Plasma monitoring of plasma-assisted nitriding of aluminium alloys

Transcription

1 Surface and Coatings Technology (1999) Plasma monitoring of plasma-assisted nitriding of aluminium alloys Michael Quast *, Peter Mayr, Heinz-Rolf Stock Stiftung Institut für Werkstofftechnik, Badgasteiner Str. 3, D Bremen, Germany Abstract To improve the plasma-assisted nitriding process of aluminium alloys it is necessary to obtain knowledge about the underlying reaction mechanisms. A suitable diagnostic tool to clarify these mechanisms is plasma monitoring, which provides mass- and energy-resolved analysis of the ions hitting the substrate surface in a glow discharge. Application of this technique to plasmaassisted nitriding of pure aluminium and the 2024 aluminium alloy is demonstrated. Mass spectra and energy distributions of the major ions were recorded during sputtering in an argon atmosphere and nitriding in pure nitrogen. The energy distributions of the gas ions are mainly determined by resonant charge exchange collisions, whereas the ions of the sputtered metal atoms reach the cathode almost without interaction. The influence of the process parameters, temperature and working pressure, on the discharge characteristic were examined. Changing the substrate temperature did not significantly affect the ion energy distributions, whereas a reduction of the working pressure increased the collision probability of the ions due to a disproportionate elongation of the cathode fall Elsevier Science S.A. All rights reserved. Keywords: Aluminium alloys; Ion energy distribution; Plasma-assisted nitriding; Plasma monitoring 1. Introduction Application of this technique to the pre-sputtering and plasma-assisted nitriding of aluminium alloys is Today aluminium alloys are used in a wide range of demonstrated here. The influence of the process parameapplications owing to their low specific weight and good ters, substrate temperature and gas pressure, on discharge characteristics and nitriding effects are discussed. machinability. However, the major drawback of these materials is their poor wear resistance. Since aluminium nitride has a high hardness in combination with a high thermal conductivity, nitriding of aluminium is of great 2. Experimental interest. Arai et al. [ 1] were the first to show that plasma-assisted nitriding of aluminium is possible, if the Mass spectra and energy distributions of ions were native oxide layer is removed by an appropriate preobtained with a commercial plasma monitoring system sputtering process. This process and the nitriding pro- ( EQP II, Hiden Analytical Ltd) sampling through a cedure were subsequently studied by several groups 50 mm aperture in the centre of the grounded cathode [2 4]. of a plasma-assisted nitriding facility ( Fig. 1). The cath- Meletis and Yan [5] were the first to use the knowlode (austenitic steel ) was covered by an aluminium edge about ion energy distributions in a glow discharge plate, on which the samples made of pure aluminium to affect the nitriding process. For a further improvement ( 1050A) and 2024 aluminium alloy (AlCuMg2) were of this process, a detailed examination of the mass and positioned. Using an ohmic resistance heater, the temenergy distributions of the ions hitting the substrate perature of the samples could be raised to 500 C indesurface during plasma-assisted nitriding is necessary. pendent of plasma parameters. Two steel plates served Plasma monitoring is a common experimental technique, as an anode; a glass cylinder and two screening plates which is often used in different applications to obtain prevented the discharge from reaching the walls of the information about reaction mechanisms or ion energies vacuum chamber. in d.c. or r.f. glow discharges [6 9]. After evacuating the vacuum chamber by a pumping * Corresponding author. address: quast@iwt.uni-bremen.de (M. Quast) system (rotary pump, roots pump and oil diffusion pump) to a pressure of 10 4 Pa, the apparatus was /99/$ see front matter 1999 Elsevier Science S.A. All rights reserved. PII: S (99)

2 M. Quast et al. / Surface and Coatings Technology (1999) Fig. 1. Schematic of the experimental set-up: (1) vacuum chamber; (2) plasma monitor; (3) cathode with samples and heating; (4) anode; (5) screening; (6) glass cylinder; (7) gas supply; (8) pumping system. heated for several hours by the substrate heating to to gain high kinetic energies, the ions coming from the reduce the water partial pressure. Then the substrates negative glow reach the cathode with the full energy of were sputtered in an argon d.c. glow discharge (pressure: the potential drop across the cathode fall. 20 Pa; applied voltage: 600 V; substrate temperature: Fig. 2a shows the mass spectrum of the low energetic 450 C) for 1 h, followed by the nitriding process in pure ions (E=20 ev ) arriving at the cathode during the nitrogen. The discharge pressure was adjusted by means sputtering process of pure aluminium. Dominating of the flow of the process gases and controlled by a species are the singly charged argon ions at a mass-tocharge capacitance manometer (MKS Baratron). The substrate ratio of 40 amu, followed by the doubly charged temperature was measured by a thermocouple. argon ions at 20 amu. These ions are formed within the Ions bombarding the cathode were extracted through cathode fall by collisions of argon ions with argon the small cathode aperture and then focused by the ion neutrals. Besides aluminium at 27 amu, only small extraction optics of the plasma monitor. Energy and amounts of impurities are detected. mass analyses of the ion beam were performed by an In contrast, the mass spectrum of the high energetic electrostatic sector field analyser, followed by a triple ions with E=600 ev (Fig. 2b) is dominated by ions of quadrupole mass filter. A channeltron secondary the sputtered material, such as aluminium at 27 amu, electron multiplier was used to detect the ions. The magnesium at 24, 25 and 26 amu, silicon at 28 amu and plasma monitor was pumped by a turbo pump; the iron at 56 amu. Apart from oxygen ions at 16 amu, no pressure did not exceed Pa. The standard resolution of the energy analyser is 0.7 ev. great number of ArH+ ions (41 amu) are formed even other fragment ions of Al O appear. In addition, a 2 3 at very low concentrations of hydrogen. Carbon ions at 12 amu probably are formed from residual gas. 3. Results and discussion Fig. 3 shows the ion energy distributions ( IEDs) of the most abundant species, i.e. the singly charged argon 3.1. Pre-sputtering ions Ar+ and the singly charged aluminium ions Al+. The IED of the Ar+ ions (Fig. 3a) exhibits a maximum The kinetic energy of ions in a d.c. glow discharge provides information about their origin. Whereas the ions formed directly in front of the cathode are not able at an energy of 30 ev and an exponential decay to higher ion energies. A weak maximum appears at an energy corresponding to the applied electrode voltage

3 246 M. Quast et al. / Surface and Coatings Technology (1999) Fig. 2. Mass spectra of ions hitting the cathode during pre-sputtering of pure aluminium: (a) low energetic ions (E=20 ev ); (b) high energetic ions (E=600 ev ). Fig. 3. Energy distributions of ions recorded during pre-sputtering of aluminium: (a) singly charged argon ions 40Ar+ ( ) and fit on Eq. (1) ( % ); (b) singly charged aluminium ions 27Al+. (600 V ). The average kinetic energy of the argon ions amounts to 97 ev, a very low value compared with this applied voltage. According to Davis and Vanderslice [6], the ion energy distributions in d.c. glow discharges are mainly determined by the probability for resonant charge exchange collisions between gas ions and atoms of the same gas. This probability may be expressed as the ratio of the length of the cathode fall d and the mean free path (referred to charge exchange collisions) l. Ifitis assumed that no electron impact ionisation takes place in the cathode sheath, the IED is then given by dn de = N 0 m d strates the important contribution of the energetic neutrals to the sputtering process. The IED of the aluminium ions (Fig. 3b) has a completely different shape. Besides the maximum at the energy corresponding to the applied voltage, a second narrow maximum of lesser intensity appears at the low energy of 10 ev. The maximum at the plasma potential indicates that most of the Al+ ions pass the cathode fall without any interaction owing to the low density of aluminium atoms in the fall region. The small maximum at low energies may result from the inhomogeneous spatial distribution of aluminium atoms. Since the mean distance between two elastic collisions at a pressure of 20 Pa is less than 1 mm, the density of the aluminium l (1 E)(1/m) 1exp C d l + d l (1 E)1/m D, (1) atoms will rapidly decrease within the cathode fall, leading to an increasing collision probability towards the surface. The time evolution of the ion intensities is of special interest, since it provides information about the pro- gression of the sputtering process. The ratio of the integrated intensities of aluminium and oxygen ions as a function of time is plotted in Fig. 4. It increases rapidly and reaches a constant value after 30 min of sputtering. This indicates that an equilibrium of sputtering and redeposition of aluminium oxide is established, and where N is the number of ions entering the cathode 0 fall, m is a parameter that describes the behaviour of the electric potential in front of the cathode and E is the relative ion energy [ 10]. A fit of the experimental data to Eq. (1) is also plotted in Fig. 3a and shows a reasonable agreement. It resulted in a d/l value of 8.9, i.e. each argon ion undergoes nine resonant charge exchange collisions and produces the same number of argon atoms with high kinetic energies. This demon-

4 M. Quast et al. / Surface and Coatings Technology (1999) Fig. 4. Ratio of integrated intensities of aluminium and oxygen ions during sputtering of pure aluminium as a function of time. continuation of the sputtering process will not change the surface composition. The high value of 475 suggests that a high amount of sputtered aluminium is accelerated back towards the cathode, whereas oxygen is removed effectively from the surface. In Fig. 5 the mass spectra of the ions arriving at the cathode during sputtering of the 2024 alloy are shown. In comparison with the sputtering of pure aluminium (Fig. 2), the lines at 24, 25 and 26 amu and between 63 and 66 amu appear with clearly higher intensities. These lines refer to magnesium, and to copper and zinc ions respectively. It should be noted that the intensity of the magnesium ions is as high as that of the aluminium ions, although the magnesium fraction in the substrate material is only 1.49 at.% (see Table 1). A similar effect is observed for copper and zinc: although the copper fraction is 30 times higher than that of zinc, the intensity of the zinc ions at 64, 66 and 68 amu exceeds that of the copper ions at 63 and 65 amu. The energy distributions of these ions were found to be similar to that of the aluminium ions ( Fig. 3b). The mass spectra differ significantly from those expected on the basis of the chemical composition of this material. This can only to a certain point be explained by the different sputtering yields of aluminium and copper compared with magnesium and zinc. The yields of magnesium and zinc calculated using the Bohdansky formula given in Ref. [11] are about three times higher than those of aluminium and copper. Therefore, the magnesium content of the sputtered material should be about 5%. Thus it could be supposed that the ionisation cross-section of magnesium has to be more than one order of magnitude greater than that of Al. Considering the cross-sections for ionisation by electron impact of Al and Mg, the reverse was found to be true [12]. This could be explained if the sputtering yields of the elements in the alloy differ from those of the pure elements. Another influence arises from the fact that both magnesium and aluminium parts of the alloy s surface are covered with their native oxide layers. Since the sputtering yield of MgO is considerably higher than that of Al O [13], the magnesium-covered parts 2 3 of the substrate surface are sputtered preferentially, resulting in a much higher magnesium fraction in the sputtered material Plasma-assisted nitriding The dominating species of the nitriding process are nitrogen ions. The mass spectrum of the low energetic ions recorded during nitriding of aluminium in 50 Pa pure nitrogen ( Fig. 6a) shows molecular nitrogen ions at 28 amu (N+) and 42 amu (N+), atomic nitrogen 2 3 ions N+ (14 amu) and a small fraction of aluminium ions at 27 amu, whereas only atomic nitrogen ions appear in the spectrum of the high energetic ions (Fig. 6b). The energy distribution of the molecular nitrogen ions N+ (Fig. 7a) is dominated by low energetic ions; 2 no ions reach the cathode without collisions. However, Fig. 5. Mass spectra of ions hitting the cathode during pre-sputtering although the average ion kinetic energy is only 46 ev, of the 2024 aluminium alloy in argon: (a) low energetic ions (E= about 60% of the ions reach the cathode with an energy 20 ev ); (b) high energetic ions (E=600 ev). above the sputtering threshold at 27 ev. A fit of the

5 248 M. Quast et al. / Surface and Coatings Technology (1999) Table 1 Chemical composition of the substrate materials (atom%) Alloy Mg Al Si Ti Cr Mn Fe Cu Zn 1050A IED using Eq. (1) leads to d/l=20.8, i.e. 21 energetic nitrogen molecules are produced by each nitrogen molecule ion. In contrast, the collision probability of the atomic nitrogen ions (Fig. 7b) is significantly lower (d/l=1.7), due to the low density of atomic nitrogen in the plasma, which can assumed to be less than 1% of the density of nitrogen molecules [14]. This leads to an average kinetic ion energy of 250 ev, which is high compared with the energy of the molecular ions. A variation of the substrate temperature during the nitriding process did not affect the IEDs of the nitrogen ions significantly. However, the intensity of the aluminium ions decreased significantly with time at higher substrate temperatures, as shown in Fig. 8. The increased diffusion coefficient at higher substrate temperatures caused a faster formation of the nitride layer and thus a stronger decrease of the aluminium content of the surface. Auger electron spectroscopy ( AES) depth profiles of the nitrided samples show a larger depth of the nitrogen signal for the higher nitriding temperatures. To examine the influence of the gas pressure on the plasma properties, aluminium was nitrided at three different nitrogen pressures. Fig. 9 shows the relevant IEDs of the atomic nitrogen ions N+. A reduction from 50 Pa to 33 and 20 Pa resulted not only in a strong decrease of the ion intensity due to a decrease of the gas density, but also changed the energy distributions drastically towards lower energies. The same behaviour was observed for the molecular nitrogen ions N+. The 2 minimum around 20 ev of the IED recorded at 20 Pa is caused by a disturbance of the electric field around the cathode aperture due to the deposition of electrically insulating material. Fig. 7. Energy distributions of ions recorded during nitriding of pure Fig. 6. Mass spectra of ions hitting the cathode during nitriding of aluminium (50 Pa N, 600 V, 450 V ): (a) nitrogen molecule ions 2 pure aluminium (50 Pa N, 600 V, 450 V ): (a) low energetic ions (E= 14N+ ( ) and fit on Eq. (1) (%); (b) atomic nitrogen ions 14N ev ); (b) high energetic ions (E=600 ev). ( ) and fit on Eq. (1) (%).

6 M. Quast et al. / Surface and Coatings Technology (1999) Conclusions Fig. 8. Intensity of aluminium ions 27Al+ at energy 20 ev during nitrid- ing of pure aluminium at two different substrate temperatures (50 Pa N, 600 V ). 2 Mass spectra and ion energy distributions were recorded during both the pre-sputtering treatment and the nitriding process of pure aluminium and the 2024 aluminium alloy. Substrate temperature and gas pressure were varied. After 30 min of pre-sputtering an equilibrium between sputtering and redeposition of aluminium oxide was established. Sputtering of 2024 aluminium alloy led to higher intensities of magnesium and zinc ions being observed, as expected from the chemical composition of the substrate material. The nitriding effect increases with increasing substrate temperature due to the enhanced diffusion. Since the collision probability depends on the pressure, a higher pressure not only leads to an increased number of particles hitting the substrate surface, but also raises the energy impact per ion. Acknowledgements The authors are grateful to Dr. H. Podlesak for AES analyses and useful discussions. In addition, we acknowledge the support of the Deutsche For-schungsge-meinschaft under grant number 217/8. References [1] T. Arai, H. Fujita, H. Tachikawa, in: T. Spalvins (Ed.), Proceedings 1st International Conference on Ion Nitriding, Cleveland, OH (1986) 37. [2] H.Y. Chen, H.-R. Stock, P. Mayr, Surf. Coat. Technol. 64 (1994) 139. Fig. 9. Energy distributions of atomic nitrogen ions 14N+ recorded [3] P.W. Wang, S. Sui, W. Wang, W. Durrer, Thin Solid Films 295 during nitriding of pure aluminium (N, 450 C, 600 V ). Variation of (1997) gas pressure. [4] B. Reinhold, J. Naumann, H.-J. Spies, F. Katzer, Härterei-Tech. Mitt. 52 (1997) 350. [5] E.I. Meletis, S. Yan, J. Vac. Sci. Technol. A: 9 (1991) [6] W.D. Davis, T.A. Vanderslice, Phys. Rev. 131 (1963) 219. Apart from changes in the ion intensities due to [7] M. Zeuner, J. Meichsner, Surf. Coat. Technol (1995) 562. [8] S. Peter, R. Pintaske, G. Hecht, F. Richter, Surf. Coat. Technol. different particle densities, a variation of the gas pressure 59 (1993) 97. should not influence the shape of the IEDs. The increase [9] S. Kadlec, C. Quaeyhaegens, G. Knuyt, L.M. Stals, Surf. Coat. of the mean free path of the ions at lower pressures is Technol. 89 (1997) 177. accompanied by a proportionate elongation of the cath- [10] J. Rickards, Vacuum 34 (1984) 559. ode fall, and thus d/l should remain constant [6,15]. [11] W. Eckstein, C. Garcia-Rosales, J. Roth, W. Ottenberger, Sputter- However, as can be seen from Fig. 9, the IEDs change ing Data, Max-Planck-Institut für Plasmaphysik, Report IPP 9/ 82, towards lower energies, i.e. the collision probability [12] R.S. Freund, R.C. Wetzel, R.J. Shul, T.R. Hayes, Phys. Rev. A increases. Fitting the measured curves to Eq. (1) leads 41 (1990) to values for d/l of 2, 4 and 8 at 50 Pa, 33 Pa and 20 Pa [13] G. Betz, G.K. Wehner, in: R. Behrisch (Ed.), Sputtering by Par- respectively. As the mean free path l is directly propor- ticle Bombardement vol. 2 Springer, Berlin, tional to 1/p, the reduced ion energy must be caused by [14] F. Becker, I.W. Rangelow, K. Maßeli, R. Kasseling, Surf. Coat. Technol (1995) 485. a disproportionate elongation of the cathode fall. A [15] B. Chapman, Glow Discharge Processes, Wiley, New York, similar behaviour was observed in argon discharges by [16] M. van Straaten, A. Bogaerts, R. Gijbels, Spectrochim. Acta Part van Straaten et al. [16]. B: 50 (1995) 583.