A process that combines neural network analysis with a human rating system likened by its creators to online dating has helped Harvard University researchers identify more than a thousand molecules they believe could dramatically improve organic light-emitting diode displays for televisions, phones and tablets.
OLED screens, which don’t need a backlight and offer dramatically improved colour contrast and energy consumption, are already replacing LCDs in high-end consumer devices. The main challenge to manufacturing affordable versions for large displays like televisions though is finding molecules capable of emitting blue light in a stable and efficient way.
The Harvard solution is a large-scale, computer-driven screening process dubbed ‘the Molecular Space Shuttle’ that incorporates theoretical and experimental chemistry, machine learning and cheminformatics.
It starts by using machine-learning algorithms to predict the best candidates to replace expensive organometallic systems that OLED producers have created using transition metals like iridium from an initial library of more than 1.6 million entirely organic alternatives.
It’s estimated that prioritising molecules for subsequent virtual testing in this way reduces the computational cost of the search by at least a factor of ten.
"We were able to model these molecules in a way that was really predictive," said researcher Rafael Gómez-Bombarelli. "We could predict the colour and the brightness of the molecules from a simple quantum chemical calculation and about 12 hours of computing per molecule. We were charting chemical space and finding the frontier of what a molecule can do by running virtual experiments."
Finding suitable molecules required human intuition as well as sheer computing power. Every month, Gómez-Bombarelli and coauthor Jorge Aguilera-Iparraguirre created ‘baseball card’ profiles containing important information about each of the most promising molecules. Collaborators at Samsung and MIT then voted on which they considered most suitable for further investigation using a voting tool nicknamed ‘molecular Tinder’ after the online dating app.
This accelerated design cycle left the team with hundreds of molecules they say perform as well as, if not better than, state-of-the-art metal-free OLEDs.
The combination of a sophisticated molecular builder, machine learning and human expertise confirmed that the family of high-performing blue OLED materials is much larger than was previously believed, said Alán Aspuru-Guzik, the Harvard professor of chemistry and chemical biology who led the project.
"Molecules are like athletes," he explained. "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."
Aspuru-Guzik believes potential applications of the technique extend far beyond OLEDs: "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."