Gas barrier properties of titanium oxynitride films deposited on polyethylene terephthalate substrates by reactive magnetron sputtering

(1)

Gas barrier properties of titanium oxynitride films deposited on

polyethylene terephthalate substrates by reactive magnetron sputtering

M.-C. Lin

a

, L.-S. Chang

a,

*

, H.C. Lin

b

a_{Department of Materials Science and Engineering, National ChungHsin University, 250, Kuo-Kung Road, 40227 Taichung, Taiwan, ROC} b_{Department of Materials Science and Engineering, National Taiwan University, 1, Roosevelt Road, Sec. 4, 106 Taipei, Taiwan, ROC}

Received 22 August 2007; received in revised form 23 November 2007; accepted 23 November 2007 Available online 15 December 2007

Abstract

Titanium oxynitride (TiNxOy) films were deposited on polyethylene terephthalate (PET) substrates by means of a reactive radio frequency (RF)

magnetron sputtering system in which the power density and substrate bias were the varied parameters. Experimental results show that the deposited TiNxOyfilms exhibited an amorphous or a columnar structure with fine crystalline dependent on power density. The deposition rate

increases significantly in conjunction as the power density increases from 2 W/cm2to 7 W/cm2. The maximum deposition rate occurs, as the substrate bias is 40 V at a certain power densities chosen in this study. The film’s roughness slightly decreases with increasing substrate bias. The TiNxOyfilms deposited at power densities above 4 W/cm

2

show a steady Ti:N:O ratio of about 1:1:0.8. The water vapor and oxygen transmission rates of the TiNxOyfilms reach values as low as 0.98 g/m2-day-atm and 0.60 cm3/m2-day-atm which are about 6 and 47 times lower than those of

the uncoated PET substrate, respectively. These transmission rates are comparable to those of DLC, carbon-based and Al2O3barrier films.

Therefore, TiNxOyfilms are potential candidates to be used as a gas permeation barrier for PET substrate.

Keywords: Titanium oxynitride; Sputtering; Polyethylene terephthalate (PET); Gas permeation

1. Introduction

The increased demand in portable devices, for example, the flexible displays and thin film batteries used in note book computers, mobile phones and various electronic instruments, has led to a great deal of work pertaining to the development of suitable substrate materials. These substrate materials are required to possess certain properties, such as portability, durability, flexibility and optical transparency. To meet these requirements, polymers are considered to be the most appropriate material. Nevertheless, most polymer substrates have some drawbacks. One of the most critical properties of polymer substrates which need to be addressed in order to achieve successful operations is their insufficient resistance to gas permeation.

The deposition of diamond-like carbon (DLC) [1–4], carbon-based [5,6] and metal oxide films [7–9] have been

considered to be a promising technique to improve the properties of polymer substrates used in portable devices by enhancing their resistance to abrasive wear, gas permeation, chemical attack, etc. These DLC, carbon-based and metal oxide films can be successfully deposited by magnetically enhanced chemical vapor deposition, plasma-source ion implantation, RF low-pressure glow discharge, magnetron sputtering and microwave techniques. An alternative coating materials class that may merit consideration are transition-metal oxynitrides (TMeNxOy), which, due to their colorific and optical properties,

chemical stability and good adhesion to polymers, have previously been widely employed as a wear resistant, anti-reflective, decorative and/or diffusion barrier coating for polymer components[10–15]. TiNxOyfilms are representative

transition-metal oxynitrides and can be deposited onto substrates by various coating techniques, which include magnetron sputtering [10,11], ion assisted deposition [14], and evaporation[15]. Magnetron sputtering is considered to be more advantageous among these different deposition techni-ques due to its low processing temperature, dense deposited film, moderate set up cost and high stability in control.

www.elsevier.com/locate/apsusc Available online at www.sciencedirect.com

Applied Surface Science 254 (2008) 3509–3516

* Corresponding author. Tel.: +886 4 2284 0500/406; fax: +886 4 2285 2433. E-mail address:[email protected](L.S. Chang).

(2)

Polyethylene terepthalate (PET) is one of the most promising polymers being used as flexible substrate due to its excellent optical transparency, chemical stability, high durability and low cost. However, to the author’s knowledge, no related paper which investigated the deposition of TiNxOyfilms on PET substrates has

been reported. In this study, the TiNxOyfilms were deposited on

PET substrates by RF reactive magnetron sputtering. The influences of power density and applied substrate bias on the deposition rate, gas permeation rate, microstructure and composition of the TiNxOyfilms were investigated.

2. Experimental details

TiNxOy films were deposited by means of a reactive

magnetron sputtering system (made by Junsun Corporation, Taiwan). A 5 cm diameter, 99.999% pure titanium target was used. The PET substrate, a C10–H8–O4compound with 3–6%

humidity (product no: BD11, Nan-Yan Corporation, Taiwan) had a thickness of 100 mm. The substrate to target distance was kept at 15 cm and the holder was rotated at a speed of 10 rpm to improve the film homogeneity. Argon was used as the sputter gas and nitrogen as the reactive gas during sputtering. Depositions were carried out in pure Ar and Ar/N2mixtures.

Meanwhile, oxygen atoms existing residually in chamber or coming from the PET substrate would contribute to form the TiNxOyfilms during deposition. The sputtering parameters used

in this study are listed inTable 1, in which the power density and dc substrate bias varied while the other parameters were fixed. To clean the substrate surface prior to the deposition and improve film’s adhesion, the PET surface was plasma pre-treated in a sputtering chamber for 10 min with a direct current (dc) bias of 300 V, which is generated by an MP-1 advanced converter magnetron power supply. The pre-treatment was carried out by using Ar gas with a pressure of 1.8 Pa. The same power supply is used for both pre-treatment of substrate and sputtering process, although their amounts of dc bias were different. The dc bias is applied between the stainless holder and stainless chamber wall. Although the electrical conductiv-ity of the TiNxOy film is low, the dc bias between the PET

substrate and stainless chamber wall can exhibit sufficient effect on the film deposition.

The water vapor and oxygen transmission rates of uncoated and TiNxOy coated PET were measured by a

Permatran-w 3/61 and an Ox-Tran 2/61model system (made by MOCON Instrument), respectively. The sample area was

4.5 cm 4.5 cm. Both transmission rate measurements were carried out at atmospheric pressure and 40 8C, and under relative humidity of 100% and 0%, respectively. A multi-function scanning probe microscope (SPM, NS4 D3100CL Digital Instrument) was used to measure the film’s roughness. A Si tip with a radius of 10 nm was used for this analysis. The scanned sample area was 1 mm2and the scanning speed was 1 Hz.

The field-emission scanning electron microscope (FESEM, JSM-6700F, JEOL) for studying the coating morphology and microstructure was operated at 1 keV accelerating voltage. The TiNxOysurface was coated with a thin platinum layer for SEM

observation to prevent charging during FESEM analysis. The crystal structures and chemical compositions of deposited TiNxOy films were analyzed by means of a high-resolution

transmission electron microscope (HRTEM, JEOL-3010, operated at 200 keV) and X-ray photoelectron spectroscopy (XPS, Theta Probe, VG), respectively. The XPS spectra were obtained using Al Ka X-rays operated at 15 kV and 400 W. The Ti, N and O concentrations in films were quantified from the areas under the Ti 2p, N 1s and O 1s characteristic signals in the photoelectron spectrum, after carrying out the Shirley back-ground subtraction. Relative sensitivity factors provided by the manufacturer were employed. The sampling size and surface etching time were 1 cm2and 30 s, respectively. The pressure in the analysis chamber was 8 10 8_{Pa. The thicknesses of the}

TiNxOy films deposited on pure silicon substrates placed

alongside the PET substrates were measured by means of an a-step profiler (Dektak3ST, Veeco).

3. Results and discussion

3.1. Chemical composition and microstructure

Fig. 1shows the XPS spectra of TiNxOyfilms deposited on

PET substrates at various power densities and with a fixed substrate bias of 40 V. The deposited films were cleaned before XPS analysis to prevent them from contamination by adventitious oxygen. The analysis of XPS spectra in Fig. 1

indicates that Ti, N and O elements coexist in the deposited TiNxOyfilms. The appearance of O element is consistent with

the reported studies on titanium oxynitride and aluminum oxynitride films by Guillot et al.[16]and Dreer et al.[17]. Their films were also deposited in a sputter chamber with very little or even no oxygen content. This indicates that at a base pressure of 5 10 6_{Pa or even below, there is enough oxygen present to}

partake in the reaction with Ti and N atoms to form the TiNxOy

films. In this work, the H2O molecules which adhere on the

surface or exist in the interior of PET substrates[18], and the residual O2impurity in the sputter chamber, are expected to

provide the O atoms during the deposition of TiNxOyfilms.

Fig. 2(a–c) shows the XPS Ti 2p lines of TiNxOy films

deposited onto PET at various power densities with a fixed substrate bias of 40 V. As shown inFig. 2(a), there occurs a predominant Ti–O2bonding for the film deposited at a power

density of 2 W/cm2. By increasing power density to 4 W/cm2or 7 W/cm2, there occurs both Ti–N and Ti–O2bonding in the

deposited film, as shown inFig. 2(b–c). This feature can be

Table 1

The sputtering parameters used in this study

Target Ti target (99.999% purity) with diameter of 5 cm

Substrate PET

Power density 2 W/cm2, 4 W/cm2, 6 W/cm2, 7 W/cm2 dc substrate bias 0 V, 20 V, 40 V, 60 V

Gas flow ratio Ar:N2= 3:2

Base pressure 1.33 10 3_Pa

Working pressure 1.33 10 1_Pa

Deposition time 60 min

M.-C. Lin et al. / Applied Surface Science 254 (2008) 3509–3516 3510

(3)

explained as below. At a low power density of 2 W/cm2, less quantity of Ti atoms are sputtered from the Ti target, and some Ti ions are created after sputtering due to charge exchange. These Ti atoms (ions) will first react with O due to their strong affinity. Therefore, the deposited film is mainly comprised of Ti and O elements. With increasing power density, more Ti atoms are sputtered from Ti target and more Ti ions are created due to charge exchange. There is more opportunity to form the Ti–N bonding besides of the Ti–O2 bonding. Therefore, the XPS

spectra intensities of Ti and N elements maintain steady values at higher power densities, as shown inFig. 1(b–d) andFig. 2(b– c). Based on the quantitative analysis of XPS, the compositions of TiNxOy films deposited at various power densities are

presented inFig. 3. As can be seen inFig. 3, the TiNxOyfilms

deposited at power densities above 4 W/cm2 show a steady Ti:N:O ratio of about 1:1:0.8.

Fig. 4(a–c) shows the HR-TEM cross-sectional images and diffraction patterns of TiNxOy films deposited onto PET at

various power densities with a fixed substrate bias of 40 V. In

Fig. 4(a), the TiNxOyfilm deposited at power densities of 2 W/

cm2 exhibits an amorphous structure. Increasing the power density to higher than 4 W/cm2, the TiNxOy film exhibits a

columnar structure with fine crystalline, as shown inFig. 4(b– c). The occurrence of columnar structure with fine crystalline in TiNxOy films may be ascribed to the rising of substrate

temperature during the deposition process. The bombardment of the sputtered atoms (ions) will raise the substrate temperature during the deposition of TiNxOyfilms. At a higher

power density, more sputtered atoms (ions) are bombarded onto the PET substrate and the substrate temperature is increased to a higher one. In the present study, the substrate temperature during deposition can reach about 80 8C at a power density of 7 W/cm2. This rising of substrate temperature will exhibit sufficient driving force for columnar structure to with fine crystalline. As illustrated clearly inFig. 4(b) and (c), the higher the power density, the more obvious the columnar structure with fine crystalline. Besides, it is worthy to mention that the indistinct trace surrounding the surface of these columnar structures inFig. 4(b–c) is only a false image introduced during the preparation of TEM specimens.

Fig. 1. XPS spectra of TiNxOyfilms deposited on PET with a fixed substrate bias of 40 V at various power densities. (a) 2 W/cm 2

, (b) 4 W/cm2, (c) 6 W/cm2, (d) 7 W/cm2.

(4)

3.2. Deposition rate

Fig. 5shows the deposition rates of TiNxOyfilms deposited

at various power densities and substrate biases. It is found that the deposition rate increases significantly with increasing power density. The deposition rate increases 4–9 times, depending on the power density in the range of 2–7 W/cm2. This phenomenon can be explained as follows. At higher power density, more Ar+ and N2+ ions are generated to impact the

titanium target and more Ti atoms (ions) are sputtered out and react with N and O atoms in the plasma to form TiNxOyand

deposit onto the PET substrate. Hence, the deposition rate of TiNxOyfilm increases with increasing power density.

One can also find, inFig. 5, that the deposition rate increases slightly from 0 V to 20 V bias, reaches a maximum at 40 V bias, and then decreases at 60 V bias. Actually, proper supply of substrate bias (such as 20 V to 40 V) will raise the negative potential between the ground and substrate. Ti ions are more guided onto the substrate surface due to this negative potential. Besides, the moving direction of other species (Ti and

N atoms) will also be more confined to around the substrate due to the collision with these guided Ti ions. All these effects will increase the adhesion of deposited species onto the substrate, and hence increase the film’s deposition rate.[19,20]. However, an over-high substrate bias (such as 60 V) will produce a significant drop of electrical potential in the plasma. This will make the species (atoms or ions) have over-high impact energy and induce re-sputtering, and hence the deposition rate of TiNxOyfilm decreases significantly. It is worthy to mention that

the maximum deposition rate of TiNxOyfilms, shown inFig. 5,

is only about 1.2 nm/min, which is much lower than those of DLC (750 nm/min)[1], carbon-based (21 nm/min) and Al2O3

(51 nm/min) films[5,7].

3.3. Surface morphology and roughness

Fig. 6(a–d) shows the FESEM micrographs of TiNxOyfilms

deposited onto PET at various power densities and without substrate bias. As shown in Fig. 6(a), many small pinholes appear in the TiNxOyfilm prepared at a power density of 2 W/ Fig. 2. The XPS Ti 2p lines of TiNxOyfilms deposited onto PET at various power densities and with a fixed substrate bias of 40 V.

(5)

Fig. 3. Chemical compositions of TiNxOyfilms deposited onto PET at various

power densities and with a fixed substrate bias of 40 V.

Fig. 5. Deposition rates of TiNxOyfilms on PET in dependence of the substrate

bias at various power densities (film thickness = deposition rate 60).

(6)

cm2. This may come from the fact that the sputter species (atoms or ions) have smaller mobility at a lower power density of 2 W/cm2, and hence the deposited TiNxOyfilm grows with a

lot of small pinholes. By increasing power density to 4 W/cm2, the quantity of pinhole decreases, as shown in Fig. 6(b). If power density is higher than 6 W/cm2, as shown inFig. 6(c–d), no pinholes are observed on the TiNxOyfilms. This indicates

that the sputter species (atoms or ions) will have enough

mobility to deposit a dense TiNxOy film without pinholes at

power densities higher than 6 W/cm2.

Fig. 7(a–d) shows the FESEM micrographs of TiNxOyfilms

deposited on PET with various substrate biases and at a fixed power density of 7 W/cm2. One can find that most TiNxOyfilms

(Fig. 7(a–c)) exhibit a cluster-type surface morphology dotted with tiny particles. Differing from that, the TiNxOy film

deposited with 60 V bias (Fig. 7(d)) exhibits a surface

Fig. 6. FESEM micrographs of TiNxOyfilms deposited on PET at various power densities without substrate bias. (a) 2 W/cm2, (b) 4 W/cm2, (c) 6 W/cm2, (d) 7 W/

cm2_.

Fig. 7. FESEM micrographs of TiNxOyfilms deposited on PET at a power density of 7 W/cm 2

with various substrate biases. (a) 0 V (b) 20 V (c) 40 V (d) 60 V. M.-C. Lin et al. / Applied Surface Science 254 (2008) 3509–3516

(7)

morphology without those closely dotted tiny particles. This feature is ascribed to the fact that the species (atoms or ions) in the plasma with 60 V bias have over-high impact energy to bombard the surface of TiNxOy films. This intense

bombard-ment may impinge away those tiny particles on TiNxOyfilms,

although some of those tiny particles may be incorporated in the film during the continuous deposition process. This phenom-enon is consistent with the result of the re-sputtering effect discussed in Section3.2.

The values of root-mean-square (RMS) roughness of TiNxOy

films deposited on PET with various substrate biases, measured by using the scanning probe microscope, are presented inFig. 8. The film’s surface roughness decreases with increasing

substrate bias. The kinetic energy of depositing atoms is increased at higher substrate bias. These depositing atoms with higher kinetic energy will easily move into the space between clusters and hence the TiNxOy films can exhibit smoother

surface morphologies at higher substrate bias. The re-sputtering effect at higher substrate bias also smoothens the film surface as shown in Fig. 7(d).

3.4. Gas permeation

The water vapor transmission rate (WVTR) and oxygen transmission rate (OTR) of TiNxOy films coated PET in

dependence on power densities and substrate biases are plotted in Fig. 9(a) and (b), respectively. The WVTR and OTR of uncoated PET are 5.53 g/m2-atm-day and 28.09 cm3/m2 -atm-day, respectively. It can be clearly seen inFig. 9(a–b) that the WVTR and OTR of the PET substrates reduce significantly after the deposition of TiNxOyfilms. From these two figures, it

can be concluded that both the film’s thickness and quality, resulting from the different power densities and substrate biases, have important effects on the film’s resistance against gas permeation.

As already mentioned, the deposition rate of TiNxOy film

increases with increasing power density, and hence the film thickness and consequent resistance against permeation of water vapor and oxygen also increase. It has also been shown in

Fig. 5 that, the deposition rate reaches a maximum at 40 V bias and then decreases due to re-sputtering at 60 V bias. This feature can explain why the WVTR and OTR reduce as the substrate bias changes from 0 V to 40 V, and then rise as the bias ranges from 40 V to 60 V, as shown inFig. 9. All these results exhibit that a thicker TiNxOyfilm will be a better gas

barrier[21,22]. InFig. 9, the OTR of TiNxOyfilms deposited at

power density of 2 W/cm2is found to have almost the same value as that of uncoated PET substrate. This phenomenon is understandable and can be explained as below. As shown in

Fig. 6(a), TiNxOyfilms deposited at power density of 2 W/cm2

Fig. 9. (a) WVTR and (b) OTR of TiNxOyfilms coated PET in dependence of the substrate biases and power densities.

Fig. 8. RMS roughness of TiNxOyfilms deposited on PET in dependence of the

(8)

have many small pinholes. These small pinholes act as the shortcut for oxygen permeation, and these films have high OTR values. It is worthwhile mentioning that these small pinholes existing in TiNxOy films still have an effective resistance to

water vapor, so their WVTR still maintains a quite low value (Fig. 9(a)).

As discussed above, properly deposited TiNxOy films can

significantly improve the barrier properties of PET substrates. In this study, the optimal sputtering parameters in depositing excellent TiNxOy films are power density of 7 W/cm2 and

40 V substrate bias. The water vapor and oxygen transmission rates of these optimal TiNxOy films reach values as low as

0.98 g/m2-day-atm and 0.60 cm3/m2-day-atm, respectively. These values of WVTR and OTR are about 6 and 47 times lower than those of the uncoated PET substrate.

It is valuable to compare briefly the gas permeation resistance of TiNxOy films and those currently used DLC,

carbon-based and Al2O3films. As presented above, the water

vapor and oxygen transmission rates of the TiNxOy films

prepared in this study can reach optimal values of 0.98 g/m2 -day-atm and 0.60 cm3/m2-day-atm, respectively. In the reported studies, the oxygen transmission rates of DLC and carbon-based films are about 0.4–4.5 cm3/m2-day-atm [1–4]

and 5.7 cm3/m2-day-atm [6], respectively, and the water transmission rate of Al2O3 is about 1.12 g/m2-day-atm [7].

Hence, based on these data, the TiNxOyfilms can exhibit similar

or even higher resistance of gas permeation as compared to DLC, carbon-based and Al2O3barrier films. Namely, TiNxOy

films are potential candidates to be used as a gas permeation barrier. Meanwhile, the TiNxOy films can have better electric

conductivity than Al2O3films, higher mechanical property than

carbon-based films and their manufacturing cost is lower than that of DLC films.

4. Conclusions

TiNxOy films have been successfully deposited on PET

substrates by means of the RF reactive magnetron sputtering technique. The influences of power density and substrate bias on the film’s properties are investigated. The TiNxOy film

deposited at power density of 2 W/cm2exhibits an amorphous structure. With increasing power density to be higher than 4 W/ cm2, the TiNxOy film exhibits a columnar structure with fine

crystalline. The deposition rate of TiNxOyfilm has an increment

of 4–9 times, when the power density rises from 2 W/cm2up to 7 W/cm2. The deposition rate reaches its maximum with the substrate bias of 40 V at all power densities chosen in this study. Most TiNxOy films exhibit a cluster-type surface

morphology with closely dotted tiny particles. Their surface roughness decreases slightly with increasing substrate bias. A lot of small pinholes are observed in TiNxOyfilms deposited at a

low power density of 2 W/cm2. The TiNxOyfilms deposited at

power densities above 4 W/cm2show a steady Ti:N:O ratio of about 1:1:0.8. The film’s resistance against gas permeation increases with increasing film’s thickness and quality, resulting from the different power densities and substrate biases. The water vapor and oxygen transmission rates of the TiNxOyfilms

reach optimal values as low as 0.98 g/m2-day-atm and 0.60 cm3/m2-day-atm and these values are about 6 and 47 times lower than those of the uncoated PET substrate, respectively, while the power density is 7 W/cm2and substrate bias is 40 V. As compared to DLC, carbon-based and Al2O3

barrier films, the sputtered TiNxOyfilms can exhibit similar or

even higher resistance of gas permeation. Besides, the TiNxOy

films can have better electric conductivity than Al2O3 films,

higher mechanical property than carbon-based films and their manufacturing cost is lower than that of DLC films.

Acknowledgement

This work was financially supported by the National Science Committee (NSC) of Taiwan/ROC, under the auspices of the Targeted Project (no. NSC93-2216-E-005-024).

References

[1] K. Inagawa, T. Zeniya, N. Hibino, Surf. Coat. Technol. 169/170 (2003) 344.

[2] S. Yamamoto, H. Kodama, T. Hasebe, A. Shirakura, T. Suzuki, Diamond Relat. Mater. 14 (2005) 1112.

[3] M. Yoshida, S. Watanabe, T. Tanaka, T. Takagi, M. Shinohara, J.W. Lee, Nucl. Instrum. Methods Phys. Res. B 206 (2003) 712.

[4] M. Yoshida, T. Tanaka, S. Watanabe, M. Shinohara, J.W. Lee, T. Takagi, Surf. Coat. Technol. 174/175 (2003) 1033.

[5] S. Watanabe, M. Shinohara, H. Kodama, T. Tanaka, M. Yoshida, T. Takagi, Thin Solid Films 420/421 (2002) 253.

[6] H. Kodama, A. Shinohara, A. Hotta, T. Suzuki, Surf. Coat. Technol. 201 (2006) 913.

[7] B.M. Henry, F. Dinelli, K.-Y. Zhao, C.R.M. Grovenor, O.V. Kolosov, G.A.D. Briggs, A.P. Roberts, R.S. Kumar, R.P. Howson, Thin Solid Films 355/356 (1999) 500.

[8] M. Benmalek, H.M. Dunlop, Surf. Coat. Technol. 76/77 (1995) 821. [9] A.S. da Silva Sobrinho, G. Czeremuszkin, M. Latreche, M.R. Wertheimer,

J. Vac. Sci. Technol. A 18 (2000) 149.

[10] F. Vaz, P. Cerqueira, L. Rebouta, S.M.C. Nascimento, E. Alves, Ph. Goudeau, J.P. Riviere, K. Psichow, J. de Rijk, Thin Solid Films 447/ 448 (2004) 449.

[11] M.H. Kazemeini, A.A. Berezin, N. Fukuhara, Thin Solid Films 372 (2000) 70.

[12] N. Martin, O. Banakh, A.M.E. Santo, S. Springer, R. Sanjines, J. Taka-doum, F. Levy, Appl. Surf. Sci. 185 (2001) 123.

[13] F. Vaz, P. Cerqueira, L. Rebouta, S.M.C. Nascimento, E. Alves, Ph. Goudeau, J.P. Riviere, Surf. Coat. Technol. 174/175 (2003) 197. [14] A. Bittar, D. Cochrane, S. Caughley, I. Vickeridge, J. Vac. Sci. Technol.

A15 (2) (1997) 223.

[15] M. Lazaror, P. Raths, H. Metager, W. Spirkl, J. Appl. Phys. 77 (5) (1995) 2133.

[16] J. Guillot, A. Jouaiti, L. Imhoff, B. Domenichini, O. Heintz, S. Zerkout, A. Mosser, S. Bourgeois, Surf. Interface Anal. 34 (2002) 577.

[17] S. Dreer, R. Krismer, P. Willhartitz, G. Friedbacher, Thin Solid Films 354 (1999) 43.

[18] M. Neuha¨user, S. Ba¨rwulf, H. Hilgers, E. Lugscheider, M. Riester, Surf. Coat. Technol. 116–119 (1999) 981.

[19] K. Koski, J. Ho¨lsa¨, P. Juliet, Surf. Coat. Technol. 120/121 (1999) 303. [20] S. Ben Amor, B. Rogier, G. Baud, M. Jacquet, M. Nardin, Mater. Sci. Eng.

B57 (1998) 28.

[21] R.S. Kumar, M. Auch, E. Ou, G. Ewald, C.S. Jin, Thin Solid Films 417 (2002) 120–126.

[22] A.P. Roberts, B.M. Henry, A.P. Sutton, C.R.M. Grovenor, G.A.D. Briggs, T. Miyamoto, M. Kano, Y. Tsukahara, M. Yanaka, J. Membr. Sci. 208 (2002) 75.