It is of great importance to develop clean and renewable energy sources due to the environmental pollution and resource shortage. Thermoelectric materials are most promising alternatives with the capacity of directly converting heat into useful electricity based on the Seebeck effect.^[1,2] Despite the great success in the commercialized application in the space exploration,^[3] the performance of thermoelectric devices is suffered from the low conversion efficiency. The thermoelectric conversion efficiency is mainly determined by the dimensionless figure of merit zT = S²σT/κ, where S is the Seebeck coefficient, σ is the electrical conductivity, T is the absolute temperature, κ is the total thermal conductivity composed of the lattice thermal conductivity κ_L and electronic thermal conductivity κ_e. Since the concept of phonon glasses and electron crystal was put forward, the filled skutterudites with high power factor, relative low thermal conductivity and excellent mechanical properties have been regarded as the promising mid-temperature TE materials.^[4–6]

The filling elements, such as rare earth, alkali metals, alkali earth and other species,^[7,8] into interstitial voids not only tune the carrier concentration by the donor doping, but also reduce the lattice thermal conductivity by strengthening the phonon scattering. Therefore, thermoelectric properties have been remarkably improved, especially for n-type skutterudites. The maximum zT value about 1.8 is obtained by adding In_0.4Co₄Sb₁₂ to triple-filled (Sr, Ba, Yb)_yCo₄Sb₁₂ and using high energy ball-milling method.^[9] As high-performance thermoelectric devices, both p-type and n-type materials with high zT values are needed. However, the reported zT values of p-type filled skutterudites are lower than zT values of n-type skutterudites.^[10–19] Therefore, development of zT values in p-type skutterudites remain challenging.

Nanocomposite is proved to be an effective method to improve the thermoelectric properties.^[20–23] Carbon-based materials have been selected as a second phase to boost the thermoelectric properties. Shi et al.^[24,25] reported the effect of fullerence (C₆₀) on the thermoelectric properties of CoSb₃ or Ba_yCo₄Sb₁₂. In such materials, the lattice thermal conductivity reduced due to the enhanced scattering of phonon. Feng et al.^[26] fabricated the CoSb₃/graphene nanocomposite by an in situ solvothermal route combined with hot pressing, remarkable increasing zT value. Such an outstanding performance could be attributable to the reduced lattice thermal conductivity and remarkable increased power factor induced by the nanostructured matrix and the second phase incorporation. Wan et al.^[27] found that the flexural strength and fracture toughness of the CeFe₄Sb₁₂ materials improved by introducing short carbon fibers while little impact on the thermoelectric performance. Zhang et al.^[28] assembled SKDs with both single- and multi-walled carbon nanotubes via a cryogenic grinding technique to enhance the thermoelectric and mechanical properties. Recently, Zong et al.^[29,30] reported the skutterudites-based composites by uniformly mixing the hydrophilic graphene oxide and skutterudites and followed by in situ reduction with spark plasma sintering. The result indicated that the introduction of graphene had minimal effect on the electrical properties, but evidently suppressed the lattice thermal conductivity, leading to enhanced figure merit of zT.

From this perspective, this study shows an improved zT value in skutterudites with the addition of graphene by plasma-enhanced chemical vapor deposition (PECVD) method. The purposely designed skutterudite/graphene nanocomposites exhibit higher Seebeck coefficient and lower lattice thermal conductivity than pure skutterudite.

Figure 1 shows the Raman spectra of La_0.8Ti_0.1Ga_0.1Fe₃-CoSb₁₂ and La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene composite powders. Compared to graphene-free powders, it can be seen that two characteristic peaks of graphene are presented at 1340 cm⁻¹ and 1580 cm⁻¹ for composite powders, which are corresponded to the D and G peaks, respectively. This indicates that graphene is successfully grew on the surface of skutterudite powders by the PECVD method.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Raman spectra of skutterudite and skutterudite/graphene composite powders.

Figure 2 shows the x-ray diffraction patterns of skutterudite/graphene composite. The peak located at 2θ around 26° indicates that a small amount of graphite phase occurred for the composites. From the results of Raman spectra and XRD, it can be concluded that the graphene is successfully introduced into skutterudite matrix. The lattice parameters of all samples determined by refined XRD data and the real filled fraction obtained via polyzonal EDS results are listed in Table 1. The similar lattice constant and real composition of both samples illustrate that the graphene is located in the boundary without causing the lattice distortion.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) XRD patterns of the sintered skutterudite/graphene sample.

Figure 3 shows the temperature-dependent electrical properties of the La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample and graphene-free sample. As shown in Fig. 3(a), the electrical conductivity σ shows similar values in the range of measurement temperature. Meanwhile, the electrical conductivity is decreased with increasing temperature for both samples, displaying typical degenerating semiconductor behavior.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) Temperature-dependent electrical properties of skutterudite and skutterudite/graphene samples (a) electrical conductivity, (b) Seebeck coefficient, (c) powder factor.

Figure 3(b) shows the temperature-dependent of Seebeck coefficient for La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample and graphene-free one. Both samples show positive value of Seebeck coefficient S, indicating p-type semiconductor. It is clearly seen that Seebeck coefficient keeps almost same below 473 K for both samples. However, for La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample, the Seebeck coefficient is slightly higher than that of reference sample above 473 K.

Table 1.

Table 1.

Table 1. La actual filling fraction and lattice parameter for skutterudite/graphene and skutterudite. . Nominal composition La (actual filling fraction) Lattice parameter/Å La0.8Ti0.1Ga0.1Fe3CoSb12/graphene 0.68 9.110 La0.8Ti0.1Ga0.1Fe3CoSb12 0.70 9.113

Table 1. La actual filling fraction and lattice parameter for skutterudite/graphene and skutterudite. .

According to the above electrical conductivity and Seebeck coefficient, the power factors of the both samples are obtained using the relationship PF = S²σ as shown in Fig. 3(c). The La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample shows the slightly improved power factor above 473 K due to the similar electrical conductivity and slightly enhanced Seebeck coefficient.

Figure 4 shows temperature-dependent thermal conductivity of La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample and graphene-free sample. From Fig. 4(a), it can be seen that the total thermal coductivity κ for both samples has similar trend, which is first reduced and then increased with increasing temperature due to the Umklapp scattering and bipolar effect. Both samples have almost same thermal conductivity temperature below 373 K. However, the total thermal conductivity of La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample is evidently lower than that of graphene-free one above 373 K. This interesting phenomenon is also same with the Seebeck coefficient. The electrical thermal conductivity κ_e is calculated by the Wiedemann–Franz law (κ_e = L₀σT), where L₀ is the Lorentz numbers and often considered to be 2 × 10⁻⁸ V² · K⁻².^[31] The lattice thermal conductivity κ_L can be obtained by subtracting electronic thermal conductivity κ_e from κ. Figure 4(b) depicts the lattice thermal conductivity κ_L as a function of temperature for La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene and graphene-free samples. Compared to reference sample, the La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample has a lower lattice thermal conductivity. It can be considered that graphene introduced by PECVD method significantly reduces the lattice thermal conductivity due to the enhanced phonon scattering.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Temperature-dependent thermal properties of skutterudite and skutterudite/graphene samples: (a) total thermal conductivity, (b) lattice thermal conductivity.

Figure 5 shows the temperature-dependent of zT value for the graphene and graphene-free skutterudite samples. While the temperature is below 523 K, the zT values of La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample and the graphene-free one are comparable. However, the zT value of the La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample is significantly higher than that of reference sample above 523 K. The difference is consistent with the trend of Seebeck coefficient, power factor and lattice thermal conductivity. At 723 K, the maximum zT value of the La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂/graphene sample is about 1.0 and increases by 25 % compared with graphene-free one.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) Temperature-dependent zT values of skutterudite and skutterudite/graphene samples.

Graphene is successfully introduced into skutterudite by a PECVD method. The introduction of graphene has no obvious effect on the electrical conductivity of p-type La_0.8Ti_0.1Ga_0.1Fe₃CoSb₁₂ skutterudite, but remarkably enhanced Seebeck coefficient and decreased lattice thermal conductivity, leading to the high power factor and low total thermal conductivity especiallyat high temperature. Ultimately, the zT value is improved from 0.8 to 1.0 due to the addition of graphene.