Effect of complexation with copper(@ on cured neat resin properties of a commercial epoxy resin using modified imidazole curing agents? Ian Hamerton,"" Brendan J. Howlin," John R. Jones," Shuyuan Liu" and John M. Bartonb "Departmentof Chemistry, University of Surrey, Guildford, Surrey, UK GU2 5XH bStructural Materials Centre, Non-metallics, Defence Research Agency, Farnborough, Hampshire, UK GUl4 6TD A commercial epoxy prepolymer (MY750) was cured with novel modified imidazole curing agents under both isothermal and dynamic scanning conditions. The incorporation of an imidazole-copper (11) chloride complex curing agent to the epoxy prepolymer effected full cure after an isothermal cure schedule and post-cure treatment. The thermal stability of polymers arising from the isothermal cure schedule were generally higher than those for the corresponding dynamic cure.For samples cured via the dynamic curing process, a lower heating rate resulted in superior thermal stability. The same findings were obtained for samples cured via the isothermal curing process where the lower initial cure temperatures were optimal. The results of the present study show that the addition of metal atoms to the polymer systems does not have an adverse effect on either the water absorption or the dielectric properties of the final product. The cured resin displayed comparable thermal stability and absorbed the same amount of water at saturation (and a marginally lower amount at lower curing agent loadings) to a similar sample cured with an unmodified imidazole adduct.Epoxy resins are of immense technological importance as they form the continuous phase that binds together many light- weight, tough composite materials. There are many factors governing the physico-mechanical properties of the final resin. For example, it has long been recognized that the chemical nature of the curing agent can have significant influence on the gel-time and the physical properties of the epoxy, largely because it determines both the morphology and the crosslink density of the growing network. The thermal stability and water absorption characteristics of the epoxy, for instance, can be markedly affected by small changes in the curing agent. Some imidazoles are highly effective, fast curing agents' and are added to commercial epoxy systems to catalyse the homo- polymerization of epoxide groups (in a polyetherification mech- anism) to yield a thermoset network.However, unmodified imidazoles have low storage stability when mixed with epoxy resins (cure occurs slowly at room temperature) making them unsuitable for use in one-pot compositions and therefore, much work has been carried out into stabilizing imidazoles for use as latent epoxy curing agents. One approach involves the preparation of transition-metal-imidazole complexe~~*~ which exhibit very good solubility in common ep~xides,~ good stab- ility at room temperature4 and a rapid cure at elevated temperature^.^,^ The addition of metal atoms to the polymer systems should not, ideally, have an adverse effect on either the water absorption or the dielectric properties of the final product.In this study the incorporation of an imidazole-copper(I1) chloride complex curing agent to an epoxy prepoly- mer effected full cure after an isothermal cure schedule and post-cure treatment. The cured resin displays comparable thermal stability and absorbs the same amount of water at saturation (and a marginally lower amount at lower curing agent loadings) to a similar sample cured with an unmodified imidazole adduct. The dielectric responses observed for this epoxy resin system are typical for the isothermal cure of an epoxy resin and an additional benefit is that the presence of copper@) salts could act as an additional probe for ionic conductivity.The preparative routes to the manufacture of these curing agents are described elsewhere.6 +Presented at the Second International Conference on Materials Chemistry, MC2, University of Kent at Canterbury, 17-21 July 1995. Experimenta1 Materials The commercial epoxy prepolymer MY750 (the structure is nominally represented by 3 in Fig. 1) was donated by Ciba- Geigy (Duxford, UK). Phenyl glycidyl ether (PGE), 2-ethyl-4- methylimidazole (EMI) and copper(n) chloride were obtained from Aldrich Chemical Company. Purities were determined using 'H NMR (with the exception of CuC12, for which elemental analysis was used) and the compounds were used as received. Sample preparation The preparative routes to the 1:1 adduct of phenyl glycidyl ether and 2-ethyl-4-methylimidazole (PGE-EMI) 2 and the corresponding metal complex Cu( PGE-EMI)4C12 1 are described elsewhere.6 Compositions containing 5 and 6.64% (by mass) of adduct 2 and 5.6 and 7.5% (by mass) of complex 1(in molar equivalence to the adduct) in MY750 were made.The adducts were dispersed directly in the resin, while the complexes were dissolved in dichloromethane prior to mixing (the solvent being removed under vacuum at 30°C for 4 h). Neat resins were obtained by the following route. The adduct 2-epoxy formulations were cured initially at 100 "C and post- cured at 150°C until no further increase in was observed from DSC measurements. The complex l-epoxy mixtures were cured initially at 120 "C and post-cured at 165 "C, again until a final Tg had been obtained.Apparatus Thermogravimetry (TG) measurements were made both iso- thermally and dynamically at a variety of heating rates (1, 5, 10, 15 and 20 K min-') under nitrogen (25 cm3 min-') using a Shimadzu TGA-50 calorimeter interfaced with a Shimadzu TA-501 thermal analyser. Samples (6&1mg) were run in an open aluminium pan. Dielectric measurements were carried out on a Solatron 1250 frequency response analyser. Neat resin and epoxy-curing agent mixtures were placed in a cell consisting of two pre- etched copper electrodes of active area 1 cm2-mounted on a glass fibre reinforced epoxy resin base. The samples to be J. Mater. Chem., 1996, 6(3), 305-310 305 formation CH3 2 +epoxy 3 ring-opened intermediate 4 4 t epoxy 3 5 polyetherification and network formation Fig.1 Structures of the compounds employed in this study and the proposed mechanism of epoxy cure involving the metal-imidazole complexes measured were injected into the cell and placed in a cryostat (DN1704) The curing temperature was 100°C for the adduct 2 and 165 "C for the complex 1 The frequency range used was 10-1-105 Hz The data were obtained as plots of relative permittivity and dielectnc loss at the frequencies given above us time Dielectric measurements on cured resin samples were carned out on samples of size 12 x 17 x 1mm3 A curometer designed at Strathclyde7 was used to monitor changes in the viscosity with time at 2 Hz The instrument was calibrated using Santovac-5 which was chosen because it exhibited a very high temperature-viscosity coefficient and formed a stable, supercooled liquid state, which has been studied extensively Water absorption measurements were performed on cured neat resin plaques (20 x 10x 1mm3) immersed in boiling water (15 cm3, reflux) over a period of 14 h, while the sample mass was monitored Results and Discussion Dielectric measurements made during the curing process and on cured resins Measurements were made on samples containing the epoxy 3 and complex 1 (7 5 and 6 5 mass%, respectively) curing iso- thermally at 165 "C,and the epoxy 3 and adduct 2 (5 mass%) curing isothermally at 100"C, three-dimensional plots were obtained The dielectric responses observed for this epoxy resin 306 J Mater Chem, 1996, 6(3), 305-310 system (Fig 2 and 3) are typical for the isothermal cure of an epoxy resin A particular feature of the resin systems studied in this work is that the presence of copper(r1) salts could act as an additional probe for ionic conductivity and, as expected, the mixtures display a distinguishable ionic conductivity for which the dielectnc loss is greater than for a conventional imidazole-cured epoxy resin The initial value of the dielectric loss (log E") is approximately 9, which is greater than that observed for a typical bisphenol A diglycidyl ether (BADGE) system (log E"= 8) At low frequency and during the initial stages of cure a large dielectric loss is observed which rapidly decreases as cure proceeds and can be attributed to blocking electrode effects lo l2 This major ionic conductivity process is revealed by the rapid decrease in dielectric loss Generally, in this region the ionic conductivity is caused by inherent impurit- ies (which are present from the preparation of the epoxy monomer) and this phenomenon dominates the dielectnc behaviour owing to the lower viscosity of the polymer The curing process leads to a reduction in the dc conductivity as a function of time [Fig 2(u) and (b)] The initial high conduc- tivity of the mixture progressively decreases with time as the behaviour becomes dominated by dipole orientation It should also be remembered that the cure of epoxy resins using imidazole compounds proceeds uza an etherification mechan- ism (Fig 1)which produces a certain amount of dipolar species (arising from the cations on imidazole rings and anions on the other end of the growing macromolecule) during the whole !5O '300 i250 Fig. 2 (a) Relative permittivity and (b)dielectric loss measured as a function of frequency and time for a mixture of a blend of epoxy 3 and curing agent 1 (5.6 mass%) at 165 "C; (c) relative permittivity and (d) dielectric loss measured as a function of frequency and time for a mixture of a blend of epoxy 3 and curing agent 2 (5.0 mass%) at 100°C course of cure.It has been postulated" that the chemical reaction leading to the formation of a gel may significantly inhibit the mobility of charge carriers, and hence that the conductivity might be used as an indication of gelation.However, if the conduction process is influenced by the segmen- tal mobility of chains forming the three-dimensional matrix, then changes in conductivity may also be indicative of vitrifi-cation. Unfortunately, in this case the isothermal cure tempera- ture (T,)required to effect cure for the epoxy4omplex mixtures (T,= 165 "C) is greater than the ultimate glass transition tem- perature for this system (q,=140 OC)I3 and consequently it is impossible for the vitrification transition to be observed for the whole process. In contrast, vitrification can be reached in the 3-2 system (q=112"C)if a T, of 100°C is used [Fig.2(c) and (41.As a result it was necessary to employ a complimen- tary technique (curometry) to analyse the cure process to yield more information concerning the gelation and vitrification. The measurements on cured resins were made by determin- ing the dielectric response against frequency and temperature. The traces showed no observable influence of the metal salt in the matrix (Fig. 3), the dielectric response being dominated by an a-relaxation between -10 and 300 "C. This phenomenon may arise because the tiny amount of metal salt, < 1%, in the system is essentially at a lower level than the interferent ion impurities and they are effectively constrained within the rigid network up to 300°C. Rheology measurements during the cure process and on cured resins Although the dielectric technique (DETA) is useful to monitor the physical changes (sol-gel-vitrification), it is probably not the most effective method to monitor gelation during cure.Dynamic viscosity measurements offer an efficient means of determining the gel-point, but our previous measurements6 could not fully examine the cure of the polymer in an undis- turbed state, because the constant shearing of the spindle effectively orders the molecules in the system, lowers entropy, and accelerates the cure process. Owing to the high shear rates of the spindle at around the gel-point, the network structure is broken down at the spindle-resin and/or cylinder wall-resin interface, making it impossible to monitor the cure after gelation has been reached.The curometer, developed in Strathclyde,' can carry out measurements under conditions which involve very little disruption to the cure mixture and allow measurement over a greater range of conversion. These J. Muter. Chem., 1996, 6(3),305-310 307 3500 1 r loo Fig.3 (a) Relative permittivity and (b)dielectric loss measured as a function of frequency and temperature for a cured mixture of a blend of epoxy 3 and curing agent 1 (7.5 mass%) measurements may even continue after vitrification has occurred so that the whole cure process may be monitored. In turn, the Tg may also be determined by examining the changes in rheological parameters in one of two ways: at cure tempera- tures above the polymer's Tg the changes in viscosity are measured as the temperature is decreased, while at cure temperatures below its Tg a temperature ramp is employed.Plots of the real and imaginary responses, and the computed viscosities for the cure are presented in Fig. 4-7. The raw data (real and imaginary responses) are shown in Fig. 4, an example in which the cure temperature is above the Tg of the polymer. Fig. 5 shows the viscosity data derived from Fig. 4. The onset of gelation occurs at about 2.8 h and the rheology value reached a plateau at about 7000 Pa s (for the polymer in a rubbery state) after cure at 165 "C for about 17 h. When the analysis temperature was decreased the viscosity gradually increased until the mixture solidified at its Tg to a value of about 11500 Pa s (Fig.6). Analysis in this manner revealed a Tg of 143°C which is in good agreement with the value measured by DSC (141 "C).The corresponding adduct mixture 3-2 showed a different viscosity behaviour during the cure process in which sol-gel-glass transitions were involved. The 308 J. Mater. Chew., 1996,6(3), 305-310 3000 a 2500 &.--a5 2000 a3 1500 J= Q .E 1000 500 0 0 I0000 20000 30000 40000 60000 60000 tls Fig. 4 Real and imaginary responses from curometry for a mixture of a blend of epoxy 3 and curing agent 1 (5.6 mass%) at 165 "C 10000 3I000 v) (0a .-2. 100 v) 0 .-> 10 I 1 1 I I I I I 0 10000 20000 30000 40000 50000 60000 tls Fig.5 Viscosity derived from curometry data for a mixture of a blend of epoxy 3 and curing agent 1 (5.6 mass%) at 165 "C 1:ooo i f--10000 a a .-'c 9000 0 v).-> 8000 7000 165 160 155 150 145 140 135 TI"C Fig. 6 Glass transition temperature derived from curometry data on cooling a mixture of a blend of epoxy 3 and curing agent 1 (5.6 mass%) from 165 to 135°C gelation started after 48 min and the viscosity value reached a plateau at about 1x lo6 Pa s (for the polymer in a solid state) after cure at 100°C for about 4 h (Fig. 7). Although the Tg obtained ( 112 "C)is higher than the cure temperature ( 100 "C) it is not the final Tg, which would require further post-cure treatment to achieve. Thermal stability of the cured epoxy resins In our earlier study we demonstratedi3 the importance of both the initial and the final stages of the cure schedule upon the nature of the network formed, and hence the effect upon the final properties of the cured resin.In particular, the magnitude 10000000 1000000 100000 -2 10000 -f I /I I I I I i 0 5000 10000 15000 20000 tis Fig. 7 Viscosity derived from curometry data for a mixture of a blend of epoxy 3 and curing agent 2 (5.0 mass%) at 100"C of Tgwas extremely sensitive to the cure schedule employed to cure the complex-cured epoxy resins. With reference to Fig. 1, earlier mechanistic studies of imidazole cure of epoxy resins14 have shown that the alkoxide (RO-) propagation is considered to be favoured at lower temperatures while the hydroxy (OH) propagation route is only active at higher temperatures, per- haps leading to different network structures.However, purely from a consideration of the molecular structure, it is not obvious how the two paths (of RO-and OH polyetherification) should lead to different types of ether-linked network. However, there is a difference between these two paths if one considers the position of the alkoxide (RO-) and hydroxy (OH) groups. RO-groups are generally located at terminal positions in the intermediate, which is produced by the initial attack of an RO-group on an oxirane ring at the terminus of the molecule, while the OH group is initially produced towards the middle of the BADGE oligomer3 (Fig.1). The geometrical structures resulting from these two paths may be slightly different: the product from the RO-etherification may result in a more linear structure, while OH etherification may result in a more branch-like structure. The influence of these two kinds of intermediate structures on the formation and final structure of the network cannot be gauged easily. The magnitude of this proposed effect may be expected to be small since the concen- tration of hydroxy groups is rather low and all the procedures were carried out under moisture-free conditions in the absence of acid contaminants. However, one point which is clear from the outcome of the results of our investigation is that this is certainly not the sole factor influencing the final structure and properties. As both the degree of cure and the network architecture have an influence on both the Tg and the thermo- oxidative stability of the polymer, in this study we will limit our analysis to the properties of specimens that have reached their ultimate conversion and yielded 'final' Tp values (Tgf).Dynamic TG experiments were performed on cured resin samples in a nitrogen atmosphere to assess the thermal stability of the cured epoxy polymer. In each case the mixture was scanned at a heating rate of 10 K min-l from ambient tempera- ture to 500°C after the final q was measured by DSC. The TG data were collected in the form of plots of residual mass us. temperature and a representative plot is shown in Fig.8. This overlay plot contains TG data from complex-containing epoxy resins cured both isothermally and using dynamic cure schedules. While some discrepancies were observed in the order of stability, in general those samples cured using a slower scanning rate were found to be more stable on subsequent rescanning of the cures resin (Le., as expected the thermal stability of the cured resin is proportional to its q).All plots from polymer samples cured isothermally displayed a similar profile to the TG data from samples cured using the dynamic cure schedules, indicating no significant differences in v)3 E 0 100.00 200.00 300.00 400.00 500.00 TI"C Fig. 8 TG plots of cured mixtures of epoxy 3 and curing agent 1 (7.5 mass%) after a variety of cure schedules: (a) 120°C, 16 h and 160"C, 12 h; (b) 1 K min-', 300°C; (c) 5 K min-', 300°C; (d) 10K min-', 300°C; (e) 15 K min-', 300°C; (f)20 K min-', 300°C.TG measure- ments were made under nitrogen at a heating rate of 10 K min-'. Note, TG data are shown offset, all data were initially 100%. the mechanism of degradation. It can be seen that those samples cured using a scanning schedule were found to be less stable on subsequent rescanning of the cured resin than those polymers cured using a comparable isothermal programme. The overall trend is that, as expected, samples exhibiting a higher final Tg value display superior thermal stability for both dynamically and isothermally cured resins. The epoxy resin cured with the imidazole adduct undergoes the onset of degradation (defined by the intersection of tangents in the mass loss data) at ca.408 "C (for a Tgf of 128 "C). This sample also gave a char yield (the residual mass remaining at 500°C) of 85%. If we now examine the properties of the complex-cured resin which has been fully cured (by an optimised isothermal cure schedule: 16 h at 120"C followed by a postcure of 8 h at 160"C), both the thermal stability (onset temperature and char yield) and the qf (140°C) are comparable with the figures from the adduct cured system (Le., an onset of 402°C and a char yield of 80%). These TG data appear to support the suggestion from the DSC experiment^'^ that the significant difference in the curing schedules is favour- ing one polymerization mechanism over another.Water absorption of the cured epoxy resins It is well known that the absorption of water in a thermosetting resin leads to a decrease in the value of Tg,and the mechanical and dielectric properties of the cured material, and this is particularly true of epoxy resins where a cured resin may sacrifice as much as 20 K of Tg for every 1% of moisture absorbed." Hence, it is of crucial importance to prepare high- performance materials which exhibit low levels of moisture absorption in a range of environments. In this study it was of particular interest to ascertain the effect of residual copper@) in the cured resin on the degree of water absorption. The immersion data for the epoxy 3 cured with both curing agents 1 and 2 show some interesting features.In the case of the higher curing agent loading both 2 (6.64%) and 1 (7.5%) display similar profiles [Fig. 9(a)]; saturation is reached at ca. 2.5 mass% after a period of ca. 8 h in boiling water. Again, in the case of the lower curing agent loading, both 2 (5%) and 1 (5.6%) display similar profiles, but in this case saturation is reached much more rapidly [ca. 4 h in boiling water, Fig. 9(b)]. In both cases saturation is reached at a level of ca. 2.5 mass%, indicating that the residual metal has no apparent effect on the eventual water uptake of the cured epoxy resin. These data have important ramifications in the possible use of these materials in microelectronic applications where the effect of moisture upon the electrical behaviour of the cured resin is of J.Muter. Chern., 1996, 6(3), 305-310 309 3 25 2 15 1 05 h s v c Y50 c 02 4 6 8 10 12 148 .c.s3 $ 25 2 15 1 05 0 0 2 4 6 8 10 12 14 th Fig. 9 Plots of water absorption (mass%) for cured epoxy resin samples after immersion in boiling water (a)Curing agents 1 (7 50 mass%, @) and 2 (664 mass%, O),(b) cunng agents 1 (5 60 mass%, 0) and 2 (500 mass%, 0) Table 1 Copper and chlonde content from ICP-MS analysis of water samples from immersion tests (after 14 h at reflux) curing agent (mass%) 35C1 content (ppb) Tu content (ppb) 2 (6 64) 2 122 5 106 1 (7 50) 2 585 134 535 1 (5 60) 2 935 298 060 prime concern It was found from an ICP-MS analysis of the water samples in which the metal-containing samples had been boiled, that these samples showed notably high copper contents when compared with the adduct-containing mixtures (Table 1) This suggests that the action of water, albeit under extreme conditions, caused some of the metal salt (CuCl,) which had been incorporated into the polymer to be leached from the resin network (to a level of ca 300 ppb) Conclusions Transition-metal-imidazole complexes have been prepared which exhibit very good solubility in common epoxides, good stability at room temperature and which effect a rapid cure at elevated temperatures The results of the present study show that the addition of metal atoms to the polymer systems does not have an adverse effect on either the water absorption or the dielectric properties of the final product The incorporation of 65 and 7 5 mass% of an imidazole-copper(I1) chloride complex curing agent to an epoxy prepolymer effected full cure after an isothermal cure schedule and post-cure treatment The cured resin displayed comparable thermal stability and absorbed the same amount of water at saturation (and a marginally lower amount at lower curing agent loadings) to a similar sample cured with an unmodified imidazole adduct Work continues to examine the great potential of this exciting family of curing agents The work of S L was generously supported by The Structural Materials Centre, Non-metallics, Defence Research Agency, Farnborough The commercial epoxy prepolymer was kindly donated by Mr Ian Gurnell and Mrs Debbie Stone of Ciba-Geigy (UK) Duxford, Cambridgeshire, UK We thank Professor Richard Pethrick and Dr David Hayward (University of Strathclyde) for their help with the operation of the curometer and DETA and advice concerning data interpret- ation At the University of Surrey we thank Dr Fadi Abou- Shakra for ICP-MS analysis References 1 W R Ashcroft, Curing Agents for epoxy resins, in Chemistry and Technology of Epoxy Resins, ed B Ellis, Blackie, Glasgow, 1993, ch 2,pp 58-59 2 R Dowbenko, C C Anderson and W H Chang, Ind Eng Chem Prod Res Dev, 1971,10,344 3 J M Barton, GB Pat 2135316B, 1984 4 J M Barton, I Hamerton, B J Howlin, J R Jones and S Liu, Polym Bull, 1994,33,347 5 J M Barton, G J Buist, I Hamerton, B J Howlin, J R Jones and S Liu, Polym Bull, 1994,33,215 6 J M Barton, G J Buist, I Hamerton, B J Howlin, J R Jones and S Liu, J Muter Chem , 1994,4,379 7 D Hayward, E Trottier, A Collins, S Affrossman and R A Pethnck, J Oil Colour Chem Assoc , 1989,452 8 A J Barlow, A Evnngsav and J Lamb, Proc R Soc London A, 1969,309,473 9 J Cochrane and G Harmon, J Phys E, 1972,547 10 J R MacDonald, Phys Rev, 1953,92,4 11 J R MacDonald, J Electrochem SOC , 1988,135,2274 12 W W Bidstrup, N F Sheppard and S D Sentuna, Polym Sci Eng , 1986,26,358 13 I Hamerton, B J Howlin, J R Jones, S Liu and J M Barton, 1995, Polymer, submitted 14 M S Heise and G C Martin, Macromolecules, 1989,22,99 15 W W Wright, Composites, 1981, 12, 201 Paper 5/04831A, Received 21st July, 1995 310 J Muter Chem , 1996, 6(3), 305-310