November/December AMP_Digital

A D V A N C E D M A T E R I A L S & P R O C E S S E S | N O V E M B E R / D E C E M B E R 2 0 1 8 1 2 ARTIFICIAL NEURAL NETWORKS IDENTIFY NEW MATERIALS Predicting the stability of materi- als has long been a challenge in chem- istry, physics, and materials science. Now, researchers at the University of California, San Diego are training ar- tificial neural networks—algorithms inspired by brain connections—to com- bine traditional chemical intuition and quantum mechanical computations to more effectively identify stable mate- rials. By training these faux neural net- works to predict a crystal’s formation energy using just two inputs, electro- negativity and ionic radius of the con- stituent atoms, researchers developed models that can identify stable mate- rials in two types of crystals—garnets and perovskites. These simulations are up to 10 times more accurate than previous machine learning models and are fast enough to efficiently screen thousands of materials in a matter of hours on a laptop PC rather than a supercomputer. The team says these neural networks have the potential to greatly accelerate discovery of new materials for applica- tions such as solar cells, rechargeable lithium-ion batteries, and LEDs. They also plan to extend the application of neural networks to other crystal proto- types and different material properties. ucsd.edu. NEW METHOD PROGRAMS MOVING HYDROGELS University of Texas at Arlington re- searchers developed a new process to program2D hydrogels that could poten- tially transform the way soft engineer- ing systems or devices are designed and fabricated. The process involves programming 2D hydrogels to expand and shrink in a space and time-con- trolled way that applies force to their surfaces, enabling formation of com- plex 3D shapes and motions. Potential applications include bioinspired soft ro- botics, artificial muscles, and program- mable matter. The concept also ap- plies to other programmable materials. The new approach uses tempera- ture-responsive hydrogels with lo- cal degrees and rates of swelling and shrinking. Those properties allow sci- entists to spatially program how the hydrogels swell or shrink in response to temperature change using a digital light 4D printing method that includes three dimensions plus time. The researchers also developed design rules based on the concept of modularity to create even more complex and dynamic struc- tures, including bioinspired designs with programmed sequential motions. uta.edu/mse. Prototype cells andmodules were truly black when they came out of production. Courtesy of Hele Savin. EMERGING TECHNOLOGY THE YEAR OF BLACK SILICON A 2011 invention by Aalto Uni- versity researchers, Finland, recently evolved from concept to reality. Just a few years ago, the team obtained record efficiency of 22% in the lab for nanostructured solar cells using atomic layer deposition. Now, with the help of industrial partners and joint European collaboration, the first prototype mod- ules have been manufactured on an in- dustrial production line. Schematic of an artificial neural network predicting a stable garnet crystal prototype. Courtesy of Weike Ye. UT Arlington researchers are fabricating lifelike materials in the lab. “Our timing could not have been better,” says lead researcher Hele Sav- in. In fact, 2018 is commonly called the Year of Black Silicon due to its rapid ex- pansion in the photovoltaic industry. The recent growth has enabled the use of diamond-wire sawing in multicrys- talline silicon, which reduces costs and environmental impact. However, there is still room for improvement as the current black silicon used in industry consists of shallow nanostructures that lead to less than ideal optical proper- ties that require a separate antireflec- tion coating. Aalto’s approach consists of us- ing deep needlelike nanostructures to create an optically perfect surface that eliminates the need for these coatings. The cells and modules were truly black when they came out of production and showed no signs of damage, and the best module produced energy with more than 20% efficiency. aalto.fi/en.