Materials Science and Engineering A 509 (2009) 57–62

Contents lists available at ScienceDirect

Materials Science and Engineering A

journa l homepage: www.e lsev ier .com/ locate /msea

Effect of sonication on thermo-mechanical properties of epoxy nanocompositeswith carboxylated-SWNT

Jaqueline Suavea, Luiz A.F. Coelhoa, Sandro C. Amicob, Sérgio H. Pezzina,∗

a Center of Technological Sciences, Santa Catarina State University, 89223-100 Joinville, SC, Brazilb Materials Engineering Department, Federal University of Rio Grande do Sul, P.O. Box 15010, 91501-970 Porto Alegre, RS, Brazil

a r t i c l e i n f o

Article history:Received 4 July 2008Received in revised form23 December 2008Accepted 14 January 2009

Keywords:Carbon nanotubes

a b s t r a c t

In this work, a 2k factorial design was used to determine the interaction between sonication time,ultrasound amplitude, and carboxylic-functionalized single-walled carbon nanotubes (c-SWNT) con-tent and their effect on the tensile, dilatometric, and dynamic-mechanical properties of casted epoxynanocomposites reinforced with c-SWNTs. The nanocomposites were prepared with the help of a sol-vent (tetrahydrofuran) to reduce resin viscosity. Tensile strength increased with sonication for higheramplitudes and shorter times (20 min). The addition of 0.25 wt% c-SWNT enhanced the mechanical prop-erties, increasing the Young’s modulus up to 30% in comparison with neat epoxy prepared under the

Mechanical propertiesFactorial designThermal analysisP

same conditions. Dilatometry tests showed that shrinkage takes place at the epoxy glass transition, whiledynamic-mechanical analyses confirmed the tensile results and suggested that the presence of nanotubesyielded a more homogeneous epoxy resin network, probably due to the minimization of residual solvent.

1

rcto

bic

bdb

morui[

ER

0d

olymer composites

. Introduction

Carbon nanotubes (NTs), reported by Iijima in 1991 [1], are cur-ently being investigated as a possible reinforcement for polymeromposites. NTs are considered candidates for this application dueo their unique structure, which exhibits a remarkable combinationf mechanical, electrical and thermal properties [2].

The use of a small amount of NTs in polymer matrices, generallyelow 1 wt% [3–5], has the potential to yield nanocomposites with

mproved properties. However, the reported effect has been, in mostases, much lower than envisaged by theoretical predictions.

The effective mechanical reinforcement by NTs in polymers maye achieved once two major challenges are faced: homogeneousispersion of NTs in the matrix, and good interfacial adhesionetween nanotubes and matrix [3,6–8].

For homogeneous dispersion, van der Waals interactions andechanical anchorage between stable NTs aggregates must be

vercome. A variety of methods to promote dispersion have been

eported, including high shear mixing, tip sonication, milling, these of surfactants and solvents, in situ polymerization of monomers

n the presence of NTs, and different chemical functionalization6,8].

∗ Corresponding author at: Centro de Ciências Tecnológicas, Universidade dostado de Santa Catarina, Campus Universitário Prof. Avelino Marcante s/n, Bometiro, 89223-100 Joinville, SC, Brazil. Tel.: +55 474009 7841; fax: +55 474009 7940.

E-mail address: pezzin@joinville.udesc.br (S.H. Pezzin).

921-5093/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.msea.2009.01.036

© 2009 Elsevier B.V. All rights reserved.

Chemical functionalization at the surface of NTs has a greatpotential for enhancing physical and chemical properties of thefinal materials by promoting adhesion and a more efficient stresstransfer at the NT/matrix interface. For instance, soluble nanotubesmay be used to produce mechanically reinforced nanocomposites[9].

In this work, a 2k factorial design was used to statisticallydetermine the interaction between the main parameters in theproduction of epoxy-matrix nanocomposites reinforced with ran-domly dispersed carboxylic-functionalized single-walled carbonnanotubes (SWNTs). The fabrication process studied here was basedon a tip sonication technique with the use of solvents to achieve auniform dispersion of SWNTs in the matrix. The influence of theSWNT content on the mechanical and thermal properties of theepoxy-matrix nanocomposites was also investigated.

2. Experimental

2.1. Materials

Single-walled carbon nanotubes functionalized with carboxylicgroups (c-SWNT) used in this work were produced by electrical

arc discharge and supplied by the Nanomaterials Laboratory at theFederal University of Minas Gerais (UFMG)-Brazil.

The bifunctional diglycidyl ether of bisphenol A (DGEBA) epoxyresin was obtained from Huntsman as Araldite GY 251®, togetherwith the hardener Aradur HY 956®, based on a polyamine. The sol-

58 J. Suave et al. / Materials Science and Engineering A 509 (2009) 57–62

Table 1Factors and levels of the 23 factorial design.

Factors Inferior level (−) Superior level (+)

(A) Sonication power 165 W 225 W((

vQ

2

eacptsoptp

2

cdtawtdawrwcla

2

mrswwutc

i(tet

o(

w

posites [12]. The loss of weight above around 300 ◦C is related to thedecomposition of the polymer.

Fig. 2 shows thermograms of c-SWNT/epoxy nanocomposites,where the presence of residual THF is again observed. The estimated

B) Sonication time 20 min 40 minC) c-SWNT content 0.10 wt% 0.25 wt%

ent used was analytical grade tetrahydrofuran (THF), from Cineticauimica-Brazil.

.2. Experimental design

In order to achieve a homogeneous dispersion of c-SWNTs in thepoxy-matrix, the nanocomposites were prepared with the help ofsolvent (THF), to reduce resin viscosity, and tip sonication. A typi-al 23 factorial design was used to study the effect of the sonicationarameters (power and time of sonication) and the c-SWNT con-ent on the final properties of the material. These factors and theirtudied levels are identified in Table 1 and the statistical analysisf the results was carried out using the software Statistica®. Thearameters and levels chosen in this work were defined based onhe former experience of the group and after testing several otherossibilities [10,11].

.3. Preparation of c-SWNT/epoxy nanocomposites

For the preparation of nanocomposites, a certain amount ofarboxylated-SWNTs (0.10 or 0.25 wt% relative to the epoxy) wasispersed in THF (10 wt%, in relation to the epoxy) under simul-aneous magnetic stirring and sonication (Sonics Vibration 750)t 165 or 225 W, for 20 or 40 min. The nanotube (NT) suspensionas then added to the epoxy under the same conditions of simul-

aneous magnetic stirring and sonication. The mixture was lateregassed under vacuum and magnetic stirring at 70 ◦C for 1.5 hnd cooled down to room temperature. After that, the hardeneras added to the c-SWNT/epoxy system under mechanical stir-

ing, at a epoxy:hardener weight ratio of 5:1. Finally, the mixtureas cast into silicone molds and left to cure at 60 ◦C for 24 h. For

omparison, another sample, called epoxy/THF, was prepared fol-owing this procedure except that in this sample no SWNT wasdded.

.4. Characterization

Thermogravimetric analysis (TG) was performed on a TA Instru-ents equipment (model 2050). The samples were heated from

oom temperature to 1000 ◦C at 10 ◦C min−1 under nitrogen atmo-phere at a flow rate of 70 cm3 min−1. Dilatometric measurementsere conducted on a Netzsch 402C equipment. The specimensere heated from room temperature to 160 ◦C at 2 ◦C min−1,nder N2 atmosphere, followed by a fast cooling down to roomemperature and a new heating (second run) under the sameonditions of the first run.

The behavior of the epoxy/THF and the nanocomposite spec-mens was evaluated by various mechanical tests. Tensile testsASTM D638M-93) were carried out in an EMIC DL 30000 universalesting machine using a 1000 kgf load-cell, at 5 mm min−1, and forvery experimental condition, a minimum of 10 specimens wereested.

The surface morphology of the tensile fractured specimens wasbserved under a Zeiss DSM 940A scanning electron microscopeSEM) operating at 10 kV.

Dynamic-mechanical analysis (DMA) on dual cantilever modeas carried out from room temperature to 250 ◦C, at 5 ◦C min−1, in

Fig. 1. Thermograms for neat epoxy and epoxy sonicated in THF.

a TA Instruments (model 2980) equipment. All measurements wereconducted at a strain of 0.01% and a frequency of 1 Hz.

The molecular weight distribution of sonified epoxy sampleswas obtained using a Waters 510 Gel Permeation Chromatographequipped with a Waters 410 refraction index detector and PS/DVBStyragel® columns thermostated at 40 ◦C. Standards of polystyrenewere used for calibration and tetrahydrofurane as eluent. The flowrate was 1.0 mL min−1 and the injection volume 200 �L.

3. Results and discussion

3.1. Thermogravimetry

Fig. 1 shows the weight loss curves for cured samples of neatepoxy and epoxy sonicated with 10 wt% THF, i.e. epoxy/THF. Ahigher weight loss in the 100–300 ◦C range was observed for theepoxy/THF, which is related to the elimination of residual solvent.This is similar to what was reported by Lau et al. who observed thepresence of residual dimethylformamide in SWNT/epoxy nanocom-

Fig. 2. Thermograms for epoxy sonicated in THF and c-SWNT/epoxy nanocom-posites prepared under different conditions [caption: sonication power/sonicationtime/c-SWNT content].

J. Suave et al. / Materials Science and Engineering A 509 (2009) 57–62 59

Table 2Residual THF concentration, Tpeak and glass transition temperatures (Tg, from DMAmeasurements) for all studied samples following experimental design: sonicationpower/sonication time/c-SWNT content.

Sample Residual THF (wt%) Tpeak (◦C) Tg (◦C)

Neat epoxy N/A 373 72Epoxy/THF 5.1 370 78165 W/20 min/0.10% 4.8 369 72225 W/20 min/0.10% 4.5 364 68165 W/40 min/0.10% 3.2 366 78225 W/40 min/0.10% 3.7 365 76165 W/20 min/0.25% 3.8 374 74225 W/20 min/0.25% 3.4 367 6812

asoTstpt

traadrct

3

tt

cttspwar

(st

omh(e(3tasta

65 W/40 min/0.25% 3.1 369 8225 W/40 min/0.25% 1.5 369 77

mount of THF trapped in each sample is presented in Table 2. Theamples with higher c-SWNT contents presented a lower amountf residual solvent. It also appears that the content of trappedHF decreases for longer sonication periods. Above ca. 350 ◦C, allamples decompose and the curves overlap. The decompositionemperatures (Tpeak in Table 2) were slightly higher for nanocom-osites with 0.25 wt% of c-SWNT than for 0.10 wt% prepared underhe same conditions.

Distinct results may be found in the literature regarding thehermal stability of NT-reinforced nanocomposites. Shen et al. [13]eported an improvement in thermal stability when the nanotubesre well dispersed and a single interfacial region between matrixnd NTs is present, whereas Miyagawa and Drzal [14] observed aecrease in the thermal stability of epoxy with the addition of fluo-inated SWNTs, which may occur due to the NTs interference in theure stoichiometry, reducing the density of crosslinks, which leadso lower degradation temperatures.

.2. Dynamic-mechanical analysis

Fig. 3(a) illustrates the DMA curves of storage modulus versusemperature for neat epoxy (prepared with or without THF addi-ion) and nanocomposites with 0.25 wt% c-SWNT.

The storage modulus was found to increase with the c-SWNTontent and the sonication time, and decrease with the sonica-ion power. For the best sonication conditions (165 W and 40 min),he addition of 0.25 wt% of c-SWNTs yielded a 57% increase intorage modulus (at 40 ◦C), in comparison with epoxy/THF sam-les, and a 42% increase in comparison with neat epoxy (preparedithout solvent). In fact, the storage modulus reached values

bove 3.0 GPa at 40 ◦C, significantly higher than similar compositesecently reported in the literature [15,16].

The loss factor (tan ı) versus temperature curves for neat epoxywith or without THF addition) and various nanocomposites arehown in Fig. 3(b). Glass transition temperatures, determined fromhe position of the tan ı peak, varied from 71 to 78 ◦C (see Table 2).

The peak factor (� ), defined as the full width at half maximumf the tan ı peak divided by its height [15], was used to esti-ate the homogeneity of the epoxy network. Lower � values, i.e.

igher homogeneity, were found for stricter sonication conditions225 W and 40 min). The variation with the c-SWNT content, how-ver, was very small, changing from 20.8 ◦C (for 0.10 wt%) to 19.9 ◦Cfor 0.25 wt%). The epoxy/THF sample showed the highest � value,9.6 ◦C, indicating a less homogeneous network, probably due to

he presence of trapped solvent. For the nanocomposites the aver-ge � value was 25.7 ◦C, showing that the presence of nanotubesomehow promoted a more homogeneous network, probably dueo the minimization of residual solvent, as confirmed by TGnalysis.

Fig. 3. DMA curves of the epoxy nanocomposites with 0.25 wt% c-SWNT, the neatepoxy, and the epoxy/THF samples [caption: sonication power/sonication time].

3.3. Dilatometry

Fig. 4(a) presents the first run (heating) dilatometric curves for

some of the c-SWNT/epoxy nanocomposites, showing volume con-traction at temperatures around Tg, related to free volume andpost-cure processes. In the second run (Fig. 4(b)), most samplesexhibited a plateau near Tg, evidencing some reorganization of

60 J. Suave et al. / Materials Science and Engineering A 509 (2009) 57–62

FS

pc

leoet

3

p

Table 4Estimated effects of (A) sonication power, (B) sonication time, and (C) c-SWNT con-tent, on the tensile parameters following a 23 factorial design.

Factors Tensile strength Elongation at break Young’s modulus

A 0.83 ± 0.66 −0.11 ± 0.04 0.13 ± 0.07B 5.32 ± 0.66 0.20 ± 0.04 −0.04 ± 0.07C 7.69 ± 0.66 0.28 ± 0.04 0.07 ± 0.07A × B −5.02 ± 0.66 −0.21 ± 0.04 0.10 ± 0.07

TT

S

E12121212

ig. 4. Dilatometry curves of the first (a) and the second (b) heatings for c-WNT/epoxy nanocomposites.

olymer chains due to stress relaxation during the heating pro-ess.

It was also noted that the nanocomposites showed a slightlyower thermal dilation than epoxy/THF. Gonnet [17] correlates thisffect to the negative thermal expansion coefficient of carbon nan-tubes, due to its sp2 network. Indeed, theoretical data suggest thatven nanocomposites with low SWNTs content should have a lowerhermal dilation than the unreinforced resin [17].

.4. Tensile properties

Table 3 summarizes the tensile tests results of the nanocom-osites and epoxy/THF. It may be observed that c-SWNT/epoxy

able 3ensile properties of the various c-SWNT/epoxy composites.

ample Tensile strength (MPa)

poxy/THF 31.3 ± 3.565 W/20 min/0.10% 20.5 ± 1.425 W/20 min/0.10% 25.0 ± 1.065 W/40 min/0.10% 31.5 ± 2.225 W/40 min/0.10% 25.4 ± 2.765 W/20 min/0.25% 28.7 ± 3.425 W/20 min/0.25% 34.9 ± 4.065 W/40 min/0.25% 39.2 ± 5.125 W/40 min/0.25% 34.6 ± 3.4

A × C 0.66 ± 0.66 0.01 ± 0.04 −0.02 ± 0.07B × C −0.60 ± 0.66 0.04 ± 0.04 −0.27 ± 0.07A × B × C −0.98 ± 0.66 −0.05 ± 0.04 0.12 ± 0.07

nanocomposites presented a higher Young’s modulus than theepoxy/THF, whereas variations in tensile strength and elongation atbreak were less significant and more dependent on the processingparameters. It has already been reported in the literature that theelastic modulus of NT/epoxy nanocomposites increases with the NTcontent regardless of the dispersion level, but tensile strength andelongation at break are more critically related to a homogeneousdistribution in the matrix [7]. Agglomerates of NTs, which maylead to rapid crack propagation [15], and poor NT/matrix interfacialadhesion are the main factors associated with low tensile strengthof nanocomposites [18].

The results of the statistical analysis are reported in Table 4. Itwas found that the c-SWNT content factor has no significant inter-action with the others (for a significance level of 5%), i.e. it can beinterpreted isolated, and is the most significant factor to enhancetensile strength. There is also a trend to increase tensile strength forlonger sonication times, which could be related to lower resin vis-cosity and better dispersion of c-SWNTs in the matrix. Similar, butless important, effects were found regarding elongation at breakand Young’s modulus.

Considering tensile properties, lower sonication power (165 W),longer sonication time (40 min) and higher c-SWNT content(0.25 wt%) were found to be the most adequate conditions to pre-pare c-SWNT/epoxy nanocomposites.

Micrographs of tensile fractured surfaces of epoxy/THF and someof its nanocomposites are shown in Fig. 5. All specimens exhib-ited a relatively smooth surface indicating a typical brittle fracturebehavior and no significant changes were observed in the nanocom-posites with c-SWNTs.

3.5. Gel permeation chromatography

The GPC results for 10 wt% epoxy/THF sonicated solutions andneat non-sonicated epoxy are presented in Table 5. The num-ber average molecular mass (Mn), the chromatogram profile, andthe concentration of monomer (DGEBA) and oligomers of thenon-sonicated epoxy were similar to those reported in the lit-

erature [19]. The concentration of oligomers tends to decreasewhen the samples are sonicated due to the scission of polymerchains. However, the most significant decrease in the oligomer con-centration occurred under milder sonication conditions (165 W),while the samples submitted to 225 W showed a lower decrease in

Elongation at break (%) Young’s modulus (GPa)

1.1 ± 0.1 3.0 ± 0.20.6 ± 0.1 3.3 ± 0.10.6 ± 0.1 3.6 ± 0.11.0 ± 0.2 3.5 ± 0.30.6 ± 0.2 3.8 ± 0.20.8 ± 0.1 3.9 ± 0.20.9 ± 0.3 3.8 ± 0.41.4 ± 0.3 3.2 ± 0.10.8 ± 0.2 3.6 ± 0.2

J. Suave et al. / Materials Science and Engineering A 509 (2009) 57–62 61

Fig. 5. Fracture surfaces of typical c-SWNT/epoxy nanocomposites [165 W/20 min/0.25 wt%] (a) and [225 W/40 min/0.25 wt%] (b), and the epoxy/THF sample (c).

Table 5Number average (Mn) molecular masses, and the concentration of monomer and oligomers obtained from GPC curves of sonicated and non-sonicated epoxy samples.

Sonication power/time Mn(monomer) (g mol−1) Mn(oligomer) (g mol−1) Concentration of monomer (%) Concentration of oligomers (%)

Non-sonicated 373 793 87.4 12.6165 W/20 min 351 751 94.3 5.7212

tacdwtt

4

pp

25 W/20 min 345 72965 W/40 min 357 66325 W/40 min 360 745

he oligomer concentration, what is possibly related to branchingnd crosslinking processes due to the recombination of radi-als. On the other hand, the number-average molecular masseso not vary significantly with the sonication parameters. Any-ay, sonication of epoxy samples under the conditions used in

his work, does not cause a considerable change in resin proper-ies.

. Conclusions

A systematic statistical study of the influence of sonicationarameters and nanotube content on thermal and mechanicalroperties of c-SWNT/epoxy nanocomposites was carried out using

89.3 10.791.5 8.589.9 10.1

a 23 factorial design. The route chosen for the preparation ofnanocomposites was proved to be not strict enough, allowing thepresence of some trapped solvent (THF) in the cured samples, asconfirmed by TG measurements.

Tensile tests showed an increase in Young’s modulus (up to30%) with the addition of 0.25 wt% c-SWNT. Tensile strength andelongation at break results, however, suggested that dispersionand/or c-SWNT/matrix adhesion should be enhanced to achieve

more satisfactory mechanical properties. The findings of the sta-tistical analysis showed that the lowest sonication power (165 W),the longest sonication time (40 min) and the highest c-SWNT con-tent (0.25 wt%) were the best conditions to prepare c-SWNT/epoxynanocomposites using THF as solvent.

6 and E

td(nbo

A

0As

R

[

[

[

[

[[

2 J. Suave et al. / Materials Science

DMA results have shown an increase in storage modulus forhe nanocomposites, reaching 3.0 GPa for the best sonication con-itions and 0.25 wt% c-SWNT content. Analysis of the peak factor� ), calculated from the tan ı curves, suggests that the presence ofanotubes promoted a more homogeneous epoxy network, proba-ly due to the minimization of residual solvent, which corroboratesther findings of this work.

cknowledgments

The authors would like to thank CAPES-PROCAD (Project303054) and the Brazilian Space Agency, (AEB Uniespaco, ProjectO-01/2006) for financial support, and CAPES-DS for the scholar-hip to Ms. Jaqueline Suave.

eferences

[1] S. Iijima, Nature 354 (6348) (1991) 56–58.[2] P.M. Ajayan, Chem. Rev. 99 (7) (1999) 1787–1799.[3] F.H. Gojny, K. Schulte, Compos. Sci. Technol. 64 (15) (2004) 2303–2308.

[[[

[

ngineering A 509 (2009) 57–62

[4] J. Sandler, M.S.P. Shaffer, T. Prasse, W. Bauhofer, K. Schulte, A.H. Windle, Polymer40 (21) (1999) 5967–5971.

[5] L.J. Ci, J.C. Bai, Compos. Sci. Technol. 66 (3–4) (2006) 599–603.[6] E.T. Thostenson, T.W. Chou, Carbon 44 (14) (2006) 3022–3029.[7] Y.S. Song, J.R. Youn, Carbon 43 (7) (2005) 1378–1385.[8] M.R. Loos, L.A.F. Coelho, S.H. Pezzin, S.C. Amico, Polímeros 18 (1) (2008)

76–80.[9] M.H. Liu, Y.L. Yang, T. Zhu, Z.F. Liu, Carbon 43 (7) (2005) 1470–1478.10] M.R. Loos, S.H. Pezzin, S.C. Amico, C.P. Bergmann, L.A.F. Coelho, J. Mater. Sci. 43

(18) (2008) 6064–6069.11] M.R. Loos, S.C. Amico, S.H. Pezzin, L.A.F. Coelho, Mater. Res. 11 (3) (2008)

347–352.12] K.T. Lau, M. Lu, C.K. Lam, H.Y. Cheung, F.L. Sheng, H.L. Li, Compos. Sci. Technol.

65 (5) (2005) 719–725.13] J.F. Shen, W.S. Huang, L.P. Wu, Y.Z. Hu, M.X. Ye, Compos. Part A - Appl. S 38 (5)

(2007) 1331–1336.14] H. Miyagawa, L.T. Drzal, Polymer 45 (15) (2004) 5163–5170.15] Y.X. Zhou, F. Pervin, L. Lewis, S. Jeelani, Mater. Sci. Eng. A - Struct. 452 (2007)

657–664.16] P.C. Ma, J.K. Kim, B.Z. Tang, Compos. Sci. Technol. 67 (14) (2007) 2965–2972.17] P. Gonnet, Masters Dissertation, The Florida State University, USA, 2004.18] F.H. Gojny, M.H.G. Wichmann, U. Kopke, B. Fiedler, K. Schulte, Compos. Sci.

Technol. 64 (15) (2004) 2363–2371.19] S.A. Garea, A.C. Corbu, C. Deleanu, H. Iovu, Polym. Test 25 (2006) 107–113.