Machine learning can accelerate the excavation of new LED lighting materials

According to foreign media reports, the university of Houston has announced that its development of machine learning algorithm to predict the characteristics of more than 100000 compounds, and can determine the most likely to become the LED lighting of efficient phosphor compound.

machine learning can speed up the new LED lighting materials excavated

among them, the researchers in the synthesis of a named & other; Sodium borate barium & throughout; Compounds can be calculated, after the tested, found that it provides a 95% efficiency and excellent thermal stability. Although boric acid barium sodium compounds produced by the light is too blue, and shall not apply to commercial, but the researchers are not discouraged. They said they can now by machine learning algorithms can find a useful emission wavelength light-emitting materials.

Jakoah Brgoch professor said: & other; Our goal is to develop a team with efficiency and good color quality and low cost LED bulbs. ”

it is understood that the research project is the first list from the database of the crystalline structure of the Pearson, 118287 kinds of potential inorganic phosphorus compounds. And machine learning fast scanning the key properties of these compounds, including the debye ( Debye） Temperature and chemical compatibility, etc. Finally the algorithm reduces the more than 110000 kinds of compounds in more than 2000.

the researchers say, by the traditional way, need a few weeks to sort out the useful materials; By machine learning algorithms, and within 30 seconds can pick out about 20 kinds of useful material.

Brgoch professor pointed out that the project for the development of high-performance materials for machine learning can bring enormous value provides strong evidence.