Enhanced Seebeck Coefficients of Thermoelectric Bi2Te3 Nanowiresas a Result of an Optimized Annealing ProcessJongmin Lee,*,† Jinwon Kim,‡ Wonjin Moon,§ Andreas Berger,∥ and Jaeyoung Lee*,†,‡

†Ertl Center for Electrochemistry and Catalysis, Laboratory for Energy Storage Systems, Research Institute of Solar and SustainableEnergies, Gwangju Institute of Science and Technology (GIST), Gwangju 500-712, South Korea‡School of Environmental Science and Engineering, Gwangju Institute of Science and Technology (GIST), Gwangju 500-712, SouthKorea§Korea Basic Science Institute (KBSI), Gwangju-Center, Gwangju, 500-757 South Korea∥Max Planck Institute of Microstructure Physics, Weinberg 2, D-06120 Halle, Germany

*S Supporting Information

ABSTRACT: Although an annealing approach for bulk or thick Bi−Te material had been done for the enhancement ofthermoelectric efficiency, this is the first time to report an annealing process for a Bi−Te nanostructure with a high aspect ratio.As is well-known, the Seebeck coefficient, one of the thermoelectric efficiency factors, is strongly influenced by the chemicalcomposition of the material. However, tellurium element tends to be easily evaporated in the high-temperature range (more than400 °C) due to its low melting point and high evaporation pressure. Furthermore, it is more serious in such a nanostructure witha high aspect ratio. In this article, we suggested the way to prevent the evaporation of tellurium element in the high annealingtemperature range (300−623 °C) and investigated why the Seebeck coefficients for annealed Bi−Te nanowires were enhanced.Seebeck coefficients for Bi−Te nanowires were improved 3 times higher than those for the as-prepared ones at the optimizedannealing process due to increased crystal defect concentrations, especially for edge dislocations. Consequently, this annealingprocess can be applied for nanostructured TE devices for the enhancement of thermopower (S).

■ INTRODUCTIONA lot of efforts have been pursued in order to enhance thedimensionless thermoelectric efficiency (figure of merit, ZT) byvarious aspects of approaches. Recent research trends can besummarized as follows: (i) looking for advanced new families ofthermoelectric materials and (ii) linking to nanoscalephenomena embedded in bulk samples as well as nanoscalesamples themselves.1,2 The former approach is mainly focusedon the significant reduction of thermal conductivity. TheKanatzidis group reported n-type thermoelectr icAgPbmSbTe2+m (LAST, ZT ∼ 2.2 at 800 K)3 and p-typethermoelectric Ag(Pb1−ySny)SbTe2+m (LASTT, ZT ∼ 1.45 at630 K)4 due to their reduction of thermal conductivity. Thelatter one is first focused on the reduction of lattice thermalconductivity owing to the introduction of plenty of nano-structured interfaces, leading to increased phonon boundaryscatterings5 and, second, focused on the enhancement of the

power factor owing to the quantum confinement effect and themodification of electronic band structures, leading to improvedthermopower (Seebeck coefficient) without a significantreduction of electrical conductivity.6 Thermopower (S) isgenerally dependent on the electronic band structure of thematerial near the Fermi energy level, which is a function of theconcentration of charge carriers and imperfections (or defectconcentrations), as well as the chemical composition ofmaterials.7 Bulk Bi−Sb−Te alloy with embedded nanostruc-tures exhibits ZT ∼ 1.4 at 373 K,8 and the Venkatasubramaniangroup9 reported superlattice structured Bi2Te3/Sb2Te3 filmwith ZT ∼ 2.4 due to a reduction of lattice thermalconductivity. In the perspective view of the modification of

Received: March 29, 2012Revised: June 23, 2012Published: August 6, 2012

Article

pubs.acs.org/JPCC

© 2012 American Chemical Society 19512 dx.doi.org/10.1021/jp3030039 | J. Phys. Chem. C 2012, 116, 19512−19516

electronic band structures for low-dimensional thermoelectricmaterials, such as nanowires10 and nanotubes,11 the annealingprocess would be a very efficient method for tuning thecrystallinity of nanostructures.Herein, since the number of reports have been very few so far

about the annealing effect on low-dimensional thermoelectricmaterials12 (almost no results reported about the annealingeffect on thermoelectric nanostructures) as well as thedemonstration of Seebeck coefficient results,13,14 we reporton the investigation of crystal defects of Bi−Te nanowires fromthe annealing process at optimized annealing conditions,leading to the corresponding Seebeck coefficient enhancementas a function of annealing temperature.

■ EXPERIMENTAL SECTION

Annealing Process for Bi−Te Nanowires. Bi−Tenanowires were fabricated by pulsed electrodeposition in ananodic aluminum oxide (AAO) membrane with a diameter of50 nm. Detailed procedures were described in our previousworks.15,16 It is well known that Te compound is easilyevaporated at certain annealing temperatures in bulk and thinfilm structures due to its high vapor pressure.17 Therefore,nanowire samples were put in a completely sealed ceramiccontainer by alumina paste (alumina paste used is resistant upto 500 °C) in order to prevent the evaporation of Te elementin nanowires. Additionally, pure Te powder was put together inthis container. All of the preparation process was performed ina glovebox. As-prepared nanowires in an alumina matrix wereannealed from 150 to 400 °C for 4 h in quartz chamber with a

nitrogen (N2) atmosphere. The annealing temperature wasincreased at every 50 °C.

Investigation on Crystalline Structures of Nanowires.The crystalline structure of nanowires was investigated by X-raydiffraction (XRD), high-resolution transmission electronmicroscopy (HR-TEM, Phillips, Technai, F20, Gwangju centerKBSI at 200 kV), and selected area electron diffraction (SAED)patterns. The composition of the nanowires was investigated byenergy-dispersive X-ray spectroscopy (EDX) attached to aTEM system.

Seebeck Coefficient Measurement for Bi−Te Nano-wires in Alumina Matrix. As-prepared and annealed oneswere placed on a sample holder, which contained wood metal.Wood metal is made up of bismuth, lead, tin, and cadmiumwith a melting point of 70 °C. The probe tip with a diameter ofseveral tens of micrometers directly contacted on the nanowiresamples within the alumina matrix while a stable temperaturegradient between the probe tip and the samples was applied for5 K. Spatial resolution of the Seebeck coefficient wascharacterized by a potential Seebeck microprobe (PSM)(Panco, Germany) with at least more than 30 points. Theaverage Seebeck coefficient was calculated based on thededuced values from more than 30 points.

■ RESULTS AND DISCUSSIONFigure 1 shows the elemental analysis of the nanowire by theline scan of EDX attached to the TEM system as a function ofannealing temperature. In particular, independent of theannealing temperature, the atomic percentage of Bi and Tefor individual nanowires exhibited stably 40% and 60%,

Figure 1. (a) Representative atomic percentage profiles of elements (Bi and Te) for an individual as-prepared nanowire and (b) an individualnanowire annealed at 350 °C. Note that the data were confirmed by EDX coupled with a TEM system performing a line scan across the nanowire.(c) EDX elemental mapping of Bi and Te for the nanowires annealed at 250 and 350 °C, respectively.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp3030039 | J. Phys. Chem. C 2012, 116, 19512−1951619513

respectively, as shown in Figure 1a,b. Additionally, EDXelemental mapping images for the nanowires annealed at 250and 350 °C revealed the uniform distribution of Bi and Tethroughout the nanowire, as shown in Figure 1c. On the basisof the above results, Bi−Te nanowires exhibit a thermodynami-cally stable Bi2Te3 phase, indicating that Te element is notevaporated during the annealing process. However, thechemical composition of the nanowires annealed above 400°C led to the imbalance of the stoichiometric composition(Bi2Te3). More details are described in the SupportingInformation (Figure S1 and Table S1).The Seebeck coefficient for Bi−Te nanowires was measured

by PSM, and the results are shown in Figure 2. Seebeck

coefficients show a negative value, indicating an n-typethermoelectric material. It is obviously demonstrated that theSeebeck coefficient increases linearly and is significantlyincreased at the annealing temperature of 350 °C. The Seebeckcoefficient (average: −119.68 (μV/K)) for the nanowiresannealed at 350 °C became 3 times higher than that (average:−38.99 (μV/K)]) for as-prepared nanowires. It is expected thatthe influence of the Seebeck coefficient by annealing ispositively attributed to the combination of enhanced crystalquality and increased defect concentrations. In addition,Yamashita et al. reported that annealing had a favorable effecton the improvement in ZT of n-type bismuth telluride bulk

compounds, but an adverse effect on the p-type ones.18 On theother hand, the Seebeck coefficient measured for the nanowiresannealed at 400 °C was decreased to −69.25 (μV/K).However, the data measured are not trustable, which will beexplained in the next figure.From the abundance distribution of the Seebeck coefficient

measured, as shown in Figure 3, it was revealed that theSeebeck coefficient values for as-prepared nanowires and thenanowires annealed up to 350 °C mostly belonged to the verynarrow regime, which exhibited a high homogeneity, but theSeebeck coefficient for the nanowires annealed at 400 °Cbelonged to the very broad regime, which showed muchinhomogeniety. It is in good agreement with the previousreport that the homogeneity was reduced despite theenhancement of the Seebeck coefficient with the increase ofannealing temperature.14 Therefore, the Seebeck coefficientvalue for the sample annealed at 400 °C is not reasonable dueto inhomogeneous variations.Figure 4 shows high-resolution transmission electron

microscopy (HRTEM) and selected area electron diffraction(SAED) patterns of as-prepared nanowires and the nanowiresannealed at different temperatures. The crystal quality of as-prepared individual nanowires and the individual nanowiresannealed up to 350 °C shows a very good single-crystallinestructure without any significant change. It will be confirmed byXRD data for the nanowires within the alumina matrix annealedat different temperatures, as shown in the SupportingInformation (Figure S2). Therefore, the influence on theenhancement of the Seebeck coefficient caused by improvedcrystal quality is insufficient. On the other hand, there aresignificant differences in the crystal defects, as shown in theHRTEM of Figure 4. As usual in the HRTEM (straight latticeplanes) of the as-prepared nanowire, a lot of dislocations(unusual lattice arrangements) were observed in the HRTEMof the nanowire annealed at 350 °C. In particular, consideringthe magnified HRTEM of the nanowire annealed at 350 °C, itwould be concluded that edge dislocations were formed duringthe annealing process due to stress and strains associated withthis temperature. Edge dislocations are a kind of crystallo-graphic defect where an extra half-plane of atoms is introducedmidway through the crystal, distorting nearby planes of atoms,like the magnified HRTEM.19 On the basis of the above results,it is good evidence that the enhancement of the Seebeck

Figure 2. Seebeck coefficient variations for Bi−Te nanowires within analumina matrix as a function of annealing temperature.

Figure 3. Abundance distribution of the Seebeck coefficient measured on as-prepared nanowires and the nanowires annealed at 350 and 400 °C.Seebeck coefficients for samples were measured by several points, respectively. Especially, the Bi−Te nanostructure annealed at 400 °C showed anonrepresentative value.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp3030039 | J. Phys. Chem. C 2012, 116, 19512−1951619514

coefficient is influenced by the increased defect concentrations(edge dislocations with the increase of annealing temperature).As the annealing temperature approached 400 °C, partial

diffusion of Te compound was observed in the nanowire.Figure 5a shows HRTEM of the individual nanowire annealed

at 400 °C. Dislocations were not observed any more, indicatingthe straight lattice arrangement like the nanowire annealedlower than 350 °C. However, the unusual feature, as indicatedby the arrow, was interestingly observed in another area of thesame nanowire, as shown in Figure 5c. To the best of ourknowledge, based on that the melting point of Te element is440 °C, the Te compound in the nanowire itself or added in

the container was possibly diffused in the nanowire at thisannealing temperature, leading to the coexistence of the Bi2Te3phase and a relatively Te-rich Bi−Te phase due to incongruentmelting of the Te element. More details are described in thebinary Bi−Te phase diagram (Supporting Information, FigureS3). According to the SAED, as shown in Figure 5b, it isrevealed that the nanowire basically possessed a single-crystalline structure with the Bi2Te3 phase as a major one.However, several weak (bright) spots, which are indicated by anarrow, were additionally observed. It indicated that another Bi−Te phase as a minor one coexisted with the Bi2Te3 phase.Unfortunately, it is not exactly understood what the Te-richBi−Te phase is. Furthermore, comparing with the FFT ofFigure 5c-1, which is investigated by the area except for theindication of an arrow (Figure 5c), several faint spots were alsoobserved, as shown in the FFT of Figure 5c-2, which isinvestigated by the indication of an arrow (Figure 5c). It is alsosupporting that Bi2Te3 and Te-rich Bi−Te phases coexisted inthe nanowire annealed at 400 °C. Consequently, thecoexistence of two kinds of phases from the diffusion of Teelement, leading to the unbalanced chemical composition ofBi−Te nanowires, is potentially influencing on the reduction ofthe Seebeck coefficient. Additionally, up to an annealingtemperature of 350 °C, the nanowires should be chemicallystable due to the reduction of the Gibbs free energy providedfrom the thermal energy during the annealing process.However, when they are annealed at more than 350 °C, the

nanowires are chemically unstable due to more Te evaporationfactor than chemical stability. It could be concluded that 350 °Cis the most appropriate annealing temperature in theperspective view of enhancement of thermopower (S).

■ CONCLUSIONSThermoelectric bismuth telluride nanowires fabricated bypulsed electrodeposition in an anodic aluminum oxidemembrane were annealed at different temperatures in orderto enhance the Seebeck coefficient. To prevent the evaporationof Te element due to high vapor pressure, the container forannealing was completely sealed with the addition of pure Te

Figure 4. (upper) Selected area diffraction pattern (SAED) and (bottom) high-resolution transmission electron microscopy (HRTEM) for as-prepared nanowires and the nanowires annealed at different temperatures as well as the magnified-view HRTEM of the nanowire annealed at 350°C. Note that at least the features of HRTEM for all the samples annealed up to 300 °C are very similar with those for the as-prepared nanowire or itis not easy to find crystal defects in the HRTEM despite the existence of dislocations.

Figure 5. (a, b) High-resolution transmission electron microscopy(HRTEM) and selected area electron diffraction (SAED) pattern forthe individual nanowire annealed at 400 °C. (c) HRTEM at otherareas for the same nanowire and the fast Fourier transformation (FFT)images depending on the selected area.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp3030039 | J. Phys. Chem. C 2012, 116, 19512−1951619515

powder. Consequently, the annealing process for nanostruc-tures at optimized conditions (350 °C) resulted in theenhancement of the Seebeck coefficient for Bi−Te nanowiresdue to increased crystal defect (edge dislocations) concen-trations, maintaining chemical stability, and the single-crystalline structure. However, with further increasing theannealing temperature, the homogeneity of the Seebeckcoefficient distribution became worse and the Seebeckcoefficient also was decreased due to the diffusion of Teelement with the high aspect ratio of nanowire structures,leading to an unbalanced chemical composition of nanowiresand the coexistence of Bi2Te3 (major one) and a relatively Te-rich Bi−Te phase (minor one).It will be expected that the annealing process at optimized

conditions can be favorably applied to the future thermoelectricnanostructured devices with high thermoelectric efficiency.

■ ASSOCIATED CONTENT

*S Supporting InformationSEM and EDX for overgrown film, XRD, and phase diagram fornanowires are shown for a better understanding of the low-dimensional Bi2Te3 structure and phase change as a function ofannealing temperature. This material is available free of chargevia the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: jongmin@gist.ac.kr, jaeyoung@gist.ac.kr. Tel: +82-62-715-2440. Fax: +82-62-715-2434.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by the Core Technology Develop-ment Program for Next-generation Energy Storage of ResearchInstitute for Solar and Sustainable Energies (RISE), GIST. Thiswork was supported by the "Basic Research Projects in High-tech Industrial Technology" Project through a grant providedby GIST in 2012.

■ REFERENCES(1) Dresselhaus, M. S.; Chen, G.; Tang, M. Y.; Yang, R.; Lee, H.;Wang, D.; Ren, Z.; Fleurial, J. P.; Gogna, P. Adv. Mater. 2007, 19,1043−1053.(2) Sootsman, J. S.; Chung, D. Y.; Kanatzdis, M. G. Angew. Chem., Int.Ed. 2009, 48, 8616−8639.(3) Hsu, K. F.; Loo, S.; Guo, F.; Chen, W.; Dyck, J. S.; Uher, C.;Hogan, T.; Polychroniadis, E. K.; Kanatzdis, M. G. Science 2004, 303,818−821.(4) Androulakis, J.; Hsu, K. F.; Pcionek, R.; Kong, H.; Uher, C.;D’Angelo, J. J.; Downey, A.; Hogan, T.; Kanatzdis, M. G. Adv. Mater.2006, 18, 1170−1173.(5) Hochbaum, A. I.; Chen, R.; Delgado, R. D.; Liang, W.; Garnett, E.C.; Najarian, M.; Majumdar, A.; Yang, P. Nature 2008, 451, 163−167.(6) Hicks, L. D.; Dresselhaus, M. S. Phys. Rev. B 1993, 47, 16631−16634.(7) Hochbaum, A. I.; Yang, P. Chem. Rev. 2010, 110, 527−546.(8) Poudel, B.; Hao, Q.; Ma, Y.; Lan, Y.; Minnich, A.; Yu, B.; Yan, X.;Wang, D.; Muto, A.; Vashaee, D.; et al. Science 2008, 320, 634−638.(9) Venkatasubramanian, R.; Siivola, E.; Colpitts, T.; O’Quinn, B.Nature 2001, 413, 597−602.(10) Lee, J.; Fahrangfar, S.; Yang, R.; Scholz, R.; Alexe, M.; Gosele,U.; Lee, J.; Nielsch, K. J. Mater. Chem. 2009, 19, 7050−7054.

(11) Cao, Y. Q.; Zhu, T. J.; Zhao, X. B. J. Alloys Compd. 2008, 449,109−112.(12) Kim, M. Y.; Oh, T. S.; Kim, J. S. J. Korean Phys. Soc. 2007, 50,670−676.(13) Huang, H.; Luan, W.; Tu, S. Thin Solid Films 2009, 517, 3731−3734.(14) Li, S.; Soliman, H. M. A.; Zhou, J.; Toprak, M. S.; Muhammed,M.; Platzek, D.; Ziolkowski, P.; Muller, E. Chem. Mater. 2008, 20,4403−4410.(15) Lee, J.; Fafrangfar, S.; Lee, J.; Cagnon, L.; Scholz, R.; Gosele, U.;Nielsch, K. Nanotechnology 2008, 19, 365701.(16) Lee, J.; Kim, Y.; Cagnon, L.; Gosele, U.; Lee, J.; Nielsch, K. Phys.Status Solidi RRL 2010, 4, 43−45.(17) Lee, J.; Berger, A.; Cagnon, L.; Gosele, U.; Lee, J.; Nielsch, K.PCCP Phys. Chem. Chem. Phys. 2010, 12, 15247−15250.(18) Taylor, A.; Mortensen, C.; Rostek, R.; Nguyen, N.; Johnson, D.C. J. Electron. Mater. 2010, 39, 1981−1986.(19) Yamashita, O.; Sugihara, S. J. Mater. Sci. 2005, 40, 6439−6444.(20) Reed-Hill, R. E.; Abbaschian, R. Physical Metallurgy Principles,3rd ed.; PWS Publishing Company: Boston, MA, 1992.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp3030039 | J. Phys. Chem. C 2012, 116, 19512−1951619516