We are very aware of the effect. We have documented the problem for Cr
in even an iron-poor host (pyrope, olivine) with chromite inclusions, and
the effect can be shown to be important for 50-100 um at 15 kV (Cr is more
strongly fluoresced by Fe than Ti).
To Horst, Eric, and others
>Anyone studying the solubility of TiO2 in garnet grains
>containing small inclusions of a Ti-rich phase, such as
>rutile or ilmenit exsolutions, needs to be aware of the
>danger of spurious effects in electron microprobe analysis
>of the host garnet. Apparent titanium contents of up to
>1 weight-% TiO2 can result from secondary fluorescence
>effects, in the vicinity of such inclusions, and the effect
>is significant to a distance of 25-30 microns from the
>nearest inclusion. The problem has been documented for
>garnet by Feenstra & Engi (1998, notably in Fig.5 and
>In essence, primary radiation generated by Fe and Mn in
>garnet causes the excitation of Ti-fluorescence radiation
>in nearby Ti-rich phases, even if no grain of the latter
>phase is hit by the electron beam. In samples that contain
>abundant such inclusions in garnet, it is virtually not
>possible to correct for this effect or indeed to avoid it.
>Ignoring this analytical artifact may produce unrealistically
>high Ti-contents for garnet (and other Fe-rich silicates).
>Feenstra A. & Engi M. (1998) An experimental study of Fe-Mn
>exchange between garnet and ilmenite. Contributions to
>Mineralogy and Petrology, vol. 131, p. 379-392.