Hyperspectral Imaging and AI Speed Up Necessary Inspection for Organic Electronics

Fraunhofer IWS researchers, working under the European Union (EU) project OledSolar, introduced an approach for the monitoring, observation, and inspection of organic electronics in the manufacturing and development stages. The hyperspectral vision and measurement technique pairs with an AI model. It considers the barrier foils — films that protect OLEDs, solar cells, circuits, and other organic electronics from environmental influences, making them more robust and durable in performance — to perform measurements on the barrier foils themselves, as opposed to the water vapor that permeates the film barriers.

The method makes the speed of the necessary measurements hundreds of times faster (hours compared to weeks) without compromising precision. By increasing the speed of this facet of the production process, the overall time it takes to manufacture organic electronics and their components could be reduced.

The newly introduced method relies on the “imanto” platform, an acronym for “imaging analysis tools.” The imanto platform relies on an optical analysis head focusing on the foil strip, and a spectrometer decomposing the light reflected from the foil in a hyperspectral imaging camera. The spectrometer visualizes the magnitude at which the foil reflects light at multiple wavelengths and at different points, delivering immediate information about the properties and quality of the foil. The system generates entire stacks of color-filtered images of the foil under examination and stores them in a so-called hypercube. These hypercubes are subjected to specific data analysis to calculate the desired parameters of the foil inspection, such as the water vapor transmission rate of the barrier foil.

The method quickly detects even minute defects to the ideal structure of barrier films, and identifies their water vapor transmission rate. That rate helps determine the service life of flexible organic electronics.

Where barrier foils have traditionally been used in food and pharmaceutical applications to protect from contaminants, Wulf Grählert, head of the Optical Inspection Technologies working group at the Fraunhofer Application Center for Optical Metrology and Surface Technologies (Fraunhofer AZOM), said that organic electronics have made the requirements for the permeability of barrier foils with respect to water vapor very stringent.

Food packaging foils are required to allow no more than 10 grams of water vapor per square meter to pass through per day. For tablet foils, the water vapor transmission rate must be one-tenth to one-hundredth of a gram per day and per square meter of foil.

The barrier foil rate is even lower for OLED devices, considering micrograms of water vapor.

Additionally, the measurement for obtaining those measurements has typically required weeks to perform and evaluate. The Fraunhofer team obtained measurements using its new method in two to three hours.

The work involves multiple Fraunhofer institutes; after partners from the Fraunhofer Institute for Organic Electronics, Electron Beam and Plasma Technology (Fraunhofer FEP) delivered 100 barrier film samples using a roll-to-roll production process, Fraunhofer IWS deployed a water vapor transmission rate measurement technique developed in-house.

AI Component
IWS measured at multiple, distinct positions to train an AI model based on the in-house-developed measurement technique that would be able to predict the water vapor transmission rate of the measured barrier foil at the desired, increased rate of speed.

“No human could evaluate this data conventionally,” Grählert said. “That is why we apply artificial intelligence methods.” As AI models grow, or evolve as they are trained, the method is able to perceive correlations between cause and effect — in this application, for example, between different defects in the barrier foils and the unique problems they are apt to cause. A trained model for the Fraunhofer collaborators would be able to detect any abnormality and make a subsequent prediction on the water vapor transmission rate involving the affected foil with a high degree of confidence.

Roll-to-roll foil inspection system at Fraunhofer FEP. Courtesy of Fraunhofer IWS.

The model was then trained to perform quickly to determine potentially faulty batches almost instantly. The full system also makes it possible to evaluate the specific points at which the barrier foils allow more (or less) vapor to permeate.

The goal is to achieve real-time quality control in the foil production process and in incoming inspection during processing operations. “This new type of sensor technology significantly shortens inspection time for barrier foils, opens up the possibility of inline quality control, reduces reject rates, and can make OLED and OPV production more cost-efficient,” Grählert said.

The team reported testing its method on frozen pizzas to inspect the presence and amount of certain toppings, and to see if other, unexpected ones are present. Beyond organic electronics, the researchers said the technology and method are applicable to applications in the electronics and semiconductor industries.

A team from the Fraunhofer Institute for Material and Beam Technology (Fraunhofer IWS) Dresden is presenting the concept at the LOPEC trade fair this month, a press release said. The OledSolar project currently involves 16 institutions. It is scheduled to run through spring 2022.