Relict grains in CAls, revisited

Abstract

Although the Type B CAI are clearly igneous rocks, they were probably not completely molten (1, 2), thus the possibility exists that preexisting materials can be recognized and characterized. Relict phases were proposed to explain high U and Th concentrations in both melilite and fassaite which would require unreasonably high partition coefficients (D} if due to crystal-liquid partitioning (3). More detailed study showed very rare perovskite grains and enigmatic Ti hot spots in melilite (4). Kuehner et al. (5) subsequently reported very high lithophile trace element contents, including actinides, for fassaite inclusions in melilite which they proposed as relict phases, but Simon et al. (6) show that the fassaite inclusions can be better explained as being the last drops of liquid crystallization. In any case, the original observations and interpretations of (3) still point to an actinide-rich relict host phase. To be able to say what levels of Ti, U, and Th in melilite can be explained by igneous partitioning, we have measured D(mel) for these elements in a synthetic CAI composition under controlled fD_2 conditions, extended down to nebular conditions by carrying out experiments in graphite crucibles in pure CO. Actinide partition coefficients are quite low: D(Th} = 0.008 and D(U) = 0.0007 (possibly a record low in measured D values). The D for trivalent U should be around 0.1-0.3 depending on Ak content as the ionic radius of U^(+3) is similar to La. Thus, for solar nebula fD_2's and CAI compositions less than 1 % of the U is trivalent. The measured D prove that igneous partitioning fails to explain the average U contents of type B CAI melilites, the difference being a factor of 600. D(Ti) is 0.018 at Ak23 and increases with Ak content. D(Ti) is relatively similar in air and at solar nebula fD_2s, surprising given the documented importance of trivalent Ti. In any case, the measured D(Ti) show that melilite Ti levels around 200-300 ppm in early crystallizing melilite can be explained by igneous partitioning, but higher levels would be indicative of resorbed Ti-rich relict phases (e.g., perovskite). To make a closer comparison ofU and Ti in CAI melilites, the fission track images of (3) have been quantitatively mapped at 20 microns resolution for two mm-sized rim and one mantle melilite. High resolution quantitative U distribution data on adjacent fassaites were also obtained. One rim grain shows several U-rich fassaites like those of(5), but the other melilites do not. There are broad U-rich regions in all grains which will be characterized in more detail. The mantle grain is especially rich in detail, but some of this may be correlated with secondary alteration. There is rough correlation ofU content and Ti in the rim grains, but the scale of the Ti analyses, based on electron probe points, is much smaller than that for U. If relict phases, e.g., perovskite, dominate the actinide distributions, they might also affect other lithophile trace elements, e.g., REE. References: (1) Wark D. (1983) thesis. (2) Stolper E. and Paque J. (1986) Geochim., Cosmochim. SO, 2159. (3) Murrell M. and Burnett D. (1987) Geochim. Cosmochim. SI, 985. (4) Johnson M., Burnett D. and Woolum D. (1988) Meteoritics 23, 276. (5) Kuehner S., Davis A. and Grossman L. (1989) Geophys. Res. Lett. 16, 775. (6) Simon S., Davis A. and Grossman L. (1990) LPSC 21, 1161.

Files

Additional details