PHASE EQUILIBRIUM CONSTRAINTS ON ANGRITE PETROGENESIS - IMPLICATIONS FROM AN EXPERIMENTAL STUDY IN THE SYSTEM CAO-FEO-MGO-AL2O3-SIO2

Abstract

Parameterizations of liquidus boundaries and solid solution in the CMAS + Fe system (Shi, 1992) have been employed to depict the liquidus equilibria relevant to the petrogenesis of angrites. Angrites are basaltic achondrites characterized by highly aluminous augite (fassaite), intermediate Mg-Fe olivine, and late-stage CaFe-olivine (kirschsteinite). Two important features of the equilibria on the olivine liquidus surface relevant to angrite petrogenesis are: 1) the presence of a thermal divide on the ol + aug + plag + liq boundary curve, which separates the compositions of source materials that produce low-silica angritic melts that crystallize highly aluminous augite from those that produce higher silica melts with tholeiitic to eucritic crystallization patterns; and 2) the change in the pseudo-invariant point on the low-silica side of the thermal divide from a plagioclase-peritectic involving spinel (ol + aug + plag + sp + liq) at high to intermediate Mg' (Mg/[Mg + Fe]) to two pseudo-eutectics involving kirschsteinite (ol + aug + plag + kir + liq and ol + kir + plag + sp + liq) at low Mg'.The fassaitic (aluminous augite) pyroxene composition in Angra Dos Reis (ADOR), the presence of minor green spinel, and the absence of primary kirschsteinite (Prinz et al., 1977) indicate that crystallization of the ADOR parental liquid was governed by the intermediate-Mg' set of equilibria such that, following crystallization of ol + aug + plag, the plagioclase reacted completely at the plagioclase-peritectic with the interstitial liquid, which subsequently crystallized beyond the plagioclase-peritectic onto the ol + aug + sp liquidus boundary curve. The ADOR bulk composition is consistent with trapping ~10% of the parental liquid in a cumulate with cotectic proportions of fassaite and olivine. Lewis Cliff (LEW)86010 crystallized from a liquid with Mg' similar to that of ADOR, but on the ol + plag cotectic closer to the thermal divide such that the first pyroxene to crystallize had much lower Al content than that of ADOR. In the late stages of crystallization the 86010 residual liquid (and that of LEW87051) encountered the low Mg' set of equilibria involving kirschsteinite. These relationships require either a higher degree of melting for the 86010 parent magma or source region different than ADOR's. These relationships are also consistent with compositionally dependent REE partition coefficients between fassaite and the ADOR liquid being as much as 1.5-2 times higher than those for the 86010 liquid at the onset of pyroxene crystallization. The combination of a trapped liquid component, higher partition coefficients, and smaller degrees of melting help to explain the observation that ADOR, an apparent cumulate, has REE concentrations twice as high as those in 86010 (Mittlefehdlt and Lindstrom, 1990), an apparent chilled liquid. The absence of a strong negative Eu-anomaly in the ADOR parent liquid, however, requires relatively high degrees of partial melting to eliminate plagioclase in the source region (resorption of plagioclase at the peritectic eliminates the Eu-anomaly that develops during crystallization), so ultimately different source regions are required.Progressive iron loss from devolatilized primitive chondrites (Allende, Murchison) produces source regions capable of producing a wide range of melt compositions with angritic to eucritic crystallization behavior. The compositions of carbonaceous and ordinary chondrite provide a similar range of potential source region compositions. However, primitive chondrite(+/-Fe) source regions that produce angrite-like melts have Mg' that is too low, whereas chondrite(+/-Fe) sources that have Mg' sufficiently high to yield the Mg' in angrite minerals have too much silica (or orthopyroxene) component to yield angrite-like liquids. No single group of meteorites +/- Fe simultaneously satisfies the constraints of Mg' and silica component. However, mixtures of Fe-depleted chondrite plus a low-silica component similar to Ca-Al-rich inclusions (CAIs) can satisfy the constraints. The absence in angrites of 48Ca and 50Ti anomalies, typical of CAIs (Lugmair and Galer, 1992), suggests that the low-silica component was not simply an enrichment of CAIs, but was the result of direct accretion of high-temperature condensate (Grossman, 1972) into sizable, thermally shielded planetesimals. Thus angrites cryptically record mixing of planetesimal-sized heterogeneities in the early solar system.