Thermoelectric Materials Discovery

Project Term:
03/13/2017 - 09/12/2018

Critical Need:

Greater than 60% of all the energy generated in the United States is lost to the environment as waste heat. Of this, only a negligible fraction of this heat, less than 0.2%, is converted back to electrical power. Harvesting waste heat through solid-state devices offers opportunities for improved energy efficiency of many technologies. Thermoelectric devices directly convert heat into electricity using solid-state materials having no moving parts, thereby lowering the probability of mechanical failure. However, the current materials and methods for manufacturing thermoelectric generators are often not cost effective, due in part to the rigid substrate and expensive semiconductor materials which may limit the application of the technology. Recapturing wasted heat without costly heat transfer systems could be facilitated by the development of inexpensive, scalable thermoelectric devices on flexible substrates.

Project Innovation + Advantages:

The Colorado School of Mines will develop a new method for the high-throughput discovery and screening of thermoelectric materials. The objective is to develop a new class of thermoelectric materials that can enable heat-to-electricity efficiencies greater than 20%. Aerosol spray deposition will be used to collect particles on the solid surfaces, allowing high throughput synthesis with finely tuned composition control. To achieve the thermoelectric performance desired, a tight feedback loop between synthesis, characterization, and theory will be employed to actively guide the design of experiments. To identify materials with high mobility and low thermal conductivity, the team developed metrics that combine experimental and computational training data. These efforts are guided by the team's existing high-throughput calculation database, which has identified specific families of previously unexplored materials with high potential for thermoelectric performance. Over the last two years, these computational methods have been applied to 10,000 compounds, yielding the most extensive database of thermoelectric performance in the world. By considering thousands of compositions within a single structural family, trends in electronic and thermal conductivity emerge that could not have been predicted from a few samples produced with traditional bulk ceramic methods. High-throughput search techniques are particularly critical because the desired qualities are likely to only occur within a narrow chemical composition. The team expects to grow and characterize more than 20 macroscopic samples per day, a significant increase in throughput compared to conventional approaches.


ARPA-E Program Director:
Dr. Joseph King
Project Contact:
Prof. Eric Toberer
Press and General Inquiries Email:
Project Contact Email:


University of Minnesota

Related Projects