Recently, the harvard joint samsung, MIT screened more than 1000 used in organic light-emitting diodes (OLED) blu-ray molecule, if the new materials used in OLED, can greatly reduce the production cost, and can improve the performance of the OLED.

The researchers developed a as a result of the computer aided Molecular Shuttle, Molecular Space Shuttle) screening process, the process is a combination of chemical experiment, machine learning, chemical informatics, can quickly find out the new OLED molecules is higher than the industry standard.

Aspuru, head of the team - Guzik said that for a long time, the researchers believe organic light-emitting molecule should come from a range of small molecules, it is hard to find. But by sophisticated molecular builder, cooperate with the most advanced machine learning technology at the same time, they found that the blue OLED materials with high performance.

According to the project researchers, to reduce the production costs of OLED, its biggest challenge is blue light.

OLED and LCD also need green, red, blue three colors of sub pixel combination to accomplish all the colors of the screen, but the existing problem is that organic molecules is difficult to effectively blue light. To enhance blue light intensity, OLED manufacturers to use some transition metal (such as iridium) with phosphorescence (phosphorescence) enhance molecules, thereby producing a kind of organic metal elements. But the high cost of the solution, and does not guarantee the stability of the blue.

Hence Aspuru - Guzik and his team want to fully use organic molecules replace these organometallic compounds. First of all, the researchers established a molecular library, more than 1.6 million candidate molecules to narrow range, Ryan Adams, assistant professor of computer science at Harvard University led the research team developed a new type of machine learning algorithms to predict which molecular can produce good results, and virtual test priority of these molecules. In this way, will search for calculate the cost more than ten times. They concluded that the combination of chemical and machine learning make it easier for researchers to predict molecular color and brightness.

Adams notes that has been used in a large number of machine learning science, machine learning and the home of interdisciplinary collaboration not only help to promote the development of computer science, but also can develop more new materials.

In order to more accurately identify these super molecules, the researchers combined with theoretical model and experimental practice to set up a web application, let the cooperation study of more than 500000 quantum chemical simulation results. This phase is completed, the team got hundreds of performance and current best without metal OLED is molecules.

From the current view, these molecules can't directly to the commercial, but with the development of the technology, OLED display to change for the future is likely to play a very big driving force.