Figure 1

(a) $M$ vs $H$ hysteresis loops measured at 300 K for the as-grown (solid line) and the UHV annealed (dotted line) ${\text{Co}}_{0.03}:{\text{TiO}}_2$ samples. (b) TRM curves taken from 5–365 K of as-deposited (solid line) and annealed (dashed line) films, illustrating the enhancement in magnetic properties with annealing. For TRM experiments, the sample is cooled to 5 K and saturated in a 1 T applied magnetic field; the field is then removed and the $M_{\text{TRM}}$, as a function of increasing temperature, is measured up to 365 K. TRM is useful as a direct measurement of the magnetic signal from the sample alone because the measurement is taken in zero applied field, and there is no contribution from the diamagnetic substrate. The increase observed below 50 K could point to a multiphase system, which could perhaps be related to the slight variations of Co distribution (always below 8%) observed in these samples (Ref. 19). Additionally, field cooled and zero-field cooled measurements of the same sample have revealed that the film exhibits no superparamagnetism; these results further support that there are no nanometer-scale metallic Co clusters present in the film, as these would contribute a peak in the zero-field cooled measurement. Note: Data adapted from Ref. 11. The moment per Co atom differs slightly from Ref. 11 because of a difference in Co concentration measurement (2% for Ref. 11, 3% here).Reuse & Permissions

Figure 2

(Color online) $Z$-contrast (high angle annular dark field) image of the annealed ${\text{Co}:\text{TiO}}_2$ film (right) on the LAO substrate (left). The image has been smoothed to remove noise (the upper inset shows the raw data). The lower inset shows a schematic of the anatase lattice with gray and red spheres representing Ti and O atoms, respectively, along the projection in the image.Reuse & Permissions

Figure 3

(Color online) (a) Magnified image of the anatase lattice after maximum entropy deconvolution (from the same area as the upper inset of Fig. 2). The HREM Research Inc. plug in for digital micrograph was used, using a physical probe with a full width at half maximum of 0.071 nm. Red circles mark O columns while gray circles correspond to Ti/O columns. (The probe was simulated using the estimated aberration parameters convoluted with a Gaussian to account for the source size. The results are not sensitive to the exact probe shape used for the deconvolution.) (b) Intensity traces along the atomic planes marked with red arrows on (a), compared to traces from the raw image, before deconvolution. The blue lines represent an average of over five lateral pixels along boxes taken on the ME processed image, such as the white box on (a). The red dots represent the intensity traces from the same regions and averaged in the same way on the raw data image. It is seen that the bright O column actually comprises two peaks: one (red arrow) in the expected position for O and the another (green arrow) displaced away from the O column. This peak is due to preferential segregation of Co interstitials.Reuse & Permissions

Figure 4

(Color online) A schematic of the interstitial atomic configuration for a single Co impurity. Ti atoms are shown in gray, O in red, and Co in blue. (a) and (b) correspond to views down the [100] and [010] projections, respectively. In (c) a projection along the [001] axis of the atomic plane where the Co is sitting is shown. The position of the O vacancy has been marked with a red dotted circle. The ${\text{Co-Ti}}^{+3}{\text{-V}}_{\text{O}}$ complex is highlighted with a yellow triangle. (d) Contour plots (in arbitrary units) for the difference between the majority- and minority-spin electronic densities in the plane of the ${\text{Co-Ti}}^{+3}{\text{-V}}_{\text{O}}$ defect complex [blue box in (c)]. Grey circles mark the position of Ti atoms while red circles mark the O positions, approximately. The blue areas correspond to the maximal negative value of this difference (majority-spin accumulation), which is located mainly at the Co atom. The two surrounding O atoms and the oxygen vacancy (dashed red circle) are also significantly spin polarized. The spin-polarization of the surrounding Ti atoms is much weaker and is opposite in sign (red color).Reuse & Permissions

Figure 5

(Color online) (a) PDOS for the majority (positive) and minority (negative) spin states on the interstitial Co atom in the ${\text{Co-Ti}}^{+3}{\text{-V}}_{\text{O}}$ defect complex (upper panel). The lower panel compares the PDOS for the Ti atom by the Co interstitial [green, Ti(1)] with the PDOS for a general Ti atom [blue, Ti(2)] in the anatase lattice. Only one spin component is shown for both Ti atoms (the spin polarization is insignificant). (b) Energy-loss near edge structure at the Ti $L_{2,3}$ edge around 455 eV (left) and the $\text{O}K$ edge around 530 eV (right) from inequivalent Ti/O atomic columns of an annealed ${\text{Co}:\text{TiO}}_2$ sample. The red spectrum was acquired by placing the electron beam on a Ti/O column adjacent to a Co interstitial, such as the one labeled on Fig. 3a. The spectrum in black comes from a Ti/O column adjacent to a conventional O column (with no Co nearby). A typical example of an averaged EELS Cobalt $L$ edge around 779 eV is shown in the inset (total acquisition time of 10 s).Reuse & Permissions

Figure 6

(Color online) $L_3$ x-ray absorption spectrum of Co in Co-doped anatase (from Ref. 7, in blue), and calculated $L_3$ spectra for Co in the interstitial ${\text{Co-Ti}}^{+3}{\text{-V}}_{\text{O}}$ complex (black line) and for substitutional Co (red).Reuse & Permissions