6.2 Discussion
6.2.2 Influence of IrMn grains
A macro-spin interpretation of the switching such as the simulation with the LLG-equation, dis-cussed in chapter 3, is unable to account for any local effects. In the study on temperature related effects on the orthogonal exchange bias in the previous chapter (section 5.5), it is already seen that a grain interpretation of the AF-layer can explain experimental observations. Hence, investi-gating the influence of such effects might be critical to observations in this chapter as well. This
section will elaborate on the influence of IrMn grain structure.
Sputtered IrMn thin films consist of a polycrystalline grain structure with a distribution of dif-ferent grains sizes [24]. No experimental evidence for coupling between these grains has been found. Hence, the grains can be described individually. This grain size distribution has a couple of important implications for the behavior of the system.
Di str ibution
Figure 6.9: Schematic distribution of grain sizes for polycrystalline IrMn after field cooling to room temperature. Thermal energy at room temperature is enough to keep the random orientation of small grains. Average sized grains are set to the desired direction by the field cooling procedure as their blocking temperature is less than the annealing temperature. The field cooling procedure did not reach the blocking temperature for larger grains and they are still bound to different crystal axes of the IrMn, as in their as deposited state. Figure adapted from [26]
An example of a grain size distribution is shown in figure 6.9. Such a distribution is experimen-tally verified by O’Grady et al. for very similar samples as the ones used in this thesis [26]. The spin orientations indicated in the distribution correspond to a sample at room temperature that is previously field-cooled with an annealing temperature of TAN. As can be seen, depending on the grain volume V , a part of the grains are either set, unset or thermally unstable. This can be explained by the crystalline (volume) anisotropy KAFwhich determines the energy barrier of the grain via ∆E = KAFV. Grains below a critical volume Vc lead to a small energy barrier, which are thermally unstable even at room temperature. Between Vcand VANthe field cooling process thermally activates all grains for which Ethermal > KAFVAN. After field cooling those grains will point along the crystalline axis closest to the field cooling direction. Grains for which V > VAN are not thermally activated due to the field cooling process and are therefore still in the unset initial orientation. Of great importance is the fact that the set and unset AF spins are bound to a crystallographic axis of the AF, as discussed in section 2.4.5. Due to the polycrystalline growth and the absence of inter-granular coupling, orientation of the grains are not identical. For the set AF grains, this means that the AF spins will align to the crystal axis closest to the applied field direction. As a result, the exchange bias direction varies between grains.
Another important effect that should be included is the experimentally observed grain-size
de-HEB
HEXT
IrMn Co
x y
Jpulse=0.5Jmax
Jpulse=Jmax Jpulse=Jmax
Jpulse=Jmax
Unset grain Thermally unstable grain
a.
b. c.
d. e.
Figure 6.10: Schematic representation of the influence of a grain structure on the switching process. a: The IrMn is split up in different grain sizes. The Co grains cor-respond to the IrMn grains and are OOP magnetized. The effective exchange bias is an average of the exchange bias of each different grain. The exchange bias direc-tion and size vary between each grain. Big grains can be unset and point in another direction (see figure 6.9), for simplicity, only the IP component of the exchange bias is shown. b: A small current is applied along the exchange bias direction. The Co grains for which a large exchange bias component along the current direction exists switch deterministically. Random switching of grains is also possible but not taken into account. c: At higher currents, most grains switch. Grains with the wrong ex-change bias direction prevent the Co from switching. d: The current is now applied perpendicular to the exchange bias direction. Most grains can’t switch deterministi-cally as the field cooling set the grains in the y-direction. e: Applying a large field along the current direction results in deterministic switching, independent of the exchange bias direction.
pendent exchange bias. Although reports vary, it is generally found that for thin AF layers (<10 nm) the exchange bias decreases in smaller grains [59] and for thick AF layers (>10 nm) it tends to increase [60][61].
By using the aforementioned grain distribution, some of the poorly understood observations in the experimental switching process in the Hall cross can be explained. Consider figure 6.10a, which shows a schematic representation of the F/AF bilayer with a grain size distribution. After field-cooling along the y-axis, most grains are set along their crystal axis closest to the y-direction.
There is also one big grain which is still unset, and one small thermally unstable grain. A small current is applied along the y-direction (figure 6.10b). The Co underneath the grains which have a large enough exchange bias component along the current direction are able to switch (red Co grains). The rest of the Co only switches non-deterministic due to the wrongly oriented AF grains (blue Co grains). At a larger current, the necessary effective symmetry-breaking field becomes less, so more grains are able to switch (figure 6.10c). The Co below the big AF grain is still prevented from switching due to its exchange bias pointing in the wrong direction.
In figure 6.10d, the current is applied along the x-direction (corresponding to in the perpendicular phase diagram in figure 6.6). This time, the big grain is able to switch as it has the correct symmetries. The other grains can only switch deterministically when an external field is applied to break the symmetry (figure 6.10e).
To summarize, a grain size distribution can explain multiple observations in the experiments:
A Pulses of equal polarity change the magnetization by a small amount Thermally unsta-ble grains or grains set in the wrong direction can result in the underlying Co to reverse non-deterministically when a current is applied, as its symmetry is not broken.
B Gradual switching Grains can have different exchange bias magnitudes and direction which results in different critical current densities for the underlying Co to switch.
F Switch at low current densities Big grains with an exchange bias direction along the cur-rent can cause the Co to already switch at significantly lower curcur-rent densities.