Virtual unrolling for analysing rolled objects

The ideal sample shape for X-ray Computed Tomography (CT) is a cylinder. When studying large and flat samples we can roll them, such that their shape gets close to a cylindric geometry. However, it is challenging to visualize the object in its original, unrolled state. Virtual unrolling is therefore a key element when considering rolling samples for imaging. Additionally, samples such as scrolls or batteries have a rolled structure, which is interesting to study unrolled as well. In this project, virtual unrolling was applied to a rolled sheet of microelectronics, consisting of a flexibel polymer layer with a lithographically coated conductive metal layer on top. For full functionality, the metal layer must not be broken, so the unrolled sheet was investigated for cracks.

2D slices through the rolled microelectronics part. a) original data, b) manually indicated start and end position for unrolling, c) layered surface detection to follow and detect the surface.

Left: 3D image of the microelectronics as obtained using X-ray CT. Right: extracted sheet of microelectronics shown as a 2D surface in 3D. The color information was collected from the X-ray CT data such that the metal coating and cracks are clearly visible on the surface

Extracted sheet of microelectronics shown as a 2D image such that the whole surface can be visually inspected. This allows detection of cracks in the metal coating or the polymer surface. The zoom in on the bottom, highlighted by the red box, shows such a crack in the polymer surface and slightly in the metal layer. Note that the length of the unrolled piece is approximately 8 times longer than the diameter of the roll. This demonstrates the potential for imaging large pieces with high resolution when being rolled.

Project

The virtual unrolling project is part of the research conducted at the 3D Imaging Centre at DTU, where surface detection is one of the focus areas. The project was part of the general development activities within the LINX project, in which researchers at leading Danish universities collaborate with scientists in industry to solve industry relevant problems using advanced neutron and X-ray techniques. Further information: DOI: 10.1109/ICPR.2018.8545715

Challenge

The output of an X-ray Computed Tomography (CT) experiment is a 3D dataset arranged in slices. However, these slices are not alligend with the shape a rolled object. To study the surface of a rolled object, e.g. to read a text in a scroll, we need to virutally unroll it. To solve this, the individual layers need to be distinguished such that the object can be unrolled virtually.

Results

A piece of rolled microelectronics was scanned by X-ray micro CT using a ZEISS Xradia 410 Versa device. The individual layers have been found by a surface detection alogrithm using a semi-automatic approach. In this pilot study we manually indicated the start and end position of the rolled surface in a first step, as it is shown in the second figure from the top. Using the surface detection, both the polymer and metal layers were detected. This allowed us to visualize the unrolled printed piece also in its individual layers. While no crack in the conductive metal layer was observed, a crack was found in the polymer layer. This crack does not influence the functionality of this individual part, but this study shows that a similar process can be used to investigate rolled materials for cracks or other questions arising in quality control.

Perspectives

Virtual unrolling brings possibilities for e.g. studying the surface conditions of anodes and cathodes in rolled batteries, for identifying text in highly fragile scrolls or for other applications in cultural heritage and industrial inspection. Furthermore, it offers the option of rolling an object to fit better with the physical imaging restrictions, such that e.g. a better spatial resolution could be obtained.

Download one pager here