Preparation and characterization of AlGaN structures for UVA LEDs

Loading...
Thumbnail Image
Files
SinghSM_PhD2024.pdf(5.41 MB)
Full Text E-thesis
Date
2024
Authors
Singh, Sandeep M.
Journal Title
Journal ISSN
Volume Title
Publisher
University College Cork
Published Version
Research Projects
Organizational Units
Journal Issue
Abstract
Visible / UV LEDs are developed from the same III-Nitride compounds semiconductor materials and share many design and technological aspects. However, the efficiency difference between the two types is quite significant, where the latter lags. Recent advancements in material technology such as the development of template/buffer, strain engineering, optimisation in doping techniques and the overall design of a UVLED structure have made it possible to achieve decent performance from UV LEDs. This sufficient performance together with the high demand for mercury-free UV light sources has led to an increase in the market share of UV LEDs, compared to other existing UV light sources. Since a comprehensive study of a UV LED in general is too broad a topic for a PhD thesis, the focus of this work is concerned with two particular research questions i.e.: a) Since AlGaN is routinely used as an electron-blocking layer in visible and UV LEDs, does it act as such and how effective is it? b) How to achieve a low dislocation density template for UV (310-350 nm) LEDs where no native substrates for homoepitaxy are available for low-mid Al content AlGaN? III-Nitride based LEDs generally use AlGaN as an electron blocking layer (EBL) to prevent surplus electron overflow from the active region of an LED into its p-doped region during device operation. The effectiveness of an AlGaN layer as an EBL has been studied in the literature using an unipolar structure (consisting of n-GaN/uid-AlGaN/n-GaN), and those results suggest that it is not as efficient as expected, which contradicts its successful routine use in LEDs. This discrepancy has generated interest in this topic, and so both simulation and experimental analysis of the proposed unipolar structure were carried out in this study. Simulation shows a rectifying nature of the AlGaN barrier, and, in contrast to previously published results, experimental results from this study corroborate the modelling. However, conditions leading to non-rectifying behaviour were also identified. Specifically, non-rectifying behaviour is observed for samples with barriers of Al content less than ~12%, and/or when a growth pause is introduced before and after AlGaN barrier growth. Based on our results, it can be concluded that AlGaN can indeed act as an efficient EBL in LEDs if the contrast in Al composition between EBL and last quantum barriers is at least ~15%. Another important aspect of an LED is a good quality template, which sets the dislocation density in the LED active region and affects significantly achievable internal quantum efficiency and thus the overall device performance. Significant effort in this regard was made to develop low dislocation density crack-free AlGaN buffer layers grown on planar c-plane substrates (PVDNC© AlN/sapphire from Kyma Technologies, Inc.). Another objective of the above efforts was to understand AlGaN growth and the effect of various growth parameters. The key achievement here was the implementation of a composition-graded layer, whose function was to partially compensate dislocation-induced tensile stress in Si-doped AlGaN, that allowed to increase in the total thickness of the buffer before it started cracking. Growth conditions which favour bright UV emission from MQWs despite the moderate crystalline quality of AlGaN thin films were also identified. This work led to the realization of a planar AlGaN template with thick (~2 µm), UV transparent and crack-free n-AlGaN buffer substrate with a total dislocation density of ~4×109 cm-2. As no further improvement in crystal quality was possible using standard planar approaches, to set a path towards an even lower dislocation density template for an efficient UV LED, AlGaN overgrowth on a patterned surface was explored. Regular arrays of hexagonally arranged µ-holes were chosen as a structure for the study (micro honeycomb, µ-HC). As µ-HC is composed of both convex and concave shapes, it makes it a good model structure for understanding surface morphology evolution and the overgrowth mechanisms underpinning it. There are three aspects to this research topic. The first one is about optimizing the µ-pattern itself, and the second relates to finding overgrowth conditions in the multi-variable growth parameter space suitable for the coalescence of µ-patterns back into planar layers of hopefully improved quality. In the third part, the growth conditions were to be optimised to direct surface morphology evolution towards the formation of regular arrays of crystallographically faceted V-pits with almost no c-plane regions. My study suggests that growth parameters influence the morphological evolution of the micro-patterned AlGaN in such a way that both above-mentioned outcomes are possible. Despite most of the work being focused on AlGaN with an Al content of ~20%, the developed approach is believed to be suitable for growing low dislocation density AlGaN across the composition range required to cover the entire UVA spectral range. I think the developed templates can act as a base for improved performance standard UV LEDs and novel-type micro-faceted semi-polar UV LEDs.
Description
Keywords
AlGaN , UV LED , GaN , Unipolar heterostructure , Microhoneycomb , Overgrowth , Template , TCAD , Silvaco
Citation
Singh, S. M. 2024. Preparation and characterization of AlGaN structures for UVA LEDs. PhD Thesis, University College Cork.
Link to publisher’s version