Supercomputing and paper cutting are the basis of research on extensible electronics

Supercomputing and paper cutting are the basis of research on extensible electronics

Some wearable electronic devices, such as sensors sewn into applicable fabrics or “leathers”, rely on the development of new, durable and stretchable electronic materials. One way to improve the elasticity of these often delicate materials is to introduce strategic cuts, creating an extensible mesh. Recently, a team of researchers from the University of Southern California tackled this material design problem with inspiration from kirigami, the Japanese art of paper cutting, and used the power of supercomputing to make their approach possible.

“Origami or kirigami design based on the ancient papermaking technique is used to modify the mechanical behavior of 2D materials,” the authors wrote in their paper. “For example, graphene is brittle in nature, but its flexibility can be greatly improved by introducing shear patterns into the graphene sheet, thereby allowing for stretchable electronics.”

The research team combined these kirigami-inspired cuts with self-reinforcement learning, with the goal of optimizing an arrangement of the cuts in a 2D molybdenum disulfide (MoS2) structure for maximum elasticity. “The question is, can we use similar behavior in material design, as in this kirigami, where your goal is to create a more textured material that is highly stretchable, one cut at a time,” explained Pankaj Rajak, a lead researcher. on the project, in an interview with John Spizzirri of the Argonne Leadership Computing Facility (ALCF). “It’s a smart strategy to understand where the cuts should go.”

To power the reinforcement learning algorithm, Rajak and his colleagues ran 98,500 simulations of the material with a range of cuts (one to six) at different lengths. The simulations were run over the course of several months on Theta, a 6.9 Linpack petaflop supercomputer from Argonne that ranked 70th on the most recent Top500 list.

“You could have two hundred people each doing five experiments a day for a month collecting data on different cuts,” said Priya Vashishta, another member of the research team. “It would be very expensive in material and time. But in this case, the model was reasonably good and produced data very similar to the experimental data.”

Based on this six-cut simulated data, the reinforcement learning algorithm actually learned to predict eight- and ten-cut structures, producing billions of possible combinations that would have taken much longer to simulate. “If it was necessary [a] A few months to do 98,500 simulations and go up three orders of magnitude is a lifetime, ”Vashishta said.

The algorithm worked admirably, producing a ten-cut structure that added 40 percent extensibility to the material in seconds (pictured in the header). “He understood things we never told him we understand,” Rajak said. “He has learned something the way he learns a human being and used his knowledge of him to do something different.”

To learn more about this research, read ALCF’s John Spizzirri report here and read the research paper here.

Leave a Comment

Your email address will not be published.