Xavier Hoffer
The aim of this thesis work is to study the spin-dependent transport in multiwalled carbon nanotubes (MWNT). To do so, the electrical resistance of MWNT contacted between ferromagnetic electrodes has been measured, as a function of the contacts magnetic configuration and temperature. An original sample fabrication process to contact MWNT has been developed. The MWNT growth is achieved via chemical vapour deposition in the pores of an alumina template, with electrodeposited nickel or cobalt nanowires as catalyst. The pores are 1,5 µm, in length, and 40 nm in diameter. The second electrical contact is sputter deposited or evaporated over the membrane surface. This geometry allows us to contact MWNT between ferromagnetic electrodes, which magnetisation states versus the applied magnetic field are known. Furthermore, the electrical current flows perpendicularly to the plane of the contacts layers. Finally, it enables us to tune the length of the MWNT between the electrodes. The features of the measured spin-dependent magnetoresistance (SD-MR) signals cannot be correctly interpreted with common tunnel or giant magnetoresistance approaches. Every MWNT which is less than 300 nm long between nickel contacts destroys the spin polarisation of the current. MWNT longer than 300 nm featured spin-dependent magnetoresistance signals, with a small amplitude and depending on the direction of the current. One 500 nm long MWNT between cobalt contacts shows 25% SD-MR signals at 2,5 K, at zero applied current. This signal is caused by a thermopower effect, and not by a spin polarised current. Another signal, from a 700 nm long between cobalt contacts is more similar to usual signals shown by magnetic tunnel junctions. In order to understand spin-dependent transport in MWNT, it is therefore necessary to determine first the electrical transport mechanisms, independent of the spin. Nowadays, it is still poorly understood. Therefore, besides spin-dependent transport measurements, we measured the conductance versus the temperature and the bias voltage. At temperatures below 50 K, the conductance diminishes as the temperature and the bias voltage decrease. This effect, called Zero-Bias Anomaly (ZBA), is a consequence of disorder and electron-electron interactions in our systems. 46 samples out of 113 have shown power law scaling laws of the ZBA. From these scaling-laws, we get the power law coefficient α. Such scaling laws have been observed many times for carbon nanotubes. However, none of these studies has such a large spectra for the values of α as we have : for our samples, α ranges between 0 and 1,7. With samples contacted via cobalt cobalt electrodes, α is usually larger than nickel contacted samples. Therefore, a single parameter α enables us to describe and characterise the electrical transport in our samples. The large number of measured samples allows us to correlate this coefficient α with other experimental parameters, such as the MWNT length, or the metallic or structural nature of the contact electrodes. We also have established the linear relation : ln G0 ~ A · α, with G0 the extrapolated conductance at 1 K. To our knowledge, this relation has not yet been predicted nor observed. The physical interpretation of this relation is interpreted in the Coulomb blockade formalism. Finally, a link between the weak localisation amplitude et the value of α has been displayed. In this case, the weak localisation amplitude not only depends on α, but also on the metallic nature of the electrodes.
EPFL2004
