Patent Application
20150139848
The present invention relates to a method for increasing the thermoelectric properties of a SiGe-type matrix by in situ silicidation.
More specifically, a second phase in the form of metal nanoparticles is introduced within the matrix, after which the assembly is submitted to a sintering at an adapted temperature, generating a modification of the composition and of the microstructure of the matrix.
The thermoelectric effect is based on two principles:
The thermoelectric effect finds multiple applications, such as thermoelectric refrigeration or electrical power generation.
However, for the time being, low efficiencies of conversion by thermoelectric effect limit commercial applications, particularly regarding electrical power generation. As an example, radioisotope thermoelectric modules are used for the supply of space probes. The generalization of thermoelectric modules on the market is conditioned by a necessary increase of the conversion efficiency, which depends on the properties of the materials used.
The efficiency of a thermoelectric refrigeration system and that of the thermoelectric effect for electrical power generation depend on temperature and on a dimensionless variable called “figure of merit”.
The properties of a thermoelectric material are quantified by figure of merit ZT according to the following relation:
ZT=σ.S
2
/λ.T
where:
Ideally, a thermoelectric material having a good efficiency will thus have a high merit coefficient Z, that is, simultaneously a high Seebeck coefficient, a good electric conductivity (a low electric resistance), and a low thermal conductivity.
The thermal conductivity represents the amount of heat transferred per surface area unit and per time unit under a temperature gradient of 1 degree per meter. The thermal conductivity is partly due to electric conductivity (charge carrier motion) and on the other hand to the very structure of the material (atom vibrations). Indeed, in a solid, the atom vibrations are not random and independent from one another, but correspond to specific vibration modes, also called “phonons”. These fundamental modes of vibration correspond to waves which may propagate in the material, if its structure is periodic (organized).
A way to increase figure of merit ZT is to decrease the lattice thermal conductivity by forming a nanostructured thermoelectric material. This effect is the consequence of a larger phonon diffusion, which may be provided:
More particularly, document Wang et al. describes an n-doped SiGe alloy synthesized from powders densified by SPS (“Spark Plasma Sintering”), having a structure with nanometric grains (10-20 nanometers). Such a specificity enables to achieve a strong improvement of the figure of merit (1.3 at 900° C.) imputable to a significant decrease of the thermal conductivity due to the phonon diffusion increase by the significant density of grain boundaries in the material.
Document Mingo et al. describes, with theoretical calculations, how the inclusion of nanoparticles of a secondary phase in a SiGe alloy results in increasing the figure of merit. 17 silicides have thus been evaluated as being potentially advantageous. At high temperature, this new class of materials may enable to achieve figure of merit values close to 2. For particles having a size in the range from 2 to 10 nanometers, the performance improvement is maximum but an improvement should be visible even with particles of greater size, particularly in the case of a thermoelectric alloy.
As already mentioned, the thermal conductivity particularly relies on phonons. A phonon is a quasi-particle which results from the displacement of one or a plurality of atoms around their position of equilibrium. The presence of nanostructures within the material decreases the mean free path of phonons, and accordingly the thermal conductivity. Thus, the phonon diffusion is disturbed when the size of the grains or of the inclusions is smaller than or comparable to the mean free path of phonons. Such a so-called nanometric size is in the range from 1 to 1,000 nanometers, but preferably in the range from 2 to 20 nanometers to disturb the diffusion of a larger number of phonons. The mean free path of a phonon is defined as being the mean distance traveled by a phonon with no collision (or between two collisions).
An adapted thermoelectric material, for example, with silicides, should meet the following constraints:
There thus is an obvious need to develop technical solutions enabling to obtain thermoelectric materials having the above-mentioned properties.
The present invention relates to a method of manufacturing a SiGe-type thermoelectric material, wherein the nanostructuring is provided by a second phase. The conditions of the manufacturing process, resulting in an effective increase of the thermoelectric performances (Z) of the material, follow a narrow selection range, defined within the framework of the present invention.
In other words, the present invention provides a method enabling to increase the thermoelectric performances of a SiGe-type alloy by a modification of its composition and of its microstructure, due to the addition of a second phase, in the case in point, nanoinclusions of metal silicides (MSi) formed in situ.
Thus, and according to a first aspect, the present invention aims at a thermoelectric material manufacturing method, which composes the steps of:
In other words, the last step comprises submitting the mixture, advantageously in solid form, to a step of sintering at an adapted temperature, during which nanometric inclusions of composition MaAb and/or MaBb are formed in the matrix.
As mentioned, the initial material, intended to be used as a matrix in the final material obtained at the end of the method according to the invention, is a material which itself has thermoelectric properties. Further, it advantageously is an at least binary material of formula AxB1-x with 0<x<1.
It may also be a ternary material (ABC), or even a quaternary material (ABCD), for which the forming of a second nanometric phase with a metal M enables to provide the microstructural modification (in the form of inclusions) capable of significantly decreasing the thermal conductivity of the material, with no decrease of the power factor, in other words, of increasing the thermoelectric properties reflected by variable Z.
As already mentioned, it advantageously is an alloy. Indeed, an alloy has a disorderly atom lattice, as compared with a material formed of a single type of atoms. Such a disorder thus enables to decrease the mean free path of part of the phonons, that is, high-frequency phonons. The consequence is a first decrease of the thermal conductivity due to the use of an alloy as a matrix.
As for low-frequency phonons, which keep a high mean free path in an alloy and are responsible for the heat transport in an alloy, they are disturbed by the addition according to the invention of nanoinclusions, which further decreases the thermal conductivity of the alloy.
Thus, nanoinclusions (or silicides in the case of a matrix containing Si) enable to lower the thermal conductivity of a non-alloyed matrix but the lowering is even stronger if the effects of an alloy and of the nanoinclusions are combined.
Advantageously, an alloy used in the context of the present invention comprises silicon (A or B═Si), the nanoinclusions formed at the end of the method according to the invention then being silicides. In the specific case of a SiGe matrix, the inclusion obtained at the end of the method thus is a silicide of formula MaSib, particularly MSi2. In the generalization to any at least binary alloy (AxB1-x), the inclusion, after its reaction in the solid state with the matrix during the sintering, then forms an alloy having composition MaAb or MaBb.
Typically, and as already mentioned, the alloy used may be SiGe at different stoichiometries and thus of general formula Si1-xGex with 0<x<1, advantageously with 0.01≦x≦0.5, more advantageously still 0.01<x≦0.2. It for example is Si0.8Ge0.2 with x=0.2 or Si0.92Ge0.08 with x=0.08.
In the case of a SiGe matrix, the proportion of germanium mainly defines the thermal conductivity thereof. Indeed, the larger its content, the more the thermal conductivity of the alloy will be advantageously decreased. It thus varies from 140-150 W/m/K when x=0 to 4-6 W/m/K when x=0.2. In practice, x designating the Ge stoichiometry is advantageously in the range from 0.01 to 0.5.
According to a specific embodiment, this material is doped with doping agents such as phosphorus (P) or boron (B).
In the case of SiGe, it advantageously is phosphorus when the alloy is n-doped and boron when the alloy is p-doped. The corresponding formulas are (Si1-x-yGexPy) and Si1-x-yGexBy, respectively, with 0.001≦y≦0.1.
As an example, and according to a preferred embodiment, a p-doped matrix has formula Si0.795Ge0.2B0.05 and an n-doped matrix has formula Si0.795Ge0.2P0.005.
Advantageously, the doping agent, such as boron or phosphorus, amounts to from 1019 to 1020, or even from 1019 to 1021 particles per cm3 of the at least binary material. The modulation of the doping rate is performed by varying the quantity of the dopant agent, for example boron in a p-doped matrix and for example phosphorus in an n-doped matrix.
An alternative binary alloy capable of being used as a matrix is Mg2Sn.
At the next step, this powder is mixed with metal particles.
Typically according to the invention, the particles mixed with the matrix are particles of pure metal M, which further have a nanometric size.
“Nanoparticles” means the fact that the particles have, in at least one of their dimensions, preferably in all their dimensions, a size smaller than 1,000 nanometers, advantageously smaller than 500 nanometers, more advantageously still in the range from 2 to 20 nanometers. It should be noted that these same size properties are desired for silicides or MaAb or MaBb alloys obtained at the end of the sintering. In practice, the initial particles of pure metal M should be from 1.5 to 5 times smaller than the size desired for the MaAb or MaBb inclusions, which should themselves respect the previously-described dimensions, that is: a size smaller than 1,000 nanometers, advantageously smaller than 500 nanometers, more advantageously still in the range from 2 to 20 nanometers.
Different metals M may be envisaged, their nature depending on the matrix into which they will be introduced in the form of nanometric particles. Indeed, a number of constraints bear on the metal particles, particularly in order to see the figure of merit increase.
One of the constraints relates to the melting point (or melting temperature) of said metal, which should be greater than the applied sintering temperature.
Indeed, during the sintering step, material AB, advantageously SiGe, and its metal inclusions M should not melt. The silicide (MA, in particular, MSi) forms during the sintering, in the solid state. It should further be stable without decomposing or melting at the sintering temperature of the matrix, advantageously SiGe. This enables to avoid a segregation, a decomposition of the silicides on cooling thereof, with, as a consequence, a non-homogeneous composite material. This condition is a significant difference with document WO 2008/140596 which provides a forming of the silicides during the cooling, after a liquid-state step, in practice after the melting.
In the context of the method according to the invention, the matrix as well as the metal nanoparticles, are in the solid state and the forming of the nanoinclusions is performed by a solid-state process.
In other words, the silicide (MaSib), or more generally the MaAb or MaBb alloy should be formed from a refractory transition metal (M), having a melting temperature higher than the sintering temperature. The sintering temperature varies according to the composition of the actual matrix and will be defined hereafter.
All the metal elements which meet these criteria may be used, for example, vanadium (V), tungsten (W), molybdenum (Mo), zirconium (Zr), and titanium (Ti), which will particularly form highly-conductive refractory silicides VSi2, WSi2, MoSi2, ZrSi2, and TiSi2.
Chromium (Cr) and iron (Fe) may be used, particularly to form semiconductor refractory silicides CrSi2 and FeSi2, respectively. Tantalum (Ta), cobalt (Co), and osmium (Os) are also possible.
Refractory materials such as Mo, Ta, and W appear to be particularly promising. All these metals are advantageously used in the case of a SiGe-type matrix.
Generally, the matrix may embark different types of silicides at a time, such as for example CrSi2 and MoSi2 inclusions. These different silicides may have different sizes (while remaining within the above-described range) to better disturb the phonon transport and thus mote advantageously further decrease the thermal conductivity. In practice, it is thus possible to use a mixture of metal powders, each powder being made of a pure but different metal.
For another example, that is, a Mg2Sn matrix, Te metal nanoparticles capable of forming the MgTe or SnTe alloys, respectively, are advantageously used.
Another criterion is the fraction by volume of the inclusions in the matrix. The fraction by volume of the metal inclusions, which will form a silicide or more generally an alloy in matrix AB, is advantageously in the range from 0.2 to 20%. Beyond, the inclusions may coalesce together and become too large to allow a phonon diffusion.
In practice, and advantageously, the nanoparticles of pure metal (M) are thus introduced by a fraction by volume of the mixture in the range from 0.2 to 20%.
Since the method according to the invention is based on a sintering step, the different components of the thermoelectric material are thus introduced in the form of powders.
The first step of this method comprises preparing, if necessary, a powder of an at least binary thermoelectric alloy (AxB1-x), intended to be used as a matrix for the final thermoelectric material.
Thus, the different components, that is, the precursors of pure A and pure B, for example, pure silicon (Si) and pure germanium (Ge), as well as the dopant, particularly phosphorus (P) for an n-doped material, or boron (B) for a p-doped material, are used in the form of powders, flakes, or chips, in stoichiometric quantities.
This alloy powder, which will form the matrix of the material, may be produced by mechanosynthesis, chemical process, melting, or also atomization.
In the specific case of mechanosynthesis, the grinding elements used, that is, the grinding bowl and the balls, are for example made of stainless steel, of zirconium, or also of tungsten carbide, and this to limit the impurity input during the grinding.
According to a preferred embodiment, the mechanosynthesis parameters are defined as follows:
In a second step, this powder of the alloy of at least A and B (AB), for example, SiGe, possibly doped, is mixed with the nanoparticles of pure metal M, judiciously selected, by the above-indicated proportions. According to an advantageous embodiment, the nanoparticles are previously dispersed, for example, by the use of a planetary grinding mill or of ultrasounds.
The powder mixture should enable to fully homogenize the two solid phases, the matrix on the one hand and the metal particles on the other hand. This step defines the inclusion distribution in the densified solid material. It should be noted that the good dispersion of the silicides or alloys in the matrix conditions, to the same degree as their size, the efficient diffusion of phonons in the composite.
Typically, the powder mixing may be achieved by means of an attrition mill or a disperser operating with rotating blades, or by the setting to motion of balls in a rotating vessel.
Typically, in an attrition mill, the rotation speed is in the range from 80 to 300 revolutions/minute for a ball mass/powder mass ratio in the range from 10:1 to 50:1, and a duration from 10 minutes to 10 hours.
For a better mixing efficiency, this powder dispersion may be performed in a liquid medium, for example, in the presence of an organic solvent chemically inert with respect to the matrix. In the case of SiGe, ethanol may be used. Dispersing agents may also be added to improve the particle dispersion and avoid the agglomeration thereof.
According to a specific embodiment, the mixing of the alloy powder with the metal nanoparticles may be performed during a phase of atomization of the powder.
The next step of the method according to the invention comprises sintering this mixture. The forming of nanoinclusions is performed in situ, during the sintering, conversely to prior art.
Thus, the method according to the invention particularly differs from the teachings of document WO 2008/140596 by the fact that it involves no step of melting the material and that accordingly, the forming of silicides does not occur during the ensuing cooling. Conversely, and according to the invention, the siliciding or the forming of the alloy of nanoparticles or nanoinclusions occurs during the compaction, by sintering of the metal particle/matrix mixture, for example, Mo/SiGe.
In practice, the sintering of the powder mixture is advantageously performed by SPS (“Spark Plasma Sintering”), uniaxial hot pressing (or UHP), P2C (“Plasma Pressure Compaction”), microwave sintering, or isostatic sintering.
The sintering is advantageously performed in vacuum, at a pressure lower than or equal to 10−2 mbar, or in an inert atmosphere, for example, in the presence of helium of argon, to avoid a reaction of oxidation with the thermoelectric material.
For pressure sintering methods, of SPS type (“Spark Plasma Sintering”), uniaxial hot pressing (or UHP), P2C (“Plasma Pressure Compaction”), or isostatic sintering, the applied pressure may vary from 15 MPa to 300 MPa.
The sintering should enable to densify the matrix comprising the inclusions, by from 85 to 100%, or even from 90 to 100% of the theoretical density. The theoretical density corresponds to the maximum density that a material is capable of having, when it no longer comprises any porosity. The theoretical density is taken back to 100%.
Crucially, this sintering step should also result in the metal particle siliciding reaction, or more generally in the simultaneous forming, in the matrix, of nanometric inclusions of composition MaAb and/or MaBb.
In the context of the invention, “nanometric inclusions” designates the structures obtained from the pure metal nanoparticles (M) at the end of the sintering step. In practice, and schematically (
A significant parameter of this sintering step is the temperature at which it is performed. The sintering temperature range is defined by taking into account the two following criteria:
In practice, the sintering temperature is determined on a curve of densification of the mixture of the initial alloy (or more generally of the at least binary material AxB1-x) and of the metal nanoparticles, and advantageously corresponds to a temperature higher than the temperature which enables to reach the top of the densification peak but lower than the melting temperature of this alloy or material. The adapted temperature is thus within a very accurately determined limited range.
A densification curve corresponds to the monitoring of the piston travel speed during the sintering of the material to be studied (while taking care of subtracting the effects due to the piston expansion). During the drawing of this densification curve, no temperature stage should be applied: the material is submitted to a simple temperature rise, until the studied material melts. One or a plurality of “densification speed peaks” which correspond to temperatures where the mixture densification is high are thus obtained. After densification of the mixture, the piston travel speed strongly decreases, or may even become zero. It subsequently drastically increases back: this increase corresponds to the melting of the mixture.
The present invention defines the sintering area to be used for the mixture of the initial material comprising the metal nanoparticles. This temperature area is located between the temperature enabling to reach the maximum densification peak (=temperature enabling to densify the material at a maximum speed) and the melting temperature of the material, at which temperature the piston travel speed drastically increases back.
It should further be noted that this temperature should be lower than the melting temperature of the metal (M) of the nanoparticles.
It is generally in the range from 60% to 99.99% of the melting temperature of the alloy, visible in the phase diagram of the studied material. As an example, the Si0.8Ge0.2 alloys melts around 1,300° C. and may be sintered in the 900-1,299° C. range. In practice, the sintering temperature can be determined on the densification curves of the (matrix+nanoparticles) mixture. Indeed, the presence of nanoparticles may shift the sintering and melting temperatures of the material towards higher or lower temperatures. The densification monitoring curves should advantageously be plotted in relation with the mixture. These curves show the densification speed versus the sintering temperature. The efficiency of the sintering is maximum between the top of the densification peak and the melting temperature.
In the context of the present application, it has been shown that for a Si0.8Ge0.2 matrix, possibly doped, and Mo nanoparticles, the sintering is advantageously carried out at a temperature in the range from 1,050 to 1,250° C., more advantageously still from 1,150 to 1,250° C.
According to a specific embodiment, for a Si0.92Ge0.08 matrix and Mo nanoparticles, the sintering step advantageously occurs at a temperature in the range:
In practice, the sintering temperatures measured during the sintering are taken close to the material: either by a pyrometer or by a thermocouple (TC) placed in the sintering container. Transposing a temperature value from one piece of equipment to another or even from one container to another or from one powder to another is thus delicate. The sintering temperature range is advantageously determined from the measured densification profile of the mixture, on given equipment, with given control means (TC or pyrometer), a given container and mixture (thermoelectric material+nanoparticles).
In other words, the invention aims at providing a solid SiGe-type material, combined with nanometric metal silicide inclusions, with a modification of the composition and of the microstructure allowing a gain of at least 10%, or even 30%, or even 100% of the figure of merit with respect to the initial matrix, that is, submicronic SiGe.
The method according to the invention should also provide an original microstructure: modification of the matrix composition on the area surrounding the nanoinclusions, presence of defects linked to the sintering of the material (twins, specific boundaries), appearing of pores during the siliciding and/or, again, presence of oxides on the perimeter of nanoinclusions.
The microstructure of the final material does not necessarily comprise all these characteristics. However, these elements all promote the increase of the figure of merit and thus the efficiency of the thermoelectric material. The material obtained at the end of the method according to the invention should exhibit at least one of these characteristics, advantageously all.
In other words, it is necessary to obtain a thermoelectric material of dense SiGe type, having nanoinclusions which do not degrade the power factor (σ.S2) of the material. To achieve this, the temperatures of the “host” material (SiGe) sintering and silicidation (with a conservation of the nanometric size of the inclusions) should be compatible. Further, the silicidation should generate a microstructural modification of the material, promoting the thermal conductivity decrease.
Indeed, as long as the silicide does not form, the presence of metal particles in the material acts as dopants: their presence directly affects the power factor. Beyond a given sintering temperature, the silicide forms and the power factor of the material is similar to that of the initial matrix. The observation of the thermal conductivity then provides direct information as to the microstructure of the material and the conservation of their nanometric size. Thus, if a given sintering temperature is exceeded, the silicide grains grow and reach too large a size to allow the expected decrease. A microstructural TEM observation then enables to validate the original structure of the material thus formed.
In practice, the parameterizing of the sintering conditions may be performed by combining thermoelectric measurements (density, Seebeck coefficient, electric conductivity, and thermal conductivity) and microstructural analyses.
In document WO 2008/140596, the composition and structure modifications of the matrix, induced by the siliciding, are not controlled and may lead to porosities, silicide agglomerations, grain growth and/or a decrease in the doping of the material. Conversely, the modifications brought to the matrix in the present invention have been demonstrated as promoting the increase of thermoelectric properties (ZT).
According to another aspect, the present invention thus relates to a thermoelectric material having at least one of the following characteristics, advantageously all of these characteristics:
Advantageously, a thermoelectric material of interest according to the invention exhibits at least the two first above-mentioned characteristics, that is:
The method according to the invention thus enables to:
The foregoing features and advantages will now be discussed in the following non-limiting description of a specific embodiment, in relation with the accompanying drawings, among which:
The present invention will be further illustrated in the case of molybdenum (Mo) nanoparticles amounting to 1.3% by weight (0.9% by volume) in a Si0.8Ge0.2 matrix (Si1-xGex with x=0.2).
At the end of the method according to the invention and such as schematized in
1/ Preparation of the Alloy Powder:
The Si1-x-yGexPy alloy powders are prepared by mechanosynthesis. The powders of pure silicon (99.99%), pure germanium (99.99%), and phosphorus (99.99%) are weighted according to the targeted stoichiometry (x=0.795; y=0.005) and then placed in a 500-mL zirconia grinding bowl. Balls having a 10-mm diameter also made of zirconia, are weighted so that the total mass of the balls placed in the bowl is equal to 15 times the powder mass (Si, Ge, or P). It is spoken of a ball/powder ratio (BPR) of 15:1.
The powders are ground in a planetary mill for 12 hours at the rotation speed set to 360 rpm. At the end of this cycle, the doped SiGe alloy powders are obtained.
2/ Addition of the Metal Particles:
The nanometric powders of pure molybdenum (99.99%) are then added up to 1.3% by weight with respect to the quantity of SiGe present in this bowl.
A step of mixing, via the grinding mill, the molybdenum and SiGe powders is then carried out. In this example, the mixing takes place for 3 h at 220 rpm.
3/ Sintering of the Mixture:
The sintering is performed by SPS.
A first determination of the ideal temperature range is performed by following the densification profile of the material during the sintering step.
The upper limit is determined on the densification profile by the drastic increase of the piston travel speed: Tsup=1,165° C.
The refining of the temperature range is performed by measurements of the electric conductivity (
The minimum sintering temperature thus is in the range from 1,120 to 1,150° C.
The monitoring of the thermal conductivity (
Document WO 2008/140596 describes a wide process range, different from and coming out of the selection range described herein. Indeed, it describes a temperature range which starts at the melting point of the material with no upper limit. In the case of the present invention, the temperature range is smaller; it is limited and narrow.
The observation by transmission electronic microscopy (TEM) of the material formed by the method according to the invention is shown in
The validity of the method according to the invention has been verified with another stoichiometry of the SiGe alloy: Si0.92Ge0.08 (Si1-xGex with x=0.08), doped with phosphorus, more exactly: Si0.913Ge0.08P0.007 for type n, and doped with boron, more exactly Si0.913Ge0.08B0.007 for type p. For these two materials, 1.3% by volume of molybdenum nanoparticles have been added by the same method as that previously described. The average nanoparticle size is 20 nm.
In this case also, an increase of the ZT has been observed, this factor passing from 0.46 to 0.77 for type p (67% increase;
Wang, X. W., Lee, H., Lan, Y. C., Zhu, G. H., Joshi, G., Wang, D. Z., Yang, J., Muto, A. J., Tang, M. Y., Klatsky, J., Song, S., Dresselhaus, M. S., Chen, G. and Ren, Z. F. 2008. Enhanced thermoelectric figure of merit in nanostructured n-type silicon germanium bulk alloy. Applied Physics Letters 93,193121.
Mingo, N., Hauser, D., Kobayashi, N. P., Plissonnier, M. and Shakouri, A. 2009. Nanoparticle-in-alloy approach to efficient thermoelectrics: silicides in SiGe. NanoLetters, Vol. 9(2), 711-715.
Scoville, N., Bajgar, C., Rolfe, J., Fleurial, J.-P., Vandersande, J. 1995. Thermal conductivity reduction in SiGe alloys by the addition of nanophase particles. Nanostructured Materials, Vol. 5(2), 207-223.
| Number | Date | Country | Kind |
|---|---|---|---|
| 12.55160 | Jun 2012 | FR | national |
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/FR2013/051237 | May 2013 | US |
| Child | 14548754 | US |
