Design and Optimization of Gradient Interface of p-TypeBa0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 Thermoelectric Materials

HONG-YU ZHOU,1 WEN-YU ZHAO,1,2 GANG LIU,1 HONG CHENG,1

and QING-JIE ZHANG1,3

1.—State Key Laboratory of Advanced Technology for Materials Synthesis and Processing, WuhanUniversity of Technology, Wuhan 430070, China. 2.—e-mail: wyzhao@whut.edu.cn. 3.—e-mail:zhangqj@whut.edu.cn

A series of p-type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 (FS/BT) thermoelectric(TE) materials containing one gradient layer (1GL) with FS/BT volume ratio of5:5, 3GLs-I with 7:3–5:5–3:7, 3GLs-II with 3:7–5:5–7:3, and 3GLs-III with 3:7–5:5–3:7 from FS to BT were fabricated by a two-step spark plasma sinteringmethod. The interface structure and mechanical properties of the p-typeFS/BT TE materials were investigated in this work. Designing the GLs at theinterface of FS and BT bulk TE materials can effectively relax the thermalstress induced by the large difference in thermal expansion coefficient andeliminate the macroscopic cracks that occur in FS/BT TE materials with noGL, hence resulting in a remarkable enhancement in the interface mechanicalproperties of the FS/BT TE materials with the GLs. The optimized gradientinterface of the FS/BT TE materials is 3GLs-II with FS/BT volume ratio of3:7–5:5–7:3. The highest flexural strength of the 3GLs-II sample reached13.68 MPa, increased by 116%.

Key words: Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 thermoelectric materials,gradient interface, microstructure, mechanical properties

INTRODUCTION

Solid-state thermoelectric (TE) devices are newenergy devices which can convert heat directly intoelectricity and vice versa. TE devices have manyadvantages, such as no moving parts, environmen-tal friendliness, reliability, and so on, and have beensuccessfully used in space power generation and avariety of cooling applications.1,2 However, there isstill a long way to large-scale applications for TEdevices because of their low conversion efficiency.3

Extending the working temperature range of TEdevices is considered as an effective way to increasetheir efficiency. Use of segmented TE modules,composed of different TE materials, has been con-firmed to enable achievement of high TE efficiencyover a wide temperature range.4 Among variousstate-of-the-art TE materials, Bi2Te3-based alloysare the most widely used materials for TE coolers

and power generation in the low temperature rangeof 300 K to 500 K,5,6 whereas filled skutterudites areidentified as some of the candidates for industrialpower generation applications in the intermediatetemperature range of 500 K to 800 K.7 In recentyears, several groups have reported segmentedBi2Te3/CoSb3 TE devices with high efficiency.8–10

However, as a key technology for fabricating seg-mented TE devices, the joining of two kinds of TEmaterials is still an obstacle. Muller et al.11 founddelamination cracks in the FeSi2 region near theinterface when Ni was selected as a joint interfacefor Bi2Te3/FeSi2. Snowden et al.12 discovered thatthe sample separated at the interface when Mo wasselected as a joint interface for Zn4Sb3 and (Bi,Sb)2

(Te,Se)3, and the same bulk materials exhibitedcracks when Fe was selected as the joint interface.The mismatch of the thermal expansion coefficientbetween two kinds of TE materials might producelarge interface thermal stress and result in macro-scopic cracks or even interface fracture. Accordingto functional gradient material (FGM) theory, a

(Received July 7, 2012; accepted September 25, 2012;published online November 10, 2012)

Journal of ELECTRONIC MATERIALS, Vol. 42, No. 7, 2013

DOI: 10.1007/s11664-012-2285-1� 2012 TMS

1436

gradient interface buffer layer can remarkably de-crease the interface thermal stress.13 Itoh et al.14

reported that designing a composite layer (mixingvolume ratio of 7:3) between Co0.92 Ni0.08Sb2.96Te0.04

and Bi2Te3 bulk materials might relax the interfacethermal stress. In previous work, we prepared aseries of n-type Ba0.4In0.4Co4Sb12/Bi2Te2.7Se0.3 TEmaterials containing three gradient layers (GLs)with Ba0.4In0.4Co4Sb12/Bi2Te2.7 Se0.3 volume ratioof 3:7–5:5–7:3, which avoided macroscopic cracksand achieved a maximum flexural strength of�12.76 MPa.15 However, a TE device generallyconsists of p-type and n-type TE materials connectedin parallel.16 It is very important to prepare p-typeFS/BT TE materials with GLs to fabricate TEdevices with high efficiency over a wide temperaturerange.

In this work, we used p-type Ba0.3In0.3FeCo3Sb12

(FS) and p-type Bi0.48Sb1.52Te3 (BT) TE materials asstarting materials to design a series of transitionlayers, and fabricated a series of p-type FS/BT TEmaterials with different GLs by employing the two-step spark plasma sintering method. Microstruc-tures of the gradient interface and their effects onthe mechanical properties of the p-type FS/BT TEmaterials were investigated. The phase compositionand stability of the 3GLs-II sample with the opti-mized gradient interface were further investigated.

EXPERIMENTAL PROCEDURES

Preparation of Ba0.3In0.3FeCo3Sb12

and Bi0.48Sb1.52Te3 Powders

A sample with nominal composition Ba0.3In0.3Fe-Co3Sb12 was prepared by the melt-quenching andannealing method. The annealed products wereground into Ba0.3In0.3FeCo3Sb12 (FS) powder. Theheat-treatment process was confirmed to be suitablefor preparing Ba/In double-filled skutterudite pow-der in our previous study.17 A mixture with nominalcomposition Bi0.48Sb1.52Te3 was heated up to 1073 Kfor 12 h, followed by quenching in oil. The quenchedingots were ground into Bi0.48Sb1.52Te3 (BT) powder.

Fabrication of p-Type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 TE Materials with GradientInterface

A two-step SPS method was developed to fabri-cate p-type FS/BT TE materials with gradientinterface. The fabrication process was presented ina previous report.15 Firstly, the Ba0.3In0.3FeCo3Sb12

powder was sintered into p-type FS bulk material bya first step of SPS at 893 K for 10 min. Secondly,homogeneously mixed powders for fabricating fourkinds of transition layer were designed fromBa0.3In0.3FeCo3Sb12 and Bi0.48Sb1.52Te3 powderswith FS/BT volume ratios from FS to BT bulkmaterials of 5:5, 7:3–5:5–3:7, 3:7–5:5–7:3, and 3:7–5:5–3:7, respectively. Thirdly, the FS bulk material,the mixed powders, and the Bi0.48Sb1.52Te3 powder

were loaded into a graphite die in turn. Finally, aseries of p-type FS/BT TE materials with differentgradient interfaces were fabricated by a second stepof SPS at 773 K for 10 min with pressure of 35 MPa.The sample names, layer numbers, and FS/BT vol-ume ratios of the transition layers are listed inTable I.

Measurement of Thermal ExpansionCoefficient of p-Type Ba0.3In0.3FeCo3Sb12

and Bi0.48Sb1.52Te3 Bulk Materials

The coefficients of thermal expansion (CTE) ofp-type FS and BT bulk materials were measured bya thermal expansion instrument (Neztsch DIL420C). Samples with dimensions of 4 mm 9 4 mm 910 mm were used for the experiments.

Measurement of Mechanical Propertiesof p-Type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3

TE Materials

Flexural strength measurements of strip sampleswith dimensions of 3 mm 9 4 mm 9 15 mm werecarried out using an INSTRON1341 electricityliquid service material testing machine. A sche-matic view of the three-point bending test setup isdepicted in Fig. 1. The modified test setup consistsof two parallel supports for the sample and a singleloading pin in the middle where the force is applied.

Table I. Sample name, layer number, and FS/BTvolume ratios of the transition layers

SampleName

LayerNumber

FS/BTVolume Ratiofrom FS to BT

0GL No gradient layer1GL 1 5:53GLs-I 3 7:3–5:5–3:73GLs-II 3 3:7–5:5–7:33GLs-III 3 3:7–5:5–3:7

Fig. 1. Schematic view of the three-point bending test unit.

Design and Optimization of Gradient Interface of p-Type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3

Thermoelectric Materials1437

The span between the supports is 10 mm. To obtainreliable flexural strength values, we tested sixsamples of each FS/BT TE material.

Textures, Phase Composition, Microstructure,and Element Distribution of the OptimizedGradient Interface

The textures of the GLs were analyzed by XJZ-6Amicroscope with MVC300 digital camera. The con-stituent phases of all samples were determined bypowder X-ray diffraction (XRD, PANalytical X’PertPRO) using Cu Ka radiation (k = 0.15418 nm). Themicrostructure and element distribution at the opti-mized gradient interface were determined by elec-tron probe microanalysis (EPMA, JXA-8230; JEOL).

RESULTS AND DISCUSSION

Coefficient of Thermal Expansion of p-TypeBa0.3In0.3FeCo3Sb12 and Bi0.48Sb1.52Te3 BulkMaterials

Figure 2 shows the temperature dependences ofthe CTEs of p-type FS and BT bulk materials in thetemperature range of 40�C to 350�C. It can be seenthat the CTEs of p-type FS and BT bulk materialsgradually increase with increasing temperature. Atthe same time, the CTEs of BT are much higherthan those of FS, by 79.7% at 300�C and by 120.1%at 350�C. Obviously, there is a large difference inCTEs between FS and BT. The interface mustexhibit large cracks when the two kinds of materialare connected together directly. Therefore, it isnecessary to relax the interface thermal stress bydesigning GLs between the FS and BT.

Effects of the GL on the Interface of p-TypeBa0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 TEMaterials

Figure 3 shows optical microscopy photographsof p-type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 TE

materials with (a) no GL (0GL) and (b) one GL(1GL) with FS/BT volume ratio of 5:5. It can be seenthat a crack of about 0.4 mm in length in the FSbulk material and a crack of about 1 mm in lengthin the BT bulk material appear next to the interfaceas shown in Fig. 3a. This is attributed to the largedifference in CTE between FS and BT. Fortunately,the length of the large cracks was significantlyshortened when 1GL with about 0.3 mm thicknesswas fabricated at the interface of the FS/BT TEmaterial, as shown in Fig. 3b. This indicates thatdesigning 1GL between p-type FS and BT bulkmaterials can effectively relax the thermal stressinduced by the large difference of CTE. We tried toeliminate the large cracks completely by increasingthe thickness of the 1GL from 0.5 mm to 4.5 mmand adjusting the FS/BT volume ratio from 5:5 to7:3. However, these attempts were not successful.This implies that it is difficult to eliminate the largecracks by fabricating only 1GL at the interface ofFS/BT TE material.

Figure 4 shows optical microscopy photographs ofp-type FS/BT TE materials containing (a) 3GLs-Iwith FS/BT volume ratio of 7:3–5:5–3:7, (b) 3GLs-IIwith FS/BT volume ratio of 3:7–5:5–7:3, and (c)3GLs-III with FS/BT volume ratio of 3:7–5:5–3:7.The thickness of the three kinds of 3GLs is about4.5 mm. Although excessive length of GLs could bedecreased the conversion efficiency, it can be seenthat the macroscopic cracks occurring in the FSbulk material side completely disappeared for theTE materials with 3GLs-II and 3GLs-III, as shownin Fig. 4b, c. Therefore, it can be concluded that3GLs is more effective for relaxing the thermalstress induced by the large difference in CTEbetween FS and BT than is 1GL. It is worth notingthat the FS/3:7-GL interface of the 3GLs-II sample

Fig. 2. Temperature dependences of the CTE for p-type FS and BTbulk materials.

Fig. 3. Optical microscopy photographs of p-type Ba0.3In0.3Fe-Co3Sb12/Bi0.48Sb1.52Te3 TE materials containing (a) no GL (0GL)and (b) one GL (1GL) with FS/BT volume ratio of 5:5.

Zhou, Zhao, Liu, Cheng, and Zhang1438

has more perfect microstructure compared withthose of the FS/7:3-GL interface of the 3GLs-I and3GLs-III samples. This experimental result wasrepeated many times, consistent with previousreports,15 indicating that increasing the BT content

Fig. 4. Optical microscopy photographs of p-type Ba0.3In0.3Fe-Co3Sb12/Bi0.48Sb1.52Te3 TE materials containing (a) 3GLs-I with FS/BT volume ratio of 7:3–5:5–3:7, (b) 3GLs-II with FS/BT volume ratioof 3:7–5:5–7:3, and (c) 3GLs-III with FS/BT volume ratio of 3:7–5:5–3:7.

Fig. 5. BEI photographs of the gradient interface of Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 TE materials containing 3GLs-II with FS/BT volumeratio of 3:7–5:5–7:3.

Fig. 6. Flexural strength of Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 TEmaterials with different GLs.

Design and Optimization of Gradient Interface of p-Type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3

Thermoelectric Materials1439

in the GL close to the FS bulk material side canmore effectively relax the thermal stress.

Microstructure of the Optimized GradientInterface

Figure 5 shows backscattered electron imaging(BEI) photographs of the gradient interface ofp-type FS/BT TE material containing 3GLs-II withFS/BT volume ratio of 3:7–5:5–7:3. The bright areasare BT, while the black ones are FS. The whole

gradient interface is closely connected and forms alenticular structure. There are no microscopiccracks on the micron scale in the lenticular struc-ture. The FS and BT grains coated each other. Thelenticular structure plays an important role ineliminating the macroscopic crack.

Effect of the GLs on the MechanicalProperties of p-Type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 TE Materials

The flexural strength values of p-type FS/BT TEmaterials with different GLs are shown in Fig. 6.The flexural strength of the FS/BT TE material withno GL is only about 6.32 MPa. The flexural strengthof the FS/BT TE material with 1GL reached8.72 MPa, increased by 38% compared with thesample with no GL. The highest flexural strengthreached 13.68 MPa for the FS/BT TE material with3GLs-II, being increased by 116%. It can be con-cluded that the GLs can remarkably improve theinterfacial mechanical properties of the FS/BT TEmaterials. The optimized gradient interface of theFS/BT TE materials is 3GLs-II with FS/BT volumeratio of 3:7–5:5–7:3. To study the fracture path ofthe 3GLs-II samples, Fig. 7 shows a photograph ofan untested 3GLs-II sample and the tested ones.The untested 3GLs-II sample shows good connec-tion with no macroscopic cracks. The 3GLs-II sam-ples separate at the FS-GLs interface after beingtested by INSTRON1341. It can be seen that failureinitiates at the FS-GLs interface region, and thefracture path is situated on the FS-GLs interface.This is attributed to the greater residual thermalstress that is distributed near the FS-GLs interfaceof all the FS/BT TE materials with GLs.

Phase Composition and Element Distributionof the Optimized Gradient Interface

XRD patterns of the GLs, BT, and FS of 3GLs-IImaterial and mixed powders of BT and FS withFS/BT volume ratio of 5:5 for fabricating the 3GLs-II (MP-GLs) are shown in Fig. 8. The phase com-positions of FS near the side of the 3GLs-II interfaceremain invariable, consisting of a main skutteruditephase with traces of FeSb2. The phase compositionof BT near the side of the 3GLs-II interface is asingle phase of BT. No other impurity phases weredetected on the 3GLs-II gradient layers. Theseresults suggest that FS, BT, and GLs materials arestable during the two-step SPS process.

BEI and EDS of the gradient interface of FS/BTTE material with 3GLs-II are shown in Fig. 9. TheEDS data show that the phase compositions of theGLs materials were stable, being the main skutt-erudite phase, Bi2Te3, and traces of FeSb2.Figure 10 shows BEI and line distributions of Ba,In, Fe, Co, Sb, Bi, and Te elements of the optimizedgradient interface of p-type FS/BT TE material with3GLs-II. The different microstructures of the three

Fig. 7. Untested and tested 3GLs-II samples.

Fig. 8. XRD patterns of BT side, 3GLs-II gradient layers, and FSside of the optimized gradient interfaces of Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 TE materials. The XRD pattern of mixed powders ofBT and FS with FS/BT volume ratio of 5:5 for fabricating the 3GLs-II(MP-GLs) is also shown to explain the formation mechanism ofinterface phases.

Zhou, Zhao, Liu, Cheng, and Zhang1440

GLs from the BT to the FS side are clearly distin-guished by BEI. The whole 3GLs-II interface is verycompact. The line distributions of Co, Fe, Sb, Bi, andTe elements show that chemical diffusion did nottake place between BT and FS material after thesecond step of SPS. No reaction of interface phasecompositions and no obvious diffusion of chemicalcompositions, occurring at the gradient interface ofp-type FS/BT TE material with 3GLs-II, indicate

that the GL materials are stable during the secondstep of SPS.

CONCLUSIONS

The gradient interface between p-type Ba0.3In0.3

FeCo3Sb12 (FS) and Bi0.48Sb1.52Te3 (BT) bulkmaterials was designed and optimized. The GLs caneffectively relax the thermal stress induced by the

Fig. 9. Backscattered electron imaging and EDS of the optimized gradient interface of p-type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 TE materialwith 3GLs-II.

Fig. 10. BEI and line distributions of Co, Sb, Bi, Te, and Se elements of the optimized gradient interface of p-type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3 TE material with 3GLs-II.

Design and Optimization of Gradient Interface of p-Type Ba0.3In0.3FeCo3Sb12/Bi0.48Sb1.52Te3

Thermoelectric Materials1441

large difference of CTE between FS and BT andthus remarkably improve the interfacial mechanicalproperties of the FS/BT TE materials. The flexuralstrength of the FS/BT TE material with 1GLreached 8.72 MPa, increased by 38% compared withthat of the FS/BT TE material with no GL.Increasing the BT content in the GL near to the FSbulk material side may effectively relax the thermalstress and completely eliminate the cracking. Theinterfacial structures and mechanical properties ofthe FS/BT TE materials with 3GLs-I, 3GLs-II, and3GLs-III indicate that the optimized gradientinterface of the FS/BT TE materials is 3GLs-II withFS/BT volume ratio of 3:7–5:5–7:3. The highestflexural strength reached 13.68 MPa, increased by116%. The interface phase compositions and thechemical compositions of GL materials are stableduring the second step of SPS. A lenticular struc-ture occurring at the optimized gradient interfaceplays an important role in relaxing the thermalstress and eliminating the macroscopic cracks.Further research will be focused on optimizing thethickness of GLs based on the FS/BT TE materialswith 3GLs-II and investigating their transportproperties by designing a special test platform.

ACKNOWLEDGEMENTS

This work was supported by the National BasicResearch Program of China (973-program) underProject No. 2013CB632505, National Natural Sci-ence Foundation of China (Nos. 50930004, 50972114,

11274248), and Program for New Century ExcellentTalents in University (NCET-09-0627).

REFERENCES

1. J. Yang and T. Caillat, MRS Bull. 31, 224 (2006).2. L.E. Bell, Science 321, 1457 (2008).3. B.C. Sales, Science 295, 1248 (2002).4. M.S. El-Genka, H.H. Sabera, and T. Caillat, Energy Con-

vers. Manage. 44, 1755 (2003).5. W.J. Xie, X.F. Tang, Y.G. Yan, Q.J. Zhang, and T.M. Tritt,

J. Appl. Phys. 105, 113713 (2009).6. H.P. Feng, B. Yu, K. Collins, C. He, Z.F. Ren, and G. Chen,

Electrochim. Acta 56, 3079 (2011).7. R.H. Liu, J. Yang, X.H. Chen, X. Shi, L.D. Chen, and C.

Uher, Intermetallics 19, 1747 (2011).8. K.T. Wojciechowski, R. Zybala, J. Leszczynski, P. Nieroda,

M. Schmidt, R. Gajerski, and E. Aleksandrova, 9th EuropeanConference on Thermoelectrics (Thessaloniki, Greece: AIPConf. Proc., 2011).

9. M.S. El-Genk, H.H. Saber, T. Caillat, and J. Sakamoto,Energy Convers. Manage. 47, 174 (2006).

10. J. Wang, X.F. Tang, H.Q. Liu, X.L. Yang, and Q.J. Zhang,J. Wuhan Univ. Technol. 21, 126 (2006).

11. E. Muller, C. Drasar, J. Schilz, and W.A. Kaysser, Mater.Sci. Eng. A 362, 17 (2003).

12. D.P. Snowden, D.T. Allen, B.A. Cook, and N.B. Elsner, 18thInternational Conference on Thermoelectrics (Pardubice,Czech Republic: IEEE, 1999).

13. T. Hirai and L. Chen, Mater. Sci. Forum 509, 308 (1999).14. T. Itoh, T. Muto, and K. Kitagawa, 25th International Con-

ference on Thermoelectrics (Vienna, Austria: IEEE, 2006).15. G. Liu, W.Y. Zhao, H.Y. Zhou, P. Wei, J. Yu, D.G. Tang, and

Q.J. Zhang, J. Electron. Mater. 41, 1376 (2012).16. G.J. Snyder and E.S. Toberer, Nat. Mater. 7, 105 (2008).17. J. Yu, W.Y. Zhao, P. Wei, D.G. Tang, and Q.J. Zhang,

J. Electron. Mater. 41, 1414 (2012).

Zhou, Zhao, Liu, Cheng, and Zhang1442