This trend does not match to the expected behaviour based on the
ionic radii of the dopants who are increasing in the order Al b Fe leading
to higher oxygen displacement and thus higher values expected for the
oxygen diffusion in the same manner. Iron in its Fe3+ form has a slightly
bigger ionic radius when it is in the same coordination as Al3+ and
should lead to higher diffusivities than the aluminium-doped sample.
As we do not know the oxidation state of the iron in the samples, it is
not trivial to explain the diffusion behaviour. That is probably the reason
for the high activation enthalpy in the iron-doped samples. It is reported
by Tolchard et al. [36] that the picture is not clear when introducing
transition metals into the crystal as the oxidation state of the dopant
is not known. The activation enthalpies are for tracer diffusion in the
order LMSO ≤ LSO b LASO b LFSO (Table 3) and confirm the predicted
behaviour for the iron-doped ATLS.