Eric Volkhardt¹, Marcus Frank², Alexey Arkhipov², Dennis Barton¹, Ralf Bandorf^1
1Fraunhofer Institute for Surface Engineering and Thin Films IST, Braunschweig, Germany
²Bühler Alzenau GmbH, Alzenau, Germany
These days, the digital twin has become an increasingly common partner to its real counterpart. The in‑silico execution of coating runs has already shown its great potential to accelerate development processes in terms of technology, economics, and ecology.
Today, we have the computational power and software to simulate and optimize our inventions before they see the light of day. Especially in the automotive industry – crash simulations, noise behavior, etc. – the future can no longer be imagined without them. However, in the coating industry, there is still considerable potential for computer‑aided engineering.
At the IST, the in‑house code PICMC is used to simulate and optimize CVD and PVD processes. Based on a digital representation of the coater in question, we simulate the plasma processes, including magnetic and electric fields and gas fluxes. On this basis, further properties are derived, e.g., target erosion and layer thickness profiles.
In former projects, the ability to accurately predict layer thickness profiles and subsequently optimize coater setups and process parameters has been demonstrated for an industrial coating platform that is used to apply a variety of optical filter designs on lenses of different shapes, e.g., a bandpass on a spherical lens or an anti‑reflection coating on aspherical lenses.
In recent industry projects, the use of the corresponding framework has been extended to large‑area coating processes. The main problem in performing these calculations is the sheer size of the units. For comparison, the length of the cathodes is generally about three meters, while in the setups calculated before they were just half a meter, and while in the circular coater the trajectory of the substrates is only about 2 or 3 meters, large‑area coaters easily extend to more than 100 meters.
As the size of the coaters exceeds the capabilities of general simulation frameworks in terms of the necessary computational resources, elaborate assumptions and model reductions have to be applied, while maintaining the precision of the computational model. Using the framework, we show how questions of layer thickness on curved substrates, influences of chamber pressure, and even variations of plasma parameters can be simulated and used for the optimization of large‑area coating units and processes.