Langmuir 2000, 16, 749-753
749
Synthesis of Teflon-like Thin Layers under Perfluoro-1-butanesulfonyl Fluoride/Methane Radio Frequency Plasma Environments C. Naum, S. Manolache, and F. Denes* Center for Plasma-Aided Manufacturing, University of WisconsinsMadison, Madison, Wisconsin 53706 Received July 5, 1999. In Final Form: September 13, 1999 Teflon-like thin layers were deposited on various substrate surfaces from perfluoro-1-butanesulfonyl fluoride/methane radio frequency pulsed plasma environments. The deposited films were characterized using attenuated total reflectance Fourier transform infrared spectroscopy, electron spectroscopy for chemical analysis, differential thermal analysis/thermogravimetric analysis (DTA/TG), and atomic force microscopy techniques. It has been demonstrated that the presence of methane is crucial for the deposition of fluorinated structures due to the simultaneously developed and fluorine atoms controlled competitive etching and deposition processes. It has been shown that plasma films with the highest fluorine concentrations can be generated under low duty cycle and high methane partial pressure conditions. DTA/TG data indicate that the Teflon-like layers have a fairly high thermal stability.
Introduction Poly(tetrafluoroethylene) (Teflon; PTFE) has properties, including chemical inertness, thermal stability, low coefficient of friction, low dielectric constant (1.9-2.1), low surface energy, and relatively low permeability for most of the gases,1 which can hardly be matched by other polymers. Conventional Teflon-production technologies do not allow the deposition of thin layer Teflon materials due to their insolubility and infusible nature. As a successful alternative, plasma-enhanced chemical vapor deposition (PECVD) has extensively been used for depositing fluorocarbon polymers on various substrates.2-10 The starting components include both monomer- and nonmonomer-type fluorocarbon derivatives. It has been shown that these approaches performed mainly by using continuous-wave (CW) PECV techniques resulted in macromolecular layers11-18 with a stoichiometry where the (1) Savage, C. R.; Timmons, R. B.; Lin, J. W. Adv. Chem. Ser. 1993, 236. (2) Anand, M.; Cohen, R. E.; Badour, R. F. Polymer 1981, 22, 361. (3) Astell-Burt, P. J.; Cairns, J. A.; Cheetham, A. K.; Hazel, R. M. J. Plasma Chem. Plasma Proc. 1986, 6 (4), 417. (4) Dedinas, J.; Feldman, M. M.; Mason, M. G.; Gerenser, L. J. First Int. Conf. Plasma Res. Technol., 1st 1982, 119. (5) Momose, Y.; Takada, T.; Okazaki, S. Proc., ACS Div. Polym. Mater. Sci. Eng. 1987, 56, 236. (6) Arefi, F.; Montazer-Rahmati, P.; Andre, V.; Amoroux, J. J. Appl. Polym. Sci: Appl. Polym. Symp. 1990, 46, 33. (7) Denes, F.; Sarmadi, A. M.; Hop, C. E. C. A.; Buncick, M.; Young, R. A. J. Appl. Polym. Sci. 1994, 52, 1419. (8) Liston, E. M.; Martinu, L.; Wertheimer, R. J. Adhes. Sci. Technol. 1993, 10, 1091. (9) Plumb, I. C.; Ryan, K. R. Plasma Chem. Plasma Process. 1986, 6 (3), 205. (10) Yagi, T.; Pavlath, A. E. J. Appl. Polym, Sci.: Appl. Polym. Symp. 1984, 38, 215. (11) Winters, H. F.; Coburn, J. W.; Kay, E. J. Appl. Phys. 1977, 48, 4973. (12) Kay, E.; Dilks, A.; Seybold, D. J. Appl. Phys. 1980, 51, 5678. (13) Kay, E.; Coburn, J. W.; Dilks, A. Topics in Current Chemistry; Plasma Chemistry, Vol. III; Springer: Berlin, 1980; Vol. 94, p 1. (14) Kay, E.; Dilks, A. Thin Solid Films 1981, 78, 309. (15) Klausner, M.; Baddour, R. F.; Cohen, R. E. Polym. Eng. Sci. 1987, 27 (11), 861. (16) Montazer Rahmati, P.; Arefi, F.; Amoroux, J. Surf. Coatings Technol. 1991, 45, 369. (17) Strobel, M.; Corn, S.; Lyons, C. S.; Korba, G. A. J. Polym. Sci.: Polym. Chem. Ed. 1985, 23, 1125.
F/C ratios were usually significantly lower than 2.19,20 This clearly reflects the presence of unsaturation, trapped free radicals, and the existence of CF functionalities in the structures of plasma-generated macromolecular layers. These structural particularities are suspected to be responsible for dielectric loss and aging effects. On the other hand, the presence at the same time, of a totally linear CF2-based macromolecular structure (e.g., conventional polytetrafluorethylene) would exhibit moderate surface adhesion characteristics21-23 (e.g., poor adhesion of metallic layers deposited on Teflon substrate surfaces). Recently it has been demonstrated1,24 that pulsed PECVD can be a solution to these problems. By starting from hexafluoropropylene oxide (HFPO) for instance, it has been shown that under controlled duty-on/duty-off cycle conditions high fluorine content fluorocarbon macromolecular layers can be synthesized with predominant CF2 units. In this contribution the formation of Teflon-like (TL) structures, by starting from perfluoro-1-butanesulfonyl fluoride (C4F10O2S; PBS) under CW- and pulsed radio frequency (RF) cold plasma conditions is investigated. Experimental Section Materials and Methods. Evaluations of the relative surface atomic compositions of plasma-deposited films were carried out using a Perkin-Elmer Physical Electronics 0 5400 small area electron spectroscopy for chemical analysis (ESCA) system: Mg source; 15 kV; 300 W; pass energy, 89.45 eV, angle: 45°. Carbon (18) d’Agostino, R.; de Benedictis, D.; Cramarossa, F. Plasma Chem. Plasma Process. 1984, 4(1), 1. (19) Denes, F.; Hua, Z. Q.; Simonsick, W. J.; Aaserud, D. J. J. Appl. Polym. Sci. 1999, 71, 1627. (20) Lee, S.-D.; Manolache, S.; Sarmadi, M.; Denes, F. Deposition of high fluorine content macromolecular thin layers under continuousflow-system corona discharge conditions, Polym. Bull., in press. (21) Millard, M. M.; Windle, J. J.; Pavlath, A. E. J. Appl. Polym. Sci. 1973, 17, 2501. (22) Yagi, T.; Pavlath, A. E.; Pittman, A. G. J. Appl. Polym. Sci. 1982, 27. (23) Yagi, T.; Pavlath, A. E.; Pittman, A. G. Proc. Int. Symp. Plasma Chem. 1981. (24) Limb, S. J.; Gleason, K. K.; Edell, D. J.; Gleason, E. F. J. Vac. Sci. Technol. 1997, 15, 4.
10.1021/la990872i CCC: $19.00 © 2000 American Chemical Society Published on Web 11/02/1999
750
Langmuir, Vol. 16, No. 2, 2000
Figure 1. Scheme of parallel plate cold-plasma reactor: 1, argon reservoir; 2, CF4 reservoir; 3, RF generator; 4, flow controller; 5, electric insulator disk; 6, gas mixing chamber; 7, drum-type stainless steel upper electrode; 8, cylidrically shaped upper part of reactor; 9, temperature controller for the built-in electric heater of the lower electrode; 10, butterfly-type valve; 11, stainless steel liquid nitrogen trap; 12, mechanical vacuum pump. (C1s), oxygen (O1s), sulfur (S2p), and fluorine (F1s) atomic compositions were evaluated, and the binding energy values of the nonequivalent C1s carbon linkages were analyzed. To correct surface-charge-origin binding energy shifts, calibrations were performed based on the well-known C1s peak. Attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) was used to identify ac-related chemical linkages of the plasma-synthesized TL layers. An ATI-Mattson Research Series IR instrument was used which was provided with a GRASEBY-Special Benchmark Series ATR in-compartment P/N/ 11160 unit. All FTIR evaluations were performed under a nitrogen blanket generated from a flow-controlled liquid nitrogen tank. Data were collected in the 600-4000 cm-1 wavenumber region with 250 scans for each sample, with a resolution of 0.4 cm-1. The surface morphologies of plasma-deposited macromolecular layers were evidenced by atomic force microscopy (Digital Instrument Nanoscope III AFM; experimental conditions: scan rate, 2.654 Hz; sampling number, 512). The thermal behavior of TL films were monitored using differential thermal analysis (DTA-TG; Seico Exstar 6200; temperature range, 0-650 °C, heating rate, 5 °C/min; nitrogen flow rate, 100.0 mL/min). Water-based static contact angle measurements were performed using a NRC contact angle goniometer (100-00 Ramihart Inc.). Plasma Reactor/Plasma Processing. The deposition of TL was performed in a parallel plate, capacitively coupled (20 cm diameter stainless steel disk-shaped electrodes; gap between the electrodes, 3 cm; grounded lower electrode), cylindrical stainless steel reactor (Figure 1) described earlier.25 The reactor is equipped with a 13.56 MHz power supply provided with pulsing capabilities. All experiments were preceded by a cleaning procedure by heating the lower electrode to 200 °C and igniting oxygen plasma (200 W; 250 mT; 5 sccm) for 10 min in order to remove possible contaminants from earlier plasma reactions. In a typical experiment silicon wafer, stainless steel, and nickel substrates (with dimensions matching the requirements of analytical instruments) were placed on the substrate holder (lower electrode), then the reaction chamber was closed and evacuated to base-pressure level. In the next step by operating the needle valve system and the large capacity valve, the working pressure of the preselected PBS/CH4 mixture was established in (25) Denes, F.; Hua, Z. Q.; Young, R. A.; Shohet, J. L. Plasmas Polym. 1997, 2 (1), 1.
Naum et al.
Figure 2. Typical survey ESCA diagrams of TL synthesized under low (S12) and high (S10) methane flow rate conditions.
Figure 3. Typical high-resolution ESCA diagrams of TL synthesized under low (S12) and high (S10) methane flow rate conditions. the reactor. The plasma state was then initiated by dissipating RF power to the electrodes and sustained under the predetermined parameters conditions. At the end of the reaction, the chamber was evacuated to base pressure followed by repressurizing the system to atmospheric conditions. The TL-coated substrates were then removed and stored under open laboratory conditions until the analytical experiments are started. During the plasma-deposition processes the following experimental conditions were employed: starting component, PBS/CH4 mixtures; base pressure in the reactor, 50 mTorr; vapor pressure in the absence of plasma, 200 mTorr; pressure in the presence of plasma, 210-220 mTorr; partial pressures PBS/CH4, Table 1; temperature of the reactor, 25 °C; RF power dissipated to the electrodes, 100 W; duty cycles period, 500 µs; plasma on, 2080%; reaction time, 0.5-10 min.
PTFE-like Thin Layers
Langmuir, Vol. 16, No. 2, 2000 751 Table 1. Experimental Parameters and Results
no.
sample label
CH4 partial pressure (mTorr)
duty cycle (%)
1 2 3 4 5 6 7 8 9
S5 S6 S7 S8 S9 S10 S11 S12 S13
10 30 30 10 20 20 20 5 35
29 71 29 71 50 20 80 50 50
ESCA relative atomic composition (%) C1s O1s F1s S2p 35.1 51.8 37.4 32.1 28.2 35.1 28.4 27.7 42.1
4.2 31.8 4.5 25.6 37.1 3.7 41.8 46.1 6.2
59.0 8.9 54.5 33.3 25.9 59.3 18.1 15.7 43.9
1.7 7.5 3.7 8.9 8.9 1.9 11.7 10.5 7.8
relative CFx bonds concentration (ESCA C1s peak %) CF3 CF2 3.6
4.7
1.8 0.3 0.1 6.5
4.7 0.4 0.2 4.2 0.1
1.1
1.9
contact angle (deg) 97 74 95 63 62 102 60 60 98
Results and Discussion ESCA Analysis. Two typical survey and high-resolution ESCA diagrams of TL synthesized under low and high methane flow rate conditions are presented in Figures 2 and 3, and Table 1. By analysis of the data, the following observations can be made: Low methane flow rates induce very thin plasma layer depositions. The presence of fairly intense Si2s (149.7 eV) and Si2p (99.2-99.8 eV) peaks indicate that the deposited layers have a thickness of less than 100 Å; the silicon peaks have a substrate origin. The relatively high oxygen and sulfur relative atomic concentrations and the moderate relative fluorine atomic concentrations of the low methane flow rate origin layers indicate that significant defluorination mechanisms accompany the deposition processes. These reactions lead to intense fluorine atom mediated etching and free-radicalformation mechanisms which are responsible probably for the deposition of only very thin layers and for the oxygen uptake developed under post plasma open laboratory environments. The presence of larger methane quantities in the reaction mixture reduces significantly the oxygen and sulfur contents of the deposited TL layers by generating probably very volatile HF, HS, and OH entities, limiting on this way the fluorine-atom based intense etching reactions. As a result thick TL layers are deposited with low oxygen and sulfur and very high fluorine atomic compositions. High-resolution deconvoluted, C1s binding energy region ESCA data indicate the presence of various nonequivalent carbon-based linkages. Films deposited at low methane flow rate environments exhibit, C-C; CdC (284.6 eV); *CH2-CF2; COO (286.3 eV), and very low surface area C-F; CdO (288.3 eV) binding energy peaks. Hydrocarbon-type macromolecular structure deposition, the presence of intense defluorination reactions, and postplasma oxidation reactions are assumed to be responsible for the existence of these linkages. High methane flow rate conditions mainly generate fluorine and carbon atom based bonds. The dominance of CF3 (294.6 eV), CF2 (292.7 eV), and C-F (287.8 eV) binding energy peak surface areas relative to the surface area of C-C and CdC (284.6 eV) binding energy peak is obvious. This nonequivalent C1s binding energy peak surface area composition is also in good agreement with the high relative fluorine atomic composition (59.3%). It also can be noted that pure PBS plasmas do not lead to detectable plasma-layer depositions. ATR-FTIR Analysis. The spectra of intensely fluorinated fluorocarbons exhibit very intense bands in the 1100-1350 cm-1 region. These vibrations are mainly associated with C-F frequency absorption. It was found
Figure 4. ATR-FTIR spectrum of a high fluorine content TL in the 500-4000 cm-1 wavenumber region.
Figure 5. ATR-FTIR spectra of low (A) and high (B) fluorine content TLs in the 500-2000 cm-1 wavenumber range.
that the frequency rises with the degree of fluorine substitution and that the absorption band usually is split into two peaks originating from symmetric and asymmetric vibrations. The CF3-CF2- and CF3-CF- groups have also a characteristic and very strong absorption range (730-750 cm-1), which is related to the CF3 deformation frequency. These structures easily can be identified by the intense absorption present in the 730-750 and 1100-
752
Langmuir, Vol. 16, No. 2, 2000
Naum et al.
Figure 7. AFM image of low (S12) and high (S10) fluorine content TL samples.
Figure 6. Influence of duty cycles and the CH4 partial pressures on the relative fluorine atomic composition (A), on the relative C-C bond concentrations (B), and on the relative C-F bond concentrations (C).
1400 cm-1 ranges and by the lack of any significant absorption above 1400 cm-1. Figures 4 and 5 exhibit a characteristic spectrum of a high and low fluorine content TL in the 500-4000 cm-1 wavenumber region and a high resolution diagram of low (Figure 5 - S12) and high fluorine content (Figure 5 (S10)) TLs in the 600-1600 cm-1 wavenumber range. The almost total absence of C-H (sp, sp2, and sp3) vibrations in the 2800-3000 cm-1 zone and the presence of strong absorption in the 1300-1400 and 750-1150 cm-1 wave-
number intervals indicate the presence of highly fluorinated macromolecular structures. It also can be observed that the relative ratios of 1300-1400 cm-1/750-1150 cm-1 peak areas have significantly higher values in the case of high fluorine content structures. This allows us to suggest that higher concentrations of CF2 functionalities are involved in the high fluorine content macromolecular layers. These structures have been synthesized under high methane partial pressure environments which allow us to suggest that the presence of hydrogen atoms is required during the plasma synthesis for diminishing the etching effects of the plasma-generated fluorine atoms. The influence of duty cycle and the CH4 partial pressure on the relative atomic composition and on the relative concentrations of C-C and C-Fx bonds of the plasmagenerated layers are presented in Figure 6. One can observe that short plasma-on periods generally result in low oxygen and sulfur atomic concentrations, regardless of the methane partial pressures employed, while longer plasma-on periods generate the highest concentration values for these elements (Figure 6A). The influence of these parameters on the C and F relative surface atomic concentrations is more complex (Figure 6B). However, it can be noted that high duty cycle values (long plasma-on periods) are associated with the highest relative carbon and lowest fluorine relative atomic concentrations. This obviously can be related to the presence of a more intense fragmentation processes of PBS associated with the consecutive formation of very volatile HF molecules, which easily can leave the system. It also can be observed that the highest relative fluorine surface atomic compositions
PTFE-like Thin Layers
Langmuir, Vol. 16, No. 2, 2000 753
Figure 8. DTA/TG diagram of high fluorine content TL, recorded under nitrogen blanket.
are produced at low duty cycle and high methane partial pressure values. This is also reflected in the relationship between the duty cycle and methane partial pressure values and the relative C-C and C-Fx bond concentrations (Figure 6C). Short plasma-on intervals and high methane partial pressures generate the lowest C-C and the highest C-Fx concentrations involved in the structures of plasmadeposited layers. AFM Analysis. AFM images show that the high fluorine-content structures are associated with rougher morphologies (Figure 7) in comparison to the low-fluorinecontent substrates. This also might be related to the presence of different deposition/etching competitive mechanisms. DTA/TG Analysis. A typical DTA/TG diagram, of high fluorine content TL, recorded under a nitrogen blanket, is presented in Figure 8. One can observe that a slight weight gain and loss is recorded in the range of 60-200 °C, then the main decomposition of the TL is initiated. It is noteworthy that at temperatures as high as 300 and
400 °C around 80% and 50%, respectively, of the sample (samples 12 and 10) are still retained. Conclusions PBS is an adequate starting component for the deposition of TL layers. These structures can only be deposited in the presence of methane when the presence of hydrogen atoms will limit the active fluorine atom based etching reactions. The partial pressure of the methane and the plasma parameters allow to control the nature of plasmadeposited fluorinated macromolecular layers. The thermal stability of TL structures is fairly good; intense decomposition reactions are initiated only at temperatures higher than 200 °C. Contact angle evaluations indicate the presence of low surface energy layers. It was shown that the structures of the plasma-deposited layers are not affected by the nature of the substrates. LA990872I
