Passive radiative cooling exploits the atmosphere's transparency in the mid-infrared "sky window" (roughly 8–13 μm) to shed heat from a surface directly to deep space without consuming any energy. Finding materials that selectively emit in this window, and integrating them into multilayer stacks that perform under realistic conditions, is a materials-design problem I'm approaching entirely from first principles.

Starting from DFT phonon data (ABINIT DDB output), I build dielectric response models that convert atomistic information into the spectral emissivity needed to predict cooling performance. A generalized multilayer optical solver based on the Transfer Matrix Method — supporting coherent and incoherent stacks, thick films, oblique incidence, and both TE/TM polarizations — handles realistic device geometries. A high-throughput screening pipeline ties these together to scan candidate compounds, and is now being extended with machine learning models trained on simulation outputs to accelerate exploration of the chemical space.

Techniques: DFT phonon calculations (ABINIT), Transfer Matrix Method, dielectric response modeling, high-throughput screening, machine learning.Status: Ongoing.