J. CHEM. SOC. FARADAY TRANS., 1994, 90(6), 883-888 Dielectric and Steric Hindrance Effects on Step-polymerization of a Diepoxide with Monoamines G. P. Johari Department of Materials Science and Engineering, McMaster University, Hamilton, Ontario, Canada L8S 4L 7 The step-polymerization reaction between monoamines and diepoxide has been investigated by measuring the permittivity and dielectric loss as a function of reaction time, t,,, at different temperatures. Hexylamine and cyclohexylamine were used both in stoichiometric and non-stoichiometric (amine-rich) compositions with the diglycidyl ether of bisphenol-A. The dielectric consequences of increase in the growth of the macromolecule during the step-polymerization are analogous to the frequency dependence of a chemically invariant sub- stances.The reaction-time dependence of &* has the functional form: @(t)= exp -[t/z(t,,)lr where Y = 0.3-0.4. Steric hindrance of the NH, group in cyclohexylamine slowed the rate of polymerization and the E* features were shifted towards longer t,, . Excess amine accelerated the polymerization process, but shifted the dielectric features towards longer reaction time. Although the reaction reached completion, the mixture did not vitrify. In this respect the dielectric behaviour differs from that of network polymerization. Reaction temperature, time, composition and chemical structure of the monoamine, all affected the strength of the sub-glass transition tem- perature (T,) relaxations of the polymerized product. The rate of chemical reactions in a polymerization process becomes controlled by the diffusion coefficient of the reacting species, especially as the reactions approach completion or viscosity becomes high.However, this diffusion coefficient decreases as a macromolecule grows, thus slowing the rate of the very chemical reaction that allows its growth. In those polymerization processes where the groups remain reactive, the extent of polymerization itself slows the rate of its progress, an occurrence referred to as negative feedback between a chemical (reaction) and physical (molecular diffusion) process.’ Cal~rirnetric,’.~ andultrasoni~~~’~ Brillouin light scattering’ studies of macromolecular network formation have shown that relax- ation characteristics of a reacting mixture change with time in a manner analogous to that observed on supercooling a chemically stable substance.Thus, for each relaxation state of the chemically stable, unreacted or partly reacted liquid (with van der Waals and other weak molecular interactions) char- acterized by its relaxation time alone, there is a correspond- ing state of its polymeric structure (containing now a certain number of intramolecular bonds) with the same relaxation time but at a higher temperature. Since molecular diffusion time is determined by EJk, T,where Ei represents the activa- tion energy for a diffusion process, and k, the Boltzmann constant, the same diffusion time can be achieved by (i) an increase in Ei at a constant temperature, T, as during the growth of a macromolecule, or (ii) a decrease in T for a con- stant Ei as on supercooling a chemically stable liquid.This means that there is a phenomenological equivalence in terms of k, T,for molecular diffusion time in the same composition, with and without covalent bonds. For dielectric studies of the growth of a macromolecule, we, as others,12-19 had chosen substances which formed a network structure on polymerization, mainly because of their well known technological importance as thermosetting resins. Nevertheless, one study of free-radical polymerization of methyl methacrylate was carried out, but without success in achieving vitrification.20 Here we report a dielectric study of linear-chain polymerization between monoamine and di-epoxide, and examine the effects of steric hindrance of the NH, group and non-stoichiometry on the rate of poly-merization ; we also report the dielectric properties of the polymerization product.This is a first report of our studies on a series of substances from which hindrance for molecular diffusion caused by a network topology is absent, because the reaction of an epoxide with a difunctional amine does not normally produce a cross-linked network. Experimental Hexylamine and cyclohexylamine of 99% purity were pur- chased from Aldrich Chemicals and diglycidyl ether of bisphenol-A (DGEBA), M, = 380 and functionality = 2, was donated by Shell Chemicals as Epon 828. All chemicals were used without further purification.Accurately weighed amounts were mixed and transferred to a 10 mm diameter glass container. A ten-plate miniature tunable capacitor was immersed in the liquid mixture, and a thermocouple was kept immersed 2 mm above the capacitor. The details of the dielectric measurement assembly, temperature controls and source of errors have been described in earlier papers.’ Results The permittivity, d, and dielectric loss, E”, were measured for a fixed frequency of 1 kHz at suitable intervals as poly-merization of stoichiometric (1 : 1 molar ratio) and monoamine-rich (2 : 1 molar ratio) mixtures of hexylamine- DGEBA and cyclohexylamine-DGEBA at different tem-peratures progressed. These are plotted against the reaction time, t,,, in Fig.1 and 2. E’ decreases at different rates in different regions, but the manner by which the decrease occurs in a given region of t,, varies with the composition and the temperature. [Note that the discontinuity of curves (e) near t,, = 3 ks in Fig. 2 was caused by an accidental failure of the temperature control unit, and is not a property of the material.] E“ decreases initially towards a local minimum followed by a peak from where it ultimately decreases to a low value. This decrease varies with the tem- perature and composition of the reacting mixture. The plots also show that on increasing the reaction temperature, T,, , d(t,, -,0) and the &”-peak height, decrease, dr(tre-+ 0) increases and the curves shift towards shorter tre.These fea- tures are qualitatively similar to those observed previously.6 E’ and E” for all monoamine4iepoxide mixtures are plotted in a complex plane in Fig. 3, where each point refers to a 884 8.51 8.04 I 7.5-7.0->. c '5 6.5-.-c *g 6.0-h 5.5-5.0-4.5-4.0-3.5 I I I I I I I l l 1 I I I I I Ill1 1 1 I I I Ill 10: 1: 0.1: 0.01 ! I I I I I IIII I I I I I IIII I I I I I (Ij 100 1000 10 000 10G lr00 t re/S Fig. 1 Permittivity and dielectric loss of hexylamine-DGEBA mix-tures measured at 1 kHz us. tre. Data shown by filled rectangles are for 2 : 1 molar ratio; all others for 1 : 1 (stoichiometric) molar ratio. (a), (b) and (c) refer to the reaction temperatures given in Table 1.certain t,, . The shape of the plot is skewed more at the long-time than the short-time end, except for the stoichiometric hexylamine mixture at 316.2 K and its amine-rich mixture at 309.5 [Fig. l(c)]. As in earlier st~dies,~-~the E* data were converted to M*(=~/E*)and the real and imaginary com-ponents of M* are plotted in a complex plane in Fig. 4. These show two relaxation processes : (i) conductivity relaxation, which appears as a semicircle with centre on the axis and (ii) Table 1 Dielectric parameters measured at 1 kHz during the poly-merization reaction of the two monoamines and DGEBA hexylamine-DGEBA 306.2 (a) 14.4 0.525 3.8 (3.9) 6.6 0.35 316.2 (b) 10.6 0.474 3.8 (3.9) 6.3 0.35 309.5" (c) ->0.024 5.4 cyclohexylamine-DGEB A 300.2 (6) 37.2 0.649 3.5 (4.0) 7.5 0.34 306.2 (e) 23.3 0.580 3.8 (3.9) 7.1 0.32 312.2 (f) 16.9 0.548 3.8 (3.9) 6.8 0.34 316.2 (g) 11.8 0.526 3.8 (4.0) 6.7 0.34 309.5" (h) 26.6 0.430 3.5 (3.6) 5.9 0.34 Notations in brackets refer to the plots in Fig.1-7. Values in brackets are those used for calculating Y and z. &'(tre+0) was obtained from the data fits shown by triangles in Fig. 3, and differs from those in Fig. 1 and 2. These values represent the dipolar contri-bution. " Monoamine saturated composition molar ratio (2 : 1). J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 8.5 I 8.0-7.5-7.0->..z 6.5-.-Z 6.0-E g 5.5-4.0 3.5 0.001 / I I I I1 Ill1 I I I I IIIII I I I IIIII 100 1000 10 000 100 000 t re/S Fig.2 Permittivity and dielectric loss of cyclohexylamine-DGEBA mixtures measured at 1 kHz us. tre. Data shown by dots are for 2 : 1 molar ratio; all others for 1 : 1 (stoichiometric) molar ratio. (4,(e), (f),(9)and (h) refer to the reaction temperatures given in Table 1. dipolar relaxation which appears as a skewed arc (t,, increases from left to right in Fig. 4). After a certain period of reaction (at which the plots in Fig. 1 and 2 terminate), the product was cooled in situ to 77 K and its E' and E" for 1 kHz measured as a function of tem-perature up to near T,, .Plots of tan 6( =E"/E') us. temperature of the polymerized product in Fig. 5 show relaxations in the glassy state. (The change in the cell constant with change in temperature is not known accurately and therefore tan 6 instead of E" plots are given.) These appear as a peak at about 250 K and a shoulder to this peak in the range 100-175 K.The height and the temperature of both vary with the nature and amount of reactants. Earlier studies have shown that for a given reactant and T,, they also vary with the reac-tion time,21*22particularly during the early period of chemi-cal reactions." The data determined from Fig. 3 are listed in Table 1. Here, ~'(t,,-+0) differs from those in Fig. 1 and 2 for the fol-lowing reasons. Values in Fig. 1 and 2 contain contributions from interfacial impedance which decreases as the reaction progresses (see ref. 6 for details), leading thus to a minimum in E" at the long-time end of the plots in Fig.3. &'(ire-+ 0) in Fig. 1 and 2 thus differs from that in Fig. 3. The values corre-spond to the dipolar process. Discussion The dielectric properties during the polymerization are deter-mined by the changing composition of the mixture in which a J. CHEM.SOC. FARADAY TRANS., 1994, VOL. 90 3.01 n A 2.5 uI 2.0 -81 Iu 0 1.5.-LU $ 1.0 .--0 0, I I 1 I I I I I -.-I CB 4.5 4.0 I 3.5 v) -8 3.0 .-0 2.5 -a a 5 2.0 1.5 1 .o 0.5 0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 perrnittivity Fig. 3 Cole-Cole plots of A, hexylamine-DGEBA and B, cyclohexylamine-DGEBA mixtures measured at 1 kHz. (a)-(h) refer to the reaction temperature, the composition and the monoamines given in Table 1.The reaction time increases from right to left. Tri- angles are the calculated values from the data in Table 1. (c) and (h)are on the scale, the axis for the others is shifted upwards, as shown. macromolecule grows.In chemical terms, it means a decrease in the number of monomers and their dipole moments, in physical terms an increase in the configurational restriction to dipolar diffusion and in the intermolecular vibrational fre- quencies. The changes in the dielectric properties during the poly- merization process are caused by at least seven effects:'e6 (i) A general decrease in the dc conductivity as the diffusion coeffi- cient of impurity ions present and zwitter ions22 formed in the reacting mixture decreases with increase in its viscosity and as any proton transfer along the H-bonded network in the mixture is virtually eliminated by the consumption of the H bonds between the monomers.(ii) An increase in the molecular diffusion or relaxation times as a result of which the permittivity measured at a fixed frequency decreases monotonically toward the value corresponding to the IR region. (iii) A change in the number of dipoles per unit volume and, therefore, in the contribution to permittivity from orientation polarization, AE, as a result of both the chemical reactions that alter the dipole moment and the volume contraction that raises the number density of dipoles. (iv) The splitting of a unimodal relaxation function into a bimodal relaxation function.' (v) A change in the dielectric relaxation function as the chemical structure of the liquid changes and its viscosity and density increase.(vi) A change in the contribution to permittivity due to IR-polarization, as the vibrational frequencies of the various 885 0.20 j 1 ............... 01 I -1. 0.30 BI t/;."---m..........*.*,ib...............1 0.20 ................. M" 0.15 21 qb .............*'......\I0.10 /.. ........ .....0 I I 4 I I 0 0.05 0.10 0.15 0.20 0.25 0.30 M' Fig. 4 Complex plane plots of M" and M' for A, hexylamine-DGEBA and B, cyclohexylamine-DGEBA measured at 1 kHz. (a)-(h) refer to the reaction temperature, the composition and the monoamines given in Table 1.(a) and (d) are on the scale, the axis for the others is shifted upwards, as shown. The reaction time increases from left to right. 0.1 *fly-I it 0.001 I I I I 0.001 100 150 200 250 300 . i0 T/K Fig. 5 Loss tangent of the polymer formed on reaction of mono- amines with diepoxide measured at 1 kHz 0s. T. A, Hexylamine and B, cyclohexylamine. (a)-(c), (e), (9) and (h) refer to the reaction tern-perature and time given in Fig. 1. modes in the structure change on polymerization and densifi- cation. (vii) A change in the optical refractive index or optical polarization as the structure densifies on polymerization. All these effects can be taken into account in eqn. (1). However, their relative magnitudes vary for a given instant reaction.Generally, the change in dielectric properties is dominated by (i), (ii) and (iv), for which substantial evidence has been provided by our earlier st~dies.~-~ As before,6 we write: Eqn. (1) is derived from time-frequency interrelations when the thermodynamic properties of a material remain constant with time. So, for a fixed t,,, the equation is valid as in most cases. It is also valid when the rate of change of thermodyna-mic properties with time is much slower than the rate of polarization and depolarization or when T is much smaller than the inverse of the reaction rate constant. For such cases : However, when the two approach each other, eqn. (1) is no longer valid, and an evaluation of the function [@(t)lt,, becomes complicated, because now two timescales are involved.This evaluation remains to be done. For convenience of analysis, we assume that (doldt), E~ and E, do not change with t,,, and replace /? by T, the reaction parameter for time-variant chemical and physical states. [It has a parallel in the physical ageing beha~iour,,~ where a similar form to eqn. (2) is used with 7 increasing irreversibly with the ageing time.] This does not imply that a direct Fourier transform of eqn. (2), now containing Y instead of j?(t,,) into a frequency domain is meaningful. It does have a technical importance for following the progress of reaction by measuring the dielectric and ultrasonic proper tie^.^,' /? (for a time-invariant state) is usually higher than T and, for network formation on polymerization, the two have opposite temperature dependence.’ Eqn.(1) is invariant on the choice of o or T as a variable, and features of E‘ and E” in Fig. 1-4 result from an irreversible increase of 7,which increases oz from a value less than one to one greater than one. When or = 1, &“ = &Lax. As in earlier the relaxation time was calcu- lated from the data in Fig. 1-4. Its value corresponds to the characteristic time 7(t,,) of eqn. (2), and not to the E” peak or average value of 7. It is plotted logarithmically against the reaction time of monoaminediepoxide mixtures in Fig. 6 and 7. The sigmoid-shaped plots shift towards longer t,, with increase in T,, ,and increase in the monoamine content of the mixture.The first effect is expected for virtually all poly- merizing reactions which are accelerated by increasing the temperature, and the second depends upon the amine used. Y and other dielectric parameters are summarized in Table 1. We now discuss the above observations in terms of the growth of a macromolecule and the steric hindrance effect of the amine on this growth. In contrast with our earlier studies of network polymers formed by the reaction of a diamine with a diep~xide,~-* the reaction product here should be a linear-chain polymer, for here the -NH, group of the mono- amine reacts with the epoxy groups to form -CH( 0H)- CH, -N( R)-CH, -CH( 0H)-linkage between two DGEBA molecules (a less probable reaction between one J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Or--1 I log (trels) Fig. 6 log 7 us. log t,,, as polymerization in hexylamine-DGEBA progresses. (a) and (6) refer to the reaction temperature given in Table 1. NH, group and two epoxide groups of the same DGEBA molecule will produce a highly strained molecular ring but not a network structure). Thus, the alkyl group, R, of the amine becomes a pendant in a linear chain of DGEBA mol- ecules linked by nitrogen atoms. Calorimetric studies of the reaction products showed that their endotherms are narrow, similar to those for linear-chain polymers. (Endotherms for network polymers tend to be broad.) A com-parison of the various plots in Fig. 1 and 2, and the data in Table 1 show that the E” peak during polymerization at the 0 -1 -2 h $ -3 Y CT,--4 -5 -6 -71 I I 3.8 4.0 4.2 4.4 4.6 4.8 log (treis) Fig.7 As for Fig. 6, but for polymerization of cyclohexylamine- DGEBA mixtures. (4-(h) refer to the conditions listed in Table 1. Data shown by dots are for 2 : 1 molar ratio; all others are for 1 : 1 molar ratio. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 same T,, appears at a shorter reaction time for hexylamine than for cyclohexylamine. This means that the composition of the polymer (and molecular weight) reach the same relax- ation time (=w-’ or 159 ps) after 14.4 ks of reaction with hexylamine and 23.3 ks with cyclohexylamine at 306.2 K. A corresponding difference is also observed at 316.2 K in Table 1.The rate of growth of the polymers is more easily examined by a comparison of the increase in the relaxation time with the reaction time given in Fig. 6 and 7, which are drawn on the same scales. Polymerization is more rapid with hexyla- mine than with cyclohexylamine. Since the reaction mecha- nism between a diepoxide and monoamine is expected to be independent of the shape of the monoamine, this implies that the probability of mutual diffusion of the reacting epoxide and -NH, groups is greater in linear-chain hexylamine than in cyclohexylamine. This is particularly important because the rate of chemical reaction is seen to be viscosity indepen- dent when the viscosity of the reaction medium is low, but viscosity dependent when it is high and molecular diffusion controls the reaction rate.Since the viscosities of the unre- acted mixtures here are not greatly different, the slower rate of polymerization must be attributed to a greater steric hin- drance of the -NH, group in cyclohexylamine than in hexylamine, which decreases the probability of mutual approach of NH, and epoxide groups. We now discuss the effect of increasing monoamine (amine- rich non-stoichiometry) on the observed dielectric behaviour. Although the E” peak shifts towards longer t,, for both mono- amines (Fig. 1 and 2), this shift is not clearly discernible for hexylamine. Both amines are less viscous than DGEBA, so an increase in the amine concentration lowers the viscosity of the unreacted mixture.Both an increase in concentration and a higher diffusion coefficient increase the polymerization rate of an amine-rich composition relative to the stoichiometric composition. Therefore, one expects that the E” peak for the amine-rich composition will appear sooner than for stoichio- metric Composition. However, even with the faster poly-merization rate, there is the requirement that for the same relaxation time, the size and molecular weight of the macro- molecule be greater in a low-viscosity solvent than in a high- viscosity solvent. This requirement is fulfilled at long t,, in a low-viscosity solvent, i.e. t,, to reach wz = 1 is greater as the increase in the polymerization rate is outweighed by the requirement for a larger macromolecular size.The relatively smaller decrease of E’ from 6.2to 5.3,the low E” as t,, 4co in Fig. l(c), and the absence of a dipolar relaxation process in Fig. 3(c),all show that although polymerization in amine-rich hexylamine-DGEBA has reached completion, the relaxation time of the solution (1 mol polymer in 1 mol unreacted hexy- lamine, or oligomers containing different numbers of -CH(OH)-CH,-NH-R linkages) at 309.5 K does not reach 159 ps (=o-’)as t --+ m. The shift in the E” peak towards longer t,, in amine-rich mixtures may therefore be interpreted either as a solvent effect on the relaxation time, or as an indication of the formation of a variety of oligomers, or a combination of both. This observation is supported by the tan 6 plots in Fig.5 which show that the a-relaxation peak of the amine-rich hexylamine-DGEBA mixture appears at ca. 285 K; for all others, it appears above t,,. Measurements, after pumping to remove the excess amine from the reaction product, will ascertain whether the effect is entirely due to the unreacted amine acting as solvent or due to the formation of oligomers as well. For the cyclohexylamine-rich mixture, wz does reach unity as the reaction progresses and an E” peak is reached (Fig. 2 and 4).As discussed, t,, for the peak is expected to be longer here than for the stoichiometric mixture, but z does not become high enough to cause vitrification (Fig. 7). Our studies of network polymerization have shown the opposite, i.e.chemical reaction in a 4,4’-diaminodiphenylmethane-rich DGEBA mixture occurred faster and the E‘’ peak appeared at a shorter time than in a stoichiometric mixture, i.e. excess diamine accelerated the reaction.24 It is also found 25*26 that when the reactants are dissolved in an inert solvent, the reac- tion is retarded and the polymer formed has a lower cross- link density and % than that formed without a solvent. Thus, network polymerization differs from step-polymerization in this respect. For a detailed understanding of these effects, we are investigating linear-chain polymerization of monoamine-diepoxide mixtures in inert non-polar solvents. When the reaction temperature is increased, the plot of 7 us. t,, for the stoichiometric cyclohexylamine-DGEBA mixture shifts towards the left, but such plots at different T,, do not cross each other, as seen in Fig.7. This occurs also for the hexylamine mixture in Fig. 6,but the plots for the two temperatures show that one crosses the other at t,, w 15.9 ks. Strictly interpreted, it means that z of the products and reac- tants mixture at t,, w 15.9 ks at 316.2K is the same as that at the same time but at 306.2 K.Thereafter, 7 at 316.2 K remains less than at 306.2K as the reaction continues further. This, of course, is also apparent in Fig. l(a) and (6) where E‘ and E“ at 316.2K (6) do not decrease to as low values, and oz does not increase to as high values as for 306.2 K.It seems that there are at least three reasons for such an occurrence.First is the extent of chemical reaction itself which, at limiting long times (when equilibrium concentration of the reactant and product is reached), is less at a high temperature than at low temperatures, because the equilibrium constant of exo-thermic reactions decreases rapidly with increase in the tem- perature. Thus, the ultimate concentration of the polymerized product is less at 316.2 K than at 306.2K,and hence r is less, although it increases with time initially more rapidly at a higher K, than at a lower T,, . Secondly, there are more oligo- mers formed at a higher than at a lower T,, and hence z remains lower and, thirdly, z of the reactant-product mixture is extremely sensitive to temperature, so that for the same composition of the reactant-product mixture a change from 306.2 K to 316.2 K decreases z by as much as two decades. The effect observed in Fig.6 may be due to all three, but if the first is the predominant cause, then all reacting mixtures which do not vitrify on polymerizing, and thereby allow the concentration of reactants and products to remain at chemi- cal equilibrium after a reasonable observation period, should show the crossover of z as in Fig. 6.We plan to examine this by further studies. The relaxation in the glassy state of fully and partially polymerized mixtures is of interest in determining the extent to which localized motions contribute to the electrical properties. In pure DGEBA, there are two such relaxation processes appearing at 156 K as a peak of height 0.025, and at 200 to 225 K,a shoulder of height 0.008 (T,= 293 K).,, In the polymerized states, both the magnitude of the sub-? relaxation processes and the shape of the features at the two temperatures are reversed.The low-temperature process appears in about the same temperature range as in pure DGEBA,,, but its strength is decreased and it appears as a shoulder; the high-temperature process also appears in the same temperature range but its strength is increased and it appears as a peak except for the hexylamine-rich mixture where the peak vanishes. It seems that changes in all the four variables namely: (i) T,, , (ii) t,, , (iii) composition, and (iv) structure of the amine, affect the sub-T, relaxations.Increasing T,, and t,, raises the height of the high-temperature peak and decreases that of the low-temperature shoulder. Substitution of hexylamine by 888 cyclohexylamine increases the strength of both the low- and high-temperature relaxations and shifts the peak towards a higher temperature. Increasing amine content lowers the height of the high-temperature, but raises that of the low- temperature relaxations. In our detailed studies of the effect of reaction temperature and time on the sub-% relaxations in a network polymer formed by the reaction of 4,4'-diaminodiphenylmethanewith DGEBA,2'*27 and pure DGEBA catalysed with N,N'-dimethylbutylamine,22 an increase in T,, increased the height of high-temperature relaxation and decreased that of the low- temperature relaxation for a fixed reaction time, as did the increase in t,, for a fixed Te.On the basis of these earlier observations we concluded that sub-T, relaxations are more appropriately understood in general physical terms without assigning, on an ad hoc basis, a particular diffusion mode or group of molecules for each relaxati~n.~~*~~ In the unreacted state there is only one, the low-temperature (145 K) relax-ation peak in the diamine-DGEBA mixtures20*21*27(there are two in pure DGEBA).What seems remarkable is that in the limits of monomeric and almost fully polymerized states, there is also only one sub-T, relaxation process. In the mono- meric state it appears at 145 K, in the polymerized state at 262 K.In the intermediate states of incomplete poly- merization both occur, and there is a gradual attrition of the first and emergence of the second as polymerization approaches completion. This evolution of sub-% relaxations is evidently a reflection of the development of structures with molecular-packing inhomogeneity as weak van der Wads, H-bonding, and dipolar interactions lose their predominance to topologically well defined, covalent bonded structures. Conclusions Dielectric measurements of a liquid in which chemical reac- tions occur to cause linear-chain polymerization provide useful information on both the irreversible change in the relaxation time and its dependence on the reaction time. In particular, the study leads to the following conclusions: (1)The irreversible decrease in the permittivity measured for a fixed frequency as the polymerization continues is phenomenologically similar to that observed by increasing the frequency under isothermal conditions of a chemically stable liquid.(2) Several effects associated with steric hindrance of the reacting group slow the rate of polymerization. (3) Excess amount of the more fluid component in the reacting mixture accelerates the process, but vitrification does not occur. (4) Reaction time, temperature, composition and chemical structure all affect the strength of sub-% relaxations of the polymerization product. I am grateful to Mr. W. Pascheto and Mrs. M. Wang for their assistance in data collection and analysis.This research J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 was supported by a grant from the Natural Sciences and Engineering Research Council of Canada. References 1 G. P. Johari and M. B. M. Mangion, J. Non-Cryst. Solids, 1991, 131-133,921. 2 M. Cassettari, G. Salvetti, E. Tombari, S. Veronesi and G. P. Johari, I1 Nuovo Cimento, 1992,14D, 763; J. Polym. Sci., Part B: Polym. Phys., 1993,31, 199; Physica A, in the press. 3 G. P. Johari, Dynamics of Irreversibly Forming Macromolecules, in Disorder Effects on Relaxation Processes, ed. R. Richert and A. Blumen, Springer-Verlag, 1994, in the press. 4 M. B. M. Mangion and G. P. Johari, J. Polym. Sci., Part B: Polym. Phys., 1991,29,1117; 1127. 5 M.G. Parthun and G. P. Johari, Macromolecules, 1992, 25, 3149; 3254; J. Polym. Sci., Part B: Polym. Phys., 1992,30,655. 6 G. P. Johari, in Chemistry and Technology of Epoxy Resins, ed. B. Ellis, Chapman and Hall, London, 1993, ch. 6, p. 175; J. Mol. Liquids, 1993,56, 153. 7 E. Tombari and G. P. Johari, J. Chem. Phys., 1992,97,6677. 8 M. Cassettari, G. Salvetti, E. Tombari, S. Veronesi and G. P. Johari, J. Mol. Liq., 1993,56, 141. 9 I. Alig, D. Lellinger and G. P. Johari, J. Polym. Sci., Part B: Polym. Phys., 1992,30,791. 10 I. Alig, K. Nancke and G. P. Johari, J. Polym. Sci., Part B: Polym. Phys., in the press. 11 M. B. M. Mangion, J. J. Vanderwal, D. Walton and G. P. Johari, J. Polym. Sci., Part B: Polym. Phys., 1991,29,723. 12 C. Huraux and A.Sellaimia, C.R. Acud. Sci. Paris, Part B, 1973, 277, 691. 13 A. Soualimia, C. Huraux and B. Dispax, Mucromol. Chem., 1982, 183,1803. 14 S. D. Senturia and N. F. Sheppard Jr., Ado. Polym. Sci., 1986, SO, 1. 15 D. R. Day, Polym. Eng. Sci., 1986,26,362. 16 See references cited in the reviews, ref. 6 and 14. 17 D. E. Kranbuehl, J. Non-Cryst. Solids, 1991,131-134,930. 18 G. M. Maistros, H. Block, C. D. Bucknall and I. K. Partridge, Polymer, 1992,33,4470. 19 C. G. Delides, D. Hayward, R. A. Pethrick and A. S. Vatalis, J. Appl. Polym. Sci., 1993,47,2037. 20 E. Tombari and G. P. Johari, J. Chem. Soc., Faruduy Trans., 1993,89,2477. 21 M. B. M. Mangion and G. P. Johari, J. Polym. Sci., Part B: Polym. Phys., 1991,29,437. 22 I. Alig and G. P. Johari, J. Polym. Sci., Part B: Polym. Phys., 1993,31,299. 23 I. M. Hodge and A. R. Berens, Macromolecules, 1981,15,762. 24 M. B. M. Mangion, M. Wang and G. P. Johari, Macromolecules, 1992,30,433. 25 K. Hofer and G. P. Johari, Macromolecules, 1991,24,4978. 26 G. Mikolajczak, J. Y.Cavaille and G. P. Johari, Polymer, 1987, 28,2023. 27 M. B. M. Mangion, M. Wang and G. P. Johari, J. Polym. Sci., Part B: Polym. Phys., 1992,30,445. 28 J. D. Keenan and J. C. Seferis, J. Appl. Polym. Sci., 1979, 2.4, 2375. 29 M. Ochi, M. Yoshizumi and J. Shimbo, J. Polym. Sci., Part B: Polym. Phys., 1987,25,1871. Paper 3/055461; Received 14th September, 1993