Liquid-Phase Hot Deformation to Enhance Thermoelectric Performance of n-type Bismuth-Telluride-Based Solid Solutions.

Bismuth-telluride-based solid solutions are the best commercial thermoelectric materials near room temperature. For their n-type polycrystalline compounds, the maximum figures of merit (zTs) are often less than 1.0 due to the degraded carrier mobility resulting from the loss of texture. Herein, a liquid-phase hot deformation procedure, during which the Bi2(Te,Se)3 ingots are directly hot deformed with the extrusion of liquid eutectic phase, is performed to enhance the thermoelectric performance of n-type Bi2(Te,Se)3 alloys. The deformation-induced dynamic recrystallization is remarkably suppressed due to the reduction of nucleation sites and the release of deformation stress by liquid phase, contributing to a weakened carrier scattering and enhanced carrier mobility. The liquid eutectic phase also facilitates the rotation of grains and enhanced (000l) texture, further improving carrier mobility. In addition, the dense dislocations and lattice distortion introduced into the matrix reduce the lattice thermal conductivity. As a result, a high zT value of 1.1 at 400 K is obtained, about 75% increment over the normal one-step hot deformed alloys. This work not only demonstrates a simple and efficient technique for achieving superior n-type Bi2Te3-based materials, but also elucidates the important role of liquid eutectic phase in hot deformation.

Introduction

Thermoelectric (TE) materials, which enable direct conversion between heat and electrical energy, have drawn much attention in the past decades. Their conversion efficiency is governed by the dimensionless figure of merit zT = α2 σT/(κe + κL), where α, σ, T, κe, and κL are the Seebeck coefficient, electrical conductivity, absolute temperature, electronic thermal conductivity, and lattice thermal conductivity, respectively.1 Several effective strategies such as carrier concentration optimization,[qv: 1b,2] band engineering,3 and phonon engineering4 have been widely utilized to boost the material's electrical properties or to reduce κL, resulting in a remarkable improvement of zT.5

As the state‐of‐the‐art TE materials, bismuth‐telluride‐based solid solutions have been the best commercial materials for solid‐state cooling near room temperature. Both p‐ and n‐type zone‐melted (ZM) Bi2Te3‐based alloys exhibit zT ≈ 1 around room temperature.5 Nevertheless, their easy‐cleavage behavior increases the cost of device fabrication. In the past decade, powder metallurgical processes, in which the samples are prepared by mechanical alloying, ball milling (BM), melt spinning, or solvothermal synthesis followed by hot pressing (HP) or spark plasma sintering (SPS), have been widely used to improve the TE performance as well as the mechanical properties for Bi2Te3‐based alloys. For p‐type (Bi,Sb)2Te3 materials, a peak zT > 1.2 can be readily realized by nanostructuring.[qv: 4d,i,6] Recently, the liquid‐phase sintering (LPS) process, which involves sintering under conditions where solid grains are surrounded by the wetting liquid,7 is also successfully applied to enhance the TE properties of p‐type Bi2Te3‐based alloys.[qv: 2a,4e,8] Nanoscale defects such as dense dislocations[qv: 4e] or Sb‐rich region[qv: 8b] could be introduced during LPS and play important roles in reducing κL. In addition, the texture could be enhanced after LPS,[qv: 8a] which is beneficial for the improvement of carrier mobility µ H.

For n‐type Bi2(Te,Se)3 alloys, the LPS process has also been employed to reduce κL via introducing nanostructures into the matrix. Through LPS, a peak zT ≈ 0.98 at 370 K and a peak zT ≈ 0.7 at 398 K have been obtained by Zhang et al. and Ge et al., respectively.[qv: 8b,9] However, compared to the p‐type counterparts, zTs in n‐type polycrystalline Bi2(Te,Se)3 alloys are usually less than 1.0 due to the deteriorated texture and the deviation from the optimal carrier concentration resulting from the pulverization of ingots and donor‐like effect.10 Recently, Park et al. prepared phase‐pure n‐type K0.06Bi2Te3.18 alloys via the hydrothermal method combined with SPS and a high zT > 1.1 at 323 K was achieved.11 Hong et al. performed a microwave‐assisted solvothermal method to get high‐quality Bi2Te3− Se nanoplates and obtained a zT of 1.23 at 480 K for the sintered sample.12 However, these two methods are too complicated to be competitive for a large‐scale production.

In recent years, the hot deformation (HD) process has been successfully applied to enhance texture as well as reduce κL in n‐type Bi2Te3‐based materials. By applying a twice HD on n‐type ZM Bi2Te2.79Se0.21 ingot, Hu et al. induced multiscale microstructures into the matrix to strongly scatter phonons and obtained a peak zT ≈ 1.2 at 357 K.13 Similarly, Li et al. reduced room temperature κL from 1.2 to 0.8 W m−1 K−1 and reported on a peak zT ≈ 1.1 at 625 K in the hot deformed Bi1.85In0.15Te2Se alloys.14 Resulting from the increased µ H by texturing in the HD process, peak zTs varying from 1.04 to 1.3 can be realized in different n‐type HD Bi2Te3‐based materials.[qv: 2b,15] Note that, most of them with peak zT ≥ 1.1 were obtained by multiple HD. Nevertheless, such an HD procedure is complicated, which includes ingot pulverization by BM, powder sintering by HP or SPS, and repeated hot deformation (up to three times), leads to considerable energy consumption. Meanwhile, the uncontrollable volatilization of Te or Se during multiple high‐temperature processing and the severely increased n H caused by the donor‐like effects during ingot pulverization[qv: 2b,10f,15b] could also result in poor repeatability in the HD Bi2Te3‐based alloys. Hence, it is necessary to develop a new technique with a simpler and milder process to prepare superior n‐type Bi2(Te,Se)3 materials.

Herein, we report a novel liquid‐phase hot deformation (LPHD) technique to enhance the TE properties of n‐type Bi2(Te,Se)3 alloys. The lamellate Te‐rich eutectic phase is introduced into the Bi2(Te,Se)3‐melted ingot. This ingot is then directly hot‐deformed in a larger graphite die at a temperature above the eutectic point. In this case, the Bi2(Te,Se)3 solid grains are initially surrounded by the liquid eutectic phase and then gradually deformed with the extrusion of liquid. Compared to the multiple HD procedure, the process flow in LPHD technique is much shortened. During one‐step LPHD, the deformation‐induced dynamic recrystallization is remarkably suppressed due to the reduction of nucleation sites at interfaces and the release of deformation stress in the matrix, contributing to a weakened electron scattering and hence an enhanced carrier mobility µ H. In addition, the µ H is further improved by texturing during LPHD. Meanwhile, a significant reduction of the lattice thermal conductivity is also realized in the LPHD sample as a result of the enhanced phonon scattering by dense dislocations as well as lattice distortion. Finally, a high zT value of 1.1 at 400 K is obtained in the n‐type LPHD Bi2(Te,Se)3 alloys. These results not only demonstrate a simple and energy‐saving technique for synthesizing high‐performance n‐type Bi2Te3‐based materials, but also emphasize the significance of liquid eutectic phase in tuning materials' microstructures.

Results and Discussion

The X‐ray diffraction (XRD) patterns in Figure a indicate that the Bi2Te2.7Se0.3 melted ingot (named as M‐0Te) has a pure rhombohedral R3m phase, while the Bi2Te2.7Se0.3 + 16 wt% Te‐melted ingot (named as M‐16Te) has a little Te second phase remaining in the Bi2Te3 matrix. The Te peaks (indicated by the red arrows) are hard to be recognized from the normal XRD patterns but can be clearly identified in the logarithmic XRD intensity versus 2θ patterns.

Figure 1

a) XRD patterns of Bi2Te2.7Se0.3 + x wt% Te melted ingot. Inset is the enlarged XRD pattern with a logarithmic intensity axis. The red arrows indicate the peaks from elemental Te. b) Phase diagram of Bi2Te3–Te system with an eutectic composition at ≈92 at% Te.[qv: 4e,16] The red arrow indicates the nominal composition of Bi2Te2.7Se0.3 + 16 wt% Te melted ingot.

The scanning electron microscopy (SEM) backscattered electron (BSE) image (Figure a) and its corresponding elemental distribution mapping results (see Figure 2c,e,g) also reveal the existence of Te‐rich phase in the M‐16Te. The brighter phase in Figure 2a is roughly identified as Bi2Te2.55Se0.33 by energy‐dispersive spectrometer (EDS; Table S1, Supporting Information) and then corrected to be Bi2Te2.66Se0.26 by electron probe microanalysis (EPMA; Table S2, Supporting Information), while the darker phase (with more Te) in Figure 2a exhibits a lamellar structure composed of two phases with different Te contents, as shown in Figure 2b and Figure S1 (Supporting Information). Further analysis on the lamellar structure (Figure 2d,f,h; Tables S1 and S2, Supporting Information) indicates that the darker region is pure Te, while the composition of brighter region is variable because the grain size is too small (≤1 µm) to be explicitly examined by EDS and EPMA. Based on the Bi2Te3–Te phase diagram in Figure 1b,[qv: 4e,16] this lamellar structure might be the eutectic phase of Bi2Te3–Te. To verify it, a differential scanning calorimetry (DSC)– thermogravimetric analysis (TGA) test is performed on the M‐16Te, and a distinct endothermic peak at 692.3 K is detected (see Figure S3 in the Supporting Information), very close to the eutectic temperature (686 K) in Figure 1b. Therefore, the microstructure of M‐16Te is concluded to be a mixture of Bi2Te2.66Se0.26 alloys and eutectic phase of Bi2(Te,Se)3–Te. Undoubtedly, the Bi2(Te,Se)3–Te eutectic phase would also exist in our other Bi2Te2.7Se0.3 + x wt% Te‐melted ingots (x from 1 to 32) according to Figure 1b.

Figure 2

a) SEM backscattered electron images for Bi2Te2.7Se0.3 + 16 wt% Te melted ingot. b) Enlarged view of boxed region in panel (a). c) Bi, e) Te, and g) Se elemental distribution in panel (a). d) Bi, f) Te, and h) Se elemental distribution in panel (b).

Since the HD temperature is above the eutectic point, the deformation of Bi2(Te,Se)3 bulk is accompanied by the extrusion of liquid eutectic phase. Hence, this procedure is named as liquid‐phase hot deformation and the picture of extruded eutectic phase is shown in Figure S3 (Supporting Information). For simplicity, the HD Bi2Te2.7Se0.3 sample is named as HD‐0Te and other HD Bi2Te2.7Se0.3 + x wt% Te (x ≥ 1) samples are named as LPHD‐xTe, respectively. The XRD patterns in Figure a,b show that all HD‐0Te or LPHD‐xTe samples are pure Bi2Te3 phase without any detectable Te second phase remaining. The compositional homogeneity in all samples is confirmed by EDS mapping and a typical result of LPHD‐16Te is shown in Figure S4 (Supporting Information). The real composition of LPHD‐16Te is identified as Bi2Te2.70Se0.27 by EPMA, which has a slightly higher Te and Se contents compared to the Bi2Te2.66Se0.26 melted precursor ingot.

Figure 3

a) In‐plane XRD patterns of the HD‐0Te and LPHD‐xTe bulk samples. b) Enlarged XRD patterns with a logarithmic intensity axis. Compared to the inset in Figure 1a, no peaks from elemental Te are detected in the red box range.

To evaluate the texture degree, orientation factor F of (000l) plane was calculated using the Lotgering method,17 and the results are summarized in Table . It could be seen that with the increase of x in LPHD‐xTe, F first increases from 0.09 to 0.20 and then slightly decreases to 0.17, indicating an initially enhanced and then weakened texture in the LPHD samples (the nonmonotonic variation of F will be discussed later). A comparison of SEM microstructure between HD‐0Te and LPHD‐16Te is also presented in Figure ; the coarse grains with random distribution are exhibited in HD‐0Te, while the grains are refined and highly oriented in the LPHD‐16Te, consistent with the F results.

Table 1

The orientation factor F of (000l) plane for the HD‐0Te and LPHD‐xTe bulk samples

x content	0	1	2	4	8	12	16	24	32
F value	0.09	0.12	0.14	0.17	0.17	0.18	0.20	0.17	0.17

Figure 4

SEM images of the cross sections parallel to the pressing direction for the a) HD‐0Te sample and b) LPHD‐16Te sample.

An increase of n H from 4.1 × 1019 to 5.0–6.4 × 1019 cm−3 after LPHD is shown in Figure a, indicating that excess Te acts as electron donor during the LPHD process, similar to the case in the n‐type Te‐doped Bi2Te2.4Se0.6 alloys.18 The compositional variation in the LPHD‐16Te sample (from Bi2Te2.66Se0.26 to Bi2Te2.70Se0.27) indicates that a small number of Te and Se atoms in the liquid eutectic phase may diffuse into the Bi2(Te,Se)3 grains during LPHD and provide electrons (in the formation of antisite defects). Nevertheless, when increasing x from 1 to 32, n H has little change, which may be caused by the relatively small equilibrium concentration of in n‐type Bi2(Te,Se)3 alloys. The variation of room temperature carrier mobility µ H with x is also presented in Figure 5a. As one can see, µ H first increases and then decreases with x, reaching µ H ≈ 196 cm−2 V−1 s−1 for the LPHD‐16Te sample (the nonmonotonic variation of µ H will be discussed later). Low‐temperature Hall measurement (Figure S5a, Supporting Information) shows that the intrinsic conduction near room temperature is obvious in the HD‐0Te sample, which will result in an incorrectly calculated µ H. Even so, the enhanced µ H by LPHD could be verified from the µ H data in the extrinsic region (T < 150 K), consistent with the F results in Table 1. To eliminate the influence of intrinsic conduction and make a better comparison between the normal HD sample and LPHD samples, a heavy SbI3‐doped Bi2Te2.7Se0.3 HD sample was also prepared, and its carrier transport properties are labeled in Figure 5a. Apparently, the µ H of LPHD samples is still higher than the HD sample, demonstrating the importance of liquid eutectic phase for enhancing µ H during the LPHD process.

Figure 5

a) Room temperature carrier concentration and mobility as a function of excess Te content x in the HD‐0Te and LPHD‐xTe samples. The green spots are the HD Bi2Te2.7Se0.3 sample with SbI3 doping. b) Nondegenerated mobility µ 0 as a function of orientation factor F in LPHD Bi2Te2.7Se0.3 samples and other reported HD Bi2Te2.7Se0.3 and HD Bi2Te2.79Se0.21 samples.[qv: 2b,13,19]

A comparison of Hall mobility µ H and orientation factor F between LPHD Bi2Te2.7Se0.3 samples and other reported HD Bi2Te2.7Se0.3 or Bi2Te2.79Se0.21 samples are summarized in Figure S5b (Supporting Information). In the single parabolic band (SPB) model, µ H is readily influenced by the reduced Fermi energy η according to the following equation20 where λ is the scattering parameter and is equal to 0 for acoustic phonon scattering, F(η) is the Fermi integral of order j, µ 0 is the nondegenerate mobility and related to the carrier relation time τ0 (affected by the carrier scattering), transport effective mass , and electron charge e as following20

In the SPB model, η could be calculated by α according to the following equation20

In order to exclude the effect of η on carrier mobility, the nondegenerate mobility µ 0 of all samples in Figure S5b (Supporting Information) was calculated according to Equations (1)–(3) with the assumption of acoustic phonon scattering dominating, and the results are presented in Figure 5b. Compared to the routine HD samples,[qv: 2b,19] our LPHD samples have higher µ 0 at a fixed F. Particularly, although the HD Bi2Te2.79Se0.21 samples13 have weaker alloy scattering and larger F values, their µ 0 is still less than that of the LPHD Bi2Te2.7Se0.3 alloys. It should be noted that at a constant F value, µ 0 is only related to τ0 and , which is described in Equation (3). Since our LPHD samples have a similar composition with other reported HD samples, the variation of among these samples could be ignored. Therefore, the remarkably enhanced µ 0 by LPHD should arise from the increased carrier relation time τ0, which indicates a reduced carrier scattering in the LPHD samples.

To verify this conjecture, further microstructural investigations on our HD‐0Te and LPHD‐16Te samples was carried out by electron backscattered diffraction (EBSD). The results are shown in Figure . Figure 6a shows that the HD‐0Te sample has the coarse grains with numerous fine grains in the vicinity of large grains. Figure 6b is a zoom‐up of the box region in Figure 6a, confirming that the dense dots around the coarse grain boundaries are fine equiaxed grains rather than other defects or holes. A highly inhomogeneous distribution of grain size in the HD‐0Te sample is presented in Figure S6a (Supporting Information); the coarsest grain size reaches ≈1 mm, while the finest grain size is only ≈4 µm and could be smaller due to the limited scanning step size of 2.5 µm. The coarse grains are considered to be generated at the melting–cooling stage and be elongated during hot deformation, while the fine equiaxed grains should be the hot deformation‐induced dynamic recrystallized grains by comparing the typical dynamic recrystallization microstructures in the plastic deformed metals.21

Figure 6

Orientation imaging microscopy maps for the a) HD‐0Te sample and c) LPHD‐16Te sample. b) The enlarged view of the box region in panel (a), and d) the enlarged view of box region in panel (c).

In contrast, the LPHD‐16Te sample has more bar‐shaped grains with a relatively homogeneous distribution of grain size, as shown in Figure 6c and Figure S6 (Supporting Information). Note that, compared to the HD‐0Te sample, the number of recrystallized grains is significantly decreased in the LPHD sample (see Figure 6d and Figure S6 in the Supporting Information) combined with a reduction of total boundary length from 2.15 m (Figure 6a) to 1.58 m (Figure 6c) in an area of 4 mm by 6 mm, which is beneficial for the reduction of carrier scattering and enhancement of µ H. The improved µ H also results from strongly texturing during LPHD. In Figure 6, a larger regime of red color represents a stronger (000l) preferred orientation, and it is easy to find that the LPHD sample has more enhanced (000l) texture. This is further demonstrated by the pole figure (POF) and inverse pole figure (IPF) results in Figure , since a higher polar density in the center of (0001) POF and near the (0001) point in IPF indicates a stronger (000l) texture.22 Furthermore, enhanced texture in the LPHD sample is also double‐checked by part of the orientation distribution function (ODF) results in Figure . For Bi2Te3 alloys, more highlighted regions located at the top of ODF sections (Φ = 0) indicate stronger texture along the (000l) direction. More detailed ODF analysis can be found in Figure S7 (Supporting Information).

Figure 7

a) Pole figure of (0001) plane and b) inverse pole figure for the HD‐0Te sample. c) Pole figure of (0001) plane and d) inverse pole figure for the LPHD‐16Te sample.

Figure 8

Partial ODF patterns for the a) HD‐0Te sample and b) LPHD‐16Te sample.

As mentioned above, both the diminution of recrystallized grains and enhancement of texture during the LPHD process are beneficial for the increase in µ H. The former could be well illustrated based on the dynamic recrystallization theory.[qv: 21b,23] In a normal hot deformation process, dynamic recrystallization takes place during straining as long as the temperature is heated above half of the melting point of samples. The recrystallized grains are produced by nucleation and growth of crystallites in the vicinity of grain boundaries and the driving force is the stored energy induced by the plastic deformation. Hence, HD‐induced recrystallized grains could be readily found in the HD‐0Te sample (see Figure 6a,b) and these grains also exist in normal hot‐deformed Bi2Te3 materials.[qv: 4c] However, during the LPHD process, most of the grains are surrounded by liquid eutectic phase until the liquid is completely extruded, resulting in a decreased nucleation sites at grain boundaries. In addition, the stress induced by deformation is timely released by the extrusion of liquid, hence the stored energy in grains maintains on a relatively low level, further inhibiting nucleation. Therefore, the dynamic crystallization is well suppressed in the LPHD‐16Te sample (see Figure 6c,d).

The enhanced texture during LPHD is also related to the eutectic phase. For HD Bi2Te3 alloys, grain rotation along the basal plane is demonstrated to be the main mechanism for texturing.13, 24 Coarse grains are quite difficult to rotate and hence the texture is relatively low in the HD‐0Te sample. Comparatively, for the LPHD sample, grains are refined by the constraint of eutectic phase at the melting–cooling stage and the rotation resistance at the HD stage is also reduced by the wetting of liquid phase, both contributing to an easier grain rotation and stronger texture during the LPHD process. Compared to the normal HD process, in which µ H could be deteriorated by recrystallization, the LPHD procedure can simultaneously enhance the texture and suppress the dynamic recrystallization, resulting in a much higher carrier mobility in the LPHD samples (see Figure 5b).

The electrical properties of HD‐0Te and LPHD‐xTe samples are displayed in Figure . The remarkable intrinsic conduction near room temperature results in a relatively low α in the HD‐0Te sample (see Figure 9a). For the SbI3‐doped HD and LPHD samples, bipolar effect is suppressed by the increased n H and hence α is improved. In addition, the slight fluctuation of α with x corresponds well to the variation of n H in Figure 5a, suggesting that the different Te contents in LPHD‐xTe samples and SbI3‐doped HD sample do not cause noticeable change in effective mass m*. This is verified by the Pisarenko plots in Figure 9d, and the m* for LPHD and SbI3‐doped HD samples is calculated to be ≈1.2 m e at 300 K, consistent well with the results reported by others.[qv: 2b,10c,25]

Figure 9

Temperature dependences of a) Seebeck coefficient, b) electrical conductivity, and c) power factor for the HD‐0Te and LPHD‐xTe samples. The green curve is the HD Bi2Te2.7Se0.3 sample with SbI3 doping. d) Room‐temperature Pisarenko plots of the LPHD‐xTe samples and SbI3‐dope HD sample with other Bi2Te2.7Se0.3 literature data.[qv: 2b,10c,25]

Attributed to the enhanced µ H at x ≤ 16 or slightly increased n H at x ≥ 24 (see Figure 5a), σ roughly increases with x in the LPHD‐xTe samples, as shown in Figure 9b. By tracing the degree of recrystallization and evolution of grains in LPHD samples, the nonmonotonic variation of µ H with x (in Figure 5a) can be interpreted as follows: in an ideal condition, the number of recrystallized grains in LPHD samples should continuously decrease with the increase of eutectic phase. However, in a certain HD process, during which the HD temperature, HD pressure and HD degree of sample are all constant, the total number of produced recrystallized grains in the matrix is limited. With the increase of eutectic phase content during LPHD, less recrystallized grains are reduced in the matrix, thus leading to less improvement of µ H. On the other hand, as mentioned above, during the melting–cooling stage, the grains are refined by the constraint of eutectic phase, hence grain size should monotonously decrease with the increase of eutectic content. Although eutectic phase is beneficial for texturing through facilitating grain rotation, it also causes significant reduction of grain size and increase of grain boundaries, which could lead to weaker texture26 (see the F values in Table 1) and enhanced carrier scattering. As a result, the gain from suppressed dynamic recrystallization is offset by the degraded texture and increased grain boundaries, leading to a compromise at x = 16 with a maximal µ H ≈ 196 cm−2 V−1 s−1. The PFs for all samples are plotted in Figure 9d. Compared to the HD‐0Te sample, PF in LPHD samples is significantly boosted by the simultaneous optimization of α and σ. The maximal PF ≈ 3.6 × 10−3 W m−1 K−2 is obtained at x = 16, about 60% increment over the HD SbI3‐doped sample.

The total thermal conductivity κ of HD‐0Te and LPHD‐xTe samples is presented in Figure a. Due to the increased n H by Te doping, bipolar conduction is suppressed at room temperature in the LPHD samples. The electronic thermal conductivity κe of all samples is calculated according to κe = LσT, where L is the Lorenz number and estimated by the SPB model. The results are shown in Figure S8 (Supporting Information). Due to the enhanced σ, κe roughly increases with x in the LPHD‐xTe samples. Figure 10b shows the temperature dependence of lattice thermal conductivity κL calculated by κ – κe. An obvious decrease of room temperature κL is observed in the LPHD samples. In particular, the minimal room temperature κL ≈ 0.43 W m−1 K−1 is achieved at x = 24, almost 50% reduction than that of HD SbI3‐doped sample. As mentioned above, compared to the HD‐0Te sample, although coarse grains are refined in the LPHD‐16Te sample, the number of fine recrystallized grains is also remarkably decreased, contributing to a reduced total grain boundary length in the LPHD‐16Te sample. In this case, low‐frequency phonon scattering by grain boundary should be weakened in the LPHD sample. Therefore, the reduction of κL by LPHD must come from other reasons.

Figure 10

Temperature dependences of in‐plane a) thermal conductivity and b) lattice thermal conductivity for the HD‐0Te and LPHD‐xTe samples. The green curve is the HD Bi2Te2.7Se0.3 sample with SbI3 doping.

In order to clearly illustrate the influence of LPHD on the κL in Bi2Te2.7Se0.3 alloys, further microstructure observation is conducted by a transmission electron microscope (TEM), and the results are shown in Figure . It can be clearly seen that compared to the HD‐0Te sample (Figure 11a), the LPHD‐16Te sample possesses more strain‐field domains (Figure 11b, marked with blue dashed curves), which are ascribed to dislocation effect in previous report.27 The bright straight line in Figure 11b is identified as a low‐angle grain boundary by the high‐resolution TEM (HRTEM) image in Figure 11c, in which two quite similar fast Fourier transform (FFT) images of adjacent grains are also displayed. More grain boundary images observed along different TEM zone axis are presented in Figure S9 (Supporting Information). To further investigate the strain‐field domains in the LPHD sample, inverse Fourier transformation is performed on the blue box region in Figure 11c, and the obtained inverse FFT (IFFT) pictures along (006) and (015) reflections are shown in Figure 11d and Figure S10 (Supporting Information), respectively. Large‐scale lattice distortion and dense dislocations (marked by red symbol) can be readily found in the IFFT, which should play important roles in scattering high‐ and medium‐frequency phonons. For HD‐0Te sample, a large number of defects are first induced by plastic deformation and then readily diminished by recrystallization, hence resulting in a relatively high κL. In contrast, for LPHD sample, although the deformation‐induced dislocations and lattice distortion are considerably decreased by the extrusion of liquid eutectic phase, these defects can always remain in the matrix due to the remarkable suppression of recrystallization. As a result, strong scattering of phonons is realized in the LPHD samples, contributing to a remarkable reduction of κL.

Figure 11

a) Low‐magnification TEM image of the HD‐0Te sample with its electron diffraction pattern (inset). b) Low‐magnification TEM image of LPHD‐16Te sample with its electron diffraction pattern (inset). c) HRTEM image from panel (b) with two FFT images of different positions. d) Inverse FFT image of the blue box in panel (c) obtained from (006) reflection.

The dimensionless figures of merit, zTs, for all HD and LPHD Bi2Te2.7Se0.3 samples are presented in Figure a. With varying x from 0 to 32, the maximum zT first increases and then decreases. The highest zT value around 1.1 at 400 K is obtained for the LPHD‐16Te sample, about 75% increment over the HD SbI3‐doped one, even though both of them have the same n H ≈ 5.0 × 1019 cm−3. As aforementioned, liquid‐phase hot deformation could effectively suppress the dynamic recrystallization and enhance the texture, both resulting in a remarkable enhancement of carrier mobility. Meanwhile, dense dislocations and lattice distortion are also introduced into the LPHD samples, contributing to an obvious reduction of κL and further enhancement of zT. A comparison of maximum zT values among LPHD‐16Te sample and other reported Bi2Te3− Se alloys[qv: 2b,10b,c,13,15a,b,d,19,24b] is presented in Figure 12b. Comparing with the normal powder metallurgical process and single HD process, higher zT could be obtained by LPHD technique. Although zTs in some multiple HD alloys are slightly higher than zT in the LPHD sample, the process flow of one‐step LPHD technique is much shorter, which makes it a more efficient and energy‐saving route for large‐scale production of superior n‐type Bi2(Te,Se)3 materials.

Figure 12

a) Temperature dependences of zT for the HD‐0Te sample and LPHD‐xTe samples. The green curve is the HD Bi2Te2.7Se0.3 sample with SbI3 doping. b) Maximum zT values for the LPHD sample and other reported Bi2Te3− Se alloys (0.21 ≤ x ≤ 0.8).[qv: 2b,10b,c,13,15a,b,d,19,24b]

Conclusion

Here, a liquid‐phase hot deformation procedure is successfully performed to enhance the thermoelectric performance of n‐type Bi2(Te,Se)3 alloys. The Te‐rich eutectic phase is introduced into the Bi2(Te,Se)3 ingot at the melting stage and plays important roles in subsequent hot deformation stage. First, the nucleation sites and stored energy for dynamic recrystallization are both reduced due to the wetting and extrusion of liquid eutectic phase, leading to a decrease of recrystallized grains and hence weakened carrier scattering, which is beneficial for an enhanced carrier mobility. Second, grain rotation along the (000l) plane becomes easier with the help of liquid phase, contributing to a stronger texture and further boosted µ H. Third, lattice thermal conductivity is also simultaneously decreased by the dense dislocations and lattice distortion introduced during LPHD. All these effects contribute to a high zT ≈ 1.1 at 400 K in the n‐type LPHD Bi2(Te,Se)3 alloys. This work demonstrates a simple technique for achieving superior n‐type Bi2Te3‐based materials, which is much more efficient and energy saving compared to the multiple HD process.

Experimental Section

Bi (5 N), Te (5 N), Se (5 N) element chunks were weighted according to the stoichiometric Bi2Te2.7Se0.3 + x wt% Te (x = 0, 1, 2, 4, 8, 12, 16, 24, 32) and sealed into φ 12.7 mm quartz tubes evacuated to 10−3 Pa. The mixtures were subsequently melted at 1173 K for 10 h in a furnace and rocked every 2 h to ensure the composition homogeneity. The quartz tubes were sufficiently rocked before taken out from the furnace and then cooled in the air. The conical bottom part of the obtained ingot was cut off and the remained φ 12.7 mm cylindrical ingot was directly hot‐deformed in a larger φ 20 mm graphite die at 773 K for 30 min with 80 MPa uniaxial pressure. The HD direction was parallel to the axial direction of the cylinder. Finally, disk‐shaped HD samples with high densities (>97% of theoretical density) were required.

The phase structures of all samples were evaluated by the XRD on a Rigaku D/MAX‐2550P diffractometer. The freshly fractured surfaces of bulk samples and the microstructure of melted ingots were observed by SEM (Hitachi S‐3700N) equipped with an EDS. The chemical compositions were investigated by EDS and checked by EPMA (JEOL, JXA‐8100) using a wavelength dispersive spectroscope (WDS). The TEM observation was performed on FEI TF20 microscopes, and the TEM samples were prepared by the dual‐beam FIB equipment (Quanta 3D FEG, FEI). The samples for EBSD measurement were first grinded and then mechanically polished by diamond paste (3 in roughness) for about 15 min and finally fine polished by oxide polishing suspension solution for 30 min with an applied load of 60 N. The EBSD analysis was performed on a dual‐beam focused ion beam (FIB, Helios NanoLab 600i, FEI) using a Hikari S/N 1040 camera (TSL/EDAX). EBSD data acquisition and analysis were performed by the OIM Data Collection software and OIM Analysis 7 software, respectively. The DSC and TGA analyses were simultaneously carried out on a TA Instrument SDT Q600 thermal analyzer with a heating rate of 10 K min−1 under Ar atmosphere.

The electrical conductivity σ and Seebeck coefficient α were measured on a commercial Linseis LSR‐3 system using a differential voltage/temperature technique and a DC four‐probe method. The thermal conductivity κ was calculated by using κ = DρC P, where ρ is the density estimated by an ordinary dimension‐and‐weight method, C P is the specific heat calculated by the Dulong–Petit law, and D is the thermal diffusivity measured by a laser flash method on a Netzsch LFA 467 instrument with a Pyroceram standard. The samples for D measurement were prepared using the same method with Xie et al.28 The low‐temperature electrical conductivity σ and Hall coefficient R H from 10 to 300 K were measured on a Mini Cryogen Free Measurement System (Cryogenic Limited, UK). Then the Hall carrier concentration n H and Hall mobility µ H were calculated via n H = 1/eR H and µ H = σR H, respectively. In particular, all thermoelectric properties were measured along the in‐plane direction of samples.

Conflict of Interest

The authors declare no conflict of interest.

Supplementary

Click here for additional data file.