“So the solution at that point was simple: We’ll just design circuits with the good combinations of logic gates and avoid using the bad combinations.”ĭREAM, the set of design rules post-doctoral researcher Gage Hills came up with, allows large-scale integration using carbon nanotubes you can purchase off-the-shelf.ģ. Amongst the many logic circuits they’d made, they found a pattern that suggested some combinations were much more susceptible to the noise problem than others. The team found that, by far, the main driver for the needed purity was not the power issue but the noise. “We thought, if we can’t process our way out of this… then somehow we had to design our way around it,” says Shulaker. But what’s needed is 99.999999 percent purity-impossibly far out of reach. The answer his team came up with was “pretty depressing.” The best today’s commercial processes could produce is 99.99 percent semiconducting nanotubes and 0.01 percent metallic. And to his surprise, it hadn’t been answered. “It’s a very basic question,” says Shulaker. But how many metallic nanotubes is too many when you’re trying to build a full-scale processor? Having some metallic nanotubes in a CNT-based logic gate means the circuit will waste power and produce a noisy signal. CNTs have always come in two basic flavors, metallic and semiconducting. While RINSE dealt with one carbon-nanotube impurity, another purity problem nearly crashed the whole project. By first coating the nanotube-covered substrate with a photo resist and then carefully washing it away-under just the right conditions-the process selectively removes the bundles but leaves the individual CNTs.Ģ. RINSE, a solution one of Shulaker’s students, Christian Lau, arrived at, relies on the fact that individual nanotubes are stuck to the substrate by Van Der Waals forces more strongly than bundles are. But for a large-scale integration like for the RV16X-NANO these nanotube-pile-ups would be common enough to mess up the whole processor. When building small-scale test circuits, this was no big deal, Shulaker explains, because even if they killed one circuit, another would work. Most of the nanotubes lie uniformly on the silicon, but every once in a while, they ball up into bundles of a thousand or more. When making CNT transistors, the nanotubes are first put into a solution and spread across a silicon wafer. Two dealt with stubborn issues of carbon nanotube purity and uniformity, and the third allowed for the creation of both n-type and p-type transistors to form complementary logic circuits.ġ. Shulaker’s team, along with engineers at Analog Devices and, later, Skywater Technology Foundry, developed three commercially-viable techniques to create the RV16X-NANO. “Now we know it is possible… and we know it can be done in commercial facilities.” “Ten years ago, we hoped this was possible,” says Shulaker. They reported the achievement this week in Nature. Naturally, the team, led by MIT assistant professor Max Shulaker, tested the chip by running a version of the obligatory “Hello, World!” program. In contrast, the new one, which is based on the open source RISC-V instruction set, is capable of working with 16-bit data and 32-bit instructions. Some of the same researchers created a modest one-bit, 178-transistor processor back in 2013. The Defense Advanced Research Projects Agency is hoping this 3D aspect will lead to commercial carbon nanotube ( CNT) chips with the performance of today’s cutting-edge silicon but without the high design and manufacturing cost. Unlike silicon transistors, nanotube devices can easily be made in multiple layers with dense 3D interconnections. The processor, called RV16X-NANO, is a milestone in the development of beyond-silicon technologies, its inventors say. It’s the most complex integration of carbon nanotube-based CMOS logic so far, with nearly 15,000 transistors, and it was done using technologies that have already been proven to work in a commercial chip-manufacturing facility. Engineers at MIT and Analog Devices have created the first fully-programmable 16-bit carbon nanotube microprocessor.
0 Comments
Leave a Reply. |