A TEM investigation of controlled magnetic behaviour in thin ferromagnetic films
Understanding the magnetic behaviour of thin film elements is of major importance for the magnetic sensor and storage industries, but also for fundamental micromagnetics. To store digital information, each memory element must support two distinct remanent magnetisation configurations that can be switched between using an applied field. In magnetoresistive random access memory (MRAM), a low switching field and reproducible reversal behaviour are desirable properties. The low field keeps the power consumption to a minimum and the reproducility enables efficient writing and read back of data. However, simple geometric structures are able to support a variety of metastable remanent configurations which can be problematic for device applications. For example, with rectangular elements, the switching fields are history dependent, and there is the possibility of flux-closure formation on repeated switching. This means different field strengths may be required to reverse the magnetisation of the same bit (binary digit) during different field cycles, and the information stored in a cell could be accidentally lost. In addition, the miniaturisation of these elements faces the problem that the coercivity is inversely proportional to element width for a given thickness; a factor which limits their use in high density arrays. The optimum geometry for supporting the stored information is therefore an important issue. In this thesis, different element shapes designed to tackle these problems have been investigated using transmission electron microscopy (TEM) backed by micromagnetic simulations. It has been found that variations in element geometry and symmetry can lead to a greater control of the states that can be formed. Alongside this work on patterned elements, continuous film multilayer samples in the form of magnetic tunnel junctions (MTJs) have also been studied. These multilayer structures serve as storage cells in MRAM devices so their successful operation is of the utmost importance to the development of this technology. At the most basic level, MTJs comprise two ferromagnetic layers separated by a layer of electrical insulator. Whilst one magnetic layer is fixed (pinned layer), the other is free to switch direction when an external field is applied (free layer). Ideally the free layer hysteresis loop would be centred at zero field, but because of magnetostatic interactions caused by layer roughness, the ferromagnets couple to one another and the hysteresis loop is offset. This shift means that the fields required to switch the cell in opposite directions are different. In collaboration with Philips Research in Eindhoven, the magnetic and physical structure of new MTJ stacks incorporating an artificial antiferromagnet (AAF) in the free layer were studied using TEM. An AAF consists of two ferromagnetic layers coupled anti-parallel through a thin layer of non-magnetic metal, typically Ru. These samples were found to reduce the offset field by up to 36% when compared to the basic MTJ stack. Whilst this research is valuable to the magnetic storage industry, the information it provides on these complicated magnetic systems is equally beneficial for solid state physics.