Evolution of thermoelectric performance for (Bi,Sb)2Te3 alloys from cutting waste powders to bulks with high figure of merit

Evolution of thermoelectric performance for (Bi,Sb)2Te3 alloys from cutting waste powders to bulks with high figure of merit

Journal of Solid State Chemistry 233 (2016) 186–193 Contents lists available at ScienceDirect Journal of Solid State Chemistry journal homepage: www...

3MB Sizes 0 Downloads 53 Views

Journal of Solid State Chemistry 233 (2016) 186–193

Contents lists available at ScienceDirect

Journal of Solid State Chemistry journal homepage: www.elsevier.com/locate/jssc

Evolution of thermoelectric performance for (Bi,Sb)2Te3 alloys from cutting waste powders to bulks with high figure of merit Xi′an Fan a,b,c,n, Xin zhi Cai a,b,c, Xue wu Han a,b,c, Cheng cheng Zhang a,b,c, Zhen zhou Rong a,b,c, Fan Yang a,b,c, Guang qiang Li a,b,c a

The State Key Laboratory of Refractories and Metallurgy, Wuhan University of Science and Technology, Wuhan 430081, PR China Key Laboratory for Ferrous Metallurgy and Resources Utilization of Ministry of Education, Wuhan University of Science and Technology, Wuhan 430081, PR China c 185#, School of Materials and Metallurgy, Wuhan University of Science and Technology, 947 Heping Road, Qingshan District, Wuhan 430081, PR China b

art ic l e i nf o

a b s t r a c t

Article history: Received 15 August 2015 Received in revised form 14 October 2015 Accepted 17 October 2015 Available online 19 October 2015

Bi2Te3 based cutting waste powders from cutting wafers were firstly selected as raw materials to prepare p-type Bi2Te3 based thermoelectric (TE) materials. Through washing, reducing, composition correction, smelting and resistance pressing sintering (RPS) process, p-type (Bi,Sb)2Te3 alloy bulks with different nominal stoichiometries were successfully obtained. The evolution of microstructure and TE performance for (Bi,Sb)2Te3 alloys were investigated in detail. All evidences confirmed that most of contaminants from line cutting process such as cutting fluid and oxides of Bi, Sb or Te could be removed by washing, reducing and smelting process used in this work. The carrier content and corresponding TE properties could be adjusted effectively by appropriate composition correction treatment. At lastly, a bulk with a nominal stoichiometry of Bi0.44Sb1.56Te3 was obtained and its' dimensionless figure of merit (ZT) was about 1.16 at 90 °C. The ZT values of Bi0.36Sb1.64Te3 and Bi0.4Sb1.6Te3 alloy bulks could also reach 0.98 and 1.08, respectively. Different from the conventional recycling technology such as hydrometallurgy extraction methods, the separation and extraction of beneficial elements such as Bi, Sb and Te did not need to be performed and the Bi2Te3 based bulks with high TE properties could be directly obtained from the cutting waste powders. In addition, the recycling technology introduced here was green and more suitable for practical industrial application. It can improve material utilization and lower raw material costs of manufacturers. & 2015 Elsevier Inc. All rights reserved.

Keywords: Bi2Te3 Thermoelectric materials Thermoelectric performance Cutting waste powders Recycle

1. Introduction As one of the best thermoelectric (TE) materials for power generators and coolers near room temperature, zone-melted Bi2Te3 and its solid-solution alloys have been commercialized for several decades [1–4]. However, the poor mechanical properties in all directions and actually more poor in the radial direction of an ingot limit their application scopes. They are easy to cleavage along basal plane due to weak Van der Waals bonding between Te (1)–Te (1) layers, which results in low material utilization efficiency of less than 50 wt% [5]. Industrially, various waste products are produced n Corresponding author at: 185#, School of Materials and Metallurgy, Wuhan University of Science and Technology, 947 Heping Road, Qingshan District, Wuhan 430081, PR China. Fax: þ 86 2768862529. E-mail addresses: [email protected] (X. Fan), [email protected] (X.z. Cai), [email protected] (X.w. Han), [email protected] (C.c. Zhang), [email protected] (Z.z. Rong), [email protected] (F. Yang), [email protected] (G.q. Li).

http://dx.doi.org/10.1016/j.jssc.2015.10.030 0022-4596/& 2015 Elsevier Inc. All rights reserved.

during zone melting and cutting processes such as tops and tails of single crystal bars due to high impurity concentration, cutting debris, cutting waste powders, broken irregular square particles and so on. There are more than 10 kinds of wastes if we distinguish in detail according to p-type and n-type semiconductors, which accounts for more than 50 wt% of the total raw materials. In addition, the scarce elements (such as Te and Se) are very expensive due to their low reserves and the increased use in metallurgy, petroleum chemistry and even photovoltaic industries, which result in high raw materials cost for TE materials manufacturers [6]. So, it is important to develop a recycle method for these waste products from zone melting and cutting processes. Generally, hydrometallurgical technology is used to extract the single element such as Te and Bi from waste products [7]. However, the average recovery is low, the process is also long and even pollutes environment. In order to overcome the pollution problems and minimize the loss of useful elementals such as Bi, Sb, Te and Se, an appropriate treatment process for these wastes is very important.

X. Fan et al. / Journal of Solid State Chemistry 233 (2016) 186–193

In this paper, Bi2Te3 based cutting waste powders from cutting wafers were firstly selected as raw materials to prepare p-type Bi2Te3 based thermoelectric materials by washing, reducing, composition correction, smelting and resistance pressing sintering (RPS) process. Their compositions and TE properties were also adjusted to meet the application requirements.

2. Experimental In this work, the initial cutting waste powders were obtained from Chang Shan Wangu electronic technologies Co., Ltd. (Zhejiang, China). These cutting waste powders are contaminated by linear cutting liquid and their own oxides. Firstly, the initial cutting waste powders were washed with deionized water and absolute ethyl alcohol to remove the contaminants of linear cutting liquid. The wet powders were then transferred to a vacuum drying oven to remove water at 60 °C [8]. And then the powders were reduced under hydrogen atmosphere at 380 °C for 1.5 h in a tubular resistance furnace. In order to adjust carrier concentrate of (Bi, Sb)2Te3 alloys, appropriate amount of high purity ( 499.99 wt%) Sb and Te granules were added into the reduced particles and then sealed into a quartz tube under vacuum [9]. The raw materials with three different nominal components of BixSb2  xTe3 (x ¼0.36, 0.40, 0.44) were sealed in quartz tubes and then placed in a rocking furnace. The smelting temperature was fixed at 710 °C and the holding time was fixed for 120 min. Then, the samples were cooled to room temperature in furnace. Subsequently, the ingots were ground into powders with a size of less than  38 μm. Finally, the grinded powders were consolidated by resistance pressing sintering (RPS) technique at 410 °C for 5 min under an axis pressure of 38 MPa in argon atmosphere [10,11]. The heating rate was fixed at 50 °C/min. For comparison, both the initial waste powders and reduced powders without adding Sb and Te were also sintered in the same condition. The thermal gravimetric analysis (TG) was performed at a heating rate of 5 °C /min in argon atmosphere (Sta449 C, Netzsch). The structure, morphology and composition of the specimens were characterized by X-ray diffraction (XRD) with Cu Ka radiation (λ ¼ 1.5418 Å) (X-Pert Philips diffractometer, D500, Siemens), fieldemission scanning electron microscopy (FESEM) (FEI, Nova400 NanoSEM) and energy dispersive X-ray spectroscopy (EDS) (IE350PentaFETX-3). The bulk samples were cut into rectangular bars with an approximate dimension of 3  3  15 mm3 for electrical transport properties. The electrical conductivity (s) were determined by four probes method [10–13]. The Seebeck coefficient (S) were measured by applying a dynamic temperature gradient (ΔT) of 5–10 °C between both ends of the bars and measuring the output voltage (ΔV) between them. The S was obtained by calculating the slope of the linear regression line through the least square method [5,12,13]. The Hall coefficient (RH) was measured by the van der Pauw method using the Hall-effect measurement system (HMS-5500, Ekopia) under a magnetic field of 0.55 T. The carrier concentration (n) and mobility (μ) were calculated according to the equations: n¼ 1/|RH|e and μ ¼s/ne, where e is electrical charge of electron [5,12,13]. The thermal conductivity κ was calculated by multiplying the measured thermal diffusivity λ, specific heat Cp and volume density (d) using the relationship κ ¼ λCpd. The λ and Cp were measured by a laser flash apparatus (LFA457, Netzsch, Germany) [10–13]. The volume densities d of the bulk samples were measured by the Archimedes method. The uncertainty in the electrical transport properties was less than  5% and in the thermal conductivity was less than 7%. In our present experiments, the electrical properties, Hall-effect and thermal properties were all measured in a direction perpendicular to the press direction during RPS process.

187

3. Results and discussion 3.1. The initial cutting waste powders and the corresponding asRPSed bulks To obtain the corresponding microstructure and composition of the initial cutting waste powders, the SEM and EDS for waste powders are performed as shown in Fig. 1(a, b). The morphology of initial cutting waste powders is flocculent and the particle size is small (Fig. 1(a)). From the EDS results, it can be found that the initial cutting waste powders contain C, O, Bi, Te and Sb elements (Fig. 1(b)). The contents of oxygen and carbon are relative high, which should come from linear cutting liquid and oxides of Bi, Sb or Te. So, the following washing and reducing process must be performed. The washing process can remove the linen cutting fluid and the reducing process can remove the element oxygen. In order to confirm the phases of the oxides in the initial cutting waste powders, the corresponding XRD pattern is shown in Fig. 1(c). It can be seen that (Bi0.5Sb0.5)2Te3, Te, Sb2O3 and Bi4TeO8 phases are observed in the XRD patterns. For further confirm the state of these phases in the as-RPSed bulks, the SEM fractographs and EDS of the as-RPSed bulks from the initial cutting waste powders were analyzed as shown in Fig. 1(d). There are three kinds of typical morphologies for the fractographs: typical lamellar structure, agglomerated submicron-sized granules and dispersed cubic particles. The corresponding EDS results show that the smooth laminated structure is the inherent Bi–Sb–Te alloy phases, the agglomerated submicron-sized granules are the oxide of bismuth (tellurium) based alloys and the dispersed cubic particles is Sb2O3, respectively [14]. So, the following reducing process is important to remove the element oxygen. 3.2. Reducing process At the beginning, in order to choose suitable temperature for reducing process, a small part of initial cutting waste powders are selected randomly for TG measurements as shown in Fig. 2(a). There is slight weight loss from room temperature to 400 °C and the weight loss is only 0.2 wt%. However, there are a large amount of volatiles when the temperature is above 400 °C. The volatiles obtained here can be confirmed as Te and their morphologies are tube or bar according to the SEM and corresponding EDS results (as shown in Fig. 2(b)) [15]. It is because that element Te has the highest saturated vapor pressure among elements Bi, Te and Sb (the saturated vapor pressures of Bi, Te and Sb are 10  5, 1 and 10  3 Pa at 367 °C, respectively) [16]. In order to remove the oxygen on the powder surface, Hong et al. also processed the gasatomized powders at 360 °C under H2 atmosphere [17,18]. Lim et al. processed the pulverized powders by H2 at 380 °C [19]. To avoid the volatilization of element Te, the following reduced temperature is selected as 380 °C according to the TG results. Fig. 2 (c, d) shows the SEM and EDS results of the reduced powders, respectively. Compared with the initial cutting waste powders, the slight agglomeration happens for the reduced powders. The EDS results show that the reducing process decrease obviously oxygen content from 10.08 to 3.74 wt%. The content of C also decrease from 3.85 to 2.26 wt%, suggesting that the washing and reducing process selected here are positive, but they are not enough. To confirm the phases in the reduced powders, the XRD patterns are shown in Fig. 2(e). It can be seen that there are mainly (Bi0.5Sb0.5)2Te3 and Sb2O3 phases in the reduced powders. Compared with the initial cutting waste powders, the Bi4TeO8 phase disappears but there are still Sb2O3 in the reduced. It means that the O in Bi4TeO8 phase can be removed but Sb2O3 can not be reduced by H2 at 380 °C [20].

188

X. Fan et al. / Journal of Solid State Chemistry 233 (2016) 186–193

Fig. 1. SEM (a) and the corresponding EDS results (b) of initial cutting waste powders, XRD pattern (c), SEM fractographs and the corresponding EDS analysis (d) of the asRPSed bulks from initial cutting waste powders.

3.3. Smelting process To further remove the Sb2O3 in the reduced powders, the smelting process is performed. Fig. 3(a, b) show the SEM and XRD of the as-RPSed samples from the grinding powders after smelting, respectively. It can be clearly seen that the preferentially oriented lamellar structure presents in the as-RPSed samples. According to the XRD results of the as-RPSed samples, the single phase of (Bi0.5Sb0.5)2Te3 is obtained and the other impurities such as Bi4TeO8 or Sb2O3 can not be found, which confirms that the smelting process can effectively remove antimony oxide. The real photos of quartz tubes after smelting process and the corresponding EDS analysis of the residues adhered to the inner wall of quartz tubes are shown in Fig. 3(c). It can be found that the gray– white or yellow solid residues adhere on the inner wall of quartz tubes after smelting process. It may be the mixtures of antimony oxide (SbxOy such as white Sb2O3, white or yellow solid Sb2O4, and yellow Sb2O5) [10]. The EDS analyses show that the residues contain elements C, O Si and Sb, indicating that the residues may be SiO2 and antimony oxide. The trace C may come from the conductive bandage substrate for SEM or linen cutting fluid. The SiO2 must come from the quartz tubes. Because the residues firmly adhere to the inner wall of quartz tubes, they can be obtained only by scraping using a knife. The antimony oxide must come from our waste powders. As we know, Sb2O5 can not exist above 525 °C [14]. In this work, the reduced powders sealed in tubes are heated to 710 °C, so only Sb2O3 or Sb2O4 can be formed. To better understand this phenomenon, a quartz tube marked with red circle are also

shown in Fig. 3(c), which is used to melt high purity Bi, Sb, and Te at the same condition. Obviously, the tube is clean and no white or yellow residues are adhered to the inner wall of quartz tubes. So, we can rationally deduce that most of mixtures of antimony oxides can be removed by smelting process and the smelting process is effective to reduce oxygen content. Why the smelting process can remove the antimony oxides? It is an interesting and open question. There may be some reasons as following. Most of antimony oxides will volatilize and present in the form of gas in the vacuum quartz tubes during smelting process because of its high saturated vapor pressure (2150 Pa, 710 °C) [21]. During cooling stage, the Sb2O3 firstly solidify because of its higher solidifying point (656 °C). They will preferentially nucleate on the inner wall of quartz tubes due to its lower temperature and the quartz tubes can provide the conditions of heterogeneous nucleation at the same time. Therefore, most of solidified Sb2O3 can gather on the inner wall of quartz tubes. Another possible reason is that the antimony oxides react with SiO2 and form SbxSiyOZ composites. We try to find the synthetic products from SiO2 and Sb2O3, but we failed. We can not find this kind of compounds in the International Centre for Diffraction Data (ICDD) file. 3.4. Composition correction and TE properties optimization According to the EDS results of the as-RPSed samples from smelting/grinding powders, the chemical formula of the samples can be writed as Bi20.11Sb20.13Te59.64. It is obvious that elements Sb and Te are lack compared with the optimal stoichiometric

X. Fan et al. / Journal of Solid State Chemistry 233 (2016) 186–193

189

Fig. 2. The TG curve of initial cutting waste powders (a), SEM images (b) of volatiles generated from reducing process and SEM (c), EDS (d) and XRD pattern (e) of the reduced powders.

composition for p-type Bi2Te3 based materials such as Bi0.4Sb1.6Te3 or Bi0.5Sb1.5Te3. In order to obtain a good TE properties, appropriate amounts of high purity elements Sb and Te is needed to adjust carrier concentration and mobility of the (Bi,Sb)2Te3 alloys. So, three kinds of samples with different nominal stoichiometry (Bi0.36Sb1.64T3, Bi0.40Sb1.60Te3 and Bi0.44Sb1.56Te3) are obtained and investigated in detail in the following parts. 3.4.1. Composition correction and microstructure Fig. 4 shows the XRD patterns of the as-RPSed samples with different nominal stoichiometry. It can be found that all elements Bi, Te and Sb are completely combined into a single phase with a rhombohedral structure and no impurity phase such as oxides are observed. The typical SEM fractographs of the as-RPSed samples with nominal stoichiometry of Bi0.40Sb1.60Te3 are shown in Fig. 5. It is clear that the partially oriented lamellar structure can be observed everywhere. In addition, there are large amounts of fine pores with nano or submicron size distributed in the as-RPSed samples. It may be resulted from the volatilization of elements Te due to its high saturated vapor pressure. Table 1 shows the actual compositions measured and relative densities for the composition correction samples with different nominal stoichiometry. It can be seen that there is a very small fluctuation for element contents between the nominal and actual compositions. The element Te is lack in all three samples due to its’ volatilization, which is the reason for the formation of fine pores as shown in Fig. 5. Moreover, the relative densities of all the samples are above 95%, confirming the RPS process is indeed a good tool for consolidating high densification bismuth tellurium alloys [10,11]. 3.4.2. Electrical transport properties Temperature dependences of s for the as-RPSed samples with different nominal stoichiometry are shown in Fig. 6(a). For

comparison, the s of a sample with a nominal stoichiometry of Bi0.44Sb1.56Te3 prepared from high purity raw materials [11] is also shown in Fig. 6(a). The negative temperature dependence of s indicates a metallic conduction behavior for all the five specimens. According to the definition equation, the s is closely related to the carrier concentration n and mobility μ by the relationship: s¼ neμ. So, the suitable n and μ is crucial for performance optimization [22,23]. For the as-RPSed bulks from initial cutting waste powders, the poor s is mainly caused by the inferior μ as shown in Table 1. It mainly results from the enriched oxygen region, which prevent the migration of carriers and reduce the μ [24]. The submicronsized Bi4TeO8 granules and cubic Sb2O3 oxides acted as scattering centers for carriers increase the carriers-grain boundaries scattering and reduce the μ of the as-RPSed bulks from initial cutting waste powders [22]. Through washing, reducing, composition correction and smelting treatment in the following process, the s increases markedly compared with the initial waste products as a result of increasing μ and appropriate n. In addition, the s of the as-RPSed samples added elements Te and Sb increase with increasing the content of element Sb (as shown in Fig. 6(a)), which agree with the results of other literature [25–27]. Possible explanations for this change can be ascribed to the increasing n. The n and μ of the as-RPSed samples with different nominal stoichiometries are also shown in Table 2. It is worth mentioning that the s of the as-RPSed sample starting from high purity raw materials is similar with that starting from cutting waste powders. As we all know, the dominative carriers (holes) are determined by antistructure defects due to the occupation of Te sites by Bi (Bi′Te) and Sb (Sb′Te) atoms for p-type Bi–Sb–Te alloys. It is noteworthy that the Sb′Te is more easily generated comparing with Bi′Te due to more analogous chemical and physical aspects between Sb with Te elements (The electro-negativities of Sb, Bi and Te are χSb ¼1.9, χBi ¼1.8 and χTe ¼ 2.1, respectively). So, the content of antisite defects increases with increasing the content of Sb, which results in the increase of n [11,28].

190

X. Fan et al. / Journal of Solid State Chemistry 233 (2016) 186–193

Fig. 3. SEM (a) and XRD patterns (b) of the as-RPSed samples from the grinding powders after smelting; real photos of quartz glass tubes after smelting reduced powders (c) and the corresponding EDS analysis of the residues adhered to the inner wall of quartz glass tubes.

Fig. 4. XRD patterns of the as-RPSed samples with different nominal stoichiometry.

The thermopower, S, as a function of temperature for the asRPSed samples with different nominal stoichiometry, are shown in Fig. 6(b). For the three as-RPSed samples added elements Te and Sb, the positive S indicates the p-type conducting behavior and the dominated carrier are hole. The doped Sb result in the change of conducting behavior from n-type to p-type compared with the initial cutting waste samples. The S of TE materials in the extrinsic regime can usually be expressed as S∝γ lnn, where n is carrier concentration and γ is scattering parameter [5,11,29]. The poor S for the as-RPSed bulks from the initial cutting waste powders (as shown in Fig. 6(b)) should ascribe to the presence of oxides of bismuth (tellurium) and the dispersed cubic Sb2O3 particles. These impurities worsen the electrical transport properties. As previously described, the following washing, reducing, composition correction and smelting process can effectively remove these impurities and adjust the type and conten of carriers, which result in a good S. In addition, the S of the as-RPSed samples added elements Te and Sb (as shown in Fig. 6(b)) decreases with increasing the content of element Sb, which should ascribe to the increase of

X. Fan et al. / Journal of Solid State Chemistry 233 (2016) 186–193

191

Fig. 5. SEM fractographs of the as-RPSed sample with nominal stoichiometry of Bi0.40Sb1.60Te3 (a) and the corresponding magnified image (b).

Table 1 The EDS results, the corresponding volume density and relative density for the composition correction samples. Samples (At%)

Nominal compositions (At%)

EDS results of bulks (At%)

Measured densities (g/cm3)

Relative densities (%)a

Bi0.36Sb1.64Te3 Bi0.40Sb1.60Te3 Bi0.44Sb1.56Te3

Bi7.20Sb32.80Te60.00 Bi8.00Sb32.00Te60.00 Bi8.80Sb31.20Te60.00

Bi7.07Sb34.76Te58.17 Bi8.23Sb32.12Te59.65 Bi8.85Sb31.66Te59.49

6.3799 6.5415 6.6610

95.50 96.34 99.71

a

With regard to the theoretical density of 6.79 g/cm3.

Fig. 6. Temperature dependence of s (a) and S (b) for the as-RPSed samples with different nominal stoichiometries.

Table 2 RH, n, μ and ρ of the as-RPSed samples with different nominal stoichiometries measured at room temperature. Samples

RH/cm3C-1

n/  1019 cm  3

μ/cm2 V-1S-1

s/  102 S m  1

Initial Reduction/Smelting Bi0.36Sb1.64Te3 Bi0.40Sb1.60Te3 Bi0.44Sb1.56Te3

 0.1797  0.0752 0.1052 0.1356 0.1930

 3.48  8.4 5.94 4.61 3.24

19.14 28.5 133.27 142.79 166.39

100 379 1266 1052 862

n (as shown in Table 2) [11,26]. It is worth mentioning that the S of the as-RPSed sample starting from high purity raw materials [11] is similar with that starting from cutting waste powders. 3.4.3. Thermal transport properties and ZT values The behaviors of thermal conductivities (k) and ZT values for the as-RPSed samples with different nominal stoichiometries are very typical for the Bi2Te3 based alloys (as shown in Fig. 7). The κ is

contributed from electronic thermal conductivity κe, lattice thermal conductivity κL and bipolar (hole–phonon) thermal conductivity κb. The κe is related to s and can be estimate by Wiedemann–Franz law (κe ¼ L0sT), where L0 is Lorenz constant, L0 ¼2.0  10  8 V2/K2 for a heavily doped semiconductor, T is the temperature in Kelvin [11,29]. Obviously, the κ of the as-RPSed sample from initial cutting waste powders show a lowest value of 0.637 Wm  K  1 at 23 °C. Firstly, the presence of impurities such

192

X. Fan et al. / Journal of Solid State Chemistry 233 (2016) 186–193

Fig. 7. Temperature dependence of k (a), λ (b), Cp (c) and ZT values (d) for the as-RPSed samples with different nominal stoichiometries.

as Bi4TeO8 and Sb2O3 as the barrier against carriers, lead to a small value of κe. Secondly, these impurities can also enhance phonon scattering and decrease the κL. Through washing, reducing, composition correction and smelting process in the following process, the k of the as-RPSed samples increases markedly compared with that of the initial cutting waste products as a result of increasing κe and κL. Besides, the k of the as-RPSed samples added elements Te and Sb increases with increasing the content of element Sb, which complies with the Sb composition correction rule for the Bi2Te3 based alloys [11,25,26]. We also present the temperature dependence of the thermal diffusivity (λ) and specific heat (Cp) of all the as-RPSed samples in Fig. 7(b) and (c), respectively. Apart from the reduction/smelting sample shows a upward trend with increasing measuring temperature, the λ firstly decreases and then increases as increasing testing temperature for all the other samples. The measured Cp (0.18–0.21 J g  1 K  1) at room temperature in this work are very close to that of p-type Bi0.52Sb1.48Te3 by Xie and p-type Bi0.3Sb1.7Te3 (0.18 J g  1 K  1) by Li [31,32]. It is worth mentioning that the k of the as-RPSed sample starting from high purity raw materials [11] is lower than that starting from cutting waste powders due to the relative low κL due to the large amounts of fine grain boundary and abundant point defects caused by the melt spinning technique [10,11]. The ZT values of the as-RPSed samples are shown in Fig. 7(d). The contaminations from the linen cutting fluid and elements’ oxides severely worsen the ZT values of the as-RPSed bulks from the initial cutting waste powders. The obviously increased ZT values after washing/reducing/composition correction/smelting suggests that these treatment processes used in this work have a positive effect on improving the thermoelectric properties. In

addition, with the increase in testing temperature, the ZT values of the three as-RPSed samples added elements Sb and Te first increase and then decrease, and show the best values at 90 °C. The maximum ZT value of 1.16 is obtained for the Bi0.44Sb1.56Te3 sample at 90 °C. The ZT values of Bi0.36Sb1.64Te3 and Bi0.4Sb1.6Te3 alloy bulks can also reach 0.98 and 1.08 at 90 °C, respectively. These results are similar with our previous work [11,25,29] and other literature reports [26,30], which start from high purity elements Bi, Sb and Te. Besides, the average ZT value of the Bi0.44Sb1.36Te3 bulks obtained in this work is all over 0.9 in the entire temperature range from room temperature to 200 °C, which is beneficial to expand the applied temperature range and achieve a stable service property for the TE devices. More importantly, the new recycle technology introduced here can also be used for the other waste products such as tops and tails of single crystal bars, cutting debris and broken irregular square particles and so on. It can increase the raw material utilization and reduce production costs. In addition, it is green and environmentally friend, which solve the problem of environmental pollution for the traditional hydrometallurgical recovery technology. The process is more likely to be beneficial for companies that are major consumers of TE materials. At the same time, it is worth mentioning that there are some shortcoming needed for further improvement such as the consumption of quartz tube, more labor and the consumption of large amount of new high purity materials such as Te and Sb. In the following work, we need to further reduce processing costs by a series of step such as replacing quartz tubes with high borosilicate glass, processing separately the p-type and n-type waste in order to reduce the consumption of new high purity materials.

X. Fan et al. / Journal of Solid State Chemistry 233 (2016) 186–193

4. Conclusions In summary, starting from Bi2Te3 based cutting waste powders, p-type (Bi,Sb)2Te3 alloy bulks with high figure of merit were successfully obtained firstly through a series of washing, reducing, composition correction, smelting and RPS processes. The evolution of microstructures and thermoelectric performances for (Bi, Sb)2Te3 alloys from cutting waste powders to bulks were investigated in detail. Most of contaminants from line cutting process such as cutting fluid and oxides of Bi, Sb or Te could be removed by the washing, reducing and smelting process used in this work. Reducing process decreased obviously oxygen content from 10.08 to 3.74 wt% and the smelting process was effective to remove antimony oxides. The carrier content and corresponding TE properties can be adjusted effectively by appropriate composition correction treatment. The maximum ZT value of 1.16 is obtained for the Bi0.44Sb1.56Te3 sample at 90 °C. This result is comparable with our previous works and other literature reports which start from high purity elements Bi, Sb and Te. Different from the traditional recycling technology such as hydrometallurgy extraction methods, the separation and extraction of beneficial elements such as Bi, Sb and Te did not need to be performed and the Bi2Te3 based bulks with high TE properties could be directly obtained from the cutting waste powders. Moreover, the recycling technology introduced here was green and more likely to be beneficial for companies that are major consumers of TE materials. It can improve material utilization and lower raw material costs of manufacturers.

Acknowledgment The authors thank the support of the National Natural Science Foundation of China (11074195).

References [1] B. Poudel, Q. Hao, Y. Ma, et al., High-thermoelectric performance of nanostructured bismuth antimony telluride bulk alloys, Science 320 (2008) 634–638. [2] S.Y. Wang, G.J. Tan, W.J. Xie, et al., Enhanced thermoelectric properties of Bi2(Te1  xSex)3 based compounds as n-type legs for low-temperature power generation, J. Mater. Chem. 22 (2012) 20943–20951. [3] C. Barreteau, D. Berardan, N. Dragoe, Studies on the thermal stability of BiCuSeO, J. Solid State Chem. 222 (2015) 53–59. [4] K. Termentzidis, O. Pokropyvnyy, M. Woda, et al., Large thermal conductivity decrease in point defective Bi2Te3 bulk materials and superlattices, J. Appl. Phys. 113 (2013) 0135062. [5] X.A. Fan, Z.Z. Rong, F. Yang, et al., Effect of process parameters of microwave activated hot pressing on the microstructure and thermoelectric properties of Bi2Te3-based alloys, J. Alloy. Compd. 630 (2015) 282–287. [6] S.Y. Wang, W.J. Xie, H. Li, et al., Effects of cooling rate on thermoelectric properties of n-type Bi2(Se0.4Te0.6)3 compounds, J. Electron. Mater. 40 (2011) 1150–1157. [7] S.B. Cai, Separation and recovery of selenium and tellurium from high grade waste material, Hydrometall. China 27 (2008) 35–37 (in Chinese). [8] C.J. Liu, Y.L. Liu, H.C. Lai, et al., Effects of reaction temperature on thermoelectric properties of p-type nanostructured Bi2  xSbxTe3 prepared using hydrothermal method and evacuated-and-encapsulated sintering, J. Electron.

193

Mater. 42 (2013) 1550–1554. [9] D.H. Lee, J.U. Lee, S.J. Jung, et al., Effect of heat treatment on the thermoelectric properties of bismuth–antimony–telluride prepared by mechanical deformation and mechanical alloying, J. Electron. Mater. 43 (2014) 2255–2261. [10] X.Z. Cai, X.A. Fan, Z.Z. Rong, et al., Improved thermoelectric properties of Bi2Te3  xSex alloys by melt spinning and resistance pressing sintering, J. Phys. D: Appl. Phys. 47 (2014) 115101. [11] X.A. Fan, X.Z. Cai, Z.Z. Rong, et al., Resistance pressing sintering: a simple, economical and practical technique and its application to p-type (Bi,Sb)2Te3 thermoelectric materials, J. Alloy. Compd. 607 (2014) 91–98. [12] Z.Z. Rong, X.A. Fan, F. Yang, et al., Microwave activated hot pressing: a new consolidation technique and its application to fine crystal bismuth telluride based compound, Powder Technol. 267 (2014) 119–125. [13] F. Yang, X.A. Fan, Z.Z. Rong, et al., Characterization and thermoelectric properties of Bi0.4Sb1.6Te3 nanostructured bulk prepared by mechanical alloying and microwave activated hot pressing, Ceram. Int. 41 (2015) 6817–6823. [14] H.S. Chin, K.Y. Cheong, K.A. Razak, Review on oxides of antimony nanoparticles: synthesis, properties, and applications, J. Mater. Sci. 45 (2010) 5993–6008. [15] T. Siciliano, E. Filippo, A. Genga, et al., Single-crystalline Te microtubes: synthesis and NO2 gas sensor application, Sens. Actuators B: Chem. 142 (2009) 185–190. [16] R.E. Honig, D.A. Kramer, Vapor-pressure data for the solid and liquid elements, RCA Rev. 30 (1969) 285–291. [17] S.J. Hong, B.S. Chun, Microstructure and thermoelectric properties of n-type 95% Bi2Te3–5% Bi2Se3 alloy produced by rapid solidification and hot extrusion, Mater. Res. Bull. 38 (2003) 599–608. [18] S.J. Hong, B.S. Chun, Microstructure and thermoelectric properties of extruded n-type 95%Bi2Te3–5%Bi2Se3 alloy along bar length, Mat. Sci. Eng. A: Struct. 356 (2003) 345–351. [19] C.H. Lim, K.T. Kim, Y.H. Kim, et al., Effect of powder mixing on thermoelectric properties in Bi2Te3-based sintered compounds, Intermetallics 14 (2006) 1370–1374. [20] C.G. Anderson, The metallurgy of antimony, Chem. Erde – Geochem. 72 (2012) 3–8. [21] K.Q. Qiu, R.L. Zhang, Research on preparation of nanometer antimony trioxide from slag containing antimony by vacuum evaporation method, Vacuum 80 (2006) 1016–1020. [22] T. Zhang, Q.S. Zhang, J. Jiang, et al., Enhanced thermoelectric performance in p-type BiSbTe bulk alloy with nanoinclusion of ZnAlO, Appl. Phys. Lett. 98 (2011) 022104. [23] S.F. Fan, J.N. Zhao, J. Ma, et al., p-type Bi0.4Sb1.6Te3 nanocomposites with enhanced Figure of merit, Appl. Phys. Lett. 96 (2010) 182104. [24] Y. Horio, A. Inoue, Effect of oxygen content on thermoelectric properties of n-Type (Bi,Sb)2(Te,Se)3 alloys prepared by rapid solidification and hot-pressing techniques, Mater. Trans. 47 (2006) 1412–1416. [25] X.A. Fan, J.Y. Yang, W. Zhu, et al., Effect of nominal Sb2Te3 content on thermoelectric properties of p-type (Bi2Te3)x(Sb2Te3)1  x alloys by MA–HP, J. Phys. D: Appl. Phys. 39 (2006) 5069. [26] J. Jiang, L.D. Chen, S.Q. Bai, et al., Thermoelectric properties of p-type (Bi2Te3)x(Sb2Te3)1  x crystals prepared via zone melting, J. Cryst. Growth 277 (2005) 258–263. [27] J. Yang, T. Aizawa, A. Yamamoto, T. Ohta, Thermoelectric properties of p-type (Bi2Te3)x(Sb2Te3)1  x prepared via bulk mechanical alloying and hot pressing, J. Alloy. Compd. 309 (2000) 225–228. [28] C. Chen, D.W. Liu, B.P. Zhang, J.F. Li, Enhanced thermoelectric properties obtained by compositional optimization in p-type BixSb2  xTe3 fabricated by mechanical alloying and spark plasma sintering, J. Electron. Mater. 40 (2011) 942–947. [29] F. Yang, X.A. Fan, Z.Z. Rong, et al., Lattice thermal conductivity reduction due to in situ-generated nano-phase in Bi0.4Sb1.6Te3 alloys by microwave-activated hot pressing, J. Electron. Mater. 43 (2014) 4327–4334. [30] J. Jiang, L.D. Chen, S.Q. Bai, et al., Thermoelectric properties of textured p-type (Bi,Sb)2Te3 fabricated by spark plasma sintering, Scr. Mater. 32 (2005) 347–351. [31] W.J. Xie, X.F. Tang, Y.G. Yan, et al., Unique nanostructures and enhanced thermoelectric performance of melt-spun BiSbTe alloys, Appl. Phys. Lett. 94 (2009) 102111. [32] J.H. Li, Q. Tan, J.F. Li, et al., BiSbTe-based nanocomposites with high ZT: the effect of SiC nanodispersion on thermoelectric properties, Adv. Funct. Mater. 23 (2013) 4317–4323.