Coalescence-driven magnetic order of the uncompensated antiferromagnetic Co doped ZnO

The evolution of the structural and magnetic properties of Co doped ZnO has been investigated over an unprecedented concentration range above the coalescence limit. ZnO films with Co concentrations from 20% to 60% of the cationic lattice have been grown by reactive magnetron sputtering. The wurtzite crystal structure was maintained even for these high dopant concentrations. By measuring the x-ray absorption at the near edge and the linear and circular dichroism of the films at the Zn and Co $K$ edge, it could be shown that Co substitutes predominantly for Zn in the lattice. No indications of metallic Co have been found in the samples. At low Co concentrations, the films are paramagnetic, but with increasing Co content, the films become antiferromagnetically ordered with increasing order temperature. Uncompensated spins, coupled to the antiferromagnetic dopant configurations, lead to a vertical exchange-bias-like effect, which increases with increasing Co concentration. In parallel, the single-ion anisotropy is gradually lost.

Figure 1

Global structural analysis shows the wurtzite crystal structure of all films. (a) XRD $2\theta -\omega$ scans. (b), (c) TEM image of the 35% Co:ZnO sample. (d) The corresponding electron-diffraction pattern.

Figure 2

(a) For all samples, the normalized isotropic XANES (left scale) at the Zn $K$ edge and the corresponding XLD for the Co:ZnO samples (right scale) are shown. (b) The normalized XANES spectra for linear polarized light at the Co $K$ edge, once perpendicular to the $c$ axis of the sample and once parallel, is shown exemplarily for 50% Co:ZnO. The direct difference of these—the XLD—is plotted again for all samples.

Figure 3

(a) The normalized and isotropic XANES at the Co $K$ edge and (b) the corresponding XMCD are shown for all Co:ZnO samples. (c) At 7.71 keV, the XMCD has been measured in dependence of an externally applied magnetic field at 2.5 K.

Figure 4

(a) SQUID M(H) measurements at 2 and 300 K, normalized to the magnetization measured at 5 T and 2 K. (b) Enlargement of the low-field region. (c), (d) The vertical exchange-bias shift for the (c) 35% and (d) 60% Co:ZnO samples. For these, the M(H) curves have been measured after being cooled in $+5$ T, $-5$ T and without applied field.

Figure 5

IP vs OOP SQUID measurements at 2 K. The arrows mark the size of the anisotropy at 5 T.

Figure 6

SQUID M(T) measurements following the FC, FH, and ZFC procedures. A separation of the curves suggests magnetic order. The arrows indicate the temperature where the transition takes place.

Figure 7

The abundance of singles, pairs, and triples are displayed (left scale), as well as the effective magnetic moment per Co (right scale). These have been evaluated from the SQUID measurements and the XMCD measurements with different applied fields. In addition, the approximation of the effective moment from the contribution of singles, pairs, and triples is shown (dashed line).

Figure 8

Summary of results for Co:ZnO. (a) The effective magnetic moment per Co from SQUID at 4 T and XMCD at 17 T. (b) The vertical exchange-bias shift increases with Co percentage, whereas (c) the anisotropy of the samples decreases with increasing Co percentage. The calculated maximum single-ion anisotropy and isotropy are marked. The comparison in (d) shows the theoretical results of Ref. [26] and our experimental results for the order temperature of Co:ZnO.