Enhancement of Thermoelectric Properties of Layered Chalcogenide Materials

Thermoelectric materials have long been proven to be effective in converting heat energy into electricity and vice versa. Since semiconductors have been used in the thermoelectric field, much work has been done to improve their efficiency. The interrelation between their thermoelectric physical parameters (Seebeck coefficient, electrical conductivity, and thermal conductivity) required special tailoring in order to get the maximum improvement in their performance. Various approaches have been reported in the research for developing thermoelectric performance, including doping and alloying, nanostructuring, and nanocompositing. Among different types of thermoelectric materials, layered chalcogenide materials are unique materials with distinctive properties. They have low self-thermal conductivity, and their layered structure allows them to be modified easily to improve their thermoelectric performance. In this review, basic knowledge of thermoelectric concepts and challenges for enhancing the figure of merit is provided. It discusses briefly different groups of layered chalcogenide thermoelectric materials with their structure and thermoelectric properties. It also reports different approaches in the literature for improving their performance and the recent progress done in this field. It highlights graphene as a promising nano additive to layered chalcogenide materials’ matrix and shows its effect on enhancing their figure of merit.


Introduction
The demand for clean and sustainable energy sources is a growing global concern as the cost of energy is rapidly increasing; fossil fuel sources have been shown to affect the environment. Considering that a large amount of our utilized energy is in the form of heat and that a large amount of other forms of utilized energy is wasted as heat, the search for a suitable technology to recover this wasted heat and limit its harmful effects is essential. Among several technologies used to meet these demands, thermoelectric energy is considered to be of the most interest due to its unique capabilities. Thermoelectric generators can convert wasted heat into electrical energy. Conversely, they can transform electrical energy into thermal energy for refrigeration applications. The thermoelectric performance of thermoelectric materials is evaluated by the dimensionless figure of merit, ZT. ZT = S 2 σ k tot (1) where S is the Seebeck coefficient, σ is electrical conductivity, k tot is total thermal conductivity, and S 2 σ is the power factor. Thermoelectric materials have several advantages; these materials are simple, solid-state devices, involve toxic-free operation, have no moving parts or chemical reactions, and demonstrate the potential for scalability. Besides, thermoelectric materials produce negligible amounts of greenhouse gas emissions. However, commercial applications of thermoelectric materials are limited because of the low efficiency relative to the cost, which is inadequate to compete with conventional power generation [1]. As a result, extensive work has been done to achieve high efficiency and widespread applications of thermoelectric materials.
The first generation of bulk thermoelectric materials has low efficiency. The interdependence of thermoelectric properties led to concerted efforts to develop strategies in the past decade to enhance thermoelectric efficiency, including improving the Seebeck coefficient and electrical conductivity as well as reducing thermal conductivity. Several strategies, including nanostructuring and nanocom-positing, have been developed and have led to remarkable progress in enhancing thermoelectric performance and the efficiency of thermoelectric materials [2].
Various materials have been used for thermoelectricity over a wide range of temperatures, including polymeric and inorganic materials. However, inorganic semiconducting materials were found to be the most efficient for thermoelectric applications [3,4]. Among inorganic materials, layered chalcogenides have attracted attention as they demonstrate high ZT values with high electrical properties and low thermal conductivity [2]. Layered materials exhibit layered structure with strong covalent bonds within the layer and weak van der Waals forces between the layers. The layered structure provides a way to separate the movement of charge carriers and phonons; charge carriers could effectively move within the layers while the phonons may scatter significantly between the layers due to the presence of weak van der Waals forces. Several strategies can be used to improve thermoelectric properties, such as doping, alloying, nanostructuring, and nanocompositing [5,6].
The ZT value has been improved by creating different nanostructured materials with different shapes and forms such as nano-bulk, quantum dots, nanowires, and nanofilms. The high density of the grain boundaries presented in those materials results in exceptional thermoelectric properties. Grain boundaries play an important role in reducing lattice thermal conductivity by phonon scattering as well as in modifying electrical properties by energy filtering low carriers. Compared with conventionalsized materials, engineering of nanostructuring boosted the efficiency of thermoelectric materials to a new level for several thermoelectric systems. The breakthrough in nanotechnology demonstrates the advantages to improve thermoelectric performance that led to a burst in the field of thermoelectric nanomaterials [7].
Nanocomposites, which include a combination of multiple materials of which at least one has a nanometer dimension, have gained considerable attention as well. Layered chalcogenide thermoelectric materials have been mixed with several additive materials. It has been found that incorporating nano additive as a second phase of the thermoelectric material matrix is a simple top-down approach that allows optimization of thermoelectric properties to improve electrical properties and reduce thermal conductivity. However, selecting the type of additive is an important step to achieve the best possible improvement in electrical properties and the best possible reduction in electrical conductivity. These requirements open the door to the search for nano additives with excellent electrical properties in order to improve power factors as well as to contribute to a reduction in thermal conductivity due to their nanosize.
Graphene is a two-dimensional sheet that is one atom thick with superior electrical and thermal properties. It has attracted the attention of researchers since its discovery in 2004 [8]. The availability of bulk quantities of high-quality graphene [9] makes it a suitable material that can be utilized as an additive for preparing different functional thermoelectric composites with different matrix materials [10,11]. A small amount of graphene embedded into the thermoelectric matrix can play an essential role in enhancing the efficiency of the thermoelectric composites [12]. However, among layered chalcogenide materials, graphene has been used recently as an additive for bismuth telluride.
The aim of this review is to conduct a comprehensive study of layered chalcogenide materials and the recent work and progress made for improving their figure of merit. First, it discusses thermoelectric properties and different types of thermoelectric materials. Then it discusses in detail, layered chalcogenide thermoelectric materials, by giving a basic knowledge of their structure and thermoelectric properties. It summarizes the effect of different mechanisms used in the literature to improve their ZT, including doping and alloying, nanostructuring, and nanocompositing. It sheds light on graphene as a distinctive material that is used in the field of thermoelectricity to improve thermoelectric properties. Finally, it overviews graphene bismuth telluride composites and how their thermoelectric behavior is affected by the addition of graphene. The limited work on graphene composites showing significant improvement has inspired further work for enhancing other layered chalcogenide materials and thermoelectric materials.

Thermoelectric Materials, Basic Theory and Categories
Thermoelectric materials were discovered in the 18 th century, and since then, many developments have been achieved. The history of thermoelectric materials is characterized by the progress of the figure of merit (ZT) [1,13,14], which is a thermoelectric quantity related to the conversion efficiency of thermoelectric materials. Three main properties are contributed to the value of ZT, Seebeck coefficient, electrical conductivity and thermal conductivity. Seebeck coefficient is an electrical property defined as the amount of voltage generated per unit of temperature differ-ence, in units of V/K. Figure 1 illustrates the concept of the Seebeck effect [3]. Electrical conductivity is the ability of the charge carriers to travel through the material under an applied electric field and is given by: where n is the concentration of charge carriers, e is the magnitude of the electron charge, and µ is the charge carrier mobility. Thermal conductivity is the energy induced from applied temperature and transported through the material by both phonons and charge carriers. Thermal conductivity is the sum of the heat transported by charge carriers (ke) and traveling of phonons through the lattice (k l ). If one type of charge carrier is predominant, then k tot = ke + k l . Good thermoelectric materials should possess as high ZT as possible; therefore, it is reasonable to expect that materials with high electrical conductivity, a high Seebeck coefficient, and low thermal conductivity are required for thermoelectric applications. However, materials that possess all these properties simultaneously are not available in nature. Metals have high electrical conductivity but low Seebeck coefficient and high thermal conductivity. Insulators exhibit low thermal conductivity and poor electrical conductivity. Hence, optimizing all thermoelectric properties is difficult due to the interdependency of the parameters. The interdependency among the parameters is represented by carrier concentration. Any favorable change in carrier concentration to improve electrical conductivity is accompanied by an unfavorable change in another property, such as increasing the electrical contribution for thermal conductivity and decreasing the Seebeck coefficient. Therefore, any optimization of the parameters should consider the carrier concentration [15]. The ideal thermoelec-tric material is called a "phonon glass electron crystal," which means it has low thermal conductivity as in glass and high electrical conductivity as in a well-ordered crystal. After a series of experiments, it was determined that heavily doped semiconductors are the most efficient bulk thermoelectric materials as they have the highest power factor together with low thermal conductivity [16]. Most of the studies in thermoelectricity and almost all commercial modules have focused on inorganic thermoelectric materials [3,17] which includes skutterudites, clathrates, oxides, nitrides, pnicogens, half-Heusler alloys, Zintl-phase, and metal chalcogenide materials. Among inorganic materials, chalcogenides are considered the prime thermoelectric candidate and have been used extensively due to their reliability and better thermoelectric performance than other materials [7]. The term chalcogenide referred mainly to tellurides (Te), selenides (Se), and sulfides (S). Layered chalcogenides have a high figure of merit mainly because of their low thermal conductivity. Low thermal conductivity is attributed to the anisotropic crystal structure and strong lattice anharmonicity. Their relatively heavy atoms compared with other thermoelectric materials efficiently help to reduce thermal conductivity [2]. Layered chalcogenides could be easily formed into different types of structures. Moreover, their layered structure enables the modification and development of their properties at several levels using a variety of strategies. On the atomic scale, reduction of thermal conductivity by scattering phonons could be attained with the help of a layered structure by intercalation of guest atoms and layers between the host layers. Further, the low energy vibration of atoms could effectively scatter phonons. In the nanoscale, any stacking faults in the layered structure could result in a further reduction in lattice thermal conductivity, whereas in the microscale, the highly oriented microstructure improves the power factor by enhancing carriers' mobility in the in-plane directions [18].

Layered Chalcogenide Materials
Layered chalcogenides represent a massive group with multiple materials. In the following section we tried to classify layered materials into groups in which each group has certain common features between the members: (1) bismuth and antimony based materials which are known for their high performance near room temperature; (2) oxychalcogenides; (3) tin-and indium-based materials [19]; (4) homologous series; (5) layered sulfides, including titanium sulfides; and (6) ternary ACrX 2 layered struc-tures [20]. The structure of chalcogenide materials is characterized by layers; the bonds in each layer are strong covalent bonds, which provides in-plane stability. The interaction that holds the layers together is van der Waals. According to the chemical bonding and anisotropic nature of the crystal structure, there is variation in thermoelectric properties with different crystallographic directions within the crystal [19]. Details of the structure of individual material and their thermoelectric properties and discussion of how the anisotropic nature of layered materials affects their thermoelectric properties are presented in the following sections.

Bismuth and Antimony-Based Layered Materials
In this section we included Bi and Sb telluride and selenide materials, whereas sulfides' compounds are excluded and discussed under the sulfide group. Although sulfur lies in the same group as Te and Se, however, bismuth sulfides adopt an orthorhombic structure different than rhombohedral structure adopted by bismuth tellurides and selenides. Sulfides show lower electrical conductivities compared to selenide and telluride because of their higher bandgap [21][22][23]. Bismuth telluride (Bi 2 Te 3 ) is a narrow bandgap n-type semiconductor of (0.21 eV) and an indirect bandgap of 0.15 eV. Bismuth telluride has a distinctive rhombohedral layered crystal structure with five atoms in each unit cell forming quintuple layers of Te (1) -Bi-Te (2) -Bi-Te (1) stacked together by weak van der Waals forces along the c-axis as shown in Figure 2a. The Te atoms are presented in two different chemical states as the subscripts 1 and 2 show. However, within the layers, Bi-Te(2) atoms are bonded by a strong covalent bond, whereas Bi-Te(1) atoms are bonded by a covalent-ionic bond [2].
Since its discovery 60 years ago, bismuth telluride was first used by Goldsmith as an effective material for refrigeration purposes. It has been used as the most promising thermoelectric material with a high figure of merit (1.14) near room temperature (200-400 K) [22]. Because of the weak van der Waals bonds between layers, Bi 2 Te 3 exhibits high thermoelectric properties and demonstrates easy cleavage along the (001) plane [2].
Bismuth selenide (Bi 2 Se 3 ) is a narrow bandgap (0.3 eV) n-type semiconductor. Analogous to bismuth telluride, it has a rhombohedral anisotropic layered structure with quintuple layers and five atoms in each unit cell [22]. The two Bi atoms and three Se atoms are packed similar to Bi 2 Te 3 pattern Se(1)-Bi-Se(2)-Bi-Se(1) with weak van der Waals bonds between neighboring Se(1) planes as shown in Figure 2b. Compared with bismuth telluride, bismuth selenide possesses high thermal conductivity because of the light atomic weight of Se. Therefore, the thermoelectric performance of Bi 2 Te 3 is much better than Bi 2 Se 3 [2].
Antimony telluride (Sb 2 Te 3 ) is a layered rhombohedral crystal structure p-type semiconductor. Its unit cell consists of five atoms packed with a similar Bi 2 Te 3 and Bi 2 Se 3 pattern, Te(1)-Sb-Te(2)-Sb-Te(1). The layers are connected by weak van der Waals bonds, so the cleavage is apt to occur perpendicular to the c-axis (cross-plane direction). As an intrinsic semiconductor, Sb 2 Te 3 has attracted attention for its high electrical conductivity and low thermal conductivity. It is the leading material for thermoelectric applications near room temperature (300-500 K) [24].

Oxychalecognide Materials
Layered oxychalcogenides thermoelectric materials have gained considerable attention due to their intrinsically low thermal conductivity, moderate Seebeck coefficient and tunable electrical conductivity [25].
The Bi-O-X system is a group of oxy-layered materials that has a complex tetragonal structure. One example is Bi 2 O 2 Se, which consists of tetragonal (BiO)n layers with Se atoms occupying interlayer positions as shown in Figure 3. Due to the strong electronegativity of oxygen atoms, oxychalcogen systems suffer from low electrical conductivity. However, they exhibit a moderate Seebeck coefficient.
BiCuOCh (Ch = S, Se, Te) is a group of oxychalcogenides that have the structure of space group = P4/nmm and consist of alternately stacked (Bi 2 O 2 ) +2 insulating lay-  ers and (Cu 2 Ch 2 ) −2 conducting layers as shown in Figure 4. They exhibit extremely low thermal conductivity and have poor electrical conductivity due to low carrier mobility. BiCuOSe was recently reported as a promising thermoelectric material for its high ZT (> 0.8) at a high-temperature range (923 K). It is a p-type material composed of alternating insulating (Bi 2 O 2 ) 2+ layers with conductive (Cu 2 Se 2 ) 2− . The thermoelectric measurements suggest that the electrical properties of BiCuSe is lower than other thermoelectric systems; thus the high thermoelectric performance of Bi-CuSe originates from its very low thermal conductivity instead of its power factor [26][27][28]. Compared with BiCuSeO, BiCuOTe has a higher power factor as it has a better hole mobility attributed to the antibonding between Cu and Te compared with antibonding between Cu and Se. Therefore hole mobility can be improved significantly in BiCuSeTe compared with BiCuSeO [26].

Homologous Series
Homologous layered materials are an interesting class of thermoelectric materials that are built on the same structure of a certain model and can be expanded by regular increments in different directions [18,22]. Examples of homologous thermoelectric series are CsBi 4 Te 6 , a member of the Cs 4 [Bi 2n +4Te 3n +6] homologous series, and Pb 5 Bi 6 Se 14 , a member of the cannizzarite homologous series.
CsBi 4 Te 6 is a p-type semiconductor with a narrow bandgap of (0.1 eV), which is nearly half the value compared with Bi 2 Te 3 . It has an anisotropic layered structure with Bi 4 Te 6 slabs interconnected by Bi-Bi bonds parallel to the b-axis and altered with Cs + ionic layers as shown in Figure 5a. CsBi 4 Te 6 is a potential candidate for thermoelectric application and exhibits a high figure of merit of 0.8 at low temperatures, which is attributed mainly to the narrow bandgap [18,22]. Pb 5 Bi 6 Se 14 is an n-type semiconductor. Its crystal structure consists of infinite alternating PbSe-and Bi 2 Se 3type layers along the c-axis, forming a three-dimensional structure, as shown in Figure 5b. The interfaces between the PbSe and Bi 2 Se 3 layers scatter phonons as well as charge carriers, leading to a highly anisotropic electrical mobility and lattice thermal conductivity between in-and cross-plane directions. Electrical conductivity is much higher in-plane direction than cross-plane direction, whereas lattice thermal conductivity is slightly higher for the in-plane direction than for the cross-plane direction. However, the Seebeck coefficient is insensitive to the crystal orientation and exhibits nearly similar values for both in-and cross-plane directions. As a result, a higher ZT of 0.46 has been achieved for the in-plane direction at 75 K. The thermoelectric properties of homologous compounds can be controlled by modifying the shape and size of the structural module. Therefore, homologous materials are a suitable platform for developing new thermoelectric materials [18].

Sn-and In-Based Layered Materials
Sn-and In-based systems are typical thermoelectric materials with low cost and good thermoelectric performance. Tin selenide (SnSe) is a p-type semiconductor with simple orthorhombic, layered-structure material with a zigzag accordion chain along the b-axis and a bandgap of 0.86 eV. Each layer consists of two-atom-thick slabs (Se and Sn) along the in-plane direction, with strong Sn-Se bonds within the layer and weak Sn-Se bonds between layers along the cross-plane direction, as shown in Figure 6a. At a temperature of 800 K, tin selenide undergoes reversible transition distortion from the orthorhombic lowtemperature Pnma space group to the orthorhombic hightemperature Cmcm space group. At this transition temperature, the power factor reaches its maximum value as the Cmcm phase exhibits a reduction in energy with enhancement in carrier mobility while preserving the low thermal conductivity [15,29]. Similarly, In 4 Se 3 forms a layered crys-tal structure of (In 3 ) +5 clusters covalently bonded to Se ions in the in-plane. Along the cross-plane direction, the planes are held together by van der Waals interactions as shown in Figure 6b. Tin selenide is a p-type semiconductor with a relatively low carrier density of 10 17 -10 18 cm −3 . At room temperature, it possesses a high electrical resistivity of 10-10 5 Ωcm. Even with high electrical resistivity, the obtained power factors of SnSe crystals are moderate at high temperatures as the crystals possess a high Seebeck coefficient and are much higher than other materials with low intrinsic thermal conductivity [29,30]. Because of the high electrical resistivity, tin selenide was ignored by thermoelectric researchers for a long time [31], until Sassi, S, et al. reported a high ZT of 2.6 for a single crystal along the b-axis, 2.3 along the c-axis, and 0.8 along cross the plane direction (a-axis) [15].
SnS recently has attracted interest as it has the same structure and chemical bond of tin selenide, as shown in Figure 6c. It is p-type semiconductor, environmentally friendly and cost-effective, and its resources are earthabundant. Similar to SnSe, SnS possesses extremely low thermal conductivity (below 1 Wm −1 K −1 ) with a high Seebeck coefficient, which makes it a promising material for thermoelectric application [32].

Layered sulfides
Multiple Sulfides materials have been discovered over the past 10 years and have proven their efficiency as environmentally friendly thermoelectric materials. TiS 2 is a misfitlayered sulfide that has a layered crystal structure connected by van der Waals forces along the cross-plane direction. TiS 2 has a wide range of chemical compositions with different concentrations of Ti atoms ranging from stoichiometric up to 1+x, where 0 ≤ x < 1. The excess Ti atoms occupy the places between layers because they are weakly bonded. TiS 2 is considered an efficient n-type material for thermoelectric applications as it is composed of light, lowtoxic, and earth-abundant elements. TiS 2 shows the best performance at high temperatures for both single as well as polycrystalline forms. The TiS 2 single-crystal possesses a high power factor; however, its ZT is limited to~0.16 at 300 K because of the high lattice thermal conductivity [18]. Therefore, the intercalation of the other layers, such as PbS, BiS, and Sn, has been found to be an effective way of reducing its thermal conductivity. Figure 7a shows a TiS 2 compound structure with the structure of other materials intercalated between its layers [33].
Misfit layered sulfides are a large layered material group with a general formula [MS] 1+m [TS 2 ]n, where M = Sn, Pb, Sb, Bi, or rare-earth metals; T = Ti, V, Cr, Nb, and Ta; and n = 1, 2, 3 [34] and 0.08 <m <0.28. These materials consist of alternating layers of MS and TS 2 where the TS 2 layer provides the electron path way and thermopower whereas the intercalated MS layer suppresses the transport of phonons [35]. The crystal structure and physical properties of this misfit family were studied extensively in the 1990s, however, not many reports on their thermoelectric properties exist to date. The crystal structure of [LaS] 1.14 NbS 2 is shown in Figure 7b as an example. Recently, misfit-layered sulfides have gained interest in thermoelectric research for their good thermoelectric behavior, where the intercalated NaCl-type MS layer is primarily responsible for the disorder (reduction in κ lat ) and the CdI 2 -type TS 2 layer behaves as a charge-carrier pathway [18]. The entire crystal structure is stabilized through charge transfer from the MS layer to the TS 2 layer. Wan et al. [33] studied a series of n-type (MS) 1+x (TiS 2 ) 2 (M = Pb, Bi, Sn) misfit layer compounds and found that they exhibit very low thermal conductivity due to the weak interlayer bond. The figure of merit values of these compounds were found to range between 0.28 and 0.37 at 700 K with a maximum ZT of 0.37 for (SnS) 1.2 (TiS 2 ) 2 . Their ZT is affected by the microstructure due to the anisotropic nature in the atomic bonds of the misfit layered compounds [18]. Therefore, tuning its microstructure, which is mainly affected by its synthesis, is one strategy to enhance its ZT. Jood et al. [34] prepared misfit-layered n-type (LaS) 1.2 CrS 2 and p-type (LaS) 1.14 NbS 2 by CS 2 sulfurization for 6 and 12 hours followed by the pressure-assisted consolidation to produce randomly and highly oriented samples whose microstructures were tunable. A highly oriented texture sample pro- duced the highest ZT of 0.14 at 950 K among (LaS) 1.2 CrS 2 systems, while the weakly oriented texture produced the highest ZT of 0.15 at 950 K among (LaS) 1.14 NbS 2 .
Recently, a new class of n-type layered sulfide materials (MnBi 4 S 7 and FeBi 4 S 7 ) has been reported. They exhibit a complex, monoclinic, low-symmetry layered structure shown in Figure 8. The complex structure and the presence of lone pair electrons with Bi +2 are the main reasons for the low thermal conductivity of these materials. In addition, they have a relatively high power factor making these materials good candidates for low and intermediatetemperature thermoelectric applications [36]. Bismuth sulfide (Bi 2 S 3 ) has received interest for use in thermoelectric applications due to the high abundance of sulfur on earth [37]. It is n-type material that crystallizes in a highly anisotropic orthorhombic structure with a space group Pnma that contains 20 atoms per unit cell. It has a wide direct bandgap of 1.3 eV; thus it has low electrical conductivity in pristine form and has a low ZT value [20,38,39].
BiS 2 -based compounds show relatively high thermoelectric performance. Their crystal structure is composed of a BiS 2 -conducting layer stacked alternately with an insulating layer. An example of BiS 2 -based thermoelectric compounds is SrFBiS 2 . SrFBiS 2 is an n-type material that belongs to the tetragonal P4/nmm space group, and its structure is built up by stacking alternately a SrF layer with NaCl-type BiS 2 layer, as shown in Figure 9. SrFBiS 2 has low thermal conductivity with a relatively high power factor at low temperatures. SrFBiS 2 is expected to be a parent for a group of material by doping it with different materials. Doping is a achieved by substituting Sr +2 site with rareearth metal (RE +3 ) [40,41].
AgBi 3 S 5 is a new nontoxic n-type thermoelectric material with high thermoelectric performance. It has a complex structure composed of NaCl-type fragments [21]. It has very low thermal conductivity in the temperature range of 300-800 K due to the unusual vibrational properties associated with Ag and Bi atoms [42].

Ternary ACrX2-Layered Structures
ACrX 2 -layered compounds, where A represents Ni, Cu, or Ag and X represents S or Se, are a group of magnetic materials that exhibit good thermoelectric performance at low to moderate temperatures [23]. ACrX 2 compounds possess a trigonal-layered structure (R3m), where Cr +3 cations are in distorted octahedral coordination to the X −2 anions, and A atoms (Ag or Cu) fill the interlayer spaces and occupy tetrahedral sites, as shown in Figure 10. Within the layer, the atoms of CrX 2 are bounded by strong ionic bonds, while the bonds are weak van der Waals between the neighboring triple layers. The thermoelectric performance of this group depends strongly on their synthesis. A highly textured p-type CuCrS 2 sample with high electrical conductivity has been obtained via a specific heat treatment process, whereas the same sample fabricated by ball milling and spark plasma sintering methods exhibited poor thermoelectric performance with a ZT of 0.11. A maximum ZT of 2 at 300 K and 1 at 848 K was reported by Tewari et al. and Gascoin et al. [43,44] for CuCrS 2 and AgCrS 2 , respectively. A ZT of 1.4 was achieved in sandwich-like p-type (AgCrSe 2 ) 0.5 (CuCrSe 2 ) 0.5 at 773 K [45].

Anisotropy
Due to the layered crystal structure of layered chalcogenide materials, the materials exhibit a significant anisotropic structure and properties of in-plane direction versus cross-plane direction. In a single crystal, the lamellar structure leads to an easy cleavage of planes that are perpendicular to the cross-plane direction. The atomic interactions within the plane are characterized by covalent bonds and the layers are held by weak van der Waals bonds along the cross-plane direction. The difference in the nature of the atomic bonds in various directions results in anisotropy in thermoelectric properties when measured in different directions. Thermal and electrical conductivity, in most of the materials, are much lower in the cross-plane than within the plane direction [46]. This is attributed mainly to weak van der Waals interactions between the planes. The weak bonds act as a thermal and electrical boundary that heavily scatter phonons and electrons [47]. In several studies, the Seebeck coefficient was found to be isotropic and independent of the crystallographic direction [46,[48][49][50][51]. However, in some other studies [34,52], the Seebeck coefficient exhibit anisotropy in different directions. Zhao et al. [50] reported on the thermoelectric performance of a SnSe single crystal with significant anisotropy in different directions. They reported ultrahigh ZT equals 2.6 in the b direction (within the plane), and a high ZT equals 2.3 in the c direction (which is inplane direction, perpendicular to b but in the same plane). They reported a significantly reduced ZT of 0.8 across the plane direction (perpendicular to the bc plane). Similarly, thermal and electrical conductivities have been found to be two and four times higher for the in-plane direction than the cross-plane direction in an n-type Bi 2 (Te 1−x Sex) 3 single crystal, whereas no significant difference in the Seebeck coefficient has been reported. In total, ZT has been found to be two times higher within the plane direction than in the cross-plane direction [48]. In polycrystalline and nanostructured materials, it has been found that for the crystals, the grains tend to be oriented in a significant preferential orientation due to the lamellar structure when they are pressed or sintered, whereas the nanostructured material parts tend to grow in a preferred direction. The anisotropic nature of the bonds, where intralayer covalent bonds are stronger than interlayer covalent bonds, led grains to grow perpendicular to the applied pressure [34,53]. Hence, they exhibit anisotropic thermoelectric properties in the direction of pressing that is not perpendicular. Bi 2 S 3 has a lamellar structure that forms a long chain parallel to the (001) direction. The XRD pattern of the prepared Bi 2 S 3 nanoribbon proved the anisotropic structure and the tendency of the ribbon to grow along the preferential (001) direction [54]. Based on the XRD pattern for the surface of the several sintered pellets of (MS) 1+x (TiS 2 ) 2 materials, Wan et al. [33] reported that the crystals preferred to align perpendicular to the pressing direction as they observed sharp (00l) peaks, which represent the planes perpendicular to the pressing direction; very few (h k l) planes were detected, as seen in Figure 11. The degree of orientation of the planes is high so that the thermoelectric properties of the polycrystalline samples approach the values of the properties reported for in-plane single crystals for the same materials. Chere et al. [51] found that the grains in the p-type SnSe polycrystalline material grew preferentially along the (400) and (111) planes. Therefore, anisotropic thermoelectric properties have been observed. Electrical conductivity was very high when measured perpendicular to the press direction, but thermal conductivity was much lower along the pressing direction leading to higher ZT along the pressing direction. Sassi et al. [53] reported anisotropic thermoelectric properties in homologous ntype Pb 5 Bi 6 Se 14−0.25 I 0.25 . A bar-shaped compressed sample was cut in parallel and perpendicular directions to the pressing direction, and thermoelectric properties were measured in both directions. ZT was calculated to be 0.5 in the perpendicular direction and about 0.35 in the parallel direction. In another study, Xu et al. [49] found that the electrical conductivity of n-type SnSe 2 is about two times higher in-plane direction than in the cross-plane direction, whereas thermal conductivity is about twice or three times higher. In another study [34], the XRD pattern of the surface perpendicular to the pressing direction of the n-type (LaS) 1.2 CrS 2 and (LaS) 1.14 NbS 2 compacted samples show that the crystals are oriented along the (001) planes. Highly anisotropic thermal and electrical properties have been  [71] observed in the in-plane and cross-plane directions. For example, electrical resistivity was~170 and 280 µΩ m at 950 K for in-plane and cross-plane directions, respectively. This is mainly because the oriented texture allows higher mobility of carriers within the plane, whereas scattering them at the interfaces between the layers. Similarly, thermal conductivity is lower in the cross-plane than in the in-plane direction because the interfaces scatter phonons. The large grains reduce the number of phonons scattered at grain boundaries. Grain alignment in a preferred orientation in polycrystalline materials has been found to be effective in enhancing the ZT by enhancing the power factor in the in-plane direction. However, in several studies [34,46] where polycrystalline materials were prepared in nanograin forms, it was found that the grains are randomly oriented, and in turn, there was no significant improvement in ZT after nanostructuring. Although nanostructuring has a significant effect on reducing thermal conductivity, the randomness in grains orientation has a negative impact on the power factor, which in turn does not show any enhancement in total ZT. Therefore, it is a challenge to keep small grains with partial alignment to improve the power factor to a level close to that of a single crystal while keeping the fine grain to reduce thermal conductivity. Yan et al. [46] achieved partial grain alignment in ntype Bi 2 Te 2.7 Se 0.3 by repressing the pressed samples with a larger diameter die at a higher temperature. By this way, small grains were forced to be aligned in-plane perpendicular to the pressing direction. They reported an enhancement in ZT from 0.85 for random grains to 1.04 after partial alignment of grains perpendicular to the press direction. Although ZT along the pressing direction decreased, this will not be considered because only the direction of high ZT is used in thermoelectric devices. Jood et al. [34] reported the effect of the degree of orientation on thermoelectric properties of n-type (LaS) 1.20 CrS 2 and p-type (LaS) 1.14 NbS 2 materials. They prepared highly oriented and randomly oriented microstructure samples of both materials. They found that both highly and randomly oriented samples of (LaS) 1.20 CrS 2 material provide ultralow thermal conductivity perpendicular to the pressing direction. The highly oriented texture provides higher electrical conductivity, which results in a higher power factor and higher ZT in the in-plane direction in both samples. Similar to (LaS) 1.20 CrS 2 , the power factor of the oriented (LaS) 1.14 NbS 2 sample was higher (22%) than the random texture in the in-plane direction as a result of higher Seebeck coefficient and higher electrical conductivity. However, it possesses higher thermal conductivity, which was 40% higher in the oriented sample than the random one, which overcomes the increment in power factor and leads to a lower ZT in the oriented sample. Table 1 illustrates examples of layered chalcogenide materials with their ZT at different temperatures.

Strategies for Improving Thermoelectric Performance
Traditional thermoelectric materials did not possess ideal thermoelectric properties. Designing and optimizing thermoelectric materials with high efficiency is a complex and challenging task. It is believed that the best thermoelectric materials provide amorphous phonon glass to achieve glass-like lattice low thermal conductivity with crystal electronic structure where electrons have high electric conductivity. The phonon glass region is suitable for hosting disorders and doping for blocking phonons without affecting the carrier mobility in the electron crystal region. Researchers have searched for thermoelectric materials that have a complex crystal structure and large molecular weight [29]. Skutterides and clathrates are, for example, cage-like materials with voids that could be occupied by foreign atoms. The inserted atoms have the ability to rattle in the cage and result in scattering the phonons and enhancing lattice thermal conductivity [72]. Layered chalcogenide materials possess low thermal conductivity because of their intrinsic lattice anharmonicity [73], which is measured by the Gruneisen parameter. The Gruneisen parameter is an indication of how much phonon vibrations are deviated from harmonic oscillation in crystal lattice. Anharmonicity of the chemical bonds plays an important role in limiting the lattice thermal conductivity [28,74]. In addition, layered chalcogenides possess low thermal conductivity in the cross-plane direction because of the weak van der Waals interactions (soft bonds) between the layers; however, for those materials to be widely used in commercial applications, further reduction of thermal conductivity is still required. Besides, the Seebeck coefficient and electrical conductivity must be improved [22]. Several strategies have been followed with the aim of decoupling the parameters and attaining a high power factor, low thermal conductivity, or a combination of both. Improving the power factor can be attained by appropriate carrier doping, quantum confinement, energy carrier filtering, lowering the effective mass, and modifying the band structure by alloying [30,75]. Thermal conductivity could be reduced by adding interfaces in order to intensify phonon scattering at the atomic, nanoscale, and microscale by forming solid solutions (alloying), nanostructuring, or all-scale hierarchi-cal architecturing [30,75]. It is worth noting that the evolution of thermoelectric properties is not limited to the use of only one method. More than one method has been utilized simultaneously in several studies to attain the highest improvement in thermoelectric performance. Methods for improving thermoelectric performance include doping, nanostructuring, and nanocompositing.

Doping
Adjusting the carrier concentration shows the best improved thermoelectric effect in semiconductors. Heavily doped semiconductors are the best thermoelectric materials with a greater power factor than the undoped counterpart [3,76]. Dopant additions have been shown to have a significant impact on the electron density distribution in thermoelectric chalcogenides [59,77]. Moreover, doping can play an important role in reducing the lattice thermal conductivity by creating defects and scattering phonons.
However, the doping level should be optimized to a certain limit in a way that increases electrical conductivity and does not affect the Seebeck coefficient [3,7,78]. Wei et al. [79] studied the effect of Na, Li, and K doping with ptype SnSe. The results showed that Na is the most efficient dopant. The addition of 1% Na to SnSe increased the carrier concentration, which led to a maximum ZT of 0.8. In this study, the dopant atoms disturb the pseudopotential of the lattice and act as scattering centers for the carriers, which in turn affect the carrier's mobility. The difference in size between the matrix atoms and the dopant leads to an extra scattering of carriers. As the difference of radius increases (in case of Li + and K + ), more scattering of the carriers is expected. This is likely the reason why the mobility of carriers decreases more than Na and Ag. Silver has been found to be an effective dopant compared with Na when added to SnSe because of its high solubility when subjected to repeated heating and cooling. According to these results, Ag was able to increase the carrier concentration to 9×10 18 cm −3 , leading to an improvement of the electrical conductivity and power factor [80,81]. However, similar to the previous study, the mobility of the doped sample has been found to be much lower than the pristine sample (12 cm 2 V −1 s −1 and 7 cm 2 V −1 s −1 for pristine and 5% Ag-doped SnSe, respectively), which suggest that Ag atoms scattering the carriers. In another study [77], it was found that doping n-type SnSe with PbBr 2 led to a significant improvement in carrier concentration whereas the carrier mobility is affected negatively by increasing carrier concentration. Overall, electrical conductivity was found to be higher for the first addition of 1% and 2% PbBr 2 than those of 3% and 4% which is due to decrease of the mobility after scattering of the carriers. However, doping with 3% PbBr 2 lead to an enhancement of ZT to 0.54 at 793 K. Zhang et al. [82] studied the effect of several dopings of p-type SnTe and found that 0.25 atom% Indoped SnTe improved the Seebeck coefficient significantly and increases ZT to 1.1 at 873 K. However, electrical conductivity was found to decrease with doping due to the reduction of carrier mobility, which indicates that dopants affect the carriers' mobility. In addition to the improvement in electrical conductivity, ultralow thermal conductivity has been achieved when doping p-type SnSe with 0.01 Cu, leading to a total improvement of ZT to 1.2 at 873 K [83]. The enhanced electrical conductivity is attributed to the high carrier concentration induced by Cu dopants, whereas the presence of Cu atoms plays an important role as scattering centers for the phonon. A significant reduction in the lattice thermal conductivity, which led to a measurable improvement in ZT, was reported by Putri et al. [59], who used chromium as a dopant for n-type (BiS) 1.2 (TiS 2 ) 2 misfit layer compound. The calculated ZT values for doped samples with different doping amounts (x = 0.025, x = 0.05, x = 0.1) exhibit an increase in ZT to 0.24-0.3. The main contribution for chromium doping in this study corresponds to its role in reducing total thermal conductivity. Indium has been added to Bi 2 Te 3 , and a maximum ZT of 1.1 was achieved at 448 K [84]. The enhancement was attributed to an effective increase of electrical conductivity because of high carrier concentration and a reduction of thermal conductivity due to phonon scattering. In this study, several amounts of In were added; however, In with x = 0.1 exhibited the best thermoelectric behavior. Although doping increases the carrier concentration, which contributes   [145] to enhancement in electrical conductivity, the effect of alloying scattering on carriers' mobility is also significant. If the effect of doping on increasing carrier concentration does not overcome the decrease of carriers' mobility, electrical conductivity cannot be improved. Therefore, doping should be optimized to a value for which increasing the carrier concentration overcomes the effect of alloying. Figure 12 shows a graph of electrical resistivity versus different iodine contents; resistivity decreases at an optimum doping level, which results in improved electrical conductivity. Doping has been found to have a significant effect on shifting the service temperature at which the peak ZT value is achieved. For example, indium doping [85] in ptype Bi 2 Te 3 showed an enhancement of ZT up to 1.4 and shifted the service temperature of Bi 2 Te 3 from room temperature range to 500 K. In this study, it was found that doping has a significant effect in broadening the bandgap, hence suppressing the bipolar effect and increasing the Seebeck coefficient. In addition, doping optimized the concentration of intrinsic point defects, which in turn led to a reduction in thermal conductivity. Shifting the maximum ZT to a high temperature, 380 K, has been reported when alloying p-type Bi 2 Te 3 with Sb [86]. Several other studies reported a shift in the service temperature. Liu et al. [87] reported a peak ZT of 0.8 at 573 K for n-type Bi 2 Te 2 S. Hua et al. [88] reported a value of 0.92 at 710 K for p-type Ag 0.01 Sb 1.85 In 0.15 Te 3 . Wang et al. [89] reported a peak ZT value of 0.86 at 600 K for n-type Bi 2 Te 1.5 Se 1.5 . In addition to doping, alloying of elements was found to be an effective method for improving thermoelectric performance but differently from the doping effect. With alloying, the power factor increases while thermal conductivity decreases. The reduction in lattice thermal conductivity is a result of the atomic disorder generated by the alloyed foreign element. The guest material acts as a scattering center for phonons, and the degree of scattering depends on the mass ratio of the alloy constituents [3,90]. Simultaneously, alloying is an efficient way to improve the power factor by modifying the band structure and creating a path for carrier transport [30]. Alloyed semiconductors with high carrier concentration are the best bulk thermoelectric materials. The thermoelectric properties of antimony telluride when alloyed with bismuth telluride improve significantly. The power factor and figure of merit of 25% Bi 2 Te 3 -75% Sb 2 Te 3 alloy at room temperature reached 3.1×10 −3 Wm −1 K −2 and 0.85, respectively [91]. In another study, the maximum power factor reached 3.81×10 −3 Wm −1 K −2 at 300 K and ZT reached 0.97 with 80% Sb 2 Te 3 [92]. Table 2 summarizes some layered chalcogenide materials doped with various dopants and their ZT.

Nanostructuring
Reducing at least one structural dimension of the thermoelectric materials to the nanoscale range by nanostructuring is a valid approach for improving the figure of merit ZT. Nanostructuring can be controlled either by producing nanomaterials such as nanofilms, nanowires, quantum dots, nanoparticles, and superlattices or by decreasing the grain size to the nanoscale range, as in the case of nanobulk materials [3]. Nanostructuring is an efficient way to reduce thermal conductivity and enhance the power factor by generating many interfaces in thermoelectric materials that can block phonon transport. Grain boundaries, especially when the grain size is comparable with the mean free path of phonons, can scatter phonons more efficiently than electrons while acting as a potential barrier for lowenergy charge carriers. Filtering of low energy charge carriers has a significant effect on keeping the potential difference through the material and can enhance the Seebeck coefficient [90,146]. However, in some cases, carrier mobility is affected by the disturbance arising from phonon scattering, leading to a reduction in electrical conductivity [7]. High ZT by nanostructure engineering was first proposed theoretically by Hicks and Dresselhause [170]. They published a theoretical model calculating the Seebeck coefficient and thermal conductivity when confining the electron in a two-dimensional quantum well. Their model predicted improvement of ZT as a result of enhancing the Seebeck coefficient and reducing thermal conductivity due to phonon scattering at the interfaces. Not long after, Venkatasubramaniam et al. synthesized Bi 2 Te 3 /Sb 2 Te 3 superlattices and reported a high ZT value of 2.4 [148]. Several studies have employed nanostructures and success-fully produced nanomaterials with a high value of ZT due to a reduction in thermal conductivity [3,149,150], whereas some other studies reported a significant improvement in the Seebeck coefficient in addition to the reduction of thermal conductivity [24,[149][150][151]. For example, low thermal conductivity with a total improvement in ZT has been achieved in ball-milled p-type SnTe as a result of phonon scattering at grain boundaries [82]. A similar improvement due to a significant reduction in thermal conductivity has been achieved in n-type Bi 2 Te 2.7 Se 0.3 ultrathin nanowires. In this study [149], the ZT value was increased to 0.95, which is 13% higher than the conventional grain size sample. In contrast, preparing nanobulk Bi 2 Te 3 represents a 30% improvement in the power factor compared with the conventional sample, especially at low temperatures (325-425 K) [151]. Wan et al. [152] prepared layered metal sulfides (BiS, SnS) with alternative stacking of TiS 2 layer and metal sulfides. In this work, they found several nanoscale stacking fault, including staging disorder for BiS and translational disorder for SnS. The presence of the nanoscale stacking fault results in a reduction of thermal conductivity along the layers demonstrating phonon blocking; the electrical mobility has not deteriorated. However, the amorphous structure with extra boundaries in polycrystals and nanostructured materials has a significant effect on the carriers' transport. Wei et al. [79] found that carrier mobility in doped SnSe was lower in the polycrystal sample than in the single-crystal sample. This was attributed to several reasons, including the extra scattering of carriers at grain boundaries and the misorientation of the crystals. Similarly, Dharmaiah and Hong [153] found that the carrier mobility of Sb 2 Te 3 nanobulk is decreased due to the scattering of charge carriers at the grain boundaries and the voids. Rarely, it was found that introducing amorphous and fine nanostructure does not affect the electrical conductivity. Xie et al. [150] found that electrical conductivity has not decreased in the sintered melt spinning Bi 0.52 Sb 1.48 Te 3 sample which exhibits a fine and amorphous structure compared with the same zone melted and sintered samples. The authors suggested that this may attributed to the fact that the coherent interfaces between grains can be beneficial to carrier mobility. To a certain extent, they can compensate for the negative effect of amorphous and nanostructure. Table 3 summarizes some layered chalcogenide nanomaterials and their ZT.

Nanocompositing
Incorporating a second-phase material, especially a nano additive, into thermoelectric materials to develop a composite is an effective strategy to combine the high electrical conductivity of the base material with the low thermal conductivity of the composite. Introducing a large structured material compared with the nano-additive atom leads to the presence of a high density of interfaces [147]. These interfaces act as barriers to scatter the phonons efficiently and cause decreasing thermal conductivity. In addition, the interfaces scatter and filter low-energy carriers, resulting in an improvement in the Seebeck coefficient without severely affecting electrical conductivity [17,90,[171][172][173].
Phonons are scattered more than electrons in nanocomposites because the phonons' mean free path is greater than the carriers' mean free path, leading to stronger scattering of phonons rather than electrons. In some cases, the introduced interfaces lead to a reduction in electrical conductivity, so it is essential to select a proper additive material [174,175]. Moreover, producing some composites is limited by the high fabrication cost as well as poor scalability and rarity of the element [176]. For several years, a considerable amount of work has been done using different nanoadditives in order to enhance the performance of layered chalcogenide materials. Here, nano-additives are classified into carbon-based and noncarbon materials.

Carbon-Based Layered Materials Composites
Carbon-based materials, including carbon black, carbon nanotubes, graphene, and graphene oxide have been introduced to layered chalcogenide materials. Yeo and Oh [177] found that the dispersion of 0.12% multiwall carbon nanotubes (MWCNT) into p-type (Bi 0.2 Sb 0.8 ) 2 Te 3 matrix improves ZT from 1.03 for a pristine sample to 1.41 for the nanocomposite at room temperature.
Although the presence of the grain boundaries and MWCNT/(Bi 0.2 Sb 0.8 ) 2 Te 3 interfaces resulted in extra scattering of the carriers and hence reduced electrical conductivity, the overall improvement is attributed mainly to the reduction of thermal conductivity as a result of strong phonon scattering. Li et al. [178] reported a high figure of merit of 1.21 for polycrystalline p-type SnSe when incorpo- rated with 2.5 vol% of carbon black. The improvement is attributed mainly to the increase in electrical conductivity resulting in a high power factor. Recently, graphene has gained significant interest for its superior thermoelectric properties. Graphene is the first two-dimensional sheet of hexagonally sp 2 hybridized carbon material arranged into a periodic ring structure. It is the thinnest material available in nature [8,[179][180][181]. It has high electrical conductivity, and its thermal conductivity is much higher than most materials measured at room temperature [10,176]. Compared with carbon nanotubes, which exhibit distinctive properties, graphene has higher thermal conductivity and higher electrical conductivity, and it possesses a large surface area that allows interaction with other atoms in the matrix [182]. Table 4 shows a comparison of properties between carbon nanotubes and graphene [181]. The charge transport properties and the carrier mobility enable graphene to be a useful material for thermoelectric devices. Utilizing graphene as an additive for several material matrices has emerged for use in thermoelectric systems in order to develop a high-efficiency composite. A small amount of graphene could play an important role in blocking phonon scattering and reducing the thermal conductivity of the composite. Due to the high electrical mobility of carriers in graphene, it can maintain the electrical conductivity of the composite [3,183]. Based on a detailed literature review, graphene has not been used with layered chalcogenide materials, except in several studies with bismuth telluride and bismuth selenide [12,13,146,175,[183][184][185][186][187][188][189][190][191][192][193][194]. However, a common trend observed for all thermoelectric properties is that there is a limited content of graphene for enhancing the properties. Despite the improvement that is achieved in composites with several graphene contents compared with the pure sample [185,191,192], most studies have achieved maximum ZT with a relatively small amount of graphene. This could be attributed to the aggregation of carriers and scatter- Figure 13: Thermal conductivity as a function of temperature for BiS-bTe + x wt% graphene. All composite samples have lower thermal conductivity than the pristine sample. Reproduced from Ref. [191].

Figure 14:
Temperature dependence of the Seebeck coeflcient of bulk Bi 2 Te 3 and Bi 2 Te 3 composite. The addition of graphene leads to an increase in the Seebeck coeflcient. Reproduced from Ref. [187].
ing at the interfacial boundaries with increasing graphene content [192]. With an optimum amount of graphene, the thermal conductivity of graphene composites has been found to reach a minimum value compared with the pristine samples [12]. Figure 13 shows the thermal conductivity of several composite samples with different graphene contents compared with the pure sample of BiSbTe [191]. The significant reduction of thermal conductivity was attributed to the high density of grain boundaries, defects, and interfaces that could effectively scatter the phonons. The Seebeck coefficient does not show a fixed trend with increasing graphene content. In some studies, the Seebeck coefficient decreased compared with the pure material when adding graphene [12,146,190,192], which is expected as the number of charge carriers accompanied  . Reproduced from Ref. [192].
by the addition of graphene increases [187]. Other studies [184,187,190,191] have shown that Seebeck coefficient followed a reverse trend. Agarwal et al. [187] found that the incorporation of monolayer graphene increases the See-beck coefficient for the composite sample compared with the pure sample as shown in Figure 14. Several studies have shown that electrical conductivity is improved with increasing the graphene content up to a certain optimum limit [146,175,192]. This finding was attributed to the presence of the highly conductive graphene [13,186]. For example, the addition of graphene from x = 0.1 wt% up to 10 wt% except at x = 0.5 wt% was found to increase electrical conductivity for both powder and nanowire n-type Bi 2 Te 3 composites [186]. At x = 0.5 wt%, the unexpected reduction in electrical conductivity could be attributed to the carrier transport at the interfaces of nanostructured Bi 2 Te 3 and graphene, which has different potential energies that affect the composite and result in decreasing carrier concentration. Suh et al. [192] found that the incorporation of 0.1% expanded graphene (EG) increases the conductivity in the whole temperature range, as shown in Figure 15. This improvement in electrical conductivity is attributed to the enhancement in both carriers' concentration and mobility indicating an excellent effect of expanded graphene. However, the extra amount of graphene beyond the optimum level leads to a reduction of both carriers' concen- tration and mobility due to the aggregation of graphene and scattering at the interfaces. The total improvement in the electrical properties, along with the reduction in the thermal conductivity of the composited samples, leads to a significant improvement in ZT. However, the maximum value of ZT is limited by the graphene content as well as the temperature where ZT is measured. It is worth mentioning that the same sample could have two maximum ZT values with different graphene contents at different temperatures. For instance, Li et al. [12] found that the sample exhibits maximum ZT at 300 K with a graphene content of 0.3 vol%, while it reaches its maximum value with a graphene content of 0.4 vol% at 440 K. Table 5 summarizes different chalcogenide materials with carbon-based composites.

Non-Carbon-Based Layered Material Composites
Besides carbon-based materials, layered materials have been mixed to improve their thermoelectric properties with other noncarbon additives like metals [162,[200][201][202], semi metals [203], metal alloys [63,[204][205][206][207][208][209], and ceramics [63,198,[210][211][212][213]. One example is the dispersion of silver nanoparticles in the bismuth telluride matrix [200], which improved ZT of the composite approximately three times compared with pristine bismuth telluride. The silver nanoparticles were uniformly dispersed on the grain boundaries and were found to play an effective role in suppressing the grain growth of bismuth telluride as well as creating lots of interfaces and defects within the matrix, forming a hierarchical structure. The hierarchical struc-ture causes strong scattering of phonons, which reduces the thermal conductivity and enhances the Seebeck coefficient through the carrier filtering effect. Ju and Kim [213] tried to enhance the thermoelectric performance of p-type SnSe by introducing SiC into the matrix. They found that the formation of the composite improved the power factor and effectively reduced thermal conductivity. As a result, ZT of the composite was larger than the pristine sample.
Liu et al. [214] introduced copper selenide (Cu 2 Se) into ptype BiCuSeO. An enhancement in thermoelectric performance was achieved with a 30% increase in ZT at 823 K. In this study, It was found that the electrical conductivity of the composite increases due to the high electrical conductivity of Cu 2 Se. The Cu 2 Se nanoparticles filled the voids among the BiCuSeO grains and acted like electrical contacts among the matrix, which in turn increased electrical conductivity. Accordingly, the Seebeck coefficient while thermal conductivity increased due to the thermal conductivity contribution of charge carriers. However, the 275% increasing in electrical conductivity of the composite overcame the reduction in the Seebeck coefficient and increment of thermal conductivity and enabled the highest ZT value of 0.44. Aside from the overall improvement in ZT, it has been found that introducing a second phase material in the matrix can play a significant role in reducing carrier concentration as well as the carriers' mobility. Ye et al. [215] prepared the n-type TiS 2 -xMoS 2 nanocomposite and found that the electrical conductivity decreased compared with the pristine sample in the whole temperature range especially when x increases above 3. They attributed the reduction in electrical conductivity to both a reduction in carriers' concentration and the reduction in carriers' mobility. As MoS 2 sample is p-type with hole charge carriers, an electron-hole combination would occur in the composite at the interfaces between MoS 2 and TiS 2 leading to a reduction in carriers' concentration. Moreover, the large difference in bandgap between the TiS 2 and MoS 2 creates a potential barrier that would scatter low energy carriers, hence reducing carriers' mobility. A similar reduction in the electrical conductivity was reported when doping Bi-CuSeO with SiC. Feng et al. [216] suggested that due to the poor electrical conductivity of SiC, the carrier concentration of the composite decreased. Moreover, the presence of SiC nanoparticles reduced the grain size and increased the boundaries where carriers are scattered and filtered, thus suppressing the mobility. Table 6 summarizes some layered material composites with noncarbon additives and their ZT.

Conclusion and Future Outlook
This paper gave an overview of the properties and structure of layered chalcogenide thermoelectric materials as well as the progress that has been achieved in measuring their thermoelectric performance. It presented a discussion of several strategies that have been used to enhance their thermoelectric properties, including doping and alloying, nanostructuring, and nanocompositing. Doping and alloying have been found to be effective for enhancing the power factor due to electron density modification as well as reducing thermal conductivity due to the scattering of carriers and phonons by the defects induced. Moreover, nanostructuring into several nanoshaped forms improves ZT. The scattering of phonons at the interfaces helps to reduce thermal conductivity, while the filtering of low-energy carriers increases the power factor. Similarly, the introduction of second-phase material into the thermoelectric matrix enhances ZT by improving the power factor and reducing thermal conductivity. The interfaces created by the additive act as barriers for filtering the low charge carriers and scattering phonons. In this context, we have discussed graphene as a promising nanoadditive material for preparing thermoelectric layered chalcogenide composites. Bismuth telluride has shown a significant improvement in power factor and thermal conductivity when mixed with graphene. Continued research is required in order to prepare low-cost and high-quality materials. As there is still a deficiency in the studies of graphene-layered chalcogenide thermoelectric materials, it is of great interest to study the cases of graphene composites with different chalcogenide materials and report its effects.