Plasma immersion ion implantation of steels

Plasma immersion ion implantation of steels

Materials Science and Engineering, A 139 ( 1991 ) 171 - 178 171 Plasma immersion ion implantation of steels G. A. Collins, R. Hutchings and J. Tendy...

632KB Sizes 1 Downloads 41 Views

Materials Science and Engineering, A 139 ( 1991 ) 171 - 178

171

Plasma immersion ion implantation of steels G. A. Collins, R. Hutchings and J. Tendys Australian Nuclear Science and Technology. Organisation, Lucas Heights Research Laboratories, Private Mail Bag 1, Menai, NS W 2234 (,4 ustralia)

Abstract Plasma immersion ion implantation (PI3) is a new technique with certain advantages over conventional ion implantation. We have developed an implanter based on an inductively coupled r.f. glow discharge plasma. Ion densities of 10 ~° cm 3 are obtained with filling pressures of about 10 -3 mbar. Ions are accelerated from the plasma by high voltage pulses (typically - 4 5 kV) applied directly to the workpiece. In this paper we report on the application of PI 3 to nitrogen implantation in steels. Dramatic increases in microhardness and wear resistance have been observed for a number of steels, ranging from 0.3 wt.% C mild steel to austenitic stainless steels. Despite the relatively low implantation energy, the modified layer can be greater than 1 /~m in thickness. The nitrogen concentration profile can be controlled by the implantation temperature and dose. Glancing-angle X-ray diffraction has been used to determine the structural changes that occur in the surface layer.

1. Introduction Ion implantation is a surface treatment technique that has been shown to be very effective in improving the wear resistance of a wide range of metals [ 1, 2]. Its widespread application has, however, been hindered by both real and perceived limitations [3]. Among these are the extremely shallow treatment depth (of the order of 100 nm) and the need for the target and/or the beam to be manipulated to achieve uniform surface coverage. In addition, the technology required for a reliable high current ion source has meant that ion implantation facilities have generally been restricted to research organizations and specialist surface treatment companies. In the past 4 years a technique has been developed that offers the possibility of taking ion implantation out of the specialized laboratory and onto the tool-room floor. This technique was first proposed by Conrad et al. [4] at the University of Wisconsin who surrounded the workpiece with a plasma and accelerated ions from the plasma by high voltage pulses applied directly to the target. This technique, which they called plasma source ion implantation (PSII), eliminates the need for cumbersome target manipulation or 0921-51/93/91/$3.50

beam rastering. It has several distinct advantages over traditional methods of implantation, including uniform coverage, low unit cost, the easing of line-of-sight restrictions and the ability to treat complex shapes. In addition, it has the ability to scale to large targets and offers the prospect of a technologically simple implanter design. The Wisconsin group has continued to develop PSII [5, 6] and has demonstrated not only that ions can be effectively implanted to the concentrations and depths required for surface modification with acceptable uniformity over non-planar targets but also that dramatic improvements can be achieved in the life of manufacturing tools in actual industrial applications. At Ansto we have developed an alternative system to PSII based on r.f.-generated plasmas, replacing the multidipole filament discharge used by the Wisconsin group with an inductively coupled r.f glow. We believe that there are some inherent advantages in our technique, which we call plasma immersion ion implantation (PI3), and will discuss these below. Early results [7] showed that nitrogen was implanted to a depth and dose consistent with expectations and, in some cases, to depths much greater than expected. Despite © Elsevier Sequoia/Printed in The Netherlands

172

the relatively low implantation energies (20 keV N2 + ions), significant increases in wear resistance and surface hardness were obtained and, in the case of the deep nitrogen penetration, exceeded those achieved by conventional ion implantation. Since those early experiments, we have significantly increased the energy of the implanted ions and have determined the conditions under which the enhanced penetration of nitrogen and the accompanying dramatic increases in surface hardness and wear resistance can be achieved. In this paper we describe the present form of our implanter and its typical operating parameters. We then discuss the application of PI 3 to a number of steels, specifically 0.3 wt.% C mild steel and austenitic stainless steel, at moderate implantation voltages between - 2 5 and - 3 5 kV. Nuclear reaction analysis was used to measure the nitrogen concentration profile, while glancing-angle X-ray diffraction allowed investigation of the structural changes that occur in the surface layer. Finally, some microhardness and weartesting results will be presented.

2. Experimental details Our implanter is based on an inductively coupled r.f. plasma produced in a cylindrical vacuum vessel (diameter, 30 cm; length, 40 cm) made of borosilicate glass. The vessel is filled with oxygen-free nitrogen to a pressure of about 1 × 10 -3 mbar. Approximately 25c~W of r.f. power at 12-13 MHz is used to produce a plasma with an ion density of about 1 × 101° cm --3. Theoretical calculations [8] indicate that the dominant ion species is N2 + with about 10% N +. The workpiece is supported in the centre of the vessel by an insulated conducting rod which

<

400

I

300

pumps

Fig. 1. Schematic diagram of the PI 3 apparatus.

also supplies the bias voltage for implantation. The target forms the cathode of the bias circuit which supplies - 4 5 kV pulses of 50-100 kts duration. One of the aluminium end plates that seals the glass cylinder is grounded and forms the anode of the implantation circuit. A schematic diagram of the apparatus is shown in Fig. 1. The pulse repetition rate is varied to control the surface temperature of the sample. Thermal equilibrium of the target is usually reached after treatment for a few minutes with radiation providing the dominant cooling mechanism. The application of a high negative voltage to the target results in the formation of an ion-rich sheath which effectively insulates the target from the surrounding neutral plasma. Nitrogen ions are accelerated across the sheath and strike the surface at angles close to normal [9]. Typical voltage and current traces are shown in Fig. 2(a) and Fig. 2(b) for implantation of a large tube-drawing die (outside diameter, 65 mm). It should be noted that the measured current contains a significant contribution from secondary electrons ejected from the surface of the target by impinging ions. From the heating characteristics of the target a reasonable estimate of the real implantation current (that due to ions actually striking the surface) can be made. Typically this will be a factor r/of between 2 and 4 below that of the measured current, depending on the implantation voltage. Note that t/= 7 + 1, where 7 is the secondary electron yield. In order to measure the extent of the sheath and its development during the implantation pulse, we have used a capacitive voltage probe [10]. This probe is suitable for potential measurements in the sheath region because of the strong secondary-electron emission that occurs from the glass surface of the probe when bombarded by the energetic ion and electron fluxes in the sheath [11]. In this case the glass surface floating potential approaches the local plasma potential and the probe behaves like an emissive probe without the problems of using thermionic emission [12]. In Fig. 2(c) we show the change in plasma potential produced in the bulk plasma by application of the implantation pulse. This is the only effect of the high negative voltage pulse seen outside the sheath when the current is low (below 1 A). The ion density as measured by a collecting Langmuir probe remains constant throughout the pulse. Even when the current is high and ion depletion would be expected to have a significant

173 ,

7v ~kv;

-[

'- /

0

'

'to;

,

i

i

i

r

r

~

1 IA~

i

{b)

t

20

~0

6o

0

80

20

~.0

60

8o

~ t h ~,~ml

100

4

90

{6j

F

0

60 -700 30

-206 i

20

4o

i

60

i

8o

o

o

L

,

20

,

,

~o

time ~s]

L

,

:

6o

80 hine @s]

Fig. 2. A series of implantation pulses on a tube-drawing die: (a) implantation voltage; (b) total current to target: (c) bulk plasma potential change; (d) sheath width.

effect, the bulk plasma parameters show surprisingly little change. In Fig. 2(d) we indicate the evolution of the sheath width. Since this width will vary around a target of complex shape [9], we have shown the sheath width at its largest extent. Figure 2(d) indicates that after a few microseconds the sheath has reached equilibrium even at the higher currents where ion depletion in the bulk plasma does occur, as evidenced by a decrease in the implantation current in Fig. 2(b). This result is in contrast with the behaviour of the sheath during PSII [4-6] where it is observed to expand continuously throughout the implantation pulse. We believe that the high ionization rate in the r.f. glow discharge is responsible for maintaining the sheath in equilibrium despite the high ion loss rate during implantation. Disks of diameter 25 mm and thickness 5 mm were cut from cold-drawn low carbon (0.3 wt.%) mild steel bar stock (AS1443-1983), while small coupons of area 35 mm x 20 mm and thickness 1 mm were cut from a sheet of type 304 stainless steel. The specimens were mechanically polished to a 1 Hm diamond finish and supported freely in the implantation chamber. Unpolished surfaces were coated with colloidal graphite to increase the efficiency of radiation cooling and to provide a surface with known emissivity so that the surface temperature could be monitored remotely.

A series of implants was performed at different pulse repetition rates, resulting in different final temperatures. The dose was controlled by counting the total number of pulses N. For a pulse width r and current density j, the implanted dose Dimp will be given by i2 Dimp = N%"-~ r/e

(1)

assuming that all ions are N2 +. The measured current density ] must be divided by the factor r; to give the real ion current to the target. The profile of retained nitrogen was measured by the use of the 14N(d, a)12C nuclear reaction. A 200 nA, 1.2 MeV deuteron beam 4 mm in diameter was used and the 6-7 MeV a particles were detected at 0 = l l 0 t Depth profiles were obtained with a deuteron dose of 150 ktC and a Mylar filter was used to eliminate the backscattered deuterons from the detector. As a result, the depth resolution (full width at halfmaximum (FWHM)) was limited to 60 nm close to the surface, increasing to 130 nm at a depth of ] ;xm. The structural changes in the surface layer were investigated by glancing-angle X-ray diffraction using a Siemens D500 diffractometer with the post-specimen monochromator acting in the parallel mode. Co K a radiation was used for all measurements. Incidence angles were typically 1°,

174

but some patterns were obtained for a range of angles from 0.5 ° to 5 °. The surface hardness was measured on a Leitz Durimet microhardness tester using a Vickers indenter at 15 gf load. These measurements include a substantial contribution from the unmodified material below the surface layer, thus providing only a qualitative assessment of the effectiveness of PI 3 with regard to changes in the surface hardness. Wear measurements were made on a pin-ondisk machine but using a fixed tungsten carbide ball (diameter, 10 mm) as the pin. The contact velocity was 0.8 m s-1 with paraffin oil continuously dropped onto the sample surface to provide light lubrication. Wear tracks produced using a 50 gf load for various contact times allowed suitable comparison between the wear properties of implanted and unimplanted samples. The wear depth profile was determined using interference microscopy. 3. Results

Nine mild steel samples were implanted at kV with pulse repetition rates of 20 Hz, 33 Hz and 50 Hz (resulting in final temperatures V = - 25

1

atom %! 20

I

l

I

l

~

I

of 200 °C, 290 °C and 340 °C respectively) with three different total implantation times of Nr =5s, 1 0 s and 15s. From eqn. (1), this corresponds to implantation doses of 2.2 × 1017 atoms cm -2, 4.4× 1017 atoms cm -2 and 6.6 x 10 ~7 atoms cm -2. In Fig. 3 we display the depth profiles obtained by nuclear reaction analysis for samples at the extremes of implantation dose and temperature. In each of the nine samples, there is a peak some 50 nm below the surface but the poor depth resolution does not allow direct comparison of the peak location with the expected implantation depth. (For 25 keV N2 ÷ ions, standard theories [13] predict a concentration profile peaking at about 15 nm with an F W H M of about 25 nm.) The observed peak broadening should be dominated by the measurement resolution but only the lowest dose, lowest temperature sample has a peak with an F W H M of the expected 60 nm. When the temperature and implantation dose increase, the peak broadens to give an F W H M of about 100 nm, appreciably higher than that expected from measurement resolution alone. More significantly, a tail appears in the profile, indicating diffusion of nitrogen to depths greater than 1 ~m. In Fig. 4 we show the total retained nitrogen

I

T= 200 °C implant time 15s dose= 2 x1017 cm-2

T = 200°C implant time 5s dose= 1x1017 cm"2

10

i..

0 I

atom % 20

I

IL

J.

I

i

~

i.

i

AL,L.i~

I

I

I

T=340°C implant time 15s dose = 7x 1017cm"2

T = 320°C imp/ant time 5s dose =2x 1017cm-2

10

0

1

depth (ym]

0

I depth [~m]

0

Fig. 3. Nitrogen depth profiles obtained from nuclear reaction analysis of four of the mild steel samples.

175

0"4

x 10178

fofal N content 6 .foms.cm'2] 4

/

/

o3/~0oOc

0"3

.

0"2

~ t ,

0"1

2

/~Ix

~

0

4

0

,/

~" 290%

xlO 17

median depth

I

200% I

0

I

xlO 17 N content of fail z, [afoms.cm-21

N content of peak [afoms.cm- 21

O f

2

,,o

/

/"

O j .

x f ~ I

00

5

I

00

10 1'5 implant time (s)

5

10 15 implant time (s]

Fig. 4. Total nitrogen content, median depth of nitrogen profile and nitrogen content of the implanted peak and diffusion enhanced tail as functions of implantation time and temperature in the mild steel samples.

40 atom%

I

(a) r=23o°c

4°I (b) atom%

20 f

,

.i

r J

1

depth [tim) 0

I i

.

,

1

depth (JJm) 0

Fig. 5. Nitrogen depth profiles obtained from nuclear reaction analysis of type 304 stainless steel implanted to a nominal dose of 15 x 1017 atoms cm : at temperatures of (a) T = 230 °C and (b) T= 340 °C.

content as a function of implantation dose and temperature, as well as the proportion of this nitrogen contained in the surface peak and the diffusion-enhanced tail. Also indicated in the figure is the median depth of the concentration profile. The nitrogen content of the peak appears to saturate at about 2 x 1017 atoms cm 2 which is well above what is expected for 25 keV N2 + ions but is consistent with the enhanced broadening of the peak. The nitrogen content of the diffusionenhanced tail, however, continues to increase with increasing implantation dose as long as the temperature is above a threshold of about 260 °C. Implantation doses greater than those in Fig. 4 have shown no sign of saturation, with

the diffusion-enhanced tail growing in height and depth as long as the temperature is above threshold. Similar results have been obtained for the type 304 stainless steel although indications are that the threshold temperature for the enhanced diffusion is somewhat higher. In Fig. 5 we show the nitrogen concentration profiles for two samples implanted at V = - 28 kV to a nominal dose D i m p = 1 5 x l 0 w atoms cm 2 at T=230 and 340 °C. While the low temperature specimen shows only a surface peak with total retained nitrogen content Dret=2.5× 1017 atoms cm-=, the amount of diffusion obtained at the higher temperature is remarkable, giving rise to a large

176

(a)

T=340°C dose = Gx lO~7crnz

1

]

!

,

h .....

40 °

50 °

~

GO o

L . . . .

..J

t. . . .

80~

70o

,

90 °

k& T=340°C dose =20 x 1017cm~

20

[ oC=0.5 o

'(+e)

!

{b)

.... x?(\ /l ........... 40 °

fl ....... 100°

20

Fig. 7. X-ray diffraction patterns obtained from the type 304 stainless steel implanted at T= 340 °C (incidence angle, 1°) with retained dose of (a) 6x1017 atoms cm -2 and (b) 20 x 10 '7 atoms cm -2.

] o:.=10 °

~=25

r\. 80 o

60 °

(d)

o

track cross-

~0 o

50 °

20

60 °

Fig. 6. (a) X-ray diffraction pattern obtained from the mild steel sample after high dose, high temperature implantation (incidence angle, 1°). (b)-(d) Diffraction patterns obtained with a range of glancing incidence angles and expanded around and normalized with respect to the 110 a-Fe peak.

section [x l O e m 2 )

5 4

untreated

/

2

/.

cm

2.

Glancing-angle X-ray diffraction has shown clear evidence of precipitates of y'-Fe4N and e-FezNl-x in the mild steel samples. A typical diffraction pattern is shown in Fig. 6(a) for the high dose, high temperature implant at an incidence angle of 1°. In Fig. 6(b)-(d) we have expanded the region around the 110 a-Fe peak and show diffraction patterns obtained over a range of incidence angles. Since a larger incidence angle will probe a deeper layer, we conclude that the e-FezNl-x is more strongly concentrated at shallower depths. The diffraction patterns obtained from the type 304 stainless steel are shown in Fig. 7. Here we compare two high temperature implants ( T = 3 4 0 °C) but at two different doses: Dret=6 x 10 t7 and 2 0 x 1017 atoms cm -2. A set of broad uncharacterized peaks appears just below each 7-Fe peak. These peaks broaden and shift towards larger lattice parameters as the dose increases. The broadness indicates considerable strain and the peak shifts are considerably greater than those expected from an austenite-based (Fe, Ni, Cr)aN. All the mild steel samples showed increases in

'

3

/

retained nitrogen dose Dre~ of 2 0 x 1017 atoms

/

0

, dose// / ,. / p!g~~ose,

/

htgh T

"-~ * 1

hlgh T 2

3

wear time

4

[hoursl

Fig. 8. Wear characteristics of implanted mild steel disks measured using a tungsten carbide ball under 50 gf load and 0.8 m s ~contact velocity with paraffin oil lubrication.

the surface hardness under a 15 gf load. The unimplanted Vickers hardness Hv was 180 while implanted specimens gave H v values of up to 240. Improvements in microhardness were more dramatic for the type 304 stainless steel. With a 15 gf load, surface hardness was observed to increase from H v = 2 2 0 for an unimplanted specimen up to Hv = 310 for the high temperature, high dose implant depicted in Fig. 5(b). Figure 8 summarizes the results of wear tests on some mild steel disks. We compare an unimplanted specimen with two specimens treated at a high temperature. Both implanted specimens showed insignificant wear for the first hour, but after 2 h the wear rate of the low dose sample ( D r e t = l . 4 X 1017 atoms cm -2) increased. This indicates that substantial breakthrough of the shallow modified layer had taken place. The high dose implant (Dre t = 4.4 x 1017 atoms cm -~) continued to show a track with negligible wear even after 4 h when some breakthrough of the modi-

177

fied layer was evident. Wear testing was not continued since a flat had been worn on the tungsten carbide ball. 4. Discussion

The most significant factor to emerge from the above results is the diffusion of nitrogen to depths of greater than 1 /~m at elevated temperatures. This is generally not the case for conventional ion beam implantation [14], where for temperatures in the range 50-200 °C the nitrogen migrates progressively towards the surface while keeping the total nitrogen content constant. Even at high doses implanted by conventional ion beams at 200 °C [15], when the nitrogen content saturates, nitrogen is still confined to its expected implantation depth. At temperatures above 200 °C, the nitrogen content decreases, corresponding to diffusion out of the implanted region. Our results show that the PP process is significantly different, with the total nitrogen content rising steadily with increasing temperature and dose, corresponding to increased diffusion at depths beyond the normal implanted layer. This implanted layer is preserved even at temperatures of 340 °C. We note that similar results have recently been reported for ultrahigh current density implantation in both pure iron and type 304 stainless steel [16, 17]. In our case we believe that the high activity of the nitrogen in the plasma is sufficient to prevent the outward diffusion that would normally occur at high temperatures, allowing the nitrogen deposited in the implanted layer to continue to move deep into the steel. The best improvements in surface hardness and wear resistance are associated with these high temperature~ high dose implants. There are similarities in the microstructural changes that we observe and those reported elsewhere [14-17]. In particular the nitrogen-rich eFezN~ .~ occurs closer to the surface where the nitrogen concentration is greater. Other workers [16, 18] have speculated that the 7'-FeaN is harder and more wear resistant than the higher nitrides, although this may simply be due to its occurrence at greater depth, thus providing better load-bearing characteristics. The X-ray diffraction patterns in the stainless steel are more curious. The most obvious interpretation is that the austenite has been expanded by nitrogen in interstitial solution. Such an expanded austenite structure was observed in the

ultrahigh current density implantations mentioned above [17] but has also been observed in plasma nitriding of type 304 stainless steel at "low" temperatures ( T = 3 5 0 - 5 0 0 °C) [19-21]. This uncharacterized phase is responsible for causing a profound hardening of the surface and has been named the "S phase" [20]. The relative positions of the broad peaks seen in Fig. 7 are consistent with a structure based on an f.c.c, metal sublattice. For the sample with the lower retained dose, shown in Fig. 7(a), the apparent lattice parameter a is approximately 0.372 nm. On the basis of the data of Jack and Jack [22], and using a slight correction for the austenite cell size of type 304 stainless steel, this corresponds to a nitrogen concentration of approximately 17 at.%. Such a phase could be referred to appropriately as expanded austenite, although the nitrogen concentration exceeds the equilibrium solubility of nitrogen in austenite. In the case of the high dose implant (Fig. 7(b)), the peaks have become so broad that it is impossible to analyse them in any detail. The maximum shift in the peaks corresponds to an f.c.c, lattice with a = 0.4 nm. This is in excess of the parameters possible for y'-Fe4N [23]. Extrapolation of the data available for the )/phase implies an apparent concentration of 27 at.% for an f.c.c, phase in the high dose implant in type 304 stainless steel. However, it is clear from Fig. 7(b) that a detailed analysis of the phase structure of such a highly strained non-equilibrium layer is impossible by the use of X-ray diffraction. The diffusion of nitrogen observed in the PI -~treated steels leads us to compare our technique with plasma nitriding. Even at the highest temperatures, PP still lies at the low temperature end of the plasma-nitriding regime. Below 500 °C, nitriding rates for type 304 stainless steel are low, although nitride layer growth has been observed at temperatures down to 280 °C [24] using a pulsed r.f. glow discharge. The gas pressures used in conventional plasma nitriding are usually greater than 1 mbar, at least three orders of magnitude higher than used in PI -~.Recently, however, there has been some nitriding work using a Penning discharge plasma at low pressures, comparable with those used in PP [25]. 5. Conclusion

The PI -~process, like the related PSII technique of Conrad and coworkers [4-6], overcomes some

178

of the disadvantages of conventional implantation that have prevented the widespread use of ion implantation as a surface modification technology. Use of an r.f. glow plasma provides an improvement on PSII in enabling a steady state sheath to form around the target. In this series of experiments on 0.3 wt.% C mild steel and type 304 stainless steel, we have shown that, despite the relatively low implantation energies, significant improvements in wear resistance and surface hardness can be obtained with high dose, high temperature treatment. In sharp contrast with ion beam implantation, there is no evidence of loss of nitrogen ~is the temperature is increased. Glancing-angle X-ray diffraction has shown that precipitates of Fe4N and Fe2N 1-x form in the mild steel, while the structures created in the stainless steel are similar to the f.c.c.-based "S phase" observed in low temperature plasma nitriding. Further development is under way to ascertain whether the hard wear-resistant surfaces produced can be incorporated into industrial components.

Acknowledgments Support for this work is provided under the Generic Technology component of the Australian Industry Research and Development Act 1986. We acknowledge the assistance of Ron Clissold with the wear testing and microhardness measurements and thank MM Metals for their interest and involvement in the project.

References 1 G. Dearnaley, Surf. Eng., 2 (1986) 213. 2 W. C. Oliver, R. Hutchings and J. B. Pethica, Metall. Trans. A, 15 (1984) 2221.

3 K. O. Legg and H. Solinick-Legg, in T. S. Sudarshan and D. G. Bhat (eds.), Proc. 1st Int. Conf. on Surface Modification Technologies, Phoenix, AZ, January 1988, Metallurgical Society of AIME, Warrendale, PA, 1988, p. 79. 4 J. R. Conrad, L. Radtke, R. A. Dodd, F. J. Worzala and N. C. Tran, J. Appl. f'hys., 62(1987)4591. 5 J. R. Conrad, S. Baumann, R. Fleming and G. P. Meeker, J. Appl, Phys., 65(1989) 1707. 6 J. R. Conrad, Mater. Sci. Eng. A, 116(1989) 197. 7 J. Tendys, I. J. Donnelly, M. J. Kenny and J. T. A. Pollock, Appl. Phys. Lett., 53(1988)2143. 8 I. J. Donnelly and E. K. Rose, in R. McWilliams (ed.), Proc. 8th Topical Conf. on R.F. Power in Plasmas, lrvine, CA, May 1989, American Institute of Physics, New York, 1989, p. 410. 9 P.A. Watterson, J. Phys. D, 22 (1989) 1300. 10 N. Benjamin, Rev. Sci. Instrum., 53(1982) 1541. 11 N. Hershkowitz and M. H. Cho, J. Vac. Sci. Technol. A, 6 (1988) 2054. 12 E. Y. Wang, N. Hershkowitz and T. Intrator, Rev. Sci. lnstrum., 57(1986) 1085. 13 J. E Ziegler, J. E Biersack and U. Littmark, The Stopping and Range of lons in Solids, Pergamon, New York, 1985. 14 N. Moncoffre, Mater. Sci. Eng., 90(1987) 99. 15 G. Terwagne, M. Piette, P. Bertrand and F. Bodart, Mater. Sci. Eng. B, 2(1989) 195, 16 R. Wei, P. L. Wilbur, W. S. Sampath, D. L. Williamson, Y. Qu and L. Wang, J. Tribol., 112 (1990) 27. 17 D. L. Williamson, L. Wang, R. Wei and P. J. Wilbur, Mater. Lett., 9(1990) 302. 18 T. Fujihana, Y. Okabe and M. Iwaki, Nucl. lnstrum. Methods B, 39 (1989) 548. 19 K. Iehii, K. Fujimura and T. Takase, Technol. Rep. Kansai Univ., 27 (1986) 135. 20 P.A. Dearnley, A. Namvar, G. G. A. Hibberd and T. Bell, in E. Broszeit, W. D. Munz, H. Oechsner, K.-T. Rie and G. K. Wolf (eds.), Proc. 1st Int. Conf. on Plasma Surface Engineering, Garmisch-Partenkirchen, September 19-23, 1988, Vol. 1, Deutsche Gesellschaft fiJr Metallkunde lnformationsgellschaft, Oberursel, 1989, p. 219. 21 S.-E Hannula, E Nenonen and J. P. Hirvonen, Thin Solid Films, 181 (1989) 343. 22 D. H. Jack and K. H. Jack, Mater. Sci. Eng., 11 (1973) 1. 23 M. A. J. Somers, N. M. van der Pers, D. Schalkoord and E. J. Mittemeijer, Metall. Trans. A, 20(1989) 1533. 24 F. El-Hossary, F. Mohammed, A. Hendry, D. J. Fabian and Z. Szaszne-Csih, Surf. Eng., 4 (1988) 150. 25 M. Nunogaki, H. Suezawa, K. Hayashi and K. Miyazaki, Appl. Surf. Sci., 33-34 ( 1988) 1135.