Spin pumping and inverse spin Hall effect in ultrathin ${\mathrm{SrRuO}}_3$ films around the percolation limit

We demonstrate spin pumping and inverse spin Hall effect (ISHE) in an all oxide heterostructure composed of two oxides with different respective crystal structures. ${\mathrm{Y}}_3{\mathrm{Fe}}_5{\mathrm{O}}_{12}$ (YIG) is used as a spin source, being a perfect candidate because of its low damping and insulating behavior. ${\mathrm{SrRuO}}_3$ (SRO), a perovskite oxide with metallic conductivity, serves as a spin sink. Spin pumping is shown by the increase of the damping in ferromagnetic resonance for the YIG in the presence of an SRO top layer and by the inverse spin Hall effect appearing in the SRO. Transmission electron microscopy shows that the growth of YIG is fully epitaxial, while the SRO grows in islands of several polycrystaline phases, a fact which strongly influences the measured ISHE. The magnetization of the YIG remains unaffected by the SRO layer while the SRO no longer shows the low temperature ferromagnetism typically observed in monocrystalline layers.

Figure 1

(a) TEM image with fast Fourier transforms (insets) of the marked areas of nominally 10 nm SRO and 20 nm YIG (sample F). YIG shows monocrystalline growth; SRO is deposited in polycrystalline and amorphous phases. In (b) the real thicknesses of sample F are determined. With a variance of 7–13 nm the SRO thickness is not homogenous. SRO grows in islands, visible in (c) (sample B). In sample B the islands are not connected with each other. This can be verified with a top view of the surface via scanning electron microscopy, shown in (e). The TEM image of a 15 nm SRO layer (sample H) is illustrated in (d).

Figure 2

RHEED images between depositions of sample F: (a) RHEED pattern after YIG deposition. (b) Change into an asymmetric pattern after SRO deposition. (c) Repetition of symmetric pattern of SRO every ${60}^{\circ }$.

Figure 3

Hysteresis loop as measured by SQUID magnetometry for a 25 nm thick bare YIG (sample Y) after deposition.

Figure 4

FMR measurmenents at 9.6 GHz. The linewidths: (a) sample Y with $(4.8\pm 1)\mathrm{Oe}$, (b) sample E with $(27.7\pm 1)\mathrm{Oe}$, and (c) sample C with $(13\pm 1)\mathrm{Oe}$. The lines are the Lorentzian fits for each curve.

Figure 5

Determination of damping constant of (a) sample Y (bare YIG) with in-plane and out-of-plane configuration, (b) sample C, (c) sample F, and (d) sample H. The lines are the linear fits to obtain the Gilbert damping.

Figure 6

Damping constant as a function of nominal SRO thickness. We have added a linear fit as a guide to the eye. The dependence should only be linear for small degrees of coverage and saturate when the layer starts to close completely.

Figure 7

ISH voltage of sample C. The change of polarity of magnetic field results in the change of polarity of ISH voltage. The measurement data have been fitted (lines).

Figure 8

Schematic illustration of different morphology layers of SRO. In (a) the the percolation threshold is not reached, in (b) the percolation limit is reached, and in (c) the islands cover almost the full layer.