Effect of dispersion method on tribological properties of carbon nanotube reinforced epoxy resin composites

Effect of dispersion method on tribological properties of carbon nanotube reinforced epoxy resin composites

ARTICLE IN PRESS POLYMER TESTING Polymer Testing 26 (2007) 351–360 www.elsevier.com/locate/polytest Material Properties Effect of dispersion method...

696KB Sizes 0 Downloads 0 Views

ARTICLE IN PRESS

POLYMER TESTING Polymer Testing 26 (2007) 351–360 www.elsevier.com/locate/polytest

Material Properties

Effect of dispersion method on tribological properties of carbon nanotube reinforced epoxy resin composites Haiyan Chena, Olaf Jacobsb,, Wei Wua, Gerrit Ru¨digerb, Birgit Scha¨delb a

East China University of Science and Technology, Xuefu Road 1300, Jingshan District Shanghai 201512, PR China b Fachhochschule Lu¨beck, Stephensonstr. 3, 23562 Lu¨beck, Germany Received 13 October 2006; accepted 25 November 2006

Abstract Epoxy/CNT nanocomposites were synthesized in various ways to examine the effects of dispersion methods on its tribological properties. The carbon nanotubes (CNTs) were pre-treated in three different ways: no pre-treatment, activation in HNO3, and activation in HNO3 plus application of a coupling agent. The dispersion and mixing methods used were: dual asymmetric centrifuge, sonication, hand mixing. The CNTs were mixed either into the hardener or the resin. The curing behaviour was studied via DSC, and the thermomecanical properties were determined using a DMA. The tribological properties were investigated in a ball-on-prism test rig under unidirectional continuous sliding against austenitic stainless steel. For the untreated CNTs, it seems that the wear resistance improves with increasing effort put into dispersion: sonication has a positive effect, sonication plus dual asymmetric centrifuge proves even better. A pre-treatment with HNO3 or a silane-coupling agent can improve the wear resistance of the composite. However, the pre-treated CNTs should be dispersed without the use of ultrasound, which seems to damage the pre-treated CNTs. The results showed that the wear resistance in general increases with improved dispersion and integrity of the CNTs. r 2006 Elsevier Ltd. All rights reserved. Keywords: Carbon nanotubes (CNTs); Epoxy resin; Polymer composites; Nano composites; Wear

1. Introduction Carbon nanotubes (CNTs) are increasingly attracting scientific and industrial interest by virtue of their outstanding characteristics. The CNT walls resemble rolled-up graphite-like sheets with strong covalent sp2 bonds. According to their graphitic structure, CNTs possess high thermal conductivity and an electrical conductivity that can be either semi-conducting or Corresponding author. Tel.: +49 451 300 5323; fax: +49 451 300 5302. E-mail address: [email protected] (O. Jacobs).

0142-9418/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.polymertesting.2006.11.004

metal like. The Young’s modulus of carbon nanotubes can be as high as 1000 GPa, which is approximately five times higher than steel. The tensile strength of carbon nanotubes can be up to 150 GPa, around 40 times higher than steel [1]. The combination of the previously mentioned material properties together with an aspect ratio in the range of several thousands makes CNTs promising candidates as reinforcement for polymer composites. In addition, the development of CNT/polymer nanocomposites opens new perspectives for multi-functional materials. Research results on the effect of CNTs on the tribological properties of several polymers have been reported. It was found

ARTICLE IN PRESS 352

H. Chen et al. / Polymer Testing 26 (2007) 351–360

that CNT-based nanocomposites exhibit a better wear resistance compared with the pure matrix [2–5]. All studies reported a beneficial effect of CNT reinforcement on the wear resistance of the polymers. Epoxy resins (EP) have been widely used in practical applications such as adhesives, construction materials, composites, laminates and coatings owing to their excellent mechanical properties, low cost, ease of processing, good adhesion to many substrates, and good chemical resistance. A previous study focused on the effect of CNT reinforcement on the tribological performance of EP composites. It was found that 1 wt% CNTs in EP yield optimal tribological performance [6]. The large diameter-dependent specific surface area of up to 1300 m2/g and the non-polar nature of this surface pose some problems concerning the dispersion of these CNTs in polymers. Various investigations focusing on the mixing methods of CNTs in the polymer matrix have been performed. Hesheng Xia et al. [7] used ultrasonic agitation and in situ emulsion polymerisation to prepare stable composite emulsions of CNTs in poly (methyl methacrylate-co-n-butyl acrylate), which allows the dispersion of CNTs in latex. Gojny et al. [8] obtained good dispersion of carbon nanotubes in epoxy resin by using a mini calender. Moreover, this method enables the manufacture of large amounts of nanocomposites. There are relatively few studies that systematically investigate the different methods of dispersing CNTs in the epoxy to research the tribological properties of the nanocomposite. In this paper, we describe the influence of different dispersion methods of CNTs in epoxy resin. In the first step, it was checked whether the CNTs could be better dispersed in the resin or in the hardener. Secondly, based on the best mixing sequence of step one, the dispersion method of CNTs in the resin was varied systematically. Two different pre-treatments of the CNTs were experimentally investigated: untreated CNTs, pre-treatment by boiling in HNO3, and the additional application of a silane as coupling agent. The tribological properties of the resulting CNT/epoxy nanocomposite were tested and will be discussed.

medium viscosity resin with good heat resistance. Neukadur T9 was used as hardener. Resin and hardener were provided by Altropol, Stockelsdorf, Germany. The CNTs were obtained from Nanmigang Co., Shenzhen, China. The properties of the CNTs were: purity495%, average diameter 10–30 nm, average length 5–15 mm and a specific surface area between 40 and 300 m2/g. A coupling agent was used to modify the CNTs’ surfaces. Dynasylan GLYMO from Degussa, Germany is a 3-glycidyloxypropyltrimethoxsilane, a bifunctional organosilane possessing a reactive organic epoxide to react with the matrix resin and hydrolysable inorganic methoxysiyl groups, which can form covalent bonds with the substrate. Fig. 1 shows the structure of the coupling agent used. 2.2. CNT treatment For all the mixtures, the amount of CNTs in epoxy resin was 1 wt%. Before mixing, the CNTs were dried at 100 1C for more than 2 h and were manually crushed with a mortar for 10 min before mixing. CNTs that were processed in this way were called ‘‘untreated’’. To enhance the dispersibility of CNTs in liquids, it is necessary to physically or chemically attach polar functional groups to the sidewalls without significantly changing the nanotubes’ desirable properties. This process is called functionalisation. Functionalisation methods such as plasma treatment, oxidation, or coating with coupling agents improve the wettability and create active bonding sites. Some mixtures were prepared with CNTs that were pre-treated in HNO3. The pure CNTs were again put into oven at 100 1C for 2 h. Then, the dried CNTs were stirred into a HNO3 solution (4 mol/l). The weight ratio of CNTs to HNO3 was 1:3. This mixture was boiled at 100 1C for 1–2 h while stirring at 300 rpm. To eliminate HNO3, the mixture was washed in distilled water until the pH

2. Experimental 2.1. Raw materials The resin used for this study was Neukadur EP571 based on bisphenol A, a highly reactive

Fig. 1. Structure of the GLYMO molecules.

ARTICLE IN PRESS H. Chen et al. / Polymer Testing 26 (2007) 351–360

value approached 7 and was put into an oven at 100 1C to dry. Some CNT samples were additionally modified by a coupling agent. First, the CNTs were again dried and pre-treated in HNO3. This pre-treatment is necessary to graft functional groups to the CNT sidewalls that can react with the silane. Then, the pre-treated CNTs were sonicated in a solution of the coupling agent in acetone (13.6 mol/l) in order to avoid agglomerates of CNTs and to evaporate the solvent. The weight ratio of coupling agent to CNTs was 1:10. The sonication process was continued until all the solution was completely evaporated. After this process, the CNTs appeared like very fine powder, without tactile lumps, that could be dispersed in the resin. 2.3. Dispersion and mixing methods A dual asymmetric centrifuge (Speed Mixers DAC 150FVZ from Hauschild, Germany) was used to mix the CNTs into the hardener or the resin. The mixture was put into a small pot, which was inserted into the Speed Mixer for 2 min at 3000 rpm. During mixing, the sample is exposed to high shear forces and impacts, while only little air is introduced into the mixture. After mixing, the samples were put into a vacuum chamber and evacuated to get the entrapped air out of the mixture. Finally, the hardener (or resin) was added into the mixture at the resin-to-hardener weight ratio of 4:1. Two methods were used to mix resin and hardener. One was stirring by hand, which had to be done gently to prevent re-mixing of air into the sample. The other mixing method was again the Speed Mixer at 1000 rpm for 1 min. The finished mixture was again evacuated to eliminate air bubbles. When no more bubbles formed during evacuation, the mixture was cast into a silicone mould. The samples were cured in an oven with a cycle of: heating from 20 to 60 1C in 2 h, heating from 60 to 80 1C in 2 h, holding at 80 1C for 4 h. Another mixing method was sonication. The CNTs were dispersed in the resin using a sonotrode. The device was a UP 400 s from Dr. Hielscher GmbH, Teltow, Germany. The cylindrical titanium sonotrode had a diameter of 7 mm. The sonication process was performed in 20 cycles, each lasting 30 s. The other processing steps were the same as mentioned above. For a better de-agglomeration, one mixture was produced by combining the above two dispersion

353

methods: after the Speed Mixer dispersion, the mixture was sonicated. 2.4. Test methods 2.4.1. Viscosity test The viscosity of the resin/CNT mixture will give some information about the degree of dispersion of CNTs in the matrix. The viscosity of the samples was measured using a cone and plate rheometer (PHYSICA UM, Germany). Firstly, a small amount of the mixture was placed between the plate and the cone. The cone used here had a diameter of 12.5 mm, an opening angle of 11 and a truncation of 50 mm. Accordingly, the distance between plate and flattened cone was 50 mm. The viscosity was measured at a shear rate ranging from 10 to 4500 1/s. For each sample, the testing time was 10 min and the measurement was conducted at 30 1C. 2.4.2. DSC The curing behaviour of the CNTs/epoxy nanocomposites was determined using a DSC 204F1, NETZSCH, Germany. About 10 mg of the samples was sealed in aluminum pans. The finished mixture including resin and hardener was heated up to 80 1C at a rate of 10 K/min. After this curing temperature was reached, the reaction enthalpy was measured during an isothermal phase of 4 h at 80 1C. 2.4.3. Mechanical tests The study of the thermo-mechanical behaviour was performed by dynamic mechanical thermal analysis (DMA), using a DMA 242C, NETZSCH, Germany. Rectangular specimens, 50 mm long, 5 mm wide, and 4 mm thick, were prepared for the measurements. The samples were exposed to an oscillating three-point bending load between 30 and 250 1C at a heating rate of 5 1C/min, the frequency was set to 1 Hz, the load amplitude was 80 mm. The impact strength of the composites was measured with unnotched specimens on a Charpy-560 impact strength-testing machine. Five specimens of each composition were tested and the average values of the data are reported. 2.4.4. Wear tests All tests were performed in a ball-on-prism tribometer from Dr. Tillwich GmbH Werner Stehr, Horb, Germany according to ISO7148-2. The specimens were subjected to uniform and unidirectional

ARTICLE IN PRESS H. Chen et al. / Polymer Testing 26 (2007) 351–360

354

sliding. The samples were glued on the inner surfaces of the prism with 901 opening angle. The prism was fixed at one end of a lever and pressed against a steel bearing ball with a diameter of d ¼ 12.7 mm. (Fig. 2). The balls consisted of X5CrNi18-10 and had a roughness Rz ¼ 0.31 mm and a hardness of 375HV. Dead weights of 30 N were attached to the levers, resulting in a normal load of FN ¼ 21.2 N acting on each specimen surface (Fig. 3). The rotational frequencywas pffiffiffi f ¼ 1 Hz, resulting in a sliding speed of v ¼ pd 2; f ¼ 28:2 mm=s. The wear tests lasted 60 h. The data acquisition was performed via inductive displacement transducers. The resulting penetration depth vs. time curve was converted into a diagram that plotted volume loss (V) as a function of sliding distance (S). After some time, the wear process entered a steady state and the V vs. S curve became linear. The specific wear rate ks was calculated by dividing the slope (dV/dS) of the linear part of the curve by the normal load, FN ks ¼

system, composites were produced in six different ways (see Table 1). For all six systems, the dispersion of CNTs was performed by a Speed Mixer. Fig. 4 compares the viscosity of some mixtures as a function of shear rate. The hardener, pure as well as filled with CNTs, exhibited a Newton flow behaviour, i.e., the viscosity is more or less independent of the shear rate. The resin (pure and with added CNTs) shows some shear thinning due to its larger molecules. Adding 1 wt% CNTs increases the viscosity in both cases, but does not significantly change the slope of the curve. In the case of the hardener, CNTs affect the viscosity more strongly than in the case of the resin. The reason for the increased viscosity is the tendency of the CNTs to form agglomerates and networks inside the mixture.

dV . F N dS

Three specimens of each sample were tested. 3. Results and discussion 3.1. Influence of mixing sequence To find out whether it is better to disperse the CNTs in hardener, resin or in the resin-hardener

Fig. 3. Sketch of the loading situation.

Fig. 2. Sketch of the ball-on-prism tribometer and prism with specimens and counterpart ball.

ARTICLE IN PRESS H. Chen et al. / Polymer Testing 26 (2007) 351–360 Table 1 Mixing procedures Sample

Mixing procedures

A1

untreated CNTs in hardener by Speed Mixer, then mix resin into hardener by Speed Mixer untreated CNTs in hardener by Speed Mixer, then stir resin into hardener by hand untreated CNTs in resin by Speed Mixer, then mix hardener into resin by Speed Mixer untreated CNTs in resin by Speed Mixer, then stir hardener into resin by hand resin and hardener homogenised by Speed Mixer, then disperse untreated CNTs into mixture by Speed Mixer resin and hardener homogenised by Speed Mixer, then untreated CNTs stirred into mixture by hand

A2 A3 A4 A5

A6

Viscosity [Pa*s]

2.5 CNTs in resin by Speed Mixer,untreated CNTs in hardener by Speed Mixer,untreated pure resin pure hardener

2 1.5 1 0.5 0 0

1000

2000

3000

4000

5000

Shear Rate[1/s] Fig. 4. Viscosity–shear rate curves of pure resin and hardener and of mixtures of CNTs in hardener and resin, respectively.

Fig. 5 shows the curing behaviour of different samples. Fig. 5(a) compares the curing behaviour of the pure epoxy resin with that of the epoxy resin including untreated CNTs. In both cases, the CNTs were dispersed with the Speed Mixer. Fig. 5(a) shows that the curing behaviour of untreated CNTs/epoxy resin is the same as that of the pure epoxy resin during the first 10 min. After that, the heat production of the untreated CNTs/epoxy resin is slightly higher than that of the pure epoxy resin. That is to say, the untreated CNTs/epoxy resin produces more heat to cure completely compared to the pure ones. This is mainly because the CNTs have a large specific surface area, and it can act as a desirable interface for stress transfer between CNTs and the resin, and also induces strong attractive forces. The mixing sequence has only little influence on the curing behaviour, as illustrated in Fig. 5(b). Almost no difference in the curing curves can be observed among the six different samples.

355

Fig. 6 shows the loss factor tan d and storage modulus as a function of temperature. All samples have virtually the same glass transition temperature (peak of loss factor curves) with the exception of A3, having a slightly higher Tg. This increase in thermal stability can be interpreted as a reduction of the mobility of the polymer molecules through the nanotubes. Dispersing the CNTs in resin and adding the hardener by Speed Mixer may improve the interfacial strength between CNTs and epoxy matrix, thus reducing the mobility of the matrix molecules. According to Fig. 6(b), the samples A1 and A3 show the highest storage modulus of up to 3800 MPa. The high storage modulus of sample A3 is in agreement with the assumed good adhesion between matrix and CNTs. A6 (dispersing CNTs by hand in homogenised mixture of resin and hardener) has the lowest modulus of about 3000 MPa. The specific wear rates of the A series samples are shown in Table 2 and are illustrated in Fig. 7. A1 (CNTs dispersed in hardener first) and A3 (CNTs dispersed in resin first) differed insignificantly from one another and delivered the best results. This means that both mixing steps (CNTs into hardener or resin, respectively, and addition of second component) should be performed with the dual asymmetric centrifuge, which effectively crushes the agglomerates and homogenises the mixture. When this is obeyed, it is of minor consequence whether the CNTs are dispersed in the resin or in the hardener first. A3, which seems to have the best interface strength, also has the lower specific wear rate. Key issues for the wear performance of CNT reinforced composites are obviously interfacial bonding and the proper dispersion of CNTs in the matrix. The results of the impact tests with unnotched specimens of the A series samples are shown in Fig. 8. The addition of CNTs tends to reduce the impact strength of the epoxy resin. Possible reasons are the inherent brittleness of the CNTs, their tendency to form agglomerates, with the resulting stress concentrations in their vicinity, and residual air that may still be entrapped inside the agglomerates, even though the mixtures were thoroughly degassed. Samples A3 and A6, which displayed good wear resistance as well, reached an impact strength comparable to that of the pure epoxy. However, there seems to be no clear correlation between impact strength and wear resistance, since the samples A1 and A2, which also had low or

ARTICLE IN PRESS H. Chen et al. / Polymer Testing 26 (2007) 351–360

356

a 1

Heat Flow exo [mW/mg]

untreated CNTs in resin, add hardener by Speed Mixer pure resin and pure hardener by Speed Mixer

0.8

0.6

0.4

0.2

0 0

20

40

60

80

100

120

Time (min)

Heat Flow exo [mW/mg]

b

0.8 A1 A2 A3 A4 A5 A6

0.6

0.4

0.2

0 0

20

40

60

80

100

120

Time (min) Fig. 5. Curing behaviour of different samples: (a) pure epoxy resin and epoxy resin with 1 wt% untreated CNTs prepared with the Speed Mixer and (b) samples of EP with 1% CNTs, prepared in various ways.

medium wear rates, had very brittle behaviour in the impact tests. 3.2. Effect of dispersion methods Three different dispersion methods were investigated in detail: the dual asymmetric centrifuge, sonication, and a combination of these two. Stirring was not further considered because of the superior results of the Speed Mixer. The three dispersion methods were applied to three series of CNTs: untreated CNTs, HNO3-treated CNTs and CNTs

pre-treated with coupling agent (Table 3). The samples of the D-series were boiled in nitric acid before the treatment with the coupling agent in order to form functional groups on the CNTs to react with the silane molecules. It should be mentioned that samples D2 and D3 were exposed twice to sonication: the first time to disperse the CNTs in the acetone/silane solution and a second time to disperse the fine powder of pre-treated CNTs in the resin. DMA results of the B-series (untreated CNTs in epoxy) are shown in Fig. 9. The glass transition

ARTICLE IN PRESS

Loss Factor, tan δ

a

1

Impact Energy [mJ/mm2]

H. Chen et al. / Polymer Testing 26 (2007) 351–360

A1 A2 A3 A4 A5 A6

0.8 0.6 0.4

357

25 20 15 10 5 0

0.2

A1

A2

A3

A4

A5

A6 Pure EP

Sample

0 0

20

40

60

80

100 120 140 160 180

Fig. 8. Impact strength of A series samples.

Temperature (°C))

Young's Modulus [MPa]

b

4500 A1 A2 A3 A4 A5 A6

4000 3500 3000 2500 2000 1500 1000 500 0 20

40

60

80

100

120

140

160

Temperature (°C)

Fig. 6. DMA result of the A series samples: (a) loss factor curve and (b) Young’s modulus curve.

Table 2 Specific wear rate of A series samples A1

A2

A3

A4

A5

A6

kS (106 mm3/Nm)

1.03

1.97

1.06

3.34

2.10

1.69

Spec.Wear Rate [10-6 mm3/Nm]

Sample

4 3 2 1 0 A1

A2

A3

A4

A5

Sample

Fig. 7. Specific wear rates of the A series samples.

A6

temperature, Tg, of B2 and B3 was increased by about 40 1C compared to B1. Both B2 and B3 were prepared by the use of ultrasound, whereas B1 was produced exclusively with the dual asymmetric centrifuge. Moreover, the sample produced with a combination of ultrasound and Speed Mixer displayed also the highest storage modulus. From these results, it can be concluded that sonication is very effective in disintegrating the agglomerates. Figs. 10 and 11, respectively, show the DMA results of the samples prepared with HNO3-treated CNTs and the composites produced with CNTs pretreated with coupling agent. The three samples of the C-series (HNO3-treated) delivered nearly identical loss curves. The Tg values are around 140 1C and virtually identical with the glass transition temperatures of B2 and B3 (untreated CNTs dispersed by ultrasound). The pre-treatment in nitric acid apparently improves the dispersion of the CNTs to a similar degree as the speed mixing and sonication of untreated CNTs. The Young’s modulus of C1 (Fig. 10(b)) is about 1000 MPa higher than that of the other two C-samples. The de-agglomerating effect of the sonication, is probably superimposed by rupture of the CNTs caused by the high local energy input, resulting in a reduction of the effective tube length [9,10]. The D-series (pre-treated with coupling agent after activation in nitric acid) showed a similar trend: D1 (CNTs dispersed in resin by Speed Mixer) displayed the highest Young’s modulus. The glass transition of D1 and D2 was again around 140 1C, while the Tg of sample D3 (sonication plus Speed Mixer) was about 20 1C lower. The different behaviour of three sample series is evidence for the influence of the chemical functionalisation of the surface on the interfacial adhesion between CNTs and the epoxy resin.

ARTICLE IN PRESS H. Chen et al. / Polymer Testing 26 (2007) 351–360

358

Table 3 Three different dispersion methods applied to three differently pre-treated CNTs in epoxy resin Pre-treatment

Dispersion method

Untreated CNTs HNO3-treated CNTs Coupling agent

CNTs in resin by US add hardener by Speed Mixer

CNTs in resin by Speed Mixer and US add hardener by Speed Mixer

B1 C1 D1

B2 C2 D2

B3 C3 D3

a

1 B1 B2 B3

0.8

Loss Factor, tan δ

Loss Factor, tan δ

a

CNTs in resin by Speed Mixer add hardener by Speed Mixer

0.6 0.4 0.2 0 0

50

100

150

0.8 C1 C2 C3

0.6 0.4 0.2 0

200

0

20

40

Temperature (°C)

b 4500 4000 3500 3000 2500 2000 1500 1000 500 0

Young's Modulus (MPa)

Young's Modulus (MPa)

b B1 B2 B3

0

50

100

150

200

Temperature(°C)

4500 4000 3500 3000 2500 2000 1500 1000 500 0

60 80 100 120 140 160 180 Temperature (°C)

C1 C2 C3

0

20

40

60 80 100 120 140 160 180 Temperature (°C)

Fig. 9. DMA results of the untreated CNT/epoxy composites: (a) loss factor and (b) Young’s modulus.

Fig. 10. DMA result of the HNO3 treated CNTs/epoxy composite: (a) loss factor and (b) Young’s modulus.

The wear test results of the B-, C-, and D-series are shown in Table 4 and Fig. 12. In comparison to the specific wear rate of the pure epoxy resin (8.51  106 mm3/N m), all the CNTs/EP nanocomposites have a lower wear rate. This shows that CNTs can significantly improve the tribological properties of the epoxy resin. The different series revealed opposite trends concerning the effect of the dispersion method on the wear rate. Overall, the untreated CNTs delivered the best results, even though the CNTs with coupling agent performed

slightly better as long as no sonication was involved. The CNTs pre-treated in nitric acid, too, delivered good results when dispersed by the Speed Mixer. For the untreated CNTs, it seems that the wear resistance improves with increasing effort put into dispersion; sonication has a positive—or at least no negative—effect. In the case of the pre-treated CNTs, sonication reduces the wear resistance of the composite. This is particularly obvious for the CNTs with coupling agent: the wear rate increases with the number of sonication cycles. An

ARTICLE IN PRESS H. Chen et al. / Polymer Testing 26 (2007) 351–360

Loss Factor, tan δ

a

1

D1 D2 D3

0.8 0.6 0.4

impact strength than the other samples. Among those samples prepared with coupling-agent-treated CNTs, D3 (sonicated and speed-mixed) had the best impact strength but the highest wear rate. It seems that there is no correlation between impact strength and wear resistance, in accordance with previous work.

Spec.Wear Rate [10-6 mm3/Nm]

explanation can be given referring to Gojny et al. [11,12] who found that ultrasound does not only disintegrate the agglomerates but, additionally, breaks the CNTs themselves. The pre-treated CNTs are apparently more susceptible to this effect because boiling in nitric acid damages the outer graphite layers and produces notches. Fig. 13 presents the effect of the dispersion method on the impact strength of unnotched specimens. Samples B3 and C1, which also displayed a good wear resistance, reached a better

359

3 2.5

Speed Mixer Sonication Speed Mixer+Sonication

2 1.5 1 0.5 0 untreated

0.2

HNO3 treated

GLYMO treated

Pre-treatment

0 0

50

100

150

200

Temperature(°C)

Fig. 12. Specific wear rates of the composites with three differently pre-treated CNTs processed by different dispersion method.

4500 4000 3500 3000 2500 2000 1500 1000 500 0 -500

D1 D2 D3

0

50

100

150

25 Impact Energy [mJ/mm2]

Young's Modulus (MPa)

b Speed Mixer Sonication Speed Mixer+Sonication

20 15 10 5 0

200

untreated

Temperature (°C)

HNO3 treated

GLYMO treated

Pre-treatment Fig. 11. DMA result of the composites including CNTs pretreated with coupling agent: (a) loss factor and (b) Young’s modulus.

Fig. 13. Impact strength of three different CNTs by different dispersion method.

Table 4 Specific wear rate of three different CNTs by different dispersion method Wear rate (106 mm3/Nm)

CNTs in resin by Speed Mixer add hardener by Speed Mixer

CNTs in resin by US add hardener by Speed Mixer

CNTs in resin by Speed Mixer and US add hardener by Speed Mixer

Untreated CNTs HNO3-treated CNTs Coupling agent

1.06 0.73 0.70

0.76 1.84 1.20

0.70 0.89 2.42

ARTICLE IN PRESS 360

H. Chen et al. / Polymer Testing 26 (2007) 351–360

Within each single series, there seems to be some correlation between Tg and storage modulus on the one hand and wear resistance on the other hand. The Tg seems to be a good parameter to characterise the dispersion state of the CNTs. Most materials with a high Tg have a high storage modulus of about 400 MPa at room temperature. In some cases, the destruction of the CNTs superimposes this effect leading to materials with high Tg but low storage modulus (ca. 3000 MPa) at room temperature. The comparison of the wear test results with the DMA tests suggests that the wear resistance in general increases with improved dispersion and integrity of the CNTs. However, this pattern changes when the pre-treatment of the CNTs is altered: Sample B1, for example, delivers fairly good wear test results although its Tg and storage modulus are low. Conclusively, there are two effective ways to improve the wear properties of the CNTs/epoxy composite. One is to apply advanced dispersion: dispersing untreated CNTs in the epoxy by Speed Mixer and ultrasound. A pre-treatment with hazardous chemicals is not necessary in this case. The other way to improve wear properties is the chemical funcationalisation of CNTs: coupling agent treatment of CNTs dispersed by Speed Mixer. CNTs pre-treated in HNO3 should not be exposed to sonication. 4. Conclusions From the above discussion, some important conclusions can be drawn: (1) The dual asymmetric centrifuge efficiently disperses the CNTs in the epoxy resin. It obviously produces sufficiently high shear stresses and impacts to disintegrate the agglomerates but is mild enough to prevent the CNTs from being destroyed. (2) Ultrasound, too, effectively disperses CNTs in the epoxy resin, a combination of dual asymmetric centrifuge and ultrasound is even slightly better. However, ultrasound should be avoided after activation of the CNTs in concentrated nitric acid. It is suspected that this pre-treatment causes some damage in the outer shells of the CNTs, and that the high stress fields in the

sonication process cause partial fracture of the pre-damaged CNTs. (3) From the tribological point of view, a pretreatment with nitric acid or coupling agent is not absolutely necessary because the untreated CNTs delivered very good results. (4) The CNTs with coupling agent performed insignificantly better as long as no sonication was involved. The CNTs pre-treated in nitric acid also delivered very good wear results when dispersed by the Speed Mixer.

References [1] M. Meyyappan, CRC Press, Boca Raton, FL, 2005. [2] Ph. Werner, V. Altsta¨dt, O. Jacobs, R. Jaskulka, K.W. Sandler, S.P. Shaffer, A.H. Windle, Tribological behaviour of carbon-nanofiber-reinforced poly(ether ether ketone), Wear 257 (2004) 1006–1014. [3] B. Dong, Z. Yang, Y. Huang, H.L. Li, Study on tribological properties of multi-walled carbon nanotubes/epoxy resin nanocomposite, Tribol. Lett. 20 (2005) 251–254. [4] Y. Xue, W. Wei, O. Jacobs, B. Scha¨del, Tribological behaviour of UHMWPE/HDPE blends reinforced with multi-wall carbon nanotubes, Polym. Test. 25 (2006) 221–229. [5] L. Chang, Z. Zhang, C. Breidt, K. Friedrich, Tribological properties of epoxy nanocomposites I. Enhancement of the wear resistance by nano-TiO2 particles, Wear 258 (2005) 141. [6] O. Jacobs, W. Xu, B. Scha¨del. Wear behaviour of carbon nanotube reinforced epoxy resin composite. Tribol. Lett. 2006, in press. [7] H. Xia, G. Qiu, Q. Wang, Polymer/carbon nanotube composite emulsion prepared through ultrasonically assisted in situ emulsion polymerization, Appl. Polym. Sci. 100 (2006) 3123–3130. [8] F.H. Gojny, M.H.G. Wichmann, U. Ko¨pke, B. Fiedler, K. Schulte, Carbon nanotube-reinforced epoxy-composite— enhanced stiffness and fracture toughness at low nanotube contents, Compos. Sci. Technol. 64 (2004) 2363–2371. [9] K.L. Lu, M. Lago, Y.K. Chen, M.L.H. Green, P.J.F. Harris, S.C. Tsang, Mechanical damage of carbon nanotubes by ultrasound, Carbon 34 (1996) 814–816. [10] M. Zhang, M. Yudasaka, A. Koshio, S. Iijima, Effect of polymer and solvent on purification and cutting of single-wall carbon nanotubes, Chem. Phys. Lett. 349 (2001) 25–30. [11] F.H. Gojny, M.H.G. Wichmann, B. Fiedler, K. Schulte, Influence of different carbon nanotubes on the mechanical properties of epoxy matrix composites—A comparative study, Compos. Sci. Technol. 45 (2005) 2300–2313. [12] K.B. Shelimov, R.O. Esenllev, A.G. Rinzler, C.B. Huffmann, R.E. Smalley, Purification of single-wall carbon nanotubes by ultrasonically assisted filtration, Chem. Phys. Lett. 282 (1998) 429–434.