Thermoelectric properties of lower concentration K-doped Ca3Co4O9 ceramics
Li Ya-Nan1, Wu Ping1, †, Zhang Shi-Ping1, Chen Sen1, Yan Dan1, Yang Jin-Guang1, Wang Li2, Huai Xiu-Lan3
Beijing Key Laboratory for Magneto-Photoelectrical Composite and Interface Science, School of Mathematics and Physics, University of Science and Technology Beijing, Beijing 100083, China
School of Energy and Environmental Engineering, University of Science and Technology Beijing, Beijing 100083, China
Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing 100190, China

 

† Corresponding author. E-mail: pingwu@sas.ustb.edu.cn

Abstract

The tuning of electron and phonon by ion doping is an effective method of improving the performances of thermoelectric materials. A series of lower concentration K-doped Ca3−xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) polycrystalline ceramic samples are prepared by combining citrate acid sol-gel method with cold-pressing sintering method. The single-phase compositions and plate-like grain morphologies of all samples are confirmed by x-ray diffraction and field emission scanning electron microscope. The effects of lower concentration K doping on the thermoelectric properties of the material are evaluated systematically at high temperatures (300–1026 K). Low concentration K doping causes electrical conductivity to increase up to 23% with little effect on the Seebeck coefficient. Simultaneously, the thermal conductivity of K-doped sample is lower than that of the undoped sample, and the total thermal conductivity reaches a minimum value of approximately , which may be suppressed mainly by the phonon thermal conduction confinement. The dimensionless figure-of-merit ZT of Ca2.95K0.05Co4O9 is close to 0.22 at 1026 K, representing an improvement of about 36% compared with that of Ca3Co4O9, suggesting that lower concentration K-doped Ca3Co4O9 series materials are promising thermoelectric oxides for high-temperature applications.

1. Introduction

Under the serious situations of resource shortage and environmental pollution, thermoelectric (TE) materials, which can directly convert heat energy to electrical power and vice versa, play a significant role in meeting energy challenges in the future.[1,2] The energy conversion efficiency of a TE material is found to depend on the dimensionless figure-of-merit, ZT which is defined as , where σ, S, T, and κ are the electrical conductivity, Seebeck coefficient, absolute temperature, and thermal conductivity, respectively. As can be observed from the formula, a combination of high electrical conductivity and Seebeck coefficient is desired, whereas thermal conductivity must be low for applicable TE materials.[3,4] Over the past few decades, intermetallic compounds, such as Bi2Te3 and SnSe, have been widely investigated because of their superior thermal performances,[5,6] but oxide TE materials, such as NaCo2O4, Ca3Co4O9, and CaMnO3, have attracted the attention of scientific workers in recent years due to their high thermal and chemical stabilities, lower cost, and non-rare or non-toxic elements.[79] Among these transition metal oxides, Ca3Co4O9 is considered to be a promising oxide TE material on account of its high electrical conductivity, low thermal conductivity, and high thermostability at approximately 1000 K in the air.[10] The Ca3Co4O9, with misfit-layered structure, is composed of NaCl-type Ca2CoO3 and CdI2-type CoO2 layers.[11] The Ca2CoO3 layer is commonly considered to be responsible for charge reservoir and structure stabilizer. Meanwhile, the charge carriers mainly transport in the CoO2 layer. These two sublattices possess the same a- and c-axis lattice parameters and β angles but different b-axis lengths, leading to a misfit along the b-direction. These strong electron correlated systems with misfit layer structures are subject to a crucial condition for the highly thermoelectric performance of Ca3Co4O9.[11,12] The ZT of single-crystalline Ca3Co4O9 can reach 0.87 at around 973 K.[13] Unfortunately, the ZT of polycrystalline Ca3Co4O9 is relatively low for practical use; thus, numerous studies have shown that the carrier conduction in polycrystalline Ca3Co4O9 TE material can be controlled by doping other elements into the system at the Ca[1416] or Co[1719] sites to improve the TE properties. However, these studies mainly use rare-earth metals, alkaline-earth metals, or toxic elements of doping to improve the performance of Ca3Co4O9 at present.[15,20,21] The K element as an environmentally friendly element was generally selected as p-type dopant for tuning the carrier concentration and microstructure, indicating its potential for achieving a high ZT.[22,23] Based on the theoretical analysis, the substitution of monovalent K+ for divalent Ca2+ of the Ca3Co4O9 may increase the carrier concentration. The K and Ca elements have similar atomic mass values, but the radius of the K ion (1.33 Å) is larger than that of the Ca ion (1.00 Å), which may cause lattice to deform, phonon scattering to be enhanced, and the material TE properties to improve. To our knowledge, Lim et al.[24] have studied the effects of higher concentration K-doping on Ca3−xKxCo4O9 (x = 0, 0.1, 0.2, 0.3), in which the conductivity of K-doped sample was found to be higher than that of undoped sample, but the most of the thermal conductivities were higher than that of the undoped sample. However, it can be seen from the results of the study that when the doping concentration was low (x = 0.1), the thermal conductivity was lower than that of the undoped sample in a temperature range of about 700–1100 K. It can be hypothesized that the lower concentration doping will be more conducive to reduce the thermal conductivity. Lan et al.[25] studied the TE properties of Bi1−xKxCuSeO ( ). When the K content was x = 0.02, the ZT was increased by 50% at 823 K compared with the undoped sample. When the doping concentration is larger, the increase is smaller or even lower than that of the original sample. Similarly, Zhang et al.[26] found that the thermal conductivities of the lower concentration of Ag-doped Ca3−xAgxCo4O9 + δ (x = 0, 0.01, 0.03, 0.05) were lower than that of the undoped sample, and when the doping content x = 0.01, the conductivity was the highest, and the conductivity decreases with the increase in the doping content. Therefore, it is necessary to investigate the effect of lower concentration K doping on the TE performance of Ca3Co4O9.

In this work, polycrystalline Ca3−xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) ceramic samples (where the x value is the apparent concentration of K element) are prepared by the combined methods of sol-gel and cold-pressing. The TE properties of the lower concentration K doping polycrystalline Ca3 −xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) ceramic materials are investigated systematically, and a method of improving the TE properties of cobalt oxides is explored.

2. Experiment

The Ca3 −xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) precursor powders were synthesized by citrate acid sol-gel method. Stoichiometric ratios of , and were dissolved thoroughly in deionized water, and a certain amount of citric acid was added. The solution was heated under continuous stirring at 353 K in the air until the gel was formed. The gel was dried at 393 K in the air for 8 h. Then, the dry gel was ground, combusted at 572 K in the air, and calcined at 1073 K for 8 h in the air to produce the Ca3 −xKxCo4O9 powder. Finally, the powder was reground and pressed at 30 MPa to form pellets, which were sintered at 1073 K for 8 h in the air. The phase compositions of all bulk specimens were analyzed using x-ray diffraction (XRD) with a RigakuD/max-2500 diffractometer equipped with Cu Kα radiation at room temperature in the air. The diffraction intensities were collected in the 2θ range from 5° to 65° in steps of 0.01°. The microstructure was observed with the field emission scanning electron microscope (FESEM, Zeiss Supra 55). Electrical conductivity and Seebeck coefficient were simultaneously conducted in Ar atmosphere from room temperature up to 1026 K on Seebeck Coefficient/Electric Conductivity Measuring System (Netzsch SBA458). The thermal conductivity was calculated by , and the density d was determined by the geometric density. The specific heat capacity Cp was determined by differential scanning calorimeter (Netzsch DSC/STA409PC), and the thermal diffusivity coefficient D was measured by the laser-flash method (Netzsch Instruments/LAF457).

3. Results and discussion
3.1. Structure and composition analysis

Figure 1(a) shows the XRD patterns of the sintered Ca3 −xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) bulk materials in the direction perpendicular to pressure direction, and the diffraction peaks are marked. All the diffraction peaks can be indexed as Ca3Co4O9 structure according to the JCPDS card with PDF No.21-0110, and no secondary phase was found, indicating that the prepared bulk materials are Ca3Co4O9 compounds, and the K has successfully entered into the lattice site.[18,27] As shown in Fig. 1(a), the diffraction peaks of (00l) planes are much stronger than other peaks. The orientation factor L is calculated by using XRD pattern to further analyze the grain orientation of the bulk sample. L can be expressed as , where , with I being the intensity of the diffraction peak, and P values are calculated from the XRD patterns of doped Ca3−xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) samples; P0 is calculated from randomly grain oriented sample by using standard PDF card JCPDS (No.21-0110). The value of L is equal to 1 when the pressure direction is oriented along the crystal c-axis completely. Figure 1(b) shows the relationship between the L and the doping content of K. The L of all the samples were around 0.85, indicating that these ceramic samples in the c-direction have an obviously preferred orientation by the cold-pressure method, that is, the grains preferred to grow in the direction perpendicular to the pressure. The density of Ca3−xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) samples is approximately 81% of the theoretical value (4.94 g/cm3)11 as shown in Fig. 1(c). Studies have shown that spark plasma sintering (SPS) technique[28,29] and hot press (HP) technique[28] each can make the sample have a higher density. Figure 2 displays the SEM images of Ca3−xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) bulk samples. Most of the particles with a grain size of several microns are flake-like. Particle size distributions (Fig. 3) of the samples were determined by the software of Nano Measurer program[30] based on the corresponding SEM images. The analyses of a number of the particle show that these samples have average sizes of , , , and , respectively. With increasing doping content, the sizes of the particles increase.

Fig. 1. (color online) (a) XRD patterns of sintered Ca3–xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) bulk samples in the direction perpendicular to pressure direction, and (b) grain orientation degree (L) and (c) density versus x value of the Ca3–xKxCo4O9 samples.
Fig. 2. SEM images of sintered sample surfaces (a) Ca3Co4O9, (b) Ca2.95K0.05Co4O9, (c) Ca2.90K0.10Co4O9, and (d) Ca2.85K0.15Co4O9.
Fig. 3. Particle size distributions of (a) Ca3Co4O9, (b) Ca2.95K0.05Co4O9, (c) Ca2.90K0.10Co4O9, and (d) Ca2.85K0.15Co4O9, obtained from the SEM images of 50 particles.
3.2. Electrical transport properties

The relationships between the electrical conductivity and the temperature of the Ca3 −xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) samples are shown in Fig. 4(a). According to the electrical conductivity formula , where n is the carrier concentration, e is the electron charge of the carrier, and μ is the carrier mobility, the electrical conductivity mainly depends on the carrier concentration and mobility of the sample. Figure 4(a) shows that all electrical conductivity values increase with increasing temperature, which exhibits remarkable semiconductor-like behavior in a temperature range from 400 K to 1026 K. At the same time, it can be seen that when the doping concentration is , the conductivity is higher than that of the undoped sample, but with the doping content increasing (x = 0.10, 0.15), the conductivity decreases and becomes lower than that of the undoped sample. In general, when K+ is substituted for Ca2+, the valence of K is lower than that of Ca, thereby increasing the carrier concentration in the sample. Simultaneously, K doping can form impurity energy level in the semiconductor band gap, reducing the band gap width. This result is conducive for impurity and intrinsic carriers transiting from the valence band to the conduction band to further increase the carrier concentration at room temperature, which leads to the higher electrical conductivity of the doped sample. These can explain why the electrical conductivity of x = 0.05 sample is higher than those of the undoped samples. However, the electrical conductivity decreases with the further increase of K doping content. It indicates that the carrier concentration increases and the carrier mobility decreases with the increase of K doping content. Moreover, the effect of carrier concentration on the electrical conductivity is weak compared with that of carrier mobility. Similarly, Pei et al.[21] found that the electrical conductivity decreases with the increase of Er-doping for Ca3−xErxCo4O9 + δ (x = 0.15, 0.30 and 0.50).

Fig. 4. (color online) Thermoelectric properties of Ca3–xKxCo4O9 samples. (a) Temperature-dependent electrical conductivities for Ca3–xKxCo4O9 samples. (b) Relationships between ln(σT) and 1000/T for Ca3–xKxCo4O9 samples. Inset: K content dependent activation energy (Ea) values. (c) Temperaturedependent Seebeck coefficients. (d) Temperature-dependent power factors. (e) Temperature-dependent total thermal conductivity k values. (f) Temperaturedependent carrier thermal conductivity kc values and the phonon thermal conductivity kp values for Ca3–xKxCo4O9 samples.

The experimental results can be further explained by the model of electrical conductivity versus temperature. Hopping conduction behavior exists in Ca3Co4O9 at high temperature. The electrical conductivity is generally described as , where A, Ea, kB, and a are the pre-exponential terms relating to the scattering mechanism, activation energy, Boltzmann constant, and intersite distance of hopping, respectively.[31] Figure 4(b) shows the linear relationship between ln(σT) and 1000/T in the temperature range above 600 K, indicating that the sample satisfies the conduction mechanism of the hopping conduction behavior. According to the above equation, the value of Ea can be obtained by linearly fitting the slope. The results are provided in the illustration (R2 reaches 0.99). The value of Ea hardly changes with the increase of doping concentration. If the K is substituted for the Co ions in the CoO2 layer, the defect would be brought into the conduction path, and thus may have a significant effect on the activation energy. However, if the substitution is in the Ca2CoO3 layer, the conduction path will not be disturbed and will have little effect on the activation energy. Thus, K ions should be substituted for the Ca sites in the crystal lattice.[18,32,33] Although further research is generally necessary, lower concentration K doping favors the increase of the electrical conductivity of Ca3Co4O9 ceramic oxide. The electrical conductivity of Ca2.95K0.05Co4O9 sample is approximately 23% higher than that of the undoped Ca3Co4O9 sample, and lower concentration K doping has a significant effect on improving the Ca3Co4O9 electric properties.

The temperature-dependent Seebeck coefficient S values of all the samples are shown in Fig. 4(c). The sign of S is positive for the entire measured temperature range, confirming a dominant p-type conduction mechanism. The Seebeck coefficients of all the samples increase with increasing temperature. This result may be related to the phonon traction effect, that is, the phonon flows from the high-temperature end of the semiconductor to the low-temperature side, transfers energy to the carrier by collision, and then forms the seam stream with the phonons, and thus increasing the Seebeck coefficient. Notably, the increase in electrical conductivity is generally accompanied by a decrease in the Seebeck coefficient; however, the Seebeck coefficients of the K-doped and the undoped samples are nearly equal. The Ca3Co4O9 is a strong electron correlation system[11] whose Seebeck coefficient can be analyzed according to Heikes expansion formula.[34] The S of the cobalt oxide system can be expressed as

where d3 and d4 are the numbers of configurations of the Co3+ and Co4+ ions, respectively, and x is the concentration of Co4+ ions in the CoO2 layer, with the low spin states of Co3+ and Co4+ yielding d3 = 1 and d4 = 6. According to the above equation, the value of S increases with decreasing x value, that is, it is assumed that the K ion fraction replaces Co4+ in CoO2, which can reduce the concentration of Co4+ ions. Simultaneously, the transition of Co4+ state to the low spin state Co3+ also reduces x, thereby increasing S. However, S does not change with K doping in the present study, and thus K ions do not enter into the CoO2 conductive layer,[35] which is consistent with previous Ea analysis. Furthermore, the test results can also be analyzed based on Mott formula[36]
where ce, kB, and μ (ε) are the electronic specific heat, Boltzmann constant, and energy correlated carrier mobility, respectively. From the above equation, S mainly depends on the n and μ(ε) at a certain temperature. The number of K+ ions increases the carrier concentration n in the alternative Ca2+ based on previous analysis, and the first term of Eq. (2) shows that S decreases as the carrier concentration n increases. However, the results of the present study show that different concentrations of K doping have little effect on the S value of the sample. When K+ replaces the Ca2+ position in the lattice, the energy band structure and the Fermi level EF of the system will change, and the μ(ε) of the second term of the formula may also change. Although dopant increases the carrier concentration, S also increases,[37,38] indicating that μ (ε) may have a critical effect on S in this layered cobalt oxide. This result indicates that when the carrier concentration increases, S is almost invariant.

The values of power factor of the Ca3−xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) series of samples are shown in Fig. 4(d) to estimate the TE properties of the material. As can be observed from the figure, PF increases with increasing temperature, and K doping can regulate the power factor of the sample. For the Ca2.95K0.05Co4O9 sample, the PF reaches at 1026 K, which is approximately 22% higher than that of the undoped Ca3Co4O9 sample. The high PF indicates a relatively high TE conversion power.[39] This result indicates that K doping is an effective method of improving the electrical properties of Ca3Co4O9 materials.

3.3. Thermal transport properties

Figure 4(e) shows the relationships between the total thermal conductivity (k) of the sample in the direction perpendicular to the pressure direction and temperature. For all samples, k decreases with increasing temperature. The value of k for the sample doped with K is less than that of the undoped sample. For the Ca3Co4O9 system, the k can be represented by the sum of carrier thermal conductivity (kc) and the phonon thermal conductivity (kp), i.e., , where the carrier thermal conductivity is defined as according to the Wiedemann–Franz law. The value of kc is small (Fig. 4(f)); it is approximately only 10% of the total thermal conductivity. Accordingly, k depends mainly on kp. In the low-order approximation, the phonon thermal conductivity can be given by using the relationship , where c, v, and lp represent the specific heat capacity of the phonon, the phonon propagation velocity, and the mean free path, respectively. Typically, such doping will increase c but reduce v and lp.[40,41] Moreover, the phonon propagation velocity and the mean free path of the phonon are positively correlated.[42] Based on the above analysis, the phonon thermal conductivity is suppressed, which can be attributed to lattice defects caused by the doping of K and thus causing lattice deformation, thereby enhancing phonon scattering and reducing the average free path of the phonon and the propagation velocity. According to Matthiessenʼs law, the phonon thermal conductivity mainly depends on the point defect scattering, grain boundary scattering, and the non-harmonic three-phonon Umklapp process (U process) scattering. As a new scattering center, point defect (such as Ca or O vacancies) can further lead to enhanced phonon scattering. Furthermore, the difference between Ca and K ions introduces the spread of stress field, which makes the point defects of the phonon increase and the phonon thermal conductivity decrease. However, there is no regular change in thermal conductivity with the increase of K doping content, which requires further study. However, the lower concentration K doping can be more effective than the higher[24] to reduce the thermal conductivity of the Ca3Co4O9.

3.4. Dimensionless thermoelectric figure of merit ZT

Figure 5 shows the relationships between the ZT of Ca3−xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) samples and the temperature. The ZT values of all samples increase with increasing temperature, which might be attributed mainly to the increase of S as the temperature increases and the thermal conductivity decreases. The values of ZT of the K-doped samples are higher than those of the undoped ZT to a large extent. The value of ZT of the Ca2.95K0.05Co4O9 sample reaches 0.22 at 1026 K, which is an increase of approximately 35% of the undoped Ca3Co4O9. The results also show that the TE properties of Ca3Co4O9 can be improved by appropriate K doping.

Fig. 5. (color online) Temperature dependent dimensionless figure-ofmerit ZT for Ca3–xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) samples.
4. Conclusions

The TE conductivities of K-doped Ca3 −xKxCo4O9 (x = 0, 0.05, 0.10, 0.15) oxide samples prepared by sol-gel method have been systematically investigated. The XRD and SEM studies show that Ca KxCo4O9 samples with x = 0, 0.05, 0.10, 0.15 are all of single-phase, and all exhibit similar flake-like morphological structures. An appropriate quantity of K ions doping at the Ca site can effectively increase the electrical conductivity and reduce thermal conductivity of the material.

In all the samples, the power factor of K-doped Ca2.95K0.05Co4O9 sample reaches at 1026 K and is approximately 22% higher than that of the undoped sample. The value of ZT of the Ca2.95K0.05Co4O9 sample reaches 0.22 at 1026 K and is approximately 35% higher than that of the undoped sample. These results indicate that lower concentration doping is can be used to improve the TE properties of Ca3Co4O9 ceramic samples.

Reference
[1] Disalvo F J 1999 Science 285 703
[2] Biswas K He J Blum I D Chun I Hogan T P Seidman D N Dravid V P Kanatzidis M G 2012 Nature 489 414
[3] Slack G A Hussain M A 1991 J. Appl. Phys. 70 2694
[4] Vining C B 1991 J. Appl. Phys. 69 331
[5] Feng S K Li S M Fu H Z 2014 Chin. Phys. 23 117202
[6] Zhao L D Lo S H Zhang Y S Sun H Tan G J Uher C 2014 Nature 508 373
[7] Terasaki I Sasago Y Uchinokura K 1997 Phys. Rev. 56 R12685
[8] Park K Lee G W 2013 Energy 60 87
[9] Kumar P Kashyap S C Sharma V K Gupta H C 2015 Chin. Phys. 24 098101
[10] Ohta H Sugiura K Koumoto K 2008 Inorg. Chem. 47 8429
[11] Masset A C Michel C Maignan A Hervieu M Toulemonde O Studer F Raveau B Hejtmanek J 2000 Phys. Rev. 62 166
[12] Limelette P Hardy V Auban-Senzier P Jérome D Flahaut D Hébert S Frésard R Simon C h Noudem J Maignan A 2005 Phys. Rev. 71 233108
[13] Shikano M Funahashi R 2003 Appl. Phys. Lett. 82 1851
[14] Zhang D W Wang Z H Tang G D Xu X B Xu X N Qiu L Zhang D X Du Y W 2013 J. Appl. Phys. 113 17E113
[15] Xu J Wei C P Jia K 2010 J. Alloy Compd. 500 227
[16] Bhaskar A Lin Z R Liu C J 2014 0. Mater Sci. 49 1359
[17] Wang Y Sui Y Wang X J Su W H Liu X Y 2010 J. Appl. Phys. 107 033708
[18] Prasoetsopha N Pinitsoontorn S Kamwanna T Amornkitbamrung V Kurosaki K Ohishi Y Muta H Yamanaka S 2014 J. Alloy Compd. 588 199
[19] Torres M A Rasekh Sh Constantinescu G Sotelo A Bosque P Madre M A Diez J C 2015 J. Mater Sci. Mater Electron 26 815
[20] Prevel M Perez O Noudem J G 2007 Solid State Sci. 9 231
[21] Pei J Chen G Zhou N Lu D Q Xiao F 2011 Physica B. 406 571
[22] Lee D S An T H Jeong M Choi H S Lim Y S Seo W S Park C H Park C Park H H 2013 Appl. Phys. Lett. 103 232110
[23] Wu H J Zhao L D Zheng F S Wu D Pei Y L Tong X 2014 Nat. Commun. 5 4515
[24] Lim C H Choi S M Seo W S 2010 J. Korean Phys. Soc. 57 1054
[25] Lan J L Ma W Q Deng C J Ren G K Lin Y H Yang X P 2007 J. Mater. Sci. 52 11569
[26] Zhang F P Zhang X Lu Q M Zhang J X Liu Y Q Zhang G Z 2011 Solid State Ion. 20 1
[27] Tian Z Y Wang X H Liu J Lin Z J Hu Y S Wu Y G Han C Hu Z Y 2016 J. Alloy Compd. 661 161
[28] Matsubara I Funahashi R Takeuchi T Sodeoka S 2001 J. Appl. Phys. 90 462
[29] Liu Y Lin Y Shi Z Nan C W Shen Z J 2005 J. Am. Ceram. Soc. 88 1337
[30] Shi Y R Wang Y H Wang D Liu B Li Y H Wei L 2012 Cryst. Growth Des. 12 1785
[31] Bosman A J Van Daal H J 1970 Adv. Phys. 9 1
[32] Zhang F P Zhang X Lu Q M Zhang J X Liu Y Q Zhang G Z 2011 Solid State Sci. 13 1443
[33] Huang C S Zhang F P Zhang X Lu Q M Zhang J X Liu Z Y 2016 J. Alloy Compd. 6 87
[34] Koshibae W Tsutsui K Maekawa S 2000 Phys. Rev. 62 6869
[35] Constantinescu G Rasekh Sh Torres M A Bosquec P Diez J C Madre M A Sotelo A 2015 Ceram. Int. 41 10897
[36] Zhao B C Sun Y P Lu W J Zhu X B Song W H 2006 Phys. Rev. B. 74 144417
[37] Bhaskar A Lin Z R Liu C J 2013 J. Mater. Sci. 49 1359
[38] Xu G J Funahashi R Shikano M Pu Q R Liu B 2002 Solid State Commun. 124 73
[39] Narducci D 2011 Appl. Phys. Lett. 99 102104
[40] Takahata K Iguchi Y Tanaka D Itoh T 2000 Phys. Rev. 61 12551
[41] Zhang F P Zhang X Lu Q M Zhang J X Liu Y Q 2011 J. Alloy Compd. 509 4171
[42] Li J C Wang C L Wang M X Peng H Zhang R Z Zhao M L Liu J Zhang J L Mei L M 2009 J. Appl. Phys. 105 043503