ii) The sizes of the europium and strontium anomalies show a general trend to decrease as the absolute abundances of incompatible elements decrease. This trend is also one of increasing degree of partial melting and implies that the source region did not possess intrinsic Eu or Sr anomalies. (iii) Titanium seems to behave largely as an incompatible element. (iv) Many mare basalts have Rb/Sr model ages of about 4.5 b.y. whereas their crystallization ages are 3.2–3.8 b.y.Recent hypotheses have proposed that mare basalts formed by equilibrium partial melting of pyroxene-rich cumulates which underlay and were complementary to the anorthositic crust. According to a variant of this category, the residual liquid resulting from fractional crystallization of the highlands and their complementary cumulates segregated to form an intermediate layer between the highlands and the underlying primary cumulates. This highly fractionated residual liquid crystallized to form a pyroxene-olivine-ilmenite assemblage. High-Ti mare basalts subsequently formed by partial melting of this layer, whereas low-Ti basalts formed by partial melting of the underlying cumulates. These hypotheses are examined in detail and are rejected on several grounds.A new hypothesis based upon partial melting under conditions of surface or local equilibrium is proposed. It is assumed that the moon accreted from material which had ultimately formed by fractional condensation from a gas phase of appropriate composition. The essential members of the condensation sequence with falling temperature were perovskite, melilite, spinel, fassaite, forsterite, enstatite, alkali felspar. As the gas cooled over an extended period (>100 yr) large megacrysts (> 1 m) were formed. Trace elements were partitioned into these phases according to equilibrium condensation and crystal chemical relationships. Trivalent rare earths and other incompatible elements mainly entered perovskite, most of the Eu and Sr entered melilite whilst Rb entered alkali felspar. Radiogenic⁸⁷Sr thus produced remained within the alkali felspar. The moon accreted from a mixture of these condensates to form a disequilibrium mineral assemblagewith a bulk composition similar to that of the pyroxenite source region of mare basalts as derived from experimental petrological considerations. After heating deep in the lunar interior, solid state reaction occurred around megacryst boundaries to form an equilibrium pyroxenite containing large unreacted cores of refractory melilite and perovskite. The latter mineral readily forms low melting point liquids when in contact with pyroxenes whereas melilite remains relatively inert. As partial melting commenced, all the perovskite and other low-melting accessory minerals (eg. alk. felspar) entered the first batch of liquid which thereby received most of the incompatible elements and⁸⁷Sr (but not Eu and common Sr) present in the source region. Further melting of the pyroxenite matrix occurred under conditions of surface equilibrium. As the degree of partial melting increased, the first batch of incompatible-element-rich liquid was diluted by major elements from the pyroxenite matrix whilst refractory melilite cores were gradually consumed, thereby supplying relatively constant amounts of Eu and Sr to liquids so produced. It is considered that this model is capable of explaining the principal minor element characteristics of mare basalts and is consistent with interpretations of the major element chemistry of their source region based upon experimental petrology.