Exploring the Thermoelectric Performance of ... - ACS Publications

Feb 8, 2017 - Despite relatively low electrical performance, further exploration of Haldane gap materials for thermoelectric applications is encourage...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/IC

Exploring the Thermoelectric Performance of BaGd2NiO5 Haldane Gap Materials Narendar Nasani,† Carlos Miguel Oliveira Rocha,† Andrei V. Kovalevsky,‡ Gonzalo Otero Irurueta,† Sascha Populoh,§,∥ Philipp Thiel,§ Anke Weidenkaff,⊥ Fernando Neto da Silva,† and Duncan P. Fagg*,† †

Centre for Mechanical Technology and Automation (TEMA), Department of Mechanical Engineering, University of Aveiro, 3810-193 Aveiro, Portugal ‡ CICECOAveiro Institute of Materials, Department of Materials and Ceramic Engineering, University of Aveiro, 3810-193 Aveiro, Portugal § Empa, Materials for Energy Conversion, Ueberlandstrasse 129, CH-8600 Duebendorf, Switzerland ∥ Semiconductors, ABB Switzerland Ltd., Fabrikstrasse 3, CH-5600 Lenzburg, Switzerland ⊥ Materials Chemistry, Institute for Materials Science, University of Stuttgart, Heisenbergstrasse 3, DE-70569 Stuttgart, Germany ABSTRACT: One-dimensional Haldane gap materials, such as the rare earth barium chain nickelates, have received great interest due to their vibrant one-dimensional spin antiferromagnetic character and unique structure. Herein we report how these 1D structural features can also be highly beneficial for thermoelectric applications by analysis of the system CaxBaGd2‑xNiO5 0 ≤ x ≤ 0.25. Attractive Seebeck coefficients of 140−280 μV K−1 at 350− 1300 K are retained even at high acceptor-substitution levels, provided by the interplay of low dimensionality and electronic correlations. Furthermore, the highly anisotropic crystal structure of Haldane gap materials allows very low thermal conductivities, reaching only 1.5 W m−1 K−1 at temperatures above 1000 K, one of the lowest values currently documented for prospective oxide thermoelectrics. Although calcium substitution in BaGd2NiO5 increases the electrical conductivity up to 5−6 S cm−1 at 1150 K < T < 1300 K, this level remains insufficient for thermoelectric applications. Hence, the combination of highly promising Seebeck coefficients and low thermal conductivities offered by this 1D material type underscores a potential new structure type for thermoelectric materials, where the main challenge will be to engineer the electronic band structure and, probably, microstructural features to further enhance the mobility of the charge carriers.

1. INTRODUCTION

The development of advanced thermoelectric systems requires both high performing materials and adequate design of the converter according to its working conditions, e.g., temperature, heat flow, etc. There are various types of materials that have been evaluated and, in some cases, commercialized for thermoelectric applications, such as Bi2Te3, Bi2Se3, PbTebased thermoelectrics, intermetallic Zintl phases and halfHeusler alloys, skutterudite nanocomposites, and silicon-based materials.1,2,6−8 While the aforementioned materials show promising thermoelectric efficiency (value ZT ≥ 1) at lower or higher temperatures, they are prone to oxidizing atmospheres and release toxic products upon decomposition.1,2,7,9 In many cases, the natural abundance of the constituent elements is also a problem limiting the range of suitable applications. Moreover, the inability of these materials to operate at intermediate and high temperatures, where higher

Alternative energy systems have received large attention due to the growing worldwide energy demand and limited availability of fossil fuel resources. Among these, thermoelectric power systems are highly promising energy converters that can generate electricity directly by utilizing waste heat from heat engines, coal-based power plants, automobiles, wood stoves, etc.1,2 In these systems, there are no moving parts involved or production of any harmful gases (e.g., CO2, NOx) during power generation.3,4 The performance of thermoelectric systems is evaluated on the basis of the term “figure of merit (ZT)” which is represented as ZT = σ × S2 × T/κ, where σ is the electrical conductivity, S is thermopower (Seebeck) coefficient, T is absolute temperature, and κ is thermal conductivity.5 High thermoelectric efficiency is achieved only when the material exhibits high thermopower, high electrical conductivity, and low thermal conductivity. A significant improvement of ZT is very challenging, since the key properties are inversely related through the charge carrier concentration.6 © XXXX American Chemical Society

Received: January 12, 2017

A

DOI: 10.1021/acs.inorgchem.7b00049 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Carnot efficiency is expected, has promoted researchers to search for new types of eco-benign and stable thermoelectrics. Oxide-based thermoelectric materials have been reported for more than 50 years, but the work has dramatically increased in intensity during the past decade.10−12 Layered cobaltites, such as Ca3Co4O9 and NaCo2O4, have been highlighted to be interesting p-type thermoelectric materials in terms of high temperature applications due to their good thermoelectric performance and low price.13−15 Layered cobaltites are also known to be stable at high temperatures and are considered to be environmentally friendly.16 Promising thermoelectric properties have also been observed for n-type A-site cation deficient/nonstoichiometric or B-site-doped SrTiO3 and CaMnO317 ceramic oxide materials. For example, rare-earthand transition-metal-substituted SrTiO3 ceramics display enhanced thermoelectric performance, controlled by defect chemistry reactions.18−25 However, the development of efficient thermoelectric modules based on the above-mentioned materials is still at the preliminary stage with obtained ZT values for the best oxide thermoelectrics still being substantially below that required by many potential applications. In the pursuit of new materials for thermoelectric applications, rare-earth-doped barium chain nickelates (BaR2NiO5, R = Gd, Y, Sm) may offer promise due to their unique structure and vibrant thermal and electrical properties.26−32 The chain nickelates offer a one-dimensional (1D) spin Heisenberg antiferromagnetic character, referred to as Haldane gap compounds, that was first discovered by Haldane in the BaY2NiO5 system.32 Among the chain nickelates, BaY2NiO5 is well-studied in terms of structural, magnetic, and electrical properties26−28,31,33−35 with the electrical conductivity of BaY2NiO5 being quite low due to the absence of oxygen vacancies, with conduction occurring through a thermally activated polaron hopping mechanism.26,31 Nonetheless, the electrical conductivity can be enhanced by doping with alkaline earth elements, such as strontium and calcium, to alter the oxidation state of the nickel cation. Another interesting material from the family of chain nickelates is gadolinium-doped barium nickelate (BaGd2NiO5). BaGd2NiO5 is isostructural with BaY2NiO5, and of orthorhombic symmetry with space group Immm.36 The crystal structure of BaGd2NiO5 reveals that it has isolated chains of NiO6 flattened octahedras, BaO10 bicapped quadrangular prisms, and GdO7 capped trigonal prisms (Figure 1). The results of infrared optical spectroscopy and Mossbauer spectroscopy have shown BaGd2NiO5 to be antiferromagnetic at TN = 58 K but to transform to magnetic spin reorientation at TR = 23 K, where the nickel magnetic moments rotate from the a-axis toward the b-axis.37−39 Although the magnetic and structural properties of BaGd2NiO5 are well-studied, no electrical and thermal properties are currently available in the literature. Nonetheless, the twisted one-dimensional nature, Haldane energy gap, and unique structure of the BaGd2NiO5 material may also provide interesting prospects toward thermoelectric applications. In this work, BaGd2NiO5 as candidate material is assessed for the first time as a possible stable 1D material for intermediate temperature thermoelectric applications. The structural, thermal, and electrical properties of pristine and calcium-substituted BaGd2NiO5 materials are explored in detail.

Figure 1. Ball-and-stick view of the BaGd2NiO5 crystal structure.

2. EXPERIMENTAL PROCEDURE 2.1. Materials Synthesis and Processing. The acceptorsubstituted CaxBaGd2‑xNiO5 materials (x = 0−0.45) were synthesized by the traditional solid state method. Stoichiometric quantities of BaO2 (Sigma-Aldrich), Gd2O3 (Sigma-Aldrich), NiO (Strem Chemicals ABCR GmbH & Co.), and CaCO3 (Sigma-Aldrich) reactants were employed. All the powders were thoroughly mixed in ethanol solvent using an agate mortar and pestle to achieve a homogeneous greenish yellow powder. The obtained powder mixture was then dried in an oven at 353 K for 30 min and, subsequently, uniaxially pressed into pellets at 40 MPa in a 1.5 cm stainless steel die and sintered at 1273 K for 10 h with a heating/cooling rate of 4 K min−1 in air. For the purpose of phase analysis, the sintered pellets were crushed into powders. Dense samples were obtained for electrical and thermal measurements by pressing the powders isostatically at 450 MPa for 15 min, followed by sintering at 1673 K for 6 h with a heating/cooling rate of 4 K min−1 under ambient atmosphere. The density of final sintered samples was found to be ∼91−95% that of theoretical density except for x = 0.05 with ∼88%, calculated by measuring geometrical parameters and weight of the samples. For easy identification, CaxBaGd2‑xNiO5 samples are named CBGN00, CBGN05, CBGN10, etc., for x = 0.00, 0.05, 0.10, etc., respectively. 2.2. Characterization. The phase purity of all final powders was analyzed using powder X-ray diffraction (XRD), Philips X’Pert diffractometer, Detector X’Celerator, active length 2.5460°, step width 0.02°, and counting time 30 s step−1, operated at 45 kV and 40 mA with Cu Kα radiation in the angular range 2θ = 20−80°. The unit cell parameters were extracted using the Rietveld refinement method with Fullprof software. The in situ high temperature XRD studies were performed using an Xpert PRO XRD instrument operating with Cu Kα radiation, wavelength of 1.541 Å. In a typical case, the heating rate was 20 K min−1 between the data acquisition steps. The microstructure of dense pellets was observed using scanning electron microscopy (SEM) (model Hitachi SU-70). The homogeneity and elemental analysis of all the samples were examined by energy dispersive X-ray spectroscopy (EDAX), model Bruker Quantax-Germany. The X-ray photoelectron spectroscopy (XPS) instrument was used to probe the surface of the pellets and was equipped with a hemispherical electron energy analyzer (SPECS Phoibos 150), a delayline detector, and a monochromatic Al Kα (1486.74 eV) X-ray source in an ultrahigh vacuum (UHV) system with a base pressure of 2 × 10−10 mbar. High resolution spectra were recorded at normal emission takeoff angle and with a pass energy of 20 eV, which provides an overall instrumental peak broadening of 0.5 eV. The samples were heated to 470 K under UHV conditions to remove atmospheric B

DOI: 10.1021/acs.inorgchem.7b00049 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry contamination. XPS spectra were fitted by using Gaussian−Lorentzian components after removing a Shirley background. The electrical conductivity and Seebeck coefficient were measured simultaneously using bar-shaped pellets with dimensions ∼1.3 × 0.3 × 0.4 cm3, as described elsewhere.22 One sample was fixed vertically by a spring load force, in connection with two spiral platinum wires of Stype thermocouples, also acting as thermal voltage probes in order to measure the thermopower. A platinum wire heater was used to create a temperature gradient along the sample length. A precise temperature control was carried out by not exceeding a 1 mm distance between the sample ends and the thermocouple junctions. The second sample was placed in crosswise position, in an isothermal plane of the cell, to measure the electrical conductivity by four-probe dc technique. All measurements were performed under air atmosphere in the temperature range 400−1273 K and decreasing temperature steps of 50−80 K. The results of the Seebeck coefficient measurements were corrected in terms of voltage offset due to the sample temperature gradient and the influence of platinum wires. An equilibrium criterion less than 0.1% min−1 for electrical conductivity and less than 0.002 mW K−1 min−1 for Seebeck coefficient was used for relaxation rates after the temperature changes. The thermal diffusivity (D) measurements were performed on 1.00 mm thick disc-shaped pellets with ∼1.2 cm of diameter by laser flash technique with a Netzsch LFA 457 Microflash equipment. The specific heat capacity (cp) was measured using Netzsch DSC 404 C equipment. The thermal conductivity was determined indirectly by eq 1.

k = D × ρ × cp

Table 1. Unit Cell Parameters of of CaxBaGd2−xNiO5 Materials Synthesized by Solid State Method lattice params (Å) x (Ca)

a

b

c

0.0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45

3.78271 3.77927 3.77596 3.77348 3.77061 3.76865 3.76721 3.76688 3.76428 3.76215

5.83214 5.83203 5.83090 5.83162 5.83040 5.82877 5.82987 5.82947 5.82915 5.82707

11.48173 11.47253 11.46211 11.45623 11.44814 11.44235 11.44003 11.43727 11.43193 11.42462

continuing to decrease at still higher Ca contents, Figure 2. The decrease in lattice parameters (Table 1) and unit cell volume (Figure 3) cannot be explained on the basis of the relative ionic

(1)

3. RESULTS AND DISCUSSION 3.1. Phase Composition and Microstructural Features. The powder X-ray diffraction patterns exhibit single phases of BaGd2NiO5 for all the studied compositions, with no observable impurity peaks; representative compositions are shown in Figure 2. All the studied Ca x BaGd 2−x NiO 5

Figure 3. Variation of unit cell volume with the substitution level x(Ca) in CaxBaGd2−xNiO5 materials.

sizes of the cations, as the ionic size of Ca2+ (0.99 Å) is larger than that of Gd3+ (0.93 Å). Nonetheless, a decrease of lattice volume has also been noted previously in the related BaY2NiO5 system with Ca substitution and ascribed to the formation of Ni3+ as a charge compensation mechanism.26 Similar trends were also observed in Y2−xCaxBaNiO5 by Lannuzel et al.28 and in Y2−xSrxBaNiO5 materials by Nasani et al.26 Due to the discontinuity in lattice volume shown by the CaxBaGd2−xNiO5 system, which may reflect a potential change in charge compensation mechanism, the current article limits the subsequent study of electrical and thermal properties to compositions in the range 0 ≤ x ≤ 0.25. Even minor structural transformations, promoted by altering the external conditions, may lead to significant changes in thermoelectric properties. Thus, the phase composition in the whole studied temperature range was confirmed by in situ XRD analysis, performed for the samples at the limits of the chosen compositional range (CBGN00 and CBGN25), and an intermediate CBGN15 composition. Figure 4 shows typical patterns for CBGN15; similar results indicating the absence of structural changes in the discussed temperature range were obtained for other samples.

Figure 2. Powder X-ray diffraction patterns of CaxBaGd2−xNiO5 samples, calcined at 1673 K for 6 h in ambient air.

compositions show an orthorhombic system with Immm space group. The lattice parameters and unit cell volume were obtained by refining the raw powder diffraction patterns by the Rietveld method using Fullprof software, Table 1. Acceptable global agreement parameters of Rietveld refinement were achieved for all the diffraction patterns. The corresponding unit cell volumes are shown to pass through an inflection at the composition x = 0.25, while C

DOI: 10.1021/acs.inorgchem.7b00049 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 6. Temperature dependence of the Seebeck coefficient for CaxBaGd2−xNiO5 materials (x = 0−0.25). Figure 4. In situ XRD of CBGN15 sample with temperature.

to the changes in oxygen content, e.g., the minimum of Seebeck coefficient of La2NiO4+δ corresponds to the maximum oxygen content and, therefore, highest concentration of Ni3+ charge carriers.40 Most likely, this is also the case for CaxBaGd2−xNiO5 materials; however, additional structural and thermogravimetric studies are necessary to uncover the relevant mechanisms behind this behavior. In contrast to the current materials, in general, nickelates have previously been considered to be unsuitable for thermoelectric applications, predominantly due to very low Seebeck coefficients. As an example, maximum absolute values for La2NiO4+δ- and LaNiO3−δ-based systems usually do not exceed 20 μV/K,40−43 being almost 10 times lower than those obtained in the present work. This striking result is likely to be promoted by the interplay of low dimensionality and electronic correlations in Haldane gap materials, making them attractive as thermoelectrics. As would be expected, the substitution level has a pronounced effect on the Seebeck coefficient; namely, it rapidly decreases to half of its original value from CBGN00 to CBGN05. However, an increase of calcium concentration above x ≥ 0.10 does not result in a further decrease of the Seebeck coefficient, suggesting that the concentration of the charge carriers is not changing significantly. Some additional guidelines on this behavior can be obtained from the results of the XPS studies. Figure 7 shows high resolution XPS spectra of the Ba 3d5/2 (a), Ni 2p3/2 (b), Ca 2p (c), and Gd 3d5/2 (d) core levels for selected compositions CBGN00, CBGN10, and CBGN15. The respective binding energies (BEs) and full width at half-maximum (fwhm) values of all core levels are summarized in Table 2. The XPS spectra show that two components are clearly detected in the barium region, ascribed to Ba atoms of the nonsubstituted (CBGN00) and Ca-substituted BaGd2NiO5 (CBGN10 and CBGN15) samples, at higher BE, and barium carbonates (BaCO3), at lower BE. The presence of this carbonate is attributed to a surface contamination that is found to be a common surface impurity in barium-based materials, previously noted in literature studies.44,45 In the case of nickel, three components were needed for fitting the spectra of all the samples. The main component appearing at the lower BE is ascribed to Ni2+ while that at intermediate BE corresponds to Ni3+. The last component, at the higher BE, corresponds with a broad satellite centered at about +6 eV with respect to the main

Figure 5 represents the microstructures of fractured cross sections of CaxBaGd2−xNiO5 ceramics, observed by SEM. The

Figure 5. SEM micrographs of fractured CaxBaGd2−xNiO5 (x = 0, 0.10, 0.15, and 0.20) samples, sintered at 1673 K for 6 h.

pristine and Ca-substituted samples show highly dense microstructures without any noticeable porosity, in accordance with their measured density values of above 90% that of the theoretical. A slight increase in density and grain size was observed on increasing the Ca content in CaxBaGd2−xNiO5, while the typical microstructural features were similar in all samples, as shown in Figure 5. Further detailed SEM-EDAX analysis (not shown) evidenced that all the CaxBaGd2−xNiO5 samples possess highly homogeneous chemical composition, without any signs of segregation of the individual oxides. 3.2. Electrical Properties. CaxBaGd2−xNiO5 materials demonstrate very attractive values of the Seebeck coefficient, from ∼130 up to ∼540 μV/K, depending on the composition and temperature, Figure 6, with the positive sign of the Seebeck coefficient in the whole studied temperature range indicative of p-type behavior. Similar temperature dependencies of the thermopower with a minimum at ∼650−900 K were previously observed for La2NiO4+δ- and LaNiO3−δ-based materials40−42 and attributed D

DOI: 10.1021/acs.inorgchem.7b00049 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 7. High resolution XPS spectra of CaxBaGd2−xNiO5 materials (x = 0, 0.10, and 0.15) in (a) Ba 3d5/2, (b) Ni 2p3/2, (c) Ca 2p, and (d) Gd 3d5/2 core-level regions, respectively.

Table 2. Core-Level Binding Energies (BEs) of Different Chemical Components Determined by Curve Fitting of XPS Spectra of CaxBaGd2−xNiO5 Materials Ba 3d5/2 (BE, eVa)

a

Ni 2p3/2 (BE, eVa) a

composition

sample

BaCO3

Gd 3d5/2 (BE, eV )

CBGN00 CBGN10 CBGN15

781 (1.7) 781 (1.7) 781 (1.7)

780 (1.5) 779.6 (1.4) 779.7 (1.4)

1187.2 (5.3) 1187.2 (5.3) 1187.2 (5.3)

2+

Ni

855.16 (2.1) 855.0 (2.0) 855.0 (1.9)

Ca 2p3/2 (BE, eVa) 3+

Ni

857.0 (1.5) 856.7 (1.45) 856.7 (1.55)

CaCO3

sample

348.1 (1.8) 347.8 (1.9)

345.6 (1.5) 345.6 (1.7)

The numbers in brackets correspond to the fwhm (full width at half-maximum) of the components used for the fit.

component.46 The Ni2+/Ni3+ ratios are, thus, estimated from the fitted data for all the measured compositions and presented in Table 3. In the Ca 2p region, two set of components were

in agreement with that of the system Y2−xCaxBaNiO5 reported by Lannuzel et al.,28 with the extent of oxygen excess decreasing with increasing Ca content. The electrical conductivities of the Ca x BaGd 2−xNiO 5 materials show semiconducting behavior, increasing on heating for all compositions (Figure 8).

Table 3. Ratio of Nickel Cations in CaxBaGd2−xNiO5+δ Materials, Determined by the Areas of the Corresponding Peaks from XPS Data sample

Ni2+/Ni3+

Ni3+ (at. %)

δ

CBGN00 CBGN10 CBGN15

8.86 5.88 5.49

10 14.5 15.4

0.051 0.023 0.002

detected. Each set is formed by the Ca 2p3/2 and Ca 2p1/2 sublevels.47 The BE of the Ca 2p3/2 component of the first set (dark gray) is in good agreement with Ca atoms intercalated in the CBGN samples. On the other hand, the Ca 2p3/2 component of the second set (light gray) appears at a higher BE and is attributed to Ca atoms in CaCO3 that are present on the surface of the sample.48 Gd 3d5/2 spectra are very similar in all the samples, and formed by a broad peak and a satellite centered at about 1187 and 1197.0 eV, respectively.49,50 The observed BE value of 1187 eV in these samples for Gd 3d5/2 is very close to that recorded in Gd2O3.49,50 It is important to note that XPS is a surface characterization technique; hence, some uncertainties can be expected for a redox-sensitive Ni2+/Ni3+ couple due to sample pretreatment procedures. Nonetheless, the results suggest that only a moderate increase in Ni3+ concentration (Table 3) occurs from nonsubstituted CBGN00 to the CBGN10 samples, which may correspond to the observed decrease in the Seebeck coefficient between these compositions. Furthermore, the negligible variation of Ni3+ content observed from CBGN10 to CBGN15 also would concur with the observed similar Seebeck coefficients noted in these materials. Table 3 further indicates the tested compositions to be oxygen excess materials,

Figure 8. Temperature dependence of the electrical conductivity of CaxBaGd2−xNiO5 materials (x = 0−0.25).

The nonsubstituted CBGN00 shows the lowest conductivity. However, CBGN25 with the highest Ca content does not represent the highest electrical conductivity. Instead, the highest conductivity value (∼5.5 S/cm at 1300 K) is obtained for the CBGN15 composition at low temperatures, while the conductivities of CBGN20 and CBGN25 samples merge to offer a similar level at the highest temperatures measured. The activation energies (Ea’s) of all samples were calculated using E

DOI: 10.1021/acs.inorgchem.7b00049 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

correlated transition-metal-oxide-based systems and low thermal conductivity provided predominantly by lattice heat transport. The values of the power factor (PF = σ × α2), presented in Figure 9, demonstrate the combined effect of the electrical conductivity and Seebeck coefficient on the thermoelectric performance.

the Arrhenius equation in the temperature range 587−1123 K, and are documented in Table 4. The activation energy values Table 4. Activation Energy of CaxBaGd2−xNiO5 Materials in the Temperature Range 587−1123 K composition

activation energy, Ea (eV)

CBGN00 CBGN05 CBGN10 CBGN15 CBGN20 CBGN25

0.166 0.179 0.174 0.169 0.191 0.193

correspond to the range 0.16−0.19 eV for all the studied compositions and are comparable to other chain nickelate Haldane family materials, such as BaY2NiO5.26,31 Such low Ea values in these materials achieved even at higher temperatures dictate that the conduction occurs predominantly through a polaronic electron hopping mechanism, while any oxide−ion conduction can be ruled out, corresponding to the typical behavior of Haldane chain nickelates.26 Assuming the defect chemistry mechanisms proposed in the above cited works, the charge compensation for Ca substitution in CaxBaGd2−xNiO5 can be expressed as (standard Kröger− Vink notation51 is used)

Figure 9. Temperature dependence of the power factor for CaxBaGd2−xNiO5 materials (x = 0−0.25).

BaO + (2 − x)GdO1.5 + xCaO + NiO × → Ba ×Ba + xCa′Gd + (2 − x)Gd Gd + β Ni•Ni x − β •• ⎛ β⎞ × V O + ⎜5 − x − ⎟OO + (1 − β)Ni ×Ni + ⎝ 2 2⎠ x+β O2 (g) + (2) 4

The power factor of all the samples increases with increasing temperature. The CBGN15 composition represents the highest PF of ∼27 μW K−1 m−2 at ∼1300 K, provided by superior electrical conductivity and moderate Seebeck coefficient. The obtained values, however, are still 1−2 orders of magnitude lower than those for benchmark Ca3Co4O9- and SrTiO3-based thermoelectric materials.18,23,54 Nonetheless, the Haldane gap CaxBaGd2−xNiO5 materials offer remarkably low thermal conductivities, as shown in Figure 10, which are attractive for thermoelectric applications. Due to the low electrical conductivity, the electronic counterpart of the thermal conductivity is less than 1%, as estimated using the Wiedemann−Franz relation, for all studied compositions, and hence, the heat is transported mainly by

where β represents the contribution provided by the formation • of the charge carriers Ni3+ (NiNi ). However, the results obtained by fitting the XPS data suggest that other defect reactions may contribute in the case of CBGN10 and, especially, nonsubstituted CBGN00, where a noticeable amount of Ni3+ was detected, apparently due to intrinsic defect disorder. At the same time, on the basis of the measured fractions of Ni3+, the charge compensation through formation of oxygen vacancies is less likely to occur in the studied range x ≤ 0.25, in accordance with the previous observations.26,28 Although the charge carrier concentrations, estimated from the XPS results, are relatively high (∼5 × 1020 cm−3), the main limitation toward high thermoelectric performance is imposed by the low charge carrier mobility, apparently due to high degree of hole localization in 1D structures. Taking into account the known relationship between electrical conductivity and mobility (μ), μ = σ/(en), where e is the elementary charge and n is the charge carrier concentration, the estimated mobilities for CBGN00, CBGN10, and CBGN15 amount to 4.7 × 10−4, 1.5 × 10−3, and 5.5 × 10−3 cm2 V−1 s−1, indicating that substitution might be favorable for increasing the mobility. Still, these values are, at least, 2−3 orders of magnitude lower than those obtained for Ca3Co4O9 (0.10−0.15 cm2 V−1 s−1)52 and SrTiO3-based materials (0.1−7.0 cm2 V−1 s−1),18,53 corresponding to known prospective thermoelectric oxide families. Low carrier mobility at high carrier concentration is a common problem for oxides that, to a certain extent, can be compensated by a relatively high Seebeck coefficient in strongly

Figure 10. Temperature dependence of the thermal conductivity of CaxBaGd2−xNiO5 materials (x = 0−0.25). F

DOI: 10.1021/acs.inorgchem.7b00049 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

CBGN15 appears to be the best performing composition, provided mainly by the highest power factor. However, the observed difference in ZT with CBGN20 and CBGN25 is rather minor and is close to the experimental error. In terms of ZT, the obtained performance is roughly 20 times lower than that for the best known oxide bulk thermoelectrics. Still, the Haldane gap CaxBaGd2−xNiO5 shows very attractive values of Seebeck coefficient and low thermal conductivities, provided by specific 1D character of the Ni−O−Ni network and high anisotropy of the crystal structure. Regarding the thermoelectrics field, further research efforts for these materials should, thus, be directed toward enhancement of the charge carrier mobility and deeper understanding of the defect chemistry, especially in the case of heavily substituted compositions where “glass-like” thermal conductivity may be promoted.

lattice vibrations. In the whole measured temperature range, the thermal conductivity is below 2.5 W m−1 K−1, reaching ∼1.5 W m−1 K−1 above 1000 K. These values are among the lowest known for prospective oxide thermoelectrics, being comparable or even slightly below that for Ca3Co4O9 and 2−3 times lower than for SrTiO3-based materials,55 while the difference further increases on decreasing the temperature. The CBGN20 and CBGN25 samples with calcium content above the solid solution limit show “glass-like” thermal conductivity behavior with both low values and weak temperature dependence, as previously observed for several A-site deficient and heavily donor-substituted strontium titanates.18,24,56 This behavior is usually promoted by the highly defective structure; the latter apparently is also the case for CaxBaGd2−xNiO5 at high acceptor-substitution levels. Highly anisotropic structures are favorable for suppressing the lattice thermal conductivity due to enhanced phonon interactions and scattering. Hence, the Haldane gap materials obviously represent an interesting pathway toward efficient thermoelectrics, assuming that the charge carrier mobility in these compositions can be improved by at least 2 orders of magnitude. The compositional dependence of the thermal conductivity is another interesting issue to explore. The common tendency of the thermal conductivity to decrease on substitution due to alloy scattering is not followed when the calcium content is increased from x = 0.05 to 0.10. This feature suggests possible alterations of local structural features or defect chemistry reactions which, however, were not detected by XRD studies, while the electrical properties show rather expected compositional dependence in this range. As an example, nonregular thermal conductivity behavior with composition was also observed for donor-substituted SrTiO3-based materials,21 where the lattice thermal transport is strongly affected by the interplay of the phonon scattering abilities of various defect types, whose concentrations have a complex dependence on the chemical composition of the sample and external redox conditions. Additional neutron diffraction and/or thermogravimetric studies would be helpful to understand the reasons for such behavior. Finally, Figure 11 summarizes the impacts of electrical and thermal properties on overall thermoelectric performance, by showing the temperature dependence of the dimensionless figure-of-merit ZT.

4. CONCLUSIONS This work assessed the prospects of Haldane gap acceptorsubstituted BaGd2NiO5-based materials toward potential thermoelectric applications. The materials were synthesized via a traditional solid state method from corresponding oxides. The maximum solubility of Ca in CaxBaGd2−xNiO5 was achieved at x = 0.25. The values of Seebeck coefficients demonstrate weak temperature dependence, remaining above 140 μV/K even at high acceptor-substitution levels. Substitution with calcium resulted in a general increase of the electrical conductivity, with a maximum achieved for the Ca0.15BaGd0.85NiO5 composition. Combined analysis of the XPS and electrical conductivity data suggested that the p-type electronic transport in studied materials is mainly limited by the low charge carrier mobilities. Remarkably low thermal conductivities were observed for all studied materials, provided by a complex crystal lattice, with a pronounced tendency to “glass-like” behavior at high acceptor-substitution levels above x ≥ 0.20. Despite relatively low electrical performance, further exploration of Haldane gap materials for thermoelectric applications is encouraged by high Seebeck coefficient and low thermal conductivity values.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: +351-234-370953. Tel: +351-234370830. ORCID

Narendar Nasani: 0000-0002-6930-5369 Andrei V. Kovalevsky: 0000-0001-5814-9797 Anke Weidenkaff: 0000-0002-7021-1765 Duncan P. Fagg: 0000-0001-6287-9223 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research work was supported by the FCT, Project PTDC/ CTM-ENE 6319 2014, CICECO-Aveiro Institute of Materials UID/CTM/50011/2013, the FCT Investigator Programme, Projects IF/01344/2014/CP1222/CT0001 and IF/00302/ 2012, QREN, FEDER and COMPETE Portugal, and the European Union. The work was co- funded through the Sinergia Programme of the Swiss National Science Foundation (SNF) Division II (Grant CRSII2_136299/1Thermoelectric oxides TEO).

Figure 11. Thermoelectric figure of merit (ZT) for CaxBaGd2−xNiO5 materials (x = 0−0.25). G

DOI: 10.1021/acs.inorgchem.7b00049 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry



(22) Kovalevsky, A. V.; Yaremchenko, A. A.; Populoh, S.; Weidenkaff, A.; Frade, J. R. Enhancement of thermoelectric performance in strontium titanate by praseodymium substitution. J. Appl. Phys. 2013, 113, 053704. (23) Kovalevsky, A. V.; Yaremchenko, A. A.; Populoh, S.; Thiel, P.; Fagg, D. P.; Weidenkaff, A.; Frade, J. R. Towards a high thermoelectric performance in rare-earth substituted SrTiO3: effects provided by strongly-reducing sintering conditions. Phys. Chem. Chem. Phys. 2014, 16, 26946−26954. (24) Kovalevsky, A. V.; Populoh, S.; Patrício, S. G.; Thiel, P.; Ferro, M. C.; Fagg, D. P.; Frade, J. R.; Weidenkaff, A. Design of SrTiO3Based Thermoelectrics by Tungsten Substitution. J. Phys. Chem. C 2015, 119, 4466−4478. (25) Sagarna, L.; Shkabko, A.; Populoh, S.; Karvonen, L.; Weidenkaff, A. Electronic structure and thermoelectric properties of nanostructured EuTi1−xNbxO3−δ (x = 0.00; 0.02). Appl. Phys. Lett. 2012, 101, 033908. (26) Nasani, N.; Ramasamy, D.; Antunes, I.; Singh, B.; Fagg, D. P. Structural and electrical properties of strontium substituted Y2BaNiO5. J. Alloys Compd. 2015, 620, 91−96. (27) Alonso, J. A.; Rasines, I.; Rodriguez-Carvajal, J.; Torrance, J. B. Hole and Electron Doping of R2BaNiO5 (R = Rare Earths). J. Solid State Chem. 1994, 109, 231−240. (28) Lannuzel, F. X.; Janod, E.; Payen, C.; Ouvrard, G.; Moreau, P.; Chauvet, O.; Parent, P.; Laffon, C. Electronic structure of a hole doped oxide with a quasi-1D crystal structure Y2−x(Sr,Ca)xBaNiO5. J. Alloys Compd. 2001, 317−318, 149−152. (29) Darriet, J.; Regnault, L. P. The compound Y2BaNiO5: A new example of a haldane gap in A S = 1 magnetic chain. Solid State Commun. 1993, 86, 409−412. (30) García-Matres, E.; Martínez, J. L.; Rodríguez-Carvajal, J.; Alonso, J. A.; Salinas-Sánchez, A.; Saez-Puche, R. Structural Characterization and Polymorphism of R2BaNiO5 (R = Nd, Gd, Dy, Y, Ho, Er, Tm, Yb) Studied by Neutron Diffraction. J. Solid State Chem. 1993, 103, 322−333. (31) Chen, J. W.; Narsinga Rao, G.; Li, K. W. Giant dielectric response of Haldane gap compound Y2BaNiO5. J. Appl. Phys. 2012, 111, 064111. (32) Haldane, F. D. M. Nonlinear Field Theory of Large-Spin Heisenberg Antiferromagnets: Semiclassically Quantized Solitons of the One-Dimensional Easy-Axis N\’eel State. Phys. Rev. Lett. 1983, 50, 1153−1156. (33) Fagot-Revurat, Y.; Malterre, D.; Lannuzel, F. X.; Janod, E.; Payen, C.; Gavioli, L.; Bertran, F. Anomalous spectral weight in photoemission spectra of the hole-doped Haldane chain Y2‑xSrxBaNiO5. Phys. Rev. B: Condens. Matter Mater. Phys. 2003, 67, 125118. (34) Castañer, R.; Prieto, C.; de Andrés, A.; Martínez, J. L.; SáezPuche, R. X-ray absorption spectroscopy and magnetization study of the Y2−x CaxBaNiO5 oxides. J. Alloys Compd. 1994, 210, 31−35. (35) Novák, P.; Boucher, F.; Gressier, P.; Blaha, P.; Schwarz, K. Electronic structure of the mixed valence system (YM)2BaNiO5 (M = Ca,Sr). Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 63, 235114. (36) Amador, J.; Gutiérrez Puebla, E.; Monge, M. A.; Rasines, I.; Campa, J. A.; Gomez de Salazar, J. M.; Ruiz Valero, C. Crystal structure of BaGd2NiO5 prepared by reaction of Ni metal with BaCO3 and Gd2O3. Solid State Ionics 1989, 32−33, 123−125. (37) Klimin, S. A.; Galkin, A. S.; Popova, M. N. Magnetic structures of the rare-earth chain nickelates BaNiO5 as studied by the optical spectroscopy of the Er probe. Phys. Lett. A 2012, 376, 1861−1865. (38) Sulewski, P. E.; Cheong, S. W. Raman-active phonons of the spin-1 chain antiferromagnets R2BaNiO5 (R = Gd and Y). Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 551−554. (39) Klimin, S. A.; Kuzmenko, A. B.; Popova, M. N.; Malkin, B. Z.; Telegina, I. V. Infrared study of lattice and magnetic dynamics in a spin-chain compound Gd2BaNiO5. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 82, 174425.

REFERENCES

(1) Zheng, X. F.; Liu, C. X.; Yan, Y. Y.; Wang, Q. A review of thermoelectrics research − Recent developments and potentials for sustainable and renewable energy applications. Renewable Sustainable Energy Rev. 2014, 32, 486−503. (2) Gayner, C.; Kar, K. K. Recent advances in thermoelectric materials. Prog. Mater. Sci. 2016, 83, 330−382. (3) Bell, L. E. Cooling, Heating, Generating Power, and Recovering Waste Heat with Thermoelectric Systems. Science 2008, 321, 1457. (4) DiSalvo, F. J. Thermoelectric Cooling and Power Generation. Science 1999, 285, 703. (5) Rowe, D. M. Modules, Systems, and Applications in Thermoelectrics, 1st ed.; CRC Press Taylor & Francis, 2012; Vol. 2, p 581. (6) Snyder, G. J.; Toberer, E. S. Complex thermoelectric materials. Nat. Mater. 2008, 7, 105−114. (7) Ohta, H.; Sugiura, K.; Koumoto, K. Recent Progress in Oxide Thermoelectric Materials: p-Type Ca3Co4O9 and n-Type SrTiO3. Inorg. Chem. 2008, 47, 8429−8436. (8) Populoh, S.; Aguirre, M. H.; Brunko, O. C.; Galazka, K.; Lu, Y.; Weidenkaff, A. High figure of merit in (Ti,Zr,Hf)NiSn half-Heusler alloys. Scr. Mater. 2012, 66, 1073−1076. (9) Gałązka, K.; Populoh, S.; Sagarna, L.; Karvonen, L.; Xie, W.; Beni, A.; Schmutz, P.; Hulliger, J.; Weidenkaff, A. Phase formation, stability, and oxidation in (Ti, Zr, Hf)NiSn half-Heusler compounds. Phys. Status Solidi A 2014, 211, 1259−1266. (10) Wood, C. High-temperature thermoelectric energy conversionII. Materials survey. Energy Convers. Manage. 1984, 24, 331− 343. (11) Vickery, R. C.; Muir, H. M. Thermoelectric properties of rare earth chalcogenides. Adv. Energy Convers. 1961, 1, 179−186. (12) Koumoto, K.; Funahashi, R.; Guilmeau, E.; Miyazaki, Y.; Weidenkaff, A.; Wang, Y.; Wan, C. Thermoelectric Ceramics for Energy Harvesting. J. Am. Ceram. Soc. 2013, 96, 1−23. (13) Wang, S.; Wang, H.; Zhang, H.; Fu, G.; Wang, J.; Zhao, X.; Yu, W. Fabrication and high temperature thermoelectric properties of caxis oriented Na0.68CoO:Ag nanocomposite thin films. Mater. Res. Bull. 2014, 50, 161−164. (14) Miyazawa, K.; Amaral, F.; Kovalevsky, A. V.; Graça, M. P. F. Hybrid microwave processing of Ca3Co4O9 thermoelectrics. Ceram. Int. 2016, 42, 9482−9487. (15) Moser, D.; Karvonen, L.; Populoh, S.; Trottmann, M.; Weidenkaff, A. Influence of the oxygen content on thermoelectric properties of Ca3−xBixCo4O9+δ system. Solid State Sci. 2011, 13, 2160− 2164. (16) Butt, S.; Liu, Y.-C.; Lan, J.-L.; Shehzad, K.; Zhan, B.; Lin, Y.; Nan, C.-W. High-temperature thermoelectric properties of La and Fe co-doped Ca−Co−O misfit-layered cobaltites consolidated by spark plasma sintering. J. Alloys Compd. 2014, 588, 277−283. (17) Bocher, L.; Aguirre, M. H.; Logvinovich, D.; Shkabko, A.; Robert, R.; Trottmann, M.; Weidenkaff, A. CaMn1−x NbxO3 (x ≤ 0.08) Perovskite-Type Phases As Promising New High-Temperature n-Type Thermoelectric Materials. Inorg. Chem. 2008, 47, 8077−8085. (18) Yaremchenko, A. A.; Populoh, S.; Patrício, S. G.; Macías, J.; Thiel, P.; Fagg, D. P.; Weidenkaff, A.; Frade, J. R.; Kovalevsky, A. V. Boosting Thermoelectric Performance by Controlled Defect Chemistry Engineering in Ta-Substituted Strontium Titanate. Chem. Mater. 2015, 27, 4995−5006. (19) Srivastava, D.; Norman, C.; Azough, F.; Schafer, M. C.; Guilmeau, E.; Kepaptsoglou, D.; Ramasse, Q. M.; Nicotra, G.; Freer, R. Tuning the thermoelectric properties of A-site deficient SrTiO3 ceramics by vacancies and carrier concentration. Phys. Chem. Chem. Phys. 2016, 18, 26475−26486. (20) Lu, Z.; Zhang, H.; Lei, W.; Sinclair, D. C.; Reaney, I. M. HighFigure-of-Merit Thermoelectric La-Doped A-Site-Deficient SrTiO3 Ceramics. Chem. Mater. 2016, 28, 925−935. (21) Kovalevsky, A. V.; Yaremchenko, A. A.; Populoh, S.; Weidenkaff, A.; Frade, J. R. Effect of A-Site Cation Deficiency on the Thermoelectric Performance of Donor-Substituted Strontium Titanate. J. Phys. Chem. C 2014, 118, 4596−4606. H

DOI: 10.1021/acs.inorgchem.7b00049 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry (40) Tichy, R.; Huang, K.; Goodenough, J. Solid-State Ionic Devices II, Ceramic Sensors: Proceedings of the International Symposium; Electrochemical Society: Pennington, NJ, 2000. (41) Zhou, N.; Chen, G.; Zhang, H. J.; Zhou, C. Synthesis and transport properties of La2NiO4. Phys. B 2009, 404, 4150−4154. (42) Tak, J.-Y.; Choi, S.-M.; Seo, W.-S.; Cho, H. K. Thermoelectric properties of a doped LaNiO3 perovskite system prepared using a spark-plasma sintering process. Electron. Mater. Lett. 2013, 9, 513−516. (43) Funahashi, R.; Urata, S.; Kitawaki, M. Exploration of n-type oxides by high throughput screening. Appl. Surf. Sci. 2004, 223, 44−48. (44) Fetisov, A. V.; Kozhina, G. A.; Estemirova, S. K.; Fetisov, V. B.; Mitrofanov, V. Y.; Uporov, S. A.; Vedmid’, L. B. XPS Study of Mechanically Activated YBa2Cu3O6+δ; and NdBa2Cu3O6+δ. J. Spectrosc. 2013, Vol. 1, pp 1−13. article no. 217268. (45) López, M. d. C. B.; Fourlaris, G.; Rand, B.; Riley, F. L. Characterization of Barium Titanate Powders: Barium Carbonate Identification. J. Am. Ceram. Soc. 1999, 82, 1777−1786. (46) Gottschall, R.; Schöllhorn, R.; Muhler, M.; Jansen, N.; Walcher, D.; Gütlich, P. Electronic State of Nickel in Barium Nickel Oxide, BaNiO3. Inorg. Chem. 1998, 37, 1513−1518. (47) Kulkarni, P.; Kulkarni, S. K.; Nigavekar, A. S.; Agarwal, S. K.; Awana, V. P. S.; Narlikar, A. V. Superconductivity in Ni substituted Bi2Ca1Sr2Cu2−xNixOy. Phys. C 1990, 166, 530−534. (48) Baer, D. R.; Moulder, J. F. High Resolution XPS Spectrum of Calcite (CaCO3). Surf. Sci. Spectra 1993, 2, 1−7. (49) Raiser, D.; Deville, J. P. Study of XPS photoemission of some gadolinium compounds. J. Electron Spectrosc. Relat. Phenom. 1991, 57, 91−97. (50) Söderlind, F.; Selegård, L.; Nordblad, P.; Uvdal, K.; Käll, P.-O. Sol−gel synthesis and characterization of polycrystalline GdFeO3 and Gd3Fe5O12 thin films. J. Sol-Gel Sci. Technol. 2009, 49, 253−259. (51) Kröger, F. A.; Vink, H. J. In Solid State Physics; Frederick, S., David, T., Ed.; Academic Press, 1956; Vol. 3, pp 307−435. (52) Schrade, M.; Fjeld, H.; Finstad, T. G.; Norby, T. Electronic Transport Properties of [Ca2CoO3−δ]q[CoO2]. J. Phys. Chem. C 2014, 118, 2908−2918. (53) Yamamoto, M.; Ohta, H.; Koumoto, K. Thermoelectric phase diagram in a CaTiO3−SrTiO3−BaTiO3 system. Appl. Phys. Lett. 2007, 90, 072101. (54) Madre, M. A.; Costa, F. M.; Ferreira, N. M.; Sotelo, A.; Torres, M. A.; Constantinescu, G.; Rasekh, S.; Diez, J. C. Preparation of highperformance Ca3Co4O9 thermoelectric ceramics produced by a new two-step method. J. Eur. Ceram. Soc. 2013, 33, 1747−1754. (55) Backhaus-Ricoult, M.; Rustad, J.; Moore, L.; Smith, C.; Brown, J. Semiconducting large bandgap oxides as potential thermoelectric materials for high-temperature power generation? Appl. Phys. A: Mater. Sci. Process. 2014, 116, 433−470. (56) Popuri, S. R.; Scott, A. J. M.; Downie, R. A.; Hall, M. A.; Suard, E.; Decourt, R.; Pollet, M.; Bos, J. W. G. Glass-like thermal conductivity in SrTiO3 thermoelectrics induced by A-site vacancies. RSC Adv. 2014, 4, 33720−33723.

I

DOI: 10.1021/acs.inorgchem.7b00049 Inorg. Chem. XXXX, XXX, XXX−XXX