I rarely find reason to argue with Chris Salter, but I think that his
conclusions here are too hasty. From the PIXE data supplied, the
composition of the slags cannot be adequately fitted to the
CaO-SiO2-TiO2 ternary equilibrium diagram, as 4 of the 5 have between 14
and 34 percent iron oxide (Jamie doesn't say whether this is calculated
as FeO or Fe2O3 - the former would be appropriate here). I have never
seen a slag like this and can't find an appropriate quaternary or
pseudo-ternary diagram (system CaO-FeO-TiO2-SiO2).
There is a puzzling discrepancy between the PIXE and the SEM data, as
the latter contain little or no iron! It guess that it would be
appropriate to squeeze the SEM data into the CaO-SiO2-TiO2 ternary - but
why the difference between the two sets of data?
I think it wholly inappropriate to guess which mineral phases are
present from the phase diagrams, and have been denouncing this approach
for years (much to the horror of Andreas Hauptmann!!). Have some
polished thin sections made. Sphene (CaTiO3) is a very easy mineral to
recognize in thin section if present as euhedral crystals rather than as
dendrites. If there is as much iron as implied by the PIXE data, then
you may also have a spinel somewhere along the continuum from magnetite
to ulvospinel (Fe2TiO4). (The more Ti present, the pinker the colour in
reflected light). Melilites are commonly seen in plain vanilla blast
furnace slags, and again are very easy to recognize in thin section, but
these need rather more alumina than is implied by the PIXE analyses.
Anyway, these materials are certainly unusual and definitely worthy of
fuller investigation - but with slags optical petrography, augmented by
SEM, is always the best place to start!
|