Data-Driven Study of Shape Memory Behavior of Multi-Component Ni-Ti Alloys in Large Compositional and Processing Space

Shape memory alloys have found wide-spread use in aerospace, automotive, biomedical, and commercial applications owing to their favorable properties and ease of operation. Binary NiTi, in particular, is known for its remarkable shape memory properties, mechanical strength, ductility, corrosion resistance, and biocompatibility. These properties can be further enhanced and better controlled through alloying NiTi with ternary, quaternary, and higher-order elements. Recently, researchers at NASA have compiled an extensive database of shape memory properties of materials, including over 8000 multi-component Ni-Ti alloys containing 37 different alloying elements. Using the Ni-Ti dataset, we train machine learning models to explore shape memory behavior of Ni-Ti alloys over a large compositional and processing space. The models predict transformation temperatures, hysteresis, and transformation strain, with low mean absolute errors of 14.8 degrees C, 7.2 degrees C, and 0.36%, respectively. We use these models to map trends and learn relationships between shape memory behavior and different parameters in the input design space. They can be used to make predictions for any multi-component alloy, without need for additional training. The combination of an extensive experimental dataset and accurate learning models, together, make our approach highly suitable for the discovery and design of new alloys with targeted properties.