J. MATER. CHEM., 1994, 4(10), 1511-1520 1511 Novel Aromatic Poly(ether ketone)s Part 1.-Synthesis and Thermal Properties of Poly(ether keto imide)s Anthony J. Lawson:+ Peter L. Pauson: David C. Sherrington,"* Stella M. Youngb* and (in part) Niall O'Brien§ a Deparfmenf of Pure and Applied Chemisfry, Universify of Sfrafhclyde, 295 Cafhedral Sfreef, Glasgo w, UKG7 7XL ICI Wilfon Maferials Research Centre, Middlesbrough, Cleveland, UK TS68JE A range of aromatic polyimides have been prepared by polycondensation of novel diamines, having four to eight benzene rings linked by ether, ketone and sulfone groups, with a number of commercially available acid dianhydrides. The effect of systematic structural alterations on the thermal properties of the polymers has been evaluated and discussed in the light of the literature.This approach has allowed the synthesis of poly(ether keto imide)s which retain the high Tg associated with polyimides (ca. 270°C) and yet have T,,, values of ca. 370°C which, in principle. would allow melt processing. Comparison is drawn with other favourable materials already described in the literature. Two important classes of thermally stable polymers are the aromatic polyimides' and the aromatic poly(ether ketone)s,2 the former in particular showing remarkably high thermo- oxidative stability. In general the polyimides have very high glass-transition temperatures, T,and decompose before melt- ing. They are also usually amorphous in nature, a factor which can influence some physical properties, e.g.toughness and tensile strength, rather adversely. Poly (ether ketone) s on the other hand have lower Tg values, but are often crystalline and melt-processable. Not surprisingly, therefore, attempting to combine the positive properties of these groups has been the centre of considerable research, most notably atby Hergenrother and co-worker~~-~ NASA. Indeed, by employing a diamine having four benzene rings linked by ether and ketone functions a semicrystalline polyimide, LARC- CPI, has been produced displaying a T, of 350°C with a Tp of 222°C. (Note: Tp of a simple polyimide is ca. 380"C, and of a simple poly(ether ketone) is ca. 150"C with T, x370 "C.) The polymer has excellent physical properties with the added crystallinity improving toughness and solvent resistance rela- tive to other polyimides, while retaining very satisfactory rigidity (q).This material has become something of a bench- mark in this area, and recently a nominally structurally identical polymer has been prepared via an alternative route employing a Friedel-Crafts acylation.' Indeed this approach has been used to generate a wide range of aromatic ether- ketone-X (EKX) polymers, where X includes imide, amide, ester, sulfone, azo and quinoxaline functionalities.*-1° The Tg and T, of a polymer are related by the phenomeno- logical Beaman" equation T,= 1.3 Tg (in K) so that reducing T, to enhance processability usually results in a fall in Tg, and loss of polymer backbone rigidity.However, since to a first approximation these thermal transitions have their origins within different regions in a polymer, i.e.the amorphous and crystalline domains, respectively, for Tg and T,, then in principle it may be possible to decouple these transitions by appropriate choice of backbone structure. Indeed for aromatic polyesters Brown et ~2.'~have demonstrated that the ratio T,/T, can be altered by increasing the proportion of rneta-t Present address: Vinamul Ltd., Mill Lane, Carshalton, Surrey SM5. $ Present address: I.C.I. plc, Fluon R & T, York House, Hillhouse International, Thornton, Cleveleys, Lancashire FY5 4QD.9 Present address: Department of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral St., Glasgow G1 1XL.substituted monomer components in the backbone, I()wering the T, significantly without any change in q. This paper describes our attempts to develop Hergenrother's approach further, to obtain semicr) stalline polyimides in which the Tp and T, are decoupled and to understand the structural design features which bring this about. The approach has been to synthesize 16 aromatic diamines with four to eight benzene rings linked together by ether, ketone and in one case sulfone groups, where, unlike in earlier poly(ether keto irnide)~,~-" the functionality adjacent to the imide unit is a ketone not an ether. We have also sought to define the contribution of rneta and para iinkages in these. Polyimides have been prepared from these using mainly pyromellitic and 3,3',4,4'-benzophenonetetracarboxylic acid dianhydrides, although three other acid dianhydrides have also been examined.Results and Discussion Polyimide Syntheses The structures of the diamines used in this work arc shown in Table 1 and their syntheses are described in Part 3 of this series. The corresponding commercially available acid dianhy- drides employed are shown in Table 2. Microanalytical data for polyimides prepared with PMDA and BTDA are shown in Tables 3 and 4 where the polymer is coded to show the precursor diamine and acid dianhydride employed. Since the polyimides are highly intractable, and in particular insoluble in organic solvents, it is impossible to determine or estimate (e.g.from reduced viscosity, RV) the molecular weights of the products. However, measurement of the RV of the precursor polyamic acid solutions is possible. From this point of view the two-step 'thermal imidisation' bynthesis (Tables 5-7) was favoured over the one-step 'solution syn- thesis' procedure using diphenyl sulfone (DPS) as the solvent (Table 8) (see Experimental section). In general therefore most syntheses used the 'thermal imidisation' procedure and, in particular, the 'thin-film casting' method (see Experimental section). It was also important that the RV of the precursor polyamic acid solution was not too high, otherwise gelation occurred and this tended to inhibit efficient imidisation. Generally RV values in the range 0.35-0.65dl 8-l were found to bc suitable (Tables 5-7), although some reached ca.3 dl 8-l very readily. The viscosity achieved was related to some extent to the reactivity of the diamine being used. Those amines with a J. MATER. CHEM., 1994, VOL. 4 Table 1 Diamines used in polyimide syntheses no. structure 1A 2A 3A 4A 5A NH2 \ 0-o-”qp””. 0 0 6A 7A 0 0 8A 9A NH2* 0 0hNH2 10A 11A 0 0 NH2-9poyJJyy7J12A ‘ \ ‘ \ 0&NH2 0 0 J. MATER. CHEM., 1994, VOL. 4 Table 1 (continued) no. structure 13A 0 0 NH2yJ0yJp0yJ&y0qP0~16A \ \ NH2 0 0 0 Table 2 Acid dianhydrides used in polyimide syntheses Table 3 Microanalytical data for polyimides synthesizcd from PMDA" code structure polymer code microanalysis found (YO) calcd.(%) PMDA CHN C II N ~~ PMDAJlA 71.6 3.0 4.7 73.2 3 1 4.7 PMDA/2A 72.5 2.9 4.7 73.2 3 1 4.7 PMDA/3Ab 72.2 3.0 4.6 73.2 3 1 4.7 PMDAJ4A 72.6 2.7 4.6 75.3 3 1 4.9 PMDAJ5 A 71.5 2.7 3.6 73.9 3 2 4.1 PMDAf6A 73.8 3.1 4.1 73.9 3.2 4.1 BTDA PMDAJ7A 73.0 2.9 4.0 73.9 3.4 4.1 PMDA/8A 72.7 3.0 3.3 74.4 1.4 3.6 PMDAJllA 74.5 3.1 2.9 75.5 3.4 3.1 PMDA/12A 74.3 3.2 3.1 75.5 ?.4 3.1 PMDAJ13A 74.3 3.0 2.9 75.5 ?.4 3.1 "'Thermal' imidisation method. "4% mol% excess of dianiine used in synthesis. ODPA Table 4 Microanalytical data for polyimides synthesized from BTDA" polymer code microanalysis BPDA found (YO) calctl. (%) ~~ CHN CHN BTDA/lA 73.6 3.0 3.6 74.4 i.2 4.0 BTDAJ2A 73.1 2.8 4.1 74.4 i.2 4.0 BTDA/3Ab 74.5 3.1 3.8 74.4 ;.2 4.0 DPSDA BTDAJSA 73.5 3.3 3.6 74.8 ;.3 3.6 BTDAJ6A 74.0 3.1 3.4 74.8 ;.3 3.6 BTDAJ7 73.7 3.1 3.4 74.8 i.3 3.6 BTDA/8A 74.2 3.4 3.5 75.2 ;.4 3.2 BTDA/l 1 A 74.9 3.4 2.7 76.0 3.4 2.8 BTDAJ12A 75.0 3.2 2.6 76.0 3.4 2.8 BTDA/13A 75.8 3.4 2.7 76.0 3.4 2.8 "'Thermal' imidisation method."5% mol% excess of diamine used in synthesis. J. MATER. CHEM., 1994, VOL. 4 Table 5 Thermal and physical properties of polyimides synthesized from PMDA using thermal imidisation natureno. of benzene RV" Tg Tm temp. of polymer code rings in diamine /dl g-l /OCb /OCb 10% Wt.loSS/°Cc of film PMDA/ 1A 4 0.44 527 663 brittle PMDA/2A 4 0.55 268 374 555 tough PMDA/3Ad 4 0.76 266 ->500 (5% wt. loss) tough --670PMDA/4A 4 0.65 tough PMDA/SA 5 0.41 230 -560 tough PMDA/6A 5 0.89 246 -561 tough PMDA/7A 5 0.46 250 449 518 brittle PMDA/8A 6 0.54 223 -528 tough PMDA/llA 7 0.56 223 42 1 504 tough PMD A/ 12A 7 0.45 234 -522 tough PMDA/13A 7 0.68 225 -5 64 tough PMDA/3Ae3f 4 244 PMDA/l 2Af 7 236 -PMDA/1 lAf 7 -223 PMDA/lA +3A( 1/1) 4 0.29 276 tough PMDA/2A + lA( 112) 4 0.36 -tough PMDA/2A + lA( 111) 4 0.36 282 tough PMDA/2A +3A( 1/1) 4 0.35 268 tough Of precursor polyamic acid.bFrom DSC trace. 'From TG curve. d4mol% excess of diamine. "5 mol% excess of diamine. fSolution imidisation in DPS. Table 6 Thermal and physical properties of polyimides synthesized from BTDA using thermal imidisation natureno. of benzene Rv" Tg Tm temp.of polymer code rings in diamine /dl g-' /oCh /oCb 10% Wt.l0SS/oCC of film BTDA/lA 4 0.58 258 376 500 (5% wt. loss) tough BTDA/2A 4 0.32 247 -560 tough BTDA/3Ad 4 1.65 226 -560 tough BTDA/5A 5 0.37 23 1 404 561 tough BTDA/6A 5 1.52 22 1 -496 tough BTDA/7A 5 0.56 22 1 320 440 tough BTDA/8A 6 0.64 205 -561 tough BTDA/llA 7 0.40 21 1 359 535 brittle BTDA/12A 7 0.42 213 -565 tough BTDA/13A 7 1.03 225 -544 tough BTDA/3A" 4 220 tough "Of precursor polyamic acid. bFrom DSC trace. 'From TG curve. d5 mol% excess of diamine. '5 mol% excess of diamine and solution imidisation in DPS. Table 7 Polymerisations using a variety of different acid dianhydrides via thermal imidisation ~~~~ ~~ no. of benzene excess of nature rings in acid solids diamine of RVh Tgcdiamine diamine dianh ydride endcapped" (%) (mol%) film /dl g-' 1°C 9A 6 PMDA 17.0 0 brittle -~ 239 9A 6 PMDA 9.2 5 brittle 1.89 237 9A 6 PMDA 13.2 4 brittle 1.8 213 9A 6 PMDA 9.9 5 creasable 0.5 227 9A 6 BTDA 15.1 5 brittle 0.7 199 9A 6 ODPA 15.2 5 creasa ble 0.4 220 9A 6 DPSDA 15.9 5 brittle 0.8 187 9A 6 BPDA 14.7 5 brittle 0.7 209 10A 6 PMDA 12.4 3 brittle 1OA 6 BTDA 12.8 5 brittle 0.7 227 1 OA 6 BPDA 12.0 5 brittle 0.8 233 1 OA 6 ODPA 11.0 5 brittle 0.6 220 15A 8 PMDA 10.0 3 creasable 2.1 210 15A 8 BTDA 12.3 3 brittle 3.1 213 15A 8 ODPA 10.0 5 brittle 0.8 190 15A 8 DPSDA 12.9 5 brittle 0.5 206 15A 8 BPDA 8.1 5 brittle -205 16A 8 PMDA 18.5 5 creasable 0.2 180 16A 8 BTDA 10.9 5 brittle 0.2 184 16A 8 ODPA 8.1 3 creasa ble -178 16A 8 BPDA 10.8 2 brittle 0.5 185 Using PA.'Of precursor polyamic acid. 'From DSC trace. J. MATER. CHEM., 1994, VOL. 4 Table 8 Polymerisations using a variety of different acid dianhydrides via solution imidisation in DPS" acid excess of nature diamine" dianhydride solids (YO) diamine (mol%) of film 'lJT 14A PMDA 15.5 creasable 250 14A BTDA 22.3 creasable (76 360) 223 14A ODPA 16.6 creasable 214 14A BPDA 14.8 creasable 226 15A PMDA 14.2 brittle 213 16A PMDA 22.3 creasable 189 16A PMDA - creasable - "PA endcapper used throughout. 'Diamines with eight benzene rings in the molecule. 'From DSC trace. carbonyl group in the immediate para position were of lower nucleophilicity. Amines with the carbonyl group in the meta position were more active, resulting in more rapid conden- sation and much higher RV values for their polyamic acid solutions.To control the rate of polymer formation and hence restrict the RV it was sometimes necessary to use phthalic anhydride (PA) as an endcapper, or to use an excess of diamine to lower the degree of polymerisation (see tables). The variation of the molecular weight of the polyamic acid has been shown previously to be related to the pK, of the diamine.'*13.14Low pK, values lower the rate constant of the forward amic acid formation step reducing the RV. With 4,4'-substituted amines, for example lA, the value of the RV was 0.44 with PMDA. Switching to 3A, with lower electron- withdrawing effect upon the amine, led to a rapid gelation of the polymeric acid with an RV of 0.76 dl g-' with the same acid dianhydride.Although molecular weights could not be determined for the final polyimide, the use of an identical thermal imidisation procedure for every polymer meant that any additional mol- ecular weight changes during the imidisation15 were expected to be more or less identical for every polymer. Thermal Properties of Polyimides All the polyimides were analysed by DSC (Tables 5-8). Typically two characteristic traces were obtained. Most poly- mers displayed a Tg transition with no evidence of a melt endotherm. Where semicrystallinity did occur, characterised by such a feature in the DSC trace, the sample was cooled and reheated to show the Tg transition more clearly.This thermal recycling generally quenched out any ~rystallinity.~ To ensure that the data were not influenced by differences in molecular weight, i.e. RV values, samples of BTDA/lA poly- amic acid were removed from the polymerisation medium at regular intervals, and the RV was measured. The samples were each cast onto borosilicate glass sheets, thermally imidised and then analysed by DSC. The results are shown in Table 9. For samples with RV ~0.2dl g-' Tg varies between 258 and 268°C. The variation seems to arise from minor differences in the nature of the polymer and also includes experimental error. For all practical purposes, however, the Tp can be considered as constant for samples with RV>O.2 dl g-'.Below this viscosity, however, the resultant polyimide has a Table 9 TJRV study for BTDA/1A RV/dl g-' 0.17 24 1 - 0.20 266 - 0.25 268 - 0.34 259 376 0.58 258 376 significantly lower q.Interestingly, when the polyamic acid was allowed to polymerise to yield a solution with an RV 30.35 dl g-' crystallinity was generated in the pollyimide product, with a discrete T, at 376 "C (entries 4 and 5, ?able 9). Structure-Property Relationships for Polyimides Most of the polyimides were essentially amorphous and those that did show a melt endotherm in the DSC did not possess any close structural relationship. Attempts to induce crystal- linity in other samples by various annealing procedures in the DSC were unsuccessf~l.~ These difficulties can be attributed to the fact that in both poly(ary1 ether ketone)s and poly(ether keto imide)s the aromatic rings are not ~oplanar'~-'~ when connected by either ether or ketone linkages, having a tor- sional angle of ca.70°L7 with respect to each other. For the polyimides an additional torsional angle of 30"16 aribes with respect to the imide. Continual repetition of these distortions is believed to produce a twisting structure in the polymer chain. Compounds modelled by O'Mahoney et all8 suggest that S-shaped structures exist for m-aminophenyl ether whilst p-aminophenyl ether structures adopt a straighter. comp- lementary self-stacking model. Although crystalline melt temperatures were not fcrund for all polyimides it was possible to look at the effect of structural changes upon the alone and those polymers that did show a melt endotherm allowed a tentative analysis of the Beaman relationship between Tp and T,.Effect of Diamine Size Increasing the size of the diamine component in order to improve the processability of polyimides has been a major aim of both Hergenrother and co-workers and Bell et Indeed, the latter group did attempt a quantitative study. The results from the present work are clear cut for the ciiamines 3A, 6A, 9A, 12A and 15A which have 4,5,6,7 and 8 benzene rings, respectively, all joined in a para configuration, and with amino groups meta to the last carbonyl linkage in each case.Table 10 shows the Tg data for polyimides prepared with PMDA and BTDA. The decrease in Tg is clearly larger for the PMDA series than for the BTDA series. It hias been suggested that the incorporation of flexible connecting groups into the acid dianhydride component (e.g. BTDA replacing PMDA) has the effect of lowering Tg more than the presence of the same flexibilising segment in the diamine component. However, increasing the size of the diamine component undoubtedly lowers the effect of the incorporation of the flexibilising group into the acid dianhydride portion of the polymer backbone to such an extent that at a diamine size of ca. eight aromatic rings changing from PMDA to BTDA does not affect the resultant polyimide at all. However, the presence of the rigid imide group, irrespective of how this is diluted by ether and ketone flexibilising linkages, raises the J.MATER. CHEM., 1994, VOL. 4 Table 10 Effect of increasing the size of the diamine upon Tgsfor PMDA and BTDA polyimides q/"cno. of aromatic code Ar rings in diamine PMDA -266 246 239 234 210 Table 11 TJTm data for semicrystalline polyimides BTDA ref. 240 26 226 this work 22 1 this work 199 this work 213 this work 213 this work no. of benzene nt polymer code rings in diamine RV/dl g-' PMDA/lA PMDA/2A PMDA/7A PMDA/llA BTDA/lA BTDA/5 A BTDA/7A BTDA/llA 0.44 0.55 0.56 0.56 0.58 0.37 0.56 0.40 LARC-CPI - by ca. 50°C from the corresponding values for poly(ary1 ether ketone)s.Clearly this in itself is of technical importance. Effect of the Substitution Pattern in the Diamine Residue In order to quantify the effect of altering the substitution pattern upon the Tg/Tmrelationship, polymerisations were carried out on a variety of isomeric four-, five- and seven-ring diamines with PMDA and BTDA. Those polyimides that showed a melt endotherm are listed in Table 11. (Note that temperatures are now in K for the Beaman relationship.) Several conclusions can be drawn from the results. First, altering the amine substitution from 4,4' to 3,3' gave polymers which failed to show any melt endotherms. This would suggest that crystallinity has been lost, although earlier X-ray diffrac- tion studies22 on polymers from compounds similar to the four-ring diamines in this work suggest that 3,3' amine substi- tution can still yield ca.16% crystallinity for PMDA-based polyimides. Altering the substitution pattern of a single amine group on the other hand, e.g. in polyimide PMDA/2A, did allow a retention of some crystallinity and a lowering of the T, from 800 K for PMDA/lA to 647 K for PMDA/2A. The Tg of PMDA/lA was unfortunately not detected by DSC so that no Tg/Tmratio could be calculated. Nevertheless, polyim- ide PMDA/2A is very interesting: its Tgis ca. 50°C higher than that of LARC CPI and yet its melting point is only ca. 25 "C higher. This is reflected in the high Beaman ratio, 0.83. Polyimides BTDA/1 A and BTDA/7A also have excellent combinations of Tgand T,.While BTDA/7A has the same Tg as LARC CPI its melting point is 30°C lower, showing an excellent combination of rigidity and potential processability. In order to look at the effect of increasing the amine meta content of Tp alone it was decided to synthesize a series of copolyimides based upon the isomers lA, 2A and 3A. Mixtures of the diamines were polymerised with PMDA. The polyamic acid solutions were cast as films and imidised thermally. The results are shown in Table 12. Clearly increasing the meta T,/TmT,/K Tm/K (Beaman relationship) -aoo -54 1 647 0.83 523 722 0.72 496 694 0.72 53 1 649 0.82 504 677 0.74 494 593 0.83 484 632 0.77 495 623 0.79 Table 12 PMDA copolyimides from isomers IA, 2A and 3A ~~ ~ ~~ ~~ amine meta mole polymer code RV/dl 8-l PMDA/lA 0.44 PMDA/2A + 1A ( 1:2) 0.36 PMDA/2A + 1A (1 :1) 0.36 PMDA/lA +3A ( 1:1) 0.29 PMDA/2A+ 3A (1: 1) 0.35 PMDA/3A" 0.76 "4 mol% excess of diamine used.fraction 7J"C 0.00 -0.17 -~ 0.25 282 0.50 276 0.75 268 1.oo 266 content of the terminal amine ring leads to an almost linear decrease in Tg. The monomer feed ratio of the diamines is a guide to the copolymer composition ratio. It can be seen that the Tp for PMDA/lA+ 3A (1 :1) (i.e. a nominal meta mole fraction of 0.5) is 276°C whilst the Tg of PMDA/2A (same equivalent meta mole fraction) is only 268 "C(Table 5). Given that the 3A component is more reactive a copolyimide with a lower Tg than that for PMDA/2A might have been expected, assuming a higher uptake of the more reactive species.The discrepancy is difficult to explain, but may be due to subtle physical differences in the polymer backbones, between the evenly spread meta linkages in PMDA/2A and the more block-like and uneven spread of the meta linkages in the copolymer. A second potentially important change in the structural pattern which might influence the thermal properties is a change from 1,4 to 1,3 aromatic substitution in the non-terminal or internal groups of the diamine. Polymers BTDA/5A and BTDA/7A were prepared from diamines con- taining five aromatic groups and differing only in the pattern of substitution of the central ring. Both polyimides were semicrystalline and offered a good opportunity for comparison with two polymers reported in the literature4 prepared from J.MATER. CHEM., 1994, VOL. 4 isomeric diamines. The latter again differ only in the pattern of substitution of the central aromatic ring, but also having the ether and ketone linkages interchanged relative to 5A and 7A. Table 13 shows the relevant Tg and T, data. Both pairs of polymers show a distinct fall in T, (ca. 70°C) for only a small fall in (ca. 10°C) when the 1,3 central pattern of substitution replaces the 1,4 pattern. Overall BTDA/7A shows the best combination of Tg and T, with a Beaman ratio of 0.83. In this case therefore the positioning of the ether linkages ‘inside’ the ketonic ones within the diamine residue seems the optimum arrangement.The close correspondence of all the Tg values for these four polymers is also interesting. However, the literature shows that the interchange of ether and ketone groups as above does not necessarily lead to identical Tg s as is the case for poly(ary1 ether ketone)^.^^ Rao and Bijim01~~ synthesized a series of ether ketone containing diamines with four aromatic groups with the structures shown in Table 14. The Tg data for polyimides prepared with PMDA and BTDA are shown in Table 14, along with the data for poly-imides prepared in this work from the diamines 1A and 3A in which the ether and ketone linkages are interchanged relative to Rao’s molecules. Clearly in this situation switching the ether linkage ‘inside’ the ketonic ones leads to an increase in Tg irrespective of the pattern of substitution of the terminal rings in the amine residue.Effect of Structural Changes in Acid Dianhydride Residue The generally lower Tgassociated with BPDA versus PMDA polyimides has been described earlier. The role of flexibilising groups in the acid dianhydride residue has been discussed in the first in terms of a comparison with flexible linkages in the diamine residue and secondly in terms of interrupting the intermolecular interactions between polymer chains. The ‘electronic isolation’ of acid anhydride residues in BTDA, ODPA, BPDA, DPSDA etc. play an important role in reducing the electron affinity of the imide rings, and the effect on reducing T, does seem to be higher than that on G.This was also the case in the present work as seen from the data in Table 15. T,/T, shifts from 0.71 to 0.77 for poljimides prepared from 11A with PMDA and BTDA, respectively, and from 0.72 to 0.83 for polymers from 7A with these acid dianhydrides; i.e. in each pair of polymers T, is lowered more than T,. With polyimides prepared from the longer diamines 9A, 10A, 14A, 15A and 16A with a variety of acid dianhydrides (Tables 7 and 8) the changes in 5 were rather small and in particular showed no regular pattern which could be corre- lated with acid dianhydride flexibility. Changing the structure of the acid dianhydride residue has also been examined previously in terms of its effect cm melt stability, itself an additional important parameter in a techno- logically exploitable species.” In the latter work none of the polymers containing PMDA or BTDA units had an acceptable melt stability, although the series containing 1,4,5,8-naphthalene imide did.All of these polymers were prepared via a Friedel-Crafts acylation route and it could be argued therefore that the different trace residues in these materials could be the source of the problem. Thermal Stability The stability of the polyimides deduced from TG analysis in air was expected to be high, but to decrease with increasing diamine size. Polymers based upon PMDA (Table 5) show a small initial decrease in thermal stability from 670°C for the highly rigid polymer based upon 4A to 663°C for LA and down as low as 518°C for the longer diamines residue containing five to seven aromatic rings.BTDA polymers Table 13 Effect of changing the substitution pattern of central aromatic ring in diamine component from 1,4 to 1,3 on the thermal properties of BTDA polyimides substitution pattern of central aroma tic diamine structure group T,/K Tm /K T,/% ref. 0 0 194 (54 504 677 } this 0.74 work 193 (74 494 593 0.83 NH2 NH2 174 700 0.72 4 \ NH2)?J0yJ+5JJJon 506 \ 0 0 NH2 1’3 495 623 0.79 4 Table 14 Effect of T,on interchanging ketone and ether linkages in the diamine component of BTDA and PMDA polyimides diamine structure amine subs ti tu tion T,rcPMDA T,rcBTDA rcf. 0 4,4’ 242 234 24 N H 2 - ~ 0 ~ 0 ~ N H 2 3,3‘ 218 213 24 4,4‘ (1A) -this NH2,, ‘0 &+NHz 3,3‘ (3A) 266 266 work& J.MATER. CHEM., 1994, VOL. 4 Table 15 Effect upon the TJT, relationship of polyimides prepared from different anhydrides PMDA BTDA diamine T,/K Tm/K T,/Tm T,/K Tm/K T,/Tm ref. 11A 496 694 0.71 484 632 0.77 this work 7A 523 722 0.72 494 593 0.83 this NH2/o-"aoa0 0 NH2 520 715 0.73 495 623 0.79 work 4 (Table 6) showed a smaller variation, being for the most part thermally stable up to ca. 560°C for polymers based upon 1 -3A. Again the longer diamines yielded BTDA polyimides of reduced stability. Overall, however, the polymers had the high stability expected. Experimental Materials N,N-Dimethylacetamide (DMAc), high-purity HPLC grade (Aldrich Chemical Co.), 99.9 +YO,was used as supplied; it was stored over molecular sieves.Diphenyl sulfone (DPS) (ICI) was recrystallised from hexane prior to use. All other solvents were used as supplied. Diamines 1-16A were synthesized as described in Part 3 of this work2* (see Table 1). Pyromellitic dianhydride (PMDA) (Aldrich) was sublimed (220 "C/lO mmHg) prior to use. 3,3'4,4'-Benzophenonetetracarboxylic acid dianhydride (BTDA) (Aldrich), high-purity sublimed grade was used as supplied. Oxydiphthalic anhydride (ODPA), biphenyl acid dianhydride (BPDA) and diphthalic acid sulfone dianhydride (DPSDA) (supplied by ICI Advanced Materials). Phthalic acid anhydride (PA) (Aldrich Chemical Co.) was used as supplied as an endcapper.Polyimide Syntheses Two methodologies were used. The first, referred to as 'thermal imidisation', involved preparation of a solution of polyamic acid from diamine and acid dianhydride in DMAc. The solvent was then removed and the polyamic acid imidised by heating the mixture under vacuum. The second method was a one-step procedure using DPS as a solvent, with water removed at high temperature. This will be referred to as 'solution imidisation'. Thermal Imidisation Typically the polyamic acid was formed initially at room temperature by charging a previously dried (220°C for 1h) three-necked round-bottomed flask with dried (under vacuum, 150 "C, 2 h) 1A (1.123 g, 2.75 x lop3mol). To this was added 50% of the DMAc solvent, (total volume of solvent 13.5 ml, 11.38% solids) in order to dissolve the diamine.The flask was then placed under a steady stream of nitrogen before the addition of the PMDA (0.600 g, 2.75 x mol) (dried as for diamine) and the remainder of the DMAc. The reaction solution was stirred with a magnetic stirrer overnight. The polymeric acid was then processed by one of two methods. 1. Thin-film casting: The polyamic acid was spread as a thin film onto a borosilicate glass plate. The film was then placed in a vacuum oven at 80°C overnight to evaporate the solvent before being imidised under vacuum, first at 150 "C for 2 h, then 200°C for a further 2 h and finally 2 h at 300°C to complete the process. 2. Polymer precipitation: The polyamic acid was precipi- tated out by adding the DMAc solution to deionised water.The solid was then macerated using a blender to produce course grains of polymer. These were collected by filtration and washed with deionised water (three times) to remove any remaining DMAc. Imidisation of the polyamic acid was then achieved identically to that of the cast film above. Solution Imidisation Typically 3A (2.667 g, 6.52~ mol, 5 mol% excess) (vacuum dried, 15OoC, 2 h), DPS C10.35 g, 30% solids and toluene (30 ml)] were loaded into the reaction vessel under nitrogen. The mixture was then heated to 110°C in order to melt the DPS and form a slurry with the diamine. At this point BTDA (2.001 g, 6.21 x mol) was added along with a further addition of toluene (10 ml) and the reaction mixture was then refluxed, with overhead stirring, at the boiling point of toluene (110°C) for 1.5 h before PA (0.138 g, 9.33 x lop4 mol) endcapper was added to the mixture.A Dean-Stark apparatus was then fitted and the polymerisation coproduct water was azeotroped off under reflux for a further 2 h before the remaining toluene was removed by distillation. The reac- tion temperature was finally raised to 170-80°C for 1 h in order to complete imidisation. At this point it was hoped that the hot reaction mixture would be a slurry; however, in all cases a gel was formed around the stirrer, yielding a hot rubbery solid which solidified on cooling. The solid was then pulverised into a fine powder and washed with acetone (three times) in order to remove residual DPS.Polymer Analysis Elemental microanalysis results are in Tables 3 and 4.Reduced viscosities (RV) of polyamic acid solutions in DMAc (1 wt.%) were measured using Ostwald Frenske viscometers (BDH BSU size B, and Fison's Scientific BSU size A) at 25°C. Solutions were pre-filtered through a porosity grade 1 glass sinter. Tp and T, were reduced from DSC traces. Analyses were carried out under nitrogen on a Du Pont 910 calorimeter controlled by a Du Pont 9900 thermal analyser. Polymer samples were heated on a dual thermal cycle. The first cycle heated the sample from 50 to 450°C at 20°C min-'. The polymer was then quenched to 100°C and the cycle was then repeated.Glass transitions were measured from the re-heat cycle. Crystalline melt transitions, where they occurred, were seen only on initial cycles and thus these values were used. J. MATER. CHEM., 1994, VOL. 4 The thermal stability of the polymers was determined by thermal gravimetry (TG). Several samples were analysed in air on a Stanton Redcroft STA 1500 instrument. All samples were heated at a constant rate of 10°C min-l up to 700°C with thermal stability being given as the temperature at which 5% weight loss occurs. The remaining samples were analysed (at Strathclyde University) on a Stanton Redcroft STA 750/770 instrument under air. Thermal stability was quoted from these results as the temperature at which 10% weight loss occurs.Heating was carried out at a rate of 10°C min-’ to 800°C. A crude estimate of the toughness of each sample was made by simply creasing the film samples by hand, and recording the behaviour as ‘brittle’ or ‘creasable’. The authors thank ICI plc for supporting this work; A.J.L. also thanks the SERC for a CASE studentship; P.L.P. thanks the Leverhulme Trust for the award of an Emeritus Fellowship which enabled him to participate in this work; D.C.S.acknowl-edges receipt of a visiting professorship at Tokyo Institute of Technology funded by Monbusho which allowed completion of this manuscript. References M. I. Bessonov, M. M. Koton, V. V. Kudryavtsev and L. A. Laius, Polyimides-Thermally Stable Polymers, ed. W. W. Wright, Plenum Press, New York, 1987.P. A. Staniland, Comprehensive Polymer Science, ed. G. Allen, J. C. Bevington, G. C. Eastmond, A. Ledwith, S. Russo and P. Sigwalt, Pergamon Press, London, 1989, vol. 5, p. 483. S. J. Havens and P. M. Hergenrother, US Pat. 4 820 791 (to NASA), 1989. P. M. Hergenrother, N. T. Wakelyn and S. J. Havens, J. Polym. Sci. A: Polym. Chem., 1987,25, 1093. P. M. Hergenrother and S. J. Havens, J. Polym. Sci. A: Polym. Chem., 1989,27, 1161. 6 P. M. Hergenrother, M. W. Beltz and S. J. Havens, J. Polym. Sci. A: Polym. Chem., 1991,29, 1483. 7 S. J. Havens and P. M. Hergenrother, J. Polym. Sci. A. Polym. Chem., 1992,30, 1209. 8 P. J. Horner and R. H. Whiteley, J. Muter. Chem., 1991, 1 271. 9 C. J. Borrill and R. H. Whiteley, J. Muter.Chem., 1991, 1, 655. 10 C. J. Borrill and R. H. Whiteley, J. Muter. Chem., 1992, 2, 997. 11 R. G. Beaman, J. Polym. Sci., 1952,9,470. 12 P. J. Brown, I. Karacan, J. Liu, J. E. McIntyre, A. H. Milburn and J. G. Tomka, Polym. Int., 1991,24,23. 13 V. M. Svetlichnyi, V. V. Krudyavtsev, N. A. Adrcva and M. M. Koton, J. Org. Chem. USSR, 1974,10,1907. 14 M. M. Koton, V. V. Krudyavtsev, N. A. Adrova, K. K. Kalnin’sh, A. M. Dubrova and V. M. Svetlichnyi, Vys. Soedin. Ser. A, 1974, 16, 2081; Polym. Sci. USSR, 1974,16,2411. 15 P. R. Young, J. R. J. Davies, A. C. Chang and J. N. Richardson, J. Polym. Sci. A: Polym. Chem., 1990,28,3107. 16 J. P. LaFemina, G. Arjavalingam and G. Haugham, I. Chem. Phps., 1989,90, 5 154. 17 S. A. Kafafi, J. P. Lafemina and J. L. Nauss, J. Am. Chzm. Soc., 1990,112,8742. 18 C. A. O’Mahoney, D. J. Williams, H. M. Colquhoun, K. Mayo, S. M. Young, A. Askari, J. Kendrick and E. Robson, Macromolecules, 1991,24,6527. 19 J. Kendrick and M. Fox, J. Mol. Graphics, 1991,9, 182. 20 V. L. Bell, W. L. Stump and H. Gager, J. Polym. Sci.: Poljm. Chem. Ed., 1976, 14, 2275. 21 V. L. Bell, L. Kilzer, E. M. Hett and G. M. Stokes, J. Apl’l. Polym. Sci.,1981,26, 3805. 22 C.E. Scroog, Macromolecular Synthesis, ed. N. G. Gaylo1 d, Wiley, New York, 1968,vol. 3, p. 83. 23 T. E. Attwood, P. C. Dawson, J. L. Freeman, L. R. J. Hoy, J. B. Rose and P. A. Staniland, Polymer, 1981,22, 1096. 24 V. L. Rao and J. Bijimol, Makromol. Chem., 1991, 192, l(125. 25 J. R. Pratt, D. A. Blackwell, T. L. St. Clair and N. L. Allphin, Polym. Prepr., 1988,29, 128. 26 C. R. Gantreauz, J. R. Pratt and T. L. St. Clair, J. Polyin. Sci. B: Polym. Phys., 1992,30, 71. 27 M. Fryd, Polyimides: Synthesis, Characterisation and Apillications, ed. K. L. Mittal, Plenum Press, New York, 1984,vol. I, p. 377. 28 Part 3: J. Muter. Chem., 1994, 4, 1527. Paper 4/02562H; Received 29th April, 1994.