In the previous sections, the magnetic properties of SAFs are investigated. In this section, EB is added to the stack by coupling an AF IrMn layer to the bottom Co layer. Since the AF coupling of the first AF peak is unstable at high temperatures, the thickness of the Ru spacer layer is choosen at 2.8 nm, which is within the second AF region. combining the EB with the synthetic antiferromagnet results in a pinned magnetic multilayer with almost no net magnetic moment. Section 2.6 discussed the theoretical hysteresis loops that are expected when an EB-SAF stack is analysed.
μ0H)
Field)cooling)direc5on)
ΔM2)
ΔM1)
HC)
H2)=)42.8)mT))
H1)=)D10.0)mT) H1+H2=) Heb=32.8)mT)
Figure 4.12: Typical hysteresis loop measured with the VSM-SQUID setup of a Ta(2)/Pt(5)/IrMn(12)/Co(3)/Ru(2.8)/Co(3)/Ta(5) sample. The sample is grown at standard conditions, except for the Co layers which are deposited at a pressure of 15 sccm and a power of 20 W. The loop shows two magnetic switches ∆M1 and ∆M2
for the top and bottom Co layer respectively. The magnetic loop is not symmetric due to the EB. The measurement is done at 100º C as part of a relaxation measurement sequence.
Section 4.4.1. presents typical magnetization loops that are experimentally observed and are compared to simulated ones.
4.4.1 Magnetic properties of EB-SAF stacks
Figure 4.12 shows a typical magnetization loop for a EB-SAF sample. The magnetic loop shows a two-step switching mechanism, as the Co layers switch at different field values. At small fields, lineair behavior of the magnetization versus applied field is observed. This linear behavior indicates that the angles of the magnetizations rotate over small angles. When no field is applied, a net magnetization is found.
Furthermore, the two switching steps vary in amplitude as indicated by ∆M1 and
∆M2 in figure 4.12. This indicates that the magnetizations in the two Co layers are not identical, and although they may align antiferromagnetically at zero applied field,
when the magnetization is switched. As explained in section 2.2.3, unstable grains contribute to the coercive field. When the applied field is large enough, one of the Co layers starts to switch completely, until saturation is reached. Section 2.6 explained that the EB field can be extracted by analysing the fields at which the Co layers are starting to rotate. The EB increases the field that is needed to start switching the magnetization of the bottom Co layer. The coercivity that is present makes it difficult to pinpoint the exact field at which the bottom Co layer starts to rotate.
The switch fields H1 and H2 are calculated by means of lineair fits through the hysteresis loops. The sum of the switching fields H2+ H1 is the EB field, which for the measurement shown in figure 4.12 is equal to 32.8 mT. Simulations are done and compared to the experimental data, as shown in figure 4.13. In this simulation, the total magnetization per unit area for the stack is minimized for the angles of the magnetizations α and β for different applied fields. This energy is given by
E = −µ0HMFtF 1cos(α) − µ0HMFtF 2cos(β) − JRKKYcos(β − α) − JEBcos(β) (4.1) The first two terms are the Zeeman energy terms of the top and bottom FM layers respectively, the third term is the RKKY coupling coëfficient coupling the two FM layers and the fourth term is the EB term, coupling the AF layer with the bottom FM layer. The difference in magnetization of the two magnetic layers as is observed experimentally is implemented by adjusting the thickness of the top FM layer. JRKKY = −0.06 mJ/m2 as calculated in the previous sections when analysing SAFs. Jeb = 0.1 mJ/m2 in order to match the observed EB field.
There are several differences when comparing the simulation with the experimen-tally observed loop. The magnetization at small negative and positive fields changes, unlike the simulations which show a constant magnetization. This means that the magnetizations are susceptible to small fields. This lineair behavior can not be re-produced using the macrospin model. Hysteresis is experimentally observed during the magnetic switching of the exchange biased layer due to the unstable AF grains.
Within the simulation model, this is not accounted for as the magnetic layers are considered to contain a macrospin and not small AF grains.
The simulations suggest the presence of spin flop when the magnetizations of the two layers are switching. Spin flop originates from the AF coupling of the two FM layers. For example, consider the magnetic switch at negative fields. Although the magnetization of the bottom FM layer is aligned with the applied field direction, it will rotate over a small angle when the magnetization of the top layer is switching.
This means that when a field is applied along the field cooling direction, the magne-tization of the exchange biased FM layer can still rotate due to this spin flop, which can influence the stability of the EB. In order to investigate this, the stability of the EB at different applied fields is investigated and presented in the next section.
μ0H) Field)cooling)direc5on)
Figure 4.13: Simulation of the magnetization (Orangle line) plotted with the angles of the magnetization in the top (red) and bottom (blue) Co-layer versus applied field.
For this simulation MF = 1.055 ∗ 10−6 J/m3, tF 1 = 2 nm, tF 2 = 3 nm, JRKKY =
−0.06 mJ/m2 and JEB = 0.1 mJ/m2.
Co 3 nm D1000)mT) D18)mT)0)mT)20)mT) 62.5)mT) 1000)mT)
Figure 4.14: Overview of fields used during the relaxation measurements. Field values are chosen at 0 field, at large negative and positive fields (+/- 1T), at (-18 mT and 20 mT) and at fields where magnetic switching occurs and spin flop can be present (-27.5 mT and 62.5 mT).