In situ high temperature EPR measurements of the growth of the signal of su
bstitutional Fe(III) ions have been used to study the diffusion of Fe in th
e rutile form of titanium dioxide. Two preparations, characteristic of the
two main processes employed for the production of titanium dioxide have bee
n studied. The first preparation, designated TiO2(SO4), was made from preci
pitated TiO2. It was calcined at ca. 850 degreesC and cooled slowly to room
temperature. The second preparation, designated TiO2(Cl), was from the gas
phase oxidation of TiCl4 at above 1200 degreesC. The resulting TiO2 was ra
pidly quenched to room temperature. The surfaces of both samples were impre
gnated with 0.030% Fe and the development of an EPR signal at g = 8.11, cha
racteristic of Fe(III) substituting for titanium ions in the rutile lattice
was monitored in situ at temperatures up to 730 degreesC by using a high t
emperature EPR cavity. For both TiO2(SO4) and TiO2(Cl) the g = 8.11 signal
showed a parabolic dependence of intensity with time typical of many diffus
ion processes. The temperature dependence of the slope of the intensity (I)
vs. time(0.5) plots allows estimates of the activation energies for the di
ffusion to be made. Values of 110 +/- 30 kJ mol(-1) for TiO2(SO4) and 50 +/
- 20 kJ mol(-1) for TiO2(Cl) are determined. The much lower value for the T
iO2(Cl) is attributed to the presence of metastable defects which, because
of the rapid cooling, persist in this rutile. This interpretation is suppor
ted by an observed increase in activation energies on heating the rapidly q
uenched TiO2(Cl) prior to the diffusion experiment. Pre-annealing at 700 de
greesC to reduce the concentration of defects, increased the activation ene
rgy for diffusion in TiO2(Cl) to 90 +/- 30 kJ mol(-1). The activation energ
y for diffusion of Fe is significantly lower than that for Cr (150 kJ mol(-
1)). Reasons for this are discussed.