Lambda Balanced Inductive Plasma Source Jvst

Characterization of a novel lambda balanced inductive plasma source G. K. Vinogradov,a) V. M. Menagarishvili, and S. Yoneyama MC Electronics Company Limited, Shirane-cho, Nakakomagun, 400-02 Yamanashi, Japan

~Received 24 September 1997; accepted 6 July 1998! A preliminary study on the 1–4.5 kW power industrial scale 27.12 MHz rf lambda-resonator oxygen asher is presented. Contact probes of several types, including single Langmuir and flat wall probes, thermocouples, and optical emission spectroscopy, are mainly used to diagnose plasma in the inductive source area, downstream chamber and in the vicinity of wafers. Electron density in a 200 mm wafer asher at 2 kW rf power varies from 231011 in the plasma source to 53107 cm23 in a downstream chamber 5–10 mm from a wafer. The ion density exceeds the electron density 10–60 times. The plasma space potential varies in a range of 14–22 V, while the floating potential of the bulk plasma and wall surface varies from 19 to 217 V. The minimum surface floating potential of 217 V exists at the maximum of the rf voltage standing wave distributed along the full lambda inductor. The wafer surface floating potential is in the range of 3–5 V depending on the reactor configuration and is constant within 61 V on the 200 mm wafer. Positive ion current density on the wafer and downstream chamber surface is less than 1 mA/cm22. The typical resist ashing nonuniformity is <2%–5% ~range, not sigma! for both 200 and 300 mm ashers at about a 6–8 m/min ashing rate. © 1998 American Vacuum Society. @S0734-2101~98!00106-7#

I. INTRODUCTION The use of inductive plasma sources has spread widely in the last decade as high density plasma etching/deposition tools.1,2 However, they have some drawbacks, including ~a! capacitive coupling of the inductor-plasma wafer, which produces a secondary capacitive plasma over the wafer and brings about wall sputtering and wafer damage; ~b! azimuthal discharge nonuniformity; ~c! too narrow a pressure and power process window; and ~d! difficult discharge ignition. A novel internally balanced rf inductive plasma source has been developed3 and realized commercially as a single wafer quasidownstream oxygen asher, l-Strip 3000.1 ~lStrip is a product of MC Electronics.! It operates in a full wave helical resonant or a lambda-resonator ~l-R! mode and represents a rf plasma source utilizing a standing wave structure in order to overcome known drawbacks and extend the applicability of inductive plasma sources for 300 mm wafer processing. The plasma phenomena in the 2 kW rf power l-R operating in a 0.001–100 Torr pressure range have been reported.4,5 The l-R plasma source, having three separate inductive zones, has substantially a three dimensional excitation structure keeping the discharge volume about constant over a wide range of pressure and power in comparison with known inductive sources. The positions of each inductive zone capable of generating of a separate plasma toroid is strictly defined by the standing wave pattern of the rf current along the coil of a one-lambda shorted spiral transmission line. The discharge structure of the novel plasma source cannot be obtained from conventional inductive sources: none of them has multiple inductive excitation zones with opposite inductive currents within the same inductor volume. Morea!

Author to whom correspondence should be addressed; electronic mail: [email protected]

3164

J. Vac. Sci. Technol. A 16„6…, Nov/Dec 1998

over, the capacitive power of the l-R plasma source is deposited mainly within the plasma source, while conventional sources deposit an essential portion of the capacitive power outside the source in a processing chamber. The l-R plasma source has not yet been well characterized in terms of plasma and process parameters. We present here some results on plasma diagnostics in the source area, in the vicinity of the wafer, and on the surface of a process chamber and wafer in high power industrial scale oxygen discharges. The ashing processes are also estimated for 200 and 300 mm single wafer ashes having 2.7 and 4.5 kW nominal rf power generators. II. EXPERIMENTAL DETAILS The l-R reactor was described previously.3 A basic configuration of the reactor is shown in Fig. 1 for scaling purposes. These reactors consist of a quartz discharge tube inserted into a water cooled copper coil representing a transmission line having total electric length of one full wave at resonance plasma loaded conditions. Roughly, the length of copper winding is about 11 m. The coil and the tube are enclosed by a cylinder water cooled copper shield. The top flanges and the chambers are made of aluminum. Flat wafer platens made of Al are mounted 120–140 mm below the bottom ground end of the coils. They can be heated up to 300 °C by resistive heaters in both chambers. The rf power at 27.12 MHz frequency in the range of 0.005–4.5 kW was supplied directly to the resonator by a 50 V coaxial cable. Three different power sources were used: 27.12 MHz, 2.7 and 4.5 kW rf power generators ~Kyosan, Japan!, and a 2.25 kW maximum rf power wide band tube amplifier IFI-410 with a sign wave signal generator HP8648A. The l-R plasma source operates without a matching network with an energy efficiency virtually up to 100%, neglecting the power losses in a coaxial cable. The resonator

0734-2101/98/16„6…/3164/6/$15.00

©1998 American Vacuum Society

3164

3165

Vinogradov, Menagarishvili, and Yoneyama: A novel lambda balanced inductive plasma source

FIG. 1. Lambda-resonator reactor configuration for 200 and 300 mm wafer processing: D5400 or 550 mm; d5235 or 420 mm; h568 or 100 mm; WD5280 or 380 mm for 200 or 300 mm wafer asher, respectively. The flat probes affixed to the chamber surface are marked by numbers.

itself has a quality factor Q of about 2360 as measured with a network analyzer HP-8712C. The O2 gas flow was maintained at 3 slm for a 200 mm machine and at about 5 slm for a 300 mm machine. The O2 pressure was 1.4 Torr unless stated otherwise. Both setups are pumped with dry pumping systems down to about 531023 Torr base pressure. Optical emission from the plasma was monitored with a Spectra Pro 275TM triple grating assembly spectrometer ~Acton Research Corp.! using an optical fiber bundle through orifices in the copper outer shield at different positions on the 200 mm setup. We used several types of electrostatic probes, described in more detail elsewhere.6 Mainly, 20 mm-diam probes of about 1–5 mm in length were used. Spherical Pt probes of about 100–130 mm diam were also used for diagnostics in a plasma toroid having a gas temperature well above 1200 K because the usual cylindrical probes sometimes melted. The 20-mm-diam Langmuir probes were inserted into fine quartz capillaries having about a 60–80 mm outside diameter. Several methods were used to protect the probe from rf fields and to examine the quality of the protection. For example, probes with different filter impedances and lengths of working wire were used in order to qualify the level of rf compensation obtained.7 The Langmuir probes were made to be moveable along and across the plasma source so that the probe I–V characteristics could be obtained at several locations. The grounded Al chamber surface was used as a reference electrode for the probe measurements in a downstream chamber. However, a special 131 cm2 Ni electrode was inserted into the center of the inductive plasma toroid to perform probe measurements in the high density plasma source6 which is essentially dc decoupled from the chamber. The plasma potential V pl was determined as the potential where the second derivative of the probe I–V characteristic crosses the potential axis. An experimental error of about 60.8 V in both the space and floating potentials was determined solely by the reproducibility of the discharge conditions. The electron and ion densities were derived from JVST A - Vacuum, Surfaces, and Films

3165

probe saturation currents at the plasma space and floating potentials, respectively.8 Flat 5-mm-diam surface probes were made of Ni and can be fixed at different locations on the process chamber surfaces, including the wafer holder, to find out the distributions of ion current densities and floating potentials over the whole chamber. The 0.5-mm thick flat probes were made as insulated Ni electrodes enclosed in Al grounded envelopes having an opening to contact plasma.6 Single flat probes and 5and 10-probe linear arrays were used. The gas temperature in the center plain of the plasma source was estimated using fine movable thermocouples inserted in quartz capillaries. The thermocouples were made of Pt-~Pt-Rh! 50-mm-diam wires. The method of two thermocouples9 having different diameters of outside capillaries ~0.6 and 1.2 mm! was used in order to estimate a correction factor for heterogeneous heat fluxes and radiative losses. The thermocouples do not interact with the inductor and do not perturb the discharge since the capacitance between the thermocouple wires and the plasma is small ~'0.2 pF/cm! and rf chokes are inserted into the signal lines. The surface electric potential of the surrounding capillaries, made of quartz, is electrically floating. Due to the large size of the plasma source and the small diameters of the thermocouples the longitudinal heat flux along the thermocouple wires or capillary can be neglected. However, the simplicity of the method is accompanied by an error that essentially increase with the gas temperature and heterogeneous heat fluxes. This error is estimated to be in the range of about 6100 K in our case.10 Resist ashing experiments were performed in both 200 and 300 mm ashers. Wafers from different manufacturers were examined. Positive photoresist MPR-4000 ~Mitsubishi Kasei! was mainly used. The 200 mm wafer resists were 1.25 mm thick. The 300 mm wafers with about 3-mm-thick resist layer from SELETE ~Semiconductor Electronics Leading Edge Technology, Japan! were used as is. The distribution of the film thickness on the wafers was measured with a Nanospec/AFT measuring system ~Nanometrics!. A wafer damage check was performed using electrostatic wafer monitors ~Kobe Steel, Japan! with metal-oxidesemiconductor ~MOS! capacitors having 90 Å-thick dielectrics. The antenna ratios were 350, 3500, 35 000, 350 000, and 700 000. III. RESULTS AND DISCUSSION The l-R discharge can be initiated from about 3 W rf power at about 0.01 Torr. Very stable 2 kW inductive discharges in Ar or O2 exist in the whole 0.001–20 Torr range. Higher pressure can cause quenching of the inductive mode due to a very high plasma impedance in an inductive current channel. However, the discharge can exist with a stable capacitive plasmoid mode up to at least 100 Torr. The capacitive plasmoid mode is a discharge with several longitudinal contracted capacitive plasma strings or banana shapes. An oxygen discharge under the standard conditions in the 200 mm machine appears as a pink plasma toroid located exactly

3166

Vinogradov, Menagarishvili, and Yoneyama: A novel lambda balanced inductive plasma source

at the center plane of the lambda coil at the zero rf voltage standing wave node. The visible diameter of the toroidal channel is about 3–4 cm. The toroid contacts the quartz wall. The rest of the discharge volume inside the coil radiates very weakly. There is no visible emission in the downstream chamber at all. The side inductive toroids typical for l-R discharge in Ar cannot be launched in oxygen in the 330mm-inner diam ~i.d.! tube even at about 3.5–4 kW rf power. A weak plasma emission in the side inductive zones can be observed at 4.5 kW in the smaller plasma source. There are two main reasons: ~1! oxygen discharges under 0.5–2 Torr pressure dissipate more than 50% of the specific plasma power into gas heating due to intrinsic mechanisms of energy dissipation;11 ~2! negative ion formation through electron capture drastically decreases the electric conductivity necessary for efficient inductive coupling.1 The specific power must be high enough to generate the necessary level of circular inductive current to overcome high electron and energy losses and support a high level of electrical conductivity for efficient rf power coupling in the inductive zones. It can be explained in the same way as the transition from the capacitive to the inductive mode in any conventional inductive source.1 The situation may occur where the central toroid efficiently absorbs an increasing power, not allowing power to reach the side toroids. The detailed mechanism of power distribution between the internal plasma structures and between the capacitive and inductive modes is presently unclear. All three plasma toroids easily appear in 2 Torr Ar discharge at only about 0.5 kW rf power. At 20 Torr Ar pressure 2 toroids exist at 1 kW power and the third one appears at about 1.2 kW. Figure 2 shows radial distributions of plasma space V pl and floating V f potentials in a plasma source at different

FIG. 2. Plasma space ~a! and floating potential ~b! distributions in a central plasma toroid of a 235 mm i.d. l-R plasma source: 1.4 Torr oxygen, 3 slm, 2 kW. The distance from the wafer platen ~in mm! is indicated. J. Vac. Sci. Technol. A, Vol. 16, No. 6, Nov/Dec 1998

3166

axial positions. The distance R 0 2R, where R 0 is an inside tube radius, is measured from the quartz wall surface in order to determine, more precisely, sharp sheath profiles. The highest plasma potential in the discharge of about 22–23 V is characteristic for the plasma toroid at about 5–15 mm from the wall. The axial area inside a toroid represents a potential well of about 5 V in depth. It is generated due to the pumping of fast electrons into this area from the toroid. This phenomenon is very similar to the formation of low electric potential on the surface of the discharge tube. There must be corresponding ambipolar diffusion of positive ions from the toroid into the potential well. They must recombine in this zone with negative ions delivering a lot of detached electrons. The electron energy distribution function should reveal specific features of this phenomenon.6 A marked change in both the plasma and floating potential is seen near the quartz wall. The plasma potential itself is not very informative unless the surface potential of the boundary wall is known. Here we have measured the floating potential up to the wall surface in several cases. Hence, the sheath voltage can be derived. The plasma floating potential is minimal at the 159 mm axial position corresponding to the position of maximum standing wave rf voltage distributed along the coil.3 The tube surface potential is most negative here. The maximum dc sheath voltage is about 30–33 V. It is obviously a low voltage sheath which cannot produce noticeable sputtering at the discharge tube surface. The mechanism of capacitive current balance is mainly responsible for a drastic decrease of sheath voltage in the l-R discharge in comparison with conventional inductive plasma sources. The plasma of the l-R discharge has enough high electric conductivity provided by the central inductive toroid to essentially shunt the area between the two opposite phase capacitive voltage halves of the coil. That is why the capacitive currents are not ‘‘looking’’ for a ground surface below the plasma source but substantially cancel one another inside the plasma source. In the case of l-R discharge the specific coil-plasma capacitance is essentially low since the coil is separated from the tube/plasma by a 30–50 mm distance thus representing a high impedance current generator with respect to the plasma load. The main portion of the rf capacitive voltage drops between the coil and the tube wall but not inside the discharge vessel. In addition, the l-R supplies antiphase currents to the plasma. Thus, the in-phase working capacitance of the coil is only 50% of the total coil-plasma capacitance. Furthermore, the l-R coil is at least 4–20 times longer than any conventional inductor, including a quarter wave helical resonator, if the same excitation frequency is used. Hence, the surface density of the capacitive current inductor plasma at the tube wall is also decreased 4–20 times for the same total power absorbed in a discharge. There is a very thin quartz wall between the coil and plasma. It does not increase the coil-plasma capacitance by introducing a high dielectric constant material like a thick flat window. Furthermore, the inductive portion of the l-R is

3167

Vinogradov, Menagarishvili, and Yoneyama: A novel lambda balanced inductive plasma source

3167

FIG. 3. Axial dependence of the floating potential in the downstream chamber as measured with a 10 MV oscilloscope divider and a data acquisition system.

at least 4–20 times more powerful in comparison with the conventional inductors because of the larger number of turns in the coil in comparison with inductively coupled plasma ~ICP! sources. Together these reasons explicitly show why there must be a low voltage sheath between the wall surface and plasma or why the capacitive coupling must not be a problem in the l-R discharge. One more factor that further decreases capacitive sheaths is the higher fundamental excitation frequency of 27.12 MHz which additionally decreases the dc sheath voltage in comparison with 13.56 MHz ICP sources. Long time ~about 103 discharge hours! observation of the discharge tube surface at 2.5 kW rf power, with about 105 ignition events, did not show any sputtering. The ignition here has a special mechanism. Since any rf discharge breakdown is capacitive in nature, there must be high initial electrical field strength inside the source. In conventional sources, the breakdown current flows from the highest rf voltage on the inductor through the dielectric wall ~window!, gas, onto the rf ground which is usually a wafer. In contrast, the l-R source realizes remote breakdown inside the plasma source between the opposite phase voltage standing waves, which generate the highest electric field strength. A rf calibrated weakly coupled capacitive probe similar to that described in Ref. 12 was used to measure rf fluctuations of the plasma potential. We found in the downstream afterglow plasma several harmonics up to the eighth at least. At resonance discharge conditions the 27, 54, 81, and 135 MHz harmonics have about a 15–20 V amplitude. The fourth, 108 MHz harmonic, is a little weaker than the third and fifth. The sixth to eighth harmonics are essentially weaker than about 5 V. So, the rf capacitive power is distributed into several high frequency harmonics. Figure 3 shows axial distributions of the floating potential V f in the downstream chamber near the wafer platen measured using a rf protected Langmuir probe and two data acquisition means: the Tektronix TEK-360P oscilloscope having a 107 V input divider and a data acquisition system6 of about 109 V input impedance. The V f at distances of ,50 mm from the wafer platen cannot be sensed with the 107 V divider because of the low level of electron density. The probe sheath has about 107 V dc impedance at a 50–60 mm JVST A - Vacuum, Surfaces, and Films

FIG. 4. Radial distributions of the discharge parameters: ~a! plasma space and floating potentials, ~b! electron and ion densities 10 mm above the wafer platen; and ~c! surface floating potential and ion current density on the wafer platen.

axial distance and drastically decreases toward the wafer. V f and the ion current at the wafer surface characterize several kinds of electric damage to wafers. The surface state is determined by a nearby plasma. Figure 4~a! shows the distribution of floating and plasma potentials 10 mm above the wafer surface. The V f is practically constant. The axial plasma potential is about 1361 V at distances of 10–50 mm from the platen when the narrow cylindrical surface of the bottom flange is insulated from the plasma and increases up to about 18–19 V with the dc conductive flange. Hence, the plasma is almost dc insulated from the grounded chamber. Even a small electrode immersed into the source plasma can control the dc plasma potential in the range of about 610 V.6 Figure 4~c! shows both the floating potential and ion current density distribution on the wafer platen. There are several independent measurements in 6X, 6Y directions. The surface floating potential V sf is about 360.8 V over the 120 mm radius in the case of the dc insulated bottom flange. So such a small variation of the distributed surface potentials cannot cause any electrical damage to the electronic structure. The same level of floating potentials was measured on the 300 mm platen at 4.5 kW rf power discharge. The bottom flange itself also has relatively low surface ion current density along the circumferential surface and some systematic azimuthal nonuniformity in the range of about 62 V which certainly cannot affect high rate thermally activated downstream chemical processes. The surface floating potentials and ion current densities were measured at several points on the chamber wall as indicated in Fig. 1 by the numbers 1–5. The ion current has measurable magnitude at only 3–5 locations. The corresponding ion current densities and floating potentials for probes 3, 4, and 5 are 0.04

3168

Vinogradov, Menagarishvili, and Yoneyama: A novel lambda balanced inductive plasma source

FIG. 5. Optical emission spectra taken across the central plasma toroid. Line identification is discussed in the text.

mA/cm2 and 9.5 V; 0.12 mA/cm2 and 6.7 V; 7 mA/cm2 and 21.1 V. The 1, 2 ion current densities are below about 0.01 mA/cm22 and could not be measured. The floating potentials were therefore undetectable. The temperature of the outside quartz surface is 74063 K at the toroid-wall contact area as measured by a thin thermocouple in close contact with the tube wall and by an infrared radiometer. There are direct indications of the thermocouples inserted into plasma showing the area of the maximum gas temperature and heterogeneous fluxes of recombining electrons and positive ions. The neutral particle density in this area is about three times lower than outside the toroid providing a more favorable low impedance condition for channeling the inductive current. The temperature drop across the 2-mm-thick quartz wall has been found to be about 4065 K thus giving a heat flux of '0.5 W/cm22 there. The thermocouple indications at the inside tube surface fit well with the results of the outside temperature measurements. Optical emission spectra from the central toroid of oxygen discharge, as shown in Fig. 5, indicate a highly dissociated plasma. The spectrum consists exclusively of atomic oxygen lines and the well known strongest OH band at 306.4 nm. Numerous atomic oxygen lines were identified. The emission bands of molecular oxygen ions were not detected. The strongest OII lines were found at 463.9 and 464.2 nm. We are not positive of some other detected OII lines because of the monochromator’s limitations. Two hydrogen lines were detected at 486.2 and 656.7 nm. The source of H and OH was the trace water content in the feed gas. The electron and ion densities and their radial distributions were measured in the toroidal area.6 The negative ion density exceeds the electron density 10–50 times depending on location. Therefore, we used a low limit ion temperature of about 0.1 eV in the Bohm formula for positive ion density.13,14 The electron density in the plasma toroid is 2 31011 cm23 15 mm from the quartz wall while the ion density is 3.431012 cm23. The plasma toroid produces a number of electrons diffusing over the plasma source. Additional rf heating of these electrons is provided by very high frequency capacitive harmonics thus creating an efficient three dimensional dissociation zone separate from the downstream process chamber. J. Vac. Sci. Technol. A, Vol. 16, No. 6, Nov/Dec 1998

3168

The plasma-wafer interaction in a l-R oxygen asher is negligible. Indeed, the positive ion flux onto the wafer is <731012 cm22 s21, while the flux of carbon atoms from the wafer surface corresponding to the normal resist ashing rate of 7 mm/min is '431017 cm22 s21. So, there is less than one positive ion striking the surface for about 63104 ashed carbon atoms. That is why at room temperature the ashing rate is only 0.004 mm/min in the same discharge corresponding to the decreased thermal activation. Chargeup damage tests were performed on the l-3000 asher. During the 5 min processing time ~2.5 kW; 0.5, 1.0, and 1.5 Torr; 1, 3, and 3 slm O2 flow! no variation of the leakage current ~typical value is 225 pA for 700 k ratio antennas! or voltage shifts of the MOS capacitors were detected. The damage parameters are comparable with those of high pressure ozone ashers. There is neither a separator nor a long bent tube between the dissociation zone and the processing surface typical for any downstream asher. It is favorable for the fastest delivery of dissociated species to wafers and decreases active particle losses. The process data were collected on the single wafer stripper l-3000. The resist stripping rate obeys an Arrhenius dependence with well known 0.5 eV activation energy typical for the thermally activated atomic oxygen ashing. The data correspond to the wafers pre-heated for about 3 min at 1.5 Torr oxygen flow at a process temperature. There are typical problems with kinetic measurements of nonstationary fast processes like dry ashing at high temperatures. Ashing results are strongly dependent on the wafer preheating conditions: atmospheric air, vacuum, or a working gas and pressure. In a usual fab process the wafer temperature in the l-3000 machine never reaches the susceptor temperature since the process time is much shorter ~typically about 10–20 s! than the characteristic heating time and neither electrostatic chuck nor mechanical clamps are used. Typical ashing rate ~AR! nonuniformity on 200 mm wafers is about 2%–3% with a 5 mm exclusion edge area @ (ARmax2ARmin!/~23ARave!#, where ARmax , ARmin, and ARave are the maximum, minimum and average ARS over 25 points measured 0, 45, and 95 mm from the wafer center. The ashing uniformity of the 300 mm asher at 4.5 kW rf power is even better than that of the 200 mm tool. There is 0.7% uniformity within a 200 mm area on the 300 mm wafer. The initial resist thickness on the 300 nm wafers was about 3 mm with about 0.3% nonuniformity. The ashing rate nonuniformity measured at about a half-thickness level was about 1.5% within a 10 mm exclusion area and about 5% within a 5 mm exclusion area.

IV. SUMMARY AND CONCLUSION Industrial scale lambda-resonator rf discharges in oxygen ashers were studied. The l-R plasma source produces one high density plasma toroid in the central area of 2 kW oxygen discharge in a 235 mm i.d. plasma source, while there are three completely developed toroids in argon discharges even at about 1 kW rf power. The number of generated tor-

3169

Vinogradov, Menagarishvili, and Yoneyama: A novel lambda balanced inductive plasma source

oids depends on the gas type, pressure, and rf power. The discharge has typically about 97%–99% energy efficiency operating without a matchbox. The highest plasma potential in the oxygen discharge under typical ashing conditions of the l-R asher is 20–22 V. The maximum dc potential difference between the plasma and the discharge tube surface is 30–33 V. There is no sputtering of the discharge tube in the capacitively balanced l-R plasma source. The parameters of the afterglow plasma in the downstream chamber are similar to those of the conventional downstream asher. The density of plasma electrons about 10 mm from the wafer surface is on the order of 5 3107 cm23. The positive ion flux on the wafer surface is <731012 cm22 s21. The wafer surface floating potential is only about 3–5 V over 200 and 300 mm wafers. The standard process examination with chargeup MOS wafer monitors having antenna ratios of up to 700 000 and oxide thicknesses of about 90 Å did not detect any damage during 5 min discharge processing. The ashing rate and uniformity of the 300 mm lambdaresonator oxygen asher are about the same as or even exceed the parameters of the 200 mm asher. Presented at the 44th National Symposium of the American Vacuum Society, San Jose, CA, 20–24 October 1997.

JVST A - Vacuum, Surfaces, and Films

1

3169

M. A. Liberman and A. J. Liechtenberg, Principles of Plasma Discharges and Materials Processing ~Wiley, New York, 1994!. 2 High Density Plasma Sources, edited by O. A. Popov ~Noyes, Park Ridge, NJ, 1995!. 3 G. K. Vinogradov and S. Yoneyama, Jpn. J. Appl. Phys., Part 2 35, L1130 ~1996!. 4 G. K. Vinogradov and S. Yoneyama, Proceedings of 13th Symposium on Plasma Processing, Tokyo, January 1996 ~unpublished!. 5 G. K. Vinogradov and S. Yoneyama, Proceedings of the 43rd National Symposium of the American Vacuum Society, Philadelphia, PA, 1996 ~unpublished!. 6 G. K. Vinogradov, V. M. Menagarishvili, and S. Yoneyama, J. Vac. Sci. Technol. A 16, 1444 ~1998!. 7 G. K. Vinogradov, G. J. Imanbaev, and D. I. Slovetsky, High Energy Chem. 19, 455 ~1985!. 8 L. Schott, in Plasma Diagnostics, edited by W. Lochte-Holtgreven ~AIP, New York, 1995!. 9 A. I. Maksimov, A. F. Sergienko, and D. I. Slovetsky, Fiz. Plasmy 4, 347 ~1978!. 10 Yu. A. Ivanov, Yu. A. Lebedev, and L. S. Polak, in Methods of Contact Diagnostics in Plasma Chemistry, edited by L. S. Polak ~Nauka, Moskow, 1981! ~in Russian!. 11 V. V. Rybkin, A. B. Bessarab, and A. I. Maximov, Teplofiz. Vys. Temp. 34, 181 ~1996! ~in Russian!. 12 R. R. J. Gagne and A. Cantin, J. Appl. Phys. 43, 2639 ~1972!. 13 R. L. F. Boyd and J. B. Thompson, Proc. R. Soc. London London A252, 102 ~1959!. 14 H. Amemiya, B. M. Annaratone, and J. E. Allen, Proceedings of the 3rd International Conference on Reactive Plasmas and 14th Symposium on Plasma Processing, Nara-ken, Japan, 1997, p. 239.