The lattice thermal conductivity is a fundamental thermal transport parameter that determines the utility of materials for specific thermal management applications. Accurate theoretical modeling of the lattice thermal conductivity is essential to numerous fields including microelectronics cooling, thermoelectrics, and even planetary science. The goal of this collaborative research effort will be to implement a theoretical approach to calculate the lattice thermal conductivity of crystalline and alloyed materials from first principles. A central feature of this approach is that it has no adjustable parameters. The focus at the Boston College site will be in using an exact numerical solution of the phonon Boltzmann equation to calculate the lattice thermal conductivity. The materials to be studied in this project include lead chalcogenides, I-V-VI2 semiconductors, and nanoparticle-in-alloy-structures. For nanostructured systems, such as the nanoparticles in alloys, a Non-Equilibrium Green?s Function approach will be implemented. The critical inputs for the transport modeling will come from harmonic and, where necessary, anharmonic interatomic force constants, whose calculation will be the focus of the Cornell site. The first principles approach has already demonstrated excellent agreement with measured high thermal conductivities of group IV semiconductors. The materials to be studied in this project are unified by their exceptionally low thermal conductivities and therefore provide an excellent test of the robustness of the theory. The measured thermal properties of many of these materials are well characterized and will provide an useful check of our calculated adjustable parameter-free results. Good agreement with measured data would further validate the predictive capability of this state-of-the-art theory in studying and understanding thermal transport and the lattice thermal conductivity in a wide range of materials for many thermal management applications.
Intellectual merit: Current theories of the lattice thermal conductivity of materials are typically based on either highly parameterized relaxation time approximations or on purely classical molecular dynamics calculations. The rigorous first principles theory proposed here has no adjustable parameters and incorporates fully the quantum mechanical phonon scattering processes. It therefore could provide currently unavailable predictive power to support ongoing and future experimental studies of thermal transport in materials, as well as contributing to the development of new highly efficient materials engineered for desired thermal management applications.
Broader impacts: The project will provide training for one postdoctoral researcher and one doctoral graduate student. In addition, several undergraduates including those from underrepresented groups will participate through NSF REU programs at both Boston College and Cornell sites. The computational tools to be developed during this project will be incorporated into the publicly available computing library of the Cornell Nanoscale Science and Technology Facility (CNF). The activity will also benefit society by aiding in the development of new materials with desired thermal transport properties. This will facilitate technological breakthroughs that may lead to the next generation of thermoelectric materials, thermal barrier coating materials, and thermal interface materials.
The thermal conductivity of a material gives a measure of how easily heat can flow through it. Materials with low thermal conductivity are used in environmentally clean thermoelectric refrigeration and power generation devices. However, the efficiencies of the best thermoelectric materials have been too low to compete with conventional, fossil-fuel based cooling and power generation systems. On the other end of the spectrum, materials with high thermal conductivity are needed for passive cooling of microelectronic devices to help channel heat away from hot spots on computer chips, thereby increasing their performance and reliability. Diamond is currently the best material for this application because of its extremely high thermal conductivity. But, diamond is expensive and difficult to make. In order to improve performance efficiencies and reduce cost, new materials with very low and very high thermal conductivities need to be found. To address this, one approach would be to try to find promising new materials by synthesizing them in the lab from different combinations of elements in the periodic table. However, the number of possible combinations is enormous, and fabrication of a desired new high-quality material is time-consuming, costly, and frequently not even possible. A better approach is to develop accurate ways to simulate the thermal conductivity using computer models because such models can rapidly and inexpensively assess a candidate material’s thermal conductivity and thus its potential utility for applications. They can also contribute fundamental scientific understanding into the nature of heat flow, which can aid in the search for new high and low thermal conductivity materials. Until recently, most models describing heat flow in materials used a number of free parameters, which were adjusted to fit measured data for each material being considered. Such models cannot be used reliably when no measured data is available, as, for example, is the case when evaluating the potential advantages of a material that has not yet been made. In this project, we have developed and applied a new "first principles" or ab initio theoretical approach to calculate the thermal conductivities for a wide range of materials. While computationally demanding, the advantage of this approach is that it requires only basic information about the constituent elements making up the material and has no free parameters. We have used our parameter-free approach to calculate the thermal conductivities of a large number of materials and found good agreement with their experimentally measured thermal conductivity values. Since the calculated results consistently match the measured thermal conductivities, the first principles model shows that it has predictive capability. Using this theoretical approach, we have obtained important results for materials whose thermal conductivities have not yet been measured. For example: 1) We have found that the thermal conductivity of the technologically important material, gallium nitride (GaN), can be significantly increased. Normally, two species of Ga isotopes occur in GaN (60% 69Ga and 40% 71Ga), but by instead using Ga atoms that all have the same isotope (e.g. 100% 69Ga), we find a large increase in thermal conductivity. Such an increase would facilitate heat dissipation in GaN-based high power devices thereby improving device performance and reliability. While isotope purification methods are costly, the potential enhancement in performance may outweigh such costs. 2) We have found an unexpected combination of properties not previously considered as criteria to be used in searching for materials with high thermal conductivity. Including these new criteria admits materials for consideration that would not previously have been thought of as possible high thermal conductivity candidates. We have identified one material, boron arsenide, that should have a remarkably high thermal conductivity, comparable to that of diamond. This finding motivates a search for other possible high thermal conductivity materials using the predictive theoretical approach. More broadly, this activity has contributed to training young researchers in state-of-the-art, predictive methods to describe heat flow in solids as well as aiding in the development of new systems with desired thermal conductivities. This may facilitate technological breakthroughs that could lead to the next generation of materials for thermal management applications.
