I welcome the opportunity to reply to Dolejs's comments regarding the nature and the origin of fluoride melts found in mantle xenoliths (Klemme, 2004). Before discussing the comments in detail, I would like to comment on the fact that the composition of the precursor melt was indeed miscalculated, and its composition should read as follows: (wt%) SiO2, 41; Al2O3, 10.3; TiO2, 0.09; FeO, 6.85; MgO, 15.3; CaO, 6.4; Na2O, 2.4; P2O5, 0.2; F, 12.0. I would like to thank Dolejs for pointing out this error. However, it should be noted that this does not in any way affect the conclusions of my paper.
Dolejs makes the suggestion that the fluoride glasses are magmatic fluorite crystals rather than quenched fluoride melt. To distinguish between fluoride glasses and fluorite crystals optical microscopy is, of course, not helpful as both glasses and fluorite crystals are isotropic. There are several arguments that the glasses are indeed quenched melts, not fluorite crystals. When using both reflected light microscopy and high resolution scanning electron microscopy, the glasses are clear, with no indication of quench crystals present. Moreover, as Figure 1 of my paper shows, textural evidence for liquid fluoride melt exists. It is hard to imagine how fluorite crystals could have formed on 20-μm thin olivine-olivine grain boundaries. Furthermore, the chemical composition of the fluoride glasses clearly shows that these materials are not fluorite crystals, as magmatic fluorites are known to be stoichiometric and contain no significant amounts of other elements except rare earth elements and Y (e.g., Marshall et al., 1998; Veksler et al., 2005), whereas evidence from petrographical and experimental studies indicates that fluoride melts contain significant Si and oxygen when in equilibrium with immiscible silicate liquids (e.g., Veksler et al., 2005).
Dolejs also suggests that liquid immiscibility between fluoride melts and silicate melts does not exist. Stating this he ignores evidence from a number of experimental studies that investigated liquid immiscibility between fluoride and silicate melts (Gramenitskiy and Shekina, 1994; Veksler et al., 1998; Webster et al., 1998; Veksler et al., 2005). Moreover, in an experimental study, Veksler et al. (2005) report high Ca partition coefficients (DCa = 20–80) between immiscible fluoride and silicate melts, which is in excellent agreement with the naturally occurring immiscible fluoride and silicate glasses found in the Kakanui xenoliths. Furthermore, experimentally determined trace element partition coefficients between immiscible fluoride melts and silicate melts (Veksler et al., 2005) also agree with chemical compositions observed in the Kakanui rocks.
To address the low totals in the fluoride glasses, these materials were reanalyzed using an electron microprobe. The fluoride glass compositions are virtually identical all across the sample. However, in an unusual matrix such as fluoride, oxygen analyses using an electron microprobe turned out to be somewhat problematic. Hence, upon replicate analysis, strong variations in the measured oxygen content of the fluoride glasses was observed which is believed to be mainly due to analytical difficulties. Based on these measurements I suggest the following representative major element composition for the fluoride glasses in the Kakanui xenoliths, standard deviations from numerous analyses are given in parentheses as last significant numbers: Si, 2.4(3); Al, 0.2(1); Fe, 0.1(1); Mg, 1.5(1); Ca, 40.0(5); Na, 0.4(2); K, 0.1(1); P, 0.1(1); O, 13(7); F, 41(3).
Dolejs claims that salt melts cannot be quenched to glass. While it seems to be true that pure CaF2 melts do not vitrify upon quenching, it is not accurate to claim that salt melts generally do not form glasses. A new emerging field of glass research with a large body of literature indicates that only small additions of network modifiers support the formation of fluoride glasses (e.g., Grande et al., 1995; Gan, 1995; and many others).
As mentioned above, there is textural and chemical evidence that the fluoride glasses are indeed quenched fluoride melts. Furthermore, chemical evidence also suggests that the fluoride glasses cannot be magmatic fluorite crystals. Spectroscopic methods may help to shed some further light on these matters, but this is clearly beyond the scope of this note.
Dolejs claims to have calculated fluorite stability in subduction zone environments. He does not, however, document in any detail how these calculations were done. Thermodynamic calculations in appropriate complex compositions would be helpful to address the stability of fluorite in subduction zones; details of Dolejs's calculations would be useful.
Concluding, there is strong textural and chemical evidence for fluoride melts in the Kakanui mantle xenoliths. In his comments Dolejs ignores a large body of literature on fluoride glasses and experimental evidence on major and trace element partitioning between immiscible fluoride and silicate melts, the latter of which further supports an origin of the fluoride melts in the Kakanui rocks by liquid immiscibility.