Research Projects

The development of new materials having high thermal conductivity (k) is attracting researchers for a long time due to their pivotal role in thermal management techniques, including thermal interface, heat sink, active cooling, power electronics and light-emitting diode (LED) driver design. During operations, LED drivers generate heat which can cause the overheating problem, potentially leading to reduced lifespan or even failure of the LED (10 K increment in temperature causes 50% reduction in lifetime). Low thermal conductivity and heat dissipation rates severely degrade the performance and energy efficiency of electronic and photonic devices. Thus, efficient thermal management is crucial which requires ultra-high thermal conductivity materials. Diamond exhibits the highest k (∼2200 W/m-K) among all naturally occurring materials. However, its application towards thermal management is limited due to its cost. Application of graphite (k~2200 W/m-K) is also limited in the semiconductor industry because of its semi-metallic behaviour. On the other hand, boron arsenide with a semiconducting bandgap and k~1300 W/m-K can be a cost-effective replacement for thermal management applications. In the past few decades, the design and discovery of high-k materials mostly depended upon trial-and-error experiments with random element substitutions employing some chemical intuitions. In the recent years, quantum mechanical calculations with density functional theory come into play. A semi-classical approach of solving the phonon Boltzmann transport equation (BTE) is required to predict lattice thermal conductivity (k_L) of crystalline materials. As this particular approach is computationally expensive, it can be limited to predict k_L for a set of materials. In this project we propose a unified framework, based on semi-classical solution of the Phonon BTE in combination with deep generative machine learning modelling, including active machine learning interatomic potentials (MLIPs) to identify a new class of emerging materials with ultra-high lattice thermal conductivity (k_L), ensuring thermal stability and suitability for thermal management. Structural symmetry properties can play a key role in filtering out low-symmetry materials. Also, it helps to identify materials that can share some common properties with known high k_L materials. The inclusion of some other descriptors, such as group velocity, specific heat and Gruneisen parameters, can also make the model more accurate in the prediction of k_L. For hybrid III-V semiconductors like Boron Arsenide (BAs), 3-phonon scattering processes cannot capture fully the phonon complexity of the material due to large gap between acoustic and optical phonon branches. In this project, based on the outcomes of ML models, k_L of the materials will be further studied with the incorporation of four-phonon scattering processes.