IMAGIC
Integrated Magnetic imAgery
based on spIntronics Components
The Giant MagnetoResistance (GMR) effect has been discovered in Fe-Cr multilayer. In absence of external field, antiferromagnetic coupling creates an alternation of moments. In this case, both electron channels are strongly difused in the layers and the resistance of the device is high. If a magnetic field is applied that tends to align the moments in the same direction, one electron channel is only slightly
difused and the resistance is low. Though historic this first device has a low sensitivity. More sensitive devices used nowadays in hard-drive read heads are Spin Valves which consist in a nanometric stack of Ferromagnetic-Nonmagnetic-Ferromagnetic layers.
Spin valve GMR sensors
The spin valve is based on a a three layers structure. The first magnetic layer is built in order to be extremely fixed in presence of an external field. This layer is called reference layer and is generally made of a antiferromagnet (PtMn, IrMn for example) coupled with a thin ferromagnet (CoFe for example). The second layer is non magnetic and its role is to decouple the magnetic response of the first and third layers. Its thickness must be as small as possible but thick enough to realize this magnetic decoupling. A typical thickness of 2.5nm of Cu is necessary. The third layer is a second magnetic layer but fabricated in order to rotate as freely as possible in the direction of the applied external field. This layer is usually composed of a soft material (NiFe) coupled to CoFe. The role of CoFe is to increase the magnetoresistance effect if it is placed near the Cu layer.
Figure 1 : example of stack for a GMR sensor. It contains an artificial antiferromagnet (CoFe/Ru/CoFe) inside the reference layer
The response of a spin valve sensor will depends on several parameters. The first one is the composition of the stack and the quality of growth. this will determine the amount of magnetoresistance and the temperature variation of this effect. The second parameter is the shape of the sensor which will determine the linearity of the response and the hysteresis of the response. The third is the creation of a specific anisotropy for the free layer by for example and external bias field. Playing on all these parameters, it is possible to optimize the response for a specific application and also the signal to noise ratio. Figure 1 gives an example of stack used for magnetic sensing and figure 2 gives a response of an optimised magnetic sensor. The MR variation is typically between 6 to 10% but values up to 15% have been reported
