J. MATER. CHEM., 1994, 4(3), 379-384 379 Preparation and Characterization of Imidazole-Metal Complexes and Evaluation of Cured Epoxy Networks John M. Barton: Gabriel J. Buist,b Ian Hamerfon,** Brendan J. Howlin,b John R. Jones* and Shuyuan Lid a Materials and Structures Department, Defence Research Agency (Aerospace Division) RAE, Farnborough, Hampshire, UK GU14 6TD b Department of Chemistry, University of Surrey, Guildford, Surrey, UK GU2 5XH A series of copper complexes of epoxy-imidazole adducts have been prepared and characterized by 'H nuclear magnetic resonance (NMR) spectroscopy. Differential scanning calorimetry (DSC) was employed to investigate the thermal behaviour of the curing agents and to investigate the medium-term storage stability of a one-pot composition of a commercial epoxy resin when mixed with the complexes. The cure onset temperatures of the mixtures containing copper complexes are ca.2040°C higher than those of the parent epoxy-imidazole adducts and the decrease of cure onset temperatures in the early stages of storage (up to 100 h) is less. The latent nature and improved storage stability of mixtures containing the copper complex were clearly demonstrated and confirmed by the viscosity bahaviour of the catalysed mixtures of the commercial epoxy resins MY720 and MY750. 'H NMR and electron paramagnetic resonance (EPR) spectroscopy were employed to monitor the thermal decomposition of the copper(i1) complexes, which were found to decompose at 120-1 30 "C and exist in equilibrium. Glass fibre-reinforced composite samples were prepared using a commercial epoxy resin cured with the complexes and their physico-mechanical properties were evaluated.Owing to their extreme versatility, epoxy resins are used extensively in industrial applications in which they are required to cure quickly and be readily formulated as one- pot compositions (i.e. the epoxide and catalyst are stored as a mixture rather than as two separate materials that have to be mixed prior to use). This in turn means that the composition must have a stability of at least several months at ambient temperature. Imidazoles are used as epoxy curing agents owing to their fast catalytic action and also the fine mechanical properties which they produce in the cured resin.Some imidazoles are highly effective epoxy curing agents,' and recent have demonstrated that epoxy resins cured with imidazoles can have superior physical properties (e.g. better heat resistance, lower tensile elongation, a higher modu- lus and a wider range of cure temperatures) than amine-cured resulting in their wide usage in the electronics industry as moulding and sealing compounds. Imidazoles are added to epoxy systems to catalyse the homopolymerization of epoxide groups (polyetherification), but unmodified imidaz- oles have low stability when mixed with epoxies (curing occurs slowly at room temperature) making them unsuitable for use in one-pot compositions. Much work has been carried out on stabilizing imidazoles for use as latent epoxy curing agents and one approach involves the preparation of metal-imidazole complexes.435 Transition metals have been used to prepare such complexes and these have exhibited good stability at room temperature and a rapid cure at elevated temperatures.Most metal-imid- azole complexes are crystalline materials with very low solubility in common epoxides.6 Solubility of the curing agent in the epoxide is very desirable because heterogenous dispersions are liable to settle out or agglomerate during storage. It is also useful to be able to form a solution containing both the epoxide and curing agent for the manufacture of pre-impregnated fibre composite materials (prepregs). Barton found6 that both the 1: 1 (1) and 2: 1 (2) adducts of phenyl glycidyl ether (PGE) and 2-ethyl-4-methylimidazole (EMI) could be cornplexed with a variety of metal salts and, in most cases, these complexes were soluble in epoxides and organic solvents.MoreoICer, the complexes were relatively unreactive at room temperature, but effective curing agents at higher temperatures. The structures of these copper(I1) complexes (denoted lc and 2c accordingly) are shown in Fig. 1. Poncipe prepared7 a range of epoxy-imidazole adducts, then complexed these adducts with metal salts and studied the co-reaction using UV spectroscopy. He found that there was a temperature-dependent induction period to the first- order reaction and that the nature of the metal ion was important in determining the length of this induction period (and hence the stability of the complex).The order Cu" >Ni" >Co" was observed for M( 1: 1),(N03l2 and M(1:1)4Cl, (M =transition metal) at a range of temperatures (which was in agreement with the order of stability predicted from the spectrochemical series).' The 1:1 adduct (I), like other imidazaoles,' coordinates to the metal through the pyridine-type tertiary nitrogen and it is the lone pair on this nitrogen which also attacks the epoxide. Complexation pre- vents the occurrence of the reaction, but when removed from the complex (lc) the 1:1 adduct (1) will react (Fig. 2). It would appear that at a given temperature, a particular com- plex has a definite lifetime during which it is stable and no reaction occurs; after this period the complex begins to break down and reaction is then able to occur.Increasxng the temperature decreases the stability of the complex and, hence, decreases the length of the induction period, until a tempera- ture is reached at which breakdown of the complex is almost instantaneous and no induction period is seen. The choice of copper as the most stable metal complex was prompted by the results of Poncipe's work. This study investigates the thermal properties of the adducts and complex mixtures in the presence of commercial epoxy systems, and in particular their storage stability. In this way it is hoped to demonstrate the usefulness of the novel cdtalysts in applications where a one-pot epoxy formulation is desirable.Aspects of the kinetic parameters associated with curing of the epoxy monomers with these catalysts and the use of high-temperature NMR as a means of monitoring currng are outlined elsewhere.lO,ll l4 J. MATER. CHEM., 1994, VOL. 4 r C CUCI, bH3aIL lc r C 1 OH i CuCI, 0-ih CH21 2c Fig. 1 Structures and ‘H NMR designations for the copper(r1) complexes lc and 2c dCH3[“;-CH2-CH-CHz-Ni; ?H : CUcIz YH2 CH3 4 lc 1 dCH3p, b-N~N:-cH,-cH-adduct formation 1 +epoxy3or4 ring-opened intermediate 5 0-1 FH-yH2dCH3 ?yyp, ACH3 ____t-NyN+-CH, -CH-CHZ-CH-k2 -N~N+-CH, -CH-polyetherification5 +epoxy3or4 and network formation Fig. 2 Structures of the compounds employed in this study and the proposed mechanism of curing J.MATER. CHEM., 1994, VOL. 4 Experimental Sample Preparation Preparation ofthe 1:1 Adduct (1) To a stirred, refluxing solution of EM1 (5.5 g, 0.05 mol) in toluene (50 ml) was added, during the course of 1 h, a solution of PGE (7.5 g, 0.1 mol) in toluene (25 ml). The mixture was refluxed for a further 2 h and then decolorized using charcoal. The product was precipitated and washed with several por- tions of 40-60 light petroleum, and dried in a vacuum oven at 40 "C to yield 10.2 g (78%) of a dark yellow liquid. The adduct was purified by column chromatography using a silica stationary phase and 4 : 1 chloroform-methanol as the eluent (analysis by TLC using the same eluent revealed a single spot, Rf=0.6, at 254 nm).After elution, the solvent was removed under vacuum and the product dried in uucuo at 40°C. The final product was characterized by 'H NMR (proton desig- nations refer to Fig. 1). 6, (300 MHz, CDCl,, ppm from TMS) 1.21-1.26 (3H, t, J=7.5 Hz, Ha), 2.12-2.19 (3H, d, J=20 Hz, Hc), 2.64-2.66 (2H, q, J=3.6Hz, Hb), 3.89-4.09 (5H, c.m., He,f,g), 6.59 (IH, S, Hd), 6.66-7.01 (2H, c.m., Hh,j), 7.26-7.32 ( lH, d, J =8.4 Hz, Hi). The preparation was subsequently repeated to produce ca. 300 g of the desired product. Prepurntion of the 2: 1 Adduct (2) To a stirred, refluxing solution of EM1 (5.5 g, 0.05 mol) in toluene (50 ml) was added, during the course of 2 h, a solution of PGE (1.5 g, 0.1 mol) in toluene (20 ml).The mixture was refluxed for a further hour, decolorized using charcoal and allowed to cool to room temperature. The product was precipitated and washed with several portions of 40-60 light petroleum, and dried in a vacuum oven at 40°C to yield 16.1 g (78%) of a dark yellow liquid. The adduct was purified by column chromatography using a silica stationary phase and 4 :1 chloroform-methanol as the eluent (analysis by TLC using 4.5 :1 chloroform-methanol revealed a single spot, R,= 0.85, at 254 nm). After elution, the solvent was removed under vacuum and the product dried in uucuo at 40°C. The final product was characterized by 'H NMR (proton designations refer to Fig. 1). 6, (300 MHz, CDCl,, ppm from TMS) 1.20-1.28 (3H, t, J=7.7 Hz, Ha), 2.12-2.18 (3H, d, J=17.9 Hz, H,), 2.63-2.67 (2H, q, J=5.5 Hz, Hb), 3.45-4.17 (5H, c.m., He,f,g),6.58 (IH, S, Hd), 6.66-6.98 (2H, c.m., Hh,j), 7.24-7.32 (lH, d, J =8.6 Hz, Hi).The preparation was subsequently repeated to produce ca. 200 g of the desired product. Preparation of the Metal Complexes of the EMI-PGE Adducts (lc and 2c) The metal complexes were all prepared using the same basic method. To a solution of CuC12.2H20 (0.85 g, 0.005 mol) was added a solution of the 1:1 adduct of PGE and EM1 (4.65 g, 0.04 mol) in absolute ethanol (10 ml). The mixture was heated gently, with stirring, for ca. 1 h. The solution was gravity filtered and the volume of the filtrate reduced on a rotary evaporator. The complex was precipitated from solution by the addition of diethyl ether and then washed thoroughly with further portions of the same.The complexes were dried in uucuo at 40°C to yield 3.87 g (70%) of the product as a brittle green glass. Apparatus 'H NMR spectra were obtained in CDC1, and C2H,]DMS0 at a range of temperatures using a Bruker AC-300 NMR spectrometer operating at 300.15 MHz. I5N NMR spectra were obtained in acetone at 25°C on the same instrument, but at 30.4 MHz. EPR spectra were obtained with DMSO as solvent at 25-90°C using a JEOL REIX EPR spectrometer operating at X-band frequencies. DSC was performed using a Du Pont 910 calorimeter interfaced with a Du Pont 9900 computer/thermal analyser. Samples of 8 f1 mg were accurately weighed into open, uncoated aluminium DSC pans.Routine DSC scans at 10 K min-' were performed from 30 to 350°C to observe the thermal properties of each of the blends. Viscometric measurements were made on the commercial epoxy-curing agent mixtures using a Brookfield viscometer operating at a range of temperatures and at a fixed shear rate of 64 Hz. Results and Discussion 'H NMR, 15NNMR and EPR Spectroscopy The 'H NMR spectra of the complexes exhibit marked changes over those of the parent adducts. In all cases, the complexation with the copper(I1) salt produced a broadening of the signals as a result of the paramagnetic effect of copper(I1). Upon coordination, the formation of a dative covalent bond between the lone pair on the pyridinyl nitrogen atom in 1 and the d2sp3 hybrid orbital on the copper atom results in a deshielding effect as the electron density is drawn away from the ring.The net result of this effect is to shift the imidazole protons (see Fig. 1) downfield. Fig. 3 shows the 'H NMR spectra for the 1 overlaid with those of the correspond- ing copper(I1) complex (lc). I5N NMR spectra were obtained for the parent imidazole (EMI) and 1. In the case of EMI, two peaks were observed at -165.1 and -176.5 ppm. Four peaks were observed for 1 (Fig. 4), indicating that adduct formation had occurred at either nitrogen to give two 1:1 adduct isomers (although presumably the steric hindrance afforded by the methyl group makes the adjacent site less accessible). The thermal stability of the complexes was studied in DMSO by recording the 'H NMR spectrum and progressively raising the temperature of the experiment.Initially the spec- trum obtained at ambient temperature (Fig. 3) displayed poor resolution, but the signals of interest at 6.6, 2.64-2.66 and 2.12-2.19 ppm (corresponding to Hd, Hb and H, in Fig. 1) rI 1 I 10.0 8.0 6.0 4.0 2.0 0.0 6 Fig. 3 300.15 MHz 'H NMR spectra (in CDCl, at 25 "C) of the (a) 1 and (b) lc J. MATER. CHEM., 1994, VOL. 4 Fig. 4 30.4 MHz 15NNMR spectrum (in acetone at 25 "C) of 1 can be seen clearly; the signal resolution improved with increasing temperature. At 120-130 "C lc appeared to undergo decomposition with the appearance of signals at 6.45-6.56, 2.64 and 2.08-2.16 ppm, corresponding to Hd, H, and H, (these changes were accompanied by a change in the colour of the DMSO solution of the complex from emerald green to brown).A parallel study carried out on these samples using EPR analysis at 25 and 90 "C indicated that the copper in the dissociated complex still existed in the Cu2+ oxidation state (rather than having undergone oxidation). When it had been kept at ambient temperature for 11days, the same sample was again scanned at 25 "Cusing 'H NMR. While the signals displayed the same poor resolution as before, the reappearance of peaks at 2.38 and 1.28 ppm (corresponding to H, and Ha) suggests that the complex had reformed to some extent (appearing to exist as an equilibrium mixture). Unfortunately, it was not possible to isolate pure crystals of either metal complex to allow a crystal determination using single-crystal X-ray diffraction techniques.As a result, the exact geometry of the metal complexes remains unsolved, although earlier work by Poncipe7 using diffuse reflectance spectroscopy on the complexes as powders indicated that they exist in octahedral form. DSC Study of Ambient-temperature Storage Stability In order to ascertain the utility of the complexes as one-pot epoxy compositions, it was of paramount importance to measure the medium-term storage stability at ambient tem- perature of, e.g., 3 when mixed with the complexes. In order to do this, DSC was employed to determine the glass-transition temperature of the uncured material and the onset of the thermal cure exotherm after progressive ambient-temperature storage.In order to achieve this the samples were stored at ambient temperature underneath inverted Petri dishes and at intervals samples were withdrawn and scanned from -70 to 280°C at 10K min-' under nitrogen (40ml min-l). The results were obtained in the form of plots of exothermic heat flow against temperature. From the plots, the glass-transition temperatures before the cure were observed as the onset of a characteristic endothermic transition (Fig. 5), and the onset of the thermal curing reactions were observed as the start of the exothermic peak due to the cure (Fig. 5). The change in Tgwith time is a useful measure of storage stability; an increase in Tgindicates early reaction leading to solidification of the stored mixture, when Tgreaches ambient temperature. After a period of ca.50 h the adduct displayed a marked increase in Tg(of ca. 20-40 "C) over the correspond- ing complex. The onset temperature for the adduct mixture decreased at a faster rate and to a greater degree in the early stages of storage, up to 100h, than the copper complex mixture. Furthermore, the cure onset temperatures of the copper complex mixtures are raised by ca. 20-50 "C, indicating -20 9 .\ -0 5 e-c . /. --20 y" --40 0 200 400 600 800 1000 storage time/h Fig. 5 Glass-transition temperature (-) and cure onset temperature (---) us. storage time for 1 (0)and lc (a) the latent nature and improved storage stability of the copper complex over the parent imidazole.Viscometric Study of the Effect of Adducts and Complexes as Catalysts The viscosity behaviour of the catalysed mixtures of the commercial epoxy resins 3 and 4 is depicted in Fig. 6, in which the latent nature of the complexes over the parent imidazole adducts is clearly demonstrated. Mixing 4 with 5 wt% of 1 led to an increase in viscosity of the resin mixture (which is initially ca. 5 CPat 100OC) after a period of ca. 5-10 min. In contrast, the corresponding copper(I1) complex (lc) causes the same increase in viscosity to occur after ca. 50min. The same commercial resin (4) exhibits similar vis- cosity profiles when catalysed with the 2: 1 adduct (2) and corresponding complex (2c), although these curing agents consistently display a lower reactivity.Similar behaviour is also observed (Fig. 6) for the second commercial epoxy system under study (3), although the reduced reactivity of this system required a higher temperature to effect curing within a reasonable timescale. Thermal Polymerization Properties of the Catalysed Epoxy Systems Prior to this study, a number of workers had published thermal investigations of imidazole-curing of epoxy resins. Heise and Martin carried o~t'~,'~ a DSC study of the behav- t i Fig. 6 Viscosity behaviour (fixed shear rate of 64 Hz) at 100"C for MY720 catalysed with 1 (O),lc (H),2 (A) and 2c (A):and at 100"C for MY750 catalysed with 1 (0),lc (@), 2 (0) and 2c ( +) J.MATER. CHEM., 1994, VOL. 4 383 Table 1 Thermal properties of catalysed epoxy resin systems from DSC measurements - 0.4W g-' curing agent T,/"C T'/"C T,/"C T3/OC TJC AHo/J g-' L i ' \ MY750 1 130 153 241 301 - 357 lc 160 197 258 307 - 428 2 150 176 246 319 - 375 2c 160 178 250 310 - 355 2c* 147 166 244 280 - 294 MY720 1 80 116 158 277 - 634 lc 130 163 231 277 - 612 2 85 115 163 239 285 548 2c 125 160 183 222 280 564 All measurements made on mixtures containing 5 wt% curing agent (except 2c* containing 7.5 wt%) using DSC at a heating rate of 10 K min-' under nitrogen. To, Onset temperature of polyn terization enthalpy; T,, nth peak maximum (i.e.TI =lowest-temperature exother- mic peak); AHo, overall polymerization enthalpy (J g-' of epoxy monomer).4-50 160 lk0 260 2iO 300 $50 TI"C Fig.7 DSC thermograms (scans at 10K min-' under nitrogen) for 3 catalysed with (a)1 and (b)lc iour of highly pure diglycidyl ether of Bisphenol A (BADGE), when catalysed with a variety of imidazoles at differing concentrations (0.5-100 mol% imidazole). They found that the thermal properties of the network were strongly dependent on the imidazole concentration. At high imidazole concen- trations (250 mol%), the adduct reaction consumed most of the epoxide groups to form adducts. As the imidazole concen- tration was decreased, Tg increased because fewer adducts were formed and more epoxide groups were available for chain growth and cross-linking.In their DSC study12,13 the OH-adduct reaction appeared at low imidazole concentrations as a shoulder on the main exotherm (and amounted to ca. 5% of the total heat of the curing reaction). In the present study, the commercial epoxy systems MY750 and MY720 were both catalysed with 5 wt% of each of the blends 1, 2, lc and 2c and a lOK rnin-' DSC scan was obtained for each of the blends during the polymerizations. Each of the thermograms are complex, often displaying several exothermic peaks, so rather than attempting to quantify each transition, the enthalpy for the entire polymerization range was calculated (Table 1). In all cases, the effect of catalysing the epoxy system with the copper(11) complexes (lc) and (2c) is to retard the onset of polymerization. As expected, the tetrafunctional monomer 4 was markedly more reactive than the difunctional 3 (undergoing an onset of polymerization ca.50°C lower in the case of the adduct, 1, and 30°C lower in the case of the complex, lc). From Table 1, the DSC results appear to be more complex than those found by Heise and Martir~'~,'~and Jisova.14 Consequently we decided to employ alternative methods of kinetic analysis (i.e. 'H NMR and FTIR spectroscopy) and these are dealt with elsewhere.lO,ll Conclusions Upon complexation of the PGE-EM1 adducts with CuCl, the 'H NMR spectra of the copper(i1) complexes exhibited broadened signals, while the imidazole protons were shifted downfield.The cure onset temperature of the copper(I1) complex mixture increases at a slower rate and to a lesser degree in the early stages of storage, up to 100 h, relative to the parent epoxy-imidazole mixture. Furthermore, the cure onset tem- peratures of the copper complex mixtures are raised by ca. 20-50 "C. After a period of ca. 50 h the adduct displays a marked increase in T, (of ca. 20-40 "C over the corresponding complex, indicating the latent nature and improved storage stability of the copper complex over the parent epoxy-imidazole adduct. The complex mixtures consistently display a greater degree of latency, and a faster rate of viscosity increase at gelation over the parent imidazoles when analysed for their dynamic viscosity in an epoxy mixture.The DSC thermograms of each of the catalytic mixtures (containing both adducts and complexes) were complex, often displaying several exothermic peaks. The effect of catalysing the epoxy system with the copper(I1) complexes lc and 2c was to retard the onset of polymerization. The tetrafunctional monomer 4 was markedly more reactive than the difunctional 3 (undergoing an onset of polymerization ca. 50°C lower in the case of 1 and 30°C lower in the case of the complex, lc. The work of Shuyuan Liu was generously supported by The Royal Society. The authors wish to thank the Materials and Structures Department, Defence Research Agency (Aerospace Division) RAE, Farnborough for the kind use of their thermal analysis facilities. The commercial epoxy monomers were kindly donated by Mr.Ian Gurnell and Mrs. Debbie Stone of Ciba-Geigy (UK) Duxford, Cambridgeshire, UK. We also thank Dr. Graham Webb and Professor Les Sutcliffe (Department of Chemistry, University of Surrey) for. helpful discussions concerning I5N NMR and EPR assignments. References W. R. Ashcroft, in Chemistry and Technology of Epoxy Resins, ed. B. Ellis, Blackie Academic and Professional, London, 1993, pp. 58, 59. M. Ito, H. Hata and K. Kamagata, J. AppI. Polyrn. Sci., 1987, 33, 1843. R. J. Jackson, A. M. Pigneri and E. C. Gaigoci, SAMPI' J., 1987, 23, 16. R. Dowbenko, W. H. Chang and C. C. Anderson, US. Pat. 3,677,978 (1972). R. Dowbenko, C. C. Anderson and W. H. Chang, Ind. Eiig. Chem. Prod. Res. Deu., 1971, 10, 344. J. M. Barton, Defence Research Agency (Aerospace Division) RAE, Farnborough, unpublished results. C. Ponciple, Ph.D. Thesis. University of Surrey, 1985. 384 J. MATER. CHEM., 1994, VOL. 4 8 H. Irving and R. J. P. Williams, J. Chem. Soc., 1953, 3192. 12 M. S. Heise and G. C. Martin, Macromolecules, 1989,22,99. 9 R. J. Sundberg and R. B. Martin, Chem. Rev., 1974,74,471. 13 M. S. Heise and G. C. Martin, J. Appf.Polpm. Sci., 1990,39, 721. 10 J. M. Barton, G. J. Buist, I. Hamerton, B. J. Howlin, J. R. Jones 14 V. Jisova, J. Appl. Pofym.Sci., 1987,34, 2547. and S. Liu, manuscript in preparation. 11 J. M. Barton, G. J. Buist, I. Hamerton, B. J. Howlin, J. R. Jones and S. Liu, Polym. Commun., submitted. Paper 3104294D; Received 21st July, 1993