Dynamic effects in CdTe quantum-dot LEDs

Abstract

In this work the electrical and electroluminescence properties CdTe nanocrystal films were analysed. The structure consisted of a multilayer of CdTe nanocrystals deposited by the layer-by-layer technique, sandwiched between an ITO anode and an aluminium cathode.

The first part of this work was dedicated to structural and process improvement. Earlier devices, produced through a layer-by-layer (LbL) manual procedure, had an average thickness of 30nm per nanocrystal monolayer, with a roughness near 30% of the overall thickness. Electrical tests showed current densities over 100 mA/cm^2, with a frequent occurrence of pinholes and short-circuits that caused erratic sample behaviour and device rupture. SEM and AFM microscopic analysis showed that the nanoparticles were aggregated in 15-20nm thick clusters bound by polymer. A high porosity and non- uniformity of the multilayer was observed, explaining the formation of short-circuits. Luminescence was obtained from a very small number of samples, and with a very short duration that did not make spectral analyses possible.

A robotic arm was programmed to carry out the LbL deposition, in an attempt to reduce the inhomogeneities of the multilayer. A key factor was the introduction of a special routine to remove the samples from the solutions. The sample withdrawal was designed to be in the vertical at a rate of 1.18 mm/s. The idea was to use gravity and the surface tension of the aqueous solutions to remove all the excess liquid from the surface. Additionally, poly(ethylenimine) (PEI) was eliminated from the process to improve homogeneity. These modifications produced multilayers with a thickness of 3nm per layer, average roughness below 5nm and CdTe packing density of 27%. Electrical measurements showed a stabilisation of the current-voltage (I-V) characteristics. A significant improvement in luminescence occurrence frequency and intensity was also achieved, enabling first spectral analyses.

Once a reliable manufacturing procedure was developed, the electrical characterisation commenced with the analyses of samples with a different number of layers, operated in air. A field dependency of the I-V curves was found. Optimal performance was obtained from 30-layer samples, and this number of layers was adopted for subsequent analyses. Best samples showed external quantum efficiencies of 0.51%, with a photometric response of 0.8 lm/W and peak brightness of 1.42 cd/m^2. However, current and electroluminescence (EL) degradation with voltage and operation time were found in the device.

Single carrier devices revealed a barrier for electron injection higher than predicted by the band diagram of the structure. The presence of an aluminium oxide layer at the multilayer/cathode interface was postulated, and confirmed through experiments in nitrogen. It was proposed that the growth of this oxide layer is the cause of device degradation during operation in air. However, it was demonstrated that the presence of the oxide favoured radiative recombination prior to degradation, with device efficiencies nearly 10 times higher than in devices without the oxide film. This was justified through three effects: charge accumulation at both sides of the oxide, field concentration across the oxide barrier and a reduction in leakage current. Unequal behaviour of samples with different electrode materials revealed that charge injection was the limiting mechanism for current flow, with a current onset field in the range of 2-3x10^7 V/m. Fowler-Nordheim plots showed that field emission was responsible for hole and electron injection into the device. Also, Fowler-Nordheim plots provided evidence of the dynamic nature of the cathode oxidation process. Dielectric breakdown of the aluminium oxide barrier at a rupture field of 3x10^7 V/m was found as a possible triggering mechanism of oxide layer growth. It was found that a critical field around 4.5x10^7 V/m caused irreversible loss of photoluminescence (PL) of the nanocrystals. This loss was attributed to an avalanche effect within the multilayer. The operational range for the devices is then found to be between 2x10^7 V/m and 4.5x10^7 V/m.