Control of Gel Time Upr

Control of gel time and exotherm behaviour

during cure of unsaturated polyester resins

Wayne D Cook,

1

*

Michelle Lau,

2

Mansour Mehrabi,

2

Katherine Dean

1

and

Marcus Zipper

2

1_{Department of Materials Engineering, Monash University, Clayton, Victoria 3168, Australia} 2_{Cooperative Research Centre for Polymers, Business Park Drive, Clayton, Victoria 3168, Australia}

Abstract: The curing behaviour of an unsaturated polyester resin has been studied by gel time and pseudo-adiabatic exotherm measurements. The gel time corresponded closely with the initial rise in exotherm temperature. Incorporation of tert-butyl catechol inhibitor increased the gel time in a linear fashion and the exotherm was similarly delayed. An increase in the concentrations of initiator (either methyl ethyl ketone peroxide or acetyl acetone peroxide) or cobalt octoate accelerator decreased the gel time in a reciprocal fashion and increased the rate of polymerization. These results are ®tted to a theoretical model for inhibition and initiation.

# 2001 Society of Chemical Industry

Keywords: gel time; exotherm; unsaturated polyester; cure kinetics; inhibition

INTRODUCTION

Many applications for thermosets, such as large mouldings and in situ castings, require that the resin be cured without external heating. The most common thermoset group, unsaturated polyester resins, are usually cured at ambient temperature by free radical polymerization through the use of peroxides and catalysts which accelerate (or promote) the peroxide decomposition to free radicals. Methyl ethyl ketone peroxide (MEKP) is one of the more frequently used organic peroxides for free radical polymerization at ambient temperatures, and a cobalt salt is commonly employed as the accelerator.1±3 _{Inhibitors, such as} phenols or quinones, react with the initiating or propagating radical, producing a radical of very low reactivity,3,4 _{thus increasing the storage life of the} polyester resin and also providing an induction stage during which the resin can be processed before cure. Because a three-dimensional network develops shortly after the start of free radical polymerization,5,6 _the control of the gelation time and cure rate of polyester resins is vital to their use in applications such as resin infusion, resin spraying and hand lay-up, and so it is important to be able to understand how the induction process depends on the initiator system.

A number of studies have qualitatively correlated the effect of initiator,1±3,5,7±9 _accelerator1,2,8±10 _and inhibitor1,3 _{on the gelation, cure rate or exotherm} behaviour of unsaturated polyester resins. There have been a few, more detailed, analyses of the effect of the concentration of initiator and accelerator on the cure behaviour of polyester resins. Yang and Suspene11

predicted the effect of MEKP and cobalt salt on the gel time of unsaturated polyester cure, while Ramis and Salla12 _{derived an expression for the cure kinetics of} polyester resins with varying levels of inhibitor in an unaccelerated polyester initiated by benzoyl peroxide. For the related styrene/divinyl benzene system, Batch and Macosko13_{also derived an expression for the cure} kinetics with varying levels of diphenylpicrylhydrazyl inhibitor and azobisisobutyronitrile initiator.

Most detailed studies of the effect of the concentra-tions of initiator, accelerator and inhibitor have been performed under isothermal or temperature-ramping conditions. Although these techniques provide a more controlled experimental environment, they cannot be related directly to the application conditions for large sections for which the polymerization occurs under almost adiabatic conditions. The importance of under-standing the curing behaviour under adiabatic condi-tions has been emphasized by Salla and Martin14 _in their work on predicting adiabatic exotherm curves. Therefore, in the present study, the kinetics of free radical initiation of polyester resins are investigated by measurement of the gel time and pseudo-adiabatic exotherm behaviour for resins initiated by varying amounts of two types of peroxide, a cobalt accelerator and a free radical inhibitor, and the results are inter-preted in terms of a gel-time model.

EXPERIMENTAL

The unsaturated polyester resin used was a com-mercial resin supplied by Huntsman Chemical Co

(Received 11 January 2000; revised version received 23 June 2000; accepted 27 July 2000)

*Correspondence to: Wayne D Cook, Department of Materials Engineering, Monash University, Clayton, Victoria 3168, Australia E-mail: wayne.cook@eng.monash.edu.au

Australia Pty Ltd as Aropol 4021 and was a poly(pro-pylene glycol±orthophthalate±diethylene glycol±fuma-rate) resin containing approximately 35wt% styrene (Fig 1). This resin was initiated with varying amounts (up to 3wt%) of methyl ethyl ketone peroxide (MEKP) supplied by Laporte Australia as MEKP-NA-1, which is a 40% solution in dimethyl phthalate and contains a mixture of species such as illustrated7_in Fig 1, or acetyl acetone peroxide (AAP; see Fig 11₎ supplied by Laporte Australia as AAP-NA-2 in a 35% solution in diacetone alcohol and was accelerated with varying amounts (up to 0.8wt%) of cobalt octoate in a 6wt% of a solution of white spirit, supplied by Thor Chemicals, Australia. Additional inhibitor (up to 0.15wt%) was added to the resin in the form of 4-tert-butyl catechol (TBC) solution (see Fig 1) supplied by Huntsman Chemical Co Australia Pty Ltd as a 40wt% solution in monopropylene glycol mono-methyl ether.

The polyester was mixed with the appropriate amount of the peroxide initiator at 25°C; then the requisite amount of cobalt accelerator was added and the mixture rapidly mixed. Approximately 70ml of this mixture was poured into a polyethylene cup at 25°C and the gel time was determined by measuring the resistance to the motion of an 8mm diameter glass tube probe using a Tecam gel-time meter (Townson and Mercer, Australia). At the same time, the cure exotherm was measured by placing a thermocouple in the centre of a 30ml quantity of the catalysed resin contained within a glass test-tube (19mm diame-ter150mm long) and initially maintained at 25°C. The tube and contents were then immediately transferred to an air-bath jacketted by a 25°C water-bath and the temperature rise due to the polymeriza-tion was recorded electronically. The exotherm rise

(T_exp) was corrected for heat loss to the environment using the heat balance method of Rojas et al.15

Tcorr ˆ Texp‡

Z

…U=Cp†…Texpÿ T0† dt …1†

where T_corr is the corrected temperature, T₀ is the ambient temperature, C_pis the speci®c heat (which is assumed to be constant), t is the reaction time and U is the global heat transfer coef®cient per unit mass (which is assumed to be constant). The U/C_p ratio was calculated from the cooling curve when the poly-merization had ceased.

At a speci®c extent of reaction (x), the heat generated up to that stage (xDH, where DH is the total heat of polymerization in J/g) can be approxi-mately equated with the heat required (C_pDT_corr) to cause a temperature rise in the sample of DT_corr. If the heat capacity of the curing resin is assumed to be constant, then the corrected exotherm curve is an approximate indication of the degree of cure as shown by the relation

x CpDT_DHcorr …2†

and so the exotherm curve provides information on the extent of cure in the resin during the polymerization. RESULTS AND DISCUSSION

Effect of inhibitor level

The important role played by the initiator on the curing behaviour of the polyester resin is illustrated in Figs 2 and 3. During a typical reaction there is a induction period in which polymerization is initially inhibited, but as time proceeds the reaction accelerates and the temperature rises to a maximum. In the

Figure 1. Typical structures of the

measured exotherm data, the temperature then decreases due to heat loss to the environment: this stage is virtually eliminated in the heat-loss corrected exotherm curves which allow for this effect. As the level of inhibitor is raised, Figs 2 and 3 reveal that the induction time is increased. A comparison of these data with the gel times measured with the gel-time meter shows that, in general, the induction period corresponds closely to the gel point, as is observed in the data of Thomas et al.7 _{This correlation is in} agreement with the commonly held view5,6,16,17 _that the gel point occurs at a low degree of reaction in multifunctional chain growth polymerization, pro-vided that the kinetic chain length is long. The maximum exotherm is also affected by the TBC concentration, and the corrected exotherm tempera-ture decreases at high TBC levels. It appears that the addition of excessive levels of inhibitor (above 0.1wt% TBC), particularly for the MEKP system shown in Fig

3, leads to depletion of the initiator before polymer-ization is complete. This phenomenon is analogous to the dead-end polymerization discussed by O'Driscoll and McArdle.18

The role of the inhibitor can best be described in terms of the initiation kinetics. It is generally con-sidered2 _{that Co}2‡_/Co3‡ _{ions reduce/oxidize} hydro-peroxides, leading to the formation of free radicals and the consequent regeneration of the Co2‡ _species during the reaction

R OOH ‡ Co2‡_ÿ!kd1 _{RO. ‡ OH}ÿ_{‡ Co}3‡ _3†

R OOH ‡ Co3‡_ÿ!kd2 _{ROO. ‡ H}‡_{‡ Co}2‡ _4†

Presumably, a similar process applies to the interaction of Co2‡_/Co3‡_{ions with peroxides; however, the exact} mechanism is not clear. Beaunez et al19 _{have noted} that the alkoxy radical (RO

.

) formed in eqn (3) is much more reactive to ethylenic monomers than the peroxy radical (formed in eqn (4) and so eqn (3) determines the rate of initiation. Equation (4) is important, however, because in this step Co2‡ _is regenerated, which results in a pseudo-steady-state for the Co2‡ _{concentration. Assuming that the rate of} consumption and regeneration of the Co2‡ _{ions are} approximately equal (ie a steady-state concentration of Co2‡ _{is attained) the rate of formation of alkoxy} radicals can be shown6_{to be}

Riˆkd1kd2‰ROOHŠ‰Co 2‡_Š

0

…kd1‡ kd2† …5†

ˆ kd‰ROOHŠ‰Co2‡Š0

As a result, the rate of radical production is predicted to be approximately constant unless the peroxide concentration is severely depleted, as noted above. When coupled with the idealized expression for the rate of free radical polymerization,20_{this gives the rate} of polymerization as ÿd‰MŠ dt ˆ kp‰MŠ…fkd†1=2‰ROOHŠ1=2‰Co2‡Š1=20 k1=2t …6† To improve storage life and for ease of application, resin systems contain varying levels of inhibitor (X) whose purpose is to consume adventitious radicals (R

.

), such as alkoxy or peroxy species, during storage and react with radicals formed in the early stages of processing, as shown by the reaction

R. ‡ X !kx RX. …7†

where RX

.

is assumed to be an inactive radical, and k_x is the rate of inhibition. If the inhibitor is ef®cient (ie the rate of inhibition is fast compared with the rate of radical production) and the loss of radicals by bi-molecular termination can be ignored, then radicals produced by the mechanism in eqns (3) and (4) will be almost immediately consumed by inhibitor (eqn (7)) and a steady-state concentration of radicals will exist.

Figure 2. Measured and corrected exotherm of the unsaturated polyester

resin containing 0.2 wt% cobalt octoate solution, 1 wt% AAP and varying amounts of TBC solution. The arrows indicate the corresponding gel times.

Figure 3. Measured and corrected exotherm of the unsaturated polyester

resin containing 0.2 wt% cobalt octoate, 1 wt% MEKP and varying amounts of TBC solution. The arrows indicate the corresponding gel times.

Thus the rate of change in radical concentration will be d‰R.Š dt ˆ kd‰ROOHŠ‰Co2‡Š ÿ kx‰R.Š‰XŠ ˆ 0 …8† and so ‰R.Š ˆkd‰ROOHŠ‰Co_k 2‡Š x‰XŠ …9†

Thus the rate of loss of inhibitor, as shown by the reaction in eqn (7), is given by4,21

ÿd‰XŠ

dt ˆ kx‰R.Š‰XŠ

ˆ kd‰ROOHŠ‰Co2‡Š …10†

When the inhibitor is ®nally consumed, the polymer-ization reaction can commence. Thus the induction, as shown in the initial stages of the exotherm curves in Figs 2 and 3, is controlled by this inhibition reaction. For an isothermal polymerization (which applies to the early stages of the experiment) and assuming that the initiator is not signi®cantly depleted, eqn (10) shows that the inhibitor concentration will decrease with time, according to4,21

‰XŠ ˆ ‰XŠ₀ÿ kd‰ROOHŠ‰Co2‡Št …11†

where [X]₀is the initial concentration of inhibitor. For active scavengers, the polymerization is effectively delayed until [X] approaches zero.4,21_{Therefore, from} eqn (11) the polymerization induction time is given by

txˆ_k ‰XŠ0

d‰ROOHŠ‰Co2‡Š …12†

This equation is similar to that derived by other workers11±13_{but it speci®cally includes the} concentra-tions of the three main species±inhibitor, initiator and accelerator.

The dependence of the gel point on the inhibitor level is shown in Fig 4. At low concentrations of inhibitor, the gel time appears to be approximately

linear with the inhibitor concentration as predicted by eqn (12). Previous rheology and DSC investigations of unsaturated polyesters11,12 _{and styrene/divinyl} ben-zene13_{showed a similar trend for the gel time or for the} reaction induction time. It is interesting to note that the gel time is not zero when the concentration of added inhibitor is zero, and that when the gel time is extrapolated to zero, the intercept on the abscissa is negative. This intercept is consistent with the presence of inhibitor added to the resin during its production in the plant, and from eqn (12) the magnitude of this intercept can be interpreted as the effective concentra-tion of pre-existing inhibitor.

The effect of varying the level of MEKP on the exotherm behaviour is shown in Fig 5. At low con-centrations of MEKP, the induction period is long and the exotherm is low. This suggests that the initiator may be partially depleted before full reaction. When the MEKP level is increased, the exotherm rises to a plateau value and the induction time is progressively reduced. Similar behaviour is also observed for the AAP system (Fig 6). The acceleration of the reaction

Figure 4. Dependence of gel time on TBC concentration for unsaturated

polyester resin initiated with 0.2 wt% Co octoate and either 1 wt% AAP (&) or 1 wt% MEKP (*).

Figure 5. Measured and corrected exotherms of the unsaturated polyester

resin containing 0.2 wt% cobalt octoate solution, a constant (undisclosed) concentration of TBC solution and varying amounts of MEKP.

Figure 6. Measured and corrected exotherm of the unsaturated polyester

resin containing 0.2 wt% cobalt octoate solution, a constant (undisclosed) concentration of TBC solution and varying amounts of AAP.

by increased levels of peroxide is con®rmed by the gel-time results shown in Fig 7. As predicted by eqn (12), the gel time is reciprocally related to the initiator concentration over most of the initiator range inves-tigated, which is consistent with the rheology and DSC studies of other workers.11±13

A similar accelerating effect of the cobalt octoate accelerator is shown for the exotherm data in Fig 8 and 9. Decreasing concentrations of cobalt salt extend the induction time and slows the reaction. It is interesting to note that in contrast to the results for varying peroxide, the corrected exotherm is independent of the cobalt salt level, suggesting that `full reaction' occurs even at very low rates of initiation because the Co2‡ species is not consumed in the reaction but is regenerated as shown in eqn (4). The plot of the gel-time data according to eqn (12) is shown in Fig 10. At low concentrations of cobalt, an approximately linear relation is exhibited between the reciprocal of the gel time and the cobalt concentration. Rheological studies by Yang and Suspene11_{on styrene±unsaturated} poly-ester systems showed an analogous relationship

between the gel time and cobalt octoate at low accelerator concentrations.

It is interesting to note that, for both the MEKP and AAP initiated systems, the gel time extrapolates to a ®nite value at zero concentration of added cobalt salt (see Fig 10), which may result from self-tion of the peroxide or by the accelerating decomposi-tion of the peroxide by catalytic impurities in the resin, in agreement with other workers.11 _{For the AAP} system, the measured gel time is longer than that expected from the extrapolated value at high levels of Co2‡_{. This may be caused by side reactions such as} reduction of the propagating radicals or the alkoxide radicals by Co2‡_{, as has been suggested elsewhere.}6,19 CONCLUSIONS

The gelation and exotherm behaviour of an unsatu-rated polyester resin with varying levels of MEKP initiator, cobalt octoate and TBC were studied. The gelation point has been found to correspond with the onset of the polymerization as expected for network

Figure 7. Reciprocal gel time versus the concentration of AAP (&) or MEKP (&) added to the polyester resin catalysed with 0.2 wt% cobalt octoate solution. The lines are guides to the data.

Figure 8. Measured and corrected exotherm of the unsaturated polyester

resin containing 1.0 wt% MEKP, a constant (undisclosed) concentration of TBC solution and varying amounts of cobalt octoate solution.

Figure 9. Measured and corrected exotherm of the unsaturated polyester

resin containing 1.0 wt% AAP, a constant (undisclosed) concentration of TBC solution and varying amounts of cobalt octoate solution.

Figure 10. Reciprocal gel time versus the concentration of cobalt octoate

solution added to the polyester resin initiated with either AAP (&) or MEKP (*). The lines are guides to the data.

formation via chain-growth polymerization. The gel time was found to be proportional to the inhibitor concentration and was inversely related to the con-centrations of peroxide and cobalt salt accelerator, as predicted by theory. The exotherm data obtained from curing under adiabatic conditions were consistent with these conclusions and complement previous work utilizing different experimental techniques.11±13 ACKNOWLEDGEMENTS

The authors would like to thank Dr John Forsythe, Dr Ference Cser, Dr Alex Kootsookos, Dr Peter Burchill, Dr Graham Durrant and Mr Enzo Palma for their helpful discussions, and Huntsman Chemical Co Australia Pty Ltd for the provision of materials. REFERENCES

1 Weatherhead RG, FPR Technology: Fibre Reinforced Resin Systems, Applied Science, London, Chapter 10 (1980).

2 Kamath VR and Gallagher RB, Developments in Reinforced Plastics, Vol 1, Ed by Pritchard G, Applied Science, London, Chapter 5 (1980).

3 Kia HG, Sheet Molding Compounds: Science and Technology, Hanser, Munich (1993).

4 Odian G, Principles of Polymerization, 3rd edn, John Wiley and Sons, New York, Chapter 3 (1991).

5 de la Cuba K, Guerrero P, Eceizal A and Mondragon I, Polymer 37:275 (1996).

6 Cook WD, Simon GP, Burchill PJ, Lau M and Fitch TJ, J Appl Polym Sci 64:769±781 (1997).

7 Thomas A, Jacyszyn O, Schmitt W and Kolczynski J, in Pro-ceedings of 32nd Annual Technical Conference, Reinforced Plastics/Composites Institute, Section 3-B, Society of Plastics Industry (1977).

8 Grentzer TH, Rust DA, Lo SK, Spencer CJ and Hackworth GW, in Proceedings of 46th Annual Conference, Composites Institute, Section 1-B, Society of Plastics Industry (1991). 9 Salla JM, Ramis X, Martin JL and Cadenato A, Thermochim Acta

134:261 (1988).

10 Cassoni JP, Harpell GA, Wang PC and Zupa AH, in Proceed-ings of 32nd Annual Technical Conference, Reinforced Plastics/Composites Institute, Section 3-E, Society of Plastics Industry, (1977).

11 Yang Y-S and Suspene L, Polym Eng Sci 31:321 (1991). 12 Ramis X and Salla JM, Polymer 36:3511 (1995).

13 Batch GL and Macosko CW, J Appl Polym Sci 44:1711 (1992). 14 Salla JM and Martin JL, J Therm Anal 42:1025 (1994). 15 Rojas AJ, Borrajo J and Williams RJJ, Polym Eng Sci 21:1122

(1981).

16 Walling C, J Am Chem Soc 67:441 (1945).

17 Yang YS and Lee LJ, Polym Process Eng 5:327±356 (1987). 18 O'Driscoll KF and McArdle SA, J Polym Sci 60:557±561 (1959). 19 Beaunez P, Helary G and Sauvet G, J Polym Sci Part A Polym

Chem 32:1459 (1994).

20 Flory PJ, Principles of Polymer Science, Cornell University Press, Ithaca (1953).

21 Burnett GM and Cowley PRE, Trans Faraday Soc 49:1490 (1953).