Alloying-Induced Structural Transition in the Promising Thermoelectric Compound CaAgSb.

AMX Zintl compounds, crystallizing in several closely related layered structures, have recently garnered attention due to their exciting thermoelectric properties. In this study, we show that orthorhombic CaAgSb can be alloyed with hexagonal CaAgBi to achieve a solid solution with a structural transformation at x ∼ 0.8. This transition can be seen as a switch from three-dimensional (3D) to two-dimensional (2D) covalent bonding in which the interlayer M-X bond distances expand while the in-plane M-X distances contract. Measurements of the elastic moduli reveal that CaAgSb1-xBix becomes softer with increasing Bi content, with the exception of a steplike 10% stiffening observed at the 3D-to-2D phase transition. Thermoelectric transport measurements reveal promising Hall mobility and a peak zT of 0.47 at 620 K for intrinsic CaAgSb, which is higher than those in previous reports for unmodified CaAgSb. However, alloying with Bi was found to increase the hole concentration beyond the optimal value, effectively lowering the zT. Interestingly, analysis of the thermal conductivity and electrical conductivity suggests that the Bi-rich alloys are low Lorenz-number (L) materials, with estimated values of L well below the nondegenerate limit of L = 1.5 × 10-8 W Ω K-2, in spite of the metallic-like transport properties. A low Lorenz number decouples lattice and electronic thermal conductivities, providing greater flexibility for enhancing thermoelectric properties.

Computational design of thermoelectric alloys through optimization of transport and dopability.

Alloying is a common technique to optimize the functional properties of materials for thermoelectrics, photovoltaics, energy storage etc. Designing thermoelectric (TE) alloys is especially challenging because it is a multi-property optimization problem, where the properties that contribute to high TE performance are interdependent. In this work, we develop a computational framework that combines first-principles calculations with alloy and point defect modeling to identify alloy compositions that optimize the electronic, thermal, and defect properties. We apply this framework to design n-type Ba2(1-x)Sr2xCdP2 Zintl thermoelectric alloys. Our predictions of the crystallographic properties such as lattice parameters and site disorder are validated with experiments. To optimize the conduction band electronic structure, we perform band unfolding to sketch the effective band structures of alloys and find a range of compositions that facilitate band convergence and minimize alloy scattering of electrons. We assess the n-type dopability of the alloys by extending the standard approach for computing point defect energetics in ordered structures. Through the application of this framework, we identify an optimal alloy composition range with the desired electronic and thermal transport properties, and n-type dopability. Such a computational framework can also be used to design alloys for other functional applications beyond TE.

Predicting energy and stability of known and hypothetical crystals using graph neural network.

The discovery of new inorganic materials in unexplored chemical spaces necessitates calculating total energy quickly and with sufficient accuracy. Machine learning models that provide such a capability for both ground-state (GS) and higher-energy structures would be instrumental in accelerated screening. Here, we demonstrate the importance of a balanced training dataset of GS and higher-energy structures to accurately predict total energies using a generic graph neural network architecture. Using ∼ 16,500 density functional theory calculations from the National Renewable Energy Laboratory (NREL) Materials Database and ∼ 11,000 calculations for hypothetical structures as our training database, we demonstrate that our model satisfactorily ranks the structures in the correct order of total energies for a given composition. Furthermore, we present a thorough error analysis to explain failure modes of the model, including both prediction outliers and occasional inconsistencies in the training data. By examining intermediate layers of the model, we analyze how the model represents learned structures and properties.