NiO/Ni wires have been investigated as a function of their width in order to investigate the size dependence of exchange bias. The samples have been prepared by e-beam lithography and ion milling of ion beam sputtered thin films. For NiO/Ni wires narrower than 3 m, the exchange bias field significantly depends on the wire width. A NiO/Ni film shows an exchange bias field of Ϫ78 Oe whereas the exchange bias field of wires narrower than 200 nm is reduced to approximately Ϫ40 Oe. The coercive field of the NiO/Ni film is 28 Oe and increases to 210 Oe for the narrowest wires. The decrease of the exchange bias field for the narrowest wires is consistent with a recent microscopic model of exchange bias where the appearance of a unidirectional anisotropy in ferromagnet/antiferromagnet bilayers has been attributed to the presence of antiferromagnetic domains in the bulk of the antiferromagnet. A possible onset of a transition from a multidomain to a single-domain state of the antiferromagnet as a function of the NiO/Ni wire width seems to be the origin for the observed decrease of the exchange bias field for narrow wires.The exchange bias ͑EB͒ effect occurs due to the exchange coupling at antiferromagnet͑AFM͒/ferromagnet͑FM͒ interfaces leading to a shift of the magnetic hysteresis loop along the field axis.1 This shift of the hysteresis loop can be established either by cooling the AFM/FM bilayers in a magnetic saturation field below the Néel temperature T N of the AFM or by depositing the bilayers in an external magnetic field.2 The exchange bias effect has been used over several decades and more recently for pinning the magnetization of one of the two electrodes in magnetoresistive devices ͑giant magnetoresistance multilayers 3 and tunnel junctions 4,5͒. Although the exchange bias effect has already been intensively exploited in micron-and submicron-sized magnetoelectronic devices, its microscopic origin is not yet fully understood.A recent experiment on Co/CoO bilayers in conjunction with a Monte Carlo simulation study has shown that the dilution of the antiferromagnet CoO with nonmagnetic impurities ͑e.g., Co 1Ϫx Mg x O͒ or defects ͑e.g., Co 1Ϫy O͒ in its volume part leads to the formation of antiferromagnetic volume domains. 6 The formation of antiferromagnetic domain walls leads to a small surplus magnetization at the AFM/FM interface which couples to the FM and results in a unidirectional anisotropy. Hence, the antiferromagnetic domains are the microscopic origin of exchange bias. This result is complementary to a previous approach attributing exchange bias to the formation of antiferromagnetic domains with domain walls ͑DW͒ perpendicular to the AFM/FM interface in the presence of only interface roughness.7 From these models one has to conclude that the exchange bias of AFM/FM bilayers vanishes when the AFM becomes single domain. From a systematic investigation of a possible finite size effect of exchange bias the microscopic role of the DW formation can be elucidated as well as the lower limit of the extension of...

Highly a-axis-textured CrO 2 films have been deposited on Al 2 O 3 (0001) and on isostructural TiO 2 (100) substrates by a chemical vapour deposition technique. For Al 2 O 3 substrates a columnar growth of CrO 2 (010) on an initial Cr 2 O 3 (0001) layer has been found in transmission electron microscopy as well as in x-ray diffraction investigations. The sixfold in-plane symmetry of a (0001)-oriented Cr 2 O 3 initial layer leads to three equivalent in-plane orientations of the CrO 2 unit cell as confirmed by electron diffraction and scanning electron microscopy. The growth can be understood by a simple model of the in-plane symmetries of the Al 2 O 3 (0001), Cr 2 O 3 (0001), and CrO 2 (010) lattices. The growth on TiO 2 (100) substrates leads to (100)-oriented CrO 2 films of higher crystalline quality than the ones grown on Al 2 O 3 (0001). Transmission electron microscope images show growth of CrO 2 (100) directly on the TiO 2 (100) substrates and no significant Cr 2 O 3 inclusions within the CrO 2 (100) layer. All contributions to the magnetoresistance (MR) due to anisotropic MR, Lorentz MR, spin disorder, and intergrain tunnelling MR have been determined and partly correlated with the crystalline properties of the samples investigated. For films of both types the intrinsic linear contribution to the high-field MR does not depend on the crystalline quality of the films and supports the suggested intrinsic doubleexchange mechanism for CrO 2 .

Exchange bias effects have been induced along the perpendicular-to-film direction in nanostructures prepared by electron beam lithography, consisting of a ferromagnetic [Pt/Co] multilayer exchange coupled to an antiferromagnet (FeMn). As a general trend, the exchange bias field and the blocking temperature decrease, whereas the coercivity increases, as the size of the nanostructures is reduced.

We report on tunnelling magnetoresistance (TMR), current-voltage (IV) characteristics and lowfrequency noise in epitaxially grown Fe(110)/MgO(111)/Fe(110) magnetic tunnel junctions (MTJs) with dimensions from 2x2 to 20x20 µm 2 . The evaluated MgO energy barrier (0.50±0.08 eV) , the barrier width (13.1±0.5 Å) as well as the resistance times area product (7±1 MΩµm 2 )show relatively small variation, confirming a high quality epitaxy and uniformity of all MTJs studied. The noise power, though exhibiting large variation, was observed to be roughly anticorrelated with the TMR. Surprisingly, for the largest junctions we observed a strong enhancement of the normalized low-frequency noise in the antiparallel magnetic configuration.This behaviour could be related to an interplay between the magnetic state and the local barrier defects structure of the epitaxial MTJs.

No abstract

We report a combined experimental-computational investigation of the electronic structure of CrO 2 . We have measured the magneto-optical Kerr spectra of CrO 2 at 10 K and 300 K. At 10 K the Kerr signal is significantly enhanced over that obtained at 300 K. We compare the measured Kerr spectra to first-principles theoretical spectra, which we computed using three different approximations to the exchange-correlation functional, i.e., the local spin-density approximation ͑LSDA͒, generalized gradient approximation ͑GGA͒, and LSDAϩU. The experimental low-temperature magneto-optical Kerr spectra are best explained by calculations employing the GGA functional. The addition of an on-site Coulomb correlation U does not lead to reasonable Kerr spectra.

The growth of (010)-oriented CrO2 thin films on Al2O3(0001) substrates leads to a higher grain boundary density than the growth of (100)-oriented CrO2 thin films on isostructural TiO2(100) substrates. For both types of films an intrinsic linear contribution to the high field magnetoresistance (MR) due to spin disorder has been determined at T=300 K. This contribution does not depend on the crystalline quality of the films and supports the suggested intrinsic double exchange mechanism for CrO2. At low temperature (T=10 K) intergrain tunneling MR and Lorentz MR appear, which strongly depend on the crystalline properties of the CrO2 films.

Fully epitaxial Fe(110)/MgO(111)/Fe(110) magnetic tunnel junctions (MTJs) have been tested with respect to symmetry-enforced spin filtering. The Fe(110) electrodes exhibit Σ1↑ and Σ1↓ spin states, both crossing the Fermi level, but with a group velocity about 50% smaller for the minority states compared to the majority ones. These epitaxial but symmetry-mismatched MTJs yield tunneling magnetoresistance (TMR) values of 54% at 1.5 K and 28% at room temperature. The TMR value and the estimated tunneling spin polarization are consistent with a partial spin filtering due to the Σ1↑ states partially compensated by the Σ1↓ states.