Thermoelectric Transport Properties of Double-Filling In_xLa_0.25Co₄Sb₁₂ Skutterudite Materials

Abstract

Filled-skutterudite materials are promising candidates for thermoelectric applications in the intermediate-temperature range to recover waste heat. In this work, mechanical alloying and spark plasma sintering (SPS) techniques were used to consolidate double-filling In_xLa_0.25Co₄Sb₁₂ (x = 0, 0.1, 0.3, 0.5) samples. X-ray diffraction, scanning electron microscopy, and energy-dispersive x-ray spectroscopy were used to analyze the microstructure of the In_xLa_0.25Co₄Sb₁₂ SPS-ed samples. The microstructure results revealed the main phase of the CoSb₃ skutterudite structure with a small amount of InSb and CoSb₂ attributed to secondary phases. Electrical resistivity remarkably decreased to 9.7 μΩ m at room temperature for the In_0.5La_0.25Co₄Sb₁₂ sample, mainly due to the effective impurity of the InSb nanoinclusions. Moreover, the InSb nanoinclusions yielded a substantial depression in lattice thermal conductivity due to phonon scattering. The reduction in lattice thermal conductivity was approximately 62% for the In_0.5La_0.25Co₄Sb₁₂ sample compared with the La_0.25Co₄Sb₁₂ sample, resulting in a remarkable enhancement in the dimensionless figure-of-merit (ZT) of 1.25.

Introduction

In recent decades, thermoelectric (TE) applications have become a promising technology for energy harvesting.¹ Waste heat can be converted into useful energy via TE generation.² A skutterudite TE material is a good candidate for TE materials in the temperature range of 300–800 K. Furthermore, filled-skutterudite TE materials have a low lattice thermal conductivity, excellent mechanical strength, and good thermal, oxidation, and chemical stability.³ However, the binary skutterudite CoSb₃ system may not be suitable for TE applications, owing to its low carrier concentration, high electrical resistivity, and high thermal conductivity.⁴ Generally, carrier effective mass, carrier concentration, and carrier mobility play an important role in TE characteristics. Three different approaches are used to improve skutterudite TE material: filling the voids,^5,6,7 doping or substitution at antimony or cobalt sites,^8,9 and nanostructuring.^10,11,12

A heavy filler atom is a promising strategy to enhance the performance of skutterudite systems, owing to its high carrier concentration and low lattice thermal conductivity. In general, the atomic radius of the filler should be smaller than the void radius of skutterudite Co₄Sb₁₂ (1.892 Å)^13,14 to be easily filled into the void space. Several experiments have been conducted on skutterudite materials by adding fillers, such as Ba, Bi, Ce, Hf, In, La, Li, Sr, Tl, and Yb,^{15,16,17,18,19,20,21,22,23,24,25} and the TE properties improved by enhancing the carrier concentration and mobility, as well as minimizing the lattice thermal conductivity. For example, Li et al.²⁶ investigated the effect of In-filling into skutterudite material. They found that In addition increased the carrier concentration due to the supply of electrons. The thermoelectric figure-of-merit (ZT) increased to the highest value of 0.67 for In_0.35Co₄Sb₁₂ at 600 K. Sesselmann et al.²⁷ examined the effect of changing the heat treatment parameters on In-filled skutterudite. The results showed a major reduction in lattice thermal conductivity due to minor changes in the synthesis parameters. Consequently, the maximum value ZT of 0.7 for the In_0.2Co₄Sb₁₂ sample at 675 K was achieved. Deng et al.²⁸ studied In_xCo₄Sb₁₂ -filled skutterudite compounds. The results showed the highest power factor of 31.3 μW/cmK² at 616 K and the lowest thermal conductivity of 2.193 W/mK at 568 K. The enhanced dimensionless ZT was 0.88 for the In_0.5Co₄Sb₁₂ sample at 665 K. Mallik et al. studied the influence of In addition into a Co₄Sb₁₂ compound. The results showed a considerable reduction in thermal conductivity probably due to the rattling of In-filling and the presence of InSb as the secondary phase at the grain boundaries. The thermoelectric ZT achieved the highest values of 1.12 and 0.96 for the In_0.25Co₄Sb₁₂ and In_0.4Co₄Sb₁₂ compounds, respectively.^29,30 The same compound of In_0.25Co₄Sb₁₂ filling skutterudite was obtained by He et al.³¹ who achieved a high ZT of 1.2 at 575 K. Lee et al.³² studied multi-filled skutterudite In_xYb_yLa_0.3-x-yCo₄Sb₁₂ compounds. The results showed that the double-filling skutterudite was most effective in reducing thermal conductivity. A maximum ZT of 0.9 was obtained for the In_0.1Yb_0.2Co₄Sb₁₂ sample at 782 K. Khovaylo et al.³³ investigated the effect of In-filled skutterudite. The results showed a substantial contribution of phonon scattering due to the InSb impurity phase that precipitated in nanometer-sized regions along the grain boundaries, and, consequently, a considerable reduction in a thermal conductivity. Thus, a maximum ZT ≈ 1.5 was recorded for the In₁Co₄Sb₁₂ sample at 725 K.

The current work aims to investigate the double-filling of In_xLa_0.25Co₄Sb₁₂ materials to boost the carrier concentration and improve the TE transport properties. A double-filled skutterudite system was synthesized with La as the dominating filling element and In as the auxiliary filler. The In atom can successfully fill the La_0.25Co₄Sb₁₂ skutterudite void because the atomic radius of In (1.55 Å) is smaller than the void radius of skutterudite Co₄Sb₁₂ (1.892 Å).¹³ The results show a remarkable reduction in lattice thermal conductivity for the In_0.5La_0.25Co₄Sb₁₂ sample compared with the La_0.25Co₄Sb₁₂ sample.

Experimental

Sample Preparation and Measurements

High-purity powder of Co (99.8%), Sb (99.5%), La (99.9%), and In (99.99%) were purchased from Alfa Aesar, and all the preparation processes are discussed in the literature.³⁴ The samples were cut into various parts for characterization. X-ray diffraction (XRD), scanning electron microscopy (SEM), and energy-dispersive x-ray spectroscopy (EDX) measurement data were collected for the SPS-ed samples. The setup of the characterization devices is described in the literature.^34,35

The temperature dependency of electrical resistivity and Seebeck coefficient was measured using a commercial instrument (ZEM-3; Ulvac-Riko), and thermal diffusivity was measured using a laser flash apparatus (TC-7000H; Ulvac-Riko). The thermal conductivity was calculated from the thermal diffusivity, D, and specific heat, C_p, using k = D C_p d, where d is the sample density. All the TE properties were measured at a temperature range between 300 and 800 K, and then the ZT value was calculated.

Results and Discussion

Microstructure Analysis of In_xLa_0.25Co₄Sb₁₂ (0 ≤ x ≤ 0.5) Skutterudites

Figure 1 shows the XRD patterns of the In_xLa_0.25Co₄Sb₁₂ SPS-ed samples. The analysis of XRD patterns revealed the major phase of CoSb₃ skutterudite for all SPS-ed samples. A small peak of impurity corresponded to CoSb₂ secondary phase. The formation of CoSb₂ phase was mainly due to the deficiency of the Sb atom and the short SPS sintering time. Moreover, the structural studies found a peak corresponding to the InSb nanoinclusion phase, as shown in Fig. 1. This nanoinclusion impurity began to occur when x ≥ 0.3, due to the excess In in the skutterudite structure. This result agrees with previous experimental research, which revealed that the maximum limit of In-filling into the CoSb₃ cages is 1.37 at.% (x = 0.22).^30,36 Furthermore, the Rietveld refinement for the In_xLa_0.25Co₄Sb₁₂ (0.1 ≤ x ≤ 0.5) samples demonstrated the skutterudite lattice structure obtained by using Jana software, as shown in Fig. 2. The results show the lattice parameters increased from 9.0411 Å for the La_0.25Co₄Sb₁₂ sample to 9.0580 Å for the In_0.3La_0.25Co₄Sb₁₂ sample. Thus, the In atom successfully filled the skutterudite void because the atomic radius of In (1.55 Å) was smaller than the void radius of skutterudite Co₄Sb₁₂ (1.892 Å).¹³ The results of the lattice parameters and the nominal and actual compositions of the SPS-ed samples are listed in Table I.

Fig. 1

XRD patterns of In_xLa_0.25Co₄Sb₁₂ (0 ≤ x ≤ 0.5) samples.

Fig. 2

Jana refinement analysis of (a) In_0.1La_0.25Co₄Sb₁₂ , (b) In_0.3La_0.25Co₄Sb₁₂ , and (c) In_0.5La_0.25Co₄Sb₁₂ SPS-ed samples.

Table I Lattice parameters, nominal and actual compositions of SPS-ed samples

The SEM morphologies of the In_xLa_0.25Co₄Sb₁₂ (0.1 ≤ x ≤ 0.5) SPS-ed samples are shown in Figs. 3a, 4, and 5(a). The samples exhibited a relatively high density of up to 99% according to the microstructures. However, pores were produced during SPS sintering, as evidenced by the surface morphology of the In_0.1La_0.25Co₄Sb₁₂ sample, which had the lowest density of about 95%. The SEM–EDX images showed skutterudite CoSb₃ as the primary phase for all SPS-ed samples, as well as CoSb₂ corresponding to the secondary phase, which match the XRD results. Moreover, the SEM–EDX results reveal that the InSb nanoinclusions were randomly distributed on the skutterudite matrix and agglomerated at the grain boundaries for the In_0.5La_0.25Co₄Sb₁₂ sample, as shown in Fig. 5a and d. Furthermore, the formation of the InSb nanoinclusions had a strong influence on reducing the lattice thermal conductivity due to the scattering of the charge carriers at the interfaces. Additionally, the La atoms were inhomogeneously distributed and appeared in all the samples, as shown by the elemental mapping in Figs. 3d, 4, and 5d, which may effectively scatter the phonons to further decrease the lattice thermal conductivity.^37,38,39,40

Fig. 3

(a) SEM morphology, and (b–e) corresponding elemental mappings of the In_0.1La_0.25Co₄Sb₁₂ SPS-ed sample.

Fig. 4

(a) SEM morphology, and (b–e) corresponding elemental mappings of the In_0.3La_0.25Co₄Sb₁₂ SPS-ed sample.

Fig. 5

(a) SEM morphology, and (b–e) corresponding elemental mappings of the In_0.5La_0.25Co₄Sb₁₂ SPS-ed sample.

Thermoelectric Properties

The temperature dependency of electrical resistivity for the In_xLa_0.25Co₄Sb₁₂ SPS-ed samples is shown in Fig. 6. At room temperature, the electrical resistivity rapidly decreased with increasing the In–filler fraction content, which clearly indicated that the In-filler acted as n-type to increase the electron concentration and that more electrons were donated to the conduction band. Furthermore, the electrical resistivity of the In_0.1La_0.25Co₄Sb₁₂ sample decreased with the increasing temperature, indicating non-degenerate semiconducting transport behavior, whereas, for the higher In content of the In_0.3La_0.25Co₄Sb₁₂ and In_0.5La_0.25Co₄Sb₁₂ samples, the electrical resistivity increased with temperature, displaying semimetal-like behavior. These results are consistent with published reports for skutterudites.^41,42 The InSb phase was highly conductive,^30,43 and, on reaching the solubility limit of In (x = 0.22), a substantial reduction in the electrical resistivity was achieved mainly due to the increasing volume fraction of InSb. Thus, the lowest electrical resistivity value of 9.67 μΩ m was obtained for the In_0.5La_0.25Co₄Sb₁₂ sample at room temperature, which was lower than that reported by He et al.³¹ for In_0.25Co₄Sb₁₂, approximately 12 μΩ m at the same temperature due to the existence of the InSb phase with relatively high conductivity, as described above. Moreover, this reduction in electrical resistivity was attributed to the contribution of the In-filler atoms that acted as electron donor on the CoSb₃ skutterudite system that increased the carrier concentration. Additionally, the addition of In above the filling fraction limit (x > 0.22) saturated the material and allowed the formation of the InSb nanoinclusion phase at the grain boundaries of the Co₄Sb₁₂ skutterudite matrix, as the results showed a considerable reduction in the electrical resistivity.

Fig. 6

Electrical resistivity for In_xLa_0.25Co₄Sb₁₂ (0 ≤ x ≤ 0.5) SPS-ed samples.

Figure 7 shows the variation of the Seebeck coefficient versus temperature difference for the In_xLa_0.25Co₄Sb₁₂ SPS-ed samples. Over the entire studied temperature range, all the double-filled samples showed a negative Seebeck coefficient, indicating that the n-type semiconductor behavior and the majority carriers were electrons. However, the Seebeck coefficient results of the single-filler La_0.25Co₄Sb₁₂ sample revealed a positive sign (p-type) semiconducting material, and these tendencies agreed well with the literature data⁴⁴ with the La fraction content form (0.1–0.5). Moreover, the In atoms acted as electron donors that led to decreasing the hole concentration of the La_0.25Co₄Sb₁₂ p-type skutterudite. The electrical conductivity of the In-filled skutterudites was inversely proportional to the Seebeck coefficients. Furthermore, the absolute Seebeck coefficient decreased with the In–filling fraction at room temperature, whereas the electrical conductivity increased owing to the higher electron carrier concentrations. The Seebeck coefficient almost monotonically increased with the temperature for x = 0.3 and x = 0.5. However, a nonmonotonic behavior was revealed for the x = 0.1 sample, hence achieving a maximum value of the Seebeck coefficient over recorded samples of 252 μVK⁻¹ at 495 K. This value is higher than the 205 μVK⁻¹ at 673 K obtained for the In_0.4Co₄Sb₁₂ compound.³⁰ Visnow⁴⁵ found that the Seebeck coefficient for the In_0.113Co₄Sb₁₂ compound was ~ 255 μVK⁻¹ at 623 K, which is almost similar to the current values. However, the InSb nanoinclusion phase had a negative influence on the Seebeck coefficient, mainly due to its high conductivity, as mentioned above, and this agreed with previous published articles.^30,43

Fig. 7

Seebeck coefficient for In_xLa_0.25Co₄Sb₁₂ (0 ≤ x ≤ 0.5) SPS-ed samples.

The temperature dependency of the power factor for the In_xLa_0.25Co₄Sb₁₂ SPS-ed samples is presented in Fig. 8. The power factor can be calculated by using the equation $PF={S}^{2}\sigma$, where S is the Seebeck coefficient, and σ is the electrical conductivity. The power factor was considerably improved with In-addition fraction content, and also the temperature, with the highest peak before declining for all samples. The highest power factor value of 3.85 10^–3 W/mK² was obtained for the In_0.5La_0.25Co₄Sb₁₂ sample at 543 K, mainly due to the considerable reduction in electrical resistivity. This result is higher than that reported in the literature of 3.25 10⁻³ W/mK² for the In_0.3La_0.5Co₄Sb₁₂ compound at the same temperature,⁴⁶ 3.5 10⁻³ W/mK² for the In_0.25Co_3.9La_0.1Sb₁₂ compound at 600 K,⁴⁷ ~ 3.2 10⁻³ W/mK² for In_0.1Yb_0.1La_0.1Co₄Sb₁₂ compound,⁴⁸ and 2.33 10⁻³ W/mK² for In_0.05Pr_0.05Co₄Sb₁₂ at 609 K.⁴²

Fig. 8

Power factor for In_xLa_0.25Co₄Sb₁₂ (0 ≤ x ≤ 0.5) SPS-ed samples.

Figure 9(a) shows the temperature dependency of the In_xLa_0.25Co₄Sb₁₂ samples as a function of thermal conductivity (k). The thermal conductivity was considerably reduced and showed nonmonotonic behavior with increasing In–filling fraction content. The lowest value of thermal conductivity of 1.27 W/mk was obtained for the In_0.1La_0.25Co₄Sb₁₂ sample at 495 K, which is substantially lower than the reported values of 1.6 W/mk for the In_0.05Co₄Sb₁₂ compound at 580 K.³¹ This result is mainly attributed to the highest porosity of the In_0.1La_0.25Co₄Sb₁₂ sample, which is consistent with the literature.^49,50,51 For the higher In–addition fraction content (x = 0.3 and x = 0.5) samples, the thermal conductivity was high mainly due to the existence of InSb impurities with a high thermal conductivity ~ 40 W/mK.^30,43 The temperature dependency of lattice thermal conductivity (k_l) for the In_xLa_0.25Co₄Sb₁₂ samples is shown in Fig. 9b. The equation k_l = k_total – k_e is used to compute the lattice thermal conductivity, where k_e is the electronic contribution, and k_total is the total thermal conductivity. Moreover, the Wiedemann–Franz law (k_e = L.σ.T, where L = 2.44 10⁻⁸ V²/K²), can be used to determine the electronic contribution. The Lorenz number as the function of temperature dependence can be calculated via a single parabolic band model equation ⁵²:

$$ L\left( T \right) = \left( {\frac{{K_{B} }}{e}} \right)^{2} \left[ {\frac{{\left( {\lambda + {\raise0.7ex\hbox{$7$} \!\mathord{\left/ {\vphantom {7 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}}} \right) F_{{\lambda + {\raise0.7ex\hbox{$5$} \!\mathord{\left/ {\vphantom {5 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}} }} \left( \xi \right)}}{{\left( {\lambda + {\raise0.7ex\hbox{$3$} \!\mathord{\left/ {\vphantom {3 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}}} \right) F_{{\lambda + {\raise0.7ex\hbox{$1$} \!\mathord{\left/ {\vphantom {1 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}} }} \left( \xi \right)}} - \left( {\frac{{\left( {\lambda + {\raise0.7ex\hbox{$5$} \!\mathord{\left/ {\vphantom {5 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}}} \right) F_{{\lambda + {\raise0.7ex\hbox{$3$} \!\mathord{\left/ {\vphantom {3 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}} }} \left( \xi \right)}}{{\left( {\lambda + {\raise0.7ex\hbox{$3$} \!\mathord{\left/ {\vphantom {3 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}}} \right) F_{{\lambda + {\raise0.7ex\hbox{$1$} \!\mathord{\left/ {\vphantom {1 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}} }} \left( \xi \right)}}} \right)^{2} } \right], $$

(1)

where K_B is the Boltzmann constant, e is the electron charge, $\lambda$ is the scattering parameter, $\upxi$=${\mathrm{E}}_{\mathrm{F}}/{\mathrm{K}}_{\mathrm{B}}\mathrm{T}$ is the reduced Fermi energy, E_F is the Fermi energy, and F_n($\upxi$) is the Fermi integral defined by:

$$ F_{n} \left( \xi \right) = \mathop \smallint \limits_{0}^{\infty } \frac{{x^{n} }}{{1 + e^{x - \xi } }}dx $$

(2)

Fig. 9

(a) Thermal conductivity and (b) lattice thermal conductivity for In_xLa_0.25Co₄Sb₁₂ (0 ≤ x ≤ 0.5) SPS-ed samples.

In this calculation, the scattering parameter is set to be − 0.5 by assuming the acoustic phonon scattering is the dominant carrier scattering mechanism at the measurement temperature range. The reduced Fermi energy is extracted from the measured Seebeck coefficient values using⁵²:

$$ S = \pm \frac{{K_{B} }}{e} \left[ {\frac{{\left( {\lambda + {\raise0.7ex\hbox{$5$} \!\mathord{\left/ {\vphantom {5 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}}} \right) F_{{\lambda + {\raise0.7ex\hbox{$3$} \!\mathord{\left/ {\vphantom {3 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}} }} \left( \xi \right)}}{{\left( {\lambda + {\raise0.7ex\hbox{$3$} \!\mathord{\left/ {\vphantom {3 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}}} \right) F_{{\lambda + {\raise0.7ex\hbox{$1$} \!\mathord{\left/ {\vphantom {1 2}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{$2$}} }} \left( \xi \right)}} - \xi } \right] $$

(3)

By using the extracted reduced Fermi energy, the Lorenz numbers were calculated to be 2.08, 1.61, 1.62, and 1.67 × 10⁻⁸ V²/K² for the proportions of x = 0, 0.1, 0.3, and 0.5, respectively, at room temperature. In general, the calculated Lorenz numbers are smaller than the degenerate limit of the Lorenz number (2.44 × 10⁻⁸ V²/K²) typically observed in TE materials with carrier concentrations between generate and non-degenerate semiconductors. The lattice thermal conductivity parameters were considerably reduced with the In–addition fraction content. Furthermore, the InSb nanoinclusion phase induced more interfaces into the system, which led to a substantial decrease in the lattice thermal conductivity through the interface scattering of phonons. The results show ~ 62% reduction of the lattice thermal conductivity compared with the La_0.25Co₄Sb₁₂ sample. The lowest value of lattice thermal conductivity, 0.83 W/m K, was achieved for the In_0.5La_0.25Co₄Sb₁₂ sample at 639 K, which is three times lower than that of samples selected from the published reports.^53,54 The lowest value of lattice thermal conductivity was obtained for the highest In–filling fraction content x = 0.5 sample, which agrees with previous published reports.^26,27,33 In addition, the lattice thermal conductivity was minimized due to the strain field scattering of high-energy phonons.^37,38,39,40 This result agrees with the previous study by Nolas et al.,⁵⁵ who found that an uneven distribution of the filler can introduce a point-defect-type phonon scattering and scatter a large spectrum of phonons.

The temperature dependency of the In_xLa_0.25Co₄Sb₁₂ SPS-ed samples as a function of the thermoelectric ZT is shown in Fig. 10. The results show a clear trend and remarkable improvement of ZT with In–filler fraction content and temperature. The formation of the InSb nanoinclusion phase at the grain boundaries of the skutterudite matrix has a substantial effect on reducing the lattice thermal conductivity, while maintaining good electrical transport properties. Consequently, the maximum ZT = 1.25 obtained for In_0.5La_0.25Co₄Sb₁₂ sample was considerably improved compared with the In-free La_0.25Co₄Sb₁₂ sample. This result is higher than the 0.79 obtained for the In_0.1Co₄Sb₁₂ compound,⁵⁶ 0.9 attained in the In_0.1Yb_0.2Co₄Sb₁₂ compound at 782 K³² and ~ 1.2 obtained for In_0.25Co₄Sb₁₂ compound.³¹ The In-filler boosts the TE properties of La_0.25Co₄Sb₁₂ skutterudite material at a medium-temperature range.

Fig. 10

The dimensionless figure of merit for In_xLa_0.25Co₄Sb₁₂ (0 ≤ x ≤ 0.5) SPSed samples.

Conclusions

MA and SPS techniques were used to successfully fabricate In_xLa_0.25Co₄Sb₁₂ skutterudite compounds. The major phase of the CoSb₃ skutterudite structure was observed by the XRD patterns for SPS-ed samples with minor amounts of CoSb₂ and InSb as the secondary phases. The InSb nanoinclusion impurity remarkably reduced the electrical resistivity and lattice thermal conductivity of the In_xLa_0.25Co₄Sb₁₂ skutterudites. The In_0.5La_0.25Co₄Sb₁₂ sample achieved the lowest electrical resistivity and lattice thermal conductivity of 9.67 μΩ m and 0.83 W/m K, respectively. Furthermore, the In-filler dramatically boosted the power factor, reaching a maximum value of 3.85 10⁻³ W/m K² at 543 K for the In_0.5La_0.25Co₄Sb₁₂ sample due to a considerable reduction in electrical resistivity. The ZT was greatly improved to 1.25 at 688 K for the In_0.5La_0.25Co₄Sb₁₂ sample, which was substantially higher than that of the unfilled sample.