Research shows how LED efficiency is far from optimal

With the UCSB group, led by Chris Van de Walle, Saulius Marcinkevicius discussed his recent experiments on iron-doped gallium nitride during his recent sabbatical at UCSB. “When they developed a theory explaining our experimental results and wrote the paper, they were kind to include me as a co-author,” he says.

At KTH, Marcinkevicius and Professor Sebastian Lourdudoss , a researcher in the Department of Material and Nanophysics, have been studying iron-doped semiconductors on and off since 1997.

Looking forward, Marcinkevicius says he will focus on building on his lab’s enhanced experimental capabilities in ultrafast near-field optical spectroscopy. “Now we are able to map optical properties of materials, among them those of gallium nitride based nanostructures, with unprecedented richness of features. That certainly benefits our collaboration with UCSB.”

“Lighting is a multibillion dollar industry, and each step in the improvement of LED efficiency produces large benefits,” says Saulius Marcinkevicius, Professor of Optics at KTH. “LEDs are very efficient lighting sources. Still, the efficiency is far from optimal, especially at high powers and in the green spectral region.”

Better understanding of the physical mechanisms that affect the LED efficiency could lead to their improvement, Marcinkevicius says. “In this paper, we are discussing one such effect.” The investigation was published in Applied Physics Letters. In LED lighting fixtures, each tiny “bulb”, or LED, contributes light through recombination of electrons and electron holes – the two charge carriers in semiconductors.

When electrons and holes recombine, they basically annihilate each other. The electrons fall into the empty state associated with the hole and what is emitted is the energy difference between the initial and final state of the electron. In the case of a radiative recombination, the emission is in the form of photons – or light.

In an ultraviolet, blue or green LED, this activity takes place inside a nanostructure based on the semiconductor material, gallium nitride (GaN). A bias voltage is applied via electric leads to inject electrons and holes into the nanostructure.

However, if the semiconductor material contains many defects and impurities, such as iron, the resulting recombination may be nonradiative, Marcinkevicius says. "That is, it will produce heat instead of light. The faster this nonradiative recombination, the lower the efficiency of the LED."

The researchers found that unintentional introduction of iron impurities – even at very modest concentrations (1015 cm–3) – leads to a substantial nonradiative recombination, with recombination rates similar to that of the radiative process.

Where does the iron come from? Apparently it may be introduced unintentionally from stainless steel reactors that are used for making LED structures. “Since even miniscule iron concentrations affect the recombination process, iron in LED structures may be a serious culprit in the quest for the ultimate LED efficiency,” he says.

The finding is actually a corroboration of ideas Marcinkevicius had already been working with, and a direct result of his recent sabbatical at UCSB. Assisting him was PhD student, Tomas Uzdavinys.

“Just before this paper, we published an experimental paper on ultrafast electron dynamics in iron-doped gallium nitride, in Journal of Applied Physics. Trying to explain our experimental results, we arrived independently at similar ideas as the UCSB group, which is working with theoretical modelling.”