Ion exchange typically involves replacing smaller ions with larger ions below the glass transition temperature. This study introduces an innovative computational approach for the inverse design of ion-exchangeable glasses. It aims to simultaneously achieve high depth of layer (DOL) and surface compressive stress (CS), a task complicated by the inherent trade-offs between these properties. Traditional ion exchange experimental methods are not only time-consuming and expensive, but they also involve complex practical difficulties. Addressing these challenges, this paper introduces a novel computational approach that synergistically combines a multi-objective genetic algorithm with machine learning models. The approach involves training models on a diverse set of glass compositions to predict DOL and CS, which then guide an evolutionary algorithm to optimize these properties concurrently. Physics-informed parameters further refine the search, enabling a flexible design framework in glass manufacturing. Comprehensive experimentation and subsequent analysis affirm the proposed method's capacity to efficiently and effectively traverse the intricate design landscape of glass materials. The findings elucidated the influence of distinct compositional and process parameters on DOL and CS, facilitating the manual design process. Additionally, the study identified sixteen glass composition candidates within the SiO2–B2O3–Al2O3–MgO–Na2O system, exhibiting remarkable properties. Specifically, some compositions achieved a DOL of up to 95 μm with a corresponding CS of 890 MPa, while others attained a CS of up to 1654 MPa with a DOL of 14.72 μm.
Read full abstract