Spin-dependent heat transport and thermal boundary resistance

Spin-dependent heat transport is a new research area and can create many future applications. The Giant Magnetoresistance (GMR) effect, which was discovered in 1988, is significant change in electric resistance due to spin-dependent electron scattering. The GMR effect has greatly impacted on techniques of data storage and magnetic sensors. For example, the areal density of hard disk drive was increased 100 times using the GMR effect. Likewise, spin-dependent heat transport, which is also called the Giant Magnetothermal Resistance (GMTR) effect, is expected to create a wealth of new applications, for example nanoscale heat or temperature detectors and spin thermoelectrics. In addition, the technique developed for this study will help with heat management in micro/nano electronics including data storage devices and heat/energy assisted magnetic recording.

In this thesis, thermal conductivity change depending on the magnetic configurations has been studied. In order to make different magnetic configurations, we developed a spin valve structure, which has high MR ratio and low saturation field. The high MR ratio was achieved using Co/Cu multilayer and 21Å or 34Å thick Cu layer. The low saturation field was obtained by implementing different coercivities of the successive ferromagnetic layers. For this purpose, Co/Cu/Cu tri-layered structure was used with the thicknesses of the Co layers; 15 Å and 30 Å. For the thermal conductivity measurement, a three-omega method was employed with a thermally isolated microscale rod. We fabricated the microscale rod using optical lithography and MEMS process. Then the rod was wire-bonded to a chip-carrier for further electrical measurement. For the thermal conductivity measurement, we built the three-omega measurement system using two lock-in amplifiers and two differential amplifiers. A custom-made electromagnet was added to the system to investigate the impact of magnetic field.

We observed titanic thermal conductivity change depending on the magnetic configurations of the Co/Cu/Co multilayer. The thermal conductivity change was closely correlated with that of the electric conductivity in terms of the spin orientation, but the thermal conductivity was much more sensitive than that of the electric conductivity. The relative thermal conductivity change was 50% meanwhile that of electric resistivity change was 8.0%. The difference between the two ratios suggests that the scattering mechanism for charge and heat transport in the Co/Cu/Co multilayer is different. The Lorentz number in Weidemann-Franz law is also spin-dependent. The application of this significant thermal conductivity change is remained for future work.

Thermal boundary resistance between metal and dielectrics was also studied in this thesis. The thermal boundary resistance becomes critical for heat transport in a nanoscale because the thermal boundary resistance can potentially determine overall heat transport in thin film structures. A transient thermoreflectance (TTR) technique can be used for measuring the thermal conductivity of thin films in cross-sectional direction. In this study, a pump-probe scheme was employed for the TTR technique. We built an optical pump-probe system by using a nanosecond pulse laser for pumping and a continuous-wave laser for probing. A short-time heating event occured at the surface of a sample by shining a laser pulse on the surface. Then the time-resolved thermoreflectance signals were detected using a photodetector and an oscilloscope. The increased temperature decreases slowly and its thermal decay depends on the thermal properties of a sample. Since the reflectivity is linearly proportional to the temperature, the time-resolved thermoreflectance signals have the information of the thermal properties of a sample. In order to extract the thermal properties of a sample, a thermal analysis was performed by fitting the experimental data with thermal models. We developed 2-layered and 3-layered thermal models using the analogies between thermal conduction and electric conduction and a transmission-line concept.

We used two sets of sample structures: Au/SiN_x/Si substrate and Au/CoFe/SiN_x/Si substrate with various thickness of SiN_x layer. Using the pump-probe system, we measured the time-resolved thermoreflectance signals for each sample. Then, the thermal conductivity and thermal boundary resistance were obtained by fitting the experimental data with the thermal models. The thermal conductivity of SiN_x films was measured to be 2.0 W/mK for both structures. In the case of the thermal boundary resistance, it was 0.81´10^-8 m²K/W at the Au/SiN_x interface and 0.54´10^-8 m²K/W at the CoFe/SiN_x interface, respectively. The difference of the thermal boundary resistance between Au/SiN_x and CoFe/SiN_x might be came from the different phonon dispersion of Au and CoFe. The thermal conductivity did not depend on the thickness of SiN_x films in the thickness range of 50-200nm. However, the thermal boundary resistance at metal/SiN_x interfaces will impact overall thermal conduction when the thickness of SiN_x thin films is in a nanometer order. For example, apparent thermal conductivity of SiN_x film becomes half of the intrinsic thermal conductivity when the thickness decreases to 16nm. Therefore, it is advised that the thermal boundary resistance between metal and dielectrics should be counted in nano-scale electronic devices.