An extensive experimental database of inorganic thin-film materials that organizes a decade’s worth of research at the Department of Energy’s National Renewable Energy Laboratory (NREL) is now publicly available (‘An open experimental database for exploring inorganic materials’ by Andriy Zakutayev et al, Scientific Data volume 5, article number 180053 (2018); doi:10.1038/sdata.2018.53).
The High Throughput Experimental Materials (HTEM) Database contains more than 140,000 sample entries collected by NREL scientists investigating inorganic materials for use in advanced energy applications, such as thin-film solar cells. The entries provide details about the structural, chemical and optoelectronic properties of the materials, and their synthesis conditions. More than half of these data are currently available online at htem.nrel.gov.All existing experimental databases either contain many entries or have all this property information, but not both,” says Andriy Zakutayev, a scientist at NREL’s Materials Science Center, which is dedicated to developing new materials and devices for solar cells and other renewable energy technologies.
In working on new materials, scientists synthesize many samples but only a fraction of what they learn along the way is ever published. Co-author John Perkins, a senior scientist in the Materials Science Center, estimates that information on about less than 10% of the samples makes it into a scientific journal. “You really only write journal articles about materials that worked,” he said, adding that the information that wasn’t published could also be useful to research efforts.
“Our belief is that putting all this data out in the public domain would accelerate the advancement of material science, in particular by researchers without access to expensive experimental equipment, both in the United States and around the world,” Perkins says.
Historically, scientists made one sample at a time, measured it, analyzed the data, and then made another one. But technological advances in research equipment and computer control have enabled researchers to collect more data faster. A square thin film measuring 2 inches on the side can have 100 data points because such sample ‘libraries’ are intentionally made with gradients in chemical composition, synthesis temperature, or film thickness, for example. “Doing such combinatorial research systematically over many years, for different projects with different goals, is what enabled creation of this database,” Zakutayev says.
This example of a high-throughput combinatorial experiment allows the data to be gathered at greater volumes and rates than ever and leads to the opportunity for machine learning to further expedite the analysis. “If you wanted to know how electrically conductive a particular combination of chemical elements was before you actually made the material and measured it, you may be able to use the machine learning algorithm to predict that quantity,” says co-author Caleb Phillips, a data scientist with the Computational Science Center at NREL.
The HTEM database draws from nearly a decade of thin-film experiments at NREL. Even data that was published, but not in a form that could be searched, was digitized and added to the database. “Once it’s all been amassed, it’s a non-trivial job to curate it and get it in a form where it can be analyzed and understood,” Phillips says, adding that this is where the future opportunities lie.
Now, Phillips, Perkins and Zakutayev are participating in a collaborative effort between NREL and the US National Institute of Standards and Technology (NIST) to deploy a network of high-throughput experimental tools that would allow researchers to collaborate virtually on the synthesis and analysis of new materials, with results being added to databases like this. The pilot project has been dubbed the High-Throughput Experimental Materials Collaboratory.
Development, curation and filling out of the HTEM database was supported by NREL’s Laboratory Directed Research and Development program, the Department of Energy’s Office of Science, and the Office of Energy Efficiency and Renewable Energy.
Operated for the DOE by The Alliance for Sustainable Energy LLC, NREL is the DOE’s primary national laboratory for renewable energy and energy efficiency R&D.
