self-healing epoxy composites – preparation and effect of the healant consisting of...

COMPOSITES

www.elsevier.com/locate/compscitech

Composites Science and Technology 67 (2007) 201–212

SCIENCE ANDTECHNOLOGY

Self-healing epoxy composites – Preparation and effect of the healantconsisting of microencapsulated epoxy and latent curing agent

Tao Yin a, Min Zhi Rong b, Ming Qiu Zhang b,*, Gui Cheng Yang a

a Key Laboratory for Polymeric Composite and Functional Materials of Ministry of Education, OFCM Institute,

School of Chemistry and Chemical Engineering, Zhongshan University, Guangzhou 510275, PR Chinab Materials Science Institute, Zhongshan University, Guangzhou 510275, PR China

Received 20 June 2006; accepted 27 July 2006Available online 26 September 2006

Abstract

To provide epoxy based composites with self-healing ability, two-component healing system consisting of urea–formaldehyde micro-capsules containing epoxy (30–70 lm in diameter) and CuBr2(2-MeIm)4 (the complex of CuBr2 and 2-methylimidazole) latent hardenerwas synthesized. When cracks were initiated or propagated in the composites, the neighbor microencapsulated epoxy healing agentwould be damaged and released. As the latent hardener is soluble in epoxy, it can be well dispersed in epoxy composites during com-posites manufacturing, and hence activate the released epoxy wherever it is. As a result, repair of the cracked sites is completed throughcuring of the released epoxy. The present paper studied the preparation of epoxy microcapsules by amino resins, and the influencingfactors as well. On the basis of this work, mechanical properties of the epoxy filled with the healing system were evaluated. It was foundthat incorporation of the two-component healing system nearly did not change the fracture toughness of the neat epoxy, as indicated bythe single-edge notched bending test. In the case of 10 wt% microcapsules and 2 wt% latent hardener, the self-healing epoxy exhibited a111% recovery of its original fracture toughness. Besides, the preliminary result of double-cantilever beam testing showed that the plainweave glass fabric laminates using the above self-healing epoxy as the matrix received a healing efficiency of 68%.� 2006 Elsevier Ltd. All rights reserved.

Keywords: A. Polymer–matrix composites (PMCs); A. Smart materials; B. Curing; B. Matrix cracking

1. Introduction

Long-term durability and reliability are critical for poly-mer–matrix composites used in structural applications [1].The exposure to harsh environment would lead to thedegradations of components made from these materials.Comparatively, microcracking is one of the fatal deteriora-tions generated in service, which would bring about cata-strophic failure of the composites and hence significantlyshorten the lifetimes of structures. Considering that thedamages inside the composites are difficult to be perceivedand to repair in particular, the materials had better to havethe ability of self-healing.

0266-3538/$ - see front matter � 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.compscitech.2006.07.028

* Corresponding author. Tel./fax: +86 20 84036576.E-mail address: [email protected] (M.Q. Zhang).

So far, the achievements in this aspect fall into two cat-egories: self-healing without and with the aid of healingagent. The works by Chen et al. represent the formerschool [2,3]. They synthesized highly cross-linked poly-meric materials with multi-furan and multi-maleimide viaDiels–Alde reaction. At temperatures above 120 �C, the‘‘intermonomer’’ linkages disconnect but then reconnectupon cooling. This process is fully reversible and can beused to restore fractured parts of the polymers. However,the low glass temperatures of these polymers (30–40,80 �C) [3] obstruct them from replacing epoxy, the thermo-setting polymer widely used as the matrix of advancedcomposites.

For healing agent aided self-mending, the agent shouldbe liquid at least at the healing temperature. It is generallyencapsulated and embedded into the composites’ matrix.

mailto:[email protected]

202 T. Yin et al. / Composites Science and Technology 67 (2007) 201–212

As soon as the cracks destroy the capsules, the healingagent would be released into the crack planes due to capil-lary effect and heals the cracks. According to the composi-tions, the healing agents can be classified as single-component and two-component ones. The single-compo-nent healants, like cyanoacrylate [4,5] and polyvinyl acetate[6], are able to be cured under the induction of air, andhence not suitable for healing damages deep in the compos-ites. The two-component system consists of polymerizableresin and hardener. When they meet, polymerization isactivated so that the cracked parts can be bonded. Usually,encapsulation of the healing agents is conducted using frag-ile-walled containers. Dry filled the glass pipette tubes witha one-part cyanoacrylate and a two-part epoxy adhesive,respectively [4,5]. Similar approach was adopted by Mot-uku et al. [7] and Zhao et al. [6]. Because the hollow glasscapillaries have diameters (on millimeter scale) much largerthan those of the reinforcing fibers in composites, they haveto act as initiation for composites failure [8]. Instead, Bleayet al. employed hollow glass fiber (with an external diame-ter of 15 lm and an internal diameter of 5 lm) to minimizethe detrimental effect associated with large diameter fibers[8], but filling of repair species into such fine tubes becomesrather difficult.

White and his co-workers sealed dicyclopentadiene(DCPD) into microcapsules made from urea–formalde-hyde resin. Then, the microencapsulated monomer wasadded into epoxy based composites together with pow-dered Grubbs’ catalyst. In the case of cracking, the releasedDCPD contacts the catalyst, a ring-opening-metathesispolymerization (ROMP) of DCPD would take place andthe cracked faces can thus be repaired [9,10]. The healingat 80 �C yields 80% recovery of virgin interlaminar fracturetoughness [11]. Comparatively, this method is superior tothe aforesaid ones in practical applicability, despite the factthat it also has some problems as viewed from long-termservice. The validity of the healing agent DCPD is limited.Higher temperature or longer resting time might result in

Microcapsulatedepoxy as healingagent

Epoxy matrix

Fiberreinforcement

Latent curing agent (dissolve

triggering polymerization ofthe released healing agent

in the matrix, invisible) for

F

F

Fig. 1. Schematic drawing of the principle

polymerization, and of course, invalidation of it. If inhibi-tor is applied, the rate of healing has to be reduced.Besides, the activity of Grubbs’ catalyst would be influ-enced by the amine curing agent contained in the surround-ing epoxy resin.

The present work also used two-component healant, butthe recipe is different from those reported previously.Epoxy was microencapsulated as polymerizable healingresin in hopes of guaranteeing miscibility between the heal-ing agent and the epoxy based composites. Besides, thecomplex of CuBr2 and 2-methylimidazole (CuBr2(2-MeIm)4) was synthesized as the latent hardener of theepoxy healing agent. The complex possesses long-term sta-bility, and would be dissociated into CuBr2 and 2-methy-limidazole again at around 130–170 �C [12–14]. Takingadvantage of this habit, curing of the released epoxy heal-ing agent catalyzed by 2-methylimidazole (i.e. cracks heal-ing) can be triggered at the dissociation temperature ofCuBr2(2-MeIm)4, which is higher than the curing tempera-ture for making the composites. Another advantage ofCuBr2(2-MeIm)4 lies in its dissolubility in uncured epoxy.As a result, the latent curing agent can be homogenouslypre-dispersed (dissolved) in the composites’ matrix onmolecular scale. It is believed that this might increase theprobability of contact between the epoxy resin from theruptured microcapsules and the dissociated imidazole.That is, the released epoxy healing agent can be activatedwherever it is. In comparison with poly(DCPD), the heal-ing system studied in this work (i.e. CuBr2(2-MeIm)4 curedepoxy) is more compatible with epoxy matrix of compos-ites that usually employs amine curing agent. Conse-quently, higher adhesion strength and better repair effectcan be expected. Fig. 1 shows the concept of the self-heal-ing epoxy based laminates.

The healing method studied here is based on manualintervention (i.e. heating). Its application is related to thedevelopment of intelligent structural health monitoring[15] or damage self-sensing techniques [16]. By accurate

d

CrackHealed crackwith cured healing agent

of self-healing epoxy based laminates.

Fig. 2. SEM micrograph of urea–formaldehyde encapsulated epoxy.

T. Yin et al. / Composites Science and Technology 67 (2007) 201–212 203

positioning of the damages, site-specific heating inducedcrack healing can be conducted. Besides, the method ismore suitable for healing of advanced composites that usu-ally possess higher Tg.

This paper discussed the preparation of microencapsu-lated epoxy and CuBr2(2-MeIm)4 latent curing agent, andevaluated feasibility of the two-component repair systemfor producing self-healing epoxy, with the objective ofimproving healing efficiency. Also, the applicability of theself-healing epoxy in fabric laminates was tested to checkwhether it works in fiber composites as expected.

2. Experimental

2.1. Materials and reagents

The bisphenol-A epoxy resin (type E-51), that acts as thehealing agent to be encapsulated and the matrix of thecomposites, was supplied by the Guangzhou DongfengChemical Industry Ltd., China. The ingredients for prepar-ing the latent curing agent are 2-MeIm (97%, AVOCAD)and CuBr2 (A.R., Sinopharm Group Chemical ReagentLtd., China), respectively. All the materials are commercialproducts and were used without further purification.

To assess the healing efficiency of fiber composites,16 · 14 plain weave glass fabric (E-glass, 0.2 mm thick,400 g/m2) supplied by the Taishan Fiber Glass Inc., Chinawas used.

2.2. Microencapsulation of epoxy healing agent andcharacterization

Epoxy healing agent was microencapsulated by a two-step approach. Firstly, 20.0 g urea was mixed with50.4 ml formaldehyde (37 wt%) and the solution wasadjusted to pH = 8.0 with 10% NaOH. After reaction for1 h at 70 �C, transparent water-soluble methylol urea pre-polymer was yielded. On the other hand, 40.0 g epoxywas added into 800 ml sodium polyacylate (PAANa) solu-tion (1.5 wt%, pH = 8.0) together with 4.0 g resorcinol,4.0 g NaCl and 0.8 g polyvinyl alcohol (PVA). Under themechanical stirring at a speed of 16,000 rpm, an oil-in-water (O/W) emulsion was formed. Secondly, the epoxyemulsion was compounded with methylol urea pre-polymerto dissolve the later in the water phase of the former. After-wards, 10% HCl solution was continuously added into thesystem by drip feed, and the system was started beingheated when pH = 4.0. Eventually, the system was heatedto 70 �C while pH reached 2.8–3.0. Having kept at this tem-perature for 1 h, the system was further adjusted topH = 1.5–2.0. One hour latter, the system was neutralizedto pH = 7 by NaOH, cooled down, filtrated and dried, giv-ing birth to urea–formaldehyde resin encapsulated epoxyhealing agent (Fig. 2).

A SingSun ZOOM 645S stereomicroscope was used tomonitor the formation of the microcapsules. The morphol-ogy of the microcapsules was observed by an XL30 FEG

scanning electron microscope (SEM), while their sizes weremeasured with a MasterSizer 2000 size analyzer. The com-positions of the microencapsulated epoxy (ground sample)were characterized by a Nexus 670 Fourier transform infra-red spectroscope (FTIR). The thermal degradation habit ofthe microcapsules was determined by a Netzsch TG-209thermogravimeter (TGA) in N2 at a heating rate of20 �C/min. The epoxy content was determined by weighingthe ground microcapsules before and after acetoneextraction.

2.3. Preparation of the latent curing agent CuBr2(2-MeIm)4

and characterization

CuBr2 (11.2 g) was dissolved in 50 ml methanol and thenthe solution was gradually added to a methanol solution(25 ml) of 2-methylimidazole (16.4 g). Having been stirredfor a period of time, the above mixture was diluted by150 ml acetone, resulting in precipitation. The sludge (i.e.the complex of CuBr2 and 2-methylimidazole) was filtrated,washed and dried. The rate of production is around 97.1%.

A Vario EL elemental analyzer was employed to deter-mine the composition of the product. A TA MDSC 2910differential scanning calorimetry (DSC) was used to studythe decomposition behavior of the complex CuBr2(2-MeIm)4 and curing kinetics of the system epoxy-CuBr2(2-MeIm)4 in N2.

2.4. Preparation of self-healing epoxy samples

A certain amount of CuBr2(2-MeIm)4 latent hardenerwas mixed with epoxy at 60 �C with stirring. Then the mix-ture was cooled down to 40 �C and compounded with agiven proportion of the urea–formaldehyde resin encapsu-lated epoxy as the healing agent. Under ultrasonic agitation,the compound was mixed with 15.2 phr tetraethylenepenta-mine curing agent and degassed. Eventually, the mixture


was cast and cured for 2 h at 60 �C, followed by 2 h at 80 �Cand 2 h at 100 �C.

2.5. Tensile test of self-healing epoxy samples

Tensile properties of the epoxy plates containing theencapsulated epoxy healing agent and the latent hardenerwere measured according to the Chinese National standardGB/T16421-1996, to assess the influence of the inclusions.A Hounsfield 10K-S universal tester was used at a cross-headspeed of 5 mm/min. Five samples (dog-bone-shaped bars60 mm in length, 2 mm in thickness and 3.2 mm in widthwith a gauge length of 6.5 mm) were tested for each case.

2.6. Fracture toughness testing of self-healing epoxy samples

To evaluate the ability of self-healing, plane-strain frac-ture toughness, KIC, of the specimens was determined bymeans of single-edge notched bending (SENB) test. Inaccordance with ASTM standard D5045-99, razor bladepre-notching and a loading rate of 10 mm/min was applied.Three samples (length = 52.8 mm, width = 12 mm,S = 48 mm and thickness = 6 mm) were tested for eachcase. The original specimens were tested to failure, givingthe fracture toughness, K0

IC, and then bound together withpaper clamps in an oven preset at 130 �C for 1 h, allowingthe curing of the healing agent coming out of the brokenmicrocapsules initiated by the catalyst CuBr2(2-MeIm)4.The healed specimens were tested again and yielded thefracture toughness, K 0IC. Accordingly, the crack healing effi-ciency, gepoxy, can be calculated by [17]:

gepoxy ¼K 0IC

K0IC

ð1Þ

2.7. Preparation of self-healing epoxy based laminates

Twelve 130 mm · 150 mm rectangles were cut from theE-glass fabric and then impregnated with epoxy to prepareprepregs. A mixture of tetraethylenepentamine and acrylo-nitrile (1.2/1 wt/wt) was used as the curing system (epoxy/curing agent = 100/30 wt/wt). The two-component self-healing agent consisting of microencapsulated epoxy andCuBr2(2-MeIm)4 latent curing agent was added to fourpieces of the prepregs (epoxy/encapsulated epoxy/CuBr2(2-MeIm)4 = 100/10/2).

During hand lay-up, a sandwich-like structure wasarranged. The four pieces of the prepregs with the self-heal-ing system acted as the central layers, while the rest eightglass fabric plies that only contain unfilled epoxy wereequally used as the two sets of the outer layers, respectively.In the middle of the laminates, a 10 lm thick and 60 mmlong Teflon film was embedded to produce the initialdelamination crack. Eventually, the 12-ply laminates werehot pressed at 80 �C for 4 h and 100 �C for 2 h. The contentof glass fabric in the laminates is 27 vol% as determined byTGA.

2.8. Interlaminar fracture toughness testing of self-healing

epoxy based laminates

Similar to Section 2.6, the self-healing ability of the fab-ric laminates was examined in terms of interlaminar frac-ture toughness, given by double-cantilever beam (DCB)method. According to ASTM D 5528-01, the laminateswere cut into 20 mm · 150 mm · 4.5 mm plates. The detailsof testing have been described elsewhere [18]. Like the caseof self-healing epoxy, interlaminar fracture toughness ofthe original specimen in terms of Mode I critical energyrelease rate, G0

IC, and that of the healed sample, G0IC, werecollected. Healing of the cracked laminates was conductedin 130 �C for 1 h. In accordance with Eq. (1), the healingefficiency of the laminates, glaminates is obtained from [11]:

glaminates ¼K 0IC

K0IC

¼ffiffiffiffiffiffiffiffiG0IC

G0IC

sð2Þ

3. Results and discussion

3.1. Preparation of microencapsulated epoxy healing agent

The technique of microencapsulation has been devel-oped rapidly since its emergency in 1950s [19–21].Although many works were conducted using urea–formal-dehyde resin as wall material, the microcapsules containingepoxy has not yet been reported to the authors’ knowledge.In the present work, urea and formaldehyde were pre-poly-merized and then formed microcapsules via in-situ conden-sation. It was found that the pH value of the reactionsystem exerted critical influence on the products.

When water-soluble methylol urea pre-polymer was syn-thesized in alkaline circumstances, excessive basicity andreaction time would lead to precipitation of white solidsas the methylol urea pre-polymer began to be polymerized.It would hinder the subsequent in-situ polymerization inthe presence of epoxy emulsion and generate large amountof unwanted cured pre-polymers that cannot be used as thewall material in the ultimate product.

Having been mixed with epoxy emulsion, eliminationreaction among methylol urea molecules took place dueto the catalysis of alkali or acid. As a result, linear orbranched pre-polymers with low relative molecular masswere connected through methylene, methylene ether, aswell as cyclic bridges between the urea units, evolving intowater-insoluble polymer networks with time, depositingonto oil-soluble epoxy droplets and bringing about micro-encapsulated epoxy. It is noteworthy that if the condensa-tion polymerization was carried out in alkali medium,methylol urea would not react with each other formingmethylene linkage but dimethylene ether linkage. It wouldlower functionality of the system, and hence the cross-link-ing density of the product, which is detrimental to the wallstrength of microencapsules. In acidic environment, theproduct of polycondensation of methylol urea was mainly

Fig. 4. Pyrolytic behaviors of bisphenol-A epoxy, urea–formaldehydeencapsulated epoxy, and urea–formaldehyde resin. Microcapsules (1#),microcapsules (2#) and microcapsules (3#) were produced in differentbatches by the same method. Their average diameters and the contents ofthe encapsulated epoxy are: 36 lm, 64%; 37 lm, 64%; and 31 lm, 61%,respectively.


bonded by methylene linkages, which facilitated chaingrowth and formation of highly cross-linking structure.Considering that high acidity at the initial stage wouldresult in too high reaction rate to be controlled, a gradualdecrease in pH value was thus followed in this work to pre-pare the microcapsules containing epoxy.

The FTIR spectra of uncured epoxy, microcapsules con-taining epoxy, and urea–formaldehyde resin that was syn-thesized under the same conditions as those employed formaking the microcapsules, were collected in Fig. 3. Clearly,strong absorptions appear at 3300–3500 cm�1 of the spec-trum of the microcapsules, which represent the stretchingmodes of AOH and ANH of urea–formaldehyde resin.Besides, the other characteristic peaks of urea–formalde-hyde resin, like amine bands (at 1600–1630 cm�1 and1530–1600 cm�1), and those of epoxy including the bandsof terminal epoxide group at 914 cm�1 and ACH2A at2873 and 2929 cm�1 are also perceivable in the spectrumof the microcapsules. The results demonstrate that theurea–formaldehyde microcapsules contains epoxy asexpected.

The thermal degradation behaviors of the materials givesupporting evidence for the above analysis. As illustratedin Fig. 4, the onset temperature of urea–formaldehyde resinlies in about 242 �C and the maximum rate of pyrolysisappears at 272 �C, while epoxy starts to be degraded at359 �C and exhibits the maximum decomposition rate at383 �C. It is interesting to see that the curves of the micro-capsules consist of two stages of weight loss with a rise intemperature. Accordingly, the temperatures of the maxi-mum rates of pyrolysis are about 266–269 �C and 404–415 �C, respectively. They should result from the thermaldegradation of the shell (urea–formaldehyde resin) andthe core (epoxy) of the microcapsules. Compared to thepristine epoxy (curve 1 in Fig. 4), thermal stability of theepoxy in the microcapsules (curves 2–4 in Fig. 4) is evi-dently increased owing to the protection of the shield ofthe urea–formaldehyde wall.

Fig. 3. FTIR spectra of bisphenol-A epoxy, urea–formaldehyde encapsu-lated epoxy, and urea–formaldehyde resin.

Since the epoxy emulsion was produced by high-speedstirring, the ultimate microcapsules’ size should be a func-tion of the stirring speed used for preparing the epoxyemulsion. The plots in Fig. 5 indicate that with increasingthe speed, the microcapsules become smaller and the sizedistribution is narrower. That is, both size and size distri-bution is inversely proportional to the stirring speed. Thisobeys the common knowledge of agitation. In addition,our experimental results showed that the stirring time isanother influencing factor. In general, 5 min of stirring issufficient for the present system to obtain microcapsulatedepoxy with stable size.

By weighing the core and shell of the microcapsules(before and after extraction), it was revealed that themicrocapsules with different sizes contain different contentsof epoxy (Fig. 6). The larger microcapsules possess largeramount of epoxy. This might result from the fact that the

Fig. 5. Size distribution of the microcapsules containing epoxy healingagent as a function of stirring speed.

Fig. 6. Influence of microcapsules’ size on content of the epoxy healingagent encapsulated inside urea–formaldehyde shell.

Fig. 8. FTIR spectra of 2-MeIm and CuBr2(2-MeIm)4.


thickness of the microcapsules’ shells is similar, regardlessof their sizes. Therefore, the microcapsules with larger sizeshould have larger portion of the core (i.e. epoxy). On thebasis of this relationship, one is able to adjust the concen-tration of the healing agent.

3.2. Curing kinetics of epoxy catalyzed by CuBr2(2-MeIm)4

According to Ref. [13], the possible structure ofCuBr2(2-MeIm)4 is shown in Fig. 7. To verify theCuBr2(2-MeIm)4 synthesized in the present work, the con-tent of the element N in it was measured. The value 20.43%is rather close to the theoretical one of 20.31%, suggestingthat the target substance has been prepared. Also, theFTIR spectrum of CuBr2(2-MeIm)4 is compared with2-MeIm in Fig. 8. It is seen that all the characteristic bandsof imidazole can be found in the former, including NAHstretching at 3400 cm�1, NAH wagging vibration at756 cm�1, CAN stretching at 1110 cm�1, C@N stretchingat 1600 cm�1, C@C stretching at 1680 cm�1 and @CHrocking vibration at 1440 cm�1. Therefore, it is evidencedthat the imidazole in CuBr2(2-MeIm)4 keeps original struc-ture as desired. On the other hand, the most significant dif-ference in the two spectra lies in that the stretchingabsorptions of CAH in imidazole ring and methyl at2500–3200 cm�1 disappear in the spectrum of CuBr2(2-MeIm)4. It manifests that the coordination might haveobstructed certain vibration modes.

Fig. 7. Structural scheme of the latent curing agent CuBr2(2-MeIm)4.

The mechanism responsible for the reaction betweenepoxy and a complex of metal salt and imidazole [22] isgiven in Fig. 9, which takes CuCl2(Im)4 (the complex ofCuCl2 and imidazole) for example and is applicable forthe system studied in this work. Having been heated, thecomplex is dissociated into CuCl2 and imidazole. Firstly,the active hydrogen of the secondary amine in imidazolereacts with epoxide group yielding affixture. Then the reac-tion of the affixture with another epoxide group is per-formed and an ionic complex is obtained. When theanionic portion further reacts with epoxide group, epoxycan be cured by ring-opening polymerization via chainreaction. As the anionic polymerization of imidazole isrestricted by the affixture, the rate of epoxy curing is lowerin comparison to the case where tertiary amine family isused as the hardener.

Fig. 10 shows the curing processes of CuBr2(2-MeIm)4/epoxy system in terms of conversion versus temperatureestimated from the non-isothermal DSC scans (Fig. 11).It is seen that curing of epoxy occurred at about 130 �C.The exothermic peaks appear at 141–176 �C, while the cor-responding conversions are still lower than 50%. It meansthat CuBr2(2-MeIm)4 is a mild curing agent. By using Kis-singer [23] and Crane et al. [24] equations, the characteristicparameters of the curing kinetics including the activationenergy, Ea, and the order of reaction, n, of CuBr2(2-MeIm)4 (1 wt%)/epoxy were calculated to be 83.7 kJ/moland 0.92, respectively. As n is a non-integral, the curingof this system should be a complicated process. On theother hand, the temperature dependence of rate constantof the curing reaction was obtained from Arrhenius equa-tion (Fig. 12). Clearly, the curing reaction proceeds veryslowly at lower temperature (e.g., 120 �C). This is con-vinced by the isothermal DSC scans of the system(Fig. 13). At a constant temperature of 120 �C, there isno detectable exothermic peak within 90 min, implyingnearly no curing takes place. In accordance with thisfinding, the curing temperature for preparing self-healing

Fig. 10. Temperature dependence of conversion of curing reaction ofepoxy activated by CuBr2(2-MeIm)4 (1 wt%) at different heating rates.

Fig. 11. Non-isothermal DSC scans of CuBr2(2-MeIm)4–epoxy system(1 wt%) at different heating rates.

Fig. 12. Temperature dependence of rate constant of curing reaction ofepoxy activated by CuBr2(2-MeIm)4 (1 wt%).

Fig. 13. Isothermal DSC scans of CuBr2(2-MeIm)4–epoxy system (1 wt%)conducted at different temperatures.

Cu

Cl

Cl

Im N

NH

Im

ImNHNCu

Cl

Cl

Im N

NH

Im

ImCuCl2 + 4

(I)

N CH2N CH

OH

RNHN

NN CH2 CH R

OH

R CH CH2

O

+

(II)

O

H2C RCH

O

H2C RCH

R CH CH2

OH

NN CH2 CH R

O CH2 CH

O

CH2R

O

H2C RCH

O

H2C RCHn

Polymer

Fig. 9. Mechanism of curing reaction between epoxy resin and the complex CuCl2(Im)4.


composites should be lower than the healing temperature,so as to avoid any change in the healing agent and thelatent hardener when the composites were cured. In other

words, the healing system developed in this work provesto be durable for long-term application under moderatetemperature.


3.3. Tensile performance of self-healing epoxy

Since the microencapsulated epoxy and the latent curingagent must be filled into the composites’ matrix, its influ-ence on the basic mechanical performance of the matrixis worth being understood. Hence tensile properties ofepoxy containing the self-repair system were measured asa function of the microcapsules’ concentration at a fixedcontent of CuBr2(2-MeIm)4.

The plots in Fig. 14(a) exhibit that tensile strength of thecompounds keeps almost unchanged with a rise in contentof the microcapsules. It is different from the results ofBrown et al. [25], who reported a continuous reduction inthe strength of epoxy with embedded microcapsules. Gen-erally, the addition of either rigid particles or rubbery par-ticles into polymers would lead to substantial decrease intensile strength of the matrices [26,27]. Nevertheless, coun-terexamples can be found in inorganic particulates com-posites when the bonding between the fillers and thematrix is strong enough [28], or the particles’ size is inthe nanometer range [29]. The strength data in Fig. 14(a)suggested that the shell material of the microcapsules,urea–formaldehyde resin, is compatible with epoxy and a

Fig. 14. Influence of microcapsules’ content on tensile properties ofepoxy. Content of CuBr2(2-MeIm)4: 2 wt%; average diameter of themicrocapsules: 37–42 lm; content of the epoxy healing agent inside themicrocapsules: 63–68%.

strong interfacial interaction was established during curing.Besides, the microcapsules behave unlike soft rubber, butare able to carry certain load transferred by the interface.These account for the dependence of tensile strength onthe content of microcapsules illustrated in Fig. 14(a).

The above analysis receives support from the variationtrend of Young’s modulus (Fig. 14(a)). As the stiffness ofthe microcapsules ranks among rigid particles and soft rub-ber, the decrease in modulus of the epoxy specimens is notremarkable, even at a filler concentration as high as20 wt%. On the other hand, the increase in failure strain(Fig. 14(b)) should be attributed to the fact that the micro-capsules have induced interfacial viscoelastic deformationand matrix yielding. The decrease in elongation at breakfor the highly loaded specimens might be due to the unevendistribution of the microcapsules, which led to stress con-centration in some parts of the specimen.

3.4. Fracture toughness of self-healing epoxy

Similar to the last section, fracture toughness of theepoxy with the healing agent is also evaluated hereinafter(Fig. 15). Fig. 15(a) indicates that the incorporation of

Fig. 15. Influence of (a) latent hardener’s content and (b) microcapsules’content on fracture toughness of epoxy. Average diameter of themicrocapsules: 37–42 lm; content of the epoxy healing agent inside themicrocapsules: 63–68%.


the latent hardener CuBr2(2-MeIm)4 tends to toughencured epoxy up to 4 wt%. It is believed that the latent cur-ing agent dissolved in epoxy plays the role of plasticizer.With increasing content of CuBr2(2-MeIm)4, those cannotbe well dissolved have to present themselves in the form oftiny particles (Fig. 16), and deteriorate the tougheningeffect. It explains the reduced toughening efficiency from0.5 to 4 wt% of CuBr2(2-MeIm)4, as compared to the dras-tic increase in K0

IC when the latent hardener contentincreases from 0 to 0.5 wt% (Fig. 15(a)).

In the case that both microcapsules and the latent curingagent were added, the fracture toughness of the system isslightly lower than that of neat epoxy at certain propor-tions of the composition. Obviously, the microcapsulescan neither hinder the crack propagation nor result inenergy consumption. Owing to the positive effect of thelatent hardener that counteracted the negative effect tothe toughening, however, fracture toughness of the blendsis similar to that of the neat epoxy on the whole (Fig. 15).

Fig. 16. SEM micrographs of cured CuBr2(2-MeIm)4–epoxy system.Content of CuBr2(2-MeIm)4: (a) 0.5 wt% and (b) 2 wt%.

From the above results, it is concluded that the two-component self-healing system developed in this workwould not significantly change the mechanical propertiesof epoxy. Instead, it brings in somewhat toughening insome cases.

3.5. Fracture toughness of repaired epoxy

As mentioned in the introductory section, healing of thepolymer composites depends on the polymerization of thecrack released healing agent (i.e., epoxy in the presentwork) activated by the latent hardener. Considering thata precise proportion of the ingredients cannot be pre-esti-mated for this specific case, we have to prepare a seriesof materials with different microcapsule-latent hardenerratios, and then compare the fracture toughness of thespecimens before and after healing (Figs. 17 and 18).

Fig. 17 shows the dependence of K 0IC on the contentof the latent hardener. Because the epoxy containing

Fig. 17. Influence of latent hardener’s content on self-healing ability ofepoxy. Average diameter of the microcapsules: 37–42 lm; content of theepoxy healing agent inside the microcapsules: 63–68%.

Fig. 18. Influence of microcapsules’ content on self-healing ability ofepoxy. Average diameter of the microcapsules: 37–42 lm; content of theepoxy healing agent inside the microcapsules: 63–68%.

Fig. 19. SEM micrographs of the fractured surfaces of the self-healingepoxy specimens: (a) virgin specimen and (b) healed specimen (the arrowindicates the direction of crack propagation). Contents of microcapsulesand CuBr2(2-MeIm)4: 10 and 2 wt%, respectively.

Fig. 20. Interlaminar fracture toughness of virgin and healed fabriclaminates as a function of crack length. Content of microcapsules andlatent hardener in the central layers: epoxy/microcapsules/CuBr2(2-MeIm)4 = 100/10/2.


different amounts of the microcapsules and latent hard-ener possesses similar K0

IC (Fig. 15), the dependence ofhealing efficiency on the content of the latent hardeneris similar to the corresponding curves of K 0IC (Fig. 17).When the microcapsules content is 5 wt%, the values ofK 0IC are relatively low within the entire loading range ofthe latent curing agent. It should result from the insuffi-cient dosage of the microcapsules, leading to insufficientquantity of the released epoxy for covering the brokensurface. Therefore, some cracked portions have to be leftunhealed and the apparent fracture toughness is not ashigh as expected. In the case of 10 wt% of the microcap-sules and 2 wt% of the latent hardener, the highest K 0IC

and gepoxy are observed, meaning the optimum propor-tion has been reached. Further increase in the contentof the latent hardener could not ensure the best cross-linking extent, and the effect of repair has to be loweraccordingly. For the system containing 15 and 20 wt%microcapsules, the areas of the cracked planes that canbe healed are reduced accordingly. The amount of thelatent hardener that might contact the released epoxybecomes insufficient, which also leads to lower repair effi-ciencies than the maximum. It is interesting to note thatthe aforesaid highest healing efficiency is 111%, whichimplies that the fracture toughness of the healed sampleis higher than that of the virgin one. For understandingthe cause, fracture toughness of epoxy cured only byCuBr2(2-MeIm)4 (5 wt%) was measured. The value0.81 MPa m1/2 is 1.23 times higher than that of bulkepoxy, which was cured by tetraethylenepentamine (cf.Section 2.4). Therefore, the bonding material not onlyhealed the cracks but also provided the damaged siteswith higher fracture toughness.

Fig. 18 shows the influence of microcapsules’ contentat certain dosage of the latent hardener. With increasingthe content of the microcapsules, K 0IC increases first,and then decreases. Besides, the content of the microcap-sules corresponding to the maximum K 0IC increases withthe latent hardener content. It demonstrates that thehighest healing efficiency can be obtained only at theoptimum microcapsule-latent hardener ratio, as foundin Fig. 17.

The SEM micrograph of the fractured surface illus-trates that after failure of the first SENB test the micro-capsules were damaged and the healing agent hadflowed off, leaving the ring-like concaves (Fig. 19(a)).Having been healed and experienced the second SENBtest, the specimen is characterized by more complicatedfracture patterns (Fig. 19(b)). The thin layers on the frac-ture surface represent the result of crack propagationthrough the cured healing agent.

3.6. Preliminary evaluation of the self-healing epoxy

laminates

The above results and discussion have clearly shown theeffectiveness of the self-healing epoxy. To examine whether

it works in fiber composites, E-glass fabric laminates wereprepared and tested using DCB method. In the central lay-ers of the laminates, the self-healing epoxy was employedto impregnate the fabric.


Fig. 20 illustrates the crack growth resistance versuscrack extension (R-curve) of the virgin and healed lami-nates. It means that the self-healing epoxy is really ableto repair the cracked laminates. The steady-state G0

IC andG0IC are 957.1 J/m2 and 448.7 J/m2, respectively. Insertingthe data into Eq. (2), it is known that glaminates = 68%.Compared to the case of self-healing epoxy itself, the lowerhealing efficiency of the laminates might result from theimproper proportion of the healing agent. Although thecontents of the microcapsules and the latent hardener inthe laminates were also kept at 10 wt% and 2 wt% relativeto the quantity of epoxy, the appearance of the glass fabricmight have change to probability for the microcapsules tomeet cracks. Therefore, further works should be done tofind out the optimum composition.

4. Conclusions

1. Urea–formaldehyde resin encapsulated epoxy healingagent (30–70 lm in diameter) can be successfully pre-pared by pre-polymerizing urea and formaldehyde,and then polycondensation of the pre-polymer in epoxyemulsion. Content of the epoxy inside the urea–formal-dehyde microcapsules is proportional to the microcap-sules’ size.

2. Latent curing agent CuBr2(2-MeIm)4 is able to be welldissolved in epoxy and cure epoxy at 130–180 �C. Tak-ing the advantage of the former characteristics, its prob-ability of encounter with epoxy released from the brokenmicrocapsules is greatly increased.

3. Incorporation of the microencapsulated epoxy intoepoxy with pre-dissolved latent hardener results in slightreduction of Young’s modulus, but improved elongationat break and unchanged tensile strength. It reveals thatthe interfacial interaction between the two-componentrepair system and matrix epoxy is strong enough totransfer stress, and the microcapsules are also able tocarry certain load transferred by the interface.

4. Fracture toughness of epoxy containing the microencap-sulated epoxy and latent hardener depends on their con-tents. Within the concentration range of interests,fracture toughness of the self-healing epoxy is nearlythe same as that of the unfilled epoxy due to the oppositeinfluence of the microencapsulated epoxy and latenthardener.

5. Fracture toughness of the healed epoxy specimens andthe corresponding healing efficiency is closely relatedto the contents of the healing agent and the latent curingagent. In the case of 10 wt% microcapsules and 2 wt%latent hardener, the healing efficiency can be as high as111% because of the higher fracture toughness of therepair ‘‘binder’’ than that of the bulk epoxy to be healed.

6. The self-healing epoxy proves to be workable in epoxybased fabric laminates. A 68% recovery of the virgininterlaminar fracture toughness preliminarily showsthe prospect of the self-healing system. More works inthis aspect will be reported in the near future.

Acknowledgements

The authors are grateful to the support of the NationalNatural Science Foundation of China (Grant: 50573093)and the Team Project of the Natural Science Foundationof Guangdong, China (Grant: 20003038).

References

[1] Riefsnider KL, Schulte K, Duke JC. Long term failure behavior ofcomposite materials. ASTM STP 1983;813:136–59.

[2] Chen XX, Dam MA, Ono K, Mal A, Shen H, Nutt SR, et al. Athermally re-mendable cross-linked polymeric material. Science2002;295:1698–702.

[3] Chen XX, Wudl F, Mal AK, Shen HB, Nutt SR. New thermallyremendable highly cross-linked polymeric materials. Macromolecules2003;36:1802–7.

[4] Dry C. Passive tunable fibers and matrices. Int J Mod Phys B1992;6:2763–71.

[5] Dry C. Procedures developed for self-repair of polymeric matrixcomposite materials. Compos Struct 1996;35:263–9.

[6] Zhao XP, Zhou BL, Luo CR, Wang JH, Liu JW. A model ofintelligent material with self-repair function. Chin J Mater Res1996;10(1):101–4 [in Chinese].

[7] Motuku M, Janowski CM, Vaidya UK. Parametric studies on self-repairing approaches for resin infused composites subjected to lowvelocity impact. Smart Mater Struct 1999;8:623–38.

[8] Bleay SM, Loader CB, Hawyes VJ, Humberstone L, Curtis PT. Asmart repair system for polymer matrix composites. Compos Part A2001;32:1767–76.

[9] White SR, Sottos NR, Geubelle PH, Moore JS, Kessler MR, SriramSR, et al. Autonomic healing of polymer composites. Nature2001;409:794–7.

[10] Brown EN, Sottos NR, White SR. Fracture testing of a self-healingpolymer composite. Exp Mech 2002;42(4):372–9.

[11] Kessler MR, Sottos NR, White SR. Self-healing structural compositematerials. Compos Part A 2003;34:743–53.

[12] Dowbenko R, Anderson CC, Chang WH. Imidazole complexes ashardeners for epoxy adhesives. Ind Eng Chem Prod Res Develop1971;10:344–51.

[13] Ibonai M, Kuramochi T. Curing of epoxy resin by use ofimidazole/metal complexs. Purasuchikkusu 1975;26(7):69–73 [inJapanese].

[14] Bi CH, Gan CL, Zhao SQ. Study on synthesis, curing reaction andproperties of imidazole salt. Thermoset Resin 1997;12(1):12–5 [inChinese].

[15] Su Z, Ye L. An intelligent signal processing and pattern recognitiontechnique for defect identification using an active sensor network.Smart Mater Struct 2004;13:957–69.

[16] Fernando GF, Degamber B, Wang LW, Doyle C, Kister G, Ralph B.Self-sensing fibre reinforced composites. Adv Compos Lett2004;13:123–9.

[17] Wool RP, O’Connor KM. A theory of crack healing in polymers. JAppl Phys 1981;52:5953–63.

[18] Rong MZ, Zhang MQ, Liu Y, Zhang ZW, Yang GC, Zeng HM.Effect of stitching on in-plane and interlaminar properties of sisal/epoxy laminates. J Compos Mater 2002;36:1505–25.

[19] Gardner GL. Manufacturing encapsulatd products. Chem Eng Progr1966;62(4):87–91.

[20] Fanger GO. What good are microcapsules. Chemtech 1974;7:397–405.

[21] Song J, Chen L, Li XJ. Microencapsulation and its application.Beijing: Chemical Engineering Press; 2001 [in Chinese].

[22] Eilbeck WJ, Holmes F, Taylor CE, Underhill AE. Cobalt (II), nickel(II), and copper (II) complexes of 2-methylimidazole. J Chem Soc (A)1968;1:128–32.


[23] Kissinger HE. Reaction kinetics in differential thermal analysis. AnalChem 1957;29:1702–6.

[24] Crane LW, Dynes PJ, Kaelble DH. Analysis of curing kinetics inpolymer composites. J Polym Sci Polym Lett Ed 1973;11:533–40.

[25] Brown EN, White SR, Sottos NR. Microcapsule induced tough-ening in a self-healing polymer composite. J Mater Sci 2004;39:1703–10.

[26] Ahmed A, Jones FR. A review of particulate reinforcement theoriesfor polymer composites. J Mater Sci 1990;25:4933–42.

[27] Kinloch AJ, Young RJ. Fracture behaviour of polymers. London:Applied Science Publishers; 1983.

[28] Jancar J, Dianselmo A, Dibenedetto AT. The yield strength ofparticulate reinforced thermoplastic composites. Polym Eng Sci1992;32:1394–9.

[29] Rong MZ, Zhang MQ, Zheng YX, Zeng HM, Walter R, Friedrich K.Irradiation graft polymerization on nano-inorganic particles: aneffective means to design polymer based nanocomposites. J Mater SciLett 2000;19:1159–61.

self-healing epoxy composites – preparation and effect of the healant consisting of...

Documents