Enhanced Thermoelectric Transport Properties of Electronegative-Element-Filled and (Ni, Te) Co-Doped Skutterudites through S Filling

Abstract

Recently, there has been a growing interest in skutterudite (SKD) compounds containing electronegative elements such as Br, Cl, S, Se, and Te, owing to their increased diversity and the versatility of filler atoms. This study focused on the thermoelectric performance of a series of (Ni, Te) co-doped SKDs filled with the electronegative element S, denoted as S_xNi_0.4Co_3.6Sb_11.2Te_0.8 (x = 0, 0.1, 0.2, and 0.3). These compounds were prepared using a combination of a solid-state reaction and spark plasma sintering techniques. The results showed that (Ni, Te) co-doping introduced excess electrons in the SKD lattice, while the incorporation of the element S into the SKD voids optimized carrier concentration. This led to a considerable increase in the absolute Seebeck coefficient to 110.6 μV K⁻¹ at ambient temperatures. The presence of S fillers induced phonon resonance scattering and point scattering, which reduced lattice thermal conductivity and ultimately improved the thermoelectric figure of merit zT, which reached 0.93 for S_0.3Ni_0.4Co_3.6Sb_11.2Te_0.8 at 823 K.

1. Introduction

Thermoelectric (TE) materials, which operate based on the Seebeck effect, are essential for converting heat energy into electrical energy, commonly utilized in waste heat recovery and cryogenic cooling applications [1,2,3,4]. The thermoelectric performance of these materials is generally quantified by the figure of merit zT = S²σT/κ, where S is the Seebeck coefficient, σ is the electrical conductivity, κ is the thermal conductivity, and T is the absolute temperature, respectively. However, achieving high zT values is challenging owing to the complex relationships and interactions among these parameters.

One of the most promising thermoelectric materials in the medium-temperature range are CoSb₃-based skutterudites (SKDs), known for their favorable thermal and mechanical stability, low toxicity, and excellent performance characteristics. These SKDs exhibit a body-centered cubic structure with a space group of $Im\overline{3}$) [5]. Their unit cells consist of 32 atoms, including 8 regular [CoSb₆] octahedral units (where Co occupies the 8c sites and Sb occupies the 24g sites) and 2 icosahedral voids (2a sites) formed by 12 Sb atoms [6,7,8,9]. However, the high thermal conductivity of binary cobaltite (~7.5 W m⁻¹ K⁻¹ at room temperature) resulting from the strong Co-Sb harmonic covalent framework leads to a low zT. To improve the thermoelectric transport properties, researchers often implement filling or doping strategies to reduce the thermal conductivity [10,11]. Filling the voids with suitable elements effectively reduces lattice thermal conductivity by scattering phonons through “rattling” in the voids. Additionally, the filler atoms modulate carrier concentration by acting as donors, adjusting the band structure [12]. Substitution at Co sites affects the regulation of the type and concentration of carriers [13,14,15,16]. Further, substituting Sb with single or multiple elements can reduce the lattice thermal conductivity by introducing defects that scatter phonons owing to differences in atomic radii and masses.

Recently proposed electronegative-element-filled SKDs (EN-SKDs) have exhibited various intriguing advantages, including thermoelectric performance comparable to traditional electropositive-element-filled SKDs (EP-SKDs) but with a lower filler content [9,13,17,18,19,20]. However, existing studies primarily focused on donor doping at the Co site (8c sites) [8,13,21], donor doping at the Sb site (24g sites) [17,22], and self-consistent charge compensation of electronegative elements [23,24,25]. Li et al. [25] analyzed the effect of ionized impurities on the energy band for S_0.07Co₄Sb_11.86S_0.14. We previously explored the effect of S fillers and (Te, Ge) co-substitution in SKDs, achieving a notable zT value of 1.45 for S_0.25Co₄Sb₁₁Ge_0.2Te_0.8 [9]. Nevertheless, investigations on EN-SKDs with donor doping at both the 8c and 24g sites are lacking, resulting in an incomplete understanding of the mechanisms. Therefore, it is crucial to explore the combined effects of filling and co-doping at these different sites on material thermoelectric performance.

It has been reported that substituting Te for Sb considerably increases electrical conductivity. Additionally, incorporating dislocations to enhance phonon scattering reduces thermal conductivity, leading to substantial improvement in the thermoelectric performance of CoSb₃ [26,27]. Further, the element Ni, as an inexpensive electron donor, has been shown to enhance material thermoelectric properties. Consequently, we implemented a strategy involving co-doping with Te and Ni in combination with S-filling to optimize material physicochemical properties and electronic transport behaviors and to improve the thermoelectric performance.

This study presents a series of electronegative-element-S-filled and (Ni, Te) co-doped SKDs, denoted as S_xNi_0.4Co_3.6Sb_11.2Te_0.8 (x = 0, 0.1, 0.2, and 0.3), which were prepared using a solid-state reaction combined with the spark plasma sintering (SPS) method. The incorporation of S into the SKD voids was facilitated by Ni and Te. Notably, the compound S_0.3Ni_0.4Co_3.6Sb_11.2Te_0.8 achieved the highest zT of 0.93 at 823 K.

2. Materials and Methods

2.1. Synthesis of S-Filled and (Ni, Te) Co-Doped SKDs

The S_xNi_0.4Co_3.6Sb_11.2Te_0.8 (x = 0, 0.1, 0.2, and 0.3) compounds were synthesized via a conventional solid-state rection route by following the methodology outlined in our previous study [9]. High-purity sulfur powder (S, 99.95%, Aladdin, Shanghai, China), nickel powder (Ni, 99.99%, Aladdin, Shanghai, China), cobalt powder (Co, 99.5%, Aladdin, Shanghai, China), antimony powder (Sb, 99.99%, Aladdin, Shanghai, China), and tellurium (Te, 99.99%, Aladdin, Shanghai, China) were used as the raw materials. These materials were weighed in stoichiometric proportions according to the formula S_xNi_0.4Co_3.6Sb_11.2Te_0.8 (x = 0, 0.1, 0.2, and 0.3). The powders were well-mixed under a N₂ atmosphere for 1 h, and then compacted into cylindrical shapes through cold-pressing. These cylinders were subsequently enclosed in vacuum quartz tubes. The vacuum-sealed quartz tubes were then placed in a melting furnace, heated to a temperature of 973 K, and held at this temperature for 3 days. Following this, the cylinders were ground into fine powders, and subjected to SPS at 853–873 K under a uniaxial pressure of 50 MPa for 10 min. Finally, the sintered products were cut into quadrangular prisms (measuring 3 × 3 × 12 mm³) and thin wafers with a size of 10 × 10 × 2 mm³ for further analysis.

2.2. Structural Characterization and Microscopy Topography

The phase compositions of the sintered ceramics were analyzed by the X-ray diffraction instrument (XRD) using a Rigaku Smartlab SE instrument (Japan) with a Cu Kα radiation source (λ = 1.5406 Å). The measurements were performed in the 2θ range of 10°–110° with a step size of 0.01° and the scanning speed of 2 °/min. The microstructures were examined via scanning electron microscopy (SEM; Hitachi Regulus 8230, Japan) and transmission electron microscopy (TEM; JEOL, JEM2100, Japan). Elemental distributions were analyzed via energy-dispersive X-ray spectroscopy (EDS; Bruker, Germany). Elemental contents were determined using Shimadzu inductively coupled plasma-atomic emission spectroscopy (Japan).

2.3. Thermoelectric Performance Measurement

The electrical resistivity (σ) values and Seebeck coefficients (S) of the samples were measured simultaneously using a thermoelectric performance measurement system (Advance Riko ZEM-3, Japan) in a He atmosphere. The Hall coefficient was determined using a Hall effect measurement system (Ecopia HMS-7000). Thermal conductivity was calculated using the following formula:

ĸ = λ ρ C_p

(1)

where λ is the thermal diffusivity, ρ is the true density, and C_p is the specific heat capacity [9]. Thermal diffusivity λ was measured using a laser flash thermal conductivity meter (NETZSCH LFA467HT, Germany). C_p was estimated using Dulong–Petit’s law and verified via a differential scanning calorimeter (DSC, NETZSCH 404 F3) with a measurement uncertainty of ±5%. The electronic thermal conductivity (κ_e) was determined by the Wiedemann–Franz law, κ_e = LσT, where the Lorenz number L is 2.0 × 10⁻⁸ V² K⁻². The density (ρ) of the samples was determined using the Archimedes method in deionized water, with a repeatability of 99%. Additionally, the thermoelectric properties were evaluated at 300–823 K with an uncertainty of ±3%.

3. Results

3.1. Crystal Structures and Microstructures

The XRD results of the S_xNi_0.4Co_3.6Sb_11.2Te_0.8 (x = 0, 0.1, 0.2, and 0.3) compounds are shown in Figure 1a. The characteristic peaks observed for the samples align well with the body-centered cubic structure of the CoSb₃ SKD, with a space group of $Im\overline{3}$ (JCPDS Card No. 85-0055). Additionally, secondary phases such as Sb₂Te₃ are present in all samples. Rietveld refinement of the XRD patterns allowed for the determination of the lattice parameter, as illustrated in Figure 1b,c. The lattice parameter increases from 9.05007(0) to 9.05365(0) Å with the addition of the S filler, indicating that the incorporation of excess filler atoms in the SKD voids leads to lattice expansion. Additionally, the relative content of the secondary phase, Sb₂Te₃, decreases with increasing S content. The reason could be attributed to the fact that the moderate amount of S reduces the system energy caused by excessive Tb doping and helps to form thermodynamically stable compounds. The crystal structure of S_xNi_0.4Co_3.6Sb_11.2Te_0.8 (x = 0, 0.1, 0.2, and 0.3) is shown in the inset of Figure 1d. The lattice parameters (a) gradually increase with the increase in S content. The lattice constant increases gently until x = 0.3, suggesting that there may be a filling limit of S in S_xNi_0.4Co_3.6Sb_11.2Te_0.8 system.

The SEM images of the cross-sectional surfaces of all samples in Figure 2a–d show a densely packed structure with numerous micropores (yellow arrow) and second-phase enrichment (rose red arrow) of varying sizes. The average grain size across all samples ranges from 4 to 10 μm, indicating that smaller grains lead to an increased number of grain boundaries. This enhances phonon scattering and helps reduce the lattice thermal conductivity. Figure 2e presents a TEM image of the S_0.3Ni_0.4Co_3.6Sb_11.2Te_0.8 compound, while Figure 2f,g depict its high-resolution TEM images. The lattice spacing measurement in Figure 2b is consistent with that of the CoSb₃ crystal structure, indicating that the CoSb₃-based compound is well crystallized and homogeneous. Figure 2g shows the lattice fringes corresponding to a secondary phase with a d-spacing of 2.15 Å, corresponding to that of the (110) plane of Sb₂Te₃. The elemental distribution analysis results of S_0.3Ni_0.4Co_3.6Sb_11.2Te_0.8 are shown in Figure 2h,i. Figure 2h shows a uniform distribution of the filling element S and doping elements Ni and Te. Figure 2i shows regions enriched with sulfides and an antimony telluride.

Table 1. Nominal composition, actual composition, carrier concentration, and Seebeck coefficient at room temperature.

Nominal Composition	Actual Composition	Carrier Concentration n (1020 cm−3)	Seebeck Coefficient S (μV K−1)
Ni0.4Co3.6Sb11.2Te0.8	Ni0.32(±0.04)Co3.6Sb11.29(±0.03)Te0.58(±0.04)	7.61	−80.65
S0.1Ni0.4Co3.6Sb11.2Te0.8	S0.10(±0.00)Ni0.37(±0.01)Co3.6Sb11.27(±0.04)Te0.52(±0.05)	6.07	−91.73
S0.2Ni0.4Co3.6Sb11.2Te0.8	S0.17(±0.01)Ni0.35(±0.04)Co3.6Sb11.08(±0.11)Te0.65(±0.07)	5.34	−104.32
S0.3Ni0.4Co3.6Sb11.2Te0.8	S0.19(±0.02)Ni0.36(±0.01)Co3.6Sb11.42(±0.05)Te0.50(±0.04)	4.85	−110.62

lists the nominal concentrations and actual concentrations, confirming that the filling limit of S is below 0.3. Generally, the as-prepared S_xNi_0.4Co_3.6Sb_11.2Te_0.8 compounds show the presence of crystal defects, such as micropores and secondary phases, which play a crucial role in phonon scattering, decreasing the thermal conductivity.

3.2. Electron Transport Properties

The temperature-dependent electrical transport properties of the S_xNi_0.4Co_3.6Sb_11.2Te_0.8 (x = 0, 0.1, 0.2, and 0.3) compounds are illustrated in Figure 3. As shown in Figure 3a, the electrical conductivities (σ) of all samples have the order of 10⁵. The increased σ of these compounds compared to that of single-donor, Ni-substituted EN-SKD is attributed to the additional electrons introduced into the crystal structure through Te and Ni doping [8,13,28,29]. At low temperatures, the σ values of all the samples decrease as the temperature rises, reflecting a behavior typical of heavily doped semiconductors, which is primarily attributed to lattice vibrations induced by thermal vibration and enhancement of lattice scattering. For samples with high sulfur filling, specifically those with x = 0.2 or 0.3, σ increases at temperatures exceeding 700 K, which can be attributed to an increase in the carrier concentration during the initial stages of intrinsic excitation. Additionally, a gradual decrease in conductivity σ is observed with an increasing S-filling ratio. For instance, the conductivity σ value decreases from 1.73 × 10⁵ S m⁻¹ for Ni_0.4Co_3.6Sb_11.2Te_0.8 to 1.14 × 10⁵ S m⁻¹ for S_0.3Ni_0.4Co_3.6Sb_11.2Te_0.8 at 300 K.

The Seebeck coefficients (S) of the compounds in the temperature range of 300–823 K is illustrated in Figure 3b. As expected, the negative S values indicate n-type semiconductor characteristics, with electrons as the majority carriers, consistent with findings of previous studies [18,21,30]. In the temperature range of 300–823 K, the absolute Seebeck coefficients |S| of all samples typically show a trend of initial increase, achievement of peak values, and subsequent decrements. This behavior is primarily attributed to bipolar diffusion and thermal excitation effects on a limited number of carriers at high temperatures. The |S| value increases with increasing S content, owing to the presence of electronegative S atoms in the SKD lattice void, which act as acceptors, and introduce two holes per S atom. This reduces the electron concentration, thereby increasing the absolute Seebeck coefficient. To further elucidate the transport behavior, a Pisarenko plot was obtained from a simple electron transport model using the following formula:

$$S={\displaystyle \frac{8{\pi }^2{k}_B^2}{3e{h}^2}}{m}^{\ast }T{\left({\displaystyle \frac{\pi }{3n}}\right)}^{2/3}$$

(2)

where k_B, h, m* and n are Boltzmann’s constant, Plank’s constant, effective mass, and carrier concentration, respectively. The m* values, shown in Figure 3c, exhibit a slight increase, indicating a minor change in the band structure owing to S filling. Figure 3d displays the temperature dependence of the power factor (PF) for all samples. Despite a reduction in electrical conductivity, the enhancement of |S| lead to an increase in the PF. In particular, the PF value increases from 11.6 mW cm⁻¹ K⁻² for Ni_0.4Co_3.6Sb_11.2Te_0.8 to 13.9 mW cm⁻¹ K⁻² for S_0.3Ni_0.4Co_3.6Sb_11.2Te_0.8 at room temperature. This improvement in PF can be attributed to the decrease in electron concentration.

3.3. Thermal Transport Properties

Figure 4 shows the thermal conductivity characteristics of the S_xNi_0.4Co_3.6Sb_11.2Te_0.8 (x = 0, 0.1, 0.2, and 0.3) compounds, including the total thermal conductivity (κ) and lattice thermal conductivity (κ_L) calculated by κ_L = κ − κ_e. The values of κ and κ_L decrease from 5.20 W m⁻¹ K⁻¹ and 4.13 W m⁻¹ K⁻¹ for x = 0 to 3.30 W m⁻¹ K⁻¹ and 1.78 W m⁻¹ K⁻¹ for x = 0.3 at room temperature, respectively. This reduction is mainly attributed to the introduction of S-fillers, which results in reduced electron thermal conductivity due to diminished electrical conductivity. The decrease in room-temperature κ_L, as shown in Figure 4b, is primarily attributed to point-defect phonon scattering. At 823 K, there is a slight increase in the κ value, which results from the enhanced bipolar diffusion effect resulting from the intrinsic excitation at high temperatures.

3.4. Thermoelectric Performance

Figure 5 illustrates the temperature-dependent zT value of the S-filled and (Ni, Te) co-doped SKDs with different S contents. The zT values continuously increase with increasing S content, owing to the combined effects of optimized electron concentration and reduced thermal conductivity. Remarkably, the maximum zT value of 0.93 was achieved at 823 K for the samples with x = 0.3 and was 1.62 times higher than that of the S-free sample. The results also indicate that even small quantities of S fillers contribute to improving the thermoelectric properties of (Ni, Te) co-doped SKDs.

4. Conclusions

Herein, a series of n-type electronegative element filled-skutterudite compounds S_xNi_0.4Co_3.6Sb_11.2Te_0.8 (x = 0, 0.1, 0.2, and 0.3) were synthesized via a solid-state reaction method combined with SPS technology. The resulting compounds exhibited a small grain size along with micropores and a precipitated phase. The (Ni, Te) co-doping introduced additional electrons and improved the electron concentration, reduced the thermal conductivity, and optimized the carrier concentration through charge compensation. S_0.3Ni_0.4Co_3.6Sb_11.2Te_0.8 achieved the highest figure of merit (zT) of 0.93, showing approximately a 62% enhancement compared to the S-free sample. These findings provide valuable insights into the microstructural changes and the behavior of S filling into 2a sites and donor doping at both the 8c and 24g sites in SKD materials.