Ulsan, South Korea, January 4, 2024 – The rapid development of 6G communications using Terahertz (THZ) electromagnetic waves has strengthened the demand for highly sensitive nanoresonators that can detect these waves.
In response to this need, researchers at the Ulsan National Institute of Science and Technology (UNIST), in collaboration with the University of Tennessee and Oak Ridge National Laboratory, optimized a THZ nanoleon agent specifically for 6G communications using artificial intelligence. Did.
According to Young-Taek Lee researchers, the optimized nanoresonar agent is versatile and has potential applications in ultrafast detectors, supramolecular detection sensors, and bolometer research.
The team’s analytical model-based approach significantly reduces the computational resources required to optimize THZ nanorocenoters and provides a practical alternative to numerical simulation-based inverse design of THZ nanodevices . To keep progress in perspective, this process allows processes that were time-consuming and demanding even on supercomputers to be efficiently engineered into personal computers.
A research team led by Professor Hyong-Ryeol Park of Unist has developed a technology that can amplify Terahertz electromagnetic waves by 30,000 times. A new nanorenonaut, developed using a rapid inverse design method based on physical models and combined with AI, could catalyze the commercialization of 6G communication frequencies.Courtesy of nano letter (www.doi.org/10.1021/acs.nanolett.3c03572).
In about 39 hours using a medium-sized PC, the researchers identified the optimal structure through 200,000 iterations and achieved an experimental electric field enhancement of 32,000 at 0.2 THz. The electric field generated by the THZ nanoleonator, which exceeded common electromagnetic waves by more than 30,000 times, showed an efficiency improvement of more than 300% compared to previously reported THZ nanoleonators.
AI-based inverse design techniques are typically used for optical device structures in the visible and infrared regions, which are just a fraction of the wavelengths. Using AI-based inverse design for the 6G communications frequency range (0.075-0.3 THz) presented significant challenges to researchers due to the small size of the range (about 1 million wavelengths in size).
The researchers worked with nanogap loop arrays, a type of resonator that has demonstrated the potential to detect THz electromagnetic waves. The unit cell of these arrays is 10 times larger, so 1,000,000 times smaller than a millimeter wavelength with a small nanogap region, accurate simulation requires significant computational resources.
To improve the efficiency of nanogup-loop arrays, researchers combined nanorocene agents with a rapid inverse design method based on physics-based machine learning. Specifically, the inverse design approach used double-deep Q-learning with an analytical model of a THz nanogap loop array.
Professor Hyun Lior Park, who led the research, emphasized the need to understand physical phenomena in conjunction with AI technology. “AI may seem like the solution to all problems, but understanding the physical phenomena is still important,” he said.
With the help of physics-based machine learning, the team evaluated the efficiency of the new nanorenone agent through a series of THZ electromagnetic wave transfer experiments conducted in simulation.
“The methodology adopted in this study is not limited to specific nanostructures, but can be extended to different studies using physical theoretical models of different wavelengths or structures,” Lee said.
This study nano letter (www.doi.org/10.1021/acs.nanolett.3c03572).
