A materials scientist from the University of Connecticut, Ramamurthy Ramprasad, decided that trial and error in materials development was old hat. Instead of randomly mixing materials to see what they produce, he decided that they should be designed rationally. He created a machine that can not only design the material, but also accurately predicts what mechanical properties it will exhibit when created.

UConn researchers design materials from the atom up. Image credit: Schematic by Chiho Kim, Ramprasad Lab/UConn ImageHis team began by analyzing 283 commonly used polymers, each of which includes a repeated unit of polyethylene, polyesters, and polyureas, theorizing that multiple varieties of polymers could be created with these basic building blocks. The team then calculated the three-dimensional atomic and electronic structures of the 283 polymers using quantum mechanics. With that knowledge, they were able to determine the properties of each of the types of polymers, and its band gap and dielectric constant.

To overcome the time it would take to make the desired material evaluation calculations on a computer, the team developed a shorthand so the computer could look at the building blocks and develop an educated guess about its mechanical properties. The shorthand consists of three strings of numbers, arranged in a 3D matrix to give the computer a numerical “fingerprint” with which to sort through the data. Once entered, the computer has been able to accurately evaluate band gap and dielectric constant for the polymers, and any variation, using only the numerical fingerprint.

UConn collaborators have created some of the novel polymers developed by the machine, and have validated their properties are as predicted.