Reversing the Training Effect in Exchange Biased $\mathrm{CoO}/\mathrm{Co}$ Bilayers

We performed a detailed study of the training effect in exchange biased $\mathrm{CoO}/\mathrm{Co}$ bilayers. High-resolution measurements of the anisotropic magnetoresistance (AMR) display an asymmetry in the first magnetization reversal process and training in the subsequent reversal processes. Surprisingly, the AMR measurements as well as magnetization measurements reveal that it is possible to partially reinduce the untrained state by performing a hysteresis measurement with an in-plane external field perpendicular to the cooling field. Indeed, the next hysteresis loop obtained in a field parallel to the cooling field resembles the initial asymmetric hysteresis loop, but with a reduced amount of spin rotation occurring at the first coercive field. This implies that the antiferromagnetic domains, which are created during the first reversal after cooling, can be partially erased.

Figure 1

SQUID magnetization measurements of a $\mathrm{CoO}/\mathrm{Co}$ bilayer at 10 K after cooling in a field of $+100\mathrm{mT}$. The upper panel (a) shows the first and second hysteresis loops with the magnetic field applied in the direction of the cooling field. Panel (b) represents the subsequent two hysteresis loops when the magnetic field is applied perpendicular to the cooling field. The lower panel (c) shows the next two hysteresis loops with the magnetic field again applied along the cooling field direction. A reentry of the untrained state without heating the sample above the blocking temperature is observed.

Figure 2

Hysteresis loops measured at 5 K with VSM magnetometry of a $\mathrm{CoO}/\mathrm{Co}$ bilayer cooled in a field of $+400\mathrm{mT}$. The first reversal at negative field is dominated by domain wall nucleation and domain wall propagation and is abrupt. All subsequent reversals are dominated by rotation of the magnetization and are more rounded.

Figure 3

Field dependence of the magnetoresistance of a $\mathrm{CoO}/\mathrm{Co}$ stripe at 10 K after cooling in a field of $+100\mathrm{mT}$ applied along the length of the stripe. A smaller resistance change (less rotation) is observed during the first reversal when compared to the subsequent reversals. The insets compare the resistance at saturation to the maximum resistance (reference line), which ideally corresponds to the case that all spins are oriented along the cooling field direction.

Figure 4

Field dependence of the magnetoresistance of the $\mathrm{CoO}/\mathrm{Co}$ stripe at 10 K after cooling in a field of $+100\mathrm{mT}$ along the stripe. The blue line illustrates the reappearance of the training effect without any heating of the sample. This reappearance is achieved by going through a hysteresis loop with the magnetic field in the sample plane but perpendicular to the cooling field direction (not shown). The insets show the resistance at saturation when compared to the maximum resistance (reference line).