b'Look-AheadNEW SHAPE-MEMORY ALLOYF unded by the National Scienceartificial intelligence materials selec- showcased high efficiency when sub-FoundationsDesigningMa- tion computational framework capa- jected to thermal cycling. The mate-terials to Revolutionize and Engi- ble of determining optimal materialrial also exhibited excellent cyclic sta-neer Our Future program, research- compositions and processing them,bility under repeated actuation.ers from the department of ma- which led to the discovery of a newA nickel-titanium-copper compo-terials science and engineering atshape-memory alloy composition. sition is typical for shape-memory Texas A&M University used an ar- Whendesigningmaterials,alloys. Nickel-titanium-copper alloys tificial intelligence materials selec- sometimes you have multiple ob- usually have titanium equal to 50% tion framework to discover a newjectives or constraints that con- and form a single-phase material. shape-memory alloy. It showed theflict, which is very difficult to workUsing machine learning, researchers highest operational efficiency thusaround, said Ibrahim Karaman,predicted a different composition far for nickel-titanium-base mate- Chevron professor I and head ofwith titanium equal to 47% and cop-rials. In addition, the data-driven framework offers proof of concept for future materials development.Shape-memory alloys are utilized in various fields when compact, lightweight and solid-state actua-tions are needed, replacing hydrau-lic or pneumatic actuators because the alloys can deform when cold and then return to their original shapes when heated. This unique property is critical for applications, such as airplane wings, jet engines and automotive components, that must withstand repeated, recover-able large-shape changes. Texas A&M UniversityThere have been advancementsDoctoral student William Trehern operates a vacuum arc melter, a synthesis method commonly in shape-memory alloys since theirused to create high-purity alloys of various compositions.beginnings in the mid-1960s but at a cost. The understanding and discov- the materials science and engineer- per equal to 21%. While this compo-ery of new shape-memory alloys hasing department. Using our ma- sition is in the two-phase region and required extensive research throughchine-learning framework, we canforms particles, they help enhance experimentation and ad hoc trial anduse experimental data to find hid- the properties of the material, ex-error. About every decade, a signifi- den correlations between differentplained doctoral student William cant shape-memory alloy composi- materials features to see if we canTrehern, graduate research assistant tion or system has been discovered.design new materials. and first author of the publication.But even with advances, the alloysThe shape-memory alloy foundIt is a revelation to use machine are hindered by their low energy ef- during the study was predicted andlearning to find connections that ficiency caused by incompatibilitiesproven to achieve the narrowest hys- our brain or known physical princi-in their microstructures during large- teresis ever recorded. In other words,ples may not be able to explain, shape changes. They also are notori- the material showed the lowest en- Karaman said. We can use data ously difficult to design from scratch. ergy loss when converting thermalscience and machine learning to To address these shortcomings,energy to mechanical work. Due toaccelerate the rate of materials Texas A&M researchers combinedits extremely small transformationdiscovery.CTEexperimentaldatatocreateantemperature window, the material Michelle Revelsctemag.com/cteguide.com45LookAhead.indd 45 9/19/22 2:28 PM'