High Thermoelectric Performance of In4Se3-Based Materials and the

3 days ago - Finally, the ZT values of all single crystalline and polycrystalline In4Se3 materials have been summarized as a function of the doping st...
0 downloads 12 Views 4MB Size
Download PDF
Article Cite This: Acc. Chem. Res. XXXX, XXX, XXX−XXX

pubs.acs.org/accounts

High Thermoelectric Performance of In4Se3‑Based Materials and the Influencing Factors Published as part of the Accounts of Chemical Research special issue “Advancing Chemistry through Intermetallic Compounds”. Xin Yin,† Jing-Yuan Liu,† Ling Chen, and Li-Ming Wu* Beijing Key Laboratory of Energy Conversion and Storage Materials, College of Chemistry, Beijing Normal University, Beijing 100875, P. R. China CONSPECTUS: Materials that can directly convert electricity into heat, i.e., thermoelectric materials, have attracted renewed attention globally for sustainable energy applications. As one of the state-of-the-art thermoelectric materials, In4Se3 features an interesting crystal structure of quasi-two-dimensional sheets comprising In/ Se chains that provide a platform to achieve a Peierls distortion and support a charge density wave instability. Single-crystal In4Se3−δ (δ = 0.65) shows strong anisotropy in its thermoelectric properties with a very high ZT of 1.48 at 705 K in the b−c plane (one of the highest values for an n-type thermoelectric material to date) but a much lower ZT of approximately 0.5 in the a−b plane. Because of the random dispersion of grains and the grain boundary effect, the electrical transport properties of polycrystalline In4Se3 are poor, which is the main impediment to improve their performance. The In4-site in the In4Se3 unit cell is substitutional for dopants such as Pb, which increases the carrier concentration by 2 orders of magnitude and the electrical conductivity to 143 S/cm. Furthermore, the electrical conductivity markedly increases to approximately 160 S/cm when Cu is doped into the interstitial site but remains as low as 30 S/cm with In1/In2/In3-site dopants, e.g., Ni, Zn, Ga, and Sn. In particular, the In4-site dopant ytterbium introduces a pinning level that highly localizes the charge carriers; thus, the electrical conductivity is maintained within an order of magnitude of 30 S/cm. Meanwhile, ytterbium also creates resonance states around the Fermi level that increase the Seebeck coefficient to −350 μV/K, the highest value at the ZT peak. However, the maximum solubility of the dopant may be limited by the Se-vacancy concentration. In addition, a Se vacancy also destroys the regular lattice vibrations and weakens phonon transport. Finally, nanoinclusions can effectively scatter the middle wavelength phonons, resulting in a decrease in the lattice thermal conductivity. Because of the multiple-dopant strategy, polycrystalline materials are competitive with single crystals regarding ZT values; for instance, Pb/Sn-co-doped In4Pb0.01Sn0.04Se3 has ZT = 1.4 at 733 K, whereas In4Se2.95(CuI)0.01 has ZT = 1.34 at 723 K. These properties illustrate the promise of polycrystalline In4Se3-based materials for various applications. Finally, the ZT values of all single crystalline and polycrystalline In4Se3 materials have been summarized as a function of the doping strategy applied at the different lattice sites. Additionally, the correlations between the electrical conductivity and the Seebeck coefficient of all the polycrystalline materials are presented. These insights may provide new ideas in the search for and selection of new thermoelectric compounds in the In/Se and related In/Te, Sn/Se, and Sn/Te systems.

1. INTRODUCTION

The performance of a TE material is theoretically determined by the figure-of-merit ZT = S2σT/κ, where S, σ, T, and κ are the Seebeck coefficient, electrical conductivity, absolute temperature, and thermal conductivity, respectively. Accordingly, an excellent TE material should possess a high S, which usually occurs in a wide band gap semiconductor or an insulator; large σ, which is an inherent characteristic of good metals; and small κ, such as those of organic or glassy solids. It is extremely difficult to have all these parameters simultaneously in a single material, as they are strongly interrelated and, at their optimal values, even reversely correlated. Therefore, decoupling their strong correlations is the key to realizing the high performance

The rapid development of society has inevitably generated the crises of energy shortage and environmental problems, which in turn have promoted renewed worldwide interest in developing thermoelectric (TE) materials to serve as a promising candidate to alleviate such problems. TE devices can directly and reversibly convert electricity into heat through the Seebeck and Peltier effects. Despite the re-examination with the new concepts and ideas of band engineering, dopant engineering, and hierarchical architecture engineering, traditional TE materials, including bismuth telluride1,2 and lead chalcogenides,3−5 remain as the best performing materials and exhibit great potential for further improvement. Interested readers can refer to several recent reviews on these topics.6−8 © XXXX American Chemical Society

Received: September 28, 2017

A

DOI: 10.1021/acs.accounts.7b00480 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 1. (a) Crystal structure of In4Se3 from the perspective of the c axis. Red and white spheres: In and Se with the atom numbers marked. Blue: interstitial site. The detailed image shows the 1D In/Se chains comprising three distinct and interlocked In−Se−In zigzag chains extending along the c axis with the In−In metallic bond distances marked. (b) Dimensionless ZT and carrier concentration n (inset) of single-crystal In4Se2.35.11 (c) STM image of the cleavage (100) surface.21 Reproduced with permission from ref 21, copyright 2008 AIP Publishing LLC and ref 11, copyright 2009 Nature Publishing Group.

of a TE material. Conversely, a TE device consists of one ntype and one p-type pellet through which electron flow and hole flow, respectively, are allowed. These two types of materials should be comparable and are equally crucial to the efficiency of the device: ηmax = [Z ] =

[S]2 [κ ][ρ]

TH − TC TH

1 + [Z]Tave − 1 1 + [Z]Tave +

TC TH

properties. For instance, there are four crystallographically independent indium atoms (In1−4) and three selenium atoms (Se1−3) at the 4g Wyckoff sites. The indium atoms are divided into two types: In4 as an In+ cation involved mostly in weak ionic bonding interactions and In1−3 as In1.667+ cations forming In−In metallic and In−Se covalent bonds.14,15 In addition, the interstitial site (Figure 1a, blue) is considered to be neutral.16 Moreover, the three selenium atoms have different local coordination environments; e.g., Se3 forms two In−Se bonds with two unoccupied coordination sites, which are considered two dangling bonds, and both Se1 and Se2 have three In−Se bonds and one dangling bond. Therefore, the formation energy of Se vacancies differs depending on the Se site.17 Single-crystal In4Se2.35 achieves a high ZT of 1.48 at 705 K in the b−c plane (one of best values among the n-type TE materials) but a much lower ZT (less than 0.5) in the a−b plane (Figure 1b).11 Despite their high ZT values, the single crystals have major issues for applications, namely, their strong anisotropy, extremely high cost, and poor mechanical properties. Therefore, the search for feasible polycrystalline In4Se3-based materials has attracted considerable attention.15 A deeper understanding of the structure−property relationships has afforded some doped polycrystalline In4Se3-based materials achieving a ZT = 1.4 at 733 K,15 which is comparable to the best values of single crystals. This Account focuses on the recent developments of In4Se3based single crystalline and polycrystalline materials. We elaborate on the crystallographic structural features, crystal structure−electronic structure−TE property relationships, and strategies for improving performance. We also summarize the major advances in polycrystalline materials according to the different doping strategies for the different lattice sites in In4Se3.

, where

([S]2 = (Sp − Sn)2, [κ] = κp + κn, and [ρ] = ρp +

ρn). However, in terms of ZT records and quantity of available materials, the n-type are generally inferior to the p-type TE materials. For instance, the ZT values of the state-of-the-art ptype materials, such as SnSe (2.62 at 923 K)9 and PbTe−SrTe (2.2 at 915 K),3 are significantly higher than those of the stateof-the-art n-type ones, such as multiple-filled skutterudites (1.7 at 850 K),10 La-doped PbTe−Ag2Te (1.6 at 775 K),4 and single-crystal In4Se3−δCl0.03 (1.53 at 705 K).11 Consequently, the development of n-type TE materials is an urgent need and has great scientific significance. In4Se3 is one of the state-of-the-art n-type TE materials in the midtemperature range (500−900 K). Its crystallographic structure was established by Hogg et al. in 1971 with a = 15.297 Å, b = 12.308 Å, c = 4.081 Å in the orthorhombic space group Pnnm (No. 58),12 endowing two-dimensional (2D) In/ Se quasi-layers (Figure 1a). Each of the In/Se layers is constructed by one-dimensional (1D) In/Se chains comprising trinuclear [(In3)5+(Se3)6−] clusters13 with strong In−Se bonds and metallic In−In bonding interactions. Property studies suggest that such a chain feature results in anisotropy in the σ. These In/Se chains undergo a Peierls distortion, which further decreases the κ in the b−c plane. In addition, each crystallographically independent atom in the unit cell has its own unique coordination environment, which provides the unique possibility of individually modifying the physical B

DOI: 10.1021/acs.accounts.7b00480 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 2. Electronic band structures of (a) In4Se3 and (b) In4Se3−δ (δ = 0.25).22 (c) Indium element-resolved calculated densities of states.23 (d) Density of states of In and Se.23 (e) Experimental band structure along the In/Se chain direction,21 and the direction perpendicular to the chain but still in the surface plane.23 Red dots at room temperature, blue dots at 150 K,21 filled dots at 15.5 eV photon energy, and open dots at 23 eV.23 (f) Sketch diagram indicating the Peierls distortion that causes the superlattice and energy gap. I and II represent the primitive periodic lattice and the superlattice after Peierls distortion, respectively.24 (g) HRTEM images and electron diffraction patterns of single crystalline In4Se2.78 in the a−b plane.11 Reproduced with permission from refs 21−23, copyright 2008, 2009 AIP Publishing LLC; ref 24, copyright 1989 John Wiley and Sons; and ref 11, copyright 2009 Nature Publishing Group.

2.2. Anisotropic Electronic Band Structures of In4Se3/In4Se3−δ

These insights may provide new insight for the exploration of new systems with excellent TE properties.

The band dispersions (Figure 2a, b) show that In4Se3 is a narrow-band gap semiconductor and In4Se3−δ is more metallic. Density of states (Figure 2c, d) illustrate that the s and p orbitals of In4 and the p orbitals of Se are dominant adjacent to the Fermi level, illustrating the interactions between the In4 cations and the neighboring In/Se layers.22,23 The band structure of In4Se3 shows anisotropy along different crystallographic directions (Figure 2e). The Γ−X direction representing the In/Se chain direction, i.e., the c axis, has a band dispersion width of approximately 1 eV near the Fermi level, indicating the existence of strong hybridization within the In/Se chains.21 The direction perpendicular to the In/Se chains (Γ−Y) within the (100) cleavage surface, i.e., the b axis, has a much smaller dispersion width of approximately 0.32 eV (Figure 2e).23 Therefore, the σ in In4Se3 is strong along the c axis, indicating anisotropy.21,23 Very interestingly, the Se-vacancy concentration also influences the electrical transport properties. For instance, with approximately 8% Se vacancies, based on the dispersive electron bands near the conduction band minimum along the Γ−Z and T−Y directions, In4Se3−δ (δ = 0.25) exhibits high electron conductivity along the c axis (Figure 2b). Moreover, the flat hole bands near the valence band maximum along the X−Γ and Y−S paths indicate that the holes are highly localized along the b axis (Figure 2b).22

2. SINGLE CRYSTALLINE In4Se3/In4Se3−δ AND THEIR THERMOELECTRIC PROPERTIES 2.1. Chemical Bonding and Structure of In4Se3

As shown in Figure 1a, In4Se3 has a quasi-2D corrugated layered structure extending in the b−c plane. Each of the In/Se layers is constructed by 1D In/Se chains comprising trinuclear (In3Se3)− clusters carrying metallic bonding interactions of In1−In2 = 2.77433(0) Å and In2−In3 = 2.75040(0) Å and covalent bonds of In1−Se1 = 2.69340(0) Å, In1−Se2 = 2.62197(0) Å, In2−Se2 = 2.80534(0) Å, In3−Se3 = 2.63689(0) Å, and In3−Se1 = 2.70880(0) Å.13,18 In the c direction, the trinuclear (In3Se3)− clusters propagate into three distinct quasi-1D In/Se zigzag chains (Figure 1a, colored in blue, green and yellow) that are interconnected via In2−In3, In1−In2, and In−Se bonds. Along the a axis, In4 cations situate between the In/Se layers, linking two neighboring layers via weak ionic bonding interactions, i.e., In4−Se3 = 2.98091(0) Å, In4−Se2 = 3.16244(0) Å, In4−Se1 = 3.40629(0) Å.18 Note that, between the layers, the position shown by the blue ball (Figure 1a) represents an interstitial site that can be occupied by different cations, the presence of which can significantly increase the σ and carrier concentration (n). In particular, the anisotropic structural features of In4Se3 are impressive; e.g., in the b−c plane where the In/Se layer extends, the covalent and metallic bonding interactions are dominant, whereas in the a direction where the In/Se layers stack, weak bonding interactions are involved.19 Such an anisotropic layered structure motif has been visualized by scanning tunneling microscopy (STM).20 As shown in Figure 1c, the STM image of the (100) cleavage surface reveals that the In/Se chains extend along the c direction and that these chains arrange in a quasi-layered pattern.21

2.3. Impact of Structure on Thermoelectric Properties

2.3.1. Peierls Distortion Influences the Thermal Conductivity and Carrier Concentration. The periodic lattice of a 1D metal chain distorts spontaneously into a superlattice. (Figure 2f, I vs II). This phenomenon, which is known as a Peierls distortion, can create an energy gap (Figure 2f). The In/Se chains in In4Se3 undergo the Peierls distortion and show a charge density wave (CDW) instability phenomenon associated with the Fermi surface nesting in the b−c plane (Figure 3a).11,20,21 The Peierls distortion and CDW C

DOI: 10.1021/acs.accounts.7b00480 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 3. (a) Fermi surface of In4Se2.75 in the b−c plane. The black square is the first Brillouin zone. Fermi nesting vector q is defined in the closed Fermi surface.11 (b) Generalized electron susceptibility χ (q) along the X (0, 1/2, 0)−U (0, 1/2, 1/2) symmetry line.11 (c) Temperature-dependent thermal conductivities κ (T) of crystalline In4Se3−δ (δ = 0.65) in the a−b and b−c planes.11 (d) Temperature-dependent κl of In4Se3 and In4Se2.5 (δ = 0.5) in different directions.17 The lattice thermal conductivity κl calculated by MD simulation and the Slack relation for In4Se317 (e), and for In4Se2.5 (δ = 0.5) (f).17 Reproduced with permission from ref 11, copyright 2009 Nature Publishing Group; ref 17, copyright 2013 American Physical Society.

Figure 4. Chart of electrical conductivity (black) and S (blue) at the corresponding peak ZT values of selected polycrystalline In4Se3-based materials. Regions I−VII marked by the dotted vertical line represent different doping strategies.

instability break the translational symmetry and lead to the κ reduction in the b−c plane.11,24 High-resolution transmission electron microscopy (HRTEM) images and electron diffraction patterns of In4Se2.78 show weak secondary superstructure peaks indicating the Peierls distortion (Figure 2g).11 For In4Se2.75, the (0, 1/2, 1/16) sharp peak of the generalized electron susceptibility originates from the well-defined commensurate nesting vector, which leads to a CDW instability in the b−c plane (Figure 3b).11 As generally accepted, layered In4Se3 should possess a higher κ in the b−c plane where the In/Se

layer extends. However, this is contrary to the experimental observation: the κ in the b−c plane, i.e., 0.74 W/mK at 705 K, is much lower than that in the a−b plane (Figure 3c).11 This result may be due to the In/Se chain undergoing the Peierls distortion in the b−c plane20,21 and the CDW instability breaking the translational symmetry, which impedes phonon transport and increases phonon scattering. Consequently, the κ in the b−c plane is eventually reduced (Figure 3c).11 The carrier concentration generally should be isotropic. However, this is not the case for In4Se3−δ crystals. By creating D

DOI: 10.1021/acs.accounts.7b00480 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research an energy gap, the Peierls distortion decreases the concentration of the itinerant carriers in the b−c plane. Thus, the carrier concentration in the b−c plane is lower than that in the a−b plane (Figure 1b, inset). Although the Peierls distortion effect reduces the carrier concentration, the κ decrease is more significant, thereby leading to a high ZT. These results suggest that a low dimensional crystalline sheet involved in strong electron−phonon coupling, such as Peierls distortion and CDW instability, should exhibit excellent TE properties. 2.3.2. Halogen Substitution Increases Electrical Conductivity. Single-crystal In4Se2.35 (δ = 0.65) has achieved a quite high ZT of 1.48, a relatively high S and low κ; however, the n (∼1018 cm−3) is 1−2 orders of magnitude lower than the optimal value for most of TE materials (1019−1020 cm−3).11,25 Chlorine could be a good electron dopant because of its 5 p valence electrons compared to the 4 of Se.26 For example, single-crystal In4Se2.67Cl0.03 with n about 1019 cm−3 reaches ZT = 1.53 at 698 K,25 which is higher than that of undoped In4Se2.35.11 The ZT enhancement originates from the 2−3 orders of magnitude enhancement in the σ, which is attributed to the increase in both carrier concentration and carrier mobility.25 Br- and I-substitution show similar effects.27 By contrast, F-substitution decreases the σ remarkably because the F atom occupies the interstitial site instead.27

migrating to the defects or the interstitial sites as electrical bridges instead of remaining at the lattice sites, resulting the n decreasing.31 Tellurium and halogens (Cl, Br, I) are effective anion dopants on the Se sites (Figure 4, III). The tellurium substitution does not increase the σ greatly because of the bipolar diffusion effect.32 In comparison, halogens with additional p electrons are more effective in improving the carrier concentration. The n of polycrystalline In4Se2.7Cl0.03 can reach 1.80 × 1019 cm−3, which is higher than that of singlecrystal In4Se2.7.26 Cation doping has a remarkable impact on the TE properties14,15,33−35 (Figure 4, IV−VI). As mentioned above, the In sites in In4Se3 are classified into two types: the In4-site as an In1+ cation and the In1/In2/In3-site as an In1.67+ cation. In addition, the interstitial site is considered as a neutral site that has tendency to incorporate different atom dopants. Atoms that substitute at In4 and interstitial sites behave more like electron donors; thus, affording higher n. For instance, Yb,36 Na,14 Sn,33,34 Pb,15,35 and Ca,14 which usually occupy the In4-site, increase the n by 2 orders of magnitude, leading to much higher σ (40−143 S/cm) (Figure 4, V). However, Ni,37 Zn, Ga and Sn,14 which preferentially substitute at the In2/In3 sites, provide low n of 1017 cm−3 and σ on approximately the same order of magnitude as the undoped materials (∼30 S/cm)28,38 (Figure 4, IV). As an In4-dopant, ytterbium forms a pinning level near the Fermi level via the f and d mixing states. In In4−xYbxSe3 (x = 0.01−0.09), Yb can act as either a +1 cation executing an isoelectronic substitution at the In1+-site, which leads to a slow n decrease at a low concentration x < 0.05, or an acceptor replacing the In1.67+-site, which leads to a sharp reduction in n when x > 0.05.36 The pinning level highly localizes the charge carriers; thus, the σ is maintained within an order of magnitude of 1017 cm−3.36 The ytterbium atoms create resonance states around the Fermi level, thereby increasing the S to −350 μV/K, the highest value at the ZT peaks.36 The heavy atom ytterbium also effectively scatters phonons and thus reduces the κ.36 Therefore, ytterbium is considerably important when searching for lead-free In4Se3-based TE materials. By contrast, Cu16 and Ag39 substitute at the interstitial site, with the former atom markedly increasing the σ to approximately 160 S/cm16 and the latter atom to only 50 S/ cm due to the extremely low solubility (In4−xAgxSe3, xm < 0.03)39 (Figure 4, VI). However, In4Se3−δ(CuI)x demonstrates that the maximum solubility of the dopant (xm) may be restricted by the Se-vacancy concentration (δ) by xm < δ. Specifically, the greater the δ, the greater the power factor (S2σ).40 As δ increases, more Cu atoms occupy the interstitial site and increase the atom packing density, inevitably increasing the κ.39 The multiple-dopant strategy is the most effective approach to increase the σ15,16,35,40−42 (Figure 4, VII). The simultaneous substitution at both the In4-site and In1/In2/In3-sites was first proposed in In4Pb0.01Sn0.03Se3.15 The successful Sn substitution at the In2/In3-sites while maintaining the Pb-substitution at the In4-site increases the σ from 125 S/cm to 200 S/cm without decreasing the S. 1 5 The resultant polycrystalline In4Pb0.01Sn0.04Se3 shows a ZT (1.4)15 similar to that of a single crystal.11,25 Other examples are the anion−cation doping in Pb/ Sn/[email protected] (notation A@formula represents that the material is doped with A),41 In4Se2.95(CuI)0.01;40 cationdeficient doping in Pb/[email protected] and In4Pb0.01Sn0.03Se2.9;42

3. POLYCRYSTALLINE In4Se3-BASED MATERIALS AND THEIR THERMOELECTRIC PROPERTIES Polycrystalline materials are feasible, isotropic and relatively low cost. Because of the random dispersion of grains and grain boundary effect, however, their electrical transport properties are relatively poor. The σ can be calculated by σ = neμ, where n is the carrier concentration, e is the electron charge, and μ is the carrier mobility. Furthermore, the S for metals or degenerate 2/3 8π 2kB2 * π semiconductors is defined by S = , where kB, 2 m T 3eh

( 3n )

h, and m* are the Boltzmann constant, Planck constant, and effective mass, respectively. Note that the σ and S are proportional to and negatively correlated with, respectively, the n; therefore, these two factors are mutually correlated and restrictive. Thus, the key to realizing a large power factor (S2σ) is to balance the S and σ. Figure 4 summarizes the recent advances in polycrystalline In4Se3-based materials, presenting both the σ and S of each material. These data show that for In4Se3-based materials, the key problem is how to enhance the σ. 3.1. Balancing Electrical Conductivity and Seebeck Coefficient

Se vacancy (δ), also known as a Se deficiency, defect, or selfdoping, leads to a higher σ (40−75 S/cm) than that of undoped polycrystalline In4Se3 (20−38 S/cm) because as an electron donor, the Se vacancy can provide extra electrons to generate a higher n (Figure 4, II vs I). For instance, without any Se vacancies, n is usually on the order magnitude of 1017 cm−3,28 and with Se vacancies, the n is 1.68 × 1018 cm−3 at δ = 0.5 or 4.35 × 1018 cm−3 at δ = 0.35.29,30 These n values are comparable to that of a single crystal (7 × 1018 cm−3 for In4Se3−δ at δ = 0.22),11 with which the polycrystalline In4Se3based material can achieve a maximum ZT to date (In4Pb0.01Sn0.04Se3, ZT = 1.4).15 However, when the Se-vacancy concentration is too high, e.g., δ = 0.8, the n in In4Se2.2 (loading ratio) dramatically decreases to 4.13 × 1016 cm−3,29 which is attributable to the possibility of the indium atoms in In4Se2.2 E

DOI: 10.1021/acs.accounts.7b00480 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 5. Current single crystalline and polycrystalline In4Se3-based materials: the thermoelectric figure of merit ZT of each material. Regions I−VII marked by the dotted vertical lines represent different doping strategy.

the lattice and total κ because of the breaking of the Peierls distortion in the In/Se chains, which plays a much more important role in the decrease in the lattice thermal conductivity than the point defects. In comparison, the anion dopants or Se vacancies influence the Peierls distortion indistinctly; thus, the κ is maintained at a value similar to those of the single crystals, such as In4Se2.7Cl0.03 (κ = 0.6 W/ mK)26 and In4(Se0.95Te0.05)2.9 (κ = ∼0.42 W/mK).44 Nanoinclusions can effectively scatter the medium wavelength (nanoscale) phonons, resulting in a decrease in lattice thermal conductivity.45,46 The nanoinclusions found in In4Se3based materials are mostly spheres with sizes of 5−100 nm, such as In nanoparticles,15,30,43 In-rich composites,44 and Cu nanoparticles.16,35,47 Only those species smaller than 30 nm can effectively scatter the phonons and decrease the κ.15 Pb and Cu are exceptions: they markedly increase the electronic thermal conductivity by 1 order of magnitude and thus slightly increase the κ as well.15,16 Although most of the heat is carried by the short and medium wavelength phonons, some long wavelength phonons also carry heat.29 More work on hierarchical architecture engineering would be worthwhile. In summary, the so-called multiple-dopant strategy utilizing In4- and/or interstitial-dopants at a suitable Se-vacancy concentration with or without doping a suitable anion is the most effective approach to improve the σ (Figure 4). However, the multiple-dopant strategy may face a dilemma, namely, that the κ may also be increased because of the increase in complexity induced by the dopants and the nanoeffects introduced by the nanoinclusions and grain boundaries. The ZT peak values of the selected single-crystal or polycrystalline materials are summarized in Figure 5. The advanced polycrystalline materials successfully exhibit comparable performances to those of the single crystals and thus show promise for various applications.

and cation-anion-deficient doping in Cu/[email protected],35 CuBr2@ In4Se2.5,16 and Pb/Cu/[email protected] Because the S and σ are negatively correlated with and proportional to, respectively, the n, their variations are somewhat opposite (Figure 4, blue vs black). In general, samples exhibit S of −250 μV/K with In4, interstitial, or multiple doping (Figure 4, blue, V, VI, VII) and above −300, or even −350 μV/K (almost the same as those of the undoped samples) with Se vacancies, anion doping or In2/In3 doping (Figure 4, blue, II, III, IV, vs I). Usually, the dopant increases the n and thus inevitably decreases the S. To obtain a higher power factor (S2σ), the inversely correlated σ and S should be balanced. Conceptually, the best strategy would be to exploit multiple dopants, in which dopants such as Sn and Pb are utilized to occupy the In1/In2/In3-site and In4-site while simultaneously introducing Cu at the interstitial site and incorporating a rare earth element, such as Yb, to suitably increase the S. 3.2. Decreasing Thermal Conductivity

The grains in a polycrystalline sample are randomly dispersed; therefore, the Peierls distortion is averaged out and weakened.26,41 In addition, the grain boundaries usually substantially scatter the phonons. Consequently, polycrystalline In4Se3-based materials are isotropic and exhibit very low thermal conductivities, i.e., less than 1 W/mK, even as low as 0.39 W/mK.43 The Se vacancy is one of the most common point defects that destroy the regular lattice vibrations and weaken the phonon transport.22,29,30,38,43 For instance, the thermal conductivities for In4Se2.95 and In4Se2.2 are ∼0.57 W/mK22 and 0.41 W/mK,29 respectively. Cation dopants also cause significant atomic mass fluctuations and reduce the lattice thermal conductivity. For example, ytterbium doping at the In4 site in In4−xYbxSe3 reduces the κ to approximately 0.43 W/mK by decreasing the lattice thermal conductivity.36 However, most cation dopants substituting at the In2/In3 sites, such as Zn,14 Ga,14 and Sn,33 increase both

4. CONCLUSION AND OUTLOOK The state-of-the-art n-type TE In4Se3 provides an interesting platform because of its unique structure, which comprises a F

DOI: 10.1021/acs.accounts.7b00480 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research crystalline sheet of 1D In/Se chains supporting the Peierls distortion and a CDW instability phenomenon involved in strong electron−phonon coupling. In addition, the Peierls distortion and the CDW instability break the periodic translational symmetry, leading to an ultralow κ in the b−c plane. Single-crystal In4Se3−δ shows strong anisotropy in its TE properties with a ZT of 1.48 in the b−c plane and less than 0.5 in the a−b plane. Because of the random grain dispersion, polycrystalline In4Se3 suffers from a very low σ. A significant σ improvement is achieved by In4 and interstitial doping15,16,35 along with contributions from Se vacancies31 and/or anion doping.26 The maximum dopant solubility may be limited by the Sevacancy concentration.39 Moreover, nanoinclusions effectively decrease the κ.15 With the efficient multiple-dopant strategy, the polycrystalline materials can reach comparable ZT values with those of single crystals, such as In4Pb0.01Sn0.04Se3 with ZT = 1.4 at 733 K15 and In4Se2.95(CuI)0.01 with ZT = 1.34 at 723 K40 (Figures 4 and 5). These advanced polycrystalline In4Se3based materials show promising prospects for a range of applications. Nevertheless, nanostructuring and all-scale hierarchical architecture engineering, as well as lead-free materials, are worthy of further investigation. As the first step toward establishing new TE materials, the discovery of new compounds with high performance is an important activity, in addition to optimizing the known systems. New insights into the In4Se3 system will encourage us and others to further explore new metal-rich compounds with complicated layered or 1D structures featuring metal−metal bonding interactions. To date, In4Se3 is the only binary metal-rich In/Se compound; however, we anticipate that there is more to be discovered in this and related systems, such as In/Te, Sn/Se, and Sn/Te.

■

professor at the Fujian Institute CAS in 2003 and moved to BNU in 2014. Her research efforts are focused on solid-state chemistry. Li-Ming Wu was born in 1973. He received his Bachelor’s and Master’s degrees in Quantum Chemistry from BNU in 1993 and 1996, respectively, and obtained his Ph.D. in Quantum Chemistry from Fuzhou University in 1999. He conducted postdoctoral work on solidstate material sciences at the Fujian Institute CAS (1999−2001) and Arizona State University (2001−2004), and was a visiting professor at Northwestern University from February to September 2015. He started his independent academic career as an associate professor at the Fujian Institute CAS in 2004, was promoted to a full professor in 2005, and moved to BNU in 2014. His research interests include thermoelectric materials, inorganic solid functional materials, inorganic structural chemistry, and solid-state theoretical chemistry.

■

ACKNOWLEDGMENTS This research was supported by the National Natural Science Foundation of China under Projects 91422303, 21571020, and 21671023.

■

REFERENCES

(1) Poudel, B.; Hao, Q.; Ma, Y.; Lan, Y.; Minnich, A.; Yu, B.; Yan, X.; Wang, D.; Muto, A.; Vashaee, D.; Chen, X.; Liu, J.; Dresselhaus, M. S.; Chen, G.; Ren, Z. High-thermoelectric performance of nanostructured bismuth antimony telluride bulk alloys. Science 2008, 320, 634−638. (2) Liu, W.-S.; Zhang, Q.; Lan, Y.; Chen, S.; Yan, X.; Zhang, Q.; Wang, H.; Wang, D.; Chen, G.; Ren, Z. Thermoelectric property studies on Cu-doped n-type CuxBi2Te2.7Se0.3 nanocomposites. Adv. Energy Mater. 2011, 1, 577−587. (3) Biswas, K.; He, J.; Blum, I. D.; Wu, C.-I.; Hogan, T. P.; Seidman, D. N.; Dravid, V. P.; Kanatzidis, M. G. High-performance bulk thermoelectrics with all-scale hierarchical architectures. Nature 2012, 489, 414−418. (4) Pei, Y.; Lensch-Falk, J.; Toberer, E. S.; Medlin, D. L.; Snyder, G. J. High thermoelectric performance in PbTe due to large nanoscale Ag2Te precipitates and La doping. Adv. Funct. Mater. 2011, 21, 241− 249. (5) Wang, H.; Gibbs, Z. M.; Takagiwa, Y.; Snyder, G. J. Tuning bands of PbSe for better thermoelectric efficiency. Energy Environ. Sci. 2014, 7, 804−811. (6) Tan, G.; Zhao, L.-D.; Kanatzidis, M. G. Rationally designing highperformance bulk thermoelectric materials. Chem. Rev. 2016, 116, 12123−12149. (7) Zhu, T.; Liu, Y.; Fu, C.; Heremans, J. P.; Snyder, J. G.; Zhao, X. Compromise and synergy in high-efficiency thermoelectric materials. Adv. Mater. 2017, 29, 1605884. (8) Zeier, W. G.; Zevalkink, A.; Gibbs, Z. M.; Hautier, G.; Kanatzidis, M. G.; Snyder, G. J. Thinking like a chemist: intuition in thermoelectric materials. Angew. Chem., Int. Ed. 2016, 55, 6826−6841. (9) Zhao, L.-D.; Lo, S.-H.; Zhang, Y.; Sun, H.; Tan, G.; Uher, C.; Wolverton, C.; Dravid, V. P.; Kanatzidis, M. G. Ultralow thermal conductivity and high thermoelectric figure of merit in SnSe crystals. Nature 2014, 508, 373−377. (10) Shi, X.; Yang, J.; Salvador, J. R.; Chi, M.; Cho, J. Y.; Wang, H.; Bai, S.; Yang, J.; Zhang, W.; Chen, L. Multiple-filled skutterudites: high thermoelectric figure of merit through separately optimizing electrical and thermal transports. J. Am. Chem. Soc. 2011, 133, 7837−7846. (11) Rhyee, J.-S.; Lee, K. H.; Lee, S. M.; Cho, E.; Kim, S. I.; Lee, E.; Kwon, Y. S.; Shim, J. H.; Kotliar, G. Peierls distortion as a route to high thermoelectric performance in In4Se3−δ crystals. Nature 2009, 459, 965−968. (12) Hogg, J. H. C.; Sutherland, H. H.; Williams, D. J. Crystallographic evidence for the existence of the phases In4Se3 and In4Te3 which contain the homonuclear triatomic cation (In3)5+. J. Chem. Soc. D 1971, 23, 1568−1569.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Ling Chen: 0000-0002-3693-4193 Li-Ming Wu: 0000-0001-8390-2138 Author Contributions †

X.Y. and J.-Y.L. contributed equally to this work.

Notes

The authors declare no competing financial interest. Biographies Xin Yin was born in Chongqing Municipality, China in 1993. She received her Bachelor of Science from Beijing Normal University (BNU) in 2016. She is currently pursuing her Master’s degree in Inorganic Chemistry at BNU, where she is working on thermoelectric materials. Jing-Yuan Liu was born in Shandong Province, China in 1995. She received her Bachelor of Science from BNU in 2016. She is currently pursuing her Master’s degree in Inorganic Chemistry at BNU, where she is working on thermoelectric materials. Ling Chen received her Master’s degree from BNU in 1996 and Ph.D. from the University of Chinese Academy of Sciences in 1999 and performed postdoctoral research at Iowa State University (2000− 2003). She started her own independent academic career as a full G

DOI: 10.1021/acs.accounts.7b00480 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

state and its effects on thermoelectric properties in In4Se3. J. Phys. D: Appl. Phys. 2013, 46, 275304. (34) Abhari, A. S.; Abdellahi, M.; Bahmanpour, M. The effects of Snsubstitution on thermoelectric properties of In4−xSnxSe3 ceramic. Ceram. Int. 2016, 42, 5593−5599. (35) Luo, Y.; Yang, J.; Liu, M.; Xiao, Y.; Fu, L.; Li, W.; Zhang, D.; Zhang, M.; Cheng, Y. Multiple heteroatom induced carrier engineering and hierarchical nanostructures for high thermoelectric performance of polycrystalline In4Se2.5. J. Mater. Chem. A 2015, 3, 1251−1257. (36) He, S.-H.; Liu, P.-F.; Lin, Z.-X.; Wang, G.-Q.; Chen, L.; Wang, Z.-X.; Wu, L.-M. The roles of Yb-substitution on thermoelectric properties of In4−xYbxSe3. Acta Mater. 2015, 101, 16−21. (37) He, S.-H.; Lin, Z.-X.; Khan, M. A.; Liu, P.-F.; Chen, L.; Wu, L.M. Thermoelectric properties of Ni-substituted polycrystalline In4Se3. Chin. J. Struct. Chem. 2015, 34, 1217−1223. (38) Yim, J.-H.; Park, H.-H.; Jang, H. W.; Yoo, M.-J.; Paik, D.-S.; Baek, S.; Kim, J.-S. Thermoelectric properties of indium-selenium nanocomposites prepared by mechanical alloying and spark plasma sintering. J. Electron. Mater. 2012, 41, 1354−1359. (39) Chen, Y.-C.; Lin, H.; Wu, L.-M. High thermoelectric performance of polycrystalline In4Se3−δ(CuI)x: synergistic effects of the Se-deficiency and CuI-doping. Inorg. Chem. Front. 2016, 3, 1566− 1571. (40) Chen, Y.-C.; Liu, P.-F.; Chen, L.; Wu, L.-M. Thermoelectric Properties of Ag-doped In4Se2.95 Polycrystalline Compounds. Chin. J. Struct. Chem. 2016, 35, 1868−1875. (41) Hee Kim, J.; Jae Kim, M.; Oh, S.; Rhyee, J.-S.; Park, S.-D.; Ahn, D. Thermoelectric properties and chlorine doping effect of In4Pb0.01Sn0.03Se2.9Clx polycrystalline compounds. Dalton Trans. 2015, 44, 3185−3189. (42) Kim, J. H.; Kim, M. J.; Oh, S.; Rhyee, J.-S. Thermoelectric properties of Se-deficient and Pb-/Sn-codoped In4Pb0.01Sn0.03Se3−x polycrystalline compounds. J. Alloys Compd. 2014, 615, 933−936. (43) Rawat, P. K.; Park, H.; Hwang, J.; Kim, W. Low thermal conductivity and high thermoelectric performance in In4Se3−x with phase-separated indium inclusions. J. Electron. Mater. 2017, 46, 1444− 1450. (44) Jeong, M.; Lim, Y. S.; Seo, W.-S.; Lee, J.-H.; Park, C.-H.; Sznajder, M.; Kharkhalis, L. Y.; Bercha, D. M.; Yang, J. Condensonrelated thermoelectric properties and formation of coherent nanoinclusions in Te-substituted In4Se3 compounds. J. Mater. Chem. A 2013, 1, 15342. (45) Vineis, C. J.; Shakouri, A.; Majumdar, A.; Kanatzidis, M. G. Nanostructured thermoelectrics: big efficiency gains from small features. Adv. Mater. 2010, 22, 3970−3980. (46) Lee, M. H.; Rhyee, J.-S.; Vaseem, M.; Hahn, Y.-B.; Park, S.-D.; Kim, H. J.; Kim, S.-J.; Lee, H. J.; Kim, C. Thermoelectric properties of SrTiO3 nano-particles dispersed indium selenide bulk composites. Appl. Phys. Lett. 2013, 102, 223901. (47) Li, G.; Yang, J.; Luo, Y.; Xiao, Y.; Fu, L.; Liu, M.; Peng, J. Improvement of thermoelectric properties of In4Se3 bulk materials with Cu nanoinclusions. J. Am. Ceram. Soc. 2013, 96, 2703−2705.

(13) Rhyee, J.-S.; Kim, J. H. Chemical potential tuning and enhancement of thermoelectric properties in indium selenides. Materials 2015, 8, 1283−1324. (14) Ahn, K.; Cho, E.; Rhyee, J.-S.; Kim, S. I.; Lee, S. M.; Lee, K. H. Effect of cationic substitution on the thermoelectric properties of In4−xMxSe2.95 compounds (M = Na, Ca, Zn, Ga, Sn, Pb; x = 0.1). Appl. Phys. Lett. 2011, 99, 102110. (15) Lin, Z. S.; Chen, L.; Wang, L.-M.; Zhao, J.-T.; Wu, L.-M. A promising mid-temperature thermoelectric material candidate: Pb/Sncodoped In4PbxSnySe3. Adv. Mater. 2013, 25, 4800−4806. (16) Luo, Y.; Yang, J.; Li, G.; Liu, M.; Xiao, Y.; Fu, L.; Li, W.; Zhu, P.; Peng, J.; Gao, S.; Zhang, J. Enhancement of the thermoelectric performance of polycrystalline In4Se2.5 by copper intercalation and bromine substitution. Adv. Energy Mater. 2014, 4, 1300599. (17) Ji, H. S.; Kim, H.; Lee, C.; Rhyee, J.-S.; Kim, M. H.; Kaviany, M.; Shim, J. H. Vacancy-suppressed lattice conductivity of high-ZT In4Se3−x. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 87, 125111. (18) Osters, O.; Blazek, G.; Nilges, T. Comments on Peierlsdistorted indium chains in In4Se3−x. Z. Anorg. Allg. Chem. 2013, 639, 497−501. (19) Hogg, J. H. C.; Sutherland, H. H.; Williams, D. J. The crystal structure of tetraindium triselenide. Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem. 1973, 29, 1590−1593. (20) Balitskii, O. A.; Savchyn, V. P.; Jaeckel, B.; Jaegermann, W. Surface characterization of In4Se3 single crystals. Phys. E 2004, 22, 921−923. (21) Losovyj, Y. B.; Klinke, M.; Cai, E.; Rodriguez, I.; Zhang, J.; Makinistian, L.; Petukhov, A. G.; Albanesi, E. A.; Galiy, P.; Fiyala, Y.; Liu, J.; Dowben, P. A. The electronic structure of surface chains in the layered semiconductor In4Se3(100). Appl. Phys. Lett. 2008, 92, 122107. (22) Rhyee, J.-S.; Cho, E.; Lee, K. H.; Lee, S. M.; Kim, S. I.; Kim, H.S.; Kwon, Y. S.; Kim, S. J. Thermoelectric properties and anisotropic electronic band structure on the In4Se3−x compounds. Appl. Phys. Lett. 2009, 95, 212106. (23) Losovyj, Y. B.; Makinistian, L.; Albanesi, E. A.; Petukhov, A. G.; Liu, J.; Galiy, P.; Dveriy, O. R.; Dowben, P. A. The anisotropic band structure of layered In4Se3(001). J. Appl. Phys. 2008, 104, 083713. (24) Hoffman, R. Solids and surfaces: A chemist’s view of bonding in extended structures; Wiley VCH, 1989; p 96. (25) Rhyee, J.-S.; Ahn, K.; Lee, K. H.; Ji, H. S.; Shim, J.-H. Enhancement of the thermoelectric figure-of-merit in a wide temperature range in In4Se3‑xCl0.03 bulk crystals. Adv. Mater. 2011, 23, 2191−2194. (26) Rhyee, J.-S.; Choi, D. Thermoelectric properties of chlorine doped compounds of In4Se2.7Clx. J. Appl. Phys. 2011, 110, 083706. (27) Ahn, K.; Cho, E.; Rhyee, J.-S.; Kim, S. I.; Hwang, S.; Kim, H.-S.; Lee, S. M.; Lee, K. H. Improvement in the thermoelectric performance of the crystals of halogen-substituted In4Se3−xH0.03 (H = F, Cl, Br, I): effect of halogen-substitution on the thermoelectric properties in In4Se3−x. J. Mater. Chem. 2012, 22, 5730. (28) Shi, X.; Cho, J. Y.; Salvador, J. R.; Yang, J.; Wang, H. Thermoelectric properties of polycrystalline In4Se3 and In4Te3. Appl. Phys. Lett. 2010, 96, 162108. (29) Zhu, G. H.; Lan, Y. C.; Wang, H.; Joshi, G.; Hao, Q.; Chen, G.; Ren, Z. F. Effect of selenium deficiency on the thermoelectric properties of n-type In4Se3−x compounds. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 115201. (30) Yang, J.-Y.; Wu, J.; Li, G.; Zhang, J.-S.; Peng, J. Preparation and thermoelectric properties of polycrystalline In4Sn3−x by mechanical alloying and hot pressing. J. Electron. Mater. 2012, 41, 1077−1080. (31) Zhao, R.; Shu, Y.-T.; Guo, F. Enhanced thermoelectric properties of n-type polycrystalline In4Se3‑x compounds via thermally induced Se deficiency. Funct. Mater. Lett. 2014, 07, 1450025. (32) Rhyee, J.-S.; Cho, E.; Ahn, K.; Lee, K. H.; Lee, S. M. Thermoelectric properties of bipolar diffusion effect on In4Se3−xTex compounds. Appl. Phys. Lett. 2010, 97, 152104. (33) Lim, Y. S.; Jeong, M.; Seo, W.-S.; Lee, J.-H.; Park, C.-H.; Sznajder, M.; Kharkhalis, L. Y.; Bercha, D. M.; Yang, J. Condenson H

DOI: 10.1021/acs.accounts.7b00480 Acc. Chem. Res. XXXX, XXX, XXX−XXX