Phase Change Materials on Metals: Crystallization Behavior and Applications in Magnetic Stacks.

摘要

In this thesis we will show that the activation energies for crystallization of phase change (PC) materials on various sputtered metal surfaces are all similar (Ru, TiW, Pt, and Fe), and are similar to oxide reference surfaces (i.e. thermally oxidized Si wafers). Additionally, the crystallization temperature on the sputtered metals was typically lower than on these reference surfaces. The principle difference between the metals and the oxide reference surfaces, however, does not appear to be the metal/oxide difference. Sputtered oxide films behave more like the metals in this study, suggesting that the effect is due to roughness or some other feature of the sputtered films. Film thickness appears to be a more important feature in determining kinetics than the nature of the PC interface. Experiments with thin Ge2Sb2Te 5 (GST) layers sandwiched between metals show a rather dramatic increase in crystallization temperature with decreasing thickness. These results are similar to literature experiments with GST sandwiched in between dielectrics. As a result, it was concluded that the presence of metal electrodes does not significantly alter GST crystallization behavior as compared to dielectric bounding materials.;In a separate set of experiments, it was found that Ru and TiW were reasonable diffusion barriers for phase change materials during crystallization. Their efficacies as diffusion barriers were determined by a novel method using the reduction of magnetization of Fe as an indicator of the extent of the diffusion. This was also corroborated with XPS depth studies. Ru and TiW were found to be good diffusion barriers at low temperatures (300 °C) but diffusion was found to occur at higher temperatures (460 °C). In comparison, significant interdiffusion was found for Pt samples at all temperatures.;Coupling experiments suggest that there is a very weak antiferromagnetic coupling when using Ru/amorphous GST/Ru composite spacers between magnetic films. This antiferromagnetic coupling is reduced when depositing crystalline GST. However, due to the thinness of the GST spacers (< 3 nm) required and the contradictory change in coupling with excimer laser pulsing, this effect cannot be proven to be unambiguously due to the crystallization or amorphization of the GST. As for magnetoresistive effects, two different phenomena were found to occur in layered magnetic structures separated with PC spacer layers: 1) granular MR and 2) spin valve MR. Both effects are much smaller than conventional MR, and so do not appear to have obvious technological application at this time. Granular MR occurs only for CoFe samples, with no such effect occurring for the Fe samples, and can be enhanced to a maximum value of 0.19% after annealing. Annealing appears to produce a microstructure consistent with granular MR, in which coupled magnetic regions are separated by non-magnetic phase change regions. The annealing process in this case enhances the granularization without causing crystallization of the phase change. As for the spin valve MR, values of MR for Ru/10 nm GST/Ru sandwiched between layers of Fe and FePt were observed to be 0.006 %, increasing to 0.008 % (1.3 X increase) concurrent with the crystallization process. This behavior is ascribed to the increase in the mean free path of the GST upon crystallization.;In summary, the role of metal boundary materials has been found to be similar to dielectrics in terms of crystallization kinetics of phase change materials. Interdiffusion is a separate issue that must be addressed when metals are in contact with PC materials, and Ru appears to be promising in this respect. Thin layers between metals show the same increase in activation energy that they show between dielectrics and this constitutes a major challenge for magnetic stack applications where a very thin PC layers might produce novel device properties. Both magnetic coupling between layers separated by phase change and GMR-like magnetoresistance with PC spacers has been shown in this work. However, the effects are small and may be challenging to increase due the need for diffusion barriers and the need for a minimum PC layer thickness within the spacer stack to permit reversible phase change. (Abstract shortened by UMI.).