The thermal conductivities of superlattices are essential to improve the properties of thermoelectrics and optoelectronics; however, limited results in relation to both the in-plane and cross-plane thermal conductivities have been reported. A convenient, effective, and accurate experimental method is required to improve the current research on the thermal properties of superlattices. We conducted an experimental research study on two GaAs/AlAs superlattice samples with a total superlattice layer thickness of 2 microm using a combination of the 2-omega and 3-omega techniques. The samples have period thicknesses of 4 nm and 10 nm, respectively.

To explore the thermal conductivities of the substrate and insulation layer of the superlattice samples indirectly, a controlled sample with the same structure, but without a superlattice layer, is used. We obtained the thermal conductivities of the GaAs substrate and insulation layer (SiO2 thin film) using the 3-omega technique and FEM simulation model. We also explored the deviation of the experimental results of the 2-omega technique from the Fourier's Law through the controlled sample. These parameters obtained from the controlled sample are used in the data analysis in the following superlattice research. In the superlattice study, we combine the 3-omega and 2-omega techniques to characterize the anisotropic thermal conductivity of GaAs/AlAs superlattice from the same wafer. The in-plane thermal conductivity, cross-plane thermal conductivity, and anisotropy are obtained from the same wafer by comparing the experimental results with the FEM simulated results. This combination works fine in general and demonstrates a significant reduction in thermal conductivity compared to that of equivalent bulk materials. Superlattices with different period thicknesses but the same total superlattice thickness present a significant difference in both the in-plane and cross-plane thermal conductivities of the superlattices.

However, we have found that the 3-omega technique is sensitive to the thermal conductivity of the insulation layer, which will affect the reliability of the results if the measured SiO2 thermal conductivity is not accurate enough. However, this effect should be able to be reduced or eliminated by using a much wider metal line than that used in the current research, and the reason for this is explained in the future work section of the last chapter. In addition, the numerical simulation results of the different thicknesses and different anisotropies of superlattices by considering the minimum and maximum SiO2 thermal conductivities are also presented in the last chapter for future reference. The thermal conductivity variance in SiO2 has a small effect in general, particularly on the 2 microm and 10 microm thick superlattices when a 10 microm wide wire is used in the 3-omega FEM model.