ASTERIX: Advanced Surface Technologies for Resilient and Innovative materials in eXtreme environment

Coordinator: Jean-Philippe POLI

CEA LIST

Keywords: PVD, CVD, symbolic AI, symbolic machine learning, extreme environment, HiPIMS, DLI-MOCVD, Digital twins, Technological sovereignty.


The development of thin-film engineering is absolutely strategic for advanced manufacturing of components that have to operate under extreme environments (nuclear energy, renewables, aeronautics in particular). These components are often exposed to a combination of constraints, thus requiring the design of multi-layer architectured systems to meet, in fine, the multi-functionality expected (diffusion barrier, resistance to oxidation and irradiation, mechanical resistance…).

Thin-film engineering is a field of technology that has made enormous progress in recent years, mainly challenged by the effective improvements demanded by the ever more uncompromising specifications of the microelectronics industry on the one hand, the booming low-carbon energies (in particular photovoltaics, hydrogen economy, as well as future of nuclear industry) and high- performance mechanics, on the other.

These advances concern PVD (Physical Vapor Deposition) technologies, in particular with the deployment of highly ionized plasma generation systems such as HiPIMS (High Power Impulse Magnetron Sputtering), which enable much better control of interface construction and coating microstructures, and, in fine, coating properties. These advances also concern CVD (Chemical Vapor Deposition) technologies, in particular thanks to the rich chemistry of organometallic compounds that can be used through the great flexibility of DLI-MOCVD (Direct Liquid InjectionMetal Organic Chemical Vapor Deposition).

The ASTERIX (Accelerated development of Surface TEchnologies for Resilient Innovative materials in eXtrem environments) project has a dual ambitious but realistic objective. Focused on the development of a complex, multi-functional technological device, a nuclear fuel cladding, it aims to develop two digital twins, one of the HiPIMS PVD process, the other of the DLI-MOCVD process, thanks in particular to an instrumented study of each process, to numerical simulation, and the training of a Machine Learning tool, EXPRESSIF MATERIALS, developed for this purpose.

The research teams involved in the consortium are at the cutting edge of the world’s state of the art, each in their field, to develop a holistic vision. It ranges from the definition of coating architectures, their elaboration using both instrumented and numerically simulated processes, and their advanced characterization. In particular, the latter will use large-scale facilities such as the ESRF and Soleil synchrotrons. Finally, implementing machine learning tools summarizes experimental data, numerical simulations, and the gathering of expertise to obtain the digital twins of these two sovereignty processes ultimately.

Work done on the HiPIMS process will also contribute to the development of a French HiPIMS plasma power supply that is more efficient than the current state-of-the-art and which will enable France to acquire or regain technological sovereignty in the field of key components for PVD technologies, thus permitting the much wider multi-sectorial industrial deployment of these technologies.

The generic nature of the methodologies used to build the two digital twins resulting from this project will enable this approach to be widely disseminated to deal with similar cases. These results will be published, along with the methods, databases, and digital tools developed, in the form of a public GitLab-type work directory.