Heat transport by phonons in crystalline materials and nanostructures

This dissertation presents experimental studies of heat transport by phonons in crystalline materials and nanostructures, and across solid-solid interfaces. Particularly, this dissertation emphasizes advancing understanding of the mean-free-paths (i.e., the distance phonons propagate without being scattered) of acoustic phonons, which are the dominant heat carriers in most crystalline semiconductor nanostructures. Two primary tools for the studies presented in this dissertation are time-domain thermoreflectance (TDTR) for measurements of thermal conductivity of nanostructures and thermal conductance of interfaces; and frequency-domain thermoreflectance (FDTR), which I developed as a direct probe of the mean-free-paths of dominant heat-carrying phonons in crystalline solids.;The foundation of FDTR is the dependence of the apparent thermal conductivity on the frequency of periodic heat sources. I find that the thermal conductivity of semiconductor alloys (InGaP, InGaAs, and SiGe) measured by TDTR depends on the modulation frequency, 0.1 ≤ f ≤ 10 MHz, used in TDTR measurements. Reduction in the thermal conductivity of the semiconductor alloys at high f compares well to the reduction in the thermal conductivity of epitaxial thin films, indicating that frequency dependence and thickness dependence of thermal conductivity are fundamentally equivalent.;I developed the frequency dependence of thermal conductivity into a convenient probe of phonon mean-free-paths, a technique which I call frequency-domain thermoreflectance (FDTR). In FDTR, I monitor the changes in the intensity of the reflected probe beam as a function of the modulation frequency. To facilitate the analysis of FDTR measurements, I developed a nonlocal theory for heat conduction by phonons at high heating frequencies. Calculations of the nonlocal theory confirm my experimental findings that phonons with mean-free-paths longer than two times the penetration depth do not contribute to the apparent thermal conductivity. I employed FDTR to study the mean-free-paths of acoustic phonons in Si1-xGex. I experimentally demonstrate that 40% of heat is carried in Si1-xGe x alloys by phonons with mean-free-path 0.5 ≤ ℓ ≤ 5 mum, and phonons with > 2 mum do not contribute to the thermal conductivity of Si.;I employed TDTR and frequency-dependent TDTR to study scattering of long- and medium-wavelength phonons in two important thermoelectric materials embedded with nanoscale precipitates. I find that the through-thickness lattice thermal conductivity of (PbTe)1-x/(PbSe)x nanodot superlattices (NDSLs) approaches the thermal conductivity of bulk homogenous PbTe1-x Sex alloys with the same average composition. On the other hand, I find that 3% of ErAs nanoparticles embedded in InGaAs is sufficient to scatter most of the phonons in InGaAs that have intermediate mean-free-paths, and thus reduces the thermal conductivity of InGaAs below the alloy limit. I find that scattering by nanoparticles approach the geometrical limit and can be readily accounted for by an additional boundary scattering which depends on the concentration of nanoparticles.;Finally, I studied the thermal conductance of Au/Ti/Graphene/SiO 2 interfaces by TDTR. I find that heat transport across the interface is dominated by phonons. Even though graphene is only one atomic layer thick, graphene interfaces should be treated as two discrete interfaces instead of one diffuse interface in thermal analysis, suggesting that direct transmission of phonons from Au to SiO2 is negligible. My study is important for thermal management of graphene devices.