The increase in computing performance in recent decades has been achieved by squeezing more transistors than ever into a tighter space on microchips.
But this downsizing means packing the wiring within microprocessors ever more tightly together, leading to effects like signal leakage between components. This can slow down communication between different parts of the chip. This delay, called “interconnect bottleneck,” is becoming an increasing problem in high-speed computing systems.
One way to tackle the interconnect bottleneck is to use light rather than wires to communicate between different parts of a microchip. This is not an easy task because silicon, the material used to build chips doesn’t emit light easily, according to MIT associate professor of physics Pablo Jarillo-Herrero.
Researchers have created a light emitter and detector that can be integrated into silicon CMOS chips.
The device is built from a semiconductor called molybdenum ditelluride. The ultrathin semiconductor belongs to an emerging group of materials known as 2D transition-metal dichalcogenides.
Unlike the conventional semiconductors, this material can be stacked on top of silicon wafers.
"Researchers have been trying to find materials that are compatible with silicon, in order to bring optoelectronics and optical communication on-chip, but so far this has proven very difficult," Jarillo-Herrero said. "For example, gallium arsenide is very good for optics, but it cannot be grown on silicon very easily because the two semiconductors are incompatible."
In contrast, the 2D molybdenum ditelluride can be mechanically attached to any material.
Another difficulty with integrating other semiconductors with silicon is the materials typically emit light in the visible range. But the light at these wavelengths is simply absorbed by silicon.
Molybdenum ditelluride emits light in the infrared range, which is not absorbed by silicon. This means it can be used for on-chip communication.
In order to use the material as a light emitter, the researchers had to convert it into a P-N junction diode, a device in which one side, the P side, is positively charged, while the other side, the N side, is negatively charged.
In conventional semiconductors, this is usually done by introducing chemical impurities into the material. With the new class of 2D materials, it can be done by applying the voltage across metallic gate electrodes placed side-by-side on top of the material.
"That is a significant breakthrough because it means we do not need to introduce chemical impurities into the material [to create the diode]. We can do it electrically," Jarillo-Herrero said.
When the diode is produced, the researchers can run a current through the device, which causes it to emit light.
"So by using diodes made of molybdenum ditelluride, we are able to fabricate light-emitting diodes (LEDs) compatible with silicon chips," Jarillo-Herrero said.
The device can be switched to operate as a photodetector, by reversing the polarity of the voltage applied to the device. This causes it to stop conducting electricity until a light shines on it when the current restarts.
This makes it so the devices are able to transmit and receive optical signals.
The researchers are currently investigating other materials that could be used for on-chip optical communication.
For example, most telecommunication systems operate using light with a wavelength of 1.3 or 1.5 micrometers.
But molybdenum ditelluride emits light at 1.1 micrometers. This makes it suitable for use in silicon chips that were in computers, but unsuitable for telecommunications systems.
"It would be highly desirable if we could develop a similar material, which could emit and detect light at 1.3 or 1.5 micrometers in wavelength, where telecommunication through optical fiber operates," said Jarillo-Herrero.
The researchers are continuing to explore another ultrathin material called black phosphorus, which can be tuned to emit light at different wavelengths by altering the number of layers used. They hope to develop devices with the necessary number of layers to allow them to emit light at the two wavelengths while remaining compatible with silicon.
"The hope is that if we are able to communicate on-chip via optical signals instead of electronic signals, we will be able to do so more quickly, and while consuming less power," Jarillo-Herrero said.
A paper on this research was published in Nature Nanotechnology.