Plasma Diagnostics Techniques for Monitoring of Plasma Oxidation of Thin Metal Films

Transcription

1 Plasma Diagnostics Techniques for Monitoring of Plasma Oxidation of Thin Metal Films J. Pavlik, Z. Stryhal, J. Matousek, and S. Novak, J.E. Purkyne University, Faculty of Science, Usti nad Labem, Czech Republic; and P. Hedbavny, Vakuum Praha, Prague, Czech Republic ABSTRACT The development of gas sensors in the last few years caused a great interest in thin tin oxide films. The SnO 2 is one of the most widely used materials for gas sensors applications. In this work the tin oxide films were produced by the thermal evaporation of Sn films followed by the in-situ plasma oxidation. The vacuum system construction and the instrumentation for diagnostics are described. The optical emission spectroscopy, the quadrupole mass spectrometry and gas chromatography were used as diagnostic methods for determination of plasma parameters and for detection of impurities. All impurities even in small concentrations can play a significant role in plasma technologies. The quadrupole gas sampling system was used for detection of impurities and neutral process gas species analysis. The analysis of optical and mass spectra during plasma oxidation were used to characterize changes in the plasma parameters. The monitoring of plasma parameters provides an information on the stability and reproducibility of the plasma process. INTRODUCTION Thin SnO 2 films have wide range of applications (electronics, special coatings, etc.). These films are usually used if high conductivity and transparency are required. They are commonly applied in gas sensors. High quality tin oxide films can be prepared by chemical vapour deposition, spray pyrolysis, evaporation or sputtering [1,2]. We prepared the SnO 2 thin films by a thermal evaporation of Sn films followed by in situ plasma oxidation [3]. The aim of this work was to investigate a low temperature plasma used for plasma oxidation. The experiments were performed in a system for plasma-chemical surface modification of thin films. In our experiments we used a direct current (DC) discharge or capacitively coupled radio frequence (RF) discharge to generate the plasma in various oxygen/argon mixtures and we used following diagnostic techniques: the optical emission spectroscopy (OES), quadrupole mass spectrometry (QMS) and gas chromatography (GC) diagnostics. The combination of different diagnostic techniques enabled us to evaluate the influence of plasma parameters during plasma oxidation upon the quality of the tin oxide films. EXPERIMENT SET-UP The experiment arrangement is shown in Figures 1, 2 and 3. Figure 1: The experimental system for plasma-chemical surface modification of thin films. The system consists of two stainless-steel chambers (HV process chamber and UHV analytical chamber) which are separated by a gate valve. An absolute capacitance manometer ( mbar) indicates variations in the pressure inside the discharge tube. The system is pumped down by an oil-free pumping system with a turbomolecular pump (240 l.s -1 ). The UHV chamber comprises an evaporation system. Thin tin films were evaporated from Mo boat. The evaporated material flow (tin is of % purity) laid out on a sample is strongly limited by inner housing and rotary shutter with different exchangeable masks. The SiO 2, Pyrex or aluminium substrates with the dimension of mm 3 were well cleaned and mounted in a sample holder and placed into a vacuum system. The tin layers were prepared by the thermal evaporation. The deposition rate and thickness were monitored during deposition by a conventional quartz crystal system (5 MHz) with a water-cooled head. Deposition rate of the metal films estimated from the film thickness was 0.5 ~ 1 nm/s. Then the tin layer was oxidized in plasma. After the plasma oxidation the sample was stored in the carrousel in the UHV chamber (Figure 2). Two types of discharges were applied: DC glow discharge and MHz RF discharge in pure oxygen or oxygen/argon mixture (purity: argon %, oxygen %) Society of Vacuum Coaters 505/ th Annual Technical Conference Proceedings (2006) ISSN

2 Figure 2: Experimental system for plasma oxidation with DC discharge. 1 HV chamber, 2 UHV chamber, 3 gate valve, 6 stepper motor, 7 rotary motion feedthrough, 8 titanium sputter ion pump, 9 linear and rotary movable transport system, 10 carrousel, 13 linear motion feedthrough, 14 power electrical feedthroughs, 15 multi pin electrical feedthrough, 17 Pirani and Penning vacuum gauges, 18 entrance for loading and unloading of samples, 19 inlet for working gases, 20 high voltage electrical feedthrough, 21 - DC glow discharge tube, 22 evaporation system, 24 rotary shutter, 25 - sample holder with sample, 26 crystal sensor head, 27 supporting starting electrode, 28 - mirror Figure 3: Experimental system for plasma oxidation with RF discharge. Plasma oxidation in the DC discharge The DC flowing discharge was produced in a Pyrex tube (inner diameter 70 mm, length 450 mm) at typical total gas pressure p = Pa, flow rate Q = l s -1 and discharge current I = 25 ma, see Figure 2. Plasma oxidation in the RF discharge The RF discharge is produced in an axial movable quartz tube (inner diameter 18 mm). The generated plasma is drifted along the quartz tube inside the HV chamber, where the sample surface is modified in the flowing post-discharge, see Figure 3. The external ring electrodes were capacitively coupled via a matching unit to an RF generator (50 W, MHz). Flow rates were adjusted so that the total mixture pressure of Pa was attained. This pressure value was found to be the optimum one for plasma ignition (oxygen/argon mixture: 5/3 and RF power: 35 W). The substrate was held on the plasma floating potential during the plasma oxidation (oxidation time min). The experimental system incorporated heater (quartz lamp 20 W mounted in polished stainless steel reflector) which allows an indirect heating of the substrate holder by the irradiation in the temperature range C, see Figure 4. Figure 4: The schematic view of the experimental set-up for plasma oxidation of thin metal films in the RF discharge DIAGNOSTIC SET-UP We mainly used OES and QMS as diagnostic techniques. The plasma gas phase reaction by-products were determined by QMS and GC methods. Optical emission spectroscopy Optical emission spectroscopy is very often used for diagnostics of the reactive plasma. The main advantage of the method is the non-invasive character of the measurement. OES is very sensitive to changes in the high-energy fraction of the electron distribution function, which is responsible for a production of radicals and ions in reactive plasmas. Measurements of the emission spectra in the visible and near infrared regions ( nm) have been carried out. The spectroscopy system was 720

3 based on the imaging spectrograph CP-200 (Jobin-Yvon). As a detector we used a CCD imaging camera ST-7I (Santa Barbara Instrument Group) with the resolution of 0.6 nm. Quadrupole mass spectrometry analysis of neutral species The sampling gas system was used for the detection of neutral species effusing from the discharge tube during plasma processing. A small amount of the process gas was extracted from the active plasma zone and transported through the pressure reduction system to the QMS sampling cell (see the schematic diagram in Figure 5). The plasma oxidation process occurs in a pressure range from 0.2 mbar to 1.5 mbar. Therefore a pressure reduction system was needed. In our case the gas pressure was reduced to 10-5 mbar. with an electron capture detector (ECD). The temperature of the detector was 300º C. The separation was performed using column HP-5 Crosslinked 5% PH ME Siloxane (30 m x 0.53 mm x 2.65 µm). The temperature program began at 60º C, after 5 min increased (6º C/min) up to 240º C and was held on this value until termination (40 min). High purity nitrogen (99.999%), filtered through the oxygen trap was used as the carrier gas with a column head pressure of 276 mbar. The direct coupling and sample injection to the gas chromatograph was not possible in our experimental set-up. Therefore a portable sampling system was used. The system was connected to the experimental apparatus, evacuated during the bake out and filled with the sample of the process gas. After addition of the high purity nitrogen ( %) to the process gas sample the pressure in the system increased above the value of the column head pressure of the chromatograph carrier gas. The sampling system was separated and transported to another laboratory with GC analyser. EXPERIMENT RESULTS Basic plasma parameters during plasma oxidation in the DC discharge were estimated with a single Langmuir probe and OES. The electron temperature was about ev. Plasma density was of the order of m -3. Figure 5: The schematic diagram of the sampling gas system The sampling gas system consists of a stainless steel tube (length 110 cm, I.D. = 4 mm), sapphire dosing valve and a closed sampling cell around the ion source of the QMS. The sampled gas is drawn through the tube to the manifold with a dosing valve and pumped through a diaphragm with an orifice 0.9 mm by a two-stage rotary pump with a trap. The pressure at the exit of the tube is about 0.1 mbar. Response time and resolution are influenced by the conductance of the gas tube (C = 9 x 10-3 l. s -1 at viscous flow) between the extraction point of discharge tube and the sample gas dosing valve and by the pumping speed of the rotary pump (S = 1.5 m 3.h -1 ). The QMS measures the composition of sampled gas with a relatively short response time due to a relatively high conductivity of the sampling system (about 10-2 l.s -1 ). Gas chromatography analysis of neutral species In addition to the quadrupole mass analysis of the process gas phase composition the method of gas chromatography was used. Analyses were performed on a Hewlett-Packard gas chromatograph (Model HP 5890 Series II) equipped Optical emission spectroscopy measurements It is well known that the plasma parameters are influenced by factors such as the argon gas pressure and the flow of oxygen in the case of Ar/O 2 mixtures. The aim of the optical emission spectroscopy measurements was to find the optimum oxidation conditions. In our case the argon pressure was 50 Pa, the oxygen flow was 0.1 %, 1 %, 5 % or 10 % of the argon flow. The oxide produced during plasma oxidation is probably caused by the existence of O - ions. The elemental oxygen ions can be created in the bulk of plasma or on the substrate surface. Several emission lines corresponding to atomic transitions in argon and oxygen were used to analyse the plasma emission spectra. The most significant oxygen lines in our OES measurement were the nm and nm lines. These lines correspond to the deexcitation of the oxygen atom in the state 5 P (O * ), whose creation is predicted by the following way, e + O 2 e + O + O and e + O e + O, i.e. dissociative excitation or direct impact excitation of oxygen atom, respectively [4]. The ratio of O atoms with respect to O 2 molecules is enhanced in oxygen/argon mixture due to the production of O atoms by the quenching reaction Ar M metastables with O 2, i.e. Ar M + O 2 Ar + O + O. The direct electron-impact excitation from the ground state is responsible for the ArI nm emission line. The dependence of the intensity ratio of lines OI nm to ArI nm (which is affected by the changes of discharge conditions) on the oxygen flow gives an information on the atomic oxygen density in the oxygen/argon mixture plasma. The dependence of the relative emission intensities of I O(844.67) /I Ar(750.4) on oxygen flow shows that the intensities ratio reaches its maximum at 1% oxygen flow in the case of 721

4 DC discharge and over 10% oxygen flow in the case of the RF discharge. In our case it is seen that argon metastables 3 P 2 (11.55 ev) and 3 P 0 (11.72 ev) play a very important role in the overall excitation mechanism, i.e. argon metastables cause an increase of the number of excited oxygen atoms. Figure 7: Mass spectrum of neutral species sampled from process chamber when the discharge was OFF (mixture oxygen/argon 5:3). Figure 6: Emission spectrum of DC discharge in the Ar/O 2 mixture in flowing regime Mass spectrometry measurements We carried out the residual gas analysis (RGA) in the sampling cell before each set of QMS measurements of the process gas. The influence of the residual gas composition on the process gas composition can be neglected. In Figures 7 and 8 the results of process gas analysis in two cases (discharge ON and discharge OFF) are shown. The peaks representing the process gas dominate in the mass spectra. The smaller peaks correspond to the impurities which can originate from the gas desorption from the inlet system surfaces and chamber surfaces. The QMS analysis of neutral species sampled from process chamber during plasma oxidation gives an information on the neutral species resulting from the interactions between plasma and chamber walls or sample surfaces. Our previous results [5] indicate that the presence of the impurities in the DC discharge is caused by teflon components inside the discharge tube (see Figure 9). The occurence of fluorocarbon radicals in the plasma may be explained by a decomposition of the teflon components. The presence of these impurities in the discharge plasma was significantly reduced after a modification of the discharge tube when teflon was eliminated. Spectrum measured for the modified system is shown in Figure 8. Figure 8; Mass spectrum of neutral species sampled from oxygen discharge during plasma oxidation of thin tin films. (oxygen, p = 0.2 mbar, the discharge was ON). 722

5 Figure 9: Mass spectrum of neutral species sampled from oxygen DC discharge before discharge tube modification showing impurities originating from teflon (oxygen, p = 0.2 mbar, the DC discharge was ON). (p = 0.2 mbar) sampled from the oxygen plasma. The peaks with retention time 1.39 min and 28.6 min are dominant. These peaks can represent the main component of process gas, i.e. oxygen. We performed a calibration of the gas chromatograph for reference gas samples of know composition like O 2, CO 2 and H 2 O vapours. In our experimental conditions we were able to identify the characteristic retention times for various gases and impurities. So some characteristic retention times were determined for oxygen (1.27 min, 1.39 min, 3.11 min, min, 13.7 min and 28.6 min), for carbon dioxide fractions (1.55 min, 2.43 min, min, 16.1 min, 18.4 min, 25.1 min and 27.5 min), for water fractions (1.45 min, 1.7 min, 1.86 min, 1.97 min, 4.79 min and 19.2 min), and for carbon monoxide and carbon dioxide (1.5 min, 2.45 min, min and 25.1 min). The qualitative results of gas chromatography measurements correspond with results of our previous mass spectrometry measurements. The comparison of the QMS spectra when the discharge was OFF and ON is shown in Figure 10. The experiments were performed in DC oxygen discharge in both flowing and stationary regime. Figure 11: Chromatogram of neutral species of process gas sampled from oxygen plasma, p = 0.2 mbar (the vessel was filled with nitrogen, final pressure p = 10 3 mbar). Figure 10: The comparison of relative concentrations of the neutral species effused from the discharge tube. Gas chromatography measurements The process gas was analysed qualitatively by gas chromatography, too. The possible admixture of nitrogen observed in QMS spectra can not occur in chromatograms since the nitrogen was used as the carrier gas. As we used the ECD detector, it was not possible to detect argon. The sensitivity for oxygen and other electronegative gas species is very high. Figure 11 shows the chromatogram of neutral species of process gas CONCLUSION The experimental system for plasma oxidation of metal thin films has been presented. The described diagnostic techniques enable the achievement of the main goal the optimization of the plasma oxidation process. Deducting from our OES measurements we suppose that the optimum ratio of the Ar/O 2 mixture can be achieved at 1% oxygen flow in the case of DC discharge and at more than 10% oxygen flow in the case of the RF discharge. The QMS and GC analysis of the process gas composition during plasma oxidation provides a basic information about changes in the plasma gas composition inside the discharge tube and consequently about the stability and reproducibility of the plasma parameters during the plasma oxidation. Both methods have sufficient selectivity and low detection limit for identification of all impurity species of the investigated process gas even at low concentrations. 723

6 ACKNOWLEDGEMENTS This work was supported by the Ministry of Education of Czech Republic (Projects COST OC 143, and LC 06041). REFERENCES 1. S. W. Lee, P. P. Tsai, H. Chen, H 2 sensing behavior of MOCVD-derived SnO 2 thin films, Sensors and Actuators B, 41, 55, P. Serrini, V. Briois, M. C. Horrillo, A.Traverse, L. Manes, Chemical composition and crystalline structure of SnO 2 thin films used as gas sensors, Thin Solid Films, 304, 113, Z. Stryhal, J. Pavlik, S. Novak, A. Mackova, V. Perina, K. Veltruska, Investigations of SnO 2 thin films prepared by plasma oxidation, Vacuum, 67 (3-4), V. Hrachova, A. Kanka, Study of admixure influence on the oxygen spectra properties, Vacuum, 48 (7-9), 689, J. Pavlik, M. Maly, Z. Stryhal, S. Novak, Study of oxygen plasma using mass spectrometry during plasma oxidation of aluminium thin films, Symposium Proc. of 13th SAPP Symposium on Application of Plasma Processes, p.124, Comenius University Press, Bratislava,