Thermoelectric Performance Optimization of n-Type La_3−xSm_xTe₄/Ni Composites via Sm Doping

Abstract

La₃Te₄-based rare-earth telluride is a kind of n-type high-temperature thermoelectric (TE) material with an operational temperature of up to 1273 K, which is a promising candidate for thermoelectric generators. In this work, the Sm substitution in La_3−xSm_xTe₄/Ni composites is reported. The electrical transport property of La_3−xSm_xTe₄ is modified by reducing carrier concentration due to the substitution of Sm²⁺ for La³⁺. The electric thermal conductivity decreases by 90% due to carrier concentration reduction, which mainly contributes to a reduction in total thermal conductivity. Lattice thermal conductivity also decreases by point-defect scattering by Sm doping. Meanwhile, based on our previous study, compositing nickel improves the thermal stability of the La_{3 − x}Sm_xTe₄ matrix. Finally, combined with carrier concentration optimization and the decreased thermal conductivity, a maximum zT of 1.1 at 1273 K and an average zT_ave value of 0.8 over 600 K–1273 K were achieved in La_2.315Sm_0.685Te₄/10 vol.% Ni composite, which is among the highest TE performance reported in La₃Te₄ compounds.

1. Introduction

Thermoelectric (TE) materials can convert heat into electricity directly, which have been applied in radioisotope thermoelectric generators (RTGs) used for deep space exploration for over half-century or in generators for waste heat recovery in industry and the human body [1,2,3,4,5]. In thermoelectric generators (TEGs), the thermal-to-electricity conversion efficiency of TE devices η_TE (η_TE = $\frac{T_{\mathrm{H}}{- T}_{\mathrm{C}}}{T_{\mathrm{H}}}\frac{\sqrt{1\text{+}z\overline{T}}}{\sqrt{1\text{+}z\overline{T}}+\frac{T_{\mathrm{C}}}{T_{\mathrm{H}}}}$) is mainly determined by Carnot efficiency η_Carnot (η_Carnot = (T_H – T_C)/T_H) and the dimensionless figure of merit (zT) of TE materials [6,7]. zT = S²σT/κ, where S, σ, T, and κ are the Seebeck coefficient, electrical conductivity, absolute temperature, and thermal conductivity, respectively [8,9,10,11]. High service temperature and the large temperature differences will benefit energy conversion efficiency. Thus, research in high-temperature TE materials has advanced significantly in the last decades, and several promising materials have been discovered [12,13]. Thereinto, lanthanum telluride (La_3−xTe₄) promotes energy conversion efficiency up to 15% in the segment couple-level prototype, roughly doubling the heritage technologies of traditional SiGe alloys [14,15].

La_{3 − x}Te₄ possesses a Th₃P₄-type structure (space group I-43d). In the last decades, the thermoelectric properties and mechanical properties of La_3−xTe₄ have been studied due to their potential applications in RTGs [16,17,18]. When changing lanthanum vacancies, the electronic behavior of La_{3 − x}Te₄ varies from metallic at x = 0 (La₃Te₄) to the semi-insulating at x = 1/3 (La₂Te₃). When 0 ≤ x < 1/3, La_3−xTe₄ exhibits n-type conducting behavior, and its carrier concentration can be widely tuned by controlling the La vacancies [19]. Additionally, La_3−xTe₄ exhibits intrinsically low lattice thermal conductivity due to its complex crystal structure, which leads to the strong electron–phonon scattering by high carrier concentrations and the strong point-defect scattering relative to phonons by vacancies. Therefore, the introduction of La vacancy is an effective method to optimize the carrier concentration and further improve the TE performance of La_{3 − x}Te₄. For example, the high zT of 1.1 at 1273 K of La_3−xTe₄ with the optimum carrier concentration n~9 × 10²⁰ cm⁻³ is obtained via stoichiometry control [15]. The nanocomposite strategy also benefits the thermoelectric performance enhancement for La_3−xTe₄ and other materials in recent years [20,21].

In addition to La vacancy regulation, it has been demonstrated that non-isoelectronic substitution is also a powerful method to tune the carrier concentration via creating vacancy-free structures, allowing for a relatively independent impact on the lattice thermal conductivity of various defects (vacancies and substitutional atoms) in the La sub-lattice and electron–phonon interactions [22]. The substitutions in the La_3−xTe₄ system either on La³⁺ sites (by Yb²⁺, Ca²⁺, etc.) or on Te²⁻ sites (by Sb³⁻, Bi³⁻, etc.) have been explored to optimize the carrier concentration in recent years [22,23,24]. In the case of cation substitutions on La³⁺ sites, the substitution of a divalent cation M²⁺ for La³⁺ (La_3−x−yM_yTe₄) results in an electronic local environment of the following:

$${\mathrm{La}}_{3-x-y}^{3+}{V}_{\mathrm{La},x}{M}_y^{2+}{\mathrm{Te}}_4^{2-}{e}_{1-3x}^{-1}$$

with a theoretical carrier concentration of n = n_max (1 − 3x − y), which can obtain a threefold improvement in the control of carrier concentrations. For instance, non-rare earth dopant Ca²⁺ modified the density of states to improve the power factor and achieved a finer control over the carrier concentration. A zT_max ~1.2 at 1273 K was obtained for La_2.2Ca_0.78Te₄ with an optimum carrier concentration of 1.1 × 10²¹ cm⁻³ [22]. A peak zT of ~1.2 at 1273 K was obtained at the carrier concentration of n ~0.3 × 10²¹ cm⁻³ by substituting Yb²⁺ for La³⁺ [23].

For this study, samarium was chosen as the dopant due to Sm²⁺ (122 pm) being similar in ionic size with La³⁺ (103.2 pm), which improves the probability of successful substitution upon the La site, which is expected to reduce the high carrier concentration in La₃Te₄. In addition, there is a possibility that Sm may exist in the form of mixed valance state Sm²⁺/Sm³⁺. Nakahara et al. revealed the mixed Sm²⁺/Sm³⁺ substitution on the La³⁺ site in the La_3−ySm_yS₄ system [25], and the valence fluctuation of Sm also exists in Sm₃Te₄ system [26]. Thus, this observation encourages us to understand the effect of Sm element doping on the electrical and thermal transport properties in lanthanum telluride.

Here, we propose a strategy to enhance the TE performance of La_{3 − x}Sm_xTe₄ by doping with samarium (Sm). When the Sm content is above 0.65, the coexistence of Sm²⁺ and Sm³⁺ ions in La_3−xSm_xTe₄ is experimentally confirmed by an X-ray absorption spectrum (XAS). The substitution of La³⁺ by Sm²⁺ results in a decrease in carrier concentration and modifies the electrical properties. Moreover, the decrease in carrier concentration greatly reduces the electric thermal conductivity (κ_e), and Sm substitution on La³⁺ can also introduce point defects to strengthen phonon scattering and further reduce lattice thermal conductivity (κ_L). The composited metallic Ni inclusions are used to improve thermal stability and oxidation resistance, as proved in our previous work [27]. Consequently, an enhanced zT value of 1.1 at 1273 K and an average zT_ave value of 0.8 over 600 K–1273 K were obtained in La_2.315Sm_0.685Te₄/Ni composites.

2. Materials and Methods

A series of La_3−xSm_xTe₄/10 vol.% Ni composite (x = 0, 0.4, 0.6, 0.65, 0.685, 0.69, 0.7) samples was synthesized by the melting–ball milling–hot pressing method. The weighted chunks of La (Alfa Aesar, 99.9%), Sm (Alfa Aesar, 99.9%), and Te (Alfa Aesar, 99.999%) elements were loaded into a carbon crucible and sealed in evacuated quartz tubes. The sealed quartz tube was then heated at 1373 K for 15 h, followed by furnace cooling to room temperature. The melted ingot was ball milled in a stainless-steel vial (MSK-SFM-3) for 3 h at 1400 rpm to obtain the La_3−xSm_xTe₄ powders. Then, the 10 vol.% metallic Ni (Alfa Aesar, 99.9%, 1–5 μm) powders were mixed in with the La_3−xSm_xTe₄ powders by using the ball milling process once more. The La_3−xSm_xTe₄/Ni bulks were sintered by hot pressing at 1400 K under 80 MPa for 90 min. To avoid oxidation, the processes of raw material weighing, quartz tube sealing, powder loading, and ball milling were performed in a glove box under an argon atmosphere.

The phase structure was determined by Powder X-ray Diffraction (PXRD, Bruker, Cu Kα: λ = 1.5406 Å). The morphology and microstructure were observed using scanning electron microscopy (SEM, ZEISS Supra 55). Element distributions were characterized using energy dispersive spectroscopy (EDS). The electrical conductivity (σ) and Seebeck coefficient (S) were measured by an electrical measurement system (SBA458, Netzsch: Bayern, Germany) under a helium atmosphere. Thermal diffusivity (D) was measured by laser flash system (LFA457, Netzsch: Bayern, Germany), and the specific heat capacity (C_p) was taken from previously published specific heat capacity results that were modified to add samarium content following the Dulong–Petit law [27]. Then, thermal conductivity (κ) was calculated by κ = ρ × C_p × D, where ρ is the measured density. The Hall coefficient (R_H) was measured under a magnetic field of −3 T to 3 T with a five-probe configuration using a Physical Property Measurement System (PPMS, Quantum Design). All samples should be preserved in a glove box under an inert atmosphere before and after testing.

3. Results and Discussion

Figure 1a shows the PXRD patterns of the as-synthesized La_3−xSm_xTe₄ (x = 0, 0.4, 0.6, 0.65, 0.685, 0.69, 0.7)/10 vol.% Ni composites. The main phase can be indexed relative to the Th₃P₄-type structure (I-43d, ICSD-642044) of La₃Te₄. The major peaks shift gradually to lower angles with the increase in Sm content from 0 to 0.65, indicating the expansion of the lattice. There is no obvious shift for the peaks with Sm content between 0.65 and 0.7. Some diffraction peaks belonging to the cubic structure of Ni (Fm-3m, ICSD-41508) are also found. Very weak diffraction peaks near 30° belonging to the impurity phases of La₂O₂Te are commonly observed due to the presence of trace oxidation [23]. The calculated average crystalline size of La_3−xSm_xTe₄/Ni composites is 93.2 nm by using the Debye–Scherrer equation [28]. Taking the as-synthesized La_2.4Sm_0.6Te₄/10 vol.% Ni composite, for example, the SEM image and EDS elemental mapping are shown in Figure 1b. All elements, La, Sm, and Te, are homogeneously distributed inside the matrix. The XRD pattern and SEM-EDS results indicated that Sm atoms successfully replaced La atoms into the lattice. The fine Ni particles (smaller than 5 μm) are also well dispersed in the sample. The solubility limit of Sm is not explored fully in La₃Te₄ because the further reduction in carrier concentration is not conducive to the TE properties of lanthanum telluride.

To better understand the crystal structure evolution of La_3−xSm_xTe₄, the Rietveld refinement (Fullprof) is performed based on the PXRD patterns of Figure 1a. The cubic lattice parameters (a = b = c) as a function of the Sm content x are presented in Figure 1c. It is noted that the lattice parameters increase discontinuously with a break near x = 0.65 and then remain the same, which is a coincidence with the shift of diffraction peaks in Figure 1a. As known, samarium and tellurium can form an inhomogeneous mixed-valence compound Sm₃Te₄ with the Th₃P₄-type structure, where Sm²⁺ and Sm³⁺ coexist with mixed valence [26]. Thus, it suggests that the valence fluctuation of Sm can be the main reason for the nonlinear change of the lattice parameters in La_3−xSm_xTe₄ samples.

To address the issue of the evolution of Sm valence states with Sm content x, the XAS spectra at the Sm L₃-edge are displayed in Figure 1d for La_2.6Sm_0.4Te₄, La_2.35Sm_0.65Te₄, and La_2.3Sm_0.7Te₄ composited with 10 vol.% Ni samples. For the sample with x = 0.4, only a single peak was observed at 6714 eV, corresponding to the divalent Sm (Sm²⁺) [29,30]. Confining attention to x = 0.65 and x = 0.7 samples, two distinct peaks, separated by the same energy difference of about 8 eV, were found in the XAS spectra. This unambiguously evidences the mixed-valence character of Sm since the high-energy peak at 6722 eV corresponds to the trivalent Sm (Sm³⁺) [29,30]. In addition, the intensity of the high-energy peak is regularly enhanced with increasing Sm content x without an energy shifts. This demonstrates that the Sm³⁺ content gradually increases with increasing Sm content x. However, it is difficult to quantify the content of Sm²⁺ and Sm³⁺ accurately because Sm²⁺ could be an oxidation relative to Sm³⁺ in the sample’s surface during the etching process before measurements [31]. The XAS spectra infer that the coexistence of Sm²⁺/Sm³⁺ when Sm content x is above ~0.65. Sm²⁺ doped on La³⁺ sites result in an increase in lattice parameters due to the radius of Sm²⁺ (122 pm) being larger than that of La³⁺ (103.2 pm). Sm³⁺ appears in these samples when Sm content x is above ~0.65. The difference of the radius between Sm³⁺ (95.8 pm) and Sm²⁺ causes the lattice parameters to deviate from the linear change with increasing Sm content (see Figure 1c).

Figure 2a,b show the temperature dependence of electrical conductivity σ and Seebeck coefficient S for all our samples. The negative S indicates that all our samples are n-type conduction dominated by electrons. σ decreases with an increase in temperature, showing the typical behavior of heavily doped semiconductors. With increasing Sm content x, a significant reduction in σ is observed throughout the entire measured temperature range. σ for the sample with x = 0.69 is ~3.95 × 10⁴ S m⁻¹ at 300 K, only about 10% of that for the sample with x = 0. Unlike σ, the absolute value of S increases with increasing Sm content x. The absolute value of S for the sample with x = 0.69 is ~256 μV K⁻¹ at 1000 K, over triple of the value for the sample with x = 0. Interestingly, for these samples with large Sm content x, the absolute value of S decreases with increasing temperature at elevated temperatures due to minority carrier activation. For the samples with low Sm content x, the absolute value of S increases linearly with increasing temperatures. A similar behavior was observed in Yb-doped La_{3 − x}Te₄, reported by May [23].

To further investigate the influence of Sm on σ and S, the room temperature Hall carrier concentration n_300K as a function of Sm content x is given in Figure 2c. Assuming that all Sm are Sm²⁺ or Sm³⁺, the expected carrier concentration is also presented for the nominal Sm-doped samples (the black dashed lines in Figure 3c). As expected, n_300K decreases with increasing Sm content due to the additional holes created by the substitution of Sm²⁺ for La³⁺. The n_300K for the sample with x = 0.69 is ~0.5 × 10²¹ cm⁻³, about one-twelfth of that for the pristine sample. The carrier concentration for all samples falls in the middle of two expected lines and follows the trend of the expected Sm²⁺ line. It is inferred that the average valence of Sm is closer to Sm²⁺, which plays a major role in optimizing carrier concentrations.

The Pisarenko relations calculated by the parabolic band model with the different effect mass m* (2.0 m_e, 3.0 m_e, and 4.0 m_e) are plotted in Figure 2d. The effect mass m* of all samples is between 2.0 m_e and 4.0 m_e, which agrees with previous works [15,32]. Notably, the data for these samples with low Sm content (x < 0.6) fall well on the gray dashed line generated by the parabolic band model with m* = 4.0 m_e, indicating that these samples possess similar m*. For the samples with large Sm content (x > 0.6), m* decreases from 4.0 m_e to 2.0 m_e with increasing Sm content. Therefore, Sm content not only affects carrier concentrations but also affects band structure.

Figure 3a shows the temperature dependence of the total thermal conductivity, κ_total, for all samples. As expected, κ_total decreases with increasing Sm content. The samples with large Sm content demonstrate low κ_total values of less than 1 W m⁻¹ K⁻¹ at 1200 K. κ_total in a solid usually consists of two parts. One is the electric thermal conductivity, κ_e, which can be calculated using the Wiedemann–Franz law κ_e = σLT (where σ is the electrical conductivity, L is the Lorenz number, and T is the absolute temperature). Here, L is calculated by the following approximation: L = 1.5 + exp(−|S|/116) [33]. The other one is the lattice thermal conductivity, κ_L, obtained by subtracting κ_e from κ_total. Figure 3b,c present κ_e and κ_L for all our samples. Similarly to κ_total, κ_e and κ_L also decrease with increasing Sm content. To more clearly depict the influence on thermal conductivity by Sm doping, κ_total, κ_e, and κ_L at 300 K are provided in Figure 3d. κ_e decreases from 2.85 to 0.25 W m⁻¹ K⁻¹ with increasing Sm content from x = 0 to 0.69 at 300 K due to a reduction in σ according to the Wiedemann–Franz law, which mainly contributes to the reduction in κ_total. In addition to κ_e, κ_L also decreases from 1.5 to 1.1 W m⁻¹ K⁻¹ at 300 K with increasing Sm content.

In order to further understand the role of Sm-doping on thermal transport, we model the low-temperature κ_L data for La_2.315Sm_0.685Te₄/10 vol.% Ni composites. Three phonon scattering mechanisms are considered in this study, including the Umklapp process (U), grain boundaries (B), and point defects (PD). Detailed calculations can be found in the Supplementary Materials. The contributions from different phonon scattering mechanisms to κ_L are shown in Figure S1a. The κ_L at low temperature is greatly suppressed by grain boundaries, while its effect can be neglected above 50 K. When point-defect scatterings are involved, the calculated curve fits well with the experimental data. This illustrates that the point-defect scattering introduced by Sm substitution contributes to κ_L reduction. In addition, the role of different phonon scattering mechanisms can be more clearly reflected by the spectral lattice thermal conductivities (κ_S) for the sample with x = 0.685 (shown in Figure S1b). Clearly, grain boundary scattering mainly affects the low-frequency phonons, while high-frequency phonon transfers are interrupted by point-defect scattering predominantly. By combining these three phonon scattering mechanisms, the κ_L about 0.5 W m⁻¹ K⁻¹ at 1273 K was obtained in La_2.315Sm_0.685Te₄/10 vol.% Ni composite, as observed in Figure 3b. It is noted that although Sm substitution leads to a decrease in κ_L, La vacancy doping is the most effective method for reducing κ_L in La₃Te₄-based materials [24].

The temperature dependence of the thermoelectric figure of merit, zT, is shown in Figure 4. Due to the optimization of carrier concentration and a reduction in thermal conductivity, significantly enhanced zTs were observed. A maximal zT of about 1.1 at 1273 K is achieved in the La_2.315Sm_0.685Te₄/10 vol.% Ni composite, which is among the highest zT reported in La₃Te₄ compounds. For the purpose of application, the conversion efficiency of thermoelectric devices largely depends on the average figure of merit, zT_ave. As shown in Figure 4b, the maximum zT_ave of 0.8 over 600 K–1273 K in La_2.315Sm_0.685Te₄/10 vol.% Ni composite is obtained in this study, which is comparable to the La_{3 − x}Te₄-based materials reported previously in this temperature range [15,22,23,24].

4. Conclusions

In this work, the successful doping of La_{3 − x}Sm_xTe₄/Ni composites with Sm was achieved, and the results indicate that doping with Sm is an effective strategy to improve zT for La_3−xSm_xTe₄. Experimental results show that the substitution of the Sm can remarkably optimize carrier concentrations. Although the coexistence of Sm²⁺/Sm³⁺ is also found in La_3−xSm_xTe₄ samples with Sm contents above ~0.65, the average valence of Sm still presents closer to +2 for all samples, creating additional holes in the samples to reduce carrier concentrations and, therefore, the electric thermal conductivity. Sm substitution can also introduce point-defect scattering, leading to a reduction in lattice thermal conductivity. Due to the optimization of carrier concentrations and reductions in thermal conductivity, a maximal zT of about 1.1 at 1273 K and zT_ave of 0.8 over 600 K–1273 K was achieved in the La_2.315Sm_0.685Te₄/10 vol.% Ni composite.