Medical applications, drone vision, autonomous vehicle vision. A credit card-sized device concentrates terahertz energy to generate high-resolution images. This advancement could make it possible to create real-time imaging devices that are smaller, cheaper and more robust than other systems.
This discovery paves the way for high-resolution, real-time imaging devices that are one-hundredth the size of other radar systems and more robust than other optical systems. Terahertz waves, located on the electromagnetic spectrum between microwaves and infrared light, exist in a “no man’s land” where neither conventional electronics nor optical devices can effectively manipulate their energy. But these high-frequency radio waves have many unique properties, such as the ability to pass through certain solid materials without the health effects of X-rays. They can also enable faster communications or vision systems that can see through foggy or dusty environments.
An ultra-precise antenna array
MIT’s Terahertz Integrated Electronics Group, led by Associate Professor Ruonan Han, seeks to bridge what is known as the terahertz divide. These researchers have just demonstrated the most precise and electronically steerable terahertz antenna array, which contains the largest number of antennas. The antenna array, called a “reflector”, works like a controllable mirror whose direction of reflection is guided by a computer. The antenna array, which packs nearly 10,000 antennas into a device the size of a credit card, can precisely focus a beam of terahertz energy onto a tiny area and quickly control it with no moving parts. Built using innovative semiconductor chips and manufacturing techniques, the reflector array is also scalable.
Performance close to LIDAR
The researchers demonstrated the device by generating in-depth 3D images of scenes. The images are similar to those generated by a LiDAR (light detection and ranging) device, but because the reflector array uses terahertz waves instead of light, it can work effectively in rain, fog or snow. This small reflector was also able to generate radar images with twice the angular resolution of that produced by a Cape Cod radar, a building so large it is visible from space. While Cape Code’s radar is capable of covering a much wider area, the new reflector is the first to bring military-grade resolution to a device for commercial smart machines. “Antenna arrays are very interesting because, by simply changing the feed delays of each antenna, you can change the direction in which the energy is concentrated, and this in a completely electronic way”, explains Nathan Monroe (13 years old , 17), first author of the paper, who recently earned his doctorate in the Department of Electrical and Computer Engineering (EECS) at MIT. “So it’s an alternative to those big radar antennas you see at the airport that move with motors. We can do the same thing, but we don’t need moving parts because we just change a few bits in a computer.” Co-authors are EECS graduate student Xibi Chen, Georgios Dogiamis, Robert Stingel, and Preston Myers of Intel Corporation, and Han, lead author of the paper. The research work is presented at the International Solid-State Circuit Conference.
Inventive manufacturing techniques
With conventional antenna arrays, each antenna generates its own radio wave power internally, which not only wastes a lot of energy, but also creates complexity and signal distribution issues that previously prevented these arrays from s adapt to the number of antennas required. Instead, the researchers built a reflective array that uses a primary energy source to send terahertz waves to the antennas, which then reflect the energy in a direction controlled by the researchers (like a rooftop satellite dish ). After receiving the energy, each antenna performs a delay before reflecting it, which concentrates the beam in a specific direction. Phase shifters that control this delay typically consume a large portion of the radio wave’s energy, sometimes as much as 90 percent, Monroe says. They designed a new phase shifter which consists of only two transistors and which therefore consumes half the energy. Also, typical phase shifters require an external power source such as a power supply or battery to operate, which creates power consumption and heating issues. The new design of the phase shifter does not consume any energy. Another problem is the orientation of the energy beam: computing and reporting enough bits to control 10,000 antennas at once would significantly slow down the performance of the reflector array. The researchers avoided this problem by integrating the antenna array directly on computer chips. Because the phase shifters are so small (only two transistors), they were able to reserve about 99% of the chip space. They use this extra space as memory, so each antenna can store a library of different phases. “Rather than telling this antenna array in real time which of 10,000 antennas should point a beam in a certain direction, just tell it once and it remembers. Then just dial the number phone to pull the page from its library. We later discovered that this allowed us to consider using this memory to implement algorithms, which could further improve the performance of the antenna array “, explains Mr. Monroe. To achieve the desired performance, the researchers needed about 10,000 antennas (more antennas allow them to direct energy more precisely), but building a computer chip large enough to contain all these antennas is a huge challenge in itself. So the researchers took an evolutionary approach, building a single, tiny chip with 49 antennae and designed to communicate with copies of itself. They then stacked the chips into a 14×14 array and stitched them together with microscopic gold wires that can communicate signals and power the array of chips, Monroe says.
The team worked with Intel to fabricate the chips and help with die assembly.
“Intel’s advanced high-reliability assembly capabilities, combined with state-of-the-art Intel 16 silicon process high-frequency transistors, have enabled our team to innovate and deliver a compact, efficient, and scalable imaging platform to sub-terahertz frequencies. Such compelling results further strengthen the research collaboration between Intel and MIT,” said Dogiamis. “Prior to this research, people didn’t really combine terahertz technologies and semiconductor chip technologies to achieve this ultra-sharp, electronically controlled beamforming,” Han says. “We saw this opportunity, and using unique circuit techniques, we developed very small but also very efficient on-chip circuits so that we could effectively control the behavior of the wave at these locations. of integrated circuit technology, we can now enable internal memory and digital behaviors, which did not exist in the past. We are convinced that by using semiconductors, we can really achieve something amazing .”
A range of applications
The researchers demonstrated the reflective array by taking measurements called radiation patterns, which describe the angular direction in which an antenna radiates its energy. They were able to focus the energy very precisely, so that the beam was only one degree wide, and were able to steer that beam in one degree steps. When used as an imager, the one-degree-wide beam travels in a zigzag pattern over each point in a scene and creates a 3D depth image. Unlike other terahertz arrays, which can take hours or even days to create an image, theirs operates in real time. Because this reflector array works quickly and is compact, it could be useful as an imager for a self-driving car, especially since terahertz waves can see through bad weather, Monroe says. The device could also be well suited for autonomous drones, as it is lightweight and has no moving parts. Additionally, the technology could be applied in security contexts, enabling a non-intrusive body scanner that could work in seconds instead of minutes, he says. Mr. Monroe is currently working with the MIT Technology Licensing Market to commercialize this technology through a startup. In the lab, Han and his collaborators hope to continue advancing this technology by using new advances in semiconductors to reduce the cost and improve the performance of chip assembly. The research is funded by Intel Corporation and the MIT Center of Integrated Circuits and Systems.