Tuning the optical and dielectric properties of calcium copper titanate Ca_xCu_3−xTi₄O₁₂ nanopowders

Abstract

Calcium copper titanate Ca_xCu_3−xTi₄O₁₂ (CCTO) nanopowders have been synthesized using the organic acid precursor method based on commercially available materials. The results revealed that cubic CCTO phase was accomplished for the formed citrate precursors annealed at 1000 °C for 2 h. The crystallite size of the formed powders was found to increase from 44.2 to 64.8 nm upon increasing the molar ratio of Ca²⁺ ion from 1.0 to 2.0. A slight increase in the lattice parameter “a” and unit cell volume were observed, while a slight decrease in the porosity was evidenced as a result of increasing Ca²⁺ ion concentration. FE-SEM observations of these powders confirmed their homogeneous regular cubic-like structure. It can be noted that the transmittance of the sample was around 85% with Ca²⁺ ratio 1.0. Furthermore, the band gap energy increased from 3.8 to 4.2 eV, and the DC resistivity was increased from 6.4 × 10⁴ to 6.8 × 10⁴ cm Ω with increasing calcium content. We demonstrate that without any dopant, only by controlling the chemistry and engineering of the interfacial regions at the grain boundaries, the dielectric loss was suppressed remarkably while retaining the giant dielectric constant. These investigations would allow the application of these materials in transparency, microelectronics and memory devices.

---

A. Introduction

Dielectric materials with high permittivity, low loss, light weight and good process ability are highly desirable for a broad range of applications including electromechanical actuators, gate dielectrics, and energy storage devices.^1–5 Although the dielectric behaviour mechanism is still open to debate, a commonly accepted theory based on the internal barrier layer capacitor (IBLC) model⁶ suggested that semiconducting and insulating grains or domain boundaries form parallel capacitors, leading to an exceptionally high dielectric permittivity (ε′). Therefore, fine calcium copper titanate (CCTO) nanoparticles smaller than 100 nm are eagerly desired. Indeed, the CCTO complex perovskite structure is very flexible, i.e. its dielectric constant (ε) and dielectric loss (tan δ) are highly dependent on the various cationic substitutions, such as La and Pr at Ca site or Ta, Cr, and Hf at the Ti site.^7–12 The cationic substitution at the Cu site and its concentration in CCTO ceramic affect the dielectric properties because Cu ion is one of the most effective inter-granular dopants for barrier layer capacitors, and it can act as an acceptor ion.¹³ Therefore, the Cu¹⁺/Cu²⁺ ratio and its correlation with oxygen vacancies has great influence on the dielectric properties of polycrystalline CCTO via the internal barrier layer capacitance (IBLC) mechanism.¹⁴

To date, CCTO materials have been generally synthesized via two major routes, i.e., the conventional high-temperature solid-state reaction^15,16 and the wet chemical methods.^17–21 In the solid-state reaction method, stoichiometric mixtures of CaCO₃, TiO₂ and CuO are usually heated up to a high temperature (1000–1150 °C) for long duration (4–50 h).¹⁶ The procedures of the solid-state synthesis are straightforward. However, the reaction products are usually not structurally and compositionally homogeneous. The solid-state reacted products often not only contain CCTO phase but also impurities, such as CaTiO₃ and CuO, were observed. Besides, it is difficult to obtain nanosized CCTO powders due to the high annealing temperature used. The high impurity and poor powder characteristics, represented by a coarse particle size, wide particle size distribution, irregular particle morphology, and a high degree of inhomogeneity have made this process unsuitable.²³ In this regard, Manik and Pradhan²² employed the ball milling technique for the synthesis of pure CCTO (18 nm) after milling for 8 h. Researchers and scientists have developed other ways to obtain high purity CaCu₃Ti₄O₁₂ phase with improved powder morphology, which will provide enhanced dielectric constant with low loss. As a result, wet chemistry methods are used to synthesize CCTO nanopowders including polymerized complex,^24,25 microwave heating,²⁶ sol–gel,²⁷ and co-precipitation methods.²⁸

Among these techniques, the combustion route appears to be a promising one due to the excellent chemical homogeneity, high purity and production of nano-sized powders. Moreover, this method has been adopted due to its various advantages compared with other methods such as energy efficiency, short reaction rate, simple reagents, easy operations, ease of introduction of dopants into the final product, low annealing temperature, better particle size distribution, high probability of the formation of single domain and limited agglomeration of powders.²⁹ As far as we know, the synthesis of CCTO powders by citrate precursor (CP) method has not been reported yet. Therefore, in this study, we investigated the details pertaining to the effect of Ca²⁺ ion variation on the crystal structure, microstructure, optical, electrical and dielectric properties of CCTO nanopowders fabricated using the citrate precursor route based on cost effective materials. Herein, the starting materials used were citric acid, calcium carbonate, copper chloride and titanium dioxide.

B. Materials and methods

1. Materials

Anhydrous analytical grade calcium carbonate CaCO₃ (98.5% Sigma Aldrich), copper chloride CuCl₂ (98.6%, BDH Chemical Ltd), titanium dioxide TiO₂ (99.8%, Fluka), and citric acid C₆H₈O₇ (98% ADWIC) were used as starting materials. Deionized water was used in the whole work.

2. Procedure

Calcium copper titanate (CCTO) nanopowders were prepared via the citrate precursor route by mixing aqueous solutions of calcium carbonate CaCO₃, copper chloride CuCl₂ and titanium dioxide TiO₂ with Ca : Cu : Ti ratios of x :3 − x : 4 with different molar ratios of Ca²⁺ ion (x = 1.0, 1.5, and 2.0) using certain amount of citric acid. The molar ratio of metals precursor to citric acid was 1 : 5. The solutions were slowly heated on a hot plate with magnetic stirring at 80 °C to form a viscous gel. Then, the swelled gel was dried at that temperature for 6 h, resulting in the dried gel citrate precursors. Subsequently, the precursors were annealed in pure alumina crucibles at 1000 °C for 2 h in a muffle furnace (in air) at a heating rate of 10 °C min⁻¹ to achieve the corresponding perovskite structure in the samples.

3. Physical characterization

X-ray powder diffraction (XRD) was carried out on a model Bruker AXS diffractometer (D8-ADVANCE Germany) with Cu Kα (λ = 1.54056 Å) radiation, operating at 40 kV and 40 mA. The diffraction data were recorded for 2θ values between 10° and 80°. Scanning electron microscopy was performed by an FE-SEM (JEOL-JSM-5410 Japan). The UV-vis absorption spectrum was measured by a UV-vis-NIR-scanning spectrophotometer (JASCO V-570 spectrophotometer, Japan). An electrometer and DC power supply (Agilent-4339B, USA) were used for the electrical resistivity measurement. Dielectric properties were measured using a network impedance analyzer (Agilent-E4991A, USA) in the frequency range of 1 MHz–3 GHz.

C. Results and discussion

1. Synthesis of CCTO nanopowders

XRD patterns of Ca_xCu_3−xTi₄O₁₂ (x = 1.0, 1.5 and 2.0) powders synthesized by the citrate precursors process calcined at 1000 °C for 2 h are shown in Fig. 1. The main peaks of calcined Ca_xCu_3−xTi₄O₁₂ powders are assigned to pure Ca_xCu_3−xTi₄O₁₂ (JCPDS card no. 75-2188). Peaks belonging to the (200), (211), (220), (013), (222), (321), (400), (422), (440), (433), and (620) planes of cubic CCTO phase were indexed. No extra secondary impurity phases were detected with increasing Ca²⁺ ions content.^30–32 Weak diffraction peaks observed in Fig. 1 are not from the secondary phases, but it is noise due to the granular morphologies of CCTO sample. Clearly, the samples with large Ca²⁺ ion concentrations exhibited a small shift in the XRD peaks towards larger 2θ (Fig. S1†). The shift was increased with increasing Ca²⁺ ion content corresponding to a decrease of the distances between the crystalline planes.

Fig. 1 XRD patterns of CCTO nanopowders with different Ca²⁺ ion content 1.0, 1.5, and 2.0 prepared by citrate precursor method and annealed at 1000 °C for 2 h.

The lattice parameter (a) and the unit cell volume (V_cell) for the cubic perovskite structure were calculated using the following equations:¹⁸

V_cell = a³ (2)

The X-ray density (d_x) was calculated using the following equation:

where N is the number of molecules per unit cell; N = 1 for CCTO, and the Avogadro's number NA = 6 × 10²³.

The apparent density (d_m) was measured in bi-distilled water according to the Archimedes principle using the following relation:

where m_s represents the mass of the sample in air, m_w is the mass of the sample in water and d_w is the density of water (1 g cm⁻³). The material porosity (P) was calculated using the following relation:

The variation in the structural parameters, such as crystallite size, lattice constant “a”, X-ray density (d_x) and the porosity, is given in Table 1. Lattice parameter slightly increased with increasing Ca²⁺ ion content. This is due to the larger ionic radii of Ca²⁺ ion (1.14 Å) as compared to Cu²⁺ ion (0.87 Å). Therefore, the unit cell volume increased from 400.15 to 401.78 ų upon increasing the molar ratio of the calcium ions from 1.0 to 2.0. Furthermore, the apparent density for all the prepared samples was smaller than that calculated with X-ray density. Besides, the percentage of porosity was decreased with increasing x value Fig. 2 of Ca²⁺ ion.

Table 1 Structural parameters of the produced (Ca_xCu_3−xTi₄O₁₂) powders with different Ca²⁺ ion content

Ca^2+ ion content	Crystallite size (nm)	Lattice parameter (Å) a	Unit cell volume (Å^3)	App. density dm (g cm^−3)	X-ray density dx (g cm^−3)	Porosity, P (%)
1.0	44	7.369 ± 0.001	400.152	3.859	5.052	23.61
1.5	52	7.378 ± 0.001	401.620	3.950	5.154	23.36
2.0	64	7.379 ± 0.001	401.783	3.997	5.215	23.31

---

Fig. 2 Lattice constant of CCTO nanopowders as a function of Ca²⁺ concentration.

Fig. 3 shows the SEM images of the CCTO powders prepared by citrate precursor method calcined at 1000 °C for 2 h. All the micro-images show nanosize particles, which gradually change their shape with the increase of Ca²⁺ ion molar ratios. All particles displayed cubic-like structure with grain size ranging between 50 and 100 nm. As observed from Fig. 3b and c micrographs, we can verify the formation of necks between the initial touching particles with the formation of elongated particles, which cause the growth of Ca_xCu_3−xTi₃O₁₂ particles relative to the amount of Ca²⁺ ion content. Moreover, all the samples have a homogeneous microstructure consisting of equiaxed grains. There was no significant difference in the microstructures of the CCTO. It can be seen that the irregularly shaped crystallites are formed with different sizes, as shown in Fig. 3a. For the x = 1.5 and 2.0 samples, the grains grow large, as shown in Fig. 3b and c. The largest and some smaller grain size are even larger than that of CCTO sample with 0.5 Ca²⁺ content. Energy dispersive X-ray (EDX) analysis of Ca_xCu_3−xTi₄O₁₂ (x = 1.0) nanopowders calcined at 1000 °C in air for 2 h prove the absence of any impurity peaks in the EDS spectrum. The close similitude of the atomic ratios of Ca, Cu and Ti to the nominal composition in Ca_xCu_3−xTi₄O₁₂ proves the elemental and phase purity of the prepared samples.^23,28

Fig. 3 Scanning electron micro-images of Ca_xCu_3−xTi₄O₁₂ at different Ca²⁺ ion content (a) 1.0, (b) 1.5 and (c) 2.0.

2. Optical properties

The optical properties of the synthesized CCTO particles annealed at 1000 °C for different Ca²⁺ molar ratios were examined by UV-Vis spectrophotometer and the results are depicted in Fig. 4. For the measurement, the synthesized powders were well dispersed in distilled water and the resultant solutions were used for the measurement. The transmittance T% spectrum recorded for the as prepared CCTO particles in the range of 200–800 nm (reflectance in ESI 2†) is given in Fig. 4b. All the samples showed transmittance increase of over 35% upon increasing the Ca²⁺ ions content in the wavelength range between 200 nm and 800 nm.

Fig. 4 (a) UV-visible transmittance spectrum (T%) of (Ca_1−xCu_xTi₄O₁₂) nanopowders synthesized using citrate precursor method annealed at 1000 °C for 2 h using different Ca²⁺ ion content (1.0, 1.5 and 2.0); (b) the band gap energy of CCTO at different Ca²⁺ ion content (1.0, 1.5, and 2.0).

The well-structured and smooth oscillations of the transmittance profiles indicate that all the samples have flat surfaces and uniform size. The band gap energy was determined by extrapolating the absorption coefficient (α) to zero from the spectral data. The absorption coefficient was calculated by the following equation

(αhν)^m = hν − E_g (6)

where α is the absorption coefficient, hν is the photon energy, E_g is the band gap energy, m = 1/2 or 3/2 for indirect allowed and indirect forbidden transitions, and m = 2 or 3 for direct allowed and direct forbidden transitions.

Liu et al.¹⁸ described that the band gap energy is direct when the electronic transitions occur from the maximum-energy states near or inside the valence band (VB) to minimum-energy states below or inside the conduction band (CB), in the same regions in the Brillouin zone. Therefore, the presence of different E values calculated from the UV-vis absorption spectra indicates the existence of intermediary energy levels between the valence and the conduction band.²⁹ The band gap energy was estimated by plotting (αhν)² of the CCTO against the photon energy (hν). The linear relationship between (αhν)² and hν supports the model of direct allowed band electronic transition. The band gap energy was determined by extrapolating the absorption coefficient (α) to zero. The absorbance (A) can be converted to the absorption coefficient using the following relationship:²⁴

where A is the absorbance of the sample, ρ is the density of CCTO, l is the length (1 cm), and c is the concentration of the CCTO nanocrystals. By plotting (αhν)² vs. hν, the optical band-gaps (E_g) of the samples were realized as the intercepts with x-axis, as shown in Fig. 4. For Ca_xCu_3−xTi₄O₁₂ (x = 1.0, 1.5 and 2.0), the band-gap energies were found to be 3.12, 3.20, and 3.26 eV, respectively. The E_g increased with increasing Ca²⁺ ion content. The electronic transitions occurred inside the Ca_xCu_3−xTi₄O₁₂ microcrystals (Fig. 4). The 3d orbitals of the copper atoms associated to the conduction band,²⁷ thus a decrease of the optical gap was observed. Moreover, structural defects such as distortions and/or strains in the CaTiO₃ lattice caused by the introduction of copper in these systems are able to induce the symmetry break of the [TiO₆] and [CaO], leading to an appearance of intermediary levels between the valence and conductions bands.^25,26,30

3. Electrical resistivity

The effect of calcium ion on the DC resistivity of CCTO was investigated. Table 2 presents the variation of DC resistivity versus Ca²⁺ ion content. It is observed that the DC resistivity was markedly dependent on the substitution of Ca ions. The resistivity at room temperature increased with the increasing of Ca²⁺ content most probably due to the high degree of crystallization observed in the XRD and the low mobility of the Ca²⁺ ions.

Table 2 Variation of DC resistivity versus Ca²⁺ ion content

Ca^2+ ion content	1.0	1.5	2.0
ρ (Ω cm)	6.4 × 10^4	6.6 × 10^4	6.8 × 10^4

---

4. Dielectric studies

Fig. 5 shows the plots of variations in dielectric constant (real part (ε′) and imaginary part (ε′′)) and dielectric loss (tan δ), measured for the silver painted samples at room temperature over the frequency range from 1 MHz to 3 GHz. The dielectric constant values obtained at the high frequency region are as high as those reported by other authors such as Masingboon et al.,¹⁹ Jin et al.¹⁷ and Liu et al.²³ They also prepared CCTO samples using wet chemical methods. At room temperature, it is seen that all samples were approximately dropped for frequencies higher than 1 GHz. The analysis of the dependence of complex permittivity on frequency at room temperature shows the existence of two relaxation processes for all the samples. The high and low frequency relaxations are usually associated with the grain and grain boundary dielectric response,^32,33 respectively.

Fig. 5 (a) Frequency-dependence of the real part permittivity, (b) imaginary part permittivity and (c) loss factor (tan δ) of CaCu₃Ti₄O₁₂ nanopowders with silver-paint contacts prepared by citrate precursor with different Ca²⁺ ion content.

The porosity of materials directly affects the dielectric characteristics such as real and imaginary permittivity. It is possible that dielectric constant increases with larger grain size. Densification of sample also plays a major part in contributing to the value of dielectric constant of CCTO. A porous sample makes the dielectric properties more difficult to penetrate and it also dissipates more heat. Hence, it will reduce the dielectric constant of the sample.^34,35

The frequency dependent dielectric constant (ε′), (ε′′) and dielectric loss (tan δ) at room temperature for the samples, namely, Ca²⁺ 1.0, Ca²⁺ 1.5 and Ca²⁺ 2.0, are shown in Fig. 5. The dielectric constant value obtained for Ca²⁺ 1.0 samples is around 0.02 × 10³ at 1.5 GHz, which is increased to 0.05 × 10³ as the frequency increased to 1.75 GHz and then decreased with further increase in frequency.

The Ca²⁺ 1.5 sample, which has higher calcium content (by 0.5 mol%) exhibited a dielectric constant value of around 0.05 × 10³ at 1.65 GHz. The Ca²⁺ 1.5 sample shows low frequency dispersion as compared to that of Ca²⁺ 1.0 samples. Interestingly, Ca²⁺ 2.0 sample, which has higher calcium content, exhibited very low dielectric constant as compared to the other samples (Ca²⁺ 1.0 and Ca²⁺ 1.5) at all the frequencies studied. The loss factor varies with the change of calcium ion content. The sample with higher calcium ion content has the higher loss factor compared to the other samples. The dielectric loss did not show any relaxation at low frequency, though there is a relaxation at high frequency.

The dielectric loss values obtained for Ca 1.0 (at 1.8 GHz) is 25 and for Ca 1.5 (at 2.0 GHz) is 50. For the rich calcium sample, it will be 300 at 2.3 GHz. It should be noted that the calcium deficient sample (Ca²⁺ 1.0), exhibited low dielectric loss while retaining the high dielectric constant. According to the literature, various dopants can be used to decrease the dielectric loss in CCTO.^{18,19,22–25,27,29} The dielectric losses reported in this study are slightly high. This may be attributed to the porosity of the materials and the measurement of the dielectric losses in high frequency range (gigahertz range). However, this matches with the frequency range measurements and the value reported by Fritsch et al.¹⁶

The reduction in the dielectric loss in CCTO also affected the dielectric constant to a large extent. However, in this work, without any dopant, the dielectric loss was suppressed remarkably while retaining the giant dielectric constant solely by controlling the chemistry and engineering the interfacial regions at the grain boundaries.

Studies are in progress to measure dielectric breakdown field strength as the function of different Ca²⁺ content.

D. Conclusions

In this study, pure Ca_xCu_3−xTi₄O₁₂ nanopowders were successfully synthesized through the citrate precursor route using low-cost starting materials. The crystallite size varies from 44.2 to 64.8 nm as a result of increasing Ca²⁺ ion content. The microstructure of the formed powders appeared as cubic-like structure. The high transparency of the formed Ca_xCu_3−xTi₄O₁₂ was considered with Ca²⁺ ions content of 1.0. It should be noted that the band gap energy was increased with increasing Ca²⁺ ion concentrations. For instance, it was increased from 3.12 to 3.26 eV by increasing the Ca²⁺ ion ratio from 1.0 to 2.0. Moreover, the imaginary part of complex permittivity and dielectric loss factor were increased with increasing Ca²⁺ ion content. These results indicate that the CCTO is a promising future material in electronic applications such as capacitors, memory device, resonators and filters.

Acknowledgements

This work was partially supported by the French Government through a fellowship granted by the French Embassy in Egypt (Institute francais d'Egypt).

Notes and references

1. F. Carpi, G. Gallone, F. Galantini and D. De Rossi, Adv. Funct. Mater., 2008, 18, 235–241.
2. C. Huang and Q. M. Zhang, Adv. Mater., 2005, 17, 1153–1158.
3. P. Kim, X.-H. Zhang, B. Domercq, S. C. Jones, P. J. Hotchkiss, S. R. Marder, B. Kippelen and J. W. Perry, Appl. Phys. Lett., 2008, 93, 013302.
4. K. H. Lee, J. Kao, S. S. Parizi, G. Caruntu and T. Xu, Nanoscale, 2014, 6, 3526–3531.
5. Q. M. Zhang, H. Li, M. Poh, F. Xia, Z. Y. Cheng, H. Xu and C. Huang, Nature, 2002, 419, 284–287.
6. W. L. Li, Y. Zhao, Q. G. Chi, Z. G. Zhang and W. D. Fei, RSC Adv., 2012, 2, 6073–6078.
7. B. Cheng, Y.-H. Lin, J. Yuan, J. Cai, C.-W. Nan, X. Xiao and J. He, J. Appl. Phys., 2009, 106, 034111.
8. M. A. de la Rubia, P. Leret, A. del Campo, R. E. Alonso, A. R. López-Garcia, J. F. Fernández and J. de Frutos, J. Eur. Ceram. Soc., 2012, 32, 1691–1699.
9. L. Feng, X. Tang, Y. Yan, X. Chen, Z. Jiao and G. Cao, Phys. Status Solidi A, 2006, 203, R22–R24.
10. Q. Zheng, H. Fan and C. Long, J. Alloys Compd., 2012, 511, 90–94.
11. P. Kum-onsa, P. Thongbai, B. Putasaeng, T. Yamwong and S. Maensiri, J. Eur. Ceram. Soc., 2015, 35, 1441–1447.
12. L. Singh, U. S. Rai, K. D. Mandal and A. Rai, Appl. Phys. A, 2013, 112, 891–900.
13. L. Singh, I. W. Kim, B. C. Sin, K. D. Mandal, U. S. Rai, A. Ullah, H. Chung and Y. Lee, RSC Adv., 2014, 4, 52770–52784.
14. L. Cao, P. Liu, J.-P. Zhou, Y.-J. Wang, L.-N. Su, C. Liu and H.-W. Zhang, J. Ceram. Process. Res., 2010, 11, 453–459.
15. W. X. Yuan, S. K. Hark and W. N. Mei, J. Ceram. Process. Res., 2009, 10, 696–699.
16. S. Guillemet-Fritsch, T. Lebey, M. Boulos and B. Durand, J. Eur. Ceram. Soc., 2006, 26, 1245–1257.
17. S. Jin, H. Xia, Y. Zhang, J. Guo and J. Xu, Mater. Lett., 2007, 61, 1404–1407.
18. J. Liu, R. W. Smith and W.-N. Mei, Chem. Mater., 2007, 19, 6020–6024.
19. C. Masingboon, P. Thongbai, S. Maensiri, T. Yamwong and S. Seraphin, Mater. Chem. Phys., 2008, 109, 262–270.
20. C. Puchmark and G. Rujijanagul, Ferroelectrics, 2011, 415, 107–112.
21. C.-M. Wang, K.-S. Kao, S.-Y. Lin, Y.-C. Chen and S.-C. Weng, J. Phys. Chem. Solids, 2008, 69, 608–610.
22. S. K. Manik and S. K. Pradhan, Phys. E, 2006, 33, 160–168.
23. L. Liu, H. Fan, P. Fang and X. Chen, Mater. Res. Bull., 2008, 43, 1800–1807.
24. A. K. Rai, N. K. Singh, S.-K. Lee, K. D. Mandal, D. Kumar and O. Parkash, J. Alloys Compd., 2011, 509, 8901–8906.
25. S. Jesurani, S. Kanagesan, T. Kalaivani and K. Ashok, J. Mater. Sci.: Mater. Electron., 2012, 23, 692–696.
26. J. Liu, Y. Sui, C.-G. Duan, W.-N. Mei, R. W. Smith and J. R. Hardy, Chem. Mater., 2006, 18, 3878–3882.
27. M. M. Rashad, A. O. Turky and A. T. Kandil, J. Mater. Sci.: Mater. Electron., 2013, 24, 3284–3291.
28. D. Xu, K. He, R. Yu, X. Sun, Y. Yang, H. Xu, H. Yuan and J. Ma, Mater. Chem. Phys., 2015, 153, 229–235.
29. B. Barbier, C. Combettes, S. Guillemet-Fritsch, T. Chartier, F. Rossignol, A. Rumeau, T. Lebey and E. Dutarde, J. Eur. Ceram. Soc., 2009, 29, 731–735.
30. P. Mishra and P. Kumar, Ceram. Int., 2015, 41, 2727–2734.
31. M. A. Ponce, M. A. Ramirez, F. Schipani, E. Joanni, J. P. Tomba and M. S. Castro, J. Eur. Ceram. Soc., 2015, 35, 153–161.
32. S. I. R. Costa, M. Li, J. R. Frade and D. C. Sinclair, RSC Adv., 2013, 3, 7030–7036.
33. R. Schmidt, S. Pandey, P. Fiorenza and D. C. Sinclair, RSC Adv., 2013, 3, 14580–14589.
34. L. Singh, I. W. Kim, B. Cheol Sin, A. Ullah, S. Kook Woo and Y. Lee, Mater. Sci. Semicond. Process., 2015, 31, 386–396.
35. A. K. Rai, J. Gim, E.-C. Shin, H.-H. Seo, V. Mathew, K. D. Mandal, O. Parkash, J.-S. Lee and J. Kim, Ceram. Int., 2014, 40, 181–189.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra15222k

---