Influence of nanoscale Ag₂Te precipitates on the thermoelectric properties of the Sn doped P-type AgSbTe₂ compound

Thermoelectric (TE) systems have recently received an increasing amount of attention because they facilitate direct and reversible conversion between heat and electrical energy, thus enabling environmentally friendly refrigeration and electric power generation.^1,2 The performance of TE materials is assessed using a dimensionless figure of merit (ZT) = α²σT/κ, where α, σ, T, α²σ, and κ are the Seebeck coefficient, electrical conductivity, absolute temperature, power factor (PF), and total thermal conductivity, respectively. Therefore, optimal TE materials must feature a high PF and low thermal conductivity.¹ AgSbTe₂ is a potential TE material because of its exceptionally low thermal conductivity (κ of approximately 0.7 W m⁻¹ K⁻¹).^3–14 AgSbTe₂ is a narrow bandgap semiconductor that has a disordered NaCl-type structure (Fm-3m), with Ag⁺ and Sb³⁺ randomly occupying the Na⁺ site.¹⁴ According to electronic structure calculations,¹⁵ the top of the valence band is Ag(4d)–Te(5p) hybridized anti-bonding states, therefore the formation energy of Ag-vacancy V(Ag) is small and can easily be compensated by Sb anti-site defects Sb(Ag), double donor at Ag-site, then following self-regeneration¹⁶ occurs: [2V(Ag) + Sb(Ag)]. When we doped Sn-acceptor at Sb-site, this is p-type doping due to the Sn-acceptor, it reduces the formation energy of [2V(Ag) + Sb(Ag)], then increases the Sb₂Te₃ and Ag₂Te, since 2[AgSbTe₂] = Sb₂Te₃ + Ag₂Te. This is caused by the spinodal nano-decomposition¹⁷ of Ag and V(Ag). Numerous recent studies have focused on the effects of nanostructures and nanoprecipitates on the TE properties of AgSbTe₂ especially alloys with other cubic compounds such as PbTe, called lead antimony silver tellurium (LAST)^18–21 and GeTe, called tellurium antimony germanium silver (TAGS)^22,23 and have successfully exhibited enhanced TE performance.

We have recently reported that doping with bismuth (Bi) and indium (In) in the AgSbTe₂ system has improved TE performance.^24,25 In this paper, Sb³⁺ was substituted with Sn²⁺ in p-type Ag(Sb_1−xSn_x)Te₂ systems to optimize the charge-carrier concentration and simultaneously reduce lattice thermal conductivity by enhancing phonon scattering through the naturally formed fine dispersion of Ag₂Te nanoprecipitates in the matrix. When x = 0.03, the sample exhibited the highest ZT value of 1.1 at 600 K; therefore, Ag(Sb_0.97Sn_0.03)Te₂ can be used as a p-type TE material.

Samples with a nominal composition of AgSb_1−xSn_xTe₂ (x = 0.01, 0.03, 0.05, and 0.07) were synthesized using the vertical Bridgman method. The synthesis procedures detailed in our previous report²⁴ were used to obtain highly dense crystalline ingots with a dark silvery metallic shine as shown in Fig. 1(a). The crystalline ingots were cut into bar shapes of approximately 3 mm × 3 mm × 14 mm, formed into circular discs of 12 mm in diameter and 2–3 mm thick, and polished before physical properties were measured (Fig. 1(b)).

The phases of the Ag(Sb_1−xSn_x)Te₂ samples were analyzed using powdered X-ray diffraction (XRD; X’Pert PRO-PANalytical, CuKα radiation) and differential scanning calorimetry (DSC, NETZSCH, STA 449, with a heating rate of 10 K min⁻¹ and a sample mass of 25 mg). The microstructure and quantitative chemical analysis of the bulk samples were determined by performing field emission electron micro-probe analysis (FE-EPMA: JEOL JXA-8500F). The sample preparation procedure for the EPMA technique is shown in Fig. S2 of the supplementary material.²⁸ The thermal conductivity κ was calculated using the equation κ = D × C_p × d, where D is the thermal diffusivity coefficient measured using the laser flash method (NETZSCH, LFA 457), C_p is the specific heat capacity obtained using DSC (NETZSCH, STA 449), and density d was determined using the Archimedes method. The relative densities of all samples were greater than 99.9% (approximately 7.11–7.12 g/cm³). The temperature dependence of the electrical conductivity σ and Seebeck coefficient α was measured simultaneously by using commercial equipment (ZEM-3, ULVAC-RIKO, Japan) in a He atmosphere at temperatures ranging from 300 to 600 K. The Hall effect measurements were performed using a five-probe configuration with the magnetic field sweeping between +5.0 and −5.0 T (Quantum Design, PPMS Ever-cool II). The average measurement errors in σ, α, and κ were estimated to be approximately ±2%, ±3%, and ±3%, respectively.

Fig. 1(c) shows the powder XRD patterns of the AgSb_1−xSn_xTe₂ (x = 0.01, 0.03, 0.05, and 0.07) samples at room temperature. All of the main peaks can be indexed into major-phase fcc AgSbTe₂ structures (reference code: 01-089-3671) except for a low-intensity peak corresponding to the Ag₂Te secondary phases (reference code: 00-004-0795) that were observed in the sample of which x = 0.05 and 0.07. The lattice parameter as a function of the Sn content x calculated using the Reitveld refinement is shown in the inset of Fig. 1(c); the lattice parameter decreased substantially from x = 0 to 0.03 as expected because the ionic radii of the dopant Sn²⁺ (0.69 Å) were slightly smaller than that of the host Sb³⁺ (0.76 Å). A slight upturn was observed in the specimens of which x ≥ 0.5 (deviation from Vegard's law-type behaviour, red dashed line). The site occupancy as a function of the Sn composition for the Ag(Sb_1−xSn_x)Te₂ series is shown in Fig. S1 of the supplementary material²⁸ and listed in Table I. Indeed, it is seen that the population of Sn atoms in Sb sites increases with the Sn content, beyond that it almost remains constant. Therefore, the existence of a second phase is highly likely for highly doped compositions. To determine the phase transition of the samples, the DSC heat flows were plotted (Fig. 1(d)). We observed endothermic peaks at 422 K and 421.5 K associated with α−β structural phase transition of Ag₂Te¹⁰ in the samples for which x = 0.05 and 0.07, respectively. The experimental results of both DSC and XRD analyses implied that Sn doping was nearly optimal in Ag(Sb_1−xSn_x)Te₂ samples of which x = 0.03.

Fig. 2 shows EPMA backscattered electron images of the AgSb_1−xSn_xTe₂ (x = 0.01, 0.03, 0.05, and 0.07) samples. The surface of the sample of which x = 0.01 was homogeneous, exhibiting no traces of secondary phases (Fig. 2(a)). Conversely, the surface of the sample of which x = 0.03 in Fig. 2(b) exhibited noticeable secondary phase precipitates Ag₂Te (dark area) embedded in the AgSbTe₂ bulk matrix (light area). As shown in the upper right inset of Fig. 2(b), the Ag₂Te precipitates, which had a feature size of 100−500 nm, were dispersed in the matrix. The precipitates coarsened in samples of which the Sn content was high (i.e., x > 0.03; Figs. 2(c) and 2(d)), as indicated by the increased precipitate size and decreased number density. Fig. 3(a) shows the elemental distribution of the AgSb_0.97Sn_0.03Te₂ sample determined using EPMA confirming that the grain boundary areas were rich in Ag and Te. The measurements were obtained at different locations on the matrix of the sample and the average total Sn content (x = approximately 0.028) measured in the matrix was close to that of the nominal composition x = 0.03 (Fig. 3(b) and Table I). However, when the Sn content was further increased (x = 0.05 and 0.07), the samples exhibited coarsening of Ag₂Te precipitates as observed using EPMA (see Figs. S3 and S4 of the supplementary material²⁸), this observation is consistent with the XRD and DSC data presented in this paper. That macro-scale characterization technique, such as XRD and DSC, cannot detect the existence of a small fraction of Ag₂Te precipitates in the AgSb_0.97Sn_0.03Te₂ samples was expected.

Fig. 4 shows the temperature dependences of the TE transport properties of the AgSb_1−xSn_xTe₂ (x = 0.01, 0.03, 0.05, and 0.07) samples. Fig. 4(a) shows the electrical conductivity σ of the Ag(Sb_1−xSn_x)Te₂ series in the temperature range of 300−600 K. The electrical conductivity decreased as the temperature increased and then increased, indicating typical degenerate semiconductor behavior. This behavior was observed in all samples except for the sample of which x = 0.03, which exhibited conventional semiconductor behavior. The sample of which x = 0.03 exhibited an electrical conductivity of value of 3 × 10⁴ S m⁻¹ at 543 K, representing the highest value amongst all of the samples. To explain the electrical transport behavior in this system (Sn-doped AgSbTe₂ samples), the Hall carrier concentration n_H (see Fig S5 of the supplementary material²⁸) and the mobility μ_H were measured on these samples, and the results are listed in Table I. Due to the carrier-carrier scattering, the mobility exhibited contrary composition dependence with carrier concentration. According to previous study, the p-type carriers in AgSbTe₂ are mainly caused by Ag vacancies¹³ and tendency to form Ag₂Te precipitates. So the maximum carrier concentration of 7.7 × 10¹⁹ cm⁻³ was obtained with the x = 0.03 because of the presence of nanoscale Ag₂Te precipitation⁶ in the bulk matrix. With the further increase of x value from 0.05 to 0.07, the carrier concentration decreased while the mobility increased due to the decrease of carrier concentrations. As shown in Fig. 4(b), the Seebeck coefficients (α) of all samples were positive throughout the temperature range indicating p-type electrical transport behavior. The substantial decrease in the Seebeck coefficient when the Sn concentration increased was caused by the decrease in the calculated effective mass obtained from the relationship between n_H and α (Table I; detailed calculations are provided in the supplementary material²⁸). The PF (α²σ) was calculated based on the values of σ and α, as shown in Fig. 4(c). The sample of which x = 0.03 exhibited the maximal value of 1.38 mW m⁻¹ K⁻² at 600 K because the σ value was enhanced at high temperatures. Overall, the electrical transport properties were markedly improved by using the optimal Sn filling fraction probably because of the presence of the nanostructural features in the bulk matrix. Fig. 4(d) shows the thermal diffusivity D measurements of the Ag(Sb_1−xSn_x)Te₂ samples. The Cp measurements (Fig S6 of the supplementary material²⁸) shows strong agreement with those reported in the literature¹² regarding typical AgSbTe₂ samples (approximately 0.205 J g⁻¹ K⁻¹). Fig. 4(e) shows the total thermal conductivity κ values for the Ag(Sb_1−xSn_x)Te₂ series in the temperature range of 300−600 K. The Sn-doped AgSbTe₂ samples exhibited lower thermal conductivity compared than did pristine AgSbTe₂ samples. The electronic thermal conductivity κ_e can be calculated by using the Wiedmann–Franz law (κ_e = L × σ × T), where L is the Lorentz number. The lattice thermal conductivity κ_L can be calculated by subtracting κ_e from κ, thus κ_L = κ − L × σ × T. As shown in Fig. 4(f), the sample of which x = 0.03 exhibited a low κ_L value of 0.35 W/m K at 510 K, which is lower than those of pristine AgSbTe₂ (0.52 W m⁻¹ K⁻¹). The κ_L of samples of which x ≤ 0.03 decreased substantially as the Sn content increased; however, the κ_L value increased sharply when x > 0.03 because the Ag₂Te precipitate in the matrix coarsened, indicating that x = 0.03 is the optimal doping amount. The low κ_L achieved in the samples of which x = 0.03 can be attributed to the combined effects of the lattice defects engendered by substituting Sn into the Sb sites in AgSbTe₂ systems^24,25 and the presence of nanometer-scale precipitates of the Ag₂Te phase with a feature size as large as 100–500 nm which was effective for enhancing phonon scattering.^9,26,27 Further studies like transmission electron microscope (TEM) analyses and ultrafast dynamics of sample of which x = 0.03 by ultrafast laser spectroscopy in order to understand nanoscale thermal transport will be subjected to future direction. The values of α²σ and κ were used to calculate the TE figure of merit ZT which is plotted in Fig. 5(a). The AgSb_0.97Sn_0.03Te₂ sample exhibited a maximal ZT value of 1.1 is achieved at 600 K because of its enhanced PF and reduced thermal conductivity at high temperatures. Fig. 5(b) shows a three-dimensional plot of recently reported ZT values for various doping elements in AgSbTe₂ systems in the temperature range of 300−700 K. This comparison shows that AgSbTe₂ compounds featuring the optimal Sn doping concentration are potential candidates for TE applications in the medium temperature range. Thermal stability can be ensured because the TE measurement properties for AgSb_0.97Sn_0.03Te₂ samples exhibited negligible change in repeated thermal cycles (heating and cooling cycles) within a temperature range of 300−600 K (Fig. S7 of the supplementary material²⁸). The main finding of this investigation is that doping an Ag(Sb_1−xSn_x)Te₂ system with an optimal amount of Sn (x = 0.03) is the most effective method for controlling the morphology and size of nanoprecipitates and achieving high ZT values, and that doping the system with amounts substantially higher than the optimal amount yields adverse effects.

In summary, the microstructure and TE properties of Sn-doped AgSbTe₂ compounds were investigated and thermal analysis was conducted. According to XRD, DSC, and EPMA analyses, the filling limit of the Sn content in AgSbTe₂ systems is approximately x = 0.03. The lattice thermal conductivity of Ag(Sb_1−xSn_x)Te₂ (x = 0.01 and 0.03) decreased as Sn content increased. The compound AgSb_0.97Sn_0.03Te₂ exhibited the lowest lattice thermal conductivity because of the enhanced phonon scattering that is generated by substitutional effects and presence of Ag₂Te nanoprecipitates in the matrix. The AgSb_0.97Sn_0.03Te₂ sample exhibited a higher PF and low thermal conductivity, resulting in a high ZT value of 1.1 at 600 K indicating that this sample is highly suitable for TE power generation applications. Our subsequent study will focus on improving the ZT value of AgSbTe₂ compounds by optimizing the doping concentration and calibrating nanoscale precipitates.

This work was supported by Academia Sinica and the National Science Council, Taiwan, Republic of China, Grant No. NSC100-2112-M-001-019-MY3.