Metallorganic chemical vapor deposition of metallic Ru thin films on biaxially textured Ni substrates using a Ru(EtCp)2 precursor

Materials Chemistry and Physics 93 (2005) 142–148

Metallorganic chemical vapor deposition of metallic Ru thin films on biaxially textured Ni substrates using a Ru(EtCp)2 precursor Hu-Yong Tian a,b,∗ , Helen-Lai-Wa Chan a , Chung-Loong Choy a , Jong-Wan Choi b , Kwang-Soo No b a b

Department of Applied Physics and Materials Research Center, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology, 373-1 Guseong-dong, Yuseong-Gu, Daejeon 305-701, South Korea Received 7 July 2004; accepted 6 March 2005

Abstract Ruthenium (Ru) films on rolling-assisted biaxially textured Ni substrates (RABiTs) were deposited by liquid source chemical vapor deposition using bis-(ethyl-␲-cyclopentadienyl)ruthenium (Ru(C2 H5 C5 H4 )2 ). The thermal decomposition process of the precursor was investigated by Fourier transform infrared spectroscopy (FTIR), mass spectroscopy, and differential scanning calorimetry/thermogravimetric analyses (DSC/TGA). The crystalline structure and resistivity of Ru thin films were investigated. The Ru films were polycrystalline and had a grainy structure. Although the thermal decomposition of the precursor required a sufficient amount of oxygen, the experimental results showed that up to a certain concentration of oxygen (i.e. O2 /Ar ∼ 30/10), Ru metal film was deposited without any detectable RuO2 impurities. A higher deposition temperature and a higher ratio of O2 /Ar will be beneficial to the growth of (0 0 2) orientation. They showed a low resistivity of about 10–20 ␮ cm, which is sufficiently low for them to be used as a buffer layer in superconductor tapes or electrode materials in dielectric capacitors. © 2005 Elsevier B.V. All rights reserved. Keywords: Metal films; Chemical vapor deposition; Auger spectrum; Resistivity

1. Introduction Ruthenium is one of the most promising candidate electrode materials for some dielectric capacitors [1–3]. Ru and RuO2 thin films have been receiving much attention because of their good conductivity, low temperature coefficient of resistance (TCR), and high thermal stability [4]. Therefore, they have been investigated for many applications such as diffusion barriers for oxygen [5,6], thin film resistors, and electrode materials for ferroelectric oxides [7,8]. Recently, there has been increasing interest in the deposition of metallic Ru films as well as in the study of RuO2 film because Ru film can be used as an adhesion layer for a diffusion barrier as well as an electrode for dynamic random access memory ∗

Corresponding author. Tel.: +852 27764616; fax: +852 23337629. E-mail address: [email protected] (H.-Y. Tian).

0254-0584/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.matchemphys.2005.03.002

(DRAM). Metallic Ru films on Si and SiO2 /Si substrates were prepared by metallorganic chemical vapor deposition (CVD) using a new precursor named (␩6 -benzene)(␩4 1,3-cyclohexadiene)ruthenium (Ru(C6 H6 )(C6 H8 )) in Ar atmosphere. The Ru films contained hydrogen that originated in the hydrogen atoms in the precursor and was involved in the CVD process due to the catalytic effect of ruthenium on hydrocarbon and hydrogen [9]. The thermal decomposition of the precursor, bis-(ethyl-␲-cyclopentadienyl)ruthenium (Ru(C2 H5 C5 H4 )2 ), required a sufficient amount of oxygen up to a certain concentration of oxygen. Ru films were deposited without any detectable RuO2 impurities [10]. The key issue of the liquid delivery technique depends on the stable vaporization of the source in the vaporizer. In order to solve problems such as decomposition and the instability of the supply, the precursor was dissolved into tetrahydrofuran (C4 H8 , THF) solvent [11].

H.-Y. Tian et al. / Materials Chemistry and Physics 93 (2005) 142–148

Recently, many studies have demonstrated that sharply biaxially textured nickel could be produced in long lengths by thermo-mechanical processing techniques [12–14]. To date, some researchers have focused their investigations on proper buffer layers, as it is well known that proper buffer layers are critical to formation of epitaxially oxide buffer layers on the rolling-assisted biaxially textured Ni substrates (RABiTs) surface that are mechanically robust and chemically inert with respect to both the substrate and the high temperature superconductor (HTS) films. It is important to utilize any of these buffer layers deposition techniques to prevent unfavorably oriented NiO on the Ni surface. Ru or RuO2 thin films as an electrode material, as mentioned above, have drawn attention because of their good conductivity as well as excellent barrier properties against oxygen diffusion [5,6,15]. However, few studies about pure metal ruthenium deposition have been reported so far, because the precursors involve oxygen atom or oxygen gas as a reaction to assist the deposition of the precursor. The purpose of the present work is to obtain information about the mechanism of the thermal decomposition of the precursor and to determine the optimum conditions for obtaining pure metal Ru films without NiO formation by oxidizing the substrate in the presence of oxygen. In this work, we investigated the microstucture and relationship between processing conditions and properties for the Ru thin film layer as a promising buffer layer for HTS applications and electrode materials in memory devices.

2. Experimental procedure In this experiment, bis-(ethyl-␲-cyclopentadienyl)ruthenium was used as the precursor, which is yellow liquid at room temperature. Tetrahydrofuran was used as the solvent to dissolve the precursor. The molar concentration of the solution was maintained at 0.1 M. The decomposition temperatures were determined by thermal analyses, such as differential scanning calorimetry (DSC) (Mac. Science DSC-3110) and a thermogravimetric analysis (TGA), combined with mass spectroscopy and Fourier transform infrared spectroscopy (FTIR) measurements. Biaxially textured Ni substrates were obtained from randomly oriented high-purity (99.9%) Ni sheets, which were first mechanically deformed by rolling and then were made into Ni tapes of 80 ␮m in thickness and 10 mm in width. Ni tapes were annealed in a vacuum chamber at 800 ◦ C, the mixed gases of H2 (10%) and Ar (90%) were flown into the chamber and the base pressure was kept to around 100 mTorr. The velocity of the Ni tapes was kept to about 1 cm min−1 . The MOCVD apparatus consisted of a vertical warm-wall reactor and a resistive substrate heater. The experimental conditions are summarized in Table 1. All of the lines to the reactor were heated to 120 ◦ C to prevent the vaporized precursor from condensing in transit. After repeated deposition runs, the lines and vaporizers were free of residues or deposits, which ensured excellent

143

Table 1 Deposition conditions of Ru thin films on Ni substrate by MOCVD Precursor Bubbler temperature (◦ C) Reaction gas O2 flow rate (sccm) Carrier gas Ar flow rate (sccm) Substrate Line temperature (◦ C) Deposition temperature (◦ C) Deposition pressure (Torr) Deposition time (min)

Ru(EtCp)2 120 0–40 10–100 Ni(2 0 0) 120 300–500 0.02–0.2 30

repeatability of the depositions. A film phase analysis and resistivity measurements were performed by X-ray diffraction (XRD) using a Rigaku D/Max-B diffractometer with Cu K␣ radiation and a four-point probe, respectively. The thickness of the film and surface morphology were observed by cross-sectional scanning electron microscopy (SEM) and atomic force microscopy (AFM). Auger electron spectroscopy (AES) was also used to measure the relative atomic concentration to confirm the variations in composition and surface electronic state, using a Perkin-Elmer phi 400 scanning Auger microprobe. 3. Results and discussion Fig. 1 shows the TGA results analyzed under O2 and Ar atmosphere. The precursor began to lose weight near 80 ◦ C regardless of the atmosphere, and the precursor decomposed at 200 ◦ C under O2 atmosphere. This also demonstrates that the weight loss, at 6.3%, results from the endothermic processing between 100 and 200 ◦ C. No more weight loss occurred above 200 ◦ C. This shows the Ru source should be completely decomposed. There are no obvious changes for the Ru source under different atmospheres, even after it has been stored for 3 months. The thermal analysis with time was carried out using differential scanning calorimetry in 200 sccm oxygen or argon. The Ru source was heated up to 300 ◦ C and cooled down to room temperature at a rate of

Fig. 1. TGA curves of the precursor under oxygen and argon. The inset shows the weight loss (%) as temperatures increase.

144

H.-Y. Tian et al. / Materials Chemistry and Physics 93 (2005) 142–148

Fig. 2. The normalized peak intensity of the mass spectroscopy of the Ru(EtCp)2 precursor.

Fig. 3. Typical IR spectra of the Ru(EtCp)2 precursor and assignment of the IR peaks.

±10 ◦ C min−1 . The DSC data of the as-received Ru source shows only an endothermic peak at 170 ◦ C upon heating, but no peak during the cooling; whereas the endothermic peak decreased slightly near 168 ◦ C after having been stored for 3 months in a sealed box. Fig. 2 shows the mass spectroscopy of Ru(EtCp)2 excluding minor peaks. The observed mass and expected molecules are listed in Table 2. All minor peaks have been excluded and the intensities normalized. Even though the precursor was stored for 3 months, the major weight that was observed was almost same, although more EtCp ligand was observed. Molecule-bonded EtCp ligand, ethylene adduct, and Ru were mainly observed. We might deduce the major decomposition mechanism of Ru(EtCp)2 . The major decomposition is not the decomposition of EtCp ligand itself but the dissociation of EtCp ligand and ethylene adduct from molecules. But mass spectroscopy data alone cannot explain the decomposition mechanism because the mass spectroscopy data was at high vacuum and the precursor was not decomposed by thermal energy. Additional analyses combined with other methods, i.e. Fourier transform, are required to understand the fragmentation of the molecular species in the ion source for the structural analysis and identification of complex molecules. Fig. 3 shows the IR absorption spectra of the Ru source. The temperature of the vaporizer was maintained at 90 ◦ C, and Ar or O2 were, respectively, used as the reaction gases inputted into the chamber. The chamber was pumped around 10−3 Torr and then maintained at 10−1 Torr with reactive gas.

In order to measure the temperature dependence of the normalized absorbance of the IR peaks with high intensity, the measuring chamber was kept between 100 and 400 ◦ C. The IR absorption bands at 2800–3200 cm−1 are due to C H stretching vibrations. There are three main absorption bands between 800 and 1600 cm−1 . They are due to the ␦(C H) in Cp ligand (1018 cm−1 ) and in ethylene (1226 cm−1 ). The peaks around 851 cm−1 are attributed to ␦(C C C) vibrations in Cp ligand, those around 902 cm−1 to ␦(C C C) vibrations in Cp ligand and ethylene, and those around 807 cm−1 ␲(C H) to ␦(C C C) vibrations in Cp ligand. Some peaks from symmetric and asymmetric bending CH3 vibration bands will be presented around 1371 and 1393 cm−1 , respectively. The 464 cm−1 peak shows ring tilt vibration between Ru and EtCp ligand. The changes in the normalized intensity of the IR peaks in a vapor state were investigated under an Ar and oxygen atmosphere. There is a tendency for the main peak intensity decrease as temperatures increase, whereas there is a little disparity between the Ar and oxygen atmosphere. It seems that the strength of the bond between Ru and EtCp ligand becomes weaker under an oxygen atmosphere. The intensity of these main peaks will change as the temperature increases. This means that these vibrations will become weaker as temperatures increase. Fig. 4 shows the variations in the normalized intensity of the IR peaks under an oxygen atmosphere. The intensity of the IR peaks will decrease sharply as the increase in temperature approaches 120 ◦ C; less change will be found above 120 ◦ C. This demonstrates that the C C and C H bonds have broken up, i.e. that EtCp ligand and organic groups in ethylene in the precursor were removed. On the other hand, the peaks around 851 and 902 cm−1 attributed to ␦(C C C) vibrations in Cp ligand and ethylene exhibited a greater decrease until 200 ◦ C. This means that much higher temperatures are needed to complete the decomposition reactions of C O bonds. We also checked the results under argon ambient, and found a similar tendency for the normalized intensity of the IR peaks to decrease (not shown here).

Table 2 Mass spectroscopy analysis of Ru(EtCp)2 excluding minor peaks Temperature (◦ C)

Ligand

95 167 193 258 273 288

[HCpCH2 CH3 ]+ [RuCp]2+ [RuCpCH2 CH]2+ [Ru(Cp)2 (CH2 )2 ]2+ [Ru(Cp)2 (CH2 CH3 )2 CH3 ]2+ [Ru(Cp)2 (CH2 CH3 )2 ]2+

H.-Y. Tian et al. / Materials Chemistry and Physics 93 (2005) 142–148

Fig. 4. Changes in the normalized intensity of the IR peaks as a function of temperatures under an oxygen atmosphere.

Fig. 5 shows the XRD patterns of the Ni substrates; it indicates a strong c-axis orientation of the Ni substrate. The strong peak observed at 2θ of 51.86◦ is indexed to be Ni(2 0 0). The textured Ni orientation was characterized by a XRD pole figure analysis. The (1 1 1) pole figure of Ni(2 0 0) before deposition is shown in the inset of Fig. 5, which shows the four-fold symmetry in it. The preferred orientation is nearly perfect, and the piercings will be grouped together in tighter and tighter bands as the texture becomes stronger. A typical (2 0 0) pole figure exhibits a c-axis orientation and good alignment in-plane of the textured Ni substrate. The FWHM of the ␻-scanning curve on the (2 0 0) peak reflection of Ni is 6.37◦ ; it also shows good out-of-plane alignment. Fig. 6 shows that crystalline Ru thin films can be deposited at temperatures as low as 250 ◦ C with a 30 sccm oxygen flow rate. The bubbler and transporting line temperature is 120 ◦ C. The orientation of the Ru thin films changed from the (1 0 1) plane to the (0 0 2) plane as the substrate temperature increased. The c-axis orientation degree α increased from 31.5 to 43.1% as the substrate temperature increased

Fig. 5. The XRD pattern of the biaxially textured Ni(2 0 0) substrate. The inset shows the pole figure of the Ni(1 1 1) direction.

145

Fig. 6. The XRD patterns of Ru films deposited at different temperatures with a 30 sccm oxygen flow rate (argon at a fixed value of 10 sccm): (a) 250 ◦ C; (b) 300 ◦ C; (c) 400 ◦ C. Peaks labeled “S” are attributed to the satellite peaks of the Ni(2 0 0) peak and consistently appear in the θ–2θ scans of the bare Ni substrate.

from 250 to 400 ◦ C. This is defined by the formula [16]: ␣ = I(0 0 2) /(I(1 0 0) + I(0 0 2) + I(1 0 1) ), where I(1 0 0) , I(0 0 2) , and I(1 0 1) represent the X-ray diffraction intensities of the (1 0 0), (0 0 2), and (1 0 1) reflections, respectively. The higher the α value, the better the crystalline quality of the Ru films. Fig. 7 shows the changes as a function of oxygen content as the oxygen flow rate is increased from 10 to 30 sccm. The substrate temperature is 400 ◦ C and the bubbler and transporting line temperature is 120 ◦ C, and the oxygen flow rates are: (a) 10 sccm; (b) 20 sccm; (c) 30 sccm. The relative intensity of the Ru thin films will increase as the contents of O2 are increased in the chamber. Moreover, the relative intensity of Ru(0 0 2) and Ru(1 0 1) will change as the oxygen flow rate is increased. It seems useful to form the Ru(0 0 2) thin films at a high ratio of O2 /(O2 + Ar). The c-axis orientation

Fig. 7. The XRD patterns of Ru films deposited at 400 ◦ C (argon at a fixed value of 10 sccm) under different oxygen flow rates: (a) 10 sccm; (b) 20 sccm; (c) 30 sccm. Peaks labeled “S” are attributed to the satellite peaks of the Ni(200) peak and consistently appear in the θ–2θ scans of the bare Ni substrate.

146

H.-Y. Tian et al. / Materials Chemistry and Physics 93 (2005) 142–148

degree α increases from 33% (a), to 38% (b) and 43.4% (c) in Fig. 7. From the observations, it may be concluded that an increase in the oxygen flow rate will benefit the texture of the Ru thin films on the textured Ni(2 0 0) substrate. The orientation of the Ru thin films changed from the (1 0 1) plane to the (0 0 2) plane, which has the lowest surface energy due to the packing in the hcp structure. This trend was similar to the change in orientation when the films were deposited at a higher temperature. The Ru atoms with a lower deposition rate were less arranged into a more stable state, inducing less tensile stress [17]. The average roughness of the AFM line scans over a 5 ␮m × 5 ␮m area of the Ru thin films deposited at 400 ◦ C decreases when the O2 contents change. The average roughness values of these samples were 11.5 nm for O2 /(O2 + Ar) = 20/30, 11.3 nm for O2 /(O2 + Ar) = 30/40, and 7.4 nm for O2 /(O2 + Ar) = 40/50. The results are not consistent with those of Hong et al., who found that RuO2 formed, leading to an abrupt increase in surface roughness [18]. Many researchers have reported that the surface of Ru is oxidized under exposure to O2 atmosphere and that Ru bottom electrodes at a high temperature deposition are oxidized by oxygen diffusion [8,19]. This shows that insufficient oxygen favored the formation of metallic Ru, not Ru-oxides, as the decomposition rate of the precursor was rapid compared with the oxidation rate of the Ru products. Fig. 8 shows the SEM images of Ru thin films at 400 ◦ C under different atmospheres (or the ratio of O2 /(O2 + Ar)): (a) O2 /(O2 + Ar) = 0.33; (b) O2 /(O2 + Ar) = 0.5; (c) O2 /

(O2 + Ar) = 0.67; and (d) O2 /(O2 + Ar) = 0.75. Obviously, the average grain size is larger for the films deposited under a higher ratio of O2 /(O2 + Ar) at the same depositing temperature. The average grain size of the Ru films is determined from the full width at half the maximum of the X-ray diffraction (0 0 2) or (1 0 0) peak by Scherrier’s equation [20]. It is about 80 nm for Ru film deposited at 400 ◦ C and O2 /(O2 + Ar) = 0.75 (Fig. 8(d)), but only 30 nm for Ru film when the gas flow rate ratio of O2 /(O2 + Ar) = 0.33 (Fig. 8(a)); whereas, there are some larger grains and aggregation as the deposition temperatures increase. The ratio of O2 /(O2 + Ar) has less of an effect on the average grain size for Ru thin films deposited at higher temperatures. The Auger depth profile for Ru thin films including the heterostructure between Ru and Ni deposited at 400 ◦ C under different atmospheres is shown in Fig. 9. The bubbler and transporting line temperature is 120 ◦ C. (a) O2 /(O2 + Ar) = 0.5; (b) O2 /(O2 + Ar) = 0.67; (c) O2 /(O2 + Ar) = 0.75. The resulting plot is the concentration of Ru, Ni, and O as a function of sputtering depth. It has an abrupt variation in Ru and Ni atomic concentration in the interface, making it obvious that Ru and Ni elements are slightly interdiffused in the interface. The amount of oxygen is very low along the etching depth, as shown in Fig. 9(a) and (b). A small amount oxygen diffusion to the substrate is observed in Fig. 9(c), which indicates the existence of oxides during the deposition. While RuO2 and NiO were not observed by XRD (shown in Figs. 6 and 7), the formation of oxides cannot be ruled out. Detailed studies of the interface reactions are required.

Fig. 8. SEM images of Ru films at different atmospheres (or the ratio of O2 /(O2 + Ar)): (a) O2 /(O2 + Ar) = 0.33; (b) O2 /(O2 + Ar) = 0.5; (c) O2 /(O2 + Ar) = 0.67; (d) O2 /(O2 + Ar) = 0.75.

H.-Y. Tian et al. / Materials Chemistry and Physics 93 (2005) 142–148

147

Fig. 10. The typical AES of the surface of Ru film. The inset shows all Ru AES peaks.

present in the inset of Fig. 10. The main peaks at 36, 153, 172, 183, 203, 234, and 275 eV with Ep = 5 keV corresponding to pure Ru metal peaks at around 37, 150, 176, 184, 200, 231, and 273 eV with Ep = 3 keV. The existence of the oxygen impurities at the surface of the as-grown Ru film might be due to the adsorption of oxygen on the Ru surface in air. On the other hand, we could not find evidence for the formation of RuO2 on the Ru surface, although RuO2 was thermodynamically more stable than the Ru under deposition [21]. Fig. 11 shows the resistivity of Ru films versus the grain size. The resistivity will decrease from 20 to 10 ␮ cm as the size of the grain increases from 30 to 80 nm. It is expected that oxygen inputted in the reaction chamber will benefit the growth of the grains, which will decrease the amount of the grain boundary, leading to low resistivity. It is obvious that the grain boundary greatly affects the resistivity, although the mechanism for this is not well known at present. The same result was found for the Ru films deposited on the Si substrate by the CVD technique in our previous work [9].

Fig. 9. Auger depth profile of Ru films at different atmospheres (or the ratio of O2 /(O2 + Ar)): (a) O2 /(O2 + Ar) = 0.5; (b) O2 /(O2 + Ar) = 0.67; (c) O2 /(O2 + Ar) = 0.75.

The typical surface composition of Ru thin films fabricated by the CVD technique is shown in Fig. 10. The sample shown in this figure was prepared at 400 ◦ C with the bubbler and transporting line temperature at 120 ◦ C, and the oxygen flow rate at 10 sccm (the same as Fig. 9(a)). The as-deposited films consist of Ru and oxygen at the surface of the sample; moreover, there are no distinct peaks belonging to Ni in the surface Auger electron spectra. All peaks belonging to Ru are

Fig. 11. The resistivity of Ru films as a function of grain size.

148

H.-Y. Tian et al. / Materials Chemistry and Physics 93 (2005) 142–148

4. Conclusions The bis-(ethyl-␲-cyclopentadienyl)ruthenium compound was found to be a suitable precursor for the deposition of metallic Ru thin film. The ruthenium precursor begins to evaporate at 80 ◦ C, and finishes the decomposition reaction before 200 ◦ C, regardless of atmosphere (Ar or O2 ). The pure Ru thin films can be deposited at 400 ◦ C with oxygen as a reactive gas by the CVD technique. These Ru films showed a dense and smooth microstructure. It would be of benefit to form high-quality Ru thin film when the oxygen content increases to up to 80% of O2 /(O2 + Ar) in the atmosphere without any ruthenium oxides, while a greater amount of oxygen will diffuse into the substrate, leading to the formation of oxides. The resistivity of metallic Ru films is about 10– 20 ␮ cm, which is near bulk value. In our work, grain size strongly affected the electrical properties of Ru films. These results show that Ru thin film is a promising buffer layer for HTS applications and electrode materials in memory devices. Acknowledgements This research was supported by the BK21 project in the Korea Advanced Institute of Science and Technology (KAIST), South Korea, and by the postdoctoral fellowship program of The Hong Kong Polytechnic University. References [1] T. Aoyama, S. Yamazaki, K. Imai, J. Electrochem. Soc. 145 (1998) 2961. [2] T. Aoyama, M. Kiyotoshi, S. Yamazaki, K. Eguchi, Jpn. J. Appl. Phys. 38 (1999) 2194. [3] J. Lin, N. Masaaki, A. Tsukune, M. Yamada, Appl. Phys. Lett. 74 (1999) 2370.

[4] Y.T. Kim, Appl. Phys. Lett. 70 (1997) 209. [5] L. Krusin-Elbaum, M. Wittmer, D.S. Yee, Appl. Phys. Lett. 50 (1987) 1879. [6] K. Yoshikawa, T. Kimura, H. Noshiro, S. Otani, M. Yamada, Y. Furumura, Jpn. J. Appl. Phys. 33 (1994) 867. [7] L.A. Bursill, I.M. Reaney, D.P. Vijay, S.B. Desu, J. Appl. Phys. 75 (1994) 1521. [8] T. Kawahara, M. Yamamuka, A. Yuuki, K. Ono, Jpn. J. Appl. Phys. 35 (1996) 4880. [9] J.W. Choi, Y.M. Choi, J.I. Hong, H.Y. Tian, J.-S. Roh, Y.S. Kim, T.-M. Chung, Y.W. Oh, Y.S. Kim, C.G. Kim, K.S. No, Jpn. J. Appl. Phys. 41 (2002) 6852. [10] S.Y. Kang, K.H. Choi, S.K. Lee, C.S. Hwang, H.J. Kim, J. Electrochem. Soc. 147 (2000) 1161. [11] T. Nabatame, M. Hiratani, M. Kadoshima, Y. Shimamoto, Y. Matsui, Y. Ohji, I. Asano, T. Fujiwara, Y. Suzuki, Jpn. J. Appl. Phys. 39 (2000) L1188. [12] Q. He, D.K. Christen, J.D. Budai, E.D. Specht, D.F. Lee, A. Goyal, D.P. Norton, M. Paranthaman, F.A. List, D.M. Kroeger, Physica C 275 (1997) 155. [13] A. Goyal, D.P. Norton, J.D. Budai, M. Paranthaman, E.D. Specht, D.M. Kroeger, D.K. Christen, Q. He, B. Saffian, F.A. List, D.F. Lee, P.M. Martin, C.E. Klabunde, E. Hartfield, V.K. Sikka, Appl. Phys. Lett. 69 (1996) 1795. [14] M. Paranthaman, D.F. Lee, A. Goyal, E.D. Specht, P.M. Martin, X. Cui, J.E. Mathis, R. Feenstra, D.K. Christen, D.M. Kroeger, Supercond. Sci. Technol. 12 (1999) 319. [15] M.L. Green, M.E. Gross, L.E. Papa, K.J. Schnoes, D. Brasen, J. Electrochem. Soc. 132 (1985) 2677. [16] K. Iijima, R. Takayama, Y. Tomita, I. Ueda, J. Appl. Phys. 60 (1986) 2914. [17] H.J. Lim, S.Y. Kang, C.S. Hwang, H.J. Kim, J. Vac. Sci. Technol. A 21 (2003) 1381. [18] S.M. Hong, H.J. Bak, I. An, O.K. Kim, Jpn. J. Appl. Phys. 39 (2000) 1796. [19] W.J. Lee, C. Basceri, S.K. Streiffer, A.I. Kingon, D.Y. Yang, Y. Park, H.G. Kim, Thin Solid Films 323 (1998) 285. [20] B.E. Warren, X-ray Diffraction, Addison-Wesley, Reading, 1969, p. 251. [21] R.S. Roth, T. Negas, L.P. Cook, in: G. Smith (Ed.), Phase Diagram for Ceramists, vol. 4, American Ceramics Society, Columbus, 1981, p. 9.