Doctor of Philosophy
Institute for Superconducting & Electronic Materials
Jood, Priyanka, Novel thermoelectric oxides for high temperature power generation, Doctor of Philosophy thesis, Institute for Superconducting & Electronic Materials, University of Wollongong, 2012. http://ro.uow.edu.au/theses/3600
Oxide based materials are believed to be the most eco-friendly materials for thermoelectric (TE) power generation applications. However, their efficiency is still very low as compared to the conventionally used semiconductor based thermoelectric materials. ZnO is a promising high figure-of-merit (ZT) thermoelectric material for power harvesting from heat due to its high melting point, high electrical conductivity (σ) and Seebeck coefficient (α). In this thesis, the main emphasis was put onto improving TE properties of ZnO by exploring various synthesis techniques and dopant elements. The microstructural and electrical properties of ZnO in its bulk form, void structured, and nanostructured forms have been thoroughly studied. Finally, one of the most efficient layered cobaltite oxides, namely Ca3Co4O9, was synthesized and studied. Pure and Bi doped Ca3Co4O9 was prepared as a Stranski-Krastanov (SK) thin film to evaluate the possibilities of obtaining high power factors in this material.
Ball milling is a well-known technique to improve the quality of various materials. Therefore, ball milling was employed as one of the ways to synthesize pure and Ga doped ZnO bulk ceramics. The optimization of ball milling conditions was done by exploring three different ball milling regimes which were low energy (200 rpm) dry milling, high energy (750 rpm) dry milling, and high energy wet milling. Samples fabricated using high energy wet ball milling (HE) were found to form highly dense ZnO matrix percolated by voids of ~900 nm average diameter. EDS analysis confirmed the presence of oxygen vacancies at the grain-void interfaces which is argued to be the main factor responsible for the increase of electron concentration in the system. The voids being electrically charged were expected to create a potential barrier, hence, resulting in energy filtering. This phenomenon was expected to contribute towards high Seebeck coefficient of -134 μVK-1 at 300 K and -185 μVK-1 at 1000 K for Ga doped ZnO, which is ~30-45% higher than the values previously reported for the ZnO:Ga system. The voids also helped in reducing thermal conductivity by behaving as extra phonon scattering centers, e.g. κ300 K for high energy wet ball milled Zn0.995Ga0.005O sample was ~40 % lower than that for the low energy wet ball milled Zn0.995Ga0.005O sample which did not have any voids. The highest power factor achieved for our high energy wet ball milled samples was 3.7 × 10-4 Wm-1K-2 at 1000 K for Zn0.995Ga0.005O.
Although ZnO is a material with tremendous potential, its practical use is limited by high lattice thermal conductivity κL. One of the aims was to overcome this problem by creating phonon scattering centres in the ZnO matrix, while maintaining high power factors. In this work, undoped and Al doped ZnO nanocrystals were synthesized using microwave mediated technique and then sintered to form nanostructured pellets. Through this process we obtained Al-containing ZnO nanocomposites with up to a factor of twenty lower κL than non-nanostructured ZnO while retaining bulk-like α and σ. We showed that enhanced phonon scattering, promoted by Al-induced grain growth inhibition and ZnAl2O4 nanoprecipitates, resulted in ultralow κ~2 Wm-1K-1 at 1000 K. High α~-300 μVK-1and high σ~1-104 Ω-1m-1 at 1000 K were achieved via offsetting of the nanostructuring-induced mobility decrease by high, but non-degenerate, carrier concentrations obtained through excitation from shallow Al donor states. The resultant projected ZT~0.44 at 1000 K was 50% higher than the best non-nanostructured counterpart at the same temperature.
After successfully obtaining Al doped nanostructured ZnO, we investigated indium (group III) and bismuth (group V) doping in ZnO through microwave synthesis. In doping lead to a high power factor of α2σ300 K~0.79×10-4 Wm-1K-2 and α2σ1000 K~12.3×10-4 Wm-1K-2, which is about ~32 % higher than Al doped ZnO. We believe that this is mainly due to the higher electrical conductivity of the samples. An effective indium substitution in Zn site was observed with no presence of secondary phase. Also, owing to the increased porosity of the samples due to In doping, a ~25% decrease in thermal conductivity was observed (compared to the pure ZnO). However, due to the absence of any secondary phase nanoprecipitates the κ300 K values were ~50 % higher than the measured κ300 K for Al doped ZnO. Bi doping, on the other hand, did not prove to be very beneficial for ZnO due to the ineffective substitution of Bi in the Zn sites. This was strongly supported by the fact of significantly lower electrical conductivity, e.g. ~16.7 Ω-1m-1 at 300 K. However, huge thermopower (-484 μV/K to -529 μV/K) to some extent compensated such low electrical conductivity. Low carrier concentrations in Bi doped ZnO were believed to be a result of the defect complexes formed by Bi with other existing point defects.
To take the benefits from both dopants, i.e. lower thermal conductivity due to nano-inclusions arising from Al doping and high power factor arising from efficient indium doping, co-doping of Al and In was employed in ZnO. In this work, various batches of co-doped ZnO with different levels of dopant concentrations were synthesized and studied. The morphology and microstructure of the co-doped ZnO pellets was found to be dependent not only on the ratio of doping levels of the two dopants, but also on the extent of total doping in the samples. Co-doping with total doping levels ≥ 1at.% proved to be beneficial in improving thermoelectric properties of this system by increasing the solubility limit of the dopants. The carrier scattering mechanism in these co-doped samples was found to be more influenced by indium rather than aluminum, with optical phonon scattering found to be the dominant scattering mechanism. Co-doped sample with 1 at. % Al and 1.5 at.% In (A1I15) was found to exhibit the highest carrier concentration (~8.3×1018 cm-3) and, hence, the highest room temperature electrical conductivity (~1.49×103 Ω-1m-1) among all our single doped and co-doped samples. Highest power factor of α2σ300 K~0.86×10-4 Wm-1K-2 was obtained for A1I15 sample. For all of the microwave synthesized ZnO samples, the monotonic increase in the α2σ300K with electron concentration n suggested the possibility of further enhancement of power factor and ZT by optimizing the doping levels at high electron concentrations n > 1018 cm-3. This work confirmed that ZnO has very high potential to be an efficient thermoelectric material in the future. Also some critical paths for further explorations have been identified and highlighted.
Lastly, thin films of Ca3Co4O9 and Ca2.8Bi0.2Co4O9 were deposited on LaAlO3 substrate through pulsed laser deposition technique (PLD). The growth mode of the films was confirmed to be Stranski-Krastanov (SK) where 3D cluster formation takes place over a thin layer (~14 nm in our case) of epitaxial film. We found that due to the Bi doping the amplitude of the undulations was reduced, which in turn restricted the ripening of the clusters. This effect resulted in a large number of smaller size clusters (about 68 % smaller in diameter) for Ca2.8Bi0.2Co4O9 thin film as compared to its pure counterpart. Unlike earlier reported trends for Bi doped Ca3Co4O9 system, measured electrical conductivity of Ca2.8Bi0.2Co4O9 (~388 Ω-1m-1) was lower than for the pure Ca3Co4O9 (~705 Ω-1m-1). We found that this was caused by the presence of comparatively large number of clusters which also represent areas of dislocations and incomplete nucleation. These imperfections obviously act as carrier scattering sites. Low electrical conductivities of the thin films were however compensated by high Seebeck coefficient of 136 μVK-1 for Ca3Co4O9 and 163 μVK-1 for Ca2.8Bi0.2Co4O9. These values are among the highest ever reported for Ca3Co4O9 thin films. The power factor measured at room temperature was ~0.13x10-4 Wm-1K-2. We believe that these values can be further improved by optimizing cluster density and improving the texture of the films.