Effects of Yttrium and Iron co-doping on the high temperature thermoelectric properties of Ca3Co4O9+δ

Accepted Manuscript Effects of Y and Fe co-doping on the High Temperature Thermoelectric Properties of Ca3Co4O9+δ NingYu Wu, Ngo Van Nong, Nini Pryds, Søren Linderoth PII: DOI: Reference:

S0925-8388(15)00637-4 http://dx.doi.org/10.1016/j.jallcom.2015.02.185 JALCOM 33564

To appear in:

Journal of Alloys and Compounds

Received Date: Revised Date: Accepted Date:

12 November 2014 21 February 2015 25 February 2015

Please cite this article as: N. Wu, N.V. Nong, N. Pryds, S. Linderoth, Effects of Y and Fe co-doping on the High Temperature Thermoelectric Properties of Ca3Co4O9+δ , Journal of Alloys and Compounds (2015), doi: http:// dx.doi.org/10.1016/j.jallcom.2015.02.185

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Effects of Y and Fe co-doping on the High Temperature Thermoelectric Properties of Ca3Co4O9+δ NingYu Wu, Ngo Van Nong, Nini Pryds, Søren Linderoth Department of Energy Conversion and Storage, Technical University of Denmark, Frederiksborgvej 399, 4000 Roskilde, Denmark Corresponding author: NingYu Wu, Tel.: +45-46774942, E-mail address: [email protected]

Abstract A series of Fe and Y co-doped Ca3-xYxCo4-yFeyO9+δ (0 £ x £ 0.3, 0 £ y £ 0.1) samples synthesized by auto-combustion reaction and followed by a spark plasma sintering (SPS) processing with the effects of Fe and Y doping on the high temperature (RT to 800 °C) thermoelectric properties were systematically investigated. For the Fe-doped system (x = 0, y £ 0.1), the electrical resistivity (ρ) decreased over the whole measured temperature range, while the Seebeck coefficient (S) remained almost the same. For the co-doped system, at any fixed Fe doping content, both ρ and S tended to increase with increasing Y dopants, however, the effect is more substantial on ρ than on S, particularly in the low temperature regime. In contrast to ρ and S, the in-plane thermal conductivity (κ) is only slightly influenced by Y and Fe substitutions. Among all the investigated samples, the co-doped sample with x = 0.1 and y = 0.03 showed a decrease of ρ, enhanced power factor over the measured temperature range, and improved ZT at 800 °C as compared to un-doped Ca3Co4O9+δ.

Keywords: thermoelectric, Ca3Co4O9, co-doped, yttrium, iron, auto-combustion, spark plasma sintering

1

1. Introduction Thermoelectric (TE) materials arouse a great deal of interest owing to the possibility of directly converting thermal energy into electrical energy via the Seebeck effect. The efficiency of TE materials can is proportional to the dimensionless figure-of-merit

=

⁄

, which

consists of the Seebeck coefficient (S), electrical resistivity (ρ), thermal conductivity (κ) and absolute temperature (T). Oxides are one of the promising candidate material classes for high temperature TE applications due to low cost, abundance of the constituent elements, and hightemperature structural and chemical stabilities in air [1,2]. In this class, p-type Ca3Co4O9+δ has gained much attention after the discovery by Shikano et al. with a ZT of 0.83 (at 800 °C) for a single crystalline sample as a promising high temperature thermoelectric p-type oxide material [1]. Ca3Co4O9+δ (abbreviated Co349 in the following text) possesses a misfit-layered structure with a CdI2-type hexagonal CoO2 subsystem and a rock salt-type Ca2CoO3 subsystem that are alternately stacked along the c-axis with identical a, c and β parameters but different and incommensurate b parameters. Therefore, this misfit-layered oxide can be described as [Ca2CoO3][CoO2](b1/b2) with a b1 to b2 ratio of approximately 1.62, where b1 to b2 are two lattice parameters for the rock salt and CoO2 subsystems respectively [3,4]. As a p-type material, the hole charge conduction takes place mainly within the CoO2 layer with the Ca2CoO3 layer serving as a charge reservoir, and the misfit between layers is expected to hinder the phonon transport [5,6]. Though Co349 single crystals have been known to exhibit competitive ZT values at high temperatures, polycrystalline Co349 exhibits a much lower ZT value of about 0.2 at 800 °C [7]. To improve the ZT of the polycrystalline Co349, many efforts have been made by atomic substitutions for Ca or Co with the purpose of adjusting the carrier concentration or hindering phonon transport via mass fluctuation in order to optimize ZT [2,8–21]. On the basis of the reported data, it shows that the substitution of Y on the Ca-site results in some improvements of the TE properties of Co349 [14,15,17]. Y3+ substitution for Ca2+ ions would reduce the Co effective valence states in Co349 leading to a decrease in the hole (i.e., Co4+) concentration and therefore an enhancement of S. Additionally, since Y3+ is heavier than Ca2+, Y substitution may diminish the phonon mean free path resulting in the reduction of κ. An optimal Y doping content of 0.2 was often reported [14,15]. Y dopants may therefore induce merits to enhance S and reduce κ, but the increase of ρ still remains a critical issue that need to be mitigated to achieve an enhancement in ZT [14,15,17]. Many reports have surveyed the transition metals such as Cr, Mn, Fe, Ni, Cu, and Zn as possible dopants for Co with respect to the TE properties of Co349 [11,13,18–21]. The influence of substitutions on the Co-site is more complicated than on the Ca-site in Co349 due to the intrinsic feature of multiple oxidation states of the transition metals as dopants as well as the presence of the two non-equivalent Co sites in the Ca2CoO3 and CoO2 subsystems [11,13,16,18,21–23]. Among the above mentioned dopants, the Fe-doped Ca3Co4-xFexO9+δ (x ≤ 0.2) system was reported to show a simultaneous increase of S and decrease of ρ in the temperature regime below RT [18,20]. The reason for S increase was interpreted to be due to a strong electron correlation effect, while the decrease in ρ was attributed to the change in carrier

2

concentration [11,18,19]. Nevertheless, to authors’ knowledge, hardly any references of the systematic investigations on Fe-doped Co349 high temperature TE properties could be found. In this work, a series of Co349 with various Fe/Y co-doping contents were prepared in order to investigate the effect of these elements on the high temperature thermoelectric properties. Firstly, a series of the Fe-doped Ca3Co4-yFeyO9+δ (0 ≤ y ≤ 0.1) was investigated, and secondly, Fe/Y co-doped Ca3-xYxCo4-yFeyO9+δ (0 £ x £ 0.3, 0 £ y £ 0.1) systems were prepared and characterized. The thermoelectric properties from RT to 800 oC and the co-doping effects are presented and discussed.

2. Experimental Procedure Polycrystalline Ca3Co4O9+δ samples with Fe doping, and Fe/Y co-doping were prepared by an auto-combustion synthesis technique similar to that which is described in Ref. [7]. A stoichiometric ratio of analytical regent grade (99+ %, Sigma-Aldrich) Ca(NO3)2·4H2O, Co(NO3)2·6H2O, Y(NO3)3·6H2O and Fe(NO3)3·9H2O were dissolved in de-ionized water with a specific amount of citric acid (99+ %, Sigma-Aldrich) to keep the citrate–metal cation molar ratio at 1. Additionally, the citrate-to-nitrate molar ratio was maintained at 0.40 by introducing NH4NO3 (98+ %, Sigma-Aldrich). After drying the solution until a uniform viscous gel was obtained, the hot plate temperature was raised to about 250 °C to initiate the auto-combustion process. The resulting powders were calcined at 750 °C for 2 hours to obtain single phase Ca3Co4O9+δ as confirmed by X-ray diffraction. Subsequently, the powders were consolidated by a spark plasma sintering (SPS, Dr. Sinter SPS-515S, Fuji Electronic Industrial Co., Ltd.). Powders were poured into a 12.7 mm diameter graphite die to fabricate 10 mm thick pellets. A pulsed electric current was then passed through the powders under vacuum (10−3 bar) to consolidate the material. Based on previous results, the sintering temperature was chosen to be 800 °C, with a fixed uniaxial pressure of 50 MPa and a ramping rate of 100 °C/min. The sintering time was kept constant for 5 min [7,24]. X-ray powder diffraction (XRD) patterns at room temperature were collected using a Bruker D8 diffractometer with Cu Kα-radiation to identify the phase composition of as-synthesized powders and SPS-sintered bulks. Microstructural analysis of the fracture surfaces of SPS samples was conducted using a ZEISS Supra 35 scanning electron microscope (SEM). TE transport properties presented in this work were measured along the direction perpendicular to the SPS pressure axis (in-plane direction) [7]. The Seebeck coefficient (S) and electrical resistivity (ρ) measurements were carried out simultaneously with an ULVAC-RIKO ZEM3 from room temperature up to 800 °C under a low-pressure helium atmosphere. The thermal conductivity was calculated using the equation κ = ρ·α·cp (where ρ, α and cp are the density, thermal diffusivity and specific heat capacity, respectively). The thermal diffusivity—along the in-plane direction perpendicular to the SPS pressure axis—was obtained under vacuum in a NETZSCH LFA-457 laser flash system. The cp in this work was taken to be the temperature independent Dulong-Petit limit. The bulk density of each sample was obtained with a Micromeritics AccuPyc 1340 gas

3

pycnometer as an average of ten measurements. All samples exhibited densities more than 99 % of the theoretical densities. The theoretical density 4.68 g/cm 3 for the un-doped Co349 was determined based on the lattice parameters from Masset et al. [3]. Variations in the TE properties (ρ, S, κ) of multiple samples taken from the same composition (sample-to-sample measurement error) remained below 1 %, which was taken into consideration in the subsequent comparison and discussion.

3. Result and discussion XRD analysis revealed that all Fe-doped and Fe/Y co-doped samples are single phase of Co349. Fig. 1 shows typical XRD patterns obtained for the as-synthesized and the SPS-sintered samples of the un-doped Co349, Fe-doped Ca3Co3.9Fe0.1O9+δ (Fig. 1(a)) and co-doped Ca2.7Y0.3Co3.9Fe0.1O9+δ (Fig. 1(b)). All peaks were indexed as Ca3Co4O9+d according to the JCPDS PDF # 21-0139 and the XRD pattern reported by Masset et al. [3]. Fig. 2 shows a set of SEM micrographs of fracture surfaces of the un-doped Co349 and co-doped Ca 2.7Y0.3Co3.9Fe0.1O9+δ after the SPS processing. A fine platelet morphology with elongation along the direction perpendicular to the SPS pressure axis is observed in these two micrographs. All samples exhibit high densities of more than 99 % of the theoretical density. These values are in agreement with the works of Liu et al. and Kenfaui et al. as well as our previous work [7,25,26]. Fig. 3 shows the resistivity (ρ) as a function of temperature for the Fe-doped Ca3Co4yFeyO9+δ

with 0 ≤ y ≤ 0.1 (Fig. 3(a)), and for the Fe/Y co-doped Ca3-xYxCo4-yFeyO9+δ series with

varied Y content (0.1 ≤ x ≤ 0.3) and fixed Y content (Fig. 3(b – d)). The resistivity of the un-doped sample is also plotted for comparison. The Fe-doped samples exhibit lower resistivity values than the un-doped Co349 over the whole measured temperature range (Fig. 3(a)). This resulting decrease in ρ upon Fe doping is similar to the reported observations below RT [11,13,18–20]. Fe substitution in Co349 is believed to be the origin of the lower ρ. Since the nature of multiple valence states of translation metals, usual valence states of Fe ion are +2 and +3 in low spin state, and the radii are 0.61 Å and 0.55 Å in six-coordination respectively. These radii are closed to those of the Co3+ 0.545 Å in low spin state and the Co4+ 0.53 Å in high spin state [27]. It indirectly justifies the speculation that Fe dopants prefer to substitute for Co ions locating in Co3+ / Co4+ mixed CoO2 subsystem thus conduce to increasing hole carriers, and the highest was found at y = 0.05 or 0.10 while it decreased with increasing Fe dopants when 0.10 ≤ y ≤ 0.20 [11,18,19]. The correlation between the unoccupied Co 3d and ρ with Fe doping was firstly found by Liu et al. [19]. It indicated that more unoccupied Co 3d accompanied higher ρ, which was opposite to the fact of Co349 as a p-type material with holes carriers. The increasing O 2p unoccupied states and oxygen content upon Fe doping should be responsible for increasing hole concentration. The subsequent research conducted by Wu et al. [11] presented the fact that Fe substitutions for Co ions locating in Ca2CoO3 subsystem and the valence states for Fe and Co ions were ~3+ and ~2+ respectively, suggesting that substitutions should be regarded as electron dopants. The resulting ρ was contributed from two factors: (1) a competition between electron doping and oxygen content change and (2) more ordered structure in the conducting CoO2 layer upon Fe doping may increase

4

carrier mobility since the suppressed carrier scattering [11]. In Fig. 3(a), the Fe-doped samples exhibited lower ρ values than the pure Co349 at all measured temperatures and no clear monotonic change in ρ was observed with Fe dopant content, confirming that the influence of Fe-doping on reducing ρ can be remained above RT. At high temperatures (e.g., above 400 oC) the effect of Fe doping on lowering the ρ seems to be minor. Schrade et al. [28] and Morita et al. [29] found that up to ≈ 5 % of the oxygen ions in the Co349 can be lost under a low oxygen partial pressure or nitrogen atmosphere. The removal of oxygen content accompanies with the change in carrier concentration and the formation of oxygen vacancies; it can be expressed as

2ℎ

·

+ O× ↔ 2

(

)/

+ V ·· + 1 2 O

( )

(1)

where p denotes the effective hole charge and V ·· is an oxygen vacancy. The vacancy formation

provides free electrons leading to the reduction of hole carriers. Since in this work the ρ measurements were performed under a low-pressure helium atmosphere, the constituent oxygen ions loss induced changes in carrier concentration and microstructure might be responsible for the observed converging ρ at high temperatures but further investigation necessitates. In Fig. 3(b – d), a clear tendency of the increasing ρ with Y dopant content can be observed for all co-doped systems. Similar observations were also found by Wang et al. on Ydoped Co349 [15]. A possible reason was suggested that the trivalent Y3+ substitutions for the divalent Ca2+ ions may reduce the Co4+ (hole) concentration in order to balance the net valence of the system. Therefore ρ values of Y-doped Co349 are usually higher than those in the un-doped Co349 [14,15,17]. In the Fe/Y co-doped system, the sample with x = 0.1 and y = 0.03 showed a lower ρ value than the un-doped Co349 over the whole measured temperature range. With the same Y dopant content of x = 0.1 (red ● in Fig. 3(b – d)), ρ tended to increase with increasing the Fe dopant content. It should be noted that the ρ of the sample with x = 0.1 and y = 0.03 (red ● in Fig. 3(b)) was similar to the Fe-doped samples (Fig. 3(a)); the structure change in Co349 subsystems might be responsible for it but more studies are needed. The temperature dependence of S for the un-doped Co349 is shown in Fig 4(a), and the similar curves were observed in the Fe-doped and Fe/Y co-doped samples (for simplicity, only the un-doped Co349 presented as a reference). Positive S values in the whole measured temperature range confirm the dominant hole carriers. Fig. 4(b – e) shows the Seebeck coefficient (S) at RT, 400 and 750 °C as a function of dopant contents for the Fe-doped Ca3Co4-yFeyO9+δ and co-doped systems. For Fe-doped system (Fig. 4(b)), although the Fe dopant content was varied, similar S values as the un-doped Co349 can be observed over the whole temperature range. Wang et al. [18] and Liu et al. [20] have demonstrated that S of the Fe-doped system below RT increases with decreasing ρ. For materials with metallic conduction, S can be expressed by Mott formula and simple Drude picture as S ~ Ce / n, which is a temperature-independent value proportional to the electron specific heat (Ce) and inversely to the carrier concentration (n). The increase in S upon Fe doping was believed mainly due to the strong electronic correlation, namely enlarged electron specific heat overwhelming the increased carrier concentration [18,30]. At high temperature region, according to the modified Heikes formula, S of Co349 can be described as

5

=−

ln

(2)

where x denotes the Co4+ concentration and the values of g3 and g4 are numbers of configurations including spin and orbit degrees of freedom for Co3+ and Co4+, respectively [23]. At high temperature the electronic correlation can be neglected, namely

≫

where t is the transfer

integral of an electron between neighboring sites and kB is Boltzmann constant, S is temperatureindependent and is only determined by the ratio g3 / g4 and relative Co4+ concentration [23,31]. Therefore, the observed insensitivity of S to Fe-substitution in Fig. 4(b) might stem from (1) no significant Co4+ concentration change which is resulted from the competition between Fe-doping induced electron doping and the constituent oxygen variation (S and ρ measurements were performed under a low-pressure helium atmosphere simultaneously), and (2) the fact of the maintained Co spin-state (i.e., no change in the ratio g3 / g4) upon Fe doping [11,19]. As for the co-doped system (Fig. 4(c – e)), a clear tendency of the increase in S with Y dopant content was observed at any typical temperatures of RT, 400 oC, and 750 oC. The increase in S coupled with the increase in ρ could be mainly due to the decrease in hole concentration caused by the substitution of Y3+ for Ca2+. A similar behavior was also observed in the case of the Y-doped Co349 [14,15,17]. This resulting trend further demonstrates that the doping merits may be inherited to the co-doped system. Fig. 5 presents the power factor (PF) as a function of temperature for the typical Fedoped and Fe/Y co-doped systems (for simplicity, only the samples with higher PF from each of Fig. 3 groups are presented). PF of the un-doped Co349 sample was also plotted as a reference. For the Fe-doped samples, as a result of the reduction in ρ combined with an unchanged S, PF was enhanced throughout the investigated temperatures. Additionally, since Y doping resulted in an increase in S, accompanying with the reduced ρ by Fe doping, the PF could be further enhanced by a proper co-doping. Accordingly, the PF was found to be enhanced in the co-doped samples. The co-doped sample with the lowest ρ, Ca2.9Y0.1Co3.97Fe0.03O9+δ, exhibited the highest PF values over the whole measured temperatures, reaching 510 µW/m·K2 at 800 °C. Wang et al. have concluded that in the doping range x ≤ 0.25, the Y dopant content of x = 0.1 greatly deteriorated the PF due to the overwhelming increase in ρ [14]. Therefore, the reduction of ρ is believed as the contribution from Fe doping and which is crucial to the PF enhancement. Fig. 6 shows the total thermal conductivity (κ) for the selected Ca3-xYxCo4-yFeyO9+δ samples (with the measurement direction as indicated in the figure, and only the samples with lower κ from each of Fig. 3 groups are presented). Since the total thermal conductivity (κ) contains both the electronic conductivity (κe) and the lattice conductivity (κL), namely κ = κe + κL. Applying the Wiedemann-Franz relationship (

=

∙ ⁄ , where the Lorenz number is taken to be L0 =

2.45×10-8 V2/K2), the highest κe was found to be around 0.14 and 0.45 W/m·K at RT and 750 oC respectively; the lattice contribution accounts for more than 80% of the total thermal conductivity as also observed in other studies [1,5,15]. All the samples for x, y £ 0.10 exhibited similar κ values to the un-doped Co349 sample over the whole temperature range. Since the minor mass difference between Fe and Co atoms, Fe substitution was not expected to be an effective phonon scattering

6

site. The decreasing κ for the samples with higher Y dopant content (e.g., x = 0.2 or 0.3) could be attributed to the heavy Y3+substitution for Ca2+ leading to a reduction in κL, as well as the observed evident microstructure changes in Fig. 2. Finally, the ZT was calculated using the in-plane (measured along the direction perpendicular to SPS pressure axis) Seebeck coefficient (S), electrical resistivity (ρ) and thermal conductivity (κ) properties in the materials, and the co-doped Ca2.9Y0.1Co3.97Fe0.03O9+δ was found with larger ZT values among all studied samples. Fig. 7 presents the PF, κ and ZT as a function of temperature for the un-doped and co-doped Ca2.9Y0.1Co3.97Fe0.03O9+δ samples, showing an enhancement in PF and ZT over the whole studied temperatures. At 800 oC, the ZT value of the Ca2.9Y0.1Co3.97Fe0.03O9+δ sample was found to be 0.22 and which is about 18% higher than that of un-doped Co349.

4. Conclusion The Fe-doped and Fe/Y co-doped Ca3-xYxCo4-yFeyO9+δ (0 £ x £ 0.3, 0 £ y £ 0.1) systems were systematically investigated in terms of Fe and Y doping at the Co- and Ca-sites of Co349, respectively. The Fe substitution at the Co-sites effectively reduces the electrical resistivity (ρ) in the high temperature region, while the Seebeck coefficient (S) is influenced only slightly. Y substitution for Ca2+ leads to an increase in the Seebeck coefficient but also in the electrical resistivity. With proper additional Fe doping, the rising ρ was compensated, together with the improved S leading to an improvement of the PF. Among all investigated systems, the co-doping effects of Fe and Y were most effective for the sample with x = 0.1 and y = 0.03 (i.e., Ca2.9Y0.1Co3.97Fe0.03O9+δ). The maximum power factor reaches a value of 510 µW/m·K2, which is among the highest value reported so far on the Co349 system at 800 °C. Although the ZT enhancement is only minor due to no significant reduction in the overall κ, the more significant increase in the PF owing to the reduction in ρ along with an increase in S renders Fe/Y co-doped Co349 a promising candidate for applications where the power output density is more important than the overall conversion efficiency.

5. Acknowledgement The authors would like to thank Tim Holgate for proof reading the article, and the Programme Commission on Energy and Environment (EnMi) which is part of the Danish Council for Strategic Research (Contract No. 10-093971) for sponsoring the OTE-POWER research work.

7

6. Reference [1]

M. Shikano, R. Funahashi, Electrical and thermal properties of single-crystalline (Ca2CoO3)0.7CoO2 with a Ca3Co4O9 structure, Appl. Phys. Lett. 82 (2003) 1851. doi:10.1063/1.1562337.

[2]

N. Van Nong, N. Pryds, S. Linderoth, M. Ohtaki, Enhancement of the thermoelectric performance of p-type layered oxide Ca3Co4O9+δ through heavy doping and metallic nanoinclusions, Adv. Mater. 23 (2011) 2484–90. doi:10.1002/adma.201004782.

[3]

A. Masset, C. Michel, A. Maignan, M. Hervieu, O. Toulemonde, F. Studer, et al., Misfitlayered cobaltite with an anisotropic giant magnetoresistance: Ca3Co4O9, Phys. Rev. B. 62 (2000) 166–175. doi:10.1103/PhysRevB.62.166.

[4]

Y. Miyazaki, M. Onoda, T. Oku, M. Kikuchi, Y. Ishii, Y. Ono, et al., Modulated Structure of the Thermoelectric Compound [Ca2CoO3]0.62CoO2, J. Phys. Soc. Japan. 71 (2002) 491– 497. doi:10.1143/JPSJ.71.491.

[5]

G. Yang, Q. Ramasse, R. Klie, Direct measurement of charge transfer in thermoelectric Ca3Co4O9, Phys. Rev. B. 78 (2008) 153109. doi:10.1103/PhysRevB.78.153109.

[6]

Y. Miyazaki, K. Kudo, M. Akoshima, Y. Ono, Y. Koike, T. Kajitani, Low-Temperature Thermoelectric Properties of the Composite Crystal [Ca2CoO3.34]0.614[CoO2], Jpn. J. Appl. Phys. 39 (2000) L531–L533. doi:10.1143/JJAP.39.L531.

[7]

N. Wu, T.C. Holgate, N. Van Nong, N. Pryds, S. Linderoth, High temperature thermoelectric properties of Ca3Co4O9+δ by auto-combustion synthesis and spark plasma sintering, J. Eur. Ceram. Soc. 34 (2014) 925–931. doi:10.1016/j.jeurceramsoc.2013.10.022.

[8]

A. Bhaskar, Z.-R. Lin, C. Liu, Thermoelectric properties of Ca2.95Bi0.05Co4−xFexO9+δ (0≤x≤0.15), Energy Convers. Manag. 76 (2013) 63–67. doi:10.1016/j.enconman.2013.07.009.

[9]

G. Constantinescu, S. Rasekh, M.A. Torres, J.C. Diez, M.A. Madre, A. Sotelo, Effect of Sr substitution for Ca on the Ca3Co4O9 thermoelectric properties, J. Alloys Compd. 577 (2013) 511–515. doi:10.1016/j.jallcom.2013.07.005.

[10]

R. Tian, R. Donelson, C.D. Ling, P.E.R. Blanchard, T. Zhang, D. Chu, et al., Ga Substitution and Oxygen Diffusion Kinetics in Ca3Co4O9+δ-Based Thermoelectric Oxides, J. Phys. Chem. C. 117 (2013) 13382–13387. doi:10.1021/jp403592s.

[11]

T. Wu, T. a. Tyson, J. Bai, K. Pandya, C. Jaye, D. Fischer, On the origin of enhanced thermoelectricity in Fe doped Ca3Co4O9, J. Mater. Chem. C. 1 (2013) 4114. doi:10.1039/c3tc30481g.

[12]

Y. Wang, Y. Sui, F. Li, L. Xu, X. Wang, W. Su, et al., Thermoelectrics in misfit-layered oxides [(Ca,Ln)2CoO3]0.62[CoO2]: From bulk to nano, Nano Energy. 1 (2012) 456–465. doi:10.1016/j.nanoen.2012.02.007.

[13]

S. Pinitsoontorn, N. Lerssongkram, N. Keawprak, V. Amornkitbamrung, Thermoelectric properties of transition metals-doped Ca3Co3.8M0.2O9+δ (M = Co, Cr, Fe, Ni, Cu and Zn), J. Mater. Sci. Mater. Electron. 23 (2011) 1050–1056. doi:10.1007/s10854-011-0546-z.

[14]

Y. Wang, Y. Sui, J. Cheng, X. Wang, W. Su, H. Fan, Influence of Y3+ doping on the high-temperature transport mechanism and thermoelectric response of misfit-layered Ca3Co4O9, Appl. Phys. A. 99 (2010) 451–458. doi:10.1007/s00339-010-5543-1.

8

[15]

Y. Wang, Y. Sui, J. Cheng, X. Wang, W. Su, X. Liu, et al., Doping-Induced Metal−Insulator Transition and the Thermal Transport Properties in Ca3−xYxCo4O9, J. Phys. Chem. C. 114 (2010) 5174–5181. doi:10.1021/jp911078h.

[16]

S.W. Nam, J.W. Choi, H.K. Hwang, K. Park, Improvement in high-temperature thermoelectric properties by adding Mn for Co in Ca3Co4O9, J. Nanosci. Nanotechnol. 10 (2010) 7689–7693. doi:10.1166/jnn.2010.2784.

[17]

H.Q. Liu, Y. Song, S.N. Zhang, X.B. Zhao, F.P. Wang, Thermoelectric properties of Ca3−xYxCo4O9+δ ceramics, J. Phys. Chem. Solids. 70 (2009) 600–603. doi:10.1016/j.jpcs.2009.01.003.

[18]

Y. Wang, Y. Sui, P. Ren, L. Wang, X. Wang, W. Su, et al., Strongly Correlated Properties and Enhanced Thermoelectric Response in Ca3Co4−xMxO9 (M = Fe, Mn, and Cu), Chem. Mater. 22 (2010) 1155–1163. doi:10.1021/cm902483a.

[19]

C.-J. Liu, J. Chen, L.-C. Huang, Z.-R. Lin, C.-L. Chang, X-ray absorption spectroscopy studies of Fe-doped misfit-layered Ca3Co4−xFexO9+δ (x=0, 0.05, 0.1, and 0.15), J. Appl. Phys. 102 (2007) 014908. doi:10.1063/1.2748722.

[20]

C.-J. Liu, L.-C. Huang, J.-S. Wang, Improvement of the thermoelectric characteristics of Fe-doped misfit-layered Ca3Co4−xFexO9+δ (x=0, 0.05, 0.1, and 0.2), Appl. Phys. Lett. 89 (2006) 204102. doi:10.1063/1.2390666.

[21]

Q. Yao, D.L. Wang, L.D. Chen, X. Shi, M. Zhou, Effects of partial substitution of transition metals for cobalt on the high-temperature thermoelectric properties of Ca3Co4O9+δ, J. Appl. Phys. 97 (2005) 103905. doi:10.1063/1.1898443.

[22]

R.F. Klie, Q. Qiao, T. Paulauskas, A. Gulec, A. Rebola, S. Öğüt, et al., Observations of Co4+ in a Higher Spin State and the Increase in the Seebeck Coefficient of Thermoelectric Ca3Co4O9, Phys. Rev. Lett. 108 (2012) 196601. doi:10.1103/PhysRevLett.108.196601.

[23]

W. Koshibae, K. Tsutsui, S. Maekawa, Thermopower in cobalt oxides, Phys. Rev. B. 62 (2000) 6869–6872. doi:10.1103/PhysRevB.62.6869.

[24]

N. Wu, T.C. Holgate, N. Nong, N. Pryds, S. Linderoth, Effects of Synthesis and Spark Plasma Sintering Conditions on the Thermoelectric Properties of Ca3Co4O9+δ, J. Electron. Mater. 42 (2013) 2134–2142. doi:10.1007/s11664-013-2546-7.

[25]

Y. Liu, Y. Lin, Z. Shi, C.-W. Nan, Z. Shen, Preparation of Ca3Co4O9 and Improvement of its Thermoelectric Properties by Spark Plasma Sintering, J. Am. Ceram. Soc. 88 (2005) 1337–1340. doi:10.1111/j.1551-2916.2005.00284.x.

[26]

D. Kenfaui, G. Bonnefont, D. Chateigner, G. Fantozzi, M. Gomina, J.G. Noudem, Ca3Co4O9 ceramics consolidated by SPS process: Optimisation of mechanical and thermoelectric properties, Mater. Res. Bull. 45 (2010) 1240–1249. doi:10.1016/j.materresbull.2010.05.006.

[27]

R.D. Shannon, Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides, Acta Crystallogr. Sect. A. 32 (1976) 751–767. doi:10.1107/S0567739476001551.

[28]

Y. Morita, J. Poulsen, K. Sakai, T. Motohashi, T. Fujii, I. Terasaki, et al., Oxygen nonstoichiometry and cobalt valence in misfit-layered cobalt oxides, J. Solid State Chem. 177 (2004) 3149–3155. doi:10.1016/j.jssc.2004.05.023.

[29]

M. Schrade, H. Fjeld, T.G. Finstad, T. Norby, Electronic Transport Properties of [Ca2CoO3−δ]q[CoO2], J. Phys. Chem. C. 118 (2014) 2908–2918. doi:10.1021/jp409581n.

9

[30]

N.W. Ashcroft, N.D. Mermin, Solid state physics, Holt, Rinehart and Winston, New York, 1976.

[31]

P. Chaikin, G. Beni, Thermopower in the correlated hopping regime, Phys. Rev. B. 13 (1976) 647–651. doi:10.1103/PhysRevB.13.647.

10

Figure Captions

Fig. 1. XRD patterns of the as-synthesized powders and SPS-sintered pellets of the un-doped Co349, Ca3Co3.9Fe0.1O9+δ (a) and Ca2.7Y0.3Co3.9Fe0.1O9+δ (b). The peaks with indices at the bottom are for the Ca3Co4O9+δ phase identified by JCPDS PDF # 21-0139.

Fig. 2. SEM images of the fracture surface of a typical SPS-sintered un-doped Co349 (left) and codoped Ca2.7Y0.3Co3.9Fe0.1O9+δ (right) samples observed along the direction perpendicular to the SPS pressure axis.

Fig. 3. The temperature dependent electrical resistivity (ρ) of the un-doped and Ca3-xYxCo4yFeyO9+δ

series. The samples with varied dopant content are sorted into four groups: (a) Fe-doped

Ca3Co4-yFeyO9+δ samples with y = 0.03, 0.05, and 0.10. Samples of the fixed Fe content (b) y = 0.03, (c) y = 0.05 and (d) y = 0.10 with x varying from 0.1, 0.2 and 0.3. Open symbol denotes ρ of the un-doped Co349 as a reference.

Fig. 4. The temperature dependent Seebeck coefficient (S) of the un-doped Co349 (a); the dependence on doping of the Seebeck coefficient of the un-doped and Ca3-xYxCo4-yFeyO9+δ series (b to e). The samples with varied dopant content are sorted into four groups: (b) Fe-doped Ca3Co4yFeyO9+δ

samples with y = 0.03, 0.05, and 0.10. Samples of the fixed Fe content (c) y = 0.03, (d) y

= 0.05 and (e) y = 0.10 with x varying from 0.1, 0.2 and 0.3. Open symbol denotes S of the undoped Co349 as a reference.

Fig. 5. The temperature dependent power factor (PF) of typical samples that are representatives of the Fe-doped (x = 0, y = 0.1) and co-doped systems (x = 0.1, y = 0.03; x = 0.2, y = 0.05; x = 0.2, y = 0.1). Open symbols denote the PF of the un-doped Co349 as a reference.

Fig. 6. The temperature dependent thermal conductivity (κ) of Ca3-xYxCo4-yFeyO9+δ samples which exhibited lower values in the four respective groups described in Fig. 3 and 4: the Fe-doped Co349 and co-doped samples with y = 0.03 to 0.10. Open symbols denote κ values of the un-doped Co349 as a reference. The measurement was performed along the direction perpendicular to the SPS pressure axis for in-plane κ.

1

Fig. 7. The temperature dependent PF, κ and ZT of the un-doped Co349 and Ca2.9Y0.1Co3.97Fe0.03O9+δ.

2

(a)

Ca3Co4-yFeyO9+δ

Intensity (a.u.)

■ Ca3Co4O9+δ

y = 0.10 SPS-sintered sample y=0 SPS-sintered sample

As-synthesized y = 0.10 powder

7

10

20

30

40

2 θ (degree)

50

(220)

(006)

(203)

(202)

(005)

(020)

(20-1)

(004)

(11-2)

(111)

(003)

(002)

(001)

As-synthesized y = 0 powder

60

(b)

Ca3-xYxCo4-yFeyO9+δ

Intensity (a.u.)

■ Ca3Co4O9+δ

x = 0.3, y = 0.10 SPS-sintered sample x, y = 0 SPS-sintered sample As-synthesized x = 0.3, y = 0.10 powder

7

10

20

30

40

2 θ (degree)

50

(220)

(006)

(203)

(202)

(005)

(020)

(20-1)

(004)

(11-2)

(111)

(003)

(002)

(001)

As-synthesized x, y = 0 powder

60

x, y = 0

x = 0.3, y = 0.1

400 nm

400 nm

SPS Pressure Axis

SPS Pressure Axis

SPS Pressure Measure

Highlights · The Fe and Fe/Y doping at the Co- and Ca-sites of Ca3 Co4O9+δ were

investigated.

· The rising ρ by Y doping can be mitigated by the coupled Fe doping.

· The increased Seebeck coefficient by Y doping can be maintained in co-doped

system.

· The co-doped system leads to an improvement of the thermoelectric

performance.

· The co-doped system may preserve the merits from each component doping.

1