Predictive Design and Experimental Realization of InAs/GaAs Superlattices with Tailored Thermal Conductivity

We demonstrate an ab initio predictive approach to computing the thermal conductivity (?) of InAs/GaAs superlattices (SLs) of varying period, thickness, and composition. Our new experimental results illustrate how this method can yield good agreement with experiment when realistic composition profiles are used as inputs for the theoretical model. Because of intrinsic limitations to the InAs thickness than can be grown, bulk-like SLs show limited sensitivity to the details of their composition profile, but the situation changes significantly when finite-thickness effects are considered. If In segregation could be minimized during the growth process, SLs with significantly higher ? than that of the random alloy with the same composition would be obtained, with the potential to improve heat dissipation in InAs/GaAs-based devices.