- Email: [email protected]
Electrochemical capacitors as attractive power sources
Solid State Ionics 265 (2014) 61–67
Contents lists available at ScienceDirect
Solid State Ionics journal homepage: www.elsevier.com/locate/ssi
Electrochemical capacitors as attractive power sources M. Meller, J. Menzel, K. Fic, D. Gastol, E. Frackowiak ⁎ Institute of Chemistry and Technical Electrochemistry, Poznan University of Technology, Piotrowo 3, 60-965 Poznan, Poland
a r t i c l e
i n f o
Article history: Received 1 April 2014 Received in revised form 6 July 2014 Accepted 18 July 2014 Available online xxxx Keywords: Supercapacitor Pseudocapacitance Iodide Vanadium species
a b s t r a c t The electrochemical performance of various carbon materials as supercapacitor electrodes with redox active electrolytes has been presented. In situ Raman investigation was carried out to show possible iodine species, polyiodides and carbon/iodine interactions during electrode polarization. Apart from the electrolyte also the kind of current collector plays an important role. Gold current collector is not adapted because of its reactivity with iodides whereas stainless steel is convenient. The conjugated iodide and vanadium species as the electrolyte were also investigated with Nafion separation. The best performance was obtained for AAC 1 and AAC 2 carbon material especially after adding 10% of carbon nanotubes. © 2014 Elsevier B.V. All rights reserved.
1. Introduction There is an extremely fast growth in market of all digital and electronic devices that are destined for portable applications. Simultaneously with their resilient development one can observe unrelenting increase of demands for high energy sources. Instead of commonly used lithium-ion (Li-ion) and nickel metal hydride (Ni-MH) batteries, there is possibility to consider electrochemical supercapacitors for this purpose. Unfortunately, there are still some difficulties to fulfill the market demand. It is well known that electrochemical double layer capacitor (EDLC) can serve as an excellent energy storage device for all applications with high power requirements. That fact is caused by the nature of the electrical charge accumulation mechanism, which is based on the electrostatic forces. Hence, it can be charged and discharged in a very short time and release enormously high peak of power [1,2]. The problem is that the energy of this system is incomparably lower than for mentioned batteries and that is the reason why they cannot be replaced so easily by the electrochemical capacitors. Because capacitance value strictly depends on the surface of the electrode/electrolyte interface, the best materials for electrodes should be made from activated carbons due to their well developed specific surface area (up to 2500 m2 g−1). They might be used in different forms, such as powders, woven cloths, felts or fibers [3–6]. Moreover, the surface functionalities of applied carbon material cannot be neglected [7].
⁎ Corresponding author at: Poznan University of Technology, Piotrowo 3, 60-965 Poznan, Poland. E-mail address: [email protected] (E. Frackowiak).
http://dx.doi.org/10.1016/j.ssi.2014.07.014 0167-2738/© 2014 Elsevier B.V. All rights reserved.
In order to achieve higher capacitance values and increase the energy of the electrochemical capacitor, the electrode materials capable to reversible redox reactions might be considered. This phenomenon of capacitance enhancement consists of pseudocapacitance effects which are related to the surface functionalities of carbon and/or to the presence of electroactive oxides from such transition metals as Ru, Ir, W, Mo, Mn, Ni and Co [8–12]. However, these materials have some limitations that do not allow them to be applied efficiently, mainly because of their slow solid state kinetics and, what is also very important, significantly higher price than for the activated carbons, especially in the case of ruthenium and iridium. Another way to increase the energy is using electrolytes that can stand as a source of redox reactions. There are some papers focused on that matter concerning various possibilities, like for example, using additives to electrolytes with oxygen functionalities, e.g. quinone/hydroquinone groups [13–20]. These groups might be able to graft on the electrode surface with generation of new oxygen functional species that can react reversibly during charging/discharging process and provide additional charge [16]. It is also possible to introduce pseudocapacitance from redox active electrolytes in the form of alkali metal iodide aqueous solutions. It has been shown by our group [21–23] that iodides supply enormous capacitance but only at the positive electrode. Our preliminary investigations indicated that a binary electrolyte based on iodides and vanadyl sulfate could be a good option to enhance capacitance of full system [24]. Taking into account all advantages of such binary redox electrolytes we decided in a present paper to investigate in more detail their activity on several activated carbon materials with different porous structure, especially, mesoporosity. A
62
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
comparison will be shown between the various activated carbons operating in the redox active electrolytes based on sodium and potassium iodide as well as vanadyl sulfate in the form of a single electrolyte component or as a redox active pair operating as catholyte and anolyte. The effect of separating such two redox couples as well as the role of nanotubes as an additive to active mass of electrodes will be elucidated. The consequence of using two different current collectors will also be discussed. The detailed study of carbon/iodine species interface by Raman spectroscopy will be presented.
2. Experimental Physicochemical properties, such as specific surface area and pore size distribution as well as oxygen content with type of surface functionalities of all carbon used, were determined by nitrogen adsorption (ASAP 2020 instrument from Micromeritics, USA) and thermogravimetry coupled with mass spectrometry in order to recognize evacuating species (TGA F1 Iris and Aeolos MS instruments by Netzsch, Germany). In some cases electrodes have been also investigated. All electrochemical experiments were made in twoand three electrode capacitor configurations using Swagelok® cell. Electrode materials were prepared from different commercially available activated carbons, i.e. SX2 (AAC 0) and DLC Super 30 Norit (AAC 1), composed of 85 wt.% of carbon, 10 wt.% of polyvinylidene fluoride (PVDF) and 5 wt.% of carbon black. In order to receive carbon material with better developed micro/mesoporosity, carbon SX2 was activated by KOH (1:4) giving carbon AAC 2. Additionally, activated carbon tissue ACC 507-20 (AAC 3) from Kynol was also applied, because of its microporous character and high specific surface area. The physicochemical properties of all utilized materials are described in more detail in the next paragraph. In the case of AAC 1 and AAC 2 carbons, their electrochemical properties were also modified by adding different amounts of carbon nanotubes as a good source of mesopores and conducting percolator. Carbon nanotubes (purity N97% C) were purchased from Sigma Aldrich (773840) and used as received. They were simply mixed with activated carbon, binder and carbon black. The composition of one pellet was 10 wt.% of CNTs, 75 wt.% of activated carbon, 10 wt.% of PVdF and 5 wt.% of carbon black. All compounds were mixed together with acetone in order to get homogenous mixture. Acetone was removed by evaporation. Each electrode was prepared in the form of pressed pellets with a geometric surface area of 0.785 cm2 per electrode. All electrode materials were assembled in Swagelok® cell in a symmetric configuration and investigated with different electrolytes, such as 1 mol L−1 KI (or 1 mol L−1 NaI) and 1 mol L−1 VOSO4, using gold current collectors and a glassy fibrous material as a separator. For comparison some experiments have been performed with stainless steel current collector. After examination of their electrochemical properties, those two electrolytes were combined in one system, where potassium iodide played a role of electrolyte for the positive electrode and vanadyl sulfate for the negative one. To preserve good condition of their separation and avoid mixing of them, Nafion® 117 membrane was used. The capacitance properties and other supercapacitor characteristics (expressed per active mass of a single electrode) were studied by applying different electrochemical techniques, i.e., galvanostatic cycling at current densities from 0.2 to 50 A g−1, cyclic voltammetry at voltage scan rates from 1 to 100 mV s−1 and electrochemical impedance spectroscopy in a frequency range from 1 mHz to 100 kHz using VMP3/Z Biologic, France. The modeling of the ions was performed with Gaussian® 03W software. An ionic behavior of iodine/ iodide species was also investigated by in-situ Raman spectroscopy, using DXR dispersive Raman microscopy (Thermofisher, USA) with 533 nm wavelength and 5 mW laser power. Fluorescence of the sample was linearly corrected with the background and prior to white light line.
3. Results and discussion Fig. 1 shows nitrogen adsorption isotherms for all examined activated carbon materials. One can observe that the most different one is AAC 0 characterized by rather poorly developed microporosity, hence, the value of BET specific surface area is not high enough (only 718 m2 g−1) to satisfy electrochemical capacitor requirements. Therefore, this carbon has been chosen for activation by potassium hydroxide. As a consequence, activated carbon (AAC 2) with very promising characteristics was obtained and dedicated for supercapacitor applications. Nitrogen adsorption profile at very low relative pressure revealed microporous character of this material, however, there is also a significant contribution of mesopores. It is obvious that both kinds of those pores are very important in charge storage mechanism, because ions are accumulated in micropores during charging process and to ensure good access to them, the presence of mesopores is necessary. By applying mentioned activation step, we were able to develop specific surface area up to 1964 m2 g−1, which is incomparably higher value than for AAC 0. In Fig. 1 one can observe that AAC 1 and AAC 3 samples demonstrate completely different porous structure than the AAC 0 and AAC 2. Whereas, the shape of their isotherms was quite similar to each other and the values of specific surface area were 1843 m2 g−1 and 2231 m2 g−1, respectively. Even if they have very well developed specific surface area it is worth noting that their applicability as the electrode material in proposed supercapacitor might be limited by the small contribution of mesopores. Lack of mesopores reduces efficiency in exploitation of such high amount of micropores to accumulate the charge by ion attraction. Insufficient quantity of small mesopores (2–5 nm) might cause difficulties in ion migration from electrolyte bulk to micropores, especially when considering very fast processes of charging–discharging, necessary to get very huge peak of power. It is important to underline that electrodes prepared from carbons have definitively lower porosity than pure carbons. For example the specific surface area of pellets dropped for the best carbon AAC 2 from 1964 m2 g− 1 to 1597 m 2 g− 1 whereas total micropore volume decreased from 1.344 cm3 g− 1 to 1.039 cm3 g − 1. On the other hand, microporous volume diminished from 0.873 cm3 g−1 to 0.7158 cm3 g−1 and mesoporous volume from 0.395 cm3 g−1 to 0.2687 cm3 g−1, respectively. Some pores are surely blocked by a binder. For this reason a small additive of carbon nanotubes (10% by mass) to electrode was very beneficial. It preserved mesoporosity of electrode (0.31 cm3 g−1) and additionally greatly enhanced conductivity. By using sodium/potassium iodide solution and vanadium species as the electrolytes in supercapacitor one needs to keep in mind that there is diversity of their possible oxidation states. All possible ions have
Fig. 1. Nitrogen adsorption isotherms (77 K) for different activated carbon materials.
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
different dimensions particularly in the solvated state, hence, some amount of mesopores is indispensable for superior capacitor performance. Fig. 2 presents Pourbaix diagram with various reactions of iodine that might occur in different pH and potential ranges. It can be easily noticed that in the case of 1 mol L−1 NaI and 2 mol L−1 KI aqueous solutions characterized by neutral pH one should consider mainly four possibilities of iodine forms i.e., I−, I2, I− 3 and eventually IO− 3 . Therefore, it is obvious that by using those solutions a few possible redox reactions are introduced to the system, giving additional charge and enhancing capacitance value. Moreover, one can use this information for more careful carbon material selection; calculating the shape and diameter of iodide ions we might easily select the electrode material in order to match the size of ions to pore distribution of the material. It is well known that iodides and iodine have a tendency to form a rich family of polyiodides, − e.g., I− 3 , I5 and even higher I2 content [25]. Our results from modeling simulation by using DFT and MP2 methods are given in Fig. 3. The − diameters of I−, I− 3 and I5 are 3.9 Å, 5.25 Å, and 6.3 Å, respectively. Additionally, I− 5 molecule is asymmetric and L-shaped [25]. Obviously, they need to be considered as the solvated ions, thus their dimensions will be higher than for non-solvated ions, depending on water molecules which surround them. Usually, there is ca. 4–6 water molecules around polyiodide ions, hence, their effective diameter will be enlarged to max. 13 Å. It has to be noticed, that among all halides, iodide ions (including polyiodides) are rather weakly solvated, mainly due to entropic effects in the solution and stronger solvent–solvent interactions. Free energy profiles of non-solvated iodide ions reach a maximum value of 189 kJ mol−1 at 2.5 Å starting from their mass center, indicating that they prefer to be non-solvated. Additionally, the values for counter ions (Na+, K+) mentioned in the study, are of 308 and 374 kJ mol−1, respectively. Enthalpy component of free energy (ca. 80 kJ mol−1) starts to be negligible at 1.5 nm (from mass center), hence, solvent–solvent interactions are almost unaffected by solvation/desolvation effects. On the other hand, both solvated and non-solvated ions match the pore size of the material (AAC 2), whereas they do not fit perfectly to the microporous character of AAC 1. Pores of carbon AAC 2 are centered
Fig. 2. Potential versus pH for iodine/H2O system in aqueous solution 25 °C.
63
− Fig. 3. Ion shape of likely iodide forms, I− (A), I− 3 (B) and I5 (C).
at 1.0 nm, 1.4 nm and 3.4 nm, especially mesopores are well pronounced. Therefore, a need for a suitable porosity of the electrode with respect to ion size has been proved. Vanadium/vanadyl sulfate solution demonstrates also a great variety of possible redox reactions, hence, using it seems to be very promising. Additionally, it needs to be mentioned that sulfate anion is the strongest solvated inorganic anion (can be solvated even by 40 molecules of water) which affects its dimension and electrochemical stability. Desolvation energy is about 108 kJ mol− 1 per one SO24 −\H2O bond [26]. All those factors and properties of investigated electrolytes should be taken into account prior to the electrode material selection. The first part of our electrochemical research was devoted to examination of iodide solutions as the electrolyte in electrochemical capacitor system. In Fig. 4 one can see a comparison of cyclic voltammograms
Fig. 4. Cyclic voltammograms for different activated carbon materials operating in 2 mol L−1 NaI aqueous solution with Au current collector, scan rate 10 mV s−1.
64
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
with current values recalculated to capacitance, recorded for different carbon materials used as the electrodes and operating in 2 mol L− 1 sodium iodide aqueous solution. Activated carbon AAC 0 was not considered due to its poor textural properties in comparison to the other carbons, mainly due to its low specific surface area. Further information on electrochemical behavior of this material and an optimal sodium iodide concentration can be found elsewhere [23]. It can be seen that the best electrochemical performance exhibits only one carbon, i.e., AAC 2. It might be explained by its well developed mesoporous character and high value of specific surface area. In view of the ion size and their high concentration, in other carbons the ions might not have enough space to freely penetrate the pores without any hindrances as it is in the case of AAC 2. Slightly aggravated shape of cyclic voltammogram with a lower capacitance value was observed for AAC 1 which stands in a good agreement with data received from nitrogen adsorption isotherms (moderate total pore volume 0.988 cm3 g− 1). In the case of carbon tissue AAC 3, as predicted, it did not show satisfying results because of lack of mesopores (only 0.05 cm3 g− 1 ). The shape of its curve and obtained capacitance value were very similar to those for AAC 1. It is necessary to mention that the results presented in Figs. 4 and 5 have been obtained with golden current collectors. It was an interesting observation that the shape of voltammograms is determined by the kind of collector, especially at the low scan rate where redox reactions are manifested. Besides sodium iodide solution, some part of this work was also focused on potassium iodide solution with its high capacitance value (Fig. 5) and electrochemical window stability (Fig. 6). It can be found that the 1 mol L− 1 KI solution gives the profitable electrochemical response. It is worth noting that apart from sodium and potassium iodides, an electrochemical behavior of different alkali metal cations (i.e., Li+, Rb+ and Cs+) was also studied and discussed [22]. A quite regular CV shape and almost the same capacitive values of all carbon materials in 1 mol L−1 KI electrolytic solution as for 2 mol L−1 sodium iodide solution (Figs. 4, 5) have been measured. Additionally, there is an important issue on possibility to extend the voltage of capacitor operating with iodide solutions even to 1.4 V (Fig. 6). It is observed that the maximum value which can be reached significantly exceeds the voltage of water decomposition, thermodynamically limited to 1.23 V. In the case of capacitor operating in iodide solution and using stainless steel current collector, it is possible to extend the voltage to 1.4 V without any evidence of the electrolyte decomposition. A three-electrode investigation demonstrated that positive electrode operates in a very narrow potential range (according to Pourbaix diagram) whereas the negative electrode works in a wider potential
range. The positive electrode approaching oxygen evolution potential exploits the charge of I−/I2 redox system but the electrolyte is not being decomposed. The increase of capacitance value achieved at higher voltages might be related to forced I2 dissolution (higher p I2) and enrichment of the electrolyte in polyiodide species, i.e., (n·I2)·I−. It should be noticed that to receive such a high value of pressure it is necessary to keep the system tightly closed, because it will prevent the iodine gas from going out and keep the partial pressure of evolved iodine as high as possible. If we do not maintain these conditions, generated iodine will not be able to react with iodide ions forming I − 3 , which might keep the reversibility of the process. It means that electrolyte will start decomposing and iodine will be gradually removed from the electrolyte. Hence, both the efficiency and cyclability of this system would be very poor and the good electrochemical behavior would be preserved only during first cycles. A good activity of positive electrode in the range of voltage from 0 to 0.2 V has been well visible for stainless steel collectors (Fig. 6) but not in the case of gold current collectors. It appears that gold becomes corrosive in the presence of iodides especially under polarization (confirmed by visual observations). Even if iodide solution has been partly investigated for capacitor performance, measurements of an ionic behavior of iodide species at electrode–electrolyte interface and carbon/iodide interactions using in-situ Raman spectroscopy (Fig. 7) are the original ones.
Fig. 5. Cyclic voltammograms for different activated carbon materials operating in 1 mol L−1 KI aqueous solution, Au current collector, scan rate 10 mV s−1.
Fig. 7. Raman spectra recorded in-situ for AAC 3 electrode at various polarization time.
Fig. 6. Cyclic voltammograms for AAC 2 activated carbon material operating in 1 mol L−1 KI, steel current collector, scan rate 5 mV s−1.
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
Raman spectroscopy profiles were recorded for AAC 3 electrode (tissue, 0.283 cm2) to eliminate binder as an electrode component. These spectra confirm that iodide I− is firstly oxidized to iodine I2 demonstrated by broad bands in 190–170 cm−1 region. Dissolved iodine reacts with iodide ions resulting in I− 3 species, giving a Raman response as I2·I− complex bands at 168 cm−1. After 1 h and 6 h of polarization at a constant potential of 750 mV vs SCE this signal is still growing. It should be mentioned that the electrochemical cell for in situ Raman investigations has a different construction and reference electrode is placed not very close to working electrode, hence, higher values of polarization are needed. Additionally the cell is not perfectly closed. It is assumed that the concentration of iodine increases and allows to recombine with I− and I− 3 , resulting in detection of the signals for higher − polyiodides (I2·I− n ) [25]. For example, I5 complex formed after 1 h of the polarization starts to appear as the signal recorded at 108 cm−1. Moreover, the signal attributed to I− 3 decreases in time, hence, one can assume, that further recombination towards higher polyiodides occurs. Apart from typical redox activity of iodide species, formation of C\I bonds on the carbon electrode surface cannot be excluded. A response at 368–270 cm−1 might be attributed not only to pure iodine, but also − −1 and C\I− to newly created C\I− n bonds, as C\I3 at 333 cm 5 at 327 cm− 1. Their presence might be also confirmed by the signals at 165 and 115 cm−1, but strong Raman activity of free polyiodides in the same region interferes with possible C\I bands. Complexity of iodine/iodide species and their mutual reactions as well as interactions with carbon causes difficulties in an interpretation of Raman spectra but provides fundamental information on iodide/carbon interface activity. Such a good capacitance performance of iodide solutions is a consequence of the presence of iodide species that can easily react and change their oxidation states. Due to this fact some faradaic reactions start to take place and provide the additional charge that significantly contributes to the total capacitance value. These phenomena were described in more detail by applying cyclic voltammetry technique in the three-electrode configuration. This method allowed to discriminate the capacitance that originates from redox reactions and typical double layer charging/discharging. It has been proved that enormous capacitance comes only from the positive electrode, giving capacitance that can reach even 1000 F g−1 [21]. Regrettably this remarkable value cannot be fully utilized because of the negative electrode that is able to provide much lower capacitance due to its electrostatic character only. Therefore, the total capacitance of the system is strongly reduced. That detrimental factor can be avoided by combining positive electrode with different one — characterized by a comparable capacitance. Therefore, it was valuable to conduct a research in completely different electrolytes which, according to Pourbaix diagram, should reveal a high redox activity in more negative potential ranges. As mentioned earlier, a number of possible reactions of vanadium species that might take place is higher than for iodide solution. As it was already explained, even if the electrochemical performance of 1 mol L−1 vanadium sulfate is not as good as for iodide solutions, its application as the electrolyte for the negative electrode seems to be reasonable. It was proved in a three-electrode investigation, that there are some faradaic reactions on both electrodes (Fig. 8). Due to this reason, a conjunction of two presented electrolytes as a redox couple in one supercapacitor has been successfully proposed by our group in previous paper [24]. One can find there description of these redox couple phenomena with the proposed reactions that can take place on the electrodes. In this paper we would like to present further research of this field that is mostly devoted to the electrochemical behavior of these electrolytes with various carbon materials, the effect of carbon nanotubes as additive to carbon electrode and the role of separation of both redox couples. From the results obtained in symmetric configurations, we decided to choose only two activated carbons that revealed the best
65
Fig. 8. Cyclic voltammograms for AAC 2 activated carbon material operating in 1 mol L−1 VOSO4. Three-electrode experiment at scan rate 5 mV s−1.
performance, i.e., AAC 1 and AAC 2. The results from cyclic voltammetry at different scan rates (1 and 10 mV s− 1) are presented in Fig. 9. In the case of both carbons we can distinguish well defined reversible oxidation/reduction peaks at the slow scan rate. It means that we introduced the additional redox reactions on the positive electrode, as well as on the negative one. However, higher capacitance is preserved for the capacitor with AAC 2 electrode, i.e., 685 F g− 1 at 1 mV s− 1 and 283 F g− 1 at 10 mV s− 1. For AAC 1 it was a little bit
Fig. 9. Cyclic voltammograms for AAC 1 and AAC 2 activated carbon materials operating in iodide/vanadium conjugated redox couples as electrolyte solutions at two scan rates.
66
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
lower in that range of scan rate: 521 F g− 1 and 273 F g− 1, respectively. Some discrepancy appeared at the highest scan rate because at 20 mV s− 1 the capacitance for AAC 1 was 197 F g− 1 and only 178 F g− 1 for AAC 2 (not in the figure). Moreover, a remarkable drop of capacitance with a scan rate increase is a good evidence for faradaic origin of mentioned peaks. It is obvious that the faster the scan rate, the smaller contribution of redox reactions can be observed due to a slow diffusion. They do not have sufficient time to occur, therefore, this system starts to resemble a typical battery behavior. One of the solutions that could help to resolve this problem is based on an increase of a mesoporous structure, since it is crucial in fast ion transportation. Carbon nanotubes (CNTs) added to electrode mass can serve as a good percolator and a source of mesopores. Fig. 10 shows how carbon nanotubes significantly contribute to a capacitance enhancement. However, a quantity added plays a crucial role in obtaining the satisfying results, therefore, an optimization was carried out in order to determine the appropriate amount of this additive. It was found that 10 wt.% of CNTs is the most profitable. Exceeding this CNT content, e.g., adding 20%, a typical mesoporous character of the nanotubes and their low capacitance (20 F/g) preserve good charge propagation but aggravate capacitance values. It can be found that nanotubes improved the capacitor performance in higher current density regimes and allow to discharge the capacitor with a much higher power peak. The addition of CNTs changed the shape of cyclic voltammograms for both carbons and made them very similar to each other. Oxidation and reduction peaks of redox couples
Fig. 10. Cyclic voltammograms for AAC 1 and AAC 2 activated carbon materials operating in iodide/vanadium conjugated redox couples as electrolyte solutions with addition of 10% of carbon nanotubes, at two scan rates.
Fig. 11. Leakage current and self-discharge profile for AAC 2 carbon separated by different membranes.
became well-defined. The difference between CVs before and after CNT addition is much more visible in the case of AAC 1, owing to the fact that this material was characterized by the significantly smaller amount of mesopores than AAC 2. An interesting behavior is presented in Fig. 11. The self-discharge profiles (without polarization) during 24 h demonstrate how it is important to use an appropriate membrane to separate anolyte from catholyte. By using a regular glass microfiber filter (Whatman®) it is almost impossible to hold the voltage for a long time. After 10 h it decreases to 0.15 V, which means that the capacitor is completely discharged, since electrolytes start to diffuse through the separator and finally mix together. To avoid this detrimental effect, it is mandatory to use the membrane which is capable of separating those two electrolytes. For this purpose we used Nafion 117 membrane. An application of the membrane prevents from mixing the electrolytes which resulted in the better self-discharge profile. After 10 h 50% of initial voltage was preserved. The beneficial effect of separation is also well observed during a long-term cycling (Fig. 12). After 5000 cycles or longer (up to 10,000 cycles) capacitance remains almost the same as at the beginning (ca. 500 F g− 1 ) in a cell with Nafion 117 separation. Therefore, we can assume that both electrodes operate in different redox-active electrolytes and without their decomposition. Two methods were used for long term testing. Cyclic voltammetry performed at 5 mV s− 1 scan rate
Fig. 12. Capacitance vs cycle number for AAC 2 carbon separated by different separators, current load 1 A g−1.
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
Fig. 13. Ragone plot for capacitors with different carbons operating in iodide/vanadium conjugated redox couples as electrolyte solutions.
demonstrated better stability of the system during cycling. However, this technique should not be applied to determine a real capacitor life performance. When comparing it with galvanostatic charging/ discharging at a current density of 1 A g−1 it was observed that values from cyclic voltammetry are usually overestimated (ca. 600 F g−1). By applying a constant current load, capacitance is reduced to 500 F g−1 after 5000 cycles. It needs to be mentioned that cyclic voltammetry is a typical method to determine the maximum operational voltage. But if one wants to preserve the same conditions as they are in the supercapacitor during its practical application, it seems to be more reasonable to apply galvanostatic cycling method. It is a more reliable technique to calculate capacitance but at the same time more adapted for cycling tests [27]. In order to determine if this system has applicable prospects, energy density vs power density, i.e., Ragone plot was prepared (Fig. 13). The most promising power/energy dependence was revealed by the capacitor with AAC 2 carbon. The addition of nanotubes slightly affects the capacitor performance giving in both cases energy density of 20 Wh kg−1 with max. power density of 2 kW kg−1. For AAC 1, nanotubes improved the charge propagation, hence the energy is maintained along with the power increase. Only at higher current densities there is a drastic energy drop but the moderate power has been retained. All the redox active electrolytes investigated here have moderate values of equivalent series resistance which never exceeded 0.5 Ω cm2. It is related with the fact that all the electrolytes are based on aqueous solutions that is additional benefit of such systems. 4. Conclusions In this paper we investigated the electrochemical behavior of various activated carbon materials and their potential application as electrodes in supercapacitors with redox active electrolytes. Taking into account relatively large dimensions of ions present in electrolyte, big micropores and some presence of mesopores were essential. In the case of iodide solutions in situ Raman investigation was carried out to show possible iodine species with polyiodides and carbon/iodine interactions under electrode polarization. A narrow potential range of pure iodide activity (0.2 V) was clearly shown whereas a total operating voltage of capacitor could be extended to 1.4 V.
67
It has been proved that apart from the type of electrolyte also the kind of current collector plays an important role. Gold current collector is not adapted for iodide solutions because of its reactivity with iodides whereas stainless steel was a suitable material without corrosion traces after long-term cycling. Additionally to the typical symmetric configurations, where three different redox active electrolytes were used, we decided also to investigate their behavior in the asymmetric system, in terms of electrolytes. It means that we used the conjugated redox couples as the electrolyte solution with different carbon materials. The best performance was obtained by AAC 1 and AAC 2 carbon material in the symmetric system. For enhancement of capacitor work carbon nanotubes (10%) were used as perfectly conducting additive and source of mesopores. Those two carbons modified with the nanotubes were employed in the capacitor operating in iodide/vanadium electrolyte solutions. AAC 2 revealed the highest capacitance value and showed good energy/power performance, i.e., 20 Wh kg−1 with max. power density of 2 kW kg−1. Effect of the nanotube addition was more visible in the case of AAC 1 because this carbon did not exhibit the mesoporous structure as AAC 2. In this work it was also demonstrated that the membrane Nafion 117 plays a crucial role when employing the redox-active couples based on iodine and vanadium species. Acknowledgments This work is supported by INGEC project in the frame of the Swiss–Polish Programme (PSPB-107/2010). The authors would like to acknowledge Polish Ministry of Science and Higher Education, grant no. NN511 474339. References [1] T. Horiba, T. Maeshima, T. Matsumura, M. Koseki, J. Arai, Y. Muranaka, J. Power Sources 146 (2005) 107–110. [2] S.L. Candelaria, Y. Shao, W. Zhou, X. Li, J. Xiao, J.-G. Zhang, Y. Wang, J. Liu, J. Li, G. Cao, Nano Energy 1 (2012) 195–220. [3] P. Chen, G. Shen, Y. Shi, H. Chen, C. Zhou, ACS Nano 4 (2010) 4403–4411. [4] X. Li, B. Wei, Nano Energy 2 (2013) 159–173. [5] Y.W.F. Wang, S. Xiao, Y. Hou, C. Hu, L. Liu, RSC Adv. 3 (2013) 13059–13084. [6] R. Kötz, M. Carlen, Electrochim. Acta 45 (2000) 2483–2498. [7] M. Mastragostino, F. Soavi, J. Power Sources 174 (2007) 89–93. [8] E. Frackowiak, F. Béguin, Carbon 39 (2001) 937–950. [9] M. Toupin, T. Brousse, D. Bélanger, Chem. Mater. 16 (2002) 3946–3952. [10] W. Sugimoto, T. Kizaki, K. Yokoshima, Y. Murakami, Y. Takasu, Electrochim. Acta 49 (2004) 313–320. [11] A. Malak, K. Fic, G. Lota, C. Vix-Guterl, E. Frackowiak, J. Solid State Electrochem. 14 (2009) 811–816. [12] J.Z.G. Wang, L. Zhang, Chem. Soc. Rev. 41 (2012) 797–828. [13] S. Roldán, M. Granda, J. Phys. Chem. C 115 (2011) 17606–17611. [14] Z. Algharaibeh, P.G. Pickup, Electrochem. Commun. 13 (2011) 147–149. [15] A.S. Kumar, S. Sornambikai, P. Gayathri, J.-M. Zen, J. Electroanal. Chem. 641 (2010) 131–135. [16] A.L. Comte, G. Pognon, T. Brousse, D. Bélanger, Electrochemistry 81 (2013) 863–866. [17] S. Roldán, C. Blanco, M. Granda, R. Menéndez, R. Santamaría, Angew. Chem. Int. Ed. 50 (2011) 1699–1701. [18] S. Roldán, Z. González, C. Blanco, M. Granda, R. Menéndez, R. Santamaría, Electrochim. Acta 56 (2011) 3401–3405. [19] S.T. Senthilkumar, R.K. Selvan, Y.S. Lee, J.S. Melo, J. Mater. Chem. A 1 (2013) 1086. [20] S.T. Senthilkumar, R.K. Selvan, N. Ponpandian, J.S. Melo, RSC Adv. 2 (2012) 8937. [21] G. Lota, E. Frackowiak, Electrochem. Commun. 11 (2009) 87–90. [22] G. Lota, K. Fic, E. Frackowiak, Electrochem. Commun. 13 (2011) 38–41. [23] J. Menzel, K. Fic, M. Meller, E. Frackowiak, J. Appl. Electrochem. 44 (2014) 439–445. [24] E. Frackowiak, K. Fic, M. Meller, G. Lota, ChemSusChem 5 (2012) 1181–1185. [25] P.H. Svensson, L. Kloo, Chem. Rev. 103 (2003) 1649–1684. [26] K. Fic, G. Lota, E. Frackowiak, Energy Environ. Sci. 5 (2012) 5842–5850. [27] D. Weingarth, H. Noh, A. Foelske-Schmitz, A. Wokaun, R. Kötz, Electrochim. Acta 103 (2013) 119–124.
Contents lists available at ScienceDirect
Solid State Ionics journal homepage: www.elsevier.com/locate/ssi
Electrochemical capacitors as attractive power sources M. Meller, J. Menzel, K. Fic, D. Gastol, E. Frackowiak ⁎ Institute of Chemistry and Technical Electrochemistry, Poznan University of Technology, Piotrowo 3, 60-965 Poznan, Poland
a r t i c l e
i n f o
Article history: Received 1 April 2014 Received in revised form 6 July 2014 Accepted 18 July 2014 Available online xxxx Keywords: Supercapacitor Pseudocapacitance Iodide Vanadium species
a b s t r a c t The electrochemical performance of various carbon materials as supercapacitor electrodes with redox active electrolytes has been presented. In situ Raman investigation was carried out to show possible iodine species, polyiodides and carbon/iodine interactions during electrode polarization. Apart from the electrolyte also the kind of current collector plays an important role. Gold current collector is not adapted because of its reactivity with iodides whereas stainless steel is convenient. The conjugated iodide and vanadium species as the electrolyte were also investigated with Nafion separation. The best performance was obtained for AAC 1 and AAC 2 carbon material especially after adding 10% of carbon nanotubes. © 2014 Elsevier B.V. All rights reserved.
1. Introduction There is an extremely fast growth in market of all digital and electronic devices that are destined for portable applications. Simultaneously with their resilient development one can observe unrelenting increase of demands for high energy sources. Instead of commonly used lithium-ion (Li-ion) and nickel metal hydride (Ni-MH) batteries, there is possibility to consider electrochemical supercapacitors for this purpose. Unfortunately, there are still some difficulties to fulfill the market demand. It is well known that electrochemical double layer capacitor (EDLC) can serve as an excellent energy storage device for all applications with high power requirements. That fact is caused by the nature of the electrical charge accumulation mechanism, which is based on the electrostatic forces. Hence, it can be charged and discharged in a very short time and release enormously high peak of power [1,2]. The problem is that the energy of this system is incomparably lower than for mentioned batteries and that is the reason why they cannot be replaced so easily by the electrochemical capacitors. Because capacitance value strictly depends on the surface of the electrode/electrolyte interface, the best materials for electrodes should be made from activated carbons due to their well developed specific surface area (up to 2500 m2 g−1). They might be used in different forms, such as powders, woven cloths, felts or fibers [3–6]. Moreover, the surface functionalities of applied carbon material cannot be neglected [7].
⁎ Corresponding author at: Poznan University of Technology, Piotrowo 3, 60-965 Poznan, Poland. E-mail address: [email protected] (E. Frackowiak).
http://dx.doi.org/10.1016/j.ssi.2014.07.014 0167-2738/© 2014 Elsevier B.V. All rights reserved.
In order to achieve higher capacitance values and increase the energy of the electrochemical capacitor, the electrode materials capable to reversible redox reactions might be considered. This phenomenon of capacitance enhancement consists of pseudocapacitance effects which are related to the surface functionalities of carbon and/or to the presence of electroactive oxides from such transition metals as Ru, Ir, W, Mo, Mn, Ni and Co [8–12]. However, these materials have some limitations that do not allow them to be applied efficiently, mainly because of their slow solid state kinetics and, what is also very important, significantly higher price than for the activated carbons, especially in the case of ruthenium and iridium. Another way to increase the energy is using electrolytes that can stand as a source of redox reactions. There are some papers focused on that matter concerning various possibilities, like for example, using additives to electrolytes with oxygen functionalities, e.g. quinone/hydroquinone groups [13–20]. These groups might be able to graft on the electrode surface with generation of new oxygen functional species that can react reversibly during charging/discharging process and provide additional charge [16]. It is also possible to introduce pseudocapacitance from redox active electrolytes in the form of alkali metal iodide aqueous solutions. It has been shown by our group [21–23] that iodides supply enormous capacitance but only at the positive electrode. Our preliminary investigations indicated that a binary electrolyte based on iodides and vanadyl sulfate could be a good option to enhance capacitance of full system [24]. Taking into account all advantages of such binary redox electrolytes we decided in a present paper to investigate in more detail their activity on several activated carbon materials with different porous structure, especially, mesoporosity. A
62
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
comparison will be shown between the various activated carbons operating in the redox active electrolytes based on sodium and potassium iodide as well as vanadyl sulfate in the form of a single electrolyte component or as a redox active pair operating as catholyte and anolyte. The effect of separating such two redox couples as well as the role of nanotubes as an additive to active mass of electrodes will be elucidated. The consequence of using two different current collectors will also be discussed. The detailed study of carbon/iodine species interface by Raman spectroscopy will be presented.
2. Experimental Physicochemical properties, such as specific surface area and pore size distribution as well as oxygen content with type of surface functionalities of all carbon used, were determined by nitrogen adsorption (ASAP 2020 instrument from Micromeritics, USA) and thermogravimetry coupled with mass spectrometry in order to recognize evacuating species (TGA F1 Iris and Aeolos MS instruments by Netzsch, Germany). In some cases electrodes have been also investigated. All electrochemical experiments were made in twoand three electrode capacitor configurations using Swagelok® cell. Electrode materials were prepared from different commercially available activated carbons, i.e. SX2 (AAC 0) and DLC Super 30 Norit (AAC 1), composed of 85 wt.% of carbon, 10 wt.% of polyvinylidene fluoride (PVDF) and 5 wt.% of carbon black. In order to receive carbon material with better developed micro/mesoporosity, carbon SX2 was activated by KOH (1:4) giving carbon AAC 2. Additionally, activated carbon tissue ACC 507-20 (AAC 3) from Kynol was also applied, because of its microporous character and high specific surface area. The physicochemical properties of all utilized materials are described in more detail in the next paragraph. In the case of AAC 1 and AAC 2 carbons, their electrochemical properties were also modified by adding different amounts of carbon nanotubes as a good source of mesopores and conducting percolator. Carbon nanotubes (purity N97% C) were purchased from Sigma Aldrich (773840) and used as received. They were simply mixed with activated carbon, binder and carbon black. The composition of one pellet was 10 wt.% of CNTs, 75 wt.% of activated carbon, 10 wt.% of PVdF and 5 wt.% of carbon black. All compounds were mixed together with acetone in order to get homogenous mixture. Acetone was removed by evaporation. Each electrode was prepared in the form of pressed pellets with a geometric surface area of 0.785 cm2 per electrode. All electrode materials were assembled in Swagelok® cell in a symmetric configuration and investigated with different electrolytes, such as 1 mol L−1 KI (or 1 mol L−1 NaI) and 1 mol L−1 VOSO4, using gold current collectors and a glassy fibrous material as a separator. For comparison some experiments have been performed with stainless steel current collector. After examination of their electrochemical properties, those two electrolytes were combined in one system, where potassium iodide played a role of electrolyte for the positive electrode and vanadyl sulfate for the negative one. To preserve good condition of their separation and avoid mixing of them, Nafion® 117 membrane was used. The capacitance properties and other supercapacitor characteristics (expressed per active mass of a single electrode) were studied by applying different electrochemical techniques, i.e., galvanostatic cycling at current densities from 0.2 to 50 A g−1, cyclic voltammetry at voltage scan rates from 1 to 100 mV s−1 and electrochemical impedance spectroscopy in a frequency range from 1 mHz to 100 kHz using VMP3/Z Biologic, France. The modeling of the ions was performed with Gaussian® 03W software. An ionic behavior of iodine/ iodide species was also investigated by in-situ Raman spectroscopy, using DXR dispersive Raman microscopy (Thermofisher, USA) with 533 nm wavelength and 5 mW laser power. Fluorescence of the sample was linearly corrected with the background and prior to white light line.
3. Results and discussion Fig. 1 shows nitrogen adsorption isotherms for all examined activated carbon materials. One can observe that the most different one is AAC 0 characterized by rather poorly developed microporosity, hence, the value of BET specific surface area is not high enough (only 718 m2 g−1) to satisfy electrochemical capacitor requirements. Therefore, this carbon has been chosen for activation by potassium hydroxide. As a consequence, activated carbon (AAC 2) with very promising characteristics was obtained and dedicated for supercapacitor applications. Nitrogen adsorption profile at very low relative pressure revealed microporous character of this material, however, there is also a significant contribution of mesopores. It is obvious that both kinds of those pores are very important in charge storage mechanism, because ions are accumulated in micropores during charging process and to ensure good access to them, the presence of mesopores is necessary. By applying mentioned activation step, we were able to develop specific surface area up to 1964 m2 g−1, which is incomparably higher value than for AAC 0. In Fig. 1 one can observe that AAC 1 and AAC 3 samples demonstrate completely different porous structure than the AAC 0 and AAC 2. Whereas, the shape of their isotherms was quite similar to each other and the values of specific surface area were 1843 m2 g−1 and 2231 m2 g−1, respectively. Even if they have very well developed specific surface area it is worth noting that their applicability as the electrode material in proposed supercapacitor might be limited by the small contribution of mesopores. Lack of mesopores reduces efficiency in exploitation of such high amount of micropores to accumulate the charge by ion attraction. Insufficient quantity of small mesopores (2–5 nm) might cause difficulties in ion migration from electrolyte bulk to micropores, especially when considering very fast processes of charging–discharging, necessary to get very huge peak of power. It is important to underline that electrodes prepared from carbons have definitively lower porosity than pure carbons. For example the specific surface area of pellets dropped for the best carbon AAC 2 from 1964 m2 g− 1 to 1597 m 2 g− 1 whereas total micropore volume decreased from 1.344 cm3 g− 1 to 1.039 cm3 g − 1. On the other hand, microporous volume diminished from 0.873 cm3 g−1 to 0.7158 cm3 g−1 and mesoporous volume from 0.395 cm3 g−1 to 0.2687 cm3 g−1, respectively. Some pores are surely blocked by a binder. For this reason a small additive of carbon nanotubes (10% by mass) to electrode was very beneficial. It preserved mesoporosity of electrode (0.31 cm3 g−1) and additionally greatly enhanced conductivity. By using sodium/potassium iodide solution and vanadium species as the electrolytes in supercapacitor one needs to keep in mind that there is diversity of their possible oxidation states. All possible ions have
Fig. 1. Nitrogen adsorption isotherms (77 K) for different activated carbon materials.
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
different dimensions particularly in the solvated state, hence, some amount of mesopores is indispensable for superior capacitor performance. Fig. 2 presents Pourbaix diagram with various reactions of iodine that might occur in different pH and potential ranges. It can be easily noticed that in the case of 1 mol L−1 NaI and 2 mol L−1 KI aqueous solutions characterized by neutral pH one should consider mainly four possibilities of iodine forms i.e., I−, I2, I− 3 and eventually IO− 3 . Therefore, it is obvious that by using those solutions a few possible redox reactions are introduced to the system, giving additional charge and enhancing capacitance value. Moreover, one can use this information for more careful carbon material selection; calculating the shape and diameter of iodide ions we might easily select the electrode material in order to match the size of ions to pore distribution of the material. It is well known that iodides and iodine have a tendency to form a rich family of polyiodides, − e.g., I− 3 , I5 and even higher I2 content [25]. Our results from modeling simulation by using DFT and MP2 methods are given in Fig. 3. The − diameters of I−, I− 3 and I5 are 3.9 Å, 5.25 Å, and 6.3 Å, respectively. Additionally, I− 5 molecule is asymmetric and L-shaped [25]. Obviously, they need to be considered as the solvated ions, thus their dimensions will be higher than for non-solvated ions, depending on water molecules which surround them. Usually, there is ca. 4–6 water molecules around polyiodide ions, hence, their effective diameter will be enlarged to max. 13 Å. It has to be noticed, that among all halides, iodide ions (including polyiodides) are rather weakly solvated, mainly due to entropic effects in the solution and stronger solvent–solvent interactions. Free energy profiles of non-solvated iodide ions reach a maximum value of 189 kJ mol−1 at 2.5 Å starting from their mass center, indicating that they prefer to be non-solvated. Additionally, the values for counter ions (Na+, K+) mentioned in the study, are of 308 and 374 kJ mol−1, respectively. Enthalpy component of free energy (ca. 80 kJ mol−1) starts to be negligible at 1.5 nm (from mass center), hence, solvent–solvent interactions are almost unaffected by solvation/desolvation effects. On the other hand, both solvated and non-solvated ions match the pore size of the material (AAC 2), whereas they do not fit perfectly to the microporous character of AAC 1. Pores of carbon AAC 2 are centered
Fig. 2. Potential versus pH for iodine/H2O system in aqueous solution 25 °C.
63
− Fig. 3. Ion shape of likely iodide forms, I− (A), I− 3 (B) and I5 (C).
at 1.0 nm, 1.4 nm and 3.4 nm, especially mesopores are well pronounced. Therefore, a need for a suitable porosity of the electrode with respect to ion size has been proved. Vanadium/vanadyl sulfate solution demonstrates also a great variety of possible redox reactions, hence, using it seems to be very promising. Additionally, it needs to be mentioned that sulfate anion is the strongest solvated inorganic anion (can be solvated even by 40 molecules of water) which affects its dimension and electrochemical stability. Desolvation energy is about 108 kJ mol− 1 per one SO24 −\H2O bond [26]. All those factors and properties of investigated electrolytes should be taken into account prior to the electrode material selection. The first part of our electrochemical research was devoted to examination of iodide solutions as the electrolyte in electrochemical capacitor system. In Fig. 4 one can see a comparison of cyclic voltammograms
Fig. 4. Cyclic voltammograms for different activated carbon materials operating in 2 mol L−1 NaI aqueous solution with Au current collector, scan rate 10 mV s−1.
64
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
with current values recalculated to capacitance, recorded for different carbon materials used as the electrodes and operating in 2 mol L− 1 sodium iodide aqueous solution. Activated carbon AAC 0 was not considered due to its poor textural properties in comparison to the other carbons, mainly due to its low specific surface area. Further information on electrochemical behavior of this material and an optimal sodium iodide concentration can be found elsewhere [23]. It can be seen that the best electrochemical performance exhibits only one carbon, i.e., AAC 2. It might be explained by its well developed mesoporous character and high value of specific surface area. In view of the ion size and their high concentration, in other carbons the ions might not have enough space to freely penetrate the pores without any hindrances as it is in the case of AAC 2. Slightly aggravated shape of cyclic voltammogram with a lower capacitance value was observed for AAC 1 which stands in a good agreement with data received from nitrogen adsorption isotherms (moderate total pore volume 0.988 cm3 g− 1). In the case of carbon tissue AAC 3, as predicted, it did not show satisfying results because of lack of mesopores (only 0.05 cm3 g− 1 ). The shape of its curve and obtained capacitance value were very similar to those for AAC 1. It is necessary to mention that the results presented in Figs. 4 and 5 have been obtained with golden current collectors. It was an interesting observation that the shape of voltammograms is determined by the kind of collector, especially at the low scan rate where redox reactions are manifested. Besides sodium iodide solution, some part of this work was also focused on potassium iodide solution with its high capacitance value (Fig. 5) and electrochemical window stability (Fig. 6). It can be found that the 1 mol L− 1 KI solution gives the profitable electrochemical response. It is worth noting that apart from sodium and potassium iodides, an electrochemical behavior of different alkali metal cations (i.e., Li+, Rb+ and Cs+) was also studied and discussed [22]. A quite regular CV shape and almost the same capacitive values of all carbon materials in 1 mol L−1 KI electrolytic solution as for 2 mol L−1 sodium iodide solution (Figs. 4, 5) have been measured. Additionally, there is an important issue on possibility to extend the voltage of capacitor operating with iodide solutions even to 1.4 V (Fig. 6). It is observed that the maximum value which can be reached significantly exceeds the voltage of water decomposition, thermodynamically limited to 1.23 V. In the case of capacitor operating in iodide solution and using stainless steel current collector, it is possible to extend the voltage to 1.4 V without any evidence of the electrolyte decomposition. A three-electrode investigation demonstrated that positive electrode operates in a very narrow potential range (according to Pourbaix diagram) whereas the negative electrode works in a wider potential
range. The positive electrode approaching oxygen evolution potential exploits the charge of I−/I2 redox system but the electrolyte is not being decomposed. The increase of capacitance value achieved at higher voltages might be related to forced I2 dissolution (higher p I2) and enrichment of the electrolyte in polyiodide species, i.e., (n·I2)·I−. It should be noticed that to receive such a high value of pressure it is necessary to keep the system tightly closed, because it will prevent the iodine gas from going out and keep the partial pressure of evolved iodine as high as possible. If we do not maintain these conditions, generated iodine will not be able to react with iodide ions forming I − 3 , which might keep the reversibility of the process. It means that electrolyte will start decomposing and iodine will be gradually removed from the electrolyte. Hence, both the efficiency and cyclability of this system would be very poor and the good electrochemical behavior would be preserved only during first cycles. A good activity of positive electrode in the range of voltage from 0 to 0.2 V has been well visible for stainless steel collectors (Fig. 6) but not in the case of gold current collectors. It appears that gold becomes corrosive in the presence of iodides especially under polarization (confirmed by visual observations). Even if iodide solution has been partly investigated for capacitor performance, measurements of an ionic behavior of iodide species at electrode–electrolyte interface and carbon/iodide interactions using in-situ Raman spectroscopy (Fig. 7) are the original ones.
Fig. 5. Cyclic voltammograms for different activated carbon materials operating in 1 mol L−1 KI aqueous solution, Au current collector, scan rate 10 mV s−1.
Fig. 7. Raman spectra recorded in-situ for AAC 3 electrode at various polarization time.
Fig. 6. Cyclic voltammograms for AAC 2 activated carbon material operating in 1 mol L−1 KI, steel current collector, scan rate 5 mV s−1.
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
Raman spectroscopy profiles were recorded for AAC 3 electrode (tissue, 0.283 cm2) to eliminate binder as an electrode component. These spectra confirm that iodide I− is firstly oxidized to iodine I2 demonstrated by broad bands in 190–170 cm−1 region. Dissolved iodine reacts with iodide ions resulting in I− 3 species, giving a Raman response as I2·I− complex bands at 168 cm−1. After 1 h and 6 h of polarization at a constant potential of 750 mV vs SCE this signal is still growing. It should be mentioned that the electrochemical cell for in situ Raman investigations has a different construction and reference electrode is placed not very close to working electrode, hence, higher values of polarization are needed. Additionally the cell is not perfectly closed. It is assumed that the concentration of iodine increases and allows to recombine with I− and I− 3 , resulting in detection of the signals for higher − polyiodides (I2·I− n ) [25]. For example, I5 complex formed after 1 h of the polarization starts to appear as the signal recorded at 108 cm−1. Moreover, the signal attributed to I− 3 decreases in time, hence, one can assume, that further recombination towards higher polyiodides occurs. Apart from typical redox activity of iodide species, formation of C\I bonds on the carbon electrode surface cannot be excluded. A response at 368–270 cm−1 might be attributed not only to pure iodine, but also − −1 and C\I− to newly created C\I− n bonds, as C\I3 at 333 cm 5 at 327 cm− 1. Their presence might be also confirmed by the signals at 165 and 115 cm−1, but strong Raman activity of free polyiodides in the same region interferes with possible C\I bands. Complexity of iodine/iodide species and their mutual reactions as well as interactions with carbon causes difficulties in an interpretation of Raman spectra but provides fundamental information on iodide/carbon interface activity. Such a good capacitance performance of iodide solutions is a consequence of the presence of iodide species that can easily react and change their oxidation states. Due to this fact some faradaic reactions start to take place and provide the additional charge that significantly contributes to the total capacitance value. These phenomena were described in more detail by applying cyclic voltammetry technique in the three-electrode configuration. This method allowed to discriminate the capacitance that originates from redox reactions and typical double layer charging/discharging. It has been proved that enormous capacitance comes only from the positive electrode, giving capacitance that can reach even 1000 F g−1 [21]. Regrettably this remarkable value cannot be fully utilized because of the negative electrode that is able to provide much lower capacitance due to its electrostatic character only. Therefore, the total capacitance of the system is strongly reduced. That detrimental factor can be avoided by combining positive electrode with different one — characterized by a comparable capacitance. Therefore, it was valuable to conduct a research in completely different electrolytes which, according to Pourbaix diagram, should reveal a high redox activity in more negative potential ranges. As mentioned earlier, a number of possible reactions of vanadium species that might take place is higher than for iodide solution. As it was already explained, even if the electrochemical performance of 1 mol L−1 vanadium sulfate is not as good as for iodide solutions, its application as the electrolyte for the negative electrode seems to be reasonable. It was proved in a three-electrode investigation, that there are some faradaic reactions on both electrodes (Fig. 8). Due to this reason, a conjunction of two presented electrolytes as a redox couple in one supercapacitor has been successfully proposed by our group in previous paper [24]. One can find there description of these redox couple phenomena with the proposed reactions that can take place on the electrodes. In this paper we would like to present further research of this field that is mostly devoted to the electrochemical behavior of these electrolytes with various carbon materials, the effect of carbon nanotubes as additive to carbon electrode and the role of separation of both redox couples. From the results obtained in symmetric configurations, we decided to choose only two activated carbons that revealed the best
65
Fig. 8. Cyclic voltammograms for AAC 2 activated carbon material operating in 1 mol L−1 VOSO4. Three-electrode experiment at scan rate 5 mV s−1.
performance, i.e., AAC 1 and AAC 2. The results from cyclic voltammetry at different scan rates (1 and 10 mV s− 1) are presented in Fig. 9. In the case of both carbons we can distinguish well defined reversible oxidation/reduction peaks at the slow scan rate. It means that we introduced the additional redox reactions on the positive electrode, as well as on the negative one. However, higher capacitance is preserved for the capacitor with AAC 2 electrode, i.e., 685 F g− 1 at 1 mV s− 1 and 283 F g− 1 at 10 mV s− 1. For AAC 1 it was a little bit
Fig. 9. Cyclic voltammograms for AAC 1 and AAC 2 activated carbon materials operating in iodide/vanadium conjugated redox couples as electrolyte solutions at two scan rates.
66
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
lower in that range of scan rate: 521 F g− 1 and 273 F g− 1, respectively. Some discrepancy appeared at the highest scan rate because at 20 mV s− 1 the capacitance for AAC 1 was 197 F g− 1 and only 178 F g− 1 for AAC 2 (not in the figure). Moreover, a remarkable drop of capacitance with a scan rate increase is a good evidence for faradaic origin of mentioned peaks. It is obvious that the faster the scan rate, the smaller contribution of redox reactions can be observed due to a slow diffusion. They do not have sufficient time to occur, therefore, this system starts to resemble a typical battery behavior. One of the solutions that could help to resolve this problem is based on an increase of a mesoporous structure, since it is crucial in fast ion transportation. Carbon nanotubes (CNTs) added to electrode mass can serve as a good percolator and a source of mesopores. Fig. 10 shows how carbon nanotubes significantly contribute to a capacitance enhancement. However, a quantity added plays a crucial role in obtaining the satisfying results, therefore, an optimization was carried out in order to determine the appropriate amount of this additive. It was found that 10 wt.% of CNTs is the most profitable. Exceeding this CNT content, e.g., adding 20%, a typical mesoporous character of the nanotubes and their low capacitance (20 F/g) preserve good charge propagation but aggravate capacitance values. It can be found that nanotubes improved the capacitor performance in higher current density regimes and allow to discharge the capacitor with a much higher power peak. The addition of CNTs changed the shape of cyclic voltammograms for both carbons and made them very similar to each other. Oxidation and reduction peaks of redox couples
Fig. 10. Cyclic voltammograms for AAC 1 and AAC 2 activated carbon materials operating in iodide/vanadium conjugated redox couples as electrolyte solutions with addition of 10% of carbon nanotubes, at two scan rates.
Fig. 11. Leakage current and self-discharge profile for AAC 2 carbon separated by different membranes.
became well-defined. The difference between CVs before and after CNT addition is much more visible in the case of AAC 1, owing to the fact that this material was characterized by the significantly smaller amount of mesopores than AAC 2. An interesting behavior is presented in Fig. 11. The self-discharge profiles (without polarization) during 24 h demonstrate how it is important to use an appropriate membrane to separate anolyte from catholyte. By using a regular glass microfiber filter (Whatman®) it is almost impossible to hold the voltage for a long time. After 10 h it decreases to 0.15 V, which means that the capacitor is completely discharged, since electrolytes start to diffuse through the separator and finally mix together. To avoid this detrimental effect, it is mandatory to use the membrane which is capable of separating those two electrolytes. For this purpose we used Nafion 117 membrane. An application of the membrane prevents from mixing the electrolytes which resulted in the better self-discharge profile. After 10 h 50% of initial voltage was preserved. The beneficial effect of separation is also well observed during a long-term cycling (Fig. 12). After 5000 cycles or longer (up to 10,000 cycles) capacitance remains almost the same as at the beginning (ca. 500 F g− 1 ) in a cell with Nafion 117 separation. Therefore, we can assume that both electrodes operate in different redox-active electrolytes and without their decomposition. Two methods were used for long term testing. Cyclic voltammetry performed at 5 mV s− 1 scan rate
Fig. 12. Capacitance vs cycle number for AAC 2 carbon separated by different separators, current load 1 A g−1.
M. Meller et al. / Solid State Ionics 265 (2014) 61–67
Fig. 13. Ragone plot for capacitors with different carbons operating in iodide/vanadium conjugated redox couples as electrolyte solutions.
demonstrated better stability of the system during cycling. However, this technique should not be applied to determine a real capacitor life performance. When comparing it with galvanostatic charging/ discharging at a current density of 1 A g−1 it was observed that values from cyclic voltammetry are usually overestimated (ca. 600 F g−1). By applying a constant current load, capacitance is reduced to 500 F g−1 after 5000 cycles. It needs to be mentioned that cyclic voltammetry is a typical method to determine the maximum operational voltage. But if one wants to preserve the same conditions as they are in the supercapacitor during its practical application, it seems to be more reasonable to apply galvanostatic cycling method. It is a more reliable technique to calculate capacitance but at the same time more adapted for cycling tests [27]. In order to determine if this system has applicable prospects, energy density vs power density, i.e., Ragone plot was prepared (Fig. 13). The most promising power/energy dependence was revealed by the capacitor with AAC 2 carbon. The addition of nanotubes slightly affects the capacitor performance giving in both cases energy density of 20 Wh kg−1 with max. power density of 2 kW kg−1. For AAC 1, nanotubes improved the charge propagation, hence the energy is maintained along with the power increase. Only at higher current densities there is a drastic energy drop but the moderate power has been retained. All the redox active electrolytes investigated here have moderate values of equivalent series resistance which never exceeded 0.5 Ω cm2. It is related with the fact that all the electrolytes are based on aqueous solutions that is additional benefit of such systems. 4. Conclusions In this paper we investigated the electrochemical behavior of various activated carbon materials and their potential application as electrodes in supercapacitors with redox active electrolytes. Taking into account relatively large dimensions of ions present in electrolyte, big micropores and some presence of mesopores were essential. In the case of iodide solutions in situ Raman investigation was carried out to show possible iodine species with polyiodides and carbon/iodine interactions under electrode polarization. A narrow potential range of pure iodide activity (0.2 V) was clearly shown whereas a total operating voltage of capacitor could be extended to 1.4 V.
67
It has been proved that apart from the type of electrolyte also the kind of current collector plays an important role. Gold current collector is not adapted for iodide solutions because of its reactivity with iodides whereas stainless steel was a suitable material without corrosion traces after long-term cycling. Additionally to the typical symmetric configurations, where three different redox active electrolytes were used, we decided also to investigate their behavior in the asymmetric system, in terms of electrolytes. It means that we used the conjugated redox couples as the electrolyte solution with different carbon materials. The best performance was obtained by AAC 1 and AAC 2 carbon material in the symmetric system. For enhancement of capacitor work carbon nanotubes (10%) were used as perfectly conducting additive and source of mesopores. Those two carbons modified with the nanotubes were employed in the capacitor operating in iodide/vanadium electrolyte solutions. AAC 2 revealed the highest capacitance value and showed good energy/power performance, i.e., 20 Wh kg−1 with max. power density of 2 kW kg−1. Effect of the nanotube addition was more visible in the case of AAC 1 because this carbon did not exhibit the mesoporous structure as AAC 2. In this work it was also demonstrated that the membrane Nafion 117 plays a crucial role when employing the redox-active couples based on iodine and vanadium species. Acknowledgments This work is supported by INGEC project in the frame of the Swiss–Polish Programme (PSPB-107/2010). The authors would like to acknowledge Polish Ministry of Science and Higher Education, grant no. NN511 474339. References [1] T. Horiba, T. Maeshima, T. Matsumura, M. Koseki, J. Arai, Y. Muranaka, J. Power Sources 146 (2005) 107–110. [2] S.L. Candelaria, Y. Shao, W. Zhou, X. Li, J. Xiao, J.-G. Zhang, Y. Wang, J. Liu, J. Li, G. Cao, Nano Energy 1 (2012) 195–220. [3] P. Chen, G. Shen, Y. Shi, H. Chen, C. Zhou, ACS Nano 4 (2010) 4403–4411. [4] X. Li, B. Wei, Nano Energy 2 (2013) 159–173. [5] Y.W.F. Wang, S. Xiao, Y. Hou, C. Hu, L. Liu, RSC Adv. 3 (2013) 13059–13084. [6] R. Kötz, M. Carlen, Electrochim. Acta 45 (2000) 2483–2498. [7] M. Mastragostino, F. Soavi, J. Power Sources 174 (2007) 89–93. [8] E. Frackowiak, F. Béguin, Carbon 39 (2001) 937–950. [9] M. Toupin, T. Brousse, D. Bélanger, Chem. Mater. 16 (2002) 3946–3952. [10] W. Sugimoto, T. Kizaki, K. Yokoshima, Y. Murakami, Y. Takasu, Electrochim. Acta 49 (2004) 313–320. [11] A. Malak, K. Fic, G. Lota, C. Vix-Guterl, E. Frackowiak, J. Solid State Electrochem. 14 (2009) 811–816. [12] J.Z.G. Wang, L. Zhang, Chem. Soc. Rev. 41 (2012) 797–828. [13] S. Roldán, M. Granda, J. Phys. Chem. C 115 (2011) 17606–17611. [14] Z. Algharaibeh, P.G. Pickup, Electrochem. Commun. 13 (2011) 147–149. [15] A.S. Kumar, S. Sornambikai, P. Gayathri, J.-M. Zen, J. Electroanal. Chem. 641 (2010) 131–135. [16] A.L. Comte, G. Pognon, T. Brousse, D. Bélanger, Electrochemistry 81 (2013) 863–866. [17] S. Roldán, C. Blanco, M. Granda, R. Menéndez, R. Santamaría, Angew. Chem. Int. Ed. 50 (2011) 1699–1701. [18] S. Roldán, Z. González, C. Blanco, M. Granda, R. Menéndez, R. Santamaría, Electrochim. Acta 56 (2011) 3401–3405. [19] S.T. Senthilkumar, R.K. Selvan, Y.S. Lee, J.S. Melo, J. Mater. Chem. A 1 (2013) 1086. [20] S.T. Senthilkumar, R.K. Selvan, N. Ponpandian, J.S. Melo, RSC Adv. 2 (2012) 8937. [21] G. Lota, E. Frackowiak, Electrochem. Commun. 11 (2009) 87–90. [22] G. Lota, K. Fic, E. Frackowiak, Electrochem. Commun. 13 (2011) 38–41. [23] J. Menzel, K. Fic, M. Meller, E. Frackowiak, J. Appl. Electrochem. 44 (2014) 439–445. [24] E. Frackowiak, K. Fic, M. Meller, G. Lota, ChemSusChem 5 (2012) 1181–1185. [25] P.H. Svensson, L. Kloo, Chem. Rev. 103 (2003) 1649–1684. [26] K. Fic, G. Lota, E. Frackowiak, Energy Environ. Sci. 5 (2012) 5842–5850. [27] D. Weingarth, H. Noh, A. Foelske-Schmitz, A. Wokaun, R. Kötz, Electrochim. Acta 103 (2013) 119–124.