The effect of hydrogen on the growth of the nitrided layer in r.f.-plasma-nitrided austenitic stainless steel AISI 316

The effect of hydrogen on the growth of the nitrided layer in r.f.-plasma-nitrided austenitic stainless steel AISI 316

Surface and Coatings Technology 123 (2000) 29–35 www.elsevier.nl/locate/surfcoat The effect of hydrogen on the growth of the nitrided layer in r.f.-p...

201KB Sizes 1 Downloads 72 Views

Surface and Coatings Technology 123 (2000) 29–35 www.elsevier.nl/locate/surfcoat

The effect of hydrogen on the growth of the nitrided layer in r.f.-plasma-nitrided austenitic stainless steel AISI 316 S. Kumar a,1, M.J. Baldwin a, M.P. Fewell a, *, S.C. Haydon a, K.T. Short b, G.A. Collins b, J. Tendys b a Division of Physics and Electronics Engineering, The University of New England, Armidale, NSW 2351, Australia b Australian Nuclear Science and Technology Organisation, Menai, NSW 2234, Australia Received 1 December 1998; accepted in revised form 6 July 1999 The remaining authors dedicate this paper to the memory of Emeritus Professor Syd Haydon, who died suddenly on 15th October 1998.

Abstract Austenitic stainless steel AISI 316 has been nitrided by low-temperature r.f. plasmas containing various nitrogen–hydrogen gas mixtures, in order to study the effect of hydrogen on the growth of the ‘expanded austenite’ layer. The layers thus produced have been characterised using cross-sectional scanning electron microscopy, X-ray diffraction and instrumented microhardness measurements; the plasmas were studied by optical emission spectroscopy. The treated layer shows higher resistance to acid etching than the bulk material. Provided that the partial pressure of nitrogen is held constant, the addition of hydrogen at concentrations in the range 5–50% results in thicker nitrided layers and enhanced surface hardness compared with treatments in pure nitrogen. An excessive amount of hydrogen (~75%), on the other hand, retards the nitriding process. Optical spectroscopy indicates that the addition of hydrogen does not increase the concentration of active nitriding species, although mass spectroscopy shows the presence of NH . The beneficial effect of hydrogen is therefore due to the action of hydrogen atoms and molecules 1–4 at the surface of the workpiece. © 2000 Elsevier Science S.A. All rights reserved. Keywords: Hydrogen; Plasma nitriding; Steel

1. Introduction Austenitic stainless steels are used widely in the chemicals and food industries owing to their excellent corrosion resistance. However, the relatively poor wear resistance and hardness of these materials need to be improved for further diversification of their applications. It is well known that plasma nitriding can achieve just such an improvement: d.c. plasma nitriding is commercially successful in hardening many types of steel. However, this process involves heating the workpiece to temperatures in excess of 500°C, and such temperatures degrade the corrosion resistance of austenitic stainless steels, owing to a precipitation of CrN that removes Cr from solid solution [1,2]. Among different variants of the plasma nitriding technique, r.f. nitriding has been shown to be effective in hardening austenitic stainless

* Corresponding author. 1 Present address: Ian Wark Research Institute, University of South Australia, The Levels Campus, Mawson Lakes, SA 5095, Australia.

steels at low temperature, thus avoiding degradation of corrosion resistance [3–9]. However, the lower treatment temperature results in reduced nitriding efficiency compared with higher-temperature treatments. That is, for a given treatment time, the nitrided layer produced at lower temperature is thinner. It is very common to nitride steels in a nitrogen– hydrogen mixture [10,11], with hydrogen concentrations varying from trace amounts [3] to over 70% [12–16 ]. However, there appear to have been few studies explicitly examining the degree to which adding hydrogen improves nitriding efficiency. In the context of nitriding AISI 316, just two previous studies are known to us; one concludes that hydrogen is beneficial [1] and the other that the addition of hydrogen has no effect [17]. The only other related work known to us is a theoretical study of nitriding mechanisms in austenitic stainless steels [18], which concludes that small amounts of adsorbed hydrogen are necessary for the process. Here, we report a study of low-temperature r.f. nitriding of austenitic stainless steel AISI 316 carried out using

0257-8972/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved. PII: S0 2 5 7- 8 9 7 2 ( 9 9 ) 0 0 39 3 - X

30

S. Kumar et al. / Surface and Coatings Technology 123 (2000) 29–35

plasmas with a range of nitrogen–hydrogen mixtures. The samples thus processed were studied for their structural and mechanical properties. This investigation shows that a significant improvement in nitriding efficiency can be obtained for this treatment process and material over a reasonably wide range of hydrogen concentrations.

2. Experimental details In order to facilitate comparison, we adhered closely to treatment conditions and diagnostic techniques used previously [6–8,19,20]. Essentially the only change in the present work was the addition of hydrogen to the process gas. To summarise the treatment process briefly: the samples consisted of discs of AISI 316 stainless steel 25 mm in diameter and 4 mm in thickness, surface ground and polished to an average surface roughness R <5 nm, as measured with a Tencor Alpha-Step 200 a stylus profilometer. They were ultrasonically cleaned in ethanol before loading onto a heated sample table in a turbomolecular-pumped stainless-steel vacuum chamber, which also contained a single-turn inductive antenna [20]. The sample table was grounded to the chamber walls. The sample was preheated to 400°C in vacuum, after which the chamber was filled with the nitrogen– hydrogen mixture of the desired composition. Total filling pressure lay in the range 150–600 mPa, as detailed in Section 3. A diffuse r.f. glow discharge, filling the chamber, was formed by applying r.f. power to the antenna. Samples were exposed to this plasma for 3.0 h. During treatment, the sample temperature was measured with both a thermocouple and an IR pyrometer viewing through a quartz window, and was controlled within ±10°C. The treatment conditions are summarised in Table 1. After the treatment, the samples were allowed to cool in vacuum. A total of ten treatments were carried out, using nitrogen–hydrogen gas mixtures in the range 0–75% (by pressure) hydrogen. As usual in nitriding, we used flowing gas, and this required an allowance for the effects of differential pumping: the pumping speed of

our system for hydrogen is more than twice that for nitrogen. We calibrated partial pressures prior to the treatments by admitting the gases separately and noting the relationship between flow rate and pressure. Plasma conditions were monitored with optical spectroscopy. Langmuir probe data cannot meaningfully be analysed in gas mixtures, but an indication of our plasma conditions is given by results from a pure nitrogen discharge at 150 mPa and 300 W r.f. power. The plasma potential is close to 30 V. The ions are essentially at room temperature. The electron energy distribution is not Maxwellian: the bulk of the electrons have a temperature of ~2.5 eV with a small fraction at rather higher temperature. The ion number density is 1.2×109 cm−3, giving a Child–Langmuir width for a planar sheath above a grounded substrate of 1.7 mm. This is the likely order of magnitude of the sheath thickness above the workpiece. The mean free path at 150 mPa, 300 K is of order 0.6 m, comparable with the dimensions of the reactor. The samples were diagnosed by X-ray diffraction ( XRD), instrumented microhardness measurements and cross-sectional scanning electron microscopy (SEM ). Glancing-angle XRD spectra were obtained with a Siemens D500 diffractometer using Co Ka radiation at an incident angle of 2°. Under these conditions, the incident X-ray beam will be 95% attenuated at a depth of 1.2 mm in AISI 316 [21]. Microhardness measurements were made using a Nano Instruments IIs indenter [19] at a load of 50 mN. The SEM was performed using a JEOL JSM-5800LV instrument in the secondaryelectron-image mode. For this, the samples were prepared in the same manner as in our previous work [8]: transversally cut slabs of the treated material were cast in an epoxy resin and the sectioned surface was ground and polished. This was followed by chemical etching for 10 s in Marble’s solution (10 g copper sulphate in 100 ml of 6 M hydrochloric acid), in order to delineate the nitrided layer. Finally, the etched samples were carbon coated to reduce the charging that generally occurs during SEM of metallic samples cast in insulating materials.

3. Results Table 1 R.f. plasma nitriding conditions Base pressure Process gases Nitrogen partial pressure Hydrogen partial pressure Nitrogen flow rate Hydrogen flow rate Treatment temperature R.f. power and frequency Treatment time

5–100 mPa N (99.99%), H (99.99999%) 2 2 38–150 mPa 0–450 mPa 3.3–13.3 mmol s−1 (4.5–18.1 sccm) 0–73.5 mmol s−1 400±10°C 300 W at 13.56 MHz 3.0 h

Initially, we maintained a constant total pressure of 150 mPa as the proportion of hydrogen was varied. This series of treatments showed negligible change in nitriding efficiency as the percentage of hydrogen was increased. However, the interpretation of this is complicated by the nitrogen dilution effect [22,23]. To avoid this, we then compared treatments in which the partial pressure of nitrogen was held constant at 150 mPa, so that the total pressure was increased by the addition of hydrogen. For example, 75% hydrogen corresponded to a total gas

S. Kumar et al. / Surface and Coatings Technology 123 (2000) 29–35

31

pressure of 600 mPa. All results presented below come from this second series of treatments. 3.1. Scanning electron microscopy — layer thickness Fig. 1(a) shows a cross-sectional scanning electron micrograph of a sample treated in pure nitrogen. The surface exposed to the plasma during the treatment is on the right. The very bright band at the sample–epoxy boundary is due to charging under the electron beam. The nitrogen-containing layer is the homogeneous band of uniform thickness with a distinct shading, less stippled than that of the bulk material. The image clearly shows a layer thickness of about 1.3 mm. The superior corrosion properties of the treated layer are demonstrated by its greater resistance to chemical etching compared with that of the bulk material. Fig. 1(b) shows a similar micrograph of a sample treated in a mixture containing 15% hydrogen, but with the same nitrogen partial pressure as in Fig. 1(a). The treated layer is much thicker (about 2.9 mm) than that shown in Fig. 1(a). This clearly demonstrates that the addition of hydrogen to the nitriding plasma can accelerate the nitriding process, resulting in thicker nitrogencontaining layers in a given treatment time. The variation of treated-layer thickness as a function of percentage hydrogen in the nitriding plasma is shown in Fig. 2. The thickness of the treated layer is a maximum at about 15% hydrogen (total pressure of 176.5 mPa), but hydrogen concentrations in the range 5–50% give a useful enhancement of nitriding efficiency. It is clearly possible to have too much hydrogen: at 75% (total

Fig. 1. Cross-sectional SEM micrographs of AISI 316 stainless-steel samples nitrided for 3.0 h at 400°C using (a) pure nitrogen (150 mPa) and (b) a gas mixture containing 15% hydrogen (150 mPa N plus 2 22.5 mPa H ). Both micrographs have the same scale. The bright band 2 on the right is due to charging at the boundary between the sectioned surface and the epoxy resin.

Fig. 2. Variation in the thickness of the modified layer of AISI 316 stainless-steel samples nitrided for 3.0 h at 400°C with the percentage of hydrogen in the gas mixture. The partial pressure of nitrogen was held at 150 mPa in all treatments.

pressure of 600 mPa), the treated layer is thinner than that with pure nitrogen at 150 mPa. 3.2. X-ray diffraction Glancing-angle XRD spectra give information on the structure of the nitrided layer that supports the broad picture from the SEM studies. These spectra are shown in Fig. 3, progressively offset for clarity. They show the presence of the nitrogen-containing ‘expanded austenite’ phase in all of the nitrided samples. The top-most spectrum corresponds to plasma treatment in pure nitrogen. This is typical of the nitrogen-containing f.c.c. phase, with a set of broad expanded-austenite peaks ( labelled c ) appearing to the left of each austenite peak N ( labelled c). The lattice planes corresponding to the peaks are also labelled in the spectra. Although the intensity-weighted average depth of the layers producing the c peaks in Fig. 2 (the ‘information depth’ defined N by Delhez et al. [24]) varies between 0.27 and 0.39 mm, diffracted X-rays are detected from the whole expanded austenite layer. The expanded austenite peaks are strongly broadened and show an asymmetry toward higher diffraction angles, suggesting a nitrogen concentration gradient in the treated layer [24]. This effect is most pronounced for the samples treated either in pure nitrogen or with 75% hydrogen, implying that the samples nitrided with 5–50% hydrogen are relatively ordered and exhibit a smaller nitrogen concentration gradient. It may also be noted that the degree of expansion is less at 75% hydrogen than under any of the other conditions. ( The lattice parameter a for the expanded-austenite peaks is 0.375 nm with 75% hydrogen and 0.381 nm for the other treatment regimes, compared with 0.361 nm for the

32

S. Kumar et al. / Surface and Coatings Technology 123 (2000) 29–35

increases the degree of coherence in the expanded austenite structure.

3.3. Microhardness

Fig. 3. Glancing-angle (2°) XRD spectra of AISI 316 stainless samples obtained after treatment for 3.0 h at 400°C in gas mixtures containing the indicated concentrations of hydrogen, with the nitrogen partial pressure held at 150 mPa. The spectra are progressively offset along the ordinate for clarity. The diffraction peaks are labelled with their origin and Miller indices: c, austenite; c , expanded austenite. N

austenite substrate.) This can be attributed to a lower concentration of nitrogen in the treated layer. The austenite (c) peaks in Fig. 3 are diffracted from the bulk material below the expanded austenite layers. Although the incident X-rays are severely attenuated when they arrive at the underlying austenite, they are diffracted from a highly coherent structure compared with the expanded austenite layer, so the c diffraction peaks are much more readily observed. The intensities of the c peaks are roughly correlated with the thickness of the expanded austenite layer, with thicker layers causing reduced diffraction intensity from the underlying material. It is not clear, however, why the addition of 5% hydrogen should lead to such a drastic reduction in the intensity of the c peaks while only causing a slight increase in the layer thickness. Like those in Fig. 2, the data in Fig. 3 suggest that the addition of 5–50% of hydrogen to an otherwise pure nitrogen plasma increases the nitriding efficiency; that is, an increased amount of the expanded-austenite phase is obtained for a given treatment time. The data of Fig. 3 also suggest that the addition of hydrogen

To study the effect of added hydrogen on the mechanical properties of treated samples, instrumented microhardness measurements were made at a load of 50 mN. The results are shown in Fig. 4; the microhardness value of an untreated sample is also shown for comparison. Each point shown in Fig. 4 is the average of five measurements at different places on a sample. The uncertainties shown are derived from the variation in the five values, and so include a component arising from the variation in hardness over the surface. As expected in view of the thickness and XRD data, the hardness values peak in the hydrogen concentration range 5– 50%, and the hardness of the sample treated in 75% hydrogen is no greater than that treated in pure nitrogen. We have previously reported the variation of microhardness with gas pressure for the case of treatments in pure nitrogen [7]. At nitrogen pressures of 150 mPa and above, the hardness of the treated layer is essentially independent of pressure, but pressures below 150 mPa lead to a significant reduction in hardness. This correlates with results from our first series of treatments, in which total pressure was held constant as the proportion of hydrogen was increased. These showed little change in microhardness with increasing nitrogen concentration; the beneficial effects of the hydrogen were being countered by the reduced availability of nitrogen. This reinforces earlier work [22,23] showing the importance of maintaining the concentration of active nitrogen species in a nitriding plasma.

Fig. 4. Variation of microhardness of AISI 316 stainless-steel samples nitrided for 3.0 h at 400°C with the percentage hydrogen in the gas mixture. The broken line shows the hardness of an untreated sample.

S. Kumar et al. / Surface and Coatings Technology 123 (2000) 29–35

3.4. Plasma spectroscopy Spectra were recorded in the wavelength range 300– 700 nm, although there was heavy attenuation below about 310 nm. The spectrometer viewed a region not far above the sample table, but beyond the estimated sheath thickness of ~2 mm. Significant differences in the concentrations of atomic species between the cathode fall and the negative glow have been observed in nitrogen glow discharges at pressures of several pascals [25–27]. However, such concentration gradients are much less marked, or even absent, at 150 mPa [26 ]. This presumably reflects the relatively long molecular mean free path at 150 mPa. From these considerations, it is expected that the species concentrations inferred from the spectrometer observations also reflect concentrations at the workpiece. Fig. 5 shows the variation with hydrogen concentration of the intensities of several spectral lines. The values shown in Fig. 5 are raw peak heights; there was no attempt to calibrate the relative efficiency of the spectrometer. However, it should be noted that the intensities shown for the atomic nitrogen line at 411.0 nm are scaled up by a factor of 10, for the line was barely discernible above the noise even in pure nitrogen. This is consistent with previous observations in pure-nitrogen glow discharges [27] and also with the results of calculations for pure-nitrogen r.f. plasmas [28]. It is clear that spectral emission from both molecular nitrogen species falls with the addition of hydrogen. However, the initial decrease in line intensity is relatively minor; it is not until the hydrogen concentration reaches ~50% that the line-intensity reduction for N+and N 2 2 becomes a significant fraction of the intensity observed

Fig. 5. Variation of the peak heights of spectral lines with the percentage of hydrogen in the gas mixture. The partial pressure of nitrogen was 150 mPa in all cases. The lines shown are as follows: N , 337.1 nm; 2 N+ , 391.4 nm; H, 656.2 nm (H ); N, 411.0 nm. The peak height of 2 a the N line was scaled up by a factor of 10. There was no evidence for spectral emission from Fe, N+ or NH.

33

with 0% hydrogen. The only spectral features which become more intense with increasing hydrogen concentration are, not unexpectedly, those from hydrogen. Hydrogen molecular lines in this wavelength range are difficult to identify because of interference from the first positive system of nitrogen. However, the H and H a b atomic lines are clearly evident, and both show the same behaviour. There is no spectral evidence for lines from Fe, consistent with the very low level of sputtering. There is also no evidence of NH lines at either 324.0 or 336.0 nm. (The region of 336 nm is difficult because of the intense band from N with a head at 337.1 nm, but 2 there is no evidence of change in spectral shape with the addition of hydrogen. The region of 324 nm is entirely clear of lines.) Nevertheless, a mass spectrum taken with 50% hydrogen showed all the species NH , with NH and NH being the most intense. 1–4 2 3 ( With pure nitrogen, there were no peaks in the mass spectrum at masses 16–18, indicating negligible levels of oxygen in our system.) Finally, there is no evidence of N+ spectral lines; the locations of several strong lines (444.7, 463.1, 500 and 566.7 nm) from this species being completely clear.

4. Discussion Figs. 2–4 indicate that, when treating AISI 316 stainless steel in a low-pressure r.f. plasma with 150 mPa of nitrogen, the addition of 50 mPa of hydrogen (25% H 2 concentration) gives significantly improved nitriding efficiency, with good improvements in efficiency being obtained for hydrogen concentrations in the range 5– 50%. Previous studies on the nitriding of this material employed different types of plasma from the present study. Zhang and Bell [1], using a conventional highertemperature d.c. discharge process, concluded that 20% H was too much, with 4–10% giving optimum results. 2 On the other hand, Renevier et al. [17], who nitrided AISI 316 in the afterglow of a high-current low-pressure Ar–N arc, found no advantage from the addition of 2 hydrogen. The same seems to be true [29] of processes such as plasma-immersion ion implantation [4,5,19,20], in which the workpiece is biased to several tens of kilovolts. The variety of these findings could be interpreted as arising from different plasma conditions. However, the effect of nitrogen dilution must be borne in mind. Although not explicitly stated, it would appear that Zhang and Bell maintained essentially constant pressure while adding hydrogen. Hence, nitrogen dilution probably led to their result that 20% hydrogen is too high for optimum nitriding, notwithstanding the fact that they used pressures of order 1000 times greater than those used in the present study. In contrast, Renevier et al.

34

S. Kumar et al. / Surface and Coatings Technology 123 (2000) 29–35

Table 2 Room-temperature reaction coefficients k for the quenching by hydrogen of selected excited states of N and other species 2 Species

k (mm3 s−1)

Reference

N ( X, v≥5) 2 N (A) 2 N (a∞) 2 N (a) 2 N+ 2 N (2D)

0.002 0.0024 26 200 2000 2.2

[43] [44] [45] [46 ] [37] [47]

used a constant nitrogen partial pressure of 240 mPa. Their conclusion that hydrogen has little effect therefore points to a significant difference between r.f. plasma nitriding and nitriding by exposure to a discharge afterglow. This presumably reflects differences in the nitriding mechanism in these two situations, and deserves further investigation. As to optical spectroscopy, there have been many studies of nitrogen and N –H plasmas, involving a 2 2 wide variety of discharge types. The species active in plasma nitriding have not yet been identified [11,30], but there is essentially universal agreement that the concentrations of almost all likely candidates for the active species fall with increasing hydrogen concentration above 10%, regardless of the type of discharge. This includes both electronically and vibrationally excited N [22,31–37], N atoms and atomic ions [11,34– 2 36 ], and metastable N (2D) atoms [36 ]. Nitrogen molecular ions are particularly strongly quenched by even low levels of hydrogen [17,34,35,37,38]. The evidence in regard to NH is varied: several studies report decreased concentrations at high percentages of hydrogen [31,34,35,37], but increased concentrations were observed in the afterglow of a microwave discharge [37]. The only other species, apart from hydrogen itself [11,17,35–37,39,40], that has been observed to benefit from high concentrations of hydrogen is ammonia [39,41], the concentration of which peaks at about 50% hydrogen. Most of the studies cited above involved d.c. glow discharges, but r.f. [40], microwave [37,40] and arc [17] discharges are included. Few of these studies report in detail the nature of the electrodes or other materials in contact with the plasma. In view of the observed strong response of AISI-316 electrodes to the products of ionisation growth [42], the possibility of particle–surface interactions that in turn influence the species in the gas phase should not be overlooked. The widely observed quenching effect of high concentrations of hydrogen for almost all species likely to be active in the nitriding process correlates with known reaction coefficients, a selection of which is listed in Table 2. For example, the very strong quenching of N+ ions is due to the high reaction coefficient for their 2

hydrogenation [37]. Detailed calculations of reaction kinetics confirm the strong quenching effect of hydrogen [48]. This effect is in addition to the nitrogen-dilution effect present in those studies for which the total pressure was held constant as the percentage of hydrogen was increased. Our data ( Fig. 5) do not show the dramatic decrease in the concentration of N+ with increasing 2 percentage of hydrogen shown by other studies. This may be due to our lower total pressure (≤1 Pa) compared with pressures typical of d.c. glow discharges (~100–1000 Pa), since reaction rates are the product of reaction coefficients and the partial pressures of the reactants. (If the production rate of a species can be maintained, its equilibrium concentration will increase as its destruction rate is reduced.) It is interesting that the behaviour of N+ ions shown in Fig. 5 is similar to 2 that observed in the only other low-pressure study [17]. Overall, the wide variety of optical spectroscopic studies carried out on N –H plasmas suggest that the 2 2 addition of hydrogen beyond 10% does not aid nitriding through increased production of active species. The observation in the present study of enhanced nitriding efficiency with 15–50% added hydrogen therefore suggests a dominant role for interactions at the workpiece surface.

5. Conclusions Low-pressure r.f. plasmas in nitrogen–hydrogen gas mixtures can be used to nitride austenitic stainless steels at low temperatures (~400°C ). The treated layer shows higher resistance to acid etching than the bulk material. The addition of hydrogen in the range 5–50% to an otherwise pure nitrogen plasma results in an increased nitrided-layer thickness, provided that the partial pressure of nitrogen is maintained. Surface microhardness is also increased correspondingly. An excessive amount of hydrogen (~75%) in the gas mixture retards the nitriding process, resulting in a reduced nitrided layer thickness and no gain in hardness compared with a treatment in pure nitrogen. The evidence of optical spectroscopy suggests that the addition of hydrogen does not increase the concentration of active nitriding species, indeed the reverse. Hence, an explanation for the beneficial effect of hydrogen must be sought in the action of hydrogen atoms and molecules at the workpiece surface.

Acknowledgements This research is supported by the Australian Research Council, the Australian Institute of Nuclear Science and Engineering and by an Australian Postgraduate Research Award.

S. Kumar et al. / Surface and Coatings Technology 123 (2000) 29–35

References [1] Z.L. Zhang, T. Bell, Surf. Eng. 1 (1985) 131–136. [2] A.M. Staines, Heat Treat. Met. 17 (1990) 85–92. [3] F. El-Hossary, F. Mohammed, A. Hendry, D.J. Fabian, Z. Szaszne-Csih, Surf. Eng. 4 (1988) 150–154. [4] G.A. Collins, R. Hutchings, K.T. Short, J. Tendys, X. Li, M. Samandi, Surf. Coat. Technol. 74–75 (1995) 417–424. [5] S. Leigh, M. Samandi, G.A. Collins, K.T. Short, P. Martin, L. Wielunski, Surf. Coat. Technol. 85 (1996) 37–43. [6 ] M.J. Baldwin, G.A. Collins, M.P. Fewell, S.C. Haydon, S. Kumar, K.T. Short, J. Tendys, Jpn. J. Appl. Phys. 36 (1997) 4941–4998. [7] M.J. Baldwin, M.P. Fewell, S.C. Haydon, S. Kumar, G.A. Collins, K.T. Short, J. Tendys, Surf. Coat. Technol. 98 (1998) 1187–1191. [8] M.J. Baldwin, S. Kumar, J.M. Priest, M.P. Fewell, K.E. Prince, K.T. Short, Thin Solid Films 345 (1999) 108–112. [9] J.M. Priest, M.J. Baldwin, M.P. Fewell, S.C. Haydon, G.A. Collins, K.T. Short, J. Tendys, Thin Solid Films 345 (1999) 113–118. [10] K.-T. Rie, E. Menthe, A. Matthews, K. Legg, J. Chin, MRS Bull. 21 (1996) 46–51. [11] A. Ricard, J. Phys. D 30 (1997) 2261–2269. [12] M. Hudis, J. Appl. Phys. 44 (1973) 1489–1496. [13] A.M. Staines, T. Bell, Thin Solid Films 86 (1981) 201–211. [14] A. Grill, D. Itzhak, Thin Solid Films 101 (1983) 219–222. [15] J. Michalski, Surf. Coat. Technol. 59 (1993) 321–324. [16 ] A. Leyland, D.B. Lewis, P.R. Stevenson, A. Matthews, Surf. Coat. Technol. 62 (1993) 608–617. [17] N. Renevier, T. Czerwiec, P. Collignon, H. Michel, Surf. Coat. Technol. 98 (1998) 1400–1405. [18] A. Szasz, D.J. Fabian, A. Hendry, Z. Szaszne-Csih, J. Appl. Phys. 66 (1989) 5598–5601. [19] R. Hutchings, K.T. Short, J. Tendys, Surf. Coat. Technol. 83 (1996) 243–249. [20] G.A. Collins, R. Hutchings, K.T. Short, J. Tendys, C.H. Van Der Valk, Surf. Coat. Technol. 84 (1996) 537–543. [21] Computer code X0hWin, available from XraySite.Com, 38881 Blackford Ave, San Jose, CA 95117, USA. [22] J. Bougdira, G. Henrion, M. Fabry, J. Phys. D 24 (1991) 1076–1080.

35

[23] K.S. Fancey, A. Leyland, D. Egerton, D. Torres, A. Matthews, Surf. Coat. Technol. 76–77 (1995) 694–699. [24] R. Delhez, Th.H. de Keijser, E.J. Mittemeijer, Surf. Eng. 3 (1987) 331–342. [25] A. Leyland, K.S. Fancey, A.S. James, A. Matthews, Surf. Coat. Technol. 41 (1990) 295–304. [26 ] K.S. Fancey, A. Matthews, Vacuum 41 (1990) 2196–2200. [27] K.S. Fancey, Vacuum 46 (1995) 695–700. [28] I.J. Donnelly, E.K. Rose, in: R. McWilliams ( Ed.), Proceedings 8th Topical Conference on RF Power in Plasmas, Irvine, CA, 1989, AIP, New York, 1989, pp. 410–411. [29] K.T. Short, G.A. Collins, personal communication. [30] H. Michel, T. Czerwiec, M. Gantois, D. Ablitzer, A. Ricard, Surf. Coat. Technol. 72 (1995) 103–111. [31] L. Petitjean, A. Ricard, J. Phys. D 17 (1984) 919–929. [32] A. Ricard, Rev. Phys. Appl. 24 (1989) 251–256. [33] J. Bougdira, G. Henrion, M. Fabry, M. Remy, J.R. Cussenot, Mater. Sci. Eng. A 139 (1991) 15–19. [34] K. Rusna´k, J. Vlcek, J. Phys. D 26 (1993) 585–589. [35] B. Kulakowska-Pawlak, W. Zyrnicki, Thin Solid Films 230 (1993) 115–120. [36 ] J. Amorim, G. Baravian, A. Ricard, Plasma Chem. Plasma Process. 15 (1995) 721–731. [37] S. Bockel, J. Amorim, G. Baravian, A. Ricard, P. Stratil, Plasma Sources Sci. Technol. (1996) 567–572. [38] S.D. Popa, P. Chiru, L. Ciobotaru, J. Phys. D 31 (1998) L53–L55. [39] J. Amorim, G. Baravian, S. Bockel, A. Ricard, G. Sultan, J. Phys. (Fr.) III 6 (1996) 1147–1155. [40] B.N. Ganguly, P. Bletzinger, J. Appl. Phys. 82 (1997) 4472–4476. [41] J. Amorim, G. Baravian, G. Sultan, Appl. Phys. Lett. 68 (1996) 1915–1917. [42] M.J. Baldwin, S. Kumar, M.P. Fewell, S.C. Haydon, G.A. Collins, K.T. Short, J. Tendys, in: Proceedings 23rd International Conference on Phenomena in Ionized Gases, Toulouse Vol. 4 (1997) 195. [43] L.G. Piper, J. Chem. Phys. 97 (1992) 270–275. [44] M.F. Golde, Int. J. Chem. Kinet. 20 (1988) 75–92. [45] L.G. Piper, J. Chem. Phys. 87 (1987) 1625–1629. [46 ] W.J. Marinelli, W.J. Kessler, B.D. Green, W.A.M. Blumberg, J. Chem. Phys. 90 (1989) 2167–2173. [47] K. Schofield, J. Phys. Chem. Ref. Data 8 (1979) 723–798. [48] A. Garscadden, R. Nagpal, Plasma Sources Sci. Technol. 4 (1995) 268–280.