Asymmetric capacitance response from the chemical characteristics of activated carbons in KOH electrolyte
Gangwei Sun,Donghui Long,Xiaojun Liu,Wenming Qiao,Liang Zhan,Xiaoyi Liang,Licheng Ling
DOI: https://doi.org/10.1016/j.jelechem.2011.05.017
IF: 4.598
2011-01-01
Journal of Electroanalytical Chemistry
Abstract:Highlights ► Effect of oxygen-containing functional groups on capacitance behaviors are compared in KOH and H 2 SO 4 electrolyte. ► Asymmetric capacitance behaviors are observed in KOH electrolyte. ► Gravimetric capacitance of positive and negative electrode is 409 F g −1 and 271 F g −1 respectively. Abstract The positive and negative capacitance behaviors of heat-treated activated carbons are separately studied in a three-electrode configuration using 6 M KOH and 1 M H 2 SO 4 as electrolytes. Heat treatment of activated carbons at 400–800 °C causes slight changes in surface area accompanied by a significant removal of oxygen-containing functional groups. Consequently, the specific surface capacitance with a 25% reduction in the negative and an 8% reduction in the positive are monitored in KOH electrolyte. The unequal capacitance reductions as well as the asymmetric CV curves reveal that the negative pseudo-capacitances are much larger than the positive ones in KOH electrolyte. These results could be ascribed to the redox reactions preferring in the negative region. The insertion process of cation also plays a significant effect on the asymmetric capacitance behavior. Compared with the identical electrochemical response in H 2 SO 4 electrolyte, huge difference in the capacitance values of both electrodes in KOH electrolyte will give a less contribution to the cell capacitance. To optimize the cell performance, a suitably increasing the weigh of active materials in the positive, especially for activated carbon with high level of oxygen-containing functional groups, is recommended. Keywords Electrochemical capacitors Oxygen-containing functional groups Pseudo-capacitance Inorganic electrolyte Asymmetric capacitance response 1 Introduction Electrochemical capacitors (ECs) are extensively studied due to their promising properties in terms of energy storage and power supply, which could fill the gap between the secondary batteries and the conventional dielectric capacitors [1–5] . In the last decade, activated carbons (ACs) have been proved as the most promising electrode materials for ECs, due to their high specific surface area, tunable pore structure, adjustable surface chemical characteristics, high chemical stability and favorable price [6–9] . For ACs-based ECs, although charge is stored mainly through the formation of the electrical double-layer, the overall capacitance can also be significantly improved through the introduction of foreign atoms, such as oxygen, nitrogen, phosphorus, etc. [10–15] . These foreign atoms are believed to form the pseudo-capacitance via fast faradic redox reactions. However, the electrochemical events of surface chemistry occurring in carbon materials are very complex, which may take place depending on, for instance, the kind and distribution of heteroatoms on carbon surface, the population of surface functional groups, history of the carbon materials, kind of electrolytes, pHs values, applied electrode potential and cell design, etc. [16–23] . It has been reported that the oxygen-containing functional groups can be categorized into CO-desorbing and CO 2 -desorbing oxides, the former (such as carbonyl, hydroxyl and quinone groups) is responsible for pseudo-capacitance rather than the latter (carboxyl, lactones and anhydrides) [20] . Conway et al. suggested that redox reactions had a large dependence on the pH values of electrolytes [21] . The redox with the quinine/hydroquinone pair required protons to proceed, whereas pyridinic group might be electrochemically active at the working potential of the basic solution [21,22] . Besides, the redox reactions occurring in the positive and negative regions should be different, as different electrolyte ions located and different potentials applied on each electrode [22] . Papers [24–26] compare positive and negative electrode characteristics in acid and alkaline environment. It is clearly indicated in these papers that in the alkaline environment (KOH), the capacitance of negative electrodes is much higher, while in the acidic environment (H 2 SO 4 ), the relationship could be reverse and the capacitances of both electrodes are similar. And a new type of high energy asymmetric capacitor has been developed to enlarges the operating voltage up to 1.5 V in acidic medium [27,28] . The present paper aimed to compare the pseudo-capacitance behaviors in H 2 SO 4 with that in KOH through the separate study of both electrodes. We used a special-designed configuration where a reference electrode was added to the symmetric two-electrode cell. The oxygen-containing functional groups of ACs were changed simply by a heat treatment. Differing from results literature reported that the contribution of pseudo-capacitance in basic medium was significantly lower than that in acidic medium [23] ; we found that the pseudo-capacitance contributed by oxygen-containing functional groups in KOH electrolyte was actually much larger than that in H 2 SO 4 electrolyte, especially for the negative electrode. It was the asymmetric capacitance distribution of both electrodes that mislead us to consider the minor effect of pseudo-capacitance in KOH electrolyte. Based on these results, the mass matching ratio of two electrodes in KOH electrolyte was theoretically optimized to improve the overall capacitance of ECs. 2 Experimental 2.1 Preparations of samples Petroleum coke was chemically activated using KOH (the mass ratio of KOH to coke is 2.5:1) at 800 °C for 2 h under nitrogen atmosphere. The resultant AC (denoted RAC) was then subjected to the heat treatment under a nitrogen flow for 2 h at 400, 500, 600, 700 and 800 °C, respectively. The heat-treated samples were labeled as N – X , where X represented as the heat-treated temperature (HTT). 2.2 Characterization of samples The pore textural characterization of samples was based on the nitrogen adsorption at 77 K (ASAP 2020 M, Micromeritics). Before the measurement, samples were degassed in vacuum at 573 K for at least 12 h. BET surface area ( S BET ) was calculated from a set of adsorption data obtained at relative pressure below 0.3. The pore size distribution was determined from non-local DFT equation using slit-pore model. Elemental analysis was performed using an elemental analyzer (Elementar Co., Vario EL III). XPS spectra were achieved with a Kratos Axis Ultra DLD photoelectron spectrometer using non-monochromatized Al Kα radiation, the source being operated at 15 kV and 10 mA. Prior to elemental analysis and XPS measurement, samples were dried for 2 h at 120 °C. 2.3 Electrochemical measurements Electrodes were made in the form of pellets composed of ACs (85%), polyterafluoroethlyene (5%) and acetylene black (10%). Two-electrode capacitors were built in a Teflon Swagelok® type system with two identical carbon pellets and a porous polymeric separator (Celgard, USA), in which Hg/HgO and Hg/Hg 2 SO 4 placed between the electrodes acted as reference electrode for 6 M KOH and 1 M H 2 SO 4 , respectively. Galvanostatic charge–discharge tests were conducted with the cell voltage range from 0 to 0.9 V using an Arbin SCTS supercapacitor testing system (Arbin instrument, TN, USA). The potential variation of each electrode was automatically recorded during charge–discharge process. The specific discharge capacitance of a single electrode (positive or negative electrode) in a capacitor was calculated according to C = It / m Δ V , where Δ V , I , t and m were the working voltage of each electrode, discharge current, discharge time, and mass of either electrode, respectively. For calculating capacitances of cell, m and Δ V should be the overall weight of both electrodes and the full voltage of the capacitor, respectively. A standard three-electrode configuration was also set up using activated carbon as working electrode, platinum foil and Hg/HgO or Hg/Hg 2 SO 4 as counter electrode and reference electrode, respectively. Cyclic voltammograms (CV) were recorded with a PCI4/300 potentiostat (Gamry, PA, USA) on three-electrode system in the potential range of −0.9 to 0.1 V vs. Hg/HgO and −0.7–0.3 V vs. Hg/Hg 2 SO 4 . Several disks were analyzed in order to obtain reliable data. 3 Results and discussion 3.1 Characteristics of pores structure Fig. 1 presents N 2 adsorption–desorption isotherms and resultant DFT pore size distributions of all samples. The porosity parameters are summarized in Table 1 . All isotherms exhibit a very similar type I isotherm according to the classification of IUPAC, with a typical microporous characteristics. The thermal treatment leads to some decreases of N 2 adsorption amount, corresponding to the textural modifications. When the RAC is treated below 700 °C, there is a slight reduction in the N 2 adsorption volume, and the S BET and V t decrease from 2023 to 1757 m 2 g −1 and from 1.09 to 0.92 cm 3 g −1 , respectively. However when the temperature increases to 800 °C, the texture changes more significantly. The DFT pore size distributions of all samples are almost similar with the pore size range from 0.4 to 3.5 nm, suggesting that the pore size is unchanged but the amount of pores decreases. Apart from the possibility of the pores collapse, the formation of micro-crystalline structure of activated carbons may also lead to the deterioration of the pores texture. 3.2 Characteristics of surface chemistry Table 2 lists the elemental analysis of the raw and heat-treated samples. The RAC contain ca. 88.9% carbon, 0.9% nitrogen, 9.3% oxygen. After the heat treatment, the percentage content of nitrogen is almost similar, being independent on the heat-treated temperature. The carbon percentage content increases gradually with the increase of the HTT, which is at the expense of the remarkable decrease of the oxygen percentage content. XPS characterization of samples is performed to further study the changes of surface chemistry during the heat treatment. The O/C atomic ratios based on the areas of C 1s and O 1s peaks are also summarized in Table 2 . RAC display the highest O 1s /C 1s ratio (0.21). This ratio decrease to 0.16 at the HTT of 400 °C and final to 0.058 at 800 °C. These results suggest that the oxygen functional groups could be removed effectively by the heat treatment. 3.3 Electrochemical characteristics 3.3.1 Asymmetric capacitance response in KOH electrolyte Galvanostatic characteristics have been recorded vs. the Hg/HgO reference electrode in a three-electrode cell. In such a configuration, the performances of each individual electrode (positive and negative) as well as the cell can be monitored separately. Fig. 2 depicts a typical example of the galvanostatic charge–discharge curves at the current density of 0.1 A g −1 . The potential of negative and positive electrodes (right Y -axis) are automatically recorded corresponding to applied cell voltage of 0.9 V (left Y -axis) in KOH electrolyte. When the cell is in the completely discharged state (0.08 V), the positive and negative potential locates at −0.47 and −0.55 V vs. Hg/HgO, respectively. While the cell is in the completely charged state (0.9 V), the positive potential reaches 0.03 V vs. Hg/HgO, whereas the negative potential decreases to −0.87 V vs. Hg/HgO. In our investigations, working potentials of capacitor electrodes shift in the range from −0.87 to 0.03 V, so within a thermodynamic stability electrochemical windows from −0.97 to 0.25 V according to the Nernst equation. A close look at Fig. 3 reveals that the potential range of positive electrode (ΔE + ) is larger than that of negative one (ΔE + > ΔE − ), indicating that the cell voltage splits unequally between two electrodes. The gravimetric capacitances of positive and negative electrodes at 0.1 A g −1 in KOH electrolyte, which are calculated from the galvanostatic charge/discharge measurement, as a functional of HTT are plotted in Fig. 3 a. All negative capacitances are obviously larger than the positive ones. RAC presents the highest capacitances both in negative and positive electrodes, which are 409 F g −1 and 271 F g −1 , respectively. Heat treatment cause unavoidable decrease in gravimetric capacitance for both electrodes, with a remarkable reduction of 182 F g −1 in the negative and relatively slight reduction of 81 F g −1 in the positive. The deterioration of gravimetric capacitance could be associated with the decrease of S BET and the removal of oxygen functional groups. We exclude the effect of S BET through the normalized specific surface capacitance by dividing the S BET of samples. As shown in Fig. 4 , specific surface capacitance of the positive experiences a slight change (<8%), indicating that it mostly depends on porosity characteristics. On the contrary, a significant reduction of specific surface capacitance (from 20.2 to15.0 μF cm −2 , over 25%) is found for the negative causing by the heat treatment. In this case, specific surface capacitance should be contributed from pure electrostatic capacitance and pseudo-capacitance; the former directly is governed by pore-ions electrostatic interaction, whereas the latter is related to the oxygen-containing functional group. Heat treatment leads to the improvement of electrical conductivity, which should play a positive effect on the pore-ions interaction and improve the capacitance values. Therefore, it is apparent that the reduction of capacitance in the negative can be related to the weakened pseudo-capacitance with HTT increasing. Fig. 3 b shows the influence of current density on the difference of gravimetric capacitance value (Δ C ) between negative and positive electrodes. As the average pore size of the active material is large enough to allow the accessible to the electrolyte ions, ion-sieving effect could be eliminated. Such huge differences with respect to gravimetric capacitance of both electrodes would be understood as a consequence of the pseudo-capacitances. Oxygen-containing functional groups exhibit an asymmetric contribution to the capacitance of each electrode, having a more sensitive effect on the improvement of capacitance in the negative electrode. As revealed by Fig. 3 b, Δ C reduces with the HTT increasing. This can be related to the thermal instability of functional groups, which decompose gradually with the intensity of heat-treatment. Consequently, the extra capacitance of the negative electrode produced from functional groups decreases, leading to the reduction of Δ C with the HTT raising. On the other hand, the decrease of Δ C as a function of current density can also be found for each sample. This can be attributed to the electrochemical instability of oxygen-containing functional groups as the cycle increasing and the kinetic limit of redox reaction with the current density increasing. It is worthy to note that Δ C for RAC presents the more prominent reduction as compared to that of other samples. This is due to the fact that most of oxygen-containing functional groups in RAC are electrochemical instability as reported in literature [20] . With the cycling increasing, a large amount of groups would be removed. Meanwhile, the high content of functional groups would also increase the electrochemical resistance of RAC. As a consequence, Δ C for RAC decreases dramatically at the current density of 4 A g −1 , being lower than that for N400 and N500. CV curves of all samples in KOH electrolyte (shown in Fig. 5 ) exhibit a clear deviation from the ideal rectangular shape, shifting to the negative potential. The response capacitances in the negative region are much higher than that in the positive ones. With the HTT increasing, the distortion degree of CV curves becomes less, and CV for sample N800, is more excellent. The variation of both positive and negative capacitances with the HTT increasing is in good agreement with the results of galvanostatic charge–discharge tests. With the voltage sweep rate increasing, response capacitances of both electrodes become similar, which is linked to the limited kinetic process involving in the redox reactions in the negative potential range. 3.3.2 Symmetric capacitance response in H 2 SO 4 electrolyte In H 2 SO 4 electrolyte, the gravimetric capacitances of positive and negative electrodes at 0.1 A g −1 are plotted in Fig. 6 . Unlike the remarkable capacitance difference between the positive and negative electrode in KOH electrolyte, the negative electrode in H 2 SO 4 electrolyte demonstrates only slightly higher capacitance as compared to positive one. Heat treatment also causes a significant capacitance reduction of both electrodes to the same extent. Fig. 7 demonstrates that CV recorded in a three-electrode cell shows a relatively symmetric shape in positive and negative regions. Well-defined redox humps can be observed from RAC both in the anodic and cathodic regions. Redox peaks are significantly small for the heat-treated samples and almost negligible for N800. In H 2 SO 4 electrolyte, the almost same capacitances of both electrodes and symmetric CV shape reveal that the pseudo-capacitance equally takes place in the positive and negative regions. 3.3.3 Comparison of capacitance behaviors in H 2 SO 4 and KOH electrolyte A separate study of capacitive performance of each electrode reveals oxygen-containing functional groups exhibit an asymmetric contribution to the capacitance, having a more sensitive effect on the improvement of capacitance of the negative electrode in KOH electrolyte, whereas, the pseudo-capacitances in H 2 SO 4 electrolyte contribute to the overall capacitance of each electrode to the same extent. The pseudo-capacitances involving in the oxygen-containing functional group in H 2 SO 4 solution are widely referred [29,30] . The most well documented case is the oxidation/reduction of hydroquinone/quinine groups. These quinine groups are responsible for a significant percentage of the cell capacitance. Even if the mechanisms of complex reactions are not thoroughly understood until now, we could conclude, at least in partial, that oxygen-containing functional groups exhibit symmetric capacitive behaviors in both electrode. On the contrary, with the presence of functional groups, an equilibrium reaction may occur in the carbon negative electrode especially in KOH medium [22,31,32] : (Reaction (1)) - C x O - + K + ↔ - C x O - ‖ K + (Reaction (2)) - C x O + e - 1 + K + ↔ - C x - OK These reactions may facilitate an excess specific capacitance in the negative electrode due to interaction between surface functional groups and cation. Additionally, as calculated according to the Nernst equation, underpotential hydrogen adsorption/desorption process might be occur at −0.9 V vs. Hg|HgO as follows [33] : (Reaction (3)) - C x - O + e - 1 + H 2 O ↔ - C x - OH + OH - In alkaline electrolytes, water is reduced to hydrogen atom which can be adsorbed on the surface and electrochemical oxidation during the discharge, thus providing extra pseudo-capacitance. Finally, the cation insertion process could also account for higher capacitance in negative region. In order to prove the occurrence of K + insertion reaction, the capacitive performances of all samples in 0.5 M K 2 SO 4 electrolyte are investigated. For all samples, except for N800, capacitance in negative electrode is higher than that in positive one as presented in Fig. 8 . And the differences in capacitance decrease with HTT increasing. For neutral medium, the effect of oxygen-containing functional groups can be negligible. The extra capacitance in negative electrode can be ascribed to insertion of K + into the lattice of activated carbons under effect the electric field. After subjected to heat treatment, the rearrangement and reorganization of carbon microcrystal would cause the passivation and closure of exposed sites. Consequently, the effect of insertion is diminished. As the aforementioned analysis, pseudo-capacitance resulted from oxygen-containing functional groups should exhibit more significant in KOH than in H 2 SO 4 electrolyte. Nevertheless, Fig. 9 shows that rather similar deterioration of capacitance in both electrolytes. Results can be ascribed to an asymmetric capacitance behavior of functional groups in KOH electrolyte which would weaken its contribution to the improvement of capacitance properties. 3.3.4 Mass matching of both electrodes in KOH electrolyte for asymmetric ECs Since pseduocapacitance exhibits an asymmetric distribution between the positive and negative electrodes in KOH electrolyte, the poor capacitive performance of the positive electrode would limit the capacitance of whole device [34,35] . Therefore, mass matching between two electrodes is necessary for maximizing the capacitance of cell according to the capacitive characteristics of samples. The capacitance of the cell ( C c ) is determined by the series equivalent circuit consisting of the positive capacitance C P and negative capacitance C N , according to the following equation: (Equation (1)) 1 / C c = 1 / C p + 1 / C N If the mass ratio of positive to negative electrode is r , the specific capacitance C (F g −1 ) of cell should be expressed as: (Equation (2)) C c = C P C N / ( C N + rC R ) 1 + 1 / r The capacitance values (e.g. C P , 274 F g −1 ; C N , 401 F g −1 for RAC) at the current density of 0.1 A g −1 are adopted for calculation. The dependence of the cell capacitance of asymmetric ECs on the mass ratio is shown in Fig. 10 . A strong correlation of specific capacitance with the mass ratio is observed, with the optimum mass ratio of 1.2 for RAC, 1.1 for N600 and 1.00 for N800. This suggests that improvement in the capacitor performance can be achieved by increasing the weight of active materials in positive electrode, especially for these materials with high level of oxygen functional groups. However, as revealed from theatrical calculation, we find that the improvement of capacitance is negligible through mass matching, even though difference in capacitance of both electrodes is huge. Additionally, it should be clear that the optimum mass ratio is not constant at various charge/discharge currents, since the variation extent of capacitance of each electrode differs along with the charge/discharge current. 4 Conclusion The electrochemical performances of activated carbons are tested in a three-electrode system to evaluate separately the positive and negative capacitive behavior. In KOH electrolyte, the negative capacitance is remarkably larger that the positive one, whereas the nearly same capacitance values for each electrode are found in H 2 SO 4 electrolyte. Heat treatment of activated carbons causes slight changes in pore texture accompanied by a significant reduction of oxygen content. Consequently, the positive and negative capacitances as well as the cell capacitance of EDLC decrease both in KOH and H 2 SO 4 electrolyte. The remarkable decrease in negative and slight decrease in positive specific surface capacitance, as well as the asymmetric shapes in CV curves, confirms that the contribution of pseudo-capacitances in the negative electrode are much larger than that in the positive electrode in KOH electrolyte. Compared with symmetric electrochemical responses on each electrode in H 2 SO 4 electrolyte, the asymmetric behavior of pseudo-capacitance in KOH electrolyte could give a less contribution to the total cell capacitance. Thus, optimum mass matching ratio between positive and negative electrodes should be recommended to improve the total supercapacitor performance. Acknowledgments This work was partly supported by National Science Foundation of China (No. 50730003 ), National Project of Scientific and Technical Supporting Programs funded by Ministry of Science & Technology of China (No. 2007BAE55B00 ), National High Technology Research and Development Program of China (No. 2007AA05Z311 ), and Program for New Century Excellent Talents in University ( NCET-07-0285 ). References [1] M. Winter R.J. Brodd Chem. Rev. 104 2004 4245 [2] H.S. Chen T.N. Cong W. Yang C.Q. Tan Y.L. Li Y.L. Ding Prog. Nat. Sci. 19 2009 291 [3] I. Hadjipaschalis A. Poullikkas V. Efthimiou Renew. Sust. Energy Rev. 13 2009 1513 [4] A. Burke Electrochim. Acta 53 2007 1083 [5] A. Lewandowski M. Galinski J. Power Sources 173 2007 822 [6] E. Frackowiak Phys. Chem. Chem. Phys. 9 2007 1774 [7] J.S. Huang B.G. Sumpter V. Meunier Chem. Eur. J. 14 2008 6614 [8] L.L. Zhang X.S. Zhao Chem. Soc. Rev. 38 2009 2520 [9] A.G. Pandolfo A.F. Hollenkamp J. Power Sources 157 2006 11 [10] D. Hulicova-Jurcakova A.M. Puziy O.I. Poddubnaya F. Suarez-Garcia J.M.D. Tascon G.Q. Lu J. Am. Chem. Soc. 131 2009 5026 [11] D. Hulicova-Jurcakova M. Seredych G.Q. Lu T.J. Bandosz Adv. Funct. Mater. 18 2008 1 [12] C.X. Zhang D.H. Long B.L. Xing W.M. Qiao R. Zhang L. Zhan X.Y. Liang L.C. Ling Electrochem. Commun. 10 2008 1809 [13] D. Hulicova-Jurcakova M. Seredych G.Q. Lu N.K.A.C. Kodiweera P.E. Stallworth S. Greenbaum T.J. Bandosz Carbon 47 2009 1576 [14] H.L. Guo Q.M. Gao J. Power Sources 186 2009 551 [15] E. Raymundo-Pinero F. Leroux F. Beguin Adv. Mater. 18 2006 1877 [16] I.Y. Jang H. Muramatsu K.C. Park Y.J. Kim M. Endo Electrochem. Commun. 11 2009 719 [17] G. Lota B. Grzyb H. Machnikowska J. Machinikowski E. Frackowiak Chem. Phys. Lett. 404 2005 53 [18] N.D. Kim W. Kim J.B. Joo S. Oh P. Kim Y. Kim J. Yi J. Power Sources 180 2008 671 [19] M. Seredych D. Hulicova-Jurcakova G.Q. Lu T.J. Bandosz Carbon 46 2008 1475 [20] P.Z. Cheng H.S. Teng Carbon 41 2003 2057 [21] H.A. Andreas B.E. Conway Electrochim. Acta 51 2006 6510 [22] D. Hulicova J. Yamashita Y. Soneda H. Hatori M. Kodama Chem. Mater. 17 2005 1241 [23] V. Ruiz C. Blanco E. Raymundo-Pinero V. Khomenko F. Beguin R. Santamaria Electrochim. Acta 52 2007 4969 [24] K. Jurewicz R. Pietrzak P. Nowicki H. Wachowska Electrochim. Acta 53 2008 5469 [25] K. Jurewicz K. Babel R. Pietrzak S. Delpeux H. Wachowska Carbon 44 2006 2368 [26] R. Pietrzak K. Jurewicz P. Nowicki K. Babel H. Wachowska Fuel 86 2007 1086 [27] E. Raymundo-Pinero K. Kierzek J. Machnikowski F. Beguin Carbon 44 2006 2498 [28] V. Khomenko E. Raymundo-Pinero F. Beguin J. Power Sources 195 2010 4234 [29] B.E. Conway Electrochemical Supercapacitors 1999 Kluwer Academic/Plenum New York [30] H. Malmberg V. Ruiz C. Blanco R. Santamaria A. Lundblad P. Bjornbom Electrochem. Commun. 9 2007 2320 [31] H.F. Li H. Xi S. Zhu Z. Wen R. Wang Micropor. Mesopor. Mater. 96 2006 357 [32] H. Oda A. Yamashita S. Minoura M. Okamoto T. Morimoto J. Power Sources 158 2006 1510 [33] V. Ruiz H. Malmberg C. Blanco A. Lundblad P. Bjornbom R. Santamaria J. Electroanal. Chem. 618 2008 33 [34] J.L. Li F. Gao J. Power Sources 194 2009 1184 [35] Y.H. Wen J. Cheng G.P. Cao Y.S. Yang J. Appl. Electrochem. 37 2007 543