Electrochimica Acta 48 (2003) 641 /649 www.elsevier.

com/locate/electacta

Polyaniline-deposited porous carbon electrode for supercapacitor

Wei-Chih Chen, Ten-Chin Wen 1,*, Hsisheng Teng
Department of Chemical Engineering, National Cheng Kung University, Tainan 701, Taiwan, ROC Received 31 October 2002

Abstract Electrodes for supercapacitors were fabricated by depositing polyaniline (PANI) on high surface area carbons. The chemical composition of the PANI-deposited carbon electrode was determined by X-ray photoelectron spectroscopy (XPS). Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were used to investigate the electrochemical properties of electrodes. An equivalent circuit was proposed to successfully fit the EIS data, and the significant contribution of pseudocapacitance from PANI was thus identified. A comparative analysis on the electrochemical properties of bare-carbon electrodes was also conducted under similar conditions. The performance of the capacitors equipped with the resulting electrodes in 1 M H2SO4 was evaluated by constant current charge /discharge cycling within a potential range from 0 to 0.6 V. The PANI-deposited electrode exhibits high specific capacitance of 180 F/g, in comparison with a value of 92 F/g for the bare-carbon electrode. # 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Supercapacitor; Double-layer capacitance; Pseudocapacitance; Polyaniline; Porous carbons

1. Introduction Supercapacitors have attracted considerable attention as an intermediate power source between conventional capacitors and rechargeable batteries. In recent years, porous carbon is the most frequently used electrode material for supercapacitors [1 /4]. Due to the large surface area and porosity of carbon, high amounts of positive and negative ionic charges within the capacitor electrolyte could accumulate on the electrode surface, resulting in an electric double-layer capacitance with high energy density. In an electric double-layer capacitor (EDLC), double-layer could be quickly formed at the electrode/electrolyte interface, providing a high power rate compared to batteries where mass transfer over long distances is necessary [4]. Sometimes, a supercapacitor of large specific capacitance forms from an electrode material not only possessing large specific surface area but also coupling

* Corresponding author. Tel.: /886-6-238-5487; fax: /886-6-2344496 E-mail address: tcwen@mail.ncku.edu.tw (T.-C. Wen). 1 ISE member.

with active species, like conducting polymers [5 /7] and metal oxides [8 /10], that can be reversibly oxidized or reduced over a potential range. Such a supercapacitor exhibits both double-layer capacitance and faradaic pseudocapacitance simultaneously in energy storage. Conducting polymers, such as polyaniline (PANI), polypyrrole, and polythiophene, exhibit oxidation /reduction properties and also have high specific surface area. Hence, conducting polymers with these properties can be suitably used as electrode materials for secondary batteries and capacitors. As a matter of factor, PANI and its derivatives have been considered promising materials in energy storage devices applications, and are used in electrochemical capacitors as electroactive materials with high faradaic pseudocapacitance owing to the existence of several different oxidation-state structures [11 /13]. The performance of such conducting polymer-based capacitors is mainly controlled by the electrochemical characteristics of the redox couples on/within the electroactive materials [14]. Up to nowadays, efforts to characterize the electrochemical properties and the stability of PANI are still required and underway in order to verify the applicability of this material in electrochemical supercapacitors.

0013-4686/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 3 - 4 6 8 6 ( 0 2 ) 0 0 7 3 4 - X

642

W.-C. Chen et al. / Electrochimica Acta 48 (2003) 641 /649

The objective of the present research work is to enhance the capacitance of an EDLC through introduction of PANI, which provides pseudocapacitance, to the carbon electrodes. However, in order to avoid the blockage of the carbon micropores, the amount of PANI loading should be adequately controlled. The chemical composition of the modified electrode will be characterized using X-ray photoelectron spectroscopy (XPS). The electrochemical characteristics of electrodes and the performances of the resulting supercapacitors are also examined by using cyclic voltammetry, ac impedance, and galvanostatic charge /discharge, in the attempt to clarify the contribution of PANI to the overall capacitance.

2. Experimental 2.1. Preparation of carbon electrodes Porous carbon powders serving as the main electrode material (specific surface area 1778 m2/g, pore volume 1.13 cm3/g and average pore diameter 2.54 nm) were added to a solution of polyvinylidenefluoride (PVdF) in N -methyl pyrrolidinone (NMP), and the mixture was stirred at ambient temperature to form a carbon slurry. Carbon electrodes were fabricated by coating the slurry on stainless steel (type 304) foils, followed by evaporating the solvent, NMP, with a blow dryer. The carbon layer consisting of 20 wt.% PVdF as binder was controlled to have a thickness of 100 mm. This form of bare-carbon electrode is denoted as BC electrode. The constant area and specific mass of BC electrodes used in this study are 1 cm2 (1 cm /1 cm) and 2.9 mg/cm2, respectively. 2.2. Loading PANI on carbon electrodes Reagent grade aniline (Merck) employed in the present work was doubly distilled and the resulting colorless liquid was kept under nitrogen in darkness at 58C. All the other reagents used in this work were also reagent grade and supplied by Merck and were used without pretreatment. Electrochemical polymerization of PANI was made by using the Autolab PGSTAT 30 (Eco. Chemie B.V., The Netherlands) with a general-purpose electrochemical system software. A three-electrode cell assembly was utilized with an Ag/AgCl reference electrode (Argenthal, 3 M KCl, 207 mV vs. SHE at 258C). Platinum wire with a diameter of 0.5 mm and an exposed area of 0.65 cm2 was used as the counter electrode. A bare-carbon (BC) electrode of area 1 cm2 was used as the working electrode. A Luggin capillary, whose tip was set at a distance of about 1 mm from the surface of the working electrode was used to minimize errors due to iR drop in

the electrolytes. The polymerization of PANI was performed under nitrogen atmosphere. The potentials reported in this work are values relative to the Ag/AgCl reference electrode. PANI were electrochemically loaded on the BC substrates by using cyclic voltammetry (CV) in a sweeping potential range between /0.2 and 0.8 V. A very slow sweep rate, 0.3 mV/s, was conducted to let the aniline monomers diffuse into the inner surface of carbon and then deposit onto it. In order to avoid blocking the entrances of carbon micropores, the amount of PANI loading should be minimized. Hence, the cyclic voltammetric polymerization of PANI was conducted for 5 cycles only. Additionally, the concentration of aniline (ANI) in 1 M H2SO4 solution used for polymerization was as low as 10 mM. Such PANIdeposited carbon electrode is denoted as PC electrode. Considering the minimum PANI loading, the specific mass of PC electrodes was assumed to be the same with BC electrodes, 2.9 mg/cm2. 2.3. X-ray photoelectron spectroscopy The chemical compositions were determined for both types of electrodes (the BC and PC electrodes) by XPS using an ESCA 210 and MICROLAB 310D (VG Scientific Ltd., UK) spectrometer. Before scanning, samples were sputtered by using Ar gas to removed absorbed impurities on the surface. For recording the spectra Mg Ka (hn /1253.6 eV) irradiation was employed as the photon source with primary tension of 12 kV and an emission current of 20 mA. The pressure of the analysis chamber during the scans was about 10 10 mbar. Low-resolution survey scans were done at 50 eV pass energy by using a step of 1 eV to obtain the atomic percentage data. After the survey spectra were obtained, higher resolution survey scans were performed at pass energy of 20 eV, with a step of 200 meV, for at least ten scans. Deconvolution of the XPS spectra was done by using Grams/32 software (Galactic). 2.4. Electrochemical measurements Electrochemical measurements (CV and ac impedance) were carried out at ambient temperature using 1 M H2SO4 aqueous solution as electrolyte to examine the electrochemical performance of BC and PC electrodes. For the cyclic voltammetric measurements, the potential scan rate was ranged from 0.3 to 1.0 mV/s within a potential range of /0.2 to 0.6 V. The cyclic voltammograms of the electrodes were recorded after stabilization. The Impedance Spectrum Analyzer, IM6 (ZAHNER, Germany), with Thales software was employed to measure and analyze the ac impedance spectra of BC and PC electrodes obtained at different applied dc

W.-C. Chen et al. / Electrochimica Acta 48 (2003) 641 /649

643

potentials. The potential amplitude of ac was kept as 10 mV and a wide frequency range of 50 mHz to 100 kHz was used. Both the CV and electrochemical impedance spectroscopy (EIS) experiments were carried out in a threecompartment cell. BC or PC electrode was used as the working electrode. Platinum wire was used as the counter electrode. Ag/AgCl coupled with a Luggin capillary was used as the reference electrode. 2.5. Capacitance measurement Sandwich-type capacitor cells were constructed with a pair of the electrodes, BC or PC, and a piece of filter paper as separator. The capacitance of the electrodes was galvanostatically measured with BT-2043 system (Arbin electrochemical instrument, USA), using 1 mA charge /discharge between 0 and 0.6 V for 1000 cycles.

3. Results and discussion 3.1. Electrochemical polymerization of PANI on carbon electrodes PC electrodes were fabricated by loading PANI on bare-carbon electrodes using cyclic voltammetry. Fig. 1 shows the voltammograms recorded continuously for five cycles during the polymerization of ANI on carbon substrates. A low ANI concentration in 1 M H2SO4 was kept as 10 mM to minimize the PANI loading on carbon. One redox pair (0.4 /0.6 V) and an anodic peak

( /0.8 V) could be observed in Fig. 1. In the first CV cycle, the anodic peak appearing at 0.78 V is attributed to the oxidation of ANI monomers and the PANI growth process begins to take place [15]. In the following growth cycles, the peak occurring at ca. 0.5 V corresponds to formation of radical cations (polaronic emeraldine), which shows a slight potential shift compared to earlier studies [16,17]. The probable reason is due to that the porous carbon was used as a working electrode in the present study, instead of a flat platinum foil used in the earlier studies. Diffusion of ANI monomers or radical cations into micropores of porous carbons is limited by the pore resistance, leading to a delayed anodic peak potential. Meanwhile, the peak occurring at ca. 0.8 V represents not only monomers oxidation but also the formation of diradical dications arising from the head-to-tail addition of monomer units. Radical cations (polaronic emeraldine) which are generated at the first anodic peak of voltammograms persist in PANI until the potential reaches the second anodic peak, where diradical dications (bipolaronic pernigraniline) are formed and exist in resonance with the quinone diimine form. The dication diradical acts as an energetic electrophile capable of interacting with a neutral ANI molecule to result in further growth [18]. 3.2. XPS analysis After the fabrication of PC electrodes by loading PANI on the BC substrates through cyclic voltammetry, XPS scans were taken for both types of electrodes to examine the surface compositions. For the purpose of fair comparison, prior to XPS analysis the BC electrode was placed in a background electrolyte, 1 M H2SO4, and subjected to CV cycling in the same manner as for PC. Survey level XPS scan clearly reflects the presence of C, N, O, and S elements in these two electrodes. The relative atomic concentrations are derived from the corresponding photoelectron peak areas and are presented in Table 1. Sensitivity factor corrections of 1.00, 1.77, 2.85 and 2.14 are applied for C 1s, N 1s, O 1s and S 2p signals to obtain the atomic percentages. From the data in Table 1, it is noticed that the atomic percentages of N 1s of BC and PC electrodes are 0 and 2.3%, respectively. The N element in the PC electrode must come from the PANI deposited on the electrode. Additionally, the PC electrode contains higher amounts of O and S elements, whose peak positions are located at 532.4 and 169.0 eV, respectively, than the BC electrode. This is probably due to the fact that sulfate molecules [19] are trapped in the polymer network of the deposited PANI. The N 1s core level for the PC electrode can be deconvoluted into four components that are identical to the N 1s deconvolution of PANI film presented in earlier literature [20]. These four components were

Fig. 1. Cyclic voltammograms of growth of PANI on carbon electrode in 1 M H2SO4 solution containing 10 mM aniline for ve cycles; scan rate /0.3 mV/s.

644

W.-C. Chen et al. / Electrochimica Acta 48 (2003) 641 /649

Table 1 Narrow scan XPS data for BC and PC electrodes. Electrodes Atom percentage (%) C1s BC PC
a

Na 1s O 1s 17.8 30.8 S 2p 3.0 6.5 C /N / 33.0 C /N / 43.4 C /N  / 14.0 C /N  / 9.6

N 1s 0 2.3

79.2 60.3 Area ratios of subpeaks.

assigned for C /N, C /N, C /N  and C /N  based on the binding energy of core electrons [20]. The N 1s spectrum is given in Fig. 2 and the area ratio of the four species are presented in Table 1. Owing to the positive charge on the nitrogen atom, PANI can attract sulfate anions as dopant, which reasonably explains why the PC electrode contains higher percentages of S and O elements than the BC electrode (Table 1). 3.3. Electrochemical characteristics of the resulting electrodes Cyclic voltammetric measurements were performed in a potential range from /0.2 to 0.6 V to analyze the electrochemical behavior of the resulting electrodes in 1 M H2SO4 solution. The voltammograms of the BC electrode with sweep rates of 0.3, 0.6 and 1.0 mV/s are shown in Fig. 3. This figure reveals that the electrode is stable in acid solution within the applied potential range. It can be seen that a wide bending of voltammograms appears just after the reversal of the potential sweep before reaching to the plateau. The voltammograms show a slight deviation from the rectangular

Fig. 3. Cyclic voltammograms of the BC electrode in 1 M H2SO4 at different sweep rates: (a) 0.3; (b) 0.6 and (c) 1.0 mV/s.

Fig. 2. XPS narrow scan of PANI-deposited carbon electrode for N 1s. The raw data (jagged curve) have been deconvoluted into Gaussian peaks.

form, which is characteristic of an ideal double-layer capacitor. Also, increasing the scan rate enhances the delay of the current to reach a constant value. These are arisen from the distributed capacitance effects in porous electrodes and the effects are enhanced upon increasing the potential sweep rate [21]. Such effects were caused by the difference between the ohmic resistance of the electrolyte at the mouth of micropores and that at the bottom of micropores [3,21]. At the bottom, the resistance is significant for ion migration and controls the charge and discharge rates at the beginning of current inversion. Consequently, increasing the scan rate would lead to an increase in the potential difference, which would enhance the lag of charge/discharge at the bottom of micropores and thus the delay of the current to reach a constant value in voltammograms (Fig. 3). The voltammograms of the PC electrode are shown in Fig. 4. In comparison with the BC electrode, PC not only presents a higher background current in the voltammograms but also displays faradaic currents, which are likely to arise from the contribution of the deposited PANI. PANI has been considered as promising materials in the application of electrochemical capacitors owing to the existence of structures of different oxidation states [7]. It can be seen from the voltammograms of Fig. 4 that, during the anodic

W.-C. Chen et al. / Electrochimica Acta 48 (2003) 641 /649

645

Fig. 4. Cyclic voltammograms of the PC electrode in 1 M H2SO4 at different sweep rates: (a) 0.3; (b) 0.6 and (c) 1.0 mV/s.

potential sweep, the current significantly increases with the potential for the sweep beyond 0.2 V and reaches a maximum at 0.6 V. This is attributable to the addition of faradaic current from PANI, which is in its conducting emeraldine state (half-oxidized state) at this potential range, 0.2 /0.6 V. On the other hand, while the potential below 0.2 V, PANI exists in the insulated leucoemeraldine state (fully-reduced state) that could not contribute faradaic current. In the reversal cathodic potential scan, the current was enhanced by the reduction of PANI. Due to the larger current in voltammograms of the PC electrode than the BC electrode, a higher capacitance for capacitors made of the PC electrodes is expected. However, the redox transitions within PANI would not be as reversible and fast as the double layer formation, and this leads to a more serious delay of current to reach the plateau (Fig. 4). Electrochemical impedance spectroscopy (EIS) was further employed to monitor the electrochemical behavior of the electrodes. Electrodes were subjected to various applied potentials and the impedance characteristics were recorded. Typical Nyquist diagrams for the BC and PC electrodes in 1 M H2SO4 solution at different potentials are given in Figs. 5 and 6, respectively. All the impedance curves of the BC electrode in Fig. 5 show a single semi-circle in the high-frequency region and a straight line in the low-frequency region. However, a vertical line was not observed in all impedance curves, which might be probably resulted from the high content (20 wt.%) of PVdF binder in the electrode materials. The inert PVdF would decrease the pore size of carbons, and thus retards the ion mobility in the micropores. Hence, the performance of an ideal capacitive behavior (a 908 line at low frequency) was possibly affected by this cause. It might also dominate the slow responses in the cyclic voltammograms shown

Fig. 5. Nyquist diagrams for the BC electrode measured at different applied potentials.

Fig. 6. Nyquist diagrams for the PC electrode measured at different applied potentials.

in Figs. 3 and 4. An earlier report has presented the rectangular shape in voltammograms with higher sweep rates and the ideal low-frequency capacitive responses shown in Nyquist plots measured in the binder-free carbon fabrics electrodes [22]. A previous study has indicated that the high-frequency arc is attributed to the impedance resulted from the electrical connection between carbon particles and as well as that between carbons and the backing current collector [22]. Accordingly, the equivalent circuit for BC

646

W.-C. Chen et al. / Electrochimica Acta 48 (2003) 641 /649

electrode should involve the following elements: the bulk solution resistance, Rs, the capacitance at the contact interface of carbon particles, CC, which was expected in parallel with the contact resistance (RC), and a Warburg diffusion element ( ) attributable to the diffusion of ions. In addition, the capacitive nature of the activated carbon electrode in the low-frequency region [22] should be reasonably expressed as a lowfrequency capacitance (CL) in parallel with a charge transfer resistance (Rct). CL here represents the doublelayer capacitance (Cdl) attributable to the large surface area of carbons. Hence, the circuit combining these elements is proposed in Fig. 7 to simulate the impedance spectra of the BC electrode. The same equivalent circuit (Fig. 7) was used for simulating the impedance spectra of the PC electrode. In this case, CL in the circuit model represents the combination of the double-layer capacitance from carbon substrate and the pseudocapacitance (CF) from PANI, which was related to the faradaic process of its redox transition. Considering the simultaneous appearance of Cdl of carbons and CF of PANI in the lowfrequency region of impedance spectra [11,22,23], CL was used alone to express Cdl/CF, though the mechanisms between these two (Cdl and CF) are different. Due to the porous surface of electrodes, a constant phase element (CPE) is employed to express the contact interface capacitance (CC) and low-frequency capacitance (CL) in the equivalent circuit (Fig. 7) [24]. CPE is defined by
a ZCPE  TCPE (jv) CPE

(1)

where T and a are frequency-independent constants and v the angular frequency. The exponent a is a correction factor related to the roughness of electrode surfaces, with values ranging between 0 and 1. Some advanced EIS papers [25 /27] have proposed a model of the transport processes in PANI and the charge transfer kinetics of the ions and solvent at the polymer/electrolyte interface. Anions, cations and solvent simultaneously crossed the polymer/electrolyte interface that anions were inserted, and solvent as well as cations were expelled to compensate the positive charges created during the redox reaction of PANI [25 /

Fig. 7. Equivalent circuit for simulating the impedance spectra of the BC and PC electrodes. CL represents Cdl and Cdl/CF for the BC and PC electrodes, respectively.

27]. In the present study, the amount of PANI loading was in minimum. And thus only the pseudocapacitance was considered to incorporate into the proposed equivalent circuit, whose construction tended to be similar to the one established for carbon electrodes [22]. In Figs. 5 and 6, the corresponding solid lines represent the best-fit curves based on the equivalent circuit shown in Fig. 7. The good fittings imply that the equivalent circuit model reasonably reflects the electrochemical process occurring on/within the BC and PC electrodes. The best fitting values of all the parameters for the respective circuit elements are listed in Table 2. The mean error of modulus is less than 0.8%, indicating that these fitting values are highly trustful. First, the parameter values corresponding to the circuit elements of the BC electrode in Table 2 are inspected. The bulk solution resistance, Rs, which is strongly dependent on the distance between Luggin capillary and the electrode, is very low and irrelevant to the applied potentials. The contact resistance, RC, decreases with the increasing applied potentials. This result reveals that the electron transfer is favored at high potentials, attributable to the resulting high driving force for electron movement among the all components of the BC electrode. The value of contact interface capacitance CC (CPE1) is negligibly small compared to the double-layer capacitance. Note that, the specific double-layer capacitance, CPE2, achieves to about 85 F/ g and does not show obvious variation within the whole applied potential range. This phenomenon may suggest that the double-layer structure would not be influenced by the applied potentials. The correction factor of CPE2, a2, shows a slight deviation from 1, informing that this double-layer element does not exhibit as an idealcapacitor (i.e. a flat plate). This may be due to the limit of ion migration in carbon micropores, thus leading a non-ideal capacitive behavior of the electrode. The element values of interest in the BC and PC electrodes are compared. RC is higher in the PC electrode than that in the BC electrode. This should result from the PANI coverage on some portions of carbon surface, and leads to an increase in the electrical resistance of the PC electrode. A corresponding increase in CPE1, due to the coverage of PANI on carbon surface, is observed as well. Interestingly, the values of CPE2 are large compared to those of BC and are a strong function of the applied potential. This is attributable to the addition of the potential-dependent pseudocapacitance provided by the deposited PANI. CPE2 is relatively small at potentials lower than 0 V, due to the insulated state (leucoemeraldine) of PANI. As the applied potential is raised above 0 V, CPE2 shows an obvious increase with the potential. The increase can be explained by the fact that PANI becomes electronically conductive (i.e. in the form of emeraldine) at the potentials higher than 0 V and the faradaic redox

W.-C. Chen et al. / Electrochimica Acta 48 (2003) 641 /649

647

Table 2 The best tting values of the equivalent circuit elements in Fig. 7 from the simulation of the impedance data for BC and PC electrodes at different applied potentials Sample Element E (V) /0.1 BC electrode Rs (V) RC (V) CPE1 (mF/g) a1 CPE2 (F/g) a2 Rct (V) Rs (V) RC (V) CPE1 (mF/g) a1 CPE2 (F/g) a2 Rct (V) 0.281 100.1 1.596 0.861 83.83 0.927 580.9 0.468 184.6 3.907 0.863 134.6 0.924 1465 0.0 0.280 95.09 1.534 0.859 84.69 0.926 580.1 0.459 182.9 3.760 0.874 137.9 0.929 1378 0.1 0.278 86.94 1.518 0.862 85.55 0.906 567.8 0.458 178.5 4.530 0.871 151.2 0.912 1310 0.2 0.277 81.55 1.548 0.865 86.72 0.913 551.1 0.463 178.3 4.107 0.862 156.5 0.902 1252 0.3 0.275 77.84 1.518 0.858 83.97 0.924 549.8 0.451 165.6 3.640 0.888 161.2 0.916 1182 0.4 0.271 76.52 1.471 0.863 84.72 0.913 549.1 0.431 163.9 3.420 0.889 163.3 0.922 1147 0.5 0.278 76.26 1.403 0.871 85.55 0.887 540.3 0.468 163.3 3.277 0.883 177.1 0.896 1104 0.6 0.277 75.87 1.359 0.859 84.14 0.906 538.6 0.485 162.4 3.377 0.865 188.4 0.914 1023

PC electrode

Mean error of modulus is less than 0.8%.

transitions occurring at these potentials are electrochemically reversible. The trend of CPE2 (or CL) variation with the applied potential is in agreement with that suggested from the cyclic voltammetry shown in Fig. 4. 3.4. Charge /discharge cycling performance of capacitors In order to examine the performance of the resulting PC as electrodes in electrochemical supercapacitors, sandwich-type capacitor cells are prepared and subjected to constant current charge /discharge cycling in 1 M H2SO4 media. Cells made with the BC electrodes are also tested as a close comparison with the PC cells. Meanwhile, for the sake of evaluating the stability of the electrodes, the charge /discharge cycling test was conducted for 1000 cycles. Typical galvanostatic charge / discharge curves (cell potential against time) at a current of 1 mA between 0 and 0.6 V are presented in Fig. 8. It is obvious to see that the PC cell has higher capacitance than the BC, which has been considered to result from the addition of faradaic pseudocapacitance of PANI. The specific discharge capacitance of the electrodes (C ) in the cells was calculated according to C  (2It)=(WDE ) (2) where I is the discharge current, t the discharge time, W the material mass on an electrode, and DE the voltage difference in discharge, eliminating the section of IR drop. The factor of 2 comes from the fact that the total capacitance measured from the test cells is the addition of two equivalent single-electrode capacitors in series. Figs. 9 and 10 show the variation of specific capacitance with cycle number for BC and PC capacitors, respectively. In Fig. 9, the result gives the information that the capacitor equipped with BC electrodes has the

Fig. 8. Charge /discharge curves at 1 mA current of capacitors equipped with (a) BC; and (b) PC electrodes.

capacitance about 92 F/g at beginning 50 cycles and remains at 90 F/g at the 1000th cycle. This illustrates the BC capacitor possessing good stability over 1000 charge /discharge tests. The specific capacitance of the PC capacitor was as high as 180 F/g at the early stage of cycling (Fig. 10). Accordingly, incorporating PANI exhibited a significant effect in capacitance enhancement. However, the capacitance decayed from 180 to 163 F/g after 1000 cycles. It is probably due to that the degradation of PANI was taking place in aqueous media. Swelling and shrinking of electroactive polymers

648

W.-C. Chen et al. / Electrochimica Acta 48 (2003) 641 /649

4. Conclusions A composite electrode fabricated by using cyclic voltammetry to implant PANI on high surface area carbons was studied for the application on supercapacitors. XPS measurements identified the chemical composition of the electrodes. Electrochemical analyses, using CV and EIS, show the significant contribution of PANI to the overall capacitance due to PANI exhibiting high faradaic pseudocapacitance. The simulation of the EIS spectra with a proposed equivalent circuit revealed that the pseudocapacitance from PANI was a strong function of applied potentials. The performance and stability of capacitors equipped with bare-carbon and PANI-deposited carbon electrodes were measured by constant current charge /discharge cycling over 1000 cycles in 1 M H2SO4. The composite electrode exhibits a large specific capacitance of 180 F/g, higher than the bare-carbon electrode exhibiting a value of 92 F/g. The high specific capacitance of PANIdeposited carbon electrodes was concluded as the combination of double-layer capacitance and pseudocapacitance.

Fig. 9. Variation of capacitance and coulombic efciency with cycle number for the BC capacitor charged and discharged at 1 mA in 1 M H2SO4.

Acknowledgements The authors are grateful to the National Science Council in Taiwan for its nancial support through NSC 91-2214-E-006-028.

References
[1] A. Lewandowski, M. Zajder, E. Frackowiak, F. Beguin, Electrochim. Acta 46 (2001) 2777. [2] M. Endo, Y.J. Kim, T. Takeda, T. Maeda, T. Hayashi, K. Koshiba, H. Hara, M.S. Dresselhaus, J. Electrochem. Soc. 148 (2001) A1135. [3] T.C. Weng, H. Teng, J. Electrochem. Soc. 148 (2001) A368. [4] H. Shi, Electrochim. Acta 41 (1996) 1633. [5] M.D. Ingram, A.J. Pappin, F. Delalande, D. Poupard, G. Terzulli, Electrochim. Acta 43 (1998) 1601. [6] A. Di Fabio, A. Giorgi, M. Mastragostino, F. Soavi, J. Electrochem. Soc. 148 (2001) A845. [7] F. Fusalba, P. Gouerec, D. Villers, D. Belanger, J. Electrochem. Soc. 148 (2001) A1. [8] J. Zhang, D. Jiang, B. Chen, J. Zhu, L. Jiang, H. Fang, J. Electrochem. Soc. 148 (2001) A1362. [9] C.C. Hu, C.C. Wang, Electrochem. Comm. 4 (2002) 554. [10] H. Kim, B.N. Popov, J. Power Sources 104 (2002) 52. [11] C.C. Hu, C.H. Chu, J. Electroanal. Chem. 503 (2001) 105. [12] K. Naoi, K.I. Kawase, M. Mori, M. Komiyama, J. Electrochem. Soc. 144 (1997) L173. [13] T. Sotomura, H. Uemachi, K. Takeyama, K. Naoi, N. Oyama, Electrochim. Acta 37 (1992) 1851. [14] B.E. Conway, Electrochemical Supercapacitors, Kluwer /Plenum, New York, 1999.

Fig. 10. Variation of capacitance and coulombic efciency with cycle number for the PC capacitor charged and discharged at 1 mA in 1 M H2SO4.

may also lead to the degradation during the long-term cycling [28]. The specific capacitance values of both BC and PC capacitors obtained from charge /discharge cycling measurements are close to those evaluated from the simulation of the impedance spectra (Table 2). Besides specific capacitance, the coulombic efficiencies [29] of the respective capacitors are also shown in Figs. 9 and 10. The coulombic efficiencies for the BC and PC electrodes based supercapacitors are as high as ca. 99% and 98%, respectively, over 1000 cycles.

W.-C. Chen et al. / Electrochimica Acta 48 (2003) 641 /649 [15] D.E. Stilwell, S.M. Park, J. Electrochem. Soc. 135 (1988) 2491. [16] Y.B. Shim, M.S. Won, S.M. Park, J. Electrochem. Soc. 137 (1990) 538. [17] D.E. Stilwell, S.M. Park, J. Electrochem. Soc. 136 (1988) 427. [18] W.C. Chen, T.C. Wen, A. Gopalan, J. Electrochem. Soc. 148 (2001) E427. [19] M.G. Han, S.S. Im, Polymer 41 (2000) 3253. [20] H.S.O. Chan, S.C. Ng, W.S. Sim, K.L. Tan, B.T.G. Tan, Macromolecules 25 (1992) 6029. [21] I. Tanahashi, A. Yoshida, A. Nishino, J. Electrochem. Soc. 137 (1990) 3052. [22] Y.R. Nian, H. Teng, J. Electrochem. Soc. 149 (2002) A1008.

649

[23] W.C. Chen, T.C. Wen, C.C. Hu, A. Gopalan, Electrochim. Acta 47 (2002) 1305. [24] M.A. Vorotyntsev, J.P. Badiali, G. Inzelt, J. Electroanal. Chem. 472 (1999) 7. [25] C. Gabrielli, M. Keddam, N. Nadi, H. Perrot, J. Electroanal. Chem. 485 (2000) 101. [26] C. Gabrielli, M. Keddam, N. Nadi, H. Perrot, Electrochim. Acta 44 (1999) 2095. [27] C. Gabrielli, M. Keddam, H. Perrot, M.C. Pham, R. Torresi, Electrochim. Acta 44 (1999) 4217. [28] R. Kotz, M. Carlen, Electrochim. Acta 45 (2000) 2483. [29] T. Osaka, X. Liu, M. Nojima, T. Momma, J. Electrochem. Soc. 146 (1999) 1724.