Research Article | Open Access

Yifan Zhu, Yi Xia, Yancheng Wang, Ye Sheng, Jiong Yang, Chenguang Fu, Airan Li, Tiejun Zhu, Jun Luo, Christopher Wolverton, G. Jeffrey Snyder, Jianjun Liu, Wenqing Zhang, "Violation of the *T*^{−1} Relationship in the Lattice Thermal Conductivity of Mg_{3}Sb_{2} with Locally Asymmetric Vibrations", *Research*, vol. 2020, Article ID 4589786, 10 pages, 2020. https://doi.org/10.34133/2020/4589786

# Violation of the *T*^{−1} Relationship in the Lattice Thermal Conductivity of Mg_{3}Sb_{2} with Locally Asymmetric Vibrations

#### Abstract

Most crystalline materials follow the guidelines of temperature-dependent lattice thermal conductivity () at elevated temperatures. Here, we observe a weak temperature dependence of in Mg_{3}Sb_{2}, from theory and from measurements, based on a comprehensive study combining *ab initio* molecular dynamics calculations and experimental measurements on single crystal Mg_{3}Sb_{2}. These results can be understood in terms of the so-called “phonon renormalization” effects due to the strong temperature dependence of the interatomic force constants (IFCs). The increasing temperature leads to the frequency upshifting for those low-frequency phonons dominating heat transport, and more importantly, the phonon-phonon interactions are weakened. In-depth analysis reveals that the phenomenon is closely related to the temperature-induced asymmetric movements of Mg atoms within MgSb_{4} tetrahedron. With increasing temperature, these Mg atoms tend to locate at the areas with relatively low force in the force profile, leading to reduced effective 3^{rd}-order IFCs. The locally asymmetrical atomic movements at elevated temperatures can be further treated as an indicator of temperature-induced variations of IFCs and thus relatively strong phonon renormalization. The present work sheds light on the fundamental origins of anomalous temperature dependence of in thermoelectrics.

#### 1. Introduction

The study of thermal conductivity has been driven by the increasing concern on both intriguing physical phenomena and powerful technical applications [1]. The thermal conductivity in most crystalline materials consists of the lattice thermal conductivity () and the electronic component (), which is related to the electrical conductivity () through the Wiedemann-Franz law, , where is the Lorenz number and is the temperature. High thermal conductivity materials such as diamond and silicon are investigated in the area of thermal management of electronics. Low thermal conductivity materials like Zintl phases [2, 3], skutterudites [4, 5], half-Heuslers [6, 7], and materials with chemical bond hierarchy [8–10] are widely used in high-performance thermoelectric energy conversion.

Phonons are quantized collective atomic vibrations. The heat capacity , the phonon velocity , and the relaxation time make contributions for [1] in the phonon gas model according to
where is the number of atoms and is the volume of the cell. The phonon relaxation time covers the influences from all different scattering mechanisms, including intrinsic mechanisms phonon-phonon interactions and electron-phonon interactions, as well as extrinsic ones, e.g., point defect and grain boundary. The anharmonic three-phonon scattering rate [11] can be expressed as
where is the occupation number of the phonon mode under the Bose-Einstein distribution. is the conservation of the momentum. The strength of three-phonon interactions is obtained by
where represents (3^{rd}-order) anharmonic interatomic force constant (IFC) tensor, denote the indexes of atoms, are the Cartesian components, and is the eigenvector for atom at the phonon mode. Above the Debye temperature, all phonon modes are excited, and the Bose-Einstein distribution is proportional to the temperature, so that the lattice thermal conductivity in crystalline materials exhibits a relationship at high temperatures.

Another precondition of the relationship of is the temperature-independent harmonic and anharmonic IFCs. It is generally applicable in high materials [12] at high temperatures. The well-known Slack model is just based on the fixed IFCs and high temperature limit to obtain the relationship. Some other studies, using phonon dispersions to predict of solids, also assume constant IFCs in their works [13, 14]. However, for some low thermal conductivity crystalline materials, the typical relationship, even at high temperatures without other extrinsic scattering mechanisms, is often violated. For these materials, the conventional first-principles theory for thermal conductivities usually fails since it is based on the lowest-order perturbative treatment and insufficient for strongly anharmonic materials. There are several possible effects existing in these materials, which change the temperature dependence of s, including the alteration of phonon scattering phase space [15], fourth-order phonon scattering [16], and off-diagonal terms in heat current operator [17]. Revealing the fundamental origin of the deviation on relationship in each case compound will advance the understanding of the effect of high-order IFCs on thermal transport.

The compound interested in this work is Mg_{3}Sb_{2}, one of the binary layered compounds. It has attracted increasing attention as a novel n-type thermoelectric material near room temperature, due to an inherent conduction band minimum with six conducting carrier pockets and a tunable bandgap using the alloy effect reported by several independent groups [18–22]. The excess Mg, compensating Mg vacancies [23], is essential to achieve n-type Mg_{3}Sb_{2}-based materials, with a peak up to 1.6 [24]. On the other hand, low of Mg_{3}Sb_{2} is also one of the significant factors for it to become applicable in thermoelectrics. Shearing of adjacent layers [25], large mode Grüneisen parameters [25], weak chemical bonding [21], and weak anharmonic potential wells [21] have been proposed to prove strong anharmonic effects in Mg_{3}Sb_{2}, which lead to low .

As shown in Fig. S1 [26–28], the experimental s of pristine Mg_{3}Sb_{2} clearly have a weaker temperature dependence, usually demonstrating . The low carrier concentration of pristine Mg_{3}Sb_{2} prohibits a strong electron-phonon scattering on , which is the origin of weaker temperature dependence of s in other materials [29–31]. As will be shown later in this work, single crystalline Mg_{3}Sb_{2} shows a similar behavior as of the polycrystalline sample for the weaker temperature dependence of , which is the research focus of this work. Our theoretical of Mg_{3}Sb_{2} shows a weaker temperature dependence as of along both and directions. This result is verified by our experimental data on single crystal Mg_{3}Sb_{2}. Detailed analysis reveals that some of the phonon vibrational modes at a low-frequency range become higher in frequency with increasing temperatures, due to the influence of the higher-order (particularly, 4^{th}-order) IFCs and the renormalization of the harmonic phonons. More importantly, we reveal that the effective 3^{rd}-order IFCs become weaker with increasing temperatures, which is the main reason for the weak temperature dependence of in Mg_{3}Sb_{2}. The surprising variation of 3^{rd}-order IFCs is due to the asymmetric displacements of intralayer Mg atoms, which play a decisive role in determining the lattice anharmonicity. The present work advances the understanding of the temperature-dependent beyond the classic three-phonon diagram in thermoelectric materials.

#### 2. Result and Discussion

The pristine Mg_{3}Sb_{2}, with only five atoms per primitive cell, is described as the Zintl-type structure, combining the ionic Mg^{2+} layer (with the Mg atoms labeled as Mg1) and the covalent (although the nature of the bonding is debatable [21]) [Mg_{2}Sb_{2}]^{2-} layer (with the Mg atoms labeled as Mg2) as shown in Figure 1(a). Mg1 is on the octahedron site, while Mg2 is on the tetrahedron site. There are two main features in the temperature-dependent calculations for Mg_{3}Sb_{2}. One is that we consider the thermal expansion (Fig. S3 (a), discussed later), the other is that the effective 2^{nd}-order and 3^{rd}-order IFCs are extracted from *ab initio* molecular dynamics (AIMD) under the corresponding volume at finite temperatures, and s can be obtained (Figure 1(b)) using the temperature-dependent effective potential (TDEP) method [32, 33]. The calculations consider all effects from finite temperatures, so we call that finite temperature method (FTM). The calculated has values of 2.05 W m^{-1} K^{-1} and 2.50 W m^{-1} K^{-1} along and directions at 300 K, respectively. The temperature dependence of follows , along both directions. In order to explore whether the off-diagonal contribution would affect the temperature dependence, we employed a unified formalism recently developed by Simoncelli et al. [17]. As shown in Figure 1(b), we find that magnitude of the off-diagonal thermal conductivity is very small over the entire temperature range and therefore has a negligible impact on the temperature dependence of in Mg_{3}Sb_{2}. Interestingly, in our FTM results, along the direction is around 20% lower than that along the direction. This is due to the differences in the group velocities (Fig. S2 (a)), i.e., the average of all modes m/s and m/s at 300 K. Previous reports in Reference [21] mentioned the nearly isotropic heat conduction of Mg_{3}Sb_{2}, with a room temperature value of approximately 1.1 W m^{-1} K^{-1}. Their isotropic properties come from the almost identical group velocities in different axes, m/s and m/s (Fig. S2 (b)). On the other hand, the lowered s in the frozen phonon method [21] are caused by the overestimated Grüneisen parameters, which will be discussed later.

To establish a convincing comparison between theory and experiment, we synthesized the Mg_{3}Sb_{2} single crystal sample (the inset image, sheet) to eliminate the impacts of defects or grain boundaries on the thermal conductivity [34] as far as possible. of the Mg_{3}Sb_{2} single crystal above 300 K is measured and also shown in Figure 1(b). The experimental heating and cooling curves are on top of each other, indicating that the single crystal Mg_{3}Sb_{2} is thermally stable below 773 K, the highest temperature experimentally performed. Notably, from our previous study [35], Mg_{3}Sb_{2} single crystals grown from the Sb flux method exhibit very high electrical resistivity (several *Ω*m at room temperature but can be made conducting with good by annealing in Mg-rich environment [36]), and thus, the electronic component of the thermal conductivity is negligible. is thus almost contributed by the lattice component, i.e., . In the following part, we simply use instead of facilitating the comparative discussion with the theoretically calculated . The experimental for Mg_{3}Sb_{2} single crystal is measured along the direction, and the value is around 0.3 W m^{-1} K^{-1} lower than our theoretical value over the entire temperature range. The small deviation might come from possible measurement uncertainty owing to the small size of the single crystal and instrument accuracy. It is noted that our experimental of Mg_{3}Sb_{2} is also weakly temperature-dependent, following approximately relationship, consistent with our calculations.

Our FTM considers the accurate volume expansion at finite temperatures. The quasiharmonic approximation [37] with two solutions of IFCs is presented in Fig. S3 (a). One is the frozen phonon method; the other is FTM with temperature 100 K. It is found that the slope of the volume-temperature curve from FTM is more consistent with the high-temperature X-ray diffraction results, especially at high temperatures, despite the general overestimation of lattice constants due to Perdew-Burke-Ernzerhof (PBE) functional [38]. The corresponding experimental lattice parameters of Mg_{3}Sb_{2} can be found in Fig. S3 (b). The calculated volumetric thermal expansion coefficient for Mg_{3}Sb_{2} is K^{-1} ( K^{-1}) at 300 K (700 K), which is close to Bi_{2}Te_{3} ( K^{-1}, 200-300 K [39]), slightly higher than PbTe ( K^{-1}, 300 K [40]), and distinctly higher than those high thermal conductivity materials, like diamond ( K^{-1}, 300 K [41]) and silicon ( K^{-1}, 300 K [42]). The relatively large thermal expansion coefficient indicates the considerable anharmonicity of Mg_{3}Sb_{2} like other low thermoelectric materials, such as Bi_{2}Te_{3} [43] and PbTe [44].

Figure 1(c) displays the frequency-dependent at 300 K and 700 K. Note that s in these two plots are the averaged ones along the three Cartesian axes. We see that phonons which dominate the s generally have frequencies below 3.5 THz and mostly are between 0.5 and 2.5 THz (Figure 1(c)) at both temperatures. Therefore, the following discussion will focus on the low-frequency region. The cumulative s in Fig. S4 for Mg_{3}Sb_{2} as a function of the mean free path (MFP) show that phonon modes in Mg_{3}Sb_{2} have a maximum MFP of around 200 nm at both temperatures, and half of the s are from those modes with MFP less than 10 nm. These are useful for the design of nanostructures in Mg_{3}Sb_{2} in order to further suppress .

Since the possible influence from electron-phonon interaction or off-diagonal contribution has been excluded, here, we focus on the variations of 2^{nd}-order and 3^{rd}-order IFCs and related physics to explore the origin of the weak temperature-dependent . Due to the fact that the low-frequency phonons dominate of Mg_{3}Sb_{2}, temperature-dependent phonon dispersions of Mg_{3}Sb_{2} in the low-frequency region are shown in Figure 2(a). (The whole phonon dispersions at 300 K and 700 K are shown in Fig. S5.) It is interesting to note that the phonons in the low-frequency region show diverse temperature dependence. Specifically, the transverse acoustic phonons become harder as temperature increases, especially at the Brillouin zone boundary , , and points. The temperature dependence along is contrary to the result in Fig. S6, similar to the result in Reference [25], which only considered the lattice thermal expansion based on the frozen phonon method. The low-lying transverse acoustic branches generally become harder with increasing temperature. For example, the zone-center speed of sound for Mg_{3}Sb_{2} along to directions at 300 K (700 K) is 1734 m/s (1909 m/s).

To provide an intuitive physical picture for the phonon hardening, we visualized the atomic vibrations for Mg_{3}Sb_{2} corresponding to the low-lying transverse acoustic phonon modes, i.e., , , and points, in Gif S1 (M.gif), Gif S2 (A.gif), and Gif S3 (L.gif). These animations demonstrate that the atomic motions within the [Mg_{2}Sb_{2}]^{2-} layer dominate the vibrational modes at these points. And since these points are at the zone boundary, Mg2 and Sb in the [Mg_{2}Sb_{2}]^{2-} layer show the largest phase difference, resulting in the “head-to-head” motions. When temperature increases, the head-to-head motion impedes the vibrations of atoms in the [Mg_{2}Sb_{2}]^{2-} layer towards the neighboring atoms, thus making the potential energy (Fig. S7) increase more rapidly than the harmonic approximation at , , and points.

The total phonon density of states from K to 700 K is shown in Figure 2(b). Due to the significant phonon hardening effects with temperature from the low-lying transverse acoustic phonons at Brillouin zone boundary points around 1 THz (Figure 2(a)), the phonon density of states around 1 THz move to a higher frequency with the increased temperature. We verified this conjecture by explicitly computing renormalized phonons using the self-consistent phonon theory that accounts for the first-order correction from quartic IFCs [45] as shown in Fig. S8. The hardening of the low-lying transverse acoustic phonon modes at the , , and points is reproduced, unambiguously confirming the essential role of quartic IFCs in the phonon renormalization.

The above analysis mainly focuses on the variations of the 2^{nd}-order IFC induced phonon dispersions. On the other hand, by examining the key quantities entering the evaluation of thermal conductivity, namely, the phonon mode-wise heat capacity, group velocity (related to 2^{nd}-order IFCs), and phonon relaxation time (inverse phonon scattering rate, related to 2^{nd}-order IFCs on the scattering phase space and 3^{rd}-order IFCs on the anharmonicity), we use controlled comparisons to clarify the influence of temperature-dependent IFCs on at 300 K, as shown in Table S1. By separately substituting the 2^{nd}-order and 3^{rd}-order IFCs, we find that 2^{nd}-order IFC related group velocity and scattering phase space have limited impact on , while 3^{rd}-order IFC related anharmonicity plays a significant role. These results are distinctly different from previous studies for other low materials, where only 2^{nd}-order IFC related quantities matter [15, 46]. Similarly, we compare the phonon scattering rates () in Mg_{3}Sb_{2} at K and 700 K, respectively, as shown in Figure 3(a). As expected, the scattering rates at 700 K are higher than those at 300 K. It is because higher temperature results in a larger phonon number (Eq. (2)), thus larger scattering magnitude. However, the scattering rates at 700 K are only slightly larger than those at 300 K, which indicates that the IFCs entering the evaluation of scattering rates might have strong temperature dependence. We examine this conjecture by comparing the scattering rates calculated at 300 K but using temperature-dependent IFCs at 300 K and 700 K, respectively, as shown in Figure 3(b). The significantly smaller scattering rates using IFCs obtained at 700 K clearly reveal the strong temperature dependence of effective IFCs. We, respectively, replace the 2^{nd}-order IFCs and the 3^{rd}-order IFCs at 700 K with those obtained at 300 K and compare the resulting scattering rates. We find that when the 3^{rd}-order IFCs are kept the same while the 2^{nd}-order IFCs are from 300 K and 700 K, the resulting scattering rates are largely similar (see Fig. S9). On the other hand, when the 2^{rd}-order IFCs are kept the same while the 3^{nd}-order IFCs are from 300 K and 700 K, the resulting scattering rates show significantly reduced magnitude for those using the 3^{rd}-order IFCs at 700 K (Figure 3(c)). These controlled comparisons confirm that (i) the strong temperature dependence of the 3^{rd}-order IFCs plays a decisive role in the unusually weak decay of scattering rates as a function of temperature and (ii) the anharmonicity is reduced at elevated temperatures.

The strength of the anharmonicity at various temperatures can also be estimated from the mode Grüneisen parameters, which are obtained as the logarithmic derivative of the phonon frequencies with respect to the volume, as shown in Figures 3(d) and 3(e). The acoustic phonons show relatively large absolute values of mode-wise Grüneisen parameters (~1.4), while the values of the optical phonons are relatively lower (~0.9), based on the results at 300 K. Specifically, the first phonon modes at and points show negative values, -1.55 and -2.47, respectively, while at the point, the value is positive as of 1.97. The negative Grüneisen parameters at and points mean that the phonon frequency at these points will be increased with lattice expansion and partly contributes to the phonon hardening effects with increasing temperature (volume), as shown in Figure 2(a). On the other hand, the positive Grüneisen parameter at the point cancels some of the phonon renormalization effect, making the frequency change at point the smallest among the three (, , and ) points. Furthermore, the Grüneisen parameters from the frozen phonon method (2.0 for acoustic phonons) are larger than those from FTM. This is the reason that previous simulations based on the frozen phonon method underestimated s, approximately 1.1 W m^{-1} K^{-1} at 300 K for Mg_{3}Sb_{2}. [21] As for the temperature dependence, the Grüneisen parameters at 300 K are higher than those at 700 K (Fig. S10), consistent with the weakened anharmonic phonon scattering in Figure 3(c). In light of this discovery, it is natural to ask what is the physical origin of the strongly temperature-dependent 3^{rd}-order IFCs.

To understand the temperature dependence of the 3^{rd}-order IFCs, we compared the 3^{rd}-order IFCs for all the triplets at 300 K and 700 K as shown in Figure 4(a). All the triplets listed are along the directions with the largest 3^{rd}-order IFCs at 300 K for Mg_{3}Sb_{2}. Among the various triplets, the ones that involve onsite interactions (e.g., (Mg2)^{3} and (Sb)^{3}) or interactions between the neighboring atoms (e.g., (Mg2)^{2}Sb and Mg2(Sb)^{2}) along the direction have the largest 3^{rd}-order IFCs at room temperature. More importantly, these triplets with large 3^{rd}-order IFCs also show strong temperature dependence. Specifically, we find that the 3^{rd}-order IFC of the onsite interactions of Mg2 atoms along the direction (denoted as ), which show the largest values among all the triplets at 300 K, decreases dramatically from 7.56 eV/Å^{3} at 300 K to 3.58 eV/Å^{3} at 700 K. In the following, we will focus on this set of IFCs to explore the microscopic origin of the reduced IFCs at elevated temperatures.

The MD trajectories at 300 K and 700 K are shown in Figure 4(b). Based on the trajectories, it can be found that the average radii of the atomic displacements at 700 K are much larger than those at 300 K. An unusual feature in Figure 4(b) is that the atomic displacements of all atoms at 300 K are approximately isotropic, while the displacements of Mg2 atoms become strongly anisotropic at 700 K, i.e., a cone-type trajectory. The pointy edge is at the opposite of the neighboring Sb along the direction (assumed to be the positive position w.r.t. Mg2), as shown in Figure 4(b). Quantitatively, according to the probability density as a function of Mg2 displacement along the direction (Fig. S11) at 700 K, there are 18% more points at the negative displacement side, which has a large tail down to over -1 Å. The same number is only 4% at 300 K. This type of Mg2 trajectory at 700 K is due to the restrictions of vibrations from the neighboring Sb atoms, and the IFCs of Mg2 are altered along with the restrictions. The energy and force profiles of Mg2 atom w.r.t. the displacement along the direction, shown in Figures 4(c) and 4(d), help to understand the alteration. As displayed in Figure 4(c), it is found that the energy of the system increases rapidly/slowly when Mg2 atom is displaced along the positive/negative direction. It is thus the asymmetric energy profile along the direction that leads to the asymmetric displacements of Mg2 atoms, and such asymmetry enhances with increasing temperature, consistent with the Boltzmann distribution. The asymmetry is also reflected in the restoring forces on Mg2 atoms, as shown in Figure 4(d). Mg2 atoms at the lower-energy side have much lower forces, and the lowered forces result in lowered 3^{rd}-order IFC of onsite since the onsite 3^{rd}-order IFC is actually the curvature of the force profile. And since the lower-energy side has more sampling at elevated temperatures, i.e., 18% more at 700 K, the 3^{rd}-order IFC for the onsite is thus lowered by the asymmetry of the force profile.

In fact, the force profile at finite temperatures is much complicated than those presented in Figure 4(d). Fig. S12 (a) shows the force profile along the direction for the same Mg2 atom at 700 K, extracted from the AIMD simulation. The general trend is like Figure 4(d), i.e., positive displacements with larger forces and negative displacements with lowered forces. It is interesting that at some very negative displacements for Mg2 along the direction, the force is close to zero. Fig. S12 (b) shows the local tetrahedron of the Mg2, with the displacement -0.93 Å and the force -0.01 eV/Å. The Mg2 locates at the center of the bottom 3 Sb atoms, and the bond with the top Sb is broken. The close-to-zero force at large negative displacements further contributes to the low .

The temperature-dependent IFCs, due to the asymmetric displacements of Mg2 atoms, provide useful guidance of exploring materials with other than relationship. The local structure of Mg2 determines an asymmetric potential energy profile, as shown in Figure 4(c); more importantly, the displacements of Mg2 atoms at finite temperatures, such as 700 K, is large enough to have uneven distributions at different sides of the atomic equilibria and finally induce the temperature-dependent IFCs. Inspired by this example, the atomic trajectories from AIMD or thermal ellipsoids can be adopted as indicators for the temperature-dependent IFCs. Strongly asymmetric trajectories or ellipsoids at reachable temperatures will cause the same uneven distributions as the Mg2 in Mg_{3}Sb_{2} and thus change the IFCs and dependence. For example, some fillers in type-I clathrates, such as Ba_{8}Ga_{16}Ge_{30}, Ba atoms show different atomic displacements along different axes at room temperature [47–49], and therefore, s show strong deviations from , sometimes even glass-like behavior. If one compound has similar asymmetric movement, it may have the temperature dependence violating on the lattice thermal transport.

#### 3. Conclusions

Layered structure Mg_{3}Sb_{2}, as a novel thermoelectric material, has achieved increasing interest due to the inherently low . Here, we find that Mg_{3}Sb_{2} has weak temperature-dependent through both simulations and experiments. The weak temperature dependence is due to the renormalizations of both 2^{nd}-order IFCs and 3^{rd}-order ones, and they can be traced back to the limitations of atomic vibrations at elevated temperatures. For 2^{nd}-order IFCs, the covalent [Mg_{2}Sb_{2}]^{2-} layer exhibits in-plane “head-to-head” motions at the Brillouin zone boundary , , and points for the first transverse acoustic phonons. The phenomenon causes that low-lying transverse acoustic phonons to become hardened. Meanwhile, the asymmetric displacements of Mg2 atoms due to the tetrahedrally bonding nature enhance with increasing temperature, and the effective 3^{rd}-order IFCs tend to have a smaller absolute value due to more sampling at the decreased curvature side in the force profile. Both the hardened low-lying phonons and weaker anharmonicity at elevated temperatures contribute to the weak temperature dependence of . And the asymmetric atomic vibrations at finite temperatures can be adopted to explore other materials with temperature-dependent IFCs and abnormal behavior.

#### 4. Methods

##### 4.1. Experiment Details

Single crystals of Mg_{3}Sb_{2} were grown by using a self-flux method with Sb as flux (details can be found in Reference [35]). Single crystallinity was checked using white-beam backscattering Laue X-ray diffraction at room temperature. Distinct diffraction spots were detected (Fig. S13), matching well with the theoretically simulated pattern based on the space group. The thermal conductivity of Mg_{3}Sb_{2} single crystals above room temperature was calculated by using the equation , where is the thermal diffusivity and measured using laser flash analysis instrument (LFA 457, Netzsch), is the theoretical density, is the specific heat, calculated using the recommended equation by Agne et al. [50]. Owing to the small size of the single crystal and instrument accuracy, an estimated uncertainty of is about 10%.

##### 4.2. Calculation Details

All first-principle calculations were carried out based on density function theory (DFT) and using the projector-augmented wave (PAW) method, as implemented in the Vienna *ab initio* Simulation Package (VASP) [51]. The thermal expansion properties were calculated with self-consistent quasiharmonic approximation (SC-QHA) [52] with IFCs (obtained by AIMD at 100 K) of Mg_{3}Sb_{2} under different volumes. Thus, temperature-dependent s for Mg_{3}Sb_{2} can be calculated under the corresponding volume at a given temperature. We performed MD simulation supercell, 320 atoms in total from 100 K to 700 K. The MD simulations for Mg_{3}Sb_{2} ran for 60 ps with a time of 2 fs, excluding 6 ps to reach the equilibrated state. The Brillouin zone integration was carried out on the point. The plane-wave energy cutoff and energy convergence criterion were set as 200 eV and 10^{-4} eV, respectively, for the MD simulations. The temperature-dependent harmonic and anharmonic IFCs were extracted by TDEP. We processed for heat transport-related properties also with the TDEP code [32, 33].

#### Conflicts of Interest

There are no conflicts to declare.

#### Authors’ Contributions

Yifan Zhu and Yi Xia contributed equally to this work. The initial idea was developed by Y. Zhu and J. Yang, and the asymmetric displacements of Mg2 atoms were proposed by Y. Xia. Y. Wang participated in its implementation. Y. Sheng performed the probability density of the MD trajectory. C. Fu synthesized single crystal Mg_{3}Sb_{2} and analyzed the experimental results. A. Li and T. Zhu provided polycrystalline powders of Mg_{3}Sb_{2}, and J. Luo measured the high-temperature X-ray diffraction for Mg_{3}Sb_{2}. J. Yang, C. Fu, and W. Zhang analyzed the reduced temperature dependence of . C. Wolverton, G. Snyder, and J. Liu provided advice in data analysis. All authors participated in the writing and reading of the paper.

#### Acknowledgments

This work was supported by the National Key Research and Development Program of China (Nos. 2018YFB0703600, 2017YFB0701600, and 2019YFA0704901), Natural Science Foundation of China (Grant Nos. 11674211, 51632005, and 51761135127), and the 111 Project D16002. W.Z. also acknowledges the support from the Guangdong Innovation Research Team Project (No. 2017ZT07C062), Guangdong Provincial Key Lab program (No. 2019B030301001), Shenzhen Municipal Key Lab program (ZDSYS20190902092905285), and Shenzhen Pengcheng-Scholarship Program. C.F. acknowledges funding support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer (392228380). Y.X. and C.W. acknowledge the financial support received from the U.S. Department of Commerce and National Institute of Standards and Technology as part of the Center for Hierarchical Materials Design (CHiMaD) under Grant No. 70NANB14H012. This research used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy (U.S. Department of Energy Contract No. DEAC02-05CH11231).

#### Supplementary Materials

Fig. S1: experimental thermal conductivity data of polycrystalline samples. Fig. S2: phonon group velocity along three directions for Mg_{3}Sb_{2} (a) at 300 K with the finite temperature method and (b) with the frozen phonon method. Fig. S3: (a) the temperature-dependent volumes of the primitive cell of Mg_{3}Sb_{2} obtained by the high-temperature X-ray diffraction method, compared to those obtained using frozen phonon and AIMD calculation methods. (b) Lattice parameters are measured by high-temperature X-ray diffraction using Mg_{3}Sb_{2} polycrystalline powders covering the range of 300 K to 700 K. Fig. S4: cumulative versus phonon mean free path for Mg_{3}Sb_{2} at 300 K and 700 K. Fig. S5: the whole phonon dispersions of Mg_{3}Sb_{2} at 300 K and 700 K. Fig. S6: calculated temperature-dependent phonon dispersions of Mg_{3}Sb_{2} from K to 700 K, only considering the lattice thermal expansion. Fig. S7: the potential energy surface corresponding to the low-lying transverse acoustic phonon modes at the Brillouin zone boundary point (a), point (b), and point (c). Fig. S8: calculated temperature-dependent phonon dispersions of Mg_{3}Sb_{2} from K to 700 K using the self-consistent phonon theory. Fig. S9: the scattering rate with 300 K IFCs is compared to 2^{nd}-order IFCs, substituted by 700 K, at K. Fig. S10: frequency-dependent mode Grüneisen parameters using the frozen phonon method, FTM at 300 K and 700 K. Fig. S11: probability density of MD trajectory deviating from equilibrium position at (a) 300 K and (b) 700 K along the direction. Fig. S12: (a) the force profile of Mg2 atom along the direction at 700 K, extracted from molecular dynamics simulation. (b) The local structure of Mg_{3}Sb_{2} with the displacement -0.93 Å along the direction and the force -0.01 eV/Å. Fig. S13: (a) Laue diffraction pattern of the studied Mg_{3}Sb_{2} single crystal and (b) the theoretically simulated pattern based on space group, matching well with the experimental data. Table S1: the calculated along and directions at 300 K with temperature-dependent IFCs.* (Supplementary Materials)*

#### References

- T. M. Tritt,
*Thermal Conductivity. Theory, Properties, and Applications*, Kluwer Academic/Plenum Publishers, New York, USA, 2004. View at: Publisher Site - J. Mao, H. Zhu, Z. Ding et al., “High thermoelectric cooling performance of n-type Mg
_{3}Bi_{2}-based materials,”*Science*, vol. 365, no. 6452, pp. 495–498, 2019. View at: Publisher Site | Google Scholar - J. Shuai, J. Mao, S. Song, Q. Zhang, G. Chen, and Z. Ren, “Recent progress and future challenges on thermoelectric Zintl materials,”
*Materials Today Physics*, vol. 1, pp. 74–95, 2017. View at: Publisher Site | Google Scholar - J. Chu, J. Huang, R. Liu et al., “Electrode interface optimization advances conversion efficiency and stability of thermoelectric devices,”
*Nature Communications*, vol. 11, no. 1, article 2723, 2020. View at: Publisher Site | Google Scholar - X. Shi, J. Yang, J. R. Salvador et al., “Multiple-filled skutterudites: high thermoelectric figure of merit through separately optimizing electrical and thermal transports,”
*Journal of the American Chemical Society*, vol. 133, no. 20, pp. 7837–7846, 2011. View at: Publisher Site | Google Scholar - H. Zhu, J. Mao, Y. Li et al., “Discovery of TaFeSb-based half-Heuslers with high thermoelectric performance,”
*Nature Communications*, vol. 10, no. 1, p. 270, 2019. View at: Publisher Site | Google Scholar - C. Fu, S. Bai, Y. Liu et al., “Realizing high figure of merit in heavy-band p-type half-Heusler thermoelectric materials,”
*Nature Communications*, vol. 6, no. 1, article 8144, 2015. View at: Publisher Site | Google Scholar - J. Yang, Y. Wang, H. Yang et al., “Thermal transport in thermoelectric materials with chemical bond hierarchy,”
*Journal of Physics: Condensed Matter*, vol. 31, no. 18, article 183002, 2019. View at: Publisher Site | Google Scholar - W. Qiu, L. Xi, P. Wei, X. Ke, J. Yang, and W. Zhang, “Part-crystalline part-liquid state and rattling-like thermal damping in materials with chemical-bond hierarchy,”
*Proceedings of the National Academy of Sciences of the United States of America*, vol. 111, no. 42, pp. 15031–15035, 2014. View at: Publisher Site | Google Scholar - H. Liu, X. Shi, F. Xu et al., “Copper ion liquid-like thermoelectrics,”
*Nature Materials*, vol. 11, no. 5, pp. 422–425, 2012. View at: Publisher Site | Google Scholar - R. A. Cowley, “Anharmonic crystals,”
*Reports on Progress in Physics*, vol. 31, no. 1, pp. 123–166, 1968. View at: Publisher Site | Google Scholar - D. T. Morelli and G. A. Slack,
*High Lattice Thermal Conductivity Solid*, Springer, 2006. View at: Publisher Site - G. J. Snyder, M. T. Agne, and R. Gurunathan, “Thermal conductivity of complex materials,”
*National Science Review*, vol. 6, no. 3, pp. 380-381, 2019. View at: Publisher Site | Google Scholar - Z. Chen, X. Zhang, S. Lin, L. Chen, and Y. Pei, “Rationalizing phonon dispersion for lattice thermal conductivity of solids,”
*National Science Review*, vol. 5, no. 6, pp. 888–894, 2018. View at: Publisher Site | Google Scholar - A. H. Romero, E. K. U. Gross, M. J. Verstraete, and O. Hellman, “Thermal conductivity in PbTe from first principles,”
*Physical Review B*, vol. 91, no. 21, article 214310, 2015. View at: Publisher Site | Google Scholar - Y. Xia, K. Pal, J. He, V. Ozoliņš, and C. Wolverton, “Particlelike phonon propagation dominates ultralow lattice thermal conductivity in Crystalline Tl
_{3}VSe_{4},”*Physical Review Letters*, vol. 124, no. 6, article 065901, 2020. View at: Publisher Site | Google Scholar - M. Simoncelli, N. Marzari, and F. Mauri, “Unified theory of thermal transport in crystals and glasses,”
*Nature Physics*, vol. 15, no. 8, pp. 809–813, 2019. View at: Publisher Site | Google Scholar - F. Zhang, C. Chen, H. Yao et al., “High-performance n-type Mg
_{3}Sb_{2}towards thermoelectric application near room temperature,”*Advanced Functional Materials*, vol. 30, no. 5, article 1906143, 2019. View at: Publisher Site | Google Scholar - X. Sun, X. Li, J. Yang et al., “Achieving band convergence by tuning the bonding ionicity in n-type Mg
_{3}Sb_{2},”*Journal of Computational Chemistry*, vol. 40, no. 18, pp. 1693–1700, 2019. View at: Publisher Site | Google Scholar - R. Shu, Y. Zhou, Q. Wang et al., “Mg
_{3+δ}SbBi_{x}_{2−x}Family: a promising substitute for the state-of-the-art n-type thermoelectric materials near room temperature,”*Advanced Functional Materials*, vol. 29, no. 4, article 1807235, 2019. View at: Publisher Site | Google Scholar - J. Zhang, L. Song, M. Sist, K. Tolborg, and B. B. Iversen, “Chemical bonding origin of the unexpected isotropic physical properties in thermoelectric Mg
_{3}Sb_{2}and related materials,”*Nature Communications*, vol. 9, no. 1, article 4716, 2018. View at: Publisher Site | Google Scholar - K. Imasato, S. D. Kang, S. Ohno, and G. J. Snyder, “Band engineering in Mg
_{3}Sb_{2}by alloying with Mg_{3}Bi_{2}for enhanced thermoelectric performance,”*Materials Horizons*, vol. 5, no. 1, pp. 59–64, 2018. View at: Publisher Site | Google Scholar - S. Ohno, K. Imasato, S. Anand et al., “Phase boundary mapping to obtain n-type Mg
_{3}Sb_{2}-based thermoelectrics,”*Joule*, vol. 2, no. 1, pp. 141–154, 2018. View at: Publisher Site | Google Scholar - J. Shuai, B. Ge, J. Mao, S. Song, Y. Wang, and Z. Ren, “Significant role of Mg stoichiometry in designing high thermoelectric performance for Mg
_{3}(Sb,Bi)_{2}-based n-type Zintls,”*Journal of the American Chemical Society*, vol. 140, no. 5, pp. 1910–1915, 2018. View at: Publisher Site | Google Scholar - W. Peng, G. Petretto, G.-M. Rignanese, G. Hautier, and A. Zevalkink, “An unlikely route to low lattice thermal conductivity: small atoms in a simple layered structure,”
*Joule*, vol. 2, no. 9, pp. 1879–1893, 2018. View at: Publisher Site | Google Scholar - L. Song, J. Zhang, and B. B. Iversen, “Simultaneous improvement of power factor and thermal conductivity via Ag doping in p-type Mg
_{3}Sb_{2}thermoelectric materials,”*Journal of Materials Chemistry A*, vol. 5, no. 10, pp. 4932–4939, 2017. View at: Publisher Site | Google Scholar - J. Shuai, Y. Wang, H. S. Kim et al., “Thermoelectric properties of Na-doped Zintl compound: Mg
_{3−x}Na_{x}Sb_{2},”*Acta Materialia*, vol. 93, pp. 187–193, 2015. View at: Publisher Site | Google Scholar - A. Bhardwaj and D. K. Misra, “Enhancing thermoelectric properties of a p-type Mg
_{3}Sb_{2}- based Zintl phase compound by Pb substitution in the anionic framework,”*RSC Advances*, vol. 4, no. 65, pp. 34552–34560, 2014. View at: Publisher Site | Google Scholar - D. D. Fan, H. J. Liu, L. Cheng, J. H. Liang, and P. H. Jiang, “A first-principles study of the effects of electron–phonon coupling on the thermoelectric properties: a case study of the SiGe compound,”
*Journal of Materials Chemistry A*, vol. 6, no. 25, pp. 12125–12131, 2018. View at: Publisher Site | Google Scholar - B. Liao, B. Qiu, J. Zhou, S. Huberman, K. Esfarjani, and G. Chen, “Significant reduction of lattice thermal conductivity by the electron-phonon interaction in silicon with high carrier concentrations: a first-principles study,”
*Physical Review Letters*, vol. 114, no. 11, article 115901, 2015. View at: Publisher Site | Google Scholar - C. Liu, M. Yao, J. Yang, J. Xi, and X. Ke, “Strong electron-phonon interaction induced significant reduction in lattice thermal conductivities for single-layer MoS
_{2}and PtSSe,”*Materials Today Physics*, vol. 15, article 100277, 2020. View at: Publisher Site | Google Scholar - O. Hellman, P. Steneteg, I. A. Abrikosov, and S. I. Simak, “Temperature dependent effective potential method for accurate free energy calculations of solids,”
*Physical Review B*, vol. 87, no. 10, article 104111, 2013. View at: Publisher Site | Google Scholar - O. Hellman, I. A. Abrikosov, and S. I. Simak, “Lattice dynamics of anharmonic solids from first principles,”
*Physical Review B*, vol. 84, no. 18, article 180301, 2011. View at: Publisher Site | Google Scholar - J. J. Kuo, M. Wood, T. J. Slade, M. G. Kanatzidis, and G. J. Snyder, “Systematic over-estimation of lattice thermal conductivity in materials with electrically-resistive grain boundaries,”
*Energy & Environmental Science*, vol. 13, no. 4, pp. 1250–1258, 2020. View at: Publisher Site | Google Scholar - J. Xin, G. Li, G. Auffermann et al., “Growth and transport properties of Mg
_{3}X_{2}(X = Sb, Bi) single crystals,”*Materials Today Physics*, vol. 7, pp. 61–68, 2018. View at: Publisher Site | Google Scholar - K. Imasato, C. Fu, Y. Pan et al., “Metallic n-type Mg
_{3}Sb_{2}single crystals demonstrate the absence of ionized impurity scattering and enhanced thermoelectric performance,”*Advanced Materials*, vol. 32, no. 16, article 1908218, 2020. View at: Publisher Site | Google Scholar - A. Togo, F. Oba, and I. Tanaka, “First-principles calculations of the ferroelastic transition between rutile-type and CaCl
_{2}-type SiO_{2}at high pressures,”*Physical Review B*, vol. 78, no. 13, article 134106, 2008. View at: Publisher Site | Google Scholar - J. P. Perdew, K. Burke, and M. Ernzerhof, “Generalized gradient approximation made simple,”
*Physical Review Letters*, vol. 77, no. 18, pp. 3865–3868, 1996. View at: Publisher Site | Google Scholar - D. Bessas, I. Sergueev, H. C. Wille, J. Perßon, D. Ebling, and R. P. Hermann, “Lattice dynamics in Bi
_{2}Te_{3}and Sb_{2}Te_{3}: Te and Sb density of phonon states,”*Physical Review B*, vol. 86, no. 22, article 224301, 2012. View at: Publisher Site | Google Scholar - S. Kastbjerg, N. Bindzus, M. Søndergaard et al., “Direct evidence of cation disorder in thermoelectric lead chalcogenides PbTe and PbS,”
*Advanced Functional Materials*, vol. 23, no. 44, pp. 5477–5483, 2013. View at: Publisher Site | Google Scholar - G. A. Slack and S. F. Bartram, “Thermal expansion of some diamondlike crystals,”
*Journal of Applied Physics*, vol. 46, no. 1, pp. 89–98, 1975. View at: Publisher Site | Google Scholar - H. Watanabe, N. Yamada, and M. Okaji, “Linear thermal expansion coefficient of silicon from 293 to 1000 K,”
*International Journal of Thermophysics*, vol. 25, no. 1, pp. 221–236, 2004. View at: Publisher Site | Google Scholar - T. Fang, X. Li, C. Hu et al., “Complex band structures and lattice dynamics of Bi
_{2}Te_{3}-based compounds and solid solutions,”*Advanced Functional Materials*, vol. 29, no. 28, article 1900677, 2019. View at: Publisher Site | Google Scholar - O. Delaire, J. Ma, K. Marty et al., “Giant anharmonic phonon scattering in PbTe,”
*Nature Materials*, vol. 10, no. 8, pp. 614–619, 2011. View at: Publisher Site | Google Scholar - I. Errea, B. Rousseau, and A. Bergara, “Anharmonic stabilization of the high-pressure simple cubic phase of calcium,”
*Physical Review Letters*, vol. 106, no. 16, article 165501, 2011. View at: Publisher Site | Google Scholar - J. S. Kang, H. Wu, M. Li, and Y. Hu, “Intrinsic low thermal conductivity and phonon renormalization due to strong anharmonicity of single-crystal tin selenide,”
*Nano Letters*, vol. 19, no. 8, pp. 4941–4948, 2019. View at: Publisher Site | Google Scholar - M. Christensen and B. B. Iversen, “Host structure engineering in thermoelectric clathrates,”
*Chemistry of Materials*, vol. 19, no. 20, pp. 4896–4905, 2007. View at: Publisher Site | Google Scholar - M. Christensen, N. Lock, J. Overgaard, and B. B. Iversen, “Crystal structures of thermoelectric
*n*-and*p*-type Ba_{8}Ga_{16}Ge_{30}studied by single crystal, multitemperature, neutron diffraction, conventional X-ray diffraction and resonant synchrotron X-ray diffraction,”*Journal of the American Chemical Society*, vol. 128, no. 49, pp. 15657–15665, 2006. View at: Publisher Site | Google Scholar - Y. Mudryk, P. Rogl, C. Paul et al., “Thermoelectricity of clathrate I Si and Ge phases,”
*Journal of Physics: Condensed Matter*, vol. 14, no. 34, pp. 7991–8004, 2002. View at: Publisher Site | Google Scholar - M. T. Agne, K. Imasato, S. Anand et al., “Heat capacity of Mg
_{3}Sb_{2}, Mg_{3}Bi_{2}, and their alloys at high temperature,”*Materials Today Physics*, vol. 6, pp. 83–88, 2018. View at: Publisher Site | Google Scholar - J. F. G. Kresse, “Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set,”
*Physical Review B*, vol. 54, no. 16, pp. 11169–11186, 1996. View at: Publisher Site | Google Scholar - L.-F. Huang, X.-Z. Lu, E. Tennessen, and J. M. Rondinelli, “An efficient ab-initio quasiharmonic approach for the thermodynamics of solids,”
*Computational Materials Science*, vol. 120, pp. 84–93, 2016. View at: Publisher Site | Google Scholar

#### Copyright

Copyright © 2020 Yifan Zhu et al. Exclusive Licensee Science and Technology Review Publishing House. Distributed under a Creative Commons Attribution License (CC BY 4.0).