Abstract
A large amount of thermal energy emitted from many industrial processes is available as waste heat. Thermoelectric (TE) power generators that convert heat directly into electricity can offer a very promising means of waste heat recovery. However, the requirements for this task place in the materials are not easily satisfied by conventional TE materials. Not only must they possess a sufficient TE performance, but they should also be stable at high temperatures, nontoxic and have low-cost comprising elements, and must be also able to be processed and shaped cheaply. Oxides are among the strongest candidate materials for this purpose. In this review, the progress in the development of two representative p- and n-type novel oxide materials based on Ca3Co4O9 and doped-ZnO is presented. TE modules built up from these oxides were fabricated, tested at high temperatures, and compared with other similar oxide modules reported in the literature. A maximum power density of 45.3 mW cm−2 was obtained for an oxide module comprising eight p–n couples at a temperature difference of 496 K, an encouraging result in the context of the present high-temperature oxide modules.
Export citation and abstract BibTeX RIS
Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
1. Introduction
In conventional energy conversion/production process, typically 50% or more of the energy used is wasted, mostly as heat. The capability of thermoelectric (TE) materials to convert waste heat into usable electrical energy in a pollution and maintenance-free and all-solid-device has attracted enormous interest over the last two decades [1]. A typical TE generator comprises two semiconductor materials, p- and n-type doped TE elements, connected to each other to form a TE couple. The TE couples are then connected electrically in series and thermally in parallel to form a module. A practical TE module can contain up to hundreds of TE couples depending on the output power needed. The key to realize an efficient TE generator for waste heat recovery lies, however, in finding good materials with high TE performance, a good durability at high temperature and preferably robustness under operation in air. The performance of a TE material is evaluated by the dimensionless figure-of-merit ZT (=S2T/ρκ, where S, T, ρ and κ are the Seebeck coefficient, absolute temperature, electrical resistivity and thermal conductivity, respectively). Currently, the most widely used TE materials are alloys, which often suffer from poor durability at high temperature, low abundance (Bi–Te), toxicity (Pb–Te) and high cost (Si–Ge) of the comprising elements as illustrated in figure 1(a) [2]. The conversion efficiency of TE generator primarily depends on the materials' TE performance ZT, and the operating temperature difference between the cold and the hot side of the generator. With higher operating temperature, both the TE conversion and Carnot efficiencies can be significantly improved [3]. Under these conditions, oxide-based TE generator would be one of the strongest candidates due to the naturally good thermal stability of TE oxide materials used at high temperatures comparing with conventional alloy-based modules.
Figure 1. (a) Various TE materials and typical waste heat sources in their respective operating temperatures (from [2]), and (b) conversion and Carnot efficiencies as a function of material thermoelectric performance and operating temperature (from [3]).
Download figure:
Standard imageOxide thermoelectrics has gained much interest during the last decade since the large TE response that was discovered in some metal oxide-based materials [4–7]. We have over the years carried out research on various TE oxide systems to search for high-temperature TE materials [8–25]. In this review, the development of doped ZnO and Ca3Co4O9 novel oxides representative of n-type and p-type materials is presented with highlights on heavily doping and nanostructuring approaches to improve their high- temperature TE properties. The performance of TE modules building up from these oxides is shown and compared with other oxide TE modules reported in the literature.
2. TE oxides
Our materials of p-type and n-type oxide systems based on Ca3Co4O9 and doped-ZnO presented in this paper were prepared by the solid state reaction followed by a hot pressing (HP) or spark plasma sintering (SPS) process. The detailed experimental procedures for producing and characterizing these materials were reported previously [8–25].
2.1. p-type layered Ca3Co4O9-based oxides
Ca3Co4O9-based oxide material has been intensively investigated due to its good TE performance (ZT =0.83 at 973 K for the single crystal) [5] and its highly thermal and chemical stabilities, even up to 1200 K in air [20, 23, 26–29]. Shown in figure 2 is an incommensurate character in the crystal structure of this compound, which can be described as [Ca2CoO3]b1/b2[CoO2], where b1 and b2 are two different periodicities along the b-axis for the rock salt-type Ca2CoO3 subsystem and the CdI2-type CoO2 subsystem, respectively [26]. However, single crystals are less likely to be applied for fabricating practical TE devices. It is hence highly desirable to achieve sufficient TE properties in a polycrystalline form of these oxides.
Figure 2. Schematic crystal structure of nanoscale misfit-layered Ca3Co4O9 along b-axis (adapted from [2 and 26]).
Download figure:
Standard imageIn fact, many attempts have been made over the years to enhance the TE performance of the polycrystalline Ca3Co4O9 by ions doping and improving the fabrication methods, e.g. employing HP or SPS techniques [20–53]. Most research has focused on the doping at Ca-site by e.g. the rare-earth elements such as Pr, Y, Nd, Gd, Nd, Eu, Yb and Lu [21, 23, 27–49], while few reports have performed the substitution on the Co-site [20, 22, 50–53]. The core idea of doping Ln (Ln = rare-earth elements) on Ca-site (Ca2+) is to increase the Seebeck coefficient due to the decrease in hole concentration by the substitution of Ln3+ for Ca2+ and further reduce the thermal conductivity by heavily doping elements. The peculiar structural arrangement of the CoO2 layers, containing cobalt cations with mixed-valence of 3 + and 4 + , is the origin of the interesting properties of the layered cobaltite system. Therefore, ions doping on the Co-site, especially Co ion in the CoO2 planes results in pronounced TE effect owing to the fact that charge carrier is transported mostly within these layers. Accordingly, the TE performance was significantly improved using this approach and the maximum ZT values were found to be about 2–2.5 times larger than that of non-doped material [21, 22, 30, 34].
Figures 3(a)–(c) displays typical SEM images of fracture surface of non-doped, Ga- and Lu-doped Ca3Co4O9 samples consolidated by HP technique, showing a rather dense microstructure. All the HP samples have a relative density of over 94% of the theoretical density (∼4.68 g/cm−3 [26]). It can be clearly observed from figures 3(a)–(c) that both the pure and doped samples exhibited a lamella structure, and this lamella-like morphology become more oriented [20, 21] with the doped samples as compared with the pure one, particularly the sample doped with Lu (figure 3(c)). Ga doping was found to act as a sintering aid [20, 22], while Lu, which has larger ionic radius substitution on the Ca-site changed the misfit ratio i.e. b1/b2 [21], which may be the reason leading to the different observed microstructure. A characteristic x-ray diffraction (XRD) spectrum of a typical Ga-doped Ca3Co4O9 sample with Rietveld refinement profile is presented in figure 3(d), confirming its isostructure and phase purity.
Figure 3. SEM images of fracture surface of HP samples for pure Ca3Co4O9 (a), Ga-doped (b), and Lu-doped (c) samples and (d) XRD patterns and Rietveld refinement profile of a typical Ga-doped sample (adapted from [20, 21]).
Download figure:
Standard imageFigures 4 and 5 show the TE properties of two systems: Ca3−xLuxCo4O9 and Ca3Co4−xGaxO9, which are representative of two doping approaches i.e. at the Ca- and Co-sites, respectively. Shown in figure 4(a) is the electrical resistivity at 300 and 1100 K as a function of Lu and Ga concentration. The electrical resistivity tends to increase with increasing Lu concentration, while it decreases for the samples with Ga doping at both low and high temperatures. The influence of Lu and Ga doping on the Seebeck coefficient shows two different tendencies: S increases with increasing Lu content, while S starts increasing for the sample with low Ga concentration i.e. x = 0.05, then it decreases with higher Ga content. It was evidenced by the Hall measurement that carrier concentration increased for x = 0.05, but it slightly decreased for x > 0.05 [20]. The simultaneous increase in the electrical resistivity and Seebeck coefficient by Lu doping was reported to be due to the decrease in hole concentration by the substitution of Lu3+ for Ca2+ [21]. As for Ga doping, the increasing Seebeck coefficient and decreasing electrical resistivity at high temperature while the carrier concentration increases suggests that this system may have more than one type of independent carrier [20, 22].
Figure 4. The electrical resistivity (a) and the Seebeck coefficient (b) as a function of Ga or Lu concentration at 300 and 1100 K.
Download figure:
Standard image
Figure 5. The thermal conductivity (a) and the dimensionless figure of merit, ZT (b) as a function of temperature for non-doped, Ga- and Lu-doped Ca3Co4O9 selected samples prepared by HP method.
Download figure:
Standard imageThe thermal conductivity and the dimensionless figure of merit are presented in figures 5(a) and (b), respectively, for non-doped and selected Ga- and Lu-doped samples. Figure 5(a) reveals that the thermal conductivity is significantly reduced by heavy elements due to the suppression in the lattice thermal conductivity, particularly with Lu doping [21]. The TE performance ZT is thus remarkably improved by Ga and Lu doping, and the effect is more substantial in the high-temperature region (T > 650 K). The maximum ZT value at ∼1100 K reached about 0.4, which is about two times higher than the non-doped material (figure 5(b)).
SPS method, which is a pressure-assisted pulsed current fast sintering process, offers an advantageous way to fabricate nanostructured bulk materials. With SPS technique, the microstructure of the material and crystal growth are well controlled while maintaining the original properties of the materials. Employing this fabrication method, we have successfully synthesized TE nanocomposites with microstructural design consisting of oriented lamellar-like nanostructures doped with heavy ions together with embedded metallic nano-inclusions such as Ag [23]. Figures 6(a)–(c) show SEM images of fracture surface (a), high-resolution transmission electron microscopy (HRTEM) (b) and bright field (BF)-TEM images (c), of a typical nanostructured bulk sample Ca2.8Lu0.15Ag0.05Co4O9 prepared by SPS. A very homogeneous and dense microstructure exhibiting a pronounced lamella-like morphology can be clearly observed (figure 6(a)). The HRTEM image shows a modulated nano-scale layered structure (figure 6(b)), and the evidence of Ag metallic nano-inclusions was found in the misfit-layered cobalt oxide system, as is clearly shown in figure 6(c). This effective approach has led to a remarkable improvement of ZT (=0.6 at 1100 K) for TE p-type oxide materials. It is more interesting to note that ZT of this material is further improved by about 6% after a subsequent heat treatment and then this high ZT value is maintained even after further extreme heating and cooling cycles (figure 6(d)). The excellent chemical and thermal stability of this material as well as the rapid increase in ZT with increasing temperature strongly suggest this system as a very promising oxide TE material for high-temperature power generation. However, there is still ongoing work in order to avoid using expensive elements such as Lu. These findings demonstrate an effective way (oriented lamella nanostructure + heavily doping + metallic nano-inclusions) to improve the TE performance of polycrystalline materials, and it may be also applied to other TE systems.
Figure 6. (a) SEM image of fracture surface of the Ca2.8Lu0.15Ag0.05Co4O9 sample prepared by SPS, (b) high-resolution transmission electron microscopy (HRTEM), (c) BF-TEM images and (d) the TE performance at 1100 K under various heating and cooling cycles in air ((c) and (d) adapted from [23]).
Download figure:
Standard image2.2. n-type ZnO-based oxides
Zinc oxide is a wide band gap semiconductor with a direct band gap of 3.2–3.5 eV [2]. ZnO crystallizes in the wurzite structure, and non-doped bulk ZnO is an n-type semiconductor i.e. its electrical resistivity decreases with increasing temperature. A small amount of Al doping could result in remarkable decreases in the electrical resistivity and change the conduction behavior from semiconducting to metallic. ZnO is high-temperature stable, non-toxic and low-cost. These properties make it a promising candidate for n-type TE materials [6, 54–60]. The extraordinary high-temperature n-type TE properties of Al-doped ZnO were first brought to public attention by Ohtaki et al in 1996 [58]. A long history of development through nanostructuring as well as heavily doping on this material for high-temperature TE oxide has recently led to successive progress in ZT = 0.47 at 1000 K and 0.65 at 1247 K for a dual doping of Al and Ga using conventional pressing techniques [6]. A recent review by Ohtaki [2] provides a deep understanding of the ZnO-based oxides system as well as the prospect of how to improve the TE performance of this type of materials. Very recently, Jood et al [54] have reported a figure-of-merit ZT of 0.44 at 1000 K for a nanostructured Al-doped ZnO, again confirming that the ZnO-based oxide is a good candidate n-type high-temperature TE material. At the Department of Energy Conversion, Denmark, single Al-doped as well as Al, Ga dually doped ZnO oxides have been studied and prepared using SPS technique. The influence of SPS conditions on the TE properties and the microstructure evolution were investigated in detail in order to understand the sintering mechanism of this system [25]. However, using this technique at 1173 K, the ZT values are in the range of 0.14–0.17, which is much smaller than that reported so far. A possible reason may be due to the inhomogeneous grains distribution of the samples densified by SPS process [25]. Further investigations are in progress. Figure 7 summarizes a historical development of n-type and p-type based on ZnO and Ca3Co4O9 TE oxides over the years, suggesting that these materials are promising candidates for high-temperature power generation.
Figure 7. The historical progress in development of two typical oxide TE materials for doped n-type ZnO and p-type Ca3Co4O9.
Download figure:
Standard image3. High-temperature oxide modules
Fabricating modules using oxide materials, is facing some critical issues, on top of their relatively low ZT, such as the mismatch of p- and n-type elements, difficulty in electrodes joining and the instability of interface contacts at high temperatures. These drawbacks make the performance of oxide modules still low as compared with the ones built from alloys. Nevertheless, oxide TE modules are still very attractive due to their capability of operating at high temperatures with low cost of materials processing.
For oxide module fabrication in this work, the Ag foil and paste were used to make electrodes. A set consisting of 16 legs (eight unicouples of p-type Ca3Co4O9 and n-type Al-doped ZnO) were attached using Ag paste onto designed alumina substrates with Ag electrodes on top. The whole module was then placed in a graphite die and hot-pressed in Ar atmosphere at 1073 K under a pressure of about 30 MPa. The power generation of the module was examined using a commercial module testing system TEGeta-PANCO in air at high temperature (heater temperature) up to 1300 K. When testing the modules at high temperature in air, boron nitride spray was used to prevent metallic contact layers and n-type elements from oxidizing.
In fact, a considerable number of oxide-based TE modules have been fabricated and examined [61–74]. Most of the previous studies have been focused on modules containing Ca3Co4O9 as p-leg and perovskite n-type legs such as CaMnO3 and LaNiO3 [61, 63, 67, 68] or Al-doped ZnO as n-leg with p-leg of NaCo2O4 [62, 64]. However, the combination of Ca3Co4O9 as p-leg and Al-doped ZnO as n-leg in a practical module has not been broadly investigated. This may be due to the fact that it is challenging to master the fabrication of these two components in a large quantity and reproducibility. Beside the choice of materials, the performance of an oxide module strongly depends on the contact resistance between TE oxide legs and metal electrodes, of which the actual internal resistance could be over 400% higher than the theoretical value [70].
Figures 8(a) and (b) show the photo of the actual p–n legs and a typical oxide module comprising eight p-Ca3Co4O9/n-Zn0.98Al0.02O unicouples and its TE power generation characteristics at different applied temperature gradients of ΔT = 358 and 496 K. At the highest applied temperature of TH/TL =906/410 K corresponding to ΔT = 496, the measured open circuit voltage (Vmax) of the module was 0.7 V, and the voltage decreased linearly with increasing current (see figure 8(b)). The measured maximum output power was 38.4 and 65.3 mW at ΔT = 358 and 496 K, respectively. The measured maximum power density of 45.3 mW/cm-2 at ΔT = 496 is comparable with some oxide modules [61, 63], and about two times higher than the p-NaCo2O4/n-ZnAlO modules [62, 64]. The higher power density of our module could be attributed to a lower internal contact resistance (1.8 Ω) of ours than that reported for p-NaCo2O4/n-ZnAlO (2.97 Ω) [64]. Listed in table 1 are the power generation characteristics of our modules and most of the oxide modules reported so far in the literature.
Figure 8. (a) Photo of the actual p–n legs and present oxide module consisting of eight unicouples, (b) TE power generation characteristics of the oxide module at different temperature gradients.
Download figure:
Standard imageTable 1. Power generation characteristics of present oxide modules as compared with the others reported.
| Reference | Materials | No. of p–n | Jointing technique | Thot (K) | ΔT (K) | V0 (V) | Pmax (mW) | Legs size (mm) | Power density (mW cm−2) |
|---|---|---|---|---|---|---|---|---|---|
| Urata et al [61] | p-Ca2.7Bi0.3Co4O9 n-CaMn0.98Mo0.02O3 | 8 | Ag | 1273 | 298 | 1.0 | 170 | 5 × 5 | 42.5 |
| Souma et al [64] | p-NaCo2O4 n-Zn0.98Al0.02O | 12 | Diffusion welding | 934 | 455 | 0.8 | 52.5 | 3 × 4 | 18.2 |
| Matsubara et al [63] | p-Ca2.75Gd0.25Co4O9 n-Ca0.92La0.08MnO3 | 8 | Pt paste | 773 | 390 | 1.0 | 63.5 | 3 × 3 | 44.1 |
| Souma et al [62] | p-NaCo2O4 n-Zn0.98Al0.02O | 12 | Ag, diffusion welding | 839 | 462 | 0.8 | 58 | 3 × 4 | 20.1 |
| Shin et al [65] | p-Li-doped NiO n-Ba0.2Sr0.8PbO3 | 4 | Sintering | 1164 | 539 | 0.4 | 34.4 | 3 × 4 | 35.8 |
| Urata et al [66] | p-Ca2.7Bi0.3Co4O9 n-CaMn0.98Mo0.02O3 | 8 | Ag paste | 897 | 565 | 1 | 170 | 5×5×4.5 | 42.5 |
| Noudem et al [67] | p-Ca3Co4O9 n-Ca0.95Sm0.05MnO3 | 2 | Ag paste | 990 | 630 | 31.5 | 4×4×10 | 49.2 | |
| Funahashi et al [68] | p-Ca2.7Bi0.3Co4O9 n-La0.9Bi0.1NiO3 | 1 | Ag paste | 1073 | 500 | 0.1 | 94 | 3.7×4–4.5 | 310 |
| Lim et al [70] | p-Ca3Co4O9 n-Ca2.9Nd0.1MnO3 | 1 | Ag paste | 1175 | 727 | 95 | 8.5×6.0 | 93.2 | |
| Han et al [69] | p-Ca3Co3.8Ag0.2O9 n-Ca0.98Sm0.02MnO3 | 2 | Ag paste +5 wt% CaMnO3 | 873 | 523 | 0.3 | 36.8 | 3×6×6 | 49.1 |
| Funahashi and Mikami [71] | p-Ca2.7Bi0.3Co4O9 n-La0.9Bi0.1NiO3 | 140 | Ag paste | 1072 | 551 | 4.5 | 150 | 1.3×1.3×5 | 31.7 |
| Choi et al [72] | p-Ca3Co4O9 n-(ZnO)7In2O3 | 44 | Ag paste | 1100 | 658 | 1.8 | 423 | 15 × 15 | 2.1 |
| Mele et al [73] | p-Ca3Co4O9 n-Zn0.98Al0.02O | 6 | Ag | 773 | 260 | 0.3 | 3.7 | 5 × 5 × 15 | <0.1 |
| This work | p-Ca3Co4O9 n-Zn0.98Al0.02O | 8 | Ag paste and foil hot-press | 906 | 496 | 0.7 | 65.3 | 3×3×8 | 45.3 |
4. Concluding remarks
We show that oxide TE materials and high-temperature oxide modules are becoming more and more important, both for research and applications, because they can be used for waste heat recovery at high temperatures. Among TE oxides, Ca3Co4O9 and Al-doped ZnO are very promising, and the development in their TE performance has marked an obvious progress by means of the nanostructuring approach. The performance of TE modules built from these oxides is getting encouraging results. However, contact resistance and the long-term stability of oxide modules still need to be further explored.
Acknowledgments
The authors thank the Programme Commission on Energy and Environment (EnMi) which is part of the Danish Council for Strategic Research (contract no 10–093971) for sponsoring the OTE-POWER research work.
Footnotes
- *
Invited talk at the 6th International Workshop on Advanced Materials Science and Nanotechnology IWAMSN2012, 30 October–2 November, 2012, Ha Long, Vietnam.