Ba1/3CoO2: A Thermoelectric Oxide Showing a Reliable ZT of ∼0.55 at 600 °C in Air.

Thermoelectric energy conversion technology has attracted attention as an energy harvesting technology that converts waste heat into electricity by means of the Seebeck effect. Oxide-based thermoelectric materials that show a high figure of merit are promising because of their good chemical and thermal stability as well as their harmless nature compared to chalcogenide-based state-of-the-art thermoelectric materials. Although several high-ZT thermoelectric oxides (ZT > 1) have been reported thus far, the reliability is low due to a lack of careful observation of their stability at elevated temperatures. Here, we show a reliable high-ZT thermoelectric oxide, Ba1/3CoO2. We fabricated Ba1/3CoO2 epitaxial films by the reactive solid-phase epitaxy method (Na3/4CoO2) followed by ion exchange (Na+ → Ba2+) treatment and performed thermal annealing of the film at high temperatures and structural and electrical measurements. The crystal structure and electrical resistivity of the Ba1/3CoO2 epitaxial films were found to be maintained up to 600 °C. The power factor gradually increased to ∼1.2 mW m-1 K-2 and the thermal conductivity gradually decreased to ∼1.9 W m-1 K-1 with increasing temperature up to 600 °C. Consequently, the ZT reached ∼0.55 at 600 °C in air.

Introduction

Thermoelectric energy conversion technology has attracted attention as an energy harvesting technology that converts waste heat into electricity by means of the Seebeck effect.[1−3] The efficiency of the conversion from the temperature difference to electricity strongly depends on the thermoelectric figure of merit ZT (=S2·σ·T·κ–1, where S is the thermopower or Seebeck coefficient, σ is the electrical conductivity, T is the absolute temperature, and κ is the total thermal conductivity) of the thermoelectric materials. At present, practically available thermoelectric materials are n-type Bi2Te3 (ZT ∼ 1 at 100 °C), n-type PbTe (ZT ∼ 0.8 at 350 °C), n-type CoSb3 (ZT ∼ 0.8 at 550 °C), n-type SiGe (ZT ∼ 1 at 900 °C), p-type Sb2Te3 (ZT ∼ 1 at 50 °C), p-type PbTe (ZT ∼ 0.7 at 400 °C), and p-type SiGe (ZT ∼ 0.6 at 800 °C).[1] Since these ZT values are ∼1, ZT = 1 is often called the threshold for practical application.

Since 2000, many efforts have been made to develop thermoelectric materials with high ZT (>1) such as lead chalcogenides,[4−6] skutterudites,[7−9] half-Heuslers,[10−12] and SnSe.[13−16] For example, an extremely high ZT of ∼2.5 at 650 °C was obtained in the PbTe-SrTe system by nonequilibrium processing.[14] More recently, an even higher ZT value of 3.1 at 510 °C was achieved in hole-doped SnSe polycrystalline samples by carefully removing performance-unfavorable tin oxides.[16] Despite the progress made in thermoelectric materials, the practical applications of these materials are still limited due to the high cost of the rare elements involved and the shortage of natural resources (Te is one of the rarest elements on earth). In addition, most sulfides, selenides, and tellurides are toxic and thermally and chemically unstable. Developing good thermoelectric materials with high ZT without the above issues is essential for widespread use.

Since most metal oxides show good chemical and thermal stability in air as well as a harmless nature compared to chalcogenide-based state-of-the-art thermoelectric materials, oxide-based thermoelectric materials that show a high ZT are promising.[17,18] Several high-ZT thermoelectric oxides have been reported thus far; Fujita et al.[19] reported that the NaCoO2 single crystal exhibited a ZT of ∼1.2 at 800 K. Acharya et al.[20] reported that Nb-doped SrTiO3 with natural graphite exhibited a ZT of ∼1.42 at 1050 K. Biswas et al.[21] reported that graphene oxide-encapsulated ZnO nanocomposites exhibited a ZT of ∼0.52 at 1100 K. However, the reliability is considerably low due to a lack of careful observation of their stability at elevated temperatures.

To prevent misunderstanding of the thermoelectric properties as well as the chemical and thermal stability of oxide-based thermoelectric materials at high temperatures, we have thus far investigated this subject using epitaxial thin films of oxide-based thermoelectric materials.[22] Since the surface area of thin films is larger than that of bulk, thin films exhibit high sensitivity to the atmosphere. For example, an Al-doped ZnO epitaxial film evaporates at high temperatures (>600 °C) in air, indicating that the reported ZT values of ZnO-related thermoelectric oxides at high temperatures might be mostly invalid. Thus, reliable oxide-based thermoelectric materials that show high ZT (>1) at high temperatures have not yet been discovered.

To discover oxide-based thermoelectric materials that show high ZT, we firstly focused layered cobalt oxide Na3/4CoO2 that was discovered by Terasaki et al.[23] Na3/4CoO2 shows rather high power factor (S2σ) whereas high κ, therefore ZT is low. We studied the thermoelectric properties of layered cobalt oxides ACoO2 (A = Na3/4, Li, Ca1/3, Sr1/3, and Ba1/3) epitaxial films step by step as follows. Firstly, we developed the epitaxial film growth method of Na3/4CoO2 namely reactive solid phase epitaxy in 2005.[24] After that, we found that the Na ions in the Na3/4CoO2 films can be exchanged for Li,[25] Sr,[26] and Ca.[27] Further, we found that the layer can be inclined by using M-plane sapphire as the substrate.[28] Recently, we began to measure the in-plane κ of the films by the time domain thermoreflectance (TDTR) method. We hypothesized that a layered cobalt oxide composed of heavy ions shows low κ due to vibrational damping caused by the heavy ions while maintaining a high S2σ.[29] As a result, we recently found that a Ba1/3CoO2 film exhibits a relatively high ZT of ∼0.11 at room temperature,[30] which is the highest among the reported oxide thermoelectric materials except for oxychalcogenide (BiO)(CuSe).[31] In this study, we systematically investigated the thermal stability of ACoO2 (A = Na3/4, Ca1/3, Sr1/3, and Ba1/3) films at high temperature in air and clarified that Ba1/3CoO2 is stable up to 600 °C. Then, we measured the thermoelectric properties of Ba1/3CoO2 films up to 600 °C. Here, we show that Ba1/3CoO2 exhibits a reliable high ZT of 0.55 at 600 °C in air. This high ZT of Ba1/3CoO2 is reproducible and reliable, the highest among oxides and comparable to that of p-type PbTe and p-type SiGe. These results reveal that Ba1/3CoO2 would be a good candidate for a high-temperature thermoelectric material.

Experimental Section

Fabrication of ACoO2 (A = Na3/4, Ca1/3, Sr1/3, and Ba1/3) Films

ACoO2 epitaxial films were fabricated on (111) YSZ or (11̅00) α-Al2O3 substrates by reactive solid-phase epitaxy (R-SPE) of Na3/4CoO2 films followed by ion exchange treatment. First, CoO films were heteroepitaxially grown on the substrate at 700 °C in an oxygen atmosphere (10–3 Pa) by a pulsed laser deposition technique (KrF excimer laser, ∼2 J cm–2 pulse–1, 10 Hz). Then, the CoO film was heated with NaHCO3 powder at 750 °C in air. This results in the formation of a Na3/4CoO2 epitaxial film. Then, the Na+ ions in the resultant films were exchanged with Ca2+, Sr2+, and Ba2+ ions by applying the appropriate ion-exchange treatment. Details of the fabrication procedure have been published elsewhere.[24,26,27,29,30]

Crystallographic Analyses

The crystalline phase, orientation, and thickness of the resultant films were analysed by X-ray diffraction (XRD) (Cu Kα1, ATX-G, Rigaku). Out-of-plane Bragg diffraction patterns, in-plane Bragg diffraction patterns, rocking curves, and X-ray reflection patterns were acquired.

Resistivity and Thermopower Measurements

At room temperature and below room temperature, the electrical resistivity (ρ) or conductivity (1/ρ, σ) of films were measured by the dc four-probe method with the van der Pauw electrode configuration. The thermopower (S) was measured by the steady method. Homemade equipment was used for these measurements. Above room temperature, σ and S values were measured by thermoelectric measurement equipment (MODEL RZ2001i, Ozawa Science Co.).

Thermal Conductivity Measurement

The thermal conductivity (κ) of the films in the direction perpendicular to the substrate surface was measured by the time domain thermoreflectance (TDTR, PicoTR, PicoTherm Co.) method. Before the measurement, a dense Pt film was deposited on the surface of the ACoO2 film as the transducer by dc sputtering at room temperature. The decay curves of TDTR signals were simulated to obtain the thermal conductivity. The specific heat capacity used for the TDTR simulation was measured using differential scanning calorimetry (DSC, DSC8500, PerkinElmer) (Supporting Information, Figure S3). The in-plane κ of the films was extracted according to the following equation as previously reported.[20,21]where φ, κobsd, κ// and κ⊥ are the inclination angle of the layers relative to the perpendicular of the film surface, observed thermal conductivity, in-plane κ, and cross-plane κ, respectively (Supporting Information, Figure S4). In this study, φ = 35°, which was proven by scanning transmission electron microscopy (STEM) observations in our previous study.[21]

Results and Discussion

Figure a schematically illustrates the crystal structure of ACoO2 (A = Ba1/3). The CoO2 layer is composed of edge-sharing CoO6 octahedra, and the A layer (A = Na3/4, Ca1/3, Sr1/3, Ba1/3) is alternately stacked along the c-axis. We fabricated 150-nm-thick c-axis oriented ACoO2 (A = Na3/4, Ca1/3, Sr1/3, and Ba1/3) epitaxial films on (111) yttria stabilized zirconia (YSZ) substrates. Details of the fabrication procedure have been reported elsewhere.[24,26,27,29,30] Only an intense 0002 ACoO2 diffraction peak is observed in the out-of-plane X-ray diffraction (XRD) patterns of the as-grown ACoO2 films (Figures b and S1a–d). The c-axis lattice parameter was calculated to be as follows: Na3/4 = 1.094 nm, Ca1/3 = 1.087 nm, Sr1/3 = 1.153 nm, and Ba1/3 = 1.223 nm, which correspond well with a previous report.[30]

Figure 1

Change in the electrical resistivity (ρ) of ACoO2 (A = Na3/4, Ca1/3, Sr1/3, and Ba1/3) films after annealing at high temperatures for 0.5 h in air. (a) Schematic crystal structure of ACoO2 (A = Ba1/3). (b) Out-of-plane XRD pattern of the ACoO2 (A = Ba1/3) film (as-grown). For other compositions, see Figure S1a–c. (c) ρ of the as-grown ACoO2 films measured at room temperature. (d) Resistivity ratio of the thermally annealed sample/as-grown sample (ρanld/ρas-grown) measured at room temperature after each annealing step. The ρanld/ρas-grown of the Na3/4CoO2 and Ca1/3CoO2 films starts to increase at 350 °C, that of the Sr1/3CoO2 film starts to increase at 450 °C, and that of the Ba1/3CoO2 film starts to increase at 650 °C.

First, we clarified the stability of the resultant ACoO2 (A = Na3/4, Ca1/3, Sr1/3, and Ba1/3) films at high temperatures in air. Before annealing, we measured the electrical resistivity (ρ) of the as-grown films at room temperature in air (Figure c). Note that the ρ of Ca1/3CoO2 was higher than that of Sr1/3CoO2 and Ba1/3CoO2. The carrier mobility of Ca1/3CoO2 decreases dramatically when the Ca arrangement is orthorhombic which occurs when the film is annealed above 300 °C.[27] Since the ion exchange treatment from Na to Ca is performed at 300 °C, there is a possibility that a tiny amount of orthorhombic arrangement of Ca formed during the ion exchange treatment. Then, we annealed the films at high temperatures and subsequently performed structural and electrical measurements at room temperature. Figure d shows the change in the ρ ratio of an annealed sample (ρanld) and the as-grown sample (ρas-grown) measured at room temperature after annealing at each temperature for 0.5 h in air. The ρanld/ρas-grown of the Na3/4CoO2 and Ca1/3CoO2 films starts to increase at 350 °C, and the ρanld/ρas-grown of the Sr1/3CoO2 film starts to increase at 450 °C, while that of the Ba1/3CoO2 film remains stable below 600 °C. Thus, the highest applicable temperatures in air for Na3/4CoO2, Ca1/3CoO2, Sr1/3CoO2, and Ba1/3CoO2 are 350, 350, 450, and 600 °C, respectively.

To clarify the origin of the resistivity increase, we performed crystallographic analyses of the annealed films. Figure a shows the changes in the out-of-plane XRD pattern of the ACoO2 films measured at room temperature after annealing at each temperature for 0.5 h in air. The 0002 Na3/4CoO2 diffraction peak intensity decreases above 450 °C, and the 111 Co3O4 peak appears after that due to evaporation of high vapor pressure Na. In contrast, the 0002 Ca1/3CoO2, 0002 Sr1/3CoO2, and 0002 Ba1/3CoO2 peaks are clearly seen below 650 °C. Thus, the out-of-plane XRD patterns of Ca1/3CoO2, Sr1/3CoO2, and Ba1/3CoO2 are not sufficient to clarify the origin of the resistivity increase.

Figure 2

Change in the crystal structure of ACoO2 (A = Na3/4, Ca1/3, Sr1/3, and Ba1/3) films after annealing at high temperatures for 0.5 h in air. (a) Out-of-plane XRD patterns measured at room temperature after each annealing step. The 0002 Na3/4CoO2 diffraction peak intensity decreases above 450 °C due to the decomposition into Co3O4 while remaining stable for Ca1/3CoO2, Sr1/3CoO2, and Ba1/3CoO2 up to 650 °C. (b) In-plane XRD patterns measured at room temperature after each annealing step. The 1/3 and 2/3 diffraction peaks of the Ca1/3CoO2 film disappear at approximately 200 °C, and the 1/2 diffraction peak appears above 200 °C. The 1/3 and 2/3 diffraction peaks of the Sr1/3CoO2 film disappear at approximately 450 °C, and the 1/2 diffraction peak appears above 450 °C. The 1/3 and 2/3 diffraction peaks of the Ba1/3CoO2 film are stable up to 600 °C.

We then analysed the structural change in the ACoO2 epitaxial films through in-plane XRD measurements. Figure S1e–h show the in-plane XRD patterns of the as-grown ACoO2 epitaxial films. An intense 112̅0 ACoO2 diffraction peak is seen at q/2π = 7.05 nm–1 together with the 22̅0 YSZ substrate peak in all cases, indicating that the lattice parameter a of the ACoO2 epitaxial films is ∼0.284 nm. Several diffraction peaks with fractional diffraction indices are seen in the in-plane XRD patterns, indicating an ordered structure of the A ions. The fractional peak of (112̅0) × 1/2 is from the orthorhombic lattice of the cation layer; the fractional peaks of (112̅0) × 1/3 and 2/3 are from the hexagonal lattice.[32] Thus, in the as-grown state, the Na3/4CoO2 film belongs to the orthorhombic lattice, and the Ca1/3CoO2 film belongs to the hexagonal lattice; the Sr1/3CoO2 and Ba1/3CoO2 films belong to the hexagonal-orthorhombic hybridized lattice.

Figure b shows the in-plane XRD patterns of the ACoO2 films measured at room temperature after annealing at each temperature for 0.5 h in air. The 1/3 and 2/3 diffraction peaks of the Ca1/3CoO2 film disappear at approximately 200 °C, and the 1/2 diffraction peak appears above 200 °C, revealing that a phase transition from hexagonal to orthorhombic occurs.[27] The Sr1/3CoO2 film shows a transition from the hexagonal-orthorhombic hybridized lattice to the orthorhombic lattice at ∼450 °C. The 1/3 and 2/3 diffraction peaks of the Ba1/3CoO2 film are stable up to 600 °C. From these results, we determined that the thermal stability of the ordered structure affects the electrical resistivity of ACoO2 films.

Then, we measured the thermoelectric properties of the ACoO2 (A = Na3/4, Ca1/3, Sr1/3, and Ba1/3) films (as-grown samples) in the in-plane direction at high temperatures in air. The room-temperature electrical conductivity (σ) of Na3/4 is ∼1400 S cm–1, Ca1/3 is ∼670 S cm–1, Sr1/3 is ∼1250 S cm–1, and Ba1/3 is 1100 S cm–1 (Figure a), which are slightly smaller than those of the films grown on (0001) α-Al2O3 substrates.[30] σ gradually decreases with temperature, showing the metallic nature of the films. In contrast, the thermopower (S) of the ACoO2 films is ∼80 μV K–1 at room temperature in all cases and linearly increases with temperature (Figure b). This T-linear increase of S is due to the metallic nature of the films. The low-temperature thermoelectric properties (Figure S2) support this conclusion. Although there is a difference in σ between Ca1/3 and the others, S does not reflect the difference in σ, indicating that the carrier relaxation time of Ca1/3 is shorter than that of the others.

Figure 3

Thermoelectric properties of ACoO2 (A = Na3/4, Ca1/3, Sr1/3, and Ba1/3) epitaxial films measured in the in-plane direction. (a) Electrical conductivity (σ). (b) Thermopower (S). (c) Power factor (S2σ). (d) Thermal conductivity (κ). In all cases, σ decreases with temperature, while S increases with temperature. The S2σ values first decrease and then increase with temperature. κ slightly decreases with temperature in all cases. Note that in the case of the Ba1/3CoO2 film, S2σ is ∼1.2 mW m–1 K–2 and κ is ∼1.9 W m–1 K–1 at 600 °C in air.

In this study, we used (111) YSZ as the substrate. The lateral grain size of the films on (111) YSZ is smaller than that of the films grown on (0001) α-Al2O3 substrates. In the case of (0001) α-Al2O3 substrates, an amorphous Na-Al-O layer is formed between the film and the substrate during Na3/4CoO2 film growth.[33] Therefore, the cobaltite films release epitaxial strain in the case of α-Al2O3 substrates, and lateral grain growth occurs. In contrast, such an amorphous layer is not formed in the case of the (111) YSZ substrate. Therefore, lateral grain growth is suppressed and the carrier mobility is suppressed due to the grain boundaries at lower temperatures.

Next, we calculated the thermoelectric power factor (S2σ) using the observed σ and S (Figure c). The room-temperature S2σ of Na3/4 is ∼0.87 mW m–1 K–2, Ca1/3 is ∼0.43 mW m–1 K–2, Sr1/3 is ∼0.72 mW m–1 K–2, and Ba1/3 is ∼0.70 mW m–1 K–2, which are smaller than those in our previous report (∼1.2 mW m–1 K–2 in all cases[30]) due to the smaller σ and S. S2σ decreases and increases with temperature. Note that the S2σ of the Ba1/3CoO2 film reaches ∼1.2 mW m–1 K–2 at approximately 600 °C in air.

Then, we estimated the temperature dependence of the thermal conductivity (κ) of ACoO2 films grown on (111) YSZ and (11̅00) α-Al2O3 substrates in the in-plane direction (Figure d). The room-temperature κ of Na3/4 is ∼5.7 W m–1 K–1, Ca1/3 is ∼6.3 W m–1 K–1, Sr1/3 is ∼3.7 W m–1 K–1, and Ba1/3 is ∼3.2 W m–1 K–1. Regardless of the composition of A, all κ values decreased gradually with increasing temperature, suggesting heat conduction dominated by phonon-phonon scattering mechanism commonly observed at elevated temperatures. The κ of the Ba1/3CoO2 film reaches ∼1.9 W m–1 K–1 at 600 °C in air.

Finally, we calculated the figure of merit ZT of the ACoO2 films (Figure a). The room-temperature ZT value of ACoO2 films is less than 0.06, but it dramatically increases with temperature. The ZT of Ba1/3CoO2 reaches 0.55 at approximately 600 °C in air. This high ZT of Ba1/3CoO2 is reproducible and reliable, the highest among oxides and comparable to that of p-type PbTe and p-type SiGe (Figure b). These results reveal that Ba1/3CoO2 would be a good candidate for a high-temperature thermoelectric material.

Figure 4

Temperature dependence of the ZT of the Ba1/3CoO2 epitaxial film in the in-plane direction. (a) Comparison among the four ACoO2 (A = Na3/4, Ca1/3, Sr1/3, and Ba1/3) films. ZT increases with increasing temperature in all cases. The Ba1/3CoO2 epitaxial film has the highest ZT among the four ACoO2 epitaxial films and reaches a high ZT value of ∼0.55 at 600 °C. (b) Comparison against commercially available p-type thermoelectric materials (ref (1)). The ZT of the Ba1/3CoO2 epitaxial film at 600 °C is comparable to that of p-type PbTe and p-type SiGe.

Conclusions

In summary, we discovered a reliable high-ZT thermoelectric oxide, Ba1/3CoO2. We clarified the stability of the Ba1/3CoO2 films at high temperatures in air and found that the crystal structure and electrical resistivity of Ba1/3CoO2 were maintained up to 600 °C. The power factor of Ba1/3CoO2 gradually increased to ∼1.2 mW m–1 K–2 and the thermal conductivity gradually decreased to ∼1.9 W m–1 K–1 with increasing temperature up to 600 °C. Consequently, ZT reached ∼0.55 at 600 °C in air. This high ZT of Ba1/3CoO2 is reproducible and reliable (see Figure S5), the highest among oxides and comparable to that of p-type PbTe and p-type SiGe. These results reveal that Ba1/3CoO2 would be a good candidate for a high-temperature thermoelectric material.