Thermoelectric modules and performance
The Seebeck and Peltier effects are the basis for many modern thermoelectric refrigeration and thermoelectric power generation devices, respectively. The versatility of thermoelectric materials is illustrated in Fig. I-4, which shows a diagram of a thermoelectric couple composed of an n-type (negative thermopower i.e. electron carriers) and a p-type (positive thermopower i.e. hole carriers) semiconducting materials connected through metallic electrical contact pads. Both refrigeration and power generation may be accomplished using the same module as shown in Fig. I-4. A thermoelectric module or device is built up of an array of these couples, which are arranged electrically in series and thermally in parallel.
Thermoelectric energy conversion utilizes the heat generated (as a result of the Peltier effect) when an electrical current is passed through a thermoelectric material to yield a temperature difference with heat being absorbed on the cold side and rejected at the sink, thus providing a refrigeration or heat pump capability (Fig. I-4a). Similarly, an imposed temperature difference applied on the material will result in a voltage (as a result of the Seebeck effect) or current, that is, power generation on a small scale (Fig. I-4b).
SURVEY OF STATE-OF-THE ART MATERIALS
A great number of materials have been examined in the last century for thermoelectric applications. Bismuth-antimony alloys, bismuth and lead telluride families, together with silicon-germanium alloys form the main part of the most established materials. Each of them operates over a particular range of temperature as illustrated in Fig. I-7. Most of the above-mentioned in §I-3 desirable features are common to all of established thermoelectric materials and the traditional semiconducting theory can generally be used to describe the transport properties.
Preparation is usually carried out by growth of single crystalline samples or by powder metallurgy techniques. Single crystals are preferable when performance is the crucial criterion, especially when the transport properties show a significant degree of anisotropy. On the other hand, mechanical strength is usually enhanced in polycrystalline materials and one also may expect lower lattice thermal conductivity values due to phonon scattering at grain boundaries. In addition, powder metallurgy is attractive in that it facilitates device fabrication and can be easily scaled up.Figure I-7: Schematic representation of the typical temperature dependence of the dimensionless figure of merit ZT as a function of temperature for several n- and p- type established materials.
The figure of merit, Zandcan be either determined by measuring individually the three transport coefficients ( ) on the same sample (see chapter IV) or estimated directly using the Harman method /GOL86, ROW95/. In the next paragraphs, a summary of the main properties of conventional thermoelectric materials is reported. For a comprehensive discussion, reviews are available in the literature /GOL86, RAV70, ROW83, ROW95, TRI01, URE61 and references therein/.
Below room temperature, interesting results were observed near liquid nitrogen temperature in n-type Bi-rich Bi1-xSbx alloys properly orientated, with ZT ~ 0.6 /LEN96, LEN98/. However, the use of Bi1-xSbx alloys in thermoelectric cooling devices has been constrained both by the absence of a good p-type material with a figure of merit comparable to that of the n-type below 150 K and by the relative brittleness of the material when used in its optimal configuration. The former can be solved by using high-Tc superconductors as passive branches in conjunction with n-type Bi-Sb thermoelements /GOL88/, while the latter can be overcome by employing polycrystalline samples produced by sintering, hot-pressing or extruding powdered materials. For instance, Martin-Lopez et al. have shown that the mechanical strength of an extruded polycrystalline Bi0.85Sb0.15 alloy is enhanced by one order of magnitude at 77 K as compared to a single crystal grown by the Czochralski method /MAR98/. However, the thermoelectric performance of the disordered structures is affected as a result of the random orientation of the grains and the presence of defects that strongly influence the transport at low temperatures.
The performance of Bi-Sb single crystals is even more important in the presence of a transverse magnetic field, as first pointed out by Wolfe and Smith /WOL62/ for a Bi0.88Sb0.12 single crystal. They reported a dimensionless figure of merit ZT greater than unity between 125 and 275 K by applying an optimum magnetic field. These values are at least twice the zero-field values. The reason for this large improvement is the presence of transverse thermomagnetic effects. The magnitude of these effects is particularly high for Bi-rich Bi-Sb alloys, making them very attractive materials for Ettingshausen cooling devices /CUF63, HOR80, JAN94/.
Near room temperature, the best materials are the (Bi,Sb)2(Te,Se)3 alloys. They constitute the most well-established materials used nowadays in thermoelectric devices for cooling applications. As these materials are at the heart of this thesis, a review of their salient properties, including their thermoelectric properties, will be presented in detail in the next chapter. Apart from cooling applications, bismuth-telluride-based alloys can be also used for power generation applications. However, the existence of a small energy gap as well as chemical stability problems limits their use for power generation up to around 500 K. At higher temperature, materials with superior thermoelectric properties are used.
Applications of materials in power generators require chemical stability at high temperature and appropriate band gaps in order to achieve reasonable Seebeck coefficients in the extrinsic regime of conduction. Alloys based on lead telluride, PbTe, satisfy these requirements and are the most suited materials to operate in the 500-800 K temperature range. They were employed in Radioisotope Thermoelectric Generators (RTGs) for space applications during fourteen years covering the period 1961-1975.
PbTe and its related compounds crystallize in a rock-salt structure and exhibit isotropic transport properties. Their electronic band structure is multi-valley with a direct band gap of 0.32 eV at 300 K. Both n- and p-type electrical conduction can be achieved in PbTe as a result of deviations from stoichiometry. Excess Pb with respect to the stoichiometric ratio results in n-type conduction while excess Te yields a p-type material. However, the maximum carrier concentration that can be achieved through these deviations is not sufficient to attain optimal electrical properties at elevated temperatures. Doping is therefore essential and can be realized upon using alkali metals (acceptors) and/or halogens (donors) /BAS80, RAV70/. A maximal ZT of about 0.8 is achieved around 700 K for both materials at optimal doping (see Fig. I-7). Recently, these materials were revisited and improved thermoelectric properties were obtained (100% increase!) in both n- and p-type materials /LAL11, PEI11/. The results were consistent with prior reports though the thermal conductivity has been found to be historically overestimated according to the authors. Another impressive result was obtained by Heremans et al. /HER08/ in PbTe doped by Tl (p-type). Thallium impurities play a specific role by adding additional electronic states in the valence band. This causes a distortion of the density of states that can be used to boost the thermopower. Such band structure engineering resulted in a twofold increase in ZT that reaches 1.5 at 773 K.
Likewise the Bi2Te3 system, the figure of merit of PbTe can be improved by the formation of solid solutions. A study of the PbTe-SnTe alloys showed that one of the most promising composition for n-type is Pb0.75Sn0.25Te /ROS61/. Due to oxidation and evaporation of Te, the usefulness of PbTe and PbSnTe solid solutions for power generation is limited to temperatures up to 800 K. More complex compounds, formed by alloying AgSbTe2 with GeTe, and referred to as TAGS compounds, have been proven to be more efficient p-type materials than PbTe alloys /SKR95/.
In this same temperature range (Fig. I-7), iron disilicide ( -FeSi2), though possessing modest ZT values, is a useful material for power generation applications and effort are currently underway to build terrestrial power generators incorporating this material. It is often preferred to other materials because of the following advantages:
• its stability with respect to oxidation, sublimation, evaporation and diffusion,
• its non-toxic elements,
• its low-cost,
• the possibility to use powder technology for its synthesis.
Moderated and high temperature bulk materials
Among materials that were suggested originally to show attributes for the realization of a PGEC material are skutterudite compounds. The name of skutterudite is derived from a naturally occurring mineral with CoAs3 structure, which was discovered in Skutterud (Norway). The general formula of skutterudite compounds is MX3, where M is one of the group 9 transition metals such as Co, Rh, or Ir and X is a pnictogen atom such as P, As, or Sb. These compounds exhibit a body-centered-cubic structure that contains 32 atoms in the unit cell described within the space group Im3. The most important point of the skutterudite structure is that there are two voids in the unit cell that can be occupied by loosely-bound atoms that are known as “rattlers”. This generates a so-called filled or partially filled skutterudite RxM4X12, where R stands for the filler atom and x (< 1) indicates the fractional occupancy on the available void site. Figure I-8 presents the crystal structure of skutterudites.
Among binary skutterudites, CoSb3 has attracted the greatest interest in electrical generation applications due to its reasonable band gap of ~ 0.2 eV, high carrier mobility, and the fact that it is composed of inexpensive and environmentally benign constituent elements as compared to other skutterudites such as CoAs3. However, the lattice thermal conductivity of CoSb3 is too high ( ~ 10 W/m.K at 300 K) to result in high ZT values. By filling the cage of the structure, it was possible to decrease significantly the thermal conductivity and thus to reach high ZT values. It has been reported that a large variety of guest atoms can be inserted, such as rare earth elements /KUZ03, MOR97, NOL98b, NOL00a, PEI08/, alkaline earth elements /CHE01b, PUY04, ZHA06b/, alkali metals /PEI06, PEI09/, or others /FUKU10, HAM10, NOL00b, NOL04, SAL00 /. The filling atoms are loosely-bound to the other atoms in the cages, leading to strong phonon scattering and significant reduction of the thermal conductivity /KEP98, HER03/. Adding filler atoms into the void of the RxCo4X12 skutterudite structure introduces also extra electrons resulting in n-type semiconductors. In order to create a p-type material, an element with fewer electrons should substitute for Co. This can be realized by using for instance Fe, which provides holes. The resultant chemical formula of p-type skutterudites is usually written in the form RyFexCo4-xSb12. The range for y and x (in both n and p-type) is determined by the electronegativity, charge states, and structural stability of the filler atoms. In addition to the single-filled system, double-filled or triple-filled compounds show great promise. Introducing two or three filler elements from different chemical groups into the cages of CoSb3 leads to two or three distinctive filler vibrational frequencies that impact a broader range of phonons, leading to a further reduction in the lattice thermal conductivity /SHI08, YAN07/. As a result, the maximum ZT values were improved to 1.3–1.5 (at 800 K) in double-filled n-type skutterudites /SHI08/. The p-type skutterudites show maximum ZT values around 0.8 – 0.9 at 800 K. While these values are lower than those of n-type analogues, they are high enough to develop modules based only on skutterudite legs. Coupled with their good mechanical properties, skutterudies materials are so far among the best candidate materials for power generation in the temperature range 500 – 800 K, which cover a large amount of wasted heat generated in industrial processes including transportation.
Table of contents :
CHAPTER I: Introduction to thermoelectricity. Survey of state-of-the-art materials
I) BASIC PRINCIPLES
I-1) Thermoelectric effects
I-2) Thermoelectric modules and performance
I-3) Selection criteria
II) SURVEY OF STATE-OF-THE ART MATERIALS
III) NEW THERMOELECTRIC MATERIALS
III-1) Moderated and high temperature bulk materials
III-1-6) Zintl phase Yb14MnSb11
III-1-7) Oxides and oxychalgogenides
III-2) Low and room temperature bulk materials
III-2-3) Organic materials
III-3) Low-dimensional structures and nanostructured bulk materials
CHAPTER II: Sb2-xBixTe3 solid solutions: general properties and state-ofthe- art
I) GENERAL PROPERTIES OF Sb2-xBixTe3 SOLID SOLUTIONS
I-1) Crystal structure
I-2) Electronic band structure
I-3) Phase diagram and defects in Sb2-xBixTe3 solid solutions
II) SOME PAST RESULTS ON SINGLE CRYSTALS AND
III) PROMISING RESULTS ON NANOSTRUCTURING OF Sb2-xBixTe3
CHAPTER III: Synthesis of Sb2-xBixTe3 based compounds. Microstructural characterizations.
I) ELABORATION OF MATERIALS
I-1) Parameters studied and prepared samples
I-2) Preparation of the initial alloys
I-3) Nanostructuration by melt-spinning
I-3-1) Principle of melt-spinner
I-3-2) Instrument description and samples preparation
I-4) Densification of the materials
I-4-1) Spark Plasma Sintering (SPS) 11111’
I-4-2) Cold pressing
II) CHARACTERIZATION TECHINQUES
II-1) X-ray diffraction
II-2) Scanning electron microscope (SEM)
II-3) Transmission electron microscope (TEM)
II-3-2) Preparation of the samples
III) PHYSICO-CHEMICAL CHARACTERIZATION OF THE MATERIALS 111
III-1) Structural analysis by X-ray diffraction
III-1-1) Melt-spun ribbons
III-1-2) Initial ingots and densified samples
III-2) Microstructure investigations by SEM
III-2-1) Melt-spun ribbons
III-2-2) MS-SPS ingots
III-2-3) MS-double SPS ingots
III-2-4) MS-aligned ribbons SPS
III-3) Microstructural investigations by TEM and HRTEM
III-3-1) MS ribbons of Sb1.6Bi0.4Te3
III-3-2) MS ribbons of Sb1.52Bi0.48Te3
III-3-3) MS-SPS samples
CHAPTER IV: Measurement techniques for thermal, electrical and galvanomagnetic properties
I) THERMOELECTRIC MEASUREMENTS AT LOW TEMPERATURES 111 116
I-1) Principle of measurement
I-1-1) Electrical resistivity
I-1-2) Thermal conductivty and thermopower
I-2) Equipments and experimental protocols
II) GALVANOMAGNETIC MEASUREMENTS AT LOW TEMPERATURES
II-1) Principle of Hall effect
II-2) Protocol of experiment
III) THERMOELECTRIC MEASUREMENTS AT HIGH TEMPERATURES
III-1) Thermal conductivity by the laser flash technique
III-1-1) Measurement principle
III-1-2) Protocol of experiment
III-2) Thermopower and electrical resistivity
IV) GALVANOMAGNETIC MEASUREMENTS AT HIGH TEMPERATURES
V) SAMPLE ORIENTATION
VI) CALIBRATION OF MEASUREMENTS AND UNCERTAINTY
CHAPTER V: Results and Discussions
I) TRANSPORT PROPERTIES OF Sb2-xBixTe3 SERIES
I-2) Thermoelectric and galvanomagnetic properties of Sb2-xBixTe3 with x = 0.4 or 0.48
I-3) Reproducibility of the processed materials
II) TRANSPORT PROPERTIES OF Sb2-xBixTe3.1 SERIES 111111 111 153
II-2) Thermoelectric and galvanomagnetic properties of Sb2-xBixTe3.1
II-4) Reproducibility of the processed materials 11111
III) INFLUENCE OF SOME PROCESSING PREPARATION ON THE TRANSPORT PROPERTIES OF MS SAMPLES
III-2) Double SPS
III-3) Influence of a post-annealing and a cold pressing
III-4) Alignment of ribbons: influence on the thermoelectric properties
IV) INTERPRETATION OF THE Sb2-xBixTe3 MEASUREMENTS
IV-2) S and approximation for T < 20 K and for 200 K < T < 300 K
IV-2-1) Seebeck coefficient S
IV-3) Lattice thermal conductivity of Sb2-xBixTe3 polycrystals
V) TRANSPORT PROPERTIES OF (Sb1.52Bi0.48)1-ySnyTe3+z
V-2) Thermoelectric and galvanomagnetic properties of (Sb1.52Bi0.48)1-ySnyTe3
V-3) Thermoelectric and galvanomagnetic properties of (Sb1.52Bi0.48)1-ySnyTe3.1
V-4) Is Sn a resonant impurity in (Sb1.52Bi0.48)1-ySnyTe3?
CONCLUSION AND FUTURE DIRECTIONS
CONCLUSIONS ET PERSPECTIVES