Defects can significantly modify the electro-optical characteristics of InGaN light-emitting diodes (LEDs); however, modeling the impact of defects on the electrical characteristics of LEDs is not straightforward. In this paper, we present an extensive investigation and modeling of the impact of defects on the electrical characteristics of InGaN-based LEDs, as a function of the thickness of the quantum well (QW). First, we demonstrate that the density of defects in the active region of III-N LEDs scales with increasing thickness of the InGaN QW. Since device layers with high indium content tend to incorporate more defects, we ascribed this experimental evidence to the increased volume of defects-rich InGaN associated to thicker InGaN layers. Second, we demonstrate that the current–voltage characteristics of the devices are significantly influenced by the presence of defects, especially in the sub turn-on region. Specifically, we show that the electrical characteristics can be effectively modeled in a wide current range (from pA to mA), by considering the existence of trap-assisted tunneling processes. A good correspondence is obtained between the experimental and simulated electrical characteristics ( I – V ), by using—in the simulation—the actual defect concentrations/activation energies extracted from steady-state photocapacitance, instead of generic fitting parameters.

The long-term stability of ultraviolet (UV)-C light-emitting diodes (LEDs) is of major importance for many applications. To improve the understanding in this field, we analyzed the degradation of AlGaN-based UVC LEDs and modeled the variation of electrical characteristics by 2D simulations based on the results of deep-level optical spectroscopy (DLOS). The increase in the forward leakage current observed during ageing was ascribed an increase in trap-assisted tunneling. The analysis of the degradation kinetics suggests the role of a defect diffusion process, possibly involving impurities coming from the p-type layers.

The lifetime of deep-ultraviolet light-emitting diodes (LEDs) is still limited by a number of factors, which are mainly related to semiconductor defects, and still need to be clarified. This paper improves the understanding of UV LED degradation, by presenting an analysis based on combined deep-level transient spectroscopy (C-DLTS), electro-optical characterization, and simulations, carried out before and during a constant current stress test. The original results of this paper are (i) C-DLTS measurements allowed us to identify three traps, two associated with Mg-related defects, also detected in the unaged device, and one related to point defects that were generated by the ageing procedure. (ii) Based on these results and on TCAD simulations, we explain the variation in the forward I–V by the degradation of the p-contact, due to Mg passivation. (iii) On the other hand, optical degradation is ascribed to an increase in defectiveness of the active region and surrounding areas, which led to a decrease in injection efficiency, to an increase in non-radiative recombination, and to an increase in trap-assisted tunneling processes.

We investigate the density of defects and the degradation rate in InGaN light-emitting diodes having identical dislocation density and epitaxial structure, but different indium content in the quantum well (QW; 12%, 16%, 20%). Our results, based on combined steady-state photocapacitance, light-capacitance voltage, and degradation measurements indicate that: (a) the density of defects in the superlattice underlayer is identical for the three wafers, indicating good and reproducible growth conditions; (b) the density of defects within the active region of the devices shows a monotonic dependence on the indium content in the QWs. These results, consistent with previous studies on the topic, prove unequivocally the important role of indium in favoring the incorporation of point defects, further clarifying the possible mechanisms of defect formation, and give a quantitative assessment of the related effect; (c) in step-stress experiments, the degradation rate was found to be much stronger for devices having high indium content in the QW. This result can be explained by considering a decrease in injection efficiency due to the generation or transport of defects, or an increment in defect-assisted Auger recombination terms due to the propagation of defects.

In this paper we investigate the reliability of AlGaN-based UV-C LEDs with an emission wavelength of 265 nm. By submitting the devices to constant current stress, two main electrical degradation processes are identified: a turn-on voltage shift and an increase in the forward leakage current. In particular, these processes were respectively attributed to: (i) a partial passivation of the Mg-doping concentration in the region adjacent to the contact, probably caused by a local hydrogen diffusion, and ii) a diffusion/generation process of defects in the interlayer, responsible for the increase in the trap-assisted tunneling. To validate these hypotheses, we employed TCAD simulations by varying only the Mg-doping concentration in the region adjacent to the p-contact and the defect density in the interlayer. Thus, we correctly reproduced the experimental variation in electrical characteristics, confirming the physical mechanisms identified.

We investigate the degradation mechanisms of In_0.2Ga_0.8N/GaN light emitting diodes through combined experimental analysis and simulations. The devices were submitted to constant current stress at 100 mA. Depending on the measuring current level, two degradation trends were observed: at high test currents (e.g. 200 mA), a monotonic decrease in optical power was observed; at low test currents (e.g. 5 mA), an initial degradation was observed, followed by an improvement in device efficiency (positive ageing). For the first time, such recovery effect was analyzed and modeled, as due to the generation of charged defects at the InGaN/GaN interface, resulting in the increase in the injection efficiency at low bias levels. The role of interface defects was validated by means of numerical simulations, with good agreement with the experimental data.