Development of the TMR-Stack

A special TMR layer stack is developed for this experiments. The main requirement is that the TMR sensor can detect a single magnetic marker. It was shown theoretically that this is generally possible [18] when the element size is about the same size as the bead. But some special properties of the layer stack are wanted to improve the detection limit. Because the stray field of a single bead is quite small, it will only influence the soft magnetic layer of the TMR sensor locally. To avoid irreproducible switching induced by the small stray field of the particle, the TMR curve (i.e. resistance vs. external field) should have a slow ascending slope around zero magnetic field. This slow ascending slope in the TMR curve means for the TMR element that the softer top ferromagnet is switching reproducible from the parallel to the anti-parallel state. Such a switching behaviour is achieved through a special orthogonal pinning of the top and bottom ferromagnetic layers.

During the sputtering of the TMR layer stack, two magnetic masks are applied. The lower part of the stack, including the bottom ferromagnet and the aluminium-oxide barrier, is sputtered with a 0° magnetic mask, and the upper part is sputtered in a 90° magnetic mask. So, the two ferromagnetic layers are aligned orthogonal to each other (confer figure 6.1a) and because of the pinning to an antiferromagnet, they stay that way when all external fields are removed.

Figure 6.1: (a) Orthogonal pinning of top to bottom magnetic electrodes. (b) Layer stack used for the TMR sensor.

0.3

Several different layer stacks are sputtered (see section 2.1 for information about the used apparatus) and tested for their magnetic properties. The sputtered samples were structured using the UV-mask lithography system (confer section 2.3) with the standard mask (figure 2.3). After ion beam milling down to the bottom electrode (see section 2.2 for more information about the Ar-ion etching process) and removing the resist, the TMR and the I/V curves for the tunnel barrier of the layer stacks are measured. Because the development of the layer stack is not in the main focus of this thesis, only the measurements of the final layer stack are presented here^6.1.

Figure 6.1(b) presents the most suitable layer stack that was chosen for the experiments. The bottom electrode consists of two tantalum and two copper layers of different thicknesses, which are necessary for the Ar-ion etching process. The TMR element itself is composed of a strongly pinned cobalt iron layer (bottom), the tunnel barrier and a weakly pinned cobalt iron layer (top). A protective tantalum layer and a gold layer on top to connect the TMR elements completes the layer stack. The full stack from bottom to top can be written as: Ta(6.4nm) / Cu(28.6nm) / Ta(20.8nm) / Cu
(7.3nm) / Mn$_{83}$Ir$_{17}$(13.2nm) / Co$_{70}$Fe$_{30}$(14.8nm) / Al$_2$O$_3$(1.8nm) / Co$_{70}$Fe$_{30}$(14.8nm) / Mn$_{83}$Ir$_{17}$
(26.5nm) / Ta(5.0nm) / Au(50nm).

The top cobalt iron is pinned by a thicker manganese iridium layer than the bottom ferromagnetic layer. Because the exchange bias correlates with the thickness of the antiferromagnet (confer figure 1.14 on page ), the magnetisation of the top ferromagnetic layer can rotate more easily (confer section 1.5.3). Together with the orthogonal directions of the pinning, this layer stack switches reproducable around zero field. So it achieves the slowly ascending slope of the TMR curve and therefore prevents any irreproducible switching of the soft magnetic layer.