Tunnel junctions and first spin-resolved photoemission spectra on CASSIOPEE

For the last several decades, information has been stored and read via a system that uses the magnetic properties of materials. The progress achieved in this field is particularly visible through the miniaturisation of computers, audio and video players, etc., thanks to the fact that the storage capacity of hard disks has shot up exponentially since half a century ago.
The field of spintronics involves the use of both the magnetic and electronic properties of materials, and has been applied, for example, to the design of non-volatile magnetic memory, which can continue to store data in the absence of an electrical power supply.

Magnetic tunnel junctions are one example of the devices to have emerged from spintronics. These are spin valves in which the central non-magnetic layer has been replaced by an insulating layer (e.g., alumina). In this case, electrons cross the central layer via a quantum phenomenon: the tunnel effect. (This effect is also present in the microscopes that have been named after it.) The amplitude of the magnetoresistance obtained in this way is greater than that of a spin valve.
R&D is being carried out very actively in this field. Tunnel junctions are not yet being used in commercial products as spin valves are, but that is sure to change within just a few years.

At the Nancy Laboratory of Materials Physics, Stéphane Andrieu belongs to a team that is working on nanomagnetism and spin electronics. They perform both fundamental and applied research, in particular for the application of tunnel junctions in MRAM (magnetic random access memory).
This group from Nancy is particularly interested in the use of monocrystalline materials for the magnetic or insulating layers of junctions. Since the early 2000s, it has been demonstrated that the magnetoresistance properties of these multilayer materials can be improved even more if monocrystalline materials are used instead of the polycrystalline materials more often used up to now. Compared to a few tens of percent magnetoresistance obtained from polycrystalline spin valves, currently found in the reading heads of computers, values of up to 500% can be obtained for certain tunnel junctions containing monocrystalline materials!

The study of these multilayers includes system polarisation measurements. Strong polarisation is required (but is not the only necessary parameter) to obtain high magnetoresistance. Crystalline symmetries also play a crucial role. These symmetries can also be selected during spin-resolved photoemission measurements on CASSIOPEE, making this beamline an extremely powerful tool in the study of this type of topic.

The Nancy team has been able to make these three-layer materials—a very tricky procedure—since 2001, and has managed to perform lithography on them since 2003. The tunnel junction system has attracted a great deal of attention in the scientific community since then, because of the high percentages of magnetoresistance measured at THALES In 2001 (25%) and then at Nancy in 2003 (70%), and, more recently, in the USA, Japan, and Nancy in 2004 (200%). The French community has therefore been working in the vanguard of this topic.

Stéphane Andrieu and Frédéric Bonell, a PhD candidate, brought their monocrystalline Fe/MgO/Fe(001) three-layer samples with them (they will one day be able to manufacture them on site at SOLEIL, in the preparation chamber which will be available on CASSIOPEE).
The experiments conducted at SOLEIL in October 2008 confirmed that the CASSIOPEE spin-resolved photoemission spectroscopy station is very useful for this type of study. After the Fe/MgO/Fe(001) three-layer samples, Stéphane Andrieu will come back with other tunnel junctions, e.g. some with impurities, such as V, in the magnetic layers. The aim will still be to understand the mechanisms leading to these strong system magnetoresistances.

The Nancy team, in collaboration with scientists from Hitachi, have applied for beamline time in order to continue this research. In less than a decade, these studies could lead to the application of this technology to reading heads in computer hard disks. 

Spectra obtained from a thick layer of epitaxial iron covered in two atomic planes of MgO (i.e. an Fe/MgO bilayer, since measurements are not carried out directly on the Fe/MgO/Fe(001) trilayer). The sample exhibits remanent magnetisation along the horizontal axis in the surface plane. Thanks to the geometry of the experimental setup, spin polarisations can be measured simultaneously along the normal to the surface and along the horizontal axis of the surface plane.
The spectra show two distinct regions: around -5 eV there are the valency states of MgO, and the 3d states of iron appear around -1 eV, in the MgO gap.
As expected, no spin polarisation is observed along the normal to the surface, whilst the 3d states of iron have approximately 45% polarisation along the magnetisation axis.

 

An electron has an electric charge, a mass, and a third intrinsic property: spin, which corresponds to an elementary magnetic moment, and can be in either of two states: 'up' or 'down'.
A non-magnetic material (whether metallic, insulating, or a semiconductor), has equal numbers of 'spin up' and 'spin down' electrons. In a magnetic material, on the other hand, the number of 'spin up' and 'spin down' electrons is different, and that difference is what makes the material magnetic. When the up/down difference is large, the material is said to have strong polarisation.

In the usual magnetic metals (iron, cobalt, nickel and their alloys), the electrons which are responsible for magnetism are the same ones which are involved in electrical transport. As a result, the electrical current flowing in a magnetic material is spin-polarised.

Research was carried out by Albert Fert in the 1970s on the movement of electrons in a ferromagnetic layer according to their polarisation. It showed that an electron does or does not pass through the layer depending on whether or not the direction of its spin matches the magnetic orientation of the layer.

This means that electron filters can be created: a thin magnetic layer (a few nanometers thick) placed in the path of a current of electrons will mainly allow the passage of those whose spin is parallel to its magnetisation, and will mostly block the others.
By alternating three thin layers: ferromagnetic / non-magnetic / ferromagnetic, we can modify the resistance to the passage of electrons. The resistance is greatly increased in multilayers such as Fe/Cr/Fe when a magnetic field is applied; this is the giant magnetoresistance effect, which earned its discoverers the 2007 Nobel Prize for Physics.
These three-layer sandwiches are called spin valves.