Enhanced Thermoelectric Performance of Te-Doped Bi 2 Se 3 − x Te x Bulks by Self-Propagating High-Temperature Synthesis

: Polycrystalline Bi 2 Se 3 − x Te x (x = 0~1.5) samples were prepared by self-propagating high-temperature synthesis (SHS) combined with spark plasma sintering (SPS) and their thermoelectric properties were investigated. The SHS-SPS process can shorten the time with few energy consumptions, and obtain almost pure Bi 2 Se 3 -based phases. Consequently, the Se vacancies and anti-site defects contribute to the converged carrier concentration of ~2 × 10 19 cm − 3 while the increased carrier effective mass enhances the Seebeck coefficient to more than − 158 µ V K − 1 over the entire temperature range. The lattice thermal conductivity is suppressed from 1.07 Wm − 1 K − 1 for the pristine specimen to ~0.6 Wm − 1 K − 1 for Te-substitution samples at 300 K because of point defects caused by the difference of mass and size between Te and Se atoms. Coupled with the enhanced power factor and reduced lattice thermal conductivity, a high ZT of 0.67 can be obtained at 473 K for the Bi 2 Se 1.5 Te 1.5 sample. Our results reveal that Te-substitution based on the SHS-SPS method is highly-efficient and can improve the thermoelectric properties of Bi 2 Se 3 -based materials largely. − (x method. In work, we successfully synthesized Bi 2 Se 3 − x Te x (x = 0, 0.3, 0.6, 0.9, 1.2, 1.5) via the SHS method followed by spark plasma sintering (SPS) and studied the thermoelectric properties from 300 K to 593 K. Our results show that the highest power factor ( PF = S 2 σ ) can achieve 11.2 µ Wcm − 1 K − 2 for Bi 2 Se 1.5 Te 1.5 at 300 K and the lattice thermal conductivity ( κ L ) could be reduced to the lowest value of 0.35 Wm − 1 K − 1 at 593 K via Te alloying for Bi 2 Se 2.1 Te 0.9 . The ZT of ~0.67 is finally achieved at 473 K for Bi 2 Se 1.5 Te 1.5 , demonstrating the potential application for energy conversion in the mid-temperature. And SHS


Introduction
With increasing attention on the environmental protection and renewable resources, thermoelectric (TE) instruments, which can directly convert heat into electricity, are considered as a potential solution for harness waste heat [1][2][3]. Considerable numbers of efforts have been devoted to improving the energy conversion efficiency and the stability of the TE materials [4]. The conversion efficiency depends positively on the dimensionless figure of merit, ZT = S 2 σT/κ, where S is the Seebeck coefficient, σ is the electrical conductivity, κ is the thermal conductivity and T is the absolute temperature, respectively [5]. To maximize the ZT value of a kind of material, a large Seebeck coefficient, electrical conductivity and low thermal conductivity are needed. However, these parameters have a strong coupling with each other, which makes it a challenging task to enhance ZT significantly. Motivated by achieving high thermoelectric performance, multiple methods have been adopted [6]. Many studies so far have focused on the atomic or molecular scales such as doping or alloying to enhance carrier concentration or carrier mobility and thus electrical conductivity to optimize TE performance [7][8][9][10]. To enhance Seebeck coefficient while maintaining high electrical conductivity, manipulating the band structure offers a new guideline [11]. Meanwhile, effective alteration at nanometer or mesoscopic scales including the quantum confinement [12] and energy filtering effect [13] can drastically elevate electrical properties.

Experimental Procedures
In the initial stage, Bi (99.99%, Aladdin), Se (99.99%, Aladdin), and Te (99.99%, Aladdin) powders were mixed meticulously in stoichiometric amounts. Then the mixture was cold-pressed into pellets with the diameter of 20 mm. The SHS process was started by heating the bottom of the pellets with a hand torch in the air. Once ignited, the hand torch is removed immediately. The heat generated by the combination reaction kept the combustion process propagating until it was finished in several seconds. Then the pellets were grounded into fine powders carefully by hand. The powders were then sintered into pellets of φ 12.7 mm by SPS (Sumitomo Coal Mining Co., Ltd., Tokyo, Japan) at the temperature of 593 K for 5 min under a uniaxial pressure of 40 MPa.
The phase structures were investigated by X-ray diffraction (XRD, RINT2000, Rigaku, Tokyo, Japan) analysis. The morphology and composition of cross-sectional bulks were checked by field-emission scanning electron microscopy (FESEM) (LEO1530, Oxford Instruments, Oxford, UK). The electrical properties including electrical conductivity and Seebeck coefficient were measured from room temperature to 593 K by ZEM-3 (ULVAC, Kanagawa, Japan). The van der Pauw method was used in an Eastchanging Hall measurement station to measure Hall coefficient (R H ). The carrier concentration (n) and mobility (µ) were estimated by the equation n = 1/eR H and µ = σR H . To ensure the accuracy, the samples were polished to be thinner than 0.5 mm for the measurements. The total thermal conductivity is determined by the equation κ = DC p ρ, where D is thermal diffusivity, C p is specific heat and ρ is the density of the bulks. The thermal diffusivity was obtained by the laser flash method and the specific heat was calculated by the Dulong-Petit relation. The density of the bulks was derived with Archimedes method. Figure 1a is the XRD result of all the Bi 2 Se 3−x Te x samples with x = 0~1.5. All the major peaks in the XRD patterns correspond to a standard card, Bi 2 Se 3 , PDF #33-0214. The additional small peaks can be identified as Bi 2 O 2 Se (PDF #29-0237), which is possibly generated by oxidation during the ultra-fast combustion process in the air. In this work, we assume that all the samples contain the same amount of Bi 2 O 2 Se, and we neglect the effect of existence of Bi 2 O 2 Se due to its small amount (small peaks in the XRD result). In Figure 1b, the lattice parameters were calculated according to the position of XRD peaks. With increasing Te content, the lattice constants a and c increase linearly, which indicates Te can successfully substitute for Se atoms to form solid solution by SHS process in a short time. Figure 1c-e show the morphology of cross-sectional bulks (x = 0, 0.3, 1.5). All the samples were sintered well with high density (94% or above). The layer structure can be seen clearly in the Bi 2 Se 3−x Te x bulks.  Figure 1a is the XRD result of all the Bi2Se3−xTex samples with x = 0~1.5. All the major peaks in the XRD patterns correspond to a standard card, Bi2Se3, PDF #33-0214. The additional small peaks can be identified as Bi2O2Se (PDF #29-0237), which is possibly generated by oxidation during the ultra-fast combustion process in the air. In this work, we assume that all the samples contain the same amount of Bi2O2Se, and we neglect the effect of existence of Bi2O2Se due to its small amount (small peaks in the XRD result). In Figure 1b, the lattice parameters were calculated according to the position of XRD peaks. With increasing Te content, the lattice constants a and c increase linearly, which indicates Te can successfully substitute for Se atoms to form solid solution by SHS process in a short time. Figure 1c-e show the morphology of cross-sectional bulks (x = 0, 0.3, 1.5). All the samples were sintered well with high density (94% or above). The layer structure can be seen clearly in the Bi2Se3−xTex bulks.  Figure 2 shows the temperature dependence of electrical conductivity and Seebeck coefficient. The electrical conductivity σ (in Figure 2a) of pristine Bi2Se3 maintains at about 400 Scm −1 from 300 K to 593 K, which is much higher than Bi2Se3 prepared by other method [22]. Se is much easier to evaporate during the combustion process because of low energy of evaporation and thus it leaves Se vacancies and free electrons, which may contribute to higher electrical conductivity. This can be indicated in the following equation:

Results and Discussion
As the Te content increases (x > 0), the electrical conductivity at 300 K initially increases to ~870 Scm −1 because of increased carrier concentration (Table 1), then decreases to ~400 Scm −1 owing to the change of carrier mobility, which is much lower than the pristine Bi2Se3 ( Table 1). The carrier concentration increases may be a result of increasing anti-site defects ( − Te Bi ) [31], which is caused by the fact that Bi can jump from Bi-site to Te-site easily because of small difference in electronegativity [10], as indicated in Equation (2).  Figure 2 shows the temperature dependence of electrical conductivity and Seebeck coefficient. The electrical conductivity σ (in Figure 2a) of pristine Bi 2 Se 3 maintains at about 400 Scm −1 from 300 K to 593 K, which is much higher than Bi 2 Se 3 prepared by other method [22]. Se is much easier to evaporate during the combustion process because of low energy of evaporation and thus it leaves Se vacancies and free electrons, which may contribute to higher electrical conductivity. This can be indicated in the following equation: As the Te content increases (x > 0), the electrical conductivity at 300 K initially increases to~870 Scm −1 because of increased carrier concentration (Table 1), then decreases to~400 Scm −1 owing to the change of carrier mobility, which is much lower than the pristine Bi 2 Se 3 ( Table 1). The carrier concentration increases may be a result of increasing anti-site defects (Bi − Te ) [31], which is caused by the fact that Bi can jump from Bi-site to Te-site easily because of small difference in electronegativity [10], as indicated in Equation (2).
On the contrary, the number of Se vacancies V 2+ Se will be fewer due to the increasing Te content. The decreased Se vacancies and increased anti-site defects make the carrier concentration converges to about 2 × 10 19 cm −3 . The carrier mobility decreases with higher Te content at 300 K in general due to the enhanced alloy scattering. Interestingly, we found µ of Bi 2 Se 1.5 Te 1.5 (x = 1.5) is slightly larger than Bi 2 Se 1.8 Te 1.2 (x = 1.2) at the room temperature, which is possibly caused by the intrinsic high mobility of Bi 2 Te 3 [31]. Table 1. Actual composition, carrier concentration (n), carrier mobility (µ), carrier effective mass (m*), Seebeck coefficient (S), Lorenz constant (L), lattice thermal conductivity (κ L ), κ L /κ, and density of Bi 2 Se 3−x Te x samples at 300 K. As shown in Figure 2b, the negative Seebeck coefficient of all the samples indicates the dominance of electrons in the transport process. Generally, the value of Seebeck coefficient can be estimated by the equation [32]: where e, k B , T, h, m * d , and n are the carrier charge, Boltzmann constant, absolute temperature, Planck constant, the effective mass of the carrier, and carrier concentration. As shown in the formula, because of largely enhanced carrier concentration with increasing Te content (x < 0.9) at 300 K, the Seebeck coefficient decreases. Then the Seebeck coefficient was improved due to the larger carrier effective mass at 300 K (Table 1). It should be noticed that each sample with x ≥ 0.9, as the temperature increases, the value of Seebeck coefficient first increases then decreases, which is caused by the intrinsic excitations. The highest Seebeck coefficient of −180 µV K −1 is achieved at 473 K for the Bi 2 Se 1.5 Te 1.5 sample. The Bi 2 Se 1.5 Te 1.5 sample attains the largest effective mass of~0.60 m 0 , which is in accord with the difference of Seebeck coefficient with different Te contents at room temperature. On the contrary, the number of Se vacancies + 2 Se V will be fewer due to the increasing Te content.
The decreased Se vacancies and increased anti-site defects make the carrier concentration converges to about 2 × 10 19 cm −3 . The carrier mobility decreases with higher Te content at 300 K in general due to the enhanced alloy scattering. Interestingly, we found μ of Bi2Se1.5Te1.5 (x = 1.5) is slightly larger than Bi2Se1.8Te1.2 (x = 1.2) at the room temperature, which is possibly caused by the intrinsic high mobility of Bi2Te3 [31]. Table 1. Actual composition, carrier concentration (n), carrier mobility (μ), carrier effective mass (m*), Seebeck coefficient (S), Lorenz constant (L), lattice thermal conductivity (κL), κL/κ, and density of Bi2Se3−xTex samples at 300 K.  As shown in Figure 2b, the negative Seebeck coefficient of all the samples indicates the dominance of electrons in the transport process. Generally, the value of Seebeck coefficient can be estimated by the equation [32]: where e , B k , T , h , * d m , and n are the carrier charge, Boltzmann constant, absolute temperature, Planck constant, the effective mass of the carrier, and carrier concentration. As shown in the formula, because of largely enhanced carrier concentration with increasing Te content (x < 0.9) at 300 K, the Seebeck coefficient decreases. Then the Seebeck coefficient was improved due to the larger carrier effective mass at 300 K (Table 1). It should be noticed that each sample with x ≥ 0.9, as the temperature increases, the value of Seebeck coefficient first increases then decreases, which is caused by the intrinsic excitations. The highest Seebeck coefficient of −180 μV K −1 is achieved at 473 K for the Bi2Se1.5Te1.5 sample. The Bi2Se1.5Te1.5 sample attains the largest effective mass of ~0.60 m0, which is in accord with the difference of Seebeck coefficient with different Te contents at room temperature. The variation of power factor (PF = S 2 σ) with increasing temperature of all the samples is shown in Figure 3. The Bi2Se1.5Te1.5 sample reaches the highest PF of 11.2 μWcm −1 K −2 at room temperature, The variation of power factor (PF = S 2 σ) with increasing temperature of all the samples is shown in Figure 3. The Bi 2 Se 1.5 Te 1.5 sample reaches the highest PF of 11.2 µWcm −1 K −2 at room temperature, which is almost twice higher than that of pristine Bi 2 Se 3 . But it drops to about 8 µWcm −1 K −2 at 593 K owing to the decreased electrical conductivity and Seebeck coefficient.
Crystals 2017, 7, 257 5 of 8 which is almost twice higher than that of pristine Bi2Se3. But it drops to about 8 μWcm −1 K −2 at 593 K owing to the decreased electrical conductivity and Seebeck coefficient.  Figure 4a illustrates the total thermal conductivity (κ) as a function of temperature from room temperature to 593 K. The κ of pristine Bi2Se3 is in the range of 1.04-1.31 Wm −1 K −1 . As x increases to 1.2, the κ drops into the range of 0.76-0.82 Wm −1 K −1 from 300 K to 593 K. With further increasing Te content, κ is much larger than the sample of x = 1.2. To have a better understanding of the thermal transport properties, κ is subsequently divided into three parts: where κe is electron thermal conductivity, κL is lattice thermal conductivity and κB is the bipolar thermal conductivity induced by intrinsic excitaions. κe can be estimated by Wiedemann-Franz relation: where L is the Lorenz constant and σ is electrical conductivity. In the single parabolic band model, L depends on the reduced chemical potential and scattering parameter. It can be estimated by fitting the values of the Seebeck coefficient and the room temperature data has been listed in Table 1. The details can be seen elsewhere [33,34]. As mentioned above (Figure 2b), intrinsic excitaions don't occur until 423 K or above. Therefore, the κB can be ignored at low temperature in Figure 4b. Consequently, κL and the reciprocal temperature, T −1 , follow a linear relationship. As shown in Figure 4c, the lattice thermal conductivity drops substantially after alloying. Note that the κL of Bi2Se2.1Te0.9 achieves the lowest value of 0.35 Wm −1 K −1 at 593 K. The effective suppression of the κL of could be attributed to point defects caused by the different mass and size between Te and Se atoms. Similar to the previous literature [9], the κL rises slightly when x is above 0.9. This may be ascribed to the relatively high κL of Bi2Te3 [31], whose effect is larger than the point defects. Figure 4d shows the temperature dependence of κB. With low Te content (x ≤ 0.6), κB is almost zero from 300 K to 593 K, because there are no intrinsic excitaions. Intrinsic excitaions occur and κB's contribution to κ becomes larger when x ≥ 0.9, which is owing to narrower band gap with increasing Te content [35].
The ZT values for all the Bi2Se3−xTex samples (x = 0, 0.3, 0.6, 0.9, 1.2, 1.5) are presented in Figure  5. The enhanced power factor and the effective suppression of lattice thermal conductivity synergistically contribute to the highest ZT value of 0.67 at 473 K for the sample of Bi2Se1.5Te1.5, which is almost twice higher than the pristine Bi2Se3. ZT values of Bi2Se3 (ball milling) [22], Bi2Se2Te (ball milling) [9], Bi2Te3 (SHS) [31] from the literature are included for comparison. Our results show that the ZT value of Te-substituted Bi2Se3-based materials prepared by SHS is much larger than that by other methods.  Figure 4a illustrates the total thermal conductivity (κ) as a function of temperature from room temperature to 593 K. The κ of pristine Bi 2 Se 3 is in the range of 1.04-1.31 Wm −1 K −1 . As x increases to 1.2, the κ drops into the range of 0.76-0.82 Wm −1 K −1 from 300 K to 593 K. With further increasing Te content, κ is much larger than the sample of x = 1.2. To have a better understanding of the thermal transport properties, κ is subsequently divided into three parts: where κ e is electron thermal conductivity, κ L is lattice thermal conductivity and κ B is the bipolar thermal conductivity induced by intrinsic excitaions. κ e can be estimated by Wiedemann-Franz relation: where L is the Lorenz constant and σ is electrical conductivity. In the single parabolic band model, L depends on the reduced chemical potential and scattering parameter. It can be estimated by fitting the values of the Seebeck coefficient and the room temperature data has been listed in Table 1. The details can be seen elsewhere [33,34]. As mentioned above (Figure 2b), intrinsic excitaions don't occur until 423 K or above. Therefore, the κ B can be ignored at low temperature in Figure 4b. Consequently, κ L and the reciprocal temperature, T −1 , follow a linear relationship. As shown in Figure 4c, the lattice thermal conductivity drops substantially after alloying. Note that the κ L of Bi 2 Se 2.1 Te 0.9 achieves the lowest value of 0.35 Wm −1 K −1 at 593 K. The effective suppression of the κ L of could be attributed to point defects caused by the different mass and size between Te and Se atoms. Similar to the previous literature [9], the κ L rises slightly when x is above 0.9. This may be ascribed to the relatively high κ L of Bi 2 Te 3 [31], whose effect is larger than the point defects. Figure 4d shows the temperature dependence of κ B . With low Te content (x ≤ 0.6), κ B is almost zero from 300 K to 593 K, because there are no intrinsic excitaions. Intrinsic excitaions occur and κ B 's contribution to κ becomes larger when x ≥ 0.9, which is owing to narrower band gap with increasing Te content [35].
The ZT values for all the Bi 2 Se 3−x Te x samples (x = 0, 0.3, 0.6, 0.9, 1.2, 1.5) are presented in Figure 5. The enhanced power factor and the effective suppression of lattice thermal conductivity synergistically contribute to the highest ZT value of 0.67 at 473 K for the sample of Bi 2 Se 1.5 Te 1.5 , which is almost twice higher than the pristine Bi 2 Se 3 . ZT values of Bi 2 Se 3 (ball milling) [22], Bi 2 Se 2 Te (ball milling) [9], Bi 2 Te 3 (SHS) [31] from the literature are included for comparison. Our results show that the ZT value of Te-substituted Bi 2 Se 3 -based materials prepared by SHS is much larger than that by other methods.

Conclusions
In summary, we have investigated the thermoelectric properties (300-593 K) of Bi2Se3−xTex samples (x = 0, 0.3, 0.6, 0.9, 1.2, 1.5), which are prepared by SHS-SPS process successfully. Compared with other methods, the SHS-SPS process is much faster and requires less energy, which is desirable in commercial application even though with small amount of second phase Bi2O2Se. Our results show that the power factor of Bi2Se1.5Te1.5 achieves 11.2 μWcm −1 K −2 at 300 K by the increased carrier concentration and the enhancement of Seebeck coefficient. The point defects originate from the difference of mass and size between Te and Se atoms significantly suppresses the lattice thermal conductivity. Benefiting from the improved power factor and the decreased lattice thermal conductivity, a high ZT of 0.67 can be obtained at 473 K for the sample of Bi2Se1.5Te1.5, which demonstrates that the Te-substitution via SHS-SPS method is highly-efficient and can enhance the thermoelectric properties of Bi2Se3-based materials.

Conclusions
In summary, we have investigated the thermoelectric properties (300-593 K) of Bi2Se3−xTex samples (x = 0, 0.3, 0.6, 0.9, 1.2, 1.5), which are prepared by SHS-SPS process successfully. Compared with other methods, the SHS-SPS process is much faster and requires less energy, which is desirable in commercial application even though with small amount of second phase Bi2O2Se. Our results show that the power factor of Bi2Se1.5Te1.5 achieves 11.2 μWcm −1 K −2 at 300 K by the increased carrier concentration and the enhancement of Seebeck coefficient. The point defects originate from the difference of mass and size between Te and Se atoms significantly suppresses the lattice thermal conductivity. Benefiting from the improved power factor and the decreased lattice thermal conductivity, a high ZT of 0.67 can be obtained at 473 K for the sample of Bi2Se1.5Te1.5, which demonstrates that the Te-substitution via SHS-SPS method is highly-efficient and can enhance the thermoelectric properties of Bi2Se3-based materials.

Conclusions
In summary, we have investigated the thermoelectric properties (300-593 K) of Bi 2 Se 3−x Te x samples (x = 0, 0.3, 0.6, 0.9, 1.2, 1.5), which are prepared by SHS-SPS process successfully. Compared with other methods, the SHS-SPS process is much faster and requires less energy, which is desirable in commercial application even though with small amount of second phase Bi 2 O 2 Se. Our results show that the power factor of Bi 2 Se 1.5 Te 1.5 achieves 11.2 µWcm −1 K −2 at 300 K by the increased carrier concentration and the enhancement of Seebeck coefficient. The point defects originate from the difference of mass and size between Te and Se atoms significantly suppresses the lattice thermal conductivity. Benefiting from the improved power factor and the decreased lattice thermal conductivity, a high ZT of 0.67 can be obtained at 473 K for the sample of Bi 2 Se 1.5 Te 1.5 , which demonstrates that the Te-substitution via SHS-SPS method is highly-efficient and can enhance the thermoelectric properties of Bi 2 Se 3 -based materials.