Conclusion

Mineral inclusions in the sapphires from the new mining area in southern Malagasy (Adranondambo) include carbonates (calcite), zirconium bearing minerals (baddeleyite, zircon?, zirconolite), Ti bearing minerals (zirconolite, rutile, F-phlogopite), calcalcali bearing minerals (K-feldspar, plagioclase, scapolite, anorthite), thorium and uranium phases (thorianite and zirconolite), hydrous minerals (Mg-hornblende, diaspore), phosphorous bearing minerals (apatite) and various types of spinels (system Mg-Al-Fe-Zn-O). Regarding the various rock types of the mining area, these minerals are typically found in various rocks belonging to the so-called U-Th skarns. Therefore, these skarns are considered as the host rocks for the sapphire formation.

The Th-U skarns were produced by circulating fluids and mineral reactions within calcsilicate marbles. Regarding the different types of inclusions found in the sapphires and comparing them with the different generations of skarns (Rakotondrazafy et al., 1996), the formation of sapphire can be more clearly defined. Mineral inclusions in sapphire can be related to at least two metasomatic events (e.g. first event includes typically spinel, second event includes typically calcite, F-phlogopite and anorthite). Sapphire therefore most probably formed during the phase of the metasomatic event I and/or II (for more details cf. Rakotondrazafy et al., 1996).

The published data on fluid inclusion analyses supports this idea as different generations of CO2-rich fluids are present (see also Bruder, 1996). The variable formation conditions for the different sapphire generations may be deduced from the data published by Bruder (1996) and Rakontondrazafy et al. (1996), ranging from approx. 850º C and 5 kbar (stage I) to lower temperatures and pressures of approx. 800º C and 3.5 kbar (stage II).

The formation of corundum during phase I is demonstrated by Rakontondrazafy et al. (1996), presenting thin section analyses. However, it is the later metasomatic event which is characterized by fluids rich in fluorine, titanium, and zirconium. Mineral inclusions composed of these elements are present in the sapphires (e.g. zirconolite and fluorite). Such sapphires are therefore formed in connection with these fluids.

Considering the mobility of Ti, a necessary trace element for the blue coloration of sapphire, the presence of water (cf. evidenced by diaspore as a daughter mineral in fluid inclusions, Bruder, 1996) and fluorine in the metasomatic fluids maybe of special interest (Peretti et al., 1996). Concentrations of water and fluorine in the fluids will, under certain formation conditions, support the mobility of titanium.

Thus, from such fluids we expect the formation of sapphire rather than colorless corundum. It is proposed that particularly the second generation of sapphires is formed from such fluids. The special formation conditions during this metasomatic event II supports the formation of baddeleyite rather than of zircon. This is in concurrence with our observation of the different type of mineral inclusions in these sapphires.

Early sapphire (phase I) does not seem to have been formed together with hibonite, which was formed during the later phase of the metasomatic event II from fluids rich in Ti, F, CO2, and REE-elements. It has been demonstrated by Rakotondrazafy et al. (1996) that hibonite is formed from corundum by mineral reactions. But such fluids also needed the high concentrations of REE-elements. It has to be further investigated whether or not the sapphires (phase II) were formed from the F- and Ti-rich fluids depleted in REE-elements.

In conclusion, we therefore have no indications for a formation of sapphires within a pegmatite-vein or a pegmatite-like vein (or generally speaking from a melt) as proposed by Kiefert et al. (1996). It follows that no similarity can be proposed between the deposit of the Madagascan (Adranondambo) sapphire and the sapphire occurrence in Kashmir (Peretti et al., 1991) as it is attempted by those authors. Metamorphic granulite facies metamorphism alone is also not the model to be accepted for the formation of the Madagascan (Adranondambo) sapphires.

In conjunction with the formation of sapphires, a mobilization of elements, such as Th, U, K, Ca, Na, Mg, Al, Si, P, Fe, Mg, Zn, Zr as well as Cl, F and Ti must have occurred. Consequently, in coincidence with Schwarz et al. (1996), sapphire formation seems to be best explained by metasomatic processes which are observed in the U-Th skarns within granulitic rocks (Rakotondrazafy et al., 1996).

While most of the sapphire formation seems to fit best in this model, it has to be examined in more detail whether some other formation process for sapphires may also have been active in some other particular zones in the mines. Some indications to that extent were obtained by the rarely occurring mineral inclusion of zircon (Kiefert et al., 1996) and quartz (this present work), which may have formed in later veins not equilibrated with the skarn rocks. The presence of other sapphire-bearing rock types, such as pegmatites in the mining area of Adranondambo (Kiefert et al., 1996) will have to be further investigated by thin section analyses.