19 May

Controlling the lattice contribution to heat flow using defects

Friday 19 May 2017, 12:00pm

ICN2 Seminar Hall, ICN2 Building, UAB

Prof. Stefan K. Estreicher - Physics Department, Texas Tech University, Lubbock, USA 

Hosted by: CSIC Research Prof. Pablo Ordejón - Theory and Simulation Group Leader and Director at ICN2 

Short Abstract: It is well known that defects impede the flow of heat in crystalline materials. Rudolf Peierls first discussed this issue almost 100 years ago. He visualized ‘lattice waves’ (the word ‘phonon’ had not been invented yet) encountering ‘static perturbations’ (defects) should behave like a water wave encountering a rock in its path: he proposed that defects scatter lattice wave. Igor Tamm later realized that lattice waves must be quantized and Jacob Frenkel coined the word ‘phonon’ to describe these energy quanta. Today, ‘phonon scattering’ by impurities, grain boundaries, interfaces, surfaces, and other defects is the accepted way to think about the interactions between thermal phonons and any disruption to the perfect crystalline order. But is it true?

Defects are not static. They have their own degrees of freedom and vibrational modes. At the atomic level, heat front – defect interactions involve the coupling between bulk-related and defect-related vibrational modes. These two types of modes differ in two fundamental ways. First, defect-related modes are localized in space while bulk modes are fully delocalized. Second, once thermally or optically excited, defect-related modes exhibit much longer long vibrational lifetimes than bulk modes, even when they have similar frequencies. We are studying these interactions using ab-initio molecular- When exposed to an incoming heat front, defects trap phonons for meaningful lengths of time. The decay of the trapped excitations depends on the availability of the receiving modes. Thus, the outcome of heat front – defect interactions depends on the temperature window and can be predicted. This dynamics (MD) simulations strongly suggests that it can also be controlled.

In this talk, I will discuss MD simulations without thermostat and excellent temperature control. I will show that defect-related modes have much longer vibrational lifetimes than bulk modes. Then, I will show the results of ‘theoretical experiments’ (for lack of a better word) which show that phonon-defect interactions are temperature dependent and can be predicted. I will conclude with ideas about a simple ‘thermal circuit’ designed to remove heat from a Si chip without the help of mechanical fans.