- Email: [email protected]
Plasma and deposition enhancement by modified arc evaporation source
Surface and Coatings Technology, 50 (1992) 103-109
103
Plasma and deposition enhancement by modified arc evaporation source* P. S a t h r u m
a n d B. F. C o i l
Multi-Arc Scientific Coatings, 200 Roundhill Drive, Rockaway, NJ 07866 (USA)
(Received March 10, 1991; accepted in final form May 21, 1991)
Abstract
The purpose of this paper is to present the "enhanced arc system" based on a modified straight plasma optics system. The magnetic field strength and orientation in the apparatus together result in an enhancement of plasma ionization, energy and density and near elimination of macroparticles. Ionization and energy characteristics of the flux impinging TiN films are described as functions of magnetic field strength. Langmuir probe measurements of voltage and current were made using a planar shielded substrate as the probe. Plasma species were observed using emission spectrometry. The ionized fraction of the titanium vapor at the substrate was determined by comparing the deposition rate while excluding ions with the deposition rate under normal conditions. Deposited TiN films were analyzed by profilometry and scanning electron microscopy. The results were an increase in the density of ionized and excited species, especially molecular and atomic nitrogen, with increasing magnetic field strength, which corresponds to an increase in the electron temperature; the average charge carried by nitrogen and titanium ions was approximately 2.1 e; 100% of the titanium vapor producing TiN films was ionized and the deposited films were nearly free of macroparticles.
1. Introduction
The cathodic arc process produces a highly ionized, energetic and dense vapor. This is advantageous for the deposition of well-adhered and dense films at a high rate. A disadvantage of the cathodic arc process is the production of macroparticles along with the ionized vapor. These macroparticles are typically incorporated into the films resulting in a degradation of film properties. Several modified arc evaporation techniques have been described which have shown that a reduction of these macroparticles along with even higher levels of vapor ionization, energy and density can be achieved by using magnetic fields [1--6]. These techniques have shown that the arc, the plasma and the deposited films can be strongly influenced by a magnetic field. Magnetic fields have also been used to enhance the synthesis in compound film formation and to increase the deposition rate, microhardness and adhesion of films [3, 7]. T h e purpose of this p a p e r is to present the " e n h a n c e d arc" system based on a modified straight plasma optics system. The magnetic field strength and orientation in the apparatus result in both a modification of the arc *Paper presented at the 18th International Conference on Metallurgical Coatings and Thin Films, San Diego, CA, 22-26 April 1991.
discharge and an increase in the density of the ionized and excited species. Titanium nitride films deposited with this system are nearly free of macroparticles. Ionization and energy characteristics of the plasma are described as functions of magnetic field strength. Langmuir probe m e a s u r e m e n t s of voltage and current were made using a planar shielded probe. Plasma species were observed using emission spectrometry. The ionized fraction of the titanium vapor and the average charge per ion were determined. Finally, the influence of the magnetic field on the surface properties of deposited films is described.
2. Enhanced arc apparatus
Figure 1 illustrates the enhanced arc system. A solenoid was mounted between the cathode and the vacuum chamber. Steel magnetic returns were m o u n t e d around the coil to contain the magnetic field. A soft magnetic core with an air gap was m o u n t e d inside the solenoid to optimize the shape and strength of the magnetic field. The arc was confined on the face of the cathode by the magnetic field produced by the solenoid and core. The solenoid, magnetic returns and core assembly will be referred to as the plasma guide.
Elsevier Sequoia
104
P. Sathrum, B. F. Coil / Plasma and deposition enhancement by modified arc
I
Ebert-type Jarrel-Ash scanning monochrometer. The monochrometer was sensitive in the range 230-990 nm with a resolution of less than 0.1 nm. The acceptance angle was 4 °. The deposition rates were determined using a microbalance with a repeatability of 20/xg. The average surface roughness was measured with a Dektak profilometer. All the experiments were done with 890 G magnetic field strength, 20 mTorr nitrogen pressure, 40 A cathode current, - 8 0 V substrate bias and a 375 mm cathodeto-substrate distance except where noted. For the conventional arc experiments the plasma guide was removed and the cathode-to-substrate distance was 300 mm. (The difference is the length of the plasma guide.)
~\\\\\\\\\\\\\\\\\'x5
Fig. 1. Enhanced arc apparatus: 1, solenoid; 2, soft magnetic core; 3, magnet returns; 4, cathode; 5, vacuum chamber; 6, substrate. To S p e c t r o m e t e r
i
To P y r o m e t e r
Fig. 2. Langmuir probe: 1, substrate; 2, ceramic fixture; 3, grounded shield; 4, cathode.
4. Results and discussion
4.1. Langrnuirprobe The substrate current was measured as a function of applied voltage and magnetic field strength. The results are shown in Fig. 3. A voltage--current curve for the conventional arc under similar conditions is also shown for reference. In Fig. 3, positive ions collected by a negative voltage are shown as a positive current and electrons collected by a positive voltage are shown as a negative current. It can be seen that the positive ion current saturates at negative voltage. The saturated ion current increases with increasing magnetic field strength at a rate that is nearly linear up to a magnetic field strength of
The length of the plasma guide was 75.0 mm and the inside diameter was 50.8 mm. Currents of up to 70 A through the solenoid produced a maximum field strength of 1500 G measured along the solenoid axis. No other magnets were used.
335 Gauss 2 b
1
3. Experimental details Figure 2 illustrates the Langmuir probe experimental system. The substrate was located 375 mm in front of the cathode. The cathode and substrate were perpendicular to their common symmetry axis. The substrate, which was 12.7 mm in diameter, was recessed into a ceramic fixture. A grounded shield 1.2 mm thick with an aperture of 11.87 mm was located 1.0 mm in front of the substrate. Plasma emission was monitored through an observation tube aimed to intersect the symmetry axis 375 mm in front of the cathode. Light from the emission was transmitted through a quartz window and through a quartz fiber optic cable to the input slit of a 0.5 m
8
670 Gauss
•
890 Gauss
•
1450 Gauss
N
Conventional Arc {0 G)
30 -300
-200 VOLTAGE
-100 (Y) -0.1 I
-0.2' o =
-0.3 -
-0.4 L
Fig. 3. Substrate current vs. vo]tage.
40
P. Sathtum, B. F. Coil / Plasma and deposition enhancement by modified arc
approximately 890 G. Above 890 G, the saturated ion current stays approximately the same. The saturated current collected at a negative bias is primarily due to the flux of titanium and nitrogen ions entering the sheath separating the plasma from the substrate [8]. The application of a magnetic field in a plasma optics system can have a significant effect on ion flux levels, as has been shown by Aksenov et al. [1]. The increase in ion flux is a result of plasma focusing and the creation of additional ions through electron impact ionization. The motion and density of electrons in the magnetic field both greatly increase the probability of electron-particle collisions. According to Aksenov and coworkers the probability of electron-molecule inelastic collisions is proportional to the square of the magnetic field strength [7]. If these collisions are energetic enough, excitation, dissociation of molecular species and ionization can occur. Electron energy was estimated from the positive current in the negative potential region before ion saturation using the method described by Clements [9]. Electron temperatures calculated by this method are listed in Table 1. The results show the electron temperature increasing nearly linearly with magnetic field strength to a maximum of 25 eV at 1450 G. Table 2 lists the ionization and dissociation threshold energies for the plasma species present in this experiment [10]. These results indicate that the electrons at higher magnetic field strengths are more than energetic enough to dissociate and ionize other species. The electron temperatures calculated here are for electrons in the vicinity of the substrate traveling perpendicular to the T A B L E 1. Electron temperature kTe, plasma density n and floating potential Vf vs. magnetic field strength Magnetic field
kTe (eV)
n ( × 10 to cm -3)
- Vt (V)
2.5±1.5 6.0±1.5 10.0±1.5 12.0±1.5 ~.0±2.0
0.80±0.15 1.01±0.26 1.48±0.32 1.66±0.32
1.0±0.4 5.8±0.4 8.9±0.4 5.5±0.4 2.5±0.4
(G) 0 335 670 8~ 1450
T A B L E 2. Ionization and dissociation threshold energies Species
N 2+ Ti ÷ N+ N
Ionization threshold
Dissociation energy
(eV)
(eV)
15.6 6.8 14.6 9.9
105
substrate. Electrons in the plasma guide closer to the cathode probably have much higher temperatures. This is indicated by the high intensity of the plasma emission coming from this area. Table 1 also lists the plasma density n and the floating potential Vf as functions of magnetic field strength. The plasma density was calculated from the electron temperature kTe, the saturated ion current and the average charge per ion calculated in Section 4.4. [11]. The floating potential was measured directly. The plasma potential can be roughly estimated from the saturation point of the negative current curves [8]. 4.2. Emission spectrometry
Emission spectrometry was used to detect and identify species present in the plasma over the wavelength range 230-900 nm. Plasma emission was monitored at the same distance as the Langmuir probe with the Langmuir probe removed. Measurements were made at different values of magnetic field strength and nitrogen pressure. The results show the presence of ionized and excited titanium and significant amounts of ionized and excited molecular and atomic nitrogen. This is in contrast with the conventional arc process where molecular nitrogen emission intensity is relatively weak and atomic nitrogen emission is not present at all [12]. Emission from the decaying species Ti, Ti +, Nz, N2 +, N and N + was observed. Ti 2+ and N 2+ emission lines were not identified. Figure 4(a) shows the variation in emission intensity with magnetic field strength for the strongest Ti, Ti +, Nz, N2 ÷, N and N ÷ lines. All lines are strongly influenced by magnetic field strength. In Fig. 4(a), it can be seen that there is an especially strong association between the intensity of emission of nitrogen species and magnetic field strength. The presence of Nz, N2 +, N and N ÷ emissions which increase nearly linearly with magnetic field strength indicates that electron impact excitation, dissociation and ionization of nitrogen is occurring. The excitation thresholds for these emissions are 11.1, 18.7, 12.0 and 20.7 eV respectively [13]. Electron temperature estimates made from the Langmuir probe experiments (Section 4.1) show that electron temperatures exceeding these excitation thresholds are reached. All nitrogen emissions including Nz +, Nz, N ÷ and N peak in the magnetic field strength range 900-1200 G. The rapid decline in N2 emission intensity and saturation of N2 ÷ emission intensity may be the result of nearly complete dissociation and ionization of Nz. Similarly, complete dissociation and ionization of nitrogen in a plasma optics system was reported by Aksenov et al. [7]. The Ti + and Ti emission lines as shown in Fig. 4(a) increase rapidly up to 335 G above which the emissions
106
P. Sathrum, B. F. Coil / Plasma and deposition enhancement by modified arc
•
N~ (391.4 n m )
---.--.--o---- N2 (357.7 rim)
00
500
(a)
1000
•
N + (568.0 rim)
•
N (746.8 n m )
.L
Ti + (334.9 nm)
.I
Ti (521.0 n m )
1500
strength of 890 G for the same emission lines as in Fig. 4(a). The characteristic features of this graph are similar to those discussed by Bergman for emission studies done with the conventional arc (zero magnetic field strength) [12]. The most notable difference is the high intensity of the nitrogen emission lines. The similarities indicate that the same reactions as reported by Bergman are taking place. Between 0 and 15 mTorr, the present results are also similar to those of Demidenko et al. [14] and Martin et al. [5] as well as Bergman. They attribute increases in the nitrogen and titanium emissions to charge exchange reactions of the type Ti"+ +NE
~ Ti ('-1)+ +N2 +
MAGNETIC FIELD (Gauss)
The rapid decrease in the intensity of the N2 + emission above 15 mTorr occurs because the initially high energy of Ti z+ species is lost in collisions with nitrogen and falls below the 14.0 eV threshold energy needed for this reaction to occur. The Ti ÷ to Ti reaction does not occur because it requires an excessively high ion threshold energy of 32.3 eV. In the region above 15 mTorr, the behavior of the Ti ÷ and Ti emissions can be attributed to the threebody reaction
4
c 3
i 2
Ti"+ + e +N2
) Ti('-l)+ + N2
1
0 0
(b)
10
20
30
PRESSURE
40
50
60
(mTorr)
Fig. 4. Variation in emission intensity with (a) magnetic field strength and (b) nitrogen pressure.
level off. From 0 to 335 G, the steep, nearly linear increase in the titanium emissions is probably due to plasma focusing and electron impact ionization. The abrupt saturation of the titanium emission lines at 335 G correlates to an electron temperature of approximately 6 eV, according to the results presented in Table 1, which is close to the ionization threshold of titanium. The abrupt saturation of the titanium emission lines may be the result of nearly complete ionization of titanium. The results presented in Section 4.3 show that the titanium vapor is 100% ionized. The presence of the excited titanium emission line which imitates the behavior of the ionized titanium emission line at a lower intensity is probably due to three-body electron recombination as described below. Figure 4(b) shows the variation in emission intensity with nitrogen pressure at a constant magnetic field
The probability of this reaction increases with increasing nitrogen concentration. This is shown in Fig. 4(b) by the decline of the Ti ÷ emission accompanied by a rise in the Ti emission and the decline of both species above 40 mTorr. These results and observations indicate that in the magnetic field of the enhanced arc apparatus, electron-particle collisions result in significant excitation, dissociation and ionization of titanium and nitrogen species. In addition to inelastic electron-particle collisions, charge exchange and three-body reactions typical of the conventional arc process are still occurring. Results and discussion in Sections 4.1 and 4.2 indicate that the contribution of nitrogen ions to the ion flux on a substrate is significant. This is in contrast with the conventional arc process where nitrogen ions contribute only a very small amount to the current on a substrate [12]. Finally, a direct comparison of the emission intensities of the Na ÷, N2, Ti + and Ti lines studied here showed that emissions in the conventional arc system were only 2.3%, 1.3%, 25% and 11% respectively of the emissions in the enhanced arc system under the same operating conditions. This supports the conclusion that significant enhancement of titanium and nitrogen excitation and ionization is occurring in the enhanced arc system.
107
P. Sathrum, B. F. Coil / Plasma and deposition enhancement by modified arc
4.3. Ionized vapor fraction The ionized titanium vapor fraction was found to be 100% using the ion exclusion method as used by Bergman [12]. The ionized fraction of the vapor impinging on the substrate was determined by comparing the deposition rate, while excluding ions, with the deposition rate when ion deposition was allowed. Ion deposition was done at a magnetic field strength of 890 G for 30 min. Ion exclusion deposition was done with the same parameters, except that a screen with a transparency of 0.903 (identical with the type used by Bergman) was placed 1 cm in front of the substrate and biased at - 100 V to collect ions. The Debye sheath surrounding the wires made the screen's effective ionstopping power much greater than the geometry suggests. The substrate was also biased at + 30 V to repel any ions that did penetrate the screen. The result was that there was no gain in mass when ions were excluded from deposition. No measurable amount of macroparticle flux was included in film deposition. Although macroparticles possess some amount of charge [6], their charge-to-mass ratio was assumed to be negligible so that macroparticle motion was not appreciably influenced by the electric fields used in this experiment. Ionization percentages of 68-83% in the conventional arc process, as reported by Bergman, reflect primarily the deposition of macroparticles. The absence of mass gain under ion exclusion deposition conditions supports the conclusion that 100% of the titanium vapor producing TiN films is ionized in the enhanced arc process.
4. 4. Average charge per ion The average charge per ion was found to be 2.08 + 0.28 e. The average charge per ion was determined from the deposition rate and the collected ion charge on a substrate. The deposition rate was measured in mass gained per second per unit area on a substrate clamped in a fixture exposing a deposition area of 3.23 cm 2. Owing to gas scattering, the area increased to approximately 3.53 cm2 at 20 mTorr. The collected ion charge was determined using a value of ion current density obtained from the Langnuir probe experiments presented in Section 4.1 for the same parameters as were used in deposition. The substrate temperature was less than 430 °C throughout the deposition. The atomic ratio of nitrogen to titanium in the film was determined using X-ray photoelectron spectrometry. The average charge per ion was determined assuming that the nitrogen mass contribution to the film was from ionized molecular nitrogen. (i.e. the ion flux producing the film included nitrogen ions as well as titanium ions). This is in contrast with the conventional arc process where the average charge per ion was deter-
mined assuming that the nitrogen ion contribution was negligible [12]. In the present experiment, if the nitrogen ion contribution is discounted, the calculated average charge per ion increases by almost 300%. The nitrogen flux is also assumed to come from ionized molecular nitrogen. If the ionized nitrogen flux is composed of 10% ionized atomic nitrogen (a reasonable estimate considering the emission spectrometry results presented in Section 4.2) the average charge per ion would be reduced to 1.79 e. The average charge per arriving atom is also a product of the sticking coefficient of the ions to the substrate. A sticking coefficient of 0.9+0.1 was used, the same as that used by Bergman, which is for low energy metal atoms impinging surfaces to which they have a high affinity [15]. An acceleration of the ions and the presence of impinging nitrogen ions in addition to titanium ions in the enhanced arc system may affect the sticking coefficient somewhat. The error in the average charge per ion primarily reflects uncertainties in the sticking coefficient and in the stoichi0metry. In the conventional arc at 20 mTorr Bergman [12] reported an average charge per ion of approximately 1.6 e. The 2.08 e reported here reflects the ionization enhancement processes occurring in the enhanced arc system. This value was also found at nearly twice the distance and half the cathode current used by Bergman. With similar parameters the average charge per ion calculated here would probably increase further.
4.5. Surface roughness Figure 5 shows the variation in surface roughness with magnetic field strength. The zero magnetic field strength value was obtained with the conventional arc system (i.e. no plasma guide). High speed steel substrates (polished to approximately 150 A finish) were placed 225 mm in front of the cathode. Films were deposited for 15 rain. The surface 2500
2OOO
A E
1500
1ooo
500
o
,
I
500
,
i
,
1000 MAGNETIC
FIELD
1
1500
(Gauss)
Fig. 5. Surface roughness vs. magnetic field strength.
108
P. Sathrum, B. F. C o l l / Plasma and deposition enhancement by modified arc
roughness was measured before and after film deposition and the difference (i.e. the change in surface roughness) plotted in Fig. 5. It can be seen in Fig. 5 that the average surface roughness decreased rapidly down to a minimum of 80 A at 1450 G magnetic field strength. Analysis by scanning electron microscopy (SEM) of the surface of the film deposited at 1450 G showed hardly any macroparticles; an effort had to be made to find them with the scanning electron microscope. Figure 6 shows a micrograph of the surface of this film and a micrograph of the surface of a conventional arc-deposited film at the same magnification for comparison. The film deposited at 890 G also showed very few macroparticles. The reduction and elimination of macroparticles in the enhanced arc system can be attributed to four effects. One effect is due to geometry and has been
(a)
reported by others [3, 16]. A simple geometric construction shows that macroparticles leaving the cathode in the enhanced arc system at less than 55° from the cathode surface and traveling in straight trajectories will collide with the plasma guide and not reach the substrate. A large percentage of the macroparticles are emitted from the cathode at angles of less than 30° according to Daalder [17]. A second effect is due to the modification of the arc discharge in the magnetic field of the enhanced arc system. The high velocity of the arc on the cathode surface results in a distinct reduction in the amount of melted area around the arc craters on the cathode [17, 18]. These melted areas are a source of macroparticles. Splitting of the arc into multiple spots also occurs in a magnetic field resulting in lower current densities and less melting of the cathode surface. SEM analysis of the surface of the cathode used in the enhanced arc system showed a distinct reduction in the amount of melted area and smaller crater sizes on average than on a cathode used in the conventional arc system. A third effect is a result of cathode "poisoning". Fewer and smaller macroparticles are produced from a cathode having a nitrided surface [12]. The production and confinement of large numbers of nitrogen ions in the vicinity of the cathode in the enhanced arc system contribute to the nitriding of the cathode surface. This is shown by an increase in the yellowing of the cathode after having been used in the enhanced arc system as compared with the conventional arc. A final effect is due to the evaporation of macroparticles in the plasma as a result of collision with electrons and ions [2, 16]. The dense energetic plasma conditions generated in the enhanced arc apparatus make this a likely explanation for the large reduction in macroparticles with increasing magnetic field strength found here.
5. Conclusions
(b) Fig. 6. Surface of (a) enhanced arc film and (b) conventional arc film.
The modified straight plasma optics system presented here enhances the ionization and energy of the flux impinging TiN films during deposition. Inelastic collisions with energetic electrons in the strong confined magnetic field of the apparatus result in an increase in the density of excited and ionized species. Dissociation and ionization of nitrogen and the resulting contribution of nitrogen ions to the plasma flux are especially apparent. The average charge per ion at the substrate is increased over the conventional arc case to approximately 2.1 e. The ionized fraction of the titanium vapor producing TiN films is also increased to 100%; the neutral vapor and macroparticle contribution to mass
P. Sathrum, B. F. Coil / Plasma and deposition enhancement by modified arc g a i n e d at the substrate was m e a s u r e d a n d f o u n d to be zero. Accordingly, such d e p o s i t e d films are n e a r l y free of macroparticles.
Acknowledgment T h e a u t h o r s would like to t h a n k R o b e r t A h a r o n o v for his assistance.
References 1 I.I. Aksenov, V. G. Padalka, V. T. Tolok and V. M. Khoroshikh, Soy. J. Plasma Phys., 6 (4) (1980) 505. 2 D. M. Sanders and E. A. Pyle, J. Vac. Sci. TechnoL A, 5 (1987) 2728--2731. 3 D. M. Sanders, J. Vac. Sci. TechnoL A, 7 (3) (1989). 4 I. I. Aksenov, V. A. Belous, V. G. Pakalka and V. M. Khoroshikh, Sov. J. Plasma Phys., 4 (4) (1978).
109
5 P. J. Martin, D. R. McKenzie, R. P. Netterfield, P. Swift, S. W. Filipczuk, K. H. Muller, C. G. Pacey and P. James, Thin Solid Films, 153 (1987) 91-102. 6 A. M. Dorodnov, Soy. Phys.-Tech. Phys., 23 (9) (1978). 7 I. I. Aksenov, Yu. P. Antuf'ev, V. G. Bren', V. G. Padalka, A. I. Popov and V. M. Khoroshikh, Soy. Phys.-Tech. Phys., 26 (2) (1981). 8 B. Chapman, Glow Discharge Processes, Wiley, New York, 1980. 9 R. M. Clements, J. Vac. Sci. Technol., 15 (2) (1978). 10 R. C. Weast, M. J. Astle, W. H. Beyer (eds.),CRC Handbook of Chemistry and Physics, CRC Press Inc., Boca Raton, FL, 1984. 11 G. A. Emmert, R. M. Weiland, A. T. Mense and J. N. Davidson, Phys. Fluids, 23 (1980) 803. 12 C. Bergrnan, Surf. Coat. Technol., 36 (1988) 243-255. 13 A. Ricard, H. Michel, P. Jacquot and M. Gantois, Thin Solid Films, 124 (1985) 67-73. 14 I. I. Demidenko, N. S. Lumind, V. D. Ducharenko, V. G. Padalka and G. N. Polvakova, Soy. Phys.-Tech. Phys., 29 (1984) 895. 15 G. Carter and D. G. Armour, Thin Solid Films, 80 (1981) 13. 16 K. Akari, H. Tamagaki, T. Kumakari and K. Tsuji, Surf. Coat. Technol., 43/44 (1990) 312-323. 17 J. E. Daalder, Physica C, 104 (1981) 91-106. 18 S. Ramalingham, U.S. Patent 4,673,477, June 1987.
103
Plasma and deposition enhancement by modified arc evaporation source* P. S a t h r u m
a n d B. F. C o i l
Multi-Arc Scientific Coatings, 200 Roundhill Drive, Rockaway, NJ 07866 (USA)
(Received March 10, 1991; accepted in final form May 21, 1991)
Abstract
The purpose of this paper is to present the "enhanced arc system" based on a modified straight plasma optics system. The magnetic field strength and orientation in the apparatus together result in an enhancement of plasma ionization, energy and density and near elimination of macroparticles. Ionization and energy characteristics of the flux impinging TiN films are described as functions of magnetic field strength. Langmuir probe measurements of voltage and current were made using a planar shielded substrate as the probe. Plasma species were observed using emission spectrometry. The ionized fraction of the titanium vapor at the substrate was determined by comparing the deposition rate while excluding ions with the deposition rate under normal conditions. Deposited TiN films were analyzed by profilometry and scanning electron microscopy. The results were an increase in the density of ionized and excited species, especially molecular and atomic nitrogen, with increasing magnetic field strength, which corresponds to an increase in the electron temperature; the average charge carried by nitrogen and titanium ions was approximately 2.1 e; 100% of the titanium vapor producing TiN films was ionized and the deposited films were nearly free of macroparticles.
1. Introduction
The cathodic arc process produces a highly ionized, energetic and dense vapor. This is advantageous for the deposition of well-adhered and dense films at a high rate. A disadvantage of the cathodic arc process is the production of macroparticles along with the ionized vapor. These macroparticles are typically incorporated into the films resulting in a degradation of film properties. Several modified arc evaporation techniques have been described which have shown that a reduction of these macroparticles along with even higher levels of vapor ionization, energy and density can be achieved by using magnetic fields [1--6]. These techniques have shown that the arc, the plasma and the deposited films can be strongly influenced by a magnetic field. Magnetic fields have also been used to enhance the synthesis in compound film formation and to increase the deposition rate, microhardness and adhesion of films [3, 7]. T h e purpose of this p a p e r is to present the " e n h a n c e d arc" system based on a modified straight plasma optics system. The magnetic field strength and orientation in the apparatus result in both a modification of the arc *Paper presented at the 18th International Conference on Metallurgical Coatings and Thin Films, San Diego, CA, 22-26 April 1991.
discharge and an increase in the density of the ionized and excited species. Titanium nitride films deposited with this system are nearly free of macroparticles. Ionization and energy characteristics of the plasma are described as functions of magnetic field strength. Langmuir probe m e a s u r e m e n t s of voltage and current were made using a planar shielded probe. Plasma species were observed using emission spectrometry. The ionized fraction of the titanium vapor and the average charge per ion were determined. Finally, the influence of the magnetic field on the surface properties of deposited films is described.
2. Enhanced arc apparatus
Figure 1 illustrates the enhanced arc system. A solenoid was mounted between the cathode and the vacuum chamber. Steel magnetic returns were m o u n t e d around the coil to contain the magnetic field. A soft magnetic core with an air gap was m o u n t e d inside the solenoid to optimize the shape and strength of the magnetic field. The arc was confined on the face of the cathode by the magnetic field produced by the solenoid and core. The solenoid, magnetic returns and core assembly will be referred to as the plasma guide.
Elsevier Sequoia
104
P. Sathrum, B. F. Coil / Plasma and deposition enhancement by modified arc
I
Ebert-type Jarrel-Ash scanning monochrometer. The monochrometer was sensitive in the range 230-990 nm with a resolution of less than 0.1 nm. The acceptance angle was 4 °. The deposition rates were determined using a microbalance with a repeatability of 20/xg. The average surface roughness was measured with a Dektak profilometer. All the experiments were done with 890 G magnetic field strength, 20 mTorr nitrogen pressure, 40 A cathode current, - 8 0 V substrate bias and a 375 mm cathodeto-substrate distance except where noted. For the conventional arc experiments the plasma guide was removed and the cathode-to-substrate distance was 300 mm. (The difference is the length of the plasma guide.)
~\\\\\\\\\\\\\\\\\'x5
Fig. 1. Enhanced arc apparatus: 1, solenoid; 2, soft magnetic core; 3, magnet returns; 4, cathode; 5, vacuum chamber; 6, substrate. To S p e c t r o m e t e r
i
To P y r o m e t e r
Fig. 2. Langmuir probe: 1, substrate; 2, ceramic fixture; 3, grounded shield; 4, cathode.
4. Results and discussion
4.1. Langrnuirprobe The substrate current was measured as a function of applied voltage and magnetic field strength. The results are shown in Fig. 3. A voltage--current curve for the conventional arc under similar conditions is also shown for reference. In Fig. 3, positive ions collected by a negative voltage are shown as a positive current and electrons collected by a positive voltage are shown as a negative current. It can be seen that the positive ion current saturates at negative voltage. The saturated ion current increases with increasing magnetic field strength at a rate that is nearly linear up to a magnetic field strength of
The length of the plasma guide was 75.0 mm and the inside diameter was 50.8 mm. Currents of up to 70 A through the solenoid produced a maximum field strength of 1500 G measured along the solenoid axis. No other magnets were used.
335 Gauss 2 b
1
3. Experimental details Figure 2 illustrates the Langmuir probe experimental system. The substrate was located 375 mm in front of the cathode. The cathode and substrate were perpendicular to their common symmetry axis. The substrate, which was 12.7 mm in diameter, was recessed into a ceramic fixture. A grounded shield 1.2 mm thick with an aperture of 11.87 mm was located 1.0 mm in front of the substrate. Plasma emission was monitored through an observation tube aimed to intersect the symmetry axis 375 mm in front of the cathode. Light from the emission was transmitted through a quartz window and through a quartz fiber optic cable to the input slit of a 0.5 m
8
670 Gauss
•
890 Gauss
•
1450 Gauss
N
Conventional Arc {0 G)
30 -300
-200 VOLTAGE
-100 (Y) -0.1 I
-0.2' o =
-0.3 -
-0.4 L
Fig. 3. Substrate current vs. vo]tage.
40
P. Sathtum, B. F. Coil / Plasma and deposition enhancement by modified arc
approximately 890 G. Above 890 G, the saturated ion current stays approximately the same. The saturated current collected at a negative bias is primarily due to the flux of titanium and nitrogen ions entering the sheath separating the plasma from the substrate [8]. The application of a magnetic field in a plasma optics system can have a significant effect on ion flux levels, as has been shown by Aksenov et al. [1]. The increase in ion flux is a result of plasma focusing and the creation of additional ions through electron impact ionization. The motion and density of electrons in the magnetic field both greatly increase the probability of electron-particle collisions. According to Aksenov and coworkers the probability of electron-molecule inelastic collisions is proportional to the square of the magnetic field strength [7]. If these collisions are energetic enough, excitation, dissociation of molecular species and ionization can occur. Electron energy was estimated from the positive current in the negative potential region before ion saturation using the method described by Clements [9]. Electron temperatures calculated by this method are listed in Table 1. The results show the electron temperature increasing nearly linearly with magnetic field strength to a maximum of 25 eV at 1450 G. Table 2 lists the ionization and dissociation threshold energies for the plasma species present in this experiment [10]. These results indicate that the electrons at higher magnetic field strengths are more than energetic enough to dissociate and ionize other species. The electron temperatures calculated here are for electrons in the vicinity of the substrate traveling perpendicular to the T A B L E 1. Electron temperature kTe, plasma density n and floating potential Vf vs. magnetic field strength Magnetic field
kTe (eV)
n ( × 10 to cm -3)
- Vt (V)
2.5±1.5 6.0±1.5 10.0±1.5 12.0±1.5 ~.0±2.0
0.80±0.15 1.01±0.26 1.48±0.32 1.66±0.32
1.0±0.4 5.8±0.4 8.9±0.4 5.5±0.4 2.5±0.4
(G) 0 335 670 8~ 1450
T A B L E 2. Ionization and dissociation threshold energies Species
N 2+ Ti ÷ N+ N
Ionization threshold
Dissociation energy
(eV)
(eV)
15.6 6.8 14.6 9.9
105
substrate. Electrons in the plasma guide closer to the cathode probably have much higher temperatures. This is indicated by the high intensity of the plasma emission coming from this area. Table 1 also lists the plasma density n and the floating potential Vf as functions of magnetic field strength. The plasma density was calculated from the electron temperature kTe, the saturated ion current and the average charge per ion calculated in Section 4.4. [11]. The floating potential was measured directly. The plasma potential can be roughly estimated from the saturation point of the negative current curves [8]. 4.2. Emission spectrometry
Emission spectrometry was used to detect and identify species present in the plasma over the wavelength range 230-900 nm. Plasma emission was monitored at the same distance as the Langmuir probe with the Langmuir probe removed. Measurements were made at different values of magnetic field strength and nitrogen pressure. The results show the presence of ionized and excited titanium and significant amounts of ionized and excited molecular and atomic nitrogen. This is in contrast with the conventional arc process where molecular nitrogen emission intensity is relatively weak and atomic nitrogen emission is not present at all [12]. Emission from the decaying species Ti, Ti +, Nz, N2 +, N and N + was observed. Ti 2+ and N 2+ emission lines were not identified. Figure 4(a) shows the variation in emission intensity with magnetic field strength for the strongest Ti, Ti +, Nz, N2 ÷, N and N ÷ lines. All lines are strongly influenced by magnetic field strength. In Fig. 4(a), it can be seen that there is an especially strong association between the intensity of emission of nitrogen species and magnetic field strength. The presence of Nz, N2 +, N and N ÷ emissions which increase nearly linearly with magnetic field strength indicates that electron impact excitation, dissociation and ionization of nitrogen is occurring. The excitation thresholds for these emissions are 11.1, 18.7, 12.0 and 20.7 eV respectively [13]. Electron temperature estimates made from the Langmuir probe experiments (Section 4.1) show that electron temperatures exceeding these excitation thresholds are reached. All nitrogen emissions including Nz +, Nz, N ÷ and N peak in the magnetic field strength range 900-1200 G. The rapid decline in N2 emission intensity and saturation of N2 ÷ emission intensity may be the result of nearly complete dissociation and ionization of Nz. Similarly, complete dissociation and ionization of nitrogen in a plasma optics system was reported by Aksenov et al. [7]. The Ti + and Ti emission lines as shown in Fig. 4(a) increase rapidly up to 335 G above which the emissions
106
P. Sathrum, B. F. Coil / Plasma and deposition enhancement by modified arc
•
N~ (391.4 n m )
---.--.--o---- N2 (357.7 rim)
00
500
(a)
1000
•
N + (568.0 rim)
•
N (746.8 n m )
.L
Ti + (334.9 nm)
.I
Ti (521.0 n m )
1500
strength of 890 G for the same emission lines as in Fig. 4(a). The characteristic features of this graph are similar to those discussed by Bergman for emission studies done with the conventional arc (zero magnetic field strength) [12]. The most notable difference is the high intensity of the nitrogen emission lines. The similarities indicate that the same reactions as reported by Bergman are taking place. Between 0 and 15 mTorr, the present results are also similar to those of Demidenko et al. [14] and Martin et al. [5] as well as Bergman. They attribute increases in the nitrogen and titanium emissions to charge exchange reactions of the type Ti"+ +NE
~ Ti ('-1)+ +N2 +
MAGNETIC FIELD (Gauss)
The rapid decrease in the intensity of the N2 + emission above 15 mTorr occurs because the initially high energy of Ti z+ species is lost in collisions with nitrogen and falls below the 14.0 eV threshold energy needed for this reaction to occur. The Ti ÷ to Ti reaction does not occur because it requires an excessively high ion threshold energy of 32.3 eV. In the region above 15 mTorr, the behavior of the Ti ÷ and Ti emissions can be attributed to the threebody reaction
4
c 3
i 2
Ti"+ + e +N2
) Ti('-l)+ + N2
1
0 0
(b)
10
20
30
PRESSURE
40
50
60
(mTorr)
Fig. 4. Variation in emission intensity with (a) magnetic field strength and (b) nitrogen pressure.
level off. From 0 to 335 G, the steep, nearly linear increase in the titanium emissions is probably due to plasma focusing and electron impact ionization. The abrupt saturation of the titanium emission lines at 335 G correlates to an electron temperature of approximately 6 eV, according to the results presented in Table 1, which is close to the ionization threshold of titanium. The abrupt saturation of the titanium emission lines may be the result of nearly complete ionization of titanium. The results presented in Section 4.3 show that the titanium vapor is 100% ionized. The presence of the excited titanium emission line which imitates the behavior of the ionized titanium emission line at a lower intensity is probably due to three-body electron recombination as described below. Figure 4(b) shows the variation in emission intensity with nitrogen pressure at a constant magnetic field
The probability of this reaction increases with increasing nitrogen concentration. This is shown in Fig. 4(b) by the decline of the Ti ÷ emission accompanied by a rise in the Ti emission and the decline of both species above 40 mTorr. These results and observations indicate that in the magnetic field of the enhanced arc apparatus, electron-particle collisions result in significant excitation, dissociation and ionization of titanium and nitrogen species. In addition to inelastic electron-particle collisions, charge exchange and three-body reactions typical of the conventional arc process are still occurring. Results and discussion in Sections 4.1 and 4.2 indicate that the contribution of nitrogen ions to the ion flux on a substrate is significant. This is in contrast with the conventional arc process where nitrogen ions contribute only a very small amount to the current on a substrate [12]. Finally, a direct comparison of the emission intensities of the Na ÷, N2, Ti + and Ti lines studied here showed that emissions in the conventional arc system were only 2.3%, 1.3%, 25% and 11% respectively of the emissions in the enhanced arc system under the same operating conditions. This supports the conclusion that significant enhancement of titanium and nitrogen excitation and ionization is occurring in the enhanced arc system.
107
P. Sathrum, B. F. Coil / Plasma and deposition enhancement by modified arc
4.3. Ionized vapor fraction The ionized titanium vapor fraction was found to be 100% using the ion exclusion method as used by Bergman [12]. The ionized fraction of the vapor impinging on the substrate was determined by comparing the deposition rate, while excluding ions, with the deposition rate when ion deposition was allowed. Ion deposition was done at a magnetic field strength of 890 G for 30 min. Ion exclusion deposition was done with the same parameters, except that a screen with a transparency of 0.903 (identical with the type used by Bergman) was placed 1 cm in front of the substrate and biased at - 100 V to collect ions. The Debye sheath surrounding the wires made the screen's effective ionstopping power much greater than the geometry suggests. The substrate was also biased at + 30 V to repel any ions that did penetrate the screen. The result was that there was no gain in mass when ions were excluded from deposition. No measurable amount of macroparticle flux was included in film deposition. Although macroparticles possess some amount of charge [6], their charge-to-mass ratio was assumed to be negligible so that macroparticle motion was not appreciably influenced by the electric fields used in this experiment. Ionization percentages of 68-83% in the conventional arc process, as reported by Bergman, reflect primarily the deposition of macroparticles. The absence of mass gain under ion exclusion deposition conditions supports the conclusion that 100% of the titanium vapor producing TiN films is ionized in the enhanced arc process.
4. 4. Average charge per ion The average charge per ion was found to be 2.08 + 0.28 e. The average charge per ion was determined from the deposition rate and the collected ion charge on a substrate. The deposition rate was measured in mass gained per second per unit area on a substrate clamped in a fixture exposing a deposition area of 3.23 cm 2. Owing to gas scattering, the area increased to approximately 3.53 cm2 at 20 mTorr. The collected ion charge was determined using a value of ion current density obtained from the Langnuir probe experiments presented in Section 4.1 for the same parameters as were used in deposition. The substrate temperature was less than 430 °C throughout the deposition. The atomic ratio of nitrogen to titanium in the film was determined using X-ray photoelectron spectrometry. The average charge per ion was determined assuming that the nitrogen mass contribution to the film was from ionized molecular nitrogen. (i.e. the ion flux producing the film included nitrogen ions as well as titanium ions). This is in contrast with the conventional arc process where the average charge per ion was deter-
mined assuming that the nitrogen ion contribution was negligible [12]. In the present experiment, if the nitrogen ion contribution is discounted, the calculated average charge per ion increases by almost 300%. The nitrogen flux is also assumed to come from ionized molecular nitrogen. If the ionized nitrogen flux is composed of 10% ionized atomic nitrogen (a reasonable estimate considering the emission spectrometry results presented in Section 4.2) the average charge per ion would be reduced to 1.79 e. The average charge per arriving atom is also a product of the sticking coefficient of the ions to the substrate. A sticking coefficient of 0.9+0.1 was used, the same as that used by Bergman, which is for low energy metal atoms impinging surfaces to which they have a high affinity [15]. An acceleration of the ions and the presence of impinging nitrogen ions in addition to titanium ions in the enhanced arc system may affect the sticking coefficient somewhat. The error in the average charge per ion primarily reflects uncertainties in the sticking coefficient and in the stoichi0metry. In the conventional arc at 20 mTorr Bergman [12] reported an average charge per ion of approximately 1.6 e. The 2.08 e reported here reflects the ionization enhancement processes occurring in the enhanced arc system. This value was also found at nearly twice the distance and half the cathode current used by Bergman. With similar parameters the average charge per ion calculated here would probably increase further.
4.5. Surface roughness Figure 5 shows the variation in surface roughness with magnetic field strength. The zero magnetic field strength value was obtained with the conventional arc system (i.e. no plasma guide). High speed steel substrates (polished to approximately 150 A finish) were placed 225 mm in front of the cathode. Films were deposited for 15 rain. The surface 2500
2OOO
A E
1500
1ooo
500
o
,
I
500
,
i
,
1000 MAGNETIC
FIELD
1
1500
(Gauss)
Fig. 5. Surface roughness vs. magnetic field strength.
108
P. Sathrum, B. F. C o l l / Plasma and deposition enhancement by modified arc
roughness was measured before and after film deposition and the difference (i.e. the change in surface roughness) plotted in Fig. 5. It can be seen in Fig. 5 that the average surface roughness decreased rapidly down to a minimum of 80 A at 1450 G magnetic field strength. Analysis by scanning electron microscopy (SEM) of the surface of the film deposited at 1450 G showed hardly any macroparticles; an effort had to be made to find them with the scanning electron microscope. Figure 6 shows a micrograph of the surface of this film and a micrograph of the surface of a conventional arc-deposited film at the same magnification for comparison. The film deposited at 890 G also showed very few macroparticles. The reduction and elimination of macroparticles in the enhanced arc system can be attributed to four effects. One effect is due to geometry and has been
(a)
reported by others [3, 16]. A simple geometric construction shows that macroparticles leaving the cathode in the enhanced arc system at less than 55° from the cathode surface and traveling in straight trajectories will collide with the plasma guide and not reach the substrate. A large percentage of the macroparticles are emitted from the cathode at angles of less than 30° according to Daalder [17]. A second effect is due to the modification of the arc discharge in the magnetic field of the enhanced arc system. The high velocity of the arc on the cathode surface results in a distinct reduction in the amount of melted area around the arc craters on the cathode [17, 18]. These melted areas are a source of macroparticles. Splitting of the arc into multiple spots also occurs in a magnetic field resulting in lower current densities and less melting of the cathode surface. SEM analysis of the surface of the cathode used in the enhanced arc system showed a distinct reduction in the amount of melted area and smaller crater sizes on average than on a cathode used in the conventional arc system. A third effect is a result of cathode "poisoning". Fewer and smaller macroparticles are produced from a cathode having a nitrided surface [12]. The production and confinement of large numbers of nitrogen ions in the vicinity of the cathode in the enhanced arc system contribute to the nitriding of the cathode surface. This is shown by an increase in the yellowing of the cathode after having been used in the enhanced arc system as compared with the conventional arc. A final effect is due to the evaporation of macroparticles in the plasma as a result of collision with electrons and ions [2, 16]. The dense energetic plasma conditions generated in the enhanced arc apparatus make this a likely explanation for the large reduction in macroparticles with increasing magnetic field strength found here.
5. Conclusions
(b) Fig. 6. Surface of (a) enhanced arc film and (b) conventional arc film.
The modified straight plasma optics system presented here enhances the ionization and energy of the flux impinging TiN films during deposition. Inelastic collisions with energetic electrons in the strong confined magnetic field of the apparatus result in an increase in the density of excited and ionized species. Dissociation and ionization of nitrogen and the resulting contribution of nitrogen ions to the plasma flux are especially apparent. The average charge per ion at the substrate is increased over the conventional arc case to approximately 2.1 e. The ionized fraction of the titanium vapor producing TiN films is also increased to 100%; the neutral vapor and macroparticle contribution to mass
P. Sathrum, B. F. Coil / Plasma and deposition enhancement by modified arc g a i n e d at the substrate was m e a s u r e d a n d f o u n d to be zero. Accordingly, such d e p o s i t e d films are n e a r l y free of macroparticles.
Acknowledgment T h e a u t h o r s would like to t h a n k R o b e r t A h a r o n o v for his assistance.
References 1 I.I. Aksenov, V. G. Padalka, V. T. Tolok and V. M. Khoroshikh, Soy. J. Plasma Phys., 6 (4) (1980) 505. 2 D. M. Sanders and E. A. Pyle, J. Vac. Sci. TechnoL A, 5 (1987) 2728--2731. 3 D. M. Sanders, J. Vac. Sci. TechnoL A, 7 (3) (1989). 4 I. I. Aksenov, V. A. Belous, V. G. Pakalka and V. M. Khoroshikh, Sov. J. Plasma Phys., 4 (4) (1978).
109
5 P. J. Martin, D. R. McKenzie, R. P. Netterfield, P. Swift, S. W. Filipczuk, K. H. Muller, C. G. Pacey and P. James, Thin Solid Films, 153 (1987) 91-102. 6 A. M. Dorodnov, Soy. Phys.-Tech. Phys., 23 (9) (1978). 7 I. I. Aksenov, Yu. P. Antuf'ev, V. G. Bren', V. G. Padalka, A. I. Popov and V. M. Khoroshikh, Soy. Phys.-Tech. Phys., 26 (2) (1981). 8 B. Chapman, Glow Discharge Processes, Wiley, New York, 1980. 9 R. M. Clements, J. Vac. Sci. Technol., 15 (2) (1978). 10 R. C. Weast, M. J. Astle, W. H. Beyer (eds.),CRC Handbook of Chemistry and Physics, CRC Press Inc., Boca Raton, FL, 1984. 11 G. A. Emmert, R. M. Weiland, A. T. Mense and J. N. Davidson, Phys. Fluids, 23 (1980) 803. 12 C. Bergrnan, Surf. Coat. Technol., 36 (1988) 243-255. 13 A. Ricard, H. Michel, P. Jacquot and M. Gantois, Thin Solid Films, 124 (1985) 67-73. 14 I. I. Demidenko, N. S. Lumind, V. D. Ducharenko, V. G. Padalka and G. N. Polvakova, Soy. Phys.-Tech. Phys., 29 (1984) 895. 15 G. Carter and D. G. Armour, Thin Solid Films, 80 (1981) 13. 16 K. Akari, H. Tamagaki, T. Kumakari and K. Tsuji, Surf. Coat. Technol., 43/44 (1990) 312-323. 17 J. E. Daalder, Physica C, 104 (1981) 91-106. 18 S. Ramalingham, U.S. Patent 4,673,477, June 1987.