Enhanced Thermoelectric Performance of Cu2SnSe3-Based Composites Incorporated with Nano-Fullerene

In this study, nano-sized fullerene C60 powder was sufficiently mixed with Cu2SnSe3 powder by ball milling method, and the C60/Cu2SnSe3 composites were prepared by spark plasma sintering technology. The fullerene C60 distributed uniformly in the form of clusters, and the average cluster size was less than 1 μm. With increasing C60 content, the electrical conductivity of C60/Cu2SnSe3 composites decreased, while the Seebeck coefficient was enhanced. The thermal conductivity of composites decreased significantly, which resulted from the phonon scattering by the C60 clusters located on the grain boundaries of the Cu2SnSe3 matrix. The highest figure of merit ZT of 0.38 was achieved at 700 K for 0.8% C60/Cu2SnSe3 composite.


Introduction
Thermoelectric (TE) materials have attracted increasing worldwide attention due to their potential application in electronic cooling, waste heat recovery, and power generation [1][2][3][4]. The conversion efficiency of TE materials is determined by the dimensionless figure of merit, ZT = α 2 σT/κ, where α is the Seebeck coefficient, σ is the electrical conductivity, T is the absolute temperature, and κ is the total thermal conductivity. The total thermal conductivity is composed of an electron part (κ E ) and a phonon part (κ L ). Therefore, to maximize the ZT value of TE materials, a large α and σ, as well as a low κ are required. In recent years, several classes of bulk materials with high ZT have been discovered and developed, such as skutterudites, clathrates, and Cu-based chalcogenide semiconductors.
Cu-based chalcogenide compounds with a diamond-like structure, such as ternary Cu 2 MSe 3 (M = Sn, Ge) and Cu 3 SbSe 4 , have attracted a lot of attention recently, due to their quite low thermal conductivity. In several Cu-based chalcogenide compound systems, the Cu 2 SnSe 3 structure has partial "phonon glass electron crystal" (PGEC) characteristic, which makes it possible to achieve high TE performance. The Cu-Se bond network dominates the electron conduction, while the contribution from the element Sn is very weak; thus, the Sn site is suitable for optimization of the TE property. Various attempts, including doping by partial substitution, have been made to improve the thermoelectric properties of Cu 2 SnSe 3 compound [5][6][7]. Shi et al. have reported the In-doped Cu 2 In x Sn 1-x Se 3 and the maximum ZT value reaches 1.14 at 850 K for Cu 2 In 0.1 Sn 0.9 Se 3 sample [8]. Fan et al. fabricated the Cu 2 Ga x Sn 1-x Se 3 samples using hot-press sintering technique and achieved the ZT value of 0.43 for Cu 2 Ga 0.075 Sn 0.925 Se 3 sample [9]. In addition, Skoug, et al. also confirmed that doping with isoelectronic Ge on the Sn site is also effective in enhancing the ZT value [10]. Aside from doping, the dispersion of nanostructure phases into the thermoelectric matrix is also an attractive approach to improving the performance of TE materials. However, the Cu 2 SnSe 3 -based thermoelectric composites are scarcely investigated, because the enhancement of ZT is unapparent compared with the doping. Although significant reduction in the lattice conductivity can be achieved via enhanced phonon scattering at grain boundaries or matrix/inclusion interfaces, the electrical conductivity of TE composites also decreases, resulting in a marginal improvement of the overall ZT value. In addition, good selection of the dispersed phase and the control of microstructure are also required for TE composite [11][12][13][14]. Therefore, effective enhancement of ZT for TE composites depends on the microstructure of composites-i.e., the distribution or the shape of the component.
Fullerene C 60 has very high elastic modulus and is a chemically-stable nonpolar fullerene molecule. C 60 and C 60 -decorated grain boundaries may provide an effective phonon scattering, which could decrease the lattice thermal conductivity. Meanwhile, the scattering of charge carriers (electrons or holes) by the C 60 could be ineffective due to the large value of electron (hole) wavelength compared to a fullerene molecule size. Blank and Kulbachinskii, et al. reported that the addition of 0.5 vol % C 60 improved the TE properties of Bi 0.5 Sb 1.5 Te 3 material, and the ZT value obtained was 1.17 at 450 K [15,16]. Shi, et al. found that adding 6.5 mass% C 60 into pure CoSb 3 can increase the ZT value, while adding amounts between 0.5% and 4.8% into CoSb 3 decreased the ZT value [17]. The similar results of Itoh, et al. showed that the maximum ZT value for the 1% C 60 /Co 0.92 Ni 0.08 Sb 2.96 Te 0.04 composite was 0.62 at 800 K, which was evidently higher than that of C 60 -free sample [18]. Nandihalli, et al. also reported that the ZT value of C 60 /Ni 0.05 Mo 3 Sb 5.4 Te 1.6 composites was enhanced in the whole temperature range due to the large decrease of κ L [19]. In this contribution, we attempted to introduce the C 60 into a Cu 2 SnSe 3 system and expected to achieve a larger reduction in the thermal conductivity of C 60 /Cu 2 SnSe 3 composites.
In the present work, the fullerene C 60 powder was incorporated into Cu 2 SnSe 3 matrix using ball milling (BM), and C 60 /Cu 2 SnSe 3 composites were fabricated by spark plasma sintering (SPS) technology. Effects of C 60 particles on the thermoelectric properties of C 60 /Cu 2 SnSe 3 composites were discussed, and the results are beneficial to the development of Cu 2 SnSe 3 -based composites with high performance using BM-SPS technology.

Experimental Procedures
The polycrystalline Cu 2 SnSe 3 samples were synthesized by melting method. The stoichiometric amount of starting materials Cu (powder, 99.95%), Sn (powder, 99.999%), and Se (shot, 99.999%) were first placed in a carbon crucible enclosed in evacuated fused-silica ampoules. The ampoules were slowly heated to 1173 K and held for 12 h in a vertical furnace. Then, the ampoules were slowly cooled to 873 K in 24 h, followed by annealing at this temperature for 2 days. Finally, the obtained ingots were ground into fine powder. Commercially available fullerene powder with average particle size of 500 nm (XFNANO, Nanjing, China) was chosen as the nanoinclusion, as shown in Figure 1. The fullerene C 60 purity is 99.98%, and the other 0.02% refers to impurities of C 70 and other carbon structures. The fullerene C 60 powder was added into the Cu 2 SnSe 3 powder at fractions of 0.4, 0.8, 1.2, and 1.6 vol %, respectively. Then, the C 60 -added Cu 2 SnSe 3 powders were mechanically ground with planetary ball milling equipment at 150 rpm for 240 min. The as-milled powders were sintered by spark plasma sintering (SPS 2040) at around 860 K for about 8 min under uniaxial pressure of 50 MPa in vacuum.
The density of the sintered C 60 /Cu 2 SnSe 3 composites was measured using the Archimedes method. The constituent phases of the samples were determined by X-ray diffractometry (Cu K α , Rigaku, Rint2000, Tokyo, Japan). The chemical composition of bulk samples was characterized using electron probe micro-analysis (EPMA, JEOL, JXA-8100, Tokyo, Japan) with a wavelength dispersive spectrometer (WDS). The composition was calculated by averaging five spots. The microstructure of all C 60 /Cu 2 SnSe 3 composites was observed by high-resolution transmission electron microscopy (HRTEM, JEM2100F, JEOL, Tokyo, Japan). The thermal diffusivity (λ) was measured by laser flash method (Netzsch LFA427) in a flowing Ar atmosphere between 300 and 700 K. The thermal conductivity was calculated from the relationship κ = ρλC p , where ρ is the density of the sintered sample and C p is the specific heat capacity. The electrical conductivity and Seebeck coefficient were measured simultaneously using commercial equipment (ZEM-3, ULVAC-RIKO, Tokyo, Japan) on a bar-type sample with a dimension of 2ˆ2ˆ10 mm 3 . The Hall coefficient (R H ) was measured using the van der Pauw's method in vacuum with a magnetic field of 2 T. The carrier concentration (p H ) and mobility (µ H ) were estimated from the relations of p H = 1/(eR H ) and µ H = σR H , based on the assumption of single band model, where e is the electronic charge. All measurements were performed in a temperature range of 300-700 K. The density of the sintered C60/Cu2SnSe3 composites was measured using the Archimedes method. The constituent phases of the samples were determined by X-ray diffractometry (Cu Kα, Rigaku, Rint2000, Tokyo, Japan). The chemical composition of bulk samples was characterized using electron probe micro-analysis (EPMA, JEOL, JXA-8100 Tokyo, Japan) with a wavelength dispersive spectrometer (WDS). The composition was calculated by averaging five spots. The microstructure of all C60/Cu2SnSe3 composites was observed by high-resolution transmission electron microscopy (HRTEM, JEM2100F, JEOL, Tokyo, Japan). The thermal diffusivity (λ) was measured by laser flash method (Netzsch LFA427) in a flowing Ar atmosphere between 300 and 700 K. The thermal conductivity was calculated from the relationship κ = ρλCp, where ρ is the density of the sintered sample and Cp is the specific heat capacity. The electrical conductivity and Seebeck coefficient were measured simultaneously using commercial equipment (ZEM-3, ULVAC-RIKO, Tokyo, Japan) on a bar-type sample with a dimension of 2 × 2 × 10 mm 3 . The Hall coefficient (RH) was measured using the van der Pauw's method in vacuum with a magnetic field of 2 T. The carrier concentration (pH) and mobility (μH) were estimated from the relations of pH = 1/(eRH) and μH = σRH, based on the assumption of single band model, where e is the electronic charge. All measurements were performed in a temperature range of 300-700 K. Figure 2 shows the SEM image of the 1.6 vol % C60-added Cu2SnSe3 powder after ball milling. It can be observed that the average particle size of milled C60/Cu2SnSe3 powder was about 100 nm. Figure 3 shows the X-ray diffraction patterns of xC60/Cu2SnSe3 composites (x = 0, 0.4, 0.8, 1.2, and 1.6 vol %) after SPS. The measured relative densities for all C60/Cu2SnSe3 composites after SPS are above 97% of the theoretical value. The diffraction peaks in Figure 3 are well-indexed based on the JCPDS 65-4145 (Joint Committee on Powder Diffraction Standards) of Cu2SnSe3. As the content of C60 in the composites is very low, the diffraction peak of C60 is not found in the XRD pattern of all C60/Cu2SnSe3 samples. Therefore, all C60/Cu2SnSe3 samples show the same XRD patterns with the pure Cu2SnSe3 sample.  Figure 2 shows the SEM image of the 1.6 vol % C 60 -added Cu 2 SnSe 3 powder after ball milling. It can be observed that the average particle size of milled C 60 /Cu 2 SnSe 3 powder was about 100 nm. Figure 3 shows the X-ray diffraction patterns of xC 60 /Cu 2 SnSe 3 composites (x = 0, 0.4, 0.8, 1.2, and 1.6 vol %) after SPS. The measured relative densities for all C 60 /Cu 2 SnSe 3 composites after SPS are above 97% of the theoretical value. The diffraction peaks in Figure 3 are well-indexed based on the JCPDS 65-4145 (Joint Committee on Powder Diffraction Standards) of Cu 2 SnSe 3 . As the content of C 60 in the composites is very low, the diffraction peak of C 60 is not found in the XRD pattern of all C 60 /Cu 2 SnSe 3 samples. Therefore, all C 60 /Cu 2 SnSe 3 samples show the same XRD patterns with the pure Cu 2 SnSe 3 sample.    Figure 4a,b show the SEM microstructure of the sintered pure Cu2SnSe3 sample and 1.6 vol % C60/Cu2SnSe3 composite, respectively. The fullerene C60 distributed uniformly in the form of clusters, and the average cluster size was lower than 1 μm. Shi, et al. reported that in the CoSb3 material, most of the C60 molecules agglomerate into irregular micrometer-size clusters located at the grain boundaries [17]. The smaller size of clusters in this study should be due to the ball milling technology. The chemical composition of C60/Cu2SnSe3 composites was characterized by SEM and energy-dispersive X-ray spectroscopy (EDS), as shown in Figure 5. The results of EDS also confirm that the matrix was analyzed to be composed of 33.53 at. %; Cu, 16.85 at. %; Sn, and 49.62 at. %; Se, corresponding to the Cu2SnSe3 phase. The black phase only contains C element, indicating C60 phase. To further analyze the C60 clusters in the C60/Cu2SnSe3 composite, HRTEM of C60/Cu2SnSe3 composite was carried out, as shown in Figure 6. The size of C60 is about 80 nm, which means the ball milling process decreases the average size of C60 particles. According to the theory proposed by Faleev and Zebardaji, et al. [20,21], nano-phases that distribute in the thermoelectric matrix can result in strain fields, which could cause some changes in the band structure of the material and then greatly influence its thermoelectric properties.  Figure 4a,b show the SEM microstructure of the sintered pure Cu 2 SnSe 3 sample and 1.6 vol % C 60 /Cu 2 SnSe 3 composite, respectively. The fullerene C 60 distributed uniformly in the form of clusters, and the average cluster size was lower than 1 µm. Shi, et al. reported that in the CoSb 3 material, most of the C 60 molecules agglomerate into irregular micrometer-size clusters located at the grain boundaries [17]. The smaller size of clusters in this study should be due to the ball milling technology. The chemical composition of C 60 /Cu 2 SnSe 3 composites was characterized by SEM and energy-dispersive X-ray spectroscopy (EDS), as shown in Figure 5. The results of EDS also confirm that the matrix was analyzed to be composed of 33.53 at. %; Cu, 16.85 at. %; Sn, and 49.62 at. %; Se, corresponding to the Cu 2 SnSe 3 phase. The black phase only contains C element, indicating C 60 phase. To further analyze the C 60 clusters in the C 60 /Cu 2 SnSe 3 composite, HRTEM of C 60 /Cu 2 SnSe 3 composite was carried out, as shown in Figure 6. The size of C 60 is about 80 nm, which means the ball milling process decreases the average size of C 60 particles. According to the theory proposed by Faleev and Zebardaji, et al. [20,21], nano-phases that distribute in the thermoelectric matrix can result in strain fields, which could cause some changes in the band structure of the material and then greatly influence its thermoelectric properties.       Figure 7 shows the temperature dependence of electrical conductivity (σ) for C60/Cu2SnSe3 composites with different vol % C60. It can be seen that the σ of the Cu2SnSe3 matrix decreases approximately linearly with rising temperature over the measured temperature range, indicating a typical behavior of a heavily-doped semiconductor. The similar tendency of σ was also observed in Figure 6. High-resolution transmission electron microscopy (HRTEM) image of a C 60 particle in the C 60 /Cu 2 SnSe 3 composite. Figure 7 shows the temperature dependence of electrical conductivity (σ) for C 60 /Cu 2 SnSe 3 composites with different vol % C 60 . It can be seen that the σ of the Cu 2 SnSe 3 matrix decreases approximately linearly with rising temperature over the measured temperature range, indicating a typical behavior of a heavily-doped semiconductor. The similar tendency of σ was also observed in C 60 /Cu 2 SnSe 3 composites. In addition, the σ of C 60 /Cu 2 SnSe 3 composites decreases with increasing C 60 content, which should be attributed to the enhanced carrier scattering at the incoherent interfaces between well-dispersed C 60 clusters and the Cu 2 SnSe 3 matrix. Generally, in the case of carriers primarily scattered by grain barriers or interfaces between the second phase and matrix in the composites, the carrier mobility can be written as [22]

Electrical Transport Properties
where b is the average grain size, k B the Boltzmann constant, m* the carrier effective mass, and E B the activation energy characterizing the barrier height between the matrix and the second phase. As the relative density of the xC 60 /Cu 2 SnSe 3 composite is higher than 97%, the porosity effect can be eliminated. Table 1 lists some physical and structural parameters of xC 60 /Cu 2 SnSe 3 composites at room temperature. As the C 60 could act as an electron acceptor in the p-type C 60 /Cu 2 SnSe 3 composite, the carrier concentration increases with increasing C 60 content, which is consistent with the results of Blank [15]. In addition, it can be noted that the carrier mobility decreases with increasing C 60 content. Therefore, the σ of xC 60 /Cu 2 SnSe 3 composites decreases compared with the σ of the Cu 2 SnSe 3 matrix.
where kB, σ(E), and n(E) are Boltzmann constant, electrical conductivity, and value of density of states (DOS), respectively. Many studies have confirmed that when nano-phases or nano-inclusions are incorporated into a semiconducting matrix material, the band bending at the inclusion/matrix interface will produce a potential energy barrier which could effectively block low energy electrons, while transmitting high energy electrons [24]. This "electron energy filter" could evidently increase the local density of states near the Fermi level (EF) and enhance the Seebeck coefficient.   Figure 8 displays the Seebeck coefficient (α) of xC 60 /Cu 2 SnSe 3 composites as a function of temperature. All composites have a positive α across the whole temperature range, indicating that the holes are major carriers. With rising temperature, the α of all xC 60 /Cu 2 SnSe 3 composites increases approximately linearly and the α of 1.6% C 60 /Cu 2 SnSe 3 composite reaches 314 µV/K at 700 K. Moreover, the α of xC 60 /Cu 2 SnSe 3 composites significantly increases with the increasing content of C 60 . At room temperature, the α increases from 130 µV/K for the Cu 2 SnSe 3 matrix to 252 µV/K for the 1.6 vol % C 60 /Cu 2 SnSe 3 composite. The enhancement of α of xC 60 /Cu 2 SnSe 3 composites should be related to the "energy filter" effect. The Seebeck coefficient can be expressed as [23], where k B , σ(E), and n(E) are Boltzmann constant, electrical conductivity, and value of density of states (DOS), respectively. Many studies have confirmed that when nano-phases or nano-inclusions are incorporated into a semiconducting matrix material, the band bending at the inclusion/matrix interface will produce a potential energy barrier which could effectively block low energy electrons, while transmitting high energy electrons [24]. This "electron energy filter" could evidently increase the local density of states near the Fermi level (E F ) and enhance the Seebeck coefficient.  Figure 8 displays the Seebeck coefficient (α) of xC60/Cu2SnSe3 composites as a function of temperature. All composites have a positive α across the whole temperature range, indicating that the holes are major carriers. With rising temperature, the α of all xC60/Cu2SnSe3 composites increases approximately linearly and the α of 1.6% C60/Cu2SnSe3 composite reaches 314 V/K at 700 K. Moreover, the α of xC60/Cu2SnSe3 composites significantly increases with the increasing content of C60. At room temperature, the α increases from 130 V/K for the Cu2SnSe3 matrix to 252 V/K for the 1.6 vol % C60/Cu2SnSe3 composite. The enhancement of α of xC60/Cu2SnSe3 composites should be related to the "energy filter" effect. The Seebeck coefficient can be expressed as [23], where kB, σ(E), and n(E) are Boltzmann constant, electrical conductivity, and value of density of states (DOS), respectively. Many studies have confirmed that when nano-phases or nano-inclusions are incorporated into a semiconducting matrix material, the band bending at the inclusion/matrix interface will produce a potential energy barrier which could effectively block low energy electrons, while transmitting high energy electrons [24]. This "electron energy filter" could evidently increase the local density of states near the Fermi level (EF) and enhance the Seebeck coefficient.  The µ H of xC 60 /Cu 2 SnSe 3 composites is shown in Figure 9. The µ H of xC 60 /Cu 2 SnSe 3 composites decreases with increasing C 60 content. In addition, the µ H of xC 60 /Cu 2 SnSe 3 composites was in the order of 10 cm 2¨V´1¨s´1 at room temperature, which was close to that of skutterudites [25,26]. It may be caused by the similar carrier effective mass of Cu 2 SnSe 3 and CoSb 3 compounds. The m* can be estimated in the single parabolic band model using the following equations, where F r , h, and r are Fermi integral, Planck's constant, and scattering parameter of relaxation time, respectively. The evaluated equivalent carrier effective mass of xC 60 /Cu 2 SnSe 3 composites at room temperature is listed in Table 1. It can also be seen from Figure 9 that the µ H of pure Cu 2 SnSe 3 shows a temperature dependence of T´1 .5 above 500 K, indicating that the acoustic phonon scattering is dominant in the temperature range from 500 to 700 K. Below 500 K, the µ H of pure Cu 2 SnSe 3 has a weak temperature dependence relationship and the relationship of µ H 9T´0 .5 is observed, suggesting that a dominative mechanism is alloy scattering. However, the µ H of xC 60 /Cu 2 SnSe 3 composites deviates from the T´1 .5 or T´0 .5 dependence over the entire temperature range, indicating that a mixed scattering mechanism dominates these samples.

(4)
where Fr, h, and r are Fermi integral, Planck's constant, and scattering parameter of relaxation time, respectively. The evaluated equivalent carrier effective mass of xC60/Cu2SnSe3 composites at room temperature is listed in Table 1. It can also be seen from Figure 9 that the μH of pure Cu2SnSe3 shows a temperature dependence of T −1.5 above 500 K, indicating that the acoustic phonon scattering is dominant in the temperature range from 500 to 700 K. Below 500 K, the μH of pure Cu2SnSe3 has a weak temperature dependence relationship and the relationship of μH∝T −0.5 is observed, suggesting that a dominative mechanism is alloy scattering. However, the μH of xC60/Cu2SnSe3 composites deviates from the T −1.5 or T −0.5 dependence over the entire temperature range, indicating that a mixed scattering mechanism dominates these samples.  Figure 10 displays the temperature dependence of total thermal conductivity (κ) and lattice thermal conductivity (κL) for C60/Cu2SnSe3 composites. The κL is estimated by subtracting the electronic contribution via the Wiedmann-Franz law (κE = L0σT, where the Lorenz number L0 is taken as a constant of 2.0 × 10 −8 V 2 /K 2 ) from the total thermal conductivity. The κ for all samples declines with increasing temperature. Moreover, the κ of xC60/Cu2SnSe3 composites decreases with increasing C60 content. The achieved κ of 1.6 vol % C60/Cu2SnSe3 composite at room temperature is 1.85 W/mK, which is 34% lower than that of pure Cu2SnSe3. The minimal κ of 1.6 vol % C60/Cu2SnSe3 composite is 0.71 W/mK at 700 K. It is well-known that the grain boundary, wide or point defects, porosity, and impurity could contribute to the decrease of κ. Owing to high relative density of C60/Cu2SnSe3 composites, the reduction of κ originating from the porosity is negligible. Meanwhile, the calculation of κE shows that the reduction of κE has a limited contribution to the decrease of κ. Therefore, the decrease of κ for C60/Cu2SnSe3 composites mainly originates from the depression of κL due to the enhancement of phonon scattering by the C60 inclusions or nano-particles in the composite. Just as shown in Figure 10b, the κL of C60/Cu2SnSe3 composites drastically decreases with  Figure 10 displays the temperature dependence of total thermal conductivity (κ) and lattice thermal conductivity (κ L ) for C 60 /Cu 2 SnSe 3 composites. The κ L is estimated by subtracting the electronic contribution via the Wiedmann-Franz law (κ E = L 0 σT, where the Lorenz number L 0 is taken as a constant of 2.0ˆ10´8 V 2 /K 2 ) from the total thermal conductivity. The κ for all samples declines with increasing temperature. Moreover, the κ of xC 60 /Cu 2 SnSe 3 composites decreases with increasing C 60 content. The achieved κ of 1.6 vol % C 60 /Cu 2 SnSe 3 composite at room temperature is 1.85 W/mK, which is 34% lower than that of pure Cu 2 SnSe 3 . The minimal κ of 1.6 vol % C 60 /Cu 2 SnSe 3 composite is 0.71 W/mK at 700 K. It is well-known that the grain boundary, wide or point defects, porosity, and impurity could contribute to the decrease of κ. Owing to high relative density of C 60 /Cu 2 SnSe 3 composites, the reduction of κ originating from the porosity is negligible. Meanwhile, the calculation of κ E shows that the reduction of κ E has a limited contribution to the decrease of κ. Therefore, the decrease of κ for C 60 /Cu 2 SnSe 3 composites mainly originates from the depression of κ L due to the enhancement of phonon scattering by the C 60 inclusions or nano-particles in the composite. Just as shown in Figure 10b, the κ L of C 60 /Cu 2 SnSe 3 composites drastically decreases with the content of C 60 increasing. The minimal κ L achieved in the present work is 0.68 W/mK at 700 K for the 1.6 vol % C 60 /Cu 2 SnSe 3 sample, which is 43% lower than that of pure Cu 2 SnSe 3 . According to the kinetic theory [27], the minimum lattice thermal conductivity κ Lmin can be obtained when the phonon mean free path reaches the shortest interatomic distance. The κ Lmin can be estimated from the formula κ L = 1/3C v ν m l, where C v is heat capacity per unit volume of the system using Dulong and Petit value, ν m the mean sound velocity, and l the mean free path of phonon. The ν m comes from the data in reference [28]. If we assume the minimum l to be the interatomic distance for Cu 2 SnSe 3 (0.238 nm), the κ Lmin is calculated as 0.52 Wm´1¨K´1, just as shown by dashed line in Figure 10b. The κ L of 1.6 vol % C 60 /Cu 2 SnSe 3 composites approaches the κ Lmin of Cu 2 SnSe 3 at high temperature.

Thermal Transport Properties
formula κL = 1/3Cvνml, where Cv is heat capacity per unit volume of the system using Dulong and Petit value, νm the mean sound velocity, and l the mean free path of phonon. The νm comes from the data in reference [28]. If we assume the minimum l to be the interatomic distance for Cu2SnSe3 (0.238 nm), the κLmin is calculated as 0.52 Wm −1 •K −1 , just as shown by dashed line in Figure 10b. The κL of 1.6 vol % C60/Cu2SnSe3 composites approaches the κLmin of Cu2SnSe3 at high temperature.

Figure of Merit
The conversion efficiency of TE materials depends on the maximum dimensionless figure of merit (ZT). Figure 11 shows the dimensionless figure of merit (ZT) of C60/Cu2SnSe3 composites as a function of temperature. Like other doped Cu2SnSe3 investigated before [8,29,30], the ZT value of C60/Cu2SnSe3 composites increases approximately linearly with increasing temperature. Compared with the ZT of the Cu2SnSe3 sample, the ZT value of C60/Cu2SnSe3 composites is enhanced. For the 0.8 vol % C60/Cu2SnSe3 sample, the maximum ZT value is 0.38 at 700 K, which is 45% higher than that of the pure Cu2SnSe3 sample. The enhancement of ZT for C60/Cu2SnSe3 composites is mainly attributed to the reduced κL and the enhanced α. The addition of C60 into the Cu2SnSe3 matrix could improve the TE properties, which is a promising process to the design Cu-based chalcogenide compounds with high TE performance. When the material with optimized carrier concentration is selected as the matrix, the higher ZT value of TE composite could be achieved.

Figure of Merit
The conversion efficiency of TE materials depends on the maximum dimensionless figure of merit (ZT). Figure 11 shows the dimensionless figure of merit (ZT) of C 60 /Cu 2 SnSe 3 composites as a function of temperature. Like other doped Cu 2 SnSe 3 investigated before [8,29,30], the ZT value of C 60 /Cu 2 SnSe 3 composites increases approximately linearly with increasing temperature. Compared with the ZT of the Cu 2 SnSe 3 sample, the ZT value of C 60 /Cu 2 SnSe 3 composites is enhanced. For the 0.8 vol % C 60 /Cu 2 SnSe 3 sample, the maximum ZT value is 0.38 at 700 K, which is 45% higher than that of the pure Cu 2 SnSe 3 sample. The enhancement of ZT for C 60 /Cu 2 SnSe 3 composites is mainly attributed to the reduced κ L and the enhanced α. The addition of C 60 into the Cu 2 SnSe 3 matrix could improve the TE properties, which is a promising process to the design Cu-based chalcogenide compounds with high TE performance. When the material with optimized carrier concentration is selected as the matrix, the higher ZT value of TE composite could be achieved.

Conclusions
In this study, C60 was incorporated into a Cu2SnSe3 matrix, and C60/Cu2SnSe3 composites were fabricated using BM-SPS method. The C60 phase distributed uniformly in the form of clusters, and the average cluster size was less than 1 μm. With increasing C60 content, the electrical conductivity of C60/Cu2SnSe3 composites decreased, while the Seebeck coefficient of C60/Cu2SnSe3 composites increased due to the "electron energy filter" of the C60 nano-phase. The thermal conductivity of C60/Cu2SnSe3 composites decreased significantly, which originated from the phonon scattering by the C60 clusters located on the grain boundaries of the Cu2SnSe3 matrix. The maximum ZT of 0.38 was obtained at 700 K for 0.8 vol % C60/Cu2SnSe3 composite.

Figure 11.
Temperature dependence of the dimensionless figure of merit of (ZT) of C 60 /Cu 2 SnSe 3 composites.

Conclusions
In this study, C 60 was incorporated into a Cu 2 SnSe 3 matrix, and C 60 /Cu 2 SnSe 3 composites were fabricated using BM-SPS method. The C 60 phase distributed uniformly in the form of clusters, and the average cluster size was less than 1 µm. With increasing C 60 content, the electrical conductivity of C 60 /Cu 2 SnSe 3 composites decreased, while the Seebeck coefficient of C 60 /Cu 2 SnSe 3 composites increased due to the "electron energy filter" of the C 60 nano-phase. The thermal conductivity of C 60 /Cu 2 SnSe 3 composites decreased significantly, which originated from the phonon scattering by the C 60 clusters located on the grain boundaries of the Cu 2 SnSe 3 matrix. The maximum ZT of 0.38 was obtained at 700 K for 0.8 vol % C 60 /Cu 2 SnSe 3 composite.