Power/Performance Bits: Oct. 11

Finer printed circuits; wavelength conversion; predicting 6G traffic.


Finer printed circuits
Researchers from the National Institute for Materials Science in Japan, Jiangnan University, Zhengzhou University, Senju Metal Industry Co., and C-INK Co. developed a way to print smaller features for printed electronics. The directed self-assembly method increases the chemical polarity of predetermined areas on a surface, which promoted selective adhesion of metallic nanoparticles.

The dual surface architectonic method enables printing of circuit lines 0.6 µm in width.

Simple photo and chemical treatments are applied to the substrate during the process. First, preselected surface areas are activated by ultraviolet irradiation. A chemical treatment is then applied to these areas which increases chemical polarity and surface energy only in the UV-activated surface areas. This increases the adhesiveness to metallic ink in those particular areas.

(a) Microcircuit patterns printed using a dual surface architectonic process. Circuits printed on polyimide (b) and transparent (c) films. (Credit: National Institute for Materials Science)

Both treatments can be performed in ambient air, and results in finer lines than can be currently made with inkjet and screen printing methods. The team said it’s also faster than photolithography and other conventional printing methods.

A metallic nanoparticle self-assembly system has been built by Priways Co. and C-INK Co., and the companies plan to offer it for sale soon, along with primers designed to improve adhesion on different substrates.

Wavelength conversion
Researchers at University of California Los Angeles, Iowa State University, and Technical University Darmstadt found a way to take advantage of an unwanted state to more efficiently convert light from one wavelength to another.

The team focused on a generally undesirable but natural phenomenon called semiconductor surface states, which occur when surface atoms have an insufficient number of other atoms to bind to, causing a breakdown in atomic structure. These dangling bonds hinder the flow of electric charges through a semiconductor, affecting performance.

“There have been many efforts to suppress the effect of surface states in semiconductor devices without realizing they have unique electrochemical properties that could enable unprecedented device functionalities,” said Mona Jarrahi, professor of electrical and computer engineering at UCLA Samueli and lead of the UCLA Terahertz Electronics Laboratory.

These incomplete bonds create a shallow electric field across the semiconductor surface. The researchers said that incoming light can hit the electrons in the semiconductor lattice and move them to a higher energy state such that they can move around the lattice. The photo-excited, high-energy electrons are further accelerated by the electric field and unload the extra energy by radiating it at different optical wavelengths.

To make this process more efficient, the team incorporated a nanoantenna array that bends incoming light so it is tightly confined around the shallow surface of the semiconductor.

Photograph, microscopy, and scanning electron microscopy images of a fabricated nanoantenna array placed at the tip of a fiber for optical-to-terahertz wavelength conversion. (Credit: Deniz Turan)

“Through this new framework, wavelength conversion happens easily and without any extra added source of energy as the incoming light crosses the field,” said Deniz Turan, a recent doctoral graduate in electrical engineering from UCLA Samueli.

The researchers successfully and efficiently converted a 1,550-nanometer wavelength light beam into the terahertz part of the spectrum, ranging from wavelengths of 100 micrometers up to 1 millimeter. The team demonstrated the wavelength-conversion efficiency by incorporating the new technology into an endoscopy probe that could be used for detailed in-vivo imaging and spectroscopy using terahertz waves.

The team said that this conversion would typically have requires 100 times the optical power to achieve the same terahertz waves, which wouldn’t be possible for the optical fibers in the endoscopy probe. In addition, they said it can apply to optical wavelength conversion in other parts of the electromagnetic spectrum, ranging from microwave to far-infrared wavelengths.

Predicting wireless traffic efficiently
Researchers from King Abdullah University of Science and Technology (KAUST) propose a way to help future wireless networks handle traffic.

AI could be used to coordinate the communication resources of upcoming 6G networks, learning from historical patterns of network usage across the network over time. However, transmitting the usage data from notes to a central database for analysis creates a substantial bandwidth overhead.

Instead, the researchers suggest a decentralized prediction model.

“Wireless traffic prediction could play a central role in network management as the basis for intelligent communication systems,” said Chuanting Zhang, a postdoctoral researcher at KAUST. “AI techniques such as deep neural networks are able to accurately model the complicated spatio-temporal nonlinear correlations in wireless traffic. However, as different base stations can have very different traffic patterns, it is quite challenging to develop a prediction model that performs well on all base stations at once.”

The approach combines a central global model with local models at each base station. The scheme weighs the influence of the local models according to network location and then sends only a limited amount of information from the base stations at each update.

“With this method, we have decentralized wireless traffic prediction and also implemented dual-attention global model optimization by paying attention to both the current knowledge of the central server and the information of local clients.” says Zhang. “Each updated global model can then be deployed to each base station to predict and adapt to new traffic patterns.”

The team said that the framework, called FedDA or dual attention-based federated learning, can provide a high-quality forecast of the spatial and temporal change in network usage over time with low overhead. It can also cluster base stations based on geolocation to improve efficiency and prediction accuracy.

Leave a Reply

(Note: This name will be displayed publicly)