1980
Chem. Mater. 1999, 11, 1980-1985
Low-Temperature Metallization of Poly(tetrafluoroethylene) and Poly(chlorotrifluoroethylene) by Chemical Vapor Deposition Alan Berry* and Noel H. Turner Chemistry Division, Code 6170, Naval Research Laboratory, Washington D.C. 20375 Received June 8, 1998. Revised Manuscript Received May 17, 1999
Aluminum films were deposited on poly(tetrafluoroethylene) and poly(chlorotrifluoroethylene) substrates pretreated with TiCl4 vapor at ambient temperature for 1 h and exposed to dimethylethylamine alane vapor for 8-72 h also at ambient temperature. The films were adherent, conducting, and polycrystalline with no preferred orientation. Scanning electron micrographs showed the surfaces to be rough and porous. X-ray photoelectron spectra contained evidence of oxidized and metallic Al, as well as F, O, C, and Cl.
Introduction The bonding of thin metal films to fluoropolymers is an area of considerable interest to electronic packaging and device fabrication in the microelectronics industry.1 Much of the device work in the past has concentrated on the use of metal/polyimide multilayer configurations composed of alternating layers of metal and polymer and on understanding the bonding at that interface. Recently the increasing demand for smaller and faster devices has led to the search for ways to reduce interconnection delay times by using lower resistivity metals and polymers with reduced dielectric constants. Fluorocarbon polymers with dielectric constants as low as 1.92 compared to that of 3.2-3.5 for polyimide3 are attractive candidates in this regard. Other properties of fluorocarbon polymers such as low surface energy, chemical inertness, and high thermal stability that have made them attractive for many applications have also been responsible for poor adhesion to other materials. Several mechanisms, such as mechanical anchoring, electrostatic interactions, and chemical bonding, have been proposed to account for the adhesion between a polymer and metal.4 Mechanical abrasion, preferential chemical etching, and ion-beam sputtering of a polymer surface can provide greater contact area between the polymer and metal for bonding, as well as irregularly shaped sites for enhanced mechanical interlocking. It has also been suggested that contact between metal and polymer surfaces can result in charge transfer and formation of charged layers at the interface, which could influence adhesion. There is considerable evidence that chemical bonding at the (1) Sacher, E. Prog. Surf. Sci. 1994, 47, 273. (2) Wu, P. K.; Yang, G.-R.; Ma, X. F.; Lu, T.-M. Appl. Phys. Lett. 1994, 65, 508. (3) Chang, C.-A.; Kim, Y.-K.; Schrott, A. G. In Metallization of Polymers; Sacher, E., Pireaux, J.-J., Kowalczyk, S. P., Eds. Am. Chem. Soc. Symp. Ser. 1990, 440, 416-422. (4) Wake, W. C. Adhesion and the Formulation of Adhesives, 2nd ed.; Applied Science Publishers: New York, 1982; Chapter 5. (5) Mittal, K. Pure Appl. Chem. 1980, 52, 1295. (6) Burkstrand, J. M. J. Appl. Phys. 1981, 52, 4795.
10.1021/cm980412v
interface is important to metal/polymer adhesion in general,5,6 and metal/polyimide adhesion in particular,7-11 where metals have been deposited by evaporation or sputtering. A similar conclusion has been reached for metal/fluoropolymer systems,12-19 although it is unclear what reaction is responsible for adhesion.1 For the less reactive fluoropolymers, it is usually necessary to modifiy the surface to enhance that interaction. A number of procedures have been used to introduce different functional groups to the surface for this purpose, including treatment with very reactive reagents such as sodium naphthalide,20 organolithium compounds,21 and the benzoin dianion,22 as well as plasma etching,23 laser processing,24 and ion beam irradiation.25 In the work reported here, we believe that charge transfer and (7) Bartha, J. W.; Hahn, P. O.; LeGoues, F.; Ho, P. S. J. Vac. Sci. Technol. A 1985, 3, 1390. (8) Sanda, P. N.; Bartha, J. W.; Clabes, J. G.; Jordan, J. L.; Feger, C.; Silverman, B. D.; Ho, P. S. J. Vac. Sci. Technol. A 1986, 4, 1035. (9) Ohuchi, F. S.; Freilich, S. C. J. Vac. Sci. Technol. A 1986, 4, 1039. (10) Vasile, M. J.; Bachman, B. J. J. Vac. Sci. Technol. A 1989, 7, 2992. (11) Pertsin, A. J.; Pashunin, Yu. M. Appl. Surf. Sci. 1991, 47, 115. (12) Wheeler, D. R.; Pepper, S. V. J. Vac. Sci. Technol. 1982, 20, 442. (13) Chang, C.-A.; Kim, Y.-K.; Schrott, A. G. J. Vac. Sci. Technol. A 1990, 8, 3304. (14) Park, J. M.; Matienzo, L. J.; Spencer, D. F. J. Adhes. Sci. Technol. 1991, 5, 153. (15) Pan, F.-M.; Huang, J.-L.; Liaw, C.-F. J. Vac. Sci. Technol. A 1993, 11, 3076. (16) Shi, M. K.; Lamontagne, B.; Selmani, A.; Martinu, L.; Sacher, E.; Wertheimer, M. R.; Yelon, A. J. Vac. Technol. A 1994, 12, 29. (17) Shi, M. K.; Lamontagne, B.; Selmani, A.; Martinu, L.; Sacher, E.; Wertheimer, M. R.; Yelon, A. J. Vac. Technol. A 1994, 12, 807. (18) Chen, T. C. S.; Mukhopadhyay, S. M. J. Appl. Phys. 1995, 78, 5422. (19) Siperko, L. M.; Thomas, R. R. J. Adhes. Sci. Technol. 1989, 3, 157. (20) Bening, R. C.; McCarthy, T. J. Macromolecules 1990, 23, 2648. (21) Dias, A. J.; McCarthy, T. J. Macromolecules 1987, 20, 2068. (22) Costello C. A.; McCarthy, T. J. Macromolecules 1987, 20, 2819. (23) Bernier, M. H.; Klemberg-Sapieha, J. E.; Martinu, L.; Wertheimer, M. R. In Metallization of Polymers; Sacher, E., Pireaux, J.J., Kowalczyk, S. P., Eds. Am. Chem. Soc. Symp. Ser. 1990, 440, 147160. (24) Niino, H.; Yabe, A. Appl. Phys. Lett. 1993, 63, 3527. (25) Takahagi, T.; Ishitani, A. Macromolecules 1987, 20, 404.
This article not subject to U.S. Copyright. Published 1999 by the American Chemical Society Published on Web 07/14/1999
Low-Temperature Metallization by CVD
chemical bonding are the most likely mechanisms operating between the Al and fluoropolymer since no enhanced roughening of the surfaces was done or observed. Polymer surfaces have been metallized by several processes, including electroless deposition, flame or arc spraying, sputtering, evaporation, and chemical vapor deposition (CVD).26 However, the smaller design requirements of electronic devices have made it necessary to find ways to deposit metal films at lower temperatures to reduce grain sizes, to prevent migration between layers, and to grow in selective areas. CVD has the advantages of low processing temperatures, conformal coverage of irregularly shaped surfaces, and high throughput, but it requires volatile precursors that decompose cleanly and on demand. Much of the work to date has focused on the use of Al and Al alloys for interconnects, resulting in the study of a variety of Al precursors including alkyls such as triisobutylaluminum (TIBA, i-Bu3Al)27 and the amine-stabilized alanes, e.g., trimethylamine alane (TMAA, Me3NAlH3),28 triethylamine alane (TEAA, Et3NAlH3),29 and dimethylethylamine alane (DMEAA, Me2EtNAlH3).30 To our knowledge, the use of CVD to metallize polymers has been limited to the plasma-enhanced CVD of metals on poly(tetrafluoroethylene) (PTFE)31,32 and the deposition of Al films from TMAA at 100 °C on polyimide treated with TiCl4.28 This work describes the deposition and characterization of Al films on PTFE and poly(chlorotrifluoroethylene) (PCTFE) treated with TiCl4 (v) and exposed to DMEAA at low pressure and ambient temperature. Experimental Procedure Titanium tetrachloride (Alfa) was vacuum distilled according to RT f -63 °C f -196 °C, and the -63 °C fraction was used. DMEAA was synthesized according to procedures described in the literature and identified by the Al-H stretching frequency at 1790 cm-1 in the gas-phase infrared spectrum.30 During routine handling, DMEAA was found to react with silicone vacuum grease to produce a volatile Si-H species; no reaction was observed with a fluorocarbon grease or halocarbon wax. Unlike TMAA, DMEAA is a pyrophoric liquid. PTFE (60 and 770 µm thick) and PCTFE (127 µm thick) substrates approximately 8 × 16 mm were cut from commercial samples obtained from Read Plastics, Inc., Rockville, MD, and were cleaned by sonication for 5 min each in aqua regia, triply distilled water, and trichloroethylene. After the samples were cleaned and dried under a stream of nitrogen, substrates were transferred to one side of a small (9 mL) double-tube Pyrex reactor with an 8-mm Kontes vacuum stopcock, and the reactor was evacuated to a minimum pressure of 5 × 10-6 Torr. For those samples to be pretreated with TiCl4, ∼0.1 mmol of the tetrachloride was condensed into the second tube of the reactor at -196 °C, warmed to ambient temperature, and left for 1 h. During this time, TiCl4 (l) was present in the second tube and (26) Krulik, G. A. In Encyclopedia of Polymer Science and Engineering, 2nd ed.; Kroschwitz, J. I., Ed.; John Wiley & Sons: New York, 1987: Vol. 9, pp 580-598. (27) Bent, B. E.; Nuzzo, R. G.; Dubois, L. H. J. Am. Chem. Soc. 1989, 111, 1634. (28) Gladfelter, W. L.; Boyd, D. C.; Jensen, K. F. Chem. Mater. 1989, 1, 339. (29) Gross, M. E.; Fleming, C. G.; Cheung, K. P.; Heimbrook, L. A. J. Appl. Phys. 1991, 69, 2589. (30) Simmonds, M. G.; Phillips, E. C.; Hwang, J.-W.; Gladfelter, W. L. Chemtronics 1991, 5, 155. (31) Haag, C.; Suhr, H. Appl. Phys. A 1988, 47, 199. (32) Meyer, H.; Schulz, R.; Suhr, H.; Haag, C.; Horn, K.; Bradshaw, A. M. In Metallized Plastics 2; Mittal, K. L., Ed.; Plenum Press: New York, 1991; pp 121-130.
Chem. Mater., Vol. 11, No. 8, 1999 1981 the substrate was exposed to TiCl4 (v). Following removal of the TiCl4, the reactor was evacuated to a minimum of 5 × 10-6 Torr before treatment with DMEAA or removal of the substrates. DMEAA (∼0.2 mmol) was transferred in vacuo to the second tube of the reactor at -196 °C and warmed to ambient temperature. Evidence of aluminum deposition was usually visible within 1 h, and the process was allowed to continue for 8-72 h. Volatile material that was not condensable at -196 °C (presumably H2 from AlH3 decomposition) was pumped away, and the condensable material was collected and identified by infrared spectroscopy as Me2EtN. Evacuation of the reactor with the substrate continued until the pressure reached a minimum of 5 × 10-6 Torr, at which time the substrate was transferred to a vial in a drybox (O2 < 1 ppm) for storage prior to characterization. No special care was used to minimize exposure to air of aluminized samples during preparation for characterization. Substrates exposed only to TiCl4 (v) and examined by X-ray photoelectron spectroscopy (XPS) were transferred to the instrument in a nitrogen-filled glovebag to minimize reaction with air and moisture. XPS data were collected on a Surface Science Instruments Model 100-03 system operating at a base pressure of ∼3 × 10-9 Torr and using Al KR1 radiation. An electron beam charge neutralizer with a nominal potential of ∼2 V was used to minimize charging effects, which were to be expected for insulating samples.33,34 In addition, non-aluminized samples were covered with a gold charging screen to further reduce charging.33 The maxima of C 1s spectra for PTFE in Figure 5A,B were shifted to 292.5 eV, the value for PTFE reported by Beamson and Briggs;34 the maxima of the F 1s peaks in Figure 6A,B were also shifted by the same amount. For the PCTFE core-level spectra in Figures 7A,B and 8A,B, two shifts were made in order to reference the peaks. First, the C 1s spectra were shifted to 291.4 eV, the midpoint of the curvefitted C 1s peaks reported by Clark et al.35 A second shift of +0.3 eV was made to account for the difference in Clark’s PTFE value of 292.2 eV and that of Beamson and Briggs. For samples with Al deposits, the maximum of the Al metal peak was shifted to a value of 72.8 eV;36 the magnitude of the shifts was always less than 1 eV. The maxima of the C 1s and F 1s peaks in these spectra were shifted by corresponding amounts. The intensities of some spectra in Figures 5-9 have been shifted and rescaled for display purposes. Depth profiles of the samples were obtained with a differentially pumped sputter gun (Physical Electronic Model 04-303). The main chamber pressure was 2 × 10-8 Torr, and 3 keV Ar+ rastered over an area of ∼2 mm2 were used. Depth profiles for the samples with Al were obtained with staggered sputter intervals, beginning with single two- and three-second sputter times and increasing to several periods of 5, 10, 20, and 30 s for a total of 180 s. Scanning electron microscopy (SEM) was carried out using a Hitachi S-800 microscope and energy dispersive spectroscopy (EDS) with an EDAX Phoenix instrument. Thin layers (∼100 D) of Au or C were sputtered or evaporated on samples for cross-sectional analysis. X-ray diffraction (XRD) data were obtained with a Scintag XDS-2000 instrument. Conductivity measurements were made with a Lucas-Signatone (Gilroy, CA) SYS-301 four-point probe apparatus. Surface roughness measurements were obtained by the stylus method using an Alpha-Step 250.
Results Aluminum films were deposited at ambient temperature on PTFE and PCTFE substrates that had been pretreated with TiCl4 vapor also at ambient temperature in a static CVD reactor using DMEAA. The films (33) Bryson, C. E., III. Surf. Sci. 1987, 189/190, 50. (34) Beamson, G.; Briggs, D. High-Resolution XPS of Organic Polymers; John Wiley & Sons: New York, 1992; p 230. (35) Clark, D. T.; Feast, W. J.; Kilcast, D.; Musgrave, W. K. R. J. Polym. Sci. Polym. Chem. Ed. 1973, 11, 389. (36) Moulder, J. F.; Stickle, W. E.; Sobol, P. E.; Bomben, K. D. Handbook of X-ray Photoelectron Spectroscopy; Chastain, J., Ed.; Perkin-Elmer Corporation: Norwalk, CT, 1992.
1982 Chem. Mater., Vol. 11, No. 8, 1999
Figure 1. SEM of (A) high magnification of aluminum film on PTFE, (B) clean PTFE surface, (C) high magnification of aluminum film on PTFE showing “balloon” produced by beam damage, and (D) cross-sectional view of aluminum film on PTFE.
Figure 2. X-ray diffraction pattern of aluminum film on PCTFE.
were dull gray in appearance and surprisingly adherent, as they could not be removed by the Scotch tape test. Scanning electron micrographs showed the film surfaces to be rough and porous with grain sizes of approximately 1-5 µm as seen in Figure 1A for the surface of a film deposited on PTFE; the surface of an untreated sample of PTFE is shown in Figure 1B for comparison. Figure 1C shows a surface with an Al film that was examined at higher magnification under the electron microscope and experienced beam damage as shown by the “balloon” marked with an arrow. Despite the observed porosity, four-point probe conductivity measurements showed the surface to be conducting with sheet resistances of 0.3 ohms/square for the best samples. Film thicknesses of ∼7 µm for an 8-h deposition were measured from scanning electron micrographs of crosssectional views of a sample that had been fractured after partially cutting and immersing in liquid nitrogen for several minutes to obtain a clean edge as shown in
Berry and Turner
Figure 3. XPS survey spectra of PTFE (A) clean, (B) treated with TiCl4, and (C) treated with TiCl4 and DMEAA. The major peaks for the observed elements are indicated, together with the X-ray excited Auger F and O peaks.
Figure 1D. This picture further shows the roughness of the surface as discussed below. Longer deposition times of about 70 h did not result in significantly thicker films, probably because H2 was not removed from the cell after the first few hours, thereby allowing an equilibrium condition to be established. X-ray diffraction patterns of substrates with Al films contained peaks corresponding to the (111) and (200) orientations for Al as seen in Figure 2.37 The I200/I111 ratio of 30/100 as compared to the reported value of 47/100 for random polycrystalline Al indicated a nearly random orientation of the film. XPS survey spectra of unsputtered samples of PTFEs clean, treated with TiCl4, and treated with TiCl4 and DMEAAsare shown in Figure 3A-C. Similar spectra for PCTFE are shown in Figure 4A-C. The clean and TiCl4-treated PTFE surfaces were essentially identical with no evidence for Ti, Cl, or O, whereas the TiCl4treated PCTFE surface showed evidence of a small O 1s peak and weak Ti 2p peaks. Both substrates with Al films contained oxidized and reduced metal species, as well as F, C, O, and Cl. The expanded C 1s and F 1s core-level spectra of the above PTFE samples are shown in Figures 5 and 6. Aluminum 2p core-level spectra of unsputtered and sputtered samples are shown in Figure 9A,B. The corresponding C 1s and F 1s spectra of clean and TiCl4treated PCTFE are shown in Figures 7 and 8, and those of PCTFE with Al in Figure 9C,D. (37) Al: Index card No. 4-0787. Powder Diffraction File, McClune, W. F., Ed.; JCPDS International Center for Diffraction Data: Swarthmore, PA.
Low-Temperature Metallization by CVD
Chem. Mater., Vol. 11, No. 8, 1999 1983
Figure 6. F 1s XPS core-level spectra of PTFE (A) clean, (B) treated with TiCl4, and (C) treated with TiCl4 and DMEAA.
Figure 4. XPS survey spectra of PCTFE (A) clean, (B) treated with TiCl4, and (C) treated with TiCl4 and DMEAA.
Figure 5. C 1s XPS core-level spectra of PTFE (A) clean, (B) treated with TiCl4, and (C) treated with TiCl4 and DMEAA.
There were no visible deposits of Al on substrates that were not treated with TiCl4, or that were pretreated with TiCl4 and exposed to air, prior to exposure to DMEAA. In each case for a PTFE substrate, the XPS spectrum contained only a very weak O peak in addition to the F and C peaks expected for PTFE. Spectra of samples of PCTFE with no TiCl4 treatment, or treated with TiCl4 and exposed to air, prior to exposure to DMEAA were similar to each other and contained the following peaks with relative intensities O (1), C (0.64), Al (0.53), F (0.09), Cl (0.06) for DMEAA exposure only, and O (1), C (0.94), Al (0.63), F (0.13), Cl (0.07) for TiCl4-air-DMEAA treatment. Sputtering with Ar+ produced a slight increase in concentration for all
Figure 7. C 1s XPS core-level spectra of PCTFE )A) cleans the cutoffs on both ends of this spectrum resulted from the detector position during data collection, (B) treated with TiCl4, and (C) treated with TiCl4 and DMEAA.
Figure 8. F 1s XPS core-level spectra of PCTFE (A) clean, (B) treated with TiCl4, and (C) treated with TiCl4 and DMEAA.
elements except C, which decreased. In these samples, there was evidence for only one Al species. XPS depth profiles for the samples with aluminum are shown in Figure 10. Aluminum proved to be the major component of the Ar+-sputtered film on PTFE, increasing in atom percent as the sputter time increased to about 140 s, after which the concentration remained constant. This was accompanied by a corresponding decrease in relative concentrations of F, O, and C. Both oxidized and reduced forms of Al were found in the film with the former being predominant on the unsputtered surface and the latter on the surface sputtered for 180 s as seen in Figure 9A,B. Aluminum also proved to be the major component of the film on PCTFE but in a smaller relative amount. As in the film on PTFE, longer sputtering times resulted in a decrease in the concen-
1984 Chem. Mater., Vol. 11, No. 8, 1999
Figure 9. Al 2p XPS core-level spectra of (A) PTFE treated with TiCl4 and DMEAAsunsputtered, (B) PTFE treated with TiCl4 and DMEAAssputtered for 180 s, (C) PCTFE treated with TiCl4 and DMEAAsunsputtered, and (D) PCTFE treated with TiCl4 and DMEAAssputtered for 180 s.
Figure 10. XPS depth profiles for (A) PTFE treated with TiCl4 and DMEAA, and (B) PCTFE treated with TiCl4 and DMEAA.
tration of the oxidized Al species at the surface and a concomitant increase in the reduced species. The relative amounts of O and C also decreased during this time, but that of F increased. Discussion DMEAA was used to deposit adherent, porous, and conducting films of Al at ambient temperature on PTFE and PCTFE pretreated with TiCl4 vapor. X-ray diffraction results were similar to those reported for a randomly oriented film of Al deposited on Au and Cu from DMEAA at temperatures ranging from 100 °C to 300 °C.30 However, they were in contrast to results from films deposited on Si and polyimide substrates at 100 °C using TMAA, which showed preferred growth in the 111 direction.28 A reviewer has suggested that the roughness of the fluorocarbon surfaces might be a factor in producing a more random orientation. We examined the comparative roughness of clean surfaces, those treated with TiCl4, and those with an Al film using the stylus method. The arithmetic average (AA) roughness, which was calculated from the areas under the trace above and below the centerline, was
Berry and Turner
determined to be 1.6 µm for a clean PTFE surface, 1.8µm for a PTFE surface treated with TiCl4, and 1.4 µm for PTFE with an Al deposit. However, using statistical averages to characterize a substrate surface can give an incomplete description of the surface texture since low values can mask the presence of sharp peaks or spikes.38 It would appear that such was the case with these samples where deviations from the centerline of as much as 6 µm were observed for the clean and TiCl4treated surfaces and upward of 13 µm for the surface with Al. Consequently, we could distinguish no difference in surface texture between the clean and TiCl4treated surfaces, but we could see gross differences in the surfaces of these samples and those with Al films. Comparisons of XPS peak positions and widths between insulating and conducting samples are often made difficult by sample charging that results in binding energy shifts and peak broadening in the spectra of insulating materials. Appropriate corrections in binding energies were made in the clean and TiCl4treated samples, which were insulating, by referencing the C 1s peaks to those reported for PTFE and PCTFE by Beamson and Briggs34 and Clark et al.35 Peaks in samples with Al coatings were close to those expected for conducting samples, indicating there was a reasonable pathway to ground potential. We have discussed observed changes in peak positions and shapes qualitatively and used no quantitative data from peak fitting operations to specify relative amounts of individual elements. Survey spectra of clean and TiCl4-treated PTFE shown in Figure 3A,B were essentially identical, as were the corresponding C 1s and F 1s core-level spectra in Figures 5A,B and 6A,B. The fwhm values of the F and C peaks of the clean sample were 1.86 and 1.50 eV, respectively, and were in good agreement with the published values of 1.7 and 1.3 eV.35 Treatment of the PTFE substrate with TiCl4 vapor for 1 h at ambient temperature gave very similar values of 1.81 and 1.34 eV. The difference in the F 1s and C 1s binding energies for both the clean and TiCl4-treated samples was 397.1 eV compared to 398.0 eV reported previously for PTFE.35 Thus we have concluded there was no detectable XPS evidence for a bonding interaction between PTFE and TiCl4 in these samples. Although the XPS survey spectra of clean and TiCl4treated PCTFE samples in Figure 4A,B contained differences due to the presence of O and Ti as noted above, the core-level C 1s and F 1s spectra in Figures 7A,B and 8A,B were qualitatively very similar. The F-C binding energy difference was 396.3 eV for the clean sample and 397.4 eV for the TiCl4-treated sample; a second sample of clean PCTFE gave a value of 397.2 eV. We attributed the variations in these binding energy differences from each other and from the value of 399.4 eV calculated for PCTFE35 to sample charging rather than a bonding interaction between PCTFE and TiCl4 since the appearances of the corresponding peaks were similar. The less intense, lower energy peaks in the C 1s region were assigned to C-CFn, C-F, CF-CFn, and various forms of adventitious C as seen by previous workers.39,40 (38) Brown, R. In Handbook of Thin Film Technology; Maissel, L. I., Glang, R., Ed.; McGraw-Hill Book Co.: New York, 1970; pp 6-146-17. (39) Shi, M. K.; Lamontagne, B.; Selmani, A.; Martinu, L. J. Vac. Technol. A 1994, 12, 44. (40) Barr, T. L.; Seal, S. J. Vac. Sci. Technol. A 1995, 13, 1239.
Low-Temperature Metallization by CVD
Deposition of Al on TiCl4-treated PTFE surfaces resulted in a decrease in intensity and broadening of the C 1s and F 1s peaks associated with fluorocarbon species, as seen in Figures 5C and 6C, and the appearance of a more intense, lower energy peak associated with adventitious material in the C spectrum. The broadness of the F peak was attributed to a differential charging effect experienced by CF2 groups located in different environments, such as those near the conducting surface and those associated more closely with the insulating bulk material.41 Observation of the PTFE surface during XPS analysis would not be unexpected, since the Al films did not have a uniform morphology and were porous as evidenced by beam damage in the SEM micrograph in Figure 1C. This was in contrast to the complete coverage exhibited by 30 Å Al films evaporated onto fluoropolymer substrates.42 Analysis of the cross-sectional samples by EDS at levels near the surface and the interface showed two to three atom percent F compared to 10 atom % observed in the XPS depth profile in Figure 10A. This indicated that most of the F seen in the XPS analysis originated from the PTFE surface, as described above, and that the small amounts of F seen near the surface may have migrated through the Al film as seen in thin, evaporated films.42 Aluminum atoms produced from the decomposition of AlH3 would be very reactive, and a comparison of the Al-F and C-F diatomic bond energies (159 ( 3 vs. 128 ( 5 kcal/mol)43 indicated the reaction of Al with C-F bonds was thermodynamically favorable. The C 1s and F 1s core-level spectra of PCTFE samples treated with TiCl4 and DMEAA also differed significantly from the clean and TiCl4-treated samples as seen in Figures 7C and 8C. The F peak was broader, but not so much as in the PTFE sample with Al, and the principal C peak was attributed to adventitious C. Both XPS and EDS data revealed the presence of O and C in the Al films. Since the samples were exposed to the atmosphere for brief periods during transfer to the XPS chamber, it was likely that contamination from oxidation and adventitious material occurred. Results from previous deposition work using amine alanes have shown no evidence of hydrocarbon contamination. Temperature-programmed desorption studies of Me3N and Et3N from Al (111) surfaces from 150 to 600 K have shown no residual C or N species on the surface.44,45 Furthermore, the level of C impurities reported in Al films deposited from DMEAA between 373 and 623 K was below Auger spectroscopy detection limits once the top layers of material were removed.30 Therefore, it was unlikely that the amine produced in our work, Me2EtN, (41) Barr, T. L. J. Vac. Sci. Technol. A 1989, 7, 1677. (42) Du, Y.; Gardella, J. A. J. Vac. Sci. Technol. A 1995, 13, 1907. (43) Handbook of Chemistry and Physics; Weast, R. C., Ed.; CRC Press: Boca Raton, 1975; pp F-216, 217. (44) Dubois, L. H.; Zegarski, B. R.; Kao, C.-T.; Nuzzo, R. G. Surf. Sci. 1990, 236, 77. (45) Dubois, L. H.; Zegarski, B. R.; Gross, M. E.; Nuzzo, R. G. Surf. Sci. 1991, 244, 89. (46) Silvain, J. F.; Arzur, A.; Alnot, M.; Ehrhardt, J. J.; Lutgen, P. Surf. Sci. 1991, 251/252, 787. (47) Gavrilenko, V. V.; Chekulaeva, L. A.; Zakharkin, L. I. Izv. Akad. Nauk SSR, Ser. Khim. 1977, 1231. (48) Levy, R. A.; Green, M. L.; Gallagher, P. K. J. Electrochem. Soc. 1984, 131, 2175. (49) Gross, M. E.; Cheung, K. P.; Fleming, C. G.; Kovalchick, J.; Heimbrook, L. A. J. Vac. Sci. Technol. A 1991, 9, 57.
Chem. Mater., Vol. 11, No. 8, 1999 1985
was responsible for residual hydrocarbon impurities. The rough surface morphology and porous structure of these samples could produce a shadowing effect during sputtering that would result in the presence of surface contaminants in data taken even after lengthy sputtering. Oxygen levels from EDS measurements of the film cross section were