- Email: [email protected]
Application of multistep electrospinning method for preparation of electrical double-layer capacitor half-cells
Electrochimica Acta 119 (2014) 72–77
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Application of multistep electrospinning method for preparation of electrical double-layer capacitor half-cells K. Tõnurist 1 , I. Vaas 1 , T. Thomberg 1 , A. Jänes 1 , H. Kurig 1 , T. Romann 1 , E. Lust ∗,1 Institute of Chemistry, University of Tartu, 14A Ravila Street, 50411 Tartu, Estonia
a r t i c l e
i n f o
Article history: Received 16 October 2013 Received in revised form 22 November 2013 Accepted 26 November 2013 Available online 12 December 2013 Keywords: Electrospinning Electrospun half-cell Electrical double-layer capacitor Micromesoporous carbon Specific energy and power
a b s t r a c t Electrochemical characteristics of the two identical electrospun micromesoporous carbon electrodes and polymer membrane based half-cells, prepared using the electrospinning method, have been studied in 1 M triethylmethylammonium tetrafluoroborate solution in acetonitrile, using cyclic voltammetry, constant current charge/discharge, constant power discharge and electrochemical impedance spectroscopy methods. The novel electrical double-layer capacitor (EDLC) half-cell preparation method consisted of two steps, at first the nanowire polymer separator layer has been prepared by electrospinning method using poly(vinylidene fluoride) (PVDF) solution in N,N − dimethylformamide and acetone mixture. Thereafter, the electrode material layer has been directly electrospun onto the separator layer from the mixture of carbon (90% commercial RP-20 carbon + 10% few layered graphene) and PVDF solution in N,N − dimethylformamide + acetone mixture. Applying the novel composite electrode preparation method we are able to prepare composite electrodes for EDLC with lower density and thickness than using the widely used traditional roll-pressing method. The wide region of ideal polarizability, high specific capacitance, low equivalent series resistance, short characteristic time constant and high specific energy and specific power values for the completed EDLC have been obtained and discussed. © 2014 Elsevier Ltd. All rights reserved.
1. Introduction Investigation of the influence of the physical and electrochemical characteristics of the micromesoprous carbon electrodes and separator materials on the performance of the electrical double-layer capacitor (EDLC) is important, taking into account the need for higher power density EDLCs [1–9]. The carbon electrode properties (thickness, density, specific surface area, microand mesoporosity, ratio of micro- and mesopores volume, etc.) have a noticeable influence on the ions adsorption/desorption rate at/inside the charged micromesoporous carbon electrodes soaked into the non − aqueous electrolyte solution or ionic liquids [10–13]. It is well-known that in order to increase the EDLC performance the porous structure, thickness, wettability and the mass transfer rate of ions in mircomesoporous carbon material, depending on the electrode preparation method, as well as in separator, have to be developed. Thus, the new carbon electrodes with optimized hierarchical structures and enhanced mass transfer and ions adsorption rate have to be designed, prepared and tested. Electrospinning is a well-known method for processing of the flexible and highly porous nanostructured (fibrous or porous
∗ Corresponding author. Tel.: +3727375165, fax: +3727375264. E-mail address: [email protected] (E. Lust). 1 ISE member 0013-4686/$ – see front matter © 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.electacta.2013.11.155
layered) scaffolds by applying a high electric field strength to a droplet of polymer solution (or melt) [7,14–16]. There have been only few attempts to use the electrospinning method for the preparation of the raw thin-layer electrodes for the energy technology devices [16] and electromechanical linear actuators. The main aim of this study was to use the multistep electrospinning method for preparation of EDLC half-cells, which could have lower density and thickness than conventional electrodes prepared applying traditional roll-pressing method. The supporting membrane material [1,2] has been prepared from 22.5% poly(vinylidene fluoride) (PVDF) solution in N,N − dimethylformamide (DMF) + acetone (AC) mixture (8020 wt/wt) by electrospinning method. Thereafter, the electrode material layer has been deposited by the electrospinning step onto the polymer membrane from the mixture of carbon (commercial RP-20 carbon (Kuraray Chemical Co.) and few layered graphene RexSheet® produced by CVD method (Pred Materials International), with 90-10 wt/wt ratio, respectively) and PVDF in DMF + AC (80-20 wt/wt) mixture to form the half-cell. Different electrode material solutions were used to establish the optimal ratio between carbon and binder (PVDF) to be used for electrospinning step. The ratio of carbon powders and binder (C-PVDF) was varied: 70-30; 75-25; 80-20 and 85-15 (wt/wt), respectively, in the DMF + AC (8020 wt/wt) solvent mixture. For completing EDLC, the two identical half-cells have been put together. The electrochemical results of the electrospun cells were compared with the results obtained by
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77 Table 1 Prepared EDLC cells with different ratios of carbon and binder material. Sample
Thickness of the cell (m)
Active mass of one electrode (g)
C-PVDF (85-15) C-PVDF (80-20) C-PVDF (75-25) C-PVDF (70-30) C-PTFE (94-6)
80 ± 5 80 ± 5 130 ± 5 110 ± 5 220 ± 5
0.0014 0.0018 0.0025 0.0024 0.015
using micromesoporous carbon electrodes prepared by the rollpressing method (roll-press HS-160 N, Hohsen Corporation, Japan) from the same carbon mixture (RP-20 + few layered graphene (9010 wt/wt)) and polytetrafluoroethylene (PTFE) binder (6 wt%). 2. Experimental PVDF (molecular weight 530 000 g mol−1 ), DMF (99.8%) (Fluka), AC (Sigma − Aldrich (puriss)), RP-20 carbon (Kuraray Chemical Co.) and few layered graphene RexSheet® (average sheet thickness 9 nm, average sheet length 15 m) (Pred Materials International) were used as received. The supporting PVDF membrane has been prepared by using PVDF solution in DMF + AC (80-20, wt/wt) mixture [1,2]. Thereafter, RP-20 and few layered graphene (90-10, wt/wt) were mixed with various amounts of PVDF solution (70-30; 75-25; 80-20 and 85-15, wt/wt) in DMF + AC mixture (80-20) and electrospun directly onto the separator material (Table 1) at electric field strength 1.4 kV cm−1 . After drying under vacuum, the electrode/membrane half-cell electrode side was covered with the pure Al layer (2 m) by applying the ac magnetron sputtering method. The total thickness of the completed electrospun cells under study varied from 80 ± 5 to 130 ± 5 m (including two half-cells and Al contact layers). The structure of the electrodes and separator were investigated by scanning electron microscopy method (SEM). The electrospun electrode layer was sputter − coated with gold before the SEM observation. The surface structure of the C-PVDF electrodes was studied using a Microtrac Semtrac system. Fig. 1 shows that the highly micromesoporous composite electrodes have been prepared. In the case of C-PVDF (75-25) and C-PVDF (80-20) electrospun electrodes, PVDF fibres can be observed between the carbon particles in the SEM images (Figs. 1a-d). However, when the PVDF concentration in C-PVDF composite electrodes is lower than 20%, we did not observed any PVDF fibres between the carbon particles (Figs. 1e-f). The porous structure of the compact fibrous electrospun separator layer can be seen in Fig. 1 g. Specific surface area (SBET ) (Table 2), pore size distribution, micropore volume (Vmicro ), micropore surface area (Smicro ) and other parameters for electrospun C-PVDF composite electrodes have been measured by ASAP 2020 system and calculated from nitrogen sorption data at liquid nitrogen temperature according to the multipoint BET and t-plot methods [17]. For later comparison the gas sorption measurement results for C-PTFE electrode materials have been obtained and added (Table 2). According to the data in Table 2, electrospun composite electrodes are mainly Table 2 Results of gas sorption measurements of electrode material prepared from carbon mixture (RP-20/graphene, 90-10 wt/wt) with different amounts of binder material. Sample
SBET (m2 g-1 )
Smicro (m2 g-1 )
Vmicro (cm3 g-1 )
Vtot (cm3 g-1 )
C-PVDF (85-15) C-PVDF (80-20) C-PVDF (75-25) C-PVDF (70-30) C-PTFE (94-6)
1172 1089 1031 943 1284
1161 1077 1028 933 1272
0.53 0.49 0.47 0.43 0.58
0.55 0.52 0.48 0.45 0.61
73
microporous and the specific surface area decreases with the increase of binder content in the mixture. From the the gas sorption measurement results it was found that the binder content did not have any noticeable influence on the pore size distribution plot (not shown for the shortness). All electrochemical experiments were carried out inside a glove box (Labmaster sp, MBraun; O2 and H2 O concentrations lower than 0.1 ppm) in a nitrogen atmosphere. 1 M (C2 H5 )3 CH3 NBF4 + AN solution was used as an electrolyte. Cyclic voltammetry (CV), constant current charge/discharge (CC) and impedance spectra over ac frequency (f) range from 3 × 105 to 1 × 10-2 Hz were recorded using a Solartron 1287 potentiostat with 1252A frequency response analyser (5 mV ac modulation). Constant power discharge tests were performed with a BT2000 testing system (Arbin Instruments, USA). All electrochemical measurements were carried out at 23 ± 0.5 ◦ C. 3. Results and discussion 3.1. Cyclic voltammetry data The cyclic voltammograms have been measured applying various cell potential scan rates. However, the CV data have been recalculated into capacitance values using Eq. (1): C = jv−1
(1)
This equation is valid if we assume that the capacitance C is constant, and if the series resistance Rs → 0, or if the current j → 0 in Eq. (1). Thus, Eq. (1) can be used to obtain the capacitance only in the case of slow and moderate cell potential scan rates (v), when the value of current is very small. At these conditions the potential drop (IR − drop) is negligible and the current response is essentially equal to that of a pure capacitor [3,5,7 − 10,13,17]. Under these conditions in a symmetrical two − electrode system, the active material specific capacitance Cs (F g−1) can be obtained from the capacitance of the completed EDLC cell by Eq. (2). Cs =
2C m
(2)
where m is the the active material weight (g) in one electrode, assuming that the positively and negatively charged electrodes have the same capacitance at fixed cell potential (V) applied. Specific capacitance Cs vs. cell potential V plots in Fig. 2 show that for all EDLC cells consisting of electrospun C-PVDF or C-PTFE composite electrodes and 1 M (C2 H5 )3 CH3 NBF4 + AN electrolyte, so − called nearly ideal capacitive behaviour (i.e. cyclic voltammograms with the nearly rectangular shape) has been established at all potential scan rates v ≤ 100 mV s-1 and cell potential V ≤ 2.7 V (Figs. 2a-c). At higher cell potentials the values of Cs start to increase, being caused by electroreduction of O2 and H2 O traces at a negatively charged electrode and by oxidation of surface functionalities at a positively charged electrode [9]. This is probably caused by the impurities found in the carbon materials, which were used as received. As it has been demonstrated by Laheäär et al. [18] that 1 M (C2 H5 )3 CH3 NBF4 + AN electrolyte is stable in much wider cell potential range V ≤ 3.2 V and based on data in Fig. 2a the electrospun C-PVDF and C-PTFE systems demonstrate similar behaviour. According to the CV data, the specific capacitance depends on the amount of PVDF added into the raw electrode material solution and Cs increases with the decrease of PVDF content in the electrospun C-PVDF composite electrodes (Fig. 2b), except in the case of electrospun C-PVDF (85-15) electrode, which had lower mechanical strength than other electrospun electrodes containing more binder (PVDF).
74
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77
Fig. 1. Scanning electron microscopy images for: (a) and (b) C-PVDF (75-25), (c) and (d) C-PVDF (80-20), (e) and (f) C-PVDF (85-15) composite electrodes, and (g) separator prepared using the electrospinning method.
3.2. Impedance complex plane plots The impedance complex plane (Z”, Z’) plots (so-called Nyquist plots, Fig. 3a) for EDLC completed with electrospun C-PVDF and C-PTFE composite electrodes and 1 M (C2 H5 )3 CH3 NBF4 + AN solution have been measured at fixed cell potentials from 0 to 2.7 V,
within the range of ac frequency f from 1 × 10-2 to 3 × 105 Hz. The shape of Z” vs. Z’ plots depends noticeably on the amount of binder in the electrodes and on the electrode preparation method (Fig. 3a) used. Similarly to other micromesoporous electrodes [6–13], the Nyquist plot consists mainly of three parts: (1) the noticeably depressed semicircle at higher ac frequency,
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77
75
electrolyte conductivity in macro/mesoporous matrix. The addition of PVDF into electrode material changes the wettability (electrowetting properties) of the electrode surface layer, influencing the capacitance of macro/mesoporous surface areas and highmedium frequency polarisation resistance of electrode. Therefore, based on the theoretical model, developed in [10], the semicircle shape is very complex variable and it is determined by the adsorption and mass transfer processes rates of ions at/inside micro/mesoporous carbon electrode, and on the series resistance of a material, mass transfer resistance inside of the macro/mesoporous carbon structure as well as on the mass transfer resistance of ions within the micropores. From the data given in Figs. 3a-b, it can be seen, that the nearly ideal capacitive behaviour has been observed at f ≤ 0.01 Hz for all EDLCs under study. However, based on the Bode phase angle plots the EDLCs completed with optimal C-PVDF (85-15) and C-PVDF (80-20) electrodes (having lower binder content and being thinner) are able to achieve the nearly ideal capacitive behaviour already at the ac frequency 0.1 Hz (Fig. 3b). To characterise further the EDLC properties, the complex power dependencies have been constructed [2,5–7]. The values of complex power can be expressed as S(ω) = P(ω) + jQ (ω),
(3)
where the real part of power
2
P(ω) = ωC"(ω)Vrms ,
(4)
and the imaginary part of power
2
Q (ω) = −ωC (ω)Vrms ,
(5)
2 √ with Vrms = Vmax / 2 (Vmax is the maximal amplitude of ac voltage) [2,6,7]. System with the ideal capacitive behaviour has no real part of the complex power, as there is only the reactive contribution to the complex power, and Eq. (3) simplifies to 2 jVrms 2 = −jωC(ω)Vrms S(ω) = jQ (ω) = − . Z"(ω)
(6)
System with the ideal |Z (ω)| resistive behaviour has no imaginary part of the complex power as this component only dissipates energy and the complex power takes the well-known form
Vrms 2 . S(ω) = Z (ω) Fig. 2. Specific capacitance (Cs ) vs. cell potential (V) curves at (a) scan rate = 1 mV s−1 at different cell potentials and Cs vs. V curves within the cell potential range from 0 to 2.7 V at scan rates (b) = 1 mV s−1 , and (c) = 100 mV s−1 for the EDLC cells based on two identical composite electrodes (noted on figure) in 1 M triethylmethylammonium tetrafluoroborate solution in acetonitrile.
(2) the linear area, so-called “porous” region with a slope of ˛ ≈ -45◦ , characteristic of the mass transfer limited process in the micro/mesoporous matrix of an electrode [8–11,17] and (3) the so-called double-layer capacitance region with a slop of ˛ ≈-90◦ (“knee” at low ac frequencies), obtained by the finite length adsorption effect [8–11,17]. According to the data in Fig. 3a, the size of the depressed semicircle increases with the increase of the binder content in the electrospun electrodes. It has been demonstrated by Eikerling et al. that the shape (i.e. width) of the high-medium frequency semicircle (i.e. high-medium frequency polarization resistance) depends noticeably on the hierarchical structure of the electrode macro/mesoporosity parameters [10]. Increase of PVDF content in carbon composite electrodes layer decreases the macro/mesoporosity of active material layer and thus, the
(7)
It should be also noted that real EDLCs balances between the two states mentioned before: resistive at high frequencies (f → ∞) and capacitive at low frequencies (f → 0). Between these two states EDLC behaves like a resistive-capacitive (RC) transmission line equivalent circuit. At the condition |P| = |Q|, i.e., when phase angle = -45◦ , we can obtain relaxation frequency fr , determining the characteristic time constant r ( r = 2 fr -1 ) of the EDLC cell. Analysis of the data in Figs. 3 and 4 show, that only half of the low-frequency capacitance is reached to establish at r [2,5–7]. Comparison of the data for the cells completed using different electrospun C-PVDF and C-PTFE composite electrodes indicate a noticeable influence of the binder content and cell thickness on the time constant calculated. EDLC based on the thinner electrospun C-PVDF composite electrodes are able to deliver their energy approximately 10 times faster than EDLC based on thick C-PTFE electrodes. However, it should be noted that there is a certain discrepancy between the CV and low frequency impedance data in Fig. 2 and 3a, respectively. This is mainly caused by lower mechanical (including electrochemical) stability of the CPVDF (85-15) electrode compared with C-PVDF (80-20) and with
76
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77
Fig. 5. Ragone plots for the EDLC based on different electrodes (noted in figure) obtained from constant power discharge testes within the cell potential range from 3.0 to 1.5 V.
3.3. Ragone plots
Fig. 3. (a) Nyquist and (b) Bode phase angle plots for the EDLC cell completed with different composite electrodes (noted on figure) at cell potential 2.7 V.
other composite systems under study. Therefore, the long testing impedance measurements after repetitive CV and constant current charge/discharge measurements demonstrated lower capacitance values for this composition due to the electrochemical degradation (expansion of electrodes during charging and compression of electrodes during discharging). In addition the microwetting properties have been exchanged if various wt% of PVDF have been used, influencing the time stability of electrochemical parameters under estimation.
The specific energy (Wh kg-1 ) and power (kW kg-1 ) relationship [2] (calculated taking into account the total active material weight of two electrodes), i.e., Ragone plots, for the EDLCs based on electrospun C-PVDF and C-PTFE electrodes have been obtained from constant power discharge tests within the cell potential range from 3.0 V to 1.5 V and are shown in Fig. 5. The Ragone plots for electrospun composite electrodes depend noticeably on the binder content in the electrodes. Electrospun electrodes containing smaller amount of binder C-PVDF (85-15 and 80-20) deliver noticeably higher power at constant energy. Very high specific energy values E ∼ 30 Wh kg-1 have been achieved at very high power loads P ∼ 100 kW kg-1 for C-PVDF 85-15 and C-PVDF 80-20 composite electrodes based EDLCs. Thus, noticeably better performance for electrospun C-PVDF composite electrodes based EDLCs has been established compared with C-PTFE electrodes based EDLCs. 3.4. Time stability tests It has been demonstrated by D. Weingarth et al. [19] that the constant voltage hold tests are more demanding compared with respective cycling tests (constant voltage charge/discharge cycling) with identical upper voltage limits. Therefore, the constant voltage hold test, i.e., the repetitive holding of capacitor cells at constant potential V = 2.7 V for fixed time t = 5 h, thereafter applying constant current charge/discharge regimes with 2 A g-1 within the cell potential range from 1.35 to 2.7 V and impedance spectroscopy at 2.7 V, have been applied to investigate the electrochemical stability of EDLC cells under study. The discharge and charge capacitances were calculated from data of the third cycle. The capacitance of the cell CCC (F cm-2 ) was obtained from the slope of the discharge curves according to Eq. 8: CCC = j
Fig. 4. Normalized complex power plots and characteristic time constant values (inset) for the EDLC based on different electrodes (noted in figure) at cell potential 2.7 V.
dt , d(V )
(8)
where dt/d(V) is the slop of the discharge curve at applied constant current density j. After 150 hours of constant voltage hold test the longest charge/discharge cycle has been observed for CPVDF (80-20) as well as for C-PTFE (94-6) composite electrodes based EDLCs (Fig. 6), corresponding to the highest specific capacitance calculated from CC-curves: CCC = 92 F g-1 and CCC = 89 F g-1 (Fig. 7), respectively. From inset in Fig. 6, it can be seen, that the C-PTFE (94-6) composite electrodes based EDLC had much bigger IR-drop compared with electrospun C-PVDF (80-20) electrodes based EDLC. Interestingly, the higher electrochemical degradation
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77
77
F g-1 ), very small characteristic time constant (0.043 s) and very high specific power (245 kW kg-1 ) at the average specific energy (22 Wh kg-1 ) applied. It was found that EDLCs, based on the new electrospun composite electrodes are able to deliver much faster and higher power at applied constant energy, than the EDLC based on the electrodes prepared by the traditional roll-pressed method. It was shown that with the novel electrode preparation method micromesoporous composite electrodes with lower thickness and density can be prepared. Acknowledgements
Fig. 6. Constant current charge/discharge cycles at current density j = 2 A g−1 for EDLCs completed using C-PVDF (80-20) and C-PTFE (94-6) electrodes.
Fig. 7. Results of floating tests for the EDLCs, completed using electrospun C-PVDF (85-15) and (80-20), and C-PTFE (94-6) electrodes at cell potential 2.7 V.
rate was observer for electrodes with higher PVDF content in electrospun C-PVDF composite electrodes (not shown for shortness), explained by the PVDF degradation inside the electrode material. 4. Conclusions Composite micromesoporous carbon electrodes based on RP20 + few layered graphene + PVDF binder have been prepared by depositing of composite micromesoporous electrode material directly onto the electrospun supporting PVDF membrane applying electrospinning method. The influence of the electrospun micromesoporous C-PVDF composite electrode characteristics (porosity, thickness), on the electrical double-layer capacitor (EDLC) parameters have been tested using cyclic voltammetry, constant current charge/discharge, constant power discharge, and electrochemical impedance spectroscopy methods in the 1 M triethylmethylammonium tetrafluoroborate in acetonitrile solution. The limits of ideal polarizability (V ≤ 2.7 V), specific capacitance, equivalent series resistance, phase angle, characteristic time constant, specific energy and power of the EDLC completed have been obtained and discussed. The optimum carbon and PVDF binder ratio for electrospun electrodes was established for C-PVDF (8020) system, which demonstrated the high specific capacitance (120
The present study was supported by The Estonian Centers of Excellence in Science: High Technology Materials for Sustainable Development; ETF Projects 8172, 8786; Estonian Ministry of Education and Research project SF0180002s08 and PUT55 project; European Regional Development Fund Project SLOKT10209 T. References [1] K. Tõnurist, T. Thomberg, A. Jänes, I. Kink, E. Lust, Specific performance of electrical double layer capacitors based on different separator materials in room temperature ionic liquid, Electrochem. Comm. 22 (2012) 77. [2] K. Tõnurist, T. Thomberg, A. Jänes, E. Lust, Specific performance of supercapacitors at lower temperatures based on different separator materials, J. Electrochem. Soc. 160 (2013) A449. [3] B.E. Conway, Electrochemical Supercapacitors: Scientific Fundamentals, Technological Applications, Kluwer Academic, Plenum Publishers, New York, 1999. [4] G. Salitra, A. Soffer, L. Eliad, Y. Cohen, D. Aurbach, Carbon electrodes for doublelayer capacitors I. Relations between ion and pore dimensions, J. Electrochem. Soc. 147 (2000) 2486. [5] J. Chmiola, G. Yushin, Y. Gogotsi, C. Portet, P. Simon, L.P. Taberna, Anomalous increase in carbon capacitance at pore sizes less than 1 nanometer, Science 313 (2006) 1760. [6] K. Tõnurist, T. Thomberg, A. Jänes, T. Romann, V. Sammelselg, E. Lust, Influence of separator properties on electrochemical performance of electrical double layer capacitors, J. Electroanal. Chem. 689 (2013) 8. [7] K. Tõnurist, A. Jänes, T. Thomberg, H. Kurig, E. Lust, Influence of mesoporous separator properties on the parameters of electrical double layer capacitor single cells, J. Electrochem. Soc. 156 (2009) A334. [8] A. Jänes, H. Kurig, E. Lust, Characterisation of activated nanoporous carbon for supercapacitor electrode materials, Carbon 45 (2007) 1226. [9] A. Jänes, L. Permann, M. Arulepp, E. Lust, Electrochemical characteristics of nanoporous carbide-derived carbon materials in nonaqueous electrolyte solutions, Electrochem. Comm. 6 (2004) 313. [10] M. Eikerling, A.A. Kornyshev, E. Lust, Optimized structure of nanoporous carbon-based double-layer capacitors, J. Electrochem. Soc. 152 (2005) E24. [11] E. Lust, A. Jänes, M. Arulepp, Influence of electrolyte characteristics on the electrochemical parameters of electrical double layer capacitors, J. Solid State Electrochem. 8 (2004) 488. [12] E. Lust, G. Nurk, A. Jänes, M. Arulepp, P. Nigu, P. Möller, S. Kallip, V. Sammelselg, Electrochemical properties of nanoporous carbon electrodes in various nonaqueous electrolytes, J. Solid State Electrochem. 7 (2003) 91. [13] E. Lust, A. Jänes, T. Pärn, P. Nigu, Influence of nanoporous carbon electrode thickness on the electrochemical characteristics of a nanoporous carbon]tetraethylammonium tetrafluoroborate in acetonitrile solution interface, J. Solid State Electrochem. 8 (2004) 224. [14] A. Formhals, US Patent 197 (1934) 5504. [15] S. Ramakrishna, K. Fujihara, W.-E. Teo, T.-C. Lim, Z. Ma, An Introduction to Electrospinning and Nanofibers, World Scientific Publishing Co. Pte. Ltd, Singapore, 2005. [16] S. Chaudhari, Y. Sharma, P.S. Archana, R. Jose, S. Ramakrishna, S. Mhaisalkar, M. Srinivasan, Electrospun polyaniline nanofibers web electrodes for supercapacitors, J. Appl. Polymer. Sci. 129 (2013) 1660. [17] A. Jänes, T. Thomberg, E. Lust, Synthesis and characterization of nanoporous carbide-derived carbon by chlorination of vanadium carbide, Carbon 45 (2007) 2717. [18] A. Laheäär, A-L. Peikolainen, M. Koel, A. Jänes, E. Lust, Comparison of carbon aerogel and carbide-derived carbon as electrode materials for non-aqueous supercapacitors with high performance, J Solid State Electrochem. 16 (2012) 2717. [19] D. Weingarth, A. Foelske-Schmitz, R. Kötz, Cycle versus voltage hold - Which is the better stability test for electrochemical double layer capacitors? J. Power Sources 225 (2013) 84.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Application of multistep electrospinning method for preparation of electrical double-layer capacitor half-cells K. Tõnurist 1 , I. Vaas 1 , T. Thomberg 1 , A. Jänes 1 , H. Kurig 1 , T. Romann 1 , E. Lust ∗,1 Institute of Chemistry, University of Tartu, 14A Ravila Street, 50411 Tartu, Estonia
a r t i c l e
i n f o
Article history: Received 16 October 2013 Received in revised form 22 November 2013 Accepted 26 November 2013 Available online 12 December 2013 Keywords: Electrospinning Electrospun half-cell Electrical double-layer capacitor Micromesoporous carbon Specific energy and power
a b s t r a c t Electrochemical characteristics of the two identical electrospun micromesoporous carbon electrodes and polymer membrane based half-cells, prepared using the electrospinning method, have been studied in 1 M triethylmethylammonium tetrafluoroborate solution in acetonitrile, using cyclic voltammetry, constant current charge/discharge, constant power discharge and electrochemical impedance spectroscopy methods. The novel electrical double-layer capacitor (EDLC) half-cell preparation method consisted of two steps, at first the nanowire polymer separator layer has been prepared by electrospinning method using poly(vinylidene fluoride) (PVDF) solution in N,N − dimethylformamide and acetone mixture. Thereafter, the electrode material layer has been directly electrospun onto the separator layer from the mixture of carbon (90% commercial RP-20 carbon + 10% few layered graphene) and PVDF solution in N,N − dimethylformamide + acetone mixture. Applying the novel composite electrode preparation method we are able to prepare composite electrodes for EDLC with lower density and thickness than using the widely used traditional roll-pressing method. The wide region of ideal polarizability, high specific capacitance, low equivalent series resistance, short characteristic time constant and high specific energy and specific power values for the completed EDLC have been obtained and discussed. © 2014 Elsevier Ltd. All rights reserved.
1. Introduction Investigation of the influence of the physical and electrochemical characteristics of the micromesoprous carbon electrodes and separator materials on the performance of the electrical double-layer capacitor (EDLC) is important, taking into account the need for higher power density EDLCs [1–9]. The carbon electrode properties (thickness, density, specific surface area, microand mesoporosity, ratio of micro- and mesopores volume, etc.) have a noticeable influence on the ions adsorption/desorption rate at/inside the charged micromesoporous carbon electrodes soaked into the non − aqueous electrolyte solution or ionic liquids [10–13]. It is well-known that in order to increase the EDLC performance the porous structure, thickness, wettability and the mass transfer rate of ions in mircomesoporous carbon material, depending on the electrode preparation method, as well as in separator, have to be developed. Thus, the new carbon electrodes with optimized hierarchical structures and enhanced mass transfer and ions adsorption rate have to be designed, prepared and tested. Electrospinning is a well-known method for processing of the flexible and highly porous nanostructured (fibrous or porous
∗ Corresponding author. Tel.: +3727375165, fax: +3727375264. E-mail address: [email protected] (E. Lust). 1 ISE member 0013-4686/$ – see front matter © 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.electacta.2013.11.155
layered) scaffolds by applying a high electric field strength to a droplet of polymer solution (or melt) [7,14–16]. There have been only few attempts to use the electrospinning method for the preparation of the raw thin-layer electrodes for the energy technology devices [16] and electromechanical linear actuators. The main aim of this study was to use the multistep electrospinning method for preparation of EDLC half-cells, which could have lower density and thickness than conventional electrodes prepared applying traditional roll-pressing method. The supporting membrane material [1,2] has been prepared from 22.5% poly(vinylidene fluoride) (PVDF) solution in N,N − dimethylformamide (DMF) + acetone (AC) mixture (8020 wt/wt) by electrospinning method. Thereafter, the electrode material layer has been deposited by the electrospinning step onto the polymer membrane from the mixture of carbon (commercial RP-20 carbon (Kuraray Chemical Co.) and few layered graphene RexSheet® produced by CVD method (Pred Materials International), with 90-10 wt/wt ratio, respectively) and PVDF in DMF + AC (80-20 wt/wt) mixture to form the half-cell. Different electrode material solutions were used to establish the optimal ratio between carbon and binder (PVDF) to be used for electrospinning step. The ratio of carbon powders and binder (C-PVDF) was varied: 70-30; 75-25; 80-20 and 85-15 (wt/wt), respectively, in the DMF + AC (8020 wt/wt) solvent mixture. For completing EDLC, the two identical half-cells have been put together. The electrochemical results of the electrospun cells were compared with the results obtained by
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77 Table 1 Prepared EDLC cells with different ratios of carbon and binder material. Sample
Thickness of the cell (m)
Active mass of one electrode (g)
C-PVDF (85-15) C-PVDF (80-20) C-PVDF (75-25) C-PVDF (70-30) C-PTFE (94-6)
80 ± 5 80 ± 5 130 ± 5 110 ± 5 220 ± 5
0.0014 0.0018 0.0025 0.0024 0.015
using micromesoporous carbon electrodes prepared by the rollpressing method (roll-press HS-160 N, Hohsen Corporation, Japan) from the same carbon mixture (RP-20 + few layered graphene (9010 wt/wt)) and polytetrafluoroethylene (PTFE) binder (6 wt%). 2. Experimental PVDF (molecular weight 530 000 g mol−1 ), DMF (99.8%) (Fluka), AC (Sigma − Aldrich (puriss)), RP-20 carbon (Kuraray Chemical Co.) and few layered graphene RexSheet® (average sheet thickness 9 nm, average sheet length 15 m) (Pred Materials International) were used as received. The supporting PVDF membrane has been prepared by using PVDF solution in DMF + AC (80-20, wt/wt) mixture [1,2]. Thereafter, RP-20 and few layered graphene (90-10, wt/wt) were mixed with various amounts of PVDF solution (70-30; 75-25; 80-20 and 85-15, wt/wt) in DMF + AC mixture (80-20) and electrospun directly onto the separator material (Table 1) at electric field strength 1.4 kV cm−1 . After drying under vacuum, the electrode/membrane half-cell electrode side was covered with the pure Al layer (2 m) by applying the ac magnetron sputtering method. The total thickness of the completed electrospun cells under study varied from 80 ± 5 to 130 ± 5 m (including two half-cells and Al contact layers). The structure of the electrodes and separator were investigated by scanning electron microscopy method (SEM). The electrospun electrode layer was sputter − coated with gold before the SEM observation. The surface structure of the C-PVDF electrodes was studied using a Microtrac Semtrac system. Fig. 1 shows that the highly micromesoporous composite electrodes have been prepared. In the case of C-PVDF (75-25) and C-PVDF (80-20) electrospun electrodes, PVDF fibres can be observed between the carbon particles in the SEM images (Figs. 1a-d). However, when the PVDF concentration in C-PVDF composite electrodes is lower than 20%, we did not observed any PVDF fibres between the carbon particles (Figs. 1e-f). The porous structure of the compact fibrous electrospun separator layer can be seen in Fig. 1 g. Specific surface area (SBET ) (Table 2), pore size distribution, micropore volume (Vmicro ), micropore surface area (Smicro ) and other parameters for electrospun C-PVDF composite electrodes have been measured by ASAP 2020 system and calculated from nitrogen sorption data at liquid nitrogen temperature according to the multipoint BET and t-plot methods [17]. For later comparison the gas sorption measurement results for C-PTFE electrode materials have been obtained and added (Table 2). According to the data in Table 2, electrospun composite electrodes are mainly Table 2 Results of gas sorption measurements of electrode material prepared from carbon mixture (RP-20/graphene, 90-10 wt/wt) with different amounts of binder material. Sample
SBET (m2 g-1 )
Smicro (m2 g-1 )
Vmicro (cm3 g-1 )
Vtot (cm3 g-1 )
C-PVDF (85-15) C-PVDF (80-20) C-PVDF (75-25) C-PVDF (70-30) C-PTFE (94-6)
1172 1089 1031 943 1284
1161 1077 1028 933 1272
0.53 0.49 0.47 0.43 0.58
0.55 0.52 0.48 0.45 0.61
73
microporous and the specific surface area decreases with the increase of binder content in the mixture. From the the gas sorption measurement results it was found that the binder content did not have any noticeable influence on the pore size distribution plot (not shown for the shortness). All electrochemical experiments were carried out inside a glove box (Labmaster sp, MBraun; O2 and H2 O concentrations lower than 0.1 ppm) in a nitrogen atmosphere. 1 M (C2 H5 )3 CH3 NBF4 + AN solution was used as an electrolyte. Cyclic voltammetry (CV), constant current charge/discharge (CC) and impedance spectra over ac frequency (f) range from 3 × 105 to 1 × 10-2 Hz were recorded using a Solartron 1287 potentiostat with 1252A frequency response analyser (5 mV ac modulation). Constant power discharge tests were performed with a BT2000 testing system (Arbin Instruments, USA). All electrochemical measurements were carried out at 23 ± 0.5 ◦ C. 3. Results and discussion 3.1. Cyclic voltammetry data The cyclic voltammograms have been measured applying various cell potential scan rates. However, the CV data have been recalculated into capacitance values using Eq. (1): C = jv−1
(1)
This equation is valid if we assume that the capacitance C is constant, and if the series resistance Rs → 0, or if the current j → 0 in Eq. (1). Thus, Eq. (1) can be used to obtain the capacitance only in the case of slow and moderate cell potential scan rates (v), when the value of current is very small. At these conditions the potential drop (IR − drop) is negligible and the current response is essentially equal to that of a pure capacitor [3,5,7 − 10,13,17]. Under these conditions in a symmetrical two − electrode system, the active material specific capacitance Cs (F g−1) can be obtained from the capacitance of the completed EDLC cell by Eq. (2). Cs =
2C m
(2)
where m is the the active material weight (g) in one electrode, assuming that the positively and negatively charged electrodes have the same capacitance at fixed cell potential (V) applied. Specific capacitance Cs vs. cell potential V plots in Fig. 2 show that for all EDLC cells consisting of electrospun C-PVDF or C-PTFE composite electrodes and 1 M (C2 H5 )3 CH3 NBF4 + AN electrolyte, so − called nearly ideal capacitive behaviour (i.e. cyclic voltammograms with the nearly rectangular shape) has been established at all potential scan rates v ≤ 100 mV s-1 and cell potential V ≤ 2.7 V (Figs. 2a-c). At higher cell potentials the values of Cs start to increase, being caused by electroreduction of O2 and H2 O traces at a negatively charged electrode and by oxidation of surface functionalities at a positively charged electrode [9]. This is probably caused by the impurities found in the carbon materials, which were used as received. As it has been demonstrated by Laheäär et al. [18] that 1 M (C2 H5 )3 CH3 NBF4 + AN electrolyte is stable in much wider cell potential range V ≤ 3.2 V and based on data in Fig. 2a the electrospun C-PVDF and C-PTFE systems demonstrate similar behaviour. According to the CV data, the specific capacitance depends on the amount of PVDF added into the raw electrode material solution and Cs increases with the decrease of PVDF content in the electrospun C-PVDF composite electrodes (Fig. 2b), except in the case of electrospun C-PVDF (85-15) electrode, which had lower mechanical strength than other electrospun electrodes containing more binder (PVDF).
74
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77
Fig. 1. Scanning electron microscopy images for: (a) and (b) C-PVDF (75-25), (c) and (d) C-PVDF (80-20), (e) and (f) C-PVDF (85-15) composite electrodes, and (g) separator prepared using the electrospinning method.
3.2. Impedance complex plane plots The impedance complex plane (Z”, Z’) plots (so-called Nyquist plots, Fig. 3a) for EDLC completed with electrospun C-PVDF and C-PTFE composite electrodes and 1 M (C2 H5 )3 CH3 NBF4 + AN solution have been measured at fixed cell potentials from 0 to 2.7 V,
within the range of ac frequency f from 1 × 10-2 to 3 × 105 Hz. The shape of Z” vs. Z’ plots depends noticeably on the amount of binder in the electrodes and on the electrode preparation method (Fig. 3a) used. Similarly to other micromesoporous electrodes [6–13], the Nyquist plot consists mainly of three parts: (1) the noticeably depressed semicircle at higher ac frequency,
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77
75
electrolyte conductivity in macro/mesoporous matrix. The addition of PVDF into electrode material changes the wettability (electrowetting properties) of the electrode surface layer, influencing the capacitance of macro/mesoporous surface areas and highmedium frequency polarisation resistance of electrode. Therefore, based on the theoretical model, developed in [10], the semicircle shape is very complex variable and it is determined by the adsorption and mass transfer processes rates of ions at/inside micro/mesoporous carbon electrode, and on the series resistance of a material, mass transfer resistance inside of the macro/mesoporous carbon structure as well as on the mass transfer resistance of ions within the micropores. From the data given in Figs. 3a-b, it can be seen, that the nearly ideal capacitive behaviour has been observed at f ≤ 0.01 Hz for all EDLCs under study. However, based on the Bode phase angle plots the EDLCs completed with optimal C-PVDF (85-15) and C-PVDF (80-20) electrodes (having lower binder content and being thinner) are able to achieve the nearly ideal capacitive behaviour already at the ac frequency 0.1 Hz (Fig. 3b). To characterise further the EDLC properties, the complex power dependencies have been constructed [2,5–7]. The values of complex power can be expressed as S(ω) = P(ω) + jQ (ω),
(3)
where the real part of power
2
P(ω) = ωC"(ω)Vrms ,
(4)
and the imaginary part of power
2
Q (ω) = −ωC (ω)Vrms ,
(5)
2 √ with Vrms = Vmax / 2 (Vmax is the maximal amplitude of ac voltage) [2,6,7]. System with the ideal capacitive behaviour has no real part of the complex power, as there is only the reactive contribution to the complex power, and Eq. (3) simplifies to 2 jVrms 2 = −jωC(ω)Vrms S(ω) = jQ (ω) = − . Z"(ω)
(6)
System with the ideal |Z (ω)| resistive behaviour has no imaginary part of the complex power as this component only dissipates energy and the complex power takes the well-known form
Vrms 2 . S(ω) = Z (ω) Fig. 2. Specific capacitance (Cs ) vs. cell potential (V) curves at (a) scan rate = 1 mV s−1 at different cell potentials and Cs vs. V curves within the cell potential range from 0 to 2.7 V at scan rates (b) = 1 mV s−1 , and (c) = 100 mV s−1 for the EDLC cells based on two identical composite electrodes (noted on figure) in 1 M triethylmethylammonium tetrafluoroborate solution in acetonitrile.
(2) the linear area, so-called “porous” region with a slope of ˛ ≈ -45◦ , characteristic of the mass transfer limited process in the micro/mesoporous matrix of an electrode [8–11,17] and (3) the so-called double-layer capacitance region with a slop of ˛ ≈-90◦ (“knee” at low ac frequencies), obtained by the finite length adsorption effect [8–11,17]. According to the data in Fig. 3a, the size of the depressed semicircle increases with the increase of the binder content in the electrospun electrodes. It has been demonstrated by Eikerling et al. that the shape (i.e. width) of the high-medium frequency semicircle (i.e. high-medium frequency polarization resistance) depends noticeably on the hierarchical structure of the electrode macro/mesoporosity parameters [10]. Increase of PVDF content in carbon composite electrodes layer decreases the macro/mesoporosity of active material layer and thus, the
(7)
It should be also noted that real EDLCs balances between the two states mentioned before: resistive at high frequencies (f → ∞) and capacitive at low frequencies (f → 0). Between these two states EDLC behaves like a resistive-capacitive (RC) transmission line equivalent circuit. At the condition |P| = |Q|, i.e., when phase angle = -45◦ , we can obtain relaxation frequency fr , determining the characteristic time constant r ( r = 2 fr -1 ) of the EDLC cell. Analysis of the data in Figs. 3 and 4 show, that only half of the low-frequency capacitance is reached to establish at r [2,5–7]. Comparison of the data for the cells completed using different electrospun C-PVDF and C-PTFE composite electrodes indicate a noticeable influence of the binder content and cell thickness on the time constant calculated. EDLC based on the thinner electrospun C-PVDF composite electrodes are able to deliver their energy approximately 10 times faster than EDLC based on thick C-PTFE electrodes. However, it should be noted that there is a certain discrepancy between the CV and low frequency impedance data in Fig. 2 and 3a, respectively. This is mainly caused by lower mechanical (including electrochemical) stability of the CPVDF (85-15) electrode compared with C-PVDF (80-20) and with
76
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77
Fig. 5. Ragone plots for the EDLC based on different electrodes (noted in figure) obtained from constant power discharge testes within the cell potential range from 3.0 to 1.5 V.
3.3. Ragone plots
Fig. 3. (a) Nyquist and (b) Bode phase angle plots for the EDLC cell completed with different composite electrodes (noted on figure) at cell potential 2.7 V.
other composite systems under study. Therefore, the long testing impedance measurements after repetitive CV and constant current charge/discharge measurements demonstrated lower capacitance values for this composition due to the electrochemical degradation (expansion of electrodes during charging and compression of electrodes during discharging). In addition the microwetting properties have been exchanged if various wt% of PVDF have been used, influencing the time stability of electrochemical parameters under estimation.
The specific energy (Wh kg-1 ) and power (kW kg-1 ) relationship [2] (calculated taking into account the total active material weight of two electrodes), i.e., Ragone plots, for the EDLCs based on electrospun C-PVDF and C-PTFE electrodes have been obtained from constant power discharge tests within the cell potential range from 3.0 V to 1.5 V and are shown in Fig. 5. The Ragone plots for electrospun composite electrodes depend noticeably on the binder content in the electrodes. Electrospun electrodes containing smaller amount of binder C-PVDF (85-15 and 80-20) deliver noticeably higher power at constant energy. Very high specific energy values E ∼ 30 Wh kg-1 have been achieved at very high power loads P ∼ 100 kW kg-1 for C-PVDF 85-15 and C-PVDF 80-20 composite electrodes based EDLCs. Thus, noticeably better performance for electrospun C-PVDF composite electrodes based EDLCs has been established compared with C-PTFE electrodes based EDLCs. 3.4. Time stability tests It has been demonstrated by D. Weingarth et al. [19] that the constant voltage hold tests are more demanding compared with respective cycling tests (constant voltage charge/discharge cycling) with identical upper voltage limits. Therefore, the constant voltage hold test, i.e., the repetitive holding of capacitor cells at constant potential V = 2.7 V for fixed time t = 5 h, thereafter applying constant current charge/discharge regimes with 2 A g-1 within the cell potential range from 1.35 to 2.7 V and impedance spectroscopy at 2.7 V, have been applied to investigate the electrochemical stability of EDLC cells under study. The discharge and charge capacitances were calculated from data of the third cycle. The capacitance of the cell CCC (F cm-2 ) was obtained from the slope of the discharge curves according to Eq. 8: CCC = j
Fig. 4. Normalized complex power plots and characteristic time constant values (inset) for the EDLC based on different electrodes (noted in figure) at cell potential 2.7 V.
dt , d(V )
(8)
where dt/d(V) is the slop of the discharge curve at applied constant current density j. After 150 hours of constant voltage hold test the longest charge/discharge cycle has been observed for CPVDF (80-20) as well as for C-PTFE (94-6) composite electrodes based EDLCs (Fig. 6), corresponding to the highest specific capacitance calculated from CC-curves: CCC = 92 F g-1 and CCC = 89 F g-1 (Fig. 7), respectively. From inset in Fig. 6, it can be seen, that the C-PTFE (94-6) composite electrodes based EDLC had much bigger IR-drop compared with electrospun C-PVDF (80-20) electrodes based EDLC. Interestingly, the higher electrochemical degradation
K. Tõnurist et al. / Electrochimica Acta 119 (2014) 72–77
77
F g-1 ), very small characteristic time constant (0.043 s) and very high specific power (245 kW kg-1 ) at the average specific energy (22 Wh kg-1 ) applied. It was found that EDLCs, based on the new electrospun composite electrodes are able to deliver much faster and higher power at applied constant energy, than the EDLC based on the electrodes prepared by the traditional roll-pressed method. It was shown that with the novel electrode preparation method micromesoporous composite electrodes with lower thickness and density can be prepared. Acknowledgements
Fig. 6. Constant current charge/discharge cycles at current density j = 2 A g−1 for EDLCs completed using C-PVDF (80-20) and C-PTFE (94-6) electrodes.
Fig. 7. Results of floating tests for the EDLCs, completed using electrospun C-PVDF (85-15) and (80-20), and C-PTFE (94-6) electrodes at cell potential 2.7 V.
rate was observer for electrodes with higher PVDF content in electrospun C-PVDF composite electrodes (not shown for shortness), explained by the PVDF degradation inside the electrode material. 4. Conclusions Composite micromesoporous carbon electrodes based on RP20 + few layered graphene + PVDF binder have been prepared by depositing of composite micromesoporous electrode material directly onto the electrospun supporting PVDF membrane applying electrospinning method. The influence of the electrospun micromesoporous C-PVDF composite electrode characteristics (porosity, thickness), on the electrical double-layer capacitor (EDLC) parameters have been tested using cyclic voltammetry, constant current charge/discharge, constant power discharge, and electrochemical impedance spectroscopy methods in the 1 M triethylmethylammonium tetrafluoroborate in acetonitrile solution. The limits of ideal polarizability (V ≤ 2.7 V), specific capacitance, equivalent series resistance, phase angle, characteristic time constant, specific energy and power of the EDLC completed have been obtained and discussed. The optimum carbon and PVDF binder ratio for electrospun electrodes was established for C-PVDF (8020) system, which demonstrated the high specific capacitance (120
The present study was supported by The Estonian Centers of Excellence in Science: High Technology Materials for Sustainable Development; ETF Projects 8172, 8786; Estonian Ministry of Education and Research project SF0180002s08 and PUT55 project; European Regional Development Fund Project SLOKT10209 T. References [1] K. Tõnurist, T. Thomberg, A. Jänes, I. Kink, E. Lust, Specific performance of electrical double layer capacitors based on different separator materials in room temperature ionic liquid, Electrochem. Comm. 22 (2012) 77. [2] K. Tõnurist, T. Thomberg, A. Jänes, E. Lust, Specific performance of supercapacitors at lower temperatures based on different separator materials, J. Electrochem. Soc. 160 (2013) A449. [3] B.E. Conway, Electrochemical Supercapacitors: Scientific Fundamentals, Technological Applications, Kluwer Academic, Plenum Publishers, New York, 1999. [4] G. Salitra, A. Soffer, L. Eliad, Y. Cohen, D. Aurbach, Carbon electrodes for doublelayer capacitors I. Relations between ion and pore dimensions, J. Electrochem. Soc. 147 (2000) 2486. [5] J. Chmiola, G. Yushin, Y. Gogotsi, C. Portet, P. Simon, L.P. Taberna, Anomalous increase in carbon capacitance at pore sizes less than 1 nanometer, Science 313 (2006) 1760. [6] K. Tõnurist, T. Thomberg, A. Jänes, T. Romann, V. Sammelselg, E. Lust, Influence of separator properties on electrochemical performance of electrical double layer capacitors, J. Electroanal. Chem. 689 (2013) 8. [7] K. Tõnurist, A. Jänes, T. Thomberg, H. Kurig, E. Lust, Influence of mesoporous separator properties on the parameters of electrical double layer capacitor single cells, J. Electrochem. Soc. 156 (2009) A334. [8] A. Jänes, H. Kurig, E. Lust, Characterisation of activated nanoporous carbon for supercapacitor electrode materials, Carbon 45 (2007) 1226. [9] A. Jänes, L. Permann, M. Arulepp, E. Lust, Electrochemical characteristics of nanoporous carbide-derived carbon materials in nonaqueous electrolyte solutions, Electrochem. Comm. 6 (2004) 313. [10] M. Eikerling, A.A. Kornyshev, E. Lust, Optimized structure of nanoporous carbon-based double-layer capacitors, J. Electrochem. Soc. 152 (2005) E24. [11] E. Lust, A. Jänes, M. Arulepp, Influence of electrolyte characteristics on the electrochemical parameters of electrical double layer capacitors, J. Solid State Electrochem. 8 (2004) 488. [12] E. Lust, G. Nurk, A. Jänes, M. Arulepp, P. Nigu, P. Möller, S. Kallip, V. Sammelselg, Electrochemical properties of nanoporous carbon electrodes in various nonaqueous electrolytes, J. Solid State Electrochem. 7 (2003) 91. [13] E. Lust, A. Jänes, T. Pärn, P. Nigu, Influence of nanoporous carbon electrode thickness on the electrochemical characteristics of a nanoporous carbon]tetraethylammonium tetrafluoroborate in acetonitrile solution interface, J. Solid State Electrochem. 8 (2004) 224. [14] A. Formhals, US Patent 197 (1934) 5504. [15] S. Ramakrishna, K. Fujihara, W.-E. Teo, T.-C. Lim, Z. Ma, An Introduction to Electrospinning and Nanofibers, World Scientific Publishing Co. Pte. Ltd, Singapore, 2005. [16] S. Chaudhari, Y. Sharma, P.S. Archana, R. Jose, S. Ramakrishna, S. Mhaisalkar, M. Srinivasan, Electrospun polyaniline nanofibers web electrodes for supercapacitors, J. Appl. Polymer. Sci. 129 (2013) 1660. [17] A. Jänes, T. Thomberg, E. Lust, Synthesis and characterization of nanoporous carbide-derived carbon by chlorination of vanadium carbide, Carbon 45 (2007) 2717. [18] A. Laheäär, A-L. Peikolainen, M. Koel, A. Jänes, E. Lust, Comparison of carbon aerogel and carbide-derived carbon as electrode materials for non-aqueous supercapacitors with high performance, J Solid State Electrochem. 16 (2012) 2717. [19] D. Weingarth, A. Foelske-Schmitz, R. Kötz, Cycle versus voltage hold - Which is the better stability test for electrochemical double layer capacitors? J. Power Sources 225 (2013) 84.