CARBON

4 9 ( 2 0 1 1 ) 3 5 6 0 3 5 7 0

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/carbon

Improving the actuating response of carbon nanotube/ionic liquid composites by the addition of conductive nanoparticles
Takushi Sugino *, Kenji Kiyohara, Ichiroh Takeuchi, Ken Mukai, Kinji Asaka
Health Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Ikeda, Osaka 563-8577, Japan

A R T I C L E I N F O

A B S T R A C T

Article history: Received 21 December 2010 Accepted 7 April 2011 Available online 27 April 2011

Polyaniline (PANI) or carbon black (CB) was used as a conductive additive in porous electrodes of a dry-type carbon nanotube (CNT)-polymer actuator to improve the actuation properties. We estimated the strain from the bending motion of the actuator in the frequency range of 2000.005 Hz. The generated stress was calculated from the Youngs modulus of the electrode lm and the maximum strain. Eight different types of electrode lms were prepared by changing the mixed amount of PANI or CB into the CNT electrode lm. When a same amount of PANI (50 mg) against CNT (50 mg) was added to the CNT electrode (in the case of CNT/PANI(50/50)), the strain of CNT/PANI(50/50) was increased to be almost three times larger than that of CNT(50) (without any additives). As a result, CNT/PANI(50/50) showed more than ve times the generated stress compared to CNT(50). Using 40 mg of CB as an additive, CNT/CB(50/40) showed a more than four times larger stress compared to CNT(50). The electrochemical properties such as conductivity and capacitance were also investigated and discussed. An asymmetrical three-layered actuator element was fabricated to investigate the actuation mechanism. The results imply that the anode contracts and the cathode expands. 2011 Elsevier Ltd. All rights reserved.

1.

Introduction

Carbon electrode is a key material for Li-battery [1,2], supercapacitors [3,4], and actuators [5,6]. Among carbon materials, especially, single-walled carbon nanotube (SWCNT) has been focused as a practically promising carbon species because of its high conductivity, large surface area, large capacitance, and high mechanical stiffness [79]. During a couple of decades, electroactive polymer (EAP) actuators represented by conductive polymers [1014], ionic polymermetal composites (IPMCs) [1518], and carbon nanotube sheets [5] have been expected as articial muscles because of their exibility, light weight, quick and large response. Recently, we have succeeded to obtain a exible polymer electrode composed by SWCNT, a base polymer (BP) and non-volatile, non-ammable room temperature ionic liquids

(ILs) and have reported dry CNT-polymer actuators with the electrode [19,20]. Our CNT actuators have a three-layered structure, that is, a gel electrolyte made up with a BP and an IL is sandwiched by two electrodes. Our actuators have characteristics as following: (i) the three-layered actuators show a bending motion toward the anode side, (ii) they are actuated with low voltages (3.0 V) in air for long time, (iii) each layer is easily obtained by a casting method, and (iv) the actuators are so tough that no delamination is seen after more than 8000 cycles bending motions [19]. And so on. In order to study the actuation mechanism, we have investigated the electrochemical and electromechanical properties of CNT actuators by using seven kinds of internal ILs [20]. It was found that the actuation response depended on the ionic resistance, the electrode resistance, and the double-layer capacitance. The volume change of cathode and anode was

* Corresponding author: Fax: +81 72 751 9628. E-mail address: takushi-sugino@aist.go.jp (T. Sugino). 0008-6223/$ - see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2011.04.056

CARBON

4 9 ( 20 1 1 ) 3 5 6 035 7 0

3561

derived from the size difference of the cation and the anion. We also have investigated the actuation mechanisms in the scale of nanometer by a method of Monte Carlo simulation [21,22] and by combination of a symmetry analysis, elasticity theory, and experimental results in the scale of micrometer [23]. From these previous works, we proposed a bending mechanism in which the steric repulsion due to the ion transfer into the electrodes besides the charge injection to SWCNT [5] is an important factor to induce the bending. Furthermore, it was revealed that the cathode expands and the anode contracts resulting in the bending motion toward the anode side [23]. Generally, actuators are required their large displacement, quick response, and high mechanical power. However, it is very difcult to improve these three parameters at the same time. For example, silver (Ag) [24] and ruthenium(IX) oxide (Ru2O) [18] nano-powders were used as additives into IMPCs in order to improve the actuation. Addition of Ag nano-powder caused the reduction of elastic modulus resulting in larger displacement. In case of using Ru2O additive, the displacement was improved at lower frequencies due to the larger capacitance derived from larger surface area of Ru2O powder. In contrast, the actuation response became slower by the addition of Ru2O because of the lower conductivity. Furthermore, the high cost and the toxicity of Ru2O are not favorable for its commercial use. Other studies have been reported to improve actuation of a conductive polymer and an IPMCs by using polymer additives as a dopant [25] and a dispersing agent [26], respectively. In order to improve the actuation performance of our CNT actuators, we have investigated the effects of additives, like polyaniline (PANI) as a conductive additive and mesoporous silica (MCM-41) as a non-conductive additive, in the electrodes [27]. Incorporation of PANI rather than MCM-41 into the electrodes induced larger capacitance, larger conductivity, and higher Youngs modulus for the electrode then the actuator with PANI showed the largest maximum strain and the largest generated stress. PANI is one of the most promising conductive polymers not only for EAP actuators [13,28,29] but also for supercapacitors [30] because of its high specic capacitance, low cost, and synthetic easiness. Especially, the capacitive behavior of PANI/nanocarbon composites has been studied [31,32]. It is interesting that the PANI/nanocarbon composites possess higher specic capacitance, specic energy, and specic power compared to those of parent PANI and nanocarbon. Furthermore, PANI has unique characteristics such as doping [33] and charge transfer [34] with CNTs. It is well known that carbon black (CB) is cheap and a commercially available conductive particle for supercapacitor [35] and Li-battery [36]. CB is also used as a ller of conductive rubbers and conductive plastics. On these backgrounds, we studied the inuence of addition of PANI into the electrode of CNT actuator in detail by changing additional amount of PANI in this paper. We can control the packing density, the electrochemical properties, and the mechanical stiffness by changing amount of PANI in the electrode. PANI used in this study was synthesised by in situ deposition on carbon black [37]. So, we also made CNT actuators by mixing different amount of CB into the electrode to compare the actuation performance in order to elucidate the role of PANI for the CNT actuator. The electro-

mechanical and electrochemical properties of eight different types of electrodes composed by CNT/PANI or CNT/CB composites were investigated and discussed.

2.

Experimental

2.1. Materials and preparation of SWCNT/IL/PVDF-HFP actuators

The SWCNTs (Puried HiPco) were purchased from Unidym Inc. Polyaniline (PANI) (20 wt.% polyaniline on carbon black) and carbon black (CB) were obtained from SigmaAldrich Co. and DENKI KAGAKU KOGYO KABUSHIKI KAISHA, respectively. The PANI has an about 40 nm particle size, conductivity of 40 S cm1, and 690 m2 g1 BET surface area. The CB (DENKA BLACK) has an about 35 nm particle size, electric resistivity (0.21 O cm), and specic surface area of 68 m2 g1. 1-Ethyl-3methylimidazolium tetrauoroborate (EMIBF4) (Fluka), polyvinylidene uoride-co-hexauoropropylene (PVDF-HFP: Kynar Flex2801) (Arkema Chemicals Inc.) were used as received without any further purication. 4-Methyl-2-pentanone (MP) P 99.5%, propylene carbonate anhydrous (PC) 99.7% (SigmaAldrich Co.), and N,N-dimethylacetamide (DMAc) 99.0% (Kishida Chem. Co.) were used as solvents. The conguration of actuator strip is depicted in Fig 1. The actuator has a three-layered structure in which an electrolyte lm is laminated by two electrode lms. The typical procedure to fabricate actuator elements is described below.

2.1.1.

Preparation of the electrode lm

In the case of CNT(50), 50.2 mg (19.9 wt.%) of SWCNTs, 121.7 mg (48.2 wt.%) of EMIBF4, and 80.6 mg (31.9 wt.%) of PVDF-HFP were dissolved into 9 mL of DMAc. The mixture was stirred at room temperature for more than 1 day and then sonicated in an ultrasonic bath (36 kHz) for more than 24 h. After sonication, a black gelatinous mixture was obtained. The gelatinous mixture was cast into a 2.5 cm 2.5 cm-size Teon mold and dried solvent on a hotplate at 50 C for 12 h, then dried furthermore at 80 C in vacuo for 3 days. Other electrode lms with SWCNT, PANI, and CB were prepared by following the same procedure mentioned above. The components of electrode lms for CNT/PANI

Electrode film: SWCNT Ionic Liquid (IL): EMIBF4 Base Polymer (BP): PVDF-HFP Additive: PANI, CB Electrolyte film: IL, BP

N CH3 BF4

N CH2CH3

EMIBF4

Gold electrode

L
Laser

CH2 CF2
0.88

CF2 CF CF3
0.12

PVDF-HFP

W
Fig. 1 Schematic view of the three-layered actuator element and chemical structures of EMIBF4 and PVDF-HFP.

3562

CARBON

4 9 ( 2 0 1 1 ) 3 5 6 0 3 5 7 0

(50/10), CNT/PANI(50/30), CNT/PANI(50/50), CB(50), CNT/ CB(50/8), CNT/CB(50/24), and CNT/CB(50/40) are summarized in Table 1. For example, CNT/PANI(50/50) means that about 50 mg of SWCNT and about 50 mg of PANI were used to make the electrode lm. The amount of SWCNT and PANI (or CB) used to make each electrode lm is represented in the parentheses. The thickness of electrode lm was 3585 lm.

electrodes was monitored by a laser displacement meter (KEYENCE model LC2100). The voltage, current, and displacement were simultaneously detected by an oscilloscope (YOKOGAWA DL708). The measured displacement (D) was converted to the strain difference (e) between two electrode layers with the following equation. e 2DW=L2 D2 1

2.1.2.

Preparation of the electrolyte lm

The gel electrolyte layer was prepared by casting 300 lL of solution composed by 100 mg of EMIBF4 and 100 mg of PVDF-HFP dissolved into the mixed solvent from MP (3 mL) and PC (250 mg). The electrolyte solution was casted into an aluminum mold (the size is 2.5 cm 2.5 cm) and dried overnight at ambient temperature then the solvent was evaporated at 80 C under reduced pressure for 3 days. The thickness of electrolyte lm was between 10 lm and 20 lm.

2.1.3.

Fabrication of three-layered actuator element

A three-layered actuator element was fabricated by heatpressing the electrode lms and the electrolyte lm. An electrolyte lm was laminated by two same electrode lms to make a symmetrical three-layered actuator element. The heat-pressing was performed at 70 C with 10 N (90 s) in rst pressing process then with 270 N (90 s) in second pressing process. Electrode lms stuck to the electrolyte lm by this thermal treating. The conguration of the three-layered actuator element was illustrated in Fig. 1. The total thickness of the actuator element was between 70 lm and 172 lm. The SEM cross-section images of CNT(50), CB(50), CNT/CB(50/40), and CNT/PANI(50/50) are shown in Fig. 2. In each actuator element, the thickness of electrolyte layer is between 11 lm and 14 lm. So, difference of total thickness of the actuator element is mainly caused by difference of thickness of the electrode layer.

where L is the free length and W is the thickness of the actuator element. The experimental detail was described elsewhere [27]. The ionic conductivity of EMIBF4 in the gel electrolyte layer was evaluated by an ac impedance spectroscopy. The impedance measurement was performed in the frequency range between 1 Hz and 10 MHz by using Solartron 1250 Impedance/Grain-phase Analyzer. The cyclic voltammetry was tested to characterize the electrochemical capacitance of the actuator electrode in a sealed two-electrode cell using a HOKUTO DENKO HSV-100 model. The conductivity of the electrode lm was studied by the four probe DC current method by means of a HOKUTO DENKO Potentio/Galvanostat HA-151 model as a galvanostat mode with a YOKOGAWA ELECTRIC FC220 model wave generator. In the four probe DC current method, the linear sweep of current was applied from the outer two probe electrodes and the voltage was detected by the inner two probe electrodes. The conductivity of the electrode lm was obtained from the slope of currentvoltage curve. A stressstrain measurement was carried out to evaluate the Youngs modulus of the electrode lm using SEIKO Instruments Inc. TMA/SS6000. The cross-section of actuator elements and morphological changes of the surfaces of electrode lms were observed by using scanning electron microscope (SEM) with JEOL JSM-6390 and JSM-6510.

3.
3.1.

Results and discussion

Displacement measurements

2.2.

Characterizations
The displacement measurements were carried out for eight actuator elements to investigate the inuence of the addition of PANI and CB nanoparticles with different amounts into the CNT electrode. As for the PANI used in this study, 20 wt.% of polyaniline was deposited and coated on the surface of carbon black particles. So, 80 wt.% of PANI consists of carbon black. Therefore, we prepared CNT/CB(50/8), CNT/CB(50/24), and CNT/CB(50/40) (these three are represented as CNT/CBsystem) against CNT/PANI(50/10), CNT/PANI(50/30), and CNT/PANI(50/50) (these three are called as CNT/PANI-system),

The actuation was evaluated by measuring the displacement of the actuator element. The displacement measurements were monitored at ambient temperature and humidity. The size of the element is 1 mm (width) 10 mm (length) clipped by gold electrodes. The alternating square-wave voltages were applied to the actuator element by a HOKUTO DENKO Potentio/Galvanostat (HA-501G) with a waveform generator (YOKOGAWA ELECTRIC model FC200). The displacement at 5 mm (free length (L)) away from the xed end of the gold

Table 1 The components for all the complexes. The wt% of each component is listed in the parentheses. Complex CNT(50) CNT/PANI(50/10) CNT/PANI(50/30) CNT/PANI(50/50) CB(50) CNT/CB(50/8) CNT/CB(50/24) CNT/CB(50/40) CNT (mg) 50.2 50.7 50.2 50.5 0 50.3 50.3 50.3 (19.9) (19.3) (17.8) (16.7) (19.4) (18.3) (17.3) PANI (mg) 0 10.4 (4.0) 30.4 (10.7) 50.5 (16.7) 0 0 0 0 CB (mg) 0 0 0 0 50.4 8.2 24.1 40.3 EMIBF4 (mg) 121.7 121.0 121.4 120.1 121.2 120.7 120.6 120.6 (48.2) (46.2) (42.9) (39.8) (48.1) (46.5) (43.8) (41.4) PVDF-HFP (mg) 80.6 80.0 80.8 80.6 80.3 80.3 80.3 80.3 (31.9) (30.5) (28.6) (26.7) (31.9) (30.9) (29.2) (27.5)

(20.0) (3.2) (8.8) (13.8)

CARBON

4 9 ( 20 1 1 ) 3 5 6 035 7 0

3563

Fig. 2 Cross-section SEM images of three-layered actuator elements: CNT(50) (a), CB(50) (b), CNT/CB(50/40) (c), and CNT/ PANI(50/50) (d). (Magnication: 650).

respectively, in order to know the pure inuence of polyaniline. The alternative square-wave voltage of 2.0 V was applied to activate the actuator element in the frequency range between 200 Hz and 0.005 Hz. The strain (e) was calculated from Eq. (1) to compare the actuation property. Fig. 3 shows the frequency dependence of the strain for the cases of the addition of PANI into the CNT electrode. Some selected strain data are listed in Table 2. The strains of CNT/ PANI-system were dramatically improved against CNT(50) (without PANI) especially when the frequency was lower than 0.1 Hz. The strains of CNT/PANI(50/10) become higher than those of CNT(50) when the frequency is lower than 10 Hz and CNT/PANI(50/10) shows the largest strains among the four actuators until the frequency is 0.05 Hz but the strain does not increase much in the lower frequency range (frequency < 0.05 Hz). On the contrary, much addition of PANI, CNT/PANI(50/30) exhibits the smallest strains in the frequency range between 200 Hz and 0.5 Hz, however, the strain increases very much as the frequency decreases then the largest strain (1.9%) is observed for CNT/PANI(50/30) at 0.005 Hz. Further addition of PANI (the case of CNT/PANI(50/

Table 2 The selected data for strains at 0.005 Hz, 0.1 Hz, 1.0 Hz, and 10 Hz for all the complexes. Complex 0.005 Hz CNT(50) CNT/PANI(50/10) CNT/PANI(50/30) CNT/PANI(50/50) CB(50) CNT/CB(50/8) CNT/CB(50/24) CNT/CB(50/40) 0.65 1.3 1.9 1.9 0.11 0.40 0.82 2.1 Strain (%) 0.1 Hz 0.37 0.88 0.52 0.78 0.045 0.38 0.59 1.1 1.0 Hz 0.15 0.19 0.092 0.19 0.021 0.13 0.29 0.33 10 Hz 0.023 0.024 0.0066 0.018 0.0055 0.029 0.060 0.053

2
1

CNT/PANI(50/50) CNT/PANI(50/30)
0.1 0.01

1.5
Strain (%)

CNT/PANI(50/10) CNT(50)
10 100 1000

1 0.5 0 0.001

0.001 1

0.01

0.1

1 10 Frequency (Hz)

100

1000

Fig. 3 Frequency dependence of strain (e) for CNT/PANIsystem. The magnication of strain in the frequency range between 1 Hz and 200 Hz is shown in the inset.

50)) improves the strain both in the middle frequency range (50.1 Hz) and in the low frequency range (0.10.005 Hz). On the contrary, the strain of CNT/PANI(50/50) becomes smaller than those of CNT(50) and CNT/PANI(50/10) in the higher frequency range (20010 Hz). CNT/PANI(50/50) possesses both advantages observed for CNT/PANI(50/10) and CNT/PANI(50/ 30) and exhibits the largest strain of 1.9% at 0.005 Hz that is almost three times larger than that of CNT(50) (0.65%). We also tried to make actuator elements with more than 50 mg of PANI as additives. However, it was quite difcult to obtain the electrode lms with more than 50 mg of PANI because the electrode lms had many cracks and wavy surface during drying process. The frequency dependence of strain for CNT/CB-system is shown in Fig. 4. CNT/CB(50/8) (the smallest addition amount of CB) shows similar strains as CNT(50) in the frequency range between 200 Hz and 0.1 Hz but the strains become smaller than those of CNT(50) when the frequency is lower than 0.1 Hz. Increased the amount of CB (the cases for CNT/

3564

CARBON

4 9 ( 2 0 1 1 ) 3 5 6 0 3 5 7 0

2.5
1

CNT/CB(50/40) CNT/CB(50/24) CNT/CB(50/8) CNT(50) CB(50)

1 10 100 1000

2 Strain (%) 1.5 1 0.5 0 0.001

0.1 0.01 0.001 0.0001

0.01

0.1

1 10 Frequency (Hz)

100

1000

Fig. 4 Frequency dependence of strain (e) for CNT/CBsystem. The magnication of strain in the frequency range between 1 Hz and 200 Hz is shown in the inset.

improvement is observed in the higher frequency range (frequency > 1 Hz) rather than in the lower frequency range (frequency < 1 Hz). On the other hand, the addition of PANI leads drastic improvement of strain in the lower frequency range (frequency < 0.1 Hz) but causes the largest depression of strain in the higher frequency range (frequency > 1 Hz), (iii) the large amount of conductive additives (50 mg of PANI and 40 mg of CB): both the addition of PANI and CB result in dramatic improvement of the strain in the lower frequency range (frequency < 1 Hz) compared to non-additive actuator (CNT(50)), however the addition of PANI reduces the strain less than CNT(50) in the higher frequency range (frequency > 5 Hz), on the contrary, the strains become larger than CNT(50) by the addition of CB in the same frequency range (frequency > 5 Hz).

3.2.
CB(50/24) and CNT/CB(50/40)), the strains are improved in the full frequency range. Compared with CNT(50), in the case of CNT/CB(50/24), larger improvement of strain is observed in higher frequency range (between 200 Hz and 1.0 Hz) but smaller improvement is seen in lower frequency range (between 0.5 Hz and 0.005 Hz)). On the other hand, compared to CNT(50), more than two times higher strains are observed in the frequency range between 10 Hz and 0.005 Hz for the case of CNT/CB(50/40). The strain is dramatically increased when the frequency is below 0.1 Hz and CNT/CB(50/40) exhibits more than three times larger strain (2.1%) than that of CNT(50) (0.65%) at 0.005 Hz although the strains in higher frequency range (frequency > 10 Hz) become smaller than those of CNT/CB(50/24). It is noteworthy that the strains of conductive additive (CB(50)) are very small (from 1/3 to 1/9) compared to that of CNT(50), nevertheless the strains of CNT/CB(50/24) and CNT/CB(50/40) are larger than those of CNT(50). In case of CNT/CB-system, further addition of CB (more than 40 mg of CB) was studied to nd optimal additional amount of CB (see Supplementary data 1). For example, we could obtain the electrodes of CNT/CB(50/50) and CNT/CB(50/100). But the strains became worse when we used 50 mg of CB (CNT/ CB(50/50)) compared with the strains of CNT/CB(50/40). Furthermore, the strains of CNT/CB(50/100) were smaller than those of CNT/CB(50/50). We also investigated the actuation performance of CNT(100). The maximum strain of CNT(100) was about 1.3 times larger than CNT(50) but was less than 1/2 compared with those of CNT/CB(50/40) and CNT/ PANI(50/50). From these results, we can say that addition of conductive nanoparticles such as PANI and CB is more effective to improve the actuation of our actuator rather than much addition of CNTs. The large (1.92.1%) strains observed for CNT/PANI(50/30), CNT/PANI(50/50), and CNT/CB(50/40) are more than twice as large as our previous result (0.9%) [19] and are comparable to those of conductive polymer actuators [38] and IPMCs actuators [18] including room temperature ionic liquids. It is summarized that (i) the small amount of conductive additives (10 mg of PANI and 8 mg of CB): the addition of PANI induces improvement of the strain (frequency < 10 Hz), however, the addition of CB reduces the strain (frequency < 0.1 Hz), (ii) the middle amount of conductive additives (30 mg of PANI and 24 mg of CB): the addition of CB affords good improvement of strain in the whole frequency range but much

Electrochemical properties Capacitances and conductivities of the electrode lms

3.2.1.

The capacitance (C) and the conductivity (j) of each electrode lm were measured and estimated by cyclic voltammogram (CV) and by the four-probe DC current method, respectively. The CVs at a scan rate of 10 mV s1 with the potential range between 0.5 V and 2.0 V for all the actuator electrodes are depicted in Fig. 5. The estimated values of C and j are listed in Table 3. The capacitance of CB(50) (0.00324 F cm2) is very small; one-magnitude less than that of CNT(50) (0.0508 F cm2). The conductivity of CB(50) (0.076 S cm1) is extremely small; about two-magnitude smaller than CNT(50) (4.1 S cm1). The capacitance increases by the addition of conductive nanoparticles and it also increases with increasing the amount of PANI or CB. The addition of PANI rather than CB is effective for the capacitance of the electrode. For example, the capacitance of CNT/PANI(50/50) is 3.9 times as large as that of CNT(50) and the capacitance of CNT/CB(50/40) is 2.3 times larger than that of CNT(50). The similar positive tendency is seen for the electrode conductivity by the addition of PANI. The conductivity of the electrode with CNT/PANI-system is improved with increasing the amount of PANI additive, on the contrary, the conductivity of CNT/CB-system become large up to 24 mg addition of CB, more than twice (9.5 S cm1) as large as CNT(50) (4.1 S cm1) then it is decreased to be 4.6 S cm1 for CNT/CB(50/40) that is almost the same conductivity as CNT(50). The CVs with potential range between 0.5 V and +0.5 V (Fig. 5(a) and (d)) have rectangular shapes. No Faradaic peaks were observed. This means that in this potential range, the electrodes consisted of SWCNT, EMIBF4, and PVDF-HFP with PANI or CB as conductive additives have a feature of electrochemical double-layer capacitor. When we compare the CVs of CNT/PANI-system to those of CNT/CB-system, all the CVs of CNT/CB-system have similar rectangular shapes with very sharp slopes at both ends of applied potential (Fig. 5(d)). This indicates that the ionic transfer of EMIBF4 occurs almost at the same speed for all the CNT/CB-system. On the contrary, the current at 0.0 V is increased very much by increasing the additional amount of PANI (Fig. 5(a)). The slope at the both ends of CNT/PANI(50/30) is not sharp compared to others. This implies that the ionic diffusion in the electrode for CNT/PANI(50/30) is slower than the others. This result supports the smaller strain of CNT/PANI(50/30) in higher fre-

CARBON

4 9 ( 20 1 1 ) 3 5 6 035 7 0

3565

(a)

0.5 0.4 0.3 0.2 0.1

(d)

0.20 0.15 0.10 0.05

I (mA)

0.0 - 0.6 - 0.4 - 0.2 - 0.1 0.0 - 0.2 - 0.3 - 0.4 - 0.5 0.2 0.4 0.6

- 0.6

- 0.4

- 0.2

0.00 - 0.05 0.0 - 0.10

0.2

0.4

0.6

CNT/PANI-all,

0.5V

- 0.15 - 0.20
CNT/CB-all, 0.5V

V (V)

(b)

1.0

(e)

0.3 0.2

0.5

0.1
0.0 - 1.5 - 1.0 - 0.5 - 0.5 0.0 0.5 1.0 1.5

0.0 - 1.5 - 1.0 - 0.5 - 0.1 - 0.2 0.0 0.5 1.0 1.5

- 1.0

CNT/PANI-all,

1.0V

CNT/CB-all,

1.0V

- 0.3

(c)

2.0 1.5 1.0 0.5 0.0

(f)

2.0 1.5 1.0 0.5

- 3.0

- 2.0

- 1.0

- 0.5 - 1.0 - 1.5 - 2.0

0.0

1.0

2.0

3.0

- 3.0

- 2.0

- 1.0

0.0 - 0.5 0.0 - 1.0

1.0

2.0

3.0

CNT/PANI-all,

2.0V

- 1.5 - 2.0

CNT/CB-all,

2.0V

Fig. 5 Cyclic voltammograms of CNT/PANI-system and CNT/CB-system. Scan rate: 10 mV/s. CVs for CNT/PANI-system; Potential limit, 0.5 V (a), 1.0 V (b), 2.0 V (c). CNT(50) (sky blue), CNT/PANI(50/10) (yellow), CNT/PANI(50/30) (blue), CNT/ PANI(50/50) (pink). CVs for CNT/CB-system; potential limit, 0.5 V (d), 1.0 V (e), 2.0 V (f). CB(50) (brown), CNT(50) (sky blue), CNT/CB(50/8) (yellow), CNT/CB(50/24) (blue), CNT/CB(50/40) (pink). (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

Table 3 Estimated capacitance (C) and conductivity (j) of the electrode. Complex CNT(50) CNT/PANI(50/10) CNT/PANI(50/30) CNT/PANI(50/50) CB(50) CNT/CB(50/8) CNT/CB(50/24) CNT/CB(50/40)
a

C (F cm2)a 0.0508 0.0852 0.152 0.196 0.00324 0.0624 0.0614 0.119

j (S cm1) 4.1 8.6 10 15 0.076 7.8 9.5 4.6

Potential limit: 2.0 V. Scan rate: 10 mV s1.

quency range (2001 Hz) (see Fig. 3). In case of 1.0 V potential range (Fig. 5(b)) and (e)), the CVs are still close to rectangular shapes but the cathodic and the anodic charging currents are increasing, especially at around limited voltage (1.0 V). The increasing of charging current is more remarkable when the potential range is 2.0 V (Fig. 5(c) and (f)). Furthermore, small peaks are seen between +1.0 V and +2.0 V (1.0 V and 2.0 V). It is considered that these increasing of charging currents and peaks are due to the electrochemical doping of imidazolium cation (EMI+) and borate anion (BF 4 ) to the CNTs in the electrodes. The cation doping (n-doping) is occurred at the cathode electrode and the anion doping (p-doping) is occurred at the anode electrode. This consideration is supported by a

drastic change of current at 0.0 V in CVs in Fig. 6. Fig. 6 shows the CVs of CNT(50) (a), CB(50) (b), CNT/CB(50/40) (c), and CNT/ PANI(50/50) (d) with three different potential limits (0.5 V, 1.0 V, and 2.0 V). Other CNT/CB-system and CNT/PANI-system have similar CVs (these are not depicted in Fig. 6). The anodic and the cathodic currents at 0.0 V in each CV dramatically increases when the potential limit becomes 2.0 V although these currents at 0.0 V with the potential limits of 0.5 V and 1.0 V are very similar in each CV. The doping induces pronounced increase of capacitance not only for CNT(50), CB(50) but also for CNT/PANI-system, CNT/CB-system. The doping characteristic of SWCNT has been found in an electrolyte solution by CV [39] and in an ionic liquid by in situ Vis/NIR and Raman spectroscopy [40]. The cation doping (n-doping) caused softening of graphene structure of SWCNT, on the other hand, stiffening of graphene structure of SWCNT was observed by the anion doping (p-doping) [40,41]. When we consider the actuation of our CNT actuators based on this reported result, expansion of cathode is caused by cation (EMI+) doping and contraction of anode arises from anion (BF 4 ) doping. This changing of graphene structure of SWCNT induces a bending motion toward the anode side of our trilayer CNT actuator. This way of bending direction is not conict with our previous results.

3.2.2.

Stored charge in the actuator composites

The loglog plots of strain vs. stored charge (anodic) are depicted in Fig. 7(a) and (b). The stored charge was calculated

3566

CARBON

4 9 ( 2 0 1 1 ) 3 5 6 0 3 5 7 0

(a)

0.10 0.08 0.06 0.04 0.02

(b)

0.8 0.6 0.4 0.2 0.0

I (mA)

0.00 - 3.0 - 2.0 - 1.0 - 0.02 0.0 - 0.04 - 0.06 - 0.08 - 0.10 1.0 2.0 3.0 - 3.0 - 2.0 - 1.0

- 0.2 - 0.4 - 0.6 - 0.8

0.0

1.0

2.0

3.0

V (V)

(c)

2.0 1.5 1.0 0.5 0.0

(d)

2.0 1.5 1.0 0.5 0.0

- 3.0

- 2.0

- 1.0

- 0.5 - 1.0 - 1.5 - 2.0

0.0

1.0

2.0

3.0

- 3.0

- 2.0

- 1.0

- 0.5 - 1.0 - 1.5 - 2.0

0.0

1.0

2.0

3.0

Fig. 6 Cyclic voltammograms of CB(50) (a), CNT(50) (b), CNT/CB(50/40) (c), CNT/PANI(50/50) (d) with different potential limits. 0.5 V (sky blue), 1.0 V (yellow), 2.0 V (blue). (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

(a)
Strain(% Strain )(%)

10 1 0.1 0.01 0.001 0.01

(b)
CNT/PANI(50/50) CNT/PANI(50/30) CNT/PANI(50/10) CNT(50)
Strain (%) Strain (%)

10 1 0.1 0.01 0.001

CNT/CB(50/40) CNT/CB(50/24) CNT/CB(50/8) CNT(50) CB(50)

0.0001

0.1

10

100

1000

0.01

0.1

10

100

1000

Anodic charge (mC) Charge+ (mC)

Ch arge+ (mC) Anodic charge (mC)

(c)
Strainefficiency efficiency (%/mC) Strain (%mC -1) 0.06 0.05 0.04 0.03 0.02 0.01 0 0.001 CNT/PANI(50/50) CNT/PANI(50/30) CNT/PANI(50/10) CNT(50)

(d)
0.06 0.06 0.05 0.05 0.04 0.04 0.03 0.03 0.02 0.02 0.01 0.01 0 0 0.001 0.001 Strainefficiency efficiency (%/mC) (%mC-1) Strain Strain efficiency
CNT/CB(50/40) CNT/CB(50/24) CNT/CB(50/8) CNT(50) CB(50)

0.1 10 Frequency (Hz)

1000

Frequency (Hz)

0.1 10 0.1 10 Frequency (Hz) Frequency (Hz)

1000 1000

Fig. 7 Strain vs. stored charge at the anode during displacement measurements. CNT/PANI-system (a), CNT/CB-system (b). Frequency dependence of strain efciency (%mC1); CNT/PANI-system (c), CNT/CB-system (d). In (a) and (c), CNT/PANI(50/50) (pink), CNT/PANI(50/30) (blue), CNT/PANI(50/10) (yellow), CNT(50) (sky blue). In (b) and (d), CNT/CB(50/40) (pink), CNT/CB(50/24) (blue), CNT/CB(50/8) (yellow), CNT(50) (sky blue), CB(50) (brown). (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

by accumulation of current during the half period of displacement measurements. In both CNT/PANI-system and CNT/CBsystem, the strains increase with increasing the stored charge. When 10 mg of PANI was added, the stored charge is improved to be more than twice as large as that of CNT(50) at 0.005 Hz causing almost double strain of CNT/PANI(50/10) compared to that of CNT(50). We introduce a parameter to evaluate the efciency of strain. This new parameter is dened as strain efciency (%mC1). The strain efciency is cal-

culated from the strain (%) and the stored charge (mC) at each frequency with Eq. (2) and frequency dependence of the stain efciency is depicted in Fig. 7(c) and (d). Strain efficiency %mC1 Strain %=Stored charge mC 2 If an actuator element has a higher strain efciency than the others, it means the stored electrical energy is converted to mechanical energy (mechanical stress) in the actuator ele-

CARBON

4 9 ( 20 1 1 ) 3 5 6 035 7 0

3567

ment more effective than the others. CNT/PANI(50/10) is charged more than CNT/PANI(50/30) and CNT/PANI(50/50) in the whole frequency range (2000.005 Hz) but the strain efciency of CNT/PANI(50/10) becomes worse rather than other two cases, especially in the lower frequency range (<0.05 Hz) (see in Fig. 7(c)). CNT/PANI(50/50) shows the best efciency among the CNT/PANI-system in the frequency range between 50 Hz and 0.05 Hz. As for the CNT/CB-system, a similar tendency is observed with respect to the strain efciency, that is, the smallest amount addition of CB (CNT/CB(50/8)) induces larger stored charge, nevertheless, the strain is smaller than other two cases. It is noteworthy that the efciency of CNT/ CB(50/8) is less than that of CNT(50) even though the CB is added into the CNT electrode. Among the CNT/CB-system, CNT/CB(50/40) possesses the highest strain efciency in the whole frequency range between 200 Hz and 0.005 Hz, except at 100, 0.1, and 0.05 Hz. It is also interesting to nd that CB(50) exhibits the best strain efciency in the frequency range between 100 Hz and 0.05 Hz in Fig. 7(d). But the stored charge of CB(50) is too small to actuate itself. For example, the stored charge of CB(50) at 0.05 Hz is about 1/44 compared to that of CNT/CB(50/40) at 0.05 Hz.

PANI(50/10). Interestingly, much amount of CB (40 mg addition of CB) softens the electrode lm. As a result, CNT/ PANI(50/50) exhibits the highest generated stress (10 MPa) that is around 29-fold compared with mammalian skeletal muscle [42] and is comparable to conductive polymers [43]. We succeeded to improve the generated stress 100 times larger than our previous result (0.1 MPa) [19]. In both cases of CNT/PANI-system and CNT/CB-system, the generated stresses increase with increasing the amount of conductive additives. The stress of CNT/PANI(50/50) is twice as large as that of CNT/PANI(50/10) and it is higher than that of CNT/CB(50/ 40). It can be said that the addition of PANI is more useful to generate higher actuation stress.

3.4.

Surface morphology of electrode lms

3.3.

Mechanical stiffness and generated stress

Not only the electrochemical property but also the mechanical property is an important factor to dene the actuation performance. The Youngs modulus of the electrode lm was evaluated by a stressstrain measurement instrument. Table 4 shows the obtained Youngs moduli of all the tested eight electrode lms in this study. The maximum generated stress (r) was calculated from the Youngs modulus (Y) and the maximum strain (emax) by using Hookes law; r = Y emax (see in Table 4). The Youngs modulus of CNT/PANI-system increases very much with the addition of PANI. Only 10 mg addition of PANI into CNT electrode stiffens the electrode lm 1.4 times compared to the electrode of CNT(50). The Youngs modulus of electrode with CNT/PANI(50/30) is almost the same as that of CNT/PANI(50/10) but the Youngs modulus of CNT/PANI(50/ 50) is almost 1.4 times as stiff as that of CNT/PANI(50/10). On the contrary, the Youngs modulus of electrode lm with CNT/ CB-system has a maximum at 24 mg addition of CB (540 MPa). This value is almost the same as the maximum value observed for CNT/PANI(50/50) (530 MPa). CNT/CB(50/8) has almost the same Youngs modulus as that observed for CNT/

Table 4 The maximum strain(emax) at 0.005 Hz, Youngs modulus (Y), and maximum generated stress (r) for all the complexes. Complex CNT(50) CNT/PANI(50/10) CNT/PANI(50/30) CNT/PANI(50/50) CB(50) CNT/CB(50/8) CNT/CB(50/24) CNT/CB(50/40) emax (%) 0.65 1.3 1.9 1.9 0.11 0.40 0.82 2.1 Y (MPa) 280 390 370 530 94 400 540 410 r (MPa) 1.8 5.0 7.0 10 0.10 1.6 4.4 8.6

Surface morphology of electrode lms was studied to know the reason of electrochemical and mechanical changes of the electrode lms by using SEM. The SEM images of eight different electrodes are shown in Fig. 8 (see high resolution SEM images in Supplementary data 2). The morphology is dramatically changed by changing carbon species (CNT or CB) and by changing the kind of conductive additives (PANI or CB) and the amounts of them. The heavily entangled bundles that form CNT ber-like mesh structure are seen for CNT(50), on the other hand, the agglomeration of CB particles is observed for CB(50). When the CB was mixed into the SWCNT, the CNT mesh structures with different concentration of agglomeration of CB are seen for CNT/CB-system (Fig. 8(c)(e)). The CB particles are agglomerated on the surface and inside of CNT mesh to form different surface morphology. A partial agglomeration of CB is seen only in small area of CNT mesh structure when the amount of CB is small (see Fig. 8(c)). The agglomerated CB and CB itself are spreading into the whole CNT mesh structure with increasing the amount of CB. In case of CB addition, it seems that the width of CNT bundles and the mesh structure are similar for all the composites between CNT and CB. We consider that changing of untangled degree of CNT-bundles and mixing state between dispersed CNTs and conductive nanoparticles affect the conductivity and the Youngs modulus of electrode lm. For example, only 8 mg addition of CB induces increasing of the conductivity (see Table 3) and the Youngs modulus (see Table 4) compared to those of CNT(50). It may imply that small additional amount of CB induce better dispersion of CNTs. As a result, the conductivity and the Youngs modulus of the electrode lm are increased. This positive tendency was seen until the additional amount of CB became 24 mg then these two parameters were decreased by further additional amount of CB such as 40 mg. This phenomenon is probably caused by agglomeration of poor conductive CB. Larger number of agglomerated CB was especially seen in the electrode of CNT/CB(50/40) compare to other two cases (CNT/CB(50/8) and CNT/CB(50/24)). We think that the agglomerated CB covers the CNT mesh structure leading to diminish the conductivity and the Youngs modulus of CNT/CB composites. In contrast, when PANI is used as a conductive additive, the morphology of electrodes with PANI is completely different from that of CNT/CB-system, CNT(50), and CB(50). As seen in Fig. 8(f)(h), the bundle width of CNT becomes narrower

3568

CARBON

4 9 ( 2 0 1 1 ) 3 5 6 0 3 5 7 0

Fig. 8 SEM images of electrode lms; CNT(50) (a), CB(50) (b), CNT/CB(50/8) (c), CNT/CB(50/24) (d), CNT/CB(50/40) (e), CNT/ PANI(50/10) (f), CNT/PANI(50/30) (g), and CNT/PANI(50/50) (h). The inset shows the magnication of each image.

than that of CNT(50) and more intricate mesh structures are seen with dispersion of PANI in the CNT mesh structure. Compared with the addition of CB, PANI used in this study does not form agglomeration but PANI is dispersed in the CNT mesh structure with each particle form. Larger amount of deposition of PANI is observed with increasing the additional amount of PANI. These results indicate that PANI rather than CB plays an important role to untangle CNT-bundles ner resulting in the ner and the intricate mesh structure,

furthermore, the dispersion of PANI particles onto and into the CNT mesh structure affords larger surface area for the electrode. These features cause advantages of the larger capacitance and the higher Youngs modulus to the electrode lm with increasing the additional amount of PANI. The larger maximum strain and the higher generated stress may be derived from the higher strain efciency and the higher Youngs modulus. From the viewpoint of actuation speed, the densely packing of conductive additives and intricate

Fig. 9 Illustration of asymmetrical three layered actuator and the expansion/contraction of the electrode of CNT/PANI(50/50).

CARBON

4 9 ( 20 1 1 ) 3 5 6 035 7 0

3569

CNT mesh structure interrupt the transfer of ion species resulting in smaller stain in the faster frequency range, especially by the addition of PANI (discussed in Section 3.1.).

3.5.

Asymmetrical three-layered actuator

An asymmetrical three-layered actuator element was fabricated to investigate actuation mechanism, especially about contraction and expansion of electrode lm. In the previous work [23], we have studied the mechanism of bending motion of our actuator by combination of a symmetrical analysis, the elasticity theory, and the displacement measurements for a ve-layered (electrode/electrolyte/electrode/electrolyte/electrode) actuator element which consists of three electrode lms and two electrolyte lms. Instead of making the ve-layered actuator, the CNT/PANI(50/50) electrode and the CB(50) electrode were chosen to fabricate the asymmetrical threelayered actuator as shown in Fig. 9 in this work. When we chose this asymmetry, the expansion and contraction of CB(50) electrode are negligibly small compared to those of CNT/PANI(50/50). For example, the strain of CB(50) is 0.045% which is about 1/17 smaller than that of CNT/PANI(50/50) (0.78%) at 0.1 Hz with 2.0 V. As a result, we can just know the expansion/contraction of CNT/PANI(50/50) electrode when the electrode is charged positively (anode) or negatively (cathode). The bending direction was checked at 0.1 Hz with 2.0 V for the asymmetrical three-layered actuator element. When the CNT/PANI(50/50) electrode was charged negatively, the actuator element bent toward the anode side (Fig. 9(a)). This means that the cathode is expanded by the cation (EMI+) charging. Oppositely, when the CNT/PANI(50/ 50) electrode was charged positively, the actuator also bent toward the anode side even for this asymmetrical actuator (Fig. 9(b)). It strongly corroborates that the anode is contracted by the anion (BF 4 ) charging. This way of expansion and contraction by ionic charging is enhanced by the CC bond expansion and contraction by ionic doping to CNTs.

ated stress (10 MPa) that is almost ve times larger than that of CNT(50) (the actuator without any additives). This is comparable to the electromechanical property of conductive polymer actuators. In contrast, the addition of CB affords an advantage for the faster actuation. The charge efciency is improved with increasing the amount of additives for both PANI and CB addition. This result indicates that the electrochemical/electromechanical energy conversion is occurred more effectively in the densely packing electrode lm with conductive nanoparticles. From the experiments with an asymmetrical three-layered actuator, it is revealed that the cathode expands and the anode contracts by cation charging and anion charging, respectively, for our CNT actuators. As for the actuation mechanism, it is suggested that the cation and the anion doping to CNTs in the electrodes besides the volume exclusion interaction between charged ions and electrodes induce the expansion of cathode and the contraction of anode resulting the bending motion of our actuators toward the anode side. The detail investigations on doping mechanism are now on progress.

Acknowledgements
This work was partially supported by New Energy and Industrial Technology Development Organization (NEDO). Takushi Sugino would like to acknowledge H. Iwazumi and Y. Shibata for their technical supports. Takushi Sugino thanks Dr. M. Fujiwara and JEOL (Osaka) for the SEM measurements.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.carbon.2011.04.056.

R E F E R E N C E S

4.

Conclusions

Conductive nanoparticles like polyaniline and carbon black were used as additives in order to improve the actuation properties of dry-type CNT-polymer actuator. The additional amount was changed to nd the optimal condition. It was found that the addition of conductive nanoparticles is more effective in order to improve the actuation performance rather than just increasing CNT amount. The SEM images of electrode lms were dramatically changed by the addition of PANI and CB. Much homogeneous ner and intricate CNT mesh structures were seen by the addition of PANI rather than the addition of CB. This may be caused by good miscibility of CNTs and PANI resulting in large capacitance and high conductivity. The electrochemical properties such as CV and conductivity were improved as increasing the amount of PANI. In contrast, the conductivity was reduced by the large amount addition of CB. This is probably due to the agglomeration of CB in the CNT mesh structure. The PANI is more effective additives rather than the CB, especially for the strong actuation. CNT/PANI(50/50) exhibits the largest gener-

[1] Claye AS, Fischer JE, Huffman CB, Rinzler AG, Smalley RE. Solid-state electrochemistry of the Li single wall carbon nanotube system. J Electrochem Soc 2000;147(8):284552. [2] Tarascon JM, Armand M. Issues and challenges facing rechargeable lithium batteries. Nature 2001;414:35967. [3] Pandolfo AG, Hollenkamp AF. Carbon properties and their role in supercapacitors. J Power Sources 2006;157:1127. [4] Futaba DN, Hata K, Yamada T, Hiraoka T, Hayamizu Y, Kakudate Y, et al. Shape-enginneerable and highly densely packed single-walled carbon nanotubes and their application as super-capacitor electrodes. Nat Mater 2006;5:98794. [5] Baughman RH, Cui C, Zakhidov AA, Iqbal Z, Barisci JN, Spinks GM, et al. Carbon nanotube actuators. Science 1999;284:13404. [6] Bar-Cohen Y, editor. Electroactive polymer (EAP) actuators as articial muscles, reality, potential, and challenges. Washington: SPIE-Press; 2001. [7] Ebbesen TW, Lezec HJ, Hiura H, Bennett JW, Ghaemi HF, Thio T. Electrical conductivity of individual carbon nanotubes. Nature 1996;382:546. [8] Shiraishi S, Kurihara H, Okabe K, Hulicova D, Oya A. Electric double layer capacitance of highly pure single-walled carbon nanotubes (HiPcoTM BuckytubesTM) in propylene carbonate electrolytes. Electrochem Commun 2002;4:5938.

3570

CARBON

4 9 ( 2 0 1 1 ) 3 5 6 0 3 5 7 0

[9] Treacy MMJ, Ebbesen TW, Gibson JM. Exceptionally high Youngs modulus observed for individual carbon nanotubes. Nature 1996;381:67880. [10] Pei Q, Ingana s O. Electrochemical applications of the bending beam method. 1. Mass transport and volume changes in polypyrrole during redox. J Phys Chem 1992;96:1050714. [11] Gandhi MR, Murray P, Spinks GM, Wallace GG. Mechanism of electromechanical actuation in polypyrrole. Synth Met 1995;73:24756. [12] Otero TF, Sansin ena JM. Soft and wet conducting polymers for articial muscles. Adv Mater 1998;10(6):4914. [13] Kaneto K, Kaneko M, Min Y, MacDiarmid AG. Articial muscle: electromechanical actuators using polyaniline lms. Synth Met 1995;71:22112. [14] Kaneto K, Fujisue H, Kunifusa M, Takashima W. Conducting polymer soft actuators based on polypyrrole lms-energy conversion efciency. Smart Mater Struct 2007;16:S2505. [15] Asaka K, Oguro K. Bending of polyelectrolyte membrane platinum composites by electric stimuli Part II. Response kinetics. J Electroanal Chem 2000;480:18698. [16] Nemat-Nasser S, Li JY. Electromechanical response of ionic polymermetal composites. J Appl Phys 2000;87(7):332131. [17] Shahinpoor M. Ionic polymer-conductor composites as biomimetic sensors, robotic actuators and articial muscles a review. Electrochim Acta 2003;48:234353. [18] Akle BJ, Bennett MD, Leo DJ. High-strain ionomeric-ionic liquid electroactive actuators. Sens Actuat A Phys 2006;126:17381. [19] Fukushima T, Asaka K, Kosaka A, Aida T. Fully plastic actuator through layer-by-layer casting with ionic-liquidbased bucky gel. Angew Chem Int Ed 2005;44:24103. [20] Takeuchi I, Asaka K, Kiyohara K, Sugino T, Terasawa N, Mukai K, et al. Electromechanical behavior of fully plastic actuators based on bucky gel containing various internal ionic liquids. Electrochim Acta 2009;54:17628. [21] Kiyohara K, Asaka K. Monte Carlo simulation of electrolytes in the constant voltage ensemble. J Chem Phys 2007;126:21470414. [22] Kiyohara K, Asaka K. Monte Carlo simulation of porous electrodes in the constant voltage ensemble. J Phys Chem C 2007;111:159039. [23] Kiyohara K, Sugino T, Takeuchi I, Mukai K, Asaka K. Expansion and contraction of polymer electrodes under applied voltage. J. Appl. Phys. 2009;105:063506-18 [Erratum: J Appl Phys 2009; 105: 119902-1]. [24] Chung CK, Fung PK, Hong YZ, Ju MS, Lin CCK, Wu TC. A novel fabrication of ionic polymer-metal composites (IPMC) actuator with silver nano-powders. Sens Actuat B 2006;117:36775. [25] Tanaka A, Takashima W, Kaneto K. Enhanced swelling behaviors of polypyrrole lm doped with sulfonated polyaniline. Chem Lett 2004;33:14701. [26] Bonomo C, Bottino M, Brunetto P, Pasquale GD, Fortuna L, Graziani S, et al. Tridimensional ionic polymer metal composites: optimization of the manufacturing techniques. Smart Mater Struct 2010;19:055022-18.

[27] Sugino T, Kiyohara K, Takeuchi I, Mukai K, Asaka K. Actuator properties of the complexes composed by carbon nanotube and ionic liquid: the effects of additives. Sens Actuat B 2009;141:17986. [28] Kaneko M, Fukui M, Takashima W, Kaneto K. Electrolyte and strain dependences of chemomechanical deformation of polyaniline lm. Synth Met 1997;84:7956. [29] Tahhan M, Truong VT, Spinks GM, Wallace GG. Carbon nanotube and polyaniline composite actuators. Smart Mater Struct 2003;12:62632. [30] Prasad KR, Munichandraiah N. Fabrication and evaluation of 450 F electrochemical redox supercapacitors using inexpensive and high-performance, polyaniline coated, stainless-steel electrodes. J Power Sources 2002;112:44351. [31] Gupta V, Miura N. Polyaniline/single-wall carbon nanotube (PANI/SWCNT) composites for high performance supercapacitors. Electrochim Acta 2006;52:17216. [32] Zhang H, Cao G, Wang W, Yuan K, Xu B, Zhang W, et al. Inuence of microstructure on the capacitive performance of polyaniline/carbon nanotube array composite electrodes. Electrochim Acta 2009;54:11539. [33] Zengin H, Zhou W, Jin J, Czerw R, Smith Jr DW, Echegoyen L. Carbon nanotube doped polyaniline. Adv Mater 2002;14(20):14803. [34] Cochet M, Maser WK, Benito AM, Callejas MA, Mart nez MT, Benoit JM, et al. Synthesis of a new polyaniline/nanotube composite: in situ polymerization and charge transfer through site-selective interaction. Chem Commun 2001:14501. [35] Pandolfo AG, Hollenkamp AF. Carbon properties and their role in supercapacitors. J Power Sources 2006;157:1127. [36] Fransson L, Eriksson T, Edstro m K, Gustafsson T, Thomas JO. Inuence of carbon black and binder on Li-ion batteries. J Power Sources 2001;101:19. [37] Avlyanov JK. Stable polyaniline and polypyrrole nanolayers on carbon surface. Synth Met 1999;102:12723. [38] Zhou D, Spinks GM, Wallace GG, Tiyapiboonchaiya C, MacFarlane DR, Forsyth M, et al. Solid state actuators based on popypyrrole and polymer-in-ionic liquid electrolytes. Electrochim Acta 2003;48:23559. [39] Kimizuka O, Tanaike O, Yamashita J, Hiraoka T, Futaba DN, Hata K, et al. Electrochemical doping of pure single-walled carbon nanotubes used as supercapacitor electrodes. Carbon 2008;46:19992001. [40] Kavan L, Dunsch L. Ionic liquid for in situ Vis/NIR and Raman spectroelectrochemistry: doping of carbon nanostructures. Chem Phys Chem 2003;4:94450. [41] Rao AM, Eklund PC, Bandow S, Thess A, Smalley RE. Evidence for charge transfer in doped carbon nanotube bundles from Raman scattering. Nature 1997;388:2579. [42] Hunter IW, Lafontaine S. A comparison of muscle with articial actuators. In: Technical digest IEEE solid-state sensor and actuator workshop. IEEE; 1992. p. 17885. [43] Hara S, Zama T, Takashima W, Kaneto K. Polypyrrolemetal coil composite actuators as articial muscle bres. Synth Met 2004;146:4755.