Origins of photoluminescence decay kinetics in CdTe colloidal quantum dots.

Abstract

Recent experimental studies have identified at least two nonradiative components in the fluorescence decay of solutions of CdTe colloidal quantum dots (CQDs). The lifetimes reported by different groups, however, differed by orders of magnitude, raising the question of whether different types of traps were at play in the different samples and experimental conditions and even whether different types of charge carriers were involved in the different trapping processes. Considering that the use of these nanomaterials in biology, optoelectronics, photonics, and photovoltaics is becoming widespread, such a gap in our understanding of carrier dynamics in these systems needs addressing. This is what we do here. Using the state-of-the-art atomistic semiempirical pseudopotential method, we calculate trapping times and nonradiative population decay curves for different CQD sizes considering up to 268 surface traps. We show that the seemingly discrepant experimental results are consistent with the trapping of the hole at unsaturated Te bonds on the dot surface in the presence of different dielectric environments. In particular, the observed increase in the trapping times following air exposure is attributed to the formation of an oxide shell on the dot surface, which increases the dielectric constant of the dot environment. Two types of traps are identified, depending on whether the unsaturated bond is single (type I) or part of a pair of dangling bonds on the same Te atom (type II). The energy landscape relative to transitions to these traps is found to be markedly different in the two cases. As a consequence, the trapping times associated with the different types of traps exhibit a strikingly contrasting sensitivity to variations in the dot environment. Based on these characteristics, we predict the presence of a sub-nanosecond component in all photoluminescence decay curves of CdTe CQDs in the size range considered here if both trap types are present. The absence of such a component is attributed to the suppression of type I traps.