Peculiarities in the solvent absorption characteristics of epoxy-siloxane hybrids.
Epoxy-silica hybrids are produced by the sol-gel method from mixtures of epoxy resins and tetralkoxysilanes [1-5]. The organic and inorganic components of these materials are present as co-continuous nanostructured phases, which are linked by covalent bonds through reactions with suitable "coupling agents" [6-8]. The formation of continuous nanostructured inorganic domains within an organic network relies on establishing the appropriate kinetics for the hydrolysis and condensation reactions of the alkoxide precursor, together with the development of strong interactions with organic component of the precursor through the formation of covalent bonds [9-12]. It is for this reason that organic-inorganic hybrids (O-I), in general, are more easily produced from tetralkoxysilanes than other metal-alkoxides [13, 14].
For the systems of this study the use of coupling agents ensures that the chemically induced spinal decomposition of the precursors, in the formation of the siloxane domains and the epoxy/hardener network, produces a co-continuous distribution of the two phases at nanoscale length [15, 16]. The presence of co-continuous inorganic nanostructured domains within an organic matrix is the most efficient way of utilizing the properties of the two components within a single material and, therefore, the use of coupling agents has a prominent role in the production of O-I hybrids.
While there are several reports on the mechanical properties of such nanocomposites, relatively little is available in the literature regarding other properties, apart from some references to barrier properties. In view of the enormous resistance of inorganic oxides to swelling by solvents, there are obvious implications for improving barrier properties of polymers through the production of O-I hybrids. Substantial improvements in barrier properties have, indeed, been reported even for nanocomposites produced from exfoliated clays, where the content of inorganic matter is usually limited to about 5% w/w and the inorganic domains are discontinuous, albeit plate-like, with a high aspect ratio [17a, b].
The feasibility of achieving a high resistance to solvent absorption in organic-inorganic hybrids has been indicated in the literature . The scope of the present study is to explore further the means of producing epoxy-siloxane hybrids by the sol-gel method and to examine their solvent absorption characteristics. In previous work on epoxy-silica hybrids  the silica precursor was prehydrolyzed using an acid catalyst and the hardener used for cross-linking the resin was an anhydride. The cured products were very brittle and the glass transition temperature (Tg) of the organic domains decreased considerably. This was attributed to a reduction in the cross-linking efficiency of the chosen system.
In the present study an organotin compound was used as a condensation catalyst for the siloxane component, together with an amine hardener for curing the epoxy resin. A previously developed procedure for the alkoxysilane functionalization of the epoxy resin was followed as a means of coupling the organic and inorganic domains of the nanocomposites . The effect of cycling the solvent absorption treatment, after drying the solvent, was also examined with a view to determining whether such treatments bring about any changes in properties through structural changes. This behavior has also been proposed by other workers [21, 22].
In the case of cross-linked glassy polymers these structural changes have been associated with physical aging and have been related to the cross-linking density. Denser networks were found to be less prone to physical aging because of the reduced free volumes and higher levels of constraints on the network to undergo configuration rearrangements .
Fillers have been found to reduce physical aging in cross-linked glassy polymers, to some extent due primarily to dilution effects . The presence of co-continuous inorganic-like domains in O-I hybrids, on the other hand, is expected to restrict even more the relaxations within the surrounding organic network and bring about more favorable conditions for the occurrence of local reactions through confinement effects .
Epoxy Resin. The resin used was a mixture of two bisphenol A type epoxy resins, commercially known as Epikote 828 and Epikote 1009, both obtained from Shell Chemicals (Houston, TX). These have a number average molecular weight of 370 and 5,000 atomic mass units, respectively, corresponding to an average degree of polymerization of 0.1 and 16 expressed in terms of central C[H.sub.2]CHOHC[H.sub.2] units per molecule.
On the basis of results obtained in previous work, which showed that higher molecular weight resins produced "better compatibility" of the two phases (i.e., finer domains), a weight ratio of Epikote 828 to Epikote 1009 equal to 9:1 was used in the preparation of the resin mixture .
Hardener. The hardener used in all cases was 4-4' methylene bis-cyclohexaneamine (PACM), obtained from Air Products as a 99% purity grade.
Coupling Agent for the Functionalization of the Epoxy Resin. The silane coupling agent used for functionalization of epoxy resin was bis-([gamma]-propyltrimethoxysilane)amine (A1170) from OSi Specialities, >95% purity
Precursors for the Siloxane Component. These were, respectively, tetraethoxysilane, as the main component (TEOS), and [gamma]-glycidoxyl trimethoxysilane (GOTMS) as an additional coupling agent. Both were obtained from ACROSS, purity >95%.
Catalyst. The catalyst was dibutyl-tin-dilaurate (DBTL), obtained from Aldrich (Milwaukee, WI), >95% pure.
Solvents for the Epoxy Resin. Solvents used to dilute the epoxy resin were xylene (Fluka, Buchs, Switzerland) and butanol (Aldrich) as a 1:1 mixture.
Swelling Solvents. The solvent used for the solvent absorption experiments were tetrahydrofuran (THF) from Fluka and methanol from Aldrich.
Functionalization of the Epoxy Resin Mixture
The epoxy resins were initially dissolved in a mixture of xylene and butanol (2:1). The composition of the solution was fixed at 60% solids to obtain a viscosity in the region of 0.1-0.2 [Nm.sup.-2]s. The weight ratio of the epoxy resins was 85% Epikote 828 and 15% Epikote 1009. The grafting reaction with the amine-silane coupling agent (A1170) were carried out at a molar ratio Epoxy:NH = 10:1. The details of the reactions have been described elsewhere [19, 20], but it is clear that the alkoxysilane functionalization takes place via the addition reaction of the amine on the epoxy groups of the resin.
Preparation of the Silica Precursor Solution
Following the procedures used in similar work [12, 15, 16], TEOS, GOTMS, ethanol, water, and DBTL were mixed using the following molar ratios: water:TEOS = 3:1, water:ethanol = 1:1, TEOS:GOTMS = 1:0.12.
The catalyst was added in amounts equal to 0.02% of the theoretical silica produced, including the equivalent amount derived from the silane functional groups in the resin. The Epoxy/TEOS weight ratios used was chosen to produce theoretical silica contents varying from 3-15% w/w. This was approximately the highest level of silica that could be produced by the outlined procedure without a propensity for the system to form large particulate silica domains. The theoretical Si[O.sub.2] content was calculated from conversion of moles to weight quantities, from which it is estimated that 28.8% is derived from TEOS and 25.4% from GOTMS.
The silica precursor solution was stirred first for 5 min at room temperature, then for 120 min at 90[degrees]C
Production of Epoxy-Silica Nanocomposites
After completion of the described reactions, the components of the precursor solution were mixed with the epoxy resin solution and prereacted under stirring conditions for 15 min at 60[degrees]C and 120 min at 80[degrees]C.
Specimen Preparation and Curing
The epoxy/alkoxysilane precursor solution and the hardener, PACM, were mixed for 1 min at room temperature, then spread on shallow molds made from supported siliconized paper to produce small plaques about 0.5-1.0 mm thick. The amount of hardener used was varied in the following stoichiometric ratio Epoxy groups : -N[H.sub.2] groups = 1 : 0.75 . The castings from the epoxy-siloxane nanocomposites precursors were left to gel and cure at room temperature for 2 days and postcured for 6 h at 80[degrees]C and then 2 h at 120[degrees]C.
Transmission Electron Microscopy (TEM). The morphological structures of the nanocomposites were characterized using a TEM-100CX apparatus, manufactured by JEOL. Examinations were made on thin slices microtomed from cast films embedded in epoxy resin.
Small Angle X-ray Scattering. A point-beam Cu X-ray source was utilized for small angle X-ray scattering (SAXS) measurements in conjunction with a GADDS 2D detector from Bruker (Billerica, MA). The scattered beam intensity was plotted as a function of the variable q, which is defined as q = 2 sin [theta]/[lambda], where [theta] is the scattered angle and [lambda] is the wave-length of the X-rays.
Thermogravimetric Measurements. Thermogravimetric analysis (TGA) was employed to estimate the actual silica content of the hybrid materials produced. Weight losses were measured with a Mettler TG50 thermobalance operating under air-flow. The samples were heated from room temperature to 750[degrees]C at a constant rate of 20 K/min.
Dynamic Mechanical Analysis (DMTA). Dynamic mechanical tests were carried out on a DMTA Thermal Analyser from Polymer Laboratories (UK) model MK III. The samples were in the form of specimens ~3 cm long, ~5 mm wide, and ~1 mm thick. The tests were performed in the scanning temperature mode from room temperature to 200[degrees]C at a constant rate of 5 K/min and with an oscillating frequency of 1 Hz.
Solvent Absorption Tests. Specimens ~20 mm long, ~10 mm wide, and ~0.5 mm thick of the samples were immersed in THF (tetrahydrofuran) and methanol to determine the solvent absorption characteristics as a complementary evaluation of the nature of the networks. The weight of the samples was measured as a function of time and the weight increase was recorded on an analytical balance after blot-drying the specimen. The samples were subsequently dried in an air circulation oven at 80[degrees]C for 2 h and the solvent absorption tests were repeated. In some cases this was done several times to determine whether there were any cumulative effects.
[FIGURE 1 OMITTED]
Thermal Expansion. Thermo-mechanical tests (TMA) were carried out using a DuPont (Wilmington, DE) 990 Thermal-Analyzer. Samples in the form of specimens 5 mm long, 5 mm wide, and about 1 mm thick. In the test the specimens were heated at a constant rate of 10 K/min from room temperature up to 150[degrees]C while a probe detected the linear expansion of the samples.
RESULTS AND DISCUSSION
The TEM micrographs in Fig. 1 indicate that the epoxy-silica nanocomposites display the typical features of organic-inorganic hybrids, which consist of diffuse silica domains dispersed within an organic matrix [9, 10, 13]. In the same figure a schematic elaboration is also shown of the morphological structure, which would be observed at higher resolutions in the TEM observations (more details below).
The SAXS plots of the scattering intensity ([I.sub.x]) against the structure parameter q (Fig. 2) confirm that these domains are co-continuous and display the fractal features of non-segregated phases. The gradient (fractal dimension, F.D.) is in the region of 3.4, as compared to an F.D. of 1.3 reported for nanocomposites containing dense silica particles  and near to zero for the epoxy matrix in the relevant length scale. The characteristic scattering intensity peak at q = 0.016, corresponding to a structural form factor of 63 [Angstrom], is to be regarded as an artifact rather than arising from scattering domains. The reduced intensity of this peak for the epoxy-silica hybrids suggest that it could be associated with the presence of low cross-linking density networks resulting from the presence of minor amounts of high molecular weight epoxy resin .
[FIGURE 2 OMITTED]
TGA measurements confirmed that the silica content after pyrolysis corresponds to the theoretical value, calculated from the composition of the precursor mixture. Furthermore, the TGA thermograms displayed the characteristic shift of the weight loss curve to higher temperatures, which is usually attributed to the presence of co-continuous "diffuse" silica domains [27, 28]. The term "diffuse" is used to indicate that the silica network is not fully formed, owing to the presence of residual SiOH and SiOR groups within the network.
Since the total siloxane content of the nanocomposites is much smaller than the epoxy components, and there are relatively large amounts of grafted trialkoxysilane groups (derived from the GOTMS in the TEOS precursor mixture and the amine silane functional groups in the epoxy resin), the resulting products are expected to contain large proportions of interpenetrating network (IPN) structures. These constitute the interphase between the epoxy-amine network and smaller amounts "diffuse" silica domains. In the sketches in Fig. 1 the three components of the nanocomposites are indicated, respectively, by light areas for the epoxy network and the very dark areas for the "diffuse" silica domains, while the intermediate shaded area denote the interphase.
The absorption kinetics for the uptake of C[H.sub.3]OH and THF epoxy-siloxane nanocomposites with siloxane contents varying from 0.5-15% w/w theoretical silica are shown in Fig. 3. The system with the lowest siloxane content corresponds to the epoxy resin systems grafted with the silane coupling agent (i.e., no TEOS was added to the resin mixture). These sorption kinetics show that there is a distinct difference in the behavior of the two solvents, which is manifest as a clear two-stage process for the absorption of THF (Fig. 3B) and a "peculiar" uptake of C[H.sub.3]OH before reaching equilibrium (Fig. 3A).
Sigmoid sorption kinetics are not uncommon for glassy polymers and have been attributed to diffusion processes controlled by the accumulation of solvent on specific sites on the surface. The adsorbed solvent will then penetrate into the network to produce a plasticized layer, whose thickness grows with time and advances towards the center of the sample, giving rise to a special type of Case II type absorption, insofar as the linear increase in absorption is preceded by an induction period. This type of absorption kinetics has been observed also by other workers for completely different systems [29-34].
[FIGURE 3 OMITTED]
The absorption kinetics for the uptake of C[H.sub.3]OH do not show the characteristic slow first stage process displayed by THF and the amount absorbed rises rapidly to equilibrium. The initial rate of C[H.sub.3]OH absorption, however, decreases significantly with increasing the siloxane content of the nanocomposite. For compositions with 15% w/w silica the absorption isotherm appears to acquire the features of a two-stage process, concurrently with the disappearance of a distinct maximum.
The equilibrium absorption values, however, decrease only slightly with the increase in nominal silica content. It is important to note that the ~1% w/w reduction in solvent uptake after the maximum is significant and the results are reproducible. Reweighing the dried samples at the end of the absorption tests has confirmed that such reduction in weight is not due to the dissolution and leaching of likely impurities or residual reactants.
It can be hypothesized that, at low silica contents, the maximum in the absorption kinetics arises from the reduction in free volumes within the epoxide network during the ingress of the solvent, which brings about a concomitant reduction in equilibrium solvent uptake. This has to result in an amount of solvent being expelled from the network during the subsequent stages, bringing about a reduction in the weight of the sample. The absorption of C[H.sub.3]OH at low nominal silica contents occurs by a Fickian mechanism, while at 15% w/w silica the density of the siloxane network is sufficiently high to prevent a rapid molecular diffusion into the epoxide domains and, therefore, penetration has to take place through the formation of a plasticized layer.
From these observations it can be deduced that the reduction in free volume within the epoxide network, resulting from the ingress of solvent before reaching equilibrium conditions, is due to physical aging. This arises from the reduction in Tg, which remains at all times greater than the environment temperature. The estimated Tg values of the fully solvated samples by applying the Gordon-Taylor's rule, using the freezing point of the solvents as their respective Tg values, vary from about 50-45[degrees]C with increasing silica content from 3 to 15% w/w silica. (The simplified Gordon-Taylor equation has been found to give reasonable estimates for other plasticized polymers .)
At higher silica contents the rate and extent of physical aging is reduced by the constraints imposed by the surrounding siloxane network to the ensuing decrease in free volumes within the main epoxide network, as evidenced by the vanishing of the maximum in the solvent absorption curves. A similar reduction in the extent of physical aging has been reported by other workers for studies on epoxylayered silicates nanocomposites, evidenced by a monotonic reduction in recovered enthalpy with increasing the nanofiller content .
Physical aging does not take place in this case during the infusion of THF because the Tg of the plasticized layers rapidly reaches values below ambient temperature. (The estimated Tg values of the fully solvated samples vary from -38 to 1[degrees]C when the silica content is increased from 3 to 15% w/w; see also Fig. 6.) (The main features of the solvent absorption data are summarized in Table 1, while the main characteristics of the solvents are shown in Table 2.)
In examining the data related to the suppression of the equilibrium solvent absorption for THF, it is noted that this takes place to a much larger extent than is predicted by the reduction in the volume of epoxy network, and confirms that the siloxane domains produce a constraint on the dilatation of the epoxy network caused by the absorption of solvent. Conversely, the equilibrium amount of absorbed C[H.sub.3]OH is fairly close to the theoretical level based on the estimated reduction of volume of epoxy-network domains present in the O-I hybrid, which is where the solvent will be residing at equilibrium (see discussion below related to data in Fig. 6). This is because the level of C[H.sub.3]OH absorbed at equilibrium is much less than THF and, accordingly, the dilatation constraint imposed by the siloxane domains on the surrounding epoxy network is substantially lower than for the THF case.
[FIGURE 4 OMITTED]
In Fig. 4 the mechanical spectra of the nanocomposites is reported, covering the entire range of compositions used in this study. These spectra indicate that there are, possibly, two stepwise increases in Tg. The first arises as a result of having additional cross-links within the network through the introduction of alkoxysilane functional groups, which can undergo condensation reactions with themselves as well as with hydroxyl groups from the epoxide chains. The second takes place when the amount of TEOS in the precursor solution is increased in order raise the theoretical silica content from 10 to 15% w/w. A similar trend is apparent with respect to the tan [delta] values at around the glass transition temperature. The very large tan [delta] values for the composition with 0.5% w/w silica, corresponding to the silane functionalized epoxy resin without any TEOS being added, are a manifestation of the presence of a looser network, due to the reduction in the amount of epoxy groups and the low yield of cross-links resulting from reactions involving alkoxysilane groups.
Solvent/Organic Network Interactions
The solvent absorption data reveal that for the absorption of THF the siloxane network formed from the amine-silane grafts on the epoxy resin decreases the solvent uptake, from about 80% to 48.3% relative to the unmodified resin, brought about simply by the introduction of the alkoxysilane groups. This results from a large change in solubility parameter for the epoxide network, which reduces the amount of THF that can remain dissolved at equilibrium. The equilibrium amount absorbed decreases further down to 27.6% w/w when the theoretical silica content of the sample is increased to 15% w/w through sol-gel reactions.
The sigmoid curve for the THF absorption resembles the effect of pressure on the successive differential absorption that has been observed by several workers, and can be related to the state of internal stresses in the "unsolvated" regions ahead of the plasticized front [22, 29, 36].
Since the samples used in the sorption experiments are very thin in relation to the surface area, it can be envisaged that the expansion produced by the plasticization of the advancing front takes place only along the diffusion direction, and consequently it creates a state of hydrostatic tension in the central "unplasticized" region of the sample. This will increase the diffusion coefficient and set up conditions for which there is a sudden infusion of solvent, corresponding to the onset of the sigmoidal rise in the second stage of the absorption process. Furthermore, it can be anticipated that the conditions for the sudden jump in solvent absorption rate requires that the advancing plasticized zone, relative to the thickness of the central "unsolvated" zone, reaches a certain (critical) value. This would depend on time (controlled by the diffusion coefficient) and on the dilatability of the epoxy domains (controlled by molecular dynamics of the network and on the constraints imposed by the surrounding siloxane domains).
Since the "induction time" for the two-stage sorption process is related to the velocity of the advancing plasticized polymer front, Vo, by the expression [[tau].sub.i] = Do/V[o.sup.2], where [[tau].sub.i] = induction time, Do = Diffusion coefficient , a plot was produced of [square root of (1/[[tau].sub.i])] against silica content, where the values for [[tau].sub.i] were estimated as the time at which the solvent uptake begins to rise rapidly (see Fig. 3B). The parameter [square root of (1/[[tau].sub.i])] can be denoted as the "front velocity factor." From these plots a linear extrapolation to [[tau].sub.i] = [proportional] (Vo = 0) can be made to estimate the approximate siloxane content required to suppress the second stage of the absorption process. This is shown in Fig. 5A, together with a plot of the "normalized solvent absorption" of THF (mass at equilibrium for the composite relative to that of the silane-grafted epoxy matrix) and "front velocity factor" against "theoretical silica content" of the composite.
Both plots indicate that a distinct relationship exists between these two diffusion-related parameters and the siloxane content of the nanocomposites. From a linear regression analysis of the experimental data it is deduced that a change in gradient occurs at about 10% w/w theoretical silica content for the normalized solvent absorption. It is worth noting that the change in gradient takes place at silica contents at which a large increase in Tg was also observed. The extrapolation of the data in these plots to the limiting conditions predicts that about 38% w/w theoretical silica is required for the front velocity to become equal to zero (i.e., the conditions for the vanishing of the two-stage absorption), and to 74% w/w silica to achieve "total resistance" to solvent penetration. These predictions are, obviously, not precise due lack of a sufficiently large number of experimental results but, nevertheless, they serve to illustrate the validity of the extrapolation procedure.
The plots in Fig. 5B of the "normalized solvent absorption" against "reduced loss factor" [ratio tan [[delta].sub.(composite)]/tan [[delta].sub.(matrix)]] confirm that there is a relationship between the mass of solvent absorbed at equilibrium and the level molecular relaxation in the epoxy network. A linear regression analysis for the THF absorption data results in a straight line passing through the origin. Although the trend line (indicated by the individual data) suggests that the relationship is more likely to be a sigmoidal type, it is inevitable that the curves would have to go either thorough the origin, or even intercept with the abscissa at low values of [tan [[delta].sub.(composite)]/tan [[delta].sub.(matrix)]]. This deduction is based on the hypothesis that there can be no penetration of THF into the material when molecular relaxations in the epoxy network are "highly" or totally suppressed by the constraints imposed by the surrounding silica domains, which corresponds to a condition for which tan [[delta].sub.(composite)]/tan [[delta].sub.(matrix] tends to zero.
[FIGURE 5 OMITTED]
The fact that in the case of the C[H.sub.3]OH absorption a linear extrapolation of the normalized solvent absorption plot does not go through the origin is concordant with the prediction that its infusion does not rely on molecular relaxation within the epoxy network, as was also predicted from the solvent absorption kinetics. On the other hand, if it is accepted that for silica contents in the region of 15% w/w and above the sorption kinetics assume the character of a two-stage solvent absorption process similar to the absorption of THF over the entire range of silica contents, then a steep downward trend for the normalized solvent absorption plot has to result at higher silica contents, passing either through the origin or intercepting the normalized solvent absorption axis at values much closer to the origin than the linear extrapolation (Fig. 5B). The latter would indicate that only a limited amount of solvent would penetrate through molecular diffusion without relaxations within the epoxide network.
The dynamic mechanical spectra obtained on the swollen samples after reaching equilibrium are shown in Fig. 6. These reveal that, for samples with 7.5% w/w theoretical silica content, the Tg is reduced from the original value of 94[degrees]C to about 0[degrees]C after absorbing 34.3% w/w THF (equilibrium amount) and to about 40[degrees]C in the case of C[H.sub.3]OH absorption, for which the equilibrium amount is about 11.7% w/w. In this respect both solvents display the typical plasticization behavior for the epoxy network including the depression and shifting of the [beta] relaxations to lower temperatures, and confirms that the absorbed C[H.sub.3]OH resides predominantly within the epoxide domains.
Structural Changes and Network Constraints
When the solvent absorption experiments were repeated on the dried samples (after having absorbed an equilibrium amount of solvent), there was again a distinct difference in the behavior of the samples towards the two solvents. The effects are shown in Fig. 7 for samples containing 15% w/w Si[O.sub.2], and demonstrate that while in the case of the C[H.sub.3]OH experiments the data for the second cycle absorption are very similar to those obtained for the first cycle, the absorption of THF in the second cycle is significantly different from that in the first cycle. In the latter case, for solvent immersions shorter than those leading to the threshold conditions for the explosive absorption, the rate of diffusion in the second cycle is reduced and the induction time is increased. At the same time the amount of solvent absorbed at equilibrium is appreciably higher than the values obtained in the first cycle.
[FIGURE 6 OMITTED]
[FIGURE 7 OMITTED]
Taking these observations in conjunction with changes in the respective dynamic mechanical spectra obtained on the dried samples after the first absorption cycle, shown in Fig. 8A,B, it can be inferred that there have been some structural changes in the material. For the samples that have absorbed an equilibrium amount of C[H.sub.3]OH, for instance, these are manifested as an upper shift in the temperatures for the tan [delta] peaks with respect to both [alpha] and [beta] transitions curves.
Particularly significant also is the increase in Tg to about 120[degrees]C and the broadening of the spectrum of relaxation times leading to Tg, which can only be associated with an increase in cross-linking density of the constituent networks. The additional cross-linking reactions take place essentially during drying of the solvent, when the temperature is substantially higher than that used for the solvent absorption experiments. If cross-linking reactions were to take place during the solvent absorption cycle there would have been a continual reduction in the weight uptake after the maximum instead of a plateau (Fig. 3A).
It can be postulated that some physical "damage" within the constituent domains can take place, which is responsible for the small increase in solvent absorption at equilibrium. This explanation is substantiated by the flattening and spreading of the [beta] transition, together with the increase in tan [delta] peak values, because both features are indicative of a greater morphological heterogeneity. This type of damage is less likely to occur, on the other hand, by the absorption and desorption of C[H.sub.3]OH because of the lower dilatational effects on the network structure, as the quantities absorbed are substantially less than for THF.
On the basis that the Tg of the samples solvated with C[H.sub.3]OH is always higher than the temperature of the absorption environment, it can be deduced that physical aging can take place during solvent absorption and in the latter stage of the drying cycle. Conversely, the samples solvated with THF will not undergo physical aging during the induction period of the absorption cycle because the Tg of the plasticized layers is lower than the temperature of the absorption environment and, therefore, the maximum in the solvent absorption kinetics is not observed. However, physical aging may occur during the last stages of the drying cycle when the Tg of the plasticized network exceeds the ambient temperature.
[FIGURE 8 OMITTED]
[FIGURE 9 OMITTED]
The stipulated increase in cross-linking density within the constituent domains during cycling solvent absorption is supported by the TMA results. The thermal expansion plots shown in Fig. 9A,B, however, contain some anomalies, primarily due to the loss of absorbed water from the silica domains and, therefore, may lead to some misinterpretations. In previous work it was found that desorption of water from O-I hybrids occurs very rapidly between 60 and 110[degrees]C . In any case, the TMA data show that the linear thermal expansion above Tg is reduced for both solvent-aged samples, even though the difference between the two aged samples may not be significant due to experimental errors.
Additional evidence for the possible occurrence of physical aging induced by solvent absorption is derived from the data in Fig. 10, where it is shown that by repeatedly soaking and drying the samples in THF for periods shorter than the induction time leading to explosive absorption, the rate of solvent uptake reduces progressively after each cycle. This is associated with physical aging in concordance with the findings of other workers who have observed a reduction in the rate of gas diffusion resulting from physical aging of different glassy polymers .
Because the amount of solvent absorption is very small (less than 1%), the glass transition of the solvated fronts is likely to remain at all times above the temperature of the environment, even by considering that the quantity of solvent in the advancing front is actually larger. During drying the solvent is likely to diffuse into the entire bulk of the sample and, therefore, the conditions for physical aging persist during the entire cycle. The possibilities for chemical reactions, on the other hand, are extremely small because there would not be sufficient chain mobility within the networks, as the material is always in the glassy state.
It should be noted that epoxy-silica hybrids with a very low siloxane content were chosen for these experiments in order to minimize the constraints on the relaxations within the epoxy network, thereby producing the most favorable conditions for the occurrence of physical aging. In other words, by limiting the amount of solvent absorbed by the samples to levels vastly lower than the equilibrium values, it is ensured that the Tg remains always well above the temperature of the environment, so that physical aging can take place readily with the ingress of solvent.
Conversely, the reduced rate of solvent absorption below the threshold conditions and the extension of the induction time, displayed by the samples in the cyclic absorption experiments reported in Fig. 7 for THF, cannot be entirely related to physical aging as the samples are swelled to equilibrium. These are conditions in which the solvated samples are in the rubbery state, except towards the end of the drying stage, when the amount of residual solvent is sufficiently small to assist physical aging.
[FIGURE 10 OMITTED]
From the results and discussion that have emanated from this study it can be concluded that the incorporation of continuous nanostructured siloxane domains in an epoxy network brings about a large reduction in the equilibrium absorption of aprotic solvents, such as THF, and a much lower reduction in the absorption of protic solvents, such as C[H.sub.3]OH.
It has also been demonstrated that the absorption of THF in epoxy-silica hybrids (bi-continuous nanocomposites) takes place via a two-stage process, consisting of an initial slow rate diffusion, followed by a step increase to equilibrium. The first stage (induction period) is prolonged with increasing the siloxane content of the nanocomposite, while the step rise in solvent uptake gradually decreases and can be eliminated altogether. Similarly, it is predicted that solvent uptake can be suppressed completely when a certain siloxane content in the networks is exceeded.
It is hypothesized that structural changes within the organic network of epoxy-silica hybrids takes place as a result of solvent absorption. These are manifest as a combination of physical aging and cross-linking reactions within the siloxane component, depending on the amount of solvent absorbed and the environment temperature.
1. L. Mascia, Trends Polym. Sci., 3, 61 (1995).
2. H.H. Wang, B. Orler, and G.L. Wilkes, Macromolecules, 20, 1322 (1987).
3. M. Nandi, J.A. Conklin, L. Salvati Jr., and A. Sean, Chem. Mat., 3, 201 (1991).
4. D. Leavy and L. Esquimus, Adv. Mat., 7, 120 (1995).
5. H. Schmidt, and H. Wolter, J. Non-Cryst. Solids, 121, 428 (1990).
6. B.M. Novak, Adv. Mat., 5, 43 (1993).
7. I.A. David and G.W. Scherer, Chem. Mat., 7, 1957 (1995).
8. L. Matejeika, K. Dusek, J. Pletsil, J. Kriz, and F. Lednicky, Polymer, 40, 171 (1998).
9. M.R. Landry, B.K. Coltran, and C.J.T. Landry, J. Polym. Sci. Part B: Polym Phys., 33, 637 (1995).
10. H.H. Wang, B. Orler, and G.L. Wilkes, Macromolecules, 20, 1322 (1987).
11. J.J. Fitzgerald, C.J.T. Landry, and J.M. Pochan, Macromolecules, 23, 3715 (1992).
12. A. Kioul and L. Mascia, J. Non-Cryst. Solids, 175, 169 (1994).
13. Z. Ahmad, S. Wang, and J.E. Mark, Polym. Prep. (Am. Chem. Soc. Div. Polym. Sci.), 34(2), 745 (1993).
14. J.F. Chen, Z. Ahmad, S. Wang, J.E. Mark, and F.E. Arnold, in Hybrids Organic-Inorganic Composites, Chap. 23, J.E. Mark, C.Y.-C. Lee, and P.A. Bianconi, eds., ACS Symposium Series, 585, 297 (1995).
15. L. Mascia and A. Kioul, Polymer, 36, 3649 (1995).
16. Y. Morikawa, M. Iyoku, M. Kakimoto, and Y. Imai, Polym. J., 24, 107 (1992).
17a. C. Chen and D. Curliss, Mater, Res. Soc. Symp. Proc., 703, 3, (2002).
17b. C. Chen and D. Curliss, Nanotechnology, 14, 23 (2003).
18. L. Martjka, Am. Chem. Soc. Polym. Preprints Div. Polym. Chem, 39, 573 (1998).
19. L. Mascia and T. Tang, J. Mat. Chem., 8, 2417 (1998).
20. L. Prezzi and L. Mascia, Adv. Polym. Technol., 24, 91 (2005).
21. J.S. Vrentas and C.M. Vrentas, Macromolecules, 24, 2404 (1991).
22. J.S. Vrentas and C.M. Vrentas, J. Polym. Sci. Part B Polym. Phys., 30, 1005 (1992).
23. D.Y. Perera, Prog. Organ. Coat., 47, 61 (2003).
24. D.Y. Perera, Prog. Organ. Coat., 24, 299 (1994).
25. H. Lu and S. Nutt, Macromolecules, 36, 4010 (2001).
26. P. Musto, L. Mascia, G. Mensitieri, and G. Ragosta, Polymer, 46, 4492 (2005).
27. L. Mascia and L. Prezzi, J. Mat. Sci. (accepted).
28. C. Xenopoulos, L. Mascia, and S.J. Shaw, High Perform. Polym., 13, 1 (2001).
29. J. Wen and J.E. Mark, J. Appl. Polym. Sci., 58, 1135 (1995).
30. H. Fuijita, Forthschr. Hochplym-Forsch, 3, 1 (1961).
31. J. Crank and G.S. Park, Diffusion in Polymers, Academic Press, London, 87 (1968).
32a. N.A. Thomas and A.H. Windle, Polymer, 19, 255 (1978).
32b. N.A. Thomas and A.H. Windle, Polymer, 23, 529 (1982).
33. P.H. Pfromm and W.J. Koros, Macromolecules, 26, 6141 (1993).
34. C.H. Hansen, Prog. Organ. Coat., 25, 55 (2004).
35. A. Friedman and G. Rossi, Macromolecules, 30, 153 (1997).
36. B.C. Hancock and G. Zografi, Pharm. Res., 11, 471 (1994).
37. P.H. Pfromm and W. Koros, Polymer, 36, 2379 (1995).
Leno Mascia, Luca Prezzi
Institute of Polymer Technology and Materials Engineering, Loughborough University, Loughborough LE11 3TU, United Kingdom
CNR-Institute of Composite Materials and Biomaterials, P.le Tecchio, 80 80125 Napoli, Italy
Correspondence to: L. Mascia; e-mail: L.Mascia@lboro.ac.uk
Contract grant sponsors: International Coatings (Akzo Nobel); Engineering and Physical Science Research Council.
TABLE 1. Extracts from solvent absorption curves in Fig. 3. Theoretical Threshold induction Equilibrium solvent Si[O.sub.2] time (X [10.sup.5] s) uptake (% w/w) content (% w/w) THF THF C[H.sub.3]OH 0.5 4.3 48.3 12.3 3.0 6.1 41.6 11.9 7.5 15.6 34.3 11.7 10.0 21.8 29.8 11.5 15.0 32.0 27.6 11.2 TABLE 2. Characteristics of the solvents. Solubility parameters (MJ/[m.sup.3])[.sup.1/2] Solvent [[delta].sub.(dispersive)] [[delta].sub.(dipolar)] C[H.sub.3]OH 15.1 12.3 THF 16.8 5.7 Solubility parameters (MJ/[m.sup.3])[.sup.1/2] Molar volume Boiling point Solvent [[delta].sub.(H-bond)] ([cm.sup.3]/mol) ([degrees]C) C[H.sub.3]OH 22.3 40 65 THF 8.0 81 67 Freezing point Solvent ([degrees]C) C[H.sub.3]OH -94 THF -108
|Printer friendly Cite/link Email Feedback|
|Author:||Mascia, Leno; Prezzi, Luca; Lavorgna, Marino|
|Publication:||Polymer Engineering and Science|
|Date:||Aug 1, 2005|
|Previous Article:||Entropic analysis of color homogeneity.|
|Next Article:||Gas assisted injection molding of a handle: three-dimensional simulation and experimental verification.|