Segregation of NiTe₂ and NbTe₂ in p-Type Thermoelectric Bi_0.5Sb_1.5Te₃ Alloys for Carrier Energy Filtering Effect by Melt Spinning

Abstract

The formation of secondary phases of NiTe₂ and NbTe₂ in p-type Bi_0.5Sb_1.5Te₃ thermoelectric alloys was investigated through in situ phase separation by using the melt spinning process. Adding stoichiometric Ni, Nb, and Te in a solid-state synthesis process of Bi_0.5Sb_1.5Te₃, followed by rapid solidification by melt spinning, successfully segregated NiTe₂ and NbTe₂ in the Bi_0.5Sb_1.5Te₃ matrix. Since heterointerfaces of Bi_0.5Sb_1.5Te₃ with NiTe₂ and NbTe₂ form potential barriers of 0.26 and 0.08 eV, respectively, a low energy carrier filtering effect can be expected; higher Seebeck coefficients and power factors were achieved for Bi_0.5Sb_1.5Te₃(NiTe₂)_0.01 (250 μV/K and 3.15 mW/mK²), compared to those of Bi_0.5Sb_1.5Te₃ (240 μV/K and 2.69 mW/mK²). However, there was no power factor increase for NbTe₂ segregated samples. The decrease in thermal conductivity was seen due to the possible additional phonon scattering by the phase segregations. Consequently, zT at room temperature was enhanced to 0.98 and 0.94 for Bi_0.5Sb_1.5Te₃(NiTe₂)_0.01 and Bi_0.5Sb_1.5Te₃(NbTe₂)_0.01, respectively, compared to 0.79 for Bi_0.5Sb_1.5Te₃. The carrier filtering effect induced by NiTe₂ segregations with an interface potential barrier of 0.26 eV effectively increased the Seebeck coefficient and power factor, thus improving the zT of p-type Bi_0.5Sb_1.5Te₃, while the interface potential barrier of 0.08 eV of NbTe₂ segregation appeared to be too small to induce an effective carrier filtering effect.

1. Introduction

Research on renewable energy generation has attracted considerable attention because traditional energy resources have become more limited. In addition, approximately 60% of the total energy is lost as waste heat during energy generation and consumption. Therefore, thermoelectric power generation has several advantages as a power generation technology to use energy more efficiently [1,2]. Thermoelectric materials can directly convert the temperature gradient into electrical energy, and they have attracted considerable attention in recent decades. In particular, bismuth telluride (Bi₂Te₃)-based alloys exhibit high thermoelectric performance at near-room temperature [1,2]. They can be applied either to power generation from low grade waste heat, or in thermoelectric solid-state cooling to replace traditional refrigeration [1,2]. The thermoelectric efficiency is determined in terms of a thermoelectric dimensionless figure of merit, expressed as zT = σ∙S²∙T/κ, where σ is the electrical conductivity, S is the Seebeck coefficient, T is the absolute temperature, and κ is the thermal conductivity [3].

Many approaches have been adopted to enhance the zT of thermoelectric materials. To improve electrical transport properties, approaches such as band convergence, carrier energy filtering, and resonance doping have been applied [4,5,6]. In addition, a variety of defect structures is introduced to the material to reduce the κ by intensifying phonon scattering [7,8,9]. For example, nanograin-composite, dense dislocations formed at the grain boundary, and substitutional doping have been known as effective approaches to reduce κ by intensifying phonon scattering in (Bi₂Te₃)-based alloys [8,9,10,11,12]. Among these approaches, introducing nanoparticles in the thermoelectric materials can improve zT by increasing the electrical transport properties and reduction in κ. The reduction in κ can be attained by additional phonon scattering by the embedded nanoparticles [13], and the increase in the electrical transport properties can be attributed to the carrier energy filtering effect caused by the nanoparticles. The carrier filtering effect is a strong energy dependence of the carrier relaxation time caused by band bending at the interface between metallic inclusions and thermoelectric materials [14,15,16].

In this study, the formation of secondary phases of NiTe₂ and NbTe₂ in p-type Bi_0.5Sb_1.5Te₃ thermoelectric alloys related with the carrier filtering effect was investigated through in situ phase segregation by using the melt spinning process. Metallic NiTe₂ and NbTe₂ were chosen for the successful segregation in Bi_0.5Sb_1.5Te₃ alloys, using the melt spinning process, which was observed in the preliminary experiment. Adding excess stoichiometric composition of NiTe₂ and NbTe₂ in the melt-process of Bi_0.5Sb_1.5Te₃ synthesis successfully segregated NiTe₂ and NbTe₂ phases through the rapid solidification process by melt spinning. As heterointerfaces of semiconducting Bi_0.5Sb_1.5Te₃ with metallic NiTe₂ and NbTe₂, a possible carrier energy filtering effect can be expected. In addition, the segregation of NiTe₂ and NbTe₂ would intensify the phonon scattering, possibly reducing κ.

2. Materials and Methods

Ingots of Bi_0.5Sb_1.5Te₃(NiTe₂)_x (x = 0, 0.003, 0.01, 0.1) and Bi_0.5Sb_1.5Te₃(NbTe₂)_x (x = 0, 0.005, 0.01, 0.1) were prepared by conventional melting and quenching techniques using high-purity elemental Bi (99.999%, shots, 5N Plus, Montreal, QC, Canada), Sb (99.999%, shots, 5N Plus), Te (99.999%, pieces, 5N Plus), Nb (99.99%, powder, Sigma Aldrich, St. Louis, MO, USA), and Ni (99.9%, powder, Sigma Aldrich), as starting materials. Stoichiometric mixtures were loaded into a quartz tube (15 mm in diameter). The tube was vacuum-sealed under 10⁻⁴ torr, and the contents were melted in a box furnace at 1423 K for 20 h, and then quenched to room temperature. The ingots were grounded using a high-energy ball mill (8000D, SPEX, Metuchen, NJ, USA) for 5 min. Subsequently, melt spinning of the prepared ingots (except for Bi_0.5Sb_1.5Te₃(NiTe₂)_0.1 and Bi_0.5Sb_1.5Te₃(NbTe₂)_0.1) was performed, under a copper wheel rotation of 3600 rpm, and the molten solution was shot at 0.03 MPa under Ar. The ribbons were grounded in an agate mortar. Then, disk-shaped polycrystalline bulk samples (10 mm in diameter and 10 mm thickness) were prepared by spark plasma sintering (SPS) at 430 °C for 5 min, under 50 MPa in vacuum.

X-ray diffraction (XRD, D8 Discover, Bruker, Billerica, MA, USA) was performed for phase analysis. The temperature-dependent S and σ parameters were measured within a temperature range from room temperature to 480 K on the samples, perpendicular to the direction of the applied pressure (ZEM-3 Advanced-RIKO, Yokohama, Japan). The nominal temperature gradients were given as 10, 20, and 30 K. The experimental error margin for S and σ are estimated as ±5% and ±3%, respectively. The carrier concentrations were determined by van der Pauw measurements under a magnetic field of 0.5 T (AHT-55T5, Ecopia, Anyang, Korea) in the same direction; the input current can be regulated between 1 nA and 20 mA. The van der Pauw method typically provides the carrier concentration with relative error within 5%. The κ values of the samples were calculated from the expression κ = ρ_s∙C_p∙λ, considering their theoretical density (ρ_s), heat capacity (C_p), and thermal diffusivity (λ) measured across the same direction (perpendicular to the pressing direction), so the zT can be determined. The C_p values from 300 K to 400 K were measured by using a physical property measurement system (PPMS, Quantum Design, San Diego, CA, USA), and 121.6 J mol⁻¹ K⁻¹ were used, which were estimated from Dulong–Petit fitting. The densities of all samples were measured as being higher than 98.5%, from the sintered samples of cylinder shape and 10 mm diameter with ~12 mm height, by the Archimedes method, and the ρ_s for Bi_0.5Sb_1.5Te₃ of 6.88 g/cm³ was used for samples with small addition (x ≤ 0.01). The λ were measured using a laser flash installation (Laser Flash Analysis, LFA 467, Netzsch, Selb, Germany). The experimental error margin for λ is estimated as ±3%.

3. Results

Figure 1a,b shows XRD patterns of the melt-spun ribbons of Bi_0.5Sb_1.5Te₃(NiTe₂)_x (x = 0.1) and Bi_0.5Sb_1.5Te₃(NbTe₂)_x (x = 0.1). Figure 1c,d shows XRD patterns of the SPS-sintered samples of Bi_0.5Sb_1.5Te₃(NiTe₂)_x (x = 0, 0.003, 0.01, and 0.1) and Bi_0.5Sb_1.5Te₃(NbTe₂)_x (x = 0, 0.005, 0.01, and 0.1). For all samples, the diffraction peaks of the major phase were related to the Bi_0.5Sb_1.5Te₃ structure (JCPDS #00-49-1713). The XRD patterns of Bi_0.5Sb_1.5Te₃(NiTe₂)_0.1 and Bi_0.5Sb_1.5Te₃(NbTe₂)_0.1 with x = 0.1 showed additional peaks of the segregated phases of NiTe₂ (JCPDS #01-089-2642) and NbTe₂ (JCPDS #00-021-0605), respectively. Thus, it was confirmed that in situ phase segregation is possible after applying a rapid solidification process. For the samples doped with small fractions (x ≤ 0.01), no secondary phase was observed. However, it can be expected that the same phase segregation of NiTe₂, and NbTe₂ occurred while the amount of the segregations was too small to be detected by XRD. Figure 1e,f shows the lattice parameters calculated from the XRD patterns for Bi_0.5Sb_1.5Te₃(NiTe₂)_x (x = 0, 0.003, and 0.01) and Bi_0.5Sb_1.5Te₃(NbTe₂)_x (x = 0, 0.005, and 0.01) with no changes in lattice parameters.

Figure 2 represents a schematic of energy band diagrams at heterointerfaces formed between Bi_0.5Sb_1.5Te₃ and the secondary phases, based on the concept of carrier energy filtering [14,15]. The main concept, which involves a Bi_0.5Sb_1.5Te₃ matrix with the precipitates, is illustrated in Figure 2. Carrier transfer induces band bending away from the heterointerfaces between the BiSbTe matrix and precipitates. The work function, electron affinity, and bandgap data for NiTe₂, NbTe₂, and Bi_0.5Sb_1.5Te₃ are shown in

Table 1. Work function (Φ) for NbTe₂ and NiTe₂, and electron affinity (χ) and band gap (E_g) for (Bi_0.5Sb_1.5)Te₃.

Phase Segregation	Properties	Work Function (Φ) or Electron Affinity (χ) and Band Gap (Eg)
NbTe2	Metallic	Φ = 4.62 eV
NiTe2	Metallic	Φ = 4.44 eV
(Bi0.5Sb1.5)Te3	Semiconducting	Eg = 0.2 eV, χ = 4.50 eV

. The electron affinity of Bi_0.5Sb_1.5Te₃ is 4.50 eV and the band gap is 0.2 eV [15]. The work functions of NiTe₂ and NbTe₂ are 4.44 eV and 4.62 eV, respectively [17,18]. Barrier potentials may form near the interface; the presence of the electrostatic potential causes energy-dependent scattering of carriers at the heterointerfaces for NiTe₂ and NbTe₂, with Bi_0.5Sb_1.5Te₃ forming potential barriers of approximately 0.26 and 0.08 eV, respectively, as depicted in Figure 2. When hole carriers pass near the surfaces of secondary phases, low energy carriers are scattered, inducing the carrier filtering effect [14,15]. However, the energy blocking effect near the interfaces between NbTe₂ would be smaller due to their low barrier height of 0.08 eV (Figure 2b).

Figure 3 shows the σ and S as a function of temperature of the samples after melt spinning with a small addition (x ≤ 0.01). Bi_0.5Sb_1.5Te₃ exhibited an σ of 465 S/cm at room temperature. As shown in Figure 3a, no change in σ was observed for x = 0.003; however, the x = 0.01 showed an increased σ (~500 S/cm at room temperature). In Figure 3b, the S values, as a function of temperature, are shown, while there is no distinct change in the error range. However, as shown in inset of Figure 3b, the power factor (σS²) was increased from 2.68 to 3.15 mW/mK² for x = 0.01 as both the σ and S increased. In Figure 3c, after the addition of a small amount of NbTe₂, there is no significant change in the σ. As in Figure 3d, the S shows no distinct change in error range by NbTe₂ addition. Therefore, power factors (σS²) also show no increase with the addition of NbTe₂ at all compositions (inset of Figure 3d).

Figure 4 shows the results from the Hall measurement (Hall carrier concentration, n_H, and Hall mobility, μ_H) and the Pisarenko plots at room temperature [14,15,19]. In Figure 4a, no clear relationship between n_H and μ_H is seen, while the n_H does not change for NiTe₂-added samples in the range of 9.5 × 10¹⁸–9.6 × 10¹⁸ cm⁻³. Figure 4b shows a Pisarenko plot (S as a function of n_H) with the effective mass m* of 0.5 m₀, 0.55 m₀, and 0.6 m₀ shown as lines, based on Equation (1).

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

(1)

where h is the Planck constant, k_B is the Boltzmann constant, e is the elemental electric charge and n is the carrier concentration. The gradual increase in m* is clearly shown as x increased for NiTe₂-added samples (Figure 4b), although the noticeable decrease in n_H is not seen (Figure 4a). As n_H could vary for various reasons, the systematic increase in the m* would suggest the presence of the carrier filtering effect due to the energy barrier of 0.26 eV (Figure 2a) as the density of interfaces between NiTe₂ and the matrix increases; for instance, the simultaneous increase in σ and S in NiTe₂-added samples (x = 0.01 compared to x = 0 of Bi_0.5Sb_1.5Te₃(NiTe₂)_x).

Figure 4c shows that the n_H of Bi_0.5Sb_1.5Te₃(NbTe₂)_x (x = 0, 0.005, 0.01) gradually decreased with x; the n_H values of 9.5 × 10¹⁸, 9.0 × 10¹⁸, and 8.7 × 10¹⁸ cm⁻³ for x = 0, 0.005, and 0.01, respectively. In contrast, the, μ_H increased gradually with increasing x, leading to values of 314, 327, and 343 cm²/Vs for x = 0, 0.005, and 0.01, respectively. However, no change in m* was shown. It seems that there is less carrier filtering influence on these samples due to the smaller barrier height (Figure 2b).

Figure 5 shows the temperature dependent κ of all samples (the uncertainty of κ would be around ±3%), while the inset shows the lattice thermal conductivity (κ_latt), calculated by subtracting the electronic contribution (κ_elec) calculated by the Wiedemann–Frantz law (κ_elec = L·σ·T), where L is Lorenz number, which can be estimated with Equation (2) (where L is in 10⁻⁸ WΩK⁻² and S in μV/K) [20],

$$L=1.5+exp\left(-\frac{\left|S\right|}{116}\right)$$

(2)

For Bi_0.5Sb_1.5Te₃(NiTe₂)_x samples, a gradual decrease in κ_latt (0.81, 0.78, and 0.75 W/mK for x = 0, 0.003, and 0.01) was seen at room temperature (inset of Figure 5a), while there seems to be no distinct change in κ in Figure 5a. At high temperatures, the different influences from the bipolar conduction may affect the κ_latt. The second phases of NiTe₂ may scatter the phonons further to reduce room temperature κ_latt. For Bi_0.5Sb_1.5Te₃(NbTe₂)_0.01, the κ (and κ_latt) decreased to 0.92 W/mK (0.70 W/mK) at room temperature, compared to 1.03 W/mK (0.81 W/mK) of Bi_0.5Sb_1.5Te₃.

The estimated zT values of the samples are shown in Figure 6. For Bi_0.5Sb_1.5Te₃(NiTe₂)_x samples (Figure 6a, the uncertainty of zT would be around ±17%), the zT was enhanced from 0.79 to 0.82 and 0.98 for x = 0.003 and 0.01, respectively, as the power factor increased due to increases in S, power factor, and m* (Figure 3b and Figure 4b). We speculate that the interface barrier height of 0.26 eV between Bi_0.5Sb_1.5Te₃ and NiTe₂ (Figure 2a) induced the carrier filtering effect, while there was slight systematic κ_latt reduction as well (inset of Figure 5a). For Bi_0.5Sb_1.5Te₃(NbTe₂)_x samples (Figure 6b), with the much smaller barrier height of 0.08 eV between Bi_0.5Sb_1.5Te₃ and NbTe2 (Figure 2b), no increases in S, power factor, and m* (Figure 3d and Figure 4d) were seen. However, the reduction in κ_latt to 0.70 W/mK at room temperature (inset of Figure 5b) was achieved for x = 0.01. Therefore, the zT was also enhanced slightly from 0.79 to 0.83 and 0.94 for x = 0.005 and 0.01; however, this was mainly due to decreased κ (and κ_latt) for x = 0.01 (Figure 5b). For Bi_0.5Sb_1.5Te₃(NiTe₂)_x samples, the improved electronic transport properties, caused by the possible carrier filtering effect, enhanced the zT, while the reduced thermal conductivity enhanced the zT in Bi_0.5Sb_1.5Te₃(NbTe₂)_x samples.

4. Conclusions

The electrical and thermal transport properties of Bi_0.5Sb_1.5Te₃ with the formation of secondary phases of NiTe₂ and NbTe₂ were analyzed in regard to the carrier filtering effect. The possible carrier filtering effect was shown with enhanced Seebeck coefficient, power factor and effective mass for the NiTe₂ added samples, where the heterointerfaces are expected to form the energy barrier of 0.26 eV with the Bi_0.5Sb_1.5Te₃ matrix. However, for the NbTe₂ added samples with the energy barrier of 0.08 eV, only slight enhancement of the power factor was seen without increase in effective mass, implying that there is only minor influence of the carrier filtering effect. While the additional phonon scattering by the segregation was anticipated, the influence on κ was slight. Consequently, zT the zT was enhanced to 0.98 for Bi_0.5Sb_1.5Te₃(NiTe₂)_0.01 from 0.79 for Bi_0.5Sb_1.5Te₃ at room temperature, mainly due to the increases in S and m* by the possible carrier filtering effect.