Melt spinning preparation of Bismuth Telluride and partially

Preparing to load PDF file. please wait...

0 of 0
Melt spinning preparation of Bismuth Telluride and partially

Transcript Of Melt spinning preparation of Bismuth Telluride and partially

Melt spinning preparation of Bismuth Telluride and partially alloying with IV-VI compounds for thermoelectric application
1H. Böttner, 1D.G. Ebling, 1A. Jacquot, 2Uta Kühn, 3J. Schmidt 1Fraunhofer-Institut für Physikalische Messtechnik IPM, Department of Thermoelectric Systems, Heidenhofstraße 8, D-
79110 Freiburg, Germany ²Leibniz-Institut für Festkörper- und Werkstoffforschung Dresden, Institut für Komplexe
Materialien IKM, Helmholtzstr. 20, D-01069 Dresden, Germany 3Fraunhofer-Institut für Fertigungstechnik und Angewandte Materialforschung IFAM, Institutsteil Dresden,
Winterbergstr. 28, D-01277 Dresden, Germany, E-mail: [email protected], Phone: +49 / 761 88 57 399, Fax: +49 / 761 8857 224

Abstract Intrinsic p- and n-doped Bismuth Telluride samples are
compared with respect to the evolution of their thermoelectric material parameters like thermal and electrical conductivity. The Seebeck coefficient is discussed in dependence on the melt spinning fabrication technique. The melt spinner used is only able to produce small thin ribbon shaped specimens, some as thin as 10 micron. This limits melt spinning to mainly production of research specimens for alloys with high critical cooling rate, which are difficult to fabricate with other techniques. Therefore this method was combined with spark plasma sintering (SPS) to produce wafer sized samples from the melt spinning flakes. Additional parameters are alloying or doping of the base material by comparing the properties as prepared to different annealing conditions. The intrinsic pand n-doped material was alloyed with up to 0.5% lead telluride by rapidly cooling the bulk material to improve the thermoelectric properties analysed from RT up to about 600 K. A Seebeck coefficient of well above 200 µV/K could be obtained for p- and n-type materials. For annealed melt spin prepared material with the composition [(Bi0,2Sb0,8)2Te3]0.97PbTe0.03 a compaction and the formation of a nanocomposite structure was obtained resulting a high powerfactor of about 34µW/K2cm.
Introduction The preparation of thermoelectric compounds of doped
bismuth telluride and its alloys has recently gained increasing interest due to their structural features showing increased values for the thermoelectric figure of merit (ZT) [1, 2].
One promising approach to improve the thermoelectric properties is to manufacture nanocomposite materials exhibiting lower thermal conductivities and higher ZT values. The use of nanostructured materials has led to recent breakthroughs in the thermoelectric field [1–3].
Beside the possibility to grow nanoscaled materials by epitaxial methods, it is attractive to produce much cheaper bulk materials with a nanoscaled structure exhibiting similar profits for thermoelectric applications. Mainly two basic methods are described. One is based on mechanical alloying and subsequent sintering of those mixtures [4], the other one is based on rapid cooling of melted materials exhibiting miscibility gaps in the phase diagram, which will form

nanoscaled precipitations in a bulk sample under these

conditions. [5-9]

Melt spinning, developed for rapid cooling of liquids like

metal melts, is used to develop materials that require

extremely high cooling rates in order to form, such as

metallic glasses or nanocomposites. The cooling rates achievable by melt-spinning are on the order of 101107Kelvin/s.






PbTe/Bi2Te3/Sb2Te3/Bi2Se3 is of special interest being

composed of good thermoelectric materials on the one hand

but with a restricted miscibility in the phase diagram e.g.

PbTe/ Bi2Te3. The miscibility in phase diagrams may even

differ severely for different authors, also indicating a high

uncertainty of the crystalline fine structure on the nanoscale.

[10-14]. Nanocomposites of „V2VI3“ with PbTe bulk

quenched material will be compared to the corresponding

melt spin material.

Melt spin n/p-“Bi2Te3“ For the melt spinning process parameters like the
rotation speed, extrusion pressure, casting temperature, distance nozzle to cooling wheel as well as the pressure and kind of gas in the recipient were varied.
The obtained samples of the melt spin process were flakes with a size of about 2 mm by 2 mm and a thickness of 30 to 100 µm. The flakes are highly oriented with the c-axis showing in the direction perpendicular to the surface, determined by oriented XRD-experiments. The specimens are very fragile. There was no difference in the composition

Seebeck coefficient [µV.K-1]
not annealed 473K 1h + 573K 1h in air 473K 1h + 573K 1h in N2 623K 3h in N2 680K 3h in N2 680K 1h sufficient

220 200 180 160 140 120 100
Figure 1: Improvement of Seebeck coefficient of p-type (Bi0.2Sb0.8)2Te3 prepared by the melt spinning method due to different annealing conditions.

of the samples measured with EDX between before or after Growth.

hour at 680 K under Nitrogen atmosphere. No influence was found of oxygen on the annealing process. The conditions hold as well for p-type material as they do for n –type material.

Figure 2: Seebeck coefficients of n-type Bi2(Se0.1Te0.9)3 prepared by the melt spinning method due to different preparation conditions of the sample. Annealing conditions were 1h at 680 K.
Measuring the thermoelectric properties of those samples results in a Seebeck coefficient for p-type material of about 140 µV/K. It is remarkable that no influence was observed from the variation of all the melt spinning process parameter mentioned above and that the variation of this value remained within the experimental error of less than 10 %. For n-type material the Seebeck coefficient became as low as -40 µV/K (fig. 2) and again there was also no evident influence of the process parameters to be seen.
A substantial improvement could be obtained for the thermoelectric properties by annealing of the samples. The annealing conditions were varied according to figure 1 in a temperature range from 473 K to 680 K and the annealing time between 1 and 3 hours.
Best conditions were found to be an annealing time of 1
Figure 3: Influence of the annealing procedure on the temperature dependence of the Seebeck coefficient for ptype (Bi0.2Sb0.8)2Te3 prepared by the melt spinning method. Annealing 1h at 680 K improves the thermal power only up to 450 K. For higher temperatures the thermoelectric properties of the not annealed sample give better results.

Figure 4: Improvement of the electrical conductivity of ntype Bi2(Se0.1Te0.9)3 prepared by the melt spinning method due to annealing 1h at 680 K.
For both materials we end up with Seebeck coefficients giving an absolute value as high as 210 µV/K (figure 2: for n-type material). It is remarkable that the initial value of the Seebeck coefficient for n-type material is only about half of that for p-type and that the n-type material shows a higher standard deviation for the as-grown material as well as for the annealed one which implies, that the n-material is less reproducible and might show a higher degree of disorder .
The temperature dependence of the Seebeck-coefficient (figure 3) shows improvement only for temperatures close to room temperature. Above 450 K the as grown material exhibits better values of the Seebeck-coefficient. This might be due to the different micro- or nanostructure of the as grown material which influences the mobility and concentration of the charge carriers as well as the heat conductivity. But the conditions are evidently not stable for the as grown material, since during measuring its temperature dependence an annealing effect is observed leading to irreproducible results. Figure 3 shows only the data of the first measurement for the not annealed sample. A second scan exhibits lower values for high temperatures and higher thermopower at room temperature.
Regarding the electrical conductivity of the melt spin samples we observe a similar behavior. While the p-material shows only a minor change of the electrical conductivity of initially 400 S/cm connected with a reduced standard deviation after annealing (+/-100 reduced to +/-10), we observe for the n-type material an increase of the electrical conductivity of more than 300% (figure 4). The drawn line indicates the increase of conductivity as a guide for the eye.
Alloying of (Bi, Sb)2Te3 with PbTe To further improve the thermoelectric properties of the
material we started to alloy the Bi2Te3 compounds with PbTe as a promising possibility to form nanocomposites. The concentration of PbTe was varied between 0.05 at% and

0.5 at%, which is still in a range were some authors predict miscibility of this compounds [10].

Seebeck-coefficient [µV/K] electrical conductivity [S/cm]

PbTe (0% Te excess) - (Bi,Sb) Te


2 3 2400





130 120 1600







0,0 0,1 0,2 0,3 0,4 0,5
concentration of PbTe [wt%]

(ZT = 0.45) compared to the samples without PbTe (fig. 6). While that improvement at RT is most likely due to the improved electrical and reduced thermal conductivity, the increase by a factor of 3 at 550 K for alloys with 0.3% PbTe might be due to a combination of increased Seebeck coefficient and a substantially reduced thermal conductivity. That leads to the highest ZT value of 0.64. Further increase of the ZT seems to be possible due to annealing of the materials.
before annealing

ZT Seebeck coeff. [µV/K] Electrical cond. [S/cm]

Figure 5: Influence of the amount of PbTe alloying on the thermoelectric properties of in water quenched p-type
(Bi0.2Sb0.8)2Te3. Lines are guides to the eye.
The samples were prepared by melting the starting materials with appropriate initial weight in a quartz ampoule together; the ampoule were quenched in cold water. The microscopic inspection of the samples showed no evidence of any precipitation or inclusions down to the sub micron range. Also annealing of the samples did not produce any precipitation in that range. TEM inspection of the material is still in preparation and will be published later on.
The thermoelectrical properties are shown in figure 5. By increasing the amount of PbTe the Seebeck coefficient is reduced from about 160 to 100 µV/K while the electrical conductivity is increased from about 800 to 1600 S/cm, which might be explained by a doping effect due to the

0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0

0,1% 0,05% 0%

0,3% 0,5%

350 400 450 500 550 600 temperature [K]

Figure 6 Figure of merit ZT for p-type (Bi0.2Sb0.8)2Te3 with different concentrations of PbTe depending on measuring temperature. A clear improvement of ZT is observed for low concentrations of PbTe close to RT and for higher concentrations at elevated temperature up to 600 K, indicated by red arrows.
PbTe. The thermoelectric properties were analysed parallel to the surface at room temperature (RT) for p-type samples with PbTe up to 0.5%. The thermal conductivity of these samples shows a minimum for a concentration of 0.1% PbTe at temperatures from RT to 600K. This leads to an improvement of the figure of merit ZT at RT of about 20%

1 µm
after annealing

10 µm

1 µm

10 µm

Figure 7 Scanning electron micrographs indicating the improvement of the melt spin prepared p-type material (Bi0.2Sb0.8)2Te3 with 0.3 at% of PbTe by annealing 1h at 680°C under nitrogen.
For comparison, samples with a concentration of PbTe of about 0.3 at% were analyzed, which were prepared by the melt spinning process.
Figure 7 exhibits a SEM picture of the surface of such a sample before and after annealing. The as grown material exhibits a network like strukture of needles with a diameter of about 1 µm. After annealing this structure has become more compact and the magnification in the inset shows clearly structures of precipitations in the nm-range.
That change of the structure is also reflected in the thermoelectrical properties of the material, shown in figure

160 150 140 130 120 110 100
bef ore


1600 1400 1200 1000 af t er

Figure 8 Improvement of the electrical conductivity for p-type (Bi0.2Sb0.8)2Te3 prepared by the melt spinning method due to annealing 1h at 680 K.

8. The electrical conductivity remains more or less the same after annealing (1000 – 1300 S/cm) and is slightly higher compared to the in water quenched material (about 1000 S/cm). More interesting is the substantial improvement of the Seebeck coefficient from about 120 µV/K (the quenched and the meltspin material are the same) to about 160 µV/K by annealing the melt spin material. By this a power factor of ~34µW/K2cm was obtained by annealing. It seems, that the nanocomposite structure improves the thermopower and eventually the thermal behaviour offering the chance of ZT>1, which is still under investigation. These results will be published elsewhere.
Conclusions It could be demonstrated that melt spin fabrication can
exhibit relatively high Seebeck coefficients of more than 200 µV/K at RT after annealing. Although the process parameters of the melt spinning process do not remarkably influence the properties of the material, both p- and n-Bi2Te3 are improved after annealing substantially to exhibit higher Seebeck coefficients and better electrical conductivity. In general, n-type material seems to remain more unstable even after annealing.
For temperatures above RT the results are still not convincing and further experiments are necessary to obtain better effects. Alloying with PbTe is another promising possibility to produce nanocomposites due to the incomplete miscibility of the materials.
For (Bi0.2Sb0.8)2Te3 alloyed with up to 0.5% PbTe and quenched after melting in water the properties could already be improved for temperatures higher than RT. The thermal conductivity is significantly reduced and Pb-rich precipitations at grain boundaries of layered material could be observed. An improvement for ZT of nearly 30% was obtained at RT and for temperatures of about 550 K a factor of 3 was obtained for the improvement giving values for ZT of about 0.64. Further enhancement seems to be possible due to annealing of the material. While the properties of in water quenched material and for melt spin material are more or less the same, a substantial improvement of the thermoelectric properties was observed due to annealing of melt spin material, resulting a power factor as high as ~34µW/K2cm. The nanocomposite structure was shown by SEM for the best material with a composition of [(Bi0.2Sb0.8)2Te3]0.97PbTe0.03 after annealing, accompanied with a substantial campaction of the material. This material offers the chance for ZT >> 1.
Acknowledgments This work was supported by the FhG Internal Programs
under Grant No. WISA 815 020. The preparation of PbTealloyed samples is also acknowledged and was performed by the group of the LPM in Nancy, France, (Dr. Bertrand Lenoir) and by BASF in Ludwigshafen, Germany (Dr. Frank Haass).

2. T.C. Harman et al., “Quantum Dot Superlattice Thermoelectric Materials and Devices”, Science Vol. 297 (2002), 2229
3. K.F. Hsu et al., “Cubic AgPbmSbTe2+m: Bulk Thermoelectric Materials with High Figure of Merit”, Science Vol. 303 (2004), 818.
4. Jing Liu et al., “Bi2Te3 and Bi2Te3/nano-SiC Prepared by Mechanical Alloying and Spark Plasma Sintering”, Key Engineering Materials Vols. 280-283 (2005), 397.
5. H. Y. Chen et al., “Microstructures and thermoelectric properties of Fe0.92Mn0.08Six alloys prepared by rapid solidification and hot pressing”, Journal of Applied Physics Vol. 94 (2003), 6621.
6. E. Koukharenko et al., “Electrical properties of Bi2xSbxTe3 materials obteined by ultrarapid quenching”, Journal of Alloys and Coumpound Vol. 327 (2001), pp. 1-4.
7. V.M. Glazov et al., “Thermoelectric properties of semiconducting solid solution, Bi2Te2.4Se0.6 Bi0.52Sb1.48Te3 prepared by ultrafast cooling of melts”, Izv. Akad. Nauk, SSSR, Neorg. Mater. Vol. 22, No. 1 (1986), pp 36-40.
8. E. Kukharenka et al., “Electrical and microstructural properties of Bi2-xSbxTe and Bi2-xSbxTe2 foils obtained by ultrarapid quenching process”, Journal of Materials Science: Materialials in Electronics Vol. 14 (2003), pp. 383-388.
9. X. Tang et al., “Preparation and thermoelectric transport properties of high-performance p-type Bi2Te3 with layered nanostructure”, Applied Physics Letters Vol. 90 (2007), 012102.
10. T. Hirai et al., “The pseudo-binary V2VI3-IV VI compounds systems, Bi2Te3-PbTe, Bi2Te3-SnTe, Sb2Te3PbTe, Sb2Te3-SnTe and Bi2Se3-SnSe”, Journal of LessCommon Metals Vol. 13 (1967), pp. 352-356,
11. E.I. Elagina et al., “An investigation of the PbTe-Bi2Te3 and the SnTe-Sb2Te3 systems”, Russian Journal of Inorganic Chemistry Vol. 4, No. 7 (1959), pp. 738-740.
12. D. Kusano et al., “Effects of PbTe Doping on the Thermoelectric Properties of (Bi2Te3)0.2(Sb2Te3)0.8”, 21st International Conference on Thermoelectrics (2002), pp. 13-16.
13. P-W. Zhu et al., “High thermoelectric properties of PbTe Doped with Bi2Te3 and Sb2Te3”, Chinese Physics Letter Vol. 28, No. 8 (2005), pp. 2103-2104
14. P. Zhu et al., “Composition-Dependent Thermoelectric Properties of (PbTe)100-x(Bi2Te3-Sb2Te3)x (0.1≤x≤5)”, Materials Transactions Vol. 46, No. 4 (2005), pp. 761764.

References 1. R. Venkatasubramanian et al., “Thin-film thermoelectric
devices with high room-temperature figures of merit”, Nature Vol. 413 (2001), 597