Electron emission spectroscopy offers new insights into the cause of droopJ. Iveland et. al. Phys. Rev. Lett. 110 177406 (2013)

A blue LED made by the Taiwanese chipmaker Walsin Lihwa has been used to study the influence of Auger processes on LED droop

Researchersfrom the University of California, Santa Barbara, and École- Polytechnique, France, say that they have conducted an experiment that enables them to unambiguously assign LED droop to an Auger process.

This is by no means the first time that an Auger process – a non-radiative interaction involving three carriers that leads to the promotion of an electron or hole to a higher energy state – has received the blame for LED droop, the decline in device efficiency as the current through the chip is cranked up. However, up until now, the evidence has been circumstantial, cliams the US-French team.

According to Claude Weisbush, who is affiliated to both institutions, one of the biggest weaknesses of circumstantial evidence is that it allows droop to be explained by many competing theories. Along with Auger, this efficiency-sapping malady has been attributed to mechanisms such as electron leakage from the quantum well, poor hole injection and localisation of carriers at defects.

Weisbush argues that the experiment that he and his co-workers have performed changes all of this. It provides a direct measurement of hot electrons, which come from an Auger process involving two electrons and a hole.

“Such hot electrons are difficult to produce in semiconductor structures,” says Weisbuch. “Very high electric fields can generate hot electrons, as can energy barriers that launch hot electrons into the semiconductor, but for the LEDs we have, there are no strong electric fields or sufficiently high energy barriers.” So, he argues, the only possible cause of hot electrons is an Auger process taking place in the LED.

To measure the energy of these electrons, the team performs a very elegant experiment. They place a commercial, conventional LED in a vacuum and bias it at a range of voltages. To measure the hot electrons produced by the device, they add one or two monolayers of caesium to the surface of the p-side of the device, so that electrons passing through the chip can exit it and be detected by a spectrometer. Simultaneous measurements of the power of the light emitted by the LED are also made.

Weishbuch explains that detecting hot electrons is not, in itself, conclusive proof that Auger recombination is the primary cause of droop.  What provides this is that the hot electrons start to appear at exactly the same time that droop kicks in.

At injected currents of 4 mA and higher, two high-energy peaks are observed: one at 0.3-0.4 eV and another at about 2 eV. These energies do not correspond exactly with those of Auger electrons at their initial kinetic energies, because these carriers have to first travel through 200 nm of p-GaN, and there is very fast longitudinal-optical phonon emission in this material.

Other research groups admire the experimental work carried out by this team, but not all are convinced that the data provides unquestionable proof that Auger recombination causes droop.

“The result is definitely a positive contribution to the droop question,” says theorist Weng Chow from Sandia National Laboratories in Albuquerque, NM. According to him, this work provides a good basis for tying up loose ends: “For example – and speaking as a non-expert – I wonder if the authors can extract from the measurement an Auger electron density relative to the low energy electron density? And from that, perhaps, they can estimate an Auger coefficient?”

Meanwhile, Fred Schubert from RensselaerPolytechnic Institute, in Troy, NY, questions whether some of the detected electrons have simply leaked out of the quantum well. Schubert says that Auger recombination is an undisputed effect and it will occurs in LEDs: “But we believe that Auger recombination is not tenable as a major contributor to the efficiency droop.” Four reasons are given to support this claim: the Auger coefficient is too small; the temperature dependence of the Auger process is opposite to the temperature dependence observed for efficiency droop; it’s not possible to fit experimental results to the well-known ABC model, which includes an Auger term; and a more general explanation for droop is needed that accounts for its occurrence in LEDs made from other material systems.

Weisbuch and his colleagues are going to be looking at other material systems, and also other devices. They hope to offer new insights into InP-based telecom lasers, and determine whether the loss mechanism is Auger recombination or intervalence band absorption.

J. Iveland et. al. Phys. Rev. Lett.

110 177406 (2013)