Researchers build a RISC-V chip from carbon nanotubes

MIT researchers took the first modest – but real – steps towards the next generation of processors by building the first RISC-V compatible processor powered by carbon nanotubes. The processor can handle 32-bit instructions and 16-bit memory addressing, and the team was able to run a Hello World! demo. It's far from what silicon chips can do, of course, but it's an important step.

Although chip makers have not finished pulling all the power of silicon, the life of silicon processors is coming to an end. Despite what the logic of Ant-Man might suggest, we can only miniaturize things. That means we have to find something else to replace silicon. Researchers have long considered carbon nanotubes as a potential new material for processors, but this possibility is as rich in pitfalls as it is full of potential.

Carbon nanotubes are natural semiconductors and they are extremely small, making them an ideal option for later miniaturization processors. However, they must be grown and it has been difficult to cultivate them accurately. For this to work, 100% purity is required, but the researchers exceeded 99.9%. It seems closer than it is; Such applications require absolute precision.

In addition, silicon semiconductors can be modified to have positive and negative biases through a process called "doping", but the microscopic size of carbon nanotubes makes the task incredibly difficult.

Bypass traps of carbon nanotubes

Researchers at MIT, in collaboration with scientists at Analog Devices, Inc., have found workarounds to these barriers.

First, the researchers let the nanotubes grow as they pleased. It's a somewhat chaotic process. In addition to useful carbon nanotubes, metallic nanotubes also grow, as do clusters of nanotubes. After allowing the forest of nanotubes to grow, the team applied a layer on the surface that was then removed by sonication. The layer took the aggregated groups with it, leaving the others in place.

Next, the team etched most of the nanotube layer to retain only those they wanted, and then applied an oxide layer as a doping agent. This is not as accurate as the process used for silicon processors, but it worked for this application.

Finally, the team worked on metal nanotubes that are not useful for a processor by modifying the RISC design tool. The team found that some functions were less sensitive to the presence of metal nanotubes and had modified the open-source design tool to account for it, avoiding functions sensitive to these metal nanotubes.

The end result is the RV16X-NANO chip. This chip has 14,000 transistors, which is far from the billions of processors of the current generation, but with this important performance of 100%. The team was then able to execute this Hello World! demo.

Not quite the answer

According to the researchers, there are a multitude of ways to improve the existing design. The design must, however, tolerate imperfections and Processors are the exact opposite of that. The chip must take into account carbon nanotubes and misdirected metallic nanotubes.

In the end, researchers want to get to a place where they can sufficiently control the growth of carbon nanotubes to make single nanotube transistors. This experience consists of working around this. This is not a solution. But this proves that it is possible to manufacture carbon nanotube chips and enhances the overall understanding of how carbon nanotubes might work in a processing environment. These are two steps in the right direction, even if they are not steps forward.