MIT develops 16-bit processor based on carbon nanotubes
Researchers from MIT and Analog Devices have created a 16-bit processor with 14,000 so-called carbon nanotube field-effect transistors. It is the largest chip based on carbon nanotubes to date.
The researchers have named the chip RV16XNano and built it on the basis of the open source RISC-V architecture. Although it is a fairly basic processor by today’s standards, it is capable of accurately executing a set of instructions. For example, the researchers managed to use a program to send the message ‘Hello, World! I am generating RV16XNano, made from CNTs’.
According to MIT, the science team is building on research that resulted in a 1-bit chip with 178 carbon nanotube field-effect transistors, or cnfets, six years ago. Significant progress has been made in the subsequent period in the field of reducing material and manufacturing defects, among other things.
Metallic contamination causes a small portion of the carbon nanotubes to delay or stop the correct switching of the transistors. Metallic CNTs are always conductive and it’s all about getting semiconductors. For a working chip of any size and complexity, the purity of the carbon nanotubes used must be 99.999999 percent.
Thanks to a set of techniques that the researchers have called Dream, they can make functional chips, even if the purity is at 99.99 percent. Dream stands for designing resiliency against metallic CNTs. The basis of the method lies in simulating different gate combinations to find out which are robust and which are not, due to defects. When designing the chip, they then used a program that automatically deploys only the robust combinations.
In addition, the researchers have improved the manufacturing process. To make cnfets, carbon nanotubes are placed in solution on a wafer containing the transistor architecture. To prevent the tubes from clumping together, the wafer is treated with an agent that promotes adhesion, and the researchers also apply a specific polymer coating, after which the wafer is immersed in a solution. This last step ensures that the polymer, and with it the clumps of tubes, is washed away. The researchers call this step Rinse, removal of incubated nanotubes through selective exfoliation.
Finally, they were able to determine in a controlled manner the properties of transistors that make them n- or p-type transistors. That was one of the challenges with transistors based on carbon nanotubes. The researchers succeeded by applying titanium or platinum, among other things. They call the technique Mixed, which stands for metal interface engineering crossed with electrostatic doping.
The tiny tubes of carbon in a hexagonal grid of chicken wire are light, strong and theoretically cheap to produce, and have good conductive properties. They have therefore been seen for years as a potential successor to silicon as a material for chips, which could be made faster with lower heat production. Researchers encountered many obstacles in actually producing chips, but according to the researchers at MIT, the question is no longer if, but when chips based on carbon nanotubes will appear. Max M. Shulaker of the research team even thinks this could be the case within five years, partly because they use the same manufacturing techniques as the current chip industry.
The researchers are publishing their work in the scientific journal Nature under the title Modern microprocessor built from complementary carbon nanotube transistors.