UCSB blames Auger for droop
The pertinent question regarding the cause of droop – the decline in GaN LED efficiency at high currents – is this: Auger or not Auger? And if it is Auger, then the obvious follow-up question is what is the exact form of this non-radiative recombination process that limits the output power of high-brightness LEDs.
In the fall of 2007, when Philips Lumileds claimed that Auger is the cause of LED droop, this chip maker failed to elaborate on the finer details. But last summer a partnership between theorists at the universities of Arizona and Marburg claimed that droop was not caused by direct Auger recombination, the interaction between an electron, a hole and a third carrier. Instead they postulated that either defect-assisted Auger recombination or phononassisted Auger recombination might be the cause.
Another group has now entered the debate. Computational scientists at the University of California, Santa Barbara (UCSB), believe that droop is caused by a form of direct Auger recombination – interband Auger recombination (figure 1). Its simulations can even explain the origin of the “green gap”, the decline in GaInN LED efficiency as the wavelength is stretched from the blue to the green.
The discrepancy between the results of the academic teams stems from differences in the modelling of the nitride band structure – the UCSB researchers identified and included a second conduction band. “When we identified this second conduction band we were thrilled,” said team member Patrick Rinke. “Then it was just a matter of working through the theory and putting the calculations of the Auger recombination rate on a rigorous footing.”
The UCSB team has not quantified LED loss due to Auger, because this would also require calculations of the radiative recombination rate and possibly other non-radiative recombination rates. “However, we think that it is significant that our Auger coefficient agrees well with that measured by Philips Lumileds,” said UCSB’s Chris Van de Walle.
One of the limitations of these simulations is that they are based on bulk structures rather than quantum wells, which are the emitting layers in LEDs. “Full first-principle calculations for realistic device structures are beyond the capabilities of current computers,” admitted Van de Walle, “but we are looking into other types of modelling.”
Arizona–Marburg team member Joerg Hader is supportive of the UCSB effort, and says that he has no reason to doubt their results. His team did not consider the transitions to higher conduction bands because the experimental results that they are aware of claim that the separation between the lowest and higher conduction bands in GaN is about 3.5 eV. “This means that for our structure, the higher conduction band cannot be reached by Auger transitions.”
Hader believes that there is an approach that can determine whether interband Auger transitions are the cause of droop. This involves the growth of an AlGaN/InGaN structure with AlGaN barriers that are tuned to prevent these transitions. “If the droop disappears, it would be a strong indication that their assumption is correct,” he said. Getting to the bottom of the cause of droop would be beneficial in several ways – it would allow closure on this debate and it would also enable the design of LED architectures that are efficient at high current densities.
