1 Introduction

1.1 Thermoelectricity

Renewable energy is a research area that is gaining global attention owing to environmental pollution concerns and the exhaustion of fossil fuel resources. Particularly, air pollution from fossil fuel used for fueling vehicles and in industries is a concerning global problem. The thermoelectric (TE) technology involves the direct and reversible conversion between thermal energy and electrical energy that is generated as electrons or holes move owing to temperature differences. Given that more than 60% of energy is wasted as heat during the energy generation and consumption process, thermoelectric power generation (TEG) is a promising energy generating technology. TEG devices present great advantages, including long life and high degree of environmental friendliness. Moreover, they can be manufactured in various areas from local to large areas. However, TEG devices are still difficult to commercialize owing to their low energy efficiency and mechanical reliability, and also their price competitiveness. The efficiency of TEG devices can be calculated as follows:

$$\eta = \frac{{T_{h} - T_{c} }}{{T_{h} }} \cdot \frac{{\sqrt {1 + Z\overline{T} } - 1}}{{\sqrt {1 + Z\overline{T} } + \frac{{T_{c} }}{{T_{h} }}}},$$
(1)

where \(T_{h}\) and \(T_{c}\) are the temperatures of the hot and cold sides, respectively, \(\overline{T}\) is the average of \(T_{h}\) and \(T_{c}\), and ZT is a dimensionless figure of merit. This dimensionless figure of merit can be used to evaluate TE materials and has been defined as:

$$ZT = \sigma S^{2} /\kappa$$
(2)

where σ, S, T, and κ are the electrical conductivity, Seebeck coefficient, absolute temperature, and total thermal conductivity of the material, respectively. TEG devices consist of a TE material, electrodes, paste materials, and diffusion barriers. Among these, the TE material is the most important component that determines their efficiency. Therefore, various strategies should be established to allow researchers to obtain materials with high ZT and excellent mechanical reliability and to develop inexpensive manufacturing processes [1].

1.2 Silicide-Based TE Materials

Recently, various TE materials featuring high ZT values have been researched, including skutterudites, half-Heusler alloys, silicides, and chalcogenides (Fig. 1 and Table 1) [2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23]. The ZT value of Bi0.52Sb1.48Te3, which presented dense dislocations, was ~ 2.0 at low temperature [2], while those of PbTe-, Ag–Pb–Sb–Te-, and SnSe-based TE materials were as high as ~ 2.5 at mid-to-high temperature [14, 16, 21, 22]. However, most TE materials that present very high ZT values contain expensive and toxic elements, and therefore, can be difficult to commercialize. Because Bi and Te are very rare, they are also very expensive. Moreover, the use of Pb and Se has been gradually reduced worldwide owing to their toxicity. Therefore, cost and toxicity considerations are essential when developing TE materials. Consequently, silicide-based materials could be promising candidates for TE materials and devices. Silicide-based TE materials have received significant attention as automotive thermoelectric generators (ATEGs) owing to their low density, low cost, and non-toxicity.

Fig. 1
figure 1

Figure of merit, ZT, as function of temperature for typical thermoelectric materials. Here, NWs nanowires and CNT carbon nanotube

Full size image
Table 1 Thermoelectric properties of state-of-art thermoelectric materials
Full size table

Magnesium silicide (Mg2Si) and higher manganese silicide (HMS) are representative n- and p-type silicide-based TE materials, respectively. Mg2Si presents cubic structure (Fm3m space group) and a narrow bandgap (~ 0.77 eV) [4]. Generally, the ZT of un-doped Mg2Si is lower than 0.1 owing to its very low electrical conductivity. However, the ZT of doped Mg2Si could reach 1.0 at 873 K [4]. Moreover, the maximum ZT of ~ 1.5 was attributed to its high S and low κ values in Mg2(Si, Sn) solid solution [24]. Additionally, controlling the Mg/Si ratio and the secondary phases, such as Si and MgO are effective route for enhancing the TE properties and ensure the thermal and chemical stability of Mg2Si [25, 26]. Despite its high ZT values, the applications of Mg2Si have been limited owing to its low mechanical reliability (fracture toughness, KIc, value of ~ 0.82 MPa m1/2), which suggested that more research into the mechanical reliability of TE materials is required to fabricate efficient TEG devices [27].

Unlike Mg2Si, HMS exhibits complex tetragonal crystal structure (Mn sub-lattices (chimneys) and interpenetrating helical Si sub-lattices (ladders)) known as the Nowotny Chimney Ladder (NCL) structure [28]. Moreover, depending on their Mn-to-Si atomic ratio, HMS presents different phases, including Mn4Si7, Mn11Si19, Mn15Si26, and Mn27Si47, which are NCL phases of various stoichiometry [28]. Although HMS phases exhibited the different lattice parameters (a = b = 5.52–5.53 Å and c = 17.46–117.90 Å) and space groups (P\(\overline{4}\)c2, P\(\overline{4}\) n2, I\(\overline{4}\)2d, and P \(\overline{4}\)n2), their electronic band structure (0.76–0.78 eV) and intrinsic TE properties (ZT = ~ 0.4 at 700–800 K) are similar [29,30,31,32]. The MnSi and Si secondary phases of HMS negatively affect their TE properties. Therefore, suppressing the secondary phases generation is essential for achieving good TE properties by optimizing the fabrication process and performing compositional tuning [33,34,35]. The ZT values of HMS doped at the Mn- and Si-sites have been reported to be as high as 0.6. Furthermore, nanocomposite engineering strategies have been used to improve the TE properties of doped HMS by manipulating their electronic and thermal transport properties [8]. Despite their low ZT values compared with those of Mg2Si, HMS exhibits excellent mechanical, chemical, and thermal stability, and therefore, novel strategies for improving their TE properties are required.

Herein, we summarize the strategies for improving the TE properties and mechanical reliability of silicide-based TE materials. We also demonstrate the importance of microstructure and band structure control to obtain the high TE properties and mechanical reliability of silicide-based TE materials. Lastly, we suggest development directions and provide material design guidelines for silicide-based TE materials to obtain highly efficient TEG devices.

2 Increasing ZT of Silicide Bulk Materials

2.1 Compositional Tuning and Optimizing n c

The individual relationships between S, σ, and κ and nc, can be expressed using the following equations:

$$S = \frac{{8\pi^{2} k_{B}^{2} }}{{3eh^{2} }}m^{*} T\left( {\frac{\pi }{{3n_{c} }}} \right)^{2/3} ,$$
(3)
$$\sigma = n_{c} e\mu_{Hall} ,$$
(4)

and

$$\kappa = L\sigma T = Ln_{c} e\mu_{Hall} T ,$$
(5)

where \(m^{*}\), ρ, and μHall are the effective mass, electric resistivity, and Hall mobility, respectively, e is the electron charge, kB is the Boltzmann constant, and L is the Lorenz number. These relationships indicate that control of nc is a key strategy for achieving high ZT values. The optimization of nc via doping is essential for improving the TE properties of silicide-based TE materials, owing to their low ZT values (~ 0.1 and ~ 0.4 for Mg2Si and HMS, respectively) [33, 36]. Sb and Bi are the main n-type doping elements that replace the Si sites, and Al is another n-type doping element that substitutes the Mg sites of Mg2Si-based compounds [4, 36,37,38,39,40]. Various other dopants, such as Y, Te, and Pb were also used, however, the increase in the TE properties of the Mg2Si-based materials was insufficient (Fig. 2 and Table 2) [41,42,43,44]. Sb and Bi doping caused nc to increase, which led to the increase in σ and decrease in the lattice thermal conductivity (κlat) via point defect phonon scattering [36,37,38]. Kim et al. were able to obtain a Mg2Si-based material featuring the ZT value of 1.0 at 873 K by optimizing its nc (0.9 × 1020–1.2 × 1020 cm−3) and controlling the solubility of Bi via a co-doping strategy [4].

Fig. 2
figure 2

Figure of merit, ZT, as function of temperature for n-type doped Mg2Si

Full size image
Table 2 Thermoelectric properties and typical synthesis methods of n-type doped Mg2Si
Full size table

Using doping element control, it is possible to obtain Mg2Si-based compounds that feature p-type characteristics. Li, Na, and Ga are representative p-type doping elements. Among them, Li and Na are substitutional dopants at the Mg site, and Ga is a dopant at the Si site (Fig. 3 and Table 3) [7, 45,46,47,48,49,50,51,52]. The p-type Mg2Si comprises doped Mg2Si–Mg2Sn and Mg2Sn-Mg2Ge solid solutions. The presence of these solid solutions increased the S value of the p-type Mg2Si owing to band convergence and decreased its κlat value owing to intensified phonon scattering [7, 45,46,47,48,49]. The ZT value of the p-type Mg2Si was 0.7 at 675 K, which was still lower than that of the n-type Mg2Si [7].

Fig. 3
figure 3

Figure of merit, ZT, as function of temperature for p-type doped Mg2Si

Full size image
Table 3 Thermoelectric properties and typical synthesis methods of the p-type doped Mg2Si
Full size table

Typically, HMS is generally known to be the p-type counterparts of Mg2Si-based TE materials. The TE properties of the undoped HMS are poor (ZT ~ 0.4) [33] and their nc are relatively high (1.1 × 1021–2.7 × 1021 cm−3) [53]. V, Cr, Fe, Co, W, and Re were investigated as doping elements at the Mn sites, and Al and Ge were studied as doping elements at the Si sites of HMS (Table 4) [9, 35, 54,55,56,57,58,59,60,61,62,63,64,65,66,67]. Chen et al. obtained HMS that presented the ZT value of 0.57 at 800 K via intensified phonon scattering. This high ZT value was attributed to the presence of point defects and the formation of ReSi1.75 nanoinclusions, which were induced by Re doping [35]. She et al. reported Ge-doped HMS that presented the ZT value of 0.62 at 840 K by optimizing nc, increasing \(m^{*}\), and reducing κlat [64]. However, because nc optimization strategies are limited owing to the doping solubility limit, novel processes such as gas atomization, rapid solidification, and shock wave were used to improve the TE properties of doped HMS by rapidly decreasing κlat [9, 55, 57, 66, 67]. Ge and Re co-doping combined with the shock wave process generated high density dislocations, which resulted in the high ZT value (~ 1.0) of the co-doped HMS (Fig. 4) [67].

Table 4 Thermoelectric properties and typical synthesis methods of p-type doped higher manganese silicide (HMS)
Full size table
Fig. 4
figure 4

Figure of merit, ZT, as function of temperature for p-type doped higher manganese silicide. Here, SSR solid state reaction, MS melt spinning, TE thermal explosion, BM ball milling, HP hot pressing, SPS spark plasma sintering, and PAS plasma activated sintering

Full size image

Controlling the nc of Mg2Si and HMS by controlling their Mg and Si contents, respectively, has been studied. However, the increase in their TE properties obtained by controlling nc was insufficient. Therefore, manipulating the electronic and thermal transport properties of Mg2Si and HMS by controlling their band structure and nanostructure should be considered to achieve materials with significantly improved TE properties.

2.2 Nanostructure Engineering

Nanostructure engineering is a conventional strategy for improving TE properties by decreasing κlat. Nanograins, heterostructures, nanophases, and nanoinclusions are effective for reducing κlat by controlling the interface properties of materials (Fig. 5a–c) [1, 68]. According to calculation results previously reported, the increase in interface density is accompanied by the decrease in κlat and μHall because for silicide-based TE materials the mean free paths of phonons and electrons are similar [69,70,71]. Therefore, nanocomposite engineering strategies, such as the introduction of nanophases and heterostructures, are more effective for manipulating the electronic and thermal transport properties of silicide-based TE materials.

Fig. 5
figure 5

Reprinted with permission from [1]. Copyright 2019 J. Korean Ceram. Soc.)

Schematic, scanning electron microscopy, transmission electron microscopy, and energy-dispersive X-ray spectroscopy images of microstructure of silicide-based thermoelectric nanocomposites: a heterostructure, b nanoparticles/nanoinclusions, and c nanosheets/nanowires. Here, rGO reduced graphene oxide.

Full size image

SiC, multi-wall carbon nanotube (MWCNT), reduced graphene oxides (rGOs), metal nanoparticles (NPs), and Si were added to Mg2Si to increase its TE properties (Fig. 6 and Table 5) [27, 34, 68, 72,73,74,75,76,77,78,79]. However, the ZT values of all nanocomposites were lower than those of pristine samples. Kim et al. investigated the relationship between the TE and mechanical properties by introducing various nanophases, such as metal NPs, rGOs, and also via dual nanoinclusion. They reported that the TE properties of all nanocomposites decreased owing to the significant decrease in their μHall values, despite the decrease in thermal conductivity [27, 68, 77, 78]. Moreover, Yin et al. confirmed the decrease in the TE properties of Mg2.16(Si0.3Sn0.7)0.98Sb0.02/SiC nanowire or Mg2.16(Si0.3Sn0.7)0.98Sb0.02/SiC nanoparticles nanocomposites [72]. The power factors of Mg2Si nanocomposites were low owing to their unfavorable band alignment induced by the low work function of Mg2Si (3.59 eV) [80]. The decrease in κlat caused by the introduction of nanophases was insufficient because the main scattering centers of Mg2Si-based TE materials were point defects.

Fig. 6
figure 6

Figure of merit, ZT, as function of temperature for n-type Mg2Si nanocomposites. Here, NWs nanowires, NPs nanoparticles, MWCNT multi-wall carbon nanotube, and rGO reduced graphene oxide

Full size image
Table 5 Thermoelectric properties and typical synthesis methods of n-type Mg2Si nanocomposites
Full size table

Conversely, HMS nanocomposites have been relatively effective in improving TE properties (Fig. 7 and Table 6) [8, 34, 81,82,83,84,85]. The ZT values of HMS nanocomposites with Te nanowires and MnS were high (~ 0.7 and 0.6, respectively at 823 K). These values were achieved by reducing κlat to the theoretical limit [8, 34]. Additionally, the power factor of the HMS nanocomposites could be increased by controlling their band alignment. Precisely controlling the band gap and work function of the matrix and nanophases, led to the minimization of the decrease in μHall at the interfaces; consequently, the power factor could be improved using the energy filtering and charge transfer effects. We fabricated metal NP-embedded HMS nanocomposites and HMS/metal silicide composites and determined that the increase in their power factors by the precise control of the band alignment and interfacial properties of these materials. Moreover, unexpected nanostructures (dislocations, precipitates, and secondary phases), induced by novel synthesis technique, led to the decrease in κlat, which, consequently, increased the ZT value [9, 55, 57, 66, 67, 84]. Nanostructure engineering, including the introduction of nanophases or heterostructures in HMS TE materials, is an effective route for increasing their power factors by controlling their interfacial properties, and also for decreasing κlat owing to the intensified phonon scattering at interfaces. Therefore, material design optimization should be implemented to maximize the increase in ZT.

Fig. 7
figure 7

Figure of merit, ZT, as function of temperature for p-type higher manganese silicide nanocomposites. Here, NWs nanowires and MWCNT multi-wall carbon nanotube

Full size image
Table 6 Thermoelectric properties and typical synthesis method of the p-type HMS nanocomposites
Full size table

2.3 Band Structure Modification

Band structure modification via the formation of point defects is an effective for increasing the power factor. Band structure modification strategies, including band convergence, band flattening, and resonant states, can increase the density of states near the Fermi level, which would result in the increase in S [86]. For Mg2S-based compounds, pseudo-binary (Mg2Si–Mg2Sn, Mg2Sn–Mg2Ge, and Mg2Sn–Mg2Pb) and pseudo-ternary (Mg2Si–Mg2Sn–Mg2Ge, Mg2Si–Mg2Sn–Mg2Pb, and Mg2Sn–Mg2Ge–Mg2Pb) systems contributed to the band convergence effect [5, 6, 24, 87,88,89,90,91,92,93,94,95,96,97,98,99]. Moreover, alloying Mg2Si with Mg2Sn, Mg2Ge, or Mg2Pb reduced the κlat values of these Mg2Si-based compounds owing to the intensified phonon scattering by point defects and nanoparticles. Therefore, the convergence of the conduction bands could lead to high ZT values. Figure 8 and Table 7 show the ZT values of the above-mentioned doped pseudo-binary and pseudo-ternary systems [5, 6, 24, 87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102]. The ZT values of these systems were higher (> 1.0) than those of doped Mg2Si (Fig. 2).

Fig. 8
figure 8

Figure of merit, ZT, as function of temperature for n-type pseudo-binary (Mg2Si–Mg2Sn, Mg2Sn–Mg2Ge, and Mg2Sn–Mg2Pb) and pseudo-ternary (Mg2Si–Mg2Sn–Mg2Ge, Mg2Si–Mg2Sn–Mg2Pb, and Mg2Sn–Mg2Ge–Mg2Pb) systems

Full size image
Table 7 Thermoelectric properties and typical synthetic method of the n-type pseudo-binary (Mg2Si–Mg2Sn, Mg2Si–Mg2Ge, Mg2Sn–Mg2Ge, and Mg2Sn–Mg2Pb) and pseudo-ternary (Mg2Si–Mg2Sn–Mg2Ge, Mg2Si–Mg2Sn–Mg2Pb, and Mg2Sn–Mg2Ge–Mg2Pb) systems
Full size table

The TE quality factor, B, was introduced to elucidate the mechanism of the increase in ZT via band convergence. The B factor, which depends on the thermal and electronic transport properties of materials, and also on ZT, can be expressed as follows [86]:

$$B = \frac{{2k_{B} h}}{3\pi }\frac{{N_{v} C_{l} }}{{m_{I}^{*} \Xi^{2} \kappa_{lat} }}T ,$$
(6)

where \(h\) is the reduced Planck constant, \(N_{v}\) is the number of degenerated valleys of the band, \(C_{l}\) is the average longitudinal elastic moduli, \(m_{I}^{*}\) is the inertial effective mass, and Ξ is the deformation potential coefficient. The increase in \(N_{v}\) owing to the small energy separation (ΔE) between the light and heavy conduction bands (Fig. 9) led to the increase in B, which indicated that ZT could also be increased.

Fig. 9
figure 9

Reprinted with permission from [5]. Copyright 2012 American Physical Society)

Schematic electronic band structure of Mg2Si–Mg2Sn system as function of Sn content. Here, CH = heavy conduction band, CL = light conduction band, Eg = band gap, ∆E = energy separation between CL and CH, and V = valence band.

Full size image

The Sb or Bi doped Mg2Si1−xSnx (0.4 ≤ x ≤ 0.7) has a high ZT (> 1.0) owing to the decrease in κlat and the increase in S [5, 6, 24, 87,88,89,90,91,92,93,94,95, 99]. Yin et al. determined that an increase in Ge content in pseudo-ternary Mg2Si–Mg2Sn–Mg2Ge system caused the improved ZT owing to smaller ΔE and the intensified phonon scattering [95]. Liu et al. investigated the pseudo-binary Mg2Sn–Mg2Ge system [96, 97]. The ZT value of Mg2Sn0.73Ge0.25Sb0.02 was higher than those of Mg2Sn systems owing to band convergence and Ge alloying. Moreover, they determined that Ge alloying caused the weighted mobility to increase, band gap to widen, and κlat to decrease [96].

According to Eq. (3), if \(m^{*}\) is high, S will also be high. Increasing \(m^{*}\) via doping could be caused by the increase in \(m_{b}^{*}\). However, \(m_{b}^{*}\) is related to the μHall as follows:

$$m^{*} = N_{v}^{2/3} m_{b}^{*} ,$$
(7)

where \(m_{b}^{*}\) is the band effective mass of a single valley.

$$\mu_{Hall} \propto \frac{1}{{m_{b}^{*5/2} }} .$$
(8)

Therefore, increasing \(m^{*}\) via doping would not cause the power factor to increase owing to the decrease in σ, which was, in turn, caused by the decrease in μHall [103]. Kim et al. [4] and Lee et al. [63] increased \(m^{*}\) by doping Mg2Si and controlling the Si contents in HMS, and consequently, the power factor increased owing to the synergistic effect of the increase in nc and \(m^{*}\) (Fig. 10a, b). Moreover, Gao et al. [67] and Shi et al. [104] increased the S value and power factor of HMS by tuning \(m^{*}\) using a new fabrication process: the rapid solidification and shock wave process.

Fig. 10
figure 10

Seebeck coefficient, S, as function of carrier concentration, nc, of a n-type Mg2Si at 873 K and b p-type higher manganese silicide at 300 K [4, 63]

Full size image

Band structure modifications require precisely controlled doping strategies to achieve high TE properties. Therefore, it is important to systematically control the composition of TE materials, because precisely manipulating their band structure could lead to the increase in their TE properties.

3 Improving Mechanical Properties of Silicide Bulk Materials

Evaluating the chemical and thermal stability, and also the mechanical reliability of TE materials is essential for the fabrication and commercialization of TE modules. The mechanical reliability of TE materials is the most important factor because thermal stress, which is induced by the temperature differences between the hot and cold sides, could cause the fracture of TE materials, and consequently the breakdown of all electric circuits of TE modules. Moreover, ATEG systems should be reliable and should be able to withstand the vibrations and shocks that occur during vehicle operation. However, Mg2Si-based compounds can be easily fractured owing to their inherent brittle fracture characteristics compared with those of other TE materials. Therefore, improving the mechanical reliability of Mg2Si-based compounds should be pursued before fabricating silicide-based TE modules.

Fracture toughness (KIc) is an important factor for evaluating the mechanical reliability of materials. Materials that present high KIc exhibit high mechanical reliability because crack propagation does not occur actively when cracks form. Therefore, the introduction of secondary phases into materials is an effective strategy for enhancing their KIc by preventing the crack propagation.

Three representative mechanisms: crack propagation deflection, crack bridging, and sheet pull-out, could be used to increase KIc [105]. When cracks propagate, propagation is interrupted when a secondary phase that presents high KIc is reached. This is the main mechanism for increasing KIc and can be observed when nanophases are introduced in nanocomposites [27, 77]. Crack bridging and pull-out are mechanisms for increasing KIc that occur mainly in nanocomposites with one-dimensional (1D) and two-dimensional (2D) nanophases. Low-dimensional nanophases can interfere with crack propagation by absorbing the crack propagation energy [68, 72]. Additionally, since the contact area of the 2D nanophase is wider than that of the 1D nanophase, crack propagation energy can be absorbed more effectively in nanocomposites with 2D nanophases [105]. Therefore, the contents, morphologies, and intrinsic properties of the nanophases should be precisely controlled, because KIc highly depends on the microstructure of nancomposites.

Kim et al. demonstrated the relationship between the TE properties and KIc values when various nanophases, such as metal NPs, rGO, and dual nanoinclusions, were introduced into Mg2Si. They also reported the dependence of the TE and mechanical properties on the dimensions of the nanophases [27, 68, 77, 78]. Figure 11 shows the KIc values of Mg2Si nanocomposites. The KIc of the 3-dimensional nanophases (metal NPs) was 1.10 MPa m1/2 and was achieved via crack propagation deflection; nonetheless, the effect was insufficient owing to interface density saturation. However, for 2D nanophases, all three KIc enhancement mechanisms were activated, and the highest achieved KIc was 1.88 MPa m1/2 [68]. Moreover, the crack propagation deflection mechanism was strengthened by introducing dual nanoinclusions (metal NPs and rGO), which resulted in the high KIc value of 2.26 MPa m1/2 [78]. Additionally, Yin et al. achieved the KIc value of 1.36 MPa m1/2 by introducing SiC nanowires at the grain boundaries and studied the effect of the nanophase morphology (nanoparticles and nanowires) on the mechanical reliability behavior of materials [72].

Fig. 11
figure 11

Fracture toughness, KIc, values of various Mg2Si nanocomposites. Here, rGO reduced graphene oxide, SiCw SiC wire, and SiCp SiC particle

Full size image

However, a significant trade-off relationship exists between ZT and KIc. Figure 12 depicts the relationship between ZT and KIc for Mg2Si nanocomposites. The increase in interface density between the nanophase and the matrix caused the decrease in TE properties of Mg2Si nanocomposites due to the decrease in μHall. Therefore, it is important to identify an optimal nanophase that could improve the mechanical properties of materials while maintaining their TE properties. Moreover, it is important to design rule that can achieve a high trade-off relationship between the TE and mechanical properties of TE materials.

Fig. 12
figure 12

Dependence of fracture toughness (KIc) on figure of merit, ZT for various Mg2Si nanocomposites

Full size image

4 Summary and Perspective

Herein, we summarize the methods for increasing the TE and mechanical properties of silicide-based TE materials. Although silicide-based TE materials have received significant attention as ATEGs, their practical applications are still hindered by many limitations. Various strategies, such as composition tuning, nc optimization, nanostructuring, nanocomposite engineering, band convergence, and effective mass tuning, have been used to overcome these limitations, and led to achieving ZT values of 1.6 and 1.0 for n-type Mg2Si and p-type HMS, respectively. Moreover, the high KIc value of 2.26 MPa m1/2 was achieved for Mg2Si via the introduction of nanophases to overcome its low mechanical reliability. However, researchers should continue to develop strategies for improving the TE properties and mechanical reliability of the silicide-bases TE materials. Band and nanostructure engineering were particularly important for overcoming the trade-off relationships between nc and S, μHall and κlat, and ZT and KIc. Moreover, the novel technique that enables us to design the material is a prerequisite. Hence, a systematic strategy that combines calculations and experimental results should be established to achieve the optimized composition and microstructure design.

Moreover, not only the development of TE materials but also the development of bonding, diffusion barrier, and electrode materials are required. Research on increasing the mechanical properties of TE materials, engineering surface coating and sealing technologies to improve the chemical stability of silicide-based materials, and developing diffusion barrier and bonding materials are important research areas for expanding the commercial applications of silicide-based TE modules.

Although researchers are still facing challenges in their attempts to improve the TE properties of silicide-based TE materials and to develop high-efficiency TE modules, systematic studies and efforts could lead to significant progress in this field.