Efficient modeling of optical spectra of nanocrystal ensembles*

One challenge in building first-principles models of novel nanocrystals (NCs) for optoelectronics applications is correlating single crystal simulations with ensemble NC properties. Current large-scale simulations often predict low energy structures using Bayesian Optimization processes or cluster expansion techniques to reduce the computational burden, but do not attempt to predict optical properties. As a result, simulated optical properties of NCs are limited to a handful of small, “magic-size” clusters that do not lend themselves to models of the vast majority of NC ensembles that exhibit structural polydispersity.

We aim to model the spectral broadening observed in ensemble NC optical spectra due to polydispersity associated with size. We simulate a set of wurtzite-phase CdSe NCs ranging from 1 nm to 3 nm in diameter and parametrize the optical response as a function of crystal size and photon energy. We then apply an energy dependent broadening scheme to produce ensemble NC spectra with various size distributions. We compare these first-principles ensemble spectra to experimentally measured CdSe NCs with a size distribution determined with transmission electron microscopy. We propose such a broadening scheme can improve the interpretability of first principles determined optical spectra of NCs, especially for high-throughput searches for novel material candidates.