Digitization of semiconductor epitaxy for power electronics
Joint project “Epi-Techno-Digital” launched with the participation of IKZ, FBH and HU Berlin
The BMFTR joint project “Digitization for semiconductor epitaxy layers with large band gaps for power electronics” was officially launched with a kick-off meeting hosted by the project coordinator AIXTRON SE in Herzogenrath in January 2026. The three-year project aims to fundamentally digitize epitaxial processes for wide bandgap semiconductors and to apply machine learning for making process adaption and optimization more efficiently.
The focus will be on process development and optimization of metal organic chemical vapor deposition (MOCVD) for the production of layers for components that can switch high voltages and currents. At the same time, a holistic digital workflow is being established: all information is systematically recorded and linked, from machine data and process data, including in-situ measurements and ex-situ characterization data, up to device performance data. This creates a digital twin of the process, which serves as a virtual representation of the real process conditions. This comprehensive database enables the targeted use of machine learning with the aim of shortening development and optimization cycles.
Project management for the established GaN-on-silicon system is handled by AIXTRON. RWTH Aachen University and the Ferdinand Braun Institute gGmbH (FBH) are responsible for manufacturing and characterizing the devices. It is planned to develop quasi-vertical GaN pn diodes with a voltage resistance of up to 1,200 volts and quasi-vertical GaN trench MOSFETs. The Max Planck Institute for Sustainable Materials (SUSMAT) in Düsseldorf is supporting the digitization process with atomistic calculations of the surface structure. The results of these calculations are incorporated into the machine learning models together with experimental data.
The further development of the Ga₂O₃ process is being carried out at the Leibniz-Institut für Kristallzüchtung (IKZ). Again, the FBH is responsible for the processing and characterization of vertical components, in this case quasi-vertical Ga₂O₃ Schottky diodes and quasi-vertical Ga₂O₃ FinFETs. Together with the Chair of Theoretical Solid State Physics at Humboldt University in Berlin, the IKZ is equally developing the digital workflow and using machine learning specifically to support process development. In addition, atomic simulations will be carried out to gain a deeper understanding of the process and further improve the data-driven models.
The digital workflow developed is designed to be as general as possible in order to enable its transfer to other material systems in the future. Technically, the workflow is mapped in the NOMAD Oasis tool, which is being further developed as part of the FAIRmat project.
The overall project is funded by the North Rhine-Westphalia Coal Phase-out Fund with a total volume of around €21.5 million. Of this, 59% comes from the Federal Ministry of Research, Technology and Space (BMFTR) funding and 41% from AIXTRON's own resources. The IKZ receives a grant of about €1.3 million.
Contact
Leibniz-Institut für Kristallzüchtung (IKZ)
Dr. Wolfram Miller
+49 30 246499-522
wolfram.miller(at)ikz-belin.de
www.ikz-berlin.de
IKZ press release, 11 March 2026