Plasma Chemistry and Plasma Processing, Vol. 25, No. 4, August 2005 (© 2005) DOI: 10.1007/s11090-004-3130-y

Effect of Ion Energy on Structure and Composition of Cathodic Arc Deposited Alumina Thin Films Johanna Ros´en,1,3 Stanislav Mra´ z,1 Ulrich Kreissig,2 Denis Music,1 and Jochen M. Schneider1 Received October 10, 2004; revised November 5, 2004

The effect of energy supplied to the growing alumina film on the composition and structure has been investigated by varying substrate temperature and substrate bias potential. The constitution and composition were studied by X-ray diffraction and elastic recoil detection analysis, respectively. Increasing the substrate bias potential from −50 to −100 V caused the amorphous or weakly crystalline films to evolve into stoichiometric, crystalline films with a mixture of the α- and γ -phase above 700 o C, and γ -phase dominated films at temperatures as low as 200 o C. All films had a grain size of <10 nm. The combined constitution and grain size data is consistent with previous work stating that γ -alumina is thermodynamically stable at grain sizes <12 nm [McHale et al., Science 277, 788 (1997)]. In order to correlate phase formation with synthesis conditions, the plasma chemistry and ion energy distributions were measured at synthesis conditions. These results indicate that for a substrate bias potential of −50 V, ion energies in excess of 100 eV are attained, both from a high energy tail and the accelerated ions with charge >1. These results are of importance for an increased understanding of the evolution of film composition and microstructure, also providing a pathway to γ -alumina growth at temperatures as low as 200 oC. KEY WORDS: Alumina; ion energy; plasma chemistry; composition; microstructure.

1. INTRODUCTION Alumina coatings exhibit advantageous properties suitable for applications in wear and corrosion protection(1,2) and as diffusion barriers.(3) The thermodynamically stable phase is α-alumina, and is as such often favoured in high temperature or high load applications. However, also the metastable κ- and γ -alumina have proven to be a suitable alternatives 1 Materials

Chemistry, RWTH Aachen, D-52056 Aachen, Germany. of Ion Beam Physics and Materials Research, Research Center Rossendorf, P.O. Box 510119, D-01314 Dresden, Germany. 3 To whom correspondence should be addressed. E-mail: [email protected] 2 Institute

303 0272-4324/05/0800-0303/0 © 2005 Springer Science+Business Media, Inc.

Ros´en et al.

304

for wear resistant coatings(4) as well as in metal cutting applications,(4,5) respectively. Various deposition techniques including chemical vapour deposition (CVD) and physical vapour deposition (PVD) have been used to synthesise alumina thin films: Typically, α-alumina is obtained by CVD at substrate temperatures above 1000 o C.(6) The high thermal load then restricts the use of substrate materials, since for example tempering of tool steels occurs already at approximately 550 o C.(7) Therefore, much attention has been paid to PVD techniques, which have successively resulted in a significant decrease in α-phase formation temperatures; 690 o C (pulsed magnetron sputtering),(8,9) 500−650 o C (filtered cathodic arc),(10,11) 580 o C (plasma assisted CVD),(12) and 280 o C (rf sputtering).(13) Correspondingly, the formation of γ -alumina has been reported at deposition temperatures of ∼ 800 o C by CVD,(14) 500 − 700 o C by plasma assisted CVD,(15) and 350−500 o C by different PVD techniques.(16−18) One approach to reduce the crystalline growth temperature is to enhance the mobility of surface species through energetic ion bombardment, see for example Refs. 1, 10–12, and 19. Hence, the plasma composition and charge state distribution is important since it affects the energy of the impinging ions during film growth. An approximate expression of the kinetic energy Ei of an ion is given by Ei = E0 + QeU

(1)

where E0 is the initial kinetic energy of the ion and Q is the ion charge state. U is the potential difference in the sheath in front of the substrate, and through U , the ion energy is controlled by selecting the substrate bias potential. Previously,(1,10−12,19) the effect of a varied substrate bias potential on the phase formation has been investigated. However, neither E0 nor Q was studied, the importance of the latter being exemplified through a difference of 200 eV ion energy for a singly and triple charged ion at −100 V substrate potential, respectively. In the present study, the effect of ion energy and substrate temperature on the structure and composition of alumina thin films, deposited by cathodic arc, have been investigated. The plasma chemistry and the charge state resolved ion energy distributions were characterised. A strong structure - ion energy dependence was found, with a mixture of α- and γ -alumina for conditions at 800 o C. Lower temperatures resulted in γ -dominant films. It is our ambition to contribute towards understanding of the evolution of film composition and microstructure, and investigate pathways to crystalline stoichiometric films at low deposition temperatures.

Effect of Ion Energy on Alumina Thin Films

305

2. EXPERIMENTAL PROCEDURES The investigation was performed using a filtered cathodic arc source, see schematic of the experimental setup in Fig. 1. The arc plasma was generated from a conical aluminium cathode (base and top diameters 51 and 12 mm, respectively, and height 38 mm), powered by a direct current (DC) arc supply, with resulting arc current of 35 A. A coil with 400 turns and current of 15 A served as a plasma filter. All measurements/depositions were performed at an O2 partial pressure of 3.4 × 10−3 Torr (base pressure approximately 3 × 10−6 Torr). Initial plasma characterisation was carried out using a mass-energyanalyser, (PPM 422, Pfeiffer Vacuum), determining the plasma chemistry through a mass-to-charge scan at a fixed energy, and ion energy distributions (IEDs) through energy scans for a fixed mass-to-charge ratio. The energy of the ions was measured with respect to ground. Alumina thin films were deposited on V2A stainless steel substrates, located at the previous position of the mass-energy-analyser. Prior to deposition, the substrate was heated up to the deposition temperature in the presence of oxygen, while the temperature was continuously monitored by means of a K-type (NiCr/NiAl) thermocouple at the substrate surface. Film depositions were made at the substrate temperature and the

Fig. 1. Experimental setup.

306

Ros´en et al.

substrate bias potential varied from 200 to 800 o C and from −50 to −300 V, respectively. The pulse frequency was set to 9.4 kHz, where the substrate bias potential was applied for half the cycle (the remaining time floating substrate potential, ∼ −45 V), see Fig. 1. Deposition time was 6 min, at a deposition rate of approximately 1 nm/s. The chemical composition of the films was obtained by elastic recoil detection analysis (ERDA). Cl7+ ions (total number 1012 ) with the incidence angle of 15o were used, having the energy of 35 MeV. Two detectors were applied: a Bragg ionization chamber (BIC) and a Si detector so as to determine the type and the energy of recoils. The BIC acquired all elemental information accept for H, which was analyzed by the Si detector. Details considering the ERDA setup can be found elsewhere.(20) The crystallographic structure was determined by X-ray diffraction (XRD) with a Siemens D500 diffractometer, operated at a grazing incidence angle of 2o . Cu–Kα radiation was used for the scans, with the generator set to 30 kV and 30 mA.

3. RESULTS AND DISCUSSION 3.1. Plasma Chemistry and Energetics In order to correlate phase formation with synthesis conditions, the plasma chemistry and ion energy distribution was measured at conditions later employed for thin film growth. A mass scan at 35 eV showed Al1+ , Al2+ , Al3+ and O+ as the most abundant ions, as well as traces of molecular oxygen, hydrogen and nitrogen ions (in total less than 1%). Consequently, the here presented IEDs are limited to the metal and atomic oxygen ions, see Fig. 2. The distribution for Al1+ can be described by an average energy of 41 eV, and a high energy tail extending above 100 eV. Al2+ shows a similar distribution, only slightly shifted towards higher energy. Additionally, there are smaller but clearly distinguishable distributions of Al3+ and O+ , with 90 and 11 eV average energy, respectively. These results indicate that for a substrate bias potential of −50 V, ion energies well in excess of 100 eV are attained, both from the high energy tail and the accelerated ions with Q >1 (see Eq. (1)). In Fig. 3, the charge dependent ion energy distributions resulting from a substrate bias potential of –100 V is exemplified. Correspondingly, under the assumption that the acceleration for the individual charge states are independent, ion energies above 900 eV are obtained at a substrate bias potential of −300 V. Varying the substrate bias potential does not only affect ion energy but also the magnitude of the current to the substrate, and hence the power density. For the depositions reported here, the power density

Effect of Ion Energy on Alumina Thin Films

307

1+

Eaverage Al =41 eV

10

0

25

50

75

100

125

150

175

200

2+

Eaverage Al =66 eV

Intensity (arb.unit)

10

0

25

50

75

100

125

150

175

200

3+

Eaverage Al =90 eV

1

0

25

50

75

100

125

150

175

200

1+

Eaverage O =11 eV

1

0

25

50

75

100

125

150

175

200

Energy (eV) Fig. 2. Ion energy distributions of Al1+ , Al2+ , Al3+ and O+ with respect to ground, measured at 3.4 × 10−3 Torr (thin film growth condition).

increased by a factor of approximately 6.7 as the substrate bias potential was changed from −50 to −250 V. It has previously been proposed that both ion flux and ion energy play an important role in the formation of crystalline alumina,(12) and hence, this is expected also for the present investigation. 3.2. Film Composition It was previously shown that the presence of plasma, may lead to both incorporation thereof in inhibition of crystalline growth.(22) The result of ERDA measurements is shown in Fig. 4, presenting

residual gas in the the film(21) and the the here performed the O/Al fraction as

Ros´en et al.

308

Al

15

3+

Al

100

10

50

5 +

O 0 0

100

200

300

400

Intensity (arb.unit)

Intensity (arb.unit)

2+

1+

Al

150

0 600

500

Ion energy (eV) Fig. 3. The charge dependent ion energy distributions resulting from a substrate bias potential of −100 V.

H (at%)

1.45

0.25

1.40

0.20

1.35

0.15

1.30

0.10

1.25

1.50

0.05

-50 V substrate bias potential

1.20 100

200

300

400

500

600

0.00 700

0.30

1.45

0.25

1.40

0.20

1.35

0.15

1.30

0.10

-150 V substrate bias potential

1.25 100

200

300

400

500

600

H (at%)

O/Al

0.30

O/Al

H (at%)

O/Al

1.50

0.05 700

o

Temperature ( C)

Fig. 4. Film composition; fraction O/Al and H content (at%) vs. substrate temperature as measured with ERDA.

Effect of Ion Energy on Alumina Thin Films

309

well as the hydrogen content vs. substrate temperature, for films grown at −50 and −150 V substrate bias potential, respectively. The hydrogen content in the films is <0.25 at%. Furthermore, the films contain no additional traces of residual gas, like carbon or nitrogen. At −50 V substrate bias potential, the O/Al fraction is temperature dependent, see top part of Fig. 4. Up to 300 o C the films are understoichiometric, while at 500 o C, the value is close to 1.5, corresponding to Al2 O3 . The oxygen content can be understood by gas adsorption on the substrate surface (see gas inlet close to the substrate in Fig. 1) with temperature dependent molecule dissociation and diffusion driven incorporation in the film. A similar trend has been observed in a study of O/Mo atomic ratio surface composition as a function of substrate temperature,(23) where the increasing oxygen content with temperature is explained by diffusion driven incorporation in the film. For −150 V substrate bias potential, the films are close to stoichiometric over the entire temperature range, as shown in the lower part of the figure. This can be explained by the increased ion energy (as compared to the −50 V case), supplying the activation energy for dissociation driven diffusion. 3.3. Film Microstructure An overview of the phase formation in films deposited at different substrate temperatures and substrate bias potentials is presented in Fig. 5. Four regions can be distinguished; amorphous and weakly crystalline films at −50 V substrate bias potential, and at lower substrate potential films containing the γ -phase, the γ -phase with small traces of the α-phase (the intensity of any identified α-peak being less than 10% of the highest γ -peak), and a mixture between the α- and the γ -phase, at temperatures ≤ 500 o C, 600–700 o C, and 800 o C, respectively. The raw data for Fig. 5 is shown in Figs. 6–9. Inserted in the top scan are the positions of the main peaks of γ - and α-alumina (according to JCPDS).(24) These lines are extended by vertical dotted lines, as a guide to the eye. In order to separate the substrate peaks from the alumina peaks, an XRD scan of a substrate (heated to 700 o C in deposition environment) is also shown. Previously reported cathodic arc film growth, shows the presence of α-alumina at 800 o C, independent of substrate bias potential used.(10,11) The present investigation therefore included depositions at this temperature, see Fig. 6. At −50 V substrate bias potential (a), the film was close to amorphous, but small traces of both α- and γ -alumina were detected. A higher substrate bias potential, −100 V (b), resulted in a drastic increase in peak intensity, with an estimated equal phase mixture of

Ros´en et al.

310

T / Tm

0.3

0.45

Substrate bias potential (V)

-50 -100 -150 -200

amorphous γ γ + α mixture γ + α-traces

-250 -300 200

300

400

500

600

700

800

Substrate temperature ( ˚C) Fig. 5. Phase formation in films deposited at different substrate temperature and with different substrate bias potential.

α- and γ -alumina. With a substrate potential of −150 V (c), the γ -phase became dominant. In Fig. 7, XRD scans for films grown at 700 o C are presented. At −50 V substrate bias potential (a), the film is close to amorphous. At a higher substrate bias potential, the peak intensity increases (b, c), and at −150 V (c) and higher (d–f) the γ -phase is clearly dominating. Traces of the α-phase are also present, with a maximum intensity at −150 V. For the specific conditions used here, the substrate bias potential required for both γ - and α-alumina formation seems to be higher than −100 V. Furthermore, the α-phase never dominates the phase mixture. This is not consistent with Refs. 10 and 11 showing a correlation between an increased substrate bias and an increased content of the α-phase, resulting in α-dominant films. However, a direct comparison is not possible due to differences in the deposition conditions, such as arc current and magnetic field strength in the plasma filter. The latter is well known to affect plasma composition(25) and ion charge states,(26) which in turn affects the ion energy, see Eq.(1). Hence, the plasma characterisation performed here is essential for progress in understanding structure evolution; one can conclude that for the ion flux specific for the here presented depositions, ion energies in excess of 100 eV (for −50 V substrate bias potential) results in amorphous structure at 700 o C. In contrast, an ion energy distribution ranging from 100 to 500 eV (as exemplified in Fig. 3 for −100 V substrate bias) is sufficient to enable crystalline growth.

Effect of Ion Energy on Alumina Thin Films

311 o

γ

Tsubstrate= 800 C

α

(a) 1

-50 V 30

40

50

60

70

80

30

40

50

60

70

80

(b)

Intensity (arb. unit)

1

-100 V (c) 1

-150 V 3

30

40

50

60

70

80

2 1

substrate

0 30

40

50

60

70

80

2θ Fig. 6. XRD scans for films deposited at 800 o C with a substrate bias potential ranging from −50 to −150 V. Bottom scan is substrate only. The dotted lines show the main peaks of α and γ alumina (JCPDS(24) ).

There are several other factors that may be of importance for the phase formation. One is the choice of substrate material, evident from low temperature crystalline depositions using a template.(13,27−28) The films analysed in Ref. 10 are deposited on vanadium foils, which prior to deposition may react with oxygen and serve as templates. Also the ion flux has been shown to be of significance.(1,12) Another deposition sequence at 600 o C, showed clearly γ -dominant films for the entire substrate bias potential range, see Fig. 8, though small traces of α-alumina are present at −100 and −150 V substrate bias potential. The high intensity of the γ -peaks motivates a study of this phase at further reduction of the substrate temperature: Fig. 9a shows phase

Ros´en et al.

312 o

Tsubstrate= 700 C

γ α

(a) 1

-50 V 30

40

50

60

70

80

30

40

50

60

70

80

30

40

50

60

70

80

30

40

50

60

70

80

30

40

50

60

70

80

(b) 1

-100 V (c)

Intensity (arb. unit)

1

-150 V (d)

1

-200 V (e) 1

-250 V (f) 1

-300 V 30

40

50

60

70

80

3

substrate

2 1 0 30

40

50

60

70

80

2θ Fig. 7. XRD scans for films grown at 700 o C with a substrate bias potential ranging from −50 to −300 V. Bottom scan is substrate only. The dotted lines show the main peaks of α and γ alumina (JCPDS(24) ).

Effect of Ion Energy on Alumina Thin Films

313 o

Tsubstrate= 600 C

γ

(a)

α

1

-50 V 30

40

50

60

70

80

(b)

Intensity (arb. unit)

1

-100 V 30

(c)

40

50

60

70

80

1

-150 V 3

30

40

50

60

70

80

2 1

substrate

0 30

40

50

60

70

80

2θ Fig. 8. XRD scans for films grown at 600 o C with a substrate bias ranging from −50 to −150 V. Bottom scan is substrate only. The dotted lines show the main peaks of α and γ alumina (JCPDS(24) ).

information for a deposition at 500 o C and −50 V substrate bias potential. The film is amorphous, with small traces of aluminium. (Corresponding depositions at further reduced temperatures, results in completely amorphous films.) The effect of the ion energy is demonstrated in Fig. 9b–e, where depositions at −150 V substrate bias potential at 200–500 o C show presence of the γ -phase. To our knowledge, the lowest temperature previously reported for the formation of this phase is 290–350 o C.(9) Based on recent reports on the application of γ -phase alumina in metal cutting(4) and steel forming,(29) underlining the (not understood) stability of these thin films, the above reported synthesis route may be of technological interest for the deposition of heat sensitive substrates.

Ros´en et al.

314

γ α

(a) 1 o

500 C 30

-50 V 40

50

60

70

80

40

50

60

70

80

40

50

60

70

80

40

50

60

70

80

(b) 1 o

500 C 30

-150 V

(c)

Intensity (arb. unit)

1 o

400 C 30

-150 V

(d) 1 o

300 C 30

-150 V

(e) 1 o

200 C 30

-150 V 40

50

60

70

80

3 2

substrate

1 0 30

40

50

60

70

80

2θ Fig. 9. XRD scans for films grown at 200–500 o C with substrate bias potential of −50 or −150 V. Bottom scan is substrate only. The dotted lines show the main peaks of γ alumina (JCPDS(24) ).

According to the structure zone model by Movchan and Demchishin,(30) further developed by Thornton,(31) surface diffusion is initiated at temperatures of T/Tm > 0.25–0.30, where Tm is the melting temperature. For α-alumina, this corresponds to approximately 310–420 o C. Correspondingly, bulk diffusion dominates for T/Tm > 0.45, which corresponds to T >

Effect of Ion Energy on Alumina Thin Films

315

770 o C. However, the phase analysis presented in this study shows that the ion energy is crucial for the resulting film microstructure. Hence, it is reasonable to assume that the activation energy required for surface diffusion, and hence for the formation of crystalline alumina, is provided by the ion bombardment. As shown in Fig. 3, the total ion energy distribution extends up to 500 eV at −100 V substrate bias potential. At these energies, defect formation upon ion–surface interaction can be expected, see for example Ref. 32, where an increased defect density is shown as the substrate bias potential is increased from −75 to −150 V. It is well known that defects can serve as nucleation sites for grain growth.(33) Therefore, small sized grains may be expected for the films deposited. Through analysis of the full width at the half maximum of all the non-overlapping peaks in the here presented diffraction data, by Scherrer’s formula,(34) the grain size was estimated to be less than 10 nm for both the α- and the γ phase as well as any mixture thereof. It has previously been shown that the lower surface energy of the γ -phase as compared to the α-phase, thermodynamically stabilizes growth of the γ -phase for nanocrystalline phase formation.(35) The maximum grain size diameter favouring γ -formation, as calculated from Ref. 35 is 12 nm. The here discussed constitution data is therefore consistent with Ref. 35.

4. CONCLUSION The effect of ion energy and substrate temperature on the film composition and microstructure has been investigated for cathodic arc deposited alumina thin films. At a substrate bias potential of −50 V, the films were amorphous or weakly crystalline, with an understoichiometric O/Al ratio for the growth temperatures below 500 o C. Correspondingly, a substrate bias potential above −100 V resulted in stoichiometric crystalline films; a mixture of the α- and γ -phase above 700 o C, and clearly γ -phase dominated films at temperatures as low as 200 o C. The grain size of all films was <10 nm. Combined constitution and grain size data is consistent with previous work stating that γ -alumina is thermodynamically stable at grain sizes <12 nm. To correlate phase formation with synthesis conditions, the plasma chemistry and ion energy distribution was measured at conditions employed for thin film growth. Due to charge dependent acceleration when using a substrate bias potential, −100 V was shown to result in a total ion energy distribution with energies up to 500 eV. These results contribute towards understanding of the evolution of film composition and microstructure, and is illustrating a pathway to γ -alumina growth at

316

Ros´en et al.

temperatures as low as 200 o C. This may be technologically important for the deposition on heat sensitive substrate materials. ACKNOWLEDGMENTS This work was supported by the Swedish Research Council (VR). J. M. S. acknowledges support by the Alexander von Humboldt Foundation, the German Federal Ministry of Education and Research, and the Program for Investment in the Future. REFERENCES 1. J. M. Schneider, W. D. Sproul, A. A. Voevodin, and A. Matthews, J. Vac. Sci. Technol. A 15, 1084 (1997). 2. S. Zhu, F. Wang, H. Lou, and W. Wu, Surf. Coat. Technol. 71, 9 (1995). 3. E. Serra, G. Benamati, and O. V. Ogorodnikova, J. Nucl. Mater. 255, 105 (1998). ¨ 4. A. Schutze and D. T. Quinto, Surf. Coat. Technol. 162, 174 (2003). 5. C. T¨aschner, B. Ljungberg, V. Alfredsson, I. Endler, and A. Leonhardt, Surf. Coat. Technol. 108–109, 257 (1998). 6. G. H. Prengel, W. Heinrich, G. Roder, and K. H. Wendt, Surf. Coat. Technol. 68–69, 217 (1994). ¨ 7. Edelstahl Handbuch (Bohler Edelstahl GmbH, Kapfenberg, Germany, 1998). 8. O. Zywitzki, G. Hoetzsch, F. Fietzke, and K. Goedicke, Surf. Coat. Technol. 82, 169 (1996). 9. O. Zywitzki and G. Hoetzsch, Surf. Coat. Technol. 86–87, 640 (1996). 10. Y. Yamada-Takamura, F. Koch, H. Maier, and H. Bolt, Surf. Coat. Technol. 142–144, 260 (2001). 11. R. Brill, F. Koch, J. Mazurelle, D. Levchuk, M. Balden, Y. Yamada-Takamura, H. Maier, and H. Bolt, Surf. Coat. Technol. 174–175, 606 (2003). 12. O. Kyrylov, D. Kurapov, and J. M. Schneider, Appl. Phys. A 80, 1657 (2005). 13. J. M. Andersson, Z. Czig´any, P. Jin, and U. Helmersson, J. Vac. Sci. Technol. A 22, 117 (2004). 14. S. Ruppi and A. Larsson, Thin Solid Films 388, 50 (2001). 15. A. Larsson and S. Ruppi, Int. J. Refract. Hard Mater. 19, 515 (2001). 16. F. Fietzke, K. Goedicke, and W. Hempel, Surf. Coat. Technol. 86, 657 (1996). 17. O. Zywitzki and G. Hoetzsch, Surf. Coat. Technol. 94–95, 303 (1997). ¨ 18. R. Cremer, M. Witthaut, D. Neuschutz, G. Erkens, T. Leyendecker, and M. Feldhege, Surf. Coat. Technol. 120–121, 213 (1999). 19. Q. Li, Y.-H. Yu, C. S. Bhatia, L. D. Marks, S. C. Lee, and Y. W. Chung, J. Vac. Sci. Technol. A 18, 2333 (2000). 20. U. Kreissig, S. Grigull, K. Lange, P. Nitzsche, and B. Schmidt, Nucl. Instrum. Meth. B 136, 674 (1998). ¨ 21. J. M. Schneider, A. Anders, B. Hjorvarsson, I. Petrov, K. Mac´ak, U. Helmersson, and J.-E. Sundgren, Appl. Phys. Lett. 74, 200 (1999). ¨ 22. J. M. Schneider, K. Larsson, J. Lu, E. Olsson, and B. Hjorvarsson, Appl. Phys. Lett. 80, 1144 (2002). ˜ 23. S. I. Castaneda, I. Montero, J. M. Ripalda, N. D´ıas, L. Gal´an, and F. Rueda, J. Appl. Phys. 85, 8415 (1999).

Effect of Ion Energy on Alumina Thin Films

317

24. Joint Committee on Powder Diffraction Standards (JCPDS). 25. J. M. Schneider, A. Anders, and G. Y. Yushkov, Appl. Phys. Lett. 78, 150 (2001). 26. E. Oks and G. Yushkov, 17th International Symposium on Discharges and Electrical Insulation in Vacuum, Berkeley (1996). 27. P. Jin, G. Xu, M. Tazawa, K. Yoshimura, D. Music, J. Alami, and U. Helmersson, J. Vac. Sci. Technol. A 20, 2134 (2002). 28. P. Jin, S. Nakao, S. X. Wang, and L. M. Wang, Appl. Phys. Lett. 82, 1024 (2003). 29. D. Kurapov and J. M. Schneider, Steel Res. Int. 75, 577 (2004). 30. B. A. Movchan and A. V. Demchishin, Phys. Met. Metallogr. 28, 83 (1969). 31. J. A. Thornton, J. Vac. Sci. Technol. 11, 666 (1974). ¨ 32. T. Savisalo, D. B. Lewis, P. Eh. Hovsepian, and W.-D. Munz, Thin Solid Films 460, 94 (2004). 33. W. D. Callister, Materials Science and Engineering An Introduction, 5th Ed., John Wiley & Sons, New York, 2000. 34. B. D. Cullity, Elements of X-Ray Diffraction, Addison-Wesley, Menlo Park. California, 1978. 35. J. M. McHale, A. Auroux, A. J. Perrotta, and A. Navrotsky, Science 277, 788 (1997).