(730g) Screening Binary Redox Pairs for Solar Thermochemical Ammonia Synthesis Using Machine Learned Predictions of Gibbs Formation Energies at Finite Temperatures

Authors: 
Bartel, C. J., University of Colorado at Boulder
Rumptz, J. R., University of Colorado at Boulder
Holder, A. M., University of Colorado at Boulder
Weimer, A. W., University of Colorado Boulder
Musgrave, C. B., University of Colorado Boulder
Solar thermochemical ammonia synthesis (STAS) is a reduction/oxidation (redox) cycle which enables the production of ammonia (NH3) from air, water, and concentrated sunlight. In this process, a metal nitride (MN) is oxidized by steam to produce a metal oxide (MO) and NH3; the resulting MO is reduced at high temperature (driven by concentrated solar radiation) and subsequently used to reduce atmospheric nitrogen (N2) and reform the MN and restart the NH3 synthesis cycle. The identification of optimal redox pairs (MO/MN) for this process has been historically limited by the lack of thermochemical data (i.e., Gibbs formation energies at finite temperatures) available for these materials, especially nitrides. Prior work by our group has demonstrated the use of machine learning to enable the prediction of Gibbs formation energies up to very high temperatures (1800 K) using low-cost DFT calculations (e.g., PBE+U), thus eliminating the need for experimentally measured thermochemistry. Utilizing this approach, we’ve screened the reaction energetics and thermodynamic stability of all known binary (i.e., monometallic) MN/MO pairs, increasing the number of redox pairs considered for this process by an order of magnitude. In addition to the consideration of new redox pairs, we also assess the effects of operating conditions and reaction scheme on the viability of candidate materials. Within this work, we gain insights into new candidate materials for NH3 synthesis, the effects of operating conditions on the viability of the overall process, and the correlated stability of metal oxides and metal nitrides.