When Human Expertise Improves the Work of Machines

Estimated reading time: 5 minutes

Using a conductive tip on an atomic force microscope, they examined the electromechanical response from a series of chemically related samples, generating as many as 2,500 time- and voltage-dependent measurements on a grid of points established on each sample. The process generated hundreds of thousands of data points and provided a good test for the stacking approach, known technically as concatenation.

“Instead of just looking at the chemical composition that provides the highest response, we looked at a range of compositions and tried to figure out the commonality,” she said. “We figured out that if we applied this data stacking with some thought process behind it, we could learn more about these interesting materials.”

Among their findings: Though the material is a single crystal, the functional response showed highly disordered behavior, reminiscent of a fully disordered material like glass. “This glassy behavior really is unexpectedly persisting beyond a small percentage of the material compositions,” said Bassiri-Gharb. “It is persisting across all of the compositions that we have looked at.”

She hopes the technique will ultimately lead to information that will improve many materials and their functionalities. Knowing which chemicals need to be included could allow the materials scientists to move to the next phase — working with chemists to put the right atoms in the right places.

“The big goal for any materials’ functionality is to find the guidelines that will provide the properties we want,” she said. “We want to find the straight path to the best compositions for the next generation of these materials.”

This research was supported by the National Science Foundation (NSF) through award DMR-1255379, by the Defense Threat Reduction Agency (DTRA) though grant HDTRA1-15-0035, by the Center for the Science and Technology of Advanced Materials and Interfaces (STAMI) at Georgia Tech, and Division II of the Swiss National Science Foundation under project 200021_178782. The piezo-response measurements were in part performed at the Center for Nanophase Materials Sciences at Oak Ridge National Laboratory, which is a U.S. Department of Energy Office of Science User Facility. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the sponsoring organizations.

• < Previous Page