The team led by Associate Professor Wang Cheng of City University of Hong Kong, in collaboration with researchers from The Chinese University of Hong Kong, has developed a microwave photon chip using lithium niobate as a platform. This chip offers faster processing speed and lower energy consumption, enabling ultra-fast analog signal processing and computation using optics. The research findings were published in Nature on February 29th.
Integrated microwave photon chips generate, transmit, and manipulate microwave signals using optical components. However, it has been challenging to simultaneously achieve chip integration and high fidelity, low power consumption for ultra-high-speed analog signal processing in integrated microwave photon systems.
"Low energy consumption is of great significance in the field of artificial intelligence. Nowadays, more and more artificial intelligence products are emerging, with faster product iterations and increasing scale and complexity of AI models. Along with this comes the increasingly prominent issue of energy consumption, as it not only leads to higher product costs but also brings about environmental problems that cannot be ignored," said Wang Cheng in an interview with the China Science News.
To address these challenges, Wang Cheng's team placed the ultra-fast electro-optic conversion module and a low-loss, multifunctional signal processing module on the same chip, creating an integrated microwave photon system. The outstanding performance of this system is thanks to the thin-film lithium niobate platform responsible for integration.
"Lithium niobate is as important for photonics as silicon is for microelectronics, earning it the nickname 'silicon of photonics'," said Wang Cheng. During his doctoral studies at Harvard University, he focused on researching integrated lithium niobate photon platforms. After joining City University of Hong Kong, his research team has continued to delve into the field of lithium niobate microwave photonics, aiming to make microwave photon chips smaller, with higher signal fidelity and lower latency performance.
"I believe that lithium niobate is a highly promising platform for large-scale, on-chip photonic integration applications. Compared to other optical materials, it possesses excellent electro-optic effects, ultra-low optical losses, and large-scale, low-cost manufacturing processes," explained Feng Hanke, the first author of the paper and a doctoral student at City University of Hong Kong.
The integrated lithium niobate microwave photon chip developed by Wang Cheng's team not only provides processing speeds 1000 times faster than traditional electronic processors, a wide processing bandwidth of 67 gigahertz, and high computational accuracy, but it also has lower energy consumption. For example, when processing a 250×250 pixel image, the integrated lithium niobate microwave photon chip only requires 3 nanojoules of energy to extract the edge information of the image, whereas a traditional electronic chip would require several hundred or even thousands of nanojoules for the same task.
For Ge Tong, co-first author of the paper and an undergraduate at City University of Hong Kong, the highlight of this research was when conducting tests for ultra-high-speed signal processing and directly inputting pulse signals with widths less than 10 picoseconds into the chip and observing the differential result of the signal on the oscilloscope. "This directly proves that our photon processor can effectively process signals of such high speeds, setting a new world record." The integrated lithium niobate microwave photon chip will enter various application scenarios with remarkable advantages, including 5G and 6G wireless communication systems, high-resolution radar systems, image and video processing, and more.
Next, the team led by Wang Cheng will further optimize and validate the chip. Key technical challenges include enhancing integration, achieving efficient packaging of the chip and control circuitry, optimizing device performance and stability, etc., to truly enter the productization stage. Related paper information: https://doi.org/10.1038/s41586-024-07078-9 Please scan the QR code below for 3 seconds.
