Improved thermoelectric performance in p-type Bi0.48Sb1.52Te3 bulk material by adding MnSb2Se4
Cao Binglei1, 2, Jian Jikang1, †, Ge Binghui2, Li Shanming2, Wang Hao2, Liu Jiao1, Zhao Huaizhou2, ‡
Physics and Optoelectronics Engineering College, Guangdong University of Technology, Guangzhou 510006, China
Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing 100190, China

 

† Corresponding author. E-mail: jianjikang@126.com hzhao@iphy.ac.cn

Abstract

Bismuth telluride (Bi Te based alloys, such as p-type Bi Sb Te , have been leading candidates for near room temperature thermoelectric applications. In this study, Bi Sb Te bulk materials with MnSb Se were prepared using high-energy ball milling and spark plasma sintering (SPS) process. The addition of MnSb Se to Bi Sb Te increased the hole concentration while slightly decreasing the Seebeck coefficient, thus optimising the electrical transport properties of the bulk material. In addition, the second phases of MnSb Se and Bi Sb Te were observed in the Bi Sb Te matrix. The nanoparticles in the semi-coherent second phase of MnSb Se behaved as scattering centres for phonons, yielding a reduction in the lattice thermal conductivity. Substantial enhancement of the figure of merit, ZT, has been achieved for Bi Sb Te by adding an Mn Cu Sb Se (2 mol%) sample, for a wide range of temperatures, with a peak value of 1.43 at 375 K, corresponding to improvement over its Bi Sb Te counterpart. Such enhancement of the thermoelectric (TE) performance of p-type Bi Te based materials is believed to be advantageous for practical applications.

1. Introduction

Direct heat-electricity conversion is an important research topic. In particular, thermoelectrics (TE) have attracted increasing attention owing to its significant potential applicability in waste heat harvesting and cooling applications.[15] Thermoelectric materials utilize the Seebeck effect to generate electricity from heat sources, and their performance is determined by the dimensionless figure of merit , where S, ρ, , , and T are the Seebeck coefficient, electrical resistivity, lattice thermal conductivity, electrical thermal conductivity, and absolute temperature, respectively. Ideally, enhancement of ZT for a TE material can be achieved by reducing the material's thermal conductivity and improving its power factor, which is defined as .[6]

As classical thermoelectric materials with a near room temperature TE energy conversion, Bi Te and its alloys have been extensively studied since the 1950s. Recently, it was suggested that the performance of Bi Te based alloys could be further improved. Till now, two approaches have been used for improving the thermoelectric properties of these materials. The first approach involves optimising the sample preparation process. For instance, Kim et al.[7] confirmed that the peak ZT value of Bi Sb Te could be increased to 1.86 ± 0.15 at 320 K owing to the increased phonon scattering by dislocation at the grain boundary of bulk materials. Xie et al. reported a high ZT of 1.56 in Bi Sb Te alloys using melt spinning.[8] Zheng et al. obtained a ZT of 1.22 for Bi Sb Te via plasma-activated sintering (MS-PAS) at 340 K.[9] An early breakthrough was reported for nanostructured BiSbTe alloys, for which a ZT of 1.4 was obtained at 373 K by ball milling and hot pressing.[10] The second approach involves the use of composites. Qin et al. increased the peak ZT value to 1.6 at 476 K by adding Cu SbSe to Bi Sb Te .[11] Li et al. reported enhanced ZT performance in Bi Sb Te accompanied by a significantly improved mechanical strength, following the addition of SiC nanoparticles.[12] The above results reveal a rich potential for further improving the TE performance of Bi Te -based alloys by tuning the alloy composition and microstructure.

In this study, we achieved substantial enhancement of the TE performance of Bi Sb Te alloys by adding MnSb Se -based materials during a combination of ball milling and hot pressing. The performance improved by compared with the undoped sample. MnSb Se has been identified as an intrinsic p-type semiconductor with an energy band gap of ∼0.31 eV, featuring a high Seebeck coefficient and low lattice thermal conductivity.[13] Based on our results, we propose that MnSb Se is decomposed during the composite's fabrication, with the decomposition products acting as dopants or secondary phases to affect the thermoelectric properties of Bi Sb Te . We observed that Cu doping on Mn sites could substantially promote electric transport and ZT.[14] Here, via the composition with MnSb Se or Mn Cu Sb Se during the synthesis, the optimal carrier concentration was reached and lattice thermal conductivity was reduced, enhancing the thermoelectric performance of Bi Sb Te in a broad range of temperatures.

2. Experimental methods

The Bi Sb Te /MnSb Se composites were prepared by mechanical alloying (MA) and spark plasma sintering (SPS). In a typical run, an individual element (Bi, Sb, Te, in purity) was firstly loaded into a stainless steel jar and milled for 8 h in the presence of Ar. Then, MnSb Se (obtained in our previous work[14]) was added into the jar and milled for 1 h, and the obtained powder was put into a graphite die (diameter, 12.7 mm) and pressed at 673 K for 5 min under 45 MPa. In a similar approach, Bi Sb Te composites were obtained by adding Mn Cu Sb Se (obtained in our previous work[14]).

The phase structure of the as-prepared samples was analysed using x-ray diffraction (XRD) operated under 40 kV and 40 mA at room temperature using Cu Kα radiation ( Å). The microstructures of the samples were examined using scanning electron microscopy (SEM, XL30S-FFG) on the samples’ cross sections, and transmission electron microscopy (TEM, Tecnai F20) specimens of the mixed BiSbTe alloys were prepared by dicing, polishing, and ion-milling. The room-temperature carrier concentration and mobility of the samples were measured using the van der Pauw technique under a reversible magnetic field of 0.5 T, using pressure-assisted W electrodes on a Nanometrics Hall measurement system with the uncertainty of for the Hall coefficient data. The electrical resistivity and the Seebeck coefficients of the samples were measured using a Linseis LSR-3 in the presence of high-purity He. These samples were cut into bars, and the measurement's deviation was . The thermal conductivities of the samples were calculated as , where D is the thermal diffusivity, is the heat capacity, and d is the measured density based on the Archimedes drainage method. All of the samples for the thermal diffusivity tests were cut into 12.7-mm-diameter and 2-mm-thickness discs, and then coated with amorphous C. The thermal diffusivities (D) were measured using Linseis XFA 500 in the presence of high-purity He.

3. Results and discussion
3.1. Structure and composition analysis

Figure 1(a) shows the XRD patterns of Bi Sb Te obtained after adding X mol% of MnSb Se (from (a) to (e), X = 0, 1, 2, 3, and 4), and those of Bi Sb Te , obtained by adding 2 mol% of Mn Cu Sb Se (Fig. 1(a), trace (f)). As seen from the XRD patterns, the XRD diffraction patterns for the doped samples are identical to that of the Bi Sb Te phase, and no secondary phases of MnSb Se or Mn Cu Sb Se were detected using our equipment. Meanwhile, some Mn and Se atoms form MnSe alloys in the host, but this cannot be resolved by XRD owing to its trivial volume ratio in the composite. The above speculation is further supported by TEM images shown below. The XRD pattern of Bi Sb Te can be indexed by the space group of R-3mH (No. 166). Figure 1(b) shows the lattice parameters ( ) of the doped Bi Sb Te , as a function of the MnSb Se content, for various X. Doping of the Bi Sb Te samples increased the values of the a and b lattice parameters, compared with the undoped sample, while the values of the c lattice parameter was reduced. Meanwhile, the cell volume was reduced by the doping. Figure 1(c) and 1(d) show the low- and high-magnification SEM images of the doped Bi Sb Te (by adding 2 mol% of MnSb Se , respectively, indicating that the sample exhibits uniform-size grain and no preferred texture.

Fig. 1. (color online) (a) XRD patterns for composites Bi Sb Te obtained by adding X mol% of MnSb Se (from (a) to (e): X = 0, 1, 2, 3, and 4; (f): X = 2 Mn Cu Sb Se . (b) Lattice parameters ( ) of Bi Sb Te composites, obtained by adding X mol% of MnSb Se . (c), (d) The low- and high-magnification SEM images of doped Bi Sb Te (obtained by adding 2 mol% of MnSb Se .

Figure 2 shows the TEM and EDX images of Bi Sb Te with added 2 mol% of MnSb Se (Figs. 2(a1)2(a3)) and 2 mol% of Mn Cu Sb Se (Figs. 2(b1)2(b3)), respectively. The low-magnification TEM image of Bi Sb Te with MnSb Se (Fig. 2(a1)) reveals uniformly sized grains and a second phase at the grain boundary, which is more clearly revealed in the amplified TEM image in Fig. 2(a2). The corresponding EDX mapping images of Bi, Sb, Te, Mn, and Se are shown in the right column in Figs. 2(a1)2(a3). It can be seen that Bi is abundantly present at the grain boundary, which arises owing to the decomposition of a small amount of Bi Sb Te in the presence of hightemperature and high pressure. Elemental Bi is in the liquid form and is expelled to the grain boundary. Figure 2(a3) shows an HRTEM image of Bi Sb Te with 2 mol% of MnSb Se . In the image, the calculated inter-planar distance is ∼0.31 nm, which is consistent with the (015) plane of Bi Sb Te (JCPDS 49-1713).

Fig. 2. (color online) Transmission electron microscopy (TEM) and energy dispersion x-ray spectrometry (EDX) mapping images of Bi Sb Te with 2 mol% of MnSb Se (column A) and with 2 mol% of Mn Cu Sb Se (column B), respectively. (a1) a low-magnification TEM image of the sample, with the main body as Bi Sb Te and the black spot corresponding to the second phase; (a2) a grain boundary TEM image of the sample; (a3) a HRTEM image of the sample matrix, the images on the right side of TEM graphs are the EDS mapping of areas in panel (a2). (b1, b2) low-magnification TEM images of the sample, with the main phase as Bi Sb Te and the spot corresponding to the second phase; (b3) a high-magnification TEM image of the sample; (b4) a HRTEM image of the sample matrix, with the images on the right side of the TEM images showing the EDS mapping images of the area in panel (b3).

Figure 2(b1)2(b4) shows the microstructural features of Bi Sb Te with 2 mol% of Mn Cu Sb Se . The grains are uniformly sized, and the second phase of nanoparticles is clear, too. This is more clearly revealed by the magnified TEM images in Figs. 2(b2) and 2(b3). The elemental EDX mapping in Fig. 2(b3) is shown in the left column. Cu, Sb, and Te were all uniformly distributed in the host matrix. However, Mn and Se were enriched in nanoparticles, suggesting the formation of the MnSe foreign phase. Figure 2(b4) shows a high-magnification TEM image of the sample, with a clearly identifiable lattice structure of MnSe. We suggest that alloying or replacing reactions occur in these samples. We assume that addition of MnSb Se or Mn Cu Sb Se to the composites results in the decomposition during the preparation process, following which Mn atoms can be incorporated into the BiSb Te lattice by replacing the atoms at BiSb sites, with Se atoms replacing Te atoms, leading to the formation of the (BiSbMn) (TeSe) alloy with the same lattice structure as that of (BiSb) Te . The EDX and HRTEM results indicate that MnSe nanoparticles emerge as the foreign second phase in the host matrix of Bi Sb Te .

3.2. Thermal and electrical transport properties

The temperature dependence of thermal conductivity, for temperatures in the 300–550 K range, for a series of samples, is shown in Fig. 3 (the data reported by Poudel et al. are provided for comparison[10]). As shown in Fig. 3(a), the overall thermal conductivity increases with increasing temperature. It is clear that all thermal conductivities of the doped samples are higher than that of Bi Sb Te for temperatures under 400 K, which can be rationalised by considering their higher electrical thermal conductivities, shown in Fig. 3(c). The thermal conductivities of the composite samples are lower than that of the undoped Bi Sb Te for K. The addition of MnSb Se or Mn Cu Sb Se increases the carrier concentration and suppresses the bipolar diffusion.[15] Figure 3(b) shows the Lorenz numbers L of the samples, which were calculated using a single parabolic band model (SPB),[1619] as follows:[20]

(1)
where L is in K−2 and S is in V/K. Figure 3(c) and 3(d) show the temperature dependence of electrical thermal conductivities and lattice conductivities for the samples. The carrier thermal conductivity can be obtained by using the Wiedemann–Franz law, , and the lattice thermal conductivity can be calculated by subtracting from the overall thermal conductivity (where L, σ, and T are the Lorenz number, the electrical conductivity, and the temperature, respectively). Figure 3(c) shows that the values for the doped samples are higher than for the Bi Sb Te , owing to the improvement in the carrier concentration. Nevertheless, the lattice thermal conductivities for the doped samples are lower than that for the undoped sample. This is mainly owing to the increased scattering from the second phase and the increased presence of structural defects in the host matrix,[21,22] as shown in Fig. 3(d). The overall thermal conductivities for the Bi Sb Te alloys with MnSb Se or Mn Cu Sb Se are lower than that of the nano-crystalline BiSbTe sample below 400 K, as reported by Poudel et al.

Fig. 3. (color online) Temperature dependence of (a) overall thermal conductivity , (b) the Lorenz number L, (c) carrier thermal conductivity , (d) lattice and bipolar contributions of thermal conductivity ( for Bi Sb Te with X mol% of MnSb Se (X = 0, 1, 2, 3, and 4), 2 mol% of Mn Cu Sb Se samples, and the reference nano-crystalline BiSbTe sample.[10]

Figure 4 shows the temperature dependence of the electrical resistivity ρ, the Seebeck coefficient S, the power factor , the ZT value, the room temperature Hall carrier concentration (n), and the carrier mobility (μ), respectively. As shown above, the electrical resistivities of the doped samples are lower than that of the undoped Bi Sb Te sample. The can be ascribed to the substitution of Bi by Mn in MnSb Se that contributed free vacancies, which is similar to Djieutedjeu's report.[23] The doping process can be described as follows:

(2)
In addition, the lowest electrical resistivity was achieved for Bi Sb Te with 2 mol% of Mn Cu Sb Se . This can also be ascribed to the substitution of Bi by Cu , resulting in two more free vacancies in the sample. This process can be described as follows:
(3)
Thus, it is expected that the carrier concentration in the Bi Sb Te sample doped with Mn Cu Sb Se is higher than those in other samples, which is supported by the data in Fig. 4(c). From Fig. 4(b), the Seebeck coefficients of the doped samples are lower than that of the undoped sample below 450 K, and the Seebeck coefficient of the undoped Bi Sb Te decreases faster than those of other samples with increasing the temperature, and this may be owing to the increasing hole concentrations for the doped samples, as shown in Fig. 4(c). Figure 4(c) shows the room temperature Hall carrier concentration and carrier mobility for the samples (the independent points of different carrier concentrations and carrier mobility represent Bi Sb Te with 2 mol% of Mn Cu Sb Se . The reference Bi Sb Te exhibits a low carrier concentration (n cm ), while the carrier concentration increases with the addition of MnSb Se . The highest carrier concentration is obtained after adding 2 mol% of MnSb Se (n cm ), and the highest carrier mobility (μ ∼152 cm V s ) is achieved for the same sample. The carrier concentration for Bi Sb Te with 2 mol% of Mn Cu Sb Se reaches a maximum of cm and the carrier mobility is 140 cm V s . However, the hole mobility is weakly dependent on the amount of MnSb Se , as shown in Fig. 4(c), because the hole moves close to the Te position, but Mn and Cu substitute for the Bi position and therefore cause weak scattering at the hole.[24]

Fig. 4. (color online) Temperature dependence of (a) the electrical resistivity ρ, (b) the Seebeck coefficient S, (d) the power factor , (e) the ZT value for Bi Sb Te samples with X mol% of MnSb Se (X = 0, 1, 2, 3, and 4), with 2 mol% of Mn Cu Sb Se , and for the nano-crystalline BiSbTe reference sample.[10] (c) Room-temperature Hall carrier concentration ( and carrier mobility (μ) for the Bi Sb Te samples with various amount of MnSb Se (X = 0, 1, 2, 3, and 4), and the independent data for a 2 mol% Mn Cu Sb Se sample.

Figure 4(d) shows the power factors ( ) of all samples, and it is obvious that the power factors of all of the doped samples are improved. The PF of Bi Sb Te with 2 mol% of Mn Cu Sb Se is higher than others, constituting an average improvement over that of the undoped Bi Sb Te . Owing to a high electrical resistivity, the power factor of the sample with 2 mol% of Mn Cu Sb Se is lower than that of the nano-crystalline (NC) BiSbTe sample.[10] Figure 4(e) shows the temperature-dependent ZTs of the samples, revealing a significant improvement of the ZT values of doped samples. In particular, the ZT value for Bi Sb Te with 2 mol% of Mn Cu Sb Se is the most striking, and the maximal ZT value of ∼1.43 at 375 K reveals a improvement over the value for the reference Bi Sb Te . Overall, incorporating a small quantity of MnSb Se or Mn Cu Sb Se into Bi Sb Te alloys significantly improves their thermoelectric performance.

Figure 5 shows the cumulative temperature dependence of thermoelectric performance, for all samples. To evaluate the energy conversion efficiency (η) and output power density ( ) of the TE material, the engineering dimensionless figure of merit ( ) and the engineering power factor ( ) have been developed by Kim et al.[26] can be expressed as follows:

(4)
where , , and are the hot-side temperature, the cold-side temperature, and , respectively. The output power density ( ) can be obtained from the engineering power factor ( ), as follows:[26]
(5)
where L is the Lorenz number and is defined as .

Fig. 5. (color online) Cumulative temperature dependence of (a) engineering dimensionless figure of merit , (b) energy conversion efficiency (η), (c) engineering power factor ( ), (d) output power density ( ).

Figure 5(a) shows the dependence of on for the doped samples, clearly showing that the values of the doped samples are substantially enhanced with respect to that of the undoped sample. In particular, a value of 0.62 (at K) for BiSbTe with 2 mol% of Mn Cu Sb Se is higher than that for the other samples, which reveals a improvement over the value for the undoped sample, mostly owing to a higher , as shown in Fig. 5(c). Figure 5(b) shows the energy conversion efficiency (η) for all of the samples. The cumulative temperature-dependent maximal efficiency is defined as[26]

(6)
where is the Carnot efficiency and is a dimensionless intensity factor of the Thompson effect (which plays an important role in increasing the efficiency[25]) defined as
in which ) is the Seebeck coefficient at the hot-side temperature . The efficiency of Bi Sb Te with 2 mol% of Mn Cu Sb Se approaches 10.24% at K, which is higher than that of the undoped Bi Se Te sample. Figure 5(c) and 5(d) show the engineering power factor ( and the output power density ( ), for all of the samples. The and for all of the doped samples have been improved, especially for the sample with Mn Cu Sb Se . Thus, a high value of engineering (0.62) and a high conversion efficiency (η) of (at K) were obtained for the Bi Sb Te alloy by adding Mn Cu Sb Se .

4. Conclusions

In summary, improvement in peak ZT, from 1.02 to 1.43 at 375 K, has been achieved for p-type Bi Sb Te by adding 2 mol% of Mn Cu Sb Se during the synthesis process. The improvement of ZT can be ascribed to the increase in the power factor by optimisation of carrier concentration, and to the reduction in the lattice thermal conductivity owing to the increased phonon scattering by the foreign second phases. This study demonstrates that incorporation of MnSb Se into p-type BiSbTe can enhance the thermoelectric performance of the latter.

Reference
1 Liu N Luo X G Zhang M L 2014 Chin. Phys. B 23 080502
2 Disalvo F J 1999 Science 285 703
3 Harman T C Taylor P J Walsh M P LaForge B E 2002 Science 297 2229
4 Wu S H Ryosuke N Masatsugu Zhang Q S Chihaya A 2014 Chin. Phys. B 23 098502
5 Lin T Zeng Z H Wang C H 2014 Acta Phys. Sin. 63 197201 in Chinese
6 Kaur K Kumar R 2016 Chin. Phys. B 25 026402
7 Kim S Lee K H Mun H A Kim H S Hwang S W Roh J W Yang D J Shin W H Li X S Lee Y H Snyder G J Kim S W 2015 Science 348 109
8 Xie W J Tang X F Yan Y G Zhang Q J Tritt T M 2009 Appl. Phys. Lett. 94 102111
9 Zheng Y Zhang Q Su X L Xie H Y Shu S C Chen T L Tan G J Yan Y G Tang X F Uher C Snyder G J 2015 Adv. Energy Mater. 5 1401391
10 Poudel B Hao Q Ma Y Lan Y C Minnich A Yu B Yan X Wang D Z Muto A Vashaee D Chen X Y Liu J M Dresslhaus M S Chen G Ren Z F 2008 Science 320 634
11 Li Y Y Li D Qin X Y Yang X H Liu Y F Zhang J Dou Y C Song C J Xin H X 2015 J. Mater. Chem. C 3 7045
12 Li J H Tan Q Li J F Liu D W Li F Li Z Y Zou M M Wang K 2013 Adv. Funct. Mater. 23 4317
13 Djieutedjeu H Makongo J P A Rotaru A Palasyuk A Takas N J Zhou X Y Ranmohotti K G S Spinu L Uber C Poudeu P F P 2011 Eur. J. Inorg. Chem. 26 3969
14 Li S M Zhao H Z Zhang H Ren G K Liu N Li D D Yang C S Jin S F Shang D S Wang W H Lin Y H Gu L Chen X L 2015 RSC Adv. 5 99065
15 Goldsmid H J 1965 The Thermal Properties of Solids vol. 1 New York Dover Publications, Inc 62 p
16 May A F Toberer E S Saramat A Snyder G J 2009 Phys. Rev. B 80 125205
17 Shi X Y Pei Y Z Snyder G J Chen L D 2011 Energy Environ. Sci. 4 4086
18 Korkosz R J Chasapis T C Lo S H Doak J W Kim Y J Wu C I Hatzikraniotis E Hogan T P Ssidman D N Wolverton C Dravid V P Kanatzidis M G 2014 J. Am. Chem. Soc. 136 3225
19 Xie H H Wang H Pei Y Z Fu C G Liu X H Snyder G J Zhao X B Zhu T J 2013 Adv. Funct. Mater 23 5123
20 Kim H S Gibbs Z M Tang Y L Wang H Snyder G J 2015 Appl. Phys. Lett. Mater 3 041506
21 Ma Y Qing H Poudel B Lan Y H Yu B Wang D Z Chen G Ren Z F 2008 Nano Lett. 8 2580
22 Hu L P Zhu T J Wang Y G Xie H H Zhao J X Zhao X B 2014 NPG Asia Mater 6 e88
23 Djieutedjeu H Zhou X Y Chi H Haldolaarachchige N Ranmohotti K G S Uber C Young D Poudeu P F P 2014 J. Mater. Chem. C 2 6199
24 Gao M Zhang J Y Rowe D M 1996 71 p
25 Hsiao Y Y Chang W C Chen S L 2010 Energy 35 1447
26 Kim H S Liu W S Chen G Chu C W Ren Z F 2015 PNAS 112 8205