Right Heterogeneous Microstructure for Achieving Excellent

Jul 16, 2018 - (1−3) Their potential for thermoelectric applications is determined by the ..... of CMO:R (R = Dy, Yb) from 300 to 1100 K is shown in...
1 downloads 0 Views 9MB Size
Download PDF
Article pubs.acs.org/IC

Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX

Right Heterogeneous Microstructure for Achieving Excellent Thermoelectric Performance in Ca0.9R0.1MnO3−δ (R = Dy, Yb) Ceramics Teng Wang,†,§ Pengfei Nan,‡,§ Hongchao Wang,*,†,⊥ Wenbin Su,† Andres Sotelo,∥ Jinze Zhai,† Xue Wang,† Yazhou Ran,† Tingting Chen,† and Chunlei Wang*,† †

School of Physics, State Key Laboratory of Crystal Materials, Shandong University, Jinan 250100, China Beijing Key Lab of Nanophotonics and Ultrafine Optoelectronic Systems, School of Physics, Beijing Institute of Technology, Beijing 100081, China ∥ Instituto de Ciencia de Materiales de Aragón (CSIC-Universidad de Zaragoza), Ma de Luna 3, Zaragoza 50018, Spain ⊥ State Key Laboratory of Metastable Materials Science and Technology, Yanshan University, Qinhuangdao 066000, China

Downloaded via TUFTS UNIV on July 17, 2018 at 00:42:50 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

S Supporting Information *

ABSTRACT: Perovskite manganite Ca0.9R0.1MnO3−δ (R = Dy, Yb) ceramics have been synthesized by a traditional solidstate reaction with multicalcination processes. A heterogeneous microstructure including large and small micrometersized grains, coherent interfaces, and oxygen defects has been formed with optimized calcination time. The carrier concentration of the third-calcined samples is enhanced approximately 3 times compared with those synthesized through conventional methods. Thus, the electrical resistivity of the third-calcined Ca0.9R0.1MnO3−δ (R = Dy, Yb) ceramic samples obviously decreases, leading to a higher power factor. Additionally, the thermal conductivity is also reduced by multiscale scattering of the heterogeneous structure. The lowest lattice thermal conductivities of Dy- or Yb-doped samples are 1.24 and 1.22 W m−1 K−1, respectively. Thus, the high thermoelectric performance for Ca0.9R0.1MnO3−δ (R = Dy, Yb) has been achieved by the multicalcination process. The highest figure of merit is almost 30% higher than that of the first-calcined samples. Therefore, a heterogeneous microstructure formed by optimized multicalcination can effectively optimize the thermoelectric performance of oxides.

1. INTRODUCTION

parable to several traditional alloys. Hence, metal oxides as thermoelectric materials have attracted considerable attention. In many researched thermoelectric oxides, their layered structure shows excellent thermoelectric performance and can be optimized by doping, nanostructuring, and so on.12−15 Through dual doping with Ag and Lu, the microstructures of Ca3Co4O9-based ceramics show metallic nanoinclusions of Ag, and thus zT is improved to ∼0.61.14 By heavy doping with Ba and refinement of the grain sizes, the layered BiCuSeO system presents a high zT of 1.1 at 923 K.15 Additionally, zT for flaky single-crystal NaxCoO2−δ, prepared by a flux technique, can exceed 1.0 because of its high PF.11 As per the above description, several high figures of merit have been achieved in some layered thermoelectric oxides, with p-type semiconducting behavior. In thermoelectric device applications, a high thermoelectric performance of n-type oxides is required and should be comparable to the p-type performance.6 Therefore, the thermoelectric performance of n-type oxides has attracted more attention, and many studies have been conducted. To

Thermoelectric materials, which can directly convert waste heat into electricity, are playing an important role in sustainable energy.1−3 Their potential for thermoelectric applications is determined by the dimensionless figure of merit, which is defined as zT = S2T/ρκ, where S is the Seebeck coefficient, ρ the electrical resistivity, κ the total thermal conductivity, and T the absolute temperature in Kelvin.4,5 In this expression, S2/ρ is the power factor (PF) and is related to the electric properties. To maximize the zT value of a material, a large S, low ρ, and low κ are needed. However, these three parameters are intercorrelated, which makes it difficult to separately control them to produce a high thermoelectric performance in one material.3 Therefore, many techniques, such as doping, band engineering, and nanostructuring, have been tried to achieve a high zT, and several high figures of merit have been found in some thermoelectric materials.6−9 Metal oxides, because of their high thermal and chemical stability, oxidation resistance, and low cost, have been promising thermoelectric materials.10 This promise has increased with the discovery of p-type NaCo2O4,11 which possesses excellent thermoelectric performance and is com© XXXX American Chemical Society

Received: April 27, 2018

A

DOI: 10.1021/acs.inorgchem.8b01163 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 1. Powder XRD patterns of multicalcined (a) Ca0.9Dy0.1MnO3−δ and (b) Ca0.9Yb0.1MnO3−δ ceramics. The black vertical bars are from the standard XRD pattern of orthorhombic perovskite CaMnO3 (ICSD: 35218).

The carrier concentration has been improved by the third calcination. This improvement leads to a lower electrical resistivity and a higher PF. The thermal conductivity is reduced by multiscattering through large and small micrometer-sized grains, coherent interfaces, and oxygen defects, leading to the highest figures of merit, zT = 0.12 and 0.14, in the third-calcined CMO:R (R = Dy, Yb) samples, respectively.

date, the thermoelectric performance of n-type thermoelectric oxides has not broken through, and the figures of merit are approximately 0.1−0.4, which are inferior to those of wellstudied p-type thermoelectric oxides. Therefore, the thermoelectric performance of n-type thermoelectric oxides should be further improved for applications with matched p-type oxides. CaMnO3 (CMO) is a typical n-type thermoelectric oxide that may achieve a high thermoelectric performance in the future through optimization. In the past few years, elemental doping has been utilized as the main method to improve zT of CMO oxides, and large relative studies have been performed. Generally, depending on the sites where the dopants are substituted, there are two types of single doping. One type is single-element doping at the Mn site,16 and the other type is doping at the Ca site. Compared with substitution at the Mn site, doping elements at the Ca site more easily achieves a high thermoelectric performance.6 Dopants such as Bi, Y, La, Ce, Dy, Nd, Tb, Ho, Yb, and Lu with tri- or tetravalent oxidation states can increase the carrier concentration.17−19 The increase in the carrier concentration leads to a marked decrease in the electrical resistivity and a moderate decrease in the Seebeck coefficient. Additionally, because of the larger mass difference with the Ca2+ ion, Dy- and Yb-doped CMO possesses lower thermal conductivity. Thus, from all of these works,6,17−20 it is clearly found that Dy- or Yb-doped samples easily raise zT. Overall, elemental doping can improve zT of CMO to a certain degree, and the effect of Dy or Yb dopants on the thermoelectric performance is especially significant. Generally, the use of synthesis technology is important to optimize physical transport in thermoelectric materials. Different quenching media can result in different microstructures and thus affect the thermoelectric performance. Heterogeneous PbTe nanocomposites are synthesized by quenching into ice water. Their zT values can be enhanced by 25% compared with homogeneous samples.21 By applying a hot-forging process to produce textured microstructures, Bi0.875Ba0.125CuSeO reaches a record zT of 1.4 at 923 K.22 As in hot-forged Bi0.5Sb1.5Te3 alloys, in situ nanostructures and high-density lattice defects were observed, which led to a higher PF, lower thermal conductivity, and, ultimately, a peak zT value of 1.3.23 Moreover, reduced SrTiO3 ceramics, prepared by controlling the heating rate of plasma sintering, have reached an impressive zT value of 0.4 at 823 K.24 This zT is the highest determined so far for bulk SrTiO3 ceramics. In summary, optimization of the synthesis conditions can effectively enhance the thermoelectric performance. Herein, heterogeneous Ca0.9R0.1MnO3−δ (CMO:R; R = Dy, Yb) samples have been prepared by a multicalcination process, and their relative thermoelectric performance was evaluated.

2. EXPERIMENTAL SECTION Ca0.9R0.1MnO3 (R = Dy, Yb) ceramics were synthesized by a solidstate reaction with a multicalcination process. The starting materials were CaCO3 (99%), MnO2 (97.5%), Dy2O3 (99.99%), and Yb2O3 (99.99%). After the raw materials were weighed in stoichiometric proportions, they were mixed by ball milling in ethanol with zirconia balls for 12 h. Then, the mixtures were dried, cold-pressed into pellets, and calcined at different times in air with intermediate grinding. The samples were assigned as CMO:R-1, -2, -3, and -4 (R = Dy, Yb) for different calcined times. The detailed calcination processes are as follows. For CMO:R-1, the cold-pressed pellets were calcined at 1373 K for 4 h in air. For CMO:R-2, the cold-pressed pellets were calcined at 1073 K for 4 h in air and then reground by hand, pressed into pellets, and calcined a second time at 1373 K for 4 h in air. For CMO:R-3 and CMO:R-4, the calcination process was similar to that of CMO:R-2 and only differed in the calcined temperature settings. For CMO:R-3, the calcined temperature settings were 1073, 1173, and then 1373 K for 4 h. For CMO:R-4, the temperature settings were 1073, 1173, 1273, and 1373 K for 4 h, sequentially performed. After multicalcination, the pellets were reground and then ball-milled for 12 h to obtain fine powders, which were pressed into pellets and sintered at 1523 K for 4 h in air and then naturally cooled to room temperature in the furnace to produce the final samples. The structures of all samples were studied using a Bruker AXS D8 ADVANCE diffractometer with Cu Kα radiation at room temperature. The X-ray diffraction (XRD) patterns for all samples are presented in Figure 1. The surface morphology and energy-dispersive spectroscopy (EDS) mapping were obtained by scanning electronic microscopy (SEM; JSM-6701F). Transmission electron microscopy (TEM) was carried out using a JEM-2100F microscope. Specimens for TEM were prepared by a focused-ion beam (FIB). The Hall resistivity was measured with a homemade Hall system using a fourprobe structure under magnetic fields of ±0.6 T. The carrier concentration was calculated based on n = 1/eRH, where e is the electronic charge and RH is the Hall coefficient, while the carrier mobility was calculated using μ = 1/ρRH, where ρ is the electrical resistivity. The electrical resistivity and Seebeck coefficient were measured in the temperature range between 300 and 1100 K with a LINSEIS LSR-3 instrument. The thermal diffusivity values (λ), as shown in Figure S1, were measured by a NETZSCH LFA-457 laser flash apparatus directly from room temperature to 1073 K. The thermal conductivity (κ) was calculated by the formula κ = λCpd, where Cp is the heat capacity. Here, Cp of CMO:Dy is an average value of all Dy-doped samples. Similarly, Cp of CMO:Yb is also the average value of all Yb-doped samples. Cp for each sample was derived B

DOI: 10.1021/acs.inorgchem.8b01163 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry using standard samples (Pyroceram) in LFA457, and these are shown in Figure S2. Furthermore, d is the density of the sample, which was measured by the Archimedes method. The PF and zT were calculated from the above-measured parameters.

boundaries are more energetic and thus less stable than the intragrain ones. As a consequence, the small grains (with higher superficial energy than the large ones) tend to disappear, while the large ones tend to grow to decrease the system energy. This change is consistent with previously reported studies regarding SrTiO3 ceramics.24 Although the small micrograins disappeared and formed larger grains after multicalcination, as shown in Figure 2, the average grain sizes of the multicalcined samples are between 1 and 3 μm in all cases, not very large. Additionally, the micrograin size distributions for all samples have been considered and are presented in Figures S3 and S4. Thus, this kind of microstructure with different relative sizes may be better to scatter the low-frequency phonons and optimize the lattice thermal conductivity. Another important feature observed in these micrographs is the very low porosity of the samples, in clear agreement with the density values previously discussed. In order to assess the element distributions as a function of the calcination time, EDS mappings for unpolished multicalcined CMO:Yb samples are presented in Figure 3. Referring to the previous work,23 it was found that the doping element or related oxides are in higher concentration at the grain boundaries. In this work, the difference in the element distributions both inside the grains and in the grain boundaries is also considered. In addition, the natural flattened surface has been obtained, as shown in Figure 2. So, the samples were only cleaned using an ultrasonic cleaner without polishing before the EDS mapping test. In the first-calcined CMO:Yb samples, the distribution of Mn and Ca is homogeneous because it is difficult to observe element-rich zones in EDS mapping, while the Yb distribution shows only slight differences. However, a heterogeneous distribution of Yb, Ca, and O can be clearly seen in the third-calcined CMO:Yb samples. This effect can be associated with the possible formation of phases with lower formation kinetics and low thermodynamical stability, resulting from the multigrinding and multicalcination processes. However, the exact composition of this phase is not exactly clear (due to the qualitative nature of EDS analysis) and will be confirmed in future works. As for the fourth-calcined samples, the element distributions become homogeneous again, which is almost the same as or a little different from the first-calcined one. Excessive grinding, calcination, and second ball milling after calcination may be responsible for the homogeneous element distribution of the fourth-calcined samples. The element distribution in the inner part of the samples was also done after polishing, as shown in Figure S5, and the heterogeneous distribution of elements is further confirmed. Furthermore, the fact that this additional phase has not been observed in the XRD patterns can be due to a small compositional difference from the thermoelectric one, which could overlap the same peaks, or due to its very small proportion in the samples. The distribution of O for physical transport is an important factor in thermoelectric oxide, as illustrated in previous reports.28,29 The heterogeneous distribution has been found in a relatively large area, as shown in Figure 3. For more clearly showing this heterogeneous distribution, the EDS line scanning at the boundary area has also been done for CMO:Yb samples, which are shown in Figure 4. The EDS line scan performed across two neighborhood grains shows the change of the O contents at the grain boundary region for all samples. With multicalcination, the O content is decreased, especially for the O defects of the third-calcined sample. The O

3. RESULTS AND DISCUSSION Powder XRD patterns shown in Figure 1 demonstrate that all Ca 0.9 R 0.1 MnO 3−δ (R = Dy, Yb) ceramics exhibit an orthorhombic perovskite structure (ICSD: 35218). Moreover, no secondary phases have been detected in any of the samples in spite of their different calcination times. From these data, lattice parameters and theoretical densities have been calculated, and the results are shown in Table S1. The relative density values for the CMO:Dy samples are the same, specifically, 92%, while the CMO:Yb ones are in the 91%− 94% range. The results are similar to previous works25,26 and clearly show that high-density CMO:R (R = Dy, Yb) samples have been obtained. Figure 2 shows SEM images of surface sections of all CMO:R (R = Dy, Yb) samples. These images indicate that the

Figure 2. SEM images of surface sections for different calcination times. The left column presents the SEM pictures for (a) CMO:Dy-1, (b) CMO:Dy-2, (c) CMO:Dy-3, and (d) CMO:Dy-4, and the right column shows the SEM pictures of (e) CMO:Yb-1, (f) CMO:Yb-2, (g) CMO:Yb-3, and (h) CMO:Yb-4.

first-calcined CMO:R (R = Dy, Yb) samples have a dual grainsize distribution. After multicalcination, the much smaller-sized grains disappeared, leading to a slight growth of the neighborhood ones. This behavior can be easily explained through Ostwald ripening.27 The cations found in the grain C

DOI: 10.1021/acs.inorgchem.8b01163 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 3. EDS elemental mappings for top to bottom are Mn (cyan), Ca (yellow), Yb (green), and O (red) for CMO:Yb-1 CMO:Yb-2, CMO:Yb3, and CMO:Yb-4 ceramics, respectively.

grain boundaries of the multicalcined CMO:Yb samples. In addition, the count intensities for Mn and Ca are about 1000 and 1500, respectively, being constant inside the grains and at the grain boundary. In other words, the effect of an uneven plane on the Mn and Ca distribution is slight. The count intensity of O is also about 1200 inside the grains. Following the Mn and Ca distribution, the defect of O at the grain boundary is not mainly coming from the uneven plane but from the heterogeneous distribution. Furthermore, the stoichiometric ratio of O vacancies for the CMO:Yb samples are also estimated according to the formula δ = (3 wtMn %/M Mn − wtO %/M O )/(wt Mn %/M Mn ), where δ is the stoichiometric ratio of O vacancies, wtMn% is the weight percentage of Mn, wtO% is the weight percentage of O, MMn is the mole mass of Mn, and MO is the mole mass of O. The weight percentages of all elements have been estimated from the EDS mapping (Figure S5) and are presented in Table S2. The δ values for CMO:Yb-1, -2, -3, and -4 are 0.14, 0.13, 0.20, and 0.15, respectively, with ±0.02 deviation. The δ value for CMO:Yb-1 is nearly the same as that previously reported.30 In the calculated O vacancies, the CMO:Yb-3 samples possess the highest value. This means that the O vacancies of the CMO:R (R = Dy, Yb) samples are increased after the third calcination and grinding. Therefore, the heterogeneous distribution of O can be formed by the multicalcination method at certain calcination times. The heterogeneous distribution of O will be relative with the changes of electrical and thermal transport. To further confirm this elemental heterogeneous distribution, TEM specimens have been extracted from the micro-

Figure 4. EDS line-scanning pictures across two grains for (a) CMO:Yb-1, (b) CMO:Yb-2, (c) CMO:Yb-3, and (d) CMO:Yb-4.

content at a certain grain boundary region of CMO:Yb-3 shows a shape decrease from that of the others, as shown in the inner picture of Figure 4c; thus, the CMO:Yb-3 samples possess the largest O defects among all Yb-doped samples. However, the O content will return to the original state of CMO:Yb-1 after the fourth calcination. This is well consistent with the EDS mapping results. Besides, the O distribution and the distribution of Ca, Mn, and Yb across two grains are also presented in Figure S6. As can be observed, the Ca and Mn distribution is practically uniform inside the grains and at the D

DOI: 10.1021/acs.inorgchem.8b01163 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

indicating that the dominant carriers are electrons. The n values obtained for the first-calcined CMO:Dy and CMO:Yb samples are 1.31 × 1020 and 2.06 × 1020 cm−3, respectively, which are both consistent with previously reported data.19 The carrier concentrations of the multicalcined CMO:R (R = Dy, Yb) samples are increased, especially for the third-calcined ones. The carrier concentration can also be affected by the synthesis process; for example, when the microstructure of Bi2Te3 alloys is modified by hot forging, applied as a new synthesis technology in previous works, 35 the carrier concentration is optimized. In addition, the rule of carrier concentrations for our all samples is similar to that of O defects found in Figure 4, and the O defects of the third-calcined samples are much more than others. Because the O defects can be seen as electron donors, more O defects mean higher carrier concentration; thus, lower electrical resistivity will be achieved. The effect of O defects on the electric-transport properties has also been illustrated in other reports.28,29 Moreover, the interfaces formed by the phase dispersion may also influence the carrier concentration, and this kind of effect has been found in previous reports.32,33 The carrier mobility at room temperature displayed in Figure 6b shows that the carrier mobilities for all multicalcined samples are not significantly changed probably because the microstructure scale is a little larger than that of the carrier mean free path and has no significant effect on the carrier mobility. Consequently, the multicalcination process leads to O defects and phase dispersion and then optimizes the carrier concentration in the CMO:R (R = Dy, Yb) samples. The temperature dependence of the electrical resistivity (ρ) of CMO:R (R = Dy, Yb) from 300 to 1100 K is shown in Figure 7a,b. All samples show metallic-like behavior (dρ/dT > 0) because ρ increases with increasing temperature in the whole measured temperature range. In addition, the values of the electrical resistivity of the multicalcined samples are much lower than those of undoped CMO17 because of the increased carrier concentration induced by R3+ for Ca2+ substitution. Furthermore, as previously discussed, the multicalcination process leads to a heterogeneous microstructure in multicalcined samples, including many O defects and coherent interfaces from phase dispersion. These microstructural modifications have resulted in an increase of the carrier concentration, and the effect of the multicalcination process may be comparable to an additional doping. Thus, the obtained high carrier concentrations lead to lower electrical resistivity values synthesized by multicalcination. Also, the electrical resistivity of multicalcined CMO:R (R = Dy, Yb) is comparable to those of some dual-doped CMO samples.25 As an example, it is clear that the third-calcined samples present

grains of these samples, not from the grain boundary by a FIB, and the images are shown in Figure 5. Although the XRD

Figure 5. TEM of CMO:Yb. (a) Low-magnification TEM image of the CMO:Yb-3 sample. (b) Medium-magnification TEM image of the CMO:Yb-3 sample. (c) HRTEM image of the CMO:Yb-3 sample. (d) HRTEM image of the CMO:Yb-1 sample.

pattern at room temperature for the third-calcined CMO:Yb sample presents pure phase (Figure 1), the phenomenon of phase dispersion is evidently seen in high- and mediummagnification TEM and high-resolution TEM (HRTEM) images. This may be related to the resolution limit of XRD, and it is in agreement with previous reports.31 Nevertheless, the first-calcined samples show a perfect HRTEM picture, and it is difficult to find phase dispersion and defects, as shown in Figure 5d. In contrast, the phase dispersion found in the thirdcalcined CMO:Yb sample is strongly related with the elemental distribution previously observed and discussed. Additionally, this phase dispersion can form more coherent interfaces, which is beneficial to optimizing, simultaneously, the lattice thermal conductivity and electrical properties, in agreement with previously published reports.32−34 Thus, the heterogeneous microstructure of multicalcined samples has been formed, which includes large and small micrometer-sized grains, O defects, and coherent interfaces. Figure 6a shows the room temperature carrier concentration for multicalcined CMO:R (R = Dy, Yb) samples. The chargecarrier concentration (n) is negative in all cases, clearly

Figure 6. Room temperature (a) carrier concentration and (b) carrier mobility for Ca0.9R0.1MnO3−δ (R = Dy, Yb) ceramics. E

DOI: 10.1021/acs.inorgchem.8b01163 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 7. Temperature dependence of the electrical resistivities (a and b) and Seebeck coefficients (c and d) for CMO:Dy and CMO:Yb.

Figure 8. (a and b) PF and (c and d) total (close symbols) and lattice (open symbols) thermal conductivity for CMO:Dy and CMO:Yb as a function of the temperature.

with previous works.9 However, the absolute values of the Seebeck coefficient are slightly decreased in the multicalcined samples at high temperatures. The reason is that more O vacancies appear at high temperature, and thus the carrier concentration increases. According to the Mott relationship,3 the Seebeck coefficient is inversely proportional to the carrier concentration; thus, a decrease of the absolute Seebeck coefficient appears at high temperatures, in agreement with many other works.36−38 Although the samples are multiground and multicalcined during the synthesis process, the Seebeck coefficient is not obviously changed in the whole measured temperature range in any case. Thus, the Seebeck coefficient for the CMO:R (R = Dy, Yb) composition is not significantly affected by the different synthesis processes and maintains the high Seebeck coefficient for CMO. Overall, the multi-

the lowest electrical resistivity for CMO:R (R = Dy, Yb), being 3.98 mΩ·cm at 326 K for CMO:Dy and 3.29 mΩ·cm at 476 K for CMO:Yb. These values can be attributed to the high carrier concentration for the third-calcined CMO:R (R = Dy, Yb) samples, as observed and discussed in Figure 6, where it can be deduced that the carrier concentration is significantly affected by multigrinding and multicalcination. Parts c and d of Figures 7 present the Seebeck coefficient for the multicalcined CMO:Dy and CMO:Yb samples, respectively. The Seebeck coefficient is negative for all samples in the whole measured temperature range, confirming the previous discussions showing that electrons are the dominant charge carriers. With an increase in the temperature, the absolute values of the Seebeck coefficient are increased, which is a typical behavior of a degenerate semiconductor, in agreement F

DOI: 10.1021/acs.inorgchem.8b01163 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 9. zT as a function of the temperature for multicalcined (a) Ca0.9Dy0.1MnO3−δ and (b) Ca0.9Yb0.1MnO3−δ ceramics.

in Figure 3, in the third-calcined sample, which has been subjected to more hand regrinding and calcination steps. These interfaces can increase middle-frequency phonon scattering. This effect of coherent interfaces on κL has been confirmed in many thermoelectric alloys in previous investigations and has become an effective way of improving the thermoelectric performance.33,34 Additionally, large O defects have been found in the third-calcined sample; the O defects, which act as point defect, can weaken the propagation of high-frequency phonons and further decrease κL. The effect of point defects on decreasing κL has also been reported in many previous works.33,39 Therefore, heterogeneous microstructures including large and small micrometer-sized grains, coherent interfaces, and O defects formed by multicalcination can effectively scatter the low-, middle-, and high-frequency phonons and obviously reduce κL. Consequently, the right heterogeneous microstructures can optimize the electrical properties and κL simultaneously and then enhance the thermoelectric zT for the CMO:R (R = Dy, Yb) samples, breaking the links between them in micrometer-grain-size bulk materials. The temperature dependence of zT for all samples is shown in Figure 9. In Figure 9a, the zT values of the multicalcined CMO:Dy samples increase with the temperature. The highest zT, 0.12, is obtained for the third-calcined sample at 971 K, which is enhanced 33% compared with the first-calcined CMO:Dy. zT for the CMO:Yb samples follow the same trend below approximately 700 K and then reach a stable value at higher temperatures, as shown in Figure 9b. Therefore, the maximum of zT, 0.14, is achieved for the third-calcined sample at 670 K, which is 27% higher than that of the first calcined CMO:Yb samples. Also, the repeated data, as shown in Figure S7, illustrated that the results are reproducible. Finally, in order to clearly show enhancement of the thermoelectric performances as a function of the multicalcination times, the maximum value of zT for all CMO:R (R = Dy, Yb) samples has been plotted in Figure 10a.

calcination process does not significantly affect the high values of the Seebeck coefficient for CMO:R (R = Dy, Yb), while the carrier concentration is modified with these processes. This is an important result because the electrical resistivity can be decreased through the multicalcination processes, while no effect on the Seebeck coefficient has been observed, unlinking these two parameters. From the measured values of the electrical resistivity and Seebeck coefficient, the PFs of CMO:R (R = Dy, Yb), as a function of the temperature, were calculated using S2/ρ, and are shown in Figure 8a,b. PF is increased up to approximately 700 K and then decreased with a further increase in the temperature in all samples. As can be easily deduced from the data presented previously, the third-calcined CMO:R (R = Dy, Yb) samples reach the highest PF, 240 μW K−2 m−1, for CMO:Dy at 676 K, and 339 μW K−2 m−1 for CMO:Yb at 678 K. These values are larger than that of the first-calcined sample, which suggests that these samples are better than the samples synthesized through conventional techniques. This enhancement of PF is due to the low electrical resistivity and high Seebeck coefficient achieved by the multicalcination process, as previously discussed. The temperature dependences of the total thermal conductivity κ (closed symbols) and lattice thermal conductivity κL (open symbols) for CMO:R (R = Dy, Yb) are shown in Figure 8c,d. κL was obtained by subtracting the electronic thermal conductivity (κe) from κ. κe was calculated using Wiedemann−Franz’s law: κe = LT/ρ, where the Lorentz constant L = 2.44 × 10−8 W Ω K−1. κL is approximately 80% of κ, being the main part of the thermal transport behavior, in agreement with previous works.19,25 κ and κL decrease with the temperature until around 700 K and then increase for higher temperatures, in all cases. The lowest κ values are 1.49 and 1.65 W m−1 K−1 and the lowest κL values are 1.24 and 1.22 W m−1 K−1 for the third-calcined CMO:Dy and CMO:Yb samples at 672 K, respectively, which are nearly 25% lower than those of the first-calcined samples. Furthermore, these κ values for multicalcined samples are almost the same as those obtained in Dy and Yb dual-doped CMO samples25 (about 1.47 W m−1 K−1). Thus, the third-calcined CMO:R (R = Dy, Yb) samples present the lowest κ and κL values among all samples, which are strongly related to the heterogeneous microstructures obtained through multicalcination, as previously mentioned. Overall, the thermal conductivity can be effectively optimized, and it is easy to decrease these values by the multicalcination process. The lowest κL value, achieved in the third-calcined CMO:R (R = Dy, Yb) samples, is associated with their right heterogeneous microstructures. The different microsized grains can enhance the scattering of low-frequency phonons. More coherent interfaces are formed by phase dispersion, as shown

Figure 10. (a) zT maximum versus calcination time plot. (b) Percentage variation of PFs, thermal conductivities (κ), and zT normalized to the CMO:R-1 (R = Dy, Yb) samples at 672 and 971 K respectively. G

DOI: 10.1021/acs.inorgchem.8b01163 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

ones. As a consequence, the multicalcination process with an optimized calcination time can be seen as a better way to form right heterogeneous microstructures, which will effectively optimize the thermoelectric performance of CMO oxides.

Obviously, these results indicate that the third-calcined samples show the highest zT in all multicalcined samples for both Dy- and Yb-doped CMO. However, the zT values of the second- and fourth-calcined samples for CMO:Dy are lower than those of the first-calcined ones, while for CMO:Yb, the zT values are hardly changed. The lower zT values for the secondand fourth-calcined CMO:Dy samples mainly result from their higher thermal conductivities than those for the first-calcined ones. This higher thermal conductivity for the second- and fourth-calcined CMO:Dy samples is related to the larger average grain size in the second-calcined samples and the lower percentage of grains in the 2−3 μm range for the fourthcalcined samples compared to the first-calcined ones, as shown in Figure S3, which lead to weaker phonon scattering and, consequently, to higher thermal conductivity. In the CMO:Yb samples, the second- and fourth-calcined zT values (0.1) are close to those of the first-calcined ones. This effect is because the carrier concentration and corresponding electrical properties are not improved. This is because of the slight differences in the element distributions for second- and fourth-calcined CMO:Yb samples, so that it does not increase the carrier concentration, compared to the first-calcined ones. To easily observe the contribution of the PF and thermal conductivity in the high zT at the third-calcined samples, the percentage variations in the PF, thermal conductivity, and zT, normalized to CMO:R-1 (R = Dy, Yb), are also shown in Figure 10b. From this graph, it is clear that the PF of the third-calcined CMO:Dy-3 sample is increased by 10.6%, and the thermal conductivity is decreased by 1.9%. Thus, the high zT of the third-calcined CMO:Dy-3 sample can be mainly associated with the increase of PF. On the other hand, in the thirdcalcined CMO:Yb sample, the PF is increased by 2.6%, but the thermal conductivity is decreased by 17.4%. Hence, the decrease of the thermal conductivity is the main factor that contributes to the higher zT. Therefore, in spite of obtaining the highest zT of the multicalcined CMO:R (R = Dy, Yb) samples at the third calcination process, the parameters responsible for the improvement are different. Essentially, the multicalcination-induced right heterogeneous microstructure is the core for thermoelectric property improvement. By optimization of multicalcination and multigrinding, the right heterogeneous microstructures are formed in the third-calcined samples, leading to synergistic optimization of the electrical and thermal properties, and finally the thermoelectric properties get enhanced.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.8b01163. Lattice parameters, relative densities, weight percentages, thermal diffusivities, heat capacities, grain-size distribution, EDS mapping of polished surfaces, EDS line scanning for all elements, and repeated data (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected] (H.W.). *E-mail: [email protected] (C.W.). ORCID

Hongchao Wang: 0000-0001-8731-9986 Andres Sotelo: 0000-0001-7056-0546 Author Contributions §

T.W. and P.N. contributed equally to this work and should be considered cofirst authors. The manuscript has been prepared by contributions of all authors. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The work is financially supported by the National Basic Research Program of China (2013CB632506), the Natural Science Fund of China under Grants 51501105, 51672159, and 51611540342, the Young Scholars Program of Shandong University under Grant 2015WLJH21, the China Postdoctoral Science Foundation under Grants 2015M580588 and 2016T90631, the Postdoctoral Innovation Foundation of Shandong Province under Grant 201603027, the Fundamental Research Funds of Shandong University under Grant 2015TB019, and the Foundation of the State Key Laboratory of Metastable Materials Science and Technology under Grant 201703.

■

4. CONCLUSIONS Dy and Yb single-doped CaMnO3−δ have been synthesized using a multicalcination method. With optimized calcination times and the distribution of grain sizes, suitable heterogeneous element distributions will be achieved. The carrier concentration is optimized in these heterogeneous microstructures, improving the PF after multiregrinding and multicalcination. At the same time, the heterogeneous microstructure with proper grain-size distributions can effectively scatter the low-, middle-, and high-frequency phonons, leading to lower κL. Therefore, the electrical performance and thermal conductivity are all optimized, and the thermoelectric performance is enhanced. The zT values for the third-calcined CMO:Dy samples reach 0.12, which is 33% higher than that of the first-calcined ones, whereas the highest zT, 0.14, is achieved for the third-calcined CMO:Yb samples, which is 27% enhanced compared with that of the first-calcined

REFERENCES

(1) Chu, S.; Cui, Y.; Liu, N. The path towards sustainable energy,. Nat. Mater. 2017, 16, 16−22. (2) Sorrell, S. Reducing energy demand: A review of issues, challenges and approaches,. Renewable Sustainable Energy Rev. 2015, 47, 74−82. (3) Snyder, G. J.; Toberer, E. S. Complex thermoelectric materials. Nat. Mater. 2008, 7, 105−114. (4) Bell, L. E. Cooling, heating, generating power, and recovering waste heat with thermoelectric systems. Science 2008, 321, 1457− 1461. (5) Wood, C. Materials for thermoelectric energy conversion. Rep. Prog. Phys. 1988, 51, 459−539. (6) Wang, H.; Su, W.; Liu, J.; Wang, C. Recent development of ntype perovskite thermoelectrics. J. Materiomics. 2016, 2, 225−236. (7) Zhu, T.; Liu, Y.; Fu, C.; Heremans, J. P.; Snyder, J. G.; Zhao, X. Compromise and synergy in high-efficiency thermoelectric materials. Adv. Mater. 2017, 29, 1605884. (8) Wang, H.; Hwang, J.; Zhang, C.; Wang, T.; Su, W.; Kim, H.; Kim, J.; Zhai, J.; Wang, X.; Park, H.; Kim, W.; Wang, C. Enhancement H

DOI: 10.1021/acs.inorgchem.8b01163 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry of the thermoelectric performances of bulk SnTe alloys via the synergistic effect of band structure modification and chemical bond softening. J. Mater. Chem. A 2017, 5, 14165−14173. (9) Wang, H.; Wang, C. Synthesis of Dy doped Yb0.1Ca0.9MnO3 ceramics with a high relative density and their thermoelectric properties. Mater. Res. Bull. 2012, 47, 2252−2256. (10) He, J.; Liu, Y.; Funahashi, R. Oxide thermoelectrics: The challenges, progress, and outlook,. J. Mater. Res. 2011, 26, 1762− 1772. (11) Terasaki, I.; Sasago, Y.; Uchinokura, K. Large thermoelectric power in NaCo2O4 single crystals. Phys. Rev. B: Condens. Matter Mater. Phys. 1997, 56, R12685. (12) Sotelo, A.; Constantinescu, G.; Rasekh, S.; Torres, M.; Diez, J.; Madre, M. Improvement of thermoelectric properties of Ca3Co4O9 using soft chemistry synthetic methods. J. Eur. Ceram. Soc. 2012, 32, 2415−2422. (13) Sotelo, A.; Rasekh, S.; Madre, M.; Guilmeau, E.; Marinel, S.; Diez, J. Solution-based synthesis routes to thermoelectric Bi2Ca2Co1.7Ox. J. Eur. Ceram. Soc. 2011, 31, 1763−1769. (14) Van Nong, N.; Pryds, N.; Linderoth, S.; Ohtaki, M. Enhancement of the thermoelectric performance of p-Type layered oxide Ca3Co4O9+δ through heavy doping and metallic nanoinclusions. Adv. Mater. 2011, 23, 2484−2490. (15) Li, J.; Sui, J.; Pei, Y.; Barreteau, C.; Berardan, D.; Dragoe, N.; Cai, W.; He, J.; Zhao, L. A high thermoelectric figure of merit zT > 1 in Ba heavily doped BiCuSeO oxyselenides,. Energy Environ. Sci. 2012, 5, 8543−8547. (16) Bocher, L.; Aguirre, M.; Logvinovich, D.; Shkabko, A.; Robert, R.; Trottmann, M.; Weidenkaff, A. CaMn1−xNbxO3 (x ≤ 0.08) perovskite-type phases as promising new high-temperature n-type thermoelectric materials. Inorg. Chem. 2008, 47, 8077−8085. (17) Ohtaki, M.; Koga, H.; Tokunaga, T.; Eguchi, K.; Arai, H. Electrical Transport Properties and High-Temperature Thermoelectric Performance of (Ca0.9M0.1 MnO3 (M= Y, La, Ce, Sm, In, Sn, Sb, Pb, Bi). J. Solid State Chem. 1995, 120, 105−111. (18) Kobayashi, T.; Takizawa, H.; Endo, T.; Sato, T.; Shimada, M.; Taguchi, H.; Nagao, M. Metal-insulator transition and thermoelectric properties in the system (R1−xCax) MnO3−δ(R: Tb, Ho, Y). J. Solid State Chem. 1991, 92, 116−129. (19) Wang, Y.; Sui, Y.; Fan, H.; Wang, X.; Su, Y.; Su, W.; Liu, X. High temperature thermoelectric response of electron-doped CaMnO3. Chem. Mater. 2009, 21, 4653−4660. (20) Wang, Y.; Sui, Y.; Su, W. High temperature thermoelectric characteristics of Ca0.9R0.1MnO3 (R = La, Pr, ..., Yb). J. Appl. Phys. 2008, 104, 093703. (21) Wang, H.; Hwang, J.; Snedaker, M. L.; Kim, I.-h.; Kang, C.; Kim, J.; Stucky, G. D.; Bowers, J.; Kim, W. High thermoelectric performance of a heterogeneous PbTe nanocomposite. Chem. Mater. 2015, 27, 944−949. (22) Sui, J.; Li, J.; He, J.; Pei, Y. L.; Berardan, D.; Wu, H.; Dragoe, N.; Cai, W.; Zhao, L. D. Texturation boosts the thermoelectric performance of BiCuSeO oxyselenides. Energy Environ. Sci. 2013, 6, 2916−2920. (23) Shen, J.; Zhu, T.; Zhao, X.; Zhang, S.; Yang, S.; Yin, Z. Recrystallization induced in situ nanostructures in bulk bismuth antimony tellurides: a simple top down route and improved thermoelectric properties. Energy Environ. Sci. 2010, 3, 1519−1523. (24) Mehdizadeh Dehkordi, A.; Bhattacharya, S.; Darroudi, T.; Graff, J. W.; Schwingenschlögl, U.; Alshareef, H. N.; Tritt, T. M. Large thermoelectric power factor in Pr-doped SrTiO3−δ ceramics via grainboundary-induced mobility enhancement. Chem. Mater. 2014, 26, 2478−2485. (25) Zhu, Y.; Wang, C.; Su, W.; Liu, J.; Li, J.; Zhang, X.; Mei, L. Influence of rare-earth elements doping on thermoelectric properties of Ca0.98Dy0.02MnO3 at high temperature. J. Solid State Chem. 2015, 225, 105−109. (26) Zhang, F.; Lu, Q.; Zhang, X.; Zhang. Preparation and improved electrical performance of the Pr-doped CaMnO3−δ thermoelectric compound. Phys. Scr. 2013, 88, 035705.

(27) De Smet, Y.; Deriemaeker, L.; Finsy, R. Ostwald ripening of alkane emulsions in the presence of surfactant micelles. Langmuir 1999, 15, 6745−6754. (28) Molinari, M.; Tompsett, D. A.; Parker, S. C.; Azough, F.; Freer, R. Structural, electronic and thermoelectric behaviour of CaMnO3 and CaMnO3−δ. J. Mater. Chem. A 2014, 2, 14109−14117. (29) Schrade, M.; Kabir, R.; Li, S.; Norby, T.; Finstad, T. G. High temperature transport properties of thermoelectric CaMnO3−δIndication of strongly interacting small polarons. J. Appl. Phys. 2014, 115, 103705. (30) Zhou, Y.; Matsubara, I.; Funahashi, R.; Xu, G.; Shikano, M. Influence of Mn-site doped with Ru on the high-temperature thermoelectric performance of CaMnO3‑δ. Mater. Res. Bull. 2003, 38, 341−346. (31) Wang, H.; Bahk, J. H.; Kang, C.; Hwang, J.; Kim, K.; Kim, J.; Burke, P.; Bowers, J. E.; Gossard, A. C.; Shakouri, A.; Kim, W. Right sizes of nano-and microstructures for high-performance and rigid bulk thermoelectrics. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 10949− 10954. (32) Medlin, D.; Snyder, G. Interfaces in bulk thermoelectric materials: a review for current opinion in colloid and interface science. Curr. Opin. Colloid Interface Sci. 2009, 14, 226−235. (33) Banik, A.; Vishal, B.; Perumal, S.; Datta, R.; Biswas, K. The origin of low thermal conductivity in Sn1‑xSbxTe: phonon scattering via layered intergrowth nanostructures,. Energy Environ. Sci. 2016, 9, 2011−2019. (34) Guin, S. N.; Negi, D. S.; Datta, R.; Biswas, K. Nanostructuring, carrier engineering and bond anharmonicity synergistically boost the thermoelectric performance of p-type AgSbSe2-ZnSe. J. Mater. Chem. A 2014, 2, 4324−4331. (35) Hu, L.; Zhu, T.; Liu, X.; Zhao, X. Point defect engineering of high performance bismuth telluride based thermoelectric materials. Adv. Funct. Mater. 2014, 24, 5211−5218. (36) Leonidov, I. A.; Konstantinova, E. I.; Patrakeev, M. V.; Chukin, A. V.; Kozhevnikov, V. L. Electron transport and mobility analysis in La/Sr co-doped CaMnO3‑δ. J. Solid State Electrochem. 2017, 21, 2099−2108. (37) Goldyreva, E. I.; Leonidov, I. A.; Patrakeev, M. V.; Kozhevnikov, V. L. Electron transport in CaMnO3−δ at elevated temperatures: a mobility analysis. J. Solid State Electrochem. 2013, 17, 1449−1454. (38) Schrade, M.; Kabir, R.; Li, S.; Norby, T.; Finstad, T. G. Hight temperature transport properties of thermoelectric CaMnO3‑δIndication of strongly interacting small polarons. J. Appl. Phys. 2014, 115, 103705. (39) Fu, L.; Yin, M.; Wu, D.; Li, W.; Feng, D.; Huang, L.; He, J. Large enhancement of thermoelectric properties in n-type PbTe via dual-site point defects. Energy Environ. Sci. 2017, 10, 2030−2040.

I

DOI: 10.1021/acs.inorgchem.8b01163 Inorg. Chem. XXXX, XXX, XXX−XXX