J O U R N A L O F C H E M I S T R Y Materials Processing stoichiometric silicon carbide fibers from polymethylsilane. Part 1 Precursor fiber processing Z-F. Zhang, C. S. Scotto and R. M. Laine* Departments of Materials Science and Engineering, Chemistry and the Macromolecular Science and Engineering Center, University of Michigan, Ann Arbor, MI 48109-2136, USA. E-mail: talsdad@umich.edu Received 8th July 1998, Accepted 9th September 1998 A highly branched form of polymethylsilane (PMS), –[MeSiH]x[MeSi]y–, has been synthsesized and successfully processed into infusible precursor fibers ca. 30–100 mm in diameter. These precursor fibers were converted into stoichiometric, nanocrystalline SiC fibers, 20–70 mm in diameter, by pyrolysis in an inert atmosphere at ramp rates up to 20 °Cmin-1 to 1000 °C.Precursor synthesis, fiber processing, fiber curing and pyrolytic processing are described. Bulk materials were characterized using TGA, DTA, chemical analysis, and XRD. (NO2) infusibility can be obtained with the incorporation of Introduction only small amounts of oxygen.7–9 The NO2 treated fibers are In the 1970s, Yajima et al.1–4 developed a polymer precursor then exposed to boron trichloride (BCl3).The incorporation route to thin (10–15 mm diameter) Si–C–O ceramic fibers that of boron permits pyrolytic processing to dense, substantially are basis of the NicalonTM and TyrannoTM fibers now used in crystalline SiC fibers. Thermally stable, substantially polycryshigh performance composite materials. In this process, precur- talline SiC fibers can be formed from PCS with 3–5 wt.% sor fibers are first produced by melt-spinning polycarbosilane boron and some titanium which binds with the excess boron (PCS), –[MeHSiCH2]x–.These precursor fibers are then air to make TiB2. The boron acts as a sintering aid to promote cured and pyrolytically transformed to ceramic fibers at ca. densification at temperatures >1400 °C.It is also required to 1200 °C. Unfortunately, this approach has several draw- retain fiber integrity. The resulting SiC fibers have O contents backs:1–6 (1) PCS is produced from polydimethylsilane, of <0.1 wt.% after heating to >1600 °C. Stoichiometric b-SiC –[Me2Si]x–, via an expensive multistep synthesis; (2) the initial fibers, as well as SiC fibers containing up to 20 wt.% excess 152 Si5C ratio leads to excess C in the final product; (3) air carbon were produced. Fiber densities approach 3.1 g cm-3 curing used to provide polymer fiber infusibility introduces (fully dense SiC#3.2 g cm-3).Average tensile strengths of up undesirably high oxygen contents; (4) the resulting non- to 3 GPa and elastic moduli of >420 GPa were obtained. stoichiometric, amorphous fibers oVer mechanical properties These properties are closer to the properties of bulk SiC than inferior to bulk, polycrystalline, stoichiometric SiC, and (5) any other precursor derived fibers reported so far.These fibers fiber application temperatures are limited to 1200 °C because represent the state-of-the-art in SiC fiber processing but still the retained oxygen reacts with C and Si to form gaseous require expensive, multistep processing.species at high temperatures. The release of gaseous species Clearly, precursor polymers with a Si5C ratio of 151 oVer degrades fiber properties as pores form coincident with exag- the best potential to obtain stoichiometric SiC fibers. gerated grain growth leading to significant decreases in tensile Polymethylsilane (PMS), –[MeSiH]x–, with a 151 Si5C stoichistrengths at >1200 °C.ometry, oVers the opportunity to meet this potential. Spinnable To overcome these problems and achieve SiC literature PMS, –[MeSiH]x[MeSi]1-x, was first synthesized by Seyferth mechanical properties, stoichiometric, dense and polycrystal- et al.,10,11 using dehalocoupling of MeHSiCl2 (Si5C=151) line SiC fibers with diameters of ca. 15 mm must be made, with Na: preferably via a simple, inexpensive route. Recent eVorts have MeHSiCl2+Na � –[MeSiH]x[MeSi]1-x–+NaCl3 (1) targeted lowering oxygen and carbon contents to obtain near stoichiometric SiC fibers, as described below. However, precursor fibers drawn from this polymer (Mn= Toreki et al. have prepared Yajima PCS withMn#5–10 kDa 400–700 Da) gave black powders on pyrolysis in N2, unless (cf. 1–2 kDa in the Yajima process).5 These polymers do not they were air cured prior to pyrolysis. Pyrolysis gave materials melt on heating, but are soluble in organic solvents and form with excess Si (ca. 25 wt.%) and low ceramic yields (<30 wt.%). viscous solutions suitable for dry-spinning. The dry-spun pre- To increase yields and C contents, compounds with vinyl cursor fibers are heated directly in an inert atmosphere to high groups, e.g.[Me(CH2NCH)SiNH]3 or [Me(CH2NCH)- temperature, without air-curing. SiC fibers produced by this SiO]4–6, were combined with this PMS. The improved ceramic method contain <2 wt.% oxygen and therefore have better yields were ca. 70 wt.%. No detailed analyses of fiber mechanithermal stability than Nicalon fibers.However, the initial 152 cal properties were reported. Si5C ratio still gives fibers with excess carbon. The Toreki PMS, –[MeSiH]x–, can also be synthesized via dehydrocoup- fibers exhibit tensile strengths of ca. 1.8 GPa at 1000 °C, and ling of MeSiH3, the Harrod reaction, using titanocene or 1.2 GPa after heating to >1500 °C in Ar.By comparison, zirconocene complexes, Cp2MMe2 (M=Ti, Zr; Cp=C5H5), Nicalon fibers heated to 1500 °C exhibit tensile strengths of as catalysts:12 only 0.3 GPa. Sacks et al. have improved on the original Toreki fibers to produce phase pure, SiC fibers with the xMeSiH3CCCCCCCCA 0.2 mol% Cp2TiMe2 cyclohexene, 45–60 °C –[MeSiH]x–+0.5xH2 appropriate properties, unfortunately the exact chemistry has (2) not been described.6 Kobayashi et al.recently described a similar PMS synthesis An alternative approach designed to minimize oxygen conusing (C5Me5)2NdCH(SiMe3)2 as catalyst.13 The synthesis is tents was developed at Dow Corning. Dow Corning patents indicate that when Yajima PCS is cured in nitrogen dioxide a one-step polymerization of MeSiH3, with ca. 90% conversion J.Mater. Chem., 1998, 8, 2715–2724 2715of monomer to PMS. Like Seyferth PMS, Harrod PMS oVers vacuum and the product purified by sublimation at 100 °C at access to nearly pure SiC upon pyrolysis.14,15 The highly 10-2 Torr. crosslinked, solid polymer does not melt on heating and gives ceramic yields of 70–80 wt.%. On heating to ca. 1000 °C in Preparation of methylsilane (MeSiH3): CAUTION, this N2, PMS provides near stoichiometric SiC (SiC0.9H0.2O0.1), material can burn on contact with air.Commercial MeSiH3 with nanosized b-SiC grains and 5–10 wt.% excess Si. This contains suYcient chlorosilane impurities to deactivate the polymer oVers the type of ceramic product desired for SiC catalyst and further purification before polymerization was fibers.necessary. Consequently, two procedures were developed to In related work, Hengge et al. used Cp2MMe2 to catalyze produce pure MeSiH3: (A) by reacting MeSiCl3 with LiAlH4 dehydropolymerization of disilanes:16 using ethylenediamine (en) as the purification agent, and (B) as a by-product of the catalytic redistribution of polymethyl- MeH2SiSiH2Me (neat) CA Cp2ZrMe2 H[MeSiH0.58]xH (3) hydridosiloxane, –[MeHSiO]x–, as described below.The low Si5H ratio is indicative of a highly crosslinked and intractable material. These polymers were reported to provide Method A. MeSiCl3 (Aldrich) was refluxed over Mg phase pure SiC, although the method of characterization was turnings and distilled under N2 to eliminate impurities. not described. Ethylenediamine was stirred over CaH2 overnight and dis- The target of the work reported here was to use Harrod tilled under N2.All solvents and reagent liquids were PMS precursor to process stoichiometric SiC fibers. The transferred via cannulae under N2. LiAlH4 reduction of aantages of using Harrod PMS are: a one step, high-yield MeSiCl3 was run in a flame dried three-necked, 1 L flask.synthesis; 151 Si5 C ratio and high ceramic yields. Dry THF (500 mL) was added and the flask was then taken Furthermore, the as-produced polymer is spinnable right inside the drybox. LiAlH4 (24 g, 0.6 mol), an addition funnel from the polymerization solution. Unfortunately MeSiH3, and a condenser were added to the flask, the assembly was the polymerization catalysts, and the resulting polymer are removed from the drybox, connected to a Schlenk line and quite pyrophoric.A nonpyrophoric PMS precursor built by pressure equalized with an N2 flush. The condenser was reverse engineering of the precursor developed here will be cooled to -46 °C with a liquid N2–MeCN slush. The outlet described at a later date.15 of the condenser was connected to a 200 mL, thick walled, The studies described below define parameters that must Pyrex trap charged with 30–50 mL of ethylenediamine (en) be considered/controlled in the overall process that leads to and cooled with liquid N2. stoichiometric, dense SiC fibers.The following sections MeSiCl3 (80 mL) was then transferred to the addition discuss: (1) synthesis of a spinnable, modified Harrod PMS; funnel and added dropwise to the reaction flask with magnetic (2) fiber spinning; (3) fiber curing; (4) control of stoichi- stirring.The MeSiH3 gas that evolved and some MeSiHCl2 ometry during pyrolysis, and (5) conversion of PMS fibers to and MeSiH2Cl were collected in the en trap over 1–2 h. After dense ceramic fibers. Studies on the polymer-to-ceramic addition, the reaction was stirred for 30 min as it warmed to transformation process for bulk polymer, using 29Si solid room temperature and was then heated slowly.Vigorous gas state NMR and FTIR, have been described.14 evolution was observed at temperatures as low as -20 °C. By employing a stepwise increase in reaction temperature (10 °C steps from 10 to 50 °C), foaming was controlled and Experimental the chemical yield maximized.During reaction, the condenser was maintained at -46 °C to condense THF and partially 1 General synthetic procedures reacted volatiles, e.g. MeSiHCl2 and MeSiH2Cl. The level of All air and moisture sensitive materials were handled using the trap bath must be high enough to prevent the excape of standard Schlenk techniques or in the argon atmosphere of a MeSiH3 out of the bubbler (w/Firestone valve), but low glove box, Vacuum Atmosphere Model No.MO40-2-Drienough to prevent clogging with frozen silane. Slight positive Lab. All solvents were distilled in Ar–N2 and degassed prior N2 pressure prevents backflow of silanes into the Schlenk to use. line. The reaction was heated at 50 °C for 4 h. During the last 0.5 h, no further gas evolved for the reaction scale Solvent purification.THF was distilled from sodium benzo- described here. Pure MeSiH3 was obtained by trap-to-trap phenone ketyl. Hexane and toluene were distilled using the distillation into an evacuated weighed metal cylinder cooled same procedure as used for THF with 10 g of tetraglyme with liquid N2. The transfer requires ca. 2 h and provides added to improve the solubility of sodium benzophenone 85–95% yields.ketyl. Diethyl ether was distilled from sliced sodium (20 g/2L flask). Acetonitrile was distilled from calcium hydride in Ar (2 g/2 L solvent). Methanol and ethanol were distilled from Method B. Reaction was carried out in a dry 2 L, threeactivated magnesium (5 g Mg, 0.5 g I2/L solvent). Small necked flash using previously published methods.17 Initially, quantities of solvent (70 mL) were added first to initiate 750–900 mL of freshly distilled toluene was added to the flask.reaction. A vigorous reaction occurs and after 15 min, 1 L of A five fold volume of toluene to –[MeHSiO]x– (Mn#2000 solvent can be added. Da, from Hu� ls) is required to avoid gelation as redistribution occurs. Then, 150 ml of –[MeHSiO]x– was placed in a dry 250 mL Schlenk flask and degassed by sparging Ar through Catalyst syntheses. Cp2TiMe2 or Cp2ZrMe2 were prepared the oligomer for 1 h.The degassed –[MeHSiO]x– was trans- by reaction of Cp2TiCl2 or Cp2ZrCl2 with MeLi according to ferred to the 2 L flask and 5–10 mL of hexane containing published methods.12 In general, ca. 100 mg of either Cp2TiCl2 20–40 mg of Cp2TiMe2 was added dropwise under Ar.or Cp2ZrCl2 (Strem Chemicals) were placed in a dry 50 mL Initiation of the redistribution reaction is slow and can require Schlenk flash with a stir bar (in the drybox). Freshly distilled 0.5–2 h depending on adventitious impurities. Reaction com- and deoxygenated hexane was added to this flask using a mences when the solution turns royal blue.17 Redistribution cannula.Approximately 1.5 mL of 1.4M of MeLi (Aldrich) to –[MeSiO1.5]x– occurs concurrent with release of MeSiH3, in either was added dropwise, and the solution was then which can be trapped without using en. The collection time filtered through Celite under Ar to remove the resulting LiCl precipitate. The filtrate was evaporated to dryness under for this reaction is 48–72 h. 2716 J. Mater. Chem., 1998, 8, 2715–27242 Polymer synthesis The amount of Me2S·BH3 added was determined by assuming that: (1) no B loss occurs during further processing, Polymerization of MeSiH3. This reaction was performed in and (2) all the Si combines with all the C to form pure SiC a 400 mL Parr reactor equipped with magnetic stirrer. The after pyrolysis.An example of the synthesis of a successful catalyst [Cp2TiMe2 or Cp2ZrMe2, 327 mg] was dissolved in spinning solution includes: 10 wt.% of TVS (0.8 g, 6 mmol), distilled cyclohexene (80 g) and added to the Parr reactor 3 wt.% Me2S·BH3 (0.24 g, 3.2 mmol), and 87 wt.% reacted under Ar–N2. The reactor was then sealed, cooled with liquid MeSiH3 (7 g, 150 mmol), which gives a molar ratio of N2 and evacuated.MeSiH3 was then transferred under vacuum Si5C5B#151.350.015. Based on the above assumptions, the from the metal container (see above) by condensing into the final B concentration in the SiC product would be 0.4 wt.%. cooled reactor. The resulting solution was warmed to 55±1 °C Note that this amount of boron makes no changes in the in a thermostatted oil bath.behaviour of TVS-PMS on pyrolysis to temperatures of Dehydrocoupling occurs with concurrent production of H2. 1000 °C. Consequently, the various discussions below do not However, the catalyst also promotes rapid hydrogenation of consider its presence or absence during pyrolysis. The eVects cyclohexene to cyclohexane. Thus, the reaction was followed of B addition are only important when fibers are processed by decreases in MeSiH3 pressure with time.The rate of above 1000 °C. This will be the subject of a following paper. pressure decrease exhibits a linear, first order dependence on initial MeSiH3 pressure. Heating was ceased when the rate of 3 Precursor fiber spinning the pressure decrease was <2 psi h-1 [ca. 25 h (1 psi#6.895× 103 Pa)] as rapid gelation occurs beyond this point.Unreac- Attempts made to melt-draw fibers at 80 °C were not successful as melting Harrod PMS rapidly leads to crosslinking and an ted MeSiH3 (with some H2 and N2) was transferred back to the attached metal cylinder cooled with liquid N2. By mass intractable material. Thus, eVorts focused on dry-spinning. diVerence, 25 g of MeSiH3 were found to react, giving a 90% yield of PMS.The PMS concentration was ca. 0.2 g mL-1. Dry-spinning (fiber extrusion). All vinylsilane modified PMS solutions were dry-spinnable when concentrated to 0.6–0.8 g The molecular weight (Mn) was ca. 1200 Da.12,14 This PMS was characterized by NMR, FTIR, TGA and DSC as mL-1 from 0.15–0.25 g mL-1. The resulting solutions were placed in an extruder (Fig. 1) mounted inside an Ar dry box. described below. Typically, 3–5 ml of viscous solution were loaded into the extruder chamber, and fibers were then extruded by applying Increasing PMS molecular weight. The reaction vessel om 100–500 psi Ar pressure to force the polymer through a above, containing ca. 25 g PMS was taken into the dry box 140 mm diameter spinneret. These extruded fibers continuously and the residual gases vented.A sample (ca. 5 ml ) was self-draw at 20–40 cm below the spinneret. Longer fibers could transferred in the dry box to a sealed, 50 ml Schenk flask and not be drawn due to dry box height limitations. Precursor heated under Ar at 60 °C until gelation occurred. Gelation fibers with 70–120 mm diameter collected across a square times were typically 7–12 h.The remaining polymer solution wooden framework with a spacing of ca. 10 cm. The fiber was then transferred into Schlenk flasks in the dry box. The diameters are not uniform because of the crude spinning sealed flasks were removed from the dry box and heated at system. In some cases, thin 30 mm diameter fibers were 60 °C (Ar) to increase the molecular weight (MW).Heating obtained. The extruded fibers were dried for at least 5 h prior was stopped after a period equivalent to 80–90% of the gel to pyrolysis. time and the solutions were stored below 0 °C. 4 Curing Star-branched polymers via vinyl modification. To further Unmodified Harrod PMS fibers melts on heating. To avoid increase the MW to improve spinnability, and to provide melting, eVorts were made to cure precursor fibers.The functionality for self-curing, PMS was modified by reaction methods examined included: (1) low temperature thermal with tetravinylsilane (TVS), Si(CHNCH2)4, 1,3,5-trimethylcures; (2) curing with traces of ammonia; (3) curing with c- 1,3,5-trivinylcyclotrisilazane [SiMe(CHNCH2)NH]3, or irradiation, and (4) curing by incorporating reactive func- dimethyldivinylsilane SiMe2(CHNCH2)2. 5–20 wt.% of these tionality.Only the last method, based on bringing the polymer compounds were added to PMS solutions brought to 80–90% close to its gelation point and then adding thermally reactive of their gelation time. The weight percentage added was based on: (MeSiH3reacted+additive)=100 wt.%. For example, 48.8 g (1.06 mol) MeSiH3 was polymerized to give 214 ml (PMS+solvent) solution.Therefore, a 30 ml portion of this solution consumed 6.84 g (0.15 mol) MeSiH3. Addition of 0.36 g (2.6 mmol), 0.76 g (5.6 mmol), or 1.71 g (12.5 mmol) TVS to a 30 mL portion of PMS solution results in 5, 10 and 20 wt.% vinylsilane-PMS solutions respectively. After adding the vinyl compounds, the PMS solutions were again heated at 60 °C (Ar, 12–24 h) to form branched, vinylmodified PMS (discussed below).After vinylsilane modifi- cation, the resulting PMS solution is stable and can be stored at room temperature for >3 d. Most research focused on TVS modified PMS solutions (TVS-PMS), to avoid incorporating N into the resulting SiC fibers. Boron modified TVS-PMS. Boron was incorporated into PMS by hydroboration of the residual vinyl groups in TVSPMS, resulting in B- and TVS-modified PMS (B-TVS-PMS).Several B additives were tested including Me2S·BH3, THF·BH3, NH3 ·BH3, Me3 N·BH3 and C4H9N2·BH3. Only Me2S·BH3 reacted with TVS-PMS to provide clear, spinnable Fig. 1 Schematic diagram of a pressure extruder made from standard swagelock fittings. solutions. J. Mater. Chem., 1998, 8, 2715–2724 2717vinyl functionality really worked.In this instance, no inter- were 5–80° 2h in 0.01° increments with a 2° 2h min-1 scan speed. Cu-Ka (l=1.54 A° ) radiation was used. mediate curing stages or green fiber pretreatments were necessary as the fibers cured even with heat ramp rates up to 20 °Cmin-1. Scanning electron microscopy (SEM). Micrographs were taken on a HITACHI S-800 microscope.SEM samples were 5 Pyrolysis prepared by breaking fibers into small segments and mounting them on the edge of an aluminum stub using double stick Precursor fibers were pyrolyzed in Ar to transform them into tape. Samples were sputter coated with a layer of Au/Pd to ceramic fibers. All pyrolysis steps below 1400 °C were carried enhance their conductivity.Micrographs of fresh fracture out in a single zone, Lindberg tube furnace (model No. 58114, surfaces were recorded to evaluate fiber microstructure. Watertown, WI) or a Thermolyne high temperature tube furnace (type 54500, Dubuque, IA). Both tube furnaces are Results and discussion equipped with Eurotherm temperature controllers (model No. 818P, Northing, England). The heating rates used were 1 Modification of PMS 5–20 °Cmin-1.Thinner fibers were cured at 20 °Cmin-1 without loss of fiber integrity. Harrod PMS typically has Mn=1000–2000 Da.12 NMR spec- Pyrolyses above 1400 °C were run in an Astro high tempera- tra provide some information on polymer structure.4,14 The ture furnace (Model 1000). All samples pyrolyzed to>1400 °C PMS 1H NMR (Fig. 2) shows only two broad PMS resonances were first pyrolyzed to 1000 °C in a Lindberg furnace, under for Si–H (3.87 ppm) and C–H (0.39 ppm); the remainder result Ar.Samples were transferred to the Astro high temperature from residual cyclohexene, cyclohexane and C6D6. The 13C furnace under ambient conditions. The Astro furnace consists NMR shows one broad Si–CH3 peak at 9.7 ppm. The 29Si of a Eurotherm temperature controller (model No. 818P, MAS spectrum exhibits resonances at -34.0 and -63.0 ppm, Northing, England), a vacuum pump, and an Ar–N2 supply. Two vacuum/Ar fill cycles were applied before starting heattreatment to ensure elimination of O2/moisture in the furnace. Samples were then heated above 1400 °C at 20–30°Cmin-1 under flowing Ar. 6 Materials characterization Thermogravimetric analyses (TGA).TGA studies of PMS precursor were carried out using a Hi-Res TGA 2950 Thermogravimetric TA Instruments Thermal Analyst 2200. Samples of 10–30 mg (chunks) were removed from the drybox and quickly placed in a Pt pan in air. The time required to load the samples was 3–5 min. Samples were then heated (10 °Cmin-1) to 1000 °C in Ar (60 cm3 min-1). DiVerential thermal analyses (DTA). DTA experiments were conducted on a DSC 2910 diVerential scanning calorimeter, with a 1600 °C DTA cell, TA Instruments Thermal Analyst 2200.Samples (10–15 mg) were removed from the drybox and quickly loaded in a Pt pan in air. The loading time was 3–5 min. Samples were heated at 5 °Cmin-1 to 1000 °C in Ar (50 cm3 min-1). Calcined alumina (Aluminum Co.of America) was used as the reference material. NMR characterization. All solution spectra were run in CDCl3 and recorded on a Bruker AM 360 MHz instrument at room temperature unless otherwise noted. Residual CHCl3 was used as an internal reference. 1H, 13C and 29Si NMR spectra were obtained with the spectrometer operating at 360, 90.6 and 71.5 MHz, respectively. Spin rates of 20 and 14 rps were used for 5 mm and 10 mm tubes respectively. 1H NMR spectra were obtained using a 4000 Hz spectral width, an acquisition time of 4.096 s and 32 K data points. 13C NMR spectra were obtained using a 20 000 Hz spectral width, an acquisition time of 0.8192 s and 32 K data points. 29Si NMR spectra were obtained using an inverse gated technique, a 20 000 Hz spectral width, an acquisition time of 0.819 s, a delay between pulses of 5–10 s and 32 K data points.Chemical analyses. Elemental analyses for selected pyrolyzed samples were performed by Galbraith Laboratories of Knoxville, TN. X-Ray diVraction ( XRD). XRD powder patterns were obtained using a Rigaku Rotating Anode Goniometer (Rigaku Denki Co. Ltd., Tokyo, Japan). Powder samples (100–200 mg) Fig. 2 (a) 1H NMR of polymethylsilane polymer synthesized via the were ground in an alumina mortar and pestle, packed in a Harrod procedure. (b) 1H NMR of 20 wt.% TVS-PMS. (c) Vinyl region of (b) expanded. glass specimen holder, and placed in the goniometer. Scans 2718 J. Mater. Chem., 1998, 8, 2715–2724corresponding to –MeSiH– units and –MeSiH2 groups respectively.14 The broad resonances indicate that the magnetic environments around H and C are not unique, suggesting branching.The OSi–CH35Si–H integration ratio in the 1H NMR spectrum is 4.5:1 rather than 351, supporting a branched structure containing cyclics. Assuming Mn=1–1.2 kDa, we can propose Fig. 3 Idealized molecular structure for PMS. an idealized PMS composition as shown in Fig. 3. In related studies that will be described in a following paper,15 we have react.If terminal –MeSiH2 groups are the most reactive Si–H identified (by mass spectral analysis) branched cyclics similar sites12 then TVS most likely reacts with these groups to form to: vinyl-capped PMS. This implies that MeSiH2 groups also cause gelation. Thus, endcapping should reduce or eliminate gelation. Indeed, following reaction with vinylsilanes, the system is stable at ambient for days to weeks without gelation.Because one TVS can react with up to four PMS molecules to form a star-branched polymer, the MW of PMS could increase as much as fourfold as suggested in Fig. 4. Initial eVorts to produce high quality SiC fibers from TVS-PMS were unsuccessful because the desired final microstructure was not obtained, as discussed below.Thus eVorts were made to introduce boron to the precursor synthesis. Small amounts of B are known to aid SiC densification during sintering.20–24 Several B containing organosilicon polymers have been synthesized.25–27 Riccitiello and coworkers25,26 and Riedel et al.27 describe reacting BH3 adducts with vinyl modified poly(diorganosilane)s to form B containing polymers.TVS-PMS is an end-functionalized oligomethylsilane (Fig. 4). However, the vinyl groups in TVS-PMS can also react with BH3 adducts, especially in the presence of catalysts, The composition shown in Fig. 3 is proposed based on the as in the TVS-PMS solution: structure shown above.15 The implication is that dehydrocoupling occurs at both chain terminating –MeSiH2 and backbone –MeSiH– groups.The formation of one tertiary Si for every three silicons suggests that the reactivity of the internal Si–H groups is (Si CH CH2)n (Si CH2 CH2 B )n BH3-adducts Cp2ZrMe2 (5) almost as high as found for the chain ends. However, dehydrocyclization is likely much favored over chain growth because This results in the formation of more complex, hyper-branched it is a unimolecular process whereas the latter is bimolecular.molecules (Fig. 5). The BH3 adduct used is (Me)2S·BH3. The implication is that the chain termini (–MeSiH2 groups) (Me)2S·BH3 and TVS are added simultaneously to a solution are much more reactive than internal Si–H groups as previously of Harrod PMS previously heated to 80–90% of its gel point. suggested by Harrod et al.12,14 One BH3 is potentially capable of reacting with three Si–vinyl The low MW polymer obtained from the Harrod procedure groups and should increase the MW and polymer viscosity as is an obstacle to fiber spinning.It is diYcult to increase MW shown by Riedel et al.27 Indeed, we have reversed engineered by simply extending the dehydropolymerization reaction time B-TVS-PMS (PMS-TVS-B) by first forming TVS-B and then because gelation occurs readily.However, stable PMS soluhydrosilylating this material with PMS to form essentially the tions can be obtained by vinyl modification of Harrod PMS. identical polymer.15 This more air stable version exhibits the Vinylsilanes react with PMS via reaction (4). Hydrosilylation GPC trace shown in Fig. 6 which supports the structure occurs thermally at 150–200 °C;10,18 however, in our system it proposed in Fig. 5; note the expected polymodality of a hyper- occurs at <60 °C most likely aided by a Cp2ZrMe2 derived branched polymer shown in Fig. 6. catalyst. The critical issues are whether or not these modifications impart spinnability, curability and sinterability, as discussed below.The 1H NMR spectrum shows no significant change in the Si–H5Si–CH3 ratio on adding TVS,19 suggesting that only a Fig. 4 Proposed molecular structure of TVS-PMS. small percentage of the Si–H bonds (e.g. –MeSiH2 groups) J. Mater. Chem., 1998, 8, 2715–2724 2719polymer for catalyst. Depolymerization was assumed to play a minor role in the reaction sequence. After venting, the polymer was reheated at 60 °C for 5–10 h under Ar to increase the MW.The solutions were brought to 80–90% of the gel time. The resulting PMS solutions, containing 0.6–0.8 g mL-1 upon partial solvent removal, were suitable for dry-spinning, although the resulting fibers were not infusible. Fig. 7 shows an SEM for a typical precursor fiber with a smooth surface and a dense interior without visible microstructure.The irregular cross-section, typical of dry-spun fibers, is due to uneven drying. Although this procedure provides improved spinnability, the fibers still melt at 100 °C. Further heating led to increased MWs; however, gels formed rapidly with loss of processability. Thus, two diVerent chemical approaches to improving infusibility, without impairing processability, were explored.One was to increase theMWlinearly to increase chain entanglement thereby improving spinnability and raising Tm above the curing point (Tc) to provide polymer infusibility. The other was to introduce reactive functional groups into the polymer structure to provide higher latent reactivity. To achieve these goals, multifunctional vinylsilanes were incorporated into the precursor synthesis.Multifunctional vinylsilanes are capable of linearly linking two or more PMS molecules together, by reaction with terminal –MeSiH2 groups, as discussed above. Linear increases in MW, obtained with divinyldimethylsilane provided better rheological properties as Fig. 5 Proposed B-TVS-PMS molecular structure. evidenced by a reduction in fiber necking during spinning; however, the fibers still melted.In contrast, the addition of tri- and tetra-vinyl functionalized silanes provided improved spinnability, and remaining unreacted vinyl groups provide sites for further branching with BH3 and higher latent reactivity to enhance polymer self-curing. Note that in all cases mentioned here, the increases in MW are assumed to occur as a consequence of the type of chemistry used; however the pyrophoric nature of the polymers produced precluded conducting MW measurements using standard techniques.A future paper will present more details about the development of higherMWPMS using a much less air sensitive derivative.15 The utility of adding vinylsilanes to increase MW is strongly dependent on the length of PMS segments linked.If vinylsilane addition occurs too early, many reactive Si–H sites are elimin- Fig. 6 GPC Trace of reverse engineered PMS and PMS-TVS-B.15 2 Precursor fiber spinning Melt-drawing was initially examined using unmodified PMS. While PMS melts and fibers can be drawn from the melt, the melted polymer crosslinks too rapidly to be considered for long term studies. Extensive eVorts were made to cure these fibers, using a variety of processes, without success. EVorts then focused on the developing stable, spinnable, and infusible polymers for dry-spinning.Toreki et al. find that Yajima PCS with molecular weights (MWs) of 5–10 kDa decomposes before it melts. Although it no longer melts, toluene solutions can be dry-spun.5 Furthermore, because it does not melt, air curing is not necessary and low oxygen content fibers result, although excess C remains a problem.Thus, eVorts were made to increase the PMS Mn to >5000 Da. Initially, eVorts were made to extend the duration of dehydrocoupling to improve PMS MWs. To increase catalyst eYciency for chain extension, residual MeSiH3 and H2 were vented, after completion of the Harrod procedure, to ensure Fig. 7 SEM of an as-spun PMS precursor fiber. that residual monomer (MeSiH3) would not compete with 2720 J. Mater. Chem., 1998, 8, 2715–2724ated at an early stage and only short chain segments are linked principle, vinyl groups that survive the incorporation process [see Fig. 2(b)] provide the requisite latent reactivity such that together. This limits the MW increases possible and impairs the processability/infusibility.For example, the addition of fiber melting no longer occurs (see below). Fig. 2(b) shows that addition of ca. 20 wt.% TVS (ca. TVS to PMS heated to <50% of the gel time did not give spinnable materials even after all the solvent was removed. 26 mol% vinyl groups) to PMS solutions provides access to vinyl-modified PMS.Fibers spun from this modified precursor, The PMS chains, before vinyl modification, were too short. This led us to add vinylsilane only at 80–90% of the gel time. TVS-PMS, were infusible. However, PMS modified with 20 wt.% Me2Si(CHNCH2)2 (ca. 17 mol% vinyl groups) still After a thorough evaluation of di- and tri-functional vinyl compounds (see Experimental section), TVS was selected for melts.One explanation is that precursor infusibility comes not only from amounts of added vinyl groups, but also from further extensive studies. TVS contains the excess carbon needed to oVset the excess silicon produced during pyrolysis macromolecular architecture. Each TVS is potentially capable of linking four PMS chains to form a star branched polymer and no nitrogen.In principle, TVS oVers the opportunity to improve spinnability by increasing the MW (therefore chain whereas Me2Si(CHNCH2)2 can only increase the PMS MW by extending the chain in one dimension to form a new, linear entanglements), and because it oVers latent reactivity which enhances self-curing. Furthermore, because it appears to cap molecular structure. The linear molecular structure may provide insuYcient chain entanglement to achieve infusibility.reactive –MeSiH2 endgroups, it stabilizes the spinning solution. Finally, B-TVS-PMS displays even better spinnability (evi- The importance of molecular architecture on curability is further supported by the fact that addition of 20 wt.% denced by the absence of necking during spinning) than TVSPMS, likely because the higher degree of branching and higher [SiMe(CHNCH2)NH]3 (ca. 11 mol% vinyl groups) also provides infusible precursor fibers. This result may arise because MW provide better chain entanglement leading to better viscoelastic properties. [SiMe(CHNCH2)NH]3, with one more vinyl group than SiMe2(CHNCH2)2, can potentially link three PMS chains to The only drawback that remains is the high degree of air sensitivity as illustrated by Fig. 8 which shows the TGA of B- form a star-branched polymer thereby providing suYcient entanglement to aid in spinning and curing.For the reasons TVS-PMS precursor fibers held at 30 °C for 24 h in dry air. The fibers exhibit a mass gain of ca. 30 wt.%, indicating mentioned above, TVS was found to be the best choice for further processing studies.extensive oxidation. This oxidation sensitivity impairs the opportunity to obtain phase pure SiC fibers unless the fibers As discussed below, pyrolysis of unmodified PMS gives a mixture of SiC and excess Si. Using TVS to modify PMS are processed in an inert atmosphere. compensates for the carbon loss by introducing additional carbon content. However, the correct balance between carbon 3 Curing unmodified PMS fibers loss and introduction must be identified to produce phase pure Before precursor fibers are subjected to high temperature SiC fibers.Fortunately, 5 wt.% TVS provides infusible PMS pyrolytic processing, they must be made infusible (cured/ precursor fibers as eVectively as 20 wt.%. This result permits crosslinked) so they will not melt on heating.Prior to our selection of an optimal quantity of TVS to produce stoichiodiscovery of the utility of vinyl-modified PMS, several curing metric SiC. studies were run on unmodified PMS fibers, including the use of low temperature long term pyrolyses, curing with traces of 5 Pyrolysis ammonia,29–31 c-irradiation32–34 and by using an excess of catalyst.10,35 None of these methods proved particularly useful, After curing, the subsequent step is pyrolytic conversion of the precursor fibers to SiC fibers.In the following section, and eVorts turned to enhancing chemical reactivity using vinyl compounds. studies on bulk pyrolytic processing of precursors using thermogravimetric analyses (TGA), diVerential thermal analyses (DTA) and powder X-ray diVraction (XRD) are discussed. 4 Curing of vinyl-modified PMS fibers Scanning electronic microscopy and chemical analysis data on Schmidt et al.36 showed that on thermolysis, vinylic polysilanes SiC fibers will also be described. heated at low temperatures generate R3Si· free radicals solely by scission of Si–Si bonds. These free radicals promote Si–H Thermogravimetric analyses (TGA) of unmodified PMS and addition across CNC double bonds (hydrosilylation) which TVS-PMS leads to crosslinking.In related work, Seyferth et al. showed that radical initiated hydrosilylative crosslinking gave Fig. 9 shows the TGA curves for PMS with selected amounts of added TVS. Both unmodified PMS and TVS-PMS precur- improved PMS ceramic yields.10,11 These results, coupled with our inability to cure unmodified PMS, led to PMS synthetic sors have higher ceramic yields (79 and 83 wt.% respectively) than typical Yajima PCS (55 wt.%).1–4 strategies that incorporated multifunctional vinyl silanes.In Fig. 9 TGAs of batches of PMS with selected amounts of TVS added. Fig. 8 TGA of TVS-PMS precursor fibers (10 wt.% TVS added) isothermed at 30 °C in dry air for 24 h.Samples were heated at 10 °Cmin-1 to 1000 °C in Ar. J. Mater. Chem., 1998, 8, 2715–2724 2721Unmodified PMS has a 151 Si5C ratio. Ideally, H2 will be the only gaseous species released during pyrolysis and the 151 Si5C stoichiometry will be retained after pyrolysis. The theoretical ceramic yield for linear, unmodified –[MeSiH]x– is 90.9 wt.%. Branched structures, e.g., –[MeSiH]x[MeSi]y–, will have slightly higher ceramic yields.However, TGAs of diVerent batches of unmodified PMS gave ceramic yields ranging from 61 to 79 wt.% (Fig. 9) indicating some loss of C and/or Si occurs concurrent with release of H2. The TGA profiles of PMS and TVS-PMS (Fig. 9) reveal several stages of mass loss. Stage 1: below 200 °C, the mass loss is insignificant (<3 wt.%), indicating good stability to this temperature.The observed mass loss is likely due to volatile low MW species. Stage 2: in the 200–400 °C range unmodified PMS undergoes mass losses in the range 13–30 wt.%, indicating that primary decomposition processes take place in this range and/or low molecular weight cyclics volatilize. Kobayashi et al. observe release of MeSiH3 in this range.13 They suggest that MeSiH3 Fig. 10 DTAs of PMS with selected amounts of TVS added. Samples were heated at 10 °Cmin-1 to 1000 °C in Ar. forms by cleavage of terminal –MeSiH2 groups. Low MW, short chain and/or branched polymers will have more terminal groups than high MW, long chain, linear polymers or species containing cyclic structures. The mass loss ranges for Harrod indicate a multitude of events that are discussed as for the TGA studies.PMS batches vary considerably depending on the reaction conditions, storage time after synthesis and solution concen- Stage 1: below 200 °C, unmodified PMS exhibits a slight endotherm, likely due to volatilization of low MW species, or tration because reaction continues under ambient conditions. TVS-PMS has smaller, quite consistent mass losses residual solvent.The TVS-PMS precursors exhibit an exotherm at low temperatures (50–150 °C) not seen for unmodified (7–10 wt.%), implying that TVS modification provides a more stabile PMS with a well defined MW and architecture. By PMS. This exotherm is ascribed to thermally promoted hydrosilylative crosslinking of remaining vinyl groups. Schmidt heating TBS-PMS to 50–150 °C, retained vinyl groups will react with remaining –MeSiH2 groups, and internal Si–H sites et al.36 and Schilling37 observed similar exotherms at 150–300 °C for such processes.The lower temperature range to crosslink the polymer. This process, seen in the DTA (below), ties up terminal –MeSiH2 groups increasing ceramic observed here is probably due to the presence of active catalyst.Stage 2: between 200 and 400 °C, a broad exotherm appears yield and coincidentally reducing gas evolution related defects in the resulting ceramic fibers. TVA-PMS also appears to in the DTA of the unmodified PMS. This transition could result from a decomposition reaction that involves cleavage of undergo Kumada rearrangement [reaction (6)] as discussed below.terminal –MeSiH2 groups, with simultaneous coupling of the resulting fragments, and evolution of MeSiH3, as proposed Stage 3: in the 400–600 °C range, unmodified PMS undergoes Kumada rearrangement (as seen by FTIR and by Kobayashi et al.13 who report the release of H2 at this stage indicating that dehydrocoupling continues with heating. NMR studies)14,15 and loses 7–8 wt.%. The mass loss in this temperature range is ascribed to release of CH4 and H2 per A similar exotherm is not observed in the DTAs of TVS-PMS precursors.The probable reason is that loss of –MeSiH2 Kobayashi et al.13 groups is minimized by crosslinking as discussed above and as supported by TGA results which show only minor mass losses for TVS-PMS in the 200–300 °C range compared to unmodified PMS.An exotherm for the TVS-PMS precursors is seen in the 300–450 °C region. Similar exotherms are also observed by Schmidt et al.36 and Schilling37 in DTAs of vinyl-substituted polysilane decomposition. The exact source of this exotherm is not clear at this point, but may be associated with the Kumada rearrangement, see below. Stage 3: between 400 and 500 °C, a strong exotherm appears in the DTA for unmodified PMS.IR and 29Si solid state NMR indicate that the Kumada rearrangement occurs in this temperature range.14,15 The Kumada rearrangement also occurs in TVS-PMS, as shown by DRIFTS studies.38 However, the TVS-PMS DTA does not show the same strong, sharp exotherm. It seems that the PMS to PCS transition exotherm shifts to lower temperature, and is partially hidden by an exotherm ascribed to loss of CH4 and H2.If this is in fact the case, then it is possible that the two events occur simul- Stage 4: in the 600–1000 °C range, only minor mass loss taneously. Another possibility is that free radicals generated (<2 wt.%) is observed, likely due to H2 evolution (see during thermally promoted hydrosilylative crosslinking pro- DRIFTS below).mote the PMS to PCS transition. More detailed studies are The TGA profile of 10 wt.% TVS-PMS precursor fibers is required to understand the exact source of these transitions very similar to that of the bulk material. and exotherms. Stage 4: the DTAs for unmodified PMS and TVS-PMS DTAs of unmodified PMS and TVS-PMS precursors show broad exotherms between 600 and 900 °C.These exotherms most likely result from H2 evolution, accord- DTAs (Ar) or PMS precursors with selected amounts of added TVS are shown in Fig. 10. As can be seen, the DTA profiles ing to the studies of Kobayashi et al.13 and hydrogenated SiC 2722 J. Mater. Chem., 1998, 8, 2715–2724results. The relatively sharp peaks at 900 °C result from however, densities increase slightly to 2.5 g cm-3.It is only on processing these fibers to higher temperatures that significant crystallization of SiC, as supported by solid state NMR, DRIFTS and XRD studies.14,18,38 changes occur. These changes and fiber mechanical properties are the subject of the next paper in this series. In future papers, we will describe the synthesis of a more air stable precursor,39 Powder X-ray diVraction ( XRD) of unmodified PMS and TVS– PMS the use of PMS for joining applications40 and for the manufacture of particle and fiber reinforced composites.41,42 In stages 1–4 (pyrolysis 800 °C), the XRD patterns of unmodified PMS and TVS-PMS exhibit amorphous features. No defined diVraction peaks can be observed in XRDs of Conclusions samples heated to <800 °C.Samples heated to 1000 °C (not Extensive eVorts were made to learn to modify polymethylsil- shown) show three broad peaks at 2h#35, 60 and 72° correane (PMS) produced directly by dehydrocoupling of MeSiH3 sponding to the (111), (220) and (311) peaks of b-SiC, and to obtain spinnable polymers. It was determined that careful indicating crystallization occurs at 1000 °C. control of the polymerization reaction conditions provided access to a branched version of PMS that exhibits viscoelastic Chemical analyses of TVS-PMS derived SiC properties that just permit it to be spun.Unfortunately, the Chemical analyses show that pyrolysis of unmodified PMS to spun fibers do not survive pyrolysis to give SiC ceramic fibers. 1000 °C (1 h, Ar) produces a material containing ca. 72.4 wt.% Furthermore, the as-synthesized PMS is highly flammable. Si, 25.4 wt.% C and trace amounts of H, indicating the The addition of chain extending/branching molecular addipresence of excess Si (Si5C=1.2251). Adding TVS permits tives containing tri- and preferably tetra-vinyl functionality, balancing the carbon loss, thus chemical analysis of 10 wt.% prior to the PMS gel point, provides access to higher molecular TVS-PMS produces a material with 68.9 wt.% Si and 31.4 wt.% weight, highly branched materials with improved viscoelastic C (Si5 C=151.06), after pyrolysis to 1000 °C (1 h, Ar).properties such that improvements in spinning are attained Chemical analysis of 1000 °C pyrolyzed fibers derived from with increased resistance to oxidation. The latter appears to this TVS-PMS gives 69.7 wt.% Si and 29.0 wt.% C (Si5C= result from capping of residual Si–H groups that are more 1.0351).Therefore, careful control of the amount of added susceptible to oxidation. Furthermore, the added vinyl com- TVS leads to near-stoichiometric SiC fibers. pounds provide reactive sites that promote thermal crosslinking and decomposition of the spun fibers before they melt.Scanning electron microscopy (SEM) of TVS-PMS/B-TVS- Finally, by controlling the amounts of additives, it is possible PMS derived SiC fibers to adjust the stoichiometry of the resulting ceramic fiber from silicon rich, to exactly stoichiometric SiC, to carbon rich fibers. Fig. 11 shows a SEM micrograph of a 10 wt.% TVS-PMS The results of these studies have suggested a new way to precursor fiber heated to 1000 °C.The fibers exhibit dense prepare related spinnable precursors that are even less air interiors and surfaces; however, at this point the fibers are sensitive and permit processing in air.41,42 nanocrystalline (2–4 nm size crystallites) and have densities (2.3 g cm-3) that are much lower than theory (3.2 g cm-3). Heating to temperatures of up to 1800 °C leads to densification Acknowledgements to ca. 3.1 g cm-3.28 SiC produced from B-TVS-PMS (0.04 wt.% B after pyrol- The authors would like to thank the Army Research ysis) behaves identically to TVS-PMS derived SiC up to Laboratories for generous support of this work through con- 1000 °C. No changes are observed by any of the analytical tract No. DOD-C-DAAL04-91-C-0068.techniques used above. Likewise fibers produced using B-TVSPMS are essentially identical to those produced without B; References 1 (a) S. Yajima, K. Okamura, J. Hayashi and M. Omori, J. Am. Ceram. Soc., 1976, 59, 324; (b) S. Yajima, J. Hayashi, M. Omori and K. Okamura, Nature, 1976, 261, 683; (c) S. Yajima, T. Shishido and H. Kayano, Nature, 1978, 273, 525; (d) S.Yajima, Y. Hasegawa, J. Hayashi and M. Iimura, J. Mater. Sci., 1978, 13, 2569; (e) Y. Hasegawa, M. Iimura and S. Yajima, J. Mater. Sci., 1980, 15, 720; ( f ) Y. Hasegawa and K. Okamura, J. Mater. Sci., 1980, 18, 3633. 2 (a) T. Yamamura, T. Ishikawa, M. Shibuya and T. Hisayuki, J. Mater. Sci., 1988, 23, 2589; (b) C. Y. Song, Y. Hasegawa, S-J. Yang and M. Sato, J. Mater. Sci., 1988, 23, 1911. 3 (a) R. M. Laine and F. Babonneau, Chem. Mater., 1993, 5, 260; (b)M. Birot, J.-P. Pillot and J. Dunogues, Chem. Rev., 1995, 1443; (c) K. J. Wynne and R. W. Rice, Annu. Rev. Mater. Sci., 1984, 14, 297; (d) R. R. Wills, R. A. Markle and S. P. Mukherjee, Ceram. Bull., 1983, 62, 904; (e) R. H. Baney and G. Chandra, in Encyclopaedia of Polymer Science and Engineering, J. Wiley and Sons, New York, 1988, vol. 13, pp. 312–44; (e) J. Bill and F. Aldinger, Adv. Mater., 1995, 7, 775; ( f ) W. Toreki, Polym. News, 1991, 16, 6; (g) G. Pouskouleli, Ceram. Int., 1989, 15, 213. 4 R.M. Laine and A. Sellinger, Si-containing ceramic precursors, in The Chemistry or Organic Silicon Compounds, ed. Z. Rappoport and Y. Apeloig, J. Wiley & Sons, London, 1998, vol. 2, pp. 2245–310. 5 See, for example: (a) W. Toreki, C. D. Batich, M. D. Sacks, M. Saleem, G. Choi and A. A. Morrone, Comput. Sci. Technol., 1994, 51, 145; (b) W. Toreki, G. J. Choi, C. D. Batich, M. D. Sacks and M. Saleem, Ceram. Eng. Sci. Proc. (July–August 1992, Cocoa Fig. 11 SEM micrograph of a TVS-PMS fiber (10 wt.% of TVS added) Beach, FL), 1992, pp. 198–208. 6 See, for example: M. D. Sacks, G.W. ScheiVele, M. Saleem, G. A. heated to 1000 °C for 1 h in Ar. J. Mater. Chem., 1998, 8, 2715–2724 2723Staab, A. A. Morrone and T. J. Williams, Mater. Res. Soc. Symp. 20 S. Prochazka and R. M. Scanlan, J. Am. Ceram. Soc., 1975, 58, 72. 21 S. Prochazka, Ceramics for High-Performance Applications, ed. Proc., 1995, 365, 3. 7 J. A. Rabe, J. Lipowitz and P. P. Lu, US Pat., 5 051 215, Sept.J. J. Burke, A. E. Gorum and R. N. Katz, Book Hill, Chestnut Hill, MA, 1974, p. 239. 1991. 8 D. C. Deleeuw, J. Lipowitz and P. P. Lu, US Pat., 5 071 600, 22 K. M. Friederich and R. L. Coble, J. Am. Ceram. Soc., 1983, 66, C141. Dec. 1991. 9 J. Lipowitz, J. A. Rabe and G. A. Zank, Ceram. Eng. Sci. Proc., 23 C. Greskovich and J. H. Rosolowski, J. Am. Ceram. Soc., 1975, 59, 336. 1991, 12, 1819. 10 (a) D. Seyferth, T. G.Wood, H. J. Tracy and J. L. Robison, J. Am. 24 J. E. Lane, C. H. Carter, Jr. and R. E. Davis, J. Am. Ceram. Soc., 1988, 71, 281. Ceram. Soc., 1992, 75, 1300; (b) D. Seyferth, H. J. Tracy and J. L. Robison, US Pat., 5 204 380, 1993; (c) D. Seyferth and Y-F. Yu, 25 M. T. S. Hsu, S. R. Riccitiello, T. S. Chen and R. Salvatore, J. Appl. Polym.Soc., 1991, 42, 851. Design of New Materials, ed. D. L. Cocke and A. Clearfield, Plenum Press, New York, 1987, p. 79. 26 S. R. Riccitiello, M. T. S. Hsu and T. S. Chen, US Pat., 4 987 201, January 1991. 11 D. Seyferth and H. Lang, Organometallics, 1991, 10, 551. 27 R. Riedel, A. Kienzle, V. Szabo and J. Mayer, J. Mater. Sci., 1993, 12 (a) Y. Mu and J. F. Harrod, in Inorganic and Organometallic 28, 3931.Oligomers and Polymers, IUPAC 33rd Symp. Macromol., ed. J. F. 28 (a) Z-F. Zhang, F. Babonneau and R. M. Laine, manuscript in Harrod and R. M. Laine, Kluwer Publ., Dordrecht, 1991, p. 23; preparation; (b) Z-F. Zhang, S. Scotto and R. M. Laine, in Mater. (b) J. F. Harrod, in Inorganic and Organometallic Polymers with Res. Soc. Symp. Proc. (Covalent Ceramics II: Non-oxides), ed.R. Special Properties, ed. R. M. Laine, NATO ASI Ser. E, vol. 206, Gottschalk, 1994, vol. 327, pp. 207–23. Kluwer Publ., Dordrecht, 1991, p. 87; (c) X. Xin, C. Aitken, J. F. 29 R. J. P. Corriu, D. Leclercq, P. H. Mutin and A. Vioux, Chem. Harrod and Y. Mu, Can. J. Chem., 1990, 68, 471; (d) H. Q. Liu Mater., 1992, 4, 711. and J. F. Harrod, Organometallics, 1992, 11, 822; (e) J. He, Q. Liu, 30 K. Okamura, M. Sato and Y. Hasegawa, Ceram. Int., 1987, 13, 55. J. F. Harrod and R. Hynes, Organometallics, 1994, 13, 336. 31 G. Burns and G. Chandra, J. Am. Ceram. Soc., 1989, 72, 333. 13 T. Kobayashi, T. Sakakura, T. Hayashi, M. Yumura and 32 P-E. Sundell, S. Jonsson and A. Hult, Radiation of curing of poly- M. Tanaka, Chem. Lett., 1992, 1157. meric materials, ACS Symp. Ser. vol. 417, ed. C. E. Hoyle and 14 (a) Z.-F. Zhang, F. Babonneau, R. M. Laine, Y. Mu, J. F. Harrod J. F. Kinstle, Am Chem. Soc.,Washington DC, 1990, pp. 459–73. and J. A. Rahn, J. Am. Ceram. Soc., 1990, 74, 670; (b) Z-F. Zhang, 33 K. Okamura, T. Matsuzawa and Y. Hasegawa, J. Mater. Sci., Y. Mu, R. M. Laine, F. Babonneau, J. F. Harrod and J. A. Rahn, 1990, 4, 55. Inorganic and Organometallic Oligomers and Polymers, IUPAC 34 M. Takamizawa, T. Kobayashi and A. Hayashida, US Pat., 33rd Symp. Macromol., ed. J. F. Harrod, R. M. Laine, Kluwer 4 604 367, August 1986. Publ., Dordrecht, 1991, pp. 127–146. 35 D. Seyferth, C. A. Sobon and J. Borm, New J. Chem., 1990, 15 (a) R. M. Laine, A. Sellinger and K. W. Chew, US Pat. pending; 14, 545. (b) A. Sellinger, Ph.D. dissertation March, 1997; (c) A. Sellinger 36 W. R. Schmidt, L. V. Interrante, R. H. Doremus, T. K. Trout, and R. M. Laine, manuscript in preparation. P. S. Marchetti and G. E. Maciel, Chem.Mater., 1991, 3, 257. 16 (a) E. Hengge, M. Weinberger and Ch. Jammegg, J Organomet. 37 C. L. Schilling Jr., Br. Polym. J., 1986, 18, 355. Chem., 1991, 410, C1; (b) E. Hengge, Organosilicon Chem. II, 38 Z-F. Zhang, Ph.D. Dissertation, Univ. of Michigan, 1995. 1996, 2, 275; (c) E. Hengge and M. Weinberger, J. Organomet. 39 A. Sellinger and R. M. Laine, manuscript in preparation. Chem., 1992, 433, 21. 40 D. R. Treadwell, R. M. Laine and R. Burzynski, to be submitted. 17 R. M. Laine, K. A. Youngdahl, F. Babonneau, J. F. Harrod, 41 K. W. Chew and R. M. Laine, J. Am. Ceram. Soc., in press. M. L. Hoppe and J. A. Rahn, Chem.Mater., 1990, 2, 464. 42 K. W. Chew, M. Nechanicki and R. M. Laine, submitted. 18 B. Boury, R. J. P. Corriu and W. E. Douglas, Chem.Mater., 1991, 3, 487. 19 C. S. Scotto, A. Sellinger and R. M. Laine, unpublished work. Paper 8/05288C 2724 J. Mater. Chem., 1998, 8, 2715–2724