Temperature
Temperature increases the rate of sorption. The diffusion coefficient has been found to increase with temperature according to the Arrhenius relationship [1981]
(
E/RT)
expD
D= 0 − ….(4.7) where E is the activation energy for diffusion. Bonniau & Bunsell (1981) calculated E to be of the order of 11000 cal/mol for (Diglycidyl ether of bisphenol A) DGEBA epoxy cured with different hardeners. Marsh et al. (1984) in their experiments calculated E for neat resins and composites to be 9500 cal/mol and 9940 cal/mol, respectively. The two values seem to be consistent. In addition, the value of activation energy does not seem to vary too much between neat resins and epoxies glass fiber composites. Verghese et al. (1999) too found that the diffusion coefficient follows an Arrhenius relation with E/R= 4650 K.
At higher temperatures (close to Tg), the sorption approaches that of the Fickian response. However, under extreme conditions (i.e. temperatures close to boiling) irreversible damage to the resin and composites was observed. Ishai [1975] found significant damage to the resin (shown by Scanning Electron Microscopic (SEM) studies of cut samples) and glass fibers at temperatures close to boiling. Significant damages included extensive cracking, debonding and degradation of the glass fibers. Bunsell and Dewimille [1983] studied sorption in DGEBA composites and found similar results. Significant damage was found to reduce fiber regions, which was attributed to imbalance of stress distribution. However, damage at lower temperatures was negligible. Samples immersed in distilled water at room temperature for over 3 years showed no damage to their structure. An important conclusion of their study was that accelerated aging could not be simulated by subjecting the resin to higher temperatures because under these conditions other phenomena (such as permanent damage to
resin) take place. This study also led to the conclusion that at higher temperatures, other factors controlled the sorption process, i.e., cracking/crazing of the resin and degradation of the interfacial layer.
Presence of --OH group in Polymer
Interactions of water with polar groups within the polymer chain of vinyl ester resins have been confirmed through FTIR (Fourier Transform Infrared Spectroscopy) studies on saturated samples. FTIR studies made on vinyl ester resins before and after aging showed changes in the shape of the peak at 1450 cm-1, which was attributed to a stretching vibration of the carbon monoxide and the bonding of the primary alcohol OH bond. Verghese et al. demonstrated the influence of polar OH groups [1999]. Specific interactions through hydrogen bonding were observed in Derakane 441-400 vinyl ester resin composites (which had an OH group in its structure) as against their model resin in which the OH groups was substituted for a CH3 group. The model resin did not exhibit
the thermal spiking phenomenon, while the saturation level of the resin was lower as compared to the Derakane 441-400 vinyl ester resin.
Nature of Water
Water may exist as either “bound” or “unbound” within the system. The existence was confirmed by 2H – NMR studies by Klotz et al. [1996] on diglycidyl ether bisphenol A resin cured with dicyandiamide, and/or diamino-diphenyl sulfone. Woo and Piggott [1987] observed that the effective dielectric constant of water is only 55-77% that of free water. Hence, the mobility of water within the polymer is between that of ice and free water. It was believed that water might cluster within a polymer, and water molecules from hydrogen bonds may become bound with polymer groups. However, contrary theories were postulated [Netravalli et al., 1984].
Changes taking place within the polymer due to the sorption of water may be reversible or irreversible. Weitsmen [1988] and Wong et al. [1985] showed that slow rearrangement of the polymer chains takes place due to the ingress of the water molecule. This rearrangement of polymer chains causes changes in the free volume and hence the second and subsequent sorption is comparatively faster.
Wong and Broutman [Part I and II, 1985] studied the sorption of water into EPON 828 resin (DGEBA cured with m-phenyldiamine/aniline) and concluded that Fickian sorption was taking place although the diffusion coefficient was concentration dependent. They proposed that the water molecules caused polymer chains to rearrange and did not cause any damage to the epoxy. However, this rearrangement was not permanent and the polymer collapsed to its original form when annealed or heated to temperatures greater than its Tg. When this sample was subjected to the sorption experiment after annealing, the behavior was found to be the same as the original sample. The researchers also observed that the post-cured sample did absorb more water.
Due to additional cure, the resin had more fractional free volume and more–OH Groups, which caused increased water absorption. McMaster and Soane [1989] also found the second and third sorption to be faster than the first. The diffusion coefficient for one such result was first sorption, 3.5 x 10-9 cm2/sec and second sorption 4.2 x 10-9 cm2/sec. Hence the history of the polymer has a definite bearing on subsequent sorption. Marsh et al. [1984] reported that the resins retained some moisture on drying. The irreversible changes that may take place are cracking/crazing, and degradation of the matrix/fiber interface are shown by Ishai [1975], and Dewimille and Bunsell [1983]. They took SEM photographs, which clearly show extensive cracking attributed mainly due to imbalance in stress and degradation of the interface. In some cases, cracks developed in the fiber rich zones. This is because the water plasticized the resin and rendered the resin rich zones more resistant to damage.
Netravalli et al. [1984] used (Differential Scanning Calorimetry) DSC technique to obtain the glass transition temperatures and curing energies of wet and dry samples.
reversible. They also found that the curing energies progressively decreased as the amount of water taken in increased. The water was found to cause the unreacted epoxide groups to react, thereby causing some permanent changes in the resin structure.
Cure
The effect of curing temperature and time of cure has been reported by a number of researchers. In all cases, the amount of water sorbed has been found to increase with the curing. Wong and Boutman [1985] found this in their study of DGEBA/m- phenyldiamine/aniline epoxy resin. Sahlin and Peppas [1991] found the post cured TGDDM/DDS resin to consistently sorb more water than the one without the post cure.
The extent of cure of a resin can be found using Torsional Braid Analysis (TBA), Torsion Pendulum and FTIR (Fourier Transform Infrared Spectroscopy). The Tg of a sample increased as the time of an isothermal cure was increased and hence the extent of cure increased. Enns and Gillham [1983] cured DGEBA with stoichiometric amount of DDS at 175°C for four different times: 50, 100, 180 and 600 minutes. Six hundred minutes was not enough to fully cure the sample, as the Tg was below the Tg∞ (approx.
215 °C) of the fully cured sample. The samples were then subjected to a humid environment at 25 °C. Four different humidities were used, 31%, 51%, 79.3% and 93%. The result of the sorption experiments was that the sample with the highest percent of cure absorbed more water at all the conditions. More cured resins exhibit higher sorption because the highly cross-linked specimen has a lower density and consequently a greater free volume. Similarly, a greater cure leads to the formation of more –OH groups. A typical result is shown in Table 4-2.
Table 4-2: Variation of Equilibrium Moisture Content with Degree of Cure Cure time (minutes) Tg °C Density (g/ml) M∞ D x 10 9 cm2/s 50 135 1.237 1.074 1.69 100 160 1.2357 1.136 1.7707 180 185 1.2339 1.65 1.58885 600 195 1.2334 1.737 1.7170
From the viewpoint of mechanical properties, it is desirable to have a fully cured polymer. This is because a partially cured resin has a lower modulus compared to a fully cured resin, and it displays more creep. The most common method of determining the extent of cure is with the help of a differential scanning calorimeter (DSC). During the course of a curing reaction, heat is liberated, and the instantaneous rate of energy evolution can be measured using a DSC. The extent of reaction, α , then is simply the ratio of the total heat evolved to the heat of reaction. When dα dtis plotted as a function
of time, t, it is found (Gupta, 2000) that data can be described by the following equation.
(
m)
(
)
n 2 1 k 1 k dt d α − α + = α …..(4.8) in which the ks are temperature-dependent rate constants while m and n aretemperature independent constants.
White and Mather [1991] used an ultrasonic cure monitor technique to assess the simultaneous extent of cure and mechanical property development during the cure on an epoxy resin EPON 815/V 140 and compared the results with DSC monitoring. The modulus extent was derived and presented as a characterization parameter similar to the degree of cure in thermal cure characterization. The results have shown that the degree of cure does not accurately reflect the mechanical property development during
significant mechanical property development is still occurring in the later stages of cure when the degree of cure is fully developed. In other words, if the goal is to determine how close a sample is to being fully cured, the ultrasonic method is likely to provide the needed information while the thermal method may not provide such information. The former method is a non-destructive method.
Hardener
A hardener is used in the cure of an epoxy resin to cross-link the epoxy chains, thus giving structural rigidity. The commonly used hardeners are Diamino Diphenyl Sulphone (DDS), diphenyl diamine, dicyandiamine (DICY), anhydride and Lewis acid hardeners.
Due to their polarity, hardeners influence the sorption of water. Sahlin and Peppas [1991] cured TGDDM resin with different amounts (5, 15, 25, 35, 45 wt %) of DDS. The sorption increased as the amount of the hardener was increased. Within a reasonable deviation, the increase was linear with the amount of water. The polarity of the hardener also had an effect.
Diamant et al. [1981] attempted to keep the polarity of the resin a constant by replacing the diamine hardener with aniline in DGEBA (Figure 4-9). By keeping the polarity a constant, morphology of the polymer has changed as the length of the matrix between cross-linking increased, thus reducing the density of cross-links in the resin without affecting the polarity and increased chain mobility.
Figure 4-9: Difference Between (a) Diamine and (b) Aniline Hardener
Contrary to expectations, the resin cured with only diamine hardener was observed to have the maximum sorption, which is attributed to (and confirmed by etching experiments) regions that had a greater cross-linking density, which surrounded areas of lower density. Higher cross-linking density caused hindrance to the movement of the water molecules and effectively reduced the sorption. However, the inherent polarity of diamine and aniline was not taken into account. Diamine is more polar than aniline and hence could have accounted for the increased uptake. In addition to polarity of individual hardeners, the presence of excess hardener increases the affinity of the resin towards water. A residual hardener, for example, DICY means greater affinity for water. DGEBA was cured with 5, 15 and 25 phr (parts per hundred parts of resin) of triethylene tetramine (TETA) at 100 °C. The results are tabulated in Table 4-3 [1982].
Table 4-3: Effect of Hardener on Equilibrium Moisture Content
%TETA
in DGEBA Tg (dry) °C Tg (Wet) °C M∞ % (70 °C) M∞ % (20 °C)
5 109 105 1.92 1.5 15 142 109 3.3 2.7
25 95 52 8.6 10.8
It is clear that the amount of water absorbed increased with the amount of the hardener and temperature. However there was a discrepancy from expected Tg. Thus, important parameters that influence sorption of water are:
• Extent of cure, like curing temperature and time of cure.
• Type of hardener used to cure the resin as well as the amount used. • Amount of free volume present in the resin.
• Sizing agent (whether it forms a good bond or not between the matrix and fiber).
• Environment like pH, temperature, etc. • Effect of Glass Reinforcement
Influence of glass fibers in polymers on the sorption of water can be determined by sorption experiments on neat samples as well as composites. The polymer resin used should be cured with the same hardener/catalyst as well as under the same conditions.
Contradictory results have been obtained on the above subject. Marsh et al. [1984] studied the sorption of water in bisphenol A and cresol novolac epoxy cured with dicyandiamide. Neat resins and composites with 40% E-type glass were studied at 75°C /100% RH. The sorption of water in the glass composite was the same as that of the neat resin. Both the neat resin as well as the glass composite showed an intermediate saturation before the onset of residual moisture. These similarities led the authors (Marsh et al., 1984) to the conclusion that water did not enter the interface between the matrix and the fiber; hence, there was no difference in the sorption of neat resin and composites. On the other hand, Ishai [1975] showed that the behavior of neat resins was quite different from composites. In the case of diffusion of moisture into Epon 828 resin with E-glass fibers, significant damage was found not only to the
interface but also to the glass fiber (confirmed by SEM pictures). The degradation was signaled by a drastic change in the sorption curve. Plotting the Relative Weight Change (RWC) Vs. Relative Length Changes (RLC) also showed the onset of degradation. The extent of degradation was more for samples exposed to extreme environments, i.e., temperatures of 80°C. Also the degradation of the interface was considerably less at lower temperatures i.e. 20°C. Similar results were obtained by Dewimille and Bunsell [1983], who found that composites (DGEB/Anhydride with E-Glass) degrade when exposed to water at high temperatures (80°C and above).
Similar sorption studies were also made on vinyl ester glass fiber reinforced composites. Pai et al. studied the effect of glass fiber lay-up sequencing in various acidic environments [Parts I and II, 1997], using six different types of resins including vinyl esters. They found that the composite with the Chopped Strand Matrix exhibited least resistance to all liquids (water, 15%, 25% and 35% Sulfuric acid). Although the diffusion process became sluggish as the concentration of sulfuric acid increased, the saturation levels were much higher. They also studied the extent of degradation on the composite. They assessed the fiber/matrix interface by performing an interlaminar shear strength Test. The loss of the shear strength increased as the concentration of sulfuric acid increased, showing an increased rate of degradation with increased acidic pH.
In order for a composite to function properly, there must be a chemical bond between the matrix and the reinforcing fibers so that the applied load (applied to the matrix) can be transferred to the fibers. In 'fiber glass' the fiber is inorganic while the matrix is organic, and these two do not bond readily unless the fibers are treated to modify their surface. Silica (SiO2) is hygroscopic, i.e., however slow it absorbs water
onto its surface where the water breaks down into hydroxyl (-OH) groups. The coupling agent takes the form of a silane (R-SiX3), where R is an organic radical that is
compatible with the polymer matrix and X is a hydrolisable organic group such as an alcohol. The most common silane couplant is tri-ethoxy-silane. Heat will force the elimination of water between the -OH pairs at the hydrated silica surface and the silane
as well as between the adjacent silane molecules forming a strong bond between the matrix and the fibers as shown in Figure 4-10.
Figure 4-10: Bonds Between Glass Fiber and Coupling Agent.
If the bonds are chemical, then the presence of the glass will have a negligible effect. If, however, the bonds are weak and can be displaced by the hydrogen bonding due to water, debonding and degradation of the glass fiber can be pronounced. This was studied by Ritter et al. [1998]. They studied the propagation of a crack in monolithic glass (soda-lime and fused silica) untreated glass-epoxy interface and glass- epoxy interface sized with 2-amino propyl triethoxy silane (3-AMPS) using a double cleavage drilled compression test.
The increase in resistance of the silane bonded epoxy interface is attributed to chemical bonding of epoxy to the glass via the coupling agent. The rupture or debonding will take place only by breaking of the Si-O-Si bonds formed between the glass and the silane-coupling agent. However, the highest threshold energy release rate, Gth for the soda-lime glass is lowered as the alkali molecules can bond with water and hence prevent the silane molecules from properly bonding to the glass surface. The Gth calculated is given in the Table 4-4.
Specimen Gth Jm-2
Soda-Lime glass (SLG) 1.69
Fused Silica (FS) 2.76
Untreated SLG-Epoxy 0.25
Untreated FS-Epoxy 0.25
Silane treated SLG-Epoxy 1.32
Silane treated FS-Epoxy 3.31