Preparation and thermoelectric properties of AgPbmSbTe2+m alloys

Preparation and thermoelectric properties of AgPbmSbTe2+m alloys

Journal of Alloys and Compounds 469 (2009) 499–503 Preparation and thermoelectric properties of AgPbmSbTe2+m alloys K.F. Cai a,∗ , C. Yan a , Z.M. He...

780KB Sizes 0 Downloads 11 Views

Journal of Alloys and Compounds 469 (2009) 499–503

Preparation and thermoelectric properties of AgPbmSbTe2+m alloys K.F. Cai a,∗ , C. Yan a , Z.M. He b , J.L. Cui c , C. Stiewe b , E. M¨uller b , H. Li a a

Tongji University, Functional Materials Research Laboratory, Shanghai 200092, China German Aerospace Center (DLR), Institute of Materials Research, D-51170 K¨oln, Germany c School of Mechanical Engineering, Ningbo University of Technology, Ningbo 315016, China b

Received 14 August 2007; received in revised form 30 January 2008; accepted 2 February 2008 Available online 26 March 2008

Abstract Hydrothermally synthesized AgPbm SbTe2+m (m = 10–18) nanopowders were pressure-less sintered at 450–480 ◦ C for 5 h in Ar. The samples show large positive Seebeck coefficient but low electrical conductivity and the AgPb18 SbTe20 sample shows higher power factors. The hot pressed AgPb10 SbTe12 sample was highly densified with grain size down to nanoscale, and the sample has inhomogeneous Seebeck coefficient. Obviously different thermoelectric properties have been observed for the AgPb18 SbTe20 samples compacted with pressure-less sintering and spark plasma sintering. © 2008 Elsevier B.V. All rights reserved. PACS: 72.15.Jf Keywords: Semiconductors; Chemical synthesis; Sintering; Electronic transport; Thermoelectric

1. Introduction The effectiveness of a thermoelectric (TE) material is determined by a dimensionless TE figure of merit, ZT = α2 σT/κ, where α, σ, T and κ are the Seebeck coefficient, the electrical conductivity, absolute temperature and the thermal conductivity, respectively. In 2004, Kanatzdis and co-workers [1] reported n-type AgPbm SbTe2+m or LAST-m (LAST stands for Lead Antimony Silver Tellurium) alloys with excellent TE properties. ZT ∼ 2.1 at 800 K for LAST-18 was obtained. This is the highest ZT value reported so far for bulk materials. It has been observed that the LAST-m alloys are inhomogeneous at nanoscale with at least two phases. The minority phase rich in Ag and Sb is in nanosize and is endotaxially embedded in the majority phase that is poor in Ag and Sb, and the electronic structure of LAST-m materials depends sensitively on the perturbation resulting from the Ag and Sb atoms [2–4]. The nanoinclusions in the alloys are considered to play a role similar to the quantum dots in the molecular beam epitaxy grown PbSe0.98 Te0.02 /PbTe superlattices [5]. Hazama et al. [6] have performed first-principles calculations for the LAST-m to clarify the effect of simultane-

Corresponding author. Tel.: +86 21 65980255; fax: +86 21 65980255. E-mail address: [email protected] (K.F. Cai).

0925-8388/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2008.02.040

ous doping of Ag and Sb on PbTe. They have concluded that there exists an optimum Ag–Sb concentration, at which the thermal conductivity is substantially reduced but a large Seebeck coefficient of PbTe is essentially preserved, so that the ZT value can be increased beyond that of PbTe at high temperatures around 800 K. Most recently, Kanatzidis and co-workers [7] again reported p-type Sn-doped LAST-m materials with high ZT up to 1.45 at 627 K. As soon as ref. [1] was published, many other groups all over the world started studying on the LAST-m system. For example, Kosuga et al. [8–11] have done a series of work, however, the maximum ZT obtained is only 1.07 [10]. Wang et al. [12] studied the system using a combined process of mechanical alloying and spark plasma sintering (SPS), found that the TE properties were much dependent on the content of Pb, and obtained ZT value of 1.37 at 673 K for a sample with composition of Ag0.8 Pb22 SbTe20 . Most recently, Karkamkar and Kanatzidis [13] synthesized LAST-m (m = 0–2, ∞) nanoparticles using a reverse micellar approach coupled with a sodium borohydride reduction. The method is convenient, low cost and environmentally friendly. However, due to a coating of an organic layer surrounding the nanoparticles it will be hard to remove the layer before sintering the nanoparticles into compact shapes. Cai and He [14] have reported a hydrothermal synthesis method to AgPbm SbSe2+m


K.F. Cai et al. / Journal of Alloys and Compounds 469 (2009) 499–503

Table 1 Characteristics of the samples

Lattice parameter, a (nm) Density, d (g/cm3 ) Relative density (%) a







0.6459 7.0861 90

0.6465 7.1545 90.5

0.6525 6.9720 90.2

0.6597 6.7650 90.1

0.6502 7.3242 93.1

0.6502 7.2534 92.2

Spark plasma sintered sample.

nanopowders. The nanopowders without an organic layer on their surface hence can be directly sintered. LAST-m nanopowders can also be hydrothermally synthesized [15]. In this work, we report on the microstructure and TE properties of the samples prepared by compacting the hydrothermally synthesized LASTm nanopowders via pressure-less sintering (PS), hot pressing (HP), and SPS. 2. Experimental Amorphous Te powder (5N) and other chemicals with analytical purity were used. The Te powder (3.6 mmol), NaBH4 (7.2 mmol) and NaOH (4 mmol, for adjusting pH value) were put into a 100 ml Teflon lined autoclave. 40 ml deionized water was poured into a 100 ml beaker, placing the beaker on the plate of a magnetic force stirring device, and then adding Pb(NO3 )2 (3 mmol), AgNO3 (0.3 mmol) and Sb(NO3 )3 (0.3 mmol), in that order, into the beaker, stirring with a magnetic bar until the materials were fully dissolved, and then the solution was gradually added into the autoclave, stirring with a glass bar. Additional deionized water was added into the autoclave until about 85% of the volume of the autoclave was filled, and then the autoclave was sealed. The autoclave was placed into an oven, heated up to 180 ◦ C and held for 20 h, and then cooled naturally to room temperature. Black precipitates were collected and washed with deionized water and absolute ethanol in sequence for several times then filtered. The black product was dried in vacuum at 60 ◦ C for 6 h. The dried nanopowders with grain size about 40 nm were uniaxially pressed into pellets (12 mm in diameter and 1 mm thick) at 12 MPa. The pellets were pressure-less sintered (∼450–480 ◦ C × 300 min) in Ar in a quartz tube furnace. The LAST-18 powders were compacted by SPS at 420 ◦ C and 40 MPa for 1 min with a heating rate of 40 ◦ C/min for comparison. The sintered samples were cut into rectangular pieces (10 mm × 1 mm × 2 mm) for TE properties measurement. Electrical conductivity measurements were performed using a steady-state four-probe technique with chopped direct current (∼10 mA), and the Seebeck coefficient was determined by the slope of the linear relationship between the thermoelectromotive force and temperature difference (10–15 K) between the two ends of the sample. The LAST-10 nanopowders were also hot pressed (400 ◦ C × 100 MPa × 1 h, heating rate 10 ◦ C/min) in Ar for comparison. Unfortunately, the sample was broken when it was being separated from graphite die. Therefore, only the local Seebeck coefficient measurement on a polished surface of a small piece was carried out on a scanning Seebeck-Microprobe (SMP) [16]. Compared with HP, the heating rate for SPS used was much faster, the holding time at the sintering temperature for SPS was much shorter, and the pressure applied was lower; therefore, the sintering temperature chosen for SPS was 20 degrees higher than that for HP. The density of the samples was measured by the water displacement method. Composition and microstructure of the sintered samples were characterized by X-ray diffraction (XRD, Rigaku, D/max2550) and scanning electron microscopy (SEM), respectively.

terns of the pressure-less sintered LAST-m samples. Most of the peaks can be indexed to PbTe (JCPDS card file, No. 77-0246) except three very weak impurity peaks. Two of the impurity peaks (2θ = ∼38.4◦ and 40.6◦ , respectively) correspond to element Te (JCPDS card file, No. 36-1452, the strongest peak for Te is just in superposition with the strongest peak for PbTe), and the other one (2θ ∼ 24.8◦ ) can be indexed to PbTeO4 (JCPDS card file, No. 23-0336). Such impurity peaks with lower intensity are also found in the XRD patterns for corresponding nanopowders. This implies that the element Te did not fully take part in the reactions and the nanopowders were slightly oxidized. Martin et al. [17] also identified impurity oxide in bulk PbTe materials prepared by compacting hydrothermally synthesized PbTe nanopowders via SPS. Fig. 2a and b are the temperature dependence of electrical conductivity and Seebeck coefficient of the pressure-less sintered LAST-m samples and spark plasma sintered LAST18 sample, respectively. For the pressure-less sintered samples, the electrical conductivities increase with increasing m value in LAST-m at a given temperature, whereas the Seebeck coefficients of the samples are close especially at high temperatures. The Seebeck coefficient of all the pressure-less sintered samples except the LAST-18 sample slightly decreases with temperature. This may be related to the LAST-18 sample with relative high density. The Seebeck coefficient values are positive in all the temperature range measured, indicating P-type conduction, which is contrary to that reported in ref. [1]. The electrical conductivity of all the pressure-less sintered samples increases with temperature, which is contrary to that reported in ref. [1]. The electrical conductivities and the Seebeck coefficients of the sam-

3. Results and discussion The densities and relative densities of all the pressure-less sintered LAST-m samples and the spark plasma sintered LAST18 sample are given in Table 1. XRD analyses reveal that all the sintered samples have very similar XRD patterns. As an example, Fig. 1 shows the XRD pat-

Fig. 1. XRD patterns of the pressure-less sintered LAST-m samples.

K.F. Cai et al. / Journal of Alloys and Compounds 469 (2009) 499–503


Fig. 3. SEM images of the fracture surface of the LAST-18 samples compacted by PS (a) and SPS (b).

Fig. 2. Temperature dependence of electrical conductivity (a), Seebeck coefficient (b) and (c) power factor for the pressure-less sintered LAST-m samples (solid symbols), the spark plasma sintered LAST-18 sample (open square symbols), the power factors for the LAST-18 sample in refs. [1,8] are also given in (c) for comparison.

ples are, respectively, much lower and higher than those reported in ref. [1] at lower temperatures. The temperature dependences of the electrical conductivity and the Seebeck coefficient of the pressurel-less sintered LAST-18 sample are also different from those of the hot pressed LAST-18 sample reported in ref. [8]. The

much different transport properties between the present work and in refs. [1,8] must be due to different processing routes employed, which result in different compositions and different microstructures. Fig. 2c shows that compared with other pressure-less sintered samples, the LAST-18 sample shows higher power factors (α2 σ) in all the temperature range measured and reaches 6.6 × 10−4 W m−1 K−2 at 480 K. At T < 450 K, the power factor for the spark plasma sintered LAST-18 sample is higher than that for the pressure-less sintered LAST-18 sample and reaches 6.87 × 10−4 W m−1 K−2 at 393 K, while at T > 450 K, the power factor of the two samples are almost the same. At a given temperature, both the two LAST-18 samples have lower power factor than the LAST-18 sample in ref. [1], whereas they have higher power factor than the LAST-18 sample in ref. [8]. The microstructure of the pressure-less sintered samples is quite similar. Some randomly distributed big pores (∼10 ␮m) were observed in the samples under SEM at a low magnification. Fig. 3a shows a typical SEM image of the pressure-less sintered LAST-18 sample at a higher magnification. Grains (∼5–10 ␮m) are mixed with a loose phase. In the spark plasma sintered LAST-18 sample, grains (∼5–10 ␮m) contact each other with


K.F. Cai et al. / Journal of Alloys and Compounds 469 (2009) 499–503

Fig. 4. SEM image of the fracture surface of the hot-pressed LAST-10 sample (a), EDX spectrum of the sample (b) and (c) room temperature Seebeck coefficient scanning image of the polished surface of the sample.

micropores (0.2–1 ␮m) at grain boundaries (see Fig. 3b). These micropores could effectively reduce thermal conductivity due to phonons scattering. The two LAST-18 samples having different microstructures could result in their significantly different TE properties (see Fig. 2). However, we believe that the much difference in TE properties between the two samples is mainly due to different compositions, as we observed that a black thin film appeared at upper inner wall at the outlet end of the quartz tube during the PS. The film consists of nanostructured Te resulted from the volatilization of element Te [18]. While the volatilization might be much suppressed due to a high pressure and very short sintering time during the SPS. SEM observations (see Fig. 4a) of the fracture surface of the hot pressed LAST-10 sample reveal that the sample is highly densified (relative density >99%). The grain size

of the sample is down to nanoscale. EDX analyses reveal that besides the signals corresponding to Pb, Te, Ag and Sb (see Fig. 4b), there are two signals corresponding to C and O, respectively. The C signal is from the carbon film beneath the sample for electrical conducting. The O signal must be due to the PbTeO4 impurity. Oxygen is an acceptor in PbTe semiconductor, which decreases the electron concentration and provides p-type carriers [12,17]. As a result, all the samples show p-type conduction (see the Seebeck coefficient for the hot pressed sample in Fig. 4c below). Fig. 4c shows the locally measured Seebeck coefficient at room temperature of the hot pressed LAST-10 sample. Different colors denote different values of the Seebeck coefficient as indicated by the color bar in Fig. 4c. The Seebeck coefficient is not uniform in the sample on a microscopic scale, varying from ∼300 to 400 ␮V/K, which is comparable with the value for the pressure-less sintered LAST-10 sample (see Fig. 2b). Chen et al. [19] also studied the LAST-m materials, prepared by the same method as described in ref. [1], by means of SMP and found the sample with inhomogeneous Seebeck coefficient. The variation of the Seebeck coefficient is even more significant: including both n-type and p-type behaviors in the same sample and the materials consist of multiphase, which is beyond the resolution of routine XRD and SEM measurements. They have deduced that this multiphase microstructure is caused by some subtle phase separation into two or more phases with very similar crystal structure and chemical composition, and that the phase separation causes local variations in dopant concentration, which leads to the variation in Seebeck coefficient. This phase separation mechanism may also apply to our sample. The Seebeck coefficient variation of our sample is gentler than that of the samples in ref. [19], which should be also due to different preparation routes and conditions. For an example, higher cooling rate will result in smaller microstructure features and less compositional variation, both of which will lead to more uniform Seebeck coefficient [19]. Some element Te exists in the samples will lead to the composition of the materials deviates from the nominal one and will lead to the samples not stable at high temperatures. Therefore, the hydrothermal synthesis conditions need to be further improved, such as increasing the temperature, prolonging the hold time, and reducing the size of the raw Te powder to increase its solubility in the solution. To prevent the nanopowder from oxidation is another key point. Preparation and compacting of pure LAST-m nanopowders are in progress. 4. Conclusions The samples prepared by compacting hydrothermally synthesized LAST-m nanopowders exhibit P-type conduction. The LAST-18 sample has better TE properties. The power factor of the pressure-less sintered LAST-18 sample and spark plasma sintered LAST-18 sample reaches 6.6 × 10−4 W m−1 K−2 at 480 K and 6.87 × 10−4 W m−1 K−2 at 393 K, respectively. Both the microstructure and electrical transport property of the two LAST-18 samples are found to be quite different. The hot pressed LAST-10 sample is highly densified and it is composed of fine

K.F. Cai et al. / Journal of Alloys and Compounds 469 (2009) 499–503

grains (down to nanoscale). SMP analysis reveals that the sample is microscopic inhomogeneous. Acknowledgements This work was sponsored by the Science and Technology Commission of Shanghai Municipality (05ZR14124), Shanghai Pujiang Program, and National Basic Research Program of China (2007CB607500). References [1] K.F. Hsu, S. Loo, F. Guo, W. Chen, J.S. Dyck, C. Uher, T. Hogan, E.K. Polychroniadis, M.G. Kanatzdis, Science 303 (2004) 818. [2] D. Bilc, S.D. Mahanti, E. Quarez, K.F. Hsu, R. Pcionek, M.G. Kanatzidis, Phys. Rev. Lett. 93 (2004) 146403. [3] H. Lin, E.S. Bozin, S.J.L. Billinge, E. Quarez, M.G. Kanatzidis, Phys. Rev. B 72 (2005) 174113. [4] E. Quarez, K.F. Hsu, R. Pcionek, N. Frangis, E.K. Poluchroniadis, M.G. kanatzidis, J. Am. Chem. Soc. 127 (2005) 9177. [5] T.C. Harman, T.J. Taylor, M.P. Walsh, B.E. Laforge, Science 297 (2002) 2229. [6] H. Hazama, U. Mizutani, R. Asahi, Phys. Rev. B 73 (2006) 115108.


[7] J. Androulakis, K.F. Hsu, R. Pcionek, C. Uher, J.J. D’Angelo, A. Downey, T. Hogan, M.G. Kanatzidis, Adv. Mater. 18 (9) (2006) 1170. [8] A. Kosuga, M. Uno, K. Kurosaki, S. Yamanaka, J. Alloys Compd. 386 (2005) 315. [9] A. Kosuga, M. Uno, K. Kurosaki, S. Yamanaka, J. Alloys Compd. 387 (2005) 52. [10] A. Kosuga, M. Uno, K. Kurosaki, S. Yamanaka, J. Alloys Compd. 391 (2005) 288. [11] A. Kosuga, K. Kurosaki, H. Muta, S. Yamanaka, J. Alloys Compd. 416 (2006) 218. [12] H. Wang, J.F. Li, C.W. Nan, M. Zhou, W.S. Liu, B.P. Zhang, T. Kita, Appl. Phys. Lett. 88 (2006) 092104–092106. [13] A.J. Karkamkar, M.G. Kanatzidis, J. Am. Chem. Soc. 128 (2006) 6002. [14] K.F. Cai, X.R. He, Mater. Lett. 60 (2006) 2461. [15] C. Yan, A.X. Zhang, X.R. He, K.F. Cai, Rare Met. Mater. Eng. 36 (Suppl. 2) (2007) 389–391. [16] D. Platzek, G. Karpinski, C. Drasar, E. M¨uller, Mater. Sci. Forum 587 (2005) 492. [17] J. Martin, G.S. Nolas, W. Zhang, L. Chen, Appl. Phys. Lett. 90 (2007) 222112. [18] K.F. Cai, Q. Lei, C. Yan, L.C. Zhang, Solid State Phenom. 121–123 (2007) 287. [19] N. Chen, F. Gascoin, G.J. Snyder, E. M¨uller, G. Karpinski, C. Stiewe, Appl. Phys. Lett. 87 (2005) 171903.