UC Davis

Thermoelectric generators (TEGs), which are fabricated from thermoelectric materials, convert heat into electricity with no moving parts, making them a robust and appealing renewable energy technology, given that approximately two thirds of the energy produced on earth is lost as waste heat. TEGs could be employed to scavenge heat from industrial plants and combustion engines to boost the efficiency of existing processes, or they could be used independently to generate usable electricity from geothermal sources and the sun. The efficiency of a TEG is dependent on the thermoelectric figure of merit, zT, of the materials used to create the device. zT is determined using a material’s Seebeck coefficient (S), temperature (T), electrical resistivity (ρ), and thermal conductivity (κ), according to the relationship: zT= (S^2 T)/ρκ. To make TEGs attractive for large scale applications, thermoelectric materials with high zT need to be designed. This dissertation investigates the chemistry of several potential thermoelectric materials to gain deeper understanding of structure-property relationships that enable the design of highly efficient materials.Layered Zintl phases are a class of materials that inherently exhibit low thermal conductivity due to their complex crystal structures and chemical tunability. They are composed of covalent networks with diverse bonding motifs separated by layers of cations, and the ionic and covalent portions can be tuned separately to optimize thermal and electronic properties. The directional nature of the covalent bonding can also give rise to low dimensional electronic structures, which are beneficial for thermoelectric properties. The first project described in Chapter 2 of this dissertation is a study on defect chemistry in a solid solution of the layered Zintl phase Yb2-xEuxCdSb2. The goal of this research was to understand the unusually low thermal conductivity, high Seebeck coefficient and low electrical resistivity of this promising thermoelectric material by examining changes in defect chemistry as Eu is substituted for Yb using pair distribution function analysis. The local structures of x = 0.2 and 0.3 deviated substantially from the average structure due to large concentrations of Eu and Sb vacancies that give rise to structural distortions that alter electronic properties. Predictions suggest that zT > 2 at 550 K could be achieved for x = 0.3 if the carrier concentration can be tuned via doping. This work highlights the importance of understanding crystallographic disorder and defect structure in thermoelectric materials. This work was published in ACS Chemistry of Materials (2022, 34, 20, 9228–9239). The next project presented in Chapter 3 focuses on Ba2-xEuxZnSb2, a Zintl phase that contains one-dimensional covalent chains and was predicted to exhibit zT > 2 at 900 K. Eu was substituted for Ba (x = 0.2, 0.3, 0.4) to reduce the air sensitivity of this compound, as well as reduce thermal conductivity through alloy scattering. Samples were prepared using binary precursors via ball milling and annealing and then consolidated by spark plasma sintering for thermoelectric property measurement. All compositions exhibited high Seebeck coefficients and low thermal conductivity, as well as high charge carrier mobility, an important quality for high efficiency thermoelectric materials, which is likely due to the one-dimensional chains. Experimental results agreed well with theory and predictions suggest higher zT is achievable through doping. Eu11Zn¬4Sn2As12 is another layered Zintl phase expected to exhibit a low-dimensional electronic structure due to high mobility hexagonal [Zn2As3]5- sheets in two crystallographic directions and ionic bonding in the third. This project presented in Chapter 4 brings theory and experiment together to understand a complex, disordered, modulated crystal structure, and develop a road map for further optimization of this promising thermoelectric material. Experimental results show high Seebeck coefficient and low thermal conductivity and electrical resistivity that decreases 5-fold with Na-doping. Denisty Functional Theory (DFT)computations show that, when doped, this material possesses a two-dimensional Fermi surface, and band structure calculations highlight the presence of multiple bands just below EF, making Eu11Zn¬4Sn2As12 a promising candidate for further optimization. Finally, theory and experiment are employed together again to characterize the electronic and structural transitions in Yb5Sb3Hx as H content varies from x = 0.25 to 1. Yb5Sb3Hx has been known to exist in two polymorphs, a hexagonal and an orthorhombic crystal structure, and previous work has suggested that H content drives the transition. In the study presented in Chapter 5, DFT calculations confirm that at lower H content (x < 0.25), the hexagonal crystal structure is energetically favored, while at x > 0.25, the orthorhombic cell is more stable and the synthesized polycrystalline samples for x = 0.25, 0.5, 0.75, and 1 support this finding. Previous theoretical work also suggests that Yb5Sb3Hx is an electride with localized anionic electrons present in zero-dimensional cavities when x < 1 and a charge balanced semiconductor at x = 1. Thermoelectric properties were used to map this transition, and they show a shift from the low charge carrier mobility associated with highly localized electride states (x = 0.25 and 0.5) to high mobility associated with free electrons in a semiconductor (x = 0.75 and 1). At x = 1, the compound exhibits high Seebeck coefficients and low thermal conductivity up at 775 K, in addition to excellent charge carrier mobility that make it a potential thermoelectric material if charge carrier concentration can be increased.