The objective of this project was to develop and then demonstrate the efficacy of a cost-effective approach for a low-defect-density substrate on which aluminum indium gallium nitride (AlInGaN) light-emitting diodes (LEDs) can be fabricated. The efficacy of this GaN-ready substrate would then be tested by growing high-efficiency, long-lifetime InxGa1-xN blue LEDs.
The approach used to meet the project objectives was to start with low-dislocation-density aluminum nitride (AlN) single-crystal substrates and grow graded AlxGa1-xN layers on top.
Pseudomorphic AlxGa1-xN epitaxial layers grown on bulk AlN substrates were used to fabricate LEDs and demonstrate better device performance as a result of the low defect density in these layers when bench-marked against state-of-the-art LEDs fabricated on sapphire substrates. The pseudomorphic LEDs showed excellent output powers compared to similar wavelength devices grown on sapphire substrates, with lifetimes exceeding 10,000 hours (which was the longest time that could reliably be estimated). In addition, high internal quantum efficiencies were demonstrated at high driving current densities even though the external quantum efficiencies were low due to poor photon extraction. Unfortunately, these pseudomorphic LEDs require high aluminum content so they emit in the ultraviolet (UV). Sapphire-based LEDs typically have threading dislocation densities (TDD) > 108 cm-2 while the pseudomorphic LEDs have TDD ≤ 105 cm-2. The resulting TDD, when grading the AlxGa1-xN layer all the way to pure GaN to produce a GaN-ready substrate, has varied between the mid 108 down to the 106 cm-2. An approach to improve the LED structures on AlN substrates for light extraction efficiency was developed by thinning and roughening the substrate.
Overall, the use of pseudomorphic epitaxial growth for the fabrication of LEDs allowed for improved performance of the device. This improvement is due to the considerably lower defect density determined on pseudomorphic layers as opposed to defect density measured on relaxed layers. This attribution is confirmed by comparing the performance of similar emission wavelength devices grown on AlN and sapphire at different drive current density, reaching up to 400 A/cm2. The external quantum efficiency for pseudomorphic LEDs is higher and flatter than that determined for sapphire-based devices with similar composition, and shows a very similar trend to that of commercial InGaN-based light-emitting diodes.
The goal of developing a cost-effective approach to generate low-dislocation density GaN substrates based on AlN was hindered by two drawbacks: Crystal IS has demonstrated high-quality, crack-free, 2-inch AlN substrates, but has not developed a consistent, high-yield product. The second critical barrier is the question of how high the gallium concentration in the AlxGa1-xN buffer layer can be pushed while maintain- ing a low TDD. Higher Ga concentrations will be needed for near-UV or visible blue LEDs. While this work suggests that low TDD are possible in epitaxial layers grown on AlN substrates even with pure GaN, these results have been difficult to reproduce across the large areas needed for cost-effective LED manufacture.
Once these hurdles are surmounted, the use of a platform conducive to lower defect density in the epitaxial layers will be reflected in the performance of LEDs with emission wavelength in the visible, as well as they are already reflected in the performance of UV LEDs.
This work was done by Sandra B. Schujman and Leo J. Schowalter of Crystal IS, Inc.
This Brief includes a Technical Support Package (TSP).
GaN-Ready Aluminum (reference GDM0016) is currently available for download from the TSP library.
Please Login at the top of the page to download.