Thermoelectric (TE) materials have been focused on in recent years, for the fact that they can convert heat and electrical energy into one another directly and environmentally friendly.^[1–4] In order to be of practical use, TE materials are required to have a high figure of merit (ZT = S²σT/κ, S is Seebeck coefficient, σ is electrical conductivity, T is absolute temperature, and κ is thermal conductivity), at the desired operating temperatures. Excellent performance of TE materials not only need to poses a high power factor (PF = S²σ) but also to show a low thermal conductivity κ. By far, the most widely used TE materials are inter-metallic compounds which have relatively high ZT values, such as Bi₂Te₃, PbTe, and SiGe.^[5–13] However, these materials also have some limitations, such as high costs, instability at high temperatures, and dependence on rare or toxic elements. Oxide TE materials may be an alternative to solve these problems, however, their ZT values are still far from practical applications.^[14,15] Therefore, it is urgent and meaningful to enhance the TE performance of oxide materials.

Recently, nonstoichiometric tungsten-bronze- (TB)-structured Sr_xBa_1−xNb₂O_6−δ (SBN) materials were found to be promising n-type oxide thermoelectric materials.^[16–21] It was reported that heavily reduced Sr_0.61Ba_0.39Nb₂O_6−δ (SBN61) single crystal shows a high thermoelectric power factor (~2000 μW/K²·m at 516 K) in the direction parallel to the c axis.^[22] As shown in Fig. 1, TB structure consists of a complex array of distorted BO₆ octahedrons sharing corners in such a way that three types of inter-sites (A₁, A₂, and C) are available for different cations with a general formula of (A₁)₄(A₂)₂B₁₀C₄O₃₀. The inter-sites could be occupied with different ions, such as K, Na, Sr, Ba, Pb, and so on, and different materials are composed with different occupation contents and different distortions of BO₆ octahedrons, for example, the piezoelectric/ferroelectric materials K₃Li₂Nb₅O₁₅^[23,24] and PbNb₂O₆.^[25,26] In (Sr, Ba)Nb₂O₆, only five out of six A-sites are occupied by Sr ion or Ba ion.^[27–30] According to previous investigations, SBN ceramics show lower electrical conductivity and thus lower power factor as compared with SrTiO₃-based TE oxides, especially at low temperature region near room temperature.^[31,32] In order to enhance electrical conductivity, lithium ions are doped into the empty inter-sites of SBN in the present paper. Sr_0.70Ba_0.30Nb₂O_6−δ (SBN70) is selected as a starting component because of its highest ZT values among Sr_1−xBa_xNb₂O₆ ceramics.^[33] And thermoelectric properties of Li-doped specimens Sr_0.70Ba_0.30Li_xNb₂O₆ (x = 0, 0.03, 0.05, 0.10) were investigated and the influences of different doping contents were discussed. It was found that the TE power factor of SBN70 ceramics could be enhanced by slight lithium doping, and reaches a maximum (486 μW/K²·m at 1073 K). And the ZT value reaches a maximum (about 0.21 at 1073 K).

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Crystalline structure of SrxBa1−xNb2O6.

Sr_0.70Ba_0.30Li_xNb₂O₆ ceramics were prepared by solid-state reaction techniques using SrCO₃ (in purity 99.8%), BaCO₃ (99.8%), Nb₂O₅ (99%), and Li₂CO₃ (99.8%) powders. Appropriate components of the starting materials were ground and pressed into pellet discs. These discs were calcined at 1400 °C for 6 hours in an air atmosphere. After intermediate grinding and pressing, the discs were sintered at 1300 °C for 6 hours in a forming gas (5-mol% hydrogen in argon) with a flow rate of 0.2 l/min. The phase structures were investigated using powder x-ray diffraction (XRD) with a Rigaku D/MAX-2550P diffractometer using Cu Kα radiation (λ = 0.154056 nm). The thermal conductivity was calculated from the thermal diffusivity, specific heat capacity, and sample density as measured on a laser flash apparatus (Netzsch LFA 427), a thermal analyzer (Netzsch DSC 200F3), and by Archimedes'method, respectively. The sintered discs were cut into rectangular columns (12 mm × 2 mm × 2 mm) to measure Seebeck coefficient and electrical conductivity using a Linseis LSR-3/1100 instrument in a helium atmosphere by a modified dynamic method.^[31] The oxygen vacancies are roughly measured by a self-made thermogravimetric apparatus, neglecting the influences of NbO₂ second phase.

The XRD data for the Sr_0.70Ba_0.30Li_xNb₂O₆ samples are shown in Fig. 2. x = 0, x = 0.03, x = 0.05, and x = 0.10 indicate the samples doped with lithium in the contents of x = 0, 0.03, 0.05, and 0.10, respectively. The profiles of XRD patterns for all samples include every diffraction peak of tungsten bronze-structured strontium barium niobate. The secondary peaks of NbO₂ are observed in all samples because of the heavily oxygen reduction.^[20] The lattice parameters and theoretical densities are calculated from the XRD data and shown in Table 1. After lithium doping, the lattice parameters of the samples are almost unchanged. All the samples show high densities that are more than 96% of the theoretical ones.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) The x-ray diffraction patterns for lithium-doped Sr0.70Ba0.30LixNb2O6 ceramics.

Table 1.

Table 1.

Table 1. Lattice parameters (a,b,c), theoretical densities (ρth), measured densities (ρm), and relative densities (ρr) of samples, contents of oxygen vacancies (δ). . x a = b/Å c/Å ρth/(g/cm3) ρm/(g/cm3) ρr/% δ 0 12.465 3.924 5.23 5.13 98.1 0.07 0.03 12.471 3.924 5.23 5.06 96.8 0.08 0.05 12.464 3.922 5.24 5.17 98.7 0.11 0.10 12.465 3.919 5.25 5.08 96.8 0.10

Table 1. Lattice parameters (a,b,c), theoretical densities (ρth), measured densities (ρm), and relative densities (ρr) of samples, contents of oxygen vacancies (δ). .

Figures 3(a) and 3(b) show the Seebeck coefficient and electrical conductivity as a function of temperature respectively. For all samples, the Seebeck coefficient increases with increasing temperature in the low temperature range and then presents a transition around 600 K, after which S decreases with increasing temperature in the high temperature range. The electrical conductivity σ increases with increasing temperature in the low temperature region, showing a semiconductor behavior. σ presents a transition around 700 K, and decreases tardily with increasing temperature in high temperatures, which presents a metal-like behavior. The two transitions appear in different temperatures in S and σ, showing that the origins are different. According to our previous studies, the transport properties are dominated by the polaron hopping behavior below 600 K and the Anderson localization behavior at high temperatures above 600 K.^[34]

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) Temperature dependence of Seebeck coefficient S (a) and electrical conductivity σ (b) of the samples Sr0.70Ba0.30LixNb2O6.

After lithium doping, the absolute Seebeck coefficient decreases and electrical conductivity increases, as compared with those the un-doped sample. It indicates that carrier concentration is enhanced by lithium doping. However, both of the magnitudes of S and σ vary non-monotonically but synchronously with the doping contents, indicating that doped lithium ions may not be fully ionized and oxygen vacancy may also contribute to carriers. Therefore, the concentration of oxygen vacancies were measured by thermogravimetric analysis, and the δ values in the formula Sr_0.70Ba_0.30Li_xNb₂O_6−δ were calculated and shown in Table 1. The dependence of δ values on the content of Li-doping is consistent with that of electrical conductivity, indicating that oxygen vacancy plays a very important role on carrier transport.

According to the measured S and σ, the thermoelectric power factor (PF) of samples with four doping contents are calculated and shown in Fig. 4. The PF values are enhanced due to lithium doping, especially at high temperatures. And the highest PF value (486 μW/K²·m at 1073 K) is obtained in the sample x = 0.05.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Temperature dependence of thermoelectric power factor (PF = S2σ) of the samples Sr0.70Ba0.30LixNb2O6.

Figure 5(a) shows the temperature dependence of thermal conductivity κ. The magnitude of κ increases after lithium doping in the high temperature region. Thermal conductivity includes two parts: lattice thermal conductivity κ_l and electronic thermal conductivity κ_e. The values of κ_e are calculated from the electrical conductivity according to the Wiedemann–Franz relation, κ_e = LTσ (where L is Lorenz number and estimated from the Seebeck coefficients.^[4]). Thus, the higher electrical conductivities of the doped samples lead to higher values of κ_e, and κ_e at 1073 K increases from 0.13 W/K·m of the un-doped sample to 0.33 W/K·m of the sample for x = 0.05.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) Temperature dependence of total thermal conductivity (a), lattice thermal conductivity (b), thermal diffusivity (c), and heat capacity (d) of the samples Sr0.70Ba0.30LixNb2O6.

Lattice thermal conductivity κ_l is calculated from κ subtracting κ_e, as shown in Fig. 5(b). The magnitude of κ_l increases firstly and then decreases when the doping content increases. To show the thermal transport clearly, thermal diffusivity D and heat capacity C_p are also shown in Fig. 5. Both of the magnitudes of D and C_p are enhanced after Li-doping, especially at high temperatures. The enhancement of heat capacity might be related to the light mass of doped lithium ions. On the other hand, thermal diffusivity increases firstly and then decrease when the doping content increases, indicating competing influence factors. The increase of electronic thermal conductivity should contribute the increase of thermal diffusivity. And doped ions are generally regarded as phonon scattering centers, which should result in the reduction of thermal diffusivity. As shown in Fig. 5(c), thermal diffusivity of un-doped sample decreases with increasing temperature in the high temperature region which may be related to the interaction between phonons. And after lithium doping, the decrease trend is depressed, indicating that the interaction between phonons would be weakened. The origin still needs further investigations. One possibility is the change of the distortion of NbO₆ octahedrons.

Figure 6 shows the thermoelectric figure of merit ZT of the Sr_0.7Ba_0.3Li_xNb₂O_6−δ samples. ZT values of all samples increase in the whole temperature region and still keep the upward trend at 1073 K, indicating a higher ZT value at higher temperatures. Although the power factor PF is enhanced obviously, the thermal conductivity is also enhanced by lithium doping. Therefore, ZT increases slightly and reaches a maximum value (0.21 at 1073 K) in the sample Sr_0.70Ba_0.30Li_0.10Nb₂O₆.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (color online) Temperature dependence of thermoelectric figure of merit ZT of the samples Sr0.70Ba0.30LixNb2O6.

In this work, the thermoelectric properties of Li-doped Sr_0.70Ba_0.30Nb₂O_6−δ ceramics were investigated in the temperature range from 323 K to 1073 K. The profiles of XRD patterns for all samples include every diffraction peak of tungsten bronze-structured strontium barium niobate and the lattice parameters of the samples are almost unchanged. After lithium doping, the non-monotonic and synchronous changes of S and σ indicate that doped lithium ions may not be fully ionized and oxygen vacancy also contribute to carriers. The lattice thermal conductivity increases firstly and then decreases when the doping content increases, which attributes to the competing factors which influence thermal diffusivity and heat capacity. The thermoelectric performance is enhanced by lithium doping due to the increase of the power factor (with a maximum of 486 μW/K²·m). The thermoelectric figure of merit reaches maximum value (0.21 at 1073 K) in the sample Sr_0.70Ba_0.30Li_0.10Nb₂O₆.