Sei sulla pagina 1di 7

Electrochemistry Communications 9 (2007) 19241930 www.elsevier.

com/locate/elecom

The eect of activation on the electrochemical behaviour of graphite felt towards the Fe3+/Fe2+ redox electrode reaction
Victor Pupkevich, Vassili Glibin, Dimitre Karamanev
*
Department of Chemical and Biochemical Engineering, University of Western Ontario, London, Ontario, Canada Received 1 March 2007; received in revised form 18 April 2007; accepted 24 April 2007 Available online 8 May 2007

Abstract This paper is dedicated to the study of the eect of graphite felt activation by thermal oxidation in air on its electrocatalytic activity towards Fe3+/Fe2+ redox electrode reaction. For the rst time, the exchange current densities and electron transfer coecients determined from the Tafel equation were obtained within the wide range of burn-o levels (050%). The maximal catalytic activity was obtained at the burn-o of 17%. The cathode having this burn-o level expressed almost three-fold enhancement in the galvanic cell performance (criterion for the performance evaluation in our case was a cell voltage at the current density of 300 mA cm2) as compared to that with the non-activated graphite felt, and allowed to obtain current densities up to 670 mA cm2 at the cathode polarization as low as 150 mV. The correlation between electrocatalytic activity and a surface oxide chemistry of graphite felt was established. The cell performance was found to be the best when the pH at a point of zero charge and the amount of surface quinoid groups per unit area were minimal. The results obtained are of signicant importance for practical applications, including the development of electrodes in redox ow batteries. 2007 Elsevier B.V. All rights reserved.
Keywords: Fe3+/Fe2+ redox electrode reaction; Exchange current; Electron transfer coecient; Activated graphite felt; Surface oxide chemistry

1. Introduction Redox ow batteries are very ecient alternative means of energy storage possessing a number of advantages over conventional ones. However, the great potential of redox battery technology has not been fully revealed so far, and a recent commercialisation of three new redox systems [1] shows the potential of this very promising direction. Carbon and graphite felts have been used extensively as electrode materials for redox ow batteries [15], in preparative scale electrolysis [6], as well as for an impurity removal and recovery of metals [7], among other applications. The kinetics of the electrochemical processes on carbon materials is quite sensitive to their physical, chemical,
* Corresponding author. Tel.: +1 519 661 2111x88230; fax: +1 519 661 3498. E-mail addresses: vpupkevi@uwo.ca (V. Pupkevich), vglibin@uwo.ca (V. Glibin), dkaramanev@eng.uwo.ca (D. Karamanev).

and electrochemical pre-treatment such as polishing [811], heat treatment [11,12], argon sputtering treatment [10], laser activation [8], specic covalent bonding of organic molecules to the carbon surfaces [811,1316], intercalation [14,17], activation under various gasses [1823], etc. Apart from the texture properties alteration (i.e. porosity, pore volume, pore size distribution, and surface area), the activation of carbon and graphite bres and other carbon materials leads to formation of functional groups on the surface [823]. The detailed distribution of the oxygen-containing groups on the activated carbon (graphite) surfaces is not fully established yet, but carboxylic, lactone, phenol, carbonyl, anhydride, quinone and ether groups are thought to be predominant [12,15,1720,24,25]. McCreery et al. [9] showed that increasing disorder during the activation allows for increasing the surface density of states at the Fermi level, which facilitates a charge transfer, as well as number of specic chemical sites consequently resulting in a boost of electrochemical activity. Five types of

1388-2481/$ - see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.elecom.2007.04.021

V. Pupkevich et al. / Electrochemistry Communications 9 (2007) 19241930

1925

inuence of oxygen-containing functional groups on electrode reactions have been suggested in the literature: electrocatalytic eects, wetting, chemosorption, double layer and resistivity eects [812,14]. According to the McCreery classication [11,26], electrode kinetics of the Fe3+/Fe2+ redox system is greatly aected by C@O functional groups present on the electrode surface. However, as far as we know, study of electrocatalytic activity of carbon brous materials with various burn-o levels towards the abovementioned redox system has not been conducted yet. In this regard, the main aim of this work was an attempt to correlate the electrocatalytic activity of the activated graphite felt with the alteration of its surface oxide chemistry depending on the burn-o level. 2. Experimental 2.1. Materials The graphite felt SIGRATHERM KFA-5 used in the experiments was purchased from SGL Carbon Corp. (USA). The hydrogen electrode GDE LT 120E-W and cation-exchange membrane Selemion HSF, used for separating cathodic and anodic compartments, were purchased from Fuel Cell Store (USA) and Asahi Glass Co. Ltd. (Japan), respectively. The solution of 0.45 M Fe2(SO4)3 was prepared by dissolving Fe2(SO4)3 (analytical grade, Merck) in deionized water and adjusting pH to 1.0 with sulphuric acid (Caledon). Series of aqueous solutions of heptanoic acid with concentrations of 102, 7 103, 5 103, 3 103, 103 M were prepared by sequential dilution of a 102 M stock solution prepared from pure heptanoic acid (SigmaAldrich). The solution of 0.015 M TiCl3 was prepared by dilution of 20% TiCl3 in 3% HCl (Alfa Aesar). The compressed hydrogen was of UHP grade and was used for carrying out the experiments with galvanic cell. 2.2. Graphite felt activation technique The graphite felt activation was performed via thermal oxidation at 450 2 C in air. For this purpose 20 20 cm specimens of the felt were placed into a mue oven (Vulcan A-550) and weighed after certain periods of time until a desired burn-o level was reached. The treatment temperature was found to be optimal both in terms of the oxidation rate and the possibility of getting uniform burn-o levels across the entire felt surface. Uniformity of oxidation was estimated by measuring the electrical resistance of samples cut from dierent spots of activated graphite felt. 2.3. Electrical resistance measurement For measuring the electrical resistance, 12.5 178 mm H-shaped samples were cut from the graphite felt sheet. The width of the felt strips at both ends was twice as wide

as the tested sample in order to reduce losses caused by the contact resistance. The wide ends of the felt were pressed between two copper plates with similar size by screw clamps. The resistance was measured using a digital multimeter Mastercraft with a precision of 1%. The term apparent specic resistance represents the electrical resistance calculated per unit geometrical (visible) crosssectional area of the sample studied. 2.4. Determination of the specic surface area The surface area of the studied samples was determined using the method of heptanoic acid adsorption [27,28]. Graphite felt samples of known mass (between 0.2 and 0.5 g) were equilibrated with aqueous solutions of heptanoic acid (125 mL) with various concentrations (102, 7 103, 5 103, 3 103 M) for 48 h, and then the surface tension (r, dyn cm2) of the solution was measured by the maximum bubble air pressure using an apparatus described in [28]. The equilibrium acid concentrations (C, mol l1) were calculated by the modied Shishkowski equation [28]: Dr 2:303a log b 2:303a log C 1 where Dr is the dierence between the surface tension of distilled water and that of the solution after adsorption; a and b are constants, equal to 14.1 and 955.0, respectively, in the case of heptanoic acid. Having assumed that the heptanoic acid adsorption (A, mol g1) on the activated felt is described by the Langmuir isotherm (monolayer equilibrium adsorption), the specic surface area of the samples (S, m2 g1) was calculated as follows: S A1 N A S m A1 C1 2

where A1 is the limiting adsorption of heptanoic acid at the graphite felt/solution interface, mol m2, NA is Avogadros number, Sm is the area occupied by one molecule of heptanoic acid, m2, and C1 is the limiting adsorption of heptanoic acid at the solution/air interface, which is equal to a number of moles per unit area, mol m2. The value of the limiting adsorption (A1, mol g1) of heptanoic acid at the solid/liquid interface was obtained from the slope of the linear dependence of C versus C. The value of C1 was calA culated by the following formula: a C1 3 RT where R is the universal gas constant, 8.314 J mol1 K1, T is temperature, K. The area occupied by one molecule of heptanoic acid (Sm, m2) could be determined as: Sm 1 C1 N A 4

14:1 103 8:314 295

In particular, C1 for the heptanoic acid was calculated as 5:7 106 mol m2 and Sm was found to be 2.91 1019 m2.

1926

V. Pupkevich et al. / Electrochemistry Communications 9 (2007) 19241930

The reliability of the technique used was tested by measuring the specic area of an activated KYNOL felt samples with known surface area. The mean error in determining the surface area by this technique was 13%.

2.5. Analysis of surface groups Recently, the most common technique, used for characterization of a chemical composition of surfaces, has become X-ray photoelectron spectroscopy (XPS) [29,30]. The XPS method has received a wide spread because of its capability of determining dierent types of oxygenated species present on the surface of various materials. However, it is not always easy to single out and assign a certain peak to a corresponding functional group [31]. Application of the Fourier transform at analysing spectra allowed to achieve fairly good precision in a peaks assignment [16,17,19,20]. However, in spite of its above advantages it mainly allows to do a surface elemental analysis (i.e. O/C atomic ratio) and relative surface concentrations (%) of given functional groups rather than determine an actual concentration (mmol g1 or mmol cm2). For this reason the functionality of carbon felt surfaces in this work was determined applying chemical techniques. The content of acidic functional groups on the carbon felt surface was determined using the Bohems methods [24]. The samples of graphite felt ($0.20.5 g) were placed into 125 mL Ehrlenmeyer asks, and then 30 mL of bases of dierent strength (0.05 M NaOH or 0.025 M Na2CO3) were introduced and allowed to equilibrate for 48 h. The excess of the base was titrated with 0.05 M HCl. The phenolic group content was determined as a dierence between the results of both titrations. The amount of basic groups on the carbon surface was determined by similar to the above technique with the only dierence that the studied samples were equilibrated with 0.05 M HCl solution and then titrated with 0.05 M NaOH. The pH values at the point of zero charge (pHpzc), or isoelectric point, which could be the measure of a total surface acidity/basicity were obtained by the mass titration method [32]. For this purpose, known mass ($0.01 g) of the activated graphite felt sample was added to 25 mL of 0.1 M NaCl solution and the equilibrium pH was measured. New portions of the felt were added until the pH stabilized. That pH reading was taken as a pH of the point of zero charge. The number of surface quinoid functional groups was determined via their reduction with titanium(III) chloride. The samples of 0.20.6 g were equilibrated with the solution of 0.015 M TiCl3 (60 mL). The residual titanium chloride, left after the reaction with quinoid groups, was titrated, in a strongly acidic medium, with 0.025 M K2CrO4 and phenylantranylic acid as an indicator. In order to prevent oxidation by air of the titanium (III) in solution, the entire procedure was carried out in inert atmosphere.

Fig. 1. Schematic of the galvanic cell.

2.6. Determination of the electrocatalytic activity The electrocatalytic activity of samples having dierent burn-o levels was estimated on the basis of exchange current density of the Fe3+/Fe2+ redox couple as well as by the electron transfer coecient determined from the Tafel equation. The exchange current density is the product of the electrochemical specic rate constant and a concentration term, and, in this respect, is useful in comparing the catalytic activities of dierent electrode materials for given reaction. For this reason, polarization curves of a galvanic cell with various ow-through cathodes were recorded. Commercially available hydrogen electrode was used as the anode. Both electrodes had geometric (visible) area of 17 cm2. The anodic and cathodic compartments were separated by cation-exchange membrane Selemion HSF (Fig. 1). This kind of a cell was used to create conditions as close to those in redox ow batteries as possible. The overall reaction taking place in the cell was Fe2 (SO4 )3 + H2 ! 2FeSO4 + H2 SO4 . 5

The catholyte, 0.45 M Fe2(SO4)3, was pumped through the cathode in a direction parallel to the membrane with the ow rate of 70 mL min1. Electrode polarization curves were measured by the steady-state galvanostatic method using an electronic load system Chroma 63103. All the experiments were carried out at 21.5 C and hydrogen pressure of 1 atm (abs). 3. Results and discussion The kinetic curve of the thermal oxidation of graphite felt in air at 450 C and the eect of the burn-o level on the electrical resistance are shown in Fig. 2. It can be seen

V. Pupkevich et al. / Electrochemistry Communications 9 (2007) 19241930

1927

Fig. 2. Plots of the burn-o level versus time (a) and the specic apparent resistance versus burn-o (b) for the SIGRATHERM KFA-5 graphite felt.

that at the very beginning of the thermal oxidation process, there is a period ($5 h) where the burn-o rate of the sample is gradually increasing (Fig. 2a). Beyond that initial period, the activation intensies and the burn-o becomes almost linear function of time (within the burn-o range of 1550%). Apparently, such behaviour points out that the oxidation follows an autocatalytic reaction mechanism. Similarly, the dependence of the apparent specic resistance (Fig. 2b) can be described by two dierent curves. Initially, it increases almost linearly, by about 45% at the burn-o of 21.8% compared to untreated graphite felt, while the further oxidation leads to an exponential growth of resistance with the burn-o level. To evaluate the electrocatalytic activity of the felt, the graphite felt cathodes were tested in the above-mentioned galvanic cell. As shown in Fig. 3, the galvanic cell with activated cathode (at 17.4% of burn-o level) showed much higher voltage at the same current density as compared to the untreated graphite felt. The treated electrode allowed to obtain more than twice as high current density at short circuit as that obtainable with the non-activated graphite felt. Such a signicant enhancement in performance is also conrmed by the exchange current densities (Fig. 4). In particular, the samples with 17.4% burn-o had a nine-fold increase in the exchange current density compared to those with no activation. These data once again show that the sample burn-o level in the vicinity of 17% is optimum in terms of electrocatalytic activity towards the Fe3+/Fe2+ redox reaction and facilitates the electron transfer (a = 0.47 at 17.4% burn-o). It should also be noted that the potential of the untreated graphite felt cathode at open circuit was by 100 mV lower as compared to that of activated, which could point out that the irreversible potential was imposed on the electrode.

In order to nd out the origin of this phenomenon, the main surface chemistry characteristics of the graphite felt samples with dierent burn-o percentages were determined (Table 1, Fig. 5). As shown in Table 1, the amounts of the individual surface functional groups gradually increase as the function of a sample burn-o level and a specic surface area. In contrast to the dependence of a surface group concentration on the burn-o level, relationship between pH of the point of zero charge, characterizing the colligative acid-base properties of a surface, and the burno level has the minimum at 21.8% (Fig. 5), and is equal to 6.01. The surface chemistry data obtained are qualitatively in good agreement with studies by XPS method, which revealed presence of the same types of functional surface groups on the pre-treated carbon and graphitized carbon bres [19,2123]. It was previously shown that the sample treatment in various gaseous atmospheres, such as CO2 or O2 [19,21], N2 and O2 mixtures [22], O3 [23] etc., results in increasing the amount of dierent types of functional groups. However, so far there is not too much consistency between XPS results on the ratios of surface functional groups available in the literature. In particular, Blyth et al. [19] determined the highest concentration of phenolic groups, and less but almost equal amounts of carbonyl and carboxylic groups. In contrast, Jin et al. [23] reported that the major surface group is a carbonyl one, while carboxylic and phenolic are present in fairly equal numbers, but are twice as less as for carbonyl. Apparently, surface functional group composition is highly sensitive to the pretreatment method as well as carbon material origin, which, in bundle, account for a signicant dierence in surface group concentrations. Similarly to Jins et al. work [23], our experimental data (Table 1) points out that the

1928

V. Pupkevich et al. / Electrochemistry Communications 9 (2007) 19241930

Fig. 3. Polarization curves and cell voltage for a galvanic cell with non-treated and activated (17.4% burn-o) graphite felt cathode at 21.5 C.

Fig. 4. Plots of the exchange current density and electron transfer coecient versus sample burn-o level.

Table 1 Surface properties of the SIGRATHERM KFA-5 graphite felt Burn-o (%) 0 9.6 17.4 21.8 34.6 49.7 62.0 Surface area (m2 g1) 12 48 16 180 23 300 38 420 53 540 26 370 46 Number of groups (mmol g1) Carboxylic 0.248 0.450 0.530 0.720 0.870 0.910 Phenolic 0.034 0.041 0.275 0.387 0.680 0.490 Quinoid 0.64 1.13 1.16 2.21 3.09 2.48 Basic 0.016 0.019 0.036 0.022 0.071 0.212

predominant groups present on the surface are quinoid (carbonyl) ones. In accordance with McCreerys studies [9,26], those make the major contribution into electrocatalytic activity of dierent carbon materials. Mainly it occurs because of formation of inner sphere complexes between surface carbonyl groups and aquated iron ions, which in turn facilitate electron transfer. Beyond that, a comparison of stability constants for the complexes of ferric and ferrous ions with dierent oxygen-containing organic ligands in the bulk revealed that the former are always more stable compared

V. Pupkevich et al. / Electrochemistry Communications 9 (2007) 19241930

1929

Fig. 5. Plots of the pHpzc and the number of surface quinoid groups versus sample burn-o level.

to the latter ones [33]. Having supposed that stability of surface complexes behaves in a similar manner, this fact can apparently be considered as another favourable factor for carrying out the electrochemical reaction. However, bridging mechanism model seems not to reect the entire complexity of the electrochemical process and another factor, such as the low double layer capacitance, could favour the electron transfer as well. Frysz et al. [18] reported that carbon bres with 1722% burno possess the lowest capacitance, which is a result of their microstructure, i.e. microholes, being formed at thermal oxidation in air, do not deeply penetrate into the interior yet. In addition, we believe that there are other factors, which could signicantly lower electrocatalytic activity. According to Kastening et al. [34] protons adsorbed on the surface @C@O groups (quinoid groups in our particular case) create holes (h+) by the following scheme: 6 In turn, the holes formed are mobile charge carriers, which account for a double layer capacity of the activated carbon electrodes, and also can be considered as traps for electrons. Having studied chemical and surface properties of dierent carbon blacks, Fabish et al. [35] established that their work function has a non-linear dependence on a pH, and its minimum falls into the pH range of 5.56.0. It is interesting to note, that in our case, the minimum pHpzc value was found to be in the same pH range for the sample with the burn-o of 17.4% (Fig. 5), which also exhibited the highest electrocatalytic activity. According to Kinoshita [36] such a minimum is observed due to the least inuence

of the basal plane electronic systems (or Cp). In this regard, it is reasonable to explain reduction in the electrochemical activity after the burn-o of 17% by a signicant increase in the amount of basal planes on the surface. This suggestion is indirectly supported by shifting the pHpzc into basic region as the burn-o level increases (Fig. 5), since, it is thought, the basal plane system behaves as a Lewis base and takes part in the following reaction [36]: Cp + 2H2 O = Cp H3 O + OH 4. Conclusions A highly ecient cathode material was obtained by thermal oxidation of the SIGRATHERM KFA-5 graphite felt. Its electrocatalytic properties towards the Fe3+/Fe2+ redox reaction allowed to obtain current densities up to 670 mA cm2 at low polarization of the cathode, which makes it very promising as an electrode in redox ow cells. The study of the surface properties of the activated graphite felt showed that observed maximum of electrocatalytic activity is a complex eect of: (a) the proper number of functional groups present on the surface, which in turn are responsible for formation of inner sphere surface complexes with iron ions, and in this way, facilitate electron transfer; (b) minimal amount of the surface @C@O groups per unit area on the activated graphite felt, which account for a double layer capacitance and (c) low electrical resistance. Acknowledgements This work was supported nancially by the Natural Science and Engineering Research Council of Canada and by the Ontario Centres for Excellence. 7

1930

V. Pupkevich et al. / Electrochemistry Communications 9 (2007) 19241930 [19] R.I.R. Blyth, H. Buqa, F.P. Netzer, M.G. Ramsey, J.O. Besenhard, P. Golob, M. Winter, Appl. Surf. Sci. 167 (2000) 99. [20] H. Buqa, R.I.R. Blyth, P. Golob, B. Evers, I. Schneider, M.V. Santis Alvarez, F. Hofer, F.P. Netzer, M.G. Ramsey, M. Winter, J.O. Besenhard, Ionics 6 (2000) 172. [21] R.I.R. Blyth, H. Buqa, F.P. Netzer, M.G. Ramsey, J.O. Besenhard, M. Winter, J. Power Sources 9798 (2001) 171. [22] W.H. Lee, J.G. Lee, P.J. Reucroft, Appl. Surf. Sci. 171 (2001) 136. [23] Z. Jin, Z. Zhang, L. Meng, Mater. Chem. Phys. 97 (2006) 167. [24] H.P. Bohem, in: D.D. Eley, H. Pines, P.B. Weisz (Eds.), Adv. Catal., vol. 16, Academic Press, New York, 1966, p. 179. [25] M. Pakua, S. Biniak, A. Swiantkowski, Langmuir 14 (1998) 3082. [26] R.L. McCreery, in: A. Wieckowski (Ed.), Interfacial Electrochemistry, Marcel Dekker, New York, 1999. [27] E. Yeager, A.J. Salkind (Eds.), Techniques of Electrochemistry, Wiley-Interscience, New York, 1972. [28] H.M. Halbouni, Development of the Fe3+/Fe2+ redox electrode for use in biofuel cells, MESc Thesis, The University of Western Ontario, London, Canada, 2005. [29] D. Briggs, Surface analysis of polymers by XPS and static SIMS, Cambridge University Press, New York, 1998. [30] J.F. Watts, An Introduction to Surface Analysis by XPS and AES, Wiley, New York, 2003. [31] H. Estrade-Szwarckopf, Carbon 42 (2004) 1713. [32] J.S. Noh, J.A. Schwarz, J. Colloid Interf. Sci. 1130 (1989) 157. [33] J. Bjerrum, G. Schwarzenbach, L.G. Sillen, Stability Consatnts, The Chemical Society, London, 1957. [34] B. Kastening, M. Hahn, J. Kremeskotter, J. Electroanal. Chem. 374 (1994) 159. [35] T.J. Fabish, D.E. Schleifer, Carbon 22 (1984) 19. [36] K. Kinoshita, Carbon: Electrochemical and Physicochemical Properties, Wiley, New York, 1988.

References
[1] C.-H. Bae, E.P.L. Roberts, R.A.W. Dryfe, Electrochim. Acta 46 (2002) 287. [2] S. Zhong, M. Skyllas-Kazacos, J. Power Sources 39 (1992) 1. [3] M. Bartolozzi, J. Power Sources 27 (1989) 219. [4] J. Gonzalez-Garcia, V. Montiel, A. Aldaz, J.A. Conesa, J.R. Perez, G. Codina, Ind. Eng. Chem. Res. 37 (1998) 4501. [5] F. Moraw, K. Fatih, D. Wilkinson, F. Girard, Adv. Mater. Res. 1517 (2007) 315. [6] A Deronsier, D. Limosin, J.-C. Moutet, Electrochim. Acta 32 (1987) 1643. [7] Y. Oren, A. Soer, Electrochim. Acta 28 (1983) 1649. [8] C.A. McDermott, K.R. Kneten, R.L. McCreery, J. Electrochem. Soc. 140 (1993) 2593. [9] R.L. McCreery, K.K. Cline, C.A. McDermott, M.T. McDermott, Colloid Surf. A 93 (1994) 211. [10] P. Chen, M.A. Fryling, R.L. McCreery, Anal. Chem. 67 (1995) 3115. [11] P. Chen, R.L. McCreery, Anal. Chem. 68 (1996) 3958. [12] L.I. Khmylko, G.V. Steinberg, V.P. Glibin, G.I. Novikov, Elektrokhimiya (Electrochemistry) 20 (1984) 649 (in Russian). [13] C. Saby, B. Ortiz, G.Y. Champagne, D. Belanger, Langmuir 13 (1997) 6805. [14] C.E. Banks, T.J. Davis, G.G. Wildgoose, R.G. Compton, Chem. Commun. 7 (2005) 829. [15] G.G. Wildgoose, N.S. Lawrence, H.C. Leventis, L. Jiang, T.G.J. Jones, R.G. Compton, J. Mater. Chem. 15 (2005) 953. [16] G. Liu, J. Liu, T. Bocking, P.K. Eggers, J.J. Gooding, Chem. Phys. 319 (2005) 136. [17] G.G. Wildgoose, M.E. Hyde, N.S. Lawrence, H.C. Leventis, L. Jiang, T.G.J. Jones, R.G. Compton, Langmuir 21 (2005) 4584. [18] C.A. Frysz, D.D.L. Chung, Carbon 35 (1997) 1111.

Potrebbero piacerti anche