Abstract
Many methods for the measurement of the thermal conductivity of thin films have been reported in the previous scientific literatures. Because each class of thin-film structure presents an almost unique set of experimental impediments to overcome, no particular measurement method has become universally accepted. Therefore, different strategies and many techniques are needed for developing a simple, convenient and reliable measurement method for each class thin-film. In this study, we had derived a set of mathematical analytical solution from a complete heater-film-substrate system model. Based on the analytical solutions, we had successfully developed three novel methods for three different thickness ranges of thin/thick film. In the range of film thickness between 50 nm to 2 μm, a novel method, called parallel-strip method, had been developed and three types of SiO2 been measured in that work. The measured results agree with that of the previous literatures. In the range between 2 μm to 10 μm, a modified parallel-strip method had been developed and four types of thermoelectric thin films fabricated by electrodeposition process had been measured, also an epoxy resin layer, substitute for SiO2 to serve as the dielectric layer, were introduced to the sample preparing process. For thick-film of 10 μm to 1000 μm in thickness, a novel method, called thickness difference method, had been built. The method is very simple because derived from a concise semi-empirical correlation. SU8 thick film had been tested by using the novel method and yielded a quite accurate result compared with the previous literatures. By using parallel-strip method and a sandwiched film structure, metal-dielectric interfacial thermal resistance had also been studied in this work. A metal layer of thickness about 10 nm, including Cr (chromium), Ti (titanium), Al (aluminum), Ni (nickel) and Pt (platinum), is sandwiched between two PECVD SiO2 layers of thickness 100 nm. The estimates, 10-10~10-9 m2 K/W, calculated with a continuum two-fluid model are significantly smaller than the measured values, ~10-8 m2 K/W. The continuum two-fluid model, which according to the phenomena of electron-phonon nonequilibrium near the interface in a metal, cannot explain completely the cause of this metal-dielectric interfacial thermal resistance. From photographs of the TEM cross section, we argue that defects at an interface likely play an important role in the magnitude of the interfacial thermal resistance.