IBM has made three breakthroughs that could help chips continue following Moore’s Law. Monday at the IEEE International Electron Devices Meeting, IBM showed off the first racetrack memory device that delivers an entirely new means to get electrons to hold data, as well as two materials breakthroughs that could lead to faster chips and even open up new spectrum bands that would be useful for delivering mobile broadband.
Racetrack memory: IBM’s racetrack memory offers the ability to store massive quantities of information like hard drives do but has no moving parts like solid state drives do, so it’s faster. It’s called a racetrack because it pushes electrons around a wire kind of like a car goes around a racetrack. After years of research, IBM said Monday it can make such devices.
This still isn’t mass production, and the big challenge here is making it power-efficient (driving those electronics through the racetrack requires a big current), but it could provide a means to a new type of faster computing. As IBM said in its release, “This breakthrough could lead to a new type of data-centric computing that allows massive amounts of stored information to be accessed in less than a billionth of a second.”
Here is a video I did back in 2010 at IBM’s Spintronics lab that explains racetrack memory and how it relates to storing more data that can be read faster.
Graphene: IBM also made two materials breakthroughs, beginning with a way to build chips using a carbon-based material called graphene. Graphene could offer better wireless chips because it could allow chips to deliver data over higher frequency bands, and also could lead to long-lasting batteries and breakthroughs in clean energy. Taking advantage of higher frequencies means we could use more of the airwaves and help assuage our growing demand for wireless broadband.
The challenge with graphene is figuring out how to use it in today’s chipmaking fabrication plants, to avoid the multi-billion-dollar costs associated with building such a plant solely to produce graphene chips. IBM said today that it had solved this problem– building a graphene-based chip that is compatible with conventional (CMOS) chipmaking technologies.
Carbon nanotubes: The carbon nanotube announcement is a bit more out there, with researchers demonstrating the first transistor with channel lengths that are smaller than 10 nanometers built using carbon nanotubes. The channel length refers to how deep the lines on a chip are etched, and the goal is to make those smaller and smaller in order to fit more chips on a wafer and continue pushing Moore’s Law forward.
But as transistor channel lengths shrink, conventional chipmaking technologies are running into a variety of problems, which is why Intel made such a big deal of its 3-D transistors earlier this year. However, it’s not clear if that advance will continue below 11 nanometers. Of course, as researchers seek ever smaller chips, some are abandoning the idea of the transistor completely with technologies such as DNA computing, Quantum computing and even brain-like computers. For the time being, IBM’s carbon nanotubes are still in that commercialization phase along with these other efforts, so don’t break out the bubbly just yet.