Exchange Bias Training Effect in Coupled All Ferromagnetic Bilayer Structures

Exchange coupled bilayers of soft and hard ferromagnetic thin films show remarkable analogies to conventional antiferromagnetic/ferromagnetic exchange bias heterostructures. Not only do all these ferromagnetic bilayers exhibit a tunable exchange bias effect, they also show a distinct training behavior upon cycling the soft layer through consecutive hysteresis loops. In contrast with conventional exchange bias systems, such all ferromagnetic bilayer structures allow the observation of training induced changes in the bias-setting hardmagnetic layer by means of simple magnetometry. Our experiments show unambiguously that the exchange bias training effect is driven by deviations from equilibrium in the pinning layer. A comparison of our experimental data with predictions from a theory based upon triggered relaxation phenomena shows excellent agreement.

Figure 1

Overall magnetic hysteresis $m$ vs ${\mu }_{0}H$ (left frame, dashed line). Thick solid lines are low field minor loops after positive and negative saturation, respectively. The horizontal line visualizes $m_r$ for the upper SL loop, the vertical line indicates the shift of the SL loop along the field axis relative to $H=0$. The inset is a schematic of the sample . The right frame shows the TE between the first (squares) and 20th (circles) hysteresis loop of the CoCr soft layer after first setting the CoPtCrB magnetization state.

Figure 2

Training effect ${\mu }_0{H}_B$ vs $n$ (left frame) of the HL-SL bilayer for set fields ${\mu }_0{H}_{\mathrm{set}}=\pm 0.36$ and $\pm 0.34\mathrm{T}$ after saturation in ${\mu }_{0}H=0.8\mathrm{T}$ for negative and $-0.8\mathrm{T}$ for positive set fields, respectively. Experimental data are shown as triangles while lines represent least-squares fits of Eq. (5) to the respective data sets. The right frame shows the change of the TE (squares) for various field amplitudes ${\mu }_0{H}_{\mathrm{Amp}}$, as measured by the HL remanence magnetization in comparison to the up magnetization branch of the SL hysteresis. The inset (open circles) evidences the linear relation between the remanent magnetization and the bias field in our training experiments. Data are obtained from 20 consecutive loops after saturation at ${\mu }_{0}H=0.8\mathrm{T}$ and initialization in a set field of ${\mu }_0{H}_{\mathrm{set}}=-0.34\mathrm{T}$. The line is the best linear fit to the data.

Figure 3

Dynamics of the TE $|{\mu }_0{H}_B(n=1)-{\mu }_0{H}_B^e|$ measured for various magnetization states of the HL. The HL magnetization at $n=1$ is linearly correlated with the bias field ${\mu }_0{H}_B(n=1)$. The maximum bias field ${\mu }_0{H}_B^{\max }\approx 0.084\mathrm{T}$ is achieved when the HL is saturated and no training appears.