Connecting Theory and Practice in Optoelectronics
GaN-LED Simulation Challenge
2015-11-25Posted by on
GaN-based light-emitting diodes (LEDs) are at the heart of modern lighting applications, but their energy efficiency deteriorates with higher current and temperature (see picture). This efficiency droop problem gave rise to intense research worldwide, including various LED modeling and simulation efforts. Different models feature different droop explanations, including Auger recombination, electron leakage, or defect-related mechanisms. But none of the LED models covers all possible mechanisms simultaneously in sufficient detail. The uncertainty of key material parameters also gives room for contradicting results. Authors often apply their model to unique LED designs which may result in unique conclusions. All this undermines the general reputation of numerical LED simulation, I think, and we should put some joint effort into finding more consensus on the GaN-LED efficiency droop.
As a first step, I propose applying different models to exactly the same LED structure, reproducing the same LED measurements, and comparing in detail all model assumptions, parameters, and results. A suitable set of measurements on industry-grade single-quantum well (SQW) devices was recently published in a German Ph.D. thesis. The measured wavelength shift with current allows for the extraction of the SQW polarization charge, which always was one of the main unknowns in numerical GaN-LED simulations. Pulsed-operation data are provided for different ambient temperatures, which also helps narrow down the dominating droop mechanisms. The APSYS simulations shown in the picture are based on Auger recombination being the sole cause of the efficiency droop. Thermionic electron leakage is unable to reproduce all the efficiency and bias characteristics measured in this case. However, other possible mechanisms related to SQW fluctuations, hot carriers, or tunneling are not considered here.
I hereby challenge all groups active in this field to apply their model to this case and to make their results available for comparison, which hopefully leads to a joint publication. Please let me know if you are interested so that I can send you more details.