(revised 08/10/2004)

Comments and Questions

Johann Kjellman has posed a series of questions that pertain to the fractionation trends among the (Mn,Fe)(Nb,Ta) oxides in pegmatites. In general, these oxides exhibit increasing Mn/Fe and Ta/Nb with progressive fractionation. Johann asks others to comment on several factors – including solubility with temperature, effects of co-crystallizing phases, and the selectivity of the Nb and Ta oxide structures – that might influence this trend:

(1) What drives pegmatite melt fractionation towards the generally increasing Ta/Nb (and is anyone aware of exceptions)

(2) Is this trend caused by Nb/Ta-selectivity of the crystallizing Nb-Ta-oxides themselves, and/or are their compositions determined in part by the co-crystallization of other phases, and if so, which?

(3) Linnen and Keppler (1997) have shown that Ta/Nb solubility in melt is an important factor for separation of Nb and Ta but is this the only one? Most Nb-Ta oxides are dimorphous compounds, for example (Fe,Mn)(Nb,Ta)₂O₆ columbite-tantalite (orth.) vs. tapiolite (tetr.) and YTiNbO₆ euxenite vs aeschynite, most likely also the fergusonite- and samarskite-groups. Often (always?) Ta (as opposed to Nb) is stabilizing the low-T modifications (Stubican 1964, fergusonite; Aleksandrov 1963, aeschynite-euxenite). If this is true it would suggest that at any given temperature the crystallizing Nb-Ta mineral actually is driving the Ta/Nb-fractionation rather than just reflecting the Ta/Nb ratio of the surrounding melt.

Reply by D. London

In a preliminary response to Johann’s question, I’ll include here an abstract of our work on Mn/Fe fractionation. We had observed that Mn was partitioned into garnet over Fe, i.e. that (Mn/Fe)Gt > (Mn/Fe)melt, so that the Mn/Fe ratio in garnet (and coexisting melt) should actually decrease with progressive fractional crystallization if garnet alone controlled the Mn/Fe ratio. This is clearly not the case in general (the Mn/Fe ratio of garnet increases with fractionation as determined by zonal position in pegmatites), so other phases appear to regulate the Mn/Fe ratio of melt – namely, tourmaline and biotite (though Mn is more compatible in biotite than in ordinary "black" tourmaline).

The same factors may influence the compositions of the primary phosphates. For example, "black" tourmaline of the typical composition (schorl-olenite-foitite solid solution) is very abundant in the border zones of Li-pegmatites in the White Picacho district, Arizona. Li(Mn,Fe)-phosphates and columbite-tantalite minerals appear later in the paragenetic sequence. Both are very near the Mn end-members, lithiophilite (LiMnPO₄) and columbite (MnNb₂O₆)(see. London and Burt, 1982), and there are no other mafic phases present.

Reply by Robert Linnen, Department of Earth Sciences, University of Waterloo, Waterloo, Ontario, Canada). Bob has used experimental methods to generate most of the existing database on the solubilities of Nb-Ta-Sn oxides in pegmatite-forming melts.

I think that I can contribute to Johann Kjellman's questions. First, there are obviously a variety of factors that control mineral solubility. Tantalite is more soluble than columbite for all melt compositions that have been investigated experimentally. However, in peralkaline melts manganotantalite is only marginally more soluble than manganocolumbite. This suggests that Nb or Ta is not strongly preferred in the columbite structure, rather that the composition of the melt controls the Nb/Ta ratio of the columbite-tantalite mineral that crystallizes from it. By contrast, in peraluminous melts manganotantalite is significantly more soluble than manganocolumbite, thus fractional crystallization leads to a trend of decreasing Nb/Ta. Adding F and Li to the melt, as well as increasing temperature, all increase columbite-tantalite solubility. However, these parameters appear to increase manganocolumbite and manganotantalite solubilities equally, i.e. this suggests that the Li-F melt composition and temperature are important to columbite-tantalite crystallization, but have a minor to negligible affect on the Nb/Ta ratio of the crystal.

As far as other Nb-Ta minerals are concerned, that's an interesting question. To my knowledge wodginite and microlite typically have very low Nb/Ta ratios, which may indicate a site preference for Ta in these minerals, although this might also just be a case of these minerals being saturated only after extreme fractionation, and the melt having a low Nb/Ta. Another point that needs to be made is that experiments that have been conducted, including my own, are not necessarily on stable phases, i.e., these could be metastable solubilities. An example of this is the tantalite-cassiterite-wodginite reaction: SnO₂ + MnTa₂O₆ = MnSnTa₂O₈ All three of these minerals can occur in the natural same sample, but it is not clear whether one or more phase are metastable.

Lastly, the Fe/Mn question is more complicated. As David London pointed out garnet, tourmaline and mica all contribute to the Fe-Mn budget of the melt and composition of the subsequent pegmatite. One area that I can add to the discussion are some preliminary experimental data that indicate that the solubilities of ferrocolumbite and ferrotapiolite are roughly an order of magnitude greater that those of the manganese counterparts (a couple of values are given in the recent GAC short course on rare element geochemistry and mineral deposits (to be published this fall) as well as an abstract submitted by myself to the upcoming pegmatite session at GSA. This means that columbite-tantalite crystallization should lead to Fe enrichment, which of course is the opposite to what is typically observed in nature. This infers that other phase(s) or processes are controlling Fe-Mn trends.

Responses are welcome in relation to the fractionation trends observed in pegmatites, and to the causes of such fractionation.