Multicomponent Chalcogenides with Diamond-Like Structure as Thermoelectrics

Abstract

Due to the advantages of environment-friendly constituent elements, relatively large Seebeck coefficient, and low thermal conductivity, multicomponent diamond-like chalcogenides (MDLCs), such as CuInTe₂, Cu₂SnSe₃, Cu₃SbSe₄ and Cu₂ZnSnSe₄, have attracted intensive attention for energy conversion as promising thermoelectric (TE) materials in recent years. This chapter provides an overview of research on MDLCs in TE field. Commencing with the crystal structure and phase transition of MDLCs, we will introduce electronic structure and lattice dynamics of MDLCs through some typical TE compounds. We then discuss new methods (i.e., band engineering, entropy engineering, in situ displacement reaction, and mosaic nanostructure) developed in MDLCs for optimizing TE performance. Finally, in addition to the performance of TE device, investigations on stability and mechanical properties of MDLCs are also presented. For future practical applications of this potential material system, the problems needed to be solved and possible directions to further promote TE performance are also explored in the outlook part.

1 Introduction

As promising candidates to fight against energy crisis and environmental pollution, thermoelectric (TE) materials have drawn worldwide attention due to the ability to achieve direct and reversible energy conversion between heat and electricity [1,2,3]. The conversion efficiency of TE material is evaluated by the figure-of-merit ZT = S²σT/κ, where S is Seebeck coefficient, σ is electrical conductivity, T is absolute temperature, and κ is thermal conductivity (including electron component κ_e and lattice component κ_L). Therefore, to get high ZT, one needs large S, high σ, and low κ. However, the inherently coupled relationships among S, σ, and κ make the optimization of ZT value challenging, which means the improvement of one TE parameter is usually achieved at the cost of another [4, 5]. Recently, some effective strategies have been adopted to obtain high ZT values greater than 2.0 in PbTe- and SnSe-based TE materials, such as electronic density of states (DOS) distortion [6], band convergence [7] and hierarchical architectures scattering [8] in polycrystal PbTe, shifting Fermi level (E_f) close to the edge of multibands, [9] and 3D charge and 2D phonon transports in single crystal SnSe [10]. However, high toxic Pb element or cleavage fracture issue in these high-efficiency TE materials will impede the commercialization process. Hence, it is of importance to alternatively develop mechanically robust TE materials with eco-friendly and earth-abundant elements [11,12,13,14].

Multicomponent diamond-like chalcogenides (MDLCs), characterized by the diamond-like structure with deformed tetrahedron building blocks, have ZT values as high as 1.6 for p-type CuInTe₂-based [15] and 1.1 for n-type AgInSe₂-based TE materials [16] (Fig. 7.1), implying great potential for TE power generation at mid-temperature range as eco-friendly TE material [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]. In this chapter, crystalline structure of MDLCs is analyzed concisely as well as the theoretical electron and phonon transport properties. Then, some new strategies to boost TE performance are discussed in MDLCs, which will provide useful reference for other TE materials. In the last part, TE device related advances in MDLCs are summarized.

Fig. 7.1

Temperature-dependent ZT of (a) p- and (b) n-type MDLC TE materials: AgGaTe₂ [17], CuGaTe₂ [18], CuInTe₂ [15], Cu₂SnS₃ [19], Cu₂GeSe₃ [20], Cu₂SnSe₃ [21], Cu₃SbS₄ [22], Cu₃SbSe₄ [23], Cu₂ZnSnS₄ [24], Cu₂ZnSnSe₄ [25], Cu₂ZnGeSe₄ [26], Cu₂CdSnSe₄ [27], Cu₂MnSnSe₄ [28], Cu₂CoSnS₄ [29], CuFeInTe₃ [30], CuFeS₂ [31], AgIn₅Se₈ [32], and AgInSe₂ [16]

2 Crystal Structure and Phase Transition

MDLCs and analogues (Fig. 7.2) can be viewed as derivatives from cubic zinc-blende binary II–VI compounds (e.g., ZnSe) through orderly substitution of II atoms with the same number of atoms of I + III, I₂ + IV, I₃ + V, and I₂ + II + IV or through other substitutions containing vacancy. Most MDLCs, such as CuInSe₂, Cu₂ZnSnSe₄, Cu₃SbSe₄, and AgIn₅Se₈ with cation deficiency, take in chalcopyrite or stannite structure. But Cu₂SnSe₃-type semiconductor crystallizes in monoclinic structure with two types of anion-centered tetrahedron rather than one type of anion-centered tetrahedron in the whole structure [33]. As the substituting atoms own different ionic radius and electronegativity, the distorted lattice with diverse bonding length and angle will present in MDLCs [25, 27], which can disrupt the transport of phonons to suppress κ_L. Meanwhile, mobility of charge carriers of MDLCs will decrease also as compared with its binary counterparts. The special electron and phonon transport properties associated in compounds with diamond-like structure will be described below.

Fig. 7.2

The representative crystal structures of multicomponent diamond-like chalcogenides (MDLCs) derived from cubic zinc-blende structure

Phase transition phenomenon, which can be easily observed in some Cu- or Ag-based TE materials with non-diamond crystal structure like Cu₂Se [34], AgBiSe₂ [35], Cu₃SbSe₃ [36], etc., is often ignored in most MDLC TE materials. But information of temperature-dependent crystal structure is of critical significance to explain the changes of TE transport nature in different temperature ranges; therefore, it is necessary to have some general information on the phase transition of MDLCs. We focus only on some typical MDLC TE materials (CuInTe₂, CuSnSe₃, and Cu₂ZnSnSe₄) with phase transition in the following.

CuInTe₂ is I-III-VI₂ ternary compound and crystallizes normally in chalcopyrite structure which can be treated as a superstructure of zinc-blende type with two types of ordered cation sublattice positions. When temperature is higher than critical value T_c (~940 K), CuInTe₂ with ordered chalcopyrite structure will undergo a first-order phase transition and revert to zinc-blende phase with cation disorder [37]. As the disordered phase contains cross substitutions between non-isovalent cation atoms, it exhibits electronic compensation between donor and acceptor states as well as a decrease in band gap [38]. The reported CuGaTe₂-, AgInSe₂-, and AgInTe₂-based TE materials also have the similar phase transition phenomenon, but this is commonly excluded from consideration in TE research because measured temperatures are obviously below T_c [37]. In addition, abnormal orthorhombic AgInSe₂ can be experimentally obtained under some special synthesis condition [39], and this metastable phase may be retained to normal chalcopyrite state after heating as the case in CuInS₂ [40].

Cu₂SnSe₃ compound is reported to own diverse crystal structures, namely, monoclinic, tetragonal, orthorhombic, and cubic [41]. The actual type of phases for Cu₂SnSe₃ at low temperatures is still controversial as different preparation conditions and compositions may lead to a variety of results. But, among mentioned phases, monoclinic is stable structure, and cubic is metastable structure which occurs at high temperature. Recently Fan et al. reported that Cu₂SnSe₃ presented in monoclinic phase at low temperature and changed to cubic zinc-blende phase with cation disorder after a phase transition at ~930 K [41]. Siyar et al. reported also that stoichiometric Cu₂SnSe₃ samples annealed at low temperatures (720–820 K) mostly possessed of monoclinic phase, while those annealed at temperature above phase transition value (960 K) were mostly of cubic phase [42]. The analogous variety in crystal structure can also be observed in Cu₂SnS₃- and Cu₂GeSe₃-based TE materials, and dimorphic or polymorphic phases can be detected simultaneously in some cases [43, 44].

As cheap and nontoxic compound Cu₂ZnSnS₄ attracts intensive research in the field of solar cell and thermoelectricity, and also its temperature-dependent change in crystal structure is explored in both theoretical and experimental aspects. It usually tends to take in kesterite structure (tetragonal) with alternating Cu/Zn and Cu/Sn planes stacked along [001] direction. As the temperature increases (Fig. 7.3a), the order-disorder transition (second-order phase transition) will take place at ~530 K, and Cu₂ZnSnS₄ will crystallize in cation-disordered kesterite structure with no change in symmetry [45]. As the further increase in temperature (Fig. 7.3b), a first-order phase transition from tetragonal-disordered kesterite to cubic zinc-blende structure will appear at ~1149 K [46].

Fig. 7.3

(a) Temperature-dependent heat capacities and (b) lattice parameters of Cu₂ZnSnS₄ [45, 46]

Researchers observed also coexistence of stannite phase (which are alternatively stacked along [001] by Cu/Cu and Zn/Sn planes) in kesterite-phased Cu₂ZnSnS₄ due to small difference in enthalpies of formation between the phases [47], resulting in the variations in the band gap on the order of 100 meV [48]. However, it appears that Cu₂GeSnSe₄, Cu₂ZnSnSe₄, Cu₂ZnSnTe₄, etc. tend to occur in stannite structure as described in the International Centre for Diffraction Data (ICDD) of X-ray diffraction. The actual stable phase of Cu₂ZnSnSe₄, etc. presented at low temperature remains debatable because of the disagreements found in previous reports. On one hand, no significant difference was observed between the theoretically calculated energetic/dynamical properties of kesterite and stannite phases of either compound [49]; also, X-ray diffraction is not sensitive in distinguishing kesterite and stannite structures. On the other hand, although Cu/Zn disorder was observed by neutron diffraction or other ways [50], the deviation of stoichiometry by forming tiny second phase (like ZnSe) [51], the preserved high temperature-disordered tetragonal phase during preparation, or the existence of point defects (like ${V}_{\mathrm{Cu}}^{-}$ + ${\mathrm{Zn}}_{\mathrm{Cu}}^{+}$) in this quaternary compound [52] can also affect the reliability of results. So, most people in TE field simply treat stannite structure as ground states for most I₂-II-IV-VI₄ selenide compounds, and more detailed structural investigation and analysis are necessary as well as the mechanism for appeared phase transition in Cu₂ZnGeSe₄ and Cu₂FeSnSe₄ [26, 28].

Thus, before application in TE device, the phase transition behavior and probably related problems, for example, ionic migration, should receive utmost attention, especially in I₂-IV-VI₃ (e.g., Cu₂SnSe₃) and I₂-II-IV-VI₄ (e.g., Cu₂ZnSnSe₄) types of TE materials.

3 Electronic Structure and Lattice Dynamics of MDLCs

Analyzing electronic structure and lattice dynamic characteristics enables us to have an in-depth understanding of electron and thermal transport properties of MDLCs, thus providing insightful guidance for TE applications. In this section, electron and phonon structure of MDLCs will be simply explored by some typical chalcogenides.

Zhang et al. analyzed comparatively electronic structure of three representative diamond-like chalcogenides (ZnS, CuGaS₂, and Cu₂ZnSnS₄) as depicted in Fig. 7.4 and summarized main features as follows [53]. (1) Binary ZnS was a simple sp semiconductor with weak hybridization between S − 3p and Zn − 3d states. (2) Cu − 3d orbitals in CuS₄ tetrahedron will split into t₂ and e suborbitals due to crystal field effect. In ternary CuGaS₂, although Cu − 3d-derived e states were localized, strong hybridization could occur between electrons in t₂ suborbitals and S − 3p states, leading to considerable decrease in band gap as well as an increase in the slop of density of states (DOS) near valence band maximum. Thus, intrinsically large Seebeck coefficient S in most MDLC TE materials can be attributed to this relatively steep slope of DOS. (3) The conduction band minimum of Cu chalcogenides was primarily derived from s orbital of Sn (or Ga) and p orbitals of sulfur S, and the deeper Sn − 5s states of Cu₂ZnSnS₄, as compared with Ga − 4s orbitals of CuGaS₂, resulted in the further reduction of band gap. (4) The valence band maximum of Cu chalcogenide was dominated by p − d anti-bonding states, in contrast to the bonding states in ZnS.

Fig. 7.4

Schematic diagram illustrating bonding for (a) ZnS, (b) CuGaS₂, and (c) Cu₂ZnSnS₄ [53]. Reproduced from Ref. [53] with permission from Elsevier

Ab initio lattice dynamics (Fig. 7.5) of AgInSe₂, CuInSe₂ and AgInTe₂ were summarized by Qiu et al. [16]. (1) The acoustic phonons (heat carrying phonons) owned low sound velocity and large Grüneisen parameter γ, which are beneficial for low κ_L in crystalline compounds [54]. (2) The low-frequency optical phonons could bring resonant scattering effect and then hindered the normal transport process of acoustic phonons with similar frequencies, leading to the further decrease of κ_L [54]. (3) The discovered Ag − Se “cluster vibrations” with low frequencies by phonon animation analysis explained of extremely low κ_L of AgInSe₂. Shao et al. reported that low κ of Cu₂GeSe₃ was related to its low Debye temperature and high bonding anharmonicity (γ ~ 1.2), based on lattice dynamic calculations [55]. Cu₃SbSe₄ compound with relatively low κ_L was reported to possess relatively large Grüneisen parameter, γ(Cu₃SbSe₄) = 1.2 > γ(ZnSe) = 0.75, and strong coupling between low-lying optical branches and acoustic phonon modes [56].

Fig. 7.5

(a) Calculated phonon spectra of AgInSe₂, CuInSe₂, and AgInTe₂. Line thickness denotes contributions from Ag or Cu atoms. (b) Energy difference ΔE as function of the phonon amplitude for the first optical mode caused by Ag − Se (or Cu − Se or Ag − Te) clusters [16]. Reproduced from Ref. [16] with permission from John Wiley and Sons

4 New Routes for High-Performance MDLC Thermoelectric Materials

In this section, we revisit only the new routes for high-performance MDLC TE materials, and other approaches, such as optimizing concentration of charge carriers and reduction of κ_L, can be found in our previous work [57].

4.1 Band Engineering

4.1.1 Pseudocubic Approach

Band convergence with increased N_v number, which is generally reported in the crystal structures with high symmetry (e.g., cubic PbTe), is desired for gaining large S without substantially degrading σ [7]. In 2014, Zhang et al. expanded this strategy into chalcopyrite with low symmetry and designed a “pseudocubic approach ” (Fig. 7.6) to achieve good electronic transport properties with effective band convergence by altering the lattice toward a more cubic crystal symmetry arrangement (i.e., $\eta =\frac{c}{2a}$ toward 1) [58]. It was also verified in experiment that CuInTe₂ in pseudocubic structure (Cu_0.75Ag_0.2InTe₂ or CuIn_0.64Ga_0.36Te₂) possessed of significantly enhanced power factor, as well as ZT values due to band convergence (Δ_CF ~ 0). As reported by Zeier et al., increasing η close to unity in stannite Cu₂ZnGeSe₄ via Cu and Fe co-doping could increase in DOS effective mass and lead in the result to increase in power factor and ZT values, providing evidence for band convergence via crystal symmetry [59]. However, this promising approach is limited to low symmetry materials with ideal band gap and low κ_L [4].

Fig. 7.6

Schematic diagram illustrating pseudocubic band convergence in ternary chalcopyrites. Symbols c and a are lattice constants. Γ_4v is nondegenerate band, and Γ_5v is double degenerate band. Δ_CF is the crystal field-induced energy split at top of bands Γ_4v and Γ_5v. The pseudocubic structure has cubic cation framework with Δ_CF embedded in noncubic distorted anion framework [58]. Reproduced from Ref. [58] with permission from John Wiley and Sons

4.1.2 Highly Efficient Doping (HED)

As for TE materials possessing intrinsically high band degeneracy and low concentration of charge carriers, heavy doping by guest atoms to reach the optimal concentration of charge carriers will damage more or less high band degeneracy, resulting in possible decrease of both S and power factor. Thus, selecting highly efficient dopants, which can provide optimal concentration of charge carriers at greatly lower doping content as compared with non-efficient dopants, should be a feasible attempt to improve σ while maintaining large S with little divergence of degenerate bands [60]. As exhibited in Fig. 7.7, serious separation of the degenerate bands in non-HED can be relieved by HED when the same level of high concentration of charge carriers is reached. One could clearly note the advantage of HED by Al/Ga/In compared with non-HED by Ge/Sn in high band degeneracy containing Cu₃SbSe₄: 0.5 mol. % Al/Ga/In substitution could provide comparable or even higher concentration of charge carriers as that of 1 mol. % Ge doping and also had little deviation from Pisarenko relation based on pristine Cu₃SbSe₄ due to nearly non-changed band structure. But high doping concentration (e.g., 5 mol. % Sn) made Seebeck coefficient S degrade obviously due to severe split of degenerate bands. Ultimately, ZT of samples with HED obtained is of higher value with respect to that of non-HED at 623 K.

Fig. 7.7

(a) Schematic diagram illustrating the maintenance of highly degenerate electronic bands of matrix phase by highly efficient doping (HED). (b) Dependence (dashed curve) of Seebeck coefficient S on concentration of charge carriers at room temperature shows that nearly non-changed band structure is obtained via HED (i.e., Al, Ga, In doping) in Cu₃SbSe₄ with high band degeneracy [60]. Reprinted with permission from Ref. [60]. Copyright (2017) American Chemical Society

4.2 Entropy Engineering

4.2.1 Spin Entropy

It was suggested that the accompanying extra entropy in magnetic ions with degeneracy of electronic spin configurations in real space (e.g., transition metal ions with unpaired 3d electrons) could play an important role in getting large Seebeck coefficient S at high σ in cobalt oxides [24, 61]. This striking fact encourages many researchers to explore the possibility of substitution using magnetic ion to enhance TE performance of the other materials . Yao et al. first observed the simultaneous optimization of all three TE parameters in Mn-doped CuInSe₂ [62]. Mn substitution increased the concentration of charge carriers by shifting E_f downward into the region of valence band, and Mn 3d level hybridization with Se 4p level resulted in additional DOS near E_f, enabling simultaneous enhancement of σ and S for CuInSe₂. Introduced local disorder after Mn substitution, which was also verified by Raman spectroscopy, could contribute to the reduction of κ_L. However, the contribution of the spin entropy has been neglected by Yao et al. Up to 2013, Xiao et al. reported that magnetic ion could decouple the strongly interrelated three TE parameters in Cu₂ZnSnS₄ (Fig. 7.8) [24]. More importantly, electron paramagnetic resonance (EPR) results implied that Ni²⁺ doping truly brought extra spin into Cu₂ZnSnS₄ and, therefore, the remarkable enhancement of Seebeck coefficient S. After fully substituted with magnetic ions (e.g., Co) in Cu₂ZnSnS₄, electron and phonon structure would undergo significant change due to the strong hybridization between 3d states of magnetic ions and S 3p states, resulting in improved TE properties. Due to the distinct band structure features of Cu₂CoSnS₄ with magnetic component, high ZT ~ 0.8 could be achieved in this sulfide due to improved electron transport properties from enhanced effective mass and depressed κ_L by weakening of covalent bonding [29].

Fig. 7.8

(a) Schematic diagram illustrating various electron spin and phonon scatterings in quaternary magnetic sulfides. (b) Comparison of ZT for Cu₂XSnS₄ nanocrystals [24]. Reproduced from Ref. [24] with permission from The Royal Society of Chemistry

4.2.2 High-Entropy Alloys

Recently, high-entropy alloys (HEAs) , which consist of multiple principal elements at equimolar or close to equimolar ratios, have been proposed as novel types of alloys with intriguing structural and functional properties [63]. Due to increased chemical complexity and configuration entropy from the occupation of multiple components at the identical atomic sites (Fig. 7.9a, b), HEA may bring significant extra phonon disorders and offer also potential way to modify band structure and result in tuning electrical and thermal transport properties [64]. HEAs of Al_xCoCrFeNi (0.0 ≤ x ≤ 3.0) were firstly introduced in the context of TE by Shafeie et al. as potential and stable TE material [65]. In this scenario, Liu et al. devised a strategy of using entropy as the global gene-like performance indicator which showed how multicomponent TE materials with high configuration entropy could be designed (Fig. 7.9c, d) [64]. Firstly, parameter δ in a novel elastic model was defined by shear modulus, lattice parameters, and mismatch in the atomic radius as a criterion for judging the feasibility of forming a complete solid solution with high entropy. Guided by δ criterion, several multicomponent TE material systems were screened out. In the end, screened candidates were tested in experiment, and enhanced ZT values approached to 1.6 and 2.23 in respective (Cu/Ag)(In/Ga)Te₂- and Cu₂(S/Se/Te)-based multicomponent materials due to the optimization of entropy. Hence, entropy engineering by forming a solid solution with multiple components works as an effective guide to improve greatly TE performance through the simultaneous optimization of electron and thermal transport properties, i.e., to lower κ_L by the presence of local mass and strain fluctuation and to improve S by increasing in crystal symmetry. Although this strategy holds promise for depressing κ_L, the simultaneous improvement of S is limited to some TE materials with increased crystal symmetry.

Fig. 7.9

(a) Schematic diagram of lattice framework in multicomponent materials compared to ordinary binary compound. (b) Schematic diagram of the entropy engineering with multicomponent TE materials. (c) The maximum configurational entropy as a function of material’s solubility parameter δ for given multicomponent TE materials, where n is the number of components. (d) Maximum ZT as a function of configurational entropy in some multicomponent TE materials [64]. Reproduced from Ref. [64] with permission from John Wiley and Sons

4.3 In Situ Displacement Reaction

In order to reduce κ_L of TE materials (e.g., skutterudite), the addition of oxide nanoinclusions is widely used to scatter phonons in nanoscale frequencies [66]. Luo et al. found firstly in situ displacement reaction between ZnO additive and CuInTe₂ matrix at 714 K by differential scanning calorimeter (DSC) and transmission electron microscopy (TEM) [15]. Consequently, this chemical reaction co-strengthened electron and thermal transport properties of CuInTe₂, i.e., to decrease in κ_L via extra phonon scattering from In₂O₃ nanoinclusions and to improve σ by formation of ${\mathrm{Zn}}_{\mathrm{In}}^{-}$ point defects. Then, 80% enhanced ZT value of 1.44 had been obtained in 1 wt. % ZnO + CuInTe₂ sample. Furthermore, by integrating anion substitution (i.e., Sb doping) and in situ displacement reaction, high power factor of 1445 μW × m⁻¹ × K⁻² and record-high ZT of 1.61 at 823 K were achieved in CuInTe₂-based TE material (Fig. 7.10). In addition, simultaneous optimization of electronic and thermal transport properties in TiO₂- or SnO₂-added CuInTe₂ illustrated also the general effectiveness of in situ displacement reaction by other oxides in enhancing TE performance [67, 68].

Fig. 7.10

(a) Schematic diagram of in situ displacement reaction. (b) Illustration of the stepping increment of ZT at 823 K in CuInTe₂ samples by Sb substitution, ZnO adding, and combined Sb substitution and ZnO adding [15]. Reproduced from Ref. [15] with permission from John Wiley and Sons

4.4 Mosaic Nanostructure

The strategy of nanoscale mosaic is reported to be a useful strategy to tune material’s TE properties [69]. As the case in Cu₂S_0.5Te_0.5, high lattice coherence in mosaic crystal helps to gain excellent electron transport, whereas, the lattice strains or interfaces of mosaic nanograins are still very effective in scattering heat conducting phonons, resulting in exceptional ZT above 2 [69]. Recently, Shen et al. discovered that matrix phase of 20 mol. % Zn-doped Cu₂SnS₃ sample presented as a distinctive mosaic-type nanostructure consisting of well-defined cation-disordered domains (the “tesserae”) coherently bonded to a surrounding network phase with semiordered cations (Fig. 7.11) [70]. The network phase Cu_4+xZn_1−xSn₂S₇ was found to have high mobility of charge carriers, and the tesserae had compositions closer to that of the nominal composition. More importantly, this distinctive mosaic nanostructure plays an important role in promoting TE properties due to simultaneous optimization of electron and phonon transport, i.e., charge carriers are smoothly transferred along crystalline network, while phonons are strongly scattered by atomic disorder and abrupt nanointerfaces of mosaic nanograins. Consequently, large power factor (~840 μW × m⁻¹ × K⁻²), ultralow κ (~0.4 W × m⁻¹ × K⁻¹), and high ZT value (~0.6) at 723 K were obtained for Zn-doped Cu₂SnS₃ sample; therefore, this mosaic nanostructure represents a new type of phonon-glass electron-crystal topology, with high-mobility semiconducting phase interwoven with phonon scattering and low thermal conductivity phase.

Fig. 7.11

Atomic-resolution HAADF-STEM (high-angle annular dark-field scanning transmission electron microscopy) images of mosaic nanostructures [70]. Reprinted with permission from Ref. [70]. Copyright (2018) American Chemical Society

5 MDLC-Based Thermoelectric Modules

Since the development of MDLC TE materials is still at the early stage, most attentions have been focused on boosting ZT; however, the feasibility to fabricate efficient TE device based on MDLCs is still an unanswered question. In this section, we will revisit the advances in TE device of MDLCs and discuss also some device-related properties which are of critical importance in practical application.

5.1 Fabrication and Performance Test

In year 2017, researchers started to face this inevitable question and finally opened the door for practical applications of MDLCs TE materials. Liu et al. used firstly (Sn, Bi) co-doped nanocrystalline Cu₃SbSe₄ samples to fabricate innovative ring-shaped TE generator, and thin insulator layers were placed between rings to alternatively contact the outer and inner rings [23]. The final assembled single p-type Cu₃SbSe₄ rings had inner and outer diameter of 28 mm and 39 mm. The test results showed that an open-circuit voltage near 20 mV was obtained for a single TE element when temperature gradient was maintained at 160 K (inner and outside ring temperatures were 250 and 90 °C, respectively) during test process. As electrical resistance of the ring <0.4 Ohm, the voltage translating into output power >1 mW for each single TE element could be obtained. Constricted by equipment setups, the actual output power density of abovementioned TE device is still unknown.

After that, the first truly TE module including both n- and p-type MDLCs had been fabricated by Qiu et al. as shown in Fig. 7.12 [16]. High-performance Ag_0.9Cd_0.1InSe₂ and Cu_0.99In_0.6Ga_0.4Te₂ TE materials were selected as respective n- and p-type legs to set up TE module. First, hot-pressed samples in cylinder shape were polished, diced, and electroplated in sequence. Then, obtained samples were welded to Mo₅₀Cu₅₀ alloy blocks with Cu − P brazing filler metal at hot side and copper clad ceramic substrates with Sn₄₂Bi₅₈ at cold side. In order to lower the heat loss, the gap in assembled TE module with two couples was finally filled with thermal insulation material. After testing, fabricated TE module exhibited maximum output power of 0.06 W under temperature difference of 520 K. Although this output power value is still inferior to that of other well-known TE materials (e.g., PbTe) [71, 72], it opens clearly the door for applications of diamond-like TE materials.

Fig. 7.12

(a) Schematic and photograph (inset) of fabricated diamond-like module. (b) Output voltage and power as function of current for TE module based on diamond-like materials [16]. Reproduced from Ref. [16] with permission from John Wiley and Sons

5.2 Operational Stability

Good stability with maintenance of desired TE properties upon cycling is essential to promote the widespread applications of TE materials. The issue of high-temperature instability in MDLC TE materials has attracted more and more attention in recent years. Skoug et al. reported that a dramatic decrease in both σ and ZT at high temperature was linked to physical deformation of Cu₂Sn_0.925In_0.075(Se_0.7S_0.3)₃ sample during measurement [73]. Allowing for instability issue, the measurement of TE properties was just below 723 K in Cu₂ZnSnSe₄ samples despite the rising trend of ZT with temperature and high melting point (~1078 K) [74]. One could note also that variation of electronic transport properties after several high-temperature cycles was not suppressed by simply coating boron nitride layer on (Sn, Bi) co-doped Cu₃SbSe₄ sample [23]. After realizing the existence of instable properties in some MDLC TE materials, researchers start to face the problem and study it also.

In order to see temperature-dependent formation phases and crystal structure at 300–800 K of hot-pressed CuGaTe₂, Fujii et al. carried out synchrotron X-ray diffraction (SXRD) measurement as well as the crystal structure analysis [75]. It was revealed that Te and CuTe could precipitate from CuGaTe₂ matrix in temperature interval 500–650 K and this would also cause Cu and Te deficiencies in matrix phase. These impurity phases would exist in the form of liquid when temperature higher than 650 K and remained even when cooled to room temperature. Deficiencies of Cu and Te in CuGaTe₂ could have influence on the tetragonal distortion as well as x-coordinate of Te. Therefore, the above changes in composition as well as crystal structure will inevitably affect high-temperature stability of CuGaTe₂, which also can be reflected from the gradual increase in S and decrease in σ over ten measurement cycles from 300 to 800 K [76].

For the purpose of observing whether there is an occurrence of Ag-ion migration in AgInSe₂ compound or not, stability test was conducted by Qiu et al. [16]. The results (Fig. 7.13) showed that (1) electron transport properties were relatively reproducible during three independent cooling and heating rounds, and (2) high-temperature relative resistance (R/R₀, where R₀ is sample’s initial resistance) under current density of 12 A/cm² was almost identical after a long time (~14 h) test, suggesting that AgInSe₂-based TE materials had a good stability, even under the condition of large current. To prevent oxidation, sublimation, or decomposition, a protective layer of glass sodium silicate (~1.5 mm in thickness) was coated on Cu_2.1Zn_0.9SnSe₄ sample, and coated sample exhibited relatively good reproducibility in electron transport properties during cooling and heating cycles [25]. In addition to protective layer, the stability can be strengthened by specific elemental doping (e.g., Ni doping in Cu₁₂Sb₄Se₁₃) [77] or incorporation of second phase (e.g., 3D graphene heterointerface in Cu_2−xS) [78] etc.

Fig. 7.13

(a–b) Reproducibility test for electron transport properties of Ag_0.9Cd_0.1InSe₂ sample. (c) Relative electrical resistance of AgInSe₂ as a function of current density of 12 A/cm² applied at 650 K for various time durations. The data for Cu₂S are included for comparison [16]. Reproduced from Ref. [16] with permission from John Wiley and Sons

5.3 Mechanical Properties

Crack or fatigue damage of TE materials, which can accelerate the deterioration of material performance as well as the failure of TE device, probably occurs because of thermomechanical stresses under temperature cycling and gradients condition [79]. So, enhancing such mechanical properties as mechanical strength and toughness is of significance for the development of MDLC TE devices.

In order to provide an insight into mechanical properties of Cu₃SbSe₄ compound, Tyagi et al. carried out hardness and fracture toughness measurements on Cu₃SbSe₄ by multimode atomic force microscope (AFM) and Vickers microhardness tester [80]. The hardness value (0.63 GPa) of Cu₃SbSe₄ compound measured by AFM agreed well with Vickers hardness value (0.65 GPa), and calculated fracture toughness was found to be 0.80 ± 0.03 MPa × m^1/2, which are comparable with those of some state-of-the-art TE materials (Fig. 7.14) mainly due to the nanoscale features in crystal size.

Fig. 7.14

FESEM (field emission scanning electron microscopy) image of Vickers indentation cracks developed in Cu₃SbSe₄. Inset image is fracture toughness of Cu₃SbSe₄, compared with the reported values of state-of-the-art and competing TE materials [80]. Reprinted with permission from Ref. [80], with permission from AIP Publishing

To understand the intrinsic mechanical properties and determine the deformation mechanism of CuInTe₂, Li et al. applied density functional theory (DFT) to investigate the shear deformation properties of single crystalline CuInTe₂ [81]. It was found that the lowest ideal shear strength was 2.43 GPa along (221) [11-1] slip system, which is obviously lower than those of other high-performance TE materials such as PbTe (3.46 GPa), CoSb₃ (7.19 GPa), and TiNiSn (10.52 GPa), suggesting that enhancing mechanical properties of CuInTe₂ is necessary for applications. The lowest ideal tensile strength was 4.88 GPa along [1–10] tension, and this value was higher than the lowest ideal shear strength. Moreover, the shear-induced failure arose mainly from the softening and breakage of covalent In − Te bond, whereas tensile failure was arisen from breakage of Cu − Te bond. The shear strength was 1.95 GPa under biaxial shear load along (221) [11-1], compression shrunk of In − Te bond and resulted in buckling of In − Te hexagonal framework. The fracture toughness of mode I (K_Ic), mode II (K_IIc), and mode III (K_IIIc) of CuInTe₂, which was estimated from ideal stress-strain relations, was equal to 0.19, 0.37, and 0.33 MPa × m^1/2, respectively. The fracture toughness of CuInTe₂ was much lower than those of ZnO, but greatly higher than that of such layered materials as BiCuSeO and NaCo₂O₄. Additionally, relatively poor mechanical properties of CuInTe₂ can be strengthened through elemental substitution (e.g., S/Se doping in PbTe) [82], grain size reduction, crystal orientation regulation [83] or addition of nanosized secondary reinforcements [84], and so on.

6 Conclusion and Outlook

Recently, MDLCs have attracted an ever-increasing attention in TE field, and TE performance, measured in terms of ZT value, was significantly enhanced to 1.6 in 2017. The effective strategies including pseudocubic structure, highly efficient doping, spin entropy, high-entropy alloys, in situ displacement reaction, and mosaic nanostructure have been successfully utilized in improving TE properties of MDLCs. Furthermore, diamond-like TE module with only two pairs can provide an output power of 0.06 W under a temperature difference of 520 K. Despite the major advances in ZT and device have been achieved in MDLCs with good prospects, more efforts are still needed to promote the widespread use of MDLC materials in power generation applications. First, ZT value, especially average one, cannot be as good as that of mainstream TE materials, so it is necessary to further increase in ZT by new strategies, such as dislocation engineering [85], resonant levels [86], and nanoheterostructure [87]. Second, as ZT reported in n-type MDLC is still inferior to that of p-type ones (Fig. 7.1), obtaining high-performance n-type MDLC is also urgently needed in future research. Third, in addition to excellent TE properties, mechanical properties, thermal stability, and choosing contact layer, which are of equal importance in commercial applications, should be also investigated intensively for MDLC materials [88]. Finally, MDLCs, not only TE material but also light absorber, probably have good application in photovoltaic-thermoelectric hybrid device if Seebeck effect and photovoltaic effect could be incorporated together in a proper way [89, 90].