Negligible Magnetism in Excellent Structural Quality ${\mathrm{Cr}}_x{\mathrm{Ti}}_{1-x}{\mathrm{O}}_2$ Anatase: Contrast with High-$T_{\mathrm{C}}$ Ferromagnetism in Structurally Defective ${\mathrm{Cr}}_x{\mathrm{Ti}}_{1-x}{\mathrm{O}}_2$

We reexamine the mechanism of ferromagnetism in doped ${\mathrm{TiO}}_2$ anatase, using epitaxial $\mathrm{Cr}:{\mathrm{TiO}}_2$ with excellent structural quality as a model system. In contrast to highly oriented but defective $\mathrm{Cr}:{\mathrm{TiO}}_2$ ($\sim 0.5{\mu }_B/\mathrm{Cr}$), these structurally superior single crystal films exhibit negligible ferromagnetism. Similar results were obtained for $\mathrm{Co}:{\mathrm{TiO}}_2$. We show for the first time that charge-compensating oxygen vacancies alone, as predicted by $F$-center mediated exchange, are not sufficient to activate ferromagnetism. Instead, the onset of ferromagnetism correlates with the presence of structural defects.

Figure 1

TEM images and EDX spectra for Cr-doped ${\mathrm{TiO}}_2$ grown at $\sim 0.015\AA /\sec$ and 550 °C (top), and $\sim 0.1\AA /\sec$ and 700 °C (bottom).

Figure 2

Anatase (004) XRD rocking curves for $\mathrm{Cr}:{\mathrm{TiO}}_2/{\mathrm{LaAlO}}_3(001)$, along with the $[110]$ azimuth RHEED patterns obtained after deposition. (a) 450 Å of slow-grown ${\mathrm{Cr}}_{0.07}{\mathrm{Ti}}_{0.93}{\mathrm{O}}_2$. (b) 560 Å of fast-grown ${\mathrm{Cr}}_{0.10}{\mathrm{Ti}}_{0.90}{\mathrm{O}}_2$.

Figure 3

(a) Cr $K$-shell XANES for 450 Å slow-grown ${\mathrm{Cr}}_{0.08}{\mathrm{Ti}}_{0.92}{\mathrm{O}}_2/{\mathrm{SrTiO}}_3(001)$ (solid line) and 400 Å fast-grown ${\mathrm{Cr}}_{0.13}{\mathrm{Ti}}_{0.87}{\mathrm{O}}_2/{\mathrm{LaAlO}}_3(001)$ (dotted line), along with epitaxial films of $\alpha \mathrm{\text{−}}{\mathrm{Cr}}_2{\mathrm{O}}_3$ (dashed line) and ${\mathrm{CrO}}_2$ (dot-dash line). (b) (i) Cr $K$-shell EXAFS for the slow-grown film, along with (ii) a model calculation for Ti lattice site occupancy; (iii) EXAFS for the fast-grown film.

Figure 4

(a) Total room-temperature $M_s$ vs Cr content for 450 Å slow-grown (open circles) and fast-grown (closed squares) ${\mathrm{Cr}}_x{\mathrm{Ti}}_{1-x}{\mathrm{O}}_2$. Cross is slow-grown undoped ${\mathrm{TiO}}_2$. Dashed line is the expected dependence based on $0.5{\mu }_B/\mathrm{Cr}$. (b) Typical room-temperature VSM hysteresis loops for $\sim 450\AA$ ${\mathrm{Cr}}_{0.07}{\mathrm{Ti}}_{0.93}{\mathrm{O}}_2$ at the two growth rates.