Technical Insight
Research Review: Standard model fails to fully account for LED droop
Plotting LED efficiency against carrier density exposes fundamental limitations in the widely used ‘ABC’ model
A TEAM led by Fred Schubert’s group at Rensselaer Polytechnic Institute have uncovered a fundamental flaw in the model used to describe carrier behaviour in an LED. This model – often referred to as the ABC model – has been widely used in discussions about the origin of droop, the decline in nitride LED efficiency at high current densities.
The ABC model states that carriers undergo one of three processes: Shockley-Reed Hall recombination, a non-radiative process that is proportional to the carrier density; radiative recombination, which is proportional to the square of the carrier density; or Auger recombination, a nonradiative recombination process that depends on the cube of the carrier density and has been claimed by several groups to be the primary cause of droop.
One of the predictions made by the ABC model is that any plot of LED efficiency as a function of carrier density (plotted on a log scale) will be symmetric.
“Real data from us and others is not symmetric,” says Schubert. “For this reason, we feel that the ABC model is insufficient for describing droop.”
The team, which includes researchers at Sandia National Labs and Samsung LED, have measured the efficiency of two different multi-quantum well devices: a commercial LED emitting at 463 nm and a 444 nm emitter.
Direct measurements of carrier density are not possible, so the researchers recorded the light output power instead. The squareroot of this is proportional to the carrier density. Plots of LED efficiency as a function of the square root of light output power show that the ABC model could not account for the behaviour of real LEDs. If a good fit is obtained for the efficiency curve at carrier densities below that of the peak LED efficiency, then the model fails to keep pace with the decline in efficiency at higher carrier densities.
The team has also studied how predictions of the ABC model vary when it accounts for phase-space filling – the filling of the conduction and valence bands in an unequal manner, due to the different masses of electrons and holes. Catering for this effect still leads to even symmetry in plots of internal quantum efficiency as a function of carrier density. Schubert says that these experiments show that at high current densities there is a significant loss process that depends on the fourth power of the carrier density. “The fourth order term suggests an active process that goes beyond the three conventional processes [non-radiative recombination, radiative recombination and Auger recombination] and could be carrier leakage.”
Q. Dai et al. (2011) Appl. Phys. Lett. 98033506
