Year after year, the explosive growth of computing power relies on manufacturers’ ability to fit more and more components into the same amount of space on a silicon chip. That progress, however, is now approaching the limits of the laws of physics, and new materials are being explored as potential replacements for the silicon semiconductors long at the heart of the computer industry.
New materials may also enable entirely new paradigms for individual chip components and their overall design. One long-promised advance is the ferroelectric field-effect transistor, or FE-FET. Such devices could switch states rapidly enough to perform computation, but also be able to hold those states without being powered, enabling them to function as long-term memory storage. Serving double duty as both RAM and ROM, FE-FET devices would make chips more space efficient and powerful.
The hurdle for making practical FE-FET devices has always been in manufacturing; the materials that best exhibit the necessary ferroelectric effect aren’t compatible with techniques for mass-producing silicon components due the high temperature requirements of the ferroelectric materials.
Now a team of researchers at the School of Engineering and Applied Science has shown a potential way around this problem. In a pair of recent studies, they have demonstrated that scandium-doped aluminum nitride (AlScN), a material recently discovered to exhibit ferroelectricity, can be used to make FE-FET as well as diode-memristor-type memory devices with commercially viable properties.
The studies were led by Deep Jariwala, assistant professor in electrical and systems engineering (ESE), and Xiwen Liu, a graduate student in his lab. They collaborated with fellow Penn Engineering faculty members Troy Olsson, also an assistant professor in ESE, and Eric Stach, professor in the Department of Materials Science and Engineering and director of the Laboratory for Research on the Structure of Matter.
“Engineers have been pursuing the concept of FE-FET memory since the 60s, since these devices could operate at extremely low powers,” says Jariwala. “The issue really has been to make their fabrication compatible with processors and make them last longer. This is where our 2D materials come in; they are so thin that once a memory bit is written in them, they could preserve that information in the form of charge for years.”
Read more at Penn Engineering Today.