Highly (001)-oriented Calcium Barium Niobate (CaxBa1-xNb2O6, CBN) thin films were grown on MgO by Radio-Frequency magnetron sputtering. Close-to-bulk film stoichiometry (Ca0.28Ba0.72Nb2O6) is obtained for an O2 fraction of 5% in the deposition chamber. At the annealing temperature of 1000 °C, (001) oriented thin films are achieved with lattice parameter in the c-direction and a chemical composition very close to that of the bulk, and a surface roughness of 2.8 nm RMS. The refractive index of the films is 2.21 at λ = 630 nm and 2.13 at λ = 1550 nm and a strong second harmonic signal can be generated nonlinearly in the films.
© 2015 Optical Society of America
Crystals of the Tetragonal Tungsten Bronze (TTB) family have been widely studied in the past years due to their good electro-optic (EO) properties and transparency at telecommunication wavelengths. In particular, Strontium Barium Niobate (SrxBa1-xNb2O6, SBN) has the highest measured EO coefficient to date i.e. 1340 pm / V in its bulk form  and 844 pm / V in the form of thin films, for x = 0.6 . These EO coefficients are much higher than those of conventional Lithium Niobate (LiNbO3, LN) that has a EO coefficient of about 30 pm / V in its bulk form . However, the relatively low Curie temperature (80 °C)  of SBN makes it unsuitable for integrated optics, where high density data packets can generate a lot of heat. Recently, some groups got interested in a new TTB material, CaxBa1-xNb2O6 (CBN) with its EO coefficient 4 times larger than LN (130 pm / V)  and its Curie temperature higher than 250 °C . Even though most of the recent studies on CBN focus on the material’s properties in its bulk form [7,8], some effort was undertaken to study this material in the form of thin films that would allow its use for high performance integrated optics [9,10]. Achieving EO thin films with high crystalline ordering is critical to obtain high EO performance. Three groups have studied the epitaxial growth of CBN on Strontium Titanate (SrTiO3, STO) [5,11] and Magnesium Oxide (MgO) [12,13] substrates using Pulsed Laser Deposition (PLD). The optimized growth of CBN results in both high crystalline quality and EO performance, paving the way to high performance EO devices such as EO modulators . However, even though PLD is a very versatile technique to grow epitaxial and high quality complex oxides, it suffers from several drawbacks such as rather poor surface quality with the presence of small particles and droplets randomly distributed over the surface. In addition, due to the high directivity and the small area of the pulsed plasma plume, it is difficult to grow thin films over large areas with very good thickness homogeneity . Therefore, these drawbacks have to be successfully addressed before very high performance EO devices could be fabricated. Radio Frequency (RF) Magnetron Sputtering is a powerful deposition technique that has been fruitfully used to grow high quality complex oxides [15–17]. It has the capability to grow thin films on large surfaces with very good homogeneity, low roughness and no droplets . In this article, we investigate the deposition of CBN by RF magnetron sputtering on MgO substrates. We show the influence of the deposition parameters and of the annealing temperature on the structural and optical properties of the thin films. We also demonstrate the growth of highly c-oriented thin films with high surface quality and good non-linear properties.
2. Experimental setup
CBN thin films were deposited on MgO substrates using a 3 inch Ca0.28Ba0.72Nb2O6 commercial target in a K.J Lesker CMS-18 sputtering system. The target was sputtered using a RF source operated in argon at a power of 300 W. Oxygen was supplied by a ring containing holes located halfway between the source and the substrate. The oxygen fraction in argon was varied from 0% to 10% by adjusting each mass flow controller accordingly. The total pressure in the chamber during deposition was maintained at 4 mTorr by automatically adjusting the total flow of the gas in the chamber while putting the gateway valve to “throttle” position. Deposition was performed at a substrate temperature of 200 °C. After deposition, the thin films were crystallized ex-situ by Rapid Thermal Annealing (RTA) using a Heatpulse AG610 for 1 min at a temperature varying from 600 °C to 1150 °C in an oxygen atmosphere. The structural characterization of the films was carried out using X-Ray Diffraction (XRD) in a Panalytical X-PERT Pro MRD system. The diffractometer was used in the Thêta-2Thêta (Bragg-Brentano) configuration for out-of-plane orientation, in the Phi-scan configuration for in-plane orientation. The chemical composition before annealing was measured using X-Ray Photoelectron Spectroscopy (XPS) in a VG Escalab 220i XL system. In order to remove surface contamination, Ar etching of the film surface was conducted prior to analysis for 10 minutes in the high vacuum chamber. The detailed chemical composition of the thin films after annealing was measured using Rutherford Backscattering Spectroscopy (RBS) and Elastic Recoil Detection Analysis (ERDA). The topography of the thin film surface was examined by Atomic Force Microscopy (AFM) using a Veeco Enviroscope in the tapping mode.
The optical characterization of CBN thin film was performed using a prism coupler (Metricon Model 2010) at laser wavelengths (λ) of 630, 960, 1310 and 1550 nm. A rutile prism was used to couple the laser beam to the guided modes of the CBN / MgO thin film structure. For this purpose, the CBN / MgO thin film was brought into contact with the base of the prism and the reflected intensity was measured as a function of the incident angle. The reflectivity dips that occur at some specific incident angles correspond to the excitation of guided modes. The ordinary guided modes of our sample with the optical axis parallel to the surface were studied by using TE polarized light. Further optical characterization consists in measuring the dispersion parameters using a VVASE spectrometric ellipsometer from J.A. Woollam in the reflection geometry. The measurements were carried from 400 nm to 1700 nm using two detectors, one for UltraViolet-Visible (UV-Vis) and another one for Near InfraRed (NIR). The ellipsometric data were then analyzed by a Sellmeier model to obtain the refractive index of the thin films. To evaluate the optical quality of our CBN thin films, we also performed transmittance measurement using the ellipsometer.
The non-linear optical properties were investigated by means of a second harmonic generation (SHG) microscope. Briefly, the excitation was provided by a Titanium:Sapphire laser (Tsunami, Spectra Physics) delivering 150 fs pulses at a 80 MHz repetition rate. The fundamental wavelength was set at 810 nm and the average power impinging the sample was adjusted using a half-wave plate and a Glan Thompson polarizer. Images were acquired with a laser scanning microscope (Till Photonics GmbH, Munich, Germany) based on galvanometric mirrors. Polarization was set at 45° for all the measurements which corresponds to the alignment with the MgO substrate (100) orientation. The focusing objective was an Olympus UplanSApo 20X air immersion microscope objective with a numerical aperture of 0.75. The beam size was increased using a telescope to fill the back pupil of the objective, provide optimal resolution of typically 600 nm (FWHM). After interaction with the thin film, the fundamental beam was rejected using two wide bandpass filters (380-700 nm, Semrock). A narrow bandpass filter (405 ± 10 nm, Semrock) is used to select the SHG wavelength. Finally, signal was collected with a condensor and detected using a photomultiplier tube (R6357, Hamamatsu Corporation) powered at 600 V. The pixel size was set at 200 nm in accordance with Nyquist criteria.
3. Results and discussion
3.1 Thin film composition and crystallization
In the case of RF sputtering, the transfer of composition from the target to the thin film for a specific substrate strongly depends both on the deposition rate and on the O2 fraction in the sputtering plasma . In the case of TTB deposition, higher deposition rates lead to more stoichiometric films. In what follows, the RF power applied to the 3 inch target was set to 300 W, which results in a suitable deposition rate (above 20 Å / min) for all the conditions studied. This enables to achieve a sufficiently high deposition rate while maintaining the ceramic target integrity that would be damaged at higher RF power. Figure 1 shows the dependence of the deposition rate on the O2 fraction (PO2) in the plasma, where PO2 is determined as:Fig. 1 that the deposition rate decreases from 84 Å / min to 21 Å / min when the oxygen fraction in the chamber increases from 0% to 10%. Such a behavior is expected as introducing oxygen in the plasma is likely to reduce the plasma density [19,20], hence the ion flux on target. Furthermore, oxygen is less efficient than argon for sputtering. In addition, the sputtering yield of each target species by Ar and O2 may differ, so that adding oxygen is expected to change both the deposition rate and the stoichiometry of the films.
The composition analysis was carried out by XPS on as-deposited samples after in situ Ar etching for 10 minutes to remove surface contamination. Figure 2 shows the ratios Ca / Nb2 and Ba / Nb2 as a function of the O2 fraction in the deposition chamber. It can be seen that the Ca / Nb2 ratio decreases slowly from 0.29 down to 0.16 when the O2 fraction increases from 0% to 10% O2. The target Ca fraction of 0.28 (indicated by the solid line in Fig. 2) is attained for about 2% oxygen fraction. In addition, the Ba/Nb2 ratio decreases from 0.89 at 0% O2 to 0.41 for 10% O2. The desired barium stoichiometry (indicated by the dashed line in Fig. 2) is attained for 5% to 6% oxygen in the chamber. Therefore, using a value of PO2 = 5% is a good compromise for obtaining Ca and Ba stoichiometry of 0.24 and 0.75, respectively. With such an oxygen fraction, we achieve an atomic ratio Ca / Ba of 0.32, which is close to the value of 0.39 expected for CBN-28. XPS analysis also reveals an excess of Nb for all the deposition conditions, regardless of the oxygen fraction in the chamber. The Nb content of the films is estimated to lie between 2.4 and 2.8, which is larger than the desired value of 2.
Figure 3 shows the XRD θ-2θ patterns of a CBN sample crystallized using RTA at 1000 °C for 1 min in O2 gas. The peak at 22.4° is for the (001) plan of CBN. For any oxygen fraction, we observe another peak at 21.8° that is attributed to the (001) reflection of a BaNbO3 (BN) phase. The presence of this parasitic phase is due to the departure from the targeted stoichiometry, because CBN only exists within a narrow composition window . A similar parasitic phase was found by Cuniot-Ponsard et al.  for the sputtering of SBN on MgO. However, it was observed to collapse for very high oxygen fraction. In our case, we did not see any intensity change of the (001) peak for BN in none of the deposition conditions studied. On the other hand, the intensity of the CBN (001) peak varies significantly with the oxygen fraction.
The highest intensity is observed for an oxygen fraction of 5% which also corresponds to the stoichiometry the closest to that of CBN-28 as shown in Fig. 2. The intensity ratio between the CBN (001) and BN (001) peaks is larger than 80, which confirms the highly preferential growth of CBN over BN. This demonstrates that the stoichiometry of the as-deposited films is required to achieve highly oriented CBN thin films after annealing.
Figure 4 shows the XRD θ-2θ patterns of the crystallized CBN thin films grown on MgO substrate as a function of the annealing temperature from 600 °C to 1150 °C using RTA for 1 min in O2 background atmosphere. Increasing the annealing temperature up to 1000 °C results in better crystallization, as demonstrated by the much higher diffracted intensity of the CBN (001) and (002) peaks.
By examining the angular position of the CBN (001) and (002) peaks, the out-of-plane lattice parameter c was calculated to be 3.968 ( ± 0.002), 3.961 ( ± 0.001) and 3.958 ( ± 0.001) Å for annealing temperatures of 800 °C, 1000 °C and 1150°C, respectively. It can be seen that the lattice parameter of the CBN thin films annealed above 1000 °C is similar to that of bulk CBN (3.958 Å). This indicates that the crystallized CBN thin films exhibit low strain as previously reported for CBN deposited on MgO by PLD 
Phi-scan analysis was also performed on the CBN / MgO as-deposited with 5% O2 fraction and further crystallized at 1000 °C. As shown in Fig. 5, the pattern for CBN (211) shows 16 peaks corresponding to the two antisymmetric orientations expected for CBN grown on MgO. The rotation of these antisymmetric domains was calculated to be 31 ° ( ± 0.5 °).
This in-plane orientation was further confirmed by the similar pattern observed for the (311) planes. This result is in very good agreement with the previously reported value of 30.5° obtained for epitaxial CBN deposited by pulsed laser deposition on MgO . No significant peak was found for BN, which indicates either no in-plane orientation of the BN crystallites or a very weak density of these crystallites into the film.
The detailed chemical composition of the annealed CBN thin films was determined by means of RBS. For this purpose, a He beam is impinging on the sample at 2.039 MeV. The detector is located at a scattering angle of 170° from the incident angle. The composition is deduced from calculations performed for a multilayer structure simulating the sample. As observed in Fig. 6, a good agreement between simulated and experimental data is found. The composition is found to be 3% Ca, 5% Ba, 22% Nb, 68% O. The RBS data indicate that the CBN layer is homogeneous except in the vicinity of the substrate surface where the film contains Nb, Mg and O. They also suggest some diffusion of Mg into the CBN film (between 5% and 2% over the first 200 nm).
In order to further study the composition of CBN, ERDA was also carried out on the same samples with Cu at 50 MeV, with both incident and detection angle at 75° from the normal to the substrate. The composition was found to be 3% Ca, 6% Ba, 24% Nb and 67% O. These values confirm that the CBN thin films are richer in Nb as compared to the target. This is likely to be due to the preferential sputtering of Nb over Ca and Ba and it could be compensated by using a non-stoichiometric target. Finally, ERDA does not show any amount of Mg on the sample surface, which suggests that Mg only diffuses in the vicinity of the film / substrate interface due to annealing.
3.2 Optical properties of the thin films
In order to examine the optical properties of the optimized thin films, the transmission spectrum was measured as shown in Fig. 7. First, the sputtered CBN thin film shows a large transmission (>75%) at wavelengths above 400 nm. Moreover, the cutoff wavelength was found to be 303 nm, which corresponds to a bandgap of 3.26 eV as calculated using the Tauc relation. This bandgap is similar to the one determined on PLD-grown CBN thin films , which confirms the good quality of the thin films grown by RF magnetron sputtering.
Figure 8 shows the TE guided mode spectrum in the CBN / MgO thin film as measured by a rutile prism at wavelengths λ = 630 and 1550 nm. Three TE modes were observed at 630 nm and only one at 1550 nm. The sharp and deep reflectivity dip for each guided mode indicates that the light is well confined within the film. Using the dispersion equation of the CBN / MgO structure  together with the angular position of the observed modes, one calculates that the ordinary refractive index of the material is 2.21 and 2.13 at λ = 630 nm and 1550 nm, respectively. In addition, we deduce that the film thickness is 0.61 µm as also confirmed by SEM measurements.
Using ellipsometric measurements at wavelengths between 400 nm and 1700 nm, we calculated the dispersion curve of the guided waves propagating in the thin films. This dispersion curve was derived from a standard Sellmeier equation  as:
In order to minimize the fitting error of the ellipsometric data, we used a data set obtained at 7 angles ranging from 45° to 75° together with a standard Sellmeier model. The fitted parameters are presented in Table 1. The parameters A, B and C are close to those measured for bulk CBN-28 as reported in . The calculated dispersion curve is presented in Fig. 9 and is in good agreement with the data obtained at 630, 960, 1310 and 1550 nm from prism coupling measurements. The refractive index is slightly below that obtained for epitaxial CBN grown by PLD  and for bulk CBN . This can be explained by the slightly different composition of the sputtered films as compared to the PLD films and to the bulk.
The surface morphology is an important factor to achieve good optical quality. Therefore, AFM measurements were carried out to investigate the surface roughness of the CBN films deposited on epi-polished MgO substrates. As shown in the inset of Fig. 9, the roughness of the crystallized thin films is as small as 2.8 nm rms which is over twice better than the reported 7 nm roughness of epitaxial CBN thin films deposited on MgO using PLD . No particles or droplets were found on the surface of the sample, which is a strong advantage of RF Sputtering over PLD. Moreover, SEM measurements show very good thickness homogeneity (better than 1%) over about 2 inches. The combination of both low surface roughness and very good thickness homogeneity is very promising to build low loss optical devices.
3.3 Non-linear optical behavior of the CBN thin films
The non-linear optical behavior of the optimized CBN thin film was studied using SHG microscopy. Images were recorded in both forward and backward directions. Using the parametric approximation (low pump depletion), one can derive the dependency of the SHG power as a function of the incident power :
As it can be noted, the second harmonic signal varies according to the square of the fundamental signal, which denotes a two-photon effect. Figure 10 shows the dependence of the 2ω beam power on the incident power. The power of both forward and backward second-harmonic images signals varies quadratically with a regression coefficient of 0.99. This confirms that the signal well follows Eq. (3) as expected for a second harmonic signal and does not contain any contributions from other processes such as fluorescence. The results are also similar to that reported in the literature for the second harmonic generation in CBN crystals . Finally, one notes that the thin film is unpoled and efficiently generates the second harmonic signal. This indicates that intrinsic polarization is very strong in CBN thin films. It could be improved further by poling the deposited thin film.
In conclusion, we have reported for the first time the successful deposition of highly c-textured CBN thin films on MgO using the RF sputtering technique. We demonstrated that a tight control of the thin film stoichiometry is required to achieve thin films of the highest quality. This control is possible by carefully choosing the O2 fraction in the chamber during deposition. We also noticed that the annealing temperature is crucial to obtain optimized (001) oriented thin films with a lattice parameter in the c-direction very close to that of the CBN bulk material. The composition of the thin films slightly differs from that of the target and could be accurately tuned by using a non-stoichiometric target. The optical properties of the thin film were found to be close to those of the bulk material with a refractive index slightly lower. The surface quality is significantly better than for PLD grown films with a roughness below 3 nm while the thickness homogeneity is very good. These results, together with the non-linear behavior of CBN thin films, are extremely promising for the large scale fabrication of high performance devices.
The authors would like to thank Martin Chicoine for RBS and ERDA measurements. They also acknowledge Stéphane Bancelin and François Légaré of the Laboratory for the Applications of Non-linear Optical Microscopy for Second Harmonic Generation measurements. This work was funded by the NSERC Strategic projects program.
References and links
1. W. R. Cook and E. Wolf, Landolt-Bornstein, New Series (Springer – Verlag, 1979).
2. P. Tayebati, D. Trivedi, and M. Tabat, “Pulsed laser deposition of SBN:75 thin films with electro-optic coefficient of 844 pm/V,” Appl. Phys. Lett. 69(8), 1023 (1996). [CrossRef]
3. K. K. Wong, Properties of Lithium Niobate (INSPEC, 2002).
4. C. David, T. Granzow, A. Tunyagi, M. Wöhlecke, T. Woike, K. Betzler, M. Ulex, M. Imlau, and R. Pankrath, “Composition dependence of the phase transition temperature in SrxBa1−xNb2O6,” Phys. Status Solidi201, R49–R52 (2004) (a). [CrossRef]
5. R. Helsten, L. Razzari, M. Ferrera, P. F. Ndione, M. Gaidi, C. Durand, M. Chaker, and R. Morandotti, “Pockels response in calcium barium niobate thin films,” Appl. Phys. Lett. 91(26), 261101 (2007). [CrossRef]
6. M. Eßer, M. Burianek, D. Klimm, and M. Mühlberg, “Single Crystal Growth of the Tetragonal Tungsten Bronze CaxBa1–xNb2O6 (x = 0.28; CBN-28),” J. Cryst. Growth 240(1-2), 1–5 (2002). [CrossRef]
7. C. S. Pandey, J. Schreuer, M. Burianek, and M. Mühlberg, “Relaxor behavior of CaxBa1−xNb2O6 (0.18≤x≤0.35) tuned by Ca/Ba ratio and investigated by resonant ultrasound spectroscopy,” Phys. Rev. B 87(9), 094101 (2013). [CrossRef]
8. K. Wieteska, W. Wierzchowski, A. Malinowska, M. Lefeld-Sosnowska, M. Swirkowicz, T. Lukasiewicz, and C. Paulmann, “Synchrotron diffraction topography of SrxBa1-xNb2O6 (SBN), CaxBa1-xNb2O6 (CBN) and mixed (Ca0.28Ba0.72)y(Sr0.61Ba0.39)1-yNb2O6 (CSBN) crystals,” Radiat. Phys. Chem. 93, 87–91 (2013). [CrossRef]
9. E. Mortazy, I. Stateikina, A. Tehranchi, S. Delprat, M. Chaker, and K. Wu, “Low-loss CaxBa1-xNb2O6 ridge waveguide for electro-optic devices,” Microelectron. Eng. 88(3), 218–221 (2011). [CrossRef]
10. P. F. Ndione, M. Ferrera, D. Duchesne, L. Razzari, M. Gaidi, M. Chaker, and R. Morandotti, “Hybrid integration of Ca 0.28 Ba 0.72 Nb2O6 thin film electro-optic waveguides with silica/silicon substrates,” Opt. Express 17(17), 15128–15133 (2009). [CrossRef] [PubMed]
11. C. L. Jia, J. Schubert, T. Heeg, S. B. Mi, H. Y. Chen, B. Joschko, M. Burianek, M. Mühlberg, and K. Urban, “Tailoring the orientations of complex niobate films on perovskite substrates,” Acta Mater. 54(9), 2383–2391 (2006). [CrossRef]
12. P. F. Ndione, M. Gaidi, C. Durand, M. Chaker, R. Morandotti, and G. Rioux, “Structural and optical properties of epitaxial CaxBa1-xNb2O6 thin films grown on MgO by pulsed laser deposition,” J. Appl. Phys. 103(3), 033510 (2008). [CrossRef]
13. S. B. Mi, C. L. Jia, K. Urban, T. Heeg, and J. Schubert, “Growth of CaxBa1−xNb2O6 thin films on MgO (1 0 0) by pulsed laser deposition,” J. Cryst. Growth 291(1), 243–248 (2006). [CrossRef]
14. R. Eason, Pulsed Laser Deposition of Thin Films (John Wiley & Sons Inc., 2007), Chap.9.
15. M. Cuniot-Ponsard, J. M. Desvignes, B. Ea-Kim, and E. Leroy, “Radio frequency magnetron sputtering deposition of hetero-epitaxial strontium barium niobate thin films (SrxBa1−xNb2O6),” J. Appl. Phys. 93(3), 1718 (2003). [CrossRef]
16. M. Cuniot-Ponsard and J. M. Desvignes, “Bellemain, Epitaxial growth of SrxBa1−xNb2O6 (SBN) thin films on Pt coated MgO substrates: the determining control of platinum crystallographic orientation,” J. Mater. Sci. 41, 5302–5309 (2006).
17. C. Wang, D. E. Laughlin, and M. H. Kryder, “Epitaxial growth of lead zirconium titanate thin films on Ag buffered Si substrates using RF sputtering,” Appl. Phys. Lett. 90(17), 172903 (2007). [CrossRef]
18. J. L. Vossen and W. Kern, Thin film processes II (Academic Press, 1991), Chap II-4.
19. K. Takechi and M. A. Lieberman, “Effect of Ar addition to an O2 plasma in an inductively coupled, traveling wave driven, large area plasma source: O2/Ar mixture plasma modeling and photoresist etching,” J. Appl. Phys. 90(7), 3205 (2001). [CrossRef]
20. D. Dorranian and M. Alizadeh, “Effect of negative oxygen ions on the characteristics of plasma in a cylindrical DC discharge,” J. Theor. Appl. Phys. 8(2), 122 (2014). [CrossRef]
21. M. Muehlberg, M. Burianek, B. Joschko, D. Klimm, A. Danilewsky, M. Gelissen, L. Bayarjargal, G. P. Görler, and B. O. Hildmann, “Phase equilibria, crystal growth and characterization of the novel ferroelectric tungsten bronzes CaxBa1-xNb2O6 (CBN) and CaxSryBa1-x-yNb2O6 (CSBN),” J. Cryst. Growth 310, 2288–2294 (2008). [CrossRef]
23. M. Eßer, M. Burianek, P. Held, J. Stade, S. Bulut, C. Wickleder, and M. Mühlberg, “Optical characterization and crystal structure of the novel bronze type CaxBa1-xNb2O6 (x=0.28; CBN-28),” Cryst. Res. Technol. 38(6), 457–464 (2003). [CrossRef]
24. S. A. Denev, T. T. A. Lummen, E. Barnes, A. Kumar, and V. Gopalan, “Probing ferroelectrics using optical second harmonic generation,” J. Am. Ceram. Soc. 94(9), 2699–2727 (2011). [CrossRef]
25. Y. Sheng, X. Chen, T. Lukasiewicz, M. Swirkowicz, K. Koynov, and W. Krolikowski, “Calcium barium niobate as a functional material for broadband optical frequency conversion,” Opt. Lett. 39(6), 1330–1332 (2014). [CrossRef] [PubMed]