Printing Nearly-Discrete Magnetic Patterns using Chemical Disorder Induced Ferromagnetism

We show that sub-50 nm ion-induced lateral patterning of magnetic structures can be enabled by disorder induced ferromagnetism.[1] Disorder is induced through exposure of the chemically ordered alloy to energetic ions; collision cascades formed by the ions knock atoms from their ordered sites and the concomitant vacancies are filled randomly via thermal diffusion of atoms at room temperature. Here we consider the case of Fe60Al40 wherein the chemically ordered B2 structure is paramagnetic, and chemical disordering leads to the formation of the A2 structure which is ferromagnetic.[2] This ion-induced transition can be exploited to induce ferromagnetism in localized regions by ion-irradiation through lithographed shadow masks. We show that this technique may be useful for fabricating novel spin-transport devices.
First we demonstrate the disorder-induced increase in saturation magnetization, Ms, in continuous films. Fe60Al40 films of 40 nm thickness were deposited on SiO2(150 nm)/Si(001) substrates by magnetron sputtering. The films were annealed at 773 K in vacuum to form the chemically ordered B2 phase structure. Hysteresis loops were measured using a vibrating sample magnetometer (VSM). Figure 1a shows that the B2 film is weakly ferromagnetic with a saturation magnetization of Ms = 20 kA m-1.
The chemically ordered films were irradiated with 6 x 1014 ions cm-2 of Ne+-ions at 10 and 30 keV respectively. After irradiation, the Ms increases to 480 and 780 kA m-1, respectively for the 10 and 30 keV samples. The Ms of the 30 keV Ne+ sample is a factor of 40 larger than that of the annealed sample. Figure 1b shows X-ray Diffraction measurements around the 100 reflection for the sample after annealing, and after subsequent 10 keV Ne+-irradiation. The 100 reflection is allowed for the B2 superstructure but vanishes for the disordered A2 phase and this transition is clearly observed since the reflection initially present in the B2 film vanishes for the 10 keV sample.
Magnetic patterning was performed on a 10 μm wide and 400 μm long wire of chemically ordered Fe60Al40 (Figure 2a). The wire was covered with a 150 nm thick resist layer, and patterned using e-beam lithography. Simulations based on the binary collision approximation (TRIM) showed that the 150 nm thick resist layer is sufficient to block impinging 10 keV Ne+-ions.[3]
Lithography was used to carve out stripe like openings of 0.5 and 2 μm widths respectively. As shown in the micrograph in Figure 2b, the stripe-openings were separated by ~ 40 nm wide (and 150 nm high) resist walls, thereby stopping the impinging Ne+ ions reaching the areas directly underneath the resist. These areas can be expected to retain chemical order after exposure to Ne+, however, only if the lateral scattering of ions is restricted. The wire covered by the shadow-mask was exposed to 10 keV Ne+-ions at a fluence of 6 x 1014 ion cm-2. Figure 2c shows the magnetic contrast image, obtained using Kerr Microscopy on the sample prior to application of saturating magnetic fields. Striped magnetic regions are clearly observed possessing random magnetic orientations.
Magnetic contrast was captured whilst sweeping the field to obtain hysteresis loops shown in Figure 2d, on a set of 32 (0.5 μm/spacer/2 μm/spacer) stripe-pairs. Magnetization reversal occurs via a two-staged process; the first reversal step occurs at ≈ ±3 mT and the second step at ≈ ±7 mT. The magneto-optic intensity changes by 80% in the first reversal step, indicating that the 2 μm wide stripes reverse collectively at the smaller field. In the stripe geometry, the internal demagnetizing field increases with the stripe width. Reversal of the 2 μm wide stripes therefore occurs at smaller externally applied field as compared with the 0.5 μm stripes, resulting in selective reversal.
Magnetic contrast images were captured at remnant points of the above hysteresis loops and are shown in Figure 2e – h. Stripes with magnetization pointing towards the left or right appear as dark and bright contrast, respectively. Images captured after applying the saturating field of +/-18 mT followed by reduction to zero field show that the parallel magnetization configuration is preserved in remanence (Figure 2e and f). The antiparallel configuration is obtained after applying a reverse field of -/+5 mT to the saturated stripes and returning to zero field. Figure 2g and h show alternating light and dark contrast of the antiparallel state in remnant fields.
Spin-resolved Photo-Emission Electron Microscopy (SPEEM) measurements were performed to observe magnetic contrast with high spatial resolution of ~ 50 nm. The SPEEM micrographs are shown in Figure 2e and g as magnified equivalent regions of the Kerr-images. The color scale varies from red to blue for magnetic moments pointing left or right respectively. Straight, low-contrast regions are found to separate the 500 nm and 2 μm wide high-contrast regions that correspond to the magnetic stripes.
The straight low-contrast regions are unresponsive to the magnetic field and occur for both parallel and anti-parallel states. Since the straight low-contrast lines also exactly follow the pattern of the shadow mask, it can be concluded that these lines correspond to the 40 nm wide un-irradiated regions. Neighboring 500 nm and 2 μm magnetic stripes are therefore isolated by a continuous weakly magnetic spacer of 40 nm nominal width.
Thus, parallel and antiparallel magnetization configurations can be programmed using the magnetic field history and are non-volatile i.e., stable when the field is switched off. Selective reversal and the existence of binary magnetic states namely is a prerequisite for spin-valves and advantageous in devices for storing data bits. In particular with respect to spin-transport devices, it is also necessary to ensure that the magnetic regions are separated by narrow spacers of zero or low magnetization. Showing that discrete magnetic nanostructures can be prepared by ion-irradiation has important consequences not previously considered in literature, such as the possibility of laterally patterned spin-transport devices – our results are a step in this direction.

[1] R. Bali et al., Nano Letters 2013 (accepted). [2] J. Fassbender, et al., Phys. Rev. B 2008, 77, 174430. [3] J. F. Ziegler et al., Nucl. Instrum. Methods B 2010, 268, 11-12, 1818.