Volume resistivity of studied materials (P – plasma sprayed, S – sintered)
Synthetic perovskite CaTiO3 (CT), geikielite MgTiO3 (MT) and their mixture MgTiO3-CaTiO3 (MCT) are materials well known and widely used as dielectrics in a sintered state. CT is high-permittivity linear dielectric material whose structure is not influenced by plasma spraying - neither chemical nor phase composition, as demonstrated earlier [Ctibor, 2003]. MT belongs to the family of low-loss microwave dielectrics and MT-CT solution is known by its temperature stability of permittivity. Namely the composition (Mg0.95Ca0.05)TiO3 is used in connection with this feature.
In recent decades plasma spraying has become a well accepted technology as the coating method for metallic and ceramic materials and has been used in a variety of fields including electrical engineering. The coatings have lamellar character of a body formed with porosity aligned with respect to the lamellas. Spraying does not require heating of the substrate to the melting point of the deposited material.
Often the research is focused on the microstructure of plasma sprayed coatings and features like porosity, unmelted particles, cracks and residual stress. Above listed characteristics are responsible for the behavior of coatings. However in the case of titanates phenomena taking place on atomic level and single crystal cell level are also important. Raman spectroscopy, infrared spectroscopy and near-field microwave microscopy are suitable techniques for this characterization.
Ca and Mg have the same charge but different ionic radii (
MCT exhibits differences from MT in the metal-oxygen bond lengths which are relevant to the stability of the compounds. While the infrared and Raman spectra of CT and MT have been measured [Jiang, 1998; Cavalcante, 2008], other cations at the
Besides the mentioned dielectric titanate materials we have sprayed also one representative of ferroelectrics - barium titanate. BaTiO3 (BT) is an interesting multifunctional oxide that exhibits complex phase appearance. Between 120°C (393K) and 1457°C (1730K) BaTiO3 has a cubic perovskite structure that consists of corner linked oxygen octahedra containing Ti4+, with Ba2+. Cooling below 120°C results in small displacements in the positions of the cations in the unit cell resulting in polar ferroelectric phase existing in the temperature interval between 5°C (278K) and 120°C [Boutinaud, 2006].
BaTiO3, due to its high dielectric constant, is used frequently as multilayer capacitor components and sensors. However, it has been found that with respect to the electrical properties BaTiO3 in the form of thin-films does not reach the qualities of bulk material. This difference was explained by a combination of the intrinsic dead layer effect, a stress effect, an effect of the microstructure within the thin film, and an effect of the stoichiometry
[Zhao, 2008]. In particular, the relative permittivity of films decreases when the film thickness is reduced [Setter, 2000]. The optimal dielectric characteristics are obtained for sintered BaTiO3-based sample with bulk density of about 5300 kg.m-3 [Jin, 2003]. Any deviation from the stoichiometric Ba/Ti ratio leads to suppression of the high relative permittivity of the ferroelectric barium titanate [Mitic, 2001]. To detect the tetragonal BaTiO3 phase by X-ray diffraction, the split of peaks of (002) and (200) reflection is a well-established indication [Waser, 1999; Yu, 2009; Simon-Seveyrat, 2007].
In general there are differences of the behavior of barium titanate in the form of a single-crystal, sintered bulk material and thin film [Boutinaud, 2006; Mitic, 2001]. Plasma spraying enables to create layers with ‘bulk-like’ thickness but adhering on a metallic substrate of various shapes. Free-standing parts of titanate ceramics can be fabricated as well by plasma spraying [Wu, 2009]. BaTiO3 itself was up to now very seldom plasma sprayed and the understanding of its behavior in the form of sprayed coating is not satisfactory. For coatings with the thickness of about 100 µm the values of relative permittivity 50 and loss factor 0.08 were reported [Dent, 2001]. The dielectric properties of the plasma sprayed BaTiO3 were related to the degree of crystallinity [Dent, 2001]. The coatings containing more crystalline material have higher relative permittivity. The relative permittivity was affected also by cracks and splat interfaces within the coating [Dent, 2001]. The reported value of relative permittivity is however surprisingly low, because one and more orders higher values are typical for bulk BaTiO3 [Buchanan, 2004].
In frame of the presented chapter we are focused on selected aspects of the dielectric characteristics of the as-sprayed barium titanate coatings and we provide comparison of them with other plasma sprayed titanates.
2.1. Feedstock materials
All materials were obtained in the form of tablets of industrial purity, produced by the sintering of micropowders. The sintering was carried out by companies Epsilon (Librice, Czech Republic), Ceramic Capacitors (Hradec Kralove, Czech Rep.) and Teceram (Hradec Kralove, Czech Rep.).
The synthetic form of perovskite CaTiO3 was produced by reactive sintering of CaO and TiO2. CaTiO3 powder used for experiments was sintered without any additives (like ZnO), normally used for decreasing the sintering temperature. Tablets were crushed and sieved into a powder of the correct size for spraying. MgTiO3 and MCT were sintered using MgO, CaCO3 and TiO2. After sieving the size distribution of the feedstocks was 63–125 microns for all three materials.
BaTiO3 feedstock powder was obtained by crushing and sieving of sintered coarse agglomerates. Those agglomerates were prepared by a reactive sintering of micrometer-sized powders of BaCO3 and TiO2 used as starting materials. After sieving the size distribution of the BT feedstock for spraying was between 20 and 63 μm with an average at 40 µm, whereas the bulk density measured by helium pycnometry was 5721 kg.m-3.
2.2. Plasma spraying
The CT, MT and MCT samples were manufactured using a high throughput Water-Stabilized Plasma (WSP) spray system WSP500® at Institute of Plasma Physics (Prague, Czech Republic) at ambient atmosphere. The WSP system operates at about 160 kW arc power and can process high amounts of material. This system can be used to fabricate deposits similar but not identical to those prepared by means of conventional atmospheric plasma-spray systems based on gas-stabilized torches. As substrates flat carbon steel coupons (Euronorm S355) were used whereas the powder was fed in by compressed air through two injectors. Just before spraying, the steel was grit blasted with Al2O3 with a mean diameter of 650 µm. The deposited thickness was about 1.5 mm for self-supporting deposits. Thick deposits were stripped from the substrate by a releasing agent or by thermal cycling between +200 and -70°C.
For manufacturing of BT samples a Gas-Stabilized Plasma gun (GPS) was used to perform Atmospheric Plasma Spraying (APS) process. The conventional d.c. GPS gun F4 consisted of a thoriated tungsten cathode of 10 mm in diameter with a conical tip and a copper anode/nozzle. The plasma gas mixture used was argon/hydrogen with the total flow rate 60 slm. The powder was injected perpendicularly to the plasma jet axis with argon as a carrier gas (at constant flow rate 5 slm at pressure 0.3 MPa for all spray experiments) through an injector located 3 mm downstream (called external injection) of the torch nozzle exit. The system can process 1 to 5 kg/hour of a ceramic powder. Barium titanate was sprayed at arc power around 30 kW. Spray distance was 100 mm and plasma spraying deposition time about five minutes to reach the thickness 0.9 to 1 mm. Substrates, rectangular shaped (120x20 mm²) 3 mm thick, were made of carbon steel (Euronorm S355). Just before spraying, they were grit blasted with Al2O3 with a mean diameter of 400 µm. The substrates were disposed on a rotating sample holder which diameter was 90 mm. This substrate holder was rotated (tangential speed of 1 m/s) with a horizontal axis and simultaneously translated back and forth orthogonally to the plasma jet axis at a velocity of 24 mm/sec, with an excursion of 160 mm, the plasma torch being stationary.
2.3. Characterization techniques
X-ray diffraction (XRD) was performed as a phase identification with SIEMENS D5000™ equipment
Raman spectra were collected by the FORAM 685 apparatus which is equipped by a 685 nm laser with output power up to 40 mW. Spectral resolution 8 cm-1 and various integration times from 10 to 60 seconds were used. Raman spectroscopy of BT was performed using a Lambda Solutions P1 apparatus – laser wavelength 785 nm, objective 50 x, integration time 25 s.The surface of the coating was polished before the test.
Infrared reflective spectra of CT samples were measured by Bruker IFS 113v Fourier transform spectrometer.
Microwave microscope Agilent AFM 5400 was used for simultaneous monitoring of the surface profile by AFM and dielectric response on external field applied at resonant frequency (approx. 2.6 MHz). This relatively new technique - near-field scanning microwave microscopy (SMM) permits characterization of the effects of inhomogeneities and defects in crystals, films, and compacts on the local dielectric behavior. By moving the tip/cavity assembly over a surface, one can map the microwave cavity resonant frequency
The complex dielectric permittivity of CT was studied in the 440 Hz – 1 MHz frequency range and 10–270K temperature interval using HP 4192A impedance analyzer with a Leybold He-flow cryostat (operating range 5–300 K). The impedance of the cylindrical sample with Au electrodes sputtered on the cylinder ends was recorded on cooling rate of 2 K/min.
All other electric measurements were performed at room temperature. The deposits were stripped of from the substrates. The surfaces were ground after spraying to eliminate surface roughness. A thin layer of aluminum as the electrode was sputtered in a reduced pressure on the ground surface [Ctibor, 2003]. A three-electrode system was used with a guarded electrode, whereas an unguarded electrode was sputtered on the entire surface of the sample opposite side. The electric field was applied parallel with the spraying direction (i.e., perpendicular to the substrate surface). Capacity was measured in the frequency range from 120 Hz to 1 MHz using the impedance analyzer 4284A (Agilent, CA, USA). Applied voltage was 1V AC whereas the stabilized electric source was equipped with a micrometric capacitor type 16451A (Agilent, CA, USA). Relative permittivity εr was calculated from measured capacities CP and specimen dimensions (Eq. 1)
where ε0 = 8.854×10-14 F cm-1; 1/k [cm] is defined as the ratio between the guarded surface and the thickness of the sample [Morey, 2003].
This same arrangement and equipment was used for the loss factor measurement at the same frequencies as capacity.
Electric resistance was measured with a special resistivity adapter – Keithley model 6105. The electric field was applied from a regulated high-voltage source and the values read by a multi-purpose electrometer (617ºC, Keithley Instruments, USA). The magnitude of the applied voltage was 100±2V DC. Volume resistivity was calculated from the measured resistance and specimen dimensions. Typically 4 - 5 specimens were measured and the average calculated.
3.1. Spectroscopic measurements
Figure 1 shows the Raman spectra of plasma sprayed CT in comparison with the sintered sample. SD means stand-off distance, in millimeters, of the as-sprayed samples and “an” denotes annealed samples. Annealing details are given elsewhere [Ctibor, 2003].
The absolute values of the intensity of the reflection are associated with surface roughness and could be omitted. From the graph we see that the positions of all peaks are identical for all samples. Wavenumbers of all three main peaks observed at 471, 495, and 640 cm-1 are in agreement with [Cavalcante, 2008]. The bands at 471 and 495 cm-1 are assigned to Ti–O torsional (bending or internal vibration of oxygen cage) modes [Hirata, 1996; Zheng, 2003]. The Ti–O stretching mode is centered at 640 cm−1 [Boutinaud, 2006]. Two small peaks at about 650 cm-1, suggesting the simultaneous presence of [TiO6] and [TiO5] clusters, however in our case they are even less pronounced compare to [Cavalcante, 2008]. Also in our plasma sprayed samples incomplete organization of the CaTiO3 lattice can be attributed to the defects in the covalent bond due to the oxygen vacancies (
The Raman spectrum of MT is displayed on the Fig. 2 in which the active modes at 478 cm-1 and 501 cm-1 [Hirata, 1996] are observed only on the coating. Next active modes are at 641 and 712 cm-1 [Hirata, 1996] and also were detected only on the coating, whereas in the sintered sample a shoulder centered at about 530 cm-1 is present.
The Raman spectrum of plasma sprayed MCT is displayed on the Fig. 3 in comparison with MT measured at exactly the same conditions. We can see for MCT a red shift of the peak centered at 712 cm-1 in the case of MT. Other pronounced maximum is at 565 cm-1. The Raman pattern of MCT is different compare to MT and CT and moreover it is not a simple combination of both of them. This is because of MgTi2O5 and Mg2TiO4 origin during the spraying, as confirmed by XRD [Ctibor, 2003]. However only two phases with different relative permittivity were detected by scanning microwave microscopy, see Fig. 4, similarly as in [Zhang, 2006]. Permittivities of MgTi2O5, Mg2TiO4 and MT are very similar together and very different from CT. Figure 4 is a superposition of the AFM contact mode image (roughness) and scanning microwave microscopy capacitance mode image (colors). The image is artificially colored – blue and red zones represent different relative permittivity.
CT (perovskite) is a high permittivity material
Raman spectrum of BT coating is displayed on Fig. 6. The spectrum with peaks at 311 and 507 cm-1 corresponds to tetragonal phase of BaTiO3 [Souza, 2006]. At low oxygen pressure which is the condition relevant for plasma spraying, the density of oxygen vacancies is higher, and then the expansion of the lattice volume is greater [Souza, 2006]. This is why the Raman modes shift to lower frequencies: 507 cm-1 in our case instead of 518 cm-1  or 532 cm-1 in [Souza, 2006] for the A1 torsion mode. Raman spectrum of the long-SD coating is practically identical.
The modes further split into longitudinal (LO) and transverse (TO) components. The spectrum in Fig. 6 shows the stretching mode of A1(TO1), A1(TO2) and A1(TO3) at around 163, 259 and 507 cm-1, respectively [Mattsson, 2010; Guo, 2005]. The stretching mode of E (TO2) appeared at 311 cm-1, while A1(LO1) stretching modes at 188 cm-1 and A1(LO2) at about 470 cm-1 [Mattsson, 2010], however the last one was not very pronounced in our case. By the Raman spectroscopy presence of TiO2 in anatase form was mentioned - the peak at about 645 cm-1 was observed in TiO2 film [Giolli, 2007] or coating [Buralcov, 2007]. Elsewhere [Ostapchuk, 2005] such a peak was shown without comments. In our case also week peak at 631 cm-1 was detected, which can correspond to anatase-TiO2 individual phase in the BaTiO3 coating.
Figure 7 shows the infrared spectra of plasma sprayed CT in comparison with the sintered sample. The infrared-active mode at 575 cm-1, assigned to the Ti–O stretch, and also the mode at 455 cm-1 , assigned to the Ti–O3 torsion, are present in both samples with the same intensity. The slightly more pronounced local valley in the case of plasma sprayed sample – localized at about 680 cm-1 is the only subtle distinction.
According X-ray diffraction measurement of CT coating, the phase composition is the same as in the feedstock powder – pure CaTiO3 (PDF2 card No. 00-022-0153). The XRD pattern of MT, Fig. 8, shows that the original metatitanate partly decomposed during the spray process on Mg2TiO4 and MgTi2O5. Al these components are present also in MCT coating, Fig. 9, whereas also CaTiO3 is present as an individual phase.
The XRD pattern of BT, Fig. 10, corresponds strictly to the tetragonal phase (PDF2 card No. 01-081-2204), which was confirmed to be the constituent of the feedstock powder as well as of the coatings. Intensity ratio of individual peaks of the feedstock as well as of the coatings is very similar to patterns reported in the literature [Wang, 2008]. In the coating, certain quantity of amorphous phase is present, manifested itself by a halo centered on 28° 2θ in the pattern, and quantified as 10 %.
3.2. Dielectric measurements
Dielectric properties of plasma sprayed CaTiO3 between 10 K and room temperature are displayed on the Figures 11 to 13. The incipient ferroelectricity of CT [Lemanov, 1999; Sudheendran, 2008] is suppressed in the case of plasma sprayed coatings, see Fig. 11. Rather relaxor type of behavior takes place. Relaxor features are believed to be connected with a situation when more than one type of ions occupies equivalent crystallographic position. One of the most important features of relaxors is their low frequency dispersion at maximum dielectric permittivity. The other features of this include dependence of phase transition temperature on the method of its determination. The origin of such properties is probably connected with a presence of microscopic polar regions (nano-domains) in these materials [Bak, 2009]. With temperature falling down also the
In the studied frequency range, the relative permittivity of all specimens exhibits significant ionic relaxation process [Dervos, 2004]. Strong relaxation effects is shown by the
Table 1 shows the volume resistivity and Fig. 16 corresponding conductivity of plasma sprayed and sintered BT, CT, MT and MCT materials. We see the different values with typically several orders of magnitude lower resistivity values for plasma sprayed specimen compare to sintered materials.
|Sample||BT (P)||BT (S)||CT (P)||CT (S)||MT (P)||MT (S)||MCT (P)|
3.3. Scanning microwave microscopy
Scanning microwave microscopy images of BT are given on Figs. 17 and 18. AFM images represent surface roughness. Since the surface was materialographically polished, the subtle differences in the roughness are associated with hardness of various phases. These phases are concentrated predominantly in individual lamellas, cf. MCT SEM micrograph in the Fig. 5. In the case of BT, the difference between lamellas is mainly in terms of Ba/Ti ratio [ Ctibor 2, 2010 ]. Such a feature is, as expectable, more easily visible on cross sections, Fig. 19, then on in-plane (i.e. spray direction) sections Fig. 18. The lamellas, red and yellow on capacitance mode images, are more Ti-rich (higher permittivity) whereas blue areas are Ba-rich (lower permittivity). The Ba-rich phase has slightly lower hardness (hardness measured by us: BaTiO3 is about 6.5 GPa versus TiO2 about 11.0 GPa; BaO not reported as a layer). The Ba-rich phase is darker on AFM contact mode image, since it is more easily worn-out by polishing process. These “steps” on border of lamellas are about 80 nm high, see also Fig. 20, which height is consistent with observations made by SEM.
BaTiO3, CaTiO3, MgTiO3 and a mixture of the last two materials with 5 wt.% of CaTiO3 in MgTiO3 were plasma sprayed by two complementary spray techniques. Raman and infrared spectroscopic measurements were carried out on plasma sprayed titanates. Coating samples were tested as mono-block capacitors. Volume dc-resistivities of the materials were summarized. Microwave microscopy was used for mapping of the dielectric response of selected samples.
CaTiO3 exhibits anomalous dielectric losses because of intrinsic conductivity. Relaxor type of behavior takes place at dielectric measurements, whereas dc-conductivity is responsible for the high dielectric losses. All coatings are slightly oxygen deficient – however under detection limits of XRD materials – the color of all of them is darker compare to sintered samples.
Polarization of grain boundaries combined with n-type conduction with oxygen vacancy as the dominant defect contributes to the observed pseudo-relaxor response of BaTiO3 and CaTiO3 to AC electric field. MgTiO3 as a low-loss material has the above mentioned effects slightly suppressed and is more similar to sintered bulk. The existence of irregularities in the crystal lattice was confirmed by spectroscopic measurements. Plasma spraying of applicable ceramic dielectrics able to substitute bulk parts will need further research. In the other hand, plasma spraying, with its possibility to cover wide range of substrate materials with thick coatings having very versatile thicknesses and areas, offers promising way to future development.
The authors thank to J. Dubsky (IPP ASCR) for XRD measurements, to M. Savinov (Inst. of Physics, ASCR) for the temperature measurement of dielectric properties of CaTiO3 and to V. Zelezny (Inst. of Physics, ASCR) for doing the infrared spectroscopy. Scanning microwave microscopy was done by M. Fenner (Agilent, Germany). The company ELMES, Lt.D. is acknowledged for lending of the Foram device. BaTiO3 was sprayed at University of Limoges, France.
This work was supported by IPP ASCR under project AV0Z20430508.