Experimental Demonstration of Inverse-Designed Nanostructures with Ultra-Low Thermal Conductivity

Tailoring heat transport at the nanoscales is crucial to several applications, such as thermal energy harvesting and heat management. First-principles calculations and mesoscales models, such as the Boltzmann transport equation (BTE), have provided tremendous help in understanding the effect of material composition on thermal transport. However, materials engineering in this space is mostly done with trial. In this talk, we will report numerical optimization and experimental demonstration of nanopatterned Si membranes with ultra-low thermal conductivity. Shape optimization is performed by density-based topology optimization, a technique commonly employed for elastic materials, and the adjoint mode-resolved BTE [1]. The fabrication of the optimized structures is carried out with Focused Ion Beam, and the effective thermal conductivity is measured with the thermal transient grating technique. Manufacturing constraints, such as minimum length scales, are taken into account by adding suitable constraints to the optimization algorithm. With a patterning resolution of about 70 nm and a film thickness of 50 nm, we record a thermal conductivity as low as 1.7 W/mK. Exploiting a deep synergy between theory and experiments, our work pushes the limits of low-thermal-conductivity materials, opening up novel routes for high-efficiency thermoelectric materials.<br/>[1]G. Romano and S.G. Johnson. Structural and Multidisciplinary Optimization 65, 297