Results and discussion

Figure 6.2: Major loop from the final layer stack of a typical 300$\times$300$\mu$m$^2$ TMR element.

Figure 6.3: Minor loop from the final layer stack of a typical 300$\times$300$\mu$m$^2$ TMR element.

0.5

Figure 6.2 presents the measurement of a typical 300$\times$300$\mu$m$^2$ TMR element. At an area resistance between 183 and 210M $\Omega\mu{\rm m}^2$, the measured TMR ratio is 13.1%. The in-plane magnetic field is applied parallel to the pinning of the bottom electrode. For further characterisation of the layer stack, I/V curves are measured and evaluated. Figure 6.4(a) shows a measured I/V curve for this element. For the evaluation, the plot was differentiated and fitted with standard software, see figure 6.4(b). Using BRINKMANS equations (confer page ) with an effective electron mass of $m_{\rm eff} = 0.4\cdot{\rm m_E}$, the barrier height has a reasonable value of $\nu =$ 2.62eV. The barrier thickness of $b =$ 1.95nm is a little bit thicker but still close to the expected 1.8nm. The asymmetry $d\nu = -0.22$eV lies also within the expected range.

Figure 6.4: I/V measurements of the used layer stack.

[Original I/V curve.] [Differentiated I/V plot that is used for the BRINKMAN fit.]

The used TMR layer stack was not developed to have the highest possible TMR ratio or to get a very robust tunnel barrier. The aim was instead a slow ascending slope around zero magnetic field. Therefore, the TMR ratio of 13.1% is not very high, but still quite good for this system and a large TMR element. An annealing of the whole sample cannot be done, because it would destroy the orthogonal pinning of the two electrodes. But an unannealed tunnel barrier is not as good as an annealed barrier. The TMR ratio is less for an unannealed TMR element and the barrier can easily be destroyed (an applied voltage of 1Volt mostly destroys the tunnel barrier). Also, the parameters given by the BRINKMAN-fit are not extremely good, but reasonable for the targeted application. In any case, the main goal, i.e. a slow ascending slope around zero field and no irreproducible switching, is achieved nicely (see minor loop in figure 6.3).