Development of epoxy/hyperbranched blends for resin transfer molding and vacuum assisted resin transfer molding applications: effect of a reactive diluent.

INTRODUCTION

Epoxy resins are widely used due to their thermal, mechanical, chemical, and corrosion resistances. The low viscosity in the uncured state is a favorable property that enables processing without the use of high pressure equipment (1). Many fiber-reinforced composites are based on the use of epoxies as a matrix. Among the different composites, production techniques using liquid molding technologies are assuming a prominent role in different fields (2). Some authors (3) have stated that the viscosity of 1 Pa-s is the upper limit at which the resin is suitable for liquid molding techniques like RTM (Resin Transfer Molding) and VARTM (Vacuum Assisted Resin Transfer Molding).

The main drawback of epoxy systems is their inherent brittleness. Most of the techniques for improving toughness are based on the addition of modifiers of an elastomeric or thermoplastic nature (4). Among the toughening agents, reactive rubbers like liquid butadiene acrylonitrile (5) or preformed rubber particles (for example, core shell particles (6)) are widely used. These modifiers result in consistent improvements in terms of fracture resistance. However, their addition to epoxy systems presents many limitations, especially in terms of the reduction of the glass transition temperature and of the stiffness of the epoxy blends (7). Moreover, the addition of reactive liquid rubbers or preformed particles results in large increases in the viscosity of uncured blends (8), (9) that, consequently, cannot be easily used as matrices for composites manufactured with liquid molding techniques.

Another possibility for the toughening of epoxies is the use of high performance engineering thermoplastics, such as poly(ether imide)s (10), (11), polycarbonate (12-14), poly(phenylene oxide) (15), and poly(ether sulphone) (16). Interesting data on carboxyl terminated polyethylene glycol adipate (17) have been also reported. Engineering thermoplastics are favorable to rubbers because of their high glass transition temperature, high elastic modulus, better solvent resistance, and toughness. However, the use of thermoplastic results in large increases in the blend's viscosity, even if low molecular mass polymers are used (9), (18).

Some authors have proposed the use of hyperbranched polymers to overcome the limitations of traditional modifiers (19). Epoxy-functionalized hyperbranched polymers have proven to be feasible as modifiers of formulations that are processed by RTM techniques (20). The hydroxyl hyperbranched polymers have been shown to increase the viscosity of blends cured by diaminodiphenylsulfone up to values that allow for their use in processes like resin film infusion but preclude their use for RTM or VARTM (21). Similar results were obtained on epoxy blends cured by diethyltoluene diamine (22).

Diluents are used in epoxy-resin technology primarily to reduce the viscosity of the mixed resin. In addition to providing for viscosity reduction, the diluent may be selected to provide modification of the cured properties. The diluents of interest to epoxy-resin technology may be divided into three basic categories: nonreactive, epoxy-containing-reactive, and reactive containing functional groups other than epoxies (23).

The aim of this article is to develop novel formulations for liquid molding modified by hydroxyl-terminated hyperbranched polymers. The proposed strategy is based on the addition of trimethylolpropane triglycidyl ether (TMTG), which acts as reactive diluent for the epoxy blend. The formulations were prepared with different percentages of TMTG and thoroughly characterized in terms of rheological and thermal properties. To the best of our knowledge, no previous study exists concerning the use of TMTG as a reactive diluent for epoxy/hyperbranched blends.

The cure cycle was chosen for practical reasons. In fact, it is known that lower cure cycle temperatures are advantageous in terms of reduction of autoclave and tooling costs, so we decided to study a cure cycle divided into two steps: a precure at medium temperatures (135[degrees]C and 155[degrees]C) for the processes in an autoclave or in a closed matched mold; a curing at higher temperature (180[degrees]C), which can be carried out inexpensively in a standard oven.

After the precure step, it was important to verify the attainment of elastic properties that could allow the precured blend to be easily demolded. The elastic properties of both the precured and postcured samples were studied by dynamic mechanical analysis (DMA).

EXPERIMENTAL

Materials

The epoxy monomer was a diglycidyl ether of bisphenol A (DGEBA), Epon828 supplied by Shell. The curing agent was a mixture of the two diethyltoluene diamine (DETDA) isomers (74%-80% 2,4 isomer and 18%-24% 2,6 isomer) (LonzaCure DETDA80 supplied by LONZA). The reactive diluent selected for the study is TMTG purchased by Aldrich.

Perstorp Specialty Chemicals kindly provided the hyperbranched polyester, which is commercialized as Boltorn[TM] H40. The polymer selected is a fourth pseudo-generation number. The characteristics of this polymer are reported in Table 1, where the molecular mass is obtained from Zagar et al. (24). The denotations for the hyperbranched polymers in the text will be H40. The structures of all the reagents are reported in Fig. 1.

[FIGURE 1 OMITTED]

TABLE 1. Properties of the hyperbranched polymer Boltorn[TM] H40.

Code  Pseudo-generation   End     OH (a)     Mw (b)   [T.sub.g]
           number        group  (mg KOH/g)  (g/mole)   (a) (C)

H40       4                OH     470-500     5100        40

(a) Data from Perstorp datasheet.
(b) Data obtained from reference (24).

Preparation of the Blends

The hyperbranched polymer was mixed with the epoxy resin and the mixture was stirred and heated at 100[degrees]C for 15 min until a homogenous solution was observed. The curing agent was then added to the mixture and stirred for 15 min. The blend obtained was kept frozen until used for the rheological test.

To prepare the cured samples, the resin mixture was poured into a preheated silicone rubber mold and degassed for 15 min at 135[degrees]C. Some samples were initially precured for 3 h at 135[degrees]C, and the temperature was increased at 2[degrees]C/min up to 180[degrees]C and maintained at this level for 3 h. At the end of the curing cycle, the panels were left to cool slowly at room temperature. From this point, we define precured samples as samples that were cured at 135[degrees]C for 3 h and postcured samples as those in which, after precuring at 135[degrees]C, the samples were then cured at 180[degrees]C for 3 h. The formulations of the blends studied (Table 2) were all calculated with stoichiometric amounts of the curing agents.

TABLE 2. Formulations used in the study [wt%].

Component/code          100D       90D10T       80D20T       60D40T

(a) Without modifier

DGBBA                   80.88       72.50        64.18        47.71
TMTG                      0          8.06        16.04        31.81
DETDA                   19.12       19.44        19.78        20.48
H40                       0           0            0            0

Component/code        100D10H40  90D10T10H40  80D20T10H40  60D40T10H40

(b) With modifier
DGEBA                   72.79       65.25        57.76        42.94
TMTG                      0          7.25        14.44        28.63
DETDA                   17.21       17.50        17.80        18.43
H40                      10           10           10           10

Characterization of the Blends

Rheological analyses were performed on a rotational rheometer (ARES by TA instruments) equipped with parallel plates. The unreacted samples were subjected to a ramp test (at 2[degrees]C/min) experiment to determine the viscous behavior over a wide temperature spectrum. For ramp testing, parallel plates 40 mm in diameter were used. In addition, isothermal experiments were carried out at several temperatures to determine the gel point of the blends. An isothermal, multiple waveform dynamic test (also termed "multiwave" rheological measurement) was used to accurately determine the gel point. The test consists of performing simultaneous time sweeps at different frequencies, within the linear viscoelastic regime, for a given temperature. Actually, with this technique, the multiwave strain generation is based upon the Boltzmann superposition principle. The total linear strain, [gramma], applied on the sample is the sum of several independent strains, each described by its corresponding Fourier series:

y = [n summation over (i = 1)][[gamma].sub.i]sin([[omega].sub.i]t) (1)

[n summation over (i = 1)][[gamma].sub.i][less than or equal to][[gamma].sub.c] (2)

The frequencies chosen, [[omega].sub.i], are harmonics of a fundamental frequency [[omega].sub.f]. The strain relation (Eq. 1) expresses the requirement that the sum of the individual strain amplitudes stay below a critical amplitude, [[lambda].sub.c], corresponding to the linear viscoelastic limit (Eq. 2). From the stress response to this compound strain, the individual stresses at each discrete frequency can be obtained by means of a discrete Fourier transform. With a single experiment, then, this procedure allows the determination of the tan [delta]-versus-time curve at different frequencies for each temperature. The crossover of the tan [delta] curves represents the gel point, according to the Winter criterion. The parallel plates used for the multiwave test were 25 mm in diameter.

DMTA (Dynamic Mechanical Thermal Analysis) tests were conducted, in torsion mode, on cured specimens by a dynamical mechanical thermal analyzer, ARES by TA instruments, at a fixed frequency of 10 rad/s with a 2[degrees] C/ min heating rate. The samples had the following dimension: 45 mm X 10 mm X 3 mm.

SEM (Scanning Electron Microscopy) micrographs were obtained using a Cambridge 90 SEM. Samples were prepared by polishing them with alumina, and then they were etched with a mixture of sulphuric acid and distilled water (3:2). The acid mixture has the role of etching the hyperbranched phase and then increasing the contrast between the phases. After the etching, the samples were washed with water and gold-coated.

RESULTS AND DISCUSSIONS

To verify the behavior of the formulations over a wide range of temperatures, dynamic rheological measurements were first carried out at 2[degrees] C/min. The results for various systems are reported in Fig. 2. The unmodified resin (100D) shows that the minimum viscosity is reached at 120[degrees]C. After this minimum, the viscosity is starts to rise because of the crosslinking reaction. The addition of 10 wt% of H40 (100D10H40) causes and increase of the overall viscosity of the blend and a shift to lower temperatures of the minimum viscosity (~ 110[degrees]C). This result is a consequence of the molar mass and of the structural characteristics of the hyperbranched polymer. In particular, the increased reactivity, manifested by the shift of the minimum viscosity to lower temperatures, is associated with the catalyst effect of the hydroxyl groups of the hyperbranched modifier. In fact, it is known that hydroxyl groups can catalyzed the epoxy curing reaction on the basis of intermolecular transition states [34 articolo 1] that stabilize the transition state and encourage the nucleophilic attack of the amine. The increase in viscosity and the higher reactivity are effects that tend to decrease the processability of the system for liquid molding applications.

[FIGURE 2 OMITTED]

The addition of the reactive diluent TMTG causes a consistent decrease of the viscosity for the formulations modified with the hyperbranched polymer H40. Analogous results were obtained for unmodified blends based on mixtures of DGEBA and TMTG. The minimum viscosities occur at 116[degrees]C, 114[degrees]C, and 107[degrees]C for the formulations containing 10%, 20%, and 40% TMTG, respectively. The decrease of temperature at which the minimum viscosity occurs is a signature of the increased reactivity of the system for higher diluent contents. This effect is due to the higher functionality and to the aliphatic structure of TMTG. The addition of TMTG is thus beneficial in terms of viscosity reduction, but higher contents are not useful because of the increased reactivity of the system. The latter effect can have a negative influence on the time available for injection. Ideally, the system for liquid molding applications must have low viscosity for a long time in order to allow complete impregnation of the dry fabrics.

The gel time is a standard parameter used in practice to give an indication of the time available for processing at the reaction temperature of a resin. Gel time occurs when one of the growing molecules reaches a mass so large that it interconnects every boundary of the system (25). When a thermoset reaches its gel time, the viscosity becomes infinite, there is a buildup of the elastic modulus, and an insoluble fraction suddenly appears. A reliable method to determine the gel point is based on the Winter criterion (26). The method is based on the determination of the point at which tan [delta] is frequency independent. Several isothermal tests were carried out in multiwave mode to determine the gel points of the studied formulations at three curing temperatures: 130[degrees]C, 140[degrees]C, and 150[degrees]C. Figure 3 shows, as an example, the results of a multiwave test at 130[degrees]C for formulations 90D10T (Fig. 3a) and 60D40T (Fig. 3b), respectively.

[FIGURE 3 OMITTED]

The gel time data for all the formulations studied are summarized in Table 3. The addition of the hyper-branched modifier causes a reduction of the gel time. This effect has been shown in previous works (21), (22), and confirms the catalyst effect of the hydroxyl groups of the H40. The blends containing TMTG present lower gel times when compared with the base resin (100D and 100D10H40). These results further confirm that TMTG has higher reactivity when compared with DGEBA.

TABLE 3. Gel times (min) from multiwave tests.

Sample/temp.  130[degrees]C  140[degrees]C  150[degrees]C

100D              46.46          28.8           19.80
90D10T            44.44          30.74          19.35
80D20T            42.70          28.25          18.03
60D40T            41.25          26.25          16.75
100D10H40         40.41          26.11          17.34
90D10T10H40       39.50          24.15          15.50
80D20T10H40       35.75          21.40          13.50
60D40T10H40       30.20          20.05          12.60

The viscoelastic properties in the cured state of the formulations were studied by DMA, and the glass transition temperature ([T.sub.g]) was derived from the peak in the tan [delta] curve. Figure 4 shows the traces of the storage modulus (Fig. 4a) and of tan [delta] (Fig. 4b) versus temperature for the formulations with various contents of TMTG after percuring. The unmodified system (100D) presents a [T.sub.g] of 144[degrees]C. Increasing the amount of TMTG over 10% results in a decrease of [T.sub.g] for the precured systems. At the highest percentage of TMTG (60D40T), a [T.sub.g] of 110[degrees]C is found. Interestingly, the 60D40T shows a clear peak at about 175[degrees]C. A clear shoulder in the same range of temperatures is clearly observed for the system with 20% TMTG (80D20T), For the base system (100D) and for the system with 10% TMTG (90D10T), a clear shoulder cannot be seen, but the tan [delta] is not very sharp but rather broad. This behavior can be explained considering the storage modulus in the rubbery region. The storage modulus is clearly increasing (Fig. 4a), and this is a typical signature of incomplete conversion that is found for thermosets that have not been fully reacted (27). The presence of a clear peak for the 60D40T system is due to the fact that, this system being highly reactive, the gel condition and the vitrification is reached earlier than in other systems, leading to the lowest conversion level. The cure conversion level partially justifies the lowest [T.sub.g] values registered for the systems containing TMTG. In fact, another factor that has an influence on the measured [T.sub.g] is the aliphatic nature of the TMTG that tends to give rise to a more flexible network and thus to lower [T.sub.g] values.

[FIGURE 4 OMITTED]

Figure 5 shows the data of DMA for the precured formulations modified with the hyperbranched modifier H40. The formulation modified with 10 wt% H40 (100D10H40) shows a [T.sub.g] of 115[degrees]C, that is 29[degrees]C less than the [T.sub.g] measured for the unmodified system 100D. Similar reductions were found in previous studies for analogous systems (22). The [T.sub.g] decrease is a consequence of the presence of dissolved modifier in the epoxy matrix. The formulations with 10%, 20%, and 40% TMTG and 10 wt% H40 show glass transitions of 138[degrees]C, 136[degrees]C, and 110[degrees]C, respectively. The addition of TMTG to the systems containing H40 seems to be beneficial in terms of containment of the glass transition decrease for the precured systems 90D10T10H40 and 80D20T10H40. This effect could be ascribed to the multifunctional nature of TMTG and to the presence of the hydroxyl group of the hyperbranched modifier that can catalyze further reactions of the epoxy groups of the TMTG diluent. This positive effect is less efficient for the precured system 60D405 10H40, thus confirming that over certain TMTG percentages the aliphatic structure of the diluent adversely affects the thermal properties of the system.

[FIGURE 5 OMITTED]

Figures 6 and 7 show the results of DMA after postcure for the system without and with the hyperbranched modifier. The postcuring at high temperature allows to extend the conversion and to reach higher glass transition temperatures, as shown in Table 4. All the formulations, with the exception of the system 60D40T 0H40, present flat moduli in the rubbery region. This behavior is a signature of the attainment of a high conversion level after postcuring. As for the precured systems, the addition of TMTG is beneficial to obtaining small decrements of [T.sub.g] when compared with the unmodified system 100D. The formulation with 10 wt% H40 (100D 0H40) shows a clear peak of tan [delta] at about 48 [degrees] C. Similar peaks were shown for other systems modified with third pseudo-generation hyperbranched modifiers (22) and were associated with the relaxation of the hyperbranched-rich domains present in the cured systems. The presence of particulate morphology is confirmed by SEM analysis, as shown in Fig. 8. Interestingly, the formulations containing TMTG do not present any clear peak in the low temperature region, and the cured specimens for these formulations appeared transparent rather than opaque as for the system 100D 0H40. These findings support the conclusion that the addition of TMTG leads to a change in the phase separation behavior of the system. The SEM analysis did not reveal any appreciable phase separation at the scale of observation.

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

TABLE 4. Glass transition temperatures calculated as peak of tan
[delta].
              [T.sub.g] ([degrees] C)

Formulation   Precured    Postcured

100D            144          188
90D 10T         146          183
80D20T          134          170
60D40T          110          152
100D 10H40      115          142
90D10T 10H40    138          180
80D20T 10H40    136          175
60D40T 10H40    110          149

CONCLUSIONS

The aim of this work was to characterize the rheological and thermomechanical behaviors of some epoxy blends modified with the addition of a reactive aliphatic diluent and of a fourth pseudo-generation hyperbranched modifier. Both precured and postcured formulations were studied.

In the uncured state, the addition of the reactive diluent significantly reduced the viscosity of the blends modified with the hyperbranched polymer. Reductions of several orders of magnitude were obtained even when low percentages (10%) were added. The viscosity reductions result in all the systems being well below the 1 Pa-s limit for a wide range of temperatures. This is an important enhancement that can be useful for the injection at low pressure as required for most of the liquid molding techniques.

Despite the use of a medium temperature (135 [degrees] C) for the precure step and the reductions due to hyperbranched dissolution in the epoxy phase, the blends presented cured properties after precuring that were still enough to guarantee the possibility of demolding.

The reactive diluent was chosen bearing three reactive epoxy groups. The multifunctional structure of the diluent proved to be effective in improving the cured properties of the blends. In this context, the aliphatic structure of the diluent can adversely affect the properties of the resin. However, the data showed that an improvement of glass transition temperature is observed when the diluent is added at percentages lower than 40%. At the highest percentage of TMTG (i.e., for 60D40T 10H40), the postcured network properties are negatively affected by the presence of TMTG, and a [T.sub.g] reduction of about 39 [degrees] C is observed when compared with the unmodified resin (100D). However, when the comparison is made with the resin modified with 10 wt% H40 (i.e., 100D 10H40), an improvement of 7 [degrees] C is still observed.

In conclusion, several formulations have been investigated and some (i.e., 90D 10T 10H40 and 80D 20T 10H) satisfy the conditions to be injected with liquid injection molding apparatus while retaining interesting cured properties for applications in the composites field.

REFERENCES

(1.) B. Ellis, Chemistry and Technology of Epoxy Resins, Blackie Academic & Professional, Glagow (1993).

(2.) P.K. Mallick, Composites Engineering Handbook, Marcel Dekker Inc., New York (1997).

(3.) C.D. Rudd, K.N. Kendall, and C. Mangin, Liquid Moulding Technologies, Woodhead Publishing, Cambridge (1997).

(4.) A.F. Yee, J. Du, and M.D. Thouless, "Toughening of Epoxies," in Polymer Blends, D.R. Paul and C.B. Bucknall, Eds., Wiley, United States of America (2000).

(5.) C.K. Riew and A.J. Kinloch. Toughened Plastics, American Chemical Society, Washington, DC, 335 (1993).

(6.) R.J. Day, P.A. Lovell, and A.A. Wazzan. Compos. Sci. Technol., 61, 41 (2001).

(7.) K. Dusek, F. Lendnicky, S. Lunak, M. Mach, and D. Duskova, "Rubber Modified Thermoset Resin." Advances in Chemistry Series, Vol. 208. American Chemical Society, Washington DC (1984).

(8.) D. Ratna, Polymer 42, 4209 (2001).

(9.) C. Carfagna, V. Ambrogi, G. Cicala, A. Pollicino. A. Recca, and G. Costa, J. Appl. Polym. Sci., 93(5), 2031 (2004).

(10.) C.B. Bucknall and A.H. Gilbert, Polymer, 30, 213 (1989).

(11.) D.J. Hourstone and J.M. Lane, Polymer, 33, 1379 (1992).

(12.) V. DiLiello, E. Martuscelli, P. Musto. G. Ragosta, and G. Scarrinzi, J. Polym. Sci. PartB: Polym. Phys., 32, 409 (1994).

(13.) M.C. Chen, D.J. Hourstone, and W.B. Sun. Eur. Polym. J., 28, 1471 (1992).

(14.) T. Ohsako, K. Natura, and I. Nozue. Polymer. 34, 5080 (1993).

(15.) R.A. Pearson and A.F. Yee, Polymer, 34, 3658 (1993).

(16.) C.B. Bucknall. C.M. Gomez, and I. Quintard, Polymer, 35, 353 (1994).

(17.) M. Murali, D. Ratna, A.B. Samui, and B.C. Chakraborty, J. Appl. Polym. Sci. 103, 1723 (2007).

(18.) G. Cicala, R. La Spina, S. Sturiale, and A. Recca. J. Appl. Polym. Sci., 101, 250 (2006).

(19.) L. Boogh, B. Pettersson, and J.E. Manson, Polymer. 40. 2249 (2004).

(20.) R. Mezzenga, L. Boogh, and J.E. Manson, Compos. Sci. Technol., 61, 787 (2001).

(21.) G. Cicala, A. Recca, and C. Restuccia, Polym. Eng. Sci., 45. 2031 (2005).

(22.) I. Blanco, G. Cicala, C. Lo Faro, O. Motta, and G. Recca, Polym. Eng. Sci., 46, 1502 (2006).

(23.) H. Lee and K. Neville, Hanbook of Epoxy Resins. McGraw-Hill, United States (1967).

(24.) E. Zagar, M. Zigon, and S. Podzimek, Polymer. 47, 166 (2006).

(25.) J.P. Pascault, H. Sautereau, J. Verdu, and R.J.J. Williams, Thermosetting Polymers, Marcel Dekker, Inc., New York (2002).

(26.) H.H. Winter, Polym. Eng. Sci., 27. 1698 (1987).

(27.) R.B. Prime, "Thermosets." in Thermal Characterization of Polymeric Materials, E.A. Turi, Ed., Academic Press, New York (1997).

G. Cicala, (1) G. Recca, (1) S. Carciotto, (1) C.L. Restuccia (2)

(1) University of Catania, DMFCI, Viale Andrea Doria 6, 95125 Catania, Italy

(2) Cytec Engineered Materials, Abenbury WayWrexham, LL 13 9UZ, UK

Correspondence to: Gianluca Cicala; e-mail: [e-mail]

Contract grant sponsor: Italian MUR (PRIN financing scheme).

DOI 10.1002/pen.21282

Published online in Wiley InterScience (www.interscience.wiley.com).

[C] 2009 Society of Plastics Engineers