Plastic packaging is designed around appearance, convenience, and function, but what happens after disposal is often considered separately—yet managing this waste is key to building a circular economy, since good recycling requires sorting plastics correctly by type. Transparent polyethylene terephthalate and polystyrene, for example, can be difficult to distinguish visually, making accurate separation challenging.
Addressing this challenge, a research team led by Mr. Juniya Yoshihara, a master's graduate student, together with Professor Tadao Tanabe and Professor Mitsuhiro Shigeri from the Department of Engineering and Design, Shibaura Institute of Technology (SIT), Japan, developed a design framework incorporating terahertz (THz) sensing into plastic packaging design. The researchers focused on transparent soba noodle containers and considered material identification alongside geometry, consumer needs, and transport efficiency. "We wanted to rethink packaging design so that recyclability becomes part of the design process while everyday usability and practical requirements are retained," says Mr. Yoshihara. The study was made available online on July 16, 2026, and will be published in Volume 32 of the journal Results in Engineering on December 01, 2026.
The research began with a community-based collection experiment in Isesaki City, Gunma Prefecture, Japan. Residents brought washed transparent food containers to a collection site equipped with an automated identification system. Among 383 collected samples, most were correctly identified, but some were misclassified. Analysis suggested that container shape and surface structure contributed to THz signal variations. The fixed model was not retrained during collection, providing an independent field-test evaluation under real-world conditions.
The researchers then explored how geometry could support stable identification. They created different bottom surface shapes and evaluated their THz transmission alongside image-based measures of complexity. The results showed that ribs, slopes, grooves, and other uneven features could affect measurements. A stable, flat region at the center helped provide a more consistent area for THz sensing, particularly when measurements were sensitive to shape.
Using these findings, the team developed four prototype soba containers. One prototype offered a strong balance between identification performance and practical use. It combined a regular triangular bottom pattern with a central flat area while retaining an uneven surface that could help prevent noodles from becoming soggy. The design also incorporated spaces for condiments and dipping sauce and considered how the container could be held during recycling.
A small survey indicated favorable impressions of the selected design, including durability, handling, and innovative appearance. The final container was designed for easy handling and placement that directs the central sensing area toward the identification system. Its stacking arrangement also allowed more containers to be transported together than the general container used for comparison.
Importantly, the final design maintained stable THz transmission while preserving the practical functions expected from a food container. The findings demonstrate that recycling considerations need not come at the expense of usability, appearance, or transport efficiency. Instead, sensing requirements can become part of product development itself.
"By integrating sensing considerations into product geometry, we can explore packaging that is easier to identify after disposal without giving up its familiar functions," says Mr. Yoshihara. The study points toward a broader shift in sustainable product development in which recyclability is considered alongside everyday performance from the beginning.
Overall, the study presents a case-based framework combining THz sensing, machine learning, image analysis, and product design to improve material identifiability after disposal. While further validation is needed across different packaging types, materials, and real-world conditions, the approach could support everyday products designed for more efficient resource recycling and reduced environmental impact.
