Chemistry of Materials 1989, 1, 14-16
14
dissolved Au/PFTA particles are smaller, having hydrodynamic radii of -7-20 nm.6 (The sizing of particles by PCS is strongly weighted toward the largest particles present, by the relationship I ( r ) 0: r6
where I is the intensity of the scattered light and r is the radius of a p a r t i ~ l e . ~ ) Another interesting feature of these Au/PFTA particles is that after dispersion in a solvent they can be deposited as a gold-appearing film or powder, allowed to dry, and then be redissolved. The particles are both lyophilic (solvent loving) and lyoselective. They also possess some of the properties of very large ligand-stabilized molecular clusters, and preliminary EXAFS and XRD measurements indicate that they consist of a metal core. Thus, fluorocarbon moieties, detected by GC-MS analysis of bromination and pyrolysis products, evidently exist as ligands coordinated only to surface metal atoms. The evidence suggests a colloid/cluster duality in the properties of these particles. A few recent accounts of large metal clusters bearing peripheral ligands or coatings have been reported by other workers,&l0and in this context we find parallels between those and the present Au/PFTA particles." The Au/PFTA colloids, both dissolved and undissolved, undergo reactions with various additives. In solution, this usually leads to coagulation, floculation, and precipitation of the particles. Addition of any ionic substance (HC1, NaI, KBr, HgClJ immediately brought this about, as did 12. Bromination of the Au/PFTA solid resulted in the evolution of perfluoroalkyl bromides. When tertiary phosphines were allowed to react with the particles, precipitation was observed; however, after 2 days the solid material had redissolved and the mixture was colorless. Removal of the solvent left a white Au-phosphine complex that has not yet been fully characterized. The reactivity of the particles appears to parallel that of organogold(1) compounds." IR, XPS, pyrolysis/GC-MS, and 19FNMR studies revealed the presence of a perfluorocarbon moiety strongly associated with the colloidal particles before and after being dissolved in acetone. Elemental analysis of Au/ PFTA powder, recovered from acetone and thoroughly dried, yielded the formula Au17CmHgEN(0,by balance). The N / F ratio in this formula is nearly the same as in PFTA; thus, there is apparently one molecule of PFTA, not necessarily intact, for every 17 Au atoms. The excess carbon, all of the hydrogen, and the assumed oxygen are probably due to coordinated acetone residues. Determination of the Au/fluorocarbon interaction has been difficult, and efforts to elucidate are continuing. Of particular importance in this work is the study of the kinetics and morphology of aggregation. Pyridine added
in very small amounts induces a controlled rate of aggregation that can be monitored by PCS, TEM, and changes in plasmon absorbance. Early qualitative results reveal three stages of growth (1) fast aggregation (amalgamation) of solvated atoms to spherical colloidal monomers, (2) pyridine-induced monomer aggregation (without amalgamation) to compact spherical clusters, and (3) aggregation of these clusters to larger, noncompact or ramified structures. Figure 2 shows the micrograph of a sample taken near the end of the second stage and the beginning of the third stage. Note the resistance of the monomer particles to merge or amalgamate together. The concept of reversibility, i.e., fragmentation of aggregates, is currently being investigated. Ongoing work in this area focuses on the development of this system as a model to which other aggregation phenomena may be compared, for example, in the synthesis of new magnetic particles in our laboratories. The surface chemistry of the colloid particles is also being explored. Acknowledgment. The support of the 3M Co. and partial support by the National Science Foundation-Materials Chemistry and the Office of Naval Research is acknowledged with gratitude. We thank Drs. Fred Behr and Richard Howells for providing PFTA samples.
Radiation Cross-Linking of Poly[ bis(2-(2-methoxyet hoxy )ethoxy )phosphazene]: Effect on Solid-state Ionic Conductivity Jordan L. Bennett, Alexa A. Dembek, and Harry R. Allcock* Department of Chemistry T h e Pennsylvania State University University Park, Pennsylvania 16802
Bruce J. Heyen and Duward F. Shriver* Department of Chemistry, Northwestern University Evanston, Illinois 60208 Received October 7, 1988
At the present time great interest exists in solid electrolyte systems for possible use in high-energy-density batteries.'P2 The polyphosphazene poly[bis(2-(2-methoxyethoxy)ethoxy)phosphazene] (I) shows excellent prom-
OCzH,0CzH40CH,
(6) The early growth stage that was observed by using microscopy and visible spectroscopy has not yet been as successfully monitored with PCS. This was primarily due to difficulties associated with intense absorbance
of the 488-nm laser line. I t is anticipated that by use of the 633-nm line of a He-Ne laser absorbance by the sample will be minimized in future experiments. ( 7 ) Taylor, T. W.; Scrivner, S. M.; Sorensen, C. M.; Merklin, J. F. Appl. O i t . 1985,24, 3713. (8) Schmid, G.;Klein, N.: Korste, L.; Kreibig, U.;Schonauer, D. Polyhedron 1988, 7, 605, and references therein. (9) Vargaftik, M. N.; Zagorodnikov, V. P.; Stolyarov, I. P.; Moiseev, I. I.; Likholobov, V. A,; Kochubey, D. I.; Chuvilin, A. L.; Zaikovsky, V. I.; Zamaraev, K. Lf Timofeeva, G. I. J. Chem. Soc., Chem. Commun. 1985, 937. (10) Bradley, J. S.; Hill, E.; Leonowicz, M. E.; Witzke, H. J. Mol. Catal. 1987, 41, 59. (11) Puddephatt, R.J. The Chemistry of Gold; Elsevier: New York, 1978; pp 98-156, and references therein.
I . n I15000
ise as a solid electrolyte host when dissolved LiCF3S03is used as the charge carrier.= Although this system shows (1) Armand, M. B.; Chabagno, J. M.; Duclot, M. J. In Fast Zon Transport in Solids; Vashista, P., Mundy, J. N., Shennoy, G. D., Eds.; North-Holland New York, 1979; p 131. (2) Shriver, D. F.; Farrington, G. C. Chem. Eng. News 1985,63 (20), 20.
(3) Blonsky, P.M.; Shriver, D. F.; Austin, P. E.; Allcock, H. R. J. Am.
Chem. SOC.1984, 106,6854.
(4) Blonsky, P. M.; Shriver, D. F.; Austin, P.; Allcock, H. R. Solid
State Ionics 1986, 18/19, 258.
(5) Allcock, H. R.;Austin, P. E.; Neenan, T. X.; Sisko, J. T.; Blonsky, P. M.; Shriver, D. F. Macromolecules 1986, 19, 1508.
0897-4756/89/2801-0014$01.50/0 0 1989 American Chemical Societv
Communications ionic conductivity 2-3 orders of magnitude higher than an analogous complex of poly(ethy1ene oxide) with LiCF3S03 a t moderate temperatures, its value as a polymeric electrolyte is limited by an inherent lack of dimensional stability. The polymer molecules are so flexible that the polymer flows slowly under light pressure. It has been shown that I can be cross-linked chemically to produce a material with increased dimensional stability and, when doped, provides comparable conductivity to the analogous un-cross-linked system.' In earlier work we showed that @'Coy-irradiation brings about cross-linking of this polymer and increases the dimensional stability.8 This was demonstrated by conversion of the water-soluble polymer into a series of water-swellable prehydrogels by irradiation over a range of doses (1.2-20.9 Mrad). The degree of water imbibition decreased with increases in irradiation dose as a consequence of increased cross-linking. Additionally, these changes were also accompanied by an increase in the toughness and mechanical strength of the water-swollen gels. In the present communication we report the effect of @'Coy-irradiation cross-linking on the solid-state ionic conductivity of [NP(OC2H40C2H40CH3)2]n doped with LiCF3S03. A key question was the degree to which radiation cross-linking of I might decrease the segmental motions of the polymer and lower the ionic mobility. The ionic conductivity of the y-irradiated cross-linked polymer/salt complexes was investigated through two complementary sets of polymer-electrolyte combinations. In one (series A) the salt was added to the polymer before irradiation, while in the second (series B) the salt was added to a solvent -swelled polymer after irradiation and the solvent was then removed. These parallel experiments were carried out to eliminate the possibility that the presence of LiCF3S03 during irradiation would markedly affect the ionic conductivity of the final material. Series A and B each consisted of six thin films of I with 0.25 equiv of LiCF3S03per polymer repeat unit (experiments 1-6). These were y-irradiated with the dose rates ranging from 0.18 to 0.19 Mrad/h. The total radiation doses were as follows: for 1 A and B, no irradiation; for 2 A and B, 0.5 Mrad; for 3 A and B, 1.5 Mrad; for 4 A and B, 3.0 Mrad; for 5 A and B, 5.0 Mrad; for 6 A and B, 20.0 Mrad.gJo The 20-Mrad dose corresponds on the average to the formation of one cross-link for every 100 repeating units.8a The polymer I used in both series was synthesized by a method described previ~usly.~ In series A, both polymer I and LiCF3S03 (4:l) were dissolved in tetrahydr0furan.l' Six thin films were cast from this solution into glass containers. The solvent was allowed to evaporate under a flow of dry air in a casting chamber for 48 h. The films were Torr) for 48 h and further dried in a vacuum (60 "C, transferred to glass containers that were evacuated before sample irradiation. In series B, polymer I was dissolved in absolute ethanol. Six thin films were cast from this solution and dried in the manner described for series A. After irradiation, a solution of LiCF3S03 in absolute ethanol was added to each of the irradiated films such that the final molar ratio of polymer repeat units to salt was (6) Blonsky, P. M.; Shriver, D. F.; Austin, P.; Allcock, H. R. Polym. Mater. Sci. Eng. 1985, 53, 118. (7) Tonge, J. S.; Shriver, D. F. J. Electrochem. SOC.1987, 134, 269. (8) (a) Allcock, H. R.; Kwon, S.; Riding, G. H.; Fitzpatrick, R. J.; Bennett, J. L. Biomateriak 1988, 9, 509. (b) Allcock, H.R.; Fitzpatrick, R. J.; Gebura, M.; Kwon, S. Polym. P r e p . 1987,28, 321. (9)%o y-irradiation experiments were carried out a t the Breazeale Nuclear Reactor a t The Pennsylvania State University. (10)The polymer/salt complex with a stoichiometry of 41 was shown to give the highest Conductivity between room temperature and 100 O C . See ref 3. (11)All manipulations were carred out under an inert atmosphere.
Chemistry of Materials, Vol. 1, No. 1, 1989 15 Table I. Conductivities i2-l cm-' ) at 25 "C of Cross-Linked Polymer Electrolytes sample A B sample A B 1 1.78 1.77 4 2.04 2.05 2 1.62 1.88 5 1.35 1.88 3 1.29 1.87 6 1.96 1.71
1 ooo/s
Figure 1. Temperature dependence of ionic conductivity of series A, log u (Q-l cm-') vs lOOO/T: A, 20 Mrad; 0 , 5 Mrad; 0 , 3 Mrad; X, 1.5 Mrad; +, 0.5 Mrad; 0,no irradiation.
4:1, as in series A. Swelling of the polymer with absolute ethanol was repeated several times to ensure thorough diffusion of the salt throughout the polymer matrix. Thus, series A and B are comparable systems, differing only in the order of LiCF3S03 incorporation and y-irradiation. The conductivities of the polymer electrolyte samples were measured by using stainless steel electrodes in spring-loaded polypropylene conductivity cells. The samples were annealed at 80 "C overnight, and impedance data were collected over a range of temperatures (10-90 "C) and frequencies (100 Hz to 3 MHz) by using a Hewlett-Packard 4129A LF impedance analyzer. The data were plotted as the imaginary component vs the real component of the impedance to determine the bulk resistance.12 Conductivities, u, were calculated on the basis of the geometric area of the electrode and thickness of the sample. Data a t room temperature are given in Table I. The conductivity of each sample was plotted as log u vs 1000/T (in kelvin). All 1 2 samples showed nonlinear behavior with a slight curvature characteristic of amorphous polymer-salt ~omp1exes.l~Figure 1 shows all six samples from series A on the same log u vs 1000/T plot. The conductivities were not affected from one polymer electrolyte to the next and did not vary more than 5 X Q-' cm-' from one sample to the next. This is well within the error of the experiment. The series B samples, in which salt was incorporated after irradiation, have almost identical conductivities. We conclude that the level of cross-linking in these irradiated polymers does not significantly suppress the polymer segmental motion responsible for ionic conductivity. A primary advantage of the cross-linking of I by the @'Co y-irradiation method, compared to the chemical crosslinking method, is that no impurities are introduced into the polymer matrix. Chemical cross-linking requires the incorporation of a difunctional reagent, for example, poly(ethy1ene glycol), while radiation cross-linking involves only side-group coupling reactions. Moreover, the ability to cross-link the system with the salt already present allows much greater control over the materials properties and the shape of devices that employ this system. Thus, although both methods provide increased dimensional stability without reducing ionic conductivity, the @'Coy-irradiation (12)Impedance Spectroscopy; .. Macdonald, J. R., Ed.; John Wiley and Sons: New York, 1987. (13)Ratner, M. A.; Shriver, D. F. Chem. Reo. 1988, 88, 109.
16
Chemistry of Materials 1989, 1, 16-18
cross-linking technique is more appealing. We have recently become aware that research workers in another group have carried out studies parallel to our own and have come to similar c o n c l ~ s i o n s . ~ ~ Acknowledgment. The synthesis and y-irradiation work at The Pennsylvania State University were supported by the Public Health Service through the National Heart, Lung, and Blood Institute. The work at Northwestern University was supported by the Office of Naval Research, and facilities were provided by the Northwestern University Materials Research Center, which is supported by the NSF. We thank Walter Johnson of the Breazeale Nuclear Reactor for his assistance. (14)Nazri, G.;Meibuhr, S. P. Abstracts of Papers, 174th Meeting of The Electrochemical Society, Chicago, IL; The Electrochemical Society: Pennington, NJ; Abstract 24. In a private communication, we were informed that a communication has been submitted on this work.
Laser-Induced Deposition of Copper from (Triethy1phosphine)cyclopentadienylcopper(1)
C. G. Dupuy, D. B. Beach,* J. E. Hurst, Jr., and J. M. Jasinski* IBM Research Division T h o m a s J . W a t s o n Research Center PO Box 218, Yorktown Heights, N e w York 10598 Received August 15, 1988 Laser-induced deposition of metals from organometallics is of interest for numerous electronics applications, including repair of open lines and direct patterning of metal lines for chip and package customization. Most of the work in this field has relied on readily obtainable alkyl, carbonyl, and acetylacetonate compounds.' Many of these compounds do not decompose to leave pure metal, or they do so only at temperatures that are too high to be compatible with advanced substrate materials or structures used in microelectronics. In the case of copper, a metal that is highly desirable for use in packaging metallization, successful deposition is further hampered by the fact that no volatile, stable, halide, carbonyl, or alkyl compounds are known. Successful laser-induced deposition of copper has been demonstrated only from acetylacetonate complexes.24 In this Communication we report the first use of (triethylphosphine)cyclopentadineylcopper(I) for the laserinduced deposition of copper. Since it contains no oxygen, this compound offers a potential advantage over the acetylacetonate compounds in producing high-purity copper films. This complex has also recently been demonstrated as a source of copper in the deposition of copper gallium ~ulfide.~ (Triethylphosphine)cyclopentadienylcopper(I) was prepared according to the procedure of Cotton and Marks.6 * Authors to whom correspondence should be addressed. (1)Ehrlich, D. J.; Tsao, J. Y. J . Vac. Sci. Technol. Bl 1983,969. (2)Houle, F.A.; Jones, C. R.; Baum, T. H.; Pico, C.; Kovac, C. A. Appl. Phys. Lett. 1985,46,204. (3)Braichotte, D.;van der Bergh, H. Springer Ser. Opt. Sci. 1985,48, 38. (4)van Hemert, R.; Spendlove, L. B.; Sievers, R. E. J. Electrochem. SOC.1965,112,1123. ( 5 ) Hara, K.; Kojima, T.; Kukimoto, H. Jpn. J . Appl. Phys. 1987,26, L1107. (6)Cotton, F. A.; Marks, T. J. J . Am. Chem. SOC.1970,92, 5114.
The white crystalline material has a room-temperature vapor pressure of 2 mTorr, sublimes without decomposition a t 65 "C, and is stable for a few hours in air. The substrates chosen for preliminary deposition studies were silicon, 2000-A Si02 on silicon, and 5-wm polyimide on silicon. The substrates were degreased prior to use and then loaded into a stainless steel cell fitted with a quartz window. The organometallic was loaded into the cell in a drybox, and the cell was evacuated with either a liquid nitrogen trapped roughing pump or a turbomolecular pump. In cases for which deposition rate data are reported below, the deposition cell was heated at 70 "C to increase the vapor pressure of the compound to -50 mTorr, thereby increasing deposition rates. For the photothermal deposition studies, the 514-nm output of a coherent Innova 20 argon ion laser was focused on the substrate through a microscope objective to a spot size of -10 pm, or with a simple lens to a spot size of 100 pm. Lines were generated by translating the sample chamber beneath the microscope with computer-controlled x-y tables. Estimates of laser power densities in our current apparatus have only order-of-magnitude accuracy. This severely limits our abilities to calculate surface temperatures from simple theory. For the excimer-laser-induced deposition, the 248-nm (KrF) output of a Lambda-Physik 203 MSC laser or the 308-nm (XeC1) output of a Lumonics TE-860-4, both operated at 5 Hz, was imaged through a mask onto the substrate. Compositional analyses were carried out by Auger spectroscopy after the sample was exposed to air. Minimum detectable levels of impurities are estimated to be 1% for carbon, oxygen, and phosphorous. In all cases, the top -200 A of the sample was sputtered off by 2-keV argon ion bombardment before compositional analysis was performed. Photothermal deposition on silicon substrates, using power densities of 104-105W/cm2 in a spot size of 100 pm, produced copper dots that were free of impurities, as determined by Auger spectroscopy. We estimate that the surface temperature of the silicon at the beginning of deposition was 200-300 0C.7 Similar results were obtained when copper dots and lines were written on polymer substrates. Laser power densities were again in the range of 104-105 W/cm2 but in a spot size of -10 pm. No detectable impurities were found below the top 200 A of the deposits. As a check for thermal damage to the polymer, copper deposits were etched away and the underlying polymer was examined with an optical microscope. Since polyimide is thermally damaged at temperatures above 400 "C and we were able to deposit copper on this substrate without damaging the polymer, the surface temperature required for decomposition of the copper complex to clean metal on polyimide is less than 400 OC. To make resistivity measurements, we wrote copper lines on silicon substrates overcoated with 2000 A of silicon dioxide and containing gold contact pads. Figure 1shows scanning electron microscope (SEM) pictures of copper lines deposited on Si-SiOp by laser-induced thermal decomposition. The polycrystalline, granular morphology is typical of laser chemical vapor deposition (CVD) grown material, with the line shape, growth rate, and morphology being very sensitive to power density and scan rate.3,4s8 The line in Figure l a was drawn with a laser power density of -5 x lo5 W/cm2 in a spot size of -10 pm scanned at
-
-
-
-
(7) (a) Sparks, M. J.Appl. Phys. 1976,47,874.(b) Lax, M. J. A p p l . Phys. 1977,48,3919. (8)Braichotte, D.;van den Bergh, H. Springer Ser. Chem. Phys. 1984, 39,183.
0897-4756/89/2801-0016$01.50/0 0 1989 American Chemical Society
