Thin-film carbon nanotubes assist novel non-destructive testing

06 Mar 2024

Chuo University develops photosensitive device able to image targets' inner structures

A project at Japan's Chuo University has developed a route to reconstruction of hidden internal structures of certain targets, an approach built around a multi-functional photosensor.

Published in Advanced Optical Materials, the research could lead to enhanced non-destructive testing (NDT) and quality assurance methods, by improving the ability of photo-imaging methods to detect concealed defects in industrial materials by reconstructing optical data.

The study builds on previous use of carbon nanotube (CNT) films as sensitive uncooled millimeter-wave infra-red sensor devices. These uses exploit the photothermoelectric (PTE) effect, a combination of photo-absorption, induced heating and thermoelectric conversion.

"CNT film PTE imagers exhibit photo-detection sensitivities in broad millimeter-wave IR (MMW-IR) regions, and demonstrate non-destructive material identification via multiple-wavelength monitoring," said the Chuo University team in its published paper.

"Employing multiple CNT film imager sheets provides omni-directional viewing angles without blind spots for monitoring 3D objects. But 3D reconstructions of the inner structures of opaque inspection target objects via broad MMW–IR monitoring are still insufficient. Information regarding the size or positioning of the respective inner elements of composite multi-layer 3D target objects is crucially missing."

What lies beneath

The team's solution combines a multi-functional CNT film sensor and computer vision (CV) in ultrabroad and multi-wavelength bands, using CV to reconstruct hidden features of target objects from the MMW-IR data captured by the senor.

In use the system first acquires two-dimensional silhouette images from different viewpoints, then hollows out the spatially overlapped area into simple structural reconstructions. This computational work flow starts with wavelength-specific 2D images extracted from the target object via the ultrabroadband sensor, followed by graphical superposition of these wavelength-specific models. The result is a non-destructive reconstruction of the 3D composite multi-layer target objects.

In trials, this synergistic combination of 3D computer vision techniques and MMW–IR ultrabroadband multiple-wavelength photo-monitoring was able to reconstruct 3D models of objects made from glass, semiconductors, plastic and metal, the major constituent materials in today's industrial components.

"THz and IR measurements and the associated silhouette views reconstructed middle- and inner-layers through the opaque outer wall respectively, and the higher permeable MMW monitoring extracts a metallic bar from the deepest center of the target," noted the project.

This first successful combination of 3D computer vision techniques and CNT film PTE-based photo-monitoring could now lead to the development of other types of non-destructive reconstruction approaches, and ultimately the extraction of functional information via deeper understandings of the imaging information.

Enhanced functional data should allow the same CNT film sensor and subsequent image reconstruction approach to be applied to "tomography, photoacoustic and LiDAR imaging," commented the project.