Connecting Theory and Practice in Optoelectronics
How to explain the low turn-on bias of blue light emitting diodes?
2015-06-03Posted by on
While the efficiency droop of GaN-based light-emitting diodes (LEDs) receives much attention, another phenomenon is hardly discussed in the literature: the surprisingly low turn-on bias of these diodes. In a recent paper, Hurni et al. report that the injected electrons have a lower energy than the emitted photons. Even advanced simulations of GaN-LEDs usually give a substantially higher turn-on bias, which puzzles the modeling community ever since this phenomenon was first discussed at NUSOD 2009. The picture on the left shows four different current-voltage (IV) curves calculated for the same blue LED structure with an experimental turn-on bias of 2.6V, as measured by Lin et al. Each calculation uses a different carrier transport model. The common drift-diffusion model employed in 2013 by Piprek et al. gives a turn-on bias of 3.2V. More recently, Wu et al. performed a percolation transport study of this device considering random alloy fluctuation. It results in a soft turn-on starting at 2.8V because carriers search along each interface for the lowest energy barrier. Li developed a unique non-local transport model accounting for high-energy (hot) carriers generated by Auger recombination which cause an IV characteristic close to the measurement. However, Kivisaari et al. also considered hot Auger electrons using a Monte-Carlo model and calculated a much higher turn-on bias. Auger recombination is expected to affect device performance only at high quantum well carrier densities, after the diode is turned on. However, I believe that this IV discrepancy is partially related to a different acceptor density inside the 45-nm-thick p-AlGaN electron blocker layer (EBL). Lin et al. did not specify this density, so that the numbers assumed in the LED simulations range from 0.2E18/cm^3 (Kivisaari et al.) to 12E18/cm^3 (Wu et al.). Stronger EBL p-doping is expected to reduce the bias but the actual EBL acceptor density is typically unknown. Anyway, in my view, further research is needed to find a convincing explanation for the low turn-on bias of these fascinating devices.
Update 11/22/16: Further analysis is now published in Optical and Quantum Electronics.