JP4894641B2 - Thermoelectric conversion material - Google Patents

Description

  The present invention relates to a thermoelectric conversion material made of cobalt oxide.

When a temperature difference (ΔT) is applied to both ends of a substance, a thermoelectromotive force (ΔV) is generated at both ends. The proportionality coefficient at this time is referred to as Seebeck coefficient (S = ΔV / ΔT). If a circuit is formed by providing terminals at both ends of the sample, the substance acts as a kind of “battery” under the difference in temperature.
That is, heat can be converted into electricity. This is called thermoelectric conversion, and its capacity and performance are called thermoelectric power. A material that efficiently performs thermoelectric conversion is referred to as a thermoelectric material.
Currently, a huge amount of waste heat is generated by the use of fuel on the earth, and almost all is released into the atmosphere. If a part of it is converted into electricity and reused effectively, there is no doubt that it will lead to significant energy savings. To achieve this, it is necessary to develop more efficient thermoelectric materials.
In order to efficiently extract a current from a circuit using a thermoelectric material, not only the Seebeck coefficient of the material substance is large, but also the electrical resistivity (ρ) needs to be low. This corresponds to the internal resistance of the battery. Further, in order to maintain the temperature difference at the sample end, it is advantageous that the heat conduction (κ) is low.

In general, the figure of merit Z = S ² / ρκ or S ² / ρ is used to represent thermoelectric performance. A practical guideline is that a dimensionless index ZT obtained by multiplying Z by temperature T exceeds 1.
On the other hand, there is a reverse process of the Seebeck effect. That is, when a current is passed through the substance, endothermic and exothermic reactions occur at the junction with the conductive wire. This is called the Peltier effect.
Some electronic cooling elements based on this principle have been put into practical use.

Conventionally known thermoelectric materials are mainly compact semiconductors such as Bi ₂ Te ₃ , PbTe, and Si—Ge. A normal semiconductor has a relatively large Seebeck coefficient, but cannot be used as a thermoelectric material because of its large electric resistance. On the other hand, the Seebeck coefficient of metals is usually small. Bi ₂ Te ₃ , PbTe, Si—Ge, etc. dope a large amount of carriers by adding impurities (N> 10 ¹⁹ cm ³ ; N: carrier amount), so that a large Seebeck coefficient and a relatively low electrical resistivity are obtained. Are compatible.

About 10 years ago, it was reported that the layered cobalt oxide NaCo ₂ O ₄ exhibits high thermoelectric power (Non-patent Document 1). Subsequently, it was found that a series of layered compounds showed high thermoelectric power as well. (For example, Ca ₃ Co ₄ O ₉ : Non-Patent Document 2, Bi ₂ Sr ₂ Co ₂ O ₉ : Non-Patent Document 3) These cobalt oxides, unlike conventional shrink-resistant semiconductor materials, have high heat The electromotive force generation mechanism is considered to be strongly related to the electron correlation occurring in the 3d orbit. (For example, Non-Patent Document 4) Because of a unique mechanism, there is a possibility that the characteristics of conventional compact semiconductor-based thermoelectric materials may be surpassed at once. Cobalt oxide includes a CdI ₂ type CoO ₂ triangular lattice layer in the crystal structure, and its high thermoelectric power is aligned with high-temperature superconductivity of copper oxide, giant magnetoresistance effect of manganese oxide, etc. As one of the quantum effects, it has recently attracted much attention among physical property researchers.

However, at room temperature or higher, the conventional layered cobalt oxide tends to be saturated, and the electrical resistivity also increases with temperature, so that there is a drawback that the efficiency (S ² / ρ) is deteriorated.

I. Terasaki, Y. et al. Sasago, and K.K. Uchinokura, Phys. Rev. B 56 (1997) R12585 Ca3Co4O9: A. C. Masset, C.I. Michel, A.M. Mainan, M.M. Hervieu, O .; Toulmonde, F.M. Studer, B.A. Raveau, and J.M. Hejtmanek, Phys. Rev. B 62 (2000) 166 F. Chen, K.K. L. Stokes, R.M. Funahashi, Appl. Phys. Lett. 81 (2002) 2379. W. Koshibae, K .; Tsutsui, and S.A. Maekawa, Phys. Rev. B 62 (2000) 6869.

  In view of such circumstances, an object of the present invention is to provide a thermoelectric conversion material having a maximum class Seebeck coefficient in a temperature region of room temperature or higher by reducing heat dependency.

The thermoelectric conversion material of the present invention has a CaFe ₂ O ₄ (calcium / ferrite) type structure, and is characterized in that the Fe is substituted with Co.

The substance CaCo ₂ O ₄ found this time has the largest class Seebeck coefficient in the temperature region above room temperature.
Conventional layered cobalt oxides tend to be saturated, and the electrical resistivity also increases with temperature, so that the efficiency (S ² / ρ) is disadvantageously deteriorated.

However, the theoretical prediction of Koshibae et al. (Previously cited document (Non-Patent Document 4)) is not limited to a layered cobalt-based material, but is extended to a transition metal oxide having an appropriate electronic state and crystal structure. As a result of material synthesis search of various transition metal oxides, we succeeded in synthesizing the new material CaCo ₂ O ₄ for the first time.
Furthermore, it has been found that this substance has a potential as a thermoelectric material not inferior to conventional layered cobalt oxide.
In particular, this substance shows a large Seebeck coefficient near room temperature. Furthermore, the Seebeck coefficient has a feature that it hardly saturates even at a temperature slightly higher than room temperature with respect to a temperature change.
Since the electrical resistivity is a little higher than that of the conventional material, it is difficult to use this material as a thermoelectric material as it is. However, if single crystallization or appropriate carrier doping is performed based on this material, the electrical resistivity can be reduced by several orders of magnitude. That can be easily guessed. The substance thus obtained is likely to exhibit performance comparable to conventional thermoelectric materials.

<Method for synthesizing CaCo ₂ O ₄ >
The CaCo ₂ O ₄ polycrystalline sample was artificially produced by a high-temperature and high-pressure synthesis method (solid-phase reaction method under high temperature and high pressure). (May not exist in nature.) This substance is considered a substance that can be synthesized only under high pressure. The raw material powder is chemically reacted under high temperature and high pressure to form this substance, but after the reaction, it is rapidly cooled (quenched) to room temperature while maintaining the high pressure, and then the pressure is released. Thereby, it can hold | maintain stably also at room temperature normal pressure.

Co ₃ O ₄ (3N) and (Ca ₂ CoO ₃ ) _0.62 CoO ₂ powder were used as starting materials for the high-pressure synthesis.
In this case, the latter ((Ca ₂ CoO ₃ ) _0.62 CoO ₂ ) was first prepared by subjecting Co ₃ O ₄ and CaCO ₃ (3N) powder to a solid-phase reaction method under normal pressure (in a 1 atm O ₂ gas stream, The reaction was performed at 900-950 ° C. In the middle, several times of pulverization and mixing were performed.
Co ₃ O ₄ : (Ca ₂ CoO ₃ ) _0.62 CoO ₂ = 0.86: A powder mixed in a molar ratio of 3 is made. This time, an agate mortar was used for mixing. Although the above raw materials were used this time, the molar ratio of the atoms may be Ca: Co: O = 1: 2: 4 + δ, and other starting materials can be used. Here, δ means an excessive amount of oxygen.
Since high-pressure synthesis proceeds in a closed system, it must not contain any other elements. For example, carbonates such as CaCO ₃ cannot be used as raw materials because there are extra C atoms. Nitrate is also impossible. The mixed powder was dried at 300 ° C. for several hours.

  Fill and seal the mixed powder in a platinum capsule. At this time, the platinum capsule had a diameter (outer diameter) of 6.8 mm, a thickness of 0.2 mm, and a height of about 5 mm. This is due to the limitation of the sample space of the high pressure synthesizer. Powder filling was performed in a glove box filled with argon gas. This platinum capsule is incorporated into a pressure cell (comprised of a pressure medium mainly composed of rock salt and a carbon heater). Using a flat belt type high-pressure press, the reaction is carried out under high temperature and high pressure (1500 ° C., 6 GPa) for 1 hour, followed by rapid cooling to room temperature and pressure release.

  The heat-treated platinum capsule is taken out, and the product (sample) inside is taken out. The sample is sintered into pellets by the reaction. Since the sample surface may have reacted with the platinum of the capsule material, the surface is scraped off. Finally, a bulk mass sample of about 0.4 g was obtained. FIG. 8 is a photograph of a small piece obtained by crushing the sample on the pellet obtained. It is an almost black mass.

  It is important to be synthesized under high temperature and pressure. The type of press and the size of the sample are not important. In addition, the reaction temperature, pressure, time, etc. are not limited to this one point, and the synthesis cannot be performed, and it seems that some range is acceptable. However, in order to obtain a good quality sample, it is necessary to guarantee the uniformity of temperature, pressure, mixing state, etc. to some extent. Other types of capsules may also be possible. This time, we chose platinum, which has a high melting point and relatively little reaction with the product.

<Sample evaluation method>
In order to obtain data for sample identification and crystal structure analysis, an ordinary laboratory powder X-ray diffraction method was used. The diffraction device used this time is Rigaku RINT-ULTIMA III (CuKα ray, Bragg-Brentano optical system, gonio radius 285 mm). The refinement of the structure was performed by Rietveld analysis (Rietan-2000 program). Measurement conditions are 2θ = 10-110 degrees, step scan mode, step width 0.02 degrees, data accumulation time 30 seconds / 1 step, diverging slit = scattering slit = 1/3 degree, light receiving slit = 0.3 mm. there were. The OFFE program was used for calculation of bond length and bond angle.
For the magnetic data measurement, a SQUID (MPMS) apparatus manufactured by Quantum Design was used. A powder sample (20.97 mg) was used. The measurement magnetic field is 1000 Oe, and the temperature range is 2 K-350 K. Zero field cool measurement and field cool measurement were performed.
For the measurement of physical properties (electric resistance, thermoelectromotive force, specific heat), a PPMS device manufactured by Quantum Design was used. Seebeck coefficient and resistivity were measured simultaneously using TTO (thermal conductivity measurement option). The size of the bulk sample is 1.5 × 2.1 × 5.5 mm ³ , and the distance between the voltage / temperature terminals is 2.9 mm. A copper wire was used for the sample electrode, and an ohmic contact was formed through a gold thin film deposited on the sample surface. Data collection is performed in a continuous mode, and the heating rate is 0.3 K / min. It is. Cernox 1050 was used for the thermometer. The temperature difference between the electrodes was controlled to be less than 3% of the measured temperature. The electrical resistance measurement is based on the 4-terminal method. An alternating current of 0.01-0.05 mA, 60-300 Hz was used. The measurement temperature range is 10 K-390 K. Specific heat was measured in the range of 302 K-2 K by relaxation using PPMS. The mass of the sample used was 12.34 mg.

<Measurement and analysis results (CaCo ₂ O ₄ crystal structure and physical properties)>
Crystal structure FIG. 1 (measurement data: Tables 4-1 to 4-6) shows a powder X-ray diffraction pattern of CaCo ₂ O ₃ . This figure is a pattern obtained by a powder X-ray diffraction method. The grating plane distance d is obtained from the diffraction angle (2θ) satisfying the Bragg reflection condition λ = 2 d sin θ (λ = 1.540593 Å, CuK _α characteristic X-ray wavelength, d: grating plane spacing, θ: Bragg reflection angle). . Furthermore, the Rietveld method is a method of refining the crystal structure model by numerically fitting measured values and calculated values of peak intensity (I _hkl ). Red + indicates the measured value, solid line indicates the theoretical value, blue solid line indicates the difference, and green line indicates the Bragg reflection position. FIG. 1 shows a pattern specific (similar) to the CaFe ₂ O ₄ structure.
The intensity of Bragg reflection and the 2θ angle are summarized in Table 1. All reflections can be indexed using an orthorhombic lattice with lattice constants of a = 8.789 (2) Å, b = 2.9006 (7) Å, c = 10.282 (3) Å. is there. This indicates that the sample purity is high (impurity phase is not included). Since the extinction rule is k + 1 = 2n (0kl) and h = 2n (hk0), the possible space groups are Pnma (No. 62) and Pn2 ₁ a (No. 33). Here, highly symmetrical Pnma is assumed.

The structural analysis of CaCo ₂ O ₃ was performed by X-ray Rietveld analysis. The diffraction pattern in FIG. 1 also shows the Rietveld analysis result. Here, the same model as CaFe ₂ O ₄ was used as the structural model. The obtained structural parameters such as atomic coordinates and crystallographic data are summarized in Table 2. Table 3 summarizes the coupling distance and the coupling angle.

2 and 3 show the crystal structures obtained. This is drawn using the values in Table 2.
FIG. 3 was drawn using the coordinate data of Table 1. In (a), the oxygen at the apex of the CoO ₆ octahedron is omitted. (B) A plurality of CoO ₆ octahedrons are bonded side by side in the b-axis direction (backward direction on the paper surface) to form a double chain. Duplexes share a vertex and join with neighboring duplexes to form a three-dimensional network.
As shown in FIG. 3, the most characteristic point in the structure of CaCo ₂ O ₃ is a Co—O double chain running in the b-axis direction. This is formed by the CoO6 octahedron sharing sides within the double chain and sharing vertices between the double chains. It can also be regarded as forming a partial lattice of a Co—O triangular lattice. The t _2g orbitals of Co atoms are oriented in the same direction, and the orbits are expected to overlap.

The Co—O bond length is roughly the same as the distance predicted from the ionic radius, and there is no structural contradiction. Although there is a slight difference between the Co-O bond lengths, the comparative analysis shows that the Co atoms in the oxygen octahedron are slightly displaced inside the double chain (the side where the distance between Co-Co becomes shorter). I understand that. It is a characteristic of the double chain structure. The bond angle within the duplex is approximately 90 degrees. This is an angle at which the Co t _2g orbits can be overlapped, which means that charge transfer is possible in the double chain. On the other hand, the bond angle between the double chains is around 130 degrees. If it is 180 degrees, the overlap of the orbits can be almost ignored, but 130 degrees as measured is that the component where the t _2g orbits overlap is not negligible even between the double chains (ie, double This means that a certain amount of charge transfer is possible even between chains. Actually, no clear anisotropy (one-dimensional property) was observed even in the physical properties described later. This means that the substance has a three-dimensional characteristic due to the structure in terms of electronic structure.

Magnetic Susceptibility FIG. 4 (measurement data: Tables 5 and 6) shows the temperature dependence of magnetic susceptibility. FIG. 4 means that the magnetism of this system can be expressed almost by Curie-Weiss paramagnetism. The Curie constant (C) is small, meaning that the ground state is a cobalt trivalent low spin state (Co ³⁺ , 3d ⁶ t _2g , S = 0). Although a small but finite C value is observed, it seems that there are (almost) isolated spins of about 12% S = 1/2. The cause of the isolated spin may be a slight (magnetic) impurity or a lattice defect, but it is difficult to explain by itself based on comparison with empirical values. Rather, it is reasonable to consider that there is a small amount (a few percent) of holes in addition to that. The existence of holes is also suggested from measured data of electrical resistance, Seebeck coefficient, and specific heat. It can be considered that the Curie temperature (θ _W ) is small, the interaction between spins is quite weak, and it is almost in an isolated state. Also, no magnetic transition is observed in this measurement temperature range. The lack of entropy release due to the phase transition is considered to be the cause of residual entropy even at low temperatures, which generates a large thermoelectric power by combining with the charge.

There are no signs of change due to magnetic transitions or low dimensionality. The temperature change is Curie-Weiss-like. The solid line in the figure is obtained by numerical fitting by Curie-Weiss approximation. Thereby, the Curie constant and the Weiss temperature are obtained. A relatively small Curie constant means that the magnetic state of Co is basically in the 3d ⁶ low spin state (S = 0). Finite spin means that there are about 12% S = 1/2 spins. The cause seems to be mainly magnetic impurities or lattice defects, but the value of 12% has a slightly large feeling. Perhaps it is reasonable to think that there is a small amount of holes and that the isolated spin is involved. A small Weiss temperature of −1.3 K suggests that the spin-spin interaction is weak and almost in an isolated state.

Electrical Resistance FIG. 5 (measurement data: Table 7) shows the temperature dependence of electrical resistivity.
FIG. 5 shows semiconductor temperature dependence (the tendency that resistance increases by a digit as temperature decreases). At low temperatures (below 100K, it shows not the Arrhenius type but the variable range hopping type temperature change (inset). From this, the electronic state of this system is not a normal bandgap semiconductor, but rather a metal It is considered to be close (there is a finite density of states at the Fermi level), and the density of states at the Fermi level is not so high (transition metal oxides exhibiting metallic conduction, as can be seen from the specific heat data) Therefore, it is considered that charge localization occurs at a low temperature, and it is presumed that the Fermi level slightly crosses near the upper end of the band mainly composed of Co t _2g orbitals.

It can be seen that semiconductor temperature dependence is shown through the measured temperature. The absolute value of the resistivity is 3 × 10 ⁻¹ ohm cm at 380 K, and is close to a typical value approximately halfway between a semiconductor and a metal in an electrically conductive oxide. It is characteristic that a temperature-dependent change in curvature is observed between 100 K and 250 K.

From this measurement result, it is presumed that the Co t _2g orbit of this system is almost filled with electrons. In order to know the electronic state in the vicinity of the Fermi level, we tried two types of plots, Arrhenius type and variable range hopping type, as shown in the inset. The results showed that the latter was linear. From this, it is expected that the Fermi level has a small but finite density of states, and a small amount of carriers are localized at low temperatures. It is presumed that carriers were generated by a slight shift in the phase ratio. Presumably, the Fermi level is slightly crossing near the top of the Co t _2g band. Thus, in essence, the electronic structure of the system should be considered metallic.
The temperature dependence near room temperature is rather close to the thermal activation type. At high temperatures (near room temperature), localized carriers are released from potential constraints, and a change in the conduction mechanism appears. At present, it cannot be determined what causes the thermally activated conduction mechanism. Possible possibilities include carrier excitation to the mobility edge, polaron, and grain boundary scattering.

Thermoelectric power (Seebeck coefficient)
FIG. 6 (measurement data: Table 7) shows the temperature dependence of the Seebeck coefficient.
FIG. 6 generally has a metallic temperature dependence (S） T). (In the band gap semiconductor, S 半導体 1 / T.) The inset shows the thermoelectromotive force in the variable range hopping state: S∝T ^1/2 (three-dimensional), S∝T ^1/3 (two-dimensional However, in practice, it showed an intermediate temperature change between 2D and 3D. Certainly not at least one-dimensional. This is consistent with the knowledge obtained from the crystal structure.
The Seebeck coefficient is generally larger than that of a normal substance, and corresponds to a layered cobalt oxide (for example, Ca ₃ Co ₄ O ₉ (Non-patent Document 2); Bi ₂ Sr ₂ Co ₂ O ₉ (Non-Patent Document 3)). The value at 380 K is S = + 147 μV / K.
The temperature dependence near (above) room temperature is quite characteristic. S in ordinary layered cobalt oxide tends to saturate in this temperature range, but in this system (CaCo ₂ O ₄ ), it hardly saturates and continues to increase as the temperature rises. (Unfortunately, it could not be measured in the higher temperature range due to the problem of the specs of the measuring device.) Since higher S and smaller ρ can be expected on the higher temperature side, it is very convenient as a thermoelectric material used in the high temperature range.
Since the negative sign is positive, it can be seen that there are mainly p-type carriers (holes). The tendency for the Seebeck coefficient to increase with increasing temperature means that the electronic structure of the system is metallic, consistent with the electrical resistance results. In particular, since variable range hopping conduction is considered to be dominant at low temperatures, the dimensionality of conduction was determined by a logarithmic plot as shown in the inset. The results suggest that the system is in a three-dimensional or two-dimensional conductor state. (It is certain that it is not at least one-dimensional.) This can be interpreted as there being charge transfer between the double strands, as expected from the structure.

When the temperature is raised, the increase in Seebeck coefficient once slows around 100K. This suggests that charge localization begins to be released near this temperature. Since the Seebeck coefficient is usually expressed by the energy derivative of logarithmic conductivity at Fermi energy, changes in the temperature curvature of the Seebeck coefficient suggest that the electronic structure undergoes significant changes.
The absolute value of the Seebeck coefficient S of CaCo ₂ O ₄ is about 147 μV / K at 380 K, which is clearly larger than that of ordinary metals, and is the value of the layered cobalt oxide expected as a thermoelectric material in recent years. Equal to or more than This suggests that CaCo ₂ O ₄ has the potential as a thermoelectric material. The cause of this large thermoelectric power generation is thought to be due to the residual entropy of spin. When the theoretical formula given by Non-Patent Document 4 is used, the thermoelectric power at the high temperature limit of this system can be estimated as S = 344 μV / K. The actually measured S (= 147 μV / K) is a value when the temperature is 380 K, and it is difficult to compare with the theoretical value of the high temperature limit, but as shown in FIG. Since the saturation tendency of S is not observed, the theoretical value may be reached at a sufficiently high temperature. Incidentally, a substance such as CaCo ₂ O ₄ in which S is not saturated (not observed) is relatively rare. (S in ordinary layered cobalt oxide tends to saturate near room temperature or higher.) This is one of the greatest features of CaCo ₂ O ₄ and is an important factor from the viewpoint of application in a high temperature region. It is considered that a large thermoelectric power is generated by sticking electric charges to the large residual entropy remaining near room temperature because it is not released by the phase transition due to electron correlation. A double chain structure such as CaCo ₂ O ₄ may be convenient for the stage.

Specific Heat FIG. 7 (measurement data: Table 8) shows the temperature change of the specific heat of CaCo ₂ O ₄ .
In a general substance, specific heat is composed of electron specific heat and lattice specific heat, and is expressed by C / T = γ + AT ² . At this time, the extrapolation of the linear portion was plotted against the C / T T ^2, the electronic specific heat coefficient from the intersection of the T = 0 axis γ is determined as shown in Figure 7. Also. Since the coefficient A = a ^{_{(12π 4/5) rN A}} k B (1 / θ D 3), Debye temperature from the slope of the straight line (A) (θ _D) is obtained. The obtained electron specific heat coefficient is γ = 4.48 (7) mJ / mol K ² and is finite. As suggested by the electrical resistance data, the electronic state of this system is not a normal bandgap semiconductor, but rather a metal-like one (there is a finite density of states N (0) at Fermi level). . However, since the value of γ is about an order of magnitude smaller than other transition metal oxides that exhibit metallic conduction, N (0) is estimated to be small. The Fermi level is presumed to have crossed slightly near the upper end of the band, presumably with a Co t _2g orbit.
The measured value at low temperature is off the straight line and increases with decreasing temperature. As shown in the inset, this can be explained by Schottky specific heat. Schottky specific heat is a decrease often seen in paramagnetic materials, and means that discrete levels due to isolated spins exist near the Fermi level. This result is consistent with the assumption that a localized level exists.

When plotting the value (C _p / T) obtained by dividing the specific heat by temperature against the square of temperature (T ² ) as shown in FIG. 7, the intersection of the extrapolated line and the vertical axis is the electronic specific heat coefficient. Give g. Further, the Debye temperature (θ _D ) can be obtained from the slope of the straight line. γ is proportional to the density of electronic states at the Fermi level. The values of γ and θ _D obtained are γ = 4.48 (7) mJ / Co-mol K ² and θ _D = 621 (1) K. Although the value of γ is about 1 smaller than an oxide showing metallic conductivity, it is not zero. Since a finite g value is observed, it can be seen that the density of electronic states at the Fermi level of this system exists finitely, that is, in a metallic state. These are consistent with the measurement results of electrical resistance and Seebeck coefficient.

    At low temperatures below 6K, the specific heat increases with decreasing temperature. This can be understood as an effect of a Schottky abnormality. The inset of FIG. 7 shows the temperature dependence of the specific heat obtained by data analysis using an equation incorporating Schottky specific heat correction. The good fit between the analytical curve and the measured data suggests that Schottky specific heat is involved in this system. In this system, Schottky specific heat is caused by splitting of levels caused by isolated spins or spins of isolated localized holes. This is consistent with the measurement results of magnetic susceptibility and electrical resistance.

FIG. 8 shows a CaCo ₂ O ₄ polycrystalline sample obtained by molding a part of the fragments from pellets obtained after synthesis under high temperature and high pressure.
The sample mass immediately after taking out from the platinum capsule is on a pellet having a diameter of about 6 mm and a height of about 3-4 mm. The sample in the photograph is a part of the pellet cut out and formed into a rectangle. The black color peculiar to electroconductive oxide.

<How to improve thermoelectric power>
As described above, the performance index Z = S ² / ρκ or S ² / ρ is used to represent the thermoelectric performance. This time, κ is not measured (usually difficult to measure), so S ² / ρ is used as an index. Compared to typical layered cobalt oxide NaCo ₂ O _4, S (Seebeck coefficient) of this CaCo ₂ O ₄ is about 1.5 culture ^{(S 2} are approximately 2-fold) is, [rho (electric resistivity) Is about 2-3 orders of magnitude higher.
As a result, S ² / ρ becomes about 2-3 orders of magnitude lower, and cannot be used as a thermoelectric material as it is. Since S is superior, how to lower ρ is the point of improvement. (Note:. NaCo ₂ O ₄ and CaCo ₂ is O composition ratio of ₄ are similar, crystal structure is different from a completely different phase former, have notation in Non-Patent Document 1, NaCo ₂ O ₄ is used However, recently, Na _0.7 Co ₂ O ₄ is usually used more often.)

Improvement measure 1: Single crystallization The sample used in this measurement is a polycrystalline (ceramics) sample. A polycrystalline sample is a product in which a large number of fine (usually several microns in size) single crystal grains are gathered and baked. For this reason, in electrical conduction, scattering at the grain boundary increases, and the absolute value of electrical resistivity increases. Using a single crystal eliminates grain scattering, so the electrical resistivity decreases significantly (perhaps by orders of magnitude). Actually, the small electrical resistivity of NaCo ₂ O ₄ in Non-Patent Document 1 is a value obtained using a single crystal. On the other hand, the electrical resistivity of Ca ₃ Co ₄ O ₉ in Non-Patent Document 2 is a value in the case of a polycrystalline sample, and can be compared with the current CaCo ₂ O ₄ polycrystalline sample.

Both show only a difference of resistivity of about one digit. (The resistivity of CaCo ₂ O _{4 is} an order of magnitude higher than that of Ca ₃ Co ₄ O ₉ ) (NaCo ₂ O ₄ and Ca ₃ Co ₄ O ₉ are both layered cobalt oxides and have similar crystal structures. .) A difference in resistivity of about an order of magnitude is a category that can be corrected by improvement measure 2. In particular, since CaCo ₂ O ₄ should have conductivity anisotropy (not as strong as a one-dimensional system) as can be inferred from the crystal structure, the electrical resistivity in the b-axis direction is higher than that in other directions. Should be orders of magnitude smaller. If a single crystal material is used in an appropriate direction, a great improvement in thermoelectric performance can be expected.

Improvement Measure 2: Carrier Doped CaCo ₂ O ₄ is presumed to be caused by a high thermal electromotive force due to the degeneration of the t _2g electron orbit of the cobalt atom and the electron correlation. In any case, there is no doubt that the double-stranded network formed by the CoO ₆ octahedron is structurally important. If carriers (holes) are injected while holding this structural portion, it can be expected that the electrical resistance will decrease while maintaining a high thermoelectromotive force. Specifically, if Ca ²⁺ ions are partially replaced with monovalent alkali metals such as Na ⁺ , K ⁺ , and Rb ⁺ (probably around 10%), holes are injected and the resistance decreases by about 1-2 digits. There is expected. At this time, there is a possibility that the thermoelectromotive force is also lowered slightly, but since the contraction of the Co t _2g orbit and the residual entropy caused thereby are maintained, it is expected that the thermoelectromotive force is suppressed to a slight decrease.

From the viewpoint of synthesis, it can be assumed that Na, K, Rb substitution at the Ca site is sufficiently possible. The ionic radius of Na is close to that of Ca and is easily dissolved. Although the ionic radius of K and Rb is slightly larger, basically, large ions shrink a lot under high pressure conditions, which is convenient for high pressure synthesis. As a result, it is estimated that K, Rb, etc. are easily dissolved. Since the same structure (CaFe ₂ O ₄ -type structure) in Nati ₂ O ₄ that substance is also present, Na substitution is not unnatural even crystallographically, but rather promising. (Note: Many CaFe ₂ O ₄ type structural materials are known to date. AB ₂ O ₄ (A = Li, Na, Mg, Ca, Sr, Ba, La, and Eu; B = Ti, V , Cr, Mn, Fe, Ru, Rh, Al, Ga, In, Tl, Sc, Y, La, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Yb, and Lu) B = Co. This is the first time.)

Improvement measures 1 and 2 are expected to improve the electrical resistivity by 2-3 orders of magnitude, and can be expected to be about the same (or more) S ² / ρ as NaCo ₂ O ₄ . Moreover, as seen in the measurement data of the temperature dependence of the Seebeck coefficient S, it seems that S is still increasing even if it exceeds 380 K. (In ordinary layered cobalt oxides, S tends to saturate at higher temperatures near room temperature.) If this increase continues, the thermoelectric performance of conventional layered cobalt oxides surpasses that of conventional layered cobalt oxides. Can be expected to do. The double-stranded network structure of this system may be a convenient structure for generating a high thermoelectromotive force in a high temperature region. This measurement data is sufficient to suggest that, and this is probably the first time for us.

As described above, many CaFe ₂ O ₄ type structural materials have been reported so far, but very few have reported the Seebeck coefficient and have not been the subject of comparison this time. The reason is considered that there were few such viewpoints in the past and the Seebeck coefficient could not be easily measured.
At present, S is reported in substances such as NaRh ₂ O ₄ , but the value of S is small.
(K. Yamaura, Q. Qingzhen Huang, M. Moldovan, D. P. Young, A. Sato, Y. Baba, T. Nagai, Y. Matsui, and E. TakamaM. 359.)

The thermoelectric material has a so-called electronic conversion mechanism that does not use a compressor or the like, and thus has an advantage of being noiseless and capable of being miniaturized.
In recent years, the demand for Peltier elements for cooling CPUs (central processing elements) and infrared detectors of computers has increased. In the future, the amount of heat generated will increase as the CPU processing speed increases, so the development of high-performance thermoelectric elements, including the development of new thermoelectric materials, is considered urgent. Furthermore, the development of electronic materials with high thermoelectric efficiency includes the realization of high-efficiency thermoelectric cooling refrigerators that do not require a compressor, conversion of automobile exhaust gas and engine waste heat by thermoelectric conversion, or waste heat generated from plants into electric power. The possibility of new applications will be expanded, and it will be a great contribution to the development of the entire thermoelectric technology.
In addition, energy conversion using thermoelectric materials is environmentally friendly. Refrigerators, coolers, etc. are cooled using chlorofluorocarbon, which has a high environmental impact. Naturally, if a thermoelectric material is used, a CFC-free cooler can be realized.
Thus, thermoelectric conversion materials are considered to be extremely important in considering energy and environmental problems in the 21st century.

CaCo ₂ O ₄ powder X-ray pattern (Rietveld analysis pattern) Schematic view showing the crystal structure of CaCo ₂ O _4. The schematic diagram which shows the double chain network of CoO ₆ octahedron. A region surrounded by a straight line indicates a unit cell (orthorhombic system). Temperature dependence of magnetic susceptibility (χ) in CaCo ₂ O ₄ . FC: Measurement data under cooling in a magnetic field. ZFC: Measurement data obtained by heating in a magnetic field after cooling without a magnetic field. The solid line is the Curie-Weiss approximate curve; χ = χ ₀ + C / (T−θ _W ): χ ₀ = 2.2 (1) × 10 ⁻⁴ emu / Co-mol, C = 0.0471 (6) emuK / Co-mol, and θ _W = −1.3 (1) K. Temperature dependence of electrical resistivity (r) in CaCo ₂ O ₄ polycrystal. The inset shows the Arrhenius plot (lnρ vs. 1 / T) and the Mott variable range hopping plot (lnρ vs. 1 / T ^0.25 ). Temperature dependence of thermoelectric power (Seebeck coefficient: S) in CaCo ₂ O ₄ . The inset shows a log-log plot (lnS vs. lnT). Temperature dependence (C _p / T vs. T ² ) of constant pressure specific heat (C _p ) in CaCo ₂ O ₄ . From the intersection of the extrapolated line and the vertical axis, the electronic specific heat coefficient g is obtained as 4.48 (7) mJ / mol K ² . The inset shows the results of analysis by Schottky specific heat approximation. A photograph of a CaCo ₂ O ₄ polycrystalline sample. What was obtained by molding a part of the fragments from pellets obtained after synthesis under high temperature and pressure.

Claims (1)

A thermoelectric conversion material comprising a cobalt oxide, having a CaFe ₂ O ₄ (calcium / ferrite) structure, wherein Fe is substituted with Co.