AI design of crystalline material characteristics?

Semiconductors and crystal materials can expect dramatic changes in characteristics, which can be achieved with small distortion and transmit light and heat. A team of researchers from MIT University, Russia and Singapore will explore ways to use artificial intelligence to predict and control these changes and use them in advanced materials research for future advanced equipment.

The team investigated the effects of strain on the band gap of silicon and diamond, and the energy region where there is no electronic state between the energy bands representing the electron state of the crystalline material. A band gap refers to a portion in which electrons do not exist in a band-like structure that is classified by an energy level in the crystal. In the case of semiconductors having a narrow portion, a current does not flow normally, but the semiconductor exhibits characteristics such that electrons can pass through by applying a specific voltage to the material.

By studying the neural network algorithm, the researchers have made it possible to accurately predict how the amount and direction of the distortion of the crystalline material affects the trenches without human knowledge or guesswork. Analysis using this technology shows that solar cells using silicon crystals can produce the same energy as a thousandths of a millimeter, or turn diamonds into practical semiconductor materials that replace silicon. .

The researchers are currently conducting studies focusing on electrical properties, but they stress that they can change optical and thermal properties. But it remains a technical challenge how to add the necessary distortion while maintaining the complexity of configuring a semiconductor chip. In the future, for example, it might be possible to build a smart-phone SoC-level chip with a diamond-based high-speed chip or an electric vehicle that can run on solar cells alone on the roof. It is a research that is only on the starting line yet, but it shows the possibility that distortions can help to improve solar energy utilization and computing performance. For more information, please click here .