The effect of annealing oxygen concentration in the transformation of Ca_xCoO₂ to thermoelectric Ca₃Co₄O₉

Abstract

Calcium cobalt oxide thin films were deposited using radio-frequency (RF) sputtering on single crystal c-axis sapphire substrates. The as-deposited calcium cobaltite crystallized in the form of Ca_xCoO₂, which has a slight calcium deficiency compared to the annealed Ca₃Co₄O₉ phase. When the films have a slight calcium deficiency, transformations to the Ca₃Co₄O₉ phase proceed with Co₃O₄ as a by-product. It was discovered that this transformation can take place with a relatively low oxygen concentration in the annealing gas, as low as 5% in this study. However, optimum thermoelectric and electrical properties were achieved in the films annealed with an oxygen concentration of about 20%. There was no significant change in such properties when the films were annealed at higher oxygen concentrations. This suggests that the optimum defect concentration induced via the transformation occurs with an oxygen concentration of about 20% in the annealing atmosphere.

---

Introduction

The layered calcium cobaltite Ca₃Co₄O₉ (CCO-349) has an intriguing crystal structure consisting of a Ca₂CoO₃ rock-salt layer sandwiched between two hexagonal CoO₂ layers.¹ This material is the subject of intense research because of its remarkable high temperature thermoelectric (TE) properties with a high figure of merit (ZT = S²/ρk, where S, ρ and k are Seebeck coefficient, resistivity and thermal conductivity) of 0.83 at 1000 K.² Apart from its remarkable TE performance at high temperature, CCO-349 is much more stable and has lower toxicity and is hence environmentally friendlier than the chalcogenides such as Bi₂Te₃.^3,4 There are many reported routes for fabricating CCO-349. A common observation among these fabrications was the formation of an intermediary phase that preceded the formation of CCO-349.⁵ This is particularly evident in the case of thin film CCO-349. This phase is commonly known as Ca_xCoO₂ (CCO-x12), which has a calcium ratio between 0.26–0.5 per atomic unit of cobalt.⁶ Instead of two hexagonal CoO₂ layers sandwiching a rock-salt layer, the crystal structure of CCO-x12 comprises of a Ca cation layer sandwiched between the CoO₂ layers.⁷

While the effect of pure oxygen and air on the properties of the resultant phase has been investigated,^8–10 these studies focused on the defect formation post CCO-349 formation. It also has been reported that the formation of CCO-349 occurs via the reaction of CCO-x12 with excess calcium oxide.⁷ However, transformations in a calcium deficient sample have not been reported. In addition, the dependence of this transformation on the annealing atmosphere with various oxygen concentrations and the properties of the resultant films have not been investigated. Seebeck and electrical properties of CCO-349 are reportedly affected by the presences of oxygen vacancies.^11,12 These vacancy defects have been created by annealing materials in the absence of oxygen.¹¹ Since there is a stoichiometric shift and structural change associated with the transformation from CCO-x12 to CCO-349, controlling the oxygen concentration during the phase transformation from CCO-x12 to CCO-349 presents an alternative route for creating vacancy defects. This work attempts to understand the influence of oxygen concentration in the annealing atmosphere of CCO-x12 on the resultant properties of CCO-349 films.

Experiment details

The CCO-349 target used in the RF sputtering process was first synthesized via a conventional solid-state reaction at 900 °C using high purity CaCO₃ (>99%) and Co₃O₄ (>99%) powders (Sigma Aldrich). The resultant pure CCO-349 powder was then crushed and re-densified at 800 °C using Spark Plasma Sintering (SPS) to produce the two-inch sputtering target with approximately 98% density. Prior to deposition, c-cut (0001) Al₂O₃ single crystal substrates were ultrasonically washed in acetone and deionised water before annealing at 1000 °C for 1 hour in air. The thermal treatment process resulted in the formation of an atomic step, thus creating a smoother surface. This is shown in the atomic force microscope image in Fig. 1. RF sputtering was carried out at 800 °C, with RF power of 100 W, at 20 mTorr, with the deposition gas consisting of 67% oxygen balanced with argon for all depositions.

Fig. 1 AFM images of Al₂O₃ substrate surface treatment.

These films were annealed in a tube furnace at 800 °C for 2 hours in flowing gases with oxygen concentrations varied between 0%–100%, balanced with nitrogen to maintain a consistent total flow rate of 200 sccm. The ratio of gases in the annealing gas mixtures were controlled using two MKS Instruments thermal mass flow controllers. The microstructure of the films was analysed using an NT-MDT Ntegra Spectral atomic force microscope (AFM) while X-ray diffraction was performed on a Bruker D8. Chemical composition of the films was examined with a Perkin Elmer NexION ICPMS integrated with an ESI-NewWave NWR213 laser ablation inductive coupled mass spectrometer (LA-ICPMS). Thermoelectric properties of the films were analyzed using Ulvac Riko ZEM3 Thermoelectric Measurement system.

Results and discussion

Morphology of the films was analysed using an atomic force microscope in semi-contact mode. As shown in Fig. 2a the as-deposited CCO-x12 crystals are approximately 100 nm in size. After the film was annealed, morphological transformation takes place. Fig. 2b shows the microstructure of films annealed in 100% flowing oxygen. As can be seen, the relatively uniform crystals of CCO-x12 transformed to platelet-like crystals of approximately 2 μm in size. The majority of these platelet crystals appeared to have grown with the c-axis perpendicular to the surface of the film. Fig. 2c is an AFM micrograph of a CCO-x12 film annealed in 5% oxygen. In general, the microstructure of this film appeared similar to the one annealed in 100% oxygen in terms of crystal morphology and orientation. This suggests that the annealed films produced in this work are highly c-axis textured. While the AFM images of films annealed in 5% and 100% oxygen bear a close resemblance to each other, the film annealed in the absence of oxygen had a microstructure that is significantly different. Fig. 2d shows the microstructure of the film annealed in the absence of oxygen; the resultant film consisted of crystals of about 100 nm in size. It appeared that the as-deposited CCO-x12 phase decomposed as shown in the XRD results in the following section.

Fig. 2 AFM images of (a) as-deposited CCO-x12, (b) annealed in 100% oxygen atmosphere, (c) annealed in 5% oxygen atmosphere and (d) annealed in pure nitrogen atmosphere (0% oxygen).

Fig. 3 shows the XRD patterns of the as-deposited film and those annealed under the various oxygen concentrations as indicated. It can be observed from the XRD patterns that the as-deposited film crystallized as a Ca_xCoO₂ phase (CCO-x12). This phase has been reported to have a calcium atomic ratio between 0.26–0.5 per unit of cobalt⁶ and c-axis spacing that varies with its calcium content.¹³ XRD analysis of the CCO-x12 suggested that the films produced in this work crystallized in the form of Ca_0.5CoO₂ (JCPDS PDF card no. 00-037-0669). This suggested that the deposited film had a slightly higher content of calcium. When the film was annealed in 100% oxygen, the XRD result showed that the CCO-x12 phase transformed into a CCO-349 phase with the simultaneous appearance of a Co₃O₄ phase. Since Co₃O₄ was not detected in the as-deposited films, it appeared that Co₃O₄ was formed as a consequence of the formation of CCO-349 under the investigated conditions. This trend of simultaneous CCO-349 and Co₃O₄ co-existence persisted with reduced oxygen concentration of the annealing gas, down to 5%. However, when the film is annealed in flowing pure nitrogen (i.e. 0% O₂), the CCO-x12 phase decomposed giving a mixture of CaO and CoO, as shown in the XRD. Analysis of the XRD results revealed that there are no significant differences in the films annealed under different oxygen concentrations, down to 5% O₂.

Fig. 3 XRD pattern of as-deposited films and films annealed in atmospheres with different oxygen concentrations.

It has been reported that transformation from CCO-x12 to CCO-349 takes place via a topotactical transformation involving the rearrangement of two calcium layers with one CoO₂ layer resulting in the rock salt Ca₂CoO₃ phase sandwiched between two hexagonal CoO₂ layers.⁷ However, given the differences in calcium content between the two phases, calcium enrichment would be necessary to facilitate formation of CCO-349.¹⁰ The chemical composition of the films was analysed using a laser ablation inductive coupled mass spectrometer (LA-ICPMS). The analysis revealed that the as-deposited films had a calcium concentration of 0.64 : 1 per atomic unit of cobalt instead of the 0.75 : 1 expected of the CCO-349 phase. As seen from the XRD, the as-deposited films appeared to comprise only of the CCO-x12 phase. This phase has a calcium concentration lower than that given by LA-ICPMS, this implies that the as-deposited film would have to contain traces of CaO that are not detected by XRD. It is estimated that the residual calcium oxide has a volume fraction of about 3%. The sputtering target used in this work had a Ca : Co ratio of 0.78 : 1 as determined by LA-ICPMS. This suggests that the sputtering yield for calcium and cobalt are not the same, resulting in films that have lower calcium concentrations. Given the differences in calcium content between CCO-x12 and CCO-349, calcium enrichment would be necessary to facilitate formation of CCO-349.⁷ It has been reported that in the transformation to CCO-349 from CCO-x12, there could be residual CCO-x12 phase remaining.^14,15 With the film having less than the theoretical optimum calcium content, this would imply that there should be residual untransformed CCO-x12 phase. However, XRD analysis indicated the presence of Co₃O₄ in the post annealed films. This suggested that formation of CCO-349 could have partially taken place via dissociation of CCO-x12 leading to calcium enrichment of the resultant CCO-349.

In this present work, it was found that an oxygen concentration as low as 5% is sufficient to facilitate CCO-349 formation from the CCO-x12 phase and that oxygen concentration in the annealing atmosphere did not appear to affect the resultant phase formations. However the impact of such treatment on the electrical properties of the films needs to be evaluated. Fig. 4 shows the room temperature resistivity fluctuation with the oxygen concentration in the annealing gas (the error bar is 5%). It can be observed that resistivity of the resultant films decreases with increases in oxygen concentration to 10% O₂. Beyond which there is only a slight ∼10% increase in the resistivity. Annealing in low oxygen favours the formation of oxygen vacancies, which can cause nonstoichiometric-oxygen in the CCO-349 system. Previous studies combined with detailed thermodynamic analysis suggested that the oxygen vacancies generally reduce the average Co valence within the rock-salt layer.^16,17 The result of such reduced Co valency is that the carrier concentration will be affected by the concentration of Co²⁺ in the rock-salt layer. It was reported that the electrical conduction mechanism of CCO-349 films in the high temperature region is related to a polaron hopping conduction model.¹⁸ The temperature dependence of electrical resistivity is given by:¹⁹

where n, a, k_B and E_a are carrier concentration, hoping distance, the Boltzmann constant and activation energy, respectively. From eqn (1), the activation energy (E_a) can be obtained from the slope of the fitted straight line of ln(ρ/T) versus 1000/T. As indicated in Fig. 5, E_a was found to be at ∼85–89 meV in the high temperature region (T > 550 K). This is slightly higher than the pure CCO-349 ceramic activation energy reported (77–85 meV).²⁰ Moreover, E_a remained in the range of 85–89 meV for different samples. It is not immediately clear what the factors contributing to the slight difference in activation energy of the CCO-349 films obtained in this experiment as compared to the reported values are. However, it can be seen that varying heat treatment on the films did not produce any differences in the activation energy for electrical conductions. This suggests that the increase in electrical resistivity could be attributed to the reduction of carrier concentration as a result of reduced Co valency.

Fig. 4 Oxygen concentration dependence of electrical resistivity for Ca₃Co₄O₉ films.

Fig. 5 Plot of ln(ρ/T) versus 1000/T for Ca₃Co₄O₉ films annealed in 5%, 10% and 100% oxygen concentration.

The Seebeck effect was measured to determine the effect of such annealing treatment on the thermoelectric performance of the films. Fig. 6 shows the variation of Seebeck voltage and power factor of the films at room temperature versus oxygen concentration (the error bar is 5%). As can be seen, Seebeck voltage rapidly increases with increasing oxygen concentration of the annealing atmosphere up to about 10% O₂. Beyond that, there is no significant change in Seebeck voltages. The trend is almost identical for the power factor. Seebeck performance has been reported to increase with the vacancy or defect content of the samples.¹⁸ However, it is unclear what causes the simultaneous increase in electrical resistivity and decrease in Seebeck coefficient. It is possible that this was contributed to by a higher fraction of secondary phases that are not evident in the XRD analysis. In this work, it was found that increasing the defect content of the films, while still retaining the phase as CCO-349, does not bring about a concomitant increase in Seebeck performance. The results suggested that a minimum level of oxygen concentration is required in the annealing gas to reduce resistivity and increase Seebeck potential.

Fig. 6 Seebeck coefficient and power factor of the films annealed under various oxygen concentrations.

Conclusion

Thin film calcium cobaltite was deposited on (0001) sapphire using the RF sputtering technique. The as-deposited films crystallized in a Ca_xCoO₂ phase along with traces of CaO. The resultant films are slightly calcium deficient with a 0.64 : 1 calcium to cobalt ratio instead of the 0.75 : 1 that was anticipated. Under these calcium deficient conditions, the transformation of CCO-x12 to CCO-349 resulted in the formation of Co₃O₄. The dependence on oxygen of the phase transformation from CCO-x12 to CCO-349 was investigated. It was revealed that the transformation could proceed with an oxygen concentration as low as 5%. However, optimum properties of CCO-349 were obtained when a minimum of 20% oxygen is present in the annealing atmosphere. In this work, it was revealed that the thermoelectric and electrical performance of the films do not appear to increase with increasing oxygen concentration of the annealing gas.

Acknowledgements

This project is supported by Australian Research Council (Grant no. DP0988687, DP110102662, FT100100956 and LP120200289) and Baosteel. Li Zhang appreciates the support of Chinese Scholarship Council.

References

1. A. C. Masset, C. Michel, A. Maignan, M. Hervieu, O. Toulemonde, F. Studer, B. Raveau and J. Hejtmanek, Phys. Rev. B: Condens. Matter Mater. Phys., 2000, 62, 166–175.
2. M. Shikano and R. Funahashi, Appl. Phys. Lett., 2003, 82, 1851–1853.
3. R. Venkatasubramanian, E. Siivola, T. Colpitts and B. O’Quinn, Nature, 2001, 413, 597–602.
4. T. Caillat, M. Carle, P. Pierrat, H. Scherrer and S. Scherrer, J. Phys. Chem. Solids, 1992, 53, 1121–1129.
5. T. Sun, J. Ma, Q. Y. Yan, Y. Z. Huang, J. L. Wang and H. H. Hng, J. Cryst. Growth, 2009, 311, 4123–4128.
6. B. L. Cushing and J. B. Wiley, J. Solid State Chem., 1998, 141, 385–391.
7. H. Itahara, W.-S. Seo, S. Lee, H. Nozaki, T. Tani and K. Koumoto, J. Am. Chem. Soc., 2005, 127, 6367–6373.
8. R. Wei, H. Jian, X. Tang, J. Yang, L. Hu, L. Chen, J. Dai, X. Zhu and Y. Sun, J. Am. Ceram. Soc., 2013, 96, 2396–2401.
9. C.-J. Liu, P. K. Nayak, Z.-R. Lin and K.-Y. Jeng, Thin Solid Films, 2008, 516, 8564–8568.
10. Y. Zhou, I. Matsubara, W. Shin, N. Izu and N. Murayama, J. Appl. Phys., 2004, 95, 625–628.
11. M. Karppinen, H. Fjellvåg, T. Konno, Y. Morita, T. Motohashi and H. Yamauchi, Chem. Mater., 2004, 16, 2790–2793.
12. M. Schrade, S. Casolo, P. J. Graham, C. Ulrich, S. Li, O.-M. Løvvik, T. G. Finstad and T. Norby, J. Phys. Chem. C, 2014, 118, 18899–18907.
13. K. Sugiura, H. Ohta, K. Nomura, H. Hosono and K. Koumoto, in Advances in Electronic Ceramics II, John Wiley & Sons, Inc., 2010, pp. 93–105.
14. T. Sun, H. Hng, Q. Yan and J. Ma, J. Electron. Mater., 2010, 39, 1611–1615.
15. R. Moubah, S. Colis, C. Ulhaq-Bouillet and A. Dinia, Appl. Phys. Lett., 2010, 96, 0419021–0419023.
16. Y. Morita, J. Poulsen, K. Sakai, T. Motohashi, T. Fujii, I. Terasaki, H. Yamauchi and M. Karppinen, J. Solid State Chem., 2004, 177, 3149–3155.
17. J. Shimoyama, S. Horii, K. Otzschi, M. Sano and K. Kishio, Jpn. J. Appl. Phys., Part 2, 2003, 42, L194–L197.
18. M. Ohtaki, H. Koga, T. Tokunaga, K. Eguchi and H. Arai, J. Solid State Chem., 1995, 120, 105–111.
19. D. P. Karim and A. T. Aldred, Phys. Rev. B: Condens. Matter Mater. Phys., 1979, 20, 2255–2263.
20. Y.-H. Lin, C.-W. Nan, Y. Liu, J. Li, T. Mizokawa and Z. Shen, J. Am. Ceram. Soc., 2007, 90, 132–136.

---