Surface & Coatings Technology 205 (2011) 2709–2713
Contents lists available at ScienceDirect
Surface & Coatings Technology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s u r f c o a t
Hydrogen permeation through TiAlN-coated Eurofer '97 steel Paul J. McGuiness a,⁎, Miha Čekada b, Vincenc Nemanič c, Bojan Zajec c, Aleksander Rečnik a a b c
Department for Nanostructured Materials, Jožef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia Department for Thin Films and Surfaces, Jožef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia Department for Surface Engineering and Optoelectronics, Jožef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia
a r t i c l e
i n f o
Article history: Received 12 April 2010 Accepted in revised form 31 August 2010 Available online 6 September 2010 Keywords: Eurofer '97 Hydrogen Permeation TiAlN
a b s t r a c t Any future fusion reactor will require a structurally sound vacuum chamber that is able to withstand the attack of hydrogen produced in the transmutation reactions, while at the same time exhibiting the rapid decay of any induced radioactivity. However, the Eurofer '97 steel, which has already been chosen for the DEMO reactor, although possessing reduced activation, remains very susceptible to hydrogen. In this study we have looked at the effectiveness of thin, TiAlN coatings with respect to the permeability of hydrogen at 400 °C. Our results reveal that the coating forms a columnar structure, with evidence of epitaxy at the substrate–coating interface, and that this coating can produce a permeation reduction factor for hydrogen of up to 20,000. This is substantially higher than any other coating reported for this type of steel. Furthermore, the relatively low costs associated with such films and the breadth of knowledge that already exists about their characteristics suggest that such a combination of a TiAlN coating and the Eurofer '97 steel could be a very promising material for reactor technologies. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Reduced-activation ferritic martensitic (RAFM) steels, such as Eurofer '97 (subsequently referred to in the text as Eurofer), will be used to form some of the structural parts of the test-blanket module of the ITER fusion reactor, scheduled to be switched on in 2018. It is also planned to incorporate this same material into the proposed demonstration fusion reactor DEMO, the first construction phase which is expected to last from 2024 to 2033. The reduced activation of Eurofer is achieved by substituting some of the conventional alloying elements in martensitic steels, i.e., Mo, Nb, and Ni, with elements that exhibit a faster decay of the induced radioactivity, i.e., Ta, W and V. The chemical composition of Eurofer was developed to achieve the best possible compromise between good metallurgical characteristics, comparable with, for example, Cr–Mo steels, and very reduced longterm radioactivity. However, one serious problem associated with steels of the Eurofer type is their susceptibility to hydrogen absorption, resulting in hydrogen embrittlement and a subsequent degradation of the mechanical properties. This has the potential to be a serious problem, as the austenitic stainless steel selected for the vacuum vessel of ITER is to be replaced in DEMO by Eurofer. The hydrogen is produced in transmutation reactions, e.g., (n,p), and because of this problem the permeation of hydrogen and its isotopes through ⁎ Corresponding author. Tel.: +386 1 477 3818. E-mail addresses:
[email protected] (P.J. McGuiness),
[email protected] (M. Čekada),
[email protected] (V. Nemanič),
[email protected] (B. Zajec),
[email protected] (A. Rečnik). 0257-8972/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2010.08.133
Eurofer under a wide range of conditions has received considerable attention in the past (see, for example, [1] and the references contained within). Unfortunately, the oxidation of Eurofer in the air does not lead to a dense and impermeable oxide film, like that which is readily formed on austenitic stainless steel. In addition to this, the hydrogen diffusivity (although not the solubility) in Eurofer is substantially higher than in austenitic steels, which might lead to the accumulation of tritium above a safe limit. With this in mind, some attempts to produce a reliable barrier material and deposit it on Eurofer have been reported in recent years. Dense alumina, only 1 μm thick, produced a permeation reduction factor (PRF) of 1000 at 800 °C, as reported by Levchuk et al. [2]. Erbium oxide, again 1 μm thick, has also been recognized as an efficient barrier, with a PRF of the order of 1000 [3]. An even more efficient barrier was made from Al–Cr–O, giving a PRF between 2000 and 3500 at 700 °C [4]. Transition-metal-nitride hard coatings, like that used in this investigation, have been employed since the early 1980s for the protection of tools. Such coatings are most commonly prepared by physical vapour deposition (PVD) at a typical deposition temperature of around 450 °C. Starting from the ubiquitous TiN, several ternary coatings have been developed, with the greatest usage probably achieved by TiAlN. This coating is distinguished by a high hardness (3300 HV), a good oxidation resistance (up to 850 °C), a chemical inertness and a low thermal conductivity [5]. In its basic form, a TiAlN coating has an atomic ratio of approximately Ti:Al = 50:50, a columnar microstructure and a B1 cubic structure. Several TiAlN-based coatings have been developed in the past 10 years or so, including AlTiN (with an increased share of aluminium) and nanolayer TiAlN/TiN. Though
2710
P.J. McGuiness et al. / Surface & Coatings Technology 205 (2011) 2709–2713
having a roughly similar stoichiometry to the classical TiAlN, their properties can be substantially different. Among the diffusion-related properties, oxidation resistance [6] and thermal stability [7] have been studied extensively. With the aim being to enhance the coating's performance in cutting operations, several options have been explored, such as multilayers or the addition of another element. However, in contrast to TiN, where there are many reports on its application as a diffusion barrier [8], the data on TiAlN is scarce. Of specific interest is the study by Man et al., who deposited TiAlN coatings on AISI 316L steel and measured the hydrogen permeation flux [9]. They obtained the lowest permeation flux of 1.2 · 10−11 mol cm−2 s−1 for an aluminium share x = 0.4. These permeation results combined with our experience in depositing dense films as hard coatings made us interested in testing the effectiveness of TiAlN barriers with respect to hydrogen when applied to Eurofer. In this work, we present the results of applying a TiAlN film of 5 μm, deposited on Eurofer samples using magnetron sputtering, which resulted in a PRF of 5800 to 20,000 at 400 °C. 2. Theory The experimental proof of a barrier's efficiency is the relative reduction of the steady permeation flux. Its definition is the ratio of the steady permeation rate through the uncoated membrane versus the steady permeation rate through the coated membrane, termed “the permeation reduction factor” (PRF): PRF =
juncoated : jcoated
ð1Þ
Both permeation rates must be obtained under identical conditions in terms of driving pressure and temperature. When the hydrogen migration through the planar, homogenous membrane is limited by hydrogen atom diffusion in the material, the steady-state permeation rate j follows the Richardson equation j=
P pffiffiffi p; d
ð2Þ
where P is the permeability coefficient, p is the hydrogen driving pressure and d is the membrane thickness. The coefficient P is a material property and is a product of the hydrogen diffusivity and solubility. Applying the Richardson equation to a composite (coated) planar membrane enables the calculation of an effective permeation coefficient. The effective permeability Peff. is based on the sum of the permeation resistances for each layer, analogous to electrical resistors in series. For a two-layered membrane it is then d d d = 1 + 2; P eff : P1 P2
ð3Þ
where the indices denote layers 1 and 2, while the membrane thickness is d = d1 + d2. The permeability coefficient of a coating material can thus be obtained from the known layer thicknesses and the steady-state permeation rates through the coated and uncoated membranes.
100-mm-diameter bar by wire electrical discharge machining 40-mm-diameter rods, which were then sliced using the same technique to produce 0.5-mm-thick discs. These discs were then ground and polished with diamond paste to obtain an optically flat surface, cleaned with a detergent in an ultrasonic bath, rinsed in demineralised water and dried in the air at 100 °C. The coatings were deposited in a CemeCon CC800/7 deposition system, which uses four unbalanced magnetron sources. The process consists of five steps: heating with resistive heaters, ion etching with argon ions in the RF mode, deposition (4 × 8 kW DC targets, −100 V bias), a second ion etching and a second deposition. The base pressure was 4 · 10−3 Pa and the working pressure during the deposition was 7 · 10−1 Pa. Argon, krypton and nitrogen were fed into the chamber during the deposition. Segmented targets were used, based on bulk titanium with aluminium plugs. The samples were mounted in a twofold rotation mode, which provided a total thickness of 5.0 μm and a deposition rate of about 2 μm/h. The deposition temperature was about 450 °C. For the cross-sectional transmission electron microscopy (TEM) observations the sample was prepared by cutting two slices from the coated Eurofer samples fixed in a face-to-face configuration within a 3-mm brass ring using acrylic resin as an adhesive. The brass ring containing the cross-section of our sample was then polished flat to approximately 120 μm and dimpled down to 20 μm at the disc centre (Dimple grinder, Gatan Inc., Warrendale, PA). The dimpled specimens were finally ion-milled (RES 010, Bal-Tec AG, Balzers) using 4 kV Ar+ ions at an incidence angle of 10° until perforation to obtain large transmissive areas for TEM investigations. TEM was performed using a 200 kV TEM (JEM-2100 FX, Jeol Inc., Tokyo, Japan) equipped with an energy-dispersive X-ray spectrometer (EDS) for the elemental composition analyses. Selected-area electron diffraction (SAED) over the multiple nanocrystals was preformed to obtain the characteristic diffraction rings with structure-specific d-values. To achieve reliable data on the permeation flux for hydrogen in a particular case, the general procedures and components applied should be selected according to the guidelines published in the “Recommended practices for measuring and reporting outgassing data” [10]. Since the permeation flux is an additional contribution to the total pressure on the downstream side, any of the two suggested methods may be applied: (i) the dynamic method or (ii) the gasaccumulation method. The latter was selected for this series of experiments since the achieved sensitivity can be made higher for two reasons: the accumulation chamber can be made small and the observation time can be extended. Its sensitivity limit is, in any case, determined by the hydrogen background outgassing from the system walls, where the major contribution comes from the walls at an elevated temperature [11]. The setup was assembled from all-metal UHV components, see Fig. 1, and kept at a stable temperature. The UHV in each chamber is achieved using its own turbo-molecular pump (TMP). The applied gauges were two capacitance manometers (MKS, 100 Pa F.S. (full scale), and 5 Pa F.S.) and a spinning rotor gauge
3. Experimental The base material on which we applied the coating was the reduced-activation ferritic martensitic (RAFM) steel Eurofer with the chemical composition (wt.%): 0.11 C, 8.7 Cr, 1.0 W, 0.10 Ta, 0.19 V, 0.44 Mn and 0.004 S, balance Fe. The steel was supplied in the form of a 100-mm-diameter bar by Forschungszentrum Karlsruhe GmbH, Karlsruhe, Germany, in the normalized-plus-tempered condition, i.e., 980 °C/110 min plus tempering at 740 °C/220 min/air-cooled. The disc-shaped samples for the coating experiments were cut from a
Fig. 1. Experimental setup for a determination of the hydrogen permeation flux through the membrane.
P.J. McGuiness et al. / Surface & Coatings Technology 205 (2011) 2709–2713
mounted in the accumulation chamber with a volume V = 0.44 L. The attachments were the calibrating volume and the hydrogen container for calibrating the quadrupole mass spectrometer (QMS). The QMS is built into the analytical chamber, which is rarely exposed to the air. This fact and the high pumping speed (S = 300 L/s) of the magnetically levitated TMP in series with the 60 L/s TMP makes it possible to achieve a base pressure of the order of 4 · 10−9 Pa. The 0.5-mm discs had a hydrogen-exposed area of A = 8.4 cm2, as determined by the I.D. (internal diameter) of the pressed Au gasket. The cell for fixing the membrane has a special construction that makes it possible to achieve a very low hydrogen background, since only thin-walled sections are exposed to the vacuum. A full description of our novel, thin-walled cell design can be found in Ref. [12]. The massive flanges necessary to exert the pressure on the seal by tightening the screws were not a part of the downstream chamber. The inserted plate and tube have a thickness of 0.3 mm and an area of approximately 65 cm2. The heating procedure for each disc membrane was a 2-hour linear increase to 400 °C followed by 22 h at this temperature. This procedure was sufficient for an uncoated membrane, but for the TiAlN-coated membrane it took several days to achieve a sufficiently low background outgassing. The pressure on the upstream side before dry hydrogen was introduced was in the range of 5 · 10− 6 Pa. At the end of the preparation stage, the achieved value of the total outgassing in the downstream chamber was between Q(400 °C) = 9 · 10−7 − 2 · 10−6 Pa L/s = 3 · 10−13 − 8 · 10−13 mol H2/s for the coated membrane. The lowest detectable permeation flux for a well outgassed membrane is thus j = 4 · 10−10 mol H2/(m2 s). Three uncoated membranes and one coated membrane with 5 μm of TiAlN were tested in several repeated cycles. These cycles consisted of: initial membrane outgassing, permeation measurement and repeated outgassing before identical or new parameters for the permeation were set. 4. Results and discussion Cross-section TEM provided valuable information on the microstructure and the phase composition of the coating. Qualitative EDS analysis confirmed the TiAlN composition of the film. Fig. 2 shows the TiAlN/Eurofer interfacial region. The interface between the substrate and the film is smooth and abrupt. TiAlN crystallites extend from the interface in the form of elongated, columnar grains. They become
2711
clearly visible after 100 nm or so. The width of these parallel grown columnar grains is 20–50 nm and they are up to 500 nm long. Close to the thin edge of the film we observe, in weak contrast, wavy horizontal bands, i.e., TiAlN growth zones. These bands are a consequence of the sample rotation. In the two-fold rotational geometry the sample travels along a complex trajectory within the deposition chamber, and thus periodically approaches towards and moves away from the target [13]. Although the targets are identical, a multilayer character becomes visible, as a consequence of the differences in the growth rate. The typical width of the bands is 10–15 nm, which corresponds to the amount of material deposited in each cycle. The SAED pattern recorded from a larger area of the TiAlN film shows the diffraction rings characteristic for a face-centred cubic TiNrelated structure with slightly larger d-spacings. Experimental dvalues are listed in Table 1 and compared with the literature data on structurally related TiN. All of the diffraction rings can be attributed to the cubic phase; therefore, we could exclude the possible presence of any hexagonal AlN- or TiN-related phase. Along the interface we can quite regularly observe a certain degree of epitaxy between the Fe grains in the substrate and the TiAlN grains in the film. One such situation is shown in the close-up in Fig. 2c, where (111) lattice planes of iron continue into (200) lattice of TiAlN. Because of the structural dissimilarity in the other directions a full-scale epitaxial growth cannot be expected in this system. The columnar texture of the TiAlN coating is retained throughout the film's cross-section all the way to the surface. Fig. 3 shows the surface area of the TiAlN film. The surface itself is smooth with a roughness comparable to the roughness of the substrate (Fig. 3a). In the ion-milled parts of the TiAlN surface we can clearly recognize the columnar grains and the growth zones, which are characteristic for the film's method of deposition (Fig. 3b). The measurements made on the uncoated membrane were performed to test the system's performance, reproducibility and to compare our data with previously published data. Since our system has a very high sensitivity, but the measurements with such a sensitivity last a long time, we decided to fix the temperature of the experiments at 400 °C. The permeation flow of hydrogen through the uncoated membranes was observed immediately on the 5 Pa FS CM (capacitance manometer) gauge after hydrogen was suddenly introduced to the upstream side. The steady flows at various values of the upstream pressure from pup = 60 – 120 kPa are given in Fig. 4 for
Fig. 2. Bottom part of the TiAlN film at the contact with the Eurofer substrate. (a) TiAlN grows in form of 20–50-nm-wide columnar grains that extend from the TiAlN/Eurofer interface. Growth zones are parallel with the interface. (b) SAED pattern recorded from the TiAlN film shows diffraction rings stemming from randomly oriented TiAlN crystallites. Measured d-values (some shown, all listed in Table 1) correspond to the TiN-related cubic structure. (c) A high-resolution TEM image of the TiAlN/Eurofer interface shows that columnar TiAlN grains in the film grow on the top of the Fe grains of the substrate in a semi-epitaxial relation.
2712
P.J. McGuiness et al. / Surface & Coatings Technology 205 (2011) 2709–2713
Table 1 Measured d-values [nm] for TiAlN from SAED pattern (from Fig. 2b).
Measured [nm] JCPDF #38-1420 Crystallogr. plane
D1
D2
D3
D4
D5
D6
D7
0.252 0.2449 {111}
0.217 0.2121 {200}
0.153 0.1450 {220}
0.132 0.1279 {311}
0.125 0.1224 {222}
0.109 0.1060 {400}
0.098 0.0973 {331}
an uncoated membrane. The measured flow is proportional to the square root of the upstream pressure, indicating a diffusion-limited permeation regime. The small differences between the flows through the three uncoated samples can be related to slight variations in the surface composition or to impurities in the hydrogen gas. These impurities were kept at a low level by leading the gas through a trap chilled by liquid nitrogen. The average time-lag of 6.5 s was determined from several repeated cycles giving us a diffusivity constant of D(400 °C) = 6.4 · 10−9 m2/s. The permeabilitypffiffiffiffiffi coefficient ffi was determined to be P(400 °C) = 1.7 · 10−11 mol H2/m s Pa and pffiffiffiffiffi ffi the −3 3 hydrogen solubility was S(400 °C) = 2.6 · 10 mol H2/m Pa. The obtained parameters are in very good agreement with published transport parameters for hydrogen and deuterium in Eurofer [14,15]. When testing the TiAlN-coated sample at 1 bar of H2 we observed that the permeation flux was both substantially lower and that the time to reach a maximum flow value was longer. The first exposure to 1 bar and the subsequent pressure change on the downstream side is given in Fig. 5. Three repeated exposures of the membrane to approximately 1 bar of hydrogen were conducted and the exposure lasted for some hours. Between those exposures the membrane was allowed to
Fig. 3. Top part of the TiAlN film. (a) Smooth surface embedded in epoxy resin visible as amorphous layer above the surface line. (b) Ion-milled surface region showing columnar TiAlN grains and distinct growth zones.
Fig. 4. Permeation rate versus square root of the upstream pressure from 62 kPa to 114 kPa.
outgas in UHV for some tens of hours, which was sufficient to restore the low background. Two values for the PRF were recorded within the first 4 days: 13,300 and 20,000. During this period the membrane was kept at 400 °C all the time. After 10 days of outgassing of both sides of the membrane and intermediate cooling down and warming up, the PRF dropped to 5800, as measured in the third exposure. At present, we have no evidence to unequivocally explain this degradation, but only minor changes to the surface or within the film may be responsible. An evaluation of a barrier's effectiveness is not simple and no unique criterion has been accepted [16]. However, using the PRF as a means of assessing the capability of a given film to act as a permeation barrier is still the most common parameter that enables such a comparison. It is generally determined in similar permeation devices, but at various temperatures and pressures. To compensate for inadequate setup sensitivity, authors have often tested coated membranes far above the temperature that is expected in any planned application within the DEMO project. This makes any PRF comparison and prediction of its long-term operation even less reliable. In any case, we made a comparison of our PRF value measured at a planned 400 °C against the highest published values of PRF for different barrier film materials deposited by various techniques.
Fig. 5. Pressure evolution and permeation flux obtained from pressure derivative versus time before and after introducing 83 kPa H2 at the upstream side. Hydrogen was introduced at t = 0.
P.J. McGuiness et al. / Surface & Coatings Technology 205 (2011) 2709–2713 Table 2 Comparison of the highest achieved PRF values for various barrier types applied on Eurofer. Material Deposition method
Thickness Evaluation PRF temperature
Ref.
Al2O3
1 μm
Filtered arc evaporation Filtered arc Er2O3 evaporation Al–Cr–O Pulse enhanced electron emission arc evaporation TiAlN Magnetron sputtering
1000
[2]
1 μm
700, 750 and 800 °C 400–700 °C
800
[3]
2 μm
600–700 °C
2000–3000
[4]
5 μm
400 °C
5800–20,000 This work
Table 2 summarizes the results on the barriers reported so far on the Eurofer substrate. Among the three published reports, all used arc evaporation for the deposition of the barrier film. One of the features of arc evaporation is the incorporation of droplets into the growing film, which may have a negative influence on the film's permeability. Nevertheless, the authors [2–4] applied two varieties of arc evaporation, where this problem is significantly reduced. In magnetron sputtering, applied in our case, the droplet problem is largely irrelevant and the resulting compact microstructure contributes to the permeability of the film. Therefore, it is not surprising that our tested TiAlN film exhibited the highest PRF of any reported barrier material applied so far to a Eurofer substrate. Despite the common usage of PRF to assess a film's capability, it is not the best quantity for an intercomparison since it depends on the substrate's properties (material and thickness) and the film thickness. When the permeation through coated membrane is assumed to be diffusion-limited, then the permeability coefficient P of the film is the proper quantity that is independent of substrate and dimensions. The permeation coefficient at 400 °C for our on the TiAlN p film pmeasurements ffiffiffiffiffiffi ffiffiffiffiffiffi give P TiAlN = 2.9 · 10−17 mol H2/m s Pa, 1.3 · 10−17 mol H2/m s Pa p ffiffiffiffiffi ffi and 8· 10−18 mol H2/m s Pa corresponding to the following PRF= 5800, 13,300 and 20,000. From Ref. [9] (the only publication on permeation measurements through a TiAlN film deposited on AISI 316L pffiffiffiffiffi ffi steel) we calculated the PTiAlN(550 °C)=4.5· 10−16 mol H2/m s Pa for the least permeable film investigated. The Refs. [2–4] used in Table 2 unfortunately do not provide sufficient data to calculate P in the temperature range close to 400 °C. The is Ref. [3], where PEr2O3 pffiffiffiffiffiexception ffi (600 °C)= 4.1· 10−16 mol H2/m s Pa could be calculated. From a rough extrapolation of Peff. for an Al2O3-coated membrane [2],pthe ffiffiffiffiffiffi permeability −16 coefficient P Al 2 O 3 (600 °C) ≈ 2 p · 10 mol H 2 /m s Pa and P Al 2 O 3 ffiffiffiffiffi ffi (400 °C) ≈ 4 · 10−17 mol H2/m s Pa could be extracted. The latter value is in good agreement with the PTiAlN obtained on our sample, pffiffiffiffiffiffi while the P values at ≈600 °C lie in the (2–5) · 10−16 mol H2/m s Pa range for TiAlN [9], Er2O3 [3] and Al2O3 [2], suggesting that all three coating materials could have similar effectiveness as a hydrogen permeation barrier. Our present approach seems a good choice, since TiAlN films are today deposited routinely as hard coatings. We also believe that
2713
the deposition parameters may be further optimized to fulfil the demands for barrier films as required at DEMO or in any other application at an even lower thickness. 5. Conclusions From our experiments involving the deposition of TiAlN coatings on Eurofer steel we can conclude the following. The coating is relatively straightforward to apply, forming a smooth interface with evidence of epitaxy. The resulting columnar structure, which extends over most of the 5-μm thickness, is sufficient to prevent the permeation of hydrogen through the coating–substrate combination almost completely. All the permeation measurements were performed at 400 °C and 1 bar upstream hydrogen pressure in the setup capable of detecting a permeation flux of the order of j = 4 · 10−10 mol H2/(m2 s). The recorded permeation reduction factor was between 5800 and 20,000—better than any value previously reported for Eurofer. A comparison of the calculated and extrapolated permeability coefficients reveals that Er2O3 and Al2O3 coatings might exhibit a similar permeation reduction at the same thickness. TiAlN films are thus another candidate material for the tritium permeation barrier in the DEMO reactor, as it is clear that they are able to suppress the hydrogen permeation through Eurofer very effectively. Acknowledgments The authors would like to thank Forschungszentrum Karlsruhe GmbH, Karlsruhe, Germany, for supplying the Eurofer, Medeja Gec for preparing the TEM samples, and the financial support of the EU 6FP ERA-NET.NMT “Hydrogen-impermeable nanomaterial coatings for steels (HY-nano-IM)” project. References [1] G.A. Esteban, A. Peña, F. Legarda, R. Lindau, Fus. Eng. Des. 82 (15–24) (2007) 2634. [2] D. Levchuk, F. Koch, H. Maier, H. Bolt, J. Nucl. Mater. 328 (2–3) (2004) 103. [3] D. Levchuk, S. Levchuk, H. Maier, H. Bolt, A. Suzuki, J. Nucl. Mater. 367–370 B (2007) 1033. [4] D. Levchuk, H. Bolt, M. Doebli, S. Eggenberger, B. Widrig, J. Ramm, Surf. Coat. Technol. 202 (2008) 5043. [5] S. PalDey, S.C. Deevi, Mater. Sci. Eng. A 342 (2003) 58. [6] S. Inoue, H. Uchida, Y. Yoshinaga, K. Koterazawa, Thin Solid Films 300 (1997) 171. [7] L. Hultman, Vacuum 57 (2000) 1. [8] M.Y. Kwak, D.H. Shin, T.W. Kang, K.N. Kim, Thin Solid Films 339 (1999) 290. [9] B.Y. Man, L. Guzman, A. Miotello, M. Adami, Surf. Coat. Technol. 180–181 (9) (2004) 9. [10] J.E. Blessing, R.E. Ellefson, B.A. Raby, G.A. Bruckner, R.K. Waits, J. Vac. Sci. Technol. A 25 (2007) 167. [11] K.S. Forcey, D.K. Ross, J.C.B. Simpson, D.S. Evans, J. Nucl. Mater. 160 (1988) 117. [12] V. Nemanič, B. Zajec, M. Žumer, J. Vac. Sci. Technol. A 28 (4) (2010) 578. [13] M. Panjan, M. Čekada, P. Panjan, A. Zalar, T. Peterman, Vacuum 82 (2008) 158. [14] G.A. Esteban, A. Peña, I. Urra, F. Legarda, B. Riccardi, J. Nucl. Mater. 367–370 (2007) 473. [15] D. Levchuk, F. Koch, H. Maier, H. Bolt, J. Nucl. Mater. 328 (2004) 103. [16] A. Perujo, K.S. Forcey, Fus. Eng. Des. 218 (1995) 224.