Synthesis of High-Density Bulk Tin Monoxide and Its Thermoelectric Properties

Abstract

SnSe exhibits exceptionally high thermoelectric (TE) figure of merit zT mainly due to its ultralow lattice thermal conductivity (κ_lat) [L.-D. Zhao et al.: Nature 508 (2014) 373.]. It is considered that strong lattice anharmonicity caused by the lone pair electrons of Sn²⁺ results in the ultralow κ_lat. Here, we focus on SnO because it has the lone pair electrons of Sn²⁺ like SnSe. Bulk samples of SnO were synthesized by low-temperature high-pressure spark plasma sintering and their TE properties were examined. The present study revealed that SnO exhibits very low κ_lat (1.44 Wm⁻¹K⁻¹ at 573 K) compared with SnO₂ which has no lone pair electrons. The Grüneisen parameter (γ) of SnO was evaluated to be 1.70 and this high γ leads to large lattice anharmonicity and thereby low κ_lat. Even though SnO has low κ_lat, the zT values were significantly low compared with SnSe. The maximum zT value of SnO was 0.00141 at 573 K. Since the main reason of this low zT is its non-optimized carrier concentration, the zT of SnO can be enhanced through the carrier concentration optimization.

1. Introduction

Thermoelectric (TE) materials can convert heat into electricity and vice versa. The conversion efficiency of TEs is mainly determined by the materials’ properties called dimensionless figure of merit, zT = S²σT/κ, where S is the Seebeck coefficient, σ is the electrical conductivity, T is the absolute temperature, and κ is the total thermal conductivity (κ = κ_lat + κ_el, sum of the contributions from lattice and electronic).¹⁾ Thus, to enhance the zT, high electrical properties, i.e. high S²σ, as well as low κ are required. Since there exists a trade-off relationship among S and σ or κ_el depending on the carrier concentration, the most effective way to enhance zT is the reduction of κ_lat, which is nearly independent of the carrier concentration.

The reduction of κ_lat with keeping high electrical properties has been demonstrated in many TE materials, including filled-skutterudites^2–4) and nanostructured materials.^5–10) In the filled-skutterudites, AM₄X₁₂, the A atoms called filler elements are bonded weakly to the other atoms and rattles, leading to strong scattering of phonons, resulting in low κ_lat.¹¹⁾ In the nanostructured materials, the nanoscale intermediates scatter phonons much more effectively than carriers, when the size of the structure is smaller than the phonon mean free path but larger than the carrier mean free path, leading to a reduction in κ_lat without a negative effect on the electrical properties.^12,13) On the other hand, a new concept of lattice anharmonicity to reduce κ_lat has been proposed recently.^14–20) For example, an extremely low κ_lat (0.23 ± 0.03 Wm⁻¹K⁻¹ at 973 K) was obtained in SnSe,²¹⁾ and it has been considered that this significantly reduced κ_lat is mainly due to the high rate of lattice anharmonicity induced by the lone pair electrons of Sn²⁺. This anharmonicity is driven by a stereochemically active lone pair electrons produced by the interaction between Sn (5p) and antibonding Sn (5s)-X (p) states.²²⁾ Owing to this quite low κ_lat, SnSe exhibits record-high zT value of 2.6 ± 0.3 at 923 K.²¹⁾ However, despite having very high zT, usability of SnSe is limited by several disadvantages of using Se, because Se is highly toxic and small production scale.^23,24)

Here, we focus on SnO which has lone pair electrons of Sn²⁺ like SnSe as O is in the same group 16 elements as Se.²²⁾ A fine bulk sample of SnO is hard to be synthesized because SnO is decomposed into SnO₂ and Sn between 700 and 1050 K.²⁵⁾ It is reported that the following disproportionation reactions of SnO occur in a N₂ atmosphere:   

\begin{equation*} \text{4SnO}(s) \to \text{Sn$_{3}$O$_{4}$}(s) + \text{Sn}(l) \end{equation*}

  

\begin{equation} \text{Sn$_{3}$O$_{4}$}(s) \to \text{2SnO$_{2}$}(s) + \text{Sn}(l), \end{equation} (1)

This is why the physical properties of bulk SnO have been hardly measured so far. In this paper, we report on the temperature-dependent κ_lat and the Grüneisen parameter (γ) at room temperature of SnO and discuss the relationship among the lone pair electrons, anharmonicity, and κ_lat. γ can be used as a measure of anharmonicity. For further discussion of the influence of the lone pair electrons on the κ_lat, we compared the κ_lat and γ values of SnO with those of SnO₂ which has no lone pair electrons. Figures 1(a) and 1(b) show the crystal structures of SnO and SnO₂, respectively. SnO has a tetragonal structure with the space group P4/nmm,²⁶⁾ while SnO₂ has a tetragonal structure with the space group P4₂/mnm.²⁷⁾ Unlike SnO₂ with a three dimensional structure, SnO has a layered structure where the lone pair electrons of Sn²⁺ above the apexes of SnO₄ square-based pyramids reside in the interstitial space between the layers.²⁸⁾ Such a structural feature is expected to be responsible for strong lattice anharmonicity.

Fig. 1

Three-dimensional crystal structures of (a) SnO and (b) SnO₂.

2. Experimental Procedure

SnO (200 mesh, 99.9%, Furuuchi Chemical Co. Ltd.) powders were used as raw materials to prepare the bulk samples. To determine the sintering conditions, high-temperature stability of SnO was checked by the thermogravimetry/differential thermal analysis (TG-DTA) measurement both in air and an Ar atmosphere with a heating rate of 20 Kmin⁻¹. The SnO powders were placed into a tungsten-carbide die, followed by spark-plasma-sintering (SPS) in an Ar flow atmosphere under four different pressure/temperature conditions (#1 500 MPa/573 K, #2 500 MPa/598 K, #3 500 MPa/623 K, and #4 300 MPa/673 K). The powders used for the SPS condition #2 were ball-milled at a rotational speed of 300 rpm for 10 min before the sintering process. In addition to SnO, commercially-available SnO₂ tablets (ϕ 10.0 mm, height 5.0 mm, 99.99%, 88.5%T.D., Kojundo Chemical Laboratory Co., Ltd.) were prepared and their physical properties were characterized to compare the data with those of SnO. The phase state and microstructure of the samples were evaluated at room temperature by powder X-ray diffraction analysis (XRD; Ultima IV, Rigaku Co.) with Cu Kα radiation and scanning electron microscopy (SEM, JSM-6500F, JEOL) equipped with an energy dispersive spectrometer (EDS). κ was calculated by κ = αC_Pd, where α, C_P, and d are the thermal diffusivity, specific heat capacity at constant pressure, and density, respectively. α parallel to the SPS (pressing) direction was measured in the temperature range from room temperature to 673 K in an Ar atmosphere using a light flash apparatus (Netzsch LFA-467). The C_P data were obtained from the literatures.^29,30) d was calculated from the measured weight and dimensions of the bulk samples. κ_lat was calculated by subtracting κ_el from κ, where κ_el was estimated by the Wiedemann-Franz law, i.e. κ_el = LTσ (L is the Lorenz number for nondegenerate limit: 1.5 × 10⁻⁸ WΩK⁻²). The volumetric thermal expansion coefficient (α_V) was evaluated by two methods, high temperature XRD analysis in a He atmosphere and direct measurement using a dilatometer in an Ar flow atmosphere. The α_V perpendicular to the press direction was measured using a dilatometer. The α_V of SnO and SnO₂ were evaluated in the temperature ranges from room temperature to 673 K and 973 K, respectively. The longitudinal (v_l) and transverse speed of sound (v_t) were measured by an ultrasonic pulse-echo method at room temperature in air using 10 MHz sound wave echogenic transducers. These properties were measured in the parallel to the press direction. S and σ of the bulk samples of SnO were simultaneously measured in the temperature range from room temperature to 573 K using a commercial apparatus (Advance Riko ZEM-3) under a He atmosphere. These properties were measured in the perpendicular to the press direction. The Hall coefficient (R_H) was measured by the Van der Pauw technique with a Hall measurement system (Toyo Resitest8300) at room temperature in vacuum under an applied magnetic field of 0.5 T. The Hall carrier concentration (n_H) and Hall mobility (μ_H) were calculated from the expressions n_H = 1/|R_H|e (e: elementary electric charge) and μ_H = |R_H|σ, respectively.

3. Results and Discussion

It was confirmed from the XRD analysis that the as-received SnO powders and SnO₂ tablet are single phase materials with no remarkable impurities (Fig. 2). Figures 3(a) and 3(b) show the TG-DTA curves of as-received SnO powders performed in an Ar-flow atmosphere and in air, respectively. As can be seen in Fig. 3(a), SnO exhibits good thermal stability in an Ar-flow atmosphere up to around 1000 K. On the other hand, SnO oxidizes in air at around 700 K. In case of the complete oxidation of SnO to SnO₂, the mass change is around 12%. However, in the present case, the mass change is only 1.6%, which can be confirmed from the TG curve in Fig. 3(b). This result means that the oxidation occurs only at the surface of the SnO powders and thus SnO₂ layers act as barriers for farther oxidation. From the TG-DTA results, we determined the SPS conditions.

Fig. 2

Powder XRD patterns of as-received SnO powders and SnO₂ tablet, together with the literature data of SnO³¹⁾ and SnO₂.³²⁾

Fig. 3

TG-DTA curves of as-received SnO powders, (a) in an Ar-flow atmosphere and (b) in air.

The powder XRD patterns of the bulk samples of SnO synthesized by SPS are shown in Fig. 4, together with the literature data of SnO,³¹⁾ SnO₂,³²⁾ and Sn.³³⁾ It can be confirmed that the samples #1 (573 K, 500 MPa) and #2 (598 K, 500 MPa) are almost pure SnO, while the samples #3 (623 K, 500 MPa) and #4 (673 K, 300 MPa) contain SnO₂ clearly as an impurity phase. At 623 K and above, the sample is stable during the TG-DTA measurement while it decomposes during the SPS process. It has been reported that the decomposition rate of SnO into SnO₂ and Sn remarkably increases under high pressure (2–15 GPa).³⁴⁾ Therefore, the decomposition rate would also increase under the SPS pressure (500 MPa), which leads to the decomposition of the sample during the SPS process. Table 1 summarizes the lattice parameters and densities of the samples. Among the SnO samples, the lattice parameters are slightly different each other, likely due to the slight differences in the O/Sn ratio. The sample #2 exhibits the highest relative density of 90.2%, thus, we used this sample for physical properties measurements. The bulk XRD pattern of sample #2 for the perpendicular plane to the SPS (pressing) direction is shown in Fig. 4. It is confirmed that the intensity of the (002) peak of the bulk sample is slightly higher than that of the powder sample, meaning that the bulk sample is slightly oriented to the [001] direction. Although the bulk sample is slightly oriented to the [001] direction, the physical properties are evaluated under the assumption that the sample is not anisotropic but isotropic. The appearance of the bulk sample of SnO (sample #2) and as-received SnO₂ tablet is shown in Figs. 5(a) and 5(b), respectively. Due to the difference in the band gap energy (E_g), SnO (E_g = 0.73 eV³⁵⁾) is black colored, while SnO₂ (E_g = 3.64 eV³⁶⁾) is white. Figure 6 shows SEM images and the EDX mapping images of the bulk sample of SnO (sample #2) and as-received SnO₂ tablet. Both samples are homogeneous and no remarkable cracks and pores. SnO₂ tablet has slightly smaller grains (< ∼5 µm) than those of the SnO bulk sample. The white area in the SEM image of SnO may correspond to SnO₂ existed as the impurity phase.

Fig. 4

Powder XRD patterns of the bulk samples of SnO synthesized by SPS and bulk XRD pattern of sample #2 for the perpendicular plane to the SPS (pressing) direction, together with the literature data of SnO,³¹⁾ SnO₂,³²⁾ and Sn.³³⁾

Table 1 Lattice parameters and relative densities of the samples. The literature data of the lattice parameters of SnO³¹⁾ and SnO₂³²⁾ are shown for comparison.

Fig. 5

Appearance of the bulk samples of (a) SnO (sample #2) and (b) as-received SnO₂ tablet.

Fig. 6

SEM and EDX mapping images of (top three images) the bulk sample of SnO (sample #2) and (bottom three images) as-received SnO₂ tablet.

Figure 7 shows the temperature dependences of the lattice thermal conductivities (κ_lat) of SnO and SnO₂. Since σ values of both SnO and SnO₂ are very small, κ_el can be neglected, i.e., κ ≈ κ_lat. In the present study, the samples contain pores, thus the Maxwell-Eucken equation was adopted for the correction of porosity in κ_lat. The equation is expressed as follows:³⁷⁾   

\begin{equation} \kappa_{0} = \kappa_{\text{P}}\frac{(1 + \beta P)}{(1 - P)}, \end{equation} (2)

where κ₀ is the thermal conductivity with no pores, κ_P is the thermal conductivity of a sample with porosity P, and β is a constant depending on the shape of the pores (here, β = 0.5³⁸⁾). The density-corrected κ_lat are plotted along with the measured data. As can be confirmed in Fig. 7, SnO has much smaller κ_lat values than SnO₂.

Fig. 7

Temperature dependence of the lattice thermal conductivity (κ_lat) of the bulk sample of SnO (sample #2) and as-received SnO₂ tablet. Density-corrected data are shown in lines.

To understand the origin of the low κ_lat of SnO compared with SnO₂, the κ_lat is analyzed by using various physical properties obtained in the present study. Figure 8 shows the temperature dependences of the lattice parameters and lattice volume of SnO and SnO₂ obtained by the HT-XRD measurement. It can be confirmed that all the parameters of both SnO and SnO₂ increase linearly with increasing temperature. The average linear thermal expansion coefficient (α_L) and the average volumetric thermal expansion coefficient (α_V) are determined by the following equations:   

\begin{equation} \alpha_{\text{L}} = \frac{1}{l_{0}}\frac{dL}{dT}, \end{equation} (3)

  

\begin{equation} \alpha_{\text{V}} = \frac{1}{V_{0}}\frac{dV}{dT}, \end{equation} (4)

where l₀ and V₀ are the lattice parameter and lattice volume at room temperature, respectively. The values of α_L along the a-axis (α_a) and c-axis (α_c) and α_V are summarized in Table 2. The α_V of SnO (9.97 × 10⁻⁵ K⁻¹) is clearly larger than that of SnO₂ (1.89 × 10⁻⁵ K⁻¹), showing that larger anharmonicity of SnO than that of SnO₂. As for SnO, the α_c (5.21 × 10⁻⁵ K⁻¹) is larger than the α_a (2.33 × 10⁻⁵ K⁻¹), which is consistent with the fact that SnO is bounded by a strong covalent bond in the a-axis direction, while by a weak van der Waals force in the c-axis direction.³⁹⁾ As shown in Fig. 1, SnO has a layered structure in the c-axis direction, while SnO₂ does not have such structure.

Fig. 8

Temperature dependences of the lattice parameters a and c, and lattice volume V of SnO (top three figures) and SnO₂ (bottom three figures).

Table 2 Linear thermal expansion coefficient along the a-axis α_a, linear thermal expansion coefficient along the c-axis α_c, linear thermal expansion coefficient α_L, volumetric thermal expansion coefficient α_V, longitudinal speed of sound v_l, transverse speed of sound v_t, average speed of sound v_m, Debye temperature θ_D, bulk modulus B, and Grüneisen parameter γ of SnO and SnO₂. The thermal expansion data are the average values in the temperature range between 300 and 700 K for SnO and 300 and 1000 K for SnO₂. The data of the speed of sound are the density-corrected ones. The literature data for SnO₂^32,42,43) are shown for comparison.

To check the validity of the values of α_L and α_V, we evaluated the parameters by direct measurements using a dilatometer (TD5000SA, Bruker AXS Inc.) under an Ar atmosphere. The temperature range was set as 300–673 K for SnO and 300–973 K for SnO₂. In the dilatometer measurement, the α_L can be defined by the following equation:   

\begin{equation} \alpha_{\text{L}} = \frac{1}{L_{0}}\frac{\Delta L}{\Delta T}, \end{equation} (5)

where L₀ is the initial length, ΔL is the change of the length, and ΔT is the change of the temperature. The α_V can be calculated using the following relationship:   

\begin{equation} \alpha_{\text{V}} = 3\alpha_{\text{L}}, \end{equation} (6)

Figure 9 shows the results of the dilatometer measurement. It can be confirmed that SnO shows larger thermal expansion than SnO₂. The α_L and α_V determined by the dilatometer measurements are summarized in Table 2. The α_V of SnO and SnO₂ are 5.57 × 10⁻⁵ K⁻¹ and 1.76 × 10⁻⁵ K⁻¹, respectively. These smaller values than those determined by the HT-XRD measurements would be caused by the effect of porosity. It has been suggested that the porosity suppressed the thermal expansion of bulk samples.⁴⁰⁾

Fig. 9

Linear thermal expansion of SnO and SnO₂, measured by using a dilatometer.

In order to obtain accurate values of the sound velocity, the porosity correction was performed by using the measured sound velocities for three bulk samples of SnO with different porosity. The value corresponding to the sample with zero porosity was evaluated by linear extrapolation.⁴¹⁾ Since it was difficult to prepare several samples of SnO₂ with different porosity, the porosity correction for SnO₂ was done by using the same parameter obtained for SnO. The values obtained by the porosity correction of the longitudinal speed of sound (v_l) and transverse speed of sound (v_t) are summarized in Table 2, together with the literature data of SnO₂.^32,42,43) The average speed of sound (v_m), Debye temperature (θ_D), and bulk modulus (B) were calculated using the following equations:^44,45)   

\begin{equation} v_{\text{m}}^{-3} = \frac{(v_{\text{l}}^{-3} + 2v_{\text{t}}^{-3})}{3}, \end{equation} (7)

  

\begin{equation} B = d_{0}\left(v_{\text{l}}^{2} - \frac{4}{3}v_{\text{t}}^{2}\right), \end{equation} (8)

  

\begin{equation} \theta_{\text{D}} = v_{\text{m}}\left(\frac{h}{k_{\text{B}}}\right)\left(\frac{3nN_{\text{A}}d_{\text{0}}}{4\pi M_{\text{W}}}\right)^{\frac{1}{3}}, \end{equation} (9)

where d₀ is the theoretical density, h is the Planck constant, k_B is the Boltzmann constant, n is the number of atoms in the molecule, N_A is the Avogadro constant, and M_W is the molecular weight. The calculated values of the v_m, B, and θ_D of SnO are summarized in Table 2, together with the literature data for SnO₂. It was confirmed that v_m, B, and θ_D of SnO are smaller than those of SnO₂.

The Grüneisen parameter (γ) which is related to the anharmonicity can be calculated as follows:⁴⁶⁾   

\begin{equation} \gamma = \frac{\alpha_{\text{V}}BV_{\text{M}}}{C_{\text{V}}}, \end{equation} (10)

where V_M is the molar volume, C_V is the specific heat at constant volume which can be estimated by using Dulong-Petit law (C_V = 3nR where R is the gas constant), and α_V is the volumetric thermal expansion coefficient evaluated by the HT-XRD method. Lattice anharmonicity is closely related with α_V, viz., large α_V means large anharmonicity. Therefore, the Grüneisen parameter (γ) can provide an indication of degree of anharmonicity as γ is proportional to α_V according to eq. (10). As summarized in Table 2, the calculated γ values were 1.70 for SnO and 1.27 for SnO₂. SnO with lone pair electrons clearly exhibits high γ value than SnO₂ with no lone pair electrons.

In order to find which parameter contributes to the difference in the κ_lat of SnO and SnO₂, we analyzed the κ_lat based on the following equation:⁴⁷⁾   

\begin{equation} \kappa_{\text{lat}} = A\frac{M\delta\theta_{\text{D}}{}^{3}}{N^{2/3}\gamma^{2}}\frac{1}{T}, \end{equation} (11)

where A is a constant independent of the material, M is the average molar mass per atom, δ³ is the average volume per atom, N is the number of atoms in the unit cell. At high temperatures above θ_D where Umklapp phonon-phonon scattering is dominant, the κ_lat can be calculated by eq. (11). Furthermore, the Grüneisen parameter (γ) is related to the relaxation time for Umklapp phonon-phonon scattering (τ_U), i.e., τ_U is proportional to γ⁻².⁴⁸⁾ Figure 10 shows the ratios of the parameters (SnO/SnO₂) of individual contributions for determination of κ_lat, where the ratio below 1 means that the parameter contributes to the lower κ_lat of SnO than that of SnO₂. As can be seen in Fig. 10, the SnO/SnO₂ ratios for γ and θ_D are less than 1, meaning that these parameters are dominant contributors to lowering κ_lat of SnO. Since the ratio for θ_D is smaller than that for γ, it can be said that the θ_D is more predominant than γ to reduce the κ_lat of SnO.

Fig. 10

Ratios of the parameters (SnO/SnO₂) of individual contributions for determination of κ_lat.

Figure 11 shows the temperature dependences of (a) σ, (b) S, (c) power factor S²σ, and (d) zT of the bulk sample of SnO (sample #2). As can be seen in Fig. 11, these parameters increase with increasing temperature. The S values are positive, indicating that the SnO sample is p-type. The maximum values of the power factor and zT are 3.76 µWm⁻¹K⁻² @ 523 K and 0.00141 @ 573 K, respectively. Very recently, Miller et al. reported the experimental results on the TE properties of polycrystalline SnO bulk samples.⁴⁹⁾ These data are summarized in Table 3, together with the data of our sample. The n_H of our sample (2.8 × 10¹⁸ cm⁻³ at room temperature) is clearly larger than the reported value (1.0 × 10¹⁶ cm⁻³ at room temperature). The difference in the n_H between our sample and the sample reported in Ref. 49 would be caused by the difference in the amount of impurities existed in the samples. Miller et al. synthesized the SnO powders through a precipitation reaction method using SnCl₂ as a precursor, where high-purity SnO powders compared with the commercial grade could be obtained. Here, we used commercially-available SnO powders (99.9%), and impurities existed in the powders would act as hole dopants and increase the n_H. Due to this large n_H, our sample shows small S and low μ_H compared with those reported in the literature. We will try to enhance the TE properties of SnO through the carrier concentration optimization. The results will be reported in near future.

Fig. 11

Temperature dependences of (a) electrical conductivity σ, (b) Seebeck coefficient S, (c) power factor S²σ, and (d) dimensionless figure of merit zT of the bulk sample of SnO (sample #2).

Table 3 Results of the Hall effect measurements and the Seebeck coefficient of the bulk sample of SnO (sample #2). The literature data for SnO⁴⁹ are shown for comparison. All the data are obtained at room temperature.

4. Conclusion

It is considered that SnO can be a good TE material because it has lone pair electrons and thereby has large anharmonicity and low κ_lat. However, a fine bulk sample of SnO has been never synthesized because SnO is easily decomposed into SnO₂ and Sn even at 700 K. In the present study, we succeeded in synthetizing high-density pure bulk SnO samples by using high-pressure and low-temperature SPS (500 MPa, 598 K). It was revealed that SnO has much smaller κ_lat values (1.44 Wm⁻¹K⁻¹ at 573 K) than SnO₂ which has no lone pair electrons. It was confirmed that not only large Grüneisen parameter (γ = 1.70) but also low Debye temperature (θ_D = 335 K) contributes to this low κ_lat of SnO. Furthermore, we reported the TE properties of SnO, where the maximum values of the power factor and zT are 3.76 µWm⁻¹K⁻² @ 523 K and 0.00141 @ 573 K, respectively. The TE properties can be enhanced through the carrier concentration optimization.

Acknowledgments

This work was supported in part by JST, PRESTO Grant Number JPMJPR15R1.

REFERENCES

• 1)   G.J.  Snyder and  E.S.  Toberer: Nat. Mater. 7 (2008) 105–114.
• 2)   B.C.  Sales,  D.  Mandrus and  R.K.  Williams: Science 272 (1996) 1325–1328.
• 3)   X.  Shi,  H.  Kong,  C.-P.  Li,  C.  Uher,  J.  Yang,  J.R.  Salvador,  H.  Wang,  L.  Chen and  W.  Zhang: Appl. Phys. Lett. 92 (2008) 182101.
• 4)   X.  Shi,  J.  Yang,  J.R.  Salvador,  M.  Chi,  J.Y.  Cho,  H.  Wang,  S.  Bai,  J.  Yang,  W.  Zhang and  L.  Chen: J. Am. Chem. Soc. 133 (2011) 7837–7846.
• 5)   A.I.  Hochbaum,  R.  Chen,  R.D.  Delgado,  W.  Liang,  E.C.  Garnett,  M.  Najarian,  A.  Majumdar and  P.  Yang: Nature 451 (2008) 163–167.
• 6)   A.I.  Boukai,  Y.  Bunimovich,  J.  Tahir-Kheli,  J.-K.  Yu,  W.A.  Goddard, III and  J.R.  Heath: Nature 451 (2008) 168–171.
• 7)   S.K.  Bux,  R.G.  Blair,  P.K.  Gogna,  H.  Lee,  G.  Chen,  M.S.  Dresselhaus,  R.B.  Kaner and  J.P.  Fleurial: Adv. Funct. Mater. 19 (2009) 2445–2452.
• 8)   P.F.P.  Poudeu,  J.  D’Angelo,  A.D.  Downey,  J.L.  Short,  T.P.  Hogan and  M.G.  Kanatzidis: Angew. Chem. Int. Ed. 45 (2006) 3835–3839.
• 9)   T.  Mokari,  M.  Zhang and  P.  Yang: J. Am. Chem. Soc. 129 (2007) 9864–9865.
• 10)   Y.  Pei,  N.A.  Heinz,  A.  LaLonde and  G.J.  Snyder: Energy Environ. Sci. 4 (2011) 3640–3645.
• 11)   M.  Christensen,  A.B.  Abrahamsen,  N.B.  Christensen,  F.  Juranyi,  N.H.  Andersen,  K.  Lefmann,  J.  Andreasson,  C.R.H.  Bahl and  B.B.  Iversen: Nat. Mater. 7 (2008) 811–815.
• 12)   A.J.  Minnich,  M.S.  Dresselhaus,  Z.F.  Ren and  G.  Chen: Energy Environ. Sci. 2 (2009) 466–479.
• 13)   B.  Qiu,  Z.  Tian,  A.  Vallabhaneni,  B.  Liao,  J.M.  Mendoza,  O.D.  Restrepo,  X.  Ruan and  G.  Chen: EPL (Europhysics Lett.) 109 (2015) 57006.
• 14)   C.  Xiao,  X.  Qin,  J.  Zhang,  R.  An,  J.  Xu,  K.  Li,  B.  Cao,  J.  Yang,  B.  Ye and  Y.  Xie: J. Am. Chem. Soc. 134 (2012) 18460–18466.
• 15)   S.N.  Guin,  A.  Chatterjee,  D.S.  Negi,  R.  Datta and  K.  Biswas: Energy Environ. Sci. 6 (2013) 2603–2608.
• 16)   S.N.  Guin and  K.  Biswas: Chem. Mater. 25 (2013) 3225–3231.
• 17)   Y.-L.  Pei,  H.  Wu,  J.  Sui,  J.  Li,  D.  Berardan,  C.  Barreteau,  L.  Pan,  N.  Dragoe,  W.-S.  Liu,  J.  He and  L.-D.  Zhao: Energy Environ. Sci. 6 (2013) 1750–1755.
• 18)   D.T.  Morelli,  V.  Jovovic and  J.P.  Heremans: Phys. Rev. Lett. 101 (2008) 035901.
• 19)   Y.  Zhang,  E.  Skoug,  J.  Cain,  V.  Ozolins,  D.  Morelli and  C.  Wolverton: Phys. Rev. B 85 (2012) 1–6.
• 20)   M.D.  Nielsen,  V.  Ozolins and  J.P.  Heremans: Energy Environ. Sci. 6 (2013) 570–578.
• 21)   L.-D.  Zhao,  S.-H.  Lo,  Y.  Zhang,  H.  Sun,  G.  Tan,  C.  Uher,  C.  Wolverton,  V.P.  Dravid and  M.G.  Kanatzidis: Nature 508 (2014) 373–377.
• 22)   A.  Walsh and  G.W.  Watson: J. Phys. Chem. B 109 (2005) 18868–18875.
• 23)  W.M. Haynes: CRC Handbook of Chemistry and Physics, 91st Edition (CRC Press, Boca Raton, Fla., 2010) p. 4.
• 24)  T. Nishiyama: RARE METAL RESOURCE (Maruzen Publishing Co. Ltd., Japan, 2009) p. 22.
• 25)   F.  Lawson: Nature 215 (1967) 955–956.
• 26)   W.J.  Moore, Jr and  L.  Pauling: J. Am. Chem. Soc. 63 (1941) 1392–1394.
• 27)   W.H.  Baur: Acta Crystallogr. 9 (1956) 515–520.
• 28)   R.  Seshadri: J. Chem. Sci. 113 (2001) 487–496.
• 29)  I. Barin, O. Knacke and O. Kubaschewski: Thermochemical Properties of Inorganic Substances (Springer-Verlag Berlin Heidelberg, 1977) p. 702.
• 30)  I. Barin, F. Sauert, E. S.-Rhonhof and W.S. Sheng: Thermochemical Data of Pure Substances (Weinheim, New York, VCH, 1989) p. 1404.
• 31)   J.  Pannetier and  G.  Denes: Acta Crystallogr. Sec. B 36 (1980) 2763–2765.
• 32)   J.  Haines and  J.M.  Léger: Phys. Rev. B 55 (1997) 11144–11154.
• 33)   H.  Swanson and  E.  Tatge: Natl. Bur. Stand. Circ. 539 (1953) 24.
• 34)   H.  Giefers,  F.  Porsch and  G.  Wortmann: Solid State Ion. 176 (2005) 1327–1332.
• 35)   J.B.  Varley,  A.  Schleife,  A.  Janotti,  C.G.  Van De Walle,  J.B.  Varley,  A.  Schleife,  A.  Janotti and  C.G.  Van De Walle: 103 (2013) 82118.
• 36)   H.  Wang,  K.  Dou,  W.Y.  Teoh,  Y.  Zhan,  T.F.  Hung,  F.  Zhang,  J.  Xu,  R.  Zhang and  A.L.  Rogach: Adv. Funct. Mater. 23 (2013) 4847–4853.
• 37)   D.S.  Smith,  A.  Alzina,  J.  Bourret,  B.  Nait-Ali,  F.  Pennec,  N.  Tessier-Doyen,  K.  Otsu,  H.  Matsubara,  P.  Elser and  U.T.  Gonzenbach: J. Mater. Res. 28 (2013) 2260–2272.
• 38)   A.  Biancheria: Trans. Am. Nucl. Soc. 9 (1966) 15.
• 39)   W.  Zhou and  N.  Umezawa: Phys. Chem. Chem. Phys. 17 (2015) 17816–17820.
• 40)   K.  Inoue,  S.  Mori and  A.  Yamaguchi: J. Ceram. Soc. Jpn. 111 (2003) 348–351.
• 41)   N.  Soga and  E.  Schreiber: J. Am. Ceram. Soc. 51 (1968) 465–467.
• 42)   D.H.  Chung: Phys. Earth Planet. Inter. 8 (1974) 113–120.
• 43)   P.  Turkes,  C.  Pluntke and  R.  Helbig: J. Phys. C 13 (1980) 4941–4951.
• 44)   O.L.  Anderson: J. Phys. Chem. Solids 24 (1963) 909–917.
• 45)   F.  Birch: J. Geophys. Res. 66 (1961) 2199–2224.
• 46)   S.  Ganesan: Philos. Mag. 7 (1962) 197–205.
• 47)   G.A.  Slack: Solid State Phys. Acad. Press. 34 (1979) 1–71.
• 48)   E.S.  Toberer,  A.  Zevalkink and  G.J.  Snyder: J. Mater. Chem. 21 (2011) 15843–15852.
• 49)   S.A.  Miller,  P.  Gorai,  U.  Aydemir,  T.O.  Mason,  V.  Stevanović,  E.S.  Toberer and  G.J.  Snyder: J. Mater. Chem. C 5 (2017) 8854–8861.