WO2022032333A1 - Thermoelectric material - Google Patents

Abstract

A composite thermoelectric material (100) comprising: a host matrix (102) comprising a semiconducting thermoelectric material; and amorphous particles (104) dispersed throughout the host matrix (102), wherein the host matrix (102) includes a polycrystalline structure (106) nucleated by the amorphous particles (104).

Description

"Thermoelectric material"

Incorporation by reference

[0001] This application claims the benefit of Australian provisional application no. 2020902822 the disclosures of which are incorporated herein by reference in their entirety.

[0002] The following published documents are incorporated herein by reference in their entirety:

[0003] Yang, G. S., Niu, R. M., Sang, L. N., Liao, X. Z., Mitchell, D. R. G., Ye, N., Pei, J., Li, J.-F., Wang, X. L., Ultra-High Thermoelectric Performance in Bulk BiSbTe/Amorphous Boron Composites with Nano-Defect Architectures. Adv. Energy Mater. 2020, 10, 2000757. https://doi.org/10.1002/aenm.202000757 (First published on 22 September 2020 in Advanced Energy Materials)

[0004] Yang, G., Sang, L., Yun, F. F., Mitchell, D. R. G., Casillas, G., Ye, N., See, K., Pei, J., Wang, X., Li, J.-F., Snyder, G. J., Wang, X., Significant Enhancement of Thermoelectric Figure of Merit in BiSbTe-Based Composites by Incorporating Carbon Microfiber. Adv. Funct. Mater. 2021, 31, 2008851. https://doi.org/10.1002/adfm.202008851 (First published on 4 February 2021 in Advanced Energy Materials)

Technical Field

[0005] The present disclosure relates to composite thermoelectric materials and methods for producing composite thermoelectric.

Background

[0006] Thermoelectric (TE) materials are capable of direct and reversible conversion of thermal to electrical energy. Operation of TE materials is based on the Seebeck and Peltier effects. Thermoelectricity, produced by TE materials, has emerged as a promising green energy technology due to the distinct advantages of zero emissions, high reliability, silent and vibration-free operation., which is primarily governed by the dimensionless figure of merit ZT = S²σT/κ, where σ is electrical conductivity, S is

Seebeck coefficient, T is the absolute temperature, and κ is the thermal conductivity (κ = electronic (κ_e) + lattice thermal conductivity

[0007] Any discussion of documents, acts, materials, devices, articles or the like which has been included in the present specification is not to be taken as an admission that any or all of these matters form part of the prior art base or were common general knowledge in the field relevant to the present disclosure as it existed before the priority date of each claim of this application.

[0008] Throughout this specification the word "comprise", or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated element, integer or step, or group of elements, integers or steps, but not the exclusion of any other element, integer or step, or group of elements, integers or steps.

Summary

[0009] There is provided a composite thermoelectric material comprising: a host matrix comprising a semiconducting thermoelectric material; and amorphous particles dispersed throughout the host matrix, wherein the host matrix includes a polycrystalline structure nucleated by the amorphous particles.

[0010] The semiconducting thermoelectric material may comprise bismuth, tellurium, and antimony.

[0011] The ratios of bismuth (Bi) to antimony (Sb) to tellurium (Te) may be given by Bi_xSb₂-_xTe₃ where x=0 to 2. [0012] The amorphous particles may be amorphous boron.

[0013] The particles of amorphous boron may be between 1 nm and 300 nm in diameter.

[0014] The host matrix may further comprise a plurality of spatially distributed phases having higher concentrations of antimony and tellurium, the plurality of spatially distributed phases having dimensions between 1 and 50 nm.

[0015] The boron may be at a concentration of 0 up to 5 wt%.

[0016] The amorphous particles may be amorphous carbon.

[0017] The amorphous carbon may be derived from carbon fibres or carbon black.

[0018] The carbon fibres may be uniformly distributed through the host matrix.

[0019] The carbon fibres may have a cross-sectional diameter of 50 μm or smaller.

[0020] The carbon fibres may have a length of 2 nm to 1 cm.

[0021] The particles of amorphous carbon may have dimensions of 1 nm to 50 μm.

[0022] The amorphous carbon particles may be derived from an evaporated liquid source of carbon.

[0023] The liquid source of carbon may comprise a solution of an organic carbonous solid in a solvent.

[0024] The amorphous carbon particles may be derived from an organic carbonous solid.

[0025] The material may further comprise a network of defects. [0026] The material may further comprise sharp interfaces between the host matrix and amorphous particles.

[0027] The amorphous particles may comprise a mixture of amorphous boron and amorphous carbon. The amorphous carbon may be derived from an organic carbonous solid or liquid.

[0028] The amorphous particles may comprise amorphous boron particles coated in amorphous carbon in a core-shell structure.

[0029] The amorphous particles may comprise amorphous carbon particles coated in amorphous boron in a core-shell structure.

[0030] There is further provided a method of producing a composite thermoelectric material, the method comprising the steps of: combining powdered constituents of a host matrix and amorphous particles, wherein the host matrix comprises a semiconducting thermoelectric material; adding powders to a die, wherein the die determines the final shape and dimensions of the composite material; and applying heat energy and pressure to the powders in the die to amalgamate the powders into the composite material.

[0031] There is further provided a method of producing a composite thermoelectric material, the method comprising the steps of: combining powdered constituents of a host matrix and a solution of organic carbonous material, wherein the host matrix comprises a semiconducting thermoelectric material; heating the combined host matrix powder and solution to evaporate solvent from the solution and decompose the organic carbonous material to amorphous carbon particles, the amorphous carbon being distributed in the powdered constituents of the host matrix; adding the combined host matrix powder and amorphous carbon particles to a die, wherein the die determines the final shape and dimensions of the composite material; and applying heat energy and pressure to the powders in the die to amalgamate the powders into the composite material.

[0032] The temperature may be below 500 Celsius.

[0033] The method may further comprise grading the combined amorphous carbon particles and powdered constituents of the host matrix before adding to the die.

Brief Description of Drawings

[0034] Fig. 1 is schematic illustration of a thermoelectric material;

[0035] Fig. 2 is a schematic illustration of a host matrix of a thermoelectric material;

[0036] Fig. 3 is a flow chart of a method for producing a thermoelectric material;

[0037] Fig. 4 is a flow chart of a method for producing a thermoelectric material;

[0038] Fig. 5a illustrates a figure of merit for different thermoelectric materials at different temperatures;

[0039] Fig. 5b illustrates a figure of merit and efficiency for different thermoelectric materials; [0040] Fig. 6a illustrates electrical conductivity for different thermoelectric materials at different temperatures;

[0041] Fig. 6b illustrates Seebeck coefficient for different thermoelectric materials at different temperatures;

[0042] Fig. 6c illustrates power factor for different thermoelectric materials at different temperatures;

[0043] Fig. 6d illustrates thermal conductivity for different thermoelectric materials at different temperatures;

[0044] Fig. 7a illustrates the difference between the total thermal conductivity and electronic thermal conductivity for different thermoelectric materials at different temperatures;

[0045] Fig. 7b illustrates lattice thermal conductivity for different thermoelectric materials at different temperatures;

[0046] Fig. 7c illustrates bipolar thermal conductivity for different thermoelectric materials at different temperatures;

[0047] Fig. 7d illustrates the frequency dependence of thermal conductivity of different defects in a thermoelectric material;

[0048] Fig. 8a is a TEM micrograph of a thermoelectric material;

[0049] Fig. 8b is a TEM micrograph of a thermoelectric material;

[0050] Fig. 8c is a high magnification TEM micrograph of a portion of the thermoelectric material illustrated in Fig. 8b;

[0051] Fig. 8d is a TEM micrograph of a thermoelectric material; [0052] Fig. 8e is a high magnification TEM micrograph of a portion of the thermoelectric material illustrated in Fig. 8d;

[0053] Fig. 8f is a high magnification TEM micrograph of a portion of the thermoelectric material illustrated in Fig. 8e;

[0054] Fig. 9a illustrates a portion of thermoelectric material where EDS measurements were taken;

[0055] Fig. 9b shows the EDS results from the areas in Fig. 9a;

[0056] Fig. 9c illustrates an area where EELS measurements were used to identify boron;

[0057] Fig. 9d illustrates the boron K-edge used to identify the boron;

[0058] Fig. 10a illustrates temperature-dependence of the thermoelectric figure of merit for different thermoelectric materials;

[0059] Fig. 10b illustrates temperature-dependence of the thermoelectric figure of merit for different thermoelectric materials;

[0060] Fig. 10c illustrates figure of merit and efficiency values for different thermoelectric materials;

[0061] Fig. 1 la illustrates a unicouple cooling module;

[0062] Fig. 1 lb illustrates the cooling effect of a unicouple at different currents compared to a reference;

[0063] Fig. 11c illustrates the cooling of unicouples as a function of its hot-side operating temperature; [0064] Fig. 12a shows the temperature of the cool side of a unicouple as a function of time for a 3.5 A current;

[0065] Fig. 12b shows the temperature of the cool side of a unicouple as a function of time for a 4A current;

[0066] Fig. 12c shows the temperature of the cool side of a unicouple as a function of time for a 4.5A current;

[0067] Fig. 13a illustrates electrical conductivity for different thermoelectric materials at different temperatures;

[0068] Fig. 13b illustrates Seebeck coefficient for different thermoelectric materials at different temperatures;

[0069] Fig. 13c illustrates power factor for different thermoelectric materials at different temperatures;

[0070] Fig. 13d illustrates thermal conductivity for different thermoelectric materials at different temperatures;

[0071] Fig. 14a illustrates the difference between the total thermal conductivity and electronic thermal conductivity for different thermoelectric materials at different temperatures;

[0072] Fig. 14b illustrates lattice thermal conductivity for different thermoelectric materials at different temperatures;

[0073] Fig. 14c illustrates bipolar thermal conductivity for different thermoelectric materials at different temperatures;

[0074] Fig. 15a shows carrier concentration and charge mobility for thermoelectric materials having different concentrations of carbon fiber; [0075] Fig. 15b illustrates the difference between the total thermal conductivity and electronic thermal conductivity for materials having different concentrations of carbon fiber;

[0076] Fig. 15c illustrates a quality factor and figure of merit for materials having different concentrations of carbon fiber;

[0077] Fig. 16a is an electron micrograph of a thermoelectric material;

[0078] Fig. 16b is a low magnification transmission electron microscope image of a thermoelectric material;

[0079] Fig. 16c is an electron micrograph of the interface between a host matrix and amorphous carbon;

[0080] Fig. 16d is an electron micrograph of the interface between a host matrix and an antimony rich phase;

[0081] Fig. 16e is an electron micrograph of an antimony rich phase;

[0082] Fig. 16f is a detailed image ofthe antimony rich phase of Fig. 16e;

[0083] Fig. 17 illustrates potential phonon scattering mechanisms;

[0084] Fig. 18a illustrates the hardness of different thermoelectric materials; and

[0085] Fig. 18b illustrates the fracture toughness of different thermoelectric materials.

Description of Embodiments

[0086] As mentioned above, thermoelectric materials are a promising candidate for green energy technology. However, its widespread application remains constrained by the relatively low conversion efficiency. The efficiency of these materials is primarily governed by the dimensionless figure of merit ZT = S²σT/κ, where σ is electrical conductivity, S is Seebeck coefficient, T is the absolute temperature, and κ is the thermal conductivity (κ = electronic (κ_e) + lattice thermal conductivity A

material with a higher ZT is more efficient at converting thermal energy to electrical energy (or vice versa).

[0087] Maximizing ZT requires the enhancement of the power factor (PF = S²σ) and the reduction of thermal conductivity, κ. The difficulty arises because S, σ and κ_e are strongly coupled with each other, which means that an improvement in one parameter usually results in a negative effect on the other two parameters. Note that KL is the only independent parameter determining ZT. However, the introduction of phonon scattering sources to reduce such as alloying, dislocations and nano/meso-architectures, also

enhances charge carrier scattering thereby reducing electrical conductivity.

OVERVIEW - MATERIALS

[0088] Fig. 1 illustrates a thermoelectric (TE) material 100 with enhanced power factor and reduced TE material 100 comprises a host matrix 102 comprising a

semiconducting thermoelectric material and amorphous particles 104 dispersed throughout host matrix 102. Host matrix 102 includes a polycrystalline structure 106 nucleated by amorphous particles 104.

[0089] In some embodiments, the semiconducting thermoelectric material comprises bismuth, telluride and antimony.

[0090] In some embodiments, amorphous particles 104 are amorphous boron (B) particles. A detailed description of TE material 100 comprising amorphous boron particles is described below. The amorphous boron particles may have a diameter between 1 nm and 300 nm. For example, in some embodiments, the amorphous boron particles have an average diameter of 100 nm. [0091] In some embodiments, as shown in Fig. 2, host matrix 102 further comprises a plurality of spatially distributed phases 202 having higher concentrations of antimony and tellurium, the plurality of spatially distributed phases having dimensions between 1 and 50 nm. A consequence of spatially distributed phases 202 having higher concentrations of antimony and tellurium is that remaining portions 204 of host matrix 102 will have relatively lower amounts of antimony and tellurium and will be relatively richer in bismuth.

[0092] As mentioned, semiconducting TE material 102 may comprise bismuth (Bi), antimony (Sb) and tellurium (Te). The ratios of bismuth (Bi) to antimony (Sb) to tellurium (Te) is given by Bi_xSb₂-_xTe₃ where x=0 to 2. For example, in some embodiments, TE material 102 comprises Bi_0.5SB_1.5Te₃.

[0093] In some embodiments, amorphous particles 104 are amorphous carbon. The amorphous carbon particles may have dimensions between 1 nm and 50 μm. The amorphous carbon may be derived from carbon fibres, carbon black, an organic carbonous solid, from an evaporated liquid source of carbon or any other suitable source of amorphous carbon. The liquid source of carbon may comprise a solution of organic carbonous material dissolved in a solvent. For example, the organic carbonous material may be a sugar and the solvent may be water. The process of deriving amorphous carbon particles from liquid carbon sources is discussed below.

[0094] In embodiments where the amorphous carbon is derived from carbon fibres, the carbon fibres, the fibres may have a cross-sectional diameter of 50 μm or smaller. The length of the fibres may be from 2 nm to 1 cm.

[0095] In some embodiments, amorphous particles 104 comprise boron particles coated in amorphous carbon in a core shell structure. Alternatively, or in combination, amorphous particles 104 may comprise carbon particles coated in amorphous boron. Amorphous particles may be included in material 100 in concentrations up to 5 wt% (that is, the amorphous particles comprise up to 5% of the total mass of the composite material). Examples of various concentrations of amorphous particles are discussed in detail below for amorphous boron particles and for amorphous carbon particles.

[0096] Thermoelectric material 100 may further comprise a network of defects 108 throughout host matrix 102. The defects are discontinuities in the material and may be, for example, crystal boundaries in the polycrystalline structure 106, interfaces between host matrix 102 and amorphous particles 104, interfaces between spatially distributed phases 202 and remaining portions 204 of host matrix 102. As discussed in detail below, defects 108 inhibit thermal conductivity, thereby enhancing the figure of merit, ZT, for the material.

OVERVIEW - FABRICATION

[0097] Methods 300 and 300’ for producing composite thermoelectric material 100 described above are illustrated as flow charts in Figs. 3 and 4 respectively. Method 300 relates to production of a composite thermoelectric material from powdered constituents, while method 300’ relates to production of a composite thermoelectric material from powdered and liquid constituents.

[0098] At step 302 of method 300, powdered constituents of a host matrix are combined with amorphous particles. As before, the host matrix comprises a semiconducting thermoelectric material. For example, the semiconducting thermoelectric material may comprise bismuth, antimony and tellurium (BST). The amorphous particles may be amorphous boron, amorphous carbon or combinations of the two.

[0099] The BST can be reduced to a powder using any appropriate technique such as a planetary ball mill. If the amorphous particle are larger than desired, they may be milled along with the BST to the desired dimensions.

[0100] At step 304, the combined powdered constituents of the host matrix and amorphous particles are added to a die. The die supports and contains the constituent materials as they are fused into a solid and determines the final shape and dimensions of the composite material.

[0101] At step 306, heat and pressure are applied to the powders in the die. The heat and pressure cause the powders in the die to amalgamate into the composite material. That is, the heat and pressure cause the powders to solidify. During the solidification, amorphous particles 104 nucleate crustal growth in host matrix 102 resulting in a polycrystalline material. The crystal boundaries in the polycrystalline material inhibits phonon transfer of energy through host matrix 102 thereby reducing thermal conductivity of composite material 100. That is, the crystal boundaries 108, as well as other defects, reduce the lattice thermal conductivity KL resulting in an overall reduction in thermal conductivity κ. Other defects, such as atomic defects and interfaces between host matrix 102 and amorphous particles 104 also reduce thermal conductivity in a similar manner.

[0102] Furthermore, differences in thermal expansion coefficients between amorphous particles 104 and host matrix 102 results in strain fields which further reduce thermal conductivity. Similarly, strain fields develop at defect sites.

[0103] A higher electrical conductivity, σ, also results from the formation of thermoelectric composite material 100. The higher electrical conductivity is primarily a result of increased charge carrier numbers, which more than accounts for a reduction in charge carrier mobility.

[0104] The overall effect in reducing thermal conductivity, κ, and increasing electrical conductivity, σ, is an increased figure of merit ZT = S²σT/κ.

[0105] Method 300’, in Fig. 4, is an alternative method for producing thermoelectric composite material 100. At step 302’, a powdered form of host matrix 102 is combined with a solution of organic carbonous material. The concentration of the carbonous material in solution, and the amount of solution added is dependent on the desired concentration of amorphous carbonous particles in the final material 100.

[0106] The mixture of powdered host matrix 102 and solution of organic carbonous material are heated at step 402 of method 300’ . The heat causes the solvent from the solution to evaporate and deposit organic carbonous material. The heat further causes the organic carbonous material to decompose into amorphous carbon particles, resulting in a mixture of amorphous carbon particles and powdered host matrix 102.

[0107] At step 304, the mixture of host matrix 102 powder and amorphous carbon particles is added to a die and heated under pressure at step 306 to amalgamate the powders into composite material 100. [0108] In some embodiments, the heating at step 402 is accomplished by the heat of step 306. That is, the heat at step 306 is used to evaporate the solvent and/or decompose the organic carbonous material into amorphous carbon particles before amalgamating the amorphous carbon particles and powdered host matrix 102 material.

[0109] In some embodiments, the solution of organic carbonous material comprises a sugar, being the organic carbonous material, dissolved in water, being the solvent. This solution is added to a powdered host matrix material such as bismuth antimony telluride (BST). The combined solution and BST are heated to not more than 500°C. At this temperature, the water evaporates leaving deposits of sugar. At the elevated temperature, the sugar decomposes to leave amorphous particles of carbon.

[0110] In some embodiments, the amorphous particles of carbon and powdered host matrix are graded before they are amalgamated into a solid at step 306. The grading helps ensure that the particles are of the desired dimensions.

Exemplary thermoelectric material using boron

[0111] Exemplary materials and their experimentally determined figures of merit will now be described.

[0112] Here we present a new strategy to significantly enhance PF while simultaneously reducing κL by incorporating a small weight fraction of non-toxic, light-weight amorphous nano-boron (B) particles into Bi_0.5SB_1.5Te₃ (BST). A record high ZT value of 1.6 at 375 K was achieved in the BST composite materials after incorporating boron (BST/B). The significant enhancement of TE performance in the BST/B can be attributed to the following synergistic strategies: i) significant reduction of κ by blocking ranges of spectrum phonon transport via various phonon scattering structures like amorphous B inclusions, in-situ high density of nano-precipitates and dislocation networks associated with severe lattice strain; ii) enhancement of σ and slight reduction of S which results in a higher power factor (PF). Our results represent an important step toward the widespread use of BST materials in TE power generation and cooling devices.

[0113] The exemplary materials comprise bismuth antimony telluride (BST) as the host matrix with amorphous particles of boron incorporated at concentrations of 0 wt%, 0.3 wt%, 0.4 wt%, 0.6 wt% (termed BST/B-0, BST/B-0.3, BST/B-0.4, BST/B-0.6).

The materials were fabricated by powder processing and followed by Sparking Plasma Sintering (SPS). In order to avoid the overestimation of ZT, all of the transport measurements were carried out in the same direction (parallel and perpendicular to the SPS pressing direction). Here the obtained TE performance in perpendicular to press direction will be discussed in detail.

[0114] The ZT of pristine BST ingot (P-BST) and BST/B samples are shown in Fig. 5a. Remarkably, the B addition significantly enhances TE performance and the maximum ZT of 1.6 at about 375 K in BST/B-0.6 sample was achieved. To ensure the reliability and repeatability of the experimental results, we provide the measurement results of the same sample by an independent third-party group (Tsinghua University). The excellent TE performance with ZT of 1.53 (Fig. 5a blue line with hexagon symbol) has been independently confirmed using the instruments of Tsinghua University, which is consistent with our experimental results when considering the existence of measurement errors.

[0115] To better highlight the advantages, as shown in Fig.5a, the ZT value (solid lines) are compared with other recently reported high performance BST systems (Poudel et al Science 2008, 320, 634, Kim et al. Science 2015, 348, 109, Deng et al. Energy Environ. Sci. 2018, 11, 1520, Pan et al. Energy Environ. Sci. 2019,12,624 - dashed lines). In contrast, the peak of ZT for our BST/B-0.6 material shifts to higher temperature and the average ZT is rather outstanding, reaching 1.5 over a broad temperature plateau from 300 to 475 K. Using the calculation method for device ZT and conversion efficiency by considering self-compatibility as described in Snyder et al. Energy Environ. Sci. 2017, 10, 2280, a record high TE device ZT of 1.2 and conversion efficiency of 11.3% are achieved in the BST/B-0.6 material when using a hot and cold source of 575 and 300 K, respectively. The performance of our BST/B material is greatly superior to current BST-based TE materials in both low- and mid- temperatures.

[0116] The temperature dependence of the TE properties for all samples are compared in Fig. 6. They all exhibit a typical degenerate semiconductor behaviour with σ decreasing monotonically with increasing temperature. Interestingly, the σ increases significantly with increasing nominal B content although B is an insulator. In particular, the room temperature σ is 850 S.cm-1 for P-BST, while it increases by 40% to 1460 S. cm-1 for BST/B-0.6.

[0117] According to the rule of mixtures in classical composites, it seems that σ should be between that of the two component materials. On that basis, the σ of the BST/B hybrid materials should be lower than that of BST/B-0 given the extremely lowσ of B albeit with the small weight fractions of B added. However, the σ of BST/B unexpectedly increases with increasing B content. To investigate the reasons for the increase, we investigated the Hall Effect to determine the carrier concentrations

and mobility We observed that the hole concentration of BST/B-0 sample is

higher than the pristine ingot (Table SI). The BST/B-0 sample was fabricated by powder processing followed by SPS processing using the ingot as raw materials.

[0118] So there should be some differences in the carrier density for the BST/B-0 and the pristine ingot. This can be related with the increased structural defects or small amount of Te evaporation during SPS processing. It should be noted that the BST/B composites exhibit a significant increase in

with the increasing B content. The enhanced carrier concentration by adding B may be associated with the enhanced lattice deficiencies. It is well known that the origin of hole concentration in ptype BST is mainly from the antisite defects. The generation of dense Sb-Te rich nano- precipitates induced by B inclusions could cause the deviation from the stoichiometry leading to the formation of more antisite defects; additionally, values of BST alloys

are sensitive to both chemical composition and processing-induced lattice defects. The similar results can be reported in the SiC-dispersed BST samples in which the concentration is increased from 1.8x 10¹⁹cm^-3 for pure sample to 3.39x 10¹⁹ cm^-3 for samples with 0.4 vol% SiC. Such structural defects are more likely to occur at interfaces between nanopreciptates and matrix, leading to the charge buildup at the interfaces, which is responsible for the increased carrier density. In comparison to the P-BST, about a 10% reduction in

is found in the BST/B-0 sample (Table S 1) -which can reflect the increased carrier-carrier scattering and the defect-related scattering. The of the BST samples with the addition of B shows a slow reduction in comparison with that of the pure BST sample, indicating that B inclusions have little influence on the carrier mobility. With increasing temperature towards 300 K, the carrier mobility of all samples is roughly a T^-1.5 dependent, indicative of the dominance of acoustic phonon scattering. Therefore, the improved σ for BST/B is mainly because the carrier concentration is sufficiently increased to compensate for the reduction of carrier mobility.

[0119] The Seebeck coefficients, S shown in Fig. 6b, are in good agreement with Hall measurements for the p-type samples. The magnitude of S initially shows an increase with increasing temperature, indicating the dominance of hole carrier transport. As the temperature increases further, S quickly decreases due to the intrinsic thermal excitation of minor electrons at elevated temperatures. The coexistence of majority and minor carriers can results in a bipolar effect. We can see that S decreases with the increasing B content, showing the inverse trend with respect to the σ. S can be estimated by the equation: where k_b is the Boltzmann

constant, e is the carrier charge, h is Planck’s constant, m* is the effective mass of the charge carrier, and n is the carrier concentration. Thus, the σ increases with carrier concentration but the S decreases.

[0120] The effective mass for BST/ B samples is about 1.1 ± 0.1 m_e, which is typical for BST samples. It is also noteworthy that the temperature corresponding to the peak value of S obviously shifts to higher temperatures for the B-doped (BST/B) samples and the decline of S value of BST/B specimens above 400 K is significantly slower than that of pristine ingot. It is well known that the bipolar effect is due to the ambipolar diffusion of electrons and holes. The minority carriers (electrons in this example) can thermally be excited across the band gap, which will degrade Seebeck coefficients. Therefore, the shift of peak S results from the reduced bipolar effect.

[0121] A plot of the calculated PF = S²σ is shown in Fig. 6c. The PF shows a considerable increase in the BST/B samples. The highest PF achieved at room temperature for the sample BST/B-0.6 is 45 μWcm^-1K^-2, which is about 13% higher than that of P-BST.

[0122] Fig. 6d plots the total thermal conductivity (κ) as a function of temperature for the BST/B materials. The κ of all samples initially decreases with increasing temperature before reaching a minimum value owing to the phonon Umklapp processes, and then increases rapidly with increasing temperature because of the additional contribution of minor carrier pairs excited at high temperature. In the case of the sample with a small amount of 0.6 wt% B, its κ shows a distinct decrease by up to 35% over the entire temperature range when compared with the P-BST sample.

[0123] Typically, the κ in BST materials consists of a lattice component a carrier

component (κ_e), and a bipolar component (κ_b). To examine the individual thermal components, we calculate κ_e according to the Wiedemann-Franz law which is expressed as κ_e = LσT, L is the Lorenz number which is estimated based on the SPB model. Assuming κ_b has no contribution to heat flow at low temperature, is

considered to be proportional to the reciprocal of temperature (1/T) above the Debye temperature. Thus, we can calculate at low temperature and extrapolate the value

corresponding to a temperature above room temperature. The temperature dependence of κ_tot-κ_e, κ_b and are shown in Fig. 7. The extremely low (κ_tot-κ_e) of 0.33 W.m- 1.K-1

at 300 K is achieved in the BST/B-0.6 sample which is about a 60% reduction when compared with the P-BST ingot. This value is comparable with that in the literature and very close to the theoretical amorphous limit (κ_min=0.31 W.m^-1.K^-1). It should be noted that introducing phonon scattering centers may potentially reduce carrier mobility. Thus the key to improving TE performance depends on if there is net gain between electrical and thermal transport. This can be evaluated by using the thermoelectric material quality factor (B), which is proportional to the ratio of The calculated thermoelectric quality factor is shown in Table S 1. We believe

the higher obtained quality factor is responsible for the high TE performance of BST/B nanocomposites.

[0124] To further understand the mechanisms of the obtained ultralow κ in the BST/B samples, detailed microstructural investigations were performed by transmission electron microscopy (TEM). Both high density dislocation networks and periodic dislocation arrays were commonly observed during the TEM examinations (Fig. 8a). The dislocations nucleate and pile up at grain boundaries, forming dislocation networks. Besides, dislocation tangles are found at the triple points of boundaries. High densities of dislocation networks and tangles associated with the severe strain in the matrix is rather effective for suppressing the phonon transmission.

[0125] Fig. 8d and Fig. 8e show some inclusions with diameters of 80-250 nm dispersed in the matrix. Energy-dispersive X-ray spectroscopy (EDS) and electron energy loss spectroscopy (EELS) confirmed that these were boron-rich (Fig. 9c, d). The EELS result shows the boron K edge at 188 eV. The matrix was confirmed to be BST. The high resolution image (Fig. 8f) shows an absence of long range order in the boron

SUBSTITUTE SHEET (RULE 26) phase indicating it to be amorphous. Probably because of the amorphous feature, interface mismatching is reduced, which does not degrade the electrical transport. Moreover, the big difference in mass density and sound velocity between the B and BST matrix will cause Kapitza thermal interfacial resistance which is effective for reducing κ.

[0126] High density of nanoscale precipitates are commonly observed in the BST/B- 0.6 matrix. Fig. 8b show a large number of spherical or elliptical nanoscale precipitates with the size in the range of 3-15 nm. We estimated the density of nanoscale precipitates is approximately 8.4x 10²⁴ m^-3 and their average size is 6 nm which is used for the calculation of the lattice thermal conductivity. The EDS line scan exhibits an increase in the Sb and Te from the precipitates compared with the matrix regions. From the EDS spectrums of precipitates and matrix (Fig. 9a, b), we can conclude that the nano-sized precipitates are Sb and Te rich phases with an atomic ratio of Sb/Te close to 1, while the corresponding (Sb+Bi)/Te ratio in the matrix is about 0.67, indicating that Sb substitutes for Te. Notably, the in-situ nano precipitates have also been found in other BST materials processed by unique methods such as ball milling and two-step sintering. Fig. 8c shows atypical phase contrast HAADF (high-angle annular dark- field) image containing the Moire Fringe of nanoscale precipitate with the interface between the precipitates and matrix. The HAADF image reveals that the precipitates are in fact due to compositional modulation-where the bright lines are richer in heavy atomic number materials than the dark bands. It can be noted that the lattice strains slightly increase with the increasing B content, indicating that the lattice strains should also contribute to minimizing the lattice thermal conductivity

The lattice strains have been demonstrated to be particularly effective for minimizing the

The nano- precipitates as well as internal strains can obviously act as scattering centers for phonons, which in turn will significantly reduce κ.

[0127] The dominant microstructures in the BST/B feature crystal defects such as dense dislocation array, nano-precipitates accompanied with lattice strains, which are less commonly observed in B-free samples . It indicates that the B nano-inclusions could modify the microstructures of BST. With regard of the dislocation array formation, firstly, it may be related with applied pressure during SPS processing. It has been reported that high density of dislocations are commonly present in any samples processed by SPS. Secondly, the inclusion of B nano-particle may induce lattice strain and thus lead to the formation of dislocations. Thirdly, there is a big difference between the thermal expansion coefficients of boron and BST, which means that it will generate strain fields around the inclusions, and this may lead to dislocations forming-localized to the vicinity of B inclusions. In terms of the generation of nano-precipitates, it has been reported that the bulk nanostructured BST could achieved by the “top-down” processing technologies, involving such methods as ball milling (BM), melt spinning (MS) , hot deformation and other advanced techniques due to the recrystallization during the sintering process. Considering that SPS is a fast-sintering method, the in-situ nanostructures may generate during the sintering process. We think the B dispersion may promote the generation of nanoscale precipitates. Firstly, precisely controlling the homogeneous nucleation density and crystal growth rate during recrystallization process would be very crucial to achieve high density of nanostructures. The light- weight element, boron, may act the role of nucleating agent which can modify the microstructure by accelerating the rate of nucleation and suppressing the crystal growth during the recrystallization process. Additionally, the presence of B may cause the Sb- Te phase to become thermodynamically favoured over BST-due to lattice strain, bonding effects etc.

[0128] Nevertheless, these defects are responsible for the significant differences in electrical and thermal transport properties compared to the sample without B inclusions.

[0129] In order to understand the obtained κL, the Debye-Callaway model was used. The total relaxation time was estimated by taking account of the Umklapp

process, point defects

grain boundaries dislocation cores and strain and

nanostructures

As shown in Fig. 7c, the calculated

values agree well with the experimental data for both the ingots and the BST/B-0.6 samples. Furthermore, the calculated frequency-dependent is shown in Fig. 7d. Firstly, atomic-scale solid-

solution alloying, severe strain, and other point defects would be effective scattering centers for high frequency phonons with comparably short wavelength.

[0130] Secondly, dense dislocation arrays, nanostructures, amorphous B, and related mismatched phonon modes between them would strongly scatter phonons with short and medium mean free paths. Thus, they impede much of the heat flow in the material. Thirdly, the additional and remaining reduction in

may come from the scattering of low frequency phonons with longer mean free paths through mesoscale grain boundaries. Therefore, the extraordinarily low

achieved in the BST/B can be primarily attributed to the phonon scattering by a combination of multiscale hierarchical scattering architectures.

[0131] In a narrow band-gap semiconductor, bipolar diffusion is detrimental to the performance at elevated temperature because it can make a considerable contribution to the thermal conductivity and degrade the Seebeck coefficient. It should be noted that κ_b in our BST/B samples is significantly reduced. The suppression of bipolar effects can be mainly attributed to the increased hole concentrations. Also the reduced bipolar effect may be resulted from the existence of an interfacial potential that scatters more electrons than holes. In summary, we have demonstrated microstructural design with superior thermoelectric properties by introducing boron inclusions into BST materials. The hierarchical microstructures significantly minimize the thermal conductivity through multiple phonon scattering mechanisms; and simultaneously enhance power factor due to the optimization of electrical transport. Because of decoupling the electrical- and thermal transport, a record high ZT of 1.6 at 375 K is achieved in the BST/B-0.6 composite materials, making it unique in being capable of a high calculated conversion efficiency of 11.3% in a device implementation.

Exemplary thermoelectric material using carbon

[0132] Here below an exemplary composite thermoelectric material 100 wherein the amorphous particles are amorphous carbon particles. The amorphous carbon particles are sections of carbon microfiber. [0133] It is demonstrated that by adding a small amount of carbon microfiber to Bi_0.5SB_1.5Te₃ (BST), we can enhance the TE performance significantly. We demonstrate that BST composites incorporating carbon microfiber (BST/CF) exhibit a significant reduction in κ and importantly, maintain a high PF. As a result, the highest ZT value of ~ 1.4 at 375 K and an average ZT of 1.25 between 300 and 500 K is obtained for BST alloys by incorporating an optimum amount of carbon microfiber. The significant reduction in κ in BST/CF composites can be attributed to the boundary modification of nanoscale and mesoscale interfaces. We show that a large interfacial thermal resistance exists between the carbon microfiber and the BST matrix, which is responsible for the ultra-low κ. At the same time, the incorporation of CFs has only a very small effect on electron transport due to the high conductivity of CFs, thus maintaining high PF. To confirm the enhanced ZT of BST/CF materials, a TE unicouple device consisting of p- type BST/CF composite materials and commercial n-type materials (referred as BST/CF unicouple) was prepared to evaluate its cooling performance. As a result, BST/CF unicouple has generated a large cooling temperature difference of 34-46 K at an operating temperature of 299-375 K and a current of 4.5 A, which is 1.6~1.9 times higher than the unicouple device made from commercially available p-type and n-type TE materials (referred as REF unicouple). In addition to excellent and repeatable TE performance, BST/CF composites show improved mechanical properties over their CF free counterparts.

[0134] Highly dense samples of BST/CF incorporating 0, 0.3, 0.4, 0.5, and 0.6 wt% carbon microfiber were fabricated by spark plasma sintering (SPS). These are denoted as: BST/CF00; BST/CF03; BST/CF04, BST/CF05, and BST/CF06, respectively. In this work, we used a commercial BiSbTe ingot (BST ingot) fabricated by a unidirectional solidification method as a raw material and a reference sample.

[0135] Powder X-ray diffraction (XRD) of BST/CF and ingot samples matches the standard Bi_0.5SB_1.5Te₃ pattern well.

[0136] Taking into account the anisotropic properties of the sample, we measured electrical and thermal transport along the in-plane and cross-plane directions. The main difference between the two directions is that the in-plane electrical and thermal conductivity are higher than those in the cross-plane direction. The thermoelectric properties obtained along the in-plane direction will be discussed in detail. The temperature-dependence of the thermoelectric ZT is shown in Fig. 10a. The peak ZT of the BST ingot fabricated by using the unidirectional solidification method at 340 K is 1.00. After powder processing and SPS sintering, the ZT value was enhanced significantly to 1.1 at 35 OK. This indicates that SPS reprocessing can cause an increase in ZT, which is consistent with reports from other groups. In fact, various manufacturing methods such as melt spinning (MS), hot deformation (HD) and self- propagating high temperature synthesis (SHS) have proven to be effective for ZT improvement of BST materials. More importantly, the addition of a small amount of CF significantly increased the ZT by 40% compared to the BST ingot sample. The maximum ZT of 1.4 was obtained at 375 K for BST/CF04 sample. It is worth noting that the peak temperature of ZT in BST/CF relative to pure BST, shifts to a higher temperature. This is advantageous for obtaining a high average ZT over a wide temperature range.

[0137] To highlight the outstanding advantages of our materials, we compare their ZT values with some of the typical p-type BST materials reported recently. As shown in Fig. 10b, our samples show a comparable ZT with most nanostructured BST materials. Although the BST/CF peak ZT is lower than that of Te-BST, it maintains a ZT above 1 over a wide temperature range of 300-500 K. More importantly, the high TE performance of BST/CF sample could be reproduced by a third independent group (Tsinghua University Lab) as shown in Fig. 10b. The average ZT (ZT_ave) is estimated by integrating the area under the ZT curve according to the relationship

[0138]

[0139] Where T_h and T_c are the temperatures of the hot and cold sides, respectively. [0140] To evaluate the energy conversion efficiency of BST/CF samples, we calculated the efficiency using the method proposed by Snyder et al. Energy Environ. Sei. 2017, 10, 2280. This method can accurately calculate the efficiency by considering the thermoelectric compatibility. It assumes that a TE material consists of a series of segments, each having a different local temperature and thermoelectric properties. The optimum efficiency (η) can be obtained by tuning the relative current density. Using the maximum conversion efficiency, the device ZT_dev (as opposed to the material TE ZT) can be defined as:

[0141]

[0142] Considering a hot side of 500 K and a cold side of 300 K, the calculated η , ZT_ave and ZT_dev are shown in Fig. 10c. We can see that the calculated average ZT of BST/CF is 1.25, which is much higher than for all prior p-type TE materials currently reported for application in low-grade waste heat recovery. Furthermore, at a temperature difference of 200 K, the maximum η exceeds 9%, corresponding to a device ZT of 1.2, which is comparable to the ZT_dev value of nanostructured Bi₂Te₃- based materials reported to date.

[0143] To further evaluate the effect of the significantly enhanced ZT value on cooling performance, two unicouple cooling modules consisting of one p and one n leg are fabricated to measure the temperature difference. The measurement setup schematic is shown in Fig. 11a. In the TE unicouple module (referred as BST/CF), BST/CF04 composite material prepared in this work and commercially available n-type bismuth telluride-based material are chosen as p-type leg and n-type leg, respectively. For comparison of the cooling performance, another TE couple (referred as REF) is also constructed by using the commercial p-type and n-type bismuth telluride-based material. The cold-side (T_c), cooling temperature difference (ΔT_c, temperature drop from initial cold-side temperature), hot-side temperature (T_h) and heating temperature difference (ΔT_h, temperature increase from initial hot-side temperature) of both two TE unicouples are measured under different working current (I) and initial hot-side temperature (T_h). The measured results are shown in Fig. 11, Fig. 12, and Table S2.

Table S2. The measured initial hot-side temperature (T_h0), initial cold-side temperature

(T_c0), minimum steady state cold-side temperature (T_cmin), maximum steady state hot- side temperature (T_hmax), cooling temperature difference (ΔT_c, temperature drop from initial cold-side temperature) at cold side, heating temperature difference (ΔT_h, , temperature increase from initial hot-side temperature), and the total temperature difference ΔT (ΔT =ΔT_c +ΔT_h).

[0144] Fig. 11b shows the relationship between the cooling ΔT_c and applied current when the starting hot side temperature fixed at room temperature (RT=299 K). It can be found that the cooling ΔT_c of BST/CF TE unicouple are significantly larger than that of REF TE unicouple under different current. Especially in the case of large current, the cooling performance of BST/CF device is obviously better than the REF device. For example, the cooling ΔT_c value of the REF module is 20.7 K, whereas cooling ΔT_c of the BST/CF module is as high as 34 K when the applied current is 4.5 A. Generally, the cooling ΔT_c increases with the increased applied current values and eventually reaches saturation. The current dependence of cooling ΔT_c curves shown in Fig. 1 lb indicates that the current saturation value (I_s) corresponding to the maximum cooling ΔT_c value of REF module is 3.5A, which is smaller than that of REF module. This can be attributed to the different thermoelectric performance of BST/CF and commercial BST materials, which determines the relationship between the Peltier effect, Fourier heat, and Joule heat. Fig. 12 presents the time-dependent cooling side T_c and the hot-side T_h for REF and BST/CF modules at a fixed current value of 3.5~4.5 A at room temperature. Because there is a heat sink (aluminium fins) connected to the hot side of TE unicouple, the heating temperature difference (ΔT_h) at hot side (T_h) induced by current injection is very small. Once the current switch is turned on, however, cooling side T_c first declines sharply in a short time due to the Peltier effect of TE elements and then gradually reach thermal equilibrium. As shown in Fig. 12a, the cold side T_c for BST/CF module drops from initial temperature of 299 K (25.8 °C) to the least temperature of 268.49 K (-4.67 °C) at I=3.5 A, but for REF module, the obtained least cold-side T_c is only 275.98 K (2.82 °C) under the same condition. It should be noted that if the current is further increased from 3.5 to 4.5 A, BST/CF module will generate a larger temperature drop (34 K) and the minimum cold-side temperature can be as low as 265.51 K (-7.65 °C). However, unlike BST/CF module, it is found that when I >3.5 A, further increase the current for the REF module will result in the increase both in the cold-side T_c and hot-side T_h, which indicates that 3.5 A is the optimal current for REF module to reach the maximum cooling temperature. The cooling performance of TE cooler is mainly depends on Peltier cooling and Joule heating induced by current, and the internal heat conduction. To obtain maximum cooling capacity, one should increase the Peltier effect, and decrease the thermal conductance and joule heat. On one hand, the internal electrical resistance of our BST/CF materials is lower than that of commercial BST ingots in this work. This means that BST/CF module will generate less joule heat than the REF module, which is why we observed that the heating ΔT_h at hot side of BST/CF module is smaller than that of REF module (Fig. 12 and table S2). On the other hand, the low thermal conductivity of BST/CF materials will result in the increased thermal resistance of modules, which has positive effect on the cooling performance.

[0145] Additionally, the cooling performance of TE modules are closely associated with the operating temperature due to the temperature dependences of the TE elements’ properties. Fig. 11c illustrates the cooling ΔT_c as a function of its hot-side operating temperature. For REF module consisting of commercial p-type and n-type Bi₂Te₃-based materials, the ΔT_c firstly increases and tend to flatten as the hot-side temperature rise from 299 to 375 K. As a result, the maximum cooling ΔT_c of 30.7 K is obtained at the hot side temperature of 350 K and I=3.5 A. It is interesting to note that the hot-side T_h dependent of cooling ΔT_c for our BST/CF TE couple is significantly higher than that of REF TE couple at all the operating hot-side temperature. Furthermore, different from REF module, BST/CF TE module shows the increase in cooling ΔT_c with the increased hot-side temperature. This should be because the peak ZT of BST/CF material shift to higher temperature compare to that of BST ingot. We can note that the BST/CF module results in the cooling ΔT_c of 46 K at T_h=375 K and I=4.5 A, which is 1.9 times higher cooling than the REF module under the same operating condition. The superior cooling performance of BST/CF module relative to the REF module also directly proves that our BST/CF materials have higher TE performance than commercial BST materials.

[0146] The temperature dependence of the electrical and thermal transport properties of the BST ingot and BST/CF samples are shown in Fig. 13. The conductivity (σ) of all samples (Fig. 13a) decreased with increasing temperature, indicating degenerate semiconductor behaviour. All the BST/CF samples after SPS processing show an increase in σ compared with the unprocessed BST ingot. By incorporating a small a small amount CFs (0.3-0.4 wt %), the electrical conductivity of the sample is significantly increased, compared with that of the sample without CF (BST/CF00). However, with increasing content of CFs when exceeding 0.4 wt%, the electrical conductivity decreases. Fig. 13b represents the temperature dependence of the Seebeck coefficient (S) for all samples. The S values of all the samples first increase with temperature before reaching a saturation point, and then they decrease because of the intrinsic thermal excitation of minority carrier electrons at high temperature. The S values of the BST/CF samples are higher than that of the ingot for T>400 K, and their peak values shift to higher temperature. The power factors (PFs) are shown in Figure 4c. It can be found that the PFs of the BST/CF samples are higher than for the ingot sample by 5~20%. The highest PFs value of 46 W.m^-1.K^-1 at room temperature could be obtained in the BST/CF04 samples.

[0147] Fig. 13d depicts the temperature dependence of the total thermal conductivity (κ) for the BST ingot and the BST/CF samples. As temperature increases, the κ for all samples initially decreases due to the phonon Umklapp processes, reaching a minima value, and then increases rapidly because of the thermally intrinsic excitation of electrons at elevated temperature. Interestingly, incorporating CF could significantly lower the κ value. At 300 K, the κ value of BST/CF05 is about 1.04 W.m^-1.K, about 17% reduction relative to the pristine ingot (1.26 W.m^-1.K at 300 K). Generally, the κ in BST materials consists of a lattice part an electronic part (κ_e), and a bipolar part

(κ_b). The κ_e can be calculated based on the Wiedemann-Franz law, namely κ_e= LσT, where L is the Lorenz number which can be calculated based on the SPB model.

Assuming that κ_b makes no contribution to κ at low temperature and

is proportional to T^-1 above the Debye temperature, we can obtain the

at low temperature according to the relationship κ- κ_e, and then extrapolated the

value at high temperature according to The temperature dependence of κ- κ_e,

and κ_b are shown in

Fig. 14. It can be noted that the values are significantly reduced in the CFs

incorporated BST materials. Moreover, in the BST/CF samples, we found that κ_b is also suppressed at higher temperature as shown in Fig. 14c.

[0148] The temperature dependence of the carrier concentration (p) and carrier mobility (μ) were determined from the Hall measurements. The p of all samples show a slowly decrease in the range of 10-340 K. The p roughly follows the T^-1.5 temperature dependence, indicating a dominant charge scattering by acoustic phonons for all the materials. Fig. 15a shows how with increasing additions of CF, the p increases slightly. The μ weakly decreases with the small amount of addition of CFs, and then rapidly reduces by 20% compared to that of the starting materials when the content of CFs up to 0.6 wt%. Because of the increased carrier density compensate the reduced carrier mobility; we observed the improved σ in the BST/CF samples. One possible reason for the variation in n can be the increased lattice defects induced during SPS processing and the addition of CFs. It is noted that the antisite defects play an important role in tuning the carrier density. We can see extensive in-situ nanoprecipitates and carbon inclusions (Fig. 16) decorating the grain boundaries. The lattice defects are more likely to exist at the interfaces between inclusions and matrix, leading to the increase of the carrier density. Meanwhile, the CFs network in the BST matrix could provide the high mobility path of carriers, thus adding a tiny amount of CFs has little effect on the carrier mobility. However, as the content of CFs further increases, the density of grain boundary would lead to enhancing the carrier scattering, which is responsible for the reduction in carrier mobility when incorporating excessive CFs. The similar results has also be reported in the SiC-dispersed BST samples in which the concentration is increased from 1.8x10¹⁹cm^-3 for pure sample to 3.39x10¹⁹ cm^-3 for samples with 0.4 vol% SiC. Based on the effective mass model, we obtained the Pisarenko plot (S vs The experimentally measured S values for the ingot and BST/CF sample can be explained by an effective mass of ≈ 1.1 +/- 0.1 m_e, which is typical for BST.

[0149] The bipolar carrier transport in bulk, narrow band-gap semiconductors severely constrains the ability to improve ZT at elevated temperature as it suppresses the S value and contributed additional thermal conductivity. The thermally excited minority carriers at higher temperature would result in the abrupt decrease in the S value due to the opposite contributions of holes and electrons to S. As the hole concentration of the BST/CF sample increases, the corresponding electron carrier concentration will reduce. Thus the electrons’ negative contribution to S will be suppressed. That is why we observed the S value of the BST/CF samples to be higher than that of the BST ingot sample above 450 K. Furthermore, it should be noted that κ_b in BST/CF samples is greatly suppressed, indicating the suppression of bipolar carrier diffusion. This can occur due to the enhanced hole carrier concentration. Due to the suppression of bipolar carrier transport, the average ZT in the BST/CF composites are greatly improved over a wide temperature range.

[0150] Fig. 15b shows the relationship between the real lattice thermal conductivity value and the CF content at 300 K. It can be clearly seen that the decreased by

as high as 34% (from 0.84 to 0.55 W.m^-1K^-1) as the contents of CF increased from 0 to 0.5 wt%. In a semiconductor, the maximum ZT could be determined by the thermoelectric material quality factor, B, defined as

[0152] Where is the weight mobility with the units of m²V^-1S^-1,

and m_e and m* are the electron rest masses and the density of states (DOS) effective. From the equation (3), the quality factor, B, is proportional to which can be

used to evaluate if there is net gain in thermoelectric performance. Fig. 15c shows the calculated thermoelectric quality factor (B) and ZT as a function of CF contents. We can conclude that a higher B obtained by incorporation of CFs, results in an ultra-high in an ultra-high ZT in BST samples containing CFs compared to a sample without CFs.

[0153] To understand the origin of low we carried out the microstructural

characterizations. Fig. 16 shows the microstructures of a typical BST/CF05 lamella produced using focused ion beam milling. SEM backscattering image shown in Fig. 16a detect a large amount of carbon fibers (dark against the bright background) that uniformly dispersed within the BST matrix without agglomeration. The size of carbon fibers ranges from several hundred nanometres to several micrometres in diameter and length. From the low magnification bright field (BF) the transmission electron microscope (TEM) image as shown in Fig. 16b, we observed that the carbon fiber particles (bright areas) and some of the nano-sized precipitates were randomly distributed in the BST matrix (circled). Energy dispersive X-ray spectroscopy (EDS) mapping was used to confirm the presence of the carbon phase. The composition of the nanoprecipitates was found to be an Sb-rich phase. The size of the grains adjacent to the carbon phase regions was smaller than that of the grains distant from such regions. This indicates that the carbon inclusion can refine the grain size of the BST, which is consistent with scanning electron microscopy (SEM) analysis. The high resolution BF image in Fig. 16c shows the interface between the BST matrix (left) and the carbon phase. The BST has good crystallinity while the carbon phase is weakly graphitic, with highly disordered planes parallel to the interface. The interface between the BST and the carbon fiber is abmpt, which is beneficial for increasing the thermal resistance and thus lowering the thermal conductivity. Fig. 16d shows a high-angle annular dark-field (HAADF) scanning TEM (STEM) image of the interface between the Sb-rich particle and BST matrix. The Sb-rich precipitates show a large misfit with the lattice structure of the BST matrix, resulting in strains at the interfaces, effectively decreasing the thermal conductivity. The nanoscale Sb-rich particles are commonly observed (Fig. 16e), and in this instance small voids (dark) are also present. Detailed imaging of such particles (Fig. 16f) shows atomic number contrast variation and lattice spacing differences. The insert FFT image shows weak additional reflections arising from an unaligned phase. There is some misalignment with the lattice structure of the matrix across the interface due to the local lattice strain. The different size of phase interfaces and the related strains can act as effective scattering centres for phonons.

[0154] Ultralow κL values can arise from lattice softening and full-spectrum phonon scattering due to defects of different dimensionalities: 0-D atomic defects, 1-D dislocations and 2-D boundaries (including intrinsic grain boundaries and phase boundaries). The potential phonon scattering mechanisms in our BST/CF sample are illustrated schematically in Fig. 17. The atomic defects and in-situ nanoscale precipitates would effectively scatter phonons with short and medium mean free paths. The mesoscale phase and intrinsic boundaries can further scatter the phonons with longer mean free paths. Grain-boundary phonon scattering has been demonstrated to be important in improving the thermoelectric performance of Bi₂Te₃-based alloy. The interfacial thermal resistance, which is defined as the ratio of the temperature discontinuity at an interface to the heat flux flowing across the interface, plays a crucial role in minimizing the

With an effective medium approximation, the of a

poly crystalline solid can be expressed as:

[0155] (4);

[0156] Where R_k is the interfacial thermal resistance and d_eff is the grain size or effective interface density.

[0157] The incorporation of CF associated with the refinement of grain size and the generation of nanoscale precipitates and CF-BST interfaces would lead to the increase in the grain boundary density. The phonons with mean free paths larger than the boundary size would therefore scatter strongly. Furthermore, due to the severely mismatched phonon structures between the carbon and the BST materials, the interfacial thermal resistance ( R_k) is considered to be extremely large. The interfacial thermal resistance can be predicated by several methods such as the commonly used the acoustic mismatch model (AMM) and the diffuse mismatch model (DMM). The R_k of CF-BST interfaces are roughly predicated to be 1.4× 10^-6 and 1.2× 10^-6 m₂K.W^-1 by using AMM and DMM models, respectively, which are much larger than that of pure grain boundary resistance. Thus, the increased interfacial thermal resistance should be responsible for the ultralow

values found in our BST/CF samples.

[0158] The incorporation of CF into BST materials not only significantly improves the TE performance, but also enhances the mechanical properties. Fig. 18 shows the Vickers hardness and fracture toughness of the BST/CF samples compared with the sample (MS40) fabricated by melt spinning in Wu et al. Adv. Mater. 2017, 29, 1606768. The hardness and the fracture toughness of pure BST is found to be about 0.3 GPa and 0.85 MPa.m^1/2, respectively. Remarkably, the hardness and fracture toughness of the BST/CF06 samples can reach as high as 0.8 GPa and 1.4 MPa.m^1/2, which is higher than that for pure BST and MS40.This indicates that the CF could be used as a reinforcing agent to effectively improve the mechanical durability. CF has superior mechanical properties and it can generate the bridging effect, which burden additional load when a crack is induced and encounters the CF, thus preventing the crack propagation and leading to the enhanced mechanical properties. The mechanical properties are very desirable in addition to ZT for TE application since the TE modules usually work under cyclic heat stress and high current concentration.

[0159] It will be appreciated by persons skilled in the art that numerous variations and/or modifications may be made to the above-described embodiments, without departing from the broad general scope of the present disclosure. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive.

Claims

CLAIMS:

1. A composite thermoelectric material comprising: a host matrix comprising a semiconducting thermoelectric material; and amorphous particles dispersed throughout the host matrix, wherein the host matrix includes a polycrystalline structure nucleated by the amorphous particles.

2. The material of claim 1 wherein the semiconducting thermoelectric material comprises bismuth, tellurium, and antimony.

3. The material of claim 2 wherein the amorphous particles are amorphous boron.

4. The material of claim 3 wherein the particles of amorphous boron are between

1 nm and 300 nm in diameter.

5. The material of claim 3 wherein the host matrix further comprises a plurality of spatially distributed phases having higher concentrations of antimony and tellurium, the plurality of spatially distributed phases having dimensions between 1 and 50 nm.

6. The material of any one of claims 2 to 5 wherein the ratios of bismuth (Bi) to antimony (Sb) to tellurium (Te) is given by Bi_xSb_2-xTe₃ where x=0 to 2.

7. The material of any one of claims 3 to 6 wherein the boron is at a concentration of 0 up to 5 wt%.

8. The material of claim 1 or claim 2 wherein the amorphous particles are amorphous carbon.

9. The material of claim 8 wherein the amorphous carbon is derived from carbon fibres or carbon black.

10. The material of claim 9 wherein the carbon fibres are uniformly distributed through the host matrix.

11. The material of claim 10 wherein the carbon fibres have a cross-sectional diameter of 50 μm or smaller.

12. The material of claim 11 wherein the carbon fibres have a length of 2 nm to

1 cm.

13. The material of claim 8 wherein the particles of amorphous carbon have dimensions of a 1 nm to 50 μm.

14. The material of claim 13 wherein the amorphous carbon particles are derived from an evaporated liquid source of carbon.

15. The material of claim 8 wherein the amorphous carbon particles are derived from an organic carbonous solid.

16. The material of claim 14 wherein the liquid source of carbon comprises a solution of an organic carbonous solid in a solvent.

17. The material of any one of the preceding claims further comprising a network of defects.

18. The material of any one of the preceding claims further comprising sharp interfaces between the host matrix and amorphous particles.

19. The material of claim 1 wherein the amorphous particles comprises a mixture of amorphous boron and amorphous carbon.

20. The material of claim 19 wherein the amorphous carbon is derived from an organic carbonous solid or liquid.

21. The material of claim 1 wherein the amorphous particles comprise amorphous boron particles coated in amorphous carbon in a core-shell structure.

22. The material of claim 1 wherein the amorphous particles comprise amorphous carbon particles coated in amorphous boron in a core-shell structure.

23. A method of producing a composite thermoelectric material comprising the steps of: combining powdered constituents of a host matrix and amorphous particles, wherein the host matrix comprises a semiconducting thermoelectric material; adding powders to a die, wherein the die determines the final shape and dimensions of the composite material; and applying heat energy and pressure to the powders in the die to amalgamate the powders into the composite material.

24. A method of producing a composite thermoelectric material comprising the steps of: combining powdered constituents of a host matrix and a solution of organic carbonous material, wherein the host matrix comprises a semiconducting thermoelectric material; heating the combined host matrix powder and solution to evaporate solvent from the solution and decompose the organic carbonous material to amorphous carbon particles, the amorphous carbon being distributed in the powdered constituents of the host matrix; adding the combined host matrix powder and amorphous carbon particles to a die, wherein the die determines the final shape and dimensions of the composite material; and applying heat energy and pressure to the powders in the die to amalgamate the powders into the composite material.

25. The method of claim 24 wherein the temperature is below 500 Celsius.

26. The method of claim 24 or 25 further comprising grading the combined amorphous carbon particles and powdered constituents of the host matrix before adding to the die.