By Professor R Leckey
Soft X-rays from a synchrotron radiation source, coupled with the technique of photoelectron spectroscopy, have played a critical role in ensuring the success of Moore's law. Gordon Moore, then President of Intel Corp. predicted in 1965 that the computational power and speed of processor chips would double every 18 months. That this has continued to be true until the present day has required ever more detailed knowledge of processes occurring at the nanometre level on the surface of materials such as silicon and gallium arsenide.
Much of our knowledge concerning the behaviour of metallic layers on semiconductors has come from the ability of electron spectroscopy, in association with synchrotron radiation, to observe the electronic and chemical consequences of the surface processing of each of the materials involved.
As an example, it may be vitally important to know if a line of cobalt, for example, when deposited on silicon, diffuses laterally across the surface, forms a resistive contact, or indeed as actually happens, reacts to form cobalt di-silicide CoSi2. This compound is a semi-metal and the track forms a useful Schottky diode barrier with the substrate ensuring that any signal stays on the track rather than being dissipated in the underlying silicon.
Photoelectron spectroscopy of such a system enables us to understand the creation of the silicide by observing changes in the binding energy of the atomic levels of the constituent atoms and in the profile of these peaks. There is a clear advantage in being able to maximise the photoemission cross-section and therefore the visibility of the atomic levels by utilising the fact that, unlike laboratory sources, synchrotron radiation can be tuned to almost any photon energy at will.
Whereas most of the increase in computational power has resulted from a steady decrease in the dimensions of the individual transistors in a processor chip, our ability to take advantage of ever more powerful software has also required the development of vastly improved storage capacity. Decreases in physical size coupled with spectacular increases in storage capacity of modern hard disc drives have also only been possible with the direct involvement of surface science in its many guises.
Early designs of read/write heads for magnetic disc storage relied on the production/detection of a local magnetic field generated by miniature coils of wire, thereby placing a limit on the extent to which they could be miniaturised. Current heads utilise the Giant Magneto Resistance (GMR) effect that occurs in multi-layer films of nanometer thickness involving dissimilar materials. Unraveling the origin of the GMR effect has required the full power of synchrotron radiation photoemission.
When photons from a synchrotron enter a solid, it emits electrons whose kinetic energy (Ep)can be measured using a specific type of spectrometer. Since the energy of the photons (Ep) is pre-determined, the binding energy ( Eb ) of each type of electron in the material can be readily calculated. The kinetic energy of an emerging photoelectron (Ek) is directly related to its binding energy within the solid (Eb) by the Einstein relation:
Ep = Eb + Ek
This process is illustrated in figure 1.
Figure 1: When a photon from a synchrotron interacts with a solid, an electron may be emitted whose kinetic energy is directly related to the binding energy of the atomic level from which it was ejected. A photoelectron spectrometer determines the distribution of such energy levels by measuring the kinetic energies and direction of travel of the emitted electrons.
In a crystalline solid, in contrast to individual atoms, the binding energy of conduction electrons varies with direction of travel. The details of this dependence, namely the band structure of a material determines its electronic, magnetic, optical and chemical behaviour. By monitoring the intensity of photoelectrons ejected from a crystal by photons of known energy from a synchrotron, as a function of their kinetic energy and also noting their direction of emission, we can determine the band structure almost directly.
It is vital to have access to a wide variety of photon energies hence such experiments cannot be performed without access to a source of synchrotron radiation. The fact that the synchrotron can produce light which is either linearly or circularly polarised is particularly important for the investigation of magnetic properties.
We accomplish our measurements using an electron spectrometer of toroidal geometry that we have developed at La Trobe University and have used for many years at the Berlin synchrotron radiation facility in Germany.
Electrons photo-emitted from a sample and having a particular kinetic energy follow paths within this spectrometer so that they arrive at a position sensitive detector where their location can be directly related to their angle of emission. For each electron detected, we therefore know both its kinetic energy and its direction of emission. By rotating the sample about the incoming light beam we can acquire information about the intensity distribution of electrons covering the entire emission hemisphere.
Armed with such information, we can 'see' almost directly one of the most important features that controls the electronic behaviour of metals, alloys and semi-conductors. This feature is known as the Fermi surface and, until the arrival of synchrotron radiation and spectrometers such as our toroidal instrument, its appearance could only be inferred from difficult, very low temperature, experiments.
It is the Fermi surface that provides the information needed to understand how the GMR effect works and consequently how to improve hard disc capacity.
By separating two thin layers of a magnetic material with a precise thickness of a non-magnetic material, the electrical resistance of the sandwich depends strongly on whether the two magnetic layers are magnetised parallel or anti-parallel to each other. If the lower magnetic film encounters a magnetised patch on the moving plate within a hard disc drive, its magnetic state changes and the resistance of the sandwich alters dramatically, but reverts to its previous value after the recorded patch has passed. The lateral dimensions of the sandwich can be reduced to sub-micron levels while retaining adequate sensitivity, and consequently ever-smaller magnetic patches may be written and read by such heads leading to ever-higher packing densities on the disc.
The surprising fact that the non-magnetic layer needs to be of a particular thickness for the GMR effect to exist posed a challenge for the surface science community. For the case of two iron layers separated by a gold film, it was observed that the GMR effect occurred only for certain thicknesses of the Au layer. In particular, a gold film thickness of about 1.5nm (corresponding to 8 atomic layers, monolayers, of Au) shows the most pronounced GMR effect.
Figure 2 shows a section through the gold Fermi surface as determined using the photoemission technique at a synchrotron radiation facility. Each arrow links two electron states which are strongly coupled. The lengths of the arrows provide us with the wavelengths of electrons which will find it possible to interact strongly with each other. It was soon realised that these wavelengths exactly correspond to the gold film thicknesses which characterise the GMR effect in iron/gold/iron devices.
Figure 2: A section through the Fermi surface of gold showing (arrows) the scattering processes responsible for the GMR effect in Fe/AU/Fe layers. The length of the arrows determines the wavelength of the electrons involved. When the gold film thickness equals one such wavelength strong interactions are possible across the gold film separating the iron layers. [Fuss et al. 1992]
The complexities of working with the electronic structure of crystalline materials may often appear unrelated to practical devices and restricted to the realm of pure fundamental research. I hope that this short article provides two examples of how modern computer technology would not exist in its present form without critical input from fundamental synchrotron radiation surface science studies.
Reference A.Fuss, S.Demokritov, P.Grunberg, W.Zinn: J.Magn.Mat.103 (1992) L221
|
