NSF Award
2428667
Shape memory alloys, such as nickel-titanium (also known as nitinol), are a type of “smart” material widely used in medical devices. Nitinol is a metal with remarkable properties including super-elasticity, which is the ability to stretch like rubber and recover the original shape after unloading, and the shape-memory effect, in which a bar bent to a “permanent” shape like a paper clip, recovers its original straight shape after heating. These characteristics are exploited in a variety of important biomedical devices, such as cardiac stents. However, repeated loading of these devices after they are implanted in a patient, for example expansion and contraction due to blood flow driven by a beating heart, can cause them to break at microscopic material defects, such as non-metallic inclusions. This NSF/FDA Scholar-in-Residence at FDA (NSF FDA SiR) research project seeks to improve the ability of engineers to design nitinol devices that are less likely to fail by developing a specialized multiscale computational method that uses artificial intelligence (AI) to model the behavior of the nitinol atoms in the small volume near a defect. The method will be validated against specially designed fracture experiments to ensure its correctness. This approach will speed development of new medical devices, improve regulatory pathways to market, and reduce the risk to patients. <br/><br/>This research approach is based on the three-dimensional quasicontinuum method (QC3D), which is concurrent multiscale method that dramatically reduces the computational cost relative to fully-atomistic methods through a coarse graining approach. Full atomistic resolution is retained in regions where “interesting” phenomena is occurring, such as phase transformations or bond breaking near a defect, whereas the rest of the device is modeled using a nonlinear finite element approximation employing Cauchy-Born kinematics. QC3D is a systematic approximation to the exact fully-atomistic result, which converges with mesh resolution. However, agreement with reality depends on the accuracy of the interatomic potential (IP) used to model the atomic interactions in the atomistic regions and as the basis of the Cauchy-Born constitutive response. To validate the QC3D approach, predictions using existing physics-based IPs and new AI-based IPs will be compared with fracture toughness experiments on single crystal and/or polycrystal nitinol samples. Experiments will include standard 3-point bending experiments as well as a novel fracture gap test that can be used to explore the effect of compressive stress on phase transformations at the crack tip. The validated QC3D method will be applied to study the effect of non-metallic inclusions on nitinol fracture.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
