AD-A253 049 NASA AVSCOM Technical Memorandum 105629 Technical Report 91-C-023 DTIC Cyclopentadiene Evolution During JUL 201992 Pyrolysis-Gas Chromatography C of PMR Polyimides William B. Alston PropulsionD irectorate U.S. Army Aviation Systems Command Lewis Research Center Cleveland, Ohio Richard E. Gluyas Lewis Research Center Cleveland, Ohio and William J. Snyder Bucknell University 92-18964 Lewisburg, Pennsylvania Prepared for the Fourth International Conference on Polyimides sponsored by the Society of Plastic Engineers Ellenville, New York, October 30-November 1, 1991 ARM~LI US AVIATIONV SYSTEMS COMMAND ....... .AVIATION R&T ACTMITY • 3,I ' . "o;; CYCLOPENTADIENE EVOLUTION DURING PYROLYSIS-GAS DTIC QUALITY r) CHROMATOGRAPHY OF PMR POLYIMIDES William B. Alston Acesi oa For Propulsion Directorate U.S. Army Aviation Systems Command W G8T& Lewis Research Center Cleveland, Ohio 44135 IJQ ,.12@*¢d . JuI~at!m' Richard E. Gluyas Materials Division NY National Aeronautics and Space Administration Distribu _n/ Lewis Research Center Aviiabilty Cod Cleveland, Ohio 44135 "'f iI-an d-/ or William J. Snyder Dist Special Department of Chemical Engineering Bucknell University Lewisburg, Pennsylvania 17837 ABSTRACT The effects of formulated molecular weight (FMW), extent of cure, and cumulative aging on the amount of cyclopentadiene (CPD) evolved from Polymerization of Mono- meric Reactants (PMR) polyimides were investigated by pyrolysis-gas chromatography (PY-GC). The PMR polyimides are addition crosslinked resins formed from an aromatic diamine, a diester of an aromatic tetracarboxylic acid and a monoester of 5-norbornene-2, 3-dicarboxylic acid. The PY-GC results were related to the degree of crosslinking and to the thermo-oxidative stability (weight loss) of PMR polyimides. Thus, PY-GC was shown to be a valid technique for the characterization of PMR polyimide resins and composites via correlation of the CPD evolved versus the thermal history of the PMR sample. INTRODUCTION The Polymerization of Monomeric Reactants (PMR) polyimides are high temperature resistant resins first reported by workers at the NASA Lewis Research Center [1,2]. Today, the resin, known as PMR-15, is mainly used as the matrix in fiber reinforced composites for a variety of advanced aerospace applications [3,4]. Currently, PMR-15 is an addition-type, thermally crosslinked resin, which is readily processed to yield fiber reinforced composites with excellent mechanical property retention for up to 2000 hr at 316 *C in air [5]. However, the effects of composition, processing conditions, and high temperature aging in air are not thoroughly understood in terms of chemical changes occurring in the resin during these processes. For example, it is known that the post- curing process (heating in air after curing) affects mechanical properties [5] but the associated chemical changes are not thoroughly understood. Unfortunately, chemical characterization of addition crosslinked PMR resins is difficult by standard analytical methods such as nuclear magnetic resonance (NMR), liquid chromatography and gel permeation chromatography because of the insolubility and intractability of these resins. Thus, a technique that does not require polymer solubility for analysis would be desirable. Pyrolysis-gas chromatography (PY-GC) is one such technique and it has been used [6,7] to characterize insoluble polymers but it has not been used to make systematic quantitative measurements on the PMR polyimides. Thus, the purposes of this Deceased. To whom correspondence should be addressed. investigation are: (1) to quantitatively measure and evaluate the significance of the evolution of cyclopentadiene (CPD) during pyrolysis of cured or postcured PMR resins; (2) to identify and obtain quantitative information required for the future development of useful life prediction methods for PMR resin/graphite fiber composites in thermo- oxidative environments; and (3) to demonstrate how the PY-GC technique can be used to characterize the cure, postcure and aging chemistry of PMR resins and PMR graphite fiber composites. EXPERIMENTAL STARTING MATERIALS-The materials used in this study are the three monomers for preparing PMR polyimides: (1) monomethyl ester of cis-5-norbornene-endo-2, 3-dicarboxylic acid (Nadic Ester abbreviated NE), mp 99-100.5 °C; (2) 4,4'-methylenedi- aniline (MDA), mp 91-93 0C; and (3) the dimethyl ester of 3,3',4,4'-benzophenonetetra- carboxylic acid (BTDE). These were obtained from commercial sources except for the BTDE which was prepared from commercially obtained dianhydride (BTDA), mp 215-217 *C, by reacting BTDA with sufficient methanol to form a 50 weight percent solution of BTDE. Other commercially obtained materials used in this study were: (1) cis-5-norbornene-endo-2,3-dicarboxylic anhydride (Nadic Anhydride or NA), mp 164-166 and (2) dicyclopentadiene used for the preparation of CPD, bp 40-42 *C, 00; by thermal decomposition. Table i shows the structural formulas for these materials. TABLE I.- STRUCTURES OF MATERIALS Structure Name Abbreviation Monomethyl ester of cis- NE cOH 5-norbornene-endo-2,3- 2 C02M dicarboxylic acid e (Nadic Ester) 1I Dimethyl ester of 3,3'4,4'- BTDE HO2C CO2H benzophenonetet racarboxylic MeO C CO2Me acid 2 H2N 0 CH2- NH2 4,4'-McthylenedianilinC MD A cis-5-Norbornene-cndo-2,3- NA (icarboxylic anhydride (Nadic Anhydride) Cyclopentadiene CPD 2 RESIN PREPARATION-Solutions containing 50 weight percent solids in anhydrous methanol were prepared using the monomers, NE, MDA, and BTDE. These solutions were prepared with formulated molecular weights (FMW) of 1000, 1250, 1500, 2000 (corresponding to PMR-10, PMR-12.5, PMR-15 and PMR-20). The expression used to calculate FMW is: FMW = n MWBTDE + (n+I)MWMDA + 2MWNE - 2(n+1)(MWwater + MWmethanol) where MWBTDE, etc., are the molecular weights of the compounds indicated in the subscripts [2]. Thus, the molar ratios of BTDA:MDA:NE equal n:(n+l):2. Also, a solution having a molar ratio of 2:1 for NE:MDA without BTDE was prepared as a model compound. These solutions were air-dried at 75 0C for 24 hr and the residues were imidized (staged) by heating for 1 hr at 204 C. The staged imides were ground into powders and a portion of each specimen of powder was used to prepare cured resins. The procedure to prepare cured resin was to weigh about 15 g of the PMR molding powder into a 5.08-cm diam cylindrical hardened-steel mold which was closed by slip-fit brass pistons. The inside of the mold and the piston surfaces were previously coated with a thin film of Frekote-33 mold release compound. Cured resin was prepared for each of the above five compositions by heating under 13.8 MPa (2000 psi) pressure in the mold for 1 hr at 316 *C. The average heatup rate to 316 *C was about 5 °C/min. In all cases, a small contact pressure was applied to barely close the press at the beginning of heatup. The temperatures at which the full 13.8 MPa pressure was applied were: 300±5 *C for 2NE/MDA and PMR-10, 275±5 0C for PMR-12.5, 250±5 *C for PMR-15 and 230±5 0C for PMR-20. Different temperatures were used because the lower the FMW composition, the greater the resin flow during curing, hence the resin must be advanced further using a higher cure temperature before final pressure could be applied. Cured resin was prepared for the PMR-15 composition at 316 0C for cure times (dwell times) of 1, 2, and 7 hr. Also cured resin specimens were prepared for the PMR-15 composition using a 1 hr cure time and varying the cure temperatures (274, 288, 302, 316, 343, and 371 'C). A portion of each of these dense cylindrical (5.08-cm diam by 0.635-cm thick) pieces of resin obtained on curing was reduced to powder by preparing turnings on a lathe, grinding in a pellet mill, and sieving to obtain a 74 to 149 pm particle size (-100/200 powder fraction from U.S. standard sieve). Samples of the PMR-15 resin powder cured at 316 *C for 1 hr were aged in air at 316 *C for 1/3, 1, 3, 6, 12, 24, and 96 hr. PMR-15 resin powder samples were also aged in air for 1 hr at temperatures of 274, 288, 302, 316, 343, and 371 *C. CROSSLINK PREPARATION-A derivative of the crosslinking material from cured (1 hr/316 * C) 2NE/MDA was prepared by digesting the resin in hydrazine monohydrate (85 percent solution w/w) at about 75 *C for an extended period of time, namely 150 hr (method adapted from [8,91), to form MDA and the N-aminoimide of nadic crosslink 110]. The mixture then was acidified with 6N-HCI solution, adjusted to pH 10 with NaOH, stirred for 12 hr at about 90 *C, and extracted four times with chloroform. The chloro- form containing the extracted material was dried, filtered, the chloroform removed by evaporation, and the residue (MDA) weighed. The recovery of MDA from the resin specimen was 98 percent of theoretical. The aqueous layer was acidified with HCl to a 3 pH of 4. This solution (about 500 ml in volume) was filtered through a Millipore molec- ular filter (with a nominal molecular weight limit of 103), to a residual of 25 ml. Then, 400 ml of acidified water was added to the residual 25 ml and the solution volume again reduced to 25 ml by filtering. This latter step was repeated and, finally, the residual of 25 ml was dried in air at 50 *C. The resulting tan-colored solid was 36 percent of theoretical yield assuming the solid to be a mixture of the N-aminoamide acid and the N-aminoimide of the crosslink. The remainder of the nadic crosslink presumably was lost through the filter as a low molecular weight fraction. In another approach, a model polymer was prepared by polymerizing NA (cis-5- norbornene-endo-2,3-dicarboxylic anhydride) with a method analogous to that used for curing 2NE/MDA. The NA was placed in a 5.08-cm diam cylindrical mold closed with slip-fit brass pistons and slowly heated to 316 *C. Between approximately 316 to 350 'C, pressure was gradually increased to 13.8 MPa. An amber-colored, brittle poly- mer was obtained and is believed to have been formed by thermally initiated crosslinking of the NA. This material was prepared to serve as a possible model structure for the crosslink formed in the PMR system. APPARATUS AND PROCEDURES-The chromatograms of the pyrolysis were obtained by using a Chemical Data Systems pyroprobe (CDS-pyroprobe) on two different GC systems, (a Chemical Data CDS-820 and a Perkin Elmer 810), each consisting of a pyrolysis type injection port, temperature programmable column, flame ionization detector, and electronic integrator. The major differences between the two sets of apparatus were the use of a delay coil and a 0.3175-cm diam by 2.43-m 10 percent SE-30 on Chromosorb WAW stainless steel column on the CDS-820 versus direct injection on a 0.3175-cm diam by 6.09-m column on the PE 810 system. The second type of column showed the better resolution but the chromatograms obtained on the two types of columns were in good agreement, particularly in the quantitative measurements of the CPD peak areas obtained from pyrolysis of NA. Solid samples of 100 to 500 Ag were weighed into 0.318-cm diam by 2.54-cm quartz tubes and pyrolyzed by heating a platinum heater/resistance thermometer to 800 ° C for a fixed time (usually 10 sec). The carrier gas for these studies was helium at 40 ml/min. The columns typically were programmed from 40 to 202 'C at 5 to 6 C/min and held at the upper temperature for 15 min. The pyrolyzer probe interface was maintained at 150 °C and the GC detector temperature was 250 *C. In addition, the CDS pyroprobe was coupled to a Finnegan 4021 GC-quadruples mass spectrometer to confirm the identity of the CPD peak. The glass transition temperatures (Tg) of PMR-15 specimens were obtained by thermomechanical analysis (on a DuPont 943 TMA) using a penetration probe loaded with 5 g and linearly programmed from ambient to 450 *C at 20 *C/min. The temperature of inflection was taken as a measure of the Tg and correlated with the extent of cure determined by pyrolysis GC. RESULTS AND DISCUSSION PMR CROSSLINKER CHEMISTRY-The monomer mixtures used in this study form norbornenyl (nadic) endcapped imides after removal of solvent and heating at 4 204 0C for 1 hr. This reaction results in an imide oligomer structure of I [2] where n determines the FMW. Subsequently, upon curing under pressure, typically at 316 'C for 1 hr, the norbornenyl endgroups, (a maleimide reacted with a CPD), undergo a thermally initiated reverse Diels-Alder reaction resulting in a crosslinked material. The crosslinking reaction has been investigated on model compounds [11-14] and it has been historically hypothe- sized that this reaction results in structures of the following types (II, III, and IV): 0=2o 040o-0 NL 1 0 0 00 x y x y II III IV Structure II results from an initial reverse Diels-Alder reaction step followed by a hetero-nuclear addition reaction while structure III results from homo-nuclear addition polymerization. The relative number of cyclopentyl and succinimide (maleimide without the double bond) groups in structure II would depend on the amount of CPD available during the polymerization reaction. Because closed, tight-fitting molds were used to cure the resins, insignificant loss of CPD occurred. Therefore, the relative numbers of these groups in the crosslink chain are expected to be equal. However, the absence of olefin resonances in the proton NMR spectra [10], and the negative results of bromination experiments for olefins in these materials [12], suggest a very low incidence of cyclopentyl type groups (olefins). Hence, struc0tu=r8e=s0 0o f type II ar=e5 unlikely to occur. Structures of type III are also considered to be unlikely because of kinetic and spatial requirements 14,15]. Instead, if one looks at the polymerization as an alternating copolymerization of a maleimide group and the olefin on the nadimide endgroup or an olefin on the higher bicyclic structure, then the lack of observed olefin characteristics [10,121 would be con- sistent with the resulting polymerization structure IV. Such a polymerization mechanism has been demonstrated in model compounds studies [16-18] and also the polymerization has been shown to be dependent on the concentration of maleimide endgroups formed by the reverse Diels-Alder reaction [19]. CPD EVOLUTION-The amount of CPD available through pyrolysis of staged, cured, postcured and aged resins depends on: (1) the initial concentration of norbornenyl endgroups, (2) the cure temperature and time, (3) the postcure temperature and time, and (4) the aging temperature and time. Although aging may be considered an extension of the postcure process, for the purpose of this paper postcure will be defined as the process that occurs during the initial aging, typically during the first 16 hr, when the Tg is increased to exceed the desired use temperature. In general, the cure time and temper- ature determines the amount of unreacted nadimide endgroups and higher bicyclic com- pounds which would quantitatively release CPD when pyrolyzed. The postcure temperature and time controls the amount of CPD incorporated into the polymer by further thermal crosslinking and initial thermo-oxidative crosslinking reactions. Finally, the aging temperature and time determine the amount of nadimide endgroups and possibly higher bicyclic compounds that are incorporated into the polymer by further thermal and more thermo-oxidative crosslinking and, now, initial thermo-oxidative degra- dation reactions. Thus, the extent of crosslinked structure of the polymer chain at any time during these processes determines the amount of CPD available during pyrolysis. The significance of the CPD measurements and the conditions to assure complete CPD evolution during pyrolysis are discussed below. The effects of the variables of pyrolysis temperature, pyrolysis time, and particle size range of the specimen were investigated in order to insure that complete evolution of CPD occurs. The amounts of CPD for staged PMR molding powders are relatively constant over the pyrolysis range of 500 to 900 *C for a pyrolysis time of 10 sec as shown in Figure 1. In order to show that no further evolution of CPD occurs, successive pyrolysis runs on the same PMR molding powder sample were performed at temperatures ranging from 300 to 1000 'C as shown in Figure 2. Thus Figure 2 shows that CPD evolution was complete in the initial 10 sec of 800 0C pyrolysis. In addition single pyrolysis runs at 800 'C were also done for 1 to 20 sec to further show that 10 sec at 800 OC results in the maximum CPD evolution from PMR molding powders. Also PMR molding powders with particle sizes of 74 to 149 pm showed no significant differences in the CPD per pg of sample evolved on pyrolysis at 800 'C for 10 sec. Thus, the pyrolysis conditions of 10 sec at 800 °C assured the com- plete evolution of the CPD which is available from the PMR specimens of particle size 74 to 149 pm used in this study. 6.00- CPO 4.00- peak area (umoles/ umoles of sample) 2.00 - SI 4 ? I I I I 0 200 400 600 800 1000 Pyrolysis temperature, °C Figure 1. - Effect of pyrolysis temperature at ten second pyrolysis time on CPD evolution for staged PMR-1 5. 6 ICPD Attenuation MlOO 3000 5000 /xl 00 Detector pyrolysis Response x40 temperatures, 7001 -ju- 8000 9000 (MOO '1000 Retention time, min Figure 2.-Successive pyrolysis runs of cured PMR-1 5. CPD 6m SE-30 column, FID 40 mVmln He. 80000C at 10 second pyrolysis time Detector- 800 4g sample response5 0C/min from 70 to 210 00 Retention time, min Figure 3.-A typical pyrogran of one hour 316 OC cured PMR-1 5. 7 A typical pyrogram of PMR-15 cured at 316 *C for 1 hr and pyrolyzed at these con- ditions is shown in Figure 3. The first large peak in Figure 3 appears at the same reten- tion time as CPD. The identification of this peak as CPD has been confirmed for the 2NE/MDA model compound and PMR-15 resins using a CDS-pyroprobe joined to a GC-mass spectrometer in the electron ionization mode. The prominent ions of masses 65 and 66 correspond to CPD minus one hydrogen and the molecular ion, respectively. Further investigation is being conducted to determine the identity of the other peaks in the pyrogram. EFFECT OF FMW ON CPD EVOLUTION-The amounts of CPD from the series of staged (1 hr at 204 *C) and cured (1 hr at 316 *C) specimens including 2NE/MDA, PMR-10, PMR-12.5, PMR-15 and PMR-20 were determined by measuring the chromato- graphic peak areas of CPD in pyrograms obtained from 10 sec runs at 800 * C. The data, summarized in Table II, are normalized for the available average amount of CPD from staged PMR material expressed as peak area per gg of sample. Staged material is used as the normalization standard because of the good sample homogeneity and no crosslink- ing should have thermally occurred, thus insuring quantitative availability of CPD for evolution in the PY-GC procedure. TABLE II.-CYCLOPENTADIENE RELEASE DATA [N = CPD evolved (normalized to 1.00 as the C average from staged PMR material resin).] Material N 2NE/MDA (1 hr/204 *C) staged 0.955 PMR-10 (1 hr/204 'C) staged 1.060 PMR-12.5 (1 hr/204 °C) staged 1.055 PMR-15 (1 hr/204 'C) staged .970 PMR-20 (1 hr/204 'C) staged .990 2NE/MDA (1 hr/316 °C) cure .171 PMR-10 (1 hr/316 *C) cure .195 PMR-12.5 (1 hr/316 *C) cure .235 PMR-15 (1 hr/316 'C) cure a.177±.015 PMR-20 (1 hr/316 *C) cure .157 Isolated crosslink .058 Polymerized NA .070 aAverage deviation from mean, 14 determinations. 8 Table II shows that the amount of CPD evolved for the cured material is 16 to 24 percent of the staged material for FMW of PMR-10 to PMR-20. This result shows that considerable reaction of norbornenyl endgroups has occurred during the curing of staged material in such a way as to make the CPD relatively unavailable by pyrolysis. It can be speculated that the CPD available after curing evolves mainly from unpolymer- ized endgroups because of the high efficiency of CPD release in the uncured staged PMR molding powders. Apparently the curing conditions of 316 0C for 1 hr were sufficient to cause the crosslinking reaction to proceed to about the same high degree of advancement (i.e., percentage of norbornenyl groups reacted) for all the compositions studied. As the available crosslink concentration decreases with increasing chain length (increasing FMW), fewer endgroups need to react at the same cure temperature to reach the same extent of cure, thus leaving the same amount of unreacted endgroups as sites for available CPD evolution. It appears that the origin of the CPD obtained from pyrolysis of cured PMR resin is mainly unreacted endgroups (see structure I). Thus, the amount of CPD obtained from cured resin compared to that from staged resin appears to be a measure of the extent of crosslinking. This is discussed in more detail in connection with the effect of curing temperature on pyrolytically available CPD and their correlation to the glass transition temperature (Tg). EFFECT OF CURE TEMPERATURE AND CURE TIME ON CPD EVOLU- TION-The amount of CPD evolved (N,) from pyrolysis of cured PMR-15 is shown in Figure 4 as a function of cure temperature. The variable N. is defined by the following expression: NC = Area of CPD peak per 9g of cured specimen = normalized amount of CPD Area of CPD peak per pg of staged specimen for a cured specimen This graph indicates a strong decrease in NC as the cure temperature increases. A linear extrapolation to a value of unity for N, gives an initial starting temperature for this decrease as approximately 270 *C. This temperature agrees well wiLh the value of 275 °C reported [11,20] as the temperature of initiation for the reverse Diels-Alder 1.00- -0 .80 _P CPD .60 - evolution (1.00 uncured PMR resin) 40 .20 0 I I I I I F 270 290 310 330 350 370 One hour cure temperature. °C Figure 4.--CPD evolved (Nc) for PMR-1 5 as a function of one hour at cure temperature. 9