Comparison of laser-induced and intrinsic tunnel magneto-Seebeck effect in $\mathrm{CoFeB}/{\mathrm{MgAl}}_2{\mathrm{O}}_4$ and CoFeB/MgO magnetic tunnel junctions

We present a comparison of the tunnel magneto-Seebeck effect for laser-induced and intrinsic heating. Therefore, ${\mathrm{Co}}_{40}{\mathrm{Fe}}_{40}{\mathrm{B}}_{20}/{\mathrm{MgAl}}_2{\mathrm{O}}_4$ and ${\mathrm{Co}}_{25}{\mathrm{Fe}}_{55}{\mathrm{B}}_{20}$/MgO magnetic tunnel junctions have been prepared. The TMS ratio of 3% in case of the MAO MTJ agrees well with ratios found for other barrier materials, while the TMS ratio of 23% of the MgO MTJ emphasizes the influence of the CoFe composition. We find results using the intrinsic method that differ in sign and magnitude in comparison to the results of the laser heating. The intrinsic contributions can alternatively be explained by the Brinkman model and the given junction properties. Especially, we are able to demonstrate that the symmetric contribution is solely influenced by the barrier asymmetry. Thus, we conclude that the symmetry analysis used for the intrinsic method is not suitable to unambiguously identify an intrinsic tunnel magneto-Seebeck effect.

Figure 1

(a) Typical TMS (measured with a laser power of 150 mW) and TMR minor loop of a junction with an area of 24 $\mu {\mathrm{m}}^2$ and a nominal MAO thickness of 1.8 nm. (b) TMS (150 mW) and TMR major loops of an CoFeB/MgO MTJ with a junction size of 2 $\mu {\mathrm{m}}^2$ and a nominal MgO thickness of 1.7 nm. (c) Antiparallel (light blue) and parallel (dark blue) Seebeck voltages for an MTJ with an MAO thickness of 1.8 nm increase linearly with the laser power, while the TMS ratio (red) is constant $(3.3\pm 0.2)\%$.

Figure 2

comsol simulation of the temperature profile across the tunnel barrier with an applied laser power of 150 mW (120 mW at the sample as deduced from calibration measurements) for different thermal conductivities of MAO. The layer position of 0 nm corresponds to the top of the stack.

Figure 3

dJ/dV curves (dark) with corresponding Brinkman fits (light) in the antiparallel (blue) and parallel (red) case. In addition, the resulting values for the barrier height $\phi$, the barrier thickness $d$, and the barrier asymmetry $\mathrm{\Delta }\phi$ are shown.

Figure 4

(a) and (b) Antisymmetric contribution of the V/I characteristics. The dashed black lines are linear fits and illustrate the nonlinearity of the experimental data. (c) and (d) Linear fits to the symmetric contributions that are performed within (e) and (f) the small, and then extended to (c) and (d) the whole range.

Figure 5

Original symmetric contribution of the MAO MTJ in the parallel case (dark red), corresponding Brinkman fit $\left(\mathrm{\Delta }\phi =-2.3\mathrm{eV}\right)$ (light red), simulated barrier asymmetry of $\mathrm{\Delta }\phi =+2.3\mathrm{eV}$ (light blue), and $\mathrm{\Delta }\phi =0\mathrm{eV}$ (dark blue). For the sake of clarity, only one in 10 data points of the original data is shown. The colored areas represent the typical error range of the Brinkman model of 10%.