Influence of Thickness and Interface on the Low-Temperature Enhancement of the Spin Seebeck Effect in YIG Films

The temperature-dependent longitudinal spin Seebeck effect (LSSE) in heavy metal $(\mathrm{HM})/{\mathrm{Y}}_3{\mathrm{Fe}}_5{\mathrm{O}}_{12}$ (YIG) hybrid structures is investigated as a function of YIG film thickness, magnetic field strength, and different HM detection materials. The LSSE signal shows a large enhancement with reductions in temperature, leading to a pronounced peak at low temperatures. We find that the LSSE peak temperature strongly depends on the film thickness as well as on the magnetic field. Our result can be well explained in the framework of magnon-driven LSSE by taking into account the temperature-dependent effective propagation length of thermally excited magnons in the bulk of the material. We further demonstrate that the LSSE peak is significantly shifted by changing the interface coupling to an adjacent detection layer, revealing a more complex behavior beyond the currently discussed bulk effect. By direct microscopic imaging of the interface, we correlate the observed temperature dependence with the interface structure between the YIG and the adjacent metal layer. Our results highlight the role of interface effects on the temperature-dependent LSSE in HM/YIG system, suggesting that the temperature-dependent spin current transparency strikingly relies on the interface conditions.

Popular Summary

The spin Seebeck effect—in which a voltage is generated by a thermal gradient across a magnetic material—exhibits a temperature dependence in magnetically insulating ${\mathrm{Y}}_3{\mathrm{Fe}}_5{\mathrm{O}}_{12}$ (YIG) films capped with heavy metals. Here, we explain the origin of the spin Seebeck effect by considering a combination of bulk and interfacial properties. Our key discovery is that the widely accepted phonon-magnon coupling mechanism in the bulk YIG is not dominant and cannot explain the low-temperature enhancement of the spin Seebeck effect signal. Instead, we show that previously neglected effects at the interface between YIG and the capping layer play a key role.

We experimentally investigate YIG films with thicknesses ranging from 150 nm to $50\mu \mathrm{m}$. By combining full electrical characterization with transmission electron microscopy, we show that phonon-magnon interactions do not significantly modulate the temperature dependence of the spin Seebeck effect. We show that the peak temperature depends on the film thickness and that higher peak temperatures are associated with thinner films. Our results complete the picture of the spin Seebeck effect by showing that it is a combination of the bulk spin current and interfacial effects that govern the size of the measured signal. This finding allows one to, for instance, optimize the signal for a device operating at a given temperature.

We expect that our findings will pave the way to engineering the spin Seebeck effect for applications in which a large signal is necessary, such as energy harvesting.

Figure 1

Temperature-dependent LSSE coefficients (${\sigma }_{\mathrm{LSSE}}$) of (a) a YIG slab and (b) YIG thin films with a wide range of thicknesses. The open symbols in (a) represent the results recorded by field-sweep mode measurements. The inset of (a) shows the peak position of the ${\sigma }_{\mathrm{LSSE}}\text{−}T$ curves as a function of film thickness.

Figure 2

Thickness-dependent ${\sigma }_{\mathrm{LSSE}}$ measured at a fixed temperature of (a) 300 K, (b) 200 K, and (c) 120 K. The solid lines are the fitting results according to the atomistic spin model. (d) Temperature-dependent effective magnon propagation length $\xi$. The solid line is the fit to the power law of $\xi \sim T^{-n}$, where the fitted exponent $n\sim 1$.

Figure 3

(a) Magnetic field-dependent normalized LSSE signals of Pt/Cu/YIG hybrids at various temperatures from 30 K to room temperature. (b) Temperature-dependent field suppression ratio of Pt/Cu/YIG hybrids at 9 T. (c) Temperature-dependent ${\sigma }_{\mathrm{LSSE}}$ of Pt/Cu/YIG hybrids measured at a fixed heating power and varied magnetic fields from 0.2 to 9 T. (d) Temperature-dependent LSSE voltages of Pt/Cu/YIG hybrids with a fixed magnetic field of $\pm 0.2\mathrm{T}$ and different heating currents.

Figure 4

(a) Temperature-dependent ${\sigma }_{\mathrm{LSSE}}$ measured at Pt/YIG, Pd/YIG, W/YIG, Pt/Cu/YIG, and W/Cu/YIG hybrids. The thickness of YIG films is $50\mu \mathrm{m}$. (b) The peak positions of the ${\sigma }_{\mathrm{LSSE}}\text{−}T$ curves measured for YIG films with different capping layers.

Figure 5

Electron microscopy analysis of (a),(c) Pt/Cu/YIG and (b),(d) Pt/YIG samples. (a),(b) TEM images of the YIG (lower) and metallic layers with (inset) colored panels indicate the distributions of (red) Fe, (green) Cu, and (blue) Pt across the interface. Elemental maps are generated by EELS in scanning-TEM mode in similar areas to those indicated. Integrated, normalized signal intensities across the data sets suggest changes in the interfacial structure (see text for details).

Figure 6

(a) AFM image of a $50\text{−}\mu \mathrm{m}$-thick LPE YIG film over an area of $2.5\times 2.5\mu {\mathrm{m}}^2$ with a rms roughness of 0.34 nm. XRR curves of (b) Pt/YIG, (c) Pd/YIG, and (d) Pt/Cu/YIG hybrid structures are illustrated. The black open symbols are the experimental data, while the red lines are the fitting curves.

Figure 7

The calculated $\mathrm{\Delta }T$ across the Pt/Cu/YIG hybrid structure as a function of system temperature. Two curves are illustrated under heating currents of 7 and 9 mA, respectively.