Project CORRTHIA aims to demonstrate the use of thermodynamic calculation for predicting corrosion properties using artificial intelligence (AI) models. Given the complexity of corrosion and the difficulty in describing it with current physical models,
understanding a material's behavior requires long-term experiments. The use of AI for predicting corrosion properties is a promising method for accelerated materials design, as it can limit the required experiments by selecting relevant materials. This research
theme aligns with the CEA's strategy for numerical material design and the objectives of the PEPR DIADEME.

Leveraging expertise in corrosion and thermodynamic calculation from the DES team and AI experience from DRT partners, we will focus on high-temperature oxidation (> 500°C) of metallic alloys to limit the range of compositions and possible environments. It is
expected that thermodynamic aspects play an essential role under these conditions. The scarcity of corrosion data necessitates work on constructing a dataset (Lot 1), based on published literature or internal S2CM data, which will be enriched with thermodynamic
calculation results. This augmented dataset will be used to train AI models.