Electrochimica Acta 53 (2008) 7730–7735

Contents lists available at ScienceDirect

Electrochimica Acta

journa l homepage: www.e lsev ier .com/ locate /e lec tac ta

Competitive effect of KOH activation on the electrochemical performances of

carbon nanotubes for EDLC: Balance between porosity and conductivity

a, S

ijing 1

e comincr

s of tmicroon elcal peoltamsurf

perfo, ACN

Bin Xua,b,∗, Feng Wub,∗, Yuefeng Sub, Gaoping Caoa Research Institute of Chemical Defense, Beijing 100083, Chinab School of Chemical Engineering & the Environment, Beijing Institute of Technology, Be

a r t i c l e i n f o

Article history:Received 17 February 2008Received in revised form 4 May 2008Accepted 10 May 2008Available online 21 May 2008

Keywords:Electric double layer capacitorCarbon nanotubesActivationSpecific surface areaCapacitance

a b s t r a c t

This work is focused on th(EDLCs) between porositynanotubes (CNTs). A serieapplication in EDLCs. TheN2 adsorption, transmissiment. Their electrochemicharge/discharge, cyclic vtion enhances the specificconductivity and the rateporosity and conductivity

1. Introduction

Electric double layer capacitor (EDLC) is an energy storage

device based on the electrical adsorption of ions at the elec-trode/electrolyte interface (non-Faradic process). EDLC has anumber of desirable characteristics, such as high power density andsafety, rapid charge/discharge capability and found applications inmemory backup, uninterruptible power sources, electric vehicles,digital communications, etc. [1]. The carbon electrode characteris-tic of high surface areas is crucial to the performance of the EDLCs[2–16]. Among the various carbon materials, CNTs are regarded aspromising electrode materials for high power-density EDLC due totheir large pore size, high electric conductivity and chemical sta-bility, and other beneficial features [9–16].

As the specific surface area of CNTs is generally lower than200 m2 g−1, their specific capacitance is much lower than thatof activated carbons with high surface area (1000–3000 m2 g−1).Some strategies have been proposed to enhance the capacitance ofthe CNTs. Of these, KOH activation [17–20] is an interesting one.The strong chemical etching effect of KOH creates lots of defects onthe nanotube walls, dramatically increasing the surface area of theCNTs, and therefore its specific capacitance. Previous efforts [17–20]were mainly focused on increasing the capacitance. However, the

∗ Corresponding author. Tel.: +86 10 68912508; fax: +86 10 68451429.E-mail addresses: xubin@bit.edu.cn (B. Xu), wufeng863@vip.sina.com (F. Wu).

0013-4686/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.doi:10.1016/j.electacta.2008.05.033

hi Chenb, Zhiming Zhoub, Yusheng Yanga

00081, China

petitive effects on the performance of the electric double layer capacitorsease and simultaneous conductivity decrease for KOH-activated carbonhe CNTs have been activated with KOH to enhance their surface areas forstructure of the activated carbon nanotubes (ACNTs) is characterized with

ectron microscopy (TEM) observation and electric conductivity measure-rformances are evaluated in aqueous KOH electrolyte with galvanostaticmetry, and ac impedance spectroscopy. It is found that the KOH activa-

ace area of the CNTs and its specific capacitance but decreases its electricrmance in EDLC. By controlling the activation of the CNTs to balance theTs with both high capacitance and good rate performance are obtained.

© 2008 Elsevier Ltd. All rights reserved.

destruction to the tubular structure may decrease the conductivityof the CNTs, deteriorating its rate capability in EDLCs. Few papersreported the effects of KOH activation on the rate performance ofthe CNTs-based EDLCs.

In this paper, we focus on the competitive effects of the poros-

ity and the conductivity on the electrochemical performance of theEDLC. Both the capacitance and the rate capability of the activatedcarbon nanotubes (ACNTs) in 7 mol L−1 KOH aqueous electrolyte areevaluated with galvanostatic charge/discharge, cyclic voltammetry,and ac impedance spectroscopy, etc. It will be seen that optimiza-tion to the KOH activation is of critical importance to obtainingACNTs with both high capacitance and good rate performance inaqueous EDLCs.

2. Experimental

2.1. Preparation and characterization of ACNTs

The CNTs were prepared by pyrolyzing C2H2/H2 with Ni powderas the catalyst. The residual Ni powder was later removed withnitric acid. The CNTs were then mixed with some KOH powder andheated to 800 ◦C at 10 ◦C min−1 in a tubular furnace under nitrogenatmosphere (99.999%) and kept there for 1 h. ACNTs were obtainedafter washing the mixture with diluted HCl and water subsequently.The sample prepared with KOH/CNT weight ratio of n is named asACNT-n.

ica Acta 53 (2008) 7730–7735 7731

B. Xu et al. / Electrochim

The specific surface area and porous structure of the ACNTs weredetermined by N2 adsorption/desorption isotherms at 77 K (Quan-tachrome NOVA 1200). All the samples were degassed at 250 ◦Cfor 8 h prior to the adsorption tests. The specific surface area wascalculated using the conventional BET (Brunauer–Emmett–Teller)method while the total pore volume was estimated by the amountof the adsorbed N2 at a relative pressure of 0.95. Transmissionelectron microscope (TEM, Gem2010) was used to observe the mor-phology of the CNTs before and after activation. The conductivitywas measured by the conventional 4-probe DC method.

2.2. Electrochemical performance evaluation

A mixture of 87 wt% of CNTs, 10 wt% of acetylene black and 3 wt%of PTFE binder was pressed into pellets (19 mm in diameter) as theelectrodes. Then the electrodes were dried under vacuum at 120 ◦Cfor 12 h. Button-type capacitor was assembled with two ACNT elec-trodes using 7 mol L−1 KOH aqueous solutions as the electrolyte.

The capacitors were galvanostatically cycled between 0 and 1.0 Von a Land cell tester. The specific capacitance (C) of a single ACNTelectrode was determined with the formula C = 2It/�Vm, where Iis the discharge current, t is the discharge time, �V is the poten-tial change in discharge and m is the mass of the active electrodematerial in one electrode. The cyclic voltammetry and impedancespectra were recorded on an electrochemistry workstation Solartry

1280B. Equivalent series resistance (ESR) was measured at 1 kHz.

3. Results and discussion

3.1. Microstructure of ACNTs

The ACNTs were obtained by activating CNTs with KOH at 800 ◦Cfor 1 h. The weight ratio of KOH to CNTs varied between 1:1 and5:1. Fig. 1 compares the N2 adsorption/desorption isotherms ofthe CNTs and ACNTs. The isotherm of the CNTs is between typeII and IV according to the IUPAC classification, indicating its meso-porous features. The isotherm of ACNTs is similar to that of theCNTs, implying that mesoporosity is still predominant after activa-tion. The enhanced N2 adsoption volume of the ACNTs means theincrease of pore volume.

The effect of KOH activation on the porosity of the CNTs is shownin Fig. 2. The BET surface area of the CNTs is only 166 m2 g−1. Itincreases drastically with the increase of KOH/CNT ratio, reach-ing 595 m2 g−1 at KOH/CNT ratio 3:1. Then it increases slowly withthe further increase of KOH/CNT ratio and reaches the maximum

Fig. 1. Nitrogen adsorption/desorption isotherms of the CNTs and ACNTs.

Fig. 2. Effect of KOH activation on the specific surface area (a), pore volume (b) andpore size (c) of the CNTs.

value (644 m2 g−1) at KOH:CNT = 5:1, thrice larger than that of theCNTs. Meanwhile, the pore volume of the ACNTs increases withthe increase of KOH/CNT ratio from 0.450 cm3 g−1 before activa-tion to 0.984 cm3 g−1 at KOH/CNT ratio 3:1. Then it decreases withthe further increase of KOH/CNT ratio. The average pore size of theACNTs decreases continuously with the increase of KOH/CNT ratiofrom 10.9 nm before activation to 5.5 nm at KOH/CNT ratio 5:1. Thismeans that the size of the new pores created by KOH activation issmaller than that of the CNTs.

The conductivity of the CNTs was measured by the conventional4-probe DC method. Fig. 3 shows that KOH activation has a neg-ative effect on the conductivity of the CNTs. The conductivity ofthe pristine CNTs is 16.3 S cm−1. It decreases after KOH activation.It drops to 12.9 S cm−1 when the KOH/CNT ratio increases to 3:1.As the KOH/CNT ratio increases to 5:1, the conductivity decreasesto 11.1 S cm−1, only about 70% of the pristine CNTs. Although the

ica Act

7732 B. Xu et al. / Electrochim

Fig. 3. Effect of KOH activation on the conductivity of the CNTs.

conductivity of the ACNTs is lower than the pristine CNTs, it is stillmuch higher than that of activated carbons, which is generally lessthan 1 S cm−1 [21].

Fig. 4 compares the TEM images of the CNTs before and afterKOH activation. Before activation, the CNTs (Fig. 4a and b) are mul-tiwalled with diameters between 10 and 20 nm. The CNTs are easily

Fig. 4. TEM imagines of the CNT (a, and b), A

a 53 (2008) 7730–7735

purified with nitric acid. Both the outer and inner walls are verysmooth. It is easy to recognize the graphene layers of the CNTs.After activation, both the morphology and the structure of the

nanotubes change obviously. The CNTs activated at KOH/CNT = 3:1(ACNT-3, Fig. 4c and d) roughly keep their nanotubular morphol-ogy. However, the tubes become short and distorted. Many defectsare generated on their walls and their surfaces become very rough.It is difficult to distinguish the graphene layers from each other.As the KOH/CNTs ratio further increases, the tubular structure isseverely damaged (Fig. 4e and f for ACNT-5). Some of nanotubesbecome fractures and even powdered. This is clearly attributed tothe strong chemical reaction between CNTs and KOH at high tem-perature. Some of the active sites on the surface of the CNTs areetched with KOH and become pores. As a result, the BET surface areaand pore volume of the CNTs increase. Destruction of the graphiticstructure of the nanotubes leads to decrease of the conductivity.The more the KOH is used, the severer the structural destructionwill be.

3.2. Electrochemical performances of ACNTs

Galvanostatic charge/discharge was performed to determinethe specific capacitance of the ACNTs. The linear voltage-timedependence (Fig. 5a) demonstrates the typical capacitive behavior

CNT-3 (c, and d) and ACNT-5 (e, and f).

B. Xu et al. / Electrochimica Act

Fig. 8 shows the cyclic voltammograms of an ACNT-3-basedcapacitor between 0 and 1.0 V scanned at 2, 50, 200 and 500 mV s−1,respectively. The CV curve at 2 mV s−1 is a well-defined rectangle,implying that its capacitive behavior is exclusively due to the elec-trostatic attraction. At higher scan rates, the CV profiles will usuallydeviate from the ideal rectangular shape due to polarization. Untilthe scan rate increases to 200 mV s−1, the CV curve still remainsits symmetrical rectangular shape, almost perfect for EDLCs. Evenif the scan rate increases to as high as 500 mV s−1, the CV curvesalmost keep rectangular. The CV for the ACNT electrodes over a widerange of scan rates confirms that the ACNT-based capacitors haveexcellent rate performance. From the TEM observation (Fig. 4), theACNTs are basically nanotubes composed of graphite sheet. There-fore, the ACNTs inherit the high electric conductivity of the CNTs(Fig. 3). The ACNTs are dominated with mesopores as deduced fromthe N2 adsorption/desorption isotherms (Fig. 1), making the ion dif-fusion easy in the ACNTs. The high conductivity and mesoporousstructure ensure the good rate capability of the ACNTs.

Fig. 5. Charge–discharge curves of the ACNT3-based EDLC, current: (a) 8.4, (b)210 mA.

of the electrode material. The negligible IR drop at the begin-ning of the charge and discharge under the current of 8.4 mA(∼200 mA g−1) actually indicates that conductivity of the ACNTs ishigh. The coulombic efficiency is as high as 99%, implying the goodcharge/discharge reversibility of the ACNTs. One of the advantagesof the EDLC over a second battery is the rapid charge/dischargecapability of the former. Even when the current density increases to5000 mA g−1, the charge/discharge voltage profile of the capacitorremains its linear shape (Fig. 5b). The discharge process finishes in

3 s, indicating the good rapid discharge capability of the capacitor.

The effect of KOH activation on the capacitance of CNTs in7 mol L−1 KOH aqueous electrolyte at a current density of 50 mA g−1

is illustrated in Fig. 6. The capacitance of the non-activated CNTsis only 18.5 F g−1. As the KOH/CNT ratio increases, the capacitanceof the activated CNTs increases continuously, consistent with thevariation of BET surface area. The ACNT-5 with the highest spe-cific surface area has the highest specific capacitance of 53.6 F g−1,twice larger than that of the non-activated CNTs. This means thatKOH activation is an effective method to reinforce the capacitanceof the CNTs. The enhanced capacitance can be mainly attributed tothe enhancement of surface area.

Theoretically, the capacitance of the carbon is proportional tothe specific surface area of the carbon material. In practice, how-ever, the situation is more complicated. Significant deviation fromthis simple law has been frequently observed [22–26]. Besides sur-face area, factors such as the precursor, pore size, conductivity andsurface chemistry are believed to influence the capacitance of theporous carbons [21–31]. Fig. 7 shows the linear dependence ofthe specific capacitance of the ACNTs on their BET surface area.As the ACNTs are obtained from the same precursor and by the

a 53 (2008) 7730–7735 7733

Fig. 6. Effect of KOH activation on the capacitance of CNTs in 7 mol L−1 KOH aqueouselectrolyte.

same method, they should have similar surface chemistry. As allthe ACNTs are highly conductive (>11 S cm−1), the voltage changeof each ACNTs sample at the beginning of discharge is actually neg-ligible at a current density 50 mA g−1. The average pore size in theACNTs is over 5.5 nm, large enough for the ions to access. Therefore,all the surface area contributes to the double layer capacitance. As aresult, the capacitance of the ACNTs is proportional to their surfacearea, consisting with the theoretical plate capacitor model.

The Nyquist plots of the ACNTs prepared with different KOH/CNTratios are shown in Fig. 9. Each plot is composed of a semicircleat high frequency and a nearly vertical line at low frequency. The

Fig. 7. Correlation between specific capacitance and BET surface area for the ACNTs.

7734 B. Xu et al. / Electrochimica Acta 53 (2008) 7730–7735

Fig. 8. Cyclic voltammetry of ACNT3-based EDLC using 7 mol L−1 KOH aqueous electrolyte at the scan rate of 2 (a), 50 (b), 200 (c), and 500 (d) mV/s.

vertical line indicates the capacitive behavior of the ACNTs. As thepore size of the ACNTs is large enough for the access of ions in KOHaqueous electrolyte, the sloping linear region at high-to-mediumfrequency related to diffusion resistance is not observed, indicat-ing that the diffusion is not the control factor in the kinetics of theelectrode process. Comparing the Nyquist plots of the EDLCs withdifferent ACNTs, it is found that the ESR increases with the KOH/CNTratio. The ERS of ACNT-1 is only 0.189 �, but it increases to 0.239 �for ACNT-3 and 0.294 � for ACNT-5. The more the amount of KOHis used, the more severe will the nanotubular morphology be dam-aged (Fig. 4). As a result, increasing the KOH/CNT ratio leads to thedecrease of the electric conductivity and hence the increase of theESR.

Fig. 9. Nyquist plots of ACNT-based EDLCs using 7 mol L−1 KOH aqueous electrolyte.Fig. 10. Capacitance retention ratio (a) and specific capacitance (b) of the ACNTs in7 mol L−1 KOH aqueous electrolyte as a function of current density.

ica Act

[[

[

[

B. Xu et al. / Electrochim

The CNTs are superior over the activated carbons as electrodematerials for EDLCs in their excellent rate performance due totheir high electric conductivity and large pore size. Fig. 10(a)shows the dependence of the capacitance of the ACNTs onthe charge/discharge current density. The capacitance retentionratio of all the ACNTs samples decreases slightly with increas-ing current density, indicating that the ACNTs allow rapid iondiffusion. This further confirms the good rate performances ofthe ACNTs, consistent with the results from CV curves. However,the ACNTs prepared with different KOH/CNT ratios show obvi-ously different rate performances. The ACNTs with higher KOH/CNTratios show lower capacitance retention ratio at the same dis-charge current density. As the current density increases from 50to 5000 mA g−1, the capacitance of ACNT-1 decreases only 15%while that of ACNT-5 decreases 40%. This is attributed to thedecrease of the electric conductivity with the increase of KOH/CNTratio.

As the KOH/CNT ratio increases, the specific capacitance of theobtained ACNTs increases because of the enhancement of the BETsurface area. On the contrary, higher KOH/CNT ratio will destroythe tubular structure more severely and the electric conductivity ofthe CNTs decreases, deteriorating the rate performance. As a result,

the dependence of capacitance of the ACNTs on the current densityvaries as shown in Fig. 10(b). As for our CNTs, the KOH/CNT = 3:1should be the best ratio to balance the specific surface area andthe conductivity so as to get both high capacitance and good rateperformance.

4. Conclusions

Enhanced capacitance of CNTs has been obtained by KOH activa-tion. The specific capacitance of the ACNTs is proportional to theirBET surface area as the pores are large enough to access for the ionsin 7 mol L−1 KOH aqueous electrolyte. The ACNTs also show goodrapid charge/discharge capability and rate performance becausethey basically inherit the high electric conductivity and large poresize of the non-activated CNTs. KOH activation has two oppositeeffects on the performance of the resulted ACNTs. It enhances thespecific capacitance but deteriorates the rate performance of theCNTs. Therefore, KOH activation has to be optimized to balancethe two effects to get both higher capacitance and better rateperformance. A KOH/CNT ratio of 3:1 is found the best for ourCNTs.

[

[[[

[[[

[

[

[

[[

[

[[[

[[

a 53 (2008) 7730–7735 7735

Acknowledgements

Financial support for this work was provided by the NationalHigh Technology Research and Development Program (863Program, 2006AA11A165, 2006AA03Z342), National Key BasicResearch and Development Program (973 Program, 2002CB211800)and the National Science Foundation of China (NSFC, 20633040).

References

[1] R. Kotza, M. Carlen, Electrochim. Acta 45 (2000) 2483.[2] E. Frackowiak, F. Beguin, Carbon 39 (2001) 937.[3] F.C. Wu, R.L. Tseng, C.C. Hu, C.C. Wang, J. Power Sources 144 (2005) 302.[4] C.C. Hu, C.C. Wang, F.C. Wu, R.L. Tseng, Electrochim. Acta 52 (2007) 2498.[5] H.Q. Li, R.L. Liu, D.Y. Zhao, Y.Y. Xia, Carbon 45 (2007) 2628.[6] B. Xu, F. Wu, S. Chen, C. Zhang, G. Cao, Y. Yang, Electrochim. Acta 52 (2007) 4595.[7] M. Sevilla, S. Alvarez, T.A. Centeno, A.B. Fuertes, F. Stoeckli, Electrochim. Acta 52

(2007) 3207.[8] Y. Zhu, H. Hu, W.C. Li, X. Zhang, J. Power Sources 162 (2006) 738.[9] C. Niu, E.K. Sichel, R. Hoch, D. Moy, H. Tennent, Appl. Phys. Lett. 70 (1997) 1480.10] C. Liu, H.M. Cheng, J. Phys. D: Appl. Phys. 38 (2005) R231.11] E. Frackowiak, K. Metenier, V. Bertagna, F. Beguin, Appl. Phys. Lett. 77 (2000)

2421.12] B. Xu, F. Wu, R. Chen, G. Cao, S. Chen, G. Wang, Y. Yang, J. Power Sources 158

(2006) 773.13] Y.T. Kim, T. Mitani, J. Power Sources 158 (2006) 1517.

14] K.H. An, W.S. Kim, Y.S. Park, Y.C. Choi, S.M. Lee, D.C. Chung, D.J. Bae, S.C. Lim,

Y.H. Lee, Adv. Mater. 13 (2001) 497.15] B. Xu, F. Wu, F. Wang, S. Chen, G. Cao, Y. Yang, Chin. J. Chem. 24 (2006) 1505.16] C.G. Liu, H.T. Fang, F. Li, M. Liu, H.M. Cheng, J. Power Sources 160 (2006) 758.17] E. Frackowiak, S. Delpeux, K. Jurewicz, K. Szostak, D. Cazorla-Amoros, F. Beguin,

Chem. Phys. Lett. 361 (2002) 35.18] Q. Jiang, M.Z. Qu, G.M. Zhou, B.L. Zhang, Z.L. Yu, Mater. Lett. 57 (2002) 988.19] B. Xu, F. Wu, Y. Su, G. Cao, S. Chen, Y. Yang, Acta Chim. Sin. 65 (2007) 2387.20] Q. Jiang, Y. Zhao, X.Y. Lu, X.T. Zhu, G.Q. Yang, L.J. Song, Y.D. Cai, X.M. Ren, L. Qian,

Chem. Phys. Lett. 410 (2005) 307.21] M.J. Bleda-Martenez, J.A. Macia-Agullo, D. Lozano-Castello, E. Morallon, D.

Cazorla-Amoros, A. Linares-Solano, Carbon 43 (2005) 2677.22] G. Gryglewicz, J. Machnikowski, J.E. Lorenc-Grabowska, G. Lota, E. Frackowiak,

Electrochim. Acta 50 (2005) 1197.23] D. Lozano-Castello, D. Cazorla-Amoros, A. Linares-Solano, S. Shiraishi, H. Kuri-

hara, A. Oya, Carbon 41 (2003) 1765.24] O. Barbier, M. Hahn, A. Herzog, R. Kotz, Carbon 43 (2005) 1303.25] M. Endo, T. Maeda, T. Takeda, Y.J. Kim, K. Koshiba, H. Hara, M.S. Dresselhaus, J.

Electrochem. Soc. 148 (2001) A910.26] G. Salitra, A. Soffer, L. Eliad, Y. Cohen, D. Aurbach, J. Electrochem. Soc. 147 (2000)

2486.27] D. Qu, H. Shi, J. Power Sources 74 (1998) 99.28] A.G. Pandolfo, A.F. Hollenkamp, J. Power Sources 57 (2006) 11.29] Y.J. Kim, Y. Horie, Y. Matsuzawa, S. Ozaki, M. Endo, M.S. Dresselhaus, Carbon 42

(2004) 2423.30] G.J. Lee, I. Pyun, Electrochim. Acta 51 (2006) 3029.31] E. Raymundo-Pinero, K. Kierze, J. Machnikowski, F. Beguin, Carbon 44 (2006)

2498.