Responsive Manufacturing of High Value Thin to Thick Films.

Thin films with a high technical specification are used in many everyday devices, including displays, solar cells, electronic devices, batteries, and sensors. Printing of the high-value flexible electronic films with insulating, dielectric, semiconducting and conducting materials used in these devices makes a major and rapidly growing contribution to UK industry.The thickness of the films required, the starting materials used and the final high-value functions desired in the finished product vary significantly.

However, the scientific principles that govern the behaviour of the printing processes for these diverse applications have many similarities, because they are all formed by selectively spreading a wet film of suspended solid particles and drying it.

At present the optimisation of the printing parameters for these films is commonly achieved through a trial and error process rather than systematic intelligent control. Individual processes are being optimised in isolation without cross-fertilization of knowledge. In a fast changing world, where disruption to supply chains or development of improved materials can change the process input materials, the need to reconfigure the formulations/printing parameters used increases.

Furthermore, desired outputs can also change rapidly as the manufacturers and customers seek to meet changing demands of their market for example requiring more precise control of film parameters such as thickness and electrical properties. Adjusting to such continually moving goal posts by relying on trial and error testing is time-consuming, wasteful and costly.

The responsive manufacturing technology we propose to develop will have sufficient flexibility to overcome such problems by utilizing intelligent machine learning to control the printing parameters in real-time and therefore maintain an optimized printing process robustly in the face of variations in feedstock materials and/or the required output. It is surprising that there has been no major attempt to implement this approach to process control and optimisation for solution printed materials.

This is despite process monitoring and feedback-based optimisation being proven enabling methods in other sectors such as additive manufacturing.

This will be achieved by developing control algorithms for the printing process that take into account our theoretical understanding of the processes occurring and utilizing high-speed (minimized and proxy) in situ data acquisition to respond autonomously and continuously to perturbations in the feedstock materials or required film properties. We will make use of the wide range of laboratory scale processing systems our project team regularly use for the production of model colloidal films, ceramic dielectrics, photovoltaics and battery electrodes to provide the datasets required to educate the machine learning algorithms, test our theoretical understanding, develop models of the printing processes and to ultimately test the autonomous control system that we develop.

Having proven the system works at a laboratory scale we plan to perform a series of demonstration runs at industrial scale in collaboration with project partners CPI who are world leading experts in production of printed electronics. This will provide the evidence needed to prove that this approach can work at an industrial scale in a highly demanding production environment (printed electronics require a high degree of control of the surface chemistry between subsequent layers to perform correctly and are typically made in cleanroom/glove-boxes within strict environmental tolerances).

We envisage a future where a deep theoretical understanding of the processes that are taking place is utilised by artificial intelligence to continuously control and optimise the manufacture of 21st century high-value printed films autonomously using the minimum number of high-speed measurements to achieve the desired results.