Effect of sintering temperature on composition, microstructure and electrical properties of K0.5Na0.5NbO3 ceramics

Effect of sintering temperature on composition, microstructure and electrical properties of K0.5Na0.5NbO3 ceramics

Physica B 434 (2014) 139–144 Contents lists available at ScienceDirect Physica B journal homepage: www.elsevier.com/locate/physb Effect of sinterin...

2MB Sizes 19 Downloads 161 Views

Physica B 434 (2014) 139–144

Contents lists available at ScienceDirect

Physica B journal homepage: www.elsevier.com/locate/physb

Effect of sintering temperature on composition, microstructure and electrical properties of K0.5Na0.5NbO3 ceramics Rajan Singh a,n, A.R. Kulkarni b, C.S. Harendranath a a b

Centre for Research in Nanotechnology & Science, Indian Institute of Technology Bombay, Mumbai 400076, India Department of Metallurgical Engineering & Materials Science, Indian Institute of Technology Bombay, Mumbai 400076, India

art ic l e i nf o

a b s t r a c t

Article history: Received 17 July 2013 Received in revised form 26 October 2013 Accepted 28 October 2013 Available online 7 November 2013

Lead free potassium sodium niobate (K0.5Na0.5NbO3) ceramic powders were synthesized by colloidal coating method. The calcined powders (800 1C) were sintered conventionally at three different temperatures (1050 1C, 1100 1C and 1150 1C) and the effect of sintering temperature on density, microstructure, composition and electrical properties was investigated. All the samples showed a single phase perovskite structure with orthorhombic symmetry similar to KNbO3 ceramics. Microstructure examined under FEG-SEM revealed an optimum microstructure, in terms of grain size, porosity and uniformity, at the sintering temperature of 1100 1C, which also showed density of 92% of ρTh. As the sintering temperature increased the X-ray diffraction peaks shifted to lower 2θ values indicating excess volatilization of Na at higher temperature as compared to K. This was further confirmed through elemental Probe X-ray microanalysis and ICP–AES studies. Dielectric constant (εr), dielectric loss (tanδ), ferroelectric (P–E loop) and piezoelectric (d33) properties showed considerable improvement and leakage current decreased with increasing sintering temperature. The sample sintered at 1100 1C showed marked improvement in maximum dielectric constant (573) at RT at 1 kHz, minimum tangent loss (0.04) at RT at 1 kHz, maximum remnant polarization (13.5 μC/cm2), lower leakage current (7.6  10  7 A/cm2) and maximum d33 value (100 pC/N). & 2013 Elsevier B.V. All rights reserved.

Keywords: Electroceramic Dielectric:ferroelectric:piezoelectric

1. Introduction Lead zirconium titanate (PZT) and modified PZT-based ceramics are widely used for many industrial applications due to their superior performance in ferroelectric, piezoelectric, and pyroelectric applications [1,2]. It is well known and widely accepted that, use of lead-based ceramics raises serious environmental and safety concerns, due to the toxicity of lead oxide and its high vapour pressure during the sintering process [3,4]. Consequently, there has been growing interest in developing alternative lead-free ferroelectric and piezoelectric materials which could eventually replace the current lead-based ceramics. Extensive research all over the world in the last two decades has resulted in several lead free candidate materials such as BaTiO3-based ceramics [5], Bi layered structures [4], alkaline niobate perovskites [4] and Bi-based perovskites [6], to name a few. Alkaline niobate perovskites, in general, potassium sodium niobate (K, Na) NbO3 (KNN) in particular has been widely studied. This is a potential candidate due to its high Curie temperature (420 1C) and promising ferroelectric and piezoelectric properties comparable with PZTs [7]. Further, KNN exhibits a morphotropic phase boundary (MPB) around 50% K and 50% Na, separating two orthorhombic phases and, as in PZT, an abrupt increase in piezoelectric coefficients

n

Corresponding author. Tel.: þ 91 22 25767636; fax: þ 91 22 25723480. E-mail addresses: [email protected], [email protected] (R. Singh).

0921-4526/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.physb.2013.10.052

near MPB [7,8]. Reports on successful processing of KNN ceramics have been with sintering using hot processing [8], spark plasma sintering (SPS) [9] and the use of sintering aids [3,10,11]. But these unconventional sintering methods are considered commercially rather nonviable. The difficulty and lack of success, in obtaining dense KNN ceramics with good electrical properties through conventional sintering has been attributed to volatilization of sodium (Na) during sintering and consequent deviation in MPB above 1100 1C as evidenced by XRD results [9]. Below 1100 1C there is no deviation from MPB, however densification is poor. Recently, we reported improvement in density, dielectric and ferroelectric properties of KNN obtained with conventional sintering of powders processed using colloidal coating method [12,13]. In this paper, we report correlation between density, microstructure, leakage current and piezoelectric properties as a function of sintering temperature for KNN ceramic powders synthesized by colloidal coating method and sintered by conventional method. Further, additional evidence for degradation of properties and deviation in MPB above 1100 1C are also examined. 2. Materials and methods The starting materials used in this study were KNO3, NaNO3 (all 99% pure, Merck India) and Nb2O5 powder (99.9%, Aldrich USA). The raw materials were dried at 200 1C for 1 h prior to use. Single

140

R. Singh et al. / Physica B 434 (2014) 139–144

phase KNN powders were prepared by colloidal coating method as earlier reported [13,14]. The phase pure powders (obtained by calcination at 800 1C) were compacted and sintered at 1050 1C, 1100 1C and 1150 1C for 2 h in air. The experimental density of the sintered samples was determined using the Archimedes method (ASTM# C 373-88). To determine the phase purity and lattice parameters of calcined powders and ceramic samples, X-ray diffraction data was recorded in the 2θ scan range of 101–901 by X-ray diffractometer (PANalytical X-ray diffractometer). X-ray source was Cu-Kα (λ ¼1.541 Å) radiation. Microstructure of sintered and polished samples was observed using a JEOL (JSM-7600F) Field Emission Scanning Electron Microscope (FEG-SEM). For composition analysis EDX was performed on the samples with the help of an EDX detector (Oxford instruments) attached with FEG-SEM. Bulk composition analysis was carried out using inductively coupled plasma–atomic emission spectroscopy (ICP–AES) using ICP–AES Spectrometer (ARCOS, Germany). For ICP–AES measurement 0.1 g of sample was dissolved in a mixture of 8 ml of aqua regia and 2 ml of HF (hydrofluoric acid) in a microwave digester at 220 1C. The sample was diluted with Di-water to 25 ml volume. For electrical measurements of samples, silver paste was applied on both sides of the sample and dried at 100 1C for 1 h to remove organic solvent. The dielectric measurements of the samples were carried out over a temperature range from 50 1C to 500 1C by using a computer interfaced NovoControl dielectric Alfa analyzer in the frequency range 1 Hz to 1 MHz. Variation of electrical polarization as a function of electric field (P–E loop) at room temperature at 2 Hz was observed using ferroelectric test system (aixACT TF analyzer, Aachen, Germany). The mechanical displacement was measured using mechanical displacement sensor (SIOS). During the polarization measurement, an electric field of 10–50 kV/cm, based on the coercive fields of the samples, was applied. To prevent the breakdown from the edges of the sample, samples were immersed in silicone oil during the measurement. The I–V characteristics were measured in “switched triangular mode” with a maximum applied electric field of 20 kV/cm.

3.2. Phase analysis Fig. 2 shows the room-temperature XRD patterns of KNN ceramic calcined at 800 1C for 6 h. It can be observed that the XRD patterns reveal a single phase perovskite structure. The diffraction peaks closely match with JCPDS card (PDF#71-0946) and all the peaks were indexed to orthorhombic perovskite unit cell. The peaks were fitted using pseudo-Voigt function in order to determine their angular positions and integral widths. The room-temperature XRD patterns of KNN ceramics sintered at 1050 1C, 1100 1C and 1150 1C for 2 h are depicted in Fig. 3. It can be seen that the diffraction peaks have shifted to lower 2θ values as the sintering temperature increases from 1050 1C to 1150 1C. The shifting of peaks to lower 2θ values suggests that the stoichiometric ratio of K and Na has deviated from the nominal stoichiometry of KNN. Since the ionic radius of K þ (rK ¼1.38 Å) is larger than Na þ (rNa ¼1.02 Å), the decrease in 2θ value or increase in d value means the K/Na ratio increases with increasing sintering temperature. This result suggests that the volatilization of Na is higher than that of K at higher temperatures. This is consistent with the reported results by Zhang et al. [9] in which K/Na ratio was tailored and found that as the K content increases the value of 2θ decreases (d value increasing) for KNN ceramics. 3.3. Density measurements

3. Results and discussion

It is well known that density and porosity have pronounced effect on properties of most electro-ceramics. The density and porosity of the samples sintered at different temperatures are listed in Table 1. It may be noted that the sample sintered at 1050 1C showed density of 3.78 g/cm3 which corresponds to 84% of the theoretical density (ρTh ¼4.51 g/cm3) [15]. The observed density for 1100 1C and 1150 1C sintered samples were 4.15 g/cm3 (92% of ρTh) and 4.25 g/cm3 (94% of ρTh), respectively. This shows that there is no appreciable difference in the density of samples sintered at 1100 1C and 1150 1C for 2 h. The improvement in density of 1100 1C and 1150 1C sintered samples over 1050 1C sintered sample can be attributed to an increase in grain size and presence of fairly uniform microstructure as described in the next section.

3.1. Raw powders morphology

3.4. Microstructure of sintered KNN ceramics

The morphology of pure Nb2O5 and precalcined (after coating) powders are shown in Fig. 1. The particle size distribution has been measured using Imge.J software and represented in inset of Fig. 1. The mean particle size of pure Nb2O5 powders is about  0.230 μm and after coating the mean particle size increased to about 0.240 μm. It can also be seen that the Nb2O5 particles are uniformly distributed and the Na and K precursors are homogeneously coated.

The effect of sintering temperature on the microstructure was examined on polished and thermally etched surface using Field Emission Scanning Electron Microscopy (FEG-SEM). Fig. 4(a)–(c) shows microstructure of pure KNN ceramics sintered in air at 1050 1C, 1100 1C and 1150 1C for 2 h, respectively. All the samples showed well developed grains and grain size increased with increasing temperature. The sample sintered at 1050 1C, having lower density, shows apparent porosity in the microstructure. The

Fig. 1. FEG-SEM Morphology of (a) Pure Nb2O5 and (b) Precursors after coating (precalcined).

R. Singh et al. / Physica B 434 (2014) 139–144

141

and is compared with nominal composition of KNN as listed in Table 2. It can be seen that atom % of Na decreased and atom % of K increased with the increasing sintering temperature. Similar results were also obtained with point analysis (not included in Table 2).These results confirm that the volatilization of Na is higher than that of K at higher sintering temperatures. Hence, the composition of the sample sintered at 1150 1C shifted closer to pure KNbO3.

Fig. 2. X-ray diffraction pattern of KNN powder calcined at 800 1C for 6 h.

Fig. 3. X-ray diffraction patterns of KNN ceramics sintered at different temperatures. The XRD pattern of calcined KNN powder (800 1C) is also shown for comparison.

Table 1 Variation of density, porosity and grain size with different sintering temperature for KNN synthesized by coating method. Sample Density (gm/cm3)

% of Theoretical density (ρTh ¼ 4.51 gm/cm3)

Porosity in % (Open poresþclose pores) (100 - % ρTh)

Grain size (lm)

1050 1C 3.78 1100 1C 4.15 1150 1C 4.25

83.81 92.02 94.23

16.19 7.98 5.77

0.3–0.5 0.8–1.1 1–1.5

density of this sample is determined to be 3.78 g/cm3 which corresponds to 84% of the theoretical density (4.51 g/cm3) [15]. Further, as the sintering temperature increases grain growth increases and porosity decreases. From Fig. 4(b) and (c) it can be observed that the sample sintered at 1100 1C and 1150 1C has less porosity and higher density. The grain size for the sample sintered at 1050 1C is less than 1 mm and increased with increasing sintering temperature to  1.5 mm at 1150 1C. 3.5. Composition analysis 3.5.1. Energy dispersive X-ray analysis (EDX) The shifting of XRD peaks to lower 2θ values, with increasing sintering temperature, was attributed to increased evaporation of sodium at higher temperatures. To support this, EDX analysis has been performed. Both aerial as well as point analysis was carried out. The atom % of K, Na, Nb and O is calculated for samples sintered at three different temperatures using EDX (aerial analysis)

3.5.2. Inductively coupled plasma–atomic emission spectroscopy (ICP–AES) As discussed in the earlier section the EDX studies showed that with increasing sintering temperature the stoichiometry of KNN shifted towards pure KNbO3. As EDX in FEG-SEM provides local composition, bulk composition analysis using inductively coupled plasma–atomic emission spectrometer (ICP–AES) was carried out to further confirm these results. The variation in atom % of K, Na, Nb and O using ICP–AES was obtained for the same specimens are listed in Table 2. These results show a similar trend to those obtained by EDX. The atom % of Na decreases and the atom % of K increases with the increasing sintering temperature. Moreover, it is known that the Nb does not easily volatilise during the sintering process of KNN. However, there is sufficient evidence in literature that the K/Na ratio increases with sintering temperature, indicating enhanced loss of Na as compared to K with increasing sintering temperature [16]. Our results concur with these earlier reported results. The small increase in K/Nb ratio is due both to a small increase in K to balance the loss of Na in the equation as also to a small deviation in Nb. All necessary standardizations and corrections were carried out during EDX and ICP–AES (ZAT etc.) measurements. 3.6. Electrical properties Fig. 5(a) and (b) show the temperature dependence of dielectric constant and tangent loss of pure KNN ceramics at 1 kHz from 50 to 500 1C for samples sintered at three temperatures 1050 1C, 1100 1C and 1150 1C. All the samples have sharp phase transition at Curie temperature, indicating normal ferroelectric behaviour. The samples showed two peaks; one near 190 1C and another near 390 1C which correspond to the orthorhombic–tetragonal phase transition (TT–O) and tetragonal cubic phase transition (Curie temperature-Tc), respectively. It can be seen that both the transition temperatures shifted to lower temperature for sample sintered at 1050 1C, viz. 175 1C and 380 1C, respectively. However there is no peak shift for samples sintered at 1100 1C and 1150 1C. It can also be noted that as the sintering temperature increases dielectric constant increases and reaches a maximum for sample sintered at 1100 1C and then starts decreasing. The increase in dielectric constant with increasing sintering temperature can be attributed to increase in density. In case of sample sintered at 1150 1C the dielectric constant decreases in spite of higher density due to dominant effect of severe volatilization of Na and resulting change in composition. In Fig. 5(b) it can be seen that the tangent loss is higher for sample sintered at 1050 1C as it has poor density. The tangent loss decreases as the sintering temperature increases and reaches a minimum for sample sintered at 1100 1C. Further increase in sintering temperature causes more volatilization of Na and the tangent loss increases. For all the samples the tangent loss increases with increase in temperature and after reaching peaks at Tc they increase rapidly due to conductivity losses [17]. The sample sintered at 1100 1C has maximum dielectric constant (573) and minimum tangent loss (0.04) at RT at 1 kHz. These values are consistent with the reported literature [7,8]. A comparison of εr,

142

R. Singh et al. / Physica B 434 (2014) 139–144

Fig. 4. FEG-SEM micrographs of polished and thermally etched KNN ceramics sintered at (a) 1050 1C (b) 1100 1C and (c) 1150 1C. Table 2 Elemental composition in KNN ceramics at different sintering temperatures in comparison to nominal composition. Element ↓

Sample -

Nominal composition (Atom%)

Experimental composition (Atom%) EDX arial analysis

Na K Nb O

ICP–AES bulk analysis

1050 1C

1100 1C

1150 1C

1050 1C

1100 1C

1150 1C

9.92 10.88 20.43 58.77

8.90 12.08 21.20 57.82

8.00 14.54 18.90 58.56

9.81 11.38 21.12 57.69

8.80 12.50 21.20 57.50

7.90 14.61 19.10 58.39

tanδ, Pr and d33 values obtained at 1 kHz of the air sintered and hot pressed of KNN [8], PZT [18] and BNT [19] is tabulated in Table 3. It may be noted that both εr and d33 values for the process adopted here are significantly higher than ceramics sintered by conventional route. Table 4 summarizes the porosity corrected dielectric constant (ε′) of pure KNN ceramics sintered at different temperatures. Porosity correction for the measured dielectric constant was carried out using Rushman and Strivens equation [20]:

εcorrected ¼ ½εobserved  ð2 þ V 2 Þ=½2ð1  V 2 Þ

ð1Þ

where V2 is the volume fraction of porosity in the sintered compact. The density, ε′max values both observed and corrected are listed in Table 4. 3.7. Ferroelectric properties Polarization versus electric field hysteresis (P–E) loop of pure KNN ceramics sintered at 1050 1C, 1100 1C and 1150 1C are shown in Fig. 6. The sample sintered at 1050 1C has higher leakage current because of poor density of about 84%. The samples sintered at 1100 1C and 1150 1C have saturated P–E loop and the loop is not leaky as these samples have higher density as compared to the sample sintered at 1050 1C. However, the sample sintered at

10 10 20 60

1150 1C has less remnant polarization (Pr) as compared to 1100 1C sintered sample due to volatilization of Na, whereas there is no change in coercive field (Ec). The observed remnant polarization (Pr) is 13.5 μC/cm2 and 11 μC/cm2 corresponding to sintering temperatures of 1100 1C and 1150 1C, respectively. 3.8. I–V characteristics In order to assess the leakage current behaviour of the ceramics, I–V characteristic of the KNN ceramics sintered at 1050 1C, 1100 1C and 1150 1C were obtained and are illustrated in Fig. 7. The sample sintered at 1050 1C has leakage current of the order of 10  5 A/cm2. The leakage current decreases with the increase in sintering temperature (reduced porosity). It can be seen that the sample sintered at 1100 1C and 1150 1C showed excellent I–V properties. For 20 kV/cm applied filed a leakage current of about 10  7 A/cm2 was observed. This is nearly  1.5 order of magnitude lower than the sample sintered at 1050 1C. The lowest value was observed for sample sintered at 1100 1C and the highest leakage current was for sample sintered at 1050 1C. Lowest leakage current (7.6  10  7 A/cm2) was observed for 1100 1C sample and the highest (1.1  10  5 A/cm2) was for 1050 1C sample. The highest leakage current in 1050 1C sample can be attributed to higher porosity.

R. Singh et al. / Physica B 434 (2014) 139–144

143

Fig. 6. P–E loop of KNN ceramics sintered at different temperatures.

Fig. 5. Temperature dependence of (a) dielectric constant and (b) tangent loss measured at 1 kHz for KNN ceramics sintered at different temperatures. In (b) inset shows temperature variation of tanδ at 1 kHz between 50 and 150 1C.

Table 3 Comparison of dielectric, ferroelectric and piezoelectric properties of KNN, BNT and PZT ceramics. Property

Air sintered KNN

Hot pressed KNN

Air sintered BNT

PZT-4

This study (1100 1C) sample

% TD εr (1 kHz) Tanδ (1 kHz) d33 (pC/N) Pr (μC/cm2)

94.24 290 –

98.89 420 0.035

– 310 0.013

– 1400 0.05

92.02 573 0.04

80 12.6

160 33

66 –

225 –

100 13.5

Table 4 Density, ε′max values both observed and corrected for pure KNN ceramics sintered at different temperatures. S. no.

Samples

Density (g/cm3)

εObserved (1 kHz)

TPorosity (%)

εCorrected (1 kHz)

1 2 3

1050 1100 1150

3.78 4.15 4.25

2725 4507 3636

16.19 7.98 5.77

3514 5093 3970

3.9. Strain behaviour The bipolar electric field-induced-strain (S–E) curves for KNN ceramics measured at room temperature are illustrated in Fig. 8(a). All the samples show a butterfly-shaped S–E curve that is typically seen in ferroelectric materials. Maximum bipolar strain (Smax) value (0.0838%) was obtained for the sample sintered at 1100 1C.

Fig. 7. Variation of Leakage current with applied electric field (I–V) for KNN ceramics sintered at different temperatures.

This is because of decrease in porosity. Further increase in temperature increases the volatilization of Na. Due to this the stiochiometry is not maintained leading to considerable drop in Smax. The unipolar electric field-induced strain curves of the KNN ceramics sintered at 1050 1C, 1100 1C and 1150 1C are shown in Fig. 8(b). The ratio of Smax/Emax defines as dn33 values are also illustrated in Fig. 7(b). The Smax and Emax represent the maximum strain and electric field, respectively. The lowest dn33 value (160 pm/V) is obtained for 1050 1C sintered sample. This is due to lower density where as the highest dn33 value (216 pm/V) is obtained for 1100 1C sintered sample. The increase in dn33 value is because of increase in density and decrease in porosity. Further, 1150 1C sintered sample has dn33 value 186 pm/V, the decrease in dn33 value is due to excess volatilization of Na at higher temperatures. As indicated in Fig. 7(b) all the samples showed some hysteresis, which is probably a consequence of domain reorientation [21].

4. Summary Conventional sintering of KNN ceramics, synthesized by colloidal coating method, was carried out at different temperatures. Density, microstructure, leakage current and piezoelectric properties, as a function of sintering temperature, was investigated. Optimum density (about 92% of ρTh) was obtained at the sintering temperature of 1100 1C. This corresponds to an optimum microstructure, in terms of grain size, porosity and uniformity, as

144

R. Singh et al. / Physica B 434 (2014) 139–144

lower leakage current (7.6  10  7 A/cm2) and maximum d33 value (100 pC/N). These results suggest that 1100 1C is the optimum temperature for KNN ceramics sintered conventionally in this study. All the ceramics, sintered at different temperatures, showed an orthorhombic phase similar to KNbO3. However, with increasing sintering temperature, the X-ray diffraction peaks shifted to lower 2θ values. X-ray microprobe analysis in an FEG-SEM and ICP–AES studies confirm the excess volatilization of sodium (Na) at higher sintering temperature leading to shifts in XRD peaks and variations in electrical properties.

References

Fig. 8. (a) Bipolar strain versus Electric field and (b) Unipolar strain versus Electric field for KNN ceramics sintered at different temperatures.

evidenced by FEG-SEM studies. Further, samples sintered at 1100 1C showed maximum dielectric constant (573), minimum tangent loss (0.04), maximum remnant polarization (13.5 mC/cm2),

[1] N. Setter, D. Damjanovic, L. Eng, G. Fox, S. Gevorgian, J. Appl. Phys. 100 (2006) 051606. [2] K. Uchino, Ferroelectric Devices, Marcel Dekker, Inc., New York, 2000. [3] E. Mensur, M. Papila, Ceram. Int. 36 (2010) 1921. [4] M.D. Maeder, D. Damjanovic, N. Setter, J. Electroceram. 13 (2004) 385. [5] Z. Yu, C. Ang, R. Guo, A.S. Bhalla, J. Appl. Phys. 92 (2002) 1489. [6] Y. Hiruma, R. Aoyagi, H. Nagata, T. Takenaka, Jpn. J. Appl. Phys. 44 (2005) 5040. [7] L. Egerton, D.M. Dillom, J. Am. Ceram. Soc. 42 (1959) 438. [8] R.E. Jaeger, L. Egerton, J. Am. Ceram. Soc. 45 (1962) 209. [9] B.P. Zhang, J.F. Li, K. Wang, H. Zhang, J. Am. Ceram. Soc. 89 (2006) 1605. [10] H.E. Mgbemere, R.P. Herber, G.A. Schneider, J. Eur. Ceram. Soc. 29 (2009) 1729. [11] F.R. Marcos, J.J. Romero, M.G.N. Rojero, J.F. Fernandez, J. Eur. Ceram. Soc. 29 (2009) 3045. [12] R. Singh, A.R. Kulkarni, C.S. Harendranath, AIP Conf. Proc. 1512 (2013) 514. [13] R. Singh, K. Kambale, A.R. Kulkarni, C.S. Harendranath, J. Mater. Chem. Phys. 138 (2013) 905. [14] H. Gu, W.Y. Shih, W.H. Shih, J. Am. Ceram. Soc. 86 (2) (2003) 217. [15] M. Matsubara, T. Yamaguchi, K. Kikuta, S. Hirano, Jpn. J. Appl. Phys. 43 (2004) 7159. [16] Y. Zhen, J.F. Li, J. Am. Ceram. Soc. 89 (12) (2006) 3669. [17] H. Yang, Y. Lin, F. Wang, H. Luo, Mater. Manuf. Processes 23 (2008) 489. [18] M.W. Hooker, NASA/CR-1998–208708. [19] X. Wang, H. Chan, C. Choy, J. Am. Ceram. Soc. 86 (10) (2003) 1809. [20] D.F. Rushman, M.A. Strivens, Proc. Phys. Soc 59 (1947) 1011. [21] E.M. Sabolsky, A.R. James, S. Kwon, S.T. Mckinstry, G.L. Messing, Appl. Phys. Lett. 78 (2001) 2551.