Main

The conversion of thermal energy to electrical energy is known as thermoelectric (TE) conversion. The TE effect can be used for both power generation and electronic refrigeration. When a temperature gradient (ΔT) is applied to a TE couple consisting of n-type (electron-transporting) and p-type (hole-transporting) elements, the mobile charge carriers at the hot end tend to diffuse to the cold end, producing an electrostatic potential (ΔV). This characteristic, known as the Seebeck effect, where α = ΔV/ΔT is defined as the Seebeck coefficient, is the basis of TE power generation, as shown in Figure 1(a). Conversely, when a voltage is applied to a TE couple, the carriers attempt to return to the electron equilibrium that existed before the current was applied by absorbing energy at one connector and releasing it at the other, an effect known as the Peltier effect, as shown in Figure 1(b). Thermoelectric technology and solid-state devices based on the TE effect have a number of advantages, including having no moving parts, and being reliable and scalable. The technology has therefore arouse worldwide interest in many fields, including waste heat recovery and solar heat utilization (power generation mode), and temperature-controlled seats, portable picnic coolers and thermal management in microprocessors (active refrigeration mode)1.

Figure 1
figure 1

Schematic illustrations of a thermoelectric module for (a) power generation (Seebeck effect) and (b) active refrigeration (Peltier effect). (a) An applied temperature difference causes charge carriers in the material (electrons or holes) to diffuse from the hot side to the cold side, resulting in current flow through the circuit. (b) Heat evolves at the upper junction and is absorbed at the lower junction when a current is made to flow through the circuit.

Full size image

The efficiency of TE devices is strongly associated with the dimensionless figure of merit (ZT) of TE materials, defined as ZT = (α2σ/κ)T, where σ, κ and T are the electrical resistivity, thermal conductivity and absolute temperature2. High electrical conductivity (corresponding to low Joule heating), a large Seebeck coefficient (corresponding to large potential difference) and low thermal conductivity (corresponding to a large temperature difference) are therefore necessary in order to realize high-performance TE materials. The ZT figure is also a very convenient indictor for evaluating the potential efficiency of TE devices. In general, good TE materials have a ZT value of close to unity. However, ZT values of up to three are considered to be essential for TE energy converters that can compete on efficiency with mechanical power generation and active refrigeration.

High-performance TE materials have been pursued since Bi2Te3-based alloys were discovered in the 1960s. Until the end of last century, moderate progress had been made in the development of TE materials. The benchmark of ZT ≈ 1 was broken in the mid-1990s by two different research approaches: one by exploring new materials with complex crystalline structures, and the other by reducing the dimensions of the materials. The search for new materials, such as skutterudites3,4 and clathrates5, is mainly motivated by the suggestion of Slack6. In those compounds, the 'rattling' motion of loosely bonded atoms within a large cage generates strong scattering against lattice phonon propagation, but has less of an impact on the transport of electrons. As a consequence, the thermal conductivity of skutterudites and clathrates can be reduced greatly while maintaining the electrical conductivity at a high level. Zintl compounds7 with a large unit cell, including Yb14MnSb11, Yb11GaSb9, Ca11GaSb9 and SrZnSb2 have also been revealed to possess an intrinsically low lattice thermal conductivity due to the high fraction of low-velocity optical phonon modes8. In research on low-dimensional material systems, Dresselhaus et al.9 suggested that the power factor (α2σ) can be enhanced through the use of quantum-confinement effects. As the system size decreases and approaches a nanometer length scale, the density of electronic states (DOS) can split and become narrow. Dresselhaus's pioneering work has shed light on various low-dimensional systems, including superlattices, nanowires and quantum dots. Venkatasubramanian et al.10 reported Bi2Te3/Sb2Te3 superlattices with a high-ZT value of up to 2.4. Subsequently, Harman et al.11 reported PbTe/PbTeSe quantum dot superlattices with a ZT value of greater than 3.0 at 600 K. However, for general applications, it is highly desired to develop high-performance bulk TE materials. Figure 2 shows the high-performance bulk TE materials developed recently on the basis of nanostructural concepts, demonstrating the significant progress that has been made in TE materials since the introduction of nanotechnology. This review covers the latest advancements in TE technology focusing on such nanostructural approaches, and in doing so aims to highlight the potential routes for improvement in the major TE material systems, including Bi–Te alloys, skutterudites, Ag–Pb–Sb–Te or 'LAST', half-Heusler alloys and some high-ZT oxides. Certainly, there are also other useful approaches for enhancing TE properties, such as band structure engineering by doping17,20 and exploring new materials with complex crystalline structure21,22, but such approaches are beyond the scope of the current review.

Figure 2
figure 2

Figure of merit (ZT) of recent high-performance bulk thermoelectric materials as a function of applied temperature. The high performance of most of these materials is related to nanostructure engineering.

Full size image

Nanostructural approaches for enhancing thermoelectric performance

The challenge to develop TE materials with superior performance is to tailor the interconnected physical parameters α, σ and κ for a crystalline system. Nanostructures provide a chance to disconnect the linkage between thermal and electrical transport by introducing some new scattering mechanisms for ZT enhancement. There are some excellent reviews focusing on nanocrystalline26 and nanostructured thermoelectric materials27 and interfaces in bulk thermoelectric materials28. Here, as illustrated in Figure 3, we attempt to categorize the morphological improvement of nanocomposites in TE materials in terms of several parameters, including dimension reduction, grain refinement and size reduction of a second phase. Single crystals usually present the best electrical conductivity because of the absence of grain boundaries that scatter charge carriers, but the ZT values of such materials can be optimized only by adjusting the carrier concentration through elemental doping. Reducing the dimensions instead offers a new possibility to individually tune the TE parameters, as illustrated in Figure 3(a–d). When the system size decreases and approaches a scale comparable to the feature length of electron behavior (e.g. mean free path, wavelength) in any direction, the DOS is increased significantly due to quantum confinement9, resulting in the enhancement of the Seebeck coefficient29. Meanwhile, the thermal conductivity is also reduced because the surface strongly scatters the propagation of phonons, as any dimension is less than the average free path of phonons. For example, Bi2Te3-based films prepared by pulsed laser deposition show κ values of 0.3–0.4 W m−1 K−1, which are 25% lower than in the bulk30. It has also been reported that a silicon nanowire with a diameter of 10–20 nm presents amazingly low thermal conductivity with a κbulk/κnanowire ratio of as high as 25–150 near room temperature31,32.

Figure 3
figure 3

Summary of thermoelectric material structure from macro- to nanoscale. (a–d) Change in dimensionality: (a) bulk, (b) thin film, (c) nanowire, (d) atomic cluster. (e–h) Grain mixtures from micro- to nanoscale: (e) normal micro-grained bulk, (f) mixture of coarse and fine grains, (g) nano-grained bulk, (h) amorphous. (i–l) Size evolution of isolated distinct phases or atoms in the composite: (i) normal composite, (j) nano-dispersions located inside grains or at grain boundaries, (k) nano-inclusions or nanodots, boundary modification, (l) atomic doping or alloying, and vacancies.

Full size image

Apart from surfaces scattering, the thermal conductivity can be reduced significantly by introducing grain boundaries or interfaces, as illustrated in Figure 3(e–h). However the suppression of lattice thermal conductivity (κlat) due to boundaries will be overshadowed by the deterioration of carrier mobility (μ). Considering this trade-off, a nano/microcomposite TE material such as that shown in Figure 3(f) might provide better TE performance compared with monolithic coarse- or fine-grained samples33,34. In this nano/micro mix-grained composite, the dispersed nanoparticles are designed to scatter the phonon, while microparticles form a connected network for electron transport, and hence decouple the connection between thermal and electrical transport to a degree according to the percolation effect. Introducing an acoustically mismatched second phase in a matrix is another way to reduce the lattice thermal conductivity through an additional scattering mechanism36.

Figure 3(i–l) shows the size reduction of isolated distinct phases or atoms in the composite, including sphere-, plate- and wire-shaped dispersed phases. Obvious ZT improvements have been obtained in (Zr,Hf)NiSn half-Heusler alloys by adding nano-ZrO236, in Bi2Te3 by adding nano-SiC particles37, and in YbyCo4Sb12 by dispersing in situ partially oxidized Yb2O3 nanoparticles38. It is also reported that more significant enhancements can be achieved by embedding metal or conductive nanoparticles into a TE material matrix: examples include lead and antimony in PbTe39, antimony in YbyCo4Sb1240, and ErAs in In0.53Ga0.47As41. This kind of nanoscale dispersed phase can be located at the grain boundary or within the grains. Shakouri et al.42 calculated the effect of nanoparticle scattering on the power factor using a relaxation-time approximation, and validated the calculated model by comparing the results with the experimental data obtained for ErAs:InGaAlAs samples. They used the theory to maximize the power factor with respect to the volume content of nanoparticles and electron concentrations, as well as the barrier height. Chen et al.43 developed a semiclassical electron transport model based on the Boltzmann transport equation to study the TE transport properties of nanocomposites. A relaxation-time model for electrical carrier–interface scattering was developed to account for the filtering effect of low-energy electrons on transport properties. After fitting the model to data for bulk TE alloys, yielding reasonable adjustment parameters for bulk alloys similar to handbook values, the model was further validated by comparing the modeled TE properties with recently reported values for p-type (BiySb2–yTe3)–(Bi0.5Sb1.5Te3) and n-type (Mg2SiyGe1–y)–(Mg2Si0.6Ge0.4) nanocomposites. The inspired work of Kanatzidis's group revealed that a new kind of intrinsic nano-inclusion, as shown schematically in Figure 3(k), formatted by the segregation of silver and antimony in the lead sublattice of PbTe, could achieve simultaneous phonon blocking and electron transmission.

Base on the same concept, Liu et al.44 obtained a similar microstructure by partially substituting antimony with tellurium and tin in CoSb3, thereby obtaining a significant improvement. The theoretical study showed that nano-inclusions could reduce the thermal conductivity of the host medium below that of the alloying case, which is defined as an atomic random distribution45, as shown in Figure 3(l). It can be noted through these examples that the combination of nano-engineering strategies has become a new trend, leading to such systems as a superlattice or quantum well wire combining reduced dimensionality with a second dispersion phase46. A current challenge in the development of nanocomposite TE materials is how to preserve the nanostructures within the bulk composite and prevent aggregation and interparticle growth. The solution to these issues significantly depends on the interfaces. A suitable interface could prohibit undesired grain-boundary growth and interdiffusion, as well as selectively scattering phonons and charge carriers, potentially resulting in a considerably enhancement of ZT values.

Representative high-performance thermoelectric materials

Bi2Te3-based alloys

Bi2Te3 has been studied extensively since 195447 and is one of the most widely used TE materials. Nevertheless, this material still attracts plenty of interest, and has undergone significant enhancements as a TE material. In 2008, Poudel et al.48 reported a nano-composite BixSb2–xTe3 bulk with ZT of ca. 1.4 at 373 K. This material was fabricated by mechanical milling of a commercial ingot into a nano-powder that was subsequently consolidated by hot pressing. Their microstructural analysis revealed that the high ZT value is partially attributable to a significant decrease in κlat due to scattering at the grain boundary and the presence of nano-precipitates49. Tang et al.14 developed a melt–quench–anneal–spark plasma sintering method to form bulk nanostructural p-type Bi0.52Sb1.48Te3 with a ZT value at of 1.56 at 300 K. Cao et al.50 obtained a maximum ZT of 1.47 at ca. 438 K for Bi2Te3/Sb2Te3 bulk nanocomposites with laminated nanostructures using a simple route involving hydrothermal synthesis and hot pressing.

The nano/micro composite has now become an important direction for research. Recently, the author's group has conducted much work on SiC/Bi2Te3 nano/microcomposite TE materials. Our studies have revealed that a 20% enhancement in ZT can be achieved by adding SiC nanoparticles at just 0.2 vol%. However, the ZT value decreased quickly at higher SiC concentrations because SiC has much higher electrical resistivity and thermal conductivity compared to Bi2Te351. The addition of SiC also obviously enhanced the micro-hardness and fracture toughness, resulting in improved machining properties for device manufacturing, particularly for miniaturized devices37.

However, a decrease in κlat by nanoparticle dispersion also leads to a decrease in carrier mobility, and as a result, no more enhancement of ZT is obtained. Recently, a percolation effect has been introduced to tune TE transport properties, and a proof-of-concept experiment has been conducted by Zhao et al.34 In their experiment, two kinds of Bi2Te3 powders with large particles size difference (coarse, ca. 1 μm; fine, ca. 100 nm) were fabricated by mechanical alloying, and mixtures with various fine/coarse ratios were consolidated by spark plasma sintering. An enhanced ZT value at room temperature was achieved with an optimized fine/coarse volume ratio of 6/4.

The mechanism of TE enhancement by the percolation effect is illustrated in Figure 4(a). The charge carriers likely 'select' a path — the low-resistivity channel connected by the large grains — whereas the phonons do not choose their path. The system of channels via coarse grains would have low electrical resistivity and thermal conductivity if more fine particles are included. As the Seebeck coefficient does not change significantly by varying the grain size, the ZT value could be tuned by adjusting the volume percentage of fine particles as shown in Figure 4(b). According to the principle of the percolation effect, a high coarse/fine size ratio in a powder would generate a more significant enhancement of ZT. The quantitative description can be given by ZT = α2/L(1 + κl/κe), where L is the Lorenz number, κl is the lattice thermal conductivity and κe is the electrical thermal conductivity. Microstructural modifications allow the fundamental transport parameters (carrier mobility and lattice thermal conductivity) to be modified for improved TE performance. For example, carrier mobility can be increased by increasing the grain size, which reduces the density of grain boundaries and results in an increase in electrical conductivity. Low lattice thermal conductivity values can be obtained by grain boundary scattering. ZT values will be enhanced by reducing the ratio of lattice thermal conductivity to electrical thermal conductivity (κl/κe) through optimization of the volume content of grain boundaries.

Figure 4
figure 4

Illustration of the principle of percolation in nano/microcomposites. (a) Presumed carrier and phonon transport path in the composite. (b) Weight fraction dependence of thermoelectric ZT values at room temperature in Bi2Te3 composite. Adapted from Ref. 34, reproduced with permission (© 2009 AIP).

Full size image

CoSb3-based skutterudites

The binary skutterudites possess a CoAs3-type structure with the general chemical formula MX3, where M is the transition-metal cobalt, rhodium or iridium, and X is phosphorus, arsenic or antimony52,53, as shown in Figure 5(a). Among the skutterudite family, CoSb3 has received the most interest to day because of its higher weighted mobility (m*)3/2μ compared to other family members, where the m* and μ are the carrier effective mass and mobility. CoSb3 can have a very high power factor, but its high thermal conductivity (10 W m−1 K−1) prevents it from competing with state-of-the-art Bi2Te3 (1.0–1.5 W m−1 K−1). However, this compound has a specific lattice structure with a large 'cage' located at the center of the unit cell, which could be filled with a small metal atom as shown in the Figure 5(a). Since the cage is much larger (1.89 Å) than most elemental ions, the filler is likely to rattle at the equilibrium position and hence generate significant scattering of phonons6. In 1996, Sales et al.3 reported Ce0.9Fe3CoSb12 and La0.9Fe3CoSb12 with a remarkably low κlat of 1.4 W m−1 K−1, one-seventh of that of unfilled CoSb3, and a consequently high ZT of greater than 1.0. Recently, broad-ranging investigations have been carried out on CoSb3-based TE materials with respect to both doping strategies and synthesis processes, and some innovative results have been obtained. In 2001, Chen et al.54 found that barium was a very good filling element, achieved a higher cage filling fraction (44%) than lanthanum (ca. 20%) without charge compensation. Barium filling not only reduced κ, but also raised σ, leading to a high ZT value in BayCo4Sb12. On the base of Chen's work, Tang et al.55 systematically studied the system RyMxCo4−xSb12 (R = Ce, Ba, Y; M = Fe, Ni), and both n- and p-type skutterudites with high performance were obtained. The p-type Ce0.28Fe1.5Co2.5Sb12 showed a peak ZT of 1.1 at 750 K, and n-type Ba0.30Ni0.05Co3.95Sb12 presented a maximum ZT of 1.25 at 900 K. From an industrial viewpoint, a good TE device requires the n-type and p-type legs to have similar mechanical and thermal properties in order to minimize the possibility of failure due to thermal stress. CoSb3 will be a very good choice for medium-temperature TE applications because both n- and p-type materials with high performance can be obtained in the same material system.

Figure 5
figure 5

(a) Structure of skutterudites showing substructure details including the [CoSb6] octahedron, [Sb4] ring and [Co8Sb12] dodecahedron. (b) InSb nanoparticles dispersed at grain boundaries. Modified after Ref. 58 (© 2009 AIP). (c) Nano-grained CoSb3 made by hydrothermal synthesis. Modified after Ref. 59 (© 2007 AIP). (d) Nanodots formed by segregation of tin and tellurium in the antimony sublattice of CoSb3. Modified after Ref. 44 (© 2008 ACS). (e) GaSb nano-inclusions dispersed in Yb0.26Co4Sb12 matrix. Modified after Ref. 18 (© 2010 Elsevier). All figures reproduced with permission.

Full size image

Theoretical research on skutterudites has at the same time made some important advancements. Shi et al.56 studied the filling limit of filled skutterudites theoretically and experimentally. They found that the filling limit is related to the electronegativity difference between antimony and the filling element (R). Only the elements meeting the electronegativity difference condition of χSb – χR > 0.8 can enter the cage position of □Co4Sb12 where □ is the cage site. According to Shi's theoretical calculations, alkali elements could fill the cage position with a high filling fraction. In 2006, Pei et al.57 successfully synthesized potassium-filled CoSb3, achieving a ZT of ca. 1.0 at 800 K in the compound K0.38Co4Sb12. Recently, a breakthrough in filled skutterudites has been made by using a melt–quench method to fabricate an indium and cerium co-filled CoSb3 nanocomposite, for which a peak ZT of 1.43 was obtained at 800 K (Figure 5(b))58. The κlat value of filled skutterudites is much lower than that of unfilled skutterudites, but still much higher than that of Bi2Te3-based TE compounds. Recent research suggests that reducing the grain size could be a fruitful route for the CoSb3-based TE materials. Hydrothermal synthesis is a versatile method for the fabrication of various nanoparticles. Zhao's group59 explored this method to synthesize many TE nanoparticles, as shown in Figure 5(c). Bulk CoSb3 prepared by hydrothermal synthesis of nano-powder followed by spark plasma sintering or hot-pressing showed a κ value of 1.61 W m−1 K−1 and a six-fold enhancement in ZT.

The author's group has used mechanical alloying and spark plasma sintering (MA-SPS) to fabricate pure CoSb3 with an average grain size of 100 nm and a κ value of 2.92 W m−1 K−160. This value is one fourth of that of the single-crystal counterpart, and is even lower than that of some filled skutterudites. Further doping of this fine-grained CoSb3 by partially co-substituting antimony with tin and tellurium raised the ZT value to 1.1 at 820 K, which is the highest value reported to date for unfilled skutterudites and is comparable to that of many filled skutterudites44. The remarkable benefit of co-doping with tin and tellurium is a further decrease in lattice thermal conductivity compared with tellurium-doped CoSb3. The presence of sparse nanodots inside grains could also be responsible for the further decrease in lattice thermal conductivity, as shown in Figure 5(d). Nanocomposites, as an important route for tuning κ, have also been used in CoSb3-based TE materials. Chen's group36 fabricated a composite containing ytterbium-filled CoSb3 and well-distributed Yb2O3 particles synthesized by in situ reaction. The Yb2O3 nanoparticles dispersed at the grain boundary and within the grains generate a combined effect that considerably scatters phonons but only slightly impacts on electrons, thereby providing a significant enhancement in peak ZT to 1.3. Using a similar idea, as shown in Figure 6(e), the same research group has made a (GaSb)0.2–Yb0.26Co4Sb12 nanocomposite exhibiting an improved ZT value of up to 1.45 at 850 K19.

Figure 6
figure 6

(a) Average Fm3¯m structure of LAST system. Adapted from Ref. 15 (© 2004 AAAS). (b) High-resolution image of a nano-inclusion in Ag0.53Pb18Sb1.2Te20 taken along the [001] zone axis. Modified after Ref. 61 (© 2009 Wilegy-VCH). (c) Temperature-dependent thermal conductivity of MA-SPS-synthesized Ag0.8Pb22SbTe20 after annealing. (d) Typical high-resolution transmission electron microscopy (HRTEM) image of an unannealed sample. (e) Typical HRTEM image of the sample after annealing for 30 days. (c–e) Adapted and modified after Ref. 64 (© 2008 ACS). All figures reproduced with permission.

Full size image

AgnPbmSbnTem+2n

In 2004, Kanatzidis's group reported a high-performance TE material system with the general chemical formula of Ag1–xPb18SbTe20, termed LAST15. As shown in Figure 6(a), the LAST compounds possess an average NaCl structure (Fm3¯m symmetry), and could thus be considered to be antimony and silver co-doped PbTe. The same group used a melting plus extremely slow cooling method to develop a nanocomposite with many nanoscale inhomogeneities embedded in the PbTe matrix, as shown in Figure 6(b). These nano-inclusions are likely to be ideal phonon-blocking and electron-transmitting centers. Due to the special nanostructure,Ag1–xPb18SbTe20 showed a very high ZT value of 2.2 at 800 K. Although lower ZT values of 1.5–1.7 were confirmed in a subsequent report62, the same study suggest that this serial compound is still the most competitive TE material. Kanatzidis's work has ignited broad research interest on the LAST system14,62,63,64,65.

The author's group has been working on the synthesis of TE materials by MA-SPS since 2003, and has used this method successfully to fabricate LAST-system TE materials. Wang et al.63 synthesized LAST compounds from lead, silver, antimony and tellurium elemental powders by MA-SPS at 673 K, achieving a respectable ZT value of 1.37 at 673 K. They found that the TE properties of LAST compounds are very sensitive to chemical composition, particularly lead content. The optimized composition is Ag0.8Pb22.5SbTe20 for the MA-SPS-synthesized sample, which is different from the composition (Ag0.8Pb18SbTe20) reported by Kanatzidis's group. Based on Wang's work, Zhou et al.64 studied the effect of annealing treatment on the TE performance and microstructure of the LAST system and found that annealing treatment not only improved the electrical conductivity, it also reduced the thermal conductivity, as shown in Figure 6(c). Usually, it is extremely difficult to achieve an increase in electrical conductivity and a decrease in thermal conductivity simultaneously. Transmission electron microscopy analysis demonstrated that the annealing treatment assisted the formation of a nanostructure that favors electron transparent and phonon scattering, i.e. AgPb and SbPb coalesce to form intrinsic quantum nanodots, as shown in Figure 6(d–e). As a result, a significantly improved ZT value of 1.5 was obtained at 703 K. The LAST system can also be regarded as a nanocomposite with AgPbTe2 nano-phase dispersed in a PbTe matrix. Recently, Wang et al.14 successfully synthesized bulk single-phase AgSbTe2 by MA-SPS. The material displayed an extremely low κ value (0.3 W m−1 K−1) and high ZT value of 1.59 at 673 K.

Many studies have been carried out on the LAST system by Kanatzidis's group, particularly with respect to microstructure characterization66 and exploration of p-type LAST systems67,68,69. They have found that a high ZT value of 1.45 at 630 K is achievable in p-type Ag(Pb1–xSnx)mSbTe2+m by partially substituting the lead in the LAST system with tin67. Subsequent research has shown that simply replacing silver in the LAST system with sodium leads to a higher ZT of 1.7 at 650 K in p-type Na0.95Pb20SbTe2268. More recently, potassium-based analogues of p-type K1–xPbmSbyTem+2 materials have been reported. A maximum ZT of 1.6 at 750 K has so far been achieved for a system with the composition K0.95Pb20Sb1.2Te2269. Both n- and p-type LAST materials can be obtained by adjusting the chemical composition, making the system is particularly promising for application in power generation.

Half-Heusler compounds

Compared with the PbTe or LAST systems, half-Heusler compounds are more environmentally benign and hence have attracted increasing levels of interest70,71,72,73,74. Half-Heusler compounds are crystallized in the MgAgAs-type structure with space group F4¯3m, which can be regarded as two interpenetrating cubic face-centered-cubic (FCC) motifs with a simple embedded cubic motif at the center75. Since half-Heusler compounds have a good combination of narrow band-gap and sharp slope of DOS near the Fermi level, a very high power factor is expected. However, on account of the relative high κ, the ZT values of the half-Heusler compounds are still much lower than those of the state-of-the-art TE materials. As there are three sublattice positions in a half-Heusler compound, isoelectronic alloying on different sublattice positions is the most common approach to reducing κ. Uher et al.70 found that the isoelectronic alloying of Zr0.5Hf0.5NiSn produced a higher ZT value than that of either ZrNiSn or HfNiSn alone due to a reduction in thermal conductivity. Shen et al.71 investigated the effect of partial substitution of nickel with palladium on the TE properties of ZrNiSn-based half-Heusler compounds. It was shown that this substitution results in a significant reduction in thermal conductivity, leading to an improvement in ZT to 0.7 at ca. 800 K for the composition Zr0.5Hf0.5Ni0.8Pd0.2Sn0.99Sb0.01. Later, Culp et al.72 reported a slightly enhanced ZT value of 0.8 at 800 K in Hf0.75Zr0.25Ni0.9Pd0.1Sn0.975Sb0.0025.

Since most of the current work on half-Heusler is focused on isoelectronic alloying, it is reasonable to anticipate that there is considerable room for further improvement in ZT through the use of various nanostructures. Systematic theoretical calculations on 36 kinds of half-Heusler compounds by Yang et al.75, demonstrates that there are potential new half-Heusler compounds beyond the well-studied n-type MNiSn (LaPdBi, NdCoSn, ZrCoBi, etc.) and p-type MCoSb (NdCoSn, TiCoSb, NdRhSb, etc.). Experimental investigations on these new compounds should be strongly encouraged.

Other systems

Oxide materials usually have high thermal stability and oxidization resistance, and are hence good for high-temperature applications. However, the extremely low electrical conductivity of these materials has resulted in them being almost entirely ignored for their potential utilization in the TE field. The discovery of NaCo2O4 as a promising TE material in 199776 completely changed the traditional understanding of oxides. NaCo2O4 posses a similar structure to the high-temperature superconductor compound YBaCuO, with a CoO2 layer and sodium layer constructing a laminated structure, as shown in Figure 7. NaCo2O4 has high electrical conductivity and a high Seebeck coefficient, resulting in a remarkably high power factor of 5,000 W m−1 K−2, even higher than that of state-of-the-art Bi2Te3 (4,000 W m−1 K−2)76. The strong electronic correlation effect in this material has been attributed to the unusually large Seebeck coefficient77. Many researchers subsequently started to explore high-performance oxides for TE applications, and some promising TE oxides have emerged, including Zn1–xAlxO (n-type, ZT ≈ 0.30 at 1273 K)78, La1–xSrxCoO3 (p-type, ZT ≈ 0.18 at room temperature)79, and La1–xYxCuO4 (p-type, ZT ≈ 0.17 at 330 K)80.

Figure 7
figure 7

Crystal structure of NaxCoO2 (left) and the corresponding nanobolock structure (right)81.

Full size image

Recently, the concept of nanoblock integration into layer-structured hybrid crystal (natural superlattice) has been proposed by Japanese researchers81,82. According to this idea, the layered NaCo2O4 is regarded as a hybrid crystal of ordered CoO2 nanosheets and disordered sodium nanosheets. The CoO2 nanosheets form a strongly connected electron system that serves as an electronic transport layer, leading to a high power factor, whereas the sodium ion nanoblock layers or calcium cobalt oxide misfit layers act as phonon scattering regions to provide low thermal conductivity. The development of hybrid crystals or natural superlattices composed of a periodic arrangement of nanoblocks or nanosheets having different functions is a new direction for achieving high TE performance. On the basis of the idea of a hybrid crystal, a series of modulated layered natural superlattices have been synthesized, including (Na,Ca)Co2O483, (Bi1.79Sr1.98Oy)0.63(RhO2)84 and (SrO)(SrTiO3)m85. Recently, Lwadaki et al.86 successfully synthesized a material with a pseudo-one-dimensional structure, Ba3Co2O6(CO3)0.7, similar to the 2H-perovskite-type BaCoO3 and containing CoO6 octahedral columns running parallel to the crystallographic c axis.

Summary and outlook

Thermoelectric conversion technology is receiving increasing attention because of the worldwide energy and environment problems, and great progress has been made in the development of high-performance TE materials, including bulk, film and nanowire materials. Nanostructures such as nano-dispersions and nanoscale heterogeneities have been found to be effective in reducing thermal conductivity to a greater degree than electrical conductivity, resulting in an enhanced figure of merit for the bulk thermoelectric material. However, additional approaches will likely be indispensable for increasing the power factor to achieve further significant ZT improvements. Theoretical modeling studies are highly anticipated for designing nanostructured or nanocomposite thermoelectric materials with enhanced Seebeck coefficient based on carrier-energy filtering or quantum confinement effects. As the cost and environmental impact of TE materials must also be taken into account for non-specialized applications, TE compounds free of tellurium and lead will become increasingly attractive targets for research in the future.