A Purdue-affiliated company is developing a new time and cost effective software technology that could offer a more efficient and realistic way to model and simulate light emitting diodes (LEDs) in order to achieve more powerful and more efficient LED light sources often used in general lighting, automobile lighting and consumer electronics.

Tillmann Kubis, a research assistant professor, and Gerhard Klimeck, a professor, both in Purdue’s School of Electrical and Computer Engineering, in Purdue’s Network for Computational Nanotechnology and Purdue’s Center for Predictive Materials and Devices, along with Junzhe Geng, a graduate student in Klimeck’s nanoelectronic modeling group, co-founded the company LEDcentral LLC to commercialize the technology. LEDcentral’s goal is to improve the design of LEDs efficiency and output power.

“The most efficient LED light bulb on the market right now has a rather dim output, so what we are trying to do is develop a way to have high output power and still achieve high efficiency. Right now that is not possible because of what is called the efficiency droop. It’s not fully understood where the droop comes from or how to solve it, so that’s where our software model comes into play,” Kubis said. “Currently, we’re specifically interested in blue LED lights, which are the basis for white light bulbs. We aim for our models to help industry develop more powerful and more efficient LED technology.”

“Instead of fabricating hundreds of devices, you can simulate thousands of them, and then pick the best 10 you actually care about,” Kubis said. “Additionally, if you have experimental data that you don’t fully understand, we can explain why the behavior is observed. We are able to explain experiments that have happened and predict experiments that have not. Users can run our technology on a local computer and we also have user interfaces so that it’s easy to use for a person who isn’t trained in the software.” Kubis said conventional models may be insufficient and expensive.

“The models existing in industry and academia are based on classical approaches where electrons are considered particles; however, their behavior should be studied on a multiscale paradigm, i.e. atomistic resolution on a micrometer length scale,” he said. “Models on the nanometer scale do a fairly good job, but important LED properties are missing. Additionally, the attempts to model the physics on a quantum level are usually very expensive. The quantum mechanics are captured and the small scale is fully covered, but these attempts are not very efficient.”

Leave a Reply