Journal of The Electrochemical Society, 150 共5兲 A645-A651 共2003兲

A645

0013-4651/2003/150共5兲/A645/7/$7.00 © The Electrochemical Society, Inc.

Activated CarbonÕConducting Polymer Hybrid Supercapacitors A. Laforgue,a P. Simon,a,*,f J. F. Fauvarque,a,* M. Mastragostino,b,*,z F. Soavi,b J. F. Sarrau,c P. Lailler,c M. Conte,d E. Rossi,d and S. Saguattie a

CNAM, Laboratoire d’Electrochimie Industrielle, 75003 Paris, France UCI Scienze Chimiche, Radiochimiche e Metallurgiche, University of Bologna, 40127 Bologna, Italy CEAC-EXIDE, 92636 Gennevilliers Cedex, France d ENEA, Italian National Agency for New Technology, Energy and the Environment, C. R. Casaccia, 00060 Roma, Italy e Arcotronics Italia S.p.A., Divisione Meccanica, 40037 Sasso Marconi, Italy b c

This paper presents the work carried out within a European Union 共EU兲 project which led to the development of 3 V and 1.5 kF preseries supercapacitor modules and 2 kW stacks based on hybrid cells with poly共3-methylthiophene兲 as positive electrode and activated carbon as the negative electrode with propylene carbonate-tetraethylammonium tetrafluoroborate electrolyte. These prototypes, which display a concept of hybrid cell operating with a high-surface-area activated carbon and a conventional electronically conducting polymer, both commercially available, and with a nontoxic and nonvolatile electrolyte, provide a successful response to the market demand for high power and energy supercapacitors operating with an environmentally friendly electrolyte. © 2003 The Electrochemical Society. 关DOI: 10.1149/1.1566411兴 All rights reserved. Manuscript submitted June 18, 2002; revised manuscript received November 26, 2002. Available electronically March 28, 2003.

Supercapacitors are energy conversion systems which fill the gap in the Ragone plot between batteries and conventional electrostatic capacitors. They provide much higher power but lower energy than batteries and fuel cells, and they are expected to serve for a significantly higher number of cycles.1 The interest in supercapacitors has increased enormously in recent years mainly for applications in nonpolluting transportation, which has become an increasingly urgent issue. Supercapacitors coupled with batteries and fuel cells are considered promising mid-term and long-term solutions for low- and zero-emission transport vehicles by providing the power peaks for start-stop integrated systems, which have been conceived and developed to limit pollution by conventional cars, and by providing power peaks for acceleration and recovering the breaking energy in electric and hybrid vehicles. Given such applications, the focus of interest is on supercapacitors with relatively high specific energy and a cycle life longer than 500,000 cycles. Three types of supercapacitors are described in the literature, carbon, metal oxide, and conducting polymer.2,3 The carbon type is the most advanced version of supercapacitors and are already on the market as high-performance devices. The energy storage in these devices is electrostatic in origin, the electrodes are high-surface-area porous carbon materials and the capacitance arises from charge separation at the interface between the solid electrode surface and the liquid electrolyte. Both aqueous and organic electrolytes are used and the cycle life of these double-layer carbon supercapacitors 共DLCSs兲 is very high (⬎100,000 cycles). The highest values of maximum energy (E MAX ⫽ 1/2 CV 2 , where C is the device capacitance and V the cell voltage at the beginning of discharge兲 and maximum power ( P MAX ⫽ V 2 /4R, where R is the equivalent series resistance兲 for kg of packaged DLCSs are 3-4 Wh kg⫺1 and 3-4 kW kg⫺1 , respectively. In the metal oxide and in the conducting polymer supercapacitors very fast and reversible faradaic charge-transfer processes take place at the electrode materials, as in batteries, which give rise to what is called pseudocapacitance to indicate their capacitance behavior. Ruthenium oxide supercapacitors have been especially studied and developed. However, despite the high pseudocapacitance values which arise from fast and reversible protonation processes on the oxide surface or in the bulk, the high cost of ruthenium does not make these supercapacitors competitive with the DLCSs.

* Electrochemical Society Active Member. f z

Present address: CIRIMAT LCMIE, CNRS UMR 5085, 31062 Toulouse Cedex, France. E-mail: [email protected]

The pseudocapacitance of electronically conducting polymers 共ECPs兲 arises from fast and reversible oxidation and reduction processes of the ␲-conjugated polymer chains, called p- and n-doping processes, respectively. Different polymer supercapacitor configurations have been suggested by Rudge et al.,4,5 although it is the n/p type, i.e., the configuration with a p-doped polymer as positive electrode and an n-doped as negative, which is able, in principle, to outperform the DLCSs at comparable cost. Given that n-doping/ undoping is not as easy a process as p-doping, few performance data are reported in the literature on the n/p type supercapacitors, which can be compared to those of commercial DLCSs. Recent results on poly共3-methylthiophene兲-based n/p type supercapacitors with optimized cell design and electrode mass loading indicate that this system delivers significantly lower specific energy than the DLCSs, mainly because of the low conductivity of the poly共3methylthiophene兲 n-doped form.6 By contrast, a successful strategy was followed by developing a hybrid supercapacitor design where the positive electrode is based on a p-doped conducting polymer and the negative electrode on a high-surface-area activated carbon.7,8 In this configuration, the charging process is not of the same type on both electrodes: energy is stored by p-doping the polymer on the positive electrode and charging the double-layer of the activated carbon on the negative. This hybrid system has been developed in the framework of an EU project 共1998-2001 SCOPE project-Joule contract no. JOE3-CT97-0047兲 involving two University groups 共CNAM and Bologna University兲, a battery manufacturer 共CEACEXIDE兲, a capacitor and machinery company 共Arcotronics Italia S.p.A.兲, and a governmental energy agency 共ENEA兲. The target of the project was the development of 3 V polymer supercapacitor prototypes which outperform the DLCSs by delivering specific energy of 8 Wh kg⫺1 of packaged device and of preseries modules with capacitance ⬎1.5 kF and 2 kW stacks. The university groups were charged with the basic studies for electrode and electrolyte materials selection, cell design, and working condition specifications, CEAC-EXIDE with cell scaling up and the development of preseries modules, ENEA with the definition of testing procedures for specific applications in electric and hybrid vehicles and the tests of preseries modules and stacks, and Arcotronics Italia S.p.A. with the development of automatic machinery for electrode lamination and cell assembly. This paper reports the main results of the SCOPE project which, starting from basic studies on the materials in laboratory-scale cells, led to preseries carbon//polymer hybrid supercapacitor modules and stacks operating in propylene carbonate 共PC兲-1 M tetraethylammonium tetrafluoroborate (Et4 NBF4 ), a nontoxic high-boiling-point electrolyte. The characterization of electrode materials in the se-

A646

Journal of The Electrochemical Society, 150 共5兲 A645-A651 共2003兲

lected electrolyte, the optimization of cell design, and the tests on laboratory-scale cells are presented and discussed. Cell scaling-up for the development of high capacitance modules and specific testing of the modules and stacks are discussed in the latter part. Technical data of the automatic machinery for electrode lamination are also reported, and prospects for further improvement of the high performing carbon//poly共3-methylthiophene兲 hybrid system are discussed. Experimental The composite electrodes were prepared as in Ref. 6-8 by mixing active material, poly共3-methylthiophene兲 共pMeT, electrosynthesized兲, poly关3-共4-fluorophenyl兲thiophene兴 共pFPT, chemically prepared兲, or activated carbon 共Spectracorp or PICA兲, graphite 共SFG44, Timcal兲, or acetylene black 共AB, Hoechst兲, carboxy methyl cellulose sodium salt 共Cmc, Aldrich兲, and polytetrafluoroethylene 共PTFE兲 共DuPont兲, followed by lamination on the current collector 共stainless steel grid, Delker, or treated aluminum兲. The pMeT was electrochemically synthesized under galvanostatic conditions at 10 mA cm⫺2 starting from 1.0 M 3-methylthiophene 共Aldrich兲 in acetonitrile 共ACN, Fluka兲-0.5 M Et4 NBF4 共Fluka兲 in a laboratory reactor. The pFPT was chemically prepared as in Ref. 8. The PICACTIF activated carbon and the pMeT composite electrodes used in modules and stacks were prepared by a semiautomatic procedure. The supercapacitor cells (1-4 cm2 ) were assembled from two composite electrodes that were kept apart by a microporous PTFE separator and with PC 共Fluka兲-1 M Et4 NBF4 . All chemicals were reagent-grade products, purified before use. Impedance measurements of the polymer and activated carbon electrodes and of the supercapacitors were carried out in opencircuit conditions in three- or two-electrode mode using a Solartron SI 1255 frequency-response analyzer coupled to a 273A PAR potentiostat/galvanostat. An ac amplitude of 5 mV was used and data were collected in the frequency range 100 kHz to 10 mHz taking 10 points per decade. An Ag quasi-reference electrode, whose potential was checked vs. a saturated calomel electrode 共SCE兲 for each set of experiments, was used during three-electrode mode impedance measurements. The cyclability performance of the supercapacitors was tested by repeated charge/discharge galvanostatic cycles at different current densities with a Perkin Elmer VMP multichannel potentiostat. The electrochemical tests on laboratory-scale cells were performed at T ⫽ 28 ⫾ 2°C in a MBraun Labmaster 150 dry box with argon atmosphere. The modules and stacks were tested at the ENEA laboratory for supercapacitors, equipped with an electronic load HP6050A, a programmable electronic power supply HP6682A, a climatic chamber, an in-house-developed data acquisition system, and precision instrumentation for specific measurements.

Table I. Anodic „E p,a… and cathodic „E p,c… peak potentials of p-doping-undoping CVs in PC-Et4 NBF4 at 20 mV sÀ1 of thin films of pFPT „0.8 mg cmÀ2 … and pMeT „1.3 mg cmÀ2 …. E p,a 共V vs. SCE兲

Polymer pFPT pMeT

E p,c

1.2 0.9

0.7 0.5

Given that the selected route in the project was electrode lamination on current collectors of composite pastes based on electroactive materials, carbon conducting additive, and binder, the evaluation of the maximum polymer content in the composites was of paramount importance for high-performing electrodes. It was estimated that the maximum polymer content was 80% w/w for both pFPT and pMeT, with a minimum carbon additive percentage of 15% w/w. As an example, Fig. 1 shows the impedance spectrum of 80% pMeT composite electrode from which a high specific capacitance of 220 F g⫺1 of pMeT and a low electrode resistance ⬍2 ⍀ cm2 can be estimated. To develop high performance activated carbon//p-doped polymer hybrid supercapacitors, the optimization of electrode mass balancing and loading is an important and intricate issue. Scheme 1 shows the charge-discharge process for the polymer hybrid supercapacitor based on pMeT.

Scheme 1.

The electrode mass balancing in these hybrid supercapacitors must take into account both different capacitance values and different potential ranges for the charging processes of the two-electrode materials. The charge of the capacitive carbon electrode (C surface) can

Results and Discussion Basic studies.—The p-doped polymers are promising materials for positive electrodes in supercapacitors because their specific capacitances are significantly higher than those of the best carbon electrodes used in DLCSs, their charging resistances are significantly lower than those of the carbons, and their cost is low, at least for conventional polymers such as polypyrrole and polythiophene or their simple derivatives. Even the cycling stability of such polymers, a much debated key feature for their use in supercapacitor technology, has been widely demonstrated for the p-doping-undoping process. Both the p-doped and undoped forms of the polythiophene derivatives are stable to oxygen and moisture, and we investigated polymers from this family, particularly pFPT and pMeT. Table I shows the anodic and cathodic peak potentials from the cyclic voltammetries 共CVs兲 in PC-Et4 NBF4 at 20 mV s⫺1 in the p-doping domain of thin films of these polymers directly electrosynthesized on current collectors to display the potential range in which their p-doping-undoping processes take place.

Figure 1. Impedance spectrum 共10 mHz to 10 kHz兲 in PC-1 M Et4 NBF4 of a composite polymer electrode (11.7 mg cm⫺2 , area 1.2 cm2 ) in the charged state 共0.8 V vs. SCE兲 with 80% pMeT (9.3 mg cm⫺2 ), 5% binder, and 15% carbon conducting additive 共acetylene black兲, recorded in three-electrode mode.

Journal of The Electrochemical Society, 150 共5兲 A645-A651 共2003兲

A647

Table II. Electrode composition „% wÕw…, total mass of the composite electrodes per cm2 „w, current collectors not included…, ratio of positive electrode active mass loading to negative „R…, and area of the tested hybrid supercapacitors. Supercapacitor configuration

w (mg cm⫺2 )

R

Area (cm2 )

ACBP25 //pMeT

18

1.56

1.2

ACBP25 //pFPT

24

1.5

4

take place in a wide potential range limited only by the electrolyte electrochemical stability, while the faradaic charge process of the polymer, i.e., the p-doping process, can take place in a limited potential range, ca. 0.5 V. Therefore, to reach cell potentials higher than 2.5 V, the potential excursion of the negative carbon electrode must be at least 2 V. Hence, on the basis of capacitance values of the polymer (⬎200 F g⫺1 ) and of the carbon (⬃130 F g⫺1 ), a ratio higher than 1 of the positive electrode active mass loading to negative, R, was chosen. In addition, total mass loading of the two composite electrodes was selected in the 20 ⫾ 4 mg cm⫺2 range, i.e., a value sufficiently high to minimize the effect of current collector weight. Higher loading values should require a higher amount of electrolyte, i.e., higher electrolyte thickness at the optimized salt concentration, so as not to limit the electrode capacitances given that, as Scheme 1 shows, the salt (C ⫹A ⫺) from the electrolyte is consumed and released during charge and discharge as in DLCSs. But a higher electrolyte thick-

Composition 共% w/w兲 Negative electrode ABBP25 Graphite SFG44 Cmc-PTFE ACBP25 Graphite SFG44 Cmc-PTFE

Positive electrode 90 5 5 80 15 5

pMeT Acetylene black Cmc-PTFE pFPT Acetylene black Cmc-PTFE

80 15 5 80 15 5

ness increases the ionic component of the equivalent series resistance 共ESR兲, adversely affecting the device power performance. Hybrid supercapacitors based on pMeT and pFPT as positive electrode active materials and with the same activated carbon Spectracorp BP25 as negative i.e., supercapacitors ACBP25 //pMeT and ACBP25 //pFPT, respectively, were assembled as in Table II. The two systems were investigated by repeated galvanostatic chargedischarge cycles at different current densities with cutoff potentials 3.0 and 1.0 V for the ACBP25 //pMeT and 3.0 and 1.5 V for the ACBP25 //pFPT. Figure 2 shows the voltage profiles of these two hybrid supercapacitors and of their positive and negative electrodes during the galvanostatic charge-discharge cycles at 5 mA cm⫺2 . The discharge curves in Fig. 2a and b indicate that device capacitance is 0.63 F cm⫺2 (35 F g⫺1 of total composite material兲 for the ACBP25 //pMeT system and that the pMeT specific capacitance is 230 F g⫺1 . The discharge curves in Fig. 2c and d show that the

Figure 2. Voltage profiles of the hybrid supercapacitors 共a兲 ACBP25 //pMeT and 共c兲 ACBP25 //pFPT and of their 共b, d兲 positive and negative electrodes at 5 mA cm⫺2 .

A648

Journal of The Electrochemical Society, 150 共5兲 A645-A651 共2003兲

Figure 4. Ragone plot for supercapacitors 共䉲兲 ACBP25 //pMeT 共cutoff: 3.01.0 V兲 and 共䉮兲 ACBP25 //ACBP25 共cutoff: 2.8-0.0 V兲. Labels indicate the current density in mA cm⫺2 .

Figure 3. 共a兲 ACBP25 //ECP capacitance (C ACBP25 //ECP) and 共b兲 capacity (Q ACBP25 //ECP), and 共c兲 ECP capacitance (C ECP), vs. cycle number for hybrid supercapacitors assembled with positive electrodes based on 共䉱兲 pMeT or 共䉭兲 pFPT.

showed good cycling stability in ACBP25 //ACBP25 double-layer cells. Therefore the pMeT was selected as positive electrode active material for the development of hybrid supercapacitors. To demonstrate the success of the hybrid configuration, Fig. 4 compares the Ragone plots of the ACBP25 //pMeT and ACBP25 //ACBP25 cells, which have almost the same total composite electrode mass loading (20 mg cm⫺2 ); the specific energy, E, and average specific power, P av , were evaluated according to Eq. 1 and 2 E⫽i

冕

V ddt/w

P av ⫽ E/⌬t d , device capacitance of the ACBP25 //pFPT supercapacitor is 0.70 F cm⫺2 (29 F g⫺1 of total composite material兲 and that the specific capacitance of the pFPT is 210 F g⫺1 . In both these hybrid supercapacitors the specific capacitance of the ACBP25 carbon at the negative electrode is ca. 130 F g⫺1 . Figure 3 reports the specific capacitance (C ACBP25 //ECP) and capacity (Q ACBP25 //ECP) per gram of total composite electrode materials and the polymer 共pMeT, pFPT兲 specific capacitance (C ECP) evaluated from repeated galvanostatic cycles at 20 mA cm⫺2 for the ACBP25 //pMeT and at 5 mA cm⫺2 for the ACBP25 //pFPT hybrid supercapacitors. The cycling coulombic efficiency was 99.8-100% for the ACBP25 //pMeT supercapacitor and 99.0-99.4% for the ACBP25 //pFPT. The performance of the two hybrid supercapacitors is completely different as to cycling stability: the ACBP25 //pFPT shows a 0.02% fade of capacitance per cycle estimated over 2000 galvanostatic cycles at 5 mA cm⫺2 , and the ACBP25 //pMeT hybrid supercapacitor, which delivers the same amount of charge per gram of composite materials at 20 mA cm⫺2 , shows a ⬍0.004% capacitance fade per cycle during 10,000 galvanostatic cycles. The poor cycling stability of the ACBP25 //pFPT is ascribed to the pFPT electrode 共the chemistry of the FPT monomer can favor ‘‘charge trapping’’ phenomena during the polymer p-doping-undoping process兲, in that the two hybrid supercapacitors have the same activated carbon in the negative electrode, and this activated carbon in turn

关1兴 关2兴

where i is the current density, V d the cell potential during the discharge curves of the galvanostatic cycles, w the total mass of the positive and negative composite electrodes per cm2 共electrolyte and current collectors not included兲, and ⌬t d the discharge time. The hybrid cell with cutoff potentials 3.0-1.0 V delivers higher specific energy and average specific power than those of the DLCS with 2.8-0.0 V cutoff potentials. In addition, Table III shows the specific energy and the average specific power of the two supercapacitors evaluated at 5 and 20 mA cm⫺2 with the same cutoff potential of discharge, i.e., 1.0 V, a more interesting potential value for practical applications. This table also reports the maximum specific power values ( P max) of the two cells, calculated as in Eq. 3 P max ⫽ 共 V 2 /4R 兲 /w,

关3兴

where V is the cell potential at the beginning of discharge, R is the ESR estimated from the ohmic drop during the galvanostatic discharges, and w is the total mass of the composite electrode materials. The data in Table III show that the ACBP25 //pMeT hybrid supercapacitor delivers significantly higher specific energy than the ACBP25 //ACBP25 , particularly when the latter is discharged down to 1.0 V, and that the maximum specific power of the hybrid supercapacitor is roughly double that of the DLCS. From the ohmic drop at 1 s the ESR of the ACBP25 //pMeT hybrid cell was ca. 12 ⍀ cm2 , a value roughly corresponding to that evaluated from impedance spec-

Journal of The Electrochemical Society, 150 共5兲 A645-A651 共2003兲

A649

Table III. Specific energy „E…, average specific power „P av… of the ACBP25 ÕÕpMeT hybrid supercapacitor and of the ACBP25 ÕÕACBP25 DLCS, evaluated at 5 and 20 mA cmÀ2 with the same cutoff potential of discharge and maximum specific power „P MAX…. Supercapacitor

Cutoff 共V兲

i (mA cm⫺2 )

E (Wh kg⫺1 )

P av (W kg⫺1 )

P M AX (kW kg⫺1 )

ACBP25 //pMeT

1.0-3.0 1.0-2.8

31 17 24 13

512 1820 460 1570

9

ACBP25 //ACBP25

5 20 5 20

troscopy at 1 Hz. Given that the impedance spectra of these hybrid cells displayed bulk resistances of ca. 3 ⍀ cm2 , and the charging resistance of the pMeT composite positive electrode was less than 1 ⍀ cm2 at 1 Hz, as shown in Fig. 1, the main contribution to ESR of the ACBP25 //pMeT hybrid cells is the carbon electrode resistance. To further improve the performance of the pMeT hybrid supercapacitor, a high surface-area carbon from PICA 共PICACTIF兲, which displays very low charging resistance in PC-1 M Et4 NBF4 , i.e., less than 1 ⍀ cm2 at 1 Hz, while maintaining high capacitance, was tested.9 Figure 5 compares the capacitance values at different frequencies, as estimated from impedance spectroscopy, of a PICACTIF carbon composite electrode prepared without conducting additive and 5% binder (ACPICACTIF) to those of a Spectracorp BP25 electrode (ACBP25) with 15% graphite SFG44 and 5% binder. Both electrodes have the same composite mass loading of ca. 7 mg cm⫺2 and were negatively charged. These data demonstrate that the ACPICACTIF composite reaches a limit capacitance value of 100 F g⫺1 , which is lower than that of the ACBP25 composite electrode 共115 F g⫺1兲 but at significantly higher frequency, i.e., in a shorter time with respect to the ACBP25 electrode. The PICACTIF carbon, with its high capacitance and low charge resistance, seems to have achieved the optimized surface area and pore-size distribution required to develop high specific energy and power supercapacitors.9 Hence, the substitution in the hybrid supercapacitor of the ACBP25 electrode with the ACPICACTIF further improves the power performance of the AC//pMeT hybrid system.

The hybrid supercapacitor modules and stacks.—Given the overall results, 3 V ACPICACTIF //pMeT module prototypes featuring ⬎1.5 kF capacitance and PC-1 M Et4 NBF4 as electrolyte were developed and tested. These prototypes were assembled using carboncoated aluminum current collectors so as to decrease contact resistance; certain characteristics of these prototypes, which require further design optimization, are listed in Table IV. Three and four modules were also connected in series to yield stacks with 2-3 kW theoretical power; Fig. 6 pictures the three modules. The modules were first individually characterized at various temperatures (⫺20°C to 60°C兲 and then connected in stacks and tested under in-house protocols to determine capacitance, ESR, specific energy at constant current 共during discharge兲, peak power, and cycle life by galvanostatic cycles with cutoff potentials V max to V max/2. Figure 7 shows the discharge curves at different currents from 5 to 100 A, with cutoff potentials 3.0-1.5 V, of a module charged at 5 A; at the lowest current the capacitance was 1.6 kF with a loss of 15% at the highest currents.

Table IV. Characteristics of ACPICACTIF ÕÕpMeT module prototypes. Cell design Electrode number Positive electrode Negative electrode R Separator Collectors Electrolyte Cell voltage Module case Total weight

Figure 5. Specific capacitance vs. frequency of 共䊏兲 ACPICACTIF and 共䊐兲 ACBP25 carbon-based electrodes in farad per gram of composite material from impedance spectroscopy in PC-1 M Et4 NBF4 . The electrode compositions 共% w/w兲 were ACPICACTIF 95% (6.6 mg cm⫺2 )-binder 5%, and ACBP25 90% (6.0 mg cm⫺2 )-SFG44 5%-binder 5%.

5

24 共prismatic configuration兲 80% pMeT, 15% acetylene black, 5% binder 95% ACPICACTIF , 5% binder 1.4 2 ⫻ 25 ␮m PTFE Treated Al PC-1M Et4 NBF4 3V Stainless steel 400 g 共case included兲

Figure 6. Picture of the prismatic ACPICACTIF //pMeT modules (6 ⫻ 2.5 ⫻ 12.5 cm) to be connected in series for 2 kW stack.

A650

Journal of The Electrochemical Society, 150 共5兲 A645-A651 共2003兲

Figure 7. Discharge curves at different currents of a ACPICACTIF //pMeT module at room temperature at various currents between 3 and 1.5 V.

The cycling stability test data for these ACPICACTIF //pMeT modules are shown in Fig. 8, where the capacitance and ESR at 1 Hz values vs. cycle number over 1600 galvanostatic cycles at 10 A with cutoff 3.0-1.5 V are reported. The initial capacitance decrease was attributed to the stabilization of the potential range in which each electrode works; thereafter cell capacitance stayed almost constant up to cycle 1550, thereby confirming the laboratory-scale cell (1-4 cm2 ) results. The ESR at 1 Hz is greater than the value 共5 m⍀兲 expected from the laboratory-scale cell data for a 1.5 kF module based on ACBP25 carbon, which is more resistive than the ACPICACTIF . This is mainly for a not yet optimized module configuration in terms of electrode connections; for this module prototype the ESR value, determined at 1 kHz, which corresponds to the bulk and current collector resistance, was 3 m⍀, a value which roughly doubles that estimated from laboratory-scale cells 共1.3 m⍀兲. From capacitance and ESR 共1 Hz兲 data 共Fig. 7 and 8兲 the maximum energy (E MAX) and maximum power ( P MAX) per kg of packaged hybrid prototype are 5 Wh kg⫺1 and 2.0 kW kg⫺1 , respectively. Figure 9 shows the Ragone plot in terms of maximum specific energy and maximum specific power per kg of packaged device for an ACPICACTIF //pMeT module and for two commercial carbon//carbon supercapacitors, namely, SC1 and SC2, tested with the same protocol. The energy values of the ACPICACTIF //pMeT prototype are higher than those of the SC2 and more or less the same as

Figure 8. 共䊏兲 Capacitance and 共䊊兲 ESR values during cycling at constant current 共15 A兲 between 3 and 1.5 V of a ACPICATIF //pMeT module prototype at room temperature.

Figure 9. Ragone plot for the 共䉱兲 ACPICACTIF //pMeT prototype at the 50th cycle 共total weight of the cell兲, and two commercial carbon//carbon cells 共䊐兲 SC1 and 共䊊兲 SC2.

for SC1 at the lowest discharge rate, whereas at the higher rates the power values of SC2 are higher than those of the hybrid supercapacitor. However, the great difference between the hybrid polymer prototype and the commercial DLCSs is that they use different electrolytes: both SC1 and SC2 operate in ACN while the hybrid operates in PC. It is well known that the conductivity of ACN-based electrolytes is significantly higher than that of PC-based ones, but the recent specification of the U.S. Department of Energy for the development of Partnership for a New Generation of Vehicles supercapacitors10 requires the substitution of ACN with high boiling point and nontoxic solvents; here we selected PC. Hence, the Ragone plot in Fig. 9 shows that the performance of the ACPICACTIF //pMeT prototype in PC is even more promising. All the data of these hybrid modules are of great interest, particularly taking into account that their design has not yet been optimized for electrolyte content and case material weight as shown in Fig. 10, which reports the weight distribution in the modules. The prototype case weight is around 30% w/w of the total weight, a factor linked to the use of stainless steel, and the substitution of this material with aluminum can decrease case weight by a factor of 3. The electrolyte weight, which represents 35% w/w of prototype weight, is in excess with respect to the amount required by electrode capacitances 共see Scheme 1兲, so that it too can be reduced. By taking into account feasible weight reduction, the estimated optimized

Figure 10. Weight distribution in ACPICACTIF //pMeT modules.

Journal of The Electrochemical Society, 150 共5兲 A645-A651 共2003兲

A651

Figure 11. Specific peak power of a stack of four series-connected ACPICACTIF //pMeT modules.

packaged weight is ca. 100 g per kF and, hence, specific maximum energy and power values become 13 Wh kg⫺1 and 5 kW kg⫺1 , respectively. The final objective of the SCOPE project was the assembly of a complete stack exceeding the nominal power of 2 kW and its characterization via a testing procedure specifically for electric vehicle applications. Accordingly, a stack of four prototype modules was series connected for a theoretical power of about 2.5 kW 共based on the calculated maximum power of each module兲. The stack showed an ESR of about 33 m⍀ at 1 kHz 共and 39 m⍀ at 1 Hz兲, i.e., about twice the sum of the ESR of each module, including contact and connection resistances 共not optimized in the assembly兲. The practical peak specific power was measured by discharging the stack at various constant currents from 12 to 4 V: this procedure was adapted by the European Car Manufacturers Committee 共EUCAR兲 to verify power capability of supercapacitors over a 5 s discharge duration. Figure 11 clearly shows the peak specific power 共the averaged values over the discharge times at various discharge currents兲, with the dashed lines intercepting the value of 475 W/kg at 5 s discharge time. Finally, Fig. 12 shows the scheme of the roll lamination machinery developed to bind the composite electrode materials to the current collectors. This unit generates a ‘‘lamination’’ line 共onedirection force/flow兲, minimizes or prevents air bubbles and wrinkles in the laminated material, and is based on a combination of fixed-gap and controlled lamination force that is achieved with a moveable upper lamination roller. This procedure provides flexibility in the adjustment of the basic process parameters such as speed, temperature, and pressure, and it is a continuous process capable of achieving an effective lamination speed from 3 to 9 m min⫺1 . The specific thermal management of this machinery overcomes the drawbacks generally related to the preheating of current collectors and electrode materials, such as the long preheating time and poor temperature uniformity. Conclusions The EU SCOPE project focused on high energy and power supercapacitors operating with an ACN-free organic electrolyte. The

Figure 12. Arcotronics continuous lamination equipment.

high energy demand for vehicle applications led to the development of supercapacitors featuring a cutting-edge hybrid configuration with activated carbon negative electrode and conducting polymer positive active material. The results of SCOPE, a project which started from basic studies for materials selection and led to the development of preseries hybrid modules and stacks with pMeT as positive electrode and PC-Et4 NBF4 electrolyte that outperform DLCSs, may provide a successful response to the market demand for high power and energy supercapacitors operating in a nontoxic and nonvolatile electrolyte. Acknowledgments This work was funded by the Commission of the European Communities, for Jan 1998 to Sept 2001, under RTD contract SCOPE no. JOE3-CT97-0047. The University of Bologna assisted in meeting the publication costs of this article.

References 1. B. E. Conway, Electrochemical Supercapacitors, Kluwer Academic/Plenum Publishers, New York 共1999兲. 2. A. Burke, J. Power Sources, 91, 37 共2000兲. 3. M. Mastragostino, F. Soavi, and C. Arbizzani, Advances in Lithium-Ion Batteries, B. Scrosati and W. van Schalkwi, Editors, p. 481, Kluwer Academic/Plenum Publishers, Dordrecht 共2002兲. 4. A. Rudge, J. Davey, F. Uribe, J. Landeros, Jr., and S. Gottesfeld, in 3rd International Seminar on Double Layer Capacitors and Similar Energy Storage Devices, Florida Educational Seminars, Inc., Deerfield Beach, FL, Dec 1993. 5. A. Rudge, J. Davey, I. Raistrick, S. Gottesfeld, and J. P. Ferraris, J. Power Sources, 47, 89 共1994兲. 6. M. Mastragostino, C. Arbizzani, and F. Soavi, Solid State Ionics, 148, 493 共2002兲. 7. A. Di Fabio, A. Giorgi, M. Mastragostino, and F. Soavi, J. Electrochem. Soc., 148, A845 共2001兲 8. A. Laforgue, P. Simon, J. F. Fauvarque, J. F. Sarrau, and P. Lailler, J. Electrochem. Soc., 148, A1130 共2001兲. 9. J. Gamby, P. L. Taberna, P. Simon, J. F. Fauvarque, and M. Chesneau, J. Power Sources, 101, 109 共2001兲. 10. R. B. Wright, D. K. Jamison, J. R. Belt, R. A. Sutula, Abstract 230, The Electrochemical Society Meeting Abstracts, Vol. 2002-1, Philadelphia, PA, May 12-17, 2002.