WO2013171260A1

WO2013171260A1 - Thermoelectric material and a method of manufacturing such a thermoelectric material

Info

Publication number: WO2013171260A1
Application number: PCT/EP2013/060030
Authority: WO
Inventors: Gregori GIULIANO; Stefan HEIZE; Piero LUPETIN; Hanns-Ulrich Habermeier; Joachim Maier
Original assignee: MAX-PLANCK-Gesellschaft zur Förderung der Wissenschaften e.V.
Priority date: 2012-05-15
Filing date: 2013-05-15
Publication date: 2013-11-21

Abstract

A thermoelectric material comprises a pressed body of adjacent nanocrystalline grains of at least one complex oxide of a metal such as a transition metal or alkali earth metal having an average grain size less than or equal to 100nm and preferably greater than 10nm. The invention is based on the finding that there is a strong change of the conduction properties of the nanocrystalline sample compared with the undisturbed bulk properties, namely a reduction of the p-type conductivity by two orders of magnitude at high oxygen partial pressure. Similarly, the Seebeck coefficient values of the nanocrystalline sample exhibit remarkable deviations from the single crystal ones: Under oxidizing conditions, values up to 2200 μV/K (at 575°C) are detected. In the nanocrystalline sample the dependence of the Seebeck coefficient on the concentration of the charge carriers is found to be four times larger than in the single crystal.

Description

Thermoelectric Material And A Method Of Manufacturing Such a

Thermoelectric Material.

The present invention relates to thermoelectric materials having a high Seebeck coefficient and the potential for high figures of merit and a method of manufacturing such thermoelectric materials. Thermoelectric materials are known for various applications. For example it is known to generate a voltage at the ends of a material exposed to a temperature gradient (the Seebeck effect) . Also it is known to pass an electrical current through a thermoelectric element to generate a temperature gradient (the Peltier effect) . For example coolers for wine or for the steering wheels of vehicles are known which make use of Peltier cooling.

There is significant interest in thermoelectric materials as power generators and/or as cooling devices because the oxide based devices can operate reliably at high temperatures and at low temperatures. There are many sources of waste heat at temperatures in the range from 100°C to 1000°C they are suitable for generating electrical power by the use of thermoelectric materials. Also there is a need to provide electrically operated cooling devices without moving parts for cooling materials down to cryogenic temperatures. For example, rather than using photovoltaic solar energy sys- terns which have a relatively low efficiency it would be possible to use focusing solar energy systems, such as parabolic mirror arrangements.

Such designs concentrate solar energy and generate temperatures of several hundred degrees at the focal point, which if a suitable thermoelectric material can be found, could be a very attractive source of electrical ener- gy. Lightweight parabolic mirrors for solar energy are for example described in European Patent 1290383.

There is therefore great interest in devising thermoelectric materials with significantly higher efficiency than is currently available.

In recent years, oxides for thermopower applications have received much attention thanks to (i) their high mechanical, electrical and thermal stability in reducing as well as oxidizing environments and (ii) their lower toxici- ty compared with conventional thermoelectric materials e.g. chalcogenides and skutterudites. An example of such an oxide material is disclosed in US patent 6376763 which discloses an example of a complex oxide material having a Seebeck coefficient of ΙΟΟμΚ. Investigations have, for example, also been carried out on SrTiO3. The main goal of recent studies has been the improvement of the figure of merit ΖΓ, which is defined by the following equation:

where a is the Seebeck coefficient, σ the electrical conductivity, T the temperature and κ the thermal conductivity.

The best results so far have been achieved through heavily donor doping the material (using lanthanum or niobium) with the aim of increasing the charge carrier concentration c and thus the electrical conductivity σ. This procedure usually has a detrimental impact on the Seebeck coefficient a which, however, is rather limited since a varies only logarithmically with c, Another possibility to influence ZT consists in reducing the thermal conductivity.

The principal objective underlying the present invention is to provide thermoelectric materials having a substantially improved figure of merit, which can be produced at favorable cost, which have excellent high temperature properties and applications, either for energy generation or cooling, which can be used not only under reducing conditions but also in a oxidizing environment (i.e. air) and which are non-toxic and thus envi- ronmentally friendly.

It is a further objective of the present invention to provide a method of manufacturing thermoelectric materials having the above recited properties and which can be generally applied to a wide range of different mate- rials.

In order to satisfy these objectives there is provided a thermoelectric material comprising a pressed body of adjacent nanocrystalline grains of at least one complex oxide of a metal, such as a transition metal or an alkali earth metal, having an average grain size less than or equal to lOOnm and preferably greater than lOnm.

Also according to the invention there is provided a thermoelectric material comprising an agglomeration of a plurality of adjacent nanocrystalline grains of at least one metal oxide having an average grain size less than or equal to lOOnm, the grains being separated by grain boundaries.

Moreover, the invention also comprises a method of manufacturing a thermoelectric material comprising the steps of taking a nanocrystalline powder of grains of at least one complex oxide of a metal, such as a transi- tion metal or an alkali earth metal having an average grain size less than or equal to lOOnm and preferably greater than lOnm and pressing the powder into a compact. Alternatively a method of manufacturing a thermoelectric material in accordance with the invention can comprise the steps of taking a nanocrys- talline powder of grains of at least one oxide of a metal having an average grain size less than or equal to lOOnm and pressing the powder into a compact, the grains being separated in the compact by grain boundaries.

The invention is based on the recognition that boundaries (grain boundaries in ceramics or interfaces in thin films) can have very significant effects on the electrical conduction properties of oxides. Particularly intriguing is the case of mixed (electronic and ionic) conductors (such as CeO2 and SrTiOs), in which the charge carrier concentrations undergo a severe redistribution in proximity of the grain boundaries (space charge zones) as a consequence of the electrostatic potential Α arising from the boundary core. Formally, the deviation from the bulk (grain interior) concentration can be written as

^l = e^~^ , (2)

where c_y(x) is the concentration of the charge carrier as a function of the distance from the boundary x, c ._∞ the concentration in the bulk (∞) , Zj the charge number of the charge carrier j, e the elementary charge and JcT the Boltzmann term. The modification of the charge carrier concentration profiles can induce striking effects such as (i) switching of the local conduction properties from ionic to electronic even in Ceo. gGdo.iO 1.95 at low temperature (280°C) and oxidizing conditions, and (ii) in the case of mesoscopic nanocrystalline SiTiOe shifting of the transition from n-type to p-type conductivity by 12 orders of magnitude in terms of oxygen partial pressure (PO2), from reducing towards oxidizing conditions.

Although such boundary effects have been extensively studied in the framework of the electrical conduction properties of oxides, much less is known about their role in other related phenomena such as the thermoelectric effect. It is worth noting nonetheless, despite the efforts aimed at improving ΖΓ by modifying σ and rc, the improvements achieved in the figure of merit are limited. No attention has been paid to the possibility of adjusting the value of a other than by varying the dopant concentration.

As noted above this is, however, counterproductive. Instead of this approach the invention takes a new direction and considers how the grain boundaries can be exploited to improve the Seebeck coefficient. So far, the only relevant result in this context has been provided by Ohta et al, who, upon the investigation of a series of multilayered heterostructures, observed very large values of ΖΓ for a 2 dimensional electronic gas at interfaces. In contrast to this the present invention exploits the effects of grain boundaries on the electrical conductivity as well as the Seebeck coefficient. To date the experimental work has been carried out by comparison of the values which can be obtained with three different internally pure forms of SrTiO3 samples including single crystals, microcrystalline and nanocrystalline ceramics. In order to clearly identify the contribution of the grain boundaries, it is crucial for the nanocrystalline grains to have an average size below about 100 nm. With such a grain size (average size below about lOOnm, typically with a normal distribution centered around below lOOnm) the concentration of the main charge carriers is exclusively controlled by boundary effects (space charge), i.e. no unperturbated bulk material is present. It should be noted that the charge carrier concentration in such un-doped samples can be easily varied by simply changing the oxygen partial pressure (pO2) for example between 1 bar and 5- 10 ⁵ bar in use of the material. Practical applications can be for power genera- tion in air.

In an embodiment of the thermoelectric material the oxide or complex oxide has electronic and ionic conductivity. Such a material can ensure a high figure of merit. Typical figures of merit which have been achieved in oxides previously do not exceed 0.5. This is due to a low conductivity and a relatively low Seebeck coefficient (~ 100-200 uV/K see e.g. the aforementioned US Patent 6,376,763). With the invention described here values of a of up to and greater than 2.2 mV/K are obtained. The order of magnitude greater values of the Seebeck coefficient arise due to the use of nanocrystalline oxide material rather than bulk crystal material. This is already a very good value and it is expected that many of the potential materials will enable values significantly higher than this to be reached. Nanocrystalline materials with good ionic and electronic conductivity are not only able to conduct electricity efficiently, but can also reduce thermal conductivity (grain boundary scattering of phonons) making them particularly benefi- cial so far as the figure of merit is concerned.

In a preferred embodiment the oxide or complex oxide is at least one of a titanate, vanadate, cuprate, manganite or similar of a metal. Such compounds can readily be made are generally non-toxic, have a low thermal conductivity and are mechanically, electrically and thermally stable. In yet a further embodiment of the thermoelectric material, the metal is selected from the alkali and earth alkali groups, as well as from the group of light rare earth elements and heavy rare earth elements and from the group of transition metal elements. The metal preferably being selected from the group comprising Li, Be, Na, Mg, K, Ca, Sc, Rb, Sr, Y, Cs, Ba, La, Ce, Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb and Lu. These metals are again readily available and non-toxic. The corresponding oxides are mechanically, electrically and thermally stable.

In a preferred embodiment of the thermoelectric material the grains of nanocrystalline material comprise a core of a first oxide or complex oxide surrounded by at least one shell of another oxide or complex oxide. Such core/ shell structures can be readily fabricated and the shell can be doped to provide n-type or p-type conductivity. Providing a doped shell structure can improve the conductivity across the grain boundaries and thus the conductivity of the material as a whole.

In a further preferred embodiment of the thermoelectric material the grains of metal oxide or complex metal oxide are sintered together, e.g. in the form of a compact or pressed body. Such a sintering further improves the conductivity across the grain boundaries. Although good conductivity can already be obtained by simple compaction without sintering. The sintering under compaction preferably takes place by the so-called SPS (spark plasma sintering) method. Other powder compaction methods known in the art can also be used, such methods include hot isostatic pressing and hot uniaxial pressing.

By fabricating the compact from powder materials a certain degree of po- rosity is inevitable and this favors the uniform oxygen partial pressure throughout the structure which is beneficial to obtaining a high uniform conductivity through the compact.

When used in a thermoelectric device the material of the invention comprises at least first and second portions of the thermoelectric material described above, the first portion having n-type conductivity and the second portion having p-type conductivity whereby to generate one or more n-p or p-n junctions, optionally with a portion of intrinsic material between the n-type and p-type portions. This embodiment recognizes that in a practical device n- and p-legs have to be connected sequentially.

Preferred embodiments of the method are defined in the dependent method claims and the following specific description.

The invention as well as the theoretical background will now be described in more detail with reference to examples and the accompanying drawings in which:

Fig la shows an XRD pattern acquired from a nanocrystalline

sample, no second phase can be recognized;

Fig. lb shows an SEM micrograph taken from the fracture surface of a nanocrystalline sample, with the average grain size L being equal to about 80 nm;

Fig. 2a - c show the characteristic impedance spectra acquired at ρθ2 = 1 bar from the single crystal sample (Fig. 2a), the microcrystal- line sample (Fig. 2b) and the nanocrystalline sample (Fig. 2c); Fig. 3 shows the ρθ2 dependence of the electrical conductivity of the single crystal as well as of the nanocrystalline;

Fig. 4a shows values of the Seebeck coefficient a measured at

575°C as a function of ρθ2 from the 3 different kind of samples considered in this study;

Fig. 4b shows a comparison between the absolute values of a collected over three different sets of SrTiO3 single crystals;

Fig. 4c shows values of the Seebeck coefficient a shown in Fig. 4a plotted as a function of the electron hole concentration calculated from the conductivity data (see Fig. 3) under the assumption of a flat concentration profile and established mobility values; and

Fig. 5 schematically shows a thermoelectric module composed of thermoelectric materials according the present invention. First of all a consideration of the theoretical background will be given. For the derivation of the Seebeck coefficient reference can be made to previous studies, according to which the following expression for , (the Seebeck coefficient associated to the charge carrier J) can be obtained:

J, = ( Vin C J: + constant) (3b) where S° is the standard partial molar entropy of the charge carrier , and Q_j the corresponding heat of transport: For electrons and electron holes these quantities can be neglected. It is worth noting that this expression is equivalent to the extensively used Heikes formula.

Consideration will now be given to the bulk defect chemistry of SrTiO3 in an oxidizing environment, which corresponds to the experimental conditions used here. In this case, the oxygen incorporation reaction (eqn. (4)), the band-band transfer (eqn. (5)) together with the condition of electro- neutrality (eqn. (6)) at each temperature and pO2 determine the concentration of the different charge carriers (n electrons, p holes and υ oxygen vacancies) .

V₀-+l<¾ = 0* + 2h-, K₀=— (4) nil = e'+h^', K_B = n-p (5)

2v+ p=m (6)

In the case of the electron holes p, one obtains from eqn. (3b) that

3(α/μν K-¹)/ai_og(p/cnY³) =198.

It is noted that for high values of ρθ2, nominally pure SiTiOe is known to be a p-type conductor. Hence, by taking into consideration also eqn. (4), it is convenient to express the dependence of the Seebeck coefficient as a function of pO2:

=— ^:— (Log (p0₂ ) + constant) (7b)

The slope θ(α/ μ\/ K^~1 ) / 3Log(p0₂ / bar) is in this case equal to -50. In the following reference will be made to the experimental procedure used to demonstrate the effect of the present invention. For this experimental procedure nominally pure, (100) oriented SrTiO3 single crystals (10 x 10 x 0.5 mm³) were purchased from CrysTec GmbH. Commercial nanocrystal- line SrTiO3 powder (Inframat^® Advanced Materials™, catalogue number 3822ON-01) with nominal grain size < 100 nm was compacted via Sparka Plasma Sintering (FCT Systeme GmbH, system type HP D 5/2) using the following parameters: maximum temperature 850°C, dwell time 5 minutes, heating rate 100°C/min, maximal pressure ca. 350 MPa, argon atmosphere in order to obtain compacts or pellets. One of the nanocystal- line samples was annealed at 1250°C for 4 hrs in air in order to let the grains become coarse and to obtain a microcrystalline specimen, i.e. the crystallites are of micron size.

An X-ray diffraction (XRD) analysis of the samples was performed using a Philips Xpert XRD diffractometer (3710 HTK, Cu-K_a = 1.54056 A). A scanning electron microscope (Zeiss Merlin^®) was employed to inspect the mi- crostructure of the nanocrystalline samples.

The electrical conductivity measurements were carried out using an Al- pha-A high resolution dielectric analyzer (Novocontrol Technologies

GmbH) (ac voltage 0.3 V, frequency range from 2 MHz to 1 Hz). The temperature range chosen for the measurements ranged between 350 and 550°C, while the ρθ2 ranged between 10 ⁴ and 1 bar. The electrical con- tacts for the conductivity measurements were made by DC-sputtering a thin platinum film (400 nm) on both sides of the pellets. The analysis of the impedance spectra was performed using the commercial software Z- View 2 by Scribner Associates Inc.

For the determination of the thermoelectric properties, an Ulvac ZEM 3- M8 Seebeck coefficient measurement system was used. The commercial set-up was modified in order to allow high temperature measurements (575°C) also in oxidizing condition [ρθ2 = 10 ⁴ - 1 bar). The measurements were carried out without depositing any platinum contacts on the surface of the sample.

For both the conductivity and Seebeck coefficient measurements, each sample was kept at the given conditions (T and PO2) for enough time to reach the equilibration (ca. 16 hours). Finally, it should be noted that by knowing the activation energy associated obtained from the impedance spectroscopy analysis and the ρθ2 dependence it was possible to compare the conductivity with the Seebeck coefficient data, even if they were acquired in slightly different experimental conditions.

In the following the results of the microstructural characterization performed on the nanocrystalline sample will be discussed with reference to the Figures. Fig. la displays the measured XRD pattern, which confirms that the measured nanocrystalline sample consists only of a single phase while the SEM micrograph (Fig. lb) illustrates that the average grain size L after spark plasma sintering is about 80 nm.

Figs. 2a to 2c shows the characteristic impedance spectra acquired at PO2 = 1 bar from the single crystal sample (Fig. 2a), from the microcrystal- line sample (Fig. 2b) and from the nanocrystalline sample (Fig. 2c). The open circles correspond to the experimental data and the equivalent circuits are shown in each Figure.

The impedance spectrum collected from the single crystal at 396°C is characterized by two contributions (Fig. 2a). The first one can be equated to an RC circuit in which C behaves like an ideal capacitor with a relative permittivity ε_Γ = 145. The activation energy associated with R is equal to 0.75 eV, which is consistent with previous values obtained for the bulk in the same temperature range. The second arc, is much distorted, and has a very high capacitance value [C2/C1 = 500). It disappears in the modulus plot and is thus assigned to interfacial effects occurring at the electrodes.

For the microcrystalline sample, the spectrum in Fig. 2(b) was fitted using the equivalent circuit consisting of one resistance R in series with an RQ circuit. Q is a constant phase element from which the capacitance C = (R¹- ⁿQ)^{1 /n} can be determined, where n is an additional fitting parameter. The presence of two contributions is characteristic of polycrystalline samples: the first contribution (the intercept obtained by extrapolating the fitting line) corresponds to the bulk contribution [Rbuik) while the second contri- bution is due to the grain boundaries [RGB). This assignment is supported by the higher capacitance value of the second contribution (ε_Γ = 350 at 544°C), which is characteristic of the grain boundaries.

It is worth emphasizing that RGB stems from the blocking character of the grain boundaries. This is consistent with finding a positive space charge potential Α in un-doped material, which induces a strong depletion of the holes (as well as the oxygen vacancies) in proximity of the grain boundaries compared with the bulk (see eqn. (1)). For the nanocrystalline sample, only a single contribution can be recognized (also in the modulus plot) , which can be fitted by an equivalent circuit consisting of a resistance in parallel with a Q element. This finding together with the large activation energy associated with this single semi- circle (1.31 eV in oxygen), indicates that this single contribution arises from the grain boundaries of the nanocrystalline sample. The absence of the bulk contribution points towards the sample being in the mesoscopic regime, meaning that the overall electrical transport properties, and therefore the charge carrier concentrations, correspond to those of the grain boundaries. Such a situation occurs when 2.X > L , λ is the space charge layer width in the Mott-Schottkly approximation and L the average grain size. For oxidizing conditions the p-type conductivity will be significantly lower in a mesoscopic sample than in the single crystal. This is precisely what can be observed in Fig. 3, which depicts the pO2 dependence of the electrical conductivity of the single crystal as well as the nanocrystalline sample. The latter sample exhibits a measured electrical conductivity which is 2 orders of magnitude lower than the single crystal (bulk) . It is worth noting that the data collected from the mesoscopic sample is in excellent agreement with the experimental data acquired from a sample having an average grain size of 50 nm and prepared starting from nanocrystalline powders synthesized in the applicants laboratory. Having regard to the gradient of the pO2 dependence, the value of dLog (a_m / S cm^"1 ) / 3Log(p0₂ / bar) for the single crystal is 0.256 while for the nanocrystalline sample it is 0.228. In both cases, the positive slope of the pO2 dependence is clear evidence that in oxidizing conditions the main charge carriers are electron holes (p-type conductivity) and is in agreement with the defect chemistry model. The somewhat lower value of the ρθ2 dependence of the mesoscopic sample (10% lower compared with the single crystal) can be explained by taking into account the partial pressure dependence of the space charge layer potential as shown by eqn (8) , which is derived for the blocking boundaries.

From the electrical conductivity and according to the hole mobility data u_p available in literature, it is possible to evaluate the electron hole concentration for the single crystal and the nanocrystalline sample as well. For the latter, the same approximation can be adopted as was used previously for the mesoscopic situation, namely one can assume the hole concentration to be flat throughout the grains and equal to po, as defined in the following (see also eqn. (2)), where the subscript 0 denotes the concentration of the holes in the very first atomic layer of the grain adjacent to the boundary:

kT (9) where Αφ{χ) = Αφ₀ . In this way, the following equalities hold:

for the single crystal p =σ_∞ I eu_p (10a) and for the nanocrystalline p =σ₀ / eu_p . (10b)

It is important to note that such an approximation cannot be applied to the microcrystalline sample. The results of the Seebeck coefficient measurements performed at different ρθ2 values are summarized in Fig. 4a. It is evident that in the ρθ2 range considered here, a is positive in agreement with the electrical conductivity results, which indicate p-type conductivity. Interestingly, de- pending on ρθ2, for the single crystal the value of a ranges between 1 150 and 1350 uV-K^{- 1}, for the microcrystalline sample between 1280 and

1576 uV-K^{" 1}, wheras the nanocrystalline specimen exhibits the largest values spanning from 1550 up to 2150 uV-K^-1. Rather large absolute values of a (which however are well below the values presented here) have been recently reported for slightly La-doped SrTiO3 epitaxial films (ranging between 980 and 100 uV-K -¹) with a charge carrier concentration between 4.8· 10¹⁷ and 2- 10²¹ cm ³.

Table I below shows slope values obtained by linearly fitting the experi- mental data shown in Fig. 4a and c.

Notably, Fig. 4 indicates that (i) a increases with decreasing ρθ2 as expected from eqn. (7b) and, more importantly, (ii) a increases with increasing volume fraction of the grain boundaries. As summarized in Table I, the slope for the single crystal investigated here (-45.6) is in excellent agreement with a theoretical value which can be predicted by eqn. (7b).

The microcrystalline sample a exhibits a ρθ2 dependence which is 20% higher than in the single crystal, for the nanocrystalline sample the slope reaches a value of - 186.2 which is 4 times larger than in the single crystal. As in the mesoscopic case the concentration of holes is determined by the space charge potential through eqn. (2) , one should consider also the ρθ2 dependence of Α (see eqn. (8)) . However, as shown in Fig. 3 such a ρθ2 dependence can account for a variation of the slope of approximately 10% and not for a factor of 4.

The hole concentration for the nanocrystalline SrTiO3 and the single crystal can now be determined from the conductivity data and according to eqns. ( 10a) and ( 10b) . It is instructive to compare the single crystal data collected in this work with previous data published in the literature. Fig. 4b depicts the absolute values of the Seebeck coefficient acquired from three different sets of samples: (i) the single crystals investigated by Fred- eriksen et al. , (ii) the La-doped single crystals studied by Okuda et al. and (iii) the crystals presently under investigation. It is important to stress here that the values of

are plotted as a function of the concentration c of the charge carriers, irrespective of whether these are electrons or holes. It is apparent that all these data are almost perfectly aligned, with a slope equal to -230 ± 3, slightly higher than the theoretical value predicted by eqn. (2b) (- 198) .

A consideration will now be given of the Seebeck coefficient values vs. the hole concentration p for the single crystal and the mesoscopic sample (Fig. 4c) . Firstly, it is worth noting that in this case the slope of the data collected from single crystal corresponds well with the theoretical value of - 198 (cf. Table I) .

Secondly, compared with Fig. 4a, the data regarding the nanocrystalline samples are obviously shifted towards the left hand side of the diagram as the hole concentration in this sample is significantly lower than in the bulk. There are two striking aspects that emerge from this plot: (i) the mesoscopic sample exhibits much larger values of a and (ii) as shown in Table I, the slope obtained from this sample is 4 times larger (-826) than the single crystal (consistent with Fig. 4a).

The key aspect that one should bear in mind here is that the nanocrystalline sample is in the mesoscopic situation and that, therefore, the corresponding Seebeck coefficient is solely due the space charge layer effects at the boundaries. This means that because of this sample one can now di- rectly access (on the basis of reasonable approximations, such as e.g. eqns. (9) and (10b)) the grain boundary properties with regard to a.

With this in mind, we can now further analyze the experimental results shown in Fig. 4c. Clearly, if one moves towards lower values of Log(p) one would expect higher values of a. Nonetheless, it is evident that by extrapolating the line with triangles as data points used to fit the single crystal data one would never reach the large value of a shown by the mesoscopic sample. The main reason for this has to be sought in the change of slope, which is, in our opinion, an indication that the dependence of a on the charge carrier concentration changes drastically in proximity to the grain boundaries compared with the situation of unperturbated bulk.

In the foregoing the electrical conduction properties and the Seebeck coefficient of a series of nominally pure SrTiO3 samples (single crystal, micro- crystalline and nanocrystalline) at temperatures ranging between 350 and 575°C and an oxygen partial pressure values between 5- 10 ⁵ and 1 bar were discussed.

All SrTiO3 specimens displayed p-type conductivity under the experi- mental conditions considered. The analysis of the impedance spectra al- lows the conclusion that the nanocrystalline sample was in the mesoscop- ic situation meaning that the charge carrier concentrations were determined only by the space charge effects. This permits a direct estimate of the Seebeck coefficient of the grain boundaries of a nanocrystalline sam- pie, the first time this has been possible. Remarkably, nanocrystalline SrTiO3 exhibited extremely large Seebeck coefficient values (up to 2200 uV-K ¹ at ρθ2 = 10 ⁴ bar) and a dependence of the Seebeck coefficient on the charge carrier concentration which was 4 times larger than in the single crystal.

These results indicate that the boundaries can have a considerable impact on the thermoelectric properties of polycrystalline ceramics. The underlying physics and our experiences to date show the effects described above with regard to SrTiO3 can be extended to a whole range of oxides and complex oxides of metals such as transition metals and alkali earth metals. In particular titanates, vanadates, cuprates and manganites of the named types of metals are expected to show significantly enhanced figures of merit of thermal electric oxides when present in compacts with nanocrystalline grains at sizes less than lOOnm. It is difficult to be specific about the lower limit of the size range which is useful, however, this is expected to be a size at which grain boundary effects no longer show the space charge characteristics useful for the present invention. This lower limit is typically assumed to be around lOnm. Finally it should be pointed out that for practical thermoelectric devices it is necessary to dope the compact so as to produce at least one n-leg and one p-leg. Such doping can readily be achieved using known dopants such as Ce for SrTiO3. Similar dopant materials are Al, Fe, Nb and La. In particular it is expected that benefit can be obtained by coating grains of nanocyrstalline grains in the desired size range with one or more shells to form shell/ core structures. The shells can be formed of doped material, whereas the cores can be intrinsic material. These can be made by making separate compacts of the individual n-type, p-type and/or i-type (intrinsic) material and placing them in contact with one another. Alternatively regions of p, n and/ or i type material grains can be placed alongside one another and simultaneously pressed and/or sintered into one compact. Fig. 5 shows a thermoelectric module 10 composed of n-type structures 12 and p-type structures 14 made of oxide materials according to the present invention. The thermoelectric module 10 can, for example, be used to convert thermal energy into electrical energy. In order to extract /provide electricity from/to the thermoelectric module 10 this is connected to an elec- trie storage device /electric power supply (not shown) via wires 16 at the base plate (Heat Sink) 20 having cooling ribs 18. The other base plate 22 is provided in the vicinity of the heat source in order to absorb heat at that side 22 to drive the electrical energy conversion. To connect the n-type structures 12 and the p-type structures 14 with one another electrodes 24 are provided therebetween. When used as a thermoelectric generator the heat source is placed in the vicinity of the base plate 22, so that heat can pass through thermoelectric generator in order to generate electricity at the heat sink side. In order to manufacture a cooling device electricity is provided at the heat sink side rather than being extracted. This causes heat to flow from the heat sink side 20 to the other side 22. The heat at the heat sink 20 provides cooling and the heat at the other side 22 can either be dissipated to the environment via cooling fins or extracted via a heat exchanger. The following references give scientific background information on the field of the invention, do not however disclose the present invention or the concept underlying the present invention: Chiang Y-M, Lavik EB, Kosacki I, Tuller HL, Ying JY (1996) Appl Phys Lett 69: 185

Tschope A, Birringer R (2001) J Electroceram 7: 169

Kim S, Maier J (2001) J Electoceram 7: 169

Balaya P, Jamnik J, Fleig J, Maier J (2006) Appl Phys Lett 88:062109 Guo X, Sigle W, Maier J (2003) J Am Ceram Soc 86:77

Vollman M, Waser R (1994) J Am Ceram Soc 77:235

Denk I, Claus J, Maier J, (1997) J Electrochem Soc 144:3526

Waser R (1995) Solid State Ionics 75:89

De Souza RA (2009) Phys Chem Chem Phys 1 1 :9939

Maier J ( 1995) Prog Solid State Chem 23: 171

Gobel MC, Gregori G, Maier J (201 1) Phys Chem Chem Phys 13; 10940

Lupetin P, Gregori G, Maier J (201 1) Angew Chem Int Ed 49: 10123

Ohta H (2007) Materials Today 10:44

Fergus JW (2012) J Eur Ceram Soc 32:525

Okuda T, Nakanishi K, Miyasaka, Tokura Y (2001) Phys Rev B 63: 1 13104

Ohta S, Nomura T, Ohta H, Koumoto K (2005) J Appl Phys 97:0341061

Ohta S, Nomura T, Ohta H, Hirano M, Hosono H, et al. (2005) Appl Phys

Lett 87:092108

Ohta S, Ohta H, Koumoto K (2006) J Ceram Soc Jpn 1 14: 102

Ohta H, Kim S, Mune Y, Mizoguchi T, Nomura K, Ohta S, Nomura T, Nakanishi Y, Ikuhara Y, Hirano M, Hosono H, Koumoto K (2007) Nature Materials 6: 129

Koumoto K, Wang Y, Zhang R, Kosuga A, Funahashi R (2010) Annual Rev Mater Res 40:363 Heikes RR, Ure RW (1961) Thermoelectricity: Science and Engineering. Interscience Publishers, New York- London

Johnson VA, Lark-Horovitz K (1953) Phys Rev 92:226

Becker JH, Frederikse HPR (1962) J Appl Phys 33:447

Jonker GH (1968) Philips Res Repts 23: 131

Vennekamp M, Janek J (1999) Solid State Ionics 1 18:43

Tschope A, Kilassonia S, Zapp B, Birringer R (2002) Solid State Ionics

149:261

Bak T, Nowotny J, Rekas M, Sorrell CC (2004) Ionics 10: 156

Denk I, Munch W, Maier J (1995) J Am Ceram Soc 78:3265

Lupetin P, (2012) PhD thesis, University of Stuttgart;

Lupetin P, Gregori G, Maier J (2012) manuscript in preparation;

Jalan B, Stemmer S (2010) Appl Phys Lett 97:042106

Frederikse HPR, Thurber WR, Hosier WR (1964) Phys Rev 134:A44

Claims

Patent Claims

1. A thermoelectric material comprising a pressed body of adjacent nanocrystalline grains of at least one complex oxide of a metal, such as a transition metal or an alkali earth metal, having an average grain size less than or equal to lOOnm and preferably greater than lOnm.

2. A thermoelectric material comprising an agglomeration of a plurality of adjacent nanocrystalline grains of at least one metal oxide having an average grain size less than or equal to lOOnm, the grains being separated by grain boundaries.

3. A thermoelectric material in accordance with either of the preceding claims wherein the oxide or complex oxide has electronic and ionic conductivity.

4. A thermoelectric material in accordance with claim 1 or claim 2, wherein the oxide or complex oxide is at least one of a titanate, vanadate, cuprate or manganite of a metal.

5. A thermoelectric material in accordance with any one of the preceding claims, wherein the metal is selected from the group of transition metals, the group of elements comprising the alkali elements, the earth alkali elements, including light rare earth elements and heavy rare earth elements, the metal in particular being selected from the group comprising Li, Be, Na, Mg, K, Ca, Sc, Rb, Sr, Y, Cs, Ba, La, Ce, Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb and Lu.

A thermoelectric material in accordance with any one of the preceding claims wherein the grains of nanocrystalline material comprise a core of a first oxide or complex oxide surrounded by at least one shell of another oxide or complex oxide.

A thermoelectric material in accordance with any one of the preceding claims wherein the grains of metal oxide or complex metal oxide are sintered together, e.g. in the form of a compact or pressed body, and/or wherein the material is doped to generate n-type and/or p- type conductivity and/or wherein the properties are dominated by the grain boundaries.

A thermoelectric device comprising at least first and second portions of thermoelectric material in accordance with one of the preceding claims, the first portion having n-type conductivity and the second portion having p-type conductivity.

A method of manufacturing a thermoelectric material comprising the steps of taking a nanocrystalline powder of grains of at least one complex oxide of a metal, such as a transition metal or an alkali earth metal, having an average grain size less than or equal to lOOnm and preferably greater than lOnm and pressing the powder into a compact.

A method of manufacturing a thermoelectric material comprising the steps of taking a nanocrystalline powder of grains of at least one ox- ide of a metal having an average grain size less than or equal to lOOnm and pressing the powder into a compact, the grains being separated in the compact by grain boundaries.

1 1. A method in accordance with claim 9 or claim 10 wherein a complex oxide or oxide is selected having electronic and ionic conductivity

12. A method in accordance with any one of the claims 9 to 1 1 , wherein at least one of a titanate, vanadate, cuprate or manganite of a metal is selected as the oxide or complex oxide.

13. A method in accordance with any one of the preceding claims 9 to 12, wherein the metal is selected from an element of the alkali groups, earth alkali groups, as well as light rare earth elements or heavy rare earth elements, the metal in particular being selected from the group comprising Li, Be, Na, Mg, K, Ca, Sc, Rb, Sr, Y, Cs, Ba, La, Ce, Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb and Lu.

14. A method in accordance with any one of the preceding claims 9 to 13, wherein the grains of nanocrystalline material comprise a core of a first oxide or complex oxide surrounded by at least one shell of another oxide or complex oxide

15. A method in accordance with any one of the preceding claims 9 to 14, wherein the grains of metal oxide or complex metal oxide are sintered together, e.g. in the form of a compact or pressed body with conducting paths extending across the grain boundaries between adjacent grains and/or wherein the material is doped to generate n- type and/or p-type conductivity.