The deposition of thin films by filtered arc evaporation

Surface arid Coatings Techn ology. 54{55 ([992) [36-[42

136

The deposition of thin films by filtered arc evaporation P . J. Martin, R. P. Netterfield, A. Bendavid and T. 1. Kinder CSJRO Diuision of Applied Physics. Sydney, NSW 2070 (A ustralia )

Abstract The principles of macroparticle filtering in arc evaporation sourc es are reviewed. The use of passive filtering, internal and external magnetic fields is discussed. The technique of filtering using the magnetic plasma duct is reviewed in detail with emphasis on the properties of deposited films. The filtered arc technique is seen to be a useful method for the deposition of metal s, nitrides and oxide films.

1. Introduction The properties of thin films deposited by con ventional physical vapour deposition (PVD) techniques are frequently inferior to those of the bulk material. The films may have a low packing density, low microhardness and in many instances poor adhesion to the substrate. Man y of these shortcomings are a direct consequence of the energy of the depositing atoms arriving at the substrate during film growth. A low particle energy leads to a reduc ed surface mobility of the atom on the substrate and the subsequent evolution of columnar film growth through self-shadowing effects. The resulting film porosity gives rise to a reduction in mechan ical strength, and in the case of dielectric optical films, a reduction in the refractive index [IJ. The situation is improved greatly when the energy of the arriving atoms is increa sed directly (e.g. by sputtering rather than thermal evaporation), or indirectly (by bomb arding the growing film with mor e energetic particles) [2]. Several methods have been developed to increase the energ y of the depositing part icles. These methods include single and dual-ion beam sputtering, unbalanced magnetron sputtering and ion-beam assisted evaporation. The technology employed in the current -arsenal of ionassisted deposition (lAD) techniques has varying levels of complexity and each has its own particular advantages and disadvantages with respect to deposition rate, deposition area , film type and quality, cost , etc. Arc evaporation is well established as a powerful, cost effective ion-b ased technique for the deposition of a wide range of metals , alloys, and compounds [3J but the presence of microdroplets (macroparticles) of cathode material in the deposited films has precluded the use of the technique in the more demanding areas of optics and electronics, restricting its application to metallurgi-

0257-8972{92{$5.00

cal and tribological applications. In recent years there has been much resear ch into improving the arc meth od, primarily aimed at reducing the macroparticle content of the deposited films by a range of methods [4]. The present review focuses on the properties of arc evaporation which render it att ractive as an effective lAD technique , the methods used to reduce macroparticles, and the properties of films deposited by the filtered arc.

2. Arc evaporation as ion-assisted deposition The cathodic spot ma y be regarded as a micron-sized source of neutral plasma which, in the absence of external fields, moves across the surface of the cathode in a random manner. There have been many investigations of the cathode spot parameters [5-11J, and Table I lists

TABLE l. Summ ary of cat hode spo t par ameters Para meter

Range

Curren t density

10 7-10 10 A m " 1020

Ref.

m- 3

?

5

(Cu)

6

Electr on density

5x

Electron tempera tu re

3- 6 eV (Cu) 6- 9 eV (AI)

6 6

Pr essure

0.[- 10 MPa

7

Crat er size

[-20 urn

8

Crater format ion time

~ lO- 7

s (Joule heat ing) 1.2- 4.5 ns [Cu) 1.6-6.2 ns (M o) (ion impact)

8 9 9

Ion energy

25- 75 eV

10

Ion fraction

0.1-1.0

II

© 1992 - Elsevier Seq uoia. All rights reserved

P. J . Martin et al.

I Thin film deposition by filtered arc evaporat ion

some of the data obtained for the current density, electron density and temperature, pressure, etc. The cathode spot ranges from 1 urn to 20 urn in diameter and is an intense source of electrons, metal atoms and micron-sized droplets of cathode material. Electronatom collisions result in the formation of positive ions in a region close to the cathode spot surface. Some of these ions are accelerated back to the cathode and may be responsible for initiating new arc sites. Two theories have been proposed to describe the formation of ions with energies greater than the cathodeanode voltage drop [5, 12, 13]. The voltage drop is 10-30 V whereas ion energies may be greater than 30 eV per charge state. In the "potential hump" theory, ions formed by electron-atom collisions are ejected towards the anode from the positive ion cloud region formed immediately above the arc site. The ion cloud region gives rise to a hump in the potential distribution above the cathode plane, and if it is some 50 V above the anode potential, is sufficient to accelerate the ions to the energies that are observed experimentally. The gas dynamic theory [13J describes the increase in ion energy in terms of momentum transfer from the arc electron flux to the ions via collisions. The precise model for ion acceleration is unknown and Sanders et al. [4J suggest that both mechanisms are probably operative. The cathode spot and interelectrode region is also an intense source of photon emission. Detailed spectroscopic studies of these regions have shown the presence of neutral particles with energies of 5 eV and ions with energies of 50-60 eV directed primarily normal to the cathode surface [14]. The instability of the detected light from the neutral particles compared with that of the ionized species indicated that vapour may also be evaporated from an extended region not localized on the actual cathode spot [15]. Eckhardt [16J has proposed that evaporation is possible from the molten tracks created by the moving cathode spot. In terms of thin film deposition, the most interesting properties of the arc process are (i) average charge state per ion, (ii) ion fraction and (iii) ion energy. The average charge state for Ti ions has been estimated to be 1.6 e by Bergman [7J, and Sathrum [17J has shown that the addition of a solenoidal magnetic field around the source can increase this value to 2.08 e. The charge state is strongly material dependent [3J, ranging from Ti 3 + to U 7 + [18J. The ion fraction is also found to be material dependent. In the case of Ti, Bergman [7J reported an ionized fraction of 68% under high vacuum conditions and 85% at a background nitrogen pressure of 0.11.5 Pa during TiN deposition. Estimates of ion fractions for other materials range from 12%-15% for Cd to 80%-100% for Mg [3]. The average ion energy for Ti has been estimated to be 1.6(10 + Vs) eV [7J where Vs

137

is the negative substrate bias . Measurements of the ion energy distribution from a Ti arc showed the peak energy of the total distribution to be 44 ± 2 eV with a full width half maximum of36 eV [19]. The ion properties of the arc evaporation process enable the technique to be regarded as an effective method of ion-assisted deposition [20]. The minimum energy of the ions from the arc source depends upon the nature of the cathode and is in the range of 28-50 eV peak energy. The energy of the ions at the substrate may be increased simply by applying a high negative bias to accelerate the ions. When the substrate bias is sufficiently high (greater than -400 V for Zr, -600 V for Ti, approximately -1500 V for Ti at 0.02 Pa N 2 , less than - 500 V for AI), the self-sputtering yield is such that no net deposition takes place [21, 22J (i.e . the sputtering yield y> 1). Under normal deposition conditions '}' ~0.02 (50 eV Ti) [20].

3. Macroparticle emission It has been shown by several researchers that most of the droplets produced by a steady vacuum arc leave the cathode at small angles to the cathode plane [23 , 24J and that the size of the particles depended upon the arc current and background gas composition and pressure. Particle sizes range up to 25 urn in diameter and macroparticle emission is higher for lower melting point materials. The particle size for Ti is .found to decrease with increasing partial pressure of N 2 as a consequence of the formation of TiN on the cathode surface. The velocity of the macroparticles is in the range 0.1-800 m S-I. The average velocity increases with the melting point of the cathode material, from 140 m S-l for Cd to 300 m s - 1 for Mo [25]. There are at least two theories of macroparticle emission . McClure [26J has proposed a model based upon plasma expansion in which ions from the ion cloud above the cathode spot are accelerated towards the liquid surface of the active cathode spot. The vapour jet recoil force presses inward on the molten metal and material is pushed towards the edge of the crater leading to the observed spatial distribution. The model predicts values for the velocity of Cu macroparticles of 20-100 m S-I . A second model has been proposed based upon the concept of explosive emission [25]. The model assumes that electron emission is concentrated on small surface protrusions on the cathode. The protrusion is rapidly heated by electron emission through it until it eventually explodes at a peak pressure of 2 x 10 -10 Pa. The impact from the explosion generates other surface defects and other sites for the process to be repeated.

P. J. Marlin et al. I Thin film deposition by filtered arc evaporation

138

4. Methods of macroparticle control

4.1. Shielding

The simplest method of reducing the number of macroparticles in the condensing film is to take advantage of the spatial distribution characteristics of the macroparticle emission and place the substrate in the deposition chamber in such a location that it is not in direct line-of-sight of the cathode. Brandolf [27J has proposed the use of a shield which is interposed between the substrate and arc source (Fig. 1(a)). The substrate and shield are then biased to attract the plasma around the shield and onto the substrate. The macroparticles

are trapped on the front of the shield and do not impinge on the substrate. The principal disadvantage of this method is that the deposition rate is reduced although macroparticle-free coatings can be produced. 4.2. Magnetic steering Ertiirk et al. [28J have concluded that droplet forma-

tion can be reduced by considering the following points: (i) The volume of molten material must be kept low. This implies small arc craters, low arc current and short cathode spot lifetimes. (ii) The arc path should be extended to reduce widespread melting of the cathode. The cathode must also

I------:~~~~ga---Cathode Magnetic

System

Bias

N~~p2tf:l

r---H,*~~;;=;;:=;;;~-\--Sh j e Id

Cathode

~'-"'-""I~:l-~~-s u bs tr ate

(a)

(b)

Plasma Duct Substrate

Arc Source

~

I~IL'I---I-

Magnetic Coils

Bias (c)

~::::::'----""':k~

(d)

Fig.!. (a) Macroparticle filtering by shielding [27]; (b) magnetic steering by means of an "internal" magnetic field [28-30]; (c) modified arc evaporation source employing an "external" solenoidal magnetic field [17, 31]; (d) macroparticie filtering system using a magnetic plasma duct system [36, 40,41,46].

P. J. Martin et al. / Thin film deposition by filtered. arc evaporation

be cooled effectively. Boxman and Goldsmith [25J have estimated the front surface temperature of a Ti 100 mm diameter cathode, 10 mm thick supporting a 100 A arc to be 41°C. (iii) A high melting point of the cathode material reduces the mean radius of the craters and a low thermal conductivity limits the extension of the molten areas. The movement of the cathode spot on the cathode surface may be controlled by the use of "internal" magnetic fields, i.e. fields that are applied directly behind the cathode itself. Erturk et al. used a permanent magnet to exert a Hall force on the cathode spot with the arc acting as the current carrying conductor. The effect is to move the arc around the surface of the cathode in a predetermined track (Fig. 1(b)). The technique has been used previously and was refined by Ramalingham [29]. The method may be used to achieve a more uniform erosion of the cathode surface or to steer the arc to specific areas on a multi-element segmented cathode in order to produce alloys or modulated layers. Erturk et al. demonstrated that the steered arc reduces the average crater size relative to that of the random arc. The macroparticle content was reduced in TiN films produced by the steered arc but the deposition rate was also reduced from 13 urn h -1 to 5 urn h -1 (80 A arc, 125 mm diameter steered arc, 63 mm diameter random arc). The steered arc reduces the number of large macroparticles, particularly in the presence of a N 2 atmosphere in TiN deposition but does not eliminate all macroparticles. Steffens et al. [30J reported a similar result for TiN deposition but found that magnetic fields up to 10 mT did not reduce droplet emission in non-reactive evaporation. Several studies have been made of the influence of "external" fields on arc evaporation. Generally these fields are applied by means of a magnetic solenoid mounted directly in front of the cathode (Fig. I(c)). In the case of Ti the solenoidal field has the effect of increasing the ionized fraction of the emission to 100% and the average charge per ion to 2.08 [17]. The enhancement in ionization results in an increase in the photon emission from Ti + and Ti2+ species as the magnetic field is increased from 0 to 10 mT [15]. The ion energy is also increased through Hall acceleration. It has been suggested that the macroparticles are evaporated by collision with electrons in this arrangement since the electron density is increased greatly by the application of the magnetic field. Akari et al. [31J reported that a 10 mT field reduced the macroparticle content by a factor of 5.5. The external solenoid source has also been adapted to produce a high intensity ion source [32]. The source was used to sputter material from a water-cooled target onto a substrate facing the target. Films of Ti, AI, Cu and Nb were found to be fine grained and free of

139

macro particles. The deposition rate ranged from 2 to 6 urn h -1 depending on the target material and bias. 4.3. Magnetic filtering

Several types of magnetic filtering devices have been described in the patent and scientific literature [33-35]. These devices used magnetic fields of varying complexity to filter the macroparticles. 4.4. Magnetic plasma duct It was first demonstrated by Aksenov and co-workers

[36-38J that a toroidal magnetic field may be used to remove neutrals and macroparticles from an arc source (Fig. led)). The technology was developed from the early work on impurity removal from hydrogen plasmas [39, 40]. The plasma duct filter is a quarter torus with a magnetic field parallel to the walls of the torus. The condition for transporting a stream of low-density plasma along a toroidal field is given by:

r

l

M c

v6

->- where u = - - - " u Vo Z eRH

(1)

u is the velocity of the centrifugal ion drift in the field, M, Z and Va are the mass, charge and longitudinal velocity respectively of the ions, R is the radius of curvature of the magnetic lines of force, l the length of the toroidal field and I' the radius of the plasma duct. If magnetic filtering alone is used then by substituting typical values of 1'=4 em, and £/Z=50 eV, then a magnetic field of the order of 2 T is necessary [36J. Aksenov and co-workers found that if the duct was biased positively to 15-20 V then the saturation Ti ion current at the exit of the system increased from 400 mA to 900 mA due to plasma-optical focusing effects. In the case of crossed electric and magnetic fields the necessary magnetic field value is obtained from the requirement: (2)

where Pe, Pi are the electron and ion Larmor radii respectively and a the radius of the duct. The magnetic field is reduced to the order of 0.02-0.08 T. Storer et al. [4lJ studied the transport of Fe through straight and curved magnetic ducts and found the transmission increased with magnetic field strength and saturated at a field at which the ion gyroradius equals the radius of the plasma duct. It was found that the optimum field strength for coupling the plasma from the source to the duct was 0.Q15 T. Sanders et al. [42J have modelled the plasma flow in a toroidal plasma guide using the flux-tube model of Morozov [43]. Individual Ti + ions with initial speeds selected from a 23 eV Maxwellian distribution centred on 65 eV were injected into the system which comprised an injection region and torus region. The torus had a radius of curvature of 21.8 em and the solenoid a radius

140

P. J. Martin et al. / Thin film deposition by filtered arc evaporation

of 3.9 em, conditions approximating those of the original Aksenov device. The main conclusion reached was that the efficiency of the filter saturated with increasing duct potential and that a maximum total efficiency of 59% was achieved for a duct potential of 70 V. Several researchers have constructed filtered arc devices based on the magnetic plasma duct with cathode diameters ranging from 15 mm to 100 mm [19, 22, 34]. The devices have been successfully used in the deposition of metals, alloys, compounds and diamond-like carbon.

5. Beam properties of filtered arcs The magnetic plasma duct method of filtering macroparticles from arc discharges provides a useful means of obtaining a low-energy, high-intensity plasma beam. When the substrate is biased to high negative values the energy of the depositing ions is increased, the substrate is heated and sputtering and diffusion processes occur. Figure 2 shows the penetration of Ti into a steel substrate as a result of biasing the steel to potentials of -50 V, -200 V and -900 V during bombardment with a 500 rnA Ti beam from a filtered arc source. The penetration is shown in the form of a depth profile obtained by monitoring the Ti + signal from an ion probe during etching of the surface. The - 50 V and - 200 V samples were heated to 400 °C as a result of the bombardment and the temperature monitored by thermocouples and a radiation pyrometer. The depth profile shows a rapid drop in the Ti + signal as the sample is etched to a depth of 0.2 urn and 0.5 urn respectively. The plateau regions are due to the thickness of the deposited layers and the rate of decrease is the result of the penetration of the titanium into the surface and a consequence of the sputter etching process in the

o

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Depth (urn) Fig. 2. Depth profile of Ti diffusion into stainless steel: (a) - 50 V, 400 °C; (b) -200 V, 400°C; (c) -900 V, 600°C.

ion probe. However, in the case of the - 900 V sample the surface temperature reached 600 °C and there was no obvious surface layer (due to self-sputtering during high bias deposition). The titanium signal is detected well above the background level even at depths of 1.6 urn. Since this degree of penetration is well above the theoretical maximum of2.7 nm for any implantation effects [44J (most extreme case is approximately 3 keV Ti + assuming I keY Ti:J+ under the present conditions), the data indicate the influence of radiation enhanced diffusion at the elevated surface temperature. This effect is used in conventional arc coating to promote adhesion prior to deposition at lower bias values. The plasma properties of the filtered arc may also be used to increase the effective area of deposition by placing suitable magnetic coils around the exit of the plasma duct and scanning the magnetic field in the x- y plane. The plasma is deflected and the area of deposition may be expanded. The undeflected ion beams have Gaussian shapes with a FWHM of 30-40 mm for sources with diameters of 58-100 mm. The deposition areas have been increased up to 100 mm with beam scanning [19, 22J.

6. Film deposition by filtered arc The filtered arc has been used to deposit a wide range of materials. Aksenov et al. [22J studied the deposition of Ti originally but gave no data on deposition rate although beam currents of 900 rnA were reported for 100 A arcs. Falabella and Sanders [34J reported a maximum Ti ion beam of 3 A using a non-toroidal "knee" filtered arc source. Other metals successfully deposited include AI, V, Cu, Nb, Zr, Ta and Mo at deposition rates up to 40 urn h- 1 [19,21,22]. The film quality is equivalent to that produced by other vapour deposition techniques. In the case of Nb the superconducting resistivity ratio has been measured at 28 for films prepared in a moderate vacuum of 10- 4 to 10- 5 Pa. The quality should be improved greatly under UHV conditions [45]. Compound films are readily synthesized by introducing reactive gases into the deposition chamber. In the case of the filtered arc this can be achieved without causing significant poisoning of the cathode since the source is remote from the region where the gas introduction takes place. Nitrides of Ti and Mo have been prepared with microhardnesses of 3600 and 2400 kg mm -2 respectively [22]. The properties of TiN are strongly dependent upon the deposition conditions, in particular the substrate temperature and bias. An increase in substrate bias (increasing ion energy) is accompanied by a decrease in lattice constant and a reduction in stress [46]. Similar results reported for d.

P. J. Martin et al. / Thin film deposition by filtered arc evaporation

sputter deposition of TiN were interpreted in terms of ion peening effects resulting in stress relief [47]. Recent theoretical studies suggest that the film. stress may be. modified according to the degree of particle momentum in the depositing atoms [48]. TiN films prepared by the filtered arc exhibit a greatly enhanced corrosion resistance relative to films prepared by unfiltered arcs. Vershina et al. [49J found that 211m thick Ti and TiN films provided maximum protection to the surface of brass and steel specimens when placed in corrosive media. Films prepared by the unfiltered arc to thicknesses of 0.6-4.5 um did not withstand corrosion testing due to the porosity caused by the presence of macroparticles. Other nitride films deposited by the filtered arc include VN, ZrN, NbN and Fe 3N4 • Carbide films may be synthesized by introducing CH 4 into the deposition chamber [46]. TiC films have been deposited at rates up to 36 urn h -1 with a microhardness of 3600 HVlO' The lattice parameter was found to increase to a maximum of approximately 4.36 nm with increasing substrate bias as reported also for r.f deposited TiC [47]. High quality oxide films suitable for optical applications have been synthesized by depositing elemental metals in the presence of oxygen. Aksenov et al. [50J deposited Al 20 3 at pressures of 10- 3-1.0 Pa at rates up to 12 urn h -1. The microhardness of Al20 3 produced by the filtered arc has been measured to be 10801500 kg mm -z. Optical quality Alz0 3 films with a refractive index of 1.67 and extinction coefficient of 6 x 10- 4 have been deposited at rates up to 3511m h- 1 [46, 50-52]. Other oxides successfully deposited by the filtered arc include rto; VOz, CrZ03' NbzOs and z-o, [46]. Table 2 lists the optical properties of oxide films deposited by the filtered arc. The filtered arc has also been used to deposit cermet films on glass surfaces suitable for thermal control, The films were prepared by depositing metal films at an oblique angle of 65° in the presence of oxygen [53]. The TABLE 2. Optical properties of materials deposited by filtered arc [19J

141

substrate was unheated during deposition. The ratio of the oxide to metal in the deposited cermets could be controlled by varying the oxygen flow rate to the deposition chamber. Cermets of AljAI20 3 , TijTiO z and CrjCrz03 were prepared with the aluminium based films exhibiting a pronounced angular selectivity in optical transmittance. High resolution microscopy revealed that the cermet structure was a modulated array of tightly packed fine columns. The filtered arc method has been shown to be a useful technique to produce energetic beams of carbon ions [54J. The energy of the carbon ions from the arc has been measured to be approximately 20 eV [55]. Strel'nitskii et al. [56J have deposited diamond-like carbon (DLC) using a filtered arc onto Cu, steel and Ti substrates. A d.c. and rJ. bias was applied to the cooled substrates and exceptional film properties were measured. The density of the DLC was measured to be 4 g cm - 3 for films condensed from ions with an energy 40 eV ~ E ~ 70 eV. The phase of this material was described as dense crystalline Cs carbon [57]. The microhardness has also been measured to be in excess of that of natural diamond [58J and electron energy loss data show the material to be more diamond-like than graphitic in nature [59]. The optical properties of the material have been determined and the refractive index found to be in the range 2.4-2.8 (0.4-10 urn) and the absorption coefficient approximately 0.1 at 0.6 urn [60]. The filtered arc technique essentially removes the graphitic material from the condensing plasma although DLC material may also be produced from unfiltered sources [60]. The main advantage of the filtered arc over other methods of depositing DLC is the high deposition rate (approximately 20 urn h- 1) , absence of hydrogen and the low substrate temperature (ambient). The arc-deposited DLC is highly stressed with compressive stresses as high as 5-]2 GPa [61]. The stress may be reduced at the expense of microhardness by the addition of hydrogen during deposition [19]. The hardness is then reduced to approximately 1600 Hv, but the optical properties are improved significantly in the visible region (Table 2). The hydrogenated films are also more adherent.

Film

DLC C+CH.. A1 20 3

AIN V0 2 (ambient)

'At 1 urn.

2.490 1.810 1.670 2.150 3.56 2.56' 2.96 1.57' 2.190 2.17

0.71 0.02 <0.0005 <0.0010

1.16 0.14' 0.41 1.72' <0.0015 <0.0010

7. Summary Filtering of the vacuum arc evaporation source with a magnetic plasma duct is seen to be the most effective method of producing macroparticIe-free coatings. Other filtering methods have varying degrees of success and many of these modifications may also be incorporated in a magnetic duct device to further improve the performance. The filtered arc technique is receiving more attention principally due to its capability of producing

P. J. Martin et aI. / Thin film deposition by filtered arc evaporation

142

high quality metal and compound films at practical rates. Future developments should see the use of multiple cathodes to produce modulated or multilayer films and the scaling of the technology for widespread industrial applications. The technique 'will ultimately complement existing PVD methods rather than compete with them.

Acknowledgment The authors wish to thank M. Petravic of the Australian National University for providing the ion microprobe depth profiling data.

References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

16 17 18 19 20 21 22 23

24 25 26 27

P. J. Martin, J. Mater. Sci.• 21 (1986) J. F. A. Smidt, Int. Mater. Rev., 35 (1990) 61. P. C. Johnson, Phys. Thin Films, 14 (1989) 128. D. M. Sanders, D. B. Boercker and S. FalabeIla, IEEE Trans. Plasma Sci., 18 (1990) 883. 1. E. Daalder, Physica C, 104 (1981) 91. R. Boxman and S. Goldsmith, J. Appl. Phys., 51 (1980) 3644. C. Bergman, Surf Coat. Techno!', 36 (1988) 243. J. E. Daalder, Thesis, Cathode Erosion of Metal Vapour Arcs in Vacuum, Eindhoven University, 1978. B. Jiittner, J. Phys. D, 14 (1981) 1265. W. D. Davis and H. C. Miller,!. Appl. Phys., 40 (1969) 2212. A. A. Plyutto, V. N. Ryzhkov and A. T. Kapin, Sou. Phys. J ETP, 20 (1965) 328. J. Kutzner and H. C. Miller, IEEE Trans. Plasma Sci.. 17 (1989) 688. N. Radic and B. Santic, IEEE Trans. Plasma Sci., 17 (1989) 679. Z. H. Wang, P. D. Swift, A. 1. Studa, D. R. McKenzie, B. James and L S. Falconer, Appl. Opt.• 29 (1990) 5145. P. J. Martin, R. P. Netterfield, D. R. McKenzie, l. S. Falconer, C. Pacey, P. Thomas and W. G. Sainty, J. Vae. Sci. Technol. A, 5 (1987) 22. G. Eckhardt, J. Appl. Phys., 46 (1975) 3282. P. Sathrum and B. F. CoIl, Surf Coat. Technol., 50 (1992) 103. l. Brown, J. E. Galvin, B. F. Gavin and R. A. MacGill, Appl. Phys. Lett., 47 (1985) 358. P. J. Martin, R. P. Netterfield and T. J. Kinder, Thin Solid Films, 153 (1990) 77. P.1. Martin, D. R. McKenzie, R. P. Netterfield, P. Swift and K. H. MUller, Thin Solid Films, 153 (1987) 91. A. A. Andreev, Fiz, Khim. Obrab. Mater., 3 (1977) 89. 1. 1. Aksenov, V. A. Belous, V. G. Padalka and V. M. Khoroshikh, Fiz. Khim. Obrab. Mater., 6 (1977) 89. 1. 1. Aksenov, L 1. Konovalov, E. E. Kudryavtseva, V. V. Kunchenko, V. G. Padalka and V, M. Khoroshikh, Sov. Phys. Tech. Phys., 29 (1984) 893A. W. Baouchi and A. 1. Perry, Surf Coat. Technol., 49 (1991) 253. R. L. Boxman and S. Goldsmith, Surf Coat. Teelmol., 52 (1992) 39. G. W. McClure, J. Appl. Phys., 45 (1974) 2078. H. Brandolf, US Patent 4 51l 593, 1985.

28 E. Ertiirk, H.-J. Heuvel and H.-G. Dedrichs, Surf Coat. Techno!.. 39 (1989) 455. 29 S. Ramalingham, US Patent 4673477, 1987. 30 H.-D. Steffens, M. Mack, K. Miihwald and K. Reichel, Surf Coat. Technol., 46 (1991) 65. 3\ K. Akari, H. Tamagaki, R. Kumakiri, K. Tsuji, E. S. Koh and C. N. Tai, Surf Coat. Technol., 43 (\990) 312. 32 D. M. Sanders, J. Vac. Sci. Techno!. A., 6 (1987) 1929. 33 l. 1. Axenov, V. A. Belous, V. G. Padalka and V. M. Khoroshikh, US Patent 4 452 686, 1984. 34 S. Falabella and D. M. Sanders, J. Vac. Sci. Technol., 10 (1992) 394. 35 G. V. Kljuchko, V. G. Padalka, L. P. Sablev and R. l. Stupak, US Patent 4 492 845, 1985. 36 l. l. Aksenov, V. A. Belous, V. G. Padalka and V. M. Khoroshikh, Sov. J. Plasma Phys., 4 (1978) 425. 37 1. 1. Aksenov, V. G. Padalka, N. S. Repalov and V. M. Khoroshikh, SOlJ. J. Plasma Phys., 6 (1980) 173. 38 I. I. Aksenov, A. N. Belokhvotikov, V. G. Padalka, N. S. Repalov and V. M. Khoroshikh, Plasma Phys. Control. Fusion, 28 (\986) 761. 39 V. S. Voitsenya, A. G. Gorbanyuk, I. N. Onishenko and B. G. Safronov, Sov. Phys. Tech. Phys., 9 (1964) 221. 40 V. S. Voitsenya, A. G. Gorbanyuk, I. N. Onishenko, B. G. Safronov, N. A. Khizhnyak and V. V. Shkoda, so». Phys. Tech. Phys.. 12 (1967) 185. 41 J. Storer, J. E. Galvin and 1. G. Brown, J. Appl. Phys., 66 (1989) 5245. 42 D. M. Sanders, D. B. Boercker and S. Falabella, IEEE TrailS. Plasma sa: 18 (1990) 883. 43 A. 1. Morozov, SOD. Phys. Dokl., 10 (\966) 775. 44 J. P. Biersack, in R. Kelly [ed.), /011 Beam Modification of Insulators, Elsevier, Amsterdam, 1987. 45 Y. Igarashi and M. Kanayama, J. Appl. Phys., 57 (1985) 849. 46 P. J. Martin, R. P. Netterfield, T. J. Kinder and L. Descotes, Surf Coat. Technol., 49 (1991) 239. 47 J.-E. Sundgren, B.-O. Johansson, S.-E. Karlsson and H. T. G. Henzell, Thin Solid Films, 105 (1983) 367. 48 H. Windischmann, J. Vac. Sci. Technol, A, 9 (1990) 2431. 49 A. K. Vershina, I. A. Bel'chin, A. A. Pitel'ko and S. D. Izotova, Fiz. Khim. Obrab. Mater., 24 (1990) 93, 50 l. 1. Aksenov, V. G. Bren, V. G. Padalka and V. M. Khoroshikh, Sov. Phys. Tech. Phys., 23 (1978) 651. 51 H. Randhawa, J. Vae. Sci. Technol. A, 7 (1989) 2346. 52 T. D. Schemmel, R. L. Cunningham and H. Randhawa, Thin Solid Films, 18/ (1989) 597. 53 G. B. Smith, M. W. Ng, R. J. Ditchburn, P. J. Martin and R. P. Netterfield, Proe. SP/E, 1536 (1991) 126. 54 V. E. Strel'nitskii, 1. 1. Aksenov, S. 1. Vakula, V. G. Padalka and V. A. Belous, Sov. Phys. Tech. Phys. Lett., 4 (1978) 546. 55 P. J. Martin, S. W. Filipczuk, R. P. Netterfield, J. S. Field, D. F. Whitnall, D. R. McKenzie, J. Mater. Sci. Lett., 7 (1988) 410. 56 V. E. Strel'nitskii, 1. 1. Aksenov, S. 1. Vakula, V. G. Padalka and V. A. Belous, SaD. Phys. Tech. Phys.. 23 (1978) 222. 57 N. N. Matyushenko, V. E. Strel'nitskii and V. A. Gusev, JETP Lett., 30 (1980) 199. 58 l. 1. Aksenov, S. 1. Vakula, V. G. Padalka, V. E. Strel'nitskii and V. M. Khoroshikh, Sov. Phys. Tech. Phys., 25 (1980) 1164. 59 S. D. Berger, D. R. McKenzie and P. J. Martin, Phil. Mag. Lett., 57 (1988) 285. 60 S. 1. Vakula, V. G. Padalka, V. E. Strel'nitskii and A. 1. Usoskin, Sov. Phys. Tech. Phys., 5 (1979) 573. 61 D. R. McKenzie, D. Muller and B. A. Pailthorpe, Phys. Rev. Lett.. 67 (199\) 773.