Imaging of lateral spin valves with soft x-ray microscopy

We investigated Co/Cu lateral spin valves by means of high-resolution transmission soft x-ray microscopy with magnetic contrast that utilizes x-ray magnetic circular dichroism (XMCD). No magnetic XMCD contrast was observed at the $\text{Cu}{\text{L}}_3$ absorption edge, which should directly image the spin accumulation in Cu, although electrical transport measurements in a nonlocal geometry clearly detected the spin accumulation in Cu, which remained unchanged during illumination with circular polarized x rays at the Co and $\text{Cu}{\text{L}}_3$ absorption edges.

Figure 1

SEM image of a Co/Cu spin valve with 200 nm separation between ferromagnetic electrodes. The bright parts of the device are the Cu wires while the darker parts are the Co electrodes.

Figure 2

(Color online) Nonlocal electrical measurements of a Co/Cu lateral spin valve with a 200 nm separation. The change in the nonlocal resistance indicates a spin accumulation in Cu and is due to the parallel vs antiparallel orientation of the magnetization in the two ferromagnetic contacts. The measurements were performed (a) without x-ray illumination, and while illuminating at the (b) $\text{Co}{\text{L}}_3$, and (c) $\text{Cu}{L}_3$ absorption edges, respectively.

Figure 3

(a) TXM image of a Co/Cu lateral spin valve with a 500 nm separation taken at the $\text{Co}{\text{L}}_3$ absorption edge in remanence after saturation in a positive field. The ferromagnetic Co electrodes are well defined in contrast to the barely visible Cu wires. (b) XMCD image of the same lateral Co/Cu spin valve. This image was obtained by subtracting images taken in remanence after positive and negative saturation, respectively. The dark XMCD contrast of $\sim 5\%$ indicates that the magnetization switched in both Co electrodes.

Figure 4

(Color online) (a) Single TXM image of a Co/Cu lateral spin valve with 500 nm separation taken at the $\text{Cu}{\text{L}}_3$ absorption edge in remanence after saturation at positive field. There is a clear contrast for the Cu wires. (b) Sum of 400 images taken at the $\text{Cu}{\text{L}}_3$ absorption edge. The area of interest around the injection contact is magnified. (c) X-ray intensity profile of the summed images along the Cu wire as indicated by the rectangular box in (b). The additional attenuation of the Co electrodes is clearly visible.

Figure 5

(Color online) (a) Sum of 400 TXM images taken at the $\text{Cu}{\text{L}}_3$ absorption edge showing the location of the Co electrodes. The electrical current path is also indicated. (b) XMCD image of the same region obtained by subtracting summed images after positive and negative saturation. No clear magnetic contrast is observed. (c) Intensity profile of the XMCD signal integrated over the width of the Cu wire. (d) Intensity profile of the XMCD signal integrated over the same width as in (c) but outside the region of the Cu wire. Any intensity variations in (c) are comparable to the noise in (d).

Figure 6

(Color online) Signal-to-noise ratio for each picture pixel plotted as a function of number $N$ of summed images. $\sqrt{N}$ fit is shown with solid line.

Figure 7

(Color online) (a) Schematic band structure of a typical transition-metal ferromagnet. The XMCD contrast results from the spin-dependent density of states at the Fermi level and the concomitant different absorption as a function of photon polarization. (b) Schematic band structure for spin accumulation in Cu. The spin accumulation results in a spin-dependent splitting of the chemical potential with nearly identical density of states.