Literature DB >> 31459118

All-Oxide Thermoelectric Module with in Situ Formed Non-Rectifying Complex p-p-n Junction and Transverse Thermoelectric Effect.

Nikola Kanas1,2, Michael Bittner2, Temesgen Debelo Desissa3, Sathya Prakash Singh1, Truls Norby3, Armin Feldhoff2, Tor Grande1, Kjell Wiik1, Mari-Ann Einarsrud1.   

Abstract

All-oxide thermoelectric modules for energy harvesting are attractive because of high-temperature stability, low cost, and the potential to use nonscarce and nontoxic elements. Thermoelectric modules are mostly fabricated in the conventional π-design, associated with the challenge of unstable metallic interconnects at high temperature. Here, we report on a novel approach for fabrication of a thermoelectric module with an in situ formed p-p-n junction made of state-of-the-art oxides Ca3Co4-x O9+δ (p-type) and CaMnO3-CaMn2O4 composite (n-type). The module was fabricated by spark plasma co-sintering of p- and n-type powders partly separated by insulating LaAlO3. Where the n- and p-type materials originally were in contact, a layer of p-type Ca3CoMnO6 was formed in situ. The hence formed p-p-n junction exhibited Ohmic behavior and a transverse thermoelectric effect, boosting the open-circuit voltage of the module. The performance of the module was characterized at 700-900 °C, with the highest power output of 5.7 mW (around 23 mW/cm2) at 900 °C and a temperature difference of 160 K. The thermoelectric properties of the p- and n-type materials were measured in the temperature range 100-900 °C, where the highest zT of 0.39 and 0.05 were obtained at 700 and 800 °C, respectively, for Ca3Co4-x O9+δ and the CaMnO3-CaMn2O4 composite.

Entities:  

Year:  2018        PMID: 31459118      PMCID: PMC6644604          DOI: 10.1021/acsomega.8b01357

Source DB:  PubMed          Journal:  ACS Omega        ISSN: 2470-1343


Introduction

Energy harvesting from waste heat from, for example, combustion engines or metallurgical processes or other industry has great potential for energy savings and emission reductions. Thermoelectric generators (TEGs) present a promising technology for such energy recovery.[1] A TEG is based on two dissimilar materials (p- and n-types) connected together and when exposed to a temperature gradient, converts thermal energy into electrical energy.[1,2] Power density (Pdensity) and the figure of merit (zT) are used to evaluate TE modules and their consisting p- and n-type materials, respectively. Figure of merit for any p- or n-type material can be calculate usingwhere S, σ, and κ represent Seebeck coefficient, electrical conductivity, and thermal conductivity, respectively.[3] In a conventional module (π-design), p- and n-type materials are connected electrically in series and thermally in parallel, with conductive metallic interconnects between legs (mostly Ag or Au).[4,5] Conventional modules normally suffer from limited temperature, which obstructs the high-temperature advantage of oxides. Furthermore, conventional modules demonstrate short lifetimes because of the instability of the metallic interconnects and even at low temperature resulting in a decrease in power output as a function of time.[6] All-oxide TE modules with direct oxideoxide p–n junction would therefore be beneficial. However, such direct p–n junction will normally have high resistance because of charge carrier depletion in the space charge regions.[7] Shin et al. demonstrated for the first time the prototype of a direct p–n oxide module using Li-doped NiO and (Ba, Sr)PbO3, and 14 mW at ΔT = 552 K was achieved.[8] Later, Hayashi et al. reported a stacked module with direct p–n oxide junctions based on p-type (La1.97Sr0.03)CuO4 and n-type (Nd1.97Ce0.03)CuO4, where the maximum power density obtained from 25 pairs was 40 mW/cm2 at 400 °C and ΔT = 360 K.[9] Moreover, Funahashi et al. demonstrated a module using a nonoxide p-type Ni0.9Mo0.1 and n-type La0.035Sr0.965TiO3,[10] where the maximum obtained power density from 50 pairs was 450 mW/cm2 at ΔT = 360 K. Furthermore, Chavez et al. showed another concept of using large-area p–n junctions (containing no insulator) as a TE module where 1.3 mW was generated from one pair at ΔT = 300 K.[11] The chemical compatibility, long-term stability, and electrical performance of the p–n junctions/modules were not considered in these studies. Ca3Co4–O9+δ (CCO) has a misfit-layered complex crystal structure containing triangular CoO2 and rock-salt Ca2CoO3 layers.[12] CCO shows p-type conductivity and exhibits excellent thermoelectric (TE) power at elevated temperatures in ambient air.[13] CaMnO3−δ (CMO) has a perovskite structure and exhibits n-type conductivity resulting from an intrinsic oxygen deficiency. By introducing a secondary CaMn2O4 spinel phase, the TE properties of the CaMnO3−δ–CaMn2O4 composite are improved compared with single phase CaMnO3−δ, as observed in our research group. TE properties of these oxides can further be improved by doping or co-doping.[14] So far, only a few reports on conventional modules based on the CCO–CMO-based system are available.[15−22] Besides high-temperature TE modules, there are recently reported promising low-temperature organic-based TE modules, containing flexible layered design.[23,24] Here, we report on high-temperature TE performance for an all-oxide TE module based on the system p-type Ca3Co4–O9+δ (CCO) and n-type CaMnO3CaMn2O4 (CMO-composite). LaAlO3 (LAO) was selected as the electrical insulating component because of its ferroelastic properties[25] and high thermal expansion coefficient (TEC) compared with other potential insulating oxide candidates. Undoped CCO and a CMO-composite were selected as model materials in this work focusing on developing processing methods of the layered module and thereby analyzing the effect of a direct oxideoxide p–n junction at high-temperature on TE performance of the module. The performance of the module critically depends on the properties of an in situ formed complex p–p–n junction, which was studied with respect to stability, interdiffusion, compatibility, and electrical conductivity. Finally, we report on an environment friendly processing method for all-oxide TE devices based on aqueous tape casting and co-sintering of the three materials using spark plasma sintering (SPS).

Results

Powder Characteristics

X-ray diffraction (XRD) patterns of CCO, CMO-composite, and LAO powders are presented in Figure a, confirming the phase purity of CCO and LAO as well as the composite nature of the CMO-composite (CaMnO3 with minor amount of CaMn2O4 secondary phase). XRD patterns of polycrystalline CCO prepared by SPS are also included in Figure a, both parallel and perpendicular to the pressing direction. A high degree of texture is observed (c oriented parallel to the pressing direction) as well as small amounts of Ca3Co2O6 and Co3O4 secondary phases. Fine-grained powders of the three starting oxides are confirmed from the scanning electron microscopy (SEM) images in Figure b. Sintering curves for the CCO, CMO-composite, and LAO powders shown in Figure c demonstrate the onset of sintering at around 650, 900, and 1000 °C for CCO, CMO-composite, and LAO, respectively. TECs of CCO (perpendicular and parallel to pressing direction during SPS), CMO-composite, and LAO are summarized in Table . CCO and CMO-composite show similar TECs while LAO has a significantly lower value.
Figure 1

(a) XRD patterns of CCO, CMO-composite, and LAO powders as well as a sample of CCO made by SPS in the directions parallel and perpendicular to the pressing direction, (b) SEM micrographs, and (c) sintering curves of CCO, CMO-composite, and LAO powders produced by spray pyrolysis. Diffraction lines from articles, LAO,[25] CCO,[26] and CMO-composite powder (CaMnO3[27] and CaMn2O4[28]) are indicated in (a).

Table 1

TEC of CCO (Perpendicular and Parallel to the Pressing Direction During SPS), CMO, and LAO

 CCO parallel to the pressing direction (K–1·10–6)CCO perpendicular to the pressing direction (K–1·10–6)LAO (K–1·10–6)CMO-composite (K–1·10–6)
400–800 °C (heating)17.014.49.718.0
700–400 °C (cooling)17.814.59.718.2
(a) XRD patterns of n class="Chemical">CCO, CMO-composite, and LAO powders as well as a sample of CCO made by SPS in the directions parallel and perpendicular to the pressing direction, (b) SEM micrographs, and (c) sintering curves of CCO, CMO-composite, and LAO powders produced by spray pyrolysis. Diffraction lines from articles, LAO,[25] CCO,[26] and CMO-composite powder (CaMnO3[27] and CaMn2O4[28]) are indicated in (a).

TE Performance of the Materials

Electrical conductivity, thermal conductivity, Seebeck coefficient, power factor, and zT as a function of temperature for the p- and n-type oxides are presented in Figure . CCO shows a maximum electrical conductivity in the range of 500–600 °C, reaching approximately 100 S cm–1 (Figure a). On the other hand, the electrical conductivity of the CMO-composite has a constant value of 7 S cm–1 up to 600 °C from where it increases to 28 S cm–1 at 900 °C. The thermal conductivity of both materials decreases with temperature, reaching minima of 0.85 W m–1 K–1 at 700 °C for CCO and 1.42 W m–1 K–1 at 800 °C for the CMO-composite (Figure b). The sudden drop in thermal conductivity for CCO above 600 °C is most probably due to phonon–phonon interactions between the a–b plane and c direction due to the anisotropic crystal structure and texturing of the sample. A maximum Seebeck coefficient of 186 μV K–1 at 500 °C for CCO and −325 μV K–1 at 400 °C for the CMO-composite (Figure c) was achieved. A maximum power factor for CCO (3.6 μW cm–1 K–2) was obtained at ∼500 °C. The figure-of-merit, zT, presented in Figure d increased with temperature, reaching 0.39 at 700 °C for CCO and 0.05 at 800 °C for the CMO-composite. The electrical conductivity and Seebeck coefficients were measured in the direction perpendicular to the pressing direction, whereas the thermal conductivity was recorded parallel to the pressing direction of the sample. The measurements in the two different orientations resulted in a higher zT than the real one because the reported thermal conductivity of CCO is strongly anisotropic.[29]
Figure 2

(a) Electrical conductivity, (b) thermal conductivity, (c) Seebeck coefficient, and (d) zT (full symbols) and power factor (open symbols) as a function of temperature for CCO and CMO-composite ceramics. Error bars represent standard deviation based on five (Seebeck coefficient) and three (thermal conductivity) measured values. The uncertainty in electrical conductivity is smaller than the symbols and less than ±1%.

(a) Electrical n class="Chemical">conductivity, (b) thermal pan> class="Chemical">conductivity, (c) Seebeck coefficient, and (d) zT (full symbols) and power factor (open symbols) as a function of temperature for CCO and CMO-composite ceramics. Error bars represent standard deviation based on five (Seebeck coefficient) and three (thermal conductivity) measured values. The uncertainty in electrical conductivity is smaller than the symbols and less than ±1%.

TE Module

Schematic of the cross section of the TE module design is shown in Figure a. The thickness of the CMO-n class="Chemical">composite after co-sintering is approximately four times larger than the thickness of the CCO layer. Resistance, R, of both conductors was calculated by the formula R = ρ·l/A where ρ is electrical resistivity (cm·S–1), l is height/length (cm), and A is area (cm2). The electrical resistance of CMO-composite at, for example, 800 °C is about 1.6 more than one of CCO, and therefore, the CMO-composite represents more electrically resistive part in spite of larger thickness compared with CCO and limits the charge carrier flow. Because of the significant increase in electrical conductivity of CMO-composite above 800 °C, and slight decrease of CCO, the electrical resistance of the CMO-composite at 900 °C is about 0.7 times less than one of CCO; hence, the CCO represents more electrically resistive part.
Figure 3

(a) Schematic of the cross section of the TE module with illustrated top part and bottom part (p- and n-type materials separated by an insulator, i) as well as flow direction of heat at top part (b) polynomial fitting of electrical power output (Pel) and linear fitting of voltage (U) as a function of the electrical current (Jq) at different Thot temperatures (700, 800, and 900 °C) and constant ΔT = 160 K. Voltage and current at maximum power U(Pma) and Jq(Pma) at different temperatures are also indicated by dotted lines.

(a) Schematic of the cross section of the TE module with illustrated top part and bottom part (p- and n-type materials separated by an insulator, i) as well as flow direction of heat at top part (b) polynpan>omial fitting of electrical power output (Pel) and linear fitting of voltage (U) as a funpan>ction of the electrical current (Jq) at different Thot temperatures (700, 800, and 900 °C) and constant ΔT = 160 K. Voltage and current at maximum power U(Pma) and Jq(Pma) at different temperatures are also indicated by dotted lines. Voltage (n class="Disease">polynomial fitting) anpan>d power output of the TE module as a funpan>ctionpan> of measured currenpan>t output are shownpan> inpan> Figure b. Dashed linpan>es represenpan>t currenpan>t anpan>d voltage at maximum power at 700, 800, anpan>d 900 °C at the hot side of the module. Power output inpan>creases with temperature, reachinpan>g a maximum of about 5.7 mW at 900 °C. The effective power denpan>sity of about 23 mW/cm2 at this temperature was calculated from the effective area of TE module (approximately 0.25 cm2). Open-circuit voltage UOC and short-circuit current Jq,SC were determined by extrapolation from the measured Jq–U line and reached 213 mV and 108 mA at 900 °C, respectively. Data from the characterization of the TE module performance are summarized in Table .
Table 2

Open-Circuit Voltage UOC, Short-Circuit Current Jq,sc, Electrical Resistance of the Module R, Electrical Power Output Pma, and Power Density of CCO–CMO TE Module at 700, 800, and 900 °C at 160 K Temperature Difference between the Hot and Cold Side

Th (°C)Uoc (mV)Jq,sc (mA)Rmodule (Ω)Pmax (mW)Pmax (mW/cm2)
700181822.24.016
8002081012.15.221
9002131082.05.723
Because the power output is dependent on load resistance, the maximum power output n class="Chemical">could be measured whenpan> the set-up-load from anpan> externpan>al circuit of 5.7 Ω became equal to the TE module’s resistanpan>ce of about 2.1 Ω (at 800 °C).[30] The calculated figure of merit of the module (ZT) is 0.01, usinpan>gwhere UOC, Jq,SC, anpan>d Rmodule represenpan>ts openpan>-circuit voltage, short-circuit currenpan>t, anpan>d resistanpan>ce of the module, respectively.[31]

p–p–n Junction

The microstructure of the CCO–CCMO–CMO p–p–n junpan>ction before and after annealing at 900 °C for 100 h is shown in Figure a. Elongated grain growth parallel to the interface is evident in CCO. A thin Ca3CoMnO6 (CCMO) layer, confirmed by energy-dispersive spectrometry (EDS), is formed in situ between CCO and CMO during the co-sintering. The CCMO layer has grown to approximately 5 μm after annealing at 900 °C for 100 h. In addition, a layer of approximately 35 μm thickness close to the interface displayed a higher density than the rest of the CCO. From the EDS profiles of the CCO–CMO interface presented in Figure b, the Ca content in this dense layer is lower than that in CCO showing a Co-rich and a Ca-deficient region at the interface. The Ca/Co ratio equals the initial Ca/Co ratio corresponding to pure CCO approximately 40 μm from the interface. A Co-oxide phase seen as grains with higher Co-content in Figure a is present both in the dense interface layer as well as in the CCO far from the interface.
Figure 4

(a) SEM micrographs of CCO–CCMO–CMO junction before and after annealing at 900 °C for 100 h and EDS maps of the magnified section of the annealed junction given by the dashed lines and (b) EDS line profiles across the CCO–CCMO–CMO junction after annealing.

(a) SEM micrographs of n class="Chemical">CCO–CCMO–CMO junpan>ctionpan> before anpan>d after anpan>nealinpan>g at 900 °C for 100 h anpan>d EDS maps of the magnpan>ified sectionpan> of the anpan>nealed junpan>ctionpan> givenpan> by the dashed linpan>es anpan>d (b) EDS linpan>e profiles across the pan> class="Chemical">CCO–CCMO–CMO junction after annealing. Current–voltage curves across the n class="Chemical">CCO–CCMO–CMO complex junction before and after annealing at 900 °C for 100 h, measured at 300, 500, and 700 °C are presented in Figure a. Ohmic behavior is observed in the whole temperature range for both as-sintered and annealed junctions, and the resistance decreases with increasing temperature. The difference in the resistance for the as-sintered and annealed samples is less pronounced as the temperature increases, and at 700 °C almost no difference is observed (inset in Figure b).
Figure 5

(a) Current–voltage curves of CCO–CMO–CMO p–p–n junctions before and after annealing at 900 °C for 100 h and (b). The activation energies for electrical conduction for the as-sintered and annealed sample. The inset shows resistance of p–p–n junctions before and after annealing as a function of temperature.

(a) Current–voltage curves of n class="Chemical">CCO–CMO–CMO p–p–n junpan>ctionpan>s before anpan>d after anpan>nealinpan>g at 900 °C for 100 h anpan>d (b). The activationpan> enpan>ergies for electrical pan> class="Chemical">conduction for the as-sintered and annealed sample. The inset shows resistance of p–p–n junctions before and after annealing as a function of temperature. Figure represents the activation energy for conduction of the as-sintered and annealed samples according to the Arrhenius equation, where the annealed sample demonstrates twice the activation energy of the as-sintered.

Discussion

Ceramic Processing

A new co-sintering route to an all-oxide TE module was developed. The maximum sintering temperature was limited by the decomposition temperature of CCO. Moreover, CMO possesses a phase transition at 896–913 °C[32] from orthorhombic to tetragonal phase associated with a volume change, which also could introduce stresses in the device. Therefore, 880 °C was selected as the maximum co-sintering temperature. Because this temperature is low for efficient densification of CMO-composite and LAO (Figure c), a maximum pressure of 75 MPa was applied in the SPS with optimal 5 min hold, which resulted in 70 and 46% relative density for CMO-composite and LAO, respectively. CCO was completely densified after 2 min at 880 °C, but due to grain growth and micro-delamination, the final density after 5 min was 91% of theoretical. The CMO-composite possesses the highest TEC of the three materials (Table ) and tensional stress is induced during cooling; hence, the CMO-composite represents the most sensitive part of the module during processing being prone to crack formation. The crack formation could be controlled by designing a thin CCO (∼0.8 mm) and thick CMO (∼3.6 mm) layer reducing the tensile stresses in the CMO-composite. The calculated tensile stresses in the CMO-composite (Supporting Information, Figure S2) developed during the cooling from 880 °C decreases both with increasing CMO-composite thickness and decreasing CCO thickness.[33]

Origins of High Power Output and Open-Circuit Voltage

As evident from Table , there are large differences in the output power of n class="Chemical">convenpan>tionpan>ally designpan>ed pan> class="Chemical">CCO–CMO TE modules dependent on whether CCO and CMO are doped or (like in our case) undoped, as well as on the applied temperatures and gradients.
Table 3

TE Performance of Conventional Modules Based on p-Type CCO and n-Type CMO Reported in the Literature

systemN-pairsThot (K)ΔT (K)Pmax (mW)Pdensity (mW cm–2)P/one pair (mW)references
Ca3Co4O9, CaMnO3124732001.983.3a0.165aSeetawan et al.[15]
Ca3Co4O9, CaMno331 2001.47 × 10–323.7 × 10–3a4.74 × 10–5aPhaga et al.[16]
Ca2.7Bi0.3Co4O9, CaMn0.98Mo0.02O3889756517042.521.2aUrata et al.[17]
Ca2.75Gd0.25Co4O9, Ca0.92La0.08MnO3877339063.544.17.9aMatsubara et al.[18]
Ca3Co4O9, Ca0.95Sm0.05MnO32100092531.549.2a15.7aReddy et al.[19]
Ca3Co4O9, Ca0.9Nd0.1MnO3111757279593.295Lim et al.[20]
Ca0.76Cu0.24Co4O9, Ca0.8Dy0.2MnO34 3468.42 2.1aPark et al.[21]
Ca3Co4O9, Ca0.95Sm0.05MnO3299063031.549.215.7aNoudem et al.[22]

The values marked with are calculated, based on the data available in the given references.

The values marked with are calculated, n class="Chemical">based onpan> the data available inpan> the givenpan> referenpan>ces. Urata et al.[17] reported as much as 42.5 mW cm–2 for a doped system with a Thot of about 900 °C and a gradient as high as 565 K, while undoped n class="Chemical">CCO–CMO systems at lower temperature anpan>d gradienpan>ts of 200 K yielded merely 3.3 mW cm–2 (Phaga et al.[16]) anpan>d 23.7 × 10–3 mW cm–2 (Seetawanpan> et al.[15]). Our prototype TE module with undoped CCO–CMO and a complex p–p–n junction could not be exposed to a larger ΔT than 160 K due to limitations of the set-up. Yet, a maximum power density of about 23 mW cm–2 was generated at 900 °C. We attribute this to the remarkably high open-circuit voltage and low interface resistance of the complex p–p–n junction, as will be discussed next. An class="Chemical">ccordinpan>g to Kanpan>as et al., electrical pan> class="Chemical">conductivity of the CCMO phase formed at the interface increases sharply with increasing temperature from 800 to 900 °C and reaches around 0.1 S cm–1 at 900 °C, where the positive Seebeck coefficient furthermore reaches as high as 668 μV K–1.[34] The three-layered CCO–CCMO–CMO junction exhibits Ohmic behavior with relatively modest interfacial resistance above 700 °C as evidenced in Figure a. We tentatively interpret this as an effect of the electron energy levels of CCMO as laying intermediate of those of CCO and CMO, hence decreasing the depletion of charge carriers at the CCO–CCMO and CCMO–CMO interfaces as compared to a hypothetical pristine CCO–CMO p–n interface. This facilitates an electrical current flow through the p–p–n interface as illustrated in Figure a. The CCMO reaction layer is thicker in the annealed sample compared with the as-sintered one (Figure a), and in the temperature range 300–500 °C, resistivity is higher for the annealed sample (Figure ). At 700 °C, the resistivity of both samples are almost equal (Figure ), showing that the same concentration of charges will be excited to conduction level because of thermal activation, demonstrating no effect of CCMO thickness on the current density above 700 °C. The CCMO layer is therefore contributing equally to the electrical conduction process at this high temperature. When two materials A and B dissimilar in thermal conductivity and Seebeck coefficient are contacted over an area exposed to a parallel thermal gradient, a voltage is generated in the transversal direction, and an effective transversal Seebeck coefficient of the couple can be expressed according to Goldsmid[35]where SA, SB, KA, and KB represent the Seebeck coefficients and thermal resistances of the two materials. In our case, the thermal conductivity of CCO at 900 °C is 0.94 W m–1 K–1 while that for CCMO is 1.3 W m–1 K–1[34] and the CMO-composite is 2.0 W m–1 K–1. This will give rise to different heat flows down the n- and p-type materials, resulting in an increasing transversal temperature gradient down the CCMO interface layer corresponding to the transversal heat flow illustrated in Figure a. For the following discussion, the module may be divided into a top hot part above the LAO insulator and a bottom conventional part. The top part contains three material layers and two subjunctions which contribute to a transversal TE voltage according to eq , whereas the bottom part contributes to standard longitudinal TE voltages. On the basis of the dimensions of the module, we may estimate a temperature difference of 40 K over the top part and the remaining 120 K cover the bottom part. Because the Seebeck coefficient of CMO is strongly influenced by temperature (Figure c), average values are used for summing up all possible contributions from materials and interfaces in the transversal and longitudinal parts. We arrive at an estimated open-circuit potential of 125 mV, as compared to the estimate of 64 mV from a regular CCO and CMO couple with a total gradient of 160 K, based on their Seebeck coefficients of +177 and −171 μV K–1 (at 900 °C) and about +179 and −270 μV K–1 (at approximately 740 °C), respectively. Average absolute values of the Seebeck coefficient and thermal conductivity of CCO and CMO for the top and bottom parts of the module used for the calculations are 178 μV K–1 (CCObottom), 177 μV K–1 (CCOtop), 0.89 W m–1 K–1 (CCOtop), 229 μV K–1 (CMObottom), 189 μV K–1 (CMOtop), and 1.88 W m–1 K–1 (CMOtop). A transversal TE effect occurs when anisotropy in the electrical and thermal transport occurs,[35] as in our top part of the module. This phenomenon is beneficially used for enhancing the voltage in transversal TE modules and related applications.[36−40] For instance, the transversal TE voltage in CaCoO2 textured thin films can be significantly higher than the ones generated by regular TE effect.[36−38] The transversal voltage significantly affected the open-circuit voltage UOC and the maximum electrical power output Pel. The presence of a transverse TE effect can also be observed by comparison of UOC of the TE module (Figure b) and Seebeck coefficient together with the power factor of individual materials from 700 to 900 °C. The Seebeck coefficients (Figure c) of the two individual materials decrease as temperature increases, as well as power factor of CCO (Figure d), whereas UOC of TE module increases with temperature (Figure b). The experimentally measured UOC of 213 mV is hence remarkably larger than both theoretical estimates. Further experimental and theoretical studies (e.g., field emission microscopy simulations) will be necessary to fully understand the effect of thin layers of materials with high Seebeck coefficients in the interface of p–n TE junctions. However, the in situ formed complex p–p–n junctions evidently improved the performance of the all-oxide TE module, hence representing a significant step forward toward the possible application of oxide TE modules in high-temperature energy recovery.

Conclusions

A novel all-oxide TE module was successfully developed and fabricated by careful processing of materials in the CCO–CMO system using LAO as an electrical insulator. Fabrication of this all-oxide TE module is simpler and faster than assembling conventional modules. The CCO–CMO module demonstrated a power of about 5.7 mW corresponding to a power density of 23 mW cm–2 at Thot = 900 °C and ΔT = 160 K, a TE performance comparable and better than some conventional CCO–CMO modules. The all-oxide layered TE module produces a large open-circuit voltage, which was attributed to the presence of a thin CCMO reaction layer and transversal TE effect across the top p–p–n part of the TE module. The effect of the CCMO reaction layer is due to the large Seebeck coefficient, working transversally and reducing the charge carrier depletion and resistance at the high-temperature p–n junction. The present investigation demonstrates an example of novel engineering of oxide TE modules without metallic interconnects.

Experimental Section

Materials and Ceramic Processing

The ceramic powders used in this work were prepared by spray pyrolysis (CerPoTech AS, Norway). Slurry for aqueous tape casting of the LAO insulator was prepared according to the schematics shown in Figure S1 (Supporting Information). The slurry was casted on a polyester (Mylar) film using a height of the doctor blade of 30 μm. After drying at ambient temperature, lamination of 8 layers was conducted by hot-pressing at 150 MPa and 80 °C for 3 min. The laminated tape was cut into 12 mm discs with one segment cut off (around 3 mm) to make the direct p–n junction. Binder burnout was done at 440 °C for 4 h in air, placing the tape between two alumina plates to avoid bending. The TE module was fabricated by SPS (Dr. Sinter 825) in a 12 mm graphite die at 880 °C and 75 MPa for 5 min using heating and cooling rates of 120 °C/min. Initially, the graphite die was filled with the CMO-composite powder (1.5 g), and then the cut LAO tape was placed on CMO-composite powder, followed by filling of CCO powder (0.5 g) onto the LAO tape. In addition, CMO-composite and CCO materials were separately sintered by SPS using the same conditions as used for the TE module. These pellets were cut into a bar-shape (20 × 5 × 2.5 mm) for electrical conductivity and Seebeck coefficient measurements and discs (12.7 mm diameter) for thermal conductivity measurements. Densities of pellets and bars were determined by Archimedes measurement in isopropanol. To do a separate analysis of the direct CCO–CMO-composite junction, two samples were prepared by SPS co-sintering at 820 °C and 50 MPa for 10 min. One of these samples was further annealed at 900 °C for 100 h in air.

Characterization

Phase composition and particle size/morphology were characterized by powder X-ray diffraction (Bruker D8 DAVINCI) and SEM (Hitachi S-3400N), respectively. Sinterability and TEC were determined by dilatometer (Netzsch DIL 402) in ambient air. TEC was measured both parallel and perpendicular to the pressing direction for CCO made by SPS. TE performance of CCO and CMO materials were analyzed by measuring Seebeck coefficient (ProboStat, NorECs AS), electrical conductivity (home-made setup) and thermal conductivity (Netzsch LFA 457 MicroFlash) at 100–900 °C in ambient air as described elsewhere.[41,42] For the power–current–voltage characterization, the layered disc-shaped TE module was placed horizontally on an n class="Chemical">alumina plate with the pan> class="Chemical">direct p–n junction at the top (hot temperature side). To ensure better physical stability of the free-standing module, a small area of about 0.25 cm2 was removed on the n class="Chemical">cold side. Both semipan> class="Chemical">conductors were contacted by Pt-wires at the lower, colder end, using gold paste (Heraeus). The alumina plate with the TE module connected to Pt-wires was heated at 700 °C for 4 h in ambient air, to establish good electrical contact between the metal electrodes and semiconducting oxides. The power output test in ambient air was performed via load resistance-dependent measurement, in a vertical furnace at 700, 800, and 900 °C, and ΔT = 160 K. The temperature gradient was established by heating the hot side of the module in the furnace, whereas the cold side was cooled by an active cooler. The temperature difference ΔT was measured by two Pt–Pt10Rh (type S) thermocouples. During the power output measurement, the voltmeter was connected parallel to the amperemeter, variable resistor (in series to each other), and to the TE module. When thermal equilibria were established at each temperature, electrical current and voltage were measured with increasing load resistance. More details could be found in ref (4). Finally, annealed and as-sintered pn junctions were characterized by current–voltage measurements at 300, 500, and 700 °C in a vertical furnace, using a ProboStat cell (NORECS, Norway). A two-electrode set-up was used, where a dc voltage was applied to the junction followed by measuring current output using Multimeter-Agilent E3642A. After the measurements, the samples were embedded in “EpoFix” resin, polished by diamond paste (DiaPro NapB1) to 1 μm and coated with carbon (Cressington Carbon Coater 208) for microstructural characterization. Interface reaction and interdiffusion at the p–n junction were investigated using SEM and EDS (Hitachi S-3400N).
  3 in total

1.  Chemical stability of Ca3Co4-x O9+δ /CaMnO3-δ p-n junction for oxide-based thermoelectric generators.

Authors:  Anette Eleonora Gunnæs; Raluca Tofan; Kristian Berland; Sandeep Gorantla; Thomas Storaas; Temesgen Debelo Desissa; Matthias Schrade; Clas Persson; Mari-Ann Einarsrud; Kjell Wiik; Truls Norby; Nikola Kanas
Journal:  RSC Adv       Date:  2020-01-30       Impact factor: 4.036

2.  Time-Enhanced Performance of Oxide Thermoelectric Modules Based on a Hybrid p-n Junction.

Authors:  Nikola Kanas; Rasmus Bjørk; Kristin Høydalsvik Wells; Raphael Schuler; Mari-Ann Einarsrud; Nini Pryds; Kjell Wiik
Journal:  ACS Omega       Date:  2020-12-28

3.  Near-Broken-Gap Alignment between FeWO4 and Fe2WO6 for Ohmic Direct p-n Junction Thermoelectrics.

Authors:  Raphael Schuler; Federico Bianchini; Truls Norby; Helmer Fjellvåg
Journal:  ACS Appl Mater Interfaces       Date:  2021-02-05       Impact factor: 9.229

  3 in total

北京卡尤迪生物科技股份有限公司 © 2022-2023.