Formation of an unusual compact Type A refractory inclusion from Allende

Abstract

Abstract— We report the results of a study of TS2, an unusual compact Type A inclusion from Allende. A distinctive, major feature of this inclusion is that many of its melilite crystals have no dominant core‐rim zoning but instead consist of 50–200 μm patches of Mg‐rich melilite (Åk32–62, median Åk51) set in or partially enclosed by, and optically continuous with, relatively Al‐rich melilite (Åk25–53, median Åk38). The Al‐rich regions have jagged, dendritic shapes but occur within crystals having straight grain boundaries. Another unusual feature of this inclusion is the size and spatial distribution of spinel. In many places, especially in the interior of the inclusion, the aluminous melilite encloses numerous, fine (0.5–5 μm) inclusions of spinel and minor perovskite and fassaite. The latter phases also occur as isolated grains throughout the inclusion. Coarse‐grained spinel, ∼50–150 μm across, occurs in clumps and chains enclosed in relatively Mg‐rich melilite, whereas none of the fine spinel grains are clumped together. The sample also contains a spinel‐free palisade body, 1.7 × 0.85 mm, that consists almost entirely of Åk‐rich (45–65 mol%) melilite. Within the palisade body are two grains of perovskite with extremely Nb‐rich (∼4–8 wt% Nb2O5) cores and rims of typical composition. All phases in this inclusion have chondrite‐normalized REE patterns that are consistent with crystal/melt partitioning superimposed upon a bulk modified Group II pattern. We suggest that TS2 had an anomalous cooling history and favor the following model for the formation of TS2. Precursors having a bulk modified Group II pattern melted. Rapid growth of large, dendritic, nonstoichiometric melilite crystals occurred. The melilite trapped pockets of melt and incorporated excess spinel components and TiO2. Bubbles formed in the residual melt. As crystallization slowed, coarse spinel grew. Some spinel grains collected against bubbles, forming spherical shells, and others formed clumps and chains. Relatively Åk‐rich melilite crystallized from the residual melt between dendritic melilite crystals and from melt trapped in pockets and between arms of dendrites, and incorporated the clumps and chains of coarse spinel. Bubbles broke and filled with late‐stage melt, their shapes preserved by their spinel shells. Slow cooling, or perhaps an episode of reheating, allowed the early melilite to become stoichiometric by exsolving fine grains of spinel, perovskite and fassaite, and allowed the melilite to form smooth grain boundaries. Dendritic crystals are indicative of rapid growth and the melilite crystals in TS2 appear to be dendritic. Coarse, dendritic melilite crystals have been grown from Type B inclusion melts cooled at ∼50–100 °C/h. If those results are applicable to Type A inclusions, we can make the first estimate of the cooling rate of a Type A inclusion, and it is outside the range (2–50 °C/h) generally inferred for Type B inclusions. The rapid cooling inferred here may be part of an anomalous thermal history for TS2, or it may be representative of part of a normal thermal history common to Types A and B that involved rapid cooling early (at high temperatures) as inferred for TS2, and slower cooling later (at lower temperatures), as inferred for Type B inclusions. We prefer the former explanation; otherwise, the unusual features of TS2 that are reported here would be common in Type A inclusions (which they are not).