Phonon thermal conduction in nanostructured materials via molecular dynamics and coarse grain dynamics simulation

Ya Zhou, Purdue University

Abstract

Thermal management in nanometer size scale devices is becoming a major challenge with the development of nanotechnology. In many applications, e.g. thermoelectrics, it is critical to be able to control electron and phonon thermal conduction separately. To gain a microscopic understanding of phonon thermal transport mechanisms, molecular dynamics (MD) and mesodynamics simulations are used to characterize thermal conduction in nanostructured metallic systems and molecular materials. MD simulations are performed on metallic nanostructures. Size and temperature effects are characterized in homogeneous samples. Effects of acoustic mismatch, strain, and defects are investigated in both coherent and incoherent nanolaminates. A thermal rectifying effect of interfaces is observed, showing that thermal conduction is more efficient when heat flows from the light to the heavy side of a material. When interface properties are determined, thermal conductivity of nanolaminates is dependent on the thickness of the layers. Two additive models are proposed, which accurately describe the MD data of nanolaminate thermal conductivity by taking size effects into account. Effects of substrate properties on thermal conductivity of nanowires are investigated in nanowire-substrate structures. Vibrational frequencies of substrate atoms are modified by changing the substrate atomic mass. Thermal conductivity of nanowires increases noticeably as the frequencies of the substrate vibrational modes are reduced, due to the dominant role of low frequency phonons during thermal transport. A model based on phonon dispersion relationships of a linear atomic chain is proposed to capture the underlying physics. For the characterization of molecular materials, all-atom MD appears to be computationally intensive. Dynamics with implicit degree of freedom (DID), a coarse grain method based on MD, is used to explore the thermal role of degrees of freedom (DoFs) internal to molecules in a model molecular material. In solid samples, internal DoFs are localized and do not contribute to thermal conduction. In liquid samples, thermal conductivity and internal specific heat are linearly correlated. The coefficient of proportionality is equal to mass diffusivity. The results of DID simulation can be used to make quantum corrections to the results of classical, all-atom MD simulations.

Degree

Ph.D.

Advisors

Strachan, Purdue University.

Subject Area

Materials science

COinS