Competition between exciton-phonon interaction and defects states in the 3.31 eV band in ZnO

Abstract

We investigate the origin of the band at 3.31 eV (A band) commonly observed in the emission spectra of various ZnO samples. This band is of prime importance for the confirmation of $p$ doping in ZnO nanostructures. We check the validity of the three main hypotheses generally evoked to explain it in undoped ZnO, namely, surface states, the 1LO phonon replica of the free exciton and a defect-related transition. Using ZnO samples structured at different scales, from macro to nano through meso (i.e., a single crystal, a nanoparticles assembly, and a microcrystalline pellet), we demonstrate that a huge surface/volume ratio does not necessarily imply a strong emission at 3.31 eV, especially for model nanoparticles which are uncapped and synthesized in ultrahigh vacuum. Furthermore, we show that the two other hypotheses are valid and can be at stake concomitantly, according to the quality of the samples. Regarding the 1LO phonon replica of the free exciton, its presence is unambiguously established using a complete model based on an exciton population at thermodynamic equilibrium, including the treatment of the interaction of the excitons with the acoustic phonon bath. We observe that the 1LO phonon replica becomes significant at temperatures higher than 80 K typically. Below this temperature, the 3.31 eV emission is only present in the microcrystalline sample and results from a defect-related transition. Since it is not observed in the nanoparticles that are made from the microcrystals, the possibility of an impurity to be the origin of the invoked defect is unlikely in our experiments. Instead, our study suggests that the defect at stake is of crystalline origin, with an activation energy of $122\ifmmode\pm\else\textpm\fi{}5\text{ }\text{meV}$. The related emission is shown to follow a free-to-bound transition mechanism.