Moore's Law and our ability (or, rather inability) to continue to shrink our silicon-based processors have engineers scrambling to find ways to move beyond silicon and find an alternative solution. One of the most promising alternatives is the carbon nanotube, a natural semiconductor that has ideal electrical properties and is extremely small.
Unfortunately, carbon nanotubes have proven very difficult to work with—it is hard to manipulate them for correct placement and they are extremely difficult to grow.
That is, until now. MIT's new carbon nanotube processor, whilst very basic and limited, is a major leap forward for this technology that could one day be used to replace traditional silicon-based chips.
About Carbon Nanotubes
Carbon nanotubes can be problematic.
First, they are very small and so it's often difficult to dope them. Also, they are difficult to grow in place and equally as difficult to manipulate and move to the right location. This makes them almost impossible to work with. As a result, when produced at scale, carbon nanotubes have several defects that make them unsuitable for use.
An illustration of the different types of carbon nanotubes. Image obtained from Wikimedia Commons.
MIT and Analog Devices Collaboration
The new processor, a general purpose RISC-V-compliant chip, is the result of a collaboration between scientists at Analog Devices, Inc., and MIT researchers who, together, found a workaround for all the issues related to carbon nanotubes.
Whilst the overall performance of the chip is nothing amazing (and, indeed, not all the researchers' solutions are going to lead to high-performance processors), the processor did manage to execute the classic "Hello world!" programming demo or, rather, a variation of it.
What Did the Researchers Discover?
The primary discovery made by the researchers responsible for the new chip was that specific logical functions are much less sensitive to metallic nanotubes than others. This allowed them to create an open-source RISC design tool to account for this information and theorize a chip design that did not include the gates that are most sensitive to metallic carbon nanotubes.
The resulting chip, the RV16X-NANO, was designed to handle 32-bit-long instructions of the RISC-V architecture with memory addressing limited to 16-bits. All in all, over 14,000 transistors were used for the chip and it demonstrated a 100% yield.
It is important to note that the manufacturing of the RISC-V-compliant processor is still in its early stages and the results they published are the best that they can do… for now. There is a huge amount of potential for this type of chip design. Given that carbon nanotubes can be made into a superconductor, they could be ideal for use in large data centers, for example.
The open-source nature of this processor and its design tools, however, is something that should aid further development of the technology.