The size and quality of starting powders are important for achieving high thermoelectric performance. The hydrothermal method is suitable to synthesizing high quality nano-powder. In this study, a hydrothermal method (HTS) and a sintering method^[29] are combined to prepare the STO/LaNb–STO composite materials. Pure STO powder was prepared by the hydrothermal method. La and Nb-doped STO precursor was prepared by using ethylene glycol, Ti (OBu)₄, NbCl₅, Sr(NO₃)₂, La(NO₃)₂· 6H₂O, and NaOH. After that, the prepared pure SrTiO₃ powder was placed in this solution. The forming solution was placed in a stainless steel reactor and heated at 180 °C for 24 h, followed by well mixing through an ultrasonic wave. The obtained powders were compacted into a disk by die pressing at 5 MPa and then cold isostatic pressing (CIP) under 20 MPa into the disk. The disk-shape samples were embedded in carbon powders in a corundum crucible, and they were placed into a muffle furnace at 1300 °C for 5-h sintering. After that, the sintered disk samples were sufficiently polished to obtain STO/LaNb–STO composites. The sample preparation process is shown in Fig. 1. Four different composites are STO/La5Nb5–STO (pure SrTiO₃: 5-mol% La and 5-mol% Nb doped SrTiO₃), STO/La5Nb10–STO, STO/La10Nb10–STO, and STO/La10Nb20–STO, respectively, where the mole ratio of pure SrTiO₃ and LaNb–SrTiO₃ is 1:1 in all composites.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Schematic diagram of preparation method.

The compositions and microstructures of the samples were characterized by scanning electron microscopy (SEM, Hitachi SU8010), x-ray diffraction (XRD, Rigaku D/Max-2500) and transmission electron microscopy (TEM, FEI Tecnai G2 F20), respectively. Ratios of Sr, La, Ti, and Nb in the sample were obtained from energy dispersive spectrometers equipped with SEM (EDS/SEM) and TEM (EDS/TEM). The electrical conductivity and the Seebeck coefficient were measured in a helium atmosphere from 300 K to 1100 K by using the Linseis LSR-3. The thermal conductivity (κ) was calculated from the equation κ = DC_pρ, where the thermal diffusivity (D) was measured by the laser flash method using the Netzsch LFA 457, the specific heat (C_p) was measured by a differential scanning calorimeter using a Netzsch DSC STA 449F3, and the density (ρ) was measured by the Archimedes method.

Figure 2(a) shows the x-ray diffraction patterns of the STO/LaNb–STO composite powder samples. The diffraction peaks of powders match with pure cubic perovskite structure, and the widening of the peaks indicates that the particle size of the powder is small. The peaks of doped samples are a little deviated from that of the pure one because the ion radii of La and Nb differ from those of Sr and Ti.^[30] The deviation depends on the La and Nb doping concentrations. In addition, from the analysis of the EDS/TEM for STO/La10Nb20STO powders (Fig. 2(b)), the presence of La and Nb besides the Sr and Ti peaks is also confirmed. These indicate that La and Nb are actually doped into the SrTiO₃ crystal lattice. The TEM image (Fig. 2(c)) of STO/La10Nb20 powder illustrates that the powder is irregularly flaky-shaped and the sizes are approximately 15 nm–25 nm. The high resolution image in Fig. 2(d) reveals that the distance between the adjacent lattice stripes is 0.27 nm, that is consistent with the typical (110) plane of SrTiO₃, which is a further proof of the cubic perovskite structure of the obtained SrTiO₃ powder.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) (a) XRD patterns of STO/LaNb–STO powders, (b) EDS/TEM, (c) TEM image, and (d) high resolution TEM image for STO/La10Nb20 powder.

The XRD patterns of disk samples are shown in Fig. 3(a). It can be seen that a small number of TiO₂ impurity peaks appear in the STO/La10Nb20–STO sample after sintering. A Ti-rich Sr-poor area has been observed from the EDS/SEM mapping image, as shown in Fig. 3(b). Both experimental facts indicate that a second phase of TiO₂ forms in the sintering process, which is in agreement with the results in a previous report.^[31] Figures 3(c)–3(f) show the SEM images of microstructures of four composites. The grain size is about 5 μ–8 μm for all samples, while the grain size decreases with the increasing of doping concentration. Meanwhile, the porosities of samples also tend to increase with doping concentration increasing. The densities of samples decrease from a relatively higher value of 4.9 g/cm³ for the STO/La5Nb5–STO sample to 4.5 g/cm³ for the STO/La10Nb20–STO sample, which may be related to the second phase of TiO₂ leading to the decrease of the density and the increase of the porosity. The occurring of the second phase and its related interface, and the increasing of grain boundary and porosity, would increase phonon scattering, which may be related to the reduction of thermal conductivity.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) (a) XRD patterns of STO/LaNb–STO disks; (b) EDS mapping for STO/La10Nb20–STO disk; SEM images for (c) STO/La5Nb5–STO, (d) STO/La5Nb10–STO, (e) STO/La10Nb10–STO, and (f) STO/La10Nb20–STO.

Figure 4 shows the temperature and doping concentration-dependent Seebeck coefficient (S), electrical conductivity (σ), and power factor (PF) for STO/LaNb–STO samples with different doping concentrations, respectively.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Temperature-dependent (a) Seebeck coefficient, (c) electrical conductivity, and (d) power factor; (b) doping concentration-dependent Seebeck coefficient.

The Seebeck coefficients (Fig. 4(a)) for all samples are negative, indicating that the samples are n-type oxides. The Seebeck coefficient increases with increasing temperature, which is likely to be due to the decrease of the chemical potential for the sample with increasing temperature.^[32] As can be seen from Fig. 4(b) the larger the doping concentration, the lower the value of |S| is. The tendency is in good agreement with the equation of S for degenerate semiconductors,^[29,33] as shown below. 1 where k_B is the Boltzmann constant, e is the electronic charge, and n is the electron concentration. The Seebeck coefficient is inversely proportional to the total carrier concentration. It also indicates that the carrier concentration increases with the increase of doping concentration.^[33]

The Seebeck coefficient of the STO/La5Nb5–STO sample has a maximum value of 350 μV/K at 1000 K, that is the largest value obtained in the present study. The value is higher than the data reported previously for the same doping concentration,^[34] which may suggest that the composites having more complex microstructure can cause interface effects such as energy filtering effect,^[35] etc. The electrical conductivity (Fig. 4(c)) first increases and then decreases with temperature increasing (except for STO/La5Nb5–STO which decreases monotonically) throughout the testing period, which is consistent with the property of a semiconductor. The figure also reveals that the electrical conductivity increases with doping concentration increasing, and its maximum value is 420 S·cm⁻¹ in the STO/La10Nb20–STO sample. In the case of a degenerated STO semiconductor, the carrier concentration is constant in a temperature range from room temperature to 1000 K.^[29] Therefore, the temperature dependence of electrical conductivity is mainly affected by the carrier mobility according to the equation σ = enμ (where e is the electron charge, n is the election concentration, and μ is the carrier mobility). As shown in the inset of Fig. 4(c), at low temperature the electrical conductivity increases proportionally to T^1.2 approximately, especially in high doping concentration, indicating a main contribution from ionized impurity scattering. The decrease of electrical conductivity is proportional nearly to T^−0.6 at high temperatures, which implies a dominant scattering by acoustic phonons.^[29,36] The power factor (Fig. 4(d)) of the sample is calculated from the equation PF = S²σ. It can be seen that the power factor decreases slowly after rapidly rising (except for STO/La5Nb5–STO) with temperature increasing. The tendency is consistent with that of electrical conductivity: as the doping concentration increases, the peak value of power factor shifts to higher temperature. Although the maximum power factor of 1.14 mW·m⁻¹·K⁻² is obtained in the STO/La10Nb10–STO sample at 520 K, from the high temperature above 570 K, the power factor increases with doping concentration increasing. These facts indicate the dominant contribution of electrical conductivity to the increase of PF.

The total thermal conductivity is composed of lattice thermal conductivity (κ_L) and electronic thermal conductivity (κ_e), κ = κ_L + κ_e. The electronic thermal conductivity is calculated from the equation κ_e = LTσ where L is the Lorenz number (2.44 × 10⁻⁸ V² · K⁻²). The total thermal conductivity and the lattice thermal conductivity are shown in Fig. 5. Both of them decrease gradually with the increasing of temperature. At high temperatures, the difference in the total thermal conductivity among samples becomes small, suggesting a dominant phonon scattering by the Umklapp process.^[37] However, with the increase of doping concentration, the lattice thermal conductivity decreases, which may be related to some other reasons of phonon scattering by the interface, defect, porosity, ionized impurity, lattice distortion, etc., as indicated in the above-mentioned microstructure analysis. The thermal conductivities of both STO/La10Nb10–STO and STO/La10Nb20–STO samples decrease to 2.57 W·m⁻¹·K⁻¹ at 1000 K, which is lower than that of mono phase STO bulk material.^[31]

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) Temperature-dependent (a) total thermal conductivities and (b) lattice thermal conductivities.

Finally, the thermoelectric figure-of-merit zT (Fig. 6) is calculated from the equation zT = S²σT/κ. With high PF and low thermal conductivity, a maximum zT value of 0.35 is obtained in the STO/La10Nb20–STO sample at 1000 K, and the value is relatively high compared with those reported previously.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (color online) Temperature-dependent zTs.