Free radical polymerization aided self-healing

Materials and reagents

The epoxy resin, diglycidyl ether of bisphenol A (trade name: E-51), was provided by Dongfeng Chemicals Co. Ltd, China. Its hardener, tetraethylenepentamine (TEPA), was supplied by Tianjin Damao Chemical Reagent Factory, China.

Melamine, analytically pure, was supplied by Shanghai First Chemical Co., China. Formaldehyde, analytically pure, 37 wt%, was supplied by Guangzhou Chemical Co., China. Styrene, maleic anhydride, and the radical initiators, benzoyl benzenecarboperoxoate (BPO) and azobisisobutyronitrile (AIBN), were purchased from Guangzhou Chemical Reagent Factory, China. The accelerant, cobaltous naphthenate, was provided by Shantou Guanghua Chemical Factory, China.

Preparation of sodium styrene–maleic anhydride copolymer emulsifier

Styrene (11.02 g), maleic anhydride (10 g), AIBN (0.11 g), and acetone (250 mL) were charged into a 1000-mL round bottom flask fitted with a condenser. The mixture was heated to 30°C–40°C for 3–4 h and then cooled down to room temperature. A measure of 200 mL of diluted NaOH (10%) was added to the system, which was further heated to 80°C with stirring. By adjusting pH value of the mixture to 5–6, the reaction was stopped. When the solvents were removed, the remained emulsion was the product we needed.

Preparation of styrene-loaded microcapsules

Cobaltous naphthenate (4.4 g) was dissolved in styrene (44.48 g) together with epoxy acrylate (7.2 g, which helps to reduce volatility of styrene during encapsulation). Afterward, the aforesaid sodium styrene–maleic anhydride copolymer (SMANa) emulsifier (14.6 mL) and gum arabic (2.3 g) were incorporated into the mixture under agitation. The prepolymer of melamine (17.5 g) and formaldehyde (28.5 g) formed at 80°C and pH = 8–9 was poured into the system with stirring. By dropwise addition of 5% citric acid, pH value of the system became about 5–6. The reaction proceeded at 50°C for 3 h. Eventually, the resultant spheres were thoroughly washed with deionized water for three times and dried. For more details about the microcapsules preparation, refer to Wang et al. (2009).

Preparation of BPO-loaded microcapsules

The BPO-loaded capsules were prepared in a way similar to that for making styrene-loaded capsules. BPO (4.42 g) was dissolved in toluene (39.8 g) together with deionized water (400 mL), SMANa emulsifier (14.5 mL), and gum arabic (2 g). Having been rigorously stirred at room temperature for 8 min, the prepolymer of melamine and formaldehyde was added. Similarly, 5% citric acid was dropwise incorporated to the system, leading to pH value of 5–6. After 4–5 h at 50°C, the produced capsules were collected, washed, and dried.

Preparation of self-healing epoxy materials

The unfilled epoxy specimens were fabricated by mixing E-51 epoxy with its curing agent TEPA at a weight proportion of 100/14. The self-healing epoxy composites were prepared by uniformly mixing certain amount of styrene- and BPO-loaded microcapsules together with the aforesaid mixture of E-51 and TEPA. To obtain cured versions, both the unfilled epoxy and filled compounds were degassed and poured into closed silicone rubber molds and cured for 7 days at room temperature.

Characterization

The microcapsules’ core contents were determined by elemental analysis (Vario EL, Germany Elementar Inc., Hanau, Germany) and acetone extraction method, following the procedures described elsewhere (Yuan et al., 2008a). Their sizes and size distributions were measured by Mastersizer 2000 (Malvern Instruments Ltd., Worcestershire, UK), laser size analyzer. Fourier transform infrared (FTIR) spectra were collected by a Nexus 670 machine (Thermo Fisher Scientific, Waltham, USA). Thermal stability of the capsules was studied by a Netzsch TG-209 thermogravimetric analyzer in N₂ at a heating rate of 20°C min⁻¹. Reactivity of the encapsulated BPO was assessed by a TA MDSC 2910 (TA Instruments, New Castle, USA) differential scanning calorimetric (DSC) equipment.

Morphology of the microcapsules and fracture surface of the self-healing composites were observed with Philips XL30 FEG (FEI Company, Hillsboro, USA). scanning electron microscope (SEM).

To examine healing ability of the healing system, the protocol proposed by Jones et al. was used, who carried out impact tests to assess matrix healing (Hayes et al., 2007)

where HE stands for efficiency of healing, and E _heal and E _init are the impact energies of healed and virgin materials, respectively. According to ASTM D256-034, notched Izod impact tests were conducted on an advanced pendulum impact (API) tester at a rate of 3.8 m s⁻¹. After testing, the samples that had been broken into two pieces were kept in alignment and intimate contact for healing at 25°C for 24 h. Then, the healed specimens were tested again to check the effect of healing. For calculating efficiencies of repeated impact–repair tests, the impact energy of samples that had not been impacted previously (i.e. before the first healing event) serves as E _init .

Optimization of the preparation processes of styrene- and BPO-loaded microcapsules

Since styrene is highly volatile and BPO is thermally sensitive, the encapsulation reactions cannot be carried out at higher temperature. Therefore, the mild two-step approach using melamine–formaldehyde resin as the wall substance was applied (refer to section “Experimental”). To obtain the optimal products, some important processing parameters should be investigated.

It is known that each melamine molecule contains three amino groups. In principle, 1 mol of melamine can react with 6 mol of formaldehyde yielding hexamethylol melamine. The subsequent condensation of hexamethylol groups results in cross-linked polymer consisting of either methylene or dimethylene ether linkages. It is understood that the cross-link density and contents of methylene or dimethylene ether linkages depend on the molar ratio of melamine to formaldehyde. Figure 1 shows the pyrolytic temperatures at 1% weight loss of the polymerization products of melamine and formaldehyde as a function of their molar ratios, which can serve as a qualitative measure of cross-linking density and composition of the cross-linked polymer. It is seen that when formaldehyde to melamine molar ratio ranges from 1 to 3, the pyrolytic temperature of the resultant resin increases as a result of raised cross-linking density. In principle, a further increase in the molar ratio (4–6) may lead to higher cross-link density, but the gradual decrease of pyrolytic temperatures is indicative of increased dimethylene ether linkages that can be easily cut off at higher temperature. Consequently, the proper formaldehyde to melamine molar ratio should be 2.5–3.

OPEN IN VIEWER

Figure 1.

Thermal decomposition temperatures at 1% weight loss of melamine–formaldehyde resin versus molar proportions of the monomers.

The pH value of the reaction system is an important factor for quality control of the ultimate product. It was found that the prepolymer of melamine and formaldehyde had better to be synthesized at a pH value of 8–9 (Wang et al., 2009). The obtained dimethylol melamine and trimethylol melamine favored further condensation at lower pH value. For the subsequent condensation process, however, higher catalyst concentration (i.e. lower pH value) may accelerate the reaction, so that there is not enough time for the polymer to encapsulate styrene or BPO solution. Depositions of melamine resin have to be produced. Contrarily, higher pH would slow down the condensation reaction and hence induce incomplete encapsulation. According to the results of our systematic research, the encapsulation should proceed at a pH value of 5–6.

Our previous studies suggested that size of microencapsulated healing agent is closely related to the healing performance (Yuan et al., 2009). In this regard, species and dosage of emulsifier should be carefully chosen. First, we used sodium dodecylbenzene sulfonate (SDBS) and found that its emulsification is not satisfied as characterized by the heterogeneous coverage of the wall materials on the capsules. We also tried Tween 80, which resulted in extremely low reaction rate and thin wall shell of the capsules, because it is electroneutral and has poor affinity to the positively charged prepolymer of melamine and formaldehyde.

When SMANa and the core substance are mixed in water, oriented arrangement of the emulsifier molecules surrounding the hydrophobic core substance would take place. That is, an adsorption layer is built up on each core droplet with inward aromatic rings and outward carboxylic groups. Consequently, a negative electric field appears at the outer surface of the core droplets, which prevents the droplets from coalescence and attracts the wall formers with positive charges. Polycondensation of the prepolymer of melamine and formaldehyde happens right at this region. In the course of emulsification, adequate emulsifier dosage should be used to separate the hydrophobic core substance into tiny droplets. Besides, gum arabic is a nonionic surfactant and is able to reduce oil–water interfacial tension. Oil droplets can be stabilized by the gum arabic adsorbed on their surfaces. Figure 2 shows the influence of emulsifier concentration on size and size distribution of the resultant microcapsules. With a rise in emulsifier concentration, the capsules’ size is reduced and the size distribution becomes narrower, and then the changes are not significant. This can be explained by the role of emulsifier as discussed earlier. That is, when all the core droplets are covered by the emulsifier layer with sufficient thickness, the excessive emulsifier naturally cannot take effect.

OPEN IN VIEWER

Figure 2.

Size distribution of (a) styrene- and (b) BPO-loaded microcapsules as a function of emulsifier (SMANa+gum arabic) concentration. The weight ratio of SMANa to gum arabic is fixed at 7.

Characterization of styrene- and BPO-loaded microcapsules

Morphology of the microcapsules is shown in Figure 3. Many small particles are adhered to the capsules’ surface. It reflects that the formation of the microcapsules is actually the accumulation of tremendous tiny particles of melamine–formaldehyde resin onto emulsified core droplets. Figure 3(c) clearly reveals the wave-like surface originating from compact stacking of the particles. According to the observation of the broken capsules (Figure 3(e) and (f)), the wall thickness is about 0.1 µm. On the whole, the microcapsules with melamine–formaldehyde resin as the wall material have relatively rough exterior surface, which would help to enhance interfacial interaction with the matrix in the subsequent manufacturing of self-healing materials.

OPEN IN VIEWER

Figure 3.

SEM images of (a, c, and e) styrene-loaded microcapsules and (b, d, and f) BPO-loaded microcapsules.

Figure 4 gives FTIR spectra of the microcapsules, core, and wall substances. The wall substance was obtained from the ground microcapsules that had been thoroughly extracted and dried. Meantime, the core substance was isolated from the extractant (i.e. acetone). On the spectra of both styrene-loaded microcapsules and their wall shell (Figure 4(a)), there is a strong absorption at 3357.3 cm⁻¹ due to stretching mode of N–H. It is worth noting that a weak peak is also perceived nearby (3357.3 cm⁻¹) on the spectrum of the core material, which can be attributed to the characteristic absorption of vinylbenzene. In addition, the peak at 1630 cm⁻¹ represents the stretching of C = C, while those at 1365 cm⁻¹ are assigned to aromatic ring. The absorptions at 1744 and 1215 cm⁻¹ result from C = O and C–O of epoxy acrylate (which helps to increase viscosity of styrene to certain extent, see section “Experimental”). The results suggest that the produced styrene-loaded capsules are indeed melamine–formaldehyde resin walled with the inclusion of styrene. On the other hand, we can also find the peak at 3357.3 cm⁻¹ on the spectra of BPO-loaded microcapsules and the wall shell, which is absent on the spectrum of the core (Figure 4(b)). In contrast, the characteristic absorption of C = O at 1759 cm⁻¹ and that of O–O stretching at 999 cm⁻¹ appear on the spectra of the capsules and core. Evidently, BPO/toluene solution has been encapsulated by melamine–formaldehyde resin as expected.

OPEN IN VIEWER

Figure 4.

FTIR spectra of (a) styrene- and (b) BPO-loaded microcapsules in comparison with those of the wall shell and core substances.

Thermal stability of the microcapsules is closely related to the manufacturing of self-healing composites. Figure 5 illustrates pyrolytic behaviors of the capsules. Both styrene- and BPO-loaded microcapsules exhibit low onset temperatures. This is because of the high volatility of the core substances, which tend to penetrate the wall shells and evaporate at elevated temperature. As a result, gradual weight loss is perceived. When temperature approaches 400°C, a drastic decrease in weight appears due to the decomposition of the wall material. It is interesting to find that the onset temperatures of the BPO-loaded microcapsules near 400°C are higher than that of neat melamine–formaldehyde resin. Probably, the presence of the core BPO takes the responsibility, because the decomposed BPO would initiate cross-linking of the capsules wall and hence improve its thermal resistance.

OPEN IN VIEWER

Figure 5.

Thermal decomposition behaviors of (a) styrene- and (b) BPO-loaded microcapsules with different sizes in N₂ in comparison with that of the wall material.

Compared to styrene, BPO is more active as viewed from their responses to heating. Considering that the microcapsules are prepared at 50°C, it should be confirmed whether the activity of BPO is maintained after encapsulation. Accordingly, BPO-loaded microcapsules were ground at room temperature and then mixed with styrene at 0°C. By heating the mixture in DSC pans, the exotherm resulting from thermal polymerization of styrene was recorded (Figure 6(a)). Clearly, the BPO inside the microcapsules is able to initiate polymerization of styrene as reflected by the obvious exothermic peaks at different heating rates, which resemble the curves of styrene/as-received BPO system (Figure 6(b)). Moreover, a quantitative treatment of the data in Figure 6(a) in terms of the well-known Kissinger method (Kissinger, 1957) gives the activation energy of 64.0 kJ mol⁻¹ (correlation coefficient of the linear regression = 0.999), which is comparable to the value estimated from Figure 6(b), that is, 57.7 kJ mol⁻¹ (correlation coefficient of the linear regression = 0.995) (Jiang, 2008). It can thus be concluded that the activity of BPO is nearly not affected by the encapsulation process.

OPEN IN VIEWER

Figure 6.

DSC heating curves of the polymerization of styrene initiated with (a) BPO released from the ground BPO-loaded microcapsules (core content = 50 wt%) and (b) as-received BPO. The weight ratio of the BPO-loaded microcapsules to styrene is 1:5.

Self-healing performance of epoxy filled with the dual capsules

On the basis of the previous study, we prepared self-healing epoxy composites with embedded styrene- and BPO-loaded microcapsules. Healing efficiency of the materials is plotted as a function of the capsules content in Figure 7, while the control composites containing individual styrene- or BPO-loaded microcapsules at the same concentrations do not offer any healing effect. The results indicate that the healing system based on styrene/BPO pair works. In addition, the healing effect is controlled by the relative amount of the two types of microcapsules. Basically, a maximum healing efficiency of about 65% is detected at 15 wt% styrene-loaded capsules and 3 wt% BPO-loaded capsules. This can be explained by the fact that when BPO dosage is excessive, the resultant PS has short chains and low molecular weight. The adhesive strength of the bonding material at the cracked interface has to be low, so that the healing efficiency is not that high. In case of excessive styrene-loaded microcapsules, the insufficient BPO leads to rather low reaction rate or even oligomer instead of macromolecules. The measured healing efficiency has to be low again.

OPEN IN VIEWER

Figure 7.

Healing efficiency of self-healing epoxy composites containing styrene- and BPO-loaded microcapsules as a function of (a) content of styrene-loaded capsules at a fixed content of BPO-loaded capsules of 3 wt% and (b) content of BPO-loaded capsules at a fixed content of styrene-loaded capsules of 15 wt%. Size and core content of the styrene-loaded capsules are 30 µm and 40 wt%, respectively. Size and core content of the BPO-loaded capsules are 50 µm and 45 wt%, respectively.

We also investigated the ability of multirepair of the materials. It is seen that the self-healing epoxy offers detectable healing effect even after the third failure. Compared with the data of the first healing, the healing efficiencies of the second and the third repairs are reduced by about 20%–30%. It means that some healed parts might be stronger than the matrix. Under the second or the third impact, cracks have to go round the specific portions and partially travel along new route with unbroken microcapsules. As a result, these capsules take effect, and the cracks are partly healed showing the capability of multirepair.

Figure 8 shows the healing efficiency of control unfilled epoxy specimens, which were healed by manually injecting premixed styrene and BPO onto the impact fracture surfaces. The results of the reference specimen testing are believed to provide an upper limit for the healing efficiency under ideal conditions and a benchmark for comparison to specimens in which the healing agent is embedded (Kessler and White, 2001). The data in Figure 8 indicate that the healing efficiency increases with increasing styrene/BPO ratio within the range of interest. Nevertheless, the healing efficiency of the reference specimens has not yet approached 100%. It seems that (a) the high volatility of styrene and (b) the lower strength of PS than that of matrix epoxy might account for the unsatisfied healing effect. The problems can be solved by increasing the content of thickening reagent like epoxy acrylate and incorporation of reinforcements like carbon nanotubes into the styrene-loaded microcapsules, which will be investigated in the future study.

OPEN IN VIEWER

Figure 8.

Healing efficiency of control unfilled epoxy specimens that were repaired by manually injecting premixed styrene and BPO.

The fracture surface of the self-healing epoxy composites is shown in Figure 9. The spherical concaves are left by the broken healant-loaded microcapsules, while the membranes on the matrix result from polymerization of the released styrene (Figure 9(a)). It means that the core materials had been liberated to the cracked parts and consolidated forming adhesive films that tightly bond the cracks. A magnified view of the broken capsules indicates that tail-like pattern appears near the edge of the concaves (Figure 9(b)). Crack pinning mechanism must take effect during the impact failure, which would facilitate energy consumption and increase toughness of the matrix.

OPEN IN VIEWER

Figure 9.

Surface of refractured healed self-healing epoxy composites containing 15 wt% styrene-loaded and 3 wt% BPO-loaded microcapsules. Size and core content of the styrene-loaded capsules are 30 µm and 40 wt%, respectively. Size and core content of the BPO-loaded capsules are 50 µm and 45 wt%, respectively. (a) Membrane of consolidated healing agent from broken capsules. (b) Magnified view of a broken capsule.