Science

Machine learning system may cheer up OLED blues

Machine learning system may cheer up OLED blues
Finding stable, blue materials for OLED displays has been challenging, but new research has utilized machine learning techniques to sift through millions of candidate molecules
Finding stable, blue materials for OLED displays has been challenging, but new research has utilized machine learning techniques to sift through millions of candidate molecules
View 1 Image
Finding stable, blue materials for OLED displays has been challenging, but new research has utilized machine learning techniques to sift through millions of candidate molecules
1/1
Finding stable, blue materials for OLED displays has been challenging, but new research has utilized machine learning techniques to sift through millions of candidate molecules

The organic light-emitting diodes in OLED TV, smartphone and tablet displays produce different colors using red, green and blue materials. Unfortunately, a lack of the latter – blue materials that are stable and efficient – has been holding the technology back. Now a team of researchers is using a machine-learning system to identify efficient blue OLED molecules, and with the discovery of over a thousand new ones, soon the B in RGB might not be such a problem.

Currently, blue light in OLED has been hard to achieve through organic materials, so developers have created organometallic molecules, which use metals like iridium to boost the molecules through phosphorescence. This workaround is not only expensive, it hasn't yet been able to produce truly stable blues.

To combat this, the team, made up of researchers from Harvard in collaboration with MIT and Samsung, developed the Molecular Space Shuttle, a machine learning process that screens a database of over 1.6 million candidate molecules, and weeds out the best of those to find organic blue molecules that perform at least as well as industry standards.

"People once believed that this family of organic light-emitting molecules was restricted to a small region of molecular space," said Alán Aspuru-Guzik, lead researcher on the project. "But by developing a sophisticated molecular builder, using state-of-the art machine learning, and drawing on the expertise of experimentalists, we discovered a large set of high-performing blue OLED materials."

The neural networks were able to prioritize which molecules to evaluate, and used a quantum chemical calculation to predict their color and brightness – a process which took 12 hours of computation for each molecule.

"Molecules are like athletes," says Aspuru-Guzik. "It's easy to find a runner, it's easy to find a swimmer, it's easy to find a cyclist but it's hard to find all three. Our molecules have to be triathletes. They have to be blue, stable and bright."

Once the algorithms have done the heavy sifting, the human engineers were able to step in and analyze the findings, using a web app to create what they called "baseball cards" (essentially profiles detailing the important information about the most promising molecules). The researchers then voted on which of the 2,500 profiles they felt were most promising, using a tool they dubbed "molecular Tinder."

In the end, the process narrowed the list down to several hundred organic molecules that perform at least as well as existing ones. While the end results may hopefully reduce the cost of OLED displays, the applications don't end there.

"This research is an intermediate stop in a trajectory towards more and more advanced organic molecules that could be used in flow batteries, solar cells, organic lasers, and more," says Aspuru-Guzik. "The future of accelerated molecular design is really, really exciting."

The research was published in the journal Nature Materials.

Source: Harvard

1 comment
1 comment
habakak
'...The future of accelerated molecular design is really, really exciting...' THIS is exciting. Building molecular space shuttles to achieve this is a fantastic idea. The problem in life is too many choices, or where to start. This helps to speed that up if I understand correctly.