Researchers create CMOS-compatible, 30nm programmable graphene transistor
By Sebastian Anthony on February 26, 2013 at 8:36 am 7 Comments
Electronic engineers at Japan’s GNC and AIST research centers have successfully created graphene transistors that are constructed and operated in a way that redefines 50 years of transistor development. These graphene transistors can be built using conventional CMOS processes, and could potentially be many times smaller, hundreds of times faster, and consume much less power than silicon transistors.
The basic design of the transistor has remained virtually unchanged since their discovery at Bell Labs in 1947: You have two terminals (usually a source and a drain), and a gate in the middle. By applying a current to the gate, electricity flows between the source and drain. This design has withstood the most dramatic miniaturization mankind has ever seen — from gate lengths measured in centimeters, down to a few nanometers — but, unfortunately, it doesn’t work with graphene gates. (See: Beyond 22nm: The processes and equipment used to scale silicon down to its theoretical limits.)
Graphene, as you may know by now, is the most awesome material on Earth. It is the strongest material yet discovered, and also the most electrically conductive. Transistors built with graphene could potentially be hundreds or thousands of times faster than their silicon counterparts, and withstand even further miniaturization to gate lengths of just a few nanometers. The problem, though, is that the gate of a transistor must be made out of a semiconductor — and pure graphene is anything but. Which leads us neatly onto this new, fundamentally different transistor from GNC and AIST.
Instead of one gate, the Japanese transistor has two gates (pictured above) — and, to create a transport gap (bandgap), the graphene between the two gates is irradiated with helium ions to introduce crystal defects. By applying a small current to the two gates, the graphene’s bandgap can be manipulated. When the polarity (positive/negative) applied to both gates is the same, the transistor turns on; when the polarities are different, the transistor is turned off. Curiously, by applying different polarities to each of the gates, the graphene can be switched between n- and p-types — in other words, the transistor can actually have its behavior altered between nFET and pFET modes at runtime. A conventional transistor, of course, is set it stone.
Beyond its new design, the most notable aspect of the Japanese transistor is that it’s compatible with conventional CMOS fabrication techniques (lithography, vapor deposition, etc.) As you can see in the image below, the proof-of-concept graphene transistor built by GNC and AIST is absolutely tiny, with a gate width of just 30nm. Compare this with another graphene transistor that we wrote about last year, which was 100,000nm wide. It’s almost unheard of for graphene research groups to begin at such a tiny scale — but that’s the advantage of using hyper-advanced CMOS techniques, rather than starting from scratch.
Moving forward, the research team will attempt to create a whole wafer of graphene transistors (to see if their process is commercially scalable), they want to improve the quality of the graphene that’s being deposited (it’s currently very low quality, due to the helium bombardment), and they also want to investigate the potential of graphene transistors (and thus computer chips) that can be electrically controlled. With these two-gate graphene transistors, we could be talking about ultra-high-speed, ultra-low-power programmable processors — which would be a rather big development in the realm of computing, and a sizable nail in the coffin of silicon chips. |
