Cite this Article
Zhang Dan, Bai Hong-Chang, Li Zhi-Liang, Wang Jiang-Long, Fu Guang-Sheng, Wang Shu-Fang. Multinary diamond-like chalcogenides for promising thermoelectric application. Chinese Physics B, 2018, 27(4): 047206  
Permissions
Multinary diamond-like chalcogenides for promising thermoelectric application
Zhang Dan, Bai Hong-Chang, Li Zhi-Liang, Wang Jiang-Long, Fu Guang-Sheng, Wang Shu-Fang †
Hebei Key Laboratory of Optic-Electronic Information and Materials, College of Physics Science and Technology, Hebei University, Baoding 071002, China

 

† Corresponding author. E-mail: sfwang@hbu.edu.cn

Abstract

Thermoelectric (TE) materials have been considered as a strong candidate for recovering the waste heat from industry and vehicles due to the ability to convert heat directly into electricity. Recently, multinary diamond-like chalcogenides (MDLCs), such as CuInTe2, Cu2SnSe3, Cu3SbSe4, Cu2ZnSnSe4, etc., are eco-friendly Pb-free TE materials with relatively large Seebeck coefficient and low thermal conductivity and have aroused intensive research as a popular theme in the TE field. In this review, we summarize the TE performance and device development of MDLCs. The features of crystalline and electronic structure are first analyzed, and then the strategies that have emerged to enhance the TE figure of merits of these materials are illustrated in detail. The final part of this review describes the advance in TE device research for MDLCs. In the outlook, the challenges and future directions are also discussed to promote the further development of MDLCs TE materials.

PACS: ;72.20.Pa;;73.50.Lw;
Keyword:;thermoelectric;;multinary diamond-like chancogenides;;decoupling;;electrical and thermal transport properties;
1. Introduction

The ever-growing concern around the energy crisis and environmental pollution have motivated scientists to study and design alternative energy technologies.[1,2] Thermoelectric (TE) materials, which can directly convert heat into electricity or vice versa, have received a great deal of attention as a robust candidate for the waste heat recovery and solid state refrigeration. Its efficiency depends on the figure of merit, , where S is the Seebeck coefficient, σ is the electrical conductivity, T is the absolute temperature, and is the thermal conductivity. Large S, high σ, and low κ are desired for getting high performance TE material with a high zT value. However, the inherent coupling effects among the S, σ, and κe by the carrier concentration make the improvement of zT value extremely difficult.[3–5] Recently, new concepts (e.g., resonant levels,[6] band convergence,[7] multiple filling,[8] hierarchical architecture scattering,[9] and nanomagnets-embedding[10]) are continually put forward to decouple these interrelated parameters, and extremely high zT values are realized in leading materials of PbTe ( ) and CoSb3 ( ). However, hazardous elements or rare earth elements in these high efficiency TE materials will hamper their wide-scale commercial application. Hence, exploring alternative materials with eco-friendly and earth-abundant elements is one of the hottest areas for researchers in the TE field.[11–13]

Multinary diamond-like chalcogenide (MDLC), such as ternary I–III–VI2, I2–IV–VI3, I3–V–VI4 compounds, and quaternary I2–II–IV–VI4 compounds (Fig. 1(a)), features the diamond-like structure with deformed tetrahedron building blocks.[14] Most of diamond-like chalcogenides, for example, Cu(In,Ga)(S,Se)2,[15] are well recognized as promising candidates for thin film solar cells, photovoltaic devices, etc., while little attention has been paid to its TE performance. Although the possibilities of MDLCs to be thermoelectric materials have been considered historically,[16–18] their good TE properties at high temperature are not demonstrated due to the difficulties in measuring or synthesizing them. Up to 2009, Pb-free earth-abundant Cu2ZnSnSe4 compound with diamond-like structure has been discovered to possess relatively high zT value (∼0.9 at 850 K) after being substituted by Cu or In.[19,20] After that, many diamond-like chalcogenides have been investigated with great interest as promising TE materials and the development of MDLCs can be reflected from the time dependent zT value in Fig. 1(b).[19–67] The zT value up to 1.26 (at 673 K) could be obtained for Bi–Sn codoped Cu3SbSe4,[59] the zT value in the (Ag, In)-dual-doped Cu2SnSe3 sample approached 1.42 at 823 K,[48] and that of CuInTe2 could increase up to a value as high as 1.61 through integrating the anion substitution and nanoinclusions,[26] moreover, an output power of 0.06 W under a temperature difference of 520 K was obtained for MDLCs thermoelectric module,[35] which reveals that MDLCs with nontoxic components are very promising and even comparable to other state-of-the-art TE materials.

Fig. 1. (color online) (a) Different crystal structures of multinary diamond-like chalcogenides (MDLCs) derived from cubic zincblende structure; (b) timeline for thermoelectric bulk materials with diamond-like structures, showing data from Cu(In, Ga)Te2,[21–32] AgGaTe2,[33] AgInSe2,[34,35] CuFeS2,[36–44] Cu2SnS3,[45,46] Cu2GeSe3,[47] Cu2SnSe3,[48,49] Cu3SbS4,[50] Cu3SbSe4,[51–60] Cu2ZnSnS4,[61] Cu2ZnSnSe4,[19,20,62,63] Cu2ZnGeSe4,[64] Cu2CdSnSe4,[65] Cu2FeSnSe4,[66] and Cu2CoSnS4.[67]

In this review, we first analyze the crystalline and electronic structure properties of MDLCs briefly, and then the strategies for improving the TE properties of MDLCs are described in detail, including carrier concentration regulation, band structure engineering, the decrease of κL and synergistic regulation. Finally, challenges and directions for this type of semiconductor in future commercial applications are also discussed.

2. Crystalline and electronic band structure

MDLCs and their analogies can be taken as derivatives from cubic zincblende binary II–VI compounds (like ZnSe) through orderly substitution of II atoms with the same number of I + III, I2+IV, I3 +V or I2 + II + IV atoms (Fig. 1(a)). Most of MDLCs, including CuInS2, Cu2ZnSnS4, etc., crystallize into chalcopyrite or stannite structure; however, the Cu2SnS3-type semiconductor takes a monoclinic structure with two types of anion-centered tetrahedra rather than one type of anion-centered tetrahedron in the whole structure.[14] As the substituting atoms own different ionic radii and electronegativities, the distorted lattice with diverse bonding length and angle will appear in MDLCs,[19,48,51,65] which can block the phonons from being transported and result in the relatively low κ. For example, the κ of Cu2ZnSnSe4 (Fig. 2(b)) is quite low in the whole temperature range, and it nearly decreases to at high temperature, which is even comparable to the κ values of other TE materials with heavy constitute elements.[19,20] Meanwhile, the carrier mobility of MDLC will also decrease after substitution as compared with those of its binary counterparts.

Fig. 2. (color online) (a) Structure of Cu2ZnSnSe4 with alternative (Cu2Se4) and (ZnSnSe4) layers; (b) the temperature-dependent κL of Cu2ZnSnSe4;[20] [(c)–(e)] the orbital-decomposed band structures of Cu2ZnSnSe4.[73] Reproduced from Ref. [73] with permission from the Royal Society of Chemistry.

Owing to the similar characteristics in the crystalline structure, MDLCs have some similarities in electronic band structure. It should also be noted that the electronic structure for MDLCs was calculated very early,[68,69] and the results have become more and more accurate recently with the development of density functional theory.[70] For Cu-based ternary and quaternary chalcogenides, the valence band maximum (VBM) mainly contains a hybridization of Cu 3d orbitals and VI p orbitals because Cu has higher d orbital energy than the other cations, whereas the conduction band minimum (CBM) is primarily derived from the II/III/IV/V s orbital and VI p orbitals due to the lower s orbital energy of III/IV/V than the other cations.[14,70–72] For stannite Cu2ZnSnSe4 compound, the (Cu2Se4 tetrahedral slabs serve as the conduction pathway for holes and the insulating (ZnSnSe4) ones can work as a charge reservoir (Fig. 2). One can also note that the valence band is relatively flat primarily due to the intrinsic localization of the Cu-3d electrons.[73] Moreover, the multiple bands at VBM can be utilized to yield a large Seebeck coefficient at high carrier concentration by band convergence, and result in the improved TE performance.[74]

3. Strategies to improve TE performance of MDLCs
3.1. Carrier concentration optimization

The intrinsic zT values for most of MDLCs are still very low primarily because their carrier concentrations are considerably lower than their corresponding optimal carrier concentrations, and thus resulting in the poor electrical transport properties. Thus, the common method of elemental doping is widely employed to increase carrier concentrations in MDLCs for achieving a high zT.[75] Importantly, the choice of doping site does lead to a difference in carrier mobility for heavily doped semiconductor, and high mobility is expected to be obtained when dopants are on the sub-lattice which makes less contributions to the formation of the charge-conducting band in principle, which is termed ‘modulation doping on the atomic level (MDAL)’ by Wang in the TE field.[76] High mobility is experimentally achieved from the anion-doped sample while the cation dopant is distinctly inferior in PbSe. We should point out that the enhanced mobility by this method can be observed when the band structure does not change (at least not significantly) with dopant. According to this concept, II/III/IV/V sites are reasonable doping places to increase the carrier concentration in p-type MDLC due to the above band structure features that the I–IV bonds dominate the p-type electrical transport while other elements (i.e., II/III/IV/V) exhibit weak contributions to the conductive framework. In experiment, substituting Sn with Indium was used as an effective guide to improve the carrier concentration in diamond-like Cu2SnSe3 compound and the generated hole would smoothly transport through three-dimensional (3D) Cu–Se conductive framework, thereby greatly enhancing the electrical properties.[48] Meanwhile, the copper deficiencies in Cu2GeSe3 are not effective in increasing the electrical properties.[77] Other investigations of GeSb/SnSb in Cu3SbSe4,[51,52,55–57] ZnM in CuMTe2 (M=Ga, In),[78–80] CuZn/InSn in Cu2ZnSnSe4,[19,20] etc. also demonstrate the effectiveness of doping at II/III/IV/V sites in the enhancement of electrical properties and thus highly improving zT s. In addition, the possible presence of donor–acceptor defect pairs in MDLCs, such as ( ) defect pairs in Cu2ZnSn (X=S, Se),[81] will make the conducting mechanisms more complicated. For example, defect pairs ( and are introduced into zinc-doped CuInTe2,[79] and the presence of antisite defect defect will cancel the p-type carrier concentration and increase the ionized impurity scattering centers, which is harmful to the increase of electrical conductivity.

3.2. Band structure engineering

Previous studies indicated that the high band degeneracy (i.e., high and spin entropy are effective for decoupling the Seebeck coefficient and electrical conductivity, and thus improving the power factor significantly.[61,75] The above-mentioned approaches toward high zT in MDLCs will be discussed below.

3.2.1. Band convergence

(i) Pseudocubic approach

A high Nv number, which is normally observed in a crystal structure with high symmetry, is desired for gaining both a large Seebeck coefficient and high electrical conductivity. Snyder et al. successfully implemented the band-convergence scheme in Se-alloyed PbTe for the first time and attained an excellent zT value of ∼1.8.[7] This scheme has also been successively applied in other highly symmetric semiconductors (e.g., Mg2Si, SnTe, half–Heusler, and CoSb3).[82–85] In 2014, Zhang et al. expanded it to chalcopyrite with low symmetry, and devised a ‘pseudocubic approach’ (Fig. 3), which has a cubic cation framework embedded in a noncubic distorted anion framework, to attain high power factor with high band convergence. They concluded a ‘unity-η rule’ ( ), which reveals that changing η toward unity would lead to band convergence for good thermoelectricity, and it was also experimentally verified that pseudocubic structured CuInTe2 possesses the significantly enhanced zT value via Ag or Ga doping. As reported by Zeier et al. the pseudocubic approach truly works in attaining an increase of the power factor and zT in stannite Cu2ZnGeSe4 via Cu and Fe co-doping.[86] However, this promising approach is limited to low symmetry material with an ideal band gap and a low κL.[3]

Fig. 3. (color online) Schematic diagrams illustrating the pseudocubic band-convergence, showing (a) crystal structure and schematic electronic bands of cubic zinc blende structure, and (b) crystal structure and schematic electronic bands of ternary chalcopyrites. The c and a are the lattice constants. is a nondegenerate band, and is a doubly degenerate band. is the crystal field-induced energy split at the top of the and bands. The pseudocubic structure has a cubic cation framework with embedded in a noncubic distorted anion framework.[23] Reproduced from Ref. [23] with permission from WILEY-VCH.

(ii) Highly-efficient-doping (HED)

As for TE materials possessing intrinsically high band degeneracy, heavy doping by guest atoms to reach the optimal carrier concentration will more or less cause the degenerate bands to destruct, leading to both the Seebeck coefficient and the power factor decreasing. Thus, selecting highly efficient dopants which can provide optimal carrier concentration at relatively low doping content, should be a feasible attempt to improve the electrical conductivity while maintaining the large Seebeck coefficient with little divergence of degenerate bands.[60] As illustrated in Fig. 4, the serious separation of the degenerate bands in non-HED can be relieved by HED when the same level of carrier concentration is reached. One could clearly note that the advantage of HED by Al/Ga/In comparing with the non-HED by Ge/Sn in high-band-degeneracy Cu3SbSe4 with 0.5% of Al/Ga/In substitution could provide comparable or even higher carrier concentration than that of 1% Ge-doping and its characteristic also has little deviation from the Pisarenko curve due to the nearly non-changed band structure. Meanwhile, the high doping concentration (e.g., 2% Ge, 5% Sn) makes the Seebeck coefficient deteriorate rapidly due to the severe split of the degenerate bands. Ultimately, the zT of the sample with HED possesses a higher value than that of the non-HED at 623 K.[60]

Fig. 4. (color online) (a) Schematic diagram illustrating the maintenance of highly degenerate electronic bands of the matrix by highly efficient doping (HED), and (b) curve of S versus carrier concentration showing that the nearly non-changed band structure can be obtained via HED (i.e., Al, Ga, In doping)[60] in Cu3SbSe4 with high band degeneracy. Reproduced from Ref. [6] with permission from the American Chemical Society.
3.2.2. Spin entropy

It was suggested that the accompanying extra entropy in magnetic ions with degeneracy of electronic spin configuration in real space, such as transition metal ions with unpaired 3d electrons, should play a key role in obtaining large S at high σ in cobalt oxide.[61,87,88] This striking fact has encouraged many researchers to explore the possibility of substitution using magnetic ions to improve the TE performances of the other materials. Yao et al. firstly investigated the TE properties of magnetic-ion-substituted chalcopyrite in 2011 and observed the simultaneous optimizing of all three TE parameters in Mn-substituted CuInSe2.[89] The Mn-doping moved the Fermi level (Ef) downward into the region of the valence band, and the Mn 3d level hybridization with the Se 4p level resulted in the finite DOS in the proximity of the Ef, enabling the considerable enhancing of S and σ for CuInSe2. The introduced Mn–Se networks could provide an additional conduction pathway for charge carriers. The increased local disorder in the crystal lattice by Mn substitution could reduce κL, which was also verified by Raman spectroscopy. However, the contribution of the spin entropy has been neglected in the work of Yao et al. In 2013, Xiao et al. reported that magnetic ions could also decouple the strongly interrelated three TE parameters in Cu2ZnSnS4 (Fig. 5(a)).[61] More importantly, the performed electron paramagnetic resonance (EPR) experiments implied that Ni2+ doping truly brought extra spin into Cu2ZnSnS4 and therefore the value of S is remarkably enhanced. After the magnetic ions (e.g., Cobalt) are fully substituted into Cu2ZnSnS4, the electronic structure and phonon structure will undergo significant change due to the strong hybridization between the 3d states of magnetic ions and S 3p states, and therefore the TE properties are improved (Figs. 5(b)–5(d)). The distinct band structure feature of Cu2CoSnS4 with magnetic component shows an extremely high zT of 0.8 achieved in this cheap and eco-friendly sulphide, which is due to the improved electrical properties by enhancing the effective mass and the depressed lattice thermal conductivity by weakening the covalent-bonding.[67]

Fig. 5. (color online) (a) Schematic diagrams illustrating various electron spin and phonon scatterings in quaternary magnetic nanocrystals and (b) comparison of the zT among Cu2XSnS4 nanocrystals.[61] Reproduced from Ref. [61] with permission from the Royal Society of Chemistry, (c) electronic and (d) phonon band structure for Cu2CoSnS4.[67] Reproduced from Ref. [67] with permission from Elsevier.

Other researches for improving TE performances of Ni-doped Cu3SbS4, Mn-doped CuGaTe2 and cobalt-doped Cu2SnS3,[46,90,91] also proved the effectiveness and good-transplanting of magnetic ion doping strategy in MDLCs. However, most of researches usually focused on explaining the enhancement in S by an additional spin entropy effect or the participation of magnetic-ion-3d states near the Ef in the electronic transport.[46,61,89–92] The above two factors, which may interconnect with each other by magnetic-ion, should be considered together and also clarify the primary cause for enhancing the value of S at different magnetic ion content (e.g., in 2.6% Ni-doped Cu2ZnSnS4, Cu2CoSnS4, and CuFeS2.[36,61,67] Additionally, as the spin-disorder scattering in magnetic ion doping more or less hampers the mobility of carriers,[93] its influence should also be taken into consideration, especially at low temperature. If the reduction in carrier mobility overweighs the increase in S, no enhancement of power factor may appear. Therefore, the relationship between magnetism and TE property still needs further exploring.

3.3. Reduction of lattice thermal conductivity
3.3.1. Point-defect scattering

It is known that the introduction of point defects is an effective approach to reducing the value of κL by enhancing the phonon scattering due to the mass contrast and strain field fluctuation between the host and guest atoms. According to the Debye–Callaway–Klemens model, the high concentration of defects possessing large mass difference and strain fluctuations can maximize the scattering parameter (Γ) and thus achieve the lowest κL.[75] Pei et al. observed that AgCu substitutional defects reduced the κL of CuGaTe2 more obviously than ZnGa/InGa defects at the same concentration, which is due to the larger mass contrast.[94] However, the maximum mass contrast ( ), which can be maximized when the dominant point defects are vacancies, seems unattainable by conventional doping but is possible when the guest or host atom represents a disordered vacancy. Thus, the compounds with inherent vacancy, such as In2Te3 (one-third of its cationic sites are vacant) and their analogues, are selected to greatly depress the κL of the parent phase by forming a solid solution.[75] It is found that the value of κL of CuGaTe2 alloyed with the proper amount of In2Te3 or Ga2Te3 was reduced more remarkably than that of the same quantity of substitutional defects (Fig. 6(a)), which could be attributed to the strong vacancy phonon scattering induced by guesting vacancies.[95] Additionally, a detailed investigation of the room temperature thermal conductivity on anion substitution series Cu2ZnGeSe4−xSx revealed a reduction of the κL of 42% (Fig. 6(b)), and the calculations quantitively revealed that mass contrast (reduction of 34%) has a greater influence than strain contrast (reduction of 8%).[96]

Fig. 6. (color online) (a) Room temperature cation defect concentration (nCD)-dependent normalized lattice thermal conductivity (κL) for CuGaTe2 solid solutions with (red) and without (black) vacancies.[95] Reproduced from Ref. [95] with permission from the Royal Society of Chemistry. (b) Plots of room-temperature κL and modeled κL against the molar fraction of selenium of the complete solid solution series Cu2ZnGeSe4−xSx with x = 0–4.[96] Reproduced from Ref. [96] with permission from the American Chemical Society.
3.3.2. Nanostructure engineering

(I) Nanocomposite

The proposed concept of low-dimensional TE materials (e.g. quantum dots, nanowires and superlattices systems) in the 1990s has revealed that the simultaneous optimization of the power factor and κL is possible in nanostructures.[97,98] After that, many experimental studies on nanostructured bulk materials, such as (Bi,Sb)2Te3, PbTe, Si80Ge20,[99–101] have also proved that the nanocomposite approach (i.e., reducing the grain size to nanometers or using nanoparticles) is indeed effective in increasing zT of 3D bulk material due to the reduction in the κL by the selective scattering of phonons or the enhancement in the Seebeck coefficient via preferential scattering of low-energy electrons (i.e., energy filtering).[12] Consequently, the nanocomposite approach is also widely adopted to achieve the expected high zT in MDLC, although it probably causes the electrical conductivity to decrease by the deterioration of carrier mobility. Liu et al. reported a solution-based scalable synthesis approach to produce several grams of monodisperse Cu3SbSe4 nanocrystals (Fig. 7),[59] and found the nano-crystalline Cu3SbSe4 had better TE performance than the micro-crystalline compound with the same constituents because the value of κL greatly decreased. Furthermore, an extremely high zT value (i.e., 1.26) was obtained in (Sn, Bi)-dual-doped Cu3SbSe4 by this improved approach. Liang et al. found that the CuFeS2 nanocrystals exhibited a significantly higher S than bulk chalcopyrite material due to quantum confinement and a tremendously low κL, due to the increased phonon scattering at the nanocrystal grain boundary, thus zT is enhanced by 77 times.[36] Chen et al. found that Cu-doped Cu2ZnSnSe4 nanocrystals prepared by hot-injection had a zT as high as 0.70 at 450 °C, which is significantly higher than that by the traditional method.[63] The obtained high TE performance in Cu2SnSe3 nanocrystal also shows the effectiveness and good prospects of the nanocomposite approach in obtaining efficient TE materials.[13] However, the problems of aggregation and growth during nanocrystals densification become challenges to further enhancing the TE performance by this approach.

Fig. 7. (color online) (a) Representative TEM micrographs of Cu3SbSe4 nanocrystallines, and the corresponding HRTEM images, electron diffraction patterns and EELS analysis results.[59] Reproduced from Ref. [59] with permission from the Royal Society of Chemistry. [(b) and (c)] Temperature-dependent κ and zT values for micro- and nano-crystalline Cu3SbSe4.[56,59]

(II) Nanoinclusion

Nanoinclusion strategies by introducing a nano-sized second phase are effectively utilized to obtain high performance TE materials (e.g., CoSb3, PbTe, SnTe)[104–106] due to the depressed κL by additional phonon scattering or the increased S through energy filtering. Most of the research of nanoinclusion about MDLCs focuses on the CuGaTe2/CuInTe2 type materials currently. In the case of CuGaTe2, the incorporation of graphite nanosheets could enhance the S and power factor due to the energy filtering effect, and also cause the value of κL to decrease below 750 K because of the increased interface scattering. The large zT value increased up to 0.93 at 873 K in CuGaTe2 composite, which had about 21% improvement on pure CuGaTe2.[107] However, the introduction of nanophase Cu2Se into CuGaTe2 reduced the value of S due to the increase of carrier concentration.[32] Benefiting from the greatly reduced value of κL, the final zT approached to 1.2 at 834 K for the CuGaTe2/3vol% Cu2Se sample. By introducing a proper amount of grapheme into CuInTe2, Chen et al. also observed that the κL successfully decreased and also the value of zT was improved.[108]

4. Synergistic regulation
4.1. Entropy engineering

Recently, high entropy alloy (HEA), which consists of multiple principal elements in an equimolar manner or in a close to equimolar ratio manner, has been proposed as a novel type of alloy with intriguing properties in structure and function.[109,110] Due to the chemical complexity (e.g., large atomic size differences) and high configuration entropy, HEA may significantly introduce extra phonon scattering and also offer a new way to adjust electrons, and resultantly affect the electrical and thermal properties. The HEA was introduced into the context of TE by Shafeie et al. as a potential candidate for future TE material.[111] In this scenario, Liu et al. devised a strategy of using entropy as the global gene-like performance indicator, which shows how multicomponent thermoelectric materials with high entropy could be designed (Fig. 8).[29] They first defined a parameter δ in a novel elastic model by shear modulus, lattice parameter and the mismatch in the atomic radius as a criterion for judging the feasibility of forming a complete solid solution with high entropy. Guided by the δ criterion, they screened out several multicomponent TE material systems. Finally, the screened candidates were experimentally tested and the 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 using multicomponent crystal structures works as an effective guide to greatly improve the TE performance through the simultaneous optimization of electrical and thermal transport properties, i.e., to lower κL by the presence of local mass and strain fluctuations and to improve the S by enhancing the crystal symmetry. Lu et al. also observed that PbSnTeSe HEAs possess quite low κL, for example, less than at 300 K, and its TE properties could be further enhanced by slight composition alteration.[112] Although this strategy holds promise for reducing the value of κL, the simultaneously observed improvement of S is limited to some TE materials with increased crystal symmetry.

Fig. 8. (color online) (a) Schematic diagrams of the lattice framework in multicomponent materials compared with an ordinary binary compound. (b) Schematic diagram of the entropy engineering with multicomponent TE materials. (c) Maximum confgurational entropy (in units of kB per formula unit) as a function of material solubility parameter δ for given multicomponent TE materials, where n is the number of components. (d) Plots of maximum zT versus confgurational entropy in Cu2(S/Se/Te)-, (Cu/Ag)(In/Ga)Te2-, and Cu8Ge(Se/Te)6-based multicomponent TE materials.[26] Reproduced from Ref. [26] with permission from WILEY-VCH.
4.2. In-situ displacement reaction

Nano-oxide additives are commonly used to reduce κL by introducing additional phonon scattering into TE materials, and the oxides are usually stable in the host material because of their intrinsically strong bonding.[113–119] In 2016, Luo et al. first discovered the in-situ displacement reaction between the ZnO additive and the CuInTe2 matrix at 714 K by differential scanning calorimeter (DSC) and transmission electron microscopy (TEM) as displayed in Figs. 9(a)–9(c),[26] and this reaction is partially understandable in the bonding strength respect: the final product of In2O3 possessing stronger bonds is more stable than ZnO due to the higher melting point (Tm of In2O3 ∼2273 K and Tm of ZnO ∼2248 K). Thus, it has two significant effects on tuning and optimizing electrical and thermal transport properties, i.e., to lower the thermal conductivity via the extra phonon scattering from In2O3 nanoinclusions, and to improve the electrical conductivity by the formation of point defects. Then, an 80% enhanced zT value of 1.44 was obtained in the 1-wt% ZnO added CuInTe2 sample. Moreover, by integrating the anion substitution (i.e., Sb doping) and in-situ displacement reaction, a high power factor of and unprecedented zT of 1.61 at 823 K were achieved in the CuInTe2 based TE material (Fig. 10(d)). In addition, the simultaneous regulation of electrical and thermal transport properties in TiO2 added CuInTe2 also proved the effectiveness of in-situ displacement reaction by other oxides in improving TE performance.[25]

Fig. 9. (color online) (a) DSC curves of CuInTe2+2.0-wt% ZnO additives before and after hot press process. (b) Highly magnified TEM images of CuInTe2 matrix with dispersed In2O3 nanoparticles. Reproduced from Ref. [25] with permission from Elsevier. (c) Scheme of in-situ displacement reaction. (d) Stepping increase of zT at 823 K in the CuInTe2 samples by Sb substitution, ZnO adding, and combined Sb substitution, and ZnO adding.[25,26] Reproduced from Ref. [26] with permission from WILEY-VCH.
Fig. 10. (color online) (a) Schematic diagram of the structure of the fabricated diamond-like module, and b) plots of output voltage (black) and power (blue) versus current for TE module based on diamond-like materials. Temperature differences between the cold side and the hot side of the module are 420, 470, and 520 K, respectively.[35]
5. TE devices

Since research on the TE properties of MDLCs is still in the early stage, most of the attention is paid to enhancing the zT s and the feasibility to fabricate an efficient TE device based on MDLCs is still an open question. In 2017, researchers started to address this inevitable question and finally opened the door for practical applications of MDLCs TE materials. Due to the obtained high TE performance in (Sn, Bi)-codoped nanocrystalline Cu3SbSe4 sample, Liu et al. used them to fabricate an innovative ring-shaped TE generator with a single p-leg, which could be used to recover the waste heat from exhaust gas pipes of vehicles.[59] The results showed that an open circuit voltage near 20 mV was obtained for a single TE element when the temperature gradient was maintained at 160 K. Then, the first TE module including both n and p-type MDLCs was fabricated by Qiu et al. as shown in Fig. 10.[35] The TE module using Ag0.9Cd0.1InSe2 and Cu0.99In0.6Ga0.4Te2 as respective n- and p-type legs presented a maximum output power of 0.06 W under a temperature difference of 520 K, clearly showing that MDLCs can be a potential candidate for TE applications.

6. Conclusions and perspectives

In this review, we mainly discuss the features in crystal structure and electronic structure, and summarize the effective strategies to enhance zT s for MDLCs. Due to the unique diamond-structure-accompanied properties: flat valence band and large local distortion, MDLCs usually possess both relatively high Seebeck coefficients and low thermal conductivities. A zT value as high as 1.6 is achieved in Pb-free MDLC by entropy engineering or by in-situ displacement reaction, thus highlighting the promising prospect of MDLCs for TE applications. However, more efforts are still needed to promote the practical TE applications of MDLCs. On the one hand, the zT value especially the average one cannot be so good as that of highly efficient PbTe-based TE material, so it is of great importance to further increase the zT s by new strategies.[9,120–122] Recently, it was demonstrated that dislocation arrays, which effectively scatter phonons and maintain high carrier mobility could concurrently optimize the electrical and thermal transport properties, leading to a great improvement of both power factor and zT in CoSb3- or (Bi,Sb)2Te3-based TE materials.[123,124] Pei et al. also noted that the dense in-grain dislocations indeed took the major responsibility for the decrease of κL through scattering the remaining mid-frequency phonons thereby resulting in the extremely low κL approaching to the amorphous limit.[125] Hence, dislocation engineering is a good direction for further obtaining advanced MDLC TE materials. Besides, the resonant levels play an important role in enhancing both the average power factor and zT value in SnTe, which is mainly due to the increased S.[126,127] Nanoheterostructure-based nanocomposites (e.g., PbTe–Ag2Te, PbTe–BiSbTe) present both the improved power factor and the depressed κL because of minority carrier blocking.[128,129] Can the resonant levels also increase S or does the nanoheterostructure block the minority carrier in MDLC?

The poor TE properties in n-type MDLCs also hamper the TE applications of this promising typed material, because a TE device requires both a high-performance n-type leg and a high-performance p-type leg. In the literature, the highest zT in reported n-type MDLCs is still inferior to that of the p-type one, such as 0.126 at 450 K for CuFeInTe3,[130] 0.264 at 500 K for CuFeS2, and 1.1 at 900 K for AgInSe2.[35,36] Meanwhile, it seems difficult to change the intrinsic p-type conduction into an n-type one for most of MDLCs with high TE performance. For instance, zinc-doping heavily at the copper site could not make the n-type Cu3SbSe4 available.[131] The origin of p-type electrical transports in most of MDLCs might be linked to the presence of intrinsic Cu vacancies.[132] Thus, discovering and realizing n-type MDLCs with high TE performance is urgently needed in the future research. In addition, the stability investigation of MDLCs under thermal cycling or large current, which is of great importance in practical applications, should also be focused on.[133]

7. Uncited references

[102,103]

Reference
[1] Bell L E 2008 Science 321 1457
[2] DiSalvo F J 1999 Science 285 703
[3] He J Tritt T M 2017 eaak9997 Science 357 10.1016/j.mtphys.2017.06.003
[4] Shuai J Mao J Song S Zhang Q Chen G Ren Z 2017 Mater. Today Phys. 10.1016/j.mtphys.2017.06.003
[5] Xiao C Li Z Li K Huang P Xie Y 2014 Acc. Chem. Res. 47 1287
[6] Heremans J P Jovovic V Toberer E S Saramat A Kurosaki K Charoenphakdee A Yamanaka S Snyder G J 2008 Science 321 554
[7] Pei Y Shi X LaLonde A Wang H Chen L Snyder G J 2011 Nature 473 66
[8] Shi X Yang J Salvador J R Chi M Cho J Y Wang H Bai S Yang J Zhang W Chen L 2011 J. Am. Chem. Soc. 133 7837
[9] Biswas K He J Blum I D Wu C I Hogan T P Seidman D N Vinayak P D Kanatzidis M G 2012 Nature 489 414
[10] Zhao W Liu Z Sun Z Zhang Q Wei P Mu X Zhou H Li C Ma S He D Ji P Zhu W Nie X Su X Tang X Shen B Dong X Yang J Liu Y Shi J 2017 Nature 549 247
[11] Qiu P Shi X Chen L 2016 Energy Storage Materials 3 85
[12] Zhang D Yang J Jiang Q Fu L Xiao Y Luo Y Zhou Z 2016 J. Mater. Chem. 4 4188
[13] Zhang D Yang J Jiang Q Fu L Xiao Y Luo Y Zhou Z 2016 Mater. Des. 98 150
[14] Xi L Zhang Y B Shi X Y Yang J Shi X Chen L D Zhang W 2012 Phys. Rev. 86 155201
[15] Contreras M A Egaas B Ramanathan K Hiltner J Swartzlander A Hasoon F Noufi R 1999 Prog. Photovoltaics Res. Appl. 7 311
[16] Mousa N H El-bahay M M Abdelghany A Abou El Ela A H 1996 Phys. Chem. Liq. 33 113
[17] Kuhn B Kaefer W Fess K Friemelt K Turner C Wendl M Bucher E 1997 Phys. Stat. Sol. 162 661
[18] Endo S Sudo I Irie T 1971 Jpn. J. Appl. Phys. 10 218
[19] Liu M L Huang F Q Chen L D Chen I W 2009 Appl. Phys. Lett. 94 202103
[20] Shi X Y Huang F Q Liu M L Chen L D 2009 Appl. Phys. Lett. 94 122103
[21] Liu R Xi L Liu H Shi X Zhang W Chen L 2012 Chem. Commun. 48 3818
[22] Li Y Meng Q Deng Y Zhou H Gao Y Li Y Yang J Cui J 2012 Appl. Phys. Lett. 100 231903
[23] Zhang J Liu R Cheng N Zhang Y Yang J Uher C Shi X Chen L Zhang W 2014 Adv. Mater. 26 3848
[24] Carr W D Morelli D T 2015 J. Alloys Compd. 630 277
[25] Luo Y Yang J Jiang Q Li W Xiao Y Fu L Zhang D Zhou Z Cheng Y 2015 Nano Energy 18 37
[26] Luo Y Yang J Jiang Q Li W Zhang D Zhou Z Cheng Y Ren Y He X 2016 Adv. Energy Mater. 6 1600007
[27] Liu R Qin Y Cheng N Zhang J Shi X Grin Y Chen L 2016 Inorg. Chem. Front. 3 1167
[28] Luo Y Jiang Q Yang J Li W Zhang D Zhou Z Cheng Y Ren Y He X Li X 2017 Nano. Energy 32 80
[29] Liu R Chen H Zhao K Qin Y Jiang B Zhang T Sha G Shi X Uher C Zhang W Chen L 2017 Adv. Mater 29 1702712
[30] Plirdpring T Kurosaki K Kosuga A Day T Firdosy S Ravi V Snyder J F Harnwunggmoung A Sugahara T Ohishi Y Muta H Yamanaka S 2012 Adv. Mater 24 3622
[31] Cui J Li Y Du Z Meng Q Zhou H 2013 J. Mater. Chem. 1 677
[32] Zhang J Qin X Li D Xin H Song C Li L Zhu X Wang Z Guo G Wang L 2014 J. Mater. Chem. 2 2891
[33] Yusufu A Kurosaki K Kosuga A Sugahara T Ohishi Y Muta H Yamanaka S 2011 Appl. Phys. Lett. 99 061902
[34] Ying P Z Zhou H Gao Y L Li Y Y Li Y P Lian X L Cui J L 2012 Key Eng. Mate. 519 188
[35] Qiu P Qin Y Zhang Q Li R Yang J Song Q Tang Y Bai S Shi X Chen L 2017 Adv. Sci. 4 1700727
[36] Liang D Ma R Jiao S Pang G Feng S 2012 Nanoscale 4 6265
[37] Tsujii N Mori T 2013 Appl. Phys. Express 6 043001
[38] Li J Tan Q Li J F 2013 J. Alloys Compd. 551 143
[39] Li Y Zhang T Qin Y Day T Jeffrey S G Shi X Chen L 2014 J. Appl. Phys. 116 203705
[40] Berthebaud D Lebedev O I Maignan A 2015 J. Materiomics. 1 68
[41] Carr W D Morelli D T 2016 J. Electronic Mater. 45 1346
[42] Lefèvre R Berthebaud D Mychinko M Y Lebedev O I Mori T Gascoin F Maignan A 2016 RSC Adv. 6 55117
[43] Xie H Su X Zheng G Zhu T Yin K Yan Y Uher C Kanatzidis M G Tang X 2017 Adv. Energy Mater. 7 1601299
[44] Xie H Su X Yan Y Liu W Chen L Fu J Yang Y Uher C Tang X 2017 NPG Asia Mater 9 e390
[45] Tan Q Sun W Li Z Li J F 2016 J. Alloys Compd. 672 558
[46] Zhao H Xu X Li C Zhang R Huang R Lv Y Li D Hu X Pan L Wang Y 2017 J. Mater. Chem. 5 23267
[47] Cho J Y Shi X Salvador J R Meisner G P Yang J Wang H Wereszczak A A Zhou X Uher C 2011 Phys. Rev. 84 085207
[48] Shi X Xi L Fan J Zhang W Chen L 2010 Chem. Mater. 22 6029
[49] Li Y Liu G Cao T Liu L Li J Chen K Li L Han Y Zhou M 2016 Adv. Funct. Mater. 26 6025
[50] Chen K Du B Bonini N Weber C Yan H Reece M J 2016 J. Phys. Chem. 120 27135
[51] Yang C Huang F Wu L Xu K 2011 J. Phy. D: Appl. Phys. 44 295404
[52] Skoug E J Cain J D Morelli D T 2011 Appl. Phys. Lett. 98 261911
[53] Li X Y Li D Xin H X Zhang J Song C J Qin X Y 2013 J. Alloys. Compd. 561 105
[54] Li D Li R Qin X Y Zhang J Song C J Wang L Xin X 2013 CrystEngComm 15 7166
[55] Li D Li R Qin X Y Song C J Xin H X Wang L Zhang J Guo G L Zou T H Liu Y F Zhu X G 2014 Dalton Trans. 43 1888
[56] Wei T R Wang H Gibbs Z M Wu C F Snyder G J Li J F 2014 J. Mater. Chem. 2 13527
[57] Chang C H Chen C L Chiu W T Chen Y Y 2017 Mater. Lett. 186 227
[58] Zhang D Yang J Jiang Q Zhou Z Li X Ren Y Xin J Basit A He X Chu W Hou J 2017 J. Alloys Compd. 724 597
[59] Liu Y García G Ortega S Cadavid D Palacios P Lu J Ibáñez M Xi L Roo J D López A M Martí-Sánchez S Cabezas I Mata M Luo Z Dun C Dobrozhan O Carroll D L Zhang W Martins J Kovalenko M V Arbiol J Noriega G Song J Wahnón P Cabot A 2017 J. Mater. Chem. 5 2592
[60] Zhang D Yang J Jiang Q Zhou Z Li X Xin J Basit A Ren Y He X Chu W Hou J 2017 ACS Appl. Mater. Interfaces 9 28558
[61] Xiao C Li K Zhang J Tong W Liu Y Li Z Huang P Pan B Su H Xie Y 2014 Mater Horiz. 1 81
[62] Dong Y Wang H Nolas G S 2014 Phys. Status Solidi-RRL 8 61
[63] Chen D Zhao Y Chen Y Wang Y Zhou J Liang Z 2015 ACS Appl. Mater. Interfaces 7 24403
[64] Zeier W G LaLonde A Gibbs Z M Heinrich C P Panthöfer M Snyder G J Tremel W 2012 J. Am. Chem. Soc. 134 7147
[65] Liu M L Chen I W Huang F Q Chen L D 2009 Adv. Mater. 21 3808
[66] Song Q Qiu P Hao F Zhao K Zhang T Ren D Shi X Chen L 2016 Adv. Electron. Mater 2 1600312
[67] Zhang D Yang J Jiang Q Zhou Z Li X Xin J Basit A Ren Y He X 2017 Nano Energy 36 156
[68] Jaffe J E Zunger A 1983 Phys. Rev. 28 5822
[69] Jaffe J E Zunger A 1984 Phys. Rev. 29 1882
[70] Zhang Y Xi L Wang Y Zhang J Zhang P Zhang W 2015 Comp. Mater. Sci. 108 239
[71] Chen S Gong X G Walsh A Wei S H 2009 Appl. Phys. Lett. 94 041903
[72] Zhang Y Zhang J Gao W Abtew T A Wang Y Zhang P Zhang W 2013 J. Chem. Phys. 139 184706
[73] Zou D Nie G Li Y Xu Y Lin J Zheng H Li J 2015 RSC Adv. 5 24908
[74] Zhao L D Tan G Hao S He J Pei Y Chi H Wang H Gong S Xu H Dravid V P Uher C Snyder G J Wolverton C Kanatzidis M G 2016 Science 351 141
[75] Tan G Zhao L D Kanatzidis M G 2016 Chem. Rev. 116 12123
[76] Wang H Cao X Takagiwa Y Snyder G J 2015 Mater. Horiz. 2 323
[77] Huang T Yan Y Peng K Tang X Guo L Wang R Lu X Zhou X Wang G 2017 J. Alloys Compd. 723 708
[78] Luo P You L Yang J Xing J Zhang J Wang C Zhao X Luo J Zhang W 2017 Chin. Phys. 26 097201
[79] Yang J Chen S Du Z Liu X Cui J 2014 Dalton Trans. 43 15228
[80] Shen J Chen Z Lin S Zheng L Li W Pei Y 2016 J. Mater. Chem. 4 209
[81] Chen S Yang J H Gong X G Walsh A Wei S H 2010 Phys. Rev. 81 245204
[82] Liu W Tan X Yin K Liu H Tang X Shi J Zhang Q Uher C 2012 Phys. Rev. Lett. 108 166601
[83] Fu C Zhu T Pei Y Xie H Wang H Snyder G J Liu Y Liu Y Zhao X 2014 Adv. Energy Mater. 4 1400600
[84] Tang Y Gibbs Z M Agapito L A Li G Kim H S Nardelli M B Curtarolo S Snyder G J 2015 Nat. Mater. 14 1223
[85] Banik A Shenoy U S Anand S Waghmare U V Biswas K 2015 Chem. Mater. 27 581
[86] Zeier W G Zhu H Gibbs Z M Ceder G Tremel W Snyder G J 2014 J. Mater. Chem. 2 10189
[87] Wang Y Rogado N S Cava R J Ong N P 2003 Nature 423 425
[88] Limelette P Hébert S Hardy V Frésard R Simon C Maignan A 2006 Phys. Rev. Lett. 97 046601
[89] Yao J Takas N J Schliefert M L Paprocki D S Blanchard P E Gou H Mar A Exstrom C L Darveau S A Poudeu P F P Aitken J A 2011 Phys. Rev. 84 075203
[90] Chen D Zhao Y Chen Y Lu T Wang Y Zhou J Liang Z 2016 Adv. Electron. Mater. 2 1500473
[91] Ahmed F Tsujii N Mori T 2017 J. Mater. Chem. 5 7545
[92] Mori T 2017 Small 13 1702013
[93] Brown S R Toberer E S Ikeda T Cox C A Gascoin F Kauzlarich S M Snyder G J 2008 Chem. Mater. 20 3412
[94] Shen J Zhang X Chen Z Lin S Li J Li W Li S Chen Y Pei Y 2017 J. Mater. Chem. 5 5314
[95] Shen J Zhang X Lin S Li J Chen Z Li W Pei Y 2016 J. Mater. Chem. 4 15464
[96] Heinrich C P Day T W Zeier W G Snyder G J Tremel W 2014 J. Am. Chem. Soc. 136 442
[97] Hicks L D Dresselhaus M S 1993 Phys. Rev. 47 12727
[98] Dresselhaus M S Chen G Tang M Y Yang R G Lee H Wang D Z Ren Z F Fleurial J P Gogna P 2007 Adv. Mater 19 1043
[99] Poudel B Hao Q Ma Y Lan Y Minnich A Yu B Yan X Wang D Muto A Vashaee D Chen X 2008 Science 320 634
[100] Finefrock S W Zhang G Bahk J H Fang H Yang H Shakouri A Wu Y 2014 Nano Lett. 14 3466
[101] Joshi G Yan X Wang H Liu W Chen G Ren Z 2011 Adv. Energy Mater 1 643
[102] Jiang Q Yang J Liu Y He H 2016 J. Adv. Dielectr. 6 1630002
[103] Ibáñez M Cadavid D Anselmi-Tamburini U Zamani R Gorsse S Li W López A M Morante J R Arbiol J Cabot A 2013 J. Mater. Chem. 1 1421
[104] Peng J Fu L Liu Q Liu M Yang J Hitchcock D Zhou M He J 2014 J.Mater. Chem. 2 73
[105] Biswas K He J Blum I D Wu C I Hogan T P Seidman D N Dravid V P Kanatzidis M G 2012 Nature 489 414
[106] Moshwan R Yang L Zou J Chen Z G 2017 Adv. Funct. Mater 27 1703278
[107] Zhang J Qin X Li D Liu Y Li Y Song C Xin H Zhu X 2016 Appl. Phys. Lett. 108 073902
[108] Chen H Yang C Liu H Zhang G Wan D Huang F 2013 CrystEngComm 15 6648
[109] Li Z Pradeep K G Deng Y Raabe D Tasan C C 2016 Nature 534 227
[110] Ye Y F Wang Q Lu J Liu C T Yang Y 2016 Mater. Today 19 349
[111] Shafeie S Guo S Hu Q Fahlquist H Erhart P Palmqvist A 2015 J. Appl. Phys. 118 184905
[112] Zhao F Wang H Wu Y Liu X Lu Z 2017 Mater Res. Lett. 5 187
[113] Moure A Rull-Bravo M Abad B Del Campo A Rojo M M Aguirre M H Jacquot A Fernandez J F Martin-Gonzalez M 2017 Nano Energy 31 393
[114] Jiang Q Yang J Xin J Zhou Z Zhang D Yan H 2017 J. Alloys Compd. 694 864
[115] Cederkrantz D Farahi N Borup K A Iversen B B Nygren M Palmqvist A E C 2012 J. Appl. Phys. 111 023701
[116] Zhou Z Yang J Jiang Q Zhang D Xin J Li X Ren Y He X 2017 J. Am. Ceram. Soc. 100 5723
[117] Dou Y C Qin X Y Li D Li L L Zou T H Wang Q Q 2013 J. Appl. Phys. 114 044906
[118] Nandihalli N Guo Q Gorsse S Khan A U Mori T Kleinke H 2016 Eur. J. Inorg. Chem. 6 853
[119] Zhou X Wang G Guo L Chi H Wang G Zhang Q Chen C Thompson T Sakamoto J Dravid V P Cao G Uher C 2014 J. Mater. Chem. 2 20629
[120] Tan G Shi F Hao S Zhao L D Chi H Zhang X Uher C Wolverton C Dravid V P Kanatzidis M G 2016 Nat. Commun. 7 12167
[121] Zhang J Wu D He D Feng D Yin M Qin X He J 2017 Adv. Mater. 29 1703148
[122] Fu L Yin M Wu D Li W Feng D Huang L He J 2017 Energy Environ. Sci. 10 2030
[123] Meng X Liu Z Cui B Qin D Geng H Cai W Fu L He J Ren Z Sui J 2017 Adv. Energy Mater 7 1602582
[124] Kim S I Lee K H Mun H A Kim H S Hwang S W Roh J W Yang D J Shin W H Li X S Lee Y H Snyder G J Kim S W 2015 Science 348 109
[125] Chen Z Jian Z Li W Chang Y Ge B Hanus R Yang J Chen Y Huang M Snyder G J Pei Y 2017 Adv. Mater. 29 1606768
[126] Zhang Q Liao B Lan Y Lukas K Liu W Esfarjani K Opeil C Broido D Chen G Ren Z 2013 Proc. Natl. Acad Sci. USA 110 13261
[127] Tan G Shi F Hao S Chi H Zhao L D Uher C Wolverton C Dravid V P Kanatzidis M G 2015 J. Am. Chem. Soc. 137 5100
[128] Yang H Bahk J H Day T Mohammed A M Snyder G J Shakouri A Wu Y 2015 Nano lett. 15 1349
[129] Xu B Agne M T Feng T Chasapis T C Ruan X Zhou Y Zheng H Bahk J H Kanatzidis M G Snyder G J Wu Y 2017 Adv. Mater. 29 1605140
[130] Cabrera H Zumeta-Dubé I Korte D Grima-Gallardo P Alvarado F Aitken J A Brant J A Zhang J H Calderónc A Marínc E Aguilar-Frutis A Erazo J E Perez-Cappe E Frankod M 2015 J. Alloys Compd. 651 490
[131] Kumar A Dhama P Saini D S Banerji P 2016 RSC Adv. 6 5528
[132] Chakrabarti D J Laughlin D E 1983 Bull. Alloys Pha. Diagr. 4 254
[133] Dennler G Chmielowski R Jacob S Capet F Roussel P Zastrow S Nielsch K Opahle I Madsen G K 2014 Adv. Energy Mater. 4 1301581