Tuesday  |  Conference Center B  |  10:00 AM–10:20 AM

#13529, Space-time-domain Optical Interferometry for Deep-Learning Assisted Big-Data-Generating Experiments Under Extreme Conditions

Recently, a framework of a big-data-generating experiment was developed using a streak camera system. To this end, we first invented a novel Dynamic Line-Image Shearing Interferometer (DL-ISI), which can detect the continuous-time history of displacement information along a line on the surface of a sample under dynamic loading. Plate impact experiments on a sample with a mid-plane crack could generate an extremely high strain rate locally near the crack tip, which is equivalent to 10^8 s^(-1). Direct measurement of the dynamic cohesive-zone behavior has not been conventionally feasible. To overcome this difficulty, we employed a convolutional neural network (CNN) based deep-learning framework that can inversely determine the accurate cohesive parameters from DL-ISI fringe images. As an application, we investigated the ultra-high fracture toughness and strength of a hierarchically nanostructured copolymer, polyurea. This big-data-generating experiment enabled us to reveal exceptionally high strength and toughness of polyurea under extreme conditions. The cohesive strength is found to be around 300MPa, which is nearly three times higher than the spall strength under the symmetric impact with the same impact speed. The dynamic toughness of polyurea was measured as approximately 12000J/m^2 under K ̇~1e6 MPa√m s^(-1). Furthermore, we found this ultra-high-strength and toughness come from the nanodomain's dynamic fragmentation and self-healing characteristics from our in-situ AFM studies and coarse-grained molecular dynamics (MD) simulations. We believe these results fill the gap in the current understanding of copolymer’s cooperative failure strength under extreme local conditions.

Hanxun Jin   Brown University
Kyung-Suk Kim   Brown University