Theory of radiative and nonradiative recombination processes in nitride-based heterostructures

Loading...
Thumbnail Image
Files
McMahonJM_PhD2024.pdf(11.77 MB)
Full Text E-thesis
Date
2023
Authors
McMahon, Joshua M.
Journal Title
Journal ISSN
Volume Title
Publisher
University College Cork
Published Version
Research Projects
Organizational Units
Journal Issue
Abstract
Most modern blue-violet short wavelength visible light-emitting diodes (LEDs) incorporate group III-nitride (III-N) semiconductor quantum wells (QWs), and ultraviolet (UV) LEDs could be developed using similar materials and heterostructures. Relative to filament and fluorescent bulb technologies, modern blue-violet short wavelength emitting III-N QW-based devices are significantly more efficient. However, they suffer from efficiency ``droop'' effects, the fundamental causes of which are under debate. For example, the efficiency of III-N LEDs drops with increasing drive current, and thus increased carrier density in the well, an effect known as ``current droop''. Furthermore, even at fixed drive current, when the temperature of the device rises, there is a corresponding drop in device efficiency, the so-called ``thermal droop''. On top of current and temperature dependent droop, III-N QW-based light emitting devices also suffer significant reductions in efficiency at longer (green to red) and shorter (deep UV) wavelengths. The reduction in efficiency as III-N devices are engineered to emit at longer wavelengths is a major contributor to the so-called ``green gap'' phenomenon, wherein neither III-N nor other III-V heterostructure-based devices (such as those based on phosphides or arsenides emitting in the red to infrared spectral range) can be designed to emit efficiently in the green to yellow spectral range. Input from theory is important for understanding the fundamental origins of these droop phenomena, as well as to guide device design to improve the efficiency of not only blue-violet visible wavelength emitters that suffer from droop, but also devices emitting at the shorter and longer wavelengths mentioned above. However, theoretical results obtained for other semiconductor structures (such as those based on II-VI and other III-V heterostructures) cannot be carried over to III-N systems due to significant differences in the fundamental material properties. For instance, alloy disorder causes strong carrier localisation in III-N semiconductor alloys, which can significantly alter the electronic and optical properties of III-N heterostructures. Although experimental data indicates the importance of such effects, only very recently have they been accounted for in theoretical studies, as they present a significant modelling challenge. The aim of this thesis is to address this challenge using atomistic modelling. Auger recombination has been discussed in the literature as an important nonradiative process in c-plane InGaN/GaN QWs in which an electron recombines with a hole, but instead of a photon being emitted (as in radiative recombination) another carrier is excited. If the rate of Auger recombination grows faster than the rate of radiative recombination as temperature or carrier density increases, overall, there will be a negative impact on device performance. For instance, there is much evidence that Auger recombination plays a significant role in carrier density dependent droop, but the exact nature of the Auger process underlying this droop phenomenon is still under debate, and several Auger recombination mechanisms have been suggested, including a defect-assisted process as well as alloy disorder enhanced Auger recombination. Auger recombination has also been explored as a possible cause of thermal droop but there is still much debate over its relevance to this droop phenomenon. Despite experimentally established links between Auger recombination and these droop phenomena, on the theoretical side the impact of alloy disorder on the Auger recombination process in III-N QW systems is widely unexplored. Our theoretical framework is based on a nearest neighbour sp3 tight-binding model, which takes input from a valence force field model to determine the equilibrium lattice positions of the alloy disordered QW structures studied. On top of this, local variations in strain and polarisation fields are accounted for in the framework, along with polarisation field screening effects at high carrier densities in the well. The tight-binding energies and wave functions are then used to calculate the radiative and Auger recombination rates. For InGaN/GaN QW systems, our predicted values for radiative and Auger recombination rates lie within the wide range of experimentally reported values, and confirm that the coefficient of Auger recombination is not as small as one may expect from the expected dependence of Auger coefficient on band gap. To study the thermal droop, we have evaluated the radiative and Auger recombination rates at a fixed carrier density but as a function of temperature. Our results reflect the unconventional but experimentally observed increase of the radiative recombination rate with increasing temperature. When focusing on the competition between radiative and alloy-enhanced Auger effects, neglecting, e.g., defect-related processes such as Shockley-Read-Hall recombination, our results indicate an improvement of device performance with increasing temperature, in contrast to experimentally determined efficiency data. Thus, we expect that alloy-enhanced Auger recombination, intrinsic to InGaN-based QWs, is not responsible for thermal droop. As a result, efficiency improvement strategies that target, for instance, factors extrinsic to the well, such as reducing defect densities, should be considered. Turning to the carrier density dependent droop (i.e. the current droop), the competition between radiative and Auger rates as a function of carrier density (but at fixed temperatures) was determined. In InGaN/GaN QW systems, already at low carrier densities our model predicts Auger recombination rates large enough to be considered significant contributors to this droop phenomenon (based on expectations from the literature). When investigating the carrier density dependence of the recombination rates, we find that the Auger rate grows faster than the radiative rate as carrier density increases in the c-plane InGaN/GaN QWs studied here. Working within a commonly used model of efficiency in LEDs, we compared the theoretically determined carrier density dependent efficiency data to experimental results from collaborators at the Universities of Manchester and Cambridge for samples with varying defect densities. We found a good agreement between theory and experiment at carrier densities where this droop phenomenon is observed; the temperature is kept constant in the experiments and calculations. Furthermore, the drop in efficiency with increasing carrier density was essentially independent of sample, and thus independent of defect density. Overall, these results indicate that defect-assisted Auger processes may be of secondary importance, and alloy-enhanced Auger recombination can be sufficient to explain the current droop. Moreover, despite the large Auger coefficients, the green emitting samples studied in our theory experiment comparison were found to have relatively high internal quantum efficiencies, suggesting that Auger recombination may not be limiting the performance of longer wavelength visible light-emitting devices. Lastly, UV LEDs based on AlGaN QW systems have attracted significant attention in recent years for the potential development of more efficient and environmentally friendly UV emitters. Despite this, and despite the experimental observation of both thermal and current droop in AlGaN-based QWs, Auger recombination and carrier localisation have also been widely unexplored in these systems, particularly from a theoretical perspective. Thus, we applied our theoretical framework to determine the radiative and Auger recombination rates at a fixed carrier density but as a function of temperature in c-plane AlGaN/AlN QWs, thus providing insight into the thermal droop of such emitters. We found Auger recombination rates on the same order of magnitude as those determined for the InGaN/GaN QWs studied above. Based on our results, we expect that Auger recombination is not the driving cause of thermal droop in AlGaN/AlN QWs. However, based on expectations from the literature, the values calculated here suggest that Auger recombination could be a significant contributor to current droop in AlGaN QWs. Overall, our calculations show that alloy-enhanced Auger recombination is indeed strongly contributing to nonradiative recombination processes in III-N heterostructures. However, the thermal droop does not seem to be driven by this intrinsic nonradiative process. Instead, other factors, such as defect-assisted Auger recombination or carrier injection deficiencies, may be responsible. Thus, device optimisation strategies that target these extrinsic effects may still improve efficiency. With regard to the current density dependent droop, our calculations indicate that alloy-enhanced Auger recombination plays a significant role. Efforts to mitigate the impact of alloy-enhanced Auger recombination on the internal quantum efficiency could consider targeting larger active regions to reduce current density, or improved current spreading through the multiple QWs employed in LEDs. Although alloy-enhanced Auger recombination may present an intrinsic roadblock to reducing current droop, our results suggest that the reduction in efficiency of longer wavelength III-N devices emitting in the visible range (i.e. green) is not driven by this Auger process. Thus, device performance optimisation may still be achievable by targeting extrinsic factors such as carrier injection efficiency and homogeneity.
Description
Keywords
Nitrides , LEDs , Quantum wells , Auger recombination , Atomistic tight-binding , Carrier localisation
Citation
McMahon, J. M. 2023. Theory of radiative and nonradiative recombination processes in nitride-based heterostructures. PhD Thesis, University College Cork.
Link to publisher’s version