Generative and Evolutionary Approaches for Safe and Sustainable by Design Material Development: A Case Study on Acid Alternatives for Advanced Material Coatings

The intersection of technological advancement and environmental sustainability presents a challenge in contemporary society. This study explores this intersection within the framework of the MOZART project, a research initiative funded by the European Union’s Horizon 2021 program. At its core, the MOZART project seeks to advance Safe and Sustainable by Design (SSbD) strategies, with a specific focus on replacing environmentally harmful Hard Chromium coatings. These coatings, while offering exceptional performance, pose significant environmental risks due to the carcinogenic nature of Chromium Trioxide, a byproduct of their production process. The proposed solution centers on nickel matrix nano-composite coatings. However, the traditional use of boric acid in the process to obtain them, contradicts sustainability principles, necessitating the exploration of sustainable alternative acids with similar performance. Using a Conditional Variational Autoencoder (CVAE), a specialized generative machine learning model, this study investigates potential boric acid substitutes. A comparative analysis of molecular representations is conducted to assess their suitability as inputs for the CVAE. Subsequently, an optimization process employing Genetic Algorithms is undertaken to identify regions within the CVAE’s latent space where sustainable molecules similar to boric acid are located.