- Email: [email protected]
Design considerations for ionic liquid based electrochemical double layer capacitors
Electrochimica Acta 270 (2018) 453e460
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Design considerations for ionic liquid based electrochemical double layer capacitors Vitor L. Martins a, b, *, Roberto M. Torresi b, Anthony J.R. Rennie a, ** a b
Chemical and Biological Engineering, University of Sheffield, Sir Robert Hadfield Building, Mappin Street, Sheffield S1 3JD, England, UK ~o Paulo, Av. Prof. Lineu Prestes 748, 05508-000 Sa ~o Paulo, SP, Brazil Depto. Química Fundamental, Instituto de Química, Universidade de Sa
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 December 2017 Received in revised form 12 March 2018 Accepted 15 March 2018 Available online 16 March 2018
A critical parameter in the design of EDLCs is the determination of electrochemical stability limits for individual combinations of electrode and electrolyte. Using [EMIm][BF4] as an example we demonstrate that the physical properties of the activated carbon used in the electrodes influence the permissible operating potential of an EDLC, and also that the binder material employed can have further effects. When [EMIm][BF4] electrolyte is coupled with electrodes containing PTFE binder, operating potentials as wide as 3.8 V can be utilised. Due to the differences in response under opposing polarisations, it is necessary to employ an asymmetric mass loading in EDLC cells in order to make full use of the operating window. By balancing the charge in the cell, the differences in stable potential limits and capacitative behaviour can be overcome, however we also found that this balance can be influenced by the rate at which these parameters are determined. Three-electrode measurements show that using an appropriate mass loading ratio results in each electrode operating within their determined stability limits. Stable cycling of a full cell at an operating potential of 3.8 V was demonstrated over 50,000 cycles. © 2018 Elsevier Ltd. All rights reserved.
Keywords: Ionic liquid Supercapacitor Mass balance
1. Introduction Electrochemical energy storage devices are key facilitators for the widespread adoption of renewable energy sources and electrochemical double layer capacitors (EDLCs) are a class of such devices that store energy at the interface between an electrode and electrolyte [1e4]. The electrochemical stability of this interface defines the safe operating potential of such devices and, in turn, determines its energy and power handling characteristics. Electrolytes that are usually found in EDLCs comprise a solution of a conducting salt in an organic solvent and have been demonstrated to withstand operating potentials as high as 3.5 V [5]. Ionic liquids (ILs) have attracted a lot of attention relating to their use as electrolytes due to their large electrochemical stability windows [6e8]. As they do not require any solvent to render them ionically conductive, and typically show a high level of thermal stability, ILs can also be considered as a safer electrolyte for EDLCs [9e11].
* Corresponding author. Chemical and Biological Engineering, University of Sheffield, Sir Robert Hadfield Building, Mappin Street, Sheffield S1 3JD, England, UK. ** Corresponding author. E-mail addresses: [email protected] (V.L. Martins), a.rennie@sheffield.ac.uk (A.J.R. Rennie). https://doi.org/10.1016/j.electacta.2018.03.094 0013-4686/© 2018 Elsevier Ltd. All rights reserved.
Among the most widely studied ILs used as EDLC electrolytes are 1-ethyl-3-methyl tetrafluoroborate ([EMIm][BF4]) [6,12e22] and N-butyl-N-methyl- pyrrolidinium bis(trifluoromethane sulfonyl imide) ([Pyr1,4][Tf2N]) [6,8,23e26]. In the latter case, operating potentials up to 3.9 V have been demonstrated [26], however performance is hindered by the relatively high viscosity and low ionic conductivity of the IL (78 mPa s and 3.0 mS cm-1 at 25 C [27]). [EMIm][BF4] typically operates over a smaller potential window (up to 3.5 V) but performs better at higher rates due to its more favourable physical properties (37 mPa s [16] and 14 mS cm-1 at 25 C [28]). Furthermore, it has been shown that mixtures of ILs can demonstrate improved characteristics when compared to the pure IL. For instance low temperature operation is made possible using IL mixtures [20,29,30]. Alternatively, Van Aken et al. [31] reported that mixtures of ILs can also extended the maximum operating voltage of EDLCs, prolonging the cycle life of symmetrical devices. A crucial parameter when assessing suitability of an IL as an EDLC electrolyte is the determination of a stable operating voltage, however there is no definitive method by which to determine this and as a result, a range of operating potentials have been reported for individual ILs [32]. Typically, stability limits have been identified using a specific current density [6,15,33] or value of coulombic efficiency [27,34e39], both methods that rely on the selection of
454
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
arbitrary cut-off values calibrated toward each individual system. However a more reliable technique for the determination of stable operating potentials has been developed by Weingarth et al. [17]. A modified version of this method is used in this work to investigate the effects that different carbon materials and binders can have on the electrochemical stability of [EMIm][BF4]. In this work it is shown that the electrochemical stability of [EMIm][BF4] in EDLCs is not independent of the electrodes used as the type of activated carbon as well as the nature of the polymeric binder play a major role in the determination of a suitable operating potential. We also show that operating at different rates can result in individual electrodes operating beyond their stability limit.
specific capacitance, CCV, was determined from the charge delivered R during the discharge process, i$dt, the operating potential window, U, and the active mass of both electrodes (total mass of activated carbon in cell), m, as shown in Equation (1):
Z i$dt CCV ¼
U$m
(1)
Cells were cycled galvanostatically between 0 V and the determined operating voltage at different rates from 0.1 to 10 A g1, using a Maccor 4000 M cell test system. Specific capacitance, CGC, was determined considering the current, i, the slope of the discharge curve after iR drop, dV/dt, and the active mass, m, as shown in Equation (2):
2. Experimental [EMIm][BF4] (>99%, ’for electrochemistry’) was purchased from Fluka and was dried by heating at >100 C with vigorous stirring for several hours in an argon-filled glovebox (H2O < 0.1 ppm, O2 < 0.1 ppm). Karl-Fischer titration (KF899, Metrohm) was used to determine the moisture content of the ILs before use and was found to be less than 10 ppm after the drying procedure. EDLC electrodes were prepared by mixing an activated carbon, conductive carbon black (Super C45, Imerys) and PVdF or PVdF-HFP in 80:10:10 ratio by mass using 1-methyl-2-pyrrolidone (anhydrous 99.5%, Sigma-Aldrich). Each slurry was then spread using an adjustable gap paint applicator to a wet film thickness of 150 mm on 15 mm thick aluminum foil. Sheets were dried at 80 C under vacuum prior overnight prior to being punched into individual electrodes with 12 mm diameter. A similar procedure was used to produce electrodes using sodium carboxymethylcellulose (CMC, Sigma-Aldrich) as binder, using deionized water rather than 1methyl-2-pyrrolidone. In order to obtain counter electrodes with at least 10 times the active mass compared with conventional electrodes, the same composition described previously was used with PTFE (Teflon 30-N, Alfa Aesar) as binder in ethanol. This produced self-supporting electrodes with a thickness of approximately 600 mm. (PTFE bound working electrodes were produced by calendering the above mixture to a thickness of roughly 50 mm). Cell were assembled using two electrode button cells (2016) with stainless steel spacers, electrodes, and glass fibre filter paper as separator (GF/F, Whatman). The separator was impregnated with the IL and the cell was placed under vacuum in the glovebox antechamber for at least 5 min to encourage thorough wetting of the electrodes. Cells were sealed inside the glovebox. Swagelok™ type three-electrode cells were assembled using the same materials as for coin cells with 0.5 mm diameter silver wire as a pseudoreference electrode. Electrochemical stability windows were investigated using a modified version of the technique proposed by Weingarth et al. [17], using electrodes with a mass at least 10 times greater than the working electrode, as described elsewhere [27,39,40]. Four cyclic voltammograms were performed in these asymmetric cells from the open circuit potential (OCP) to 0.5 V (vs OCP) at 5 mV s1. After that, the window was increased in increments of 0.1 V to a maximum of þ2.0 V (vs OCP). The same procedure was performed using fresh cells from OCP to 1.0 V (vs OCP), to a maximum of 3.0 V (vs OCP). Dq-values were calculated from the last cycle at each electrochemical window, and stability limits identified by a sharp rise in the value of d2(Dq)/dV2. Cyclic voltammetry was performed using a Solartron Analytical 1470E Multi-channel Potentiostat/Galvanostat. The electrochemical performance of EDLCs were evaluated by cyclic voltammetry using scan rates from 5 to 500 mV s1. The
CGC ¼
i ðdV=dtÞ$m
(2)
Specific energy, Eave and specific power, Pave, were determined from the galvanostatic experiments considering the current, i, operating voltage, U, time of discharge, td, and both electrodes active mass, m, using Equations (3) and (4):
Z Eave ¼ i$
Pave ¼
U $dt m$3:6 d
Eave $3600 td
(3)
(4)
Cycle life was investigated by cycling cells between 0 and the stated operating potential at 2.0 A g1 using a Maccor 4000 M cell test system. Cells were maintained at 25 C using an environmental chamber unless stated otherwise. 3. Results & discussion Five different carbon materials were used to produce electrodes using PVdF binder. Three of the materials (A, B and C) are commercially available carbons used widely in EDLC electrodes. Carbons D and E are described as conductive carbon blacks with high specific surface area. Electrodes containing carbons A, B and C were fabricated using 10% wt. of carbon black and 10% wt. PVdF binder, however carbons D and E required a greater mass fraction of PVdF to produce mechanically stable electrodes and no carbon black was needed. Results from carbons D and E were obtained from electrodes comprising 80% wt. carbon and 20% wt. PVdF. The carbons possess important different physical and textural properties that are considered in the following discussions; they are described and illustrated in Supplementary Material, in Table S1, Figs. S1eS3. 3.1. Electrochemical stability of [EMIm][BF4] with different carbon materials Fig. 1 shows the results of stability determinations for [EMIm] [BF4] conducted in a manner similar to that described by Weingarth et al. [17]. In this method, cells with a highly asymmetrical mass loading are cycled potentiodynamically to increasing vertex potentials, with the quantity of charge passed in each step being recorded. These measurements are then expressed in terms of a stability factor, or ‘S- value’ which is equivalent to 1- (Qdischarge/ Qcharge). However, during the course of this work it was necessary to adjust the criteria to allow for the use of different activated carbon materials and alternative binders, as well as different electrolytes. For this reason, the results are presented in a slightly
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
Fig. 1. Operating potential determination for [EMIm][BF4] using electrodes containing different carbon materials and PVdF binder. (a) Dq vs. vertex potential measured in asymmetric cells at 5 mV s1. Positive/negative polarisations were performed in separate cells at 25 C. (b) Corresponding cyclic voltammograms at the determined stability limits.
different format where the quantity on the y-axis in Fig. 1 represents the difference in the specific quantity of charge associated with the charging and discharging steps of each cyclic voltammogram (i.e. Dq ¼ jqchargejjqdischargej). The plot in Fig. 1 can be thought of to represent the efficiency, and therefore stability, of a given electrode/electrolyte system with all irreversible processes resulting in a deviation in Dq from the ideal of zero. The physical and chemical properties of the carbon material used to produce electrodes are known to influence their electrochemical behaviour, especially when used in EDLCs. For instance, pore characteristics have been shown to have a large impact not only the values of specific capacitance attained with IL electrolytes, but also contributes to their impedance and defines their rate dependent performance [15]. Considering Fig. 1 it is also clear that the characteristics of the carbon used in the electrode can affect the stability of individual ILs. The clearest distinction in Fig. 1 occurs as the potential increases
455
above þ1.0 V (vs OCP) where the behaviour of carbons D and E diverge substantially from the other carbons. This increase in instability may be due to the larger fraction of binder used, however direct comparisons between these carbon black based electrodes and carbons A, B and C should be treated with caution. When polarized negatively all of the carbons appear to behave similarly, with instability increasing after 2.0 V is reached, also there appears to be distinct differences between carbons A, B and C when positively polarized beyond þ1.0 V. There is a significant change in gradient for these carbons with B showing the steepest change and C showing the most gradual as the vertex potential moves away from the OCP. Stability limits determined from Fig. 1 are summarized in Table 1. The limits were defined by considering the second derivative of Dq with respect to vertex potential (similar to the method proposed by Weingarth et al. [17]) which negates the effect of background and/or leakage currents arising from electrolyte resistance or competing self-discharge processes. Normalizing Dq on the basis of electrode mass allows for comparisons between different systems to be made and by considering Dq (rather than coulombic efficiency or ‘S-value’) with respect to vertex potential, changes in behaviour and limits were easier to identify. Limits were identified as the last potential moving away from the OCP before a peak value in d2(Dq)/dV2 was observed. It is thought that such a peak indicates the potential where the current that can be attributed to electrolyte decomposition becomes significant. It is also worth noting that the behaviour of identical cells is highly reproducible until this point and a greater degree of variation between cells is observed after the limit has been exceeded. This is likely due to different products arising from electrolyte decomposition which may be electroactive or may passivate regions of the electrode in subsequent cycles. From Fig. 1 it is seen that carbons D and E produce a more drastic response to increases in potential when compared with the more microporous materials. As the electrolytic decomposition of the ILs involves the transport of species to and from the electrode surface, it is proposed that the presence of a network of mesopores in the electrode facilitates this mass transfer. Once initiated, decomposition of electrolyte within a mesoporous carbon can proceed at a faster rate than in a microporous electrode and it is possible that the relatively high adsorption potentials present in mesopores can contribute not only to electrolyte decomposition but can enhance the electrochemical stability of an electrolyte (in comparison with a planar electrode). Alternative physical differences that could account for the variation in electrochemical stability were explored. For instance, we have previously reported that the particle size distribution of the activated carbon in an electrode can influence the electrochemical stability of ILs [39]. However, this effect was not responsible for the differences in Fig. 1 as the particle size distributions do not show any significant volumes associated with submicron particles (see Fig. S2 in Supplementary Material). It is also possible that the degree of graphitization present in the electrode materials can have an influence on the ESWs determined as it has been shown that different graphitic planes present distinctly different behaviour in the formation of the double layer [41]. A greater number of edge sites may effectively catalyse the decomposition of the electrolyte. Again, this was discounted as an influencing factor in this case as the ratios of disordered to graphitic carbon in each of the carbons were found to be similar using Raman scattering (see Fig. S3 in Supplementary Material). Symmetrical EDLCs were assembled using carbon A with PVdF binder electrodes and [EMIm][BF4] as electrolyte. Cells were then tested at a rate of 2 A g1 using different operating potentials either side of the determined stability limits (i.e. 2.0 V, 2.5 V, 3.0 V and
456
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
Table 1 Operating potential limits determined for [EMIm][BF4] using electrodes containing different carbon materials and/or different binder materials. (cf. Figs. 1 and 3). Carbon/Binder
A/PVdF
B/PVdF
C/PVdF
D/PVdF
E/PVdF
A/PVdF-HFP
A/CMC
A/PTFE
Positive limit, Uþ[V (vs OCP)] Negative limit, U[V (vs OCP)] Total Stability Window [V]
0.9 1.8 2.7
1.1 1.8 2.9
1.0 1.7 2.7
1.1 1.7 2.8
1.0 1.8 2.8
0.9 1.8 2.7
0.7 1.9 2.6
1.7 2.1 3.8
3.5 V). From Table 1 the stability limits of this combination indicate a stable operating potential of 2.7 V, and it can be expected that the specific capacitance should be fairly constant when cycled below this limit and will drop quickly if cycled above this limit. Fig. 2(a) shows that in the initial cycles, higher values of specific capacitance are associated with higher operating potentials, for example at 3.5 V the initial specific capacitance is 26.3 F g1. Fig. 2(b) portrays the same results expressed as a fraction of their initial specific capacitance, highlighting the fact that the specific capacitance drops more rapidly in the initial stages as the operating potential increases. When an operating voltage of 3.5 V is used it can be seen that the specific capacitance drops drastically over the first few hundred cycles indicating that a substantial degree of electrolyte decomposition occurs in the early stages. At an operating potential of 2.5 V, 70% of the initial capacitance was retained and at 2.0 V the specific capacitance is seen to fade gradually upon
Fig. 3. Operating potential determination for [EMIm][BF4] using electrodes containing carbon A and different binder materials.
cycling, retaining 80% of the initial capacitance after 50,000 cycles. Even at the relatively low operating potential of 2.0 V, which should be safely within the ESW, a significant drop in the specific capacitance is seen during the initial stages of cycling in Fig. 2(b). This behaviour can be explained by considering the balance of charge in the cell, i.e. the amount of charge stored in the double layer of the positive electrode must be matched by the quantity of charge in the double layer of the negative electrode. As the assembled cells are symmetrical (where the mass of the positive electrode is equal to the mass of the negative electrode) this means that if the quantity of charge passed to the positive electrode results in the potential exceeding the positive limit, electrolyte decomposition will occur even if the potential across the cell is lower than the determined operating window. Under the assumption that the specific capacitance is the same when the material is charged positively and negatively, this essentially limits the operating potential to twice the value of the lesser limit, therefore in the case of the cells in Fig. 2 the stable operating potential for a symmetrical cell is as low as 1.8 V. 3.2. Electrochemical stability of [EMIm][BF4] with different binder materials
Fig. 2. Capacitance vs. cycle number for symmetrical EDLCs using [EMIm][BF4] electrolyte operating over voltage ranges of 2.0e3.5 V under galvanostatic cycling at 2 A g1. Expressed as (a) specific cell capacitance and (b) a fraction of the initial cell capacitance. (Electrode composition was 80:10:10 of carbon A, carbon black and PVdF binder.)
Clearly an operating potential of 1.8 V is far below that of conventional electrolytes and even if the whole operating windows in Table 1 were to be used, the determined operating potentials fall far below that usually reported for [EMIm][BF4]. From a previous study we reported that when a polymeric IL was used as an electrode binder it could operate reliably at a higher potential [42] by interacting with the electrolyte. In order to investigate the effect of using different inert binders on the stability of [EMIm][BF4], electrodes were produced using some commonly utilised materials; PVdFHFP, CMC and PTFE. Fig. 3 illustrates the results using electrodes containing different binders using the same method as Fig. 1. The most apparent feature
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
in Fig. 3 is that cells using PTFE binder are substantially more stable than those using other binders as these cells were associated with the smallest values of Dq over the whole potential range investigated. At a potential of around þ1.0 V (vs OCP) the behaviour of PTFE bound cells clearly diverges from that of cells using the other binders. The instability at positive potentials for PVdF-based and CMC binders may indicate that reactive species are generated when using [EMIm][BF4] as an electrolyte. The labile proton present on the cation may be responsible for such decomposition reactions as it is known to limit both electrochemical stability and the stability of the dialkylimidazolium cation with metallic lithium [7,26,43]. The generation of protons and radicals is likely to produce intermediates that can degrade the binders, or the imidazolium cations themselves may polymerize. It is also possible that [EMIm] [BF4] acts as a plasticizer when combined with PVdF and PVdF-HFP polymers [44] and, in contrast with the other binder materials studied, PTFE is known to bind particles together by forming a network of fibrils rather than by producing a film [45] which may also influence the results. As a result, the operating potential of devices coupling these binders with [EMIm][BF4] are very limited when compared with those employing PTFE. Stability limits for each of the binders coupled with carbon A were derived from the information in Fig. 3 and are presented in Table 1. It is worth noting that when expressed in terms of S-value, stability limits determined using the criteria that d2S/dV2 > 0.05, namely 2.1 V and þ1.5 V (vs OCP) correlate well with those reported by Weingarth et al. [17] for [EMIm] [BF4]. Electrodes that use PTFE binder show a substantial improvement in operating potential when compared against the other binders studied, with a total of 3.8 V arising from stability limits of 2.1 V and þ1.7 V (vs OCP). 3.3. Asymmetric mass loading However, a feature of the determined stability limits for a vast majority of electrolytes is that they are asymmetrical with respect to the OCP of the cell. Typically, the limit determined for positive deviations from the OCP is lower than that under negative polarisations. To adjust for this and permit the use of wider operating potentials, it has become common practice [26,27,36,37,39,40,46e51] to employ asymmetric mass loadings based on the balance of charge in the cell. This is typically represented by Equation (5) where mþ represents the mass of the positive electrode, cþ the specific capacitance of the positive electrode and Uþ the positive stability limit. (m, c and U represent the same quantities for the negative electrode)
mþ cþ Uþ ¼ m c U
(5)
Equation (5) is often applied under the assumption that the material exhibits the same specific capacitance when either a positive or negative potential is applied. This however is not found often in practice, especially concerning ILs, as the capacitive response from the same material polarized in opposite directions can vary substantially. For example, using the values of discharge capacity determined during cyclic voltammetry, the relationship between specific capacitance and applied potential using [EMIm][BF4] for each of the carbon materials is illustrated in Fig. 4(a). Carbon A clearly displays the highest specific capacitances over the operating potential region (i.e. 2.0e1.2 V (vs OCP)) which can be attributed to this material having a high specific surface area but also containing a significant mesopore volume that facilitates electrolyte access. It is unsurprising that carbon C has a similar specific capacitance to carbon A when positively polarized as this material displays the
457
highest specific surface area. The difference seen during negative polarization may be a result of the incompatibility of the PVdF binder with the IL electrolyte and the charge attributed to the double layer capacitance may actually arise from the decomposition process and/or generation of electroactive species. Carbon D performs poorest, which can be expected as it possesses a low surface area and lesser mesopore volume than carbon E, however it is worth reiterating that these carbons were tested in composite electrodes possessing 20% wt. binder which can be expected to adversely influence their specific capacitances. From Fig. 4(a) it can be seen that in general, as the vertex potential of the cell is swept to increasing values the specific capacitance rises. Increasing electrode potential can be seen as a driving force for the adsorption of ions on the surface and this in turn displaces a greater amount of charge within the electrode. This effect has also been reported by Grey and collaborators [52,53] where NMR and a quartz crystal microbalance showed that the populations of ions in pores are dependent on the cell voltage. The accelerated decomposition of electrolyte in carbon E may result from its relatively high mesopore volume (1.19 cm3 g1). It is likely that improved mass transfer through the mesopore network facilitates the increased rate of IL decomposition in these electrodes. This could result in the formation of a passivating layer on the electrode surface or the blocking of pores, reducing the capacitive response from the electrode. Therefore, the specific capacitance of the material at the electrochemical stability limits, in addition to the potential limits themselves are required to determine suitable mass balance ratios for EDLCs. From Fig. 4(b) it is clear that the electrode binder can also have a substantial effect on the values of specific capacitance attained at a given potential with variations of up to 65 F g1 being observed between the binders studied. It is also clear that larger values of specific capacitance are obtained when the electrode is negatively polarized. It is unsurprising that different specific capacitances are observed when the opposite polarization is applied as the size of the cations and anions present in the IL are significantly different; EMImþ is substantially larger than BF 4. In Fig. 4(b) there are clear differences in the specific capacitance obtained when negatively polarized; PVdF bound electrodes produce values as high as 205 F g1 (at 2.8 V (vs OCP)) whereas PTFE bound electrodes yield 137 F g-1 at the same voltage. As mentioned previously, it is likely that there are different adhesion mechanisms present when using PVdF based binders and PTFE, as the method of electrode production differs; in the former case the polymer is left behind after evaporation of a solvent, whereas in the latter a network of polymer fibrils is established through kneading and mixing. These different methods can be expected to result in the composite electrodes possessing different degrees of electrical conductivity and accessible porosity. Also, polymer binder was added as a constant fraction of mass of the dried composite electrode and differences in binder density is likely to influence the behaviour of composite electrodes. The asymmetry between the positive and negative stability limits is often disregarded when using [EMIm][BF4] electrolyte and symmetrical cells are typically used for EDLC testing. In order to make full use of the operating potential, the use of asymmetrical mass loadings was investigated. In Fig. 4(b) the determined stability limits of 2.1 V and þ1.7 V (vs OCP) for PTFE electrodes correspond with capacitance values of 124.7 F g1 and 111.6 F g1 respectively. However, these values of specific capacitance were determined at a rate of 5 mV s1 which is equivalent to a very low current in a full cell. Specific capacitance is typically seen to diminish with increasing rates of charge, especially when ILs are used as electrolytes, due to the relatively poor transport properties of the electrolyte, which kinetically limits the
458
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
Fig. 4. Discharge specific capacitance at each voltage step in stability determination (a) using different carbons and PVdF as binder and (b) using carbon A and different binders.
charge storage process. As EDLCs are high power devices, typically delivering all of their charge in tens of seconds, it can be expected that the specific capacitances achieved under the expected operating conditions will be substantially lower than those measured using cyclic voltammetry at low sweep rates. Asymmetry between the specific capacitances measured at low rates may be due to the long charge and discharge times that enable small ions to access a greater fraction of the electrode surface than those at higher rates where ions face increased resistance to migration. Therefore, the application of Equation (5) to define a suitable mass loading should use values of specific capacitance that are more representative of full cells operating at practical currents. For example, cells designed using the limits of 2.1 V and þ1.7 V and the corresponding capacitance values determined at 5 mV s1 results in a mass loading ratio, mþ/m, of 1.38, however this would only be suited to a discharge time of over 12 min. Asymmetric cells coupling carbon A with PVdF and PTFE binders were tested using cyclic voltammetry between the OCP and stability limit using a range of sweep rates from 5 to 500 mV s1. The mass ratios calculated from these cyclic voltammograms are illustrated in Fig. 5. In the case of PVdF binder, Fig. 5 shows that a mass loading ratio
Fig. 5. Mass ratio of positive electrode and negative electrode determined at different scan rates, using carbon A and PVdF (black square) or PTFE (green diamond) binders. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
of around 2.3 is suitable for sweep rates lower than 25 mV s1, however as the sweep rate increases towards 500 mV s1 the mass loading ratio is seen to increase. This arises not only due to the asymmetry in stability limits, but also due to the rate dependent values of specific capacitance. In a cell designed using the capacitance values obtained at low rates (i.e. mþ/m ¼ 2.3) the system is balanced when cycling at low rates, however as the rate increases the amount of charge that can be stored on the positive electrode becomes limited and to compensate for this the relative potential across the positive electrode increases. Consequently, this results in the electrode operating outside the stability limits of the electrolyte. Fig. 5 illustrates different behaviour for electrodes using PTFE binder. The required mass loading ratio is much closer to symmetry than for PVdF based electrodes and at low rates a suitable ratio is around 1.3. Also, in contrast with PVdF based electrodes, as the scan rate increases, the mass loading ratio is seen to decrease slightly. This means that if a cell with a mass loading ratio of 1.3 is cycled as higher rates the negative electrode, rather than the positive electrode, becomes limiting. This is evident in Fig. 6 where we employed three-electrode Swagelok™ type cells that enable the potential of each electrode to be measured during cycling vs a Ag quasi-reference electrode. Fig. 6(aec) shows the behaviour of electrodes in asymmetric cells with different mass loading ratios when cycled at 0.1 A g1. In the case of the symmetrical cell in Fig. 6(a) it is seen that the positive electrode operates slightly above the determined stability limit of þ1.7 V as anticipated. When using a ratio of 1.3 in Fig. 6(b) both electrodes remain within their stability limits, and at a ratio of 2.0 the negative electrode becomes limiting and operates below 2.1 V in Fig. 6(c). This effect is more apparent at the higher rate of 1.0 Ag1 shown in Fig. 6(f) which also shows a large increase in iR drop across the positive electrode resulting from the higher mass loading ratio. Although the differences between the determined stability limits appear trivial, these points deviations represent instances where electrolyte decomposition occurs and quickly become evident during charge/discharge cycling. Romann et al. [22] used in-situ IR to show that [EMIm][BF4] decomposes when an EDLC is operating outside the voltage limits, and determined that 2.8 V is stable in a symmetrical cell, and 3.5 V is permissible when using asymmetric cell. It is clear that these maximum operating voltages depend on the carbon and binder utilised. EDLCs performance using asymmetric mass balance is shown in Supplementary Material in Section EDLC Performance and Fig. S4.
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
459
Fig. 6. Potential (vs. OCP) of positive (blue line) and negative (red line) electrodes measured in a two electrode cell with a silver quasi-reference third electrode. Galvanostatic charge-discharge at 0.1 A g1 (aec) and 1.0 A g1 (def) for cells with mþ/m of 1.0 (a, d), 1.3 (b, e) and 2.0 (c, f). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
4. Conclusions
https://doi.org/10.1016/j.electacta.2018.03.094.
In this work we have outlined some of the considerations necessary for the design of EDLCs employing IL electrolytes. Of crucial importance is the determination of electrochemical stability limits for individual combinations of electrode and electrolyte. Using [EMIm][BF4] as an example we demonstrated that the physical properties of the activated carbon used in the electrodes influence the permissible operating potential of an EDLC, and also that the binder material employed can have further effects. It was shown that in the case of [EMIm][BF4], using PTFE binder allows for operating potentials as wide as 3.8 V. Due to the differences in response under opposing polarisations, it is necessary to employ an asymmetric mass loading in EDLC cells in order to make full use of the operating window. By balancing the charge in the cell the differences in stable potential limits and capacitative behaviour can be overcome, however we also found that this balance can be influenced by the rate at which these parameters are determined. Three-electrode measurements show that using a mass loading ratio of 1.3 results in each electrode operating within their determined stability limits and stable cycling of a full cell at an operating potential of 3.8 V was demonstrated over 50,000 cycles. Premature ageing of the cell can result from only considering the capacitance at low rates.
References
Acknowledgements The authors would like to thank the EPSRC (EP/K021192/1) and FAPESP (15/26308-7, 13/22748-7 and 14/14690-1) for funding. Appendix A. Supplementary data Supplementary data related to this article can be found at
€tz, M. Carlen, Principles and applications of electrochemical capacitors, [1] R. Ko Electrochim. Acta 45 (2000) 2483e2498, https://doi.org/10.1016/S00134686(00)00354-6. [2] P.J. Hall, M. Mirzaeian, S.I. Fletcher, F.B. Sillars, A.J.R. Rennie, G.O. Shitta-Bey, et al., Energy storage in electrochemical capacitors: designing functional materials to improve performance, Energy Environ. Sci. 3 (2010) 1238e1251, https://doi.org/10.1039/c0ee00004c. [3] P. Simon, Y. Gogotsi, Capacitive energy storage in nanostructured carboneelectrolyte systems, Acc. Chem. Res. 46 (2013) 1094e1103, https:// doi.org/10.1021/ar200306b. guin, V. Presser, A. Balducci, E. Frackowiak, Carbons and electrolytes for [4] F. Be advanced supercapacitors, Adv. Mater. 26 (2014) 2219e2251, https://doi.org/ 10.1002/adma.201304137. [5] A. Krause, A. Balducci, High voltage electrochemical double layer capacitor containing mixtures of ionic liquids and organic carbonate as electrolytes, Electrochem. Commun. 13 (2011) 814e817, https://doi.org/10.1016/ j.elecom.2011.05.010. ski, A. Lewandowski, I. Stepniak, Ionic liquids as electrolytes, Elec[6] M. Galin trochim. Acta 51 (2006) 5567e5580, https://doi.org/10.1016/ j.electacta.2006.03.016. [7] M. Armand, F. Endres, D.R. MacFarlane, H. Ohno, B. Scrosati, Ionic-liquid materials for the electrochemical challenges of the future, Nat. Mater. 8 (2009) 621e629, https://doi.org/10.1038/nmat2448. [8] A. Eftekhari, Supercapacitors utilising ionic liquids, Energy Storage Mater 9 (2017) 47e69, https://doi.org/10.1016/j.ensm.2017.06.009. [9] D.R. MacFarlane, M. Forsyth, P.C. Howlett, M. Kar, S. Passerini, J.M. Pringle, et al., Ionic liquids and their solid-state analogues as materials for energy generation and storage, Nat. Rev. Mater 1 (2016) 15005, https://doi.org/10.1038/ natrevmats.2015.5. [10] M. Salanne, Ionic liquids for supercapacitor applications, Top. Curr. Chem. 375 (2017) 63, https://doi.org/10.1007/s41061-017-0150-7. [11] D.R. MacFarlane, A.L. Chong, M. Forsyth, M. Kar, R. Vijayaraghavan, A. Somers, et al., New dimensions in saltesolvent mixtures: a 4th evolution of ionic liquids, Faraday Discuss (2017), https://doi.org/10.1039/C7FD00189D. €nes, E. Lust, Electrochemical characteristics of carbide-derived [12] H. Kurig, A. Ja carbonr1-ethyl-3-methylimidazolium tetrafluoroborate supercapacitor cells, J. Electrochem. Soc. 157 (2010) A272, https://doi.org/10.1149/1.3274208. €nes, E. Lust, Electrical double layer capacitors based on [13] H. Kurig, M. Vestli, A. Ja
460
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460 two 1-ethyl-3-methylimidazolium ionic liquids with different anions, Electrochem. Solid state Lett. 14 (2011) A120, https://doi.org/10.1149/1.3596722. H. Kurig, M. Vestli, K. Tonurist, A. Janes, E. Lust, Influence of room temperature ionic liquid anion chemical composition and electrical charge delocalization on the supercapacitor properties, J. Electrochem. Soc. 159 (2012) A944eA951, https://doi.org/10.1149/2.022207jes. F.B. Sillars, S.I. Fletcher, M. Mirzaeian, P.J. Hall, Effect of activated carbon xerogel pore size on the capacitance performance of ionic liquid electrolytes, Energy Environ. Sci. 4 (2011) 695e706, https://doi.org/10.1039/C0EE00337A. F.B. Sillars, S.I. Fletcher, M. Mirzaeian, P.J. Hall, Variation of electrochemical capacitor performance with room temperature ionic liquid electrolyte viscosity and ion size, Phys. Chem. Chem. Phys. 14 (2012) 6094, https://doi.org/ 10.1039/c2cp40089h. €tz, A reliable D. Weingarth, H. Noh, A. Foelske-Schmitz, A. Wokaun, R. Ko determination method of stability limits for electrochemical double layer capacitors, Electrochim. Acta 103 (2013) 119e124, https://doi.org/10.1016/ j.electacta.2013.04.057. R. Palm, H. Kurig, K. Tonurist, A. Janes, E. Lust, Influence of different organic solvent additives on 1-ethyl-3-methylimidazolium tetrafluoroborate electrolyte based electrical double layer capacitors, J. Electrochem. Soc. 160 (2013) A1741eA1745, https://doi.org/10.1149/2.046310jes. ~ nurist, A. Ja €nes, E. Lust, Electrical double layer capacR. Palm, H. Kurig, K. To itors based on 1-ethyl-3-methylimidazolium tetrafluoroborate with small addition of acetonitrile, Electrochim. Acta 85 (2012) 139e144, https://doi.org/ 10.1016/j.electacta.2012.08.030. ~ nurist, A. J€ R. Palm, H. Kurig, K. To anes, E. Lust, Is the mixture of 1-ethyl-3methylimidazolium tetrafluoroborate and 1-butyl-3-methylimidazolium tetrafluoroborate applicable as electrolyte in electrical double layer capacitors? Electrochem. Commun. 22 (2012) 203e206, https://doi.org/10.1016/ j.elecom.2012.06.029. ~ nurist, T. Thomberg, A. Ja €nes, T. Romann, V. Sammelselg, E. Lust, InfluK. To ence of separator properties on electrochemical performance of electrical double-layer capacitors, J. Electroanal. Chem. 689 (2013) 8e20, https:// doi.org/10.1016/j.jelechem.2012.11.024. T. Romann, O. Oll, P. Pikma, H. Tamme, E. Lust, Surface chemistry of carbon electrodes in 1-ethyl-3-methylimidazolium tetrafluoroborate ionic liquid e an in situ infrared study, Electrochim. Acta 125 (2014) 183e190, https:// doi.org/10.1016/j.electacta.2014.01.077. M. Lazzari, F. Soavi, M. Mastragostino, Mesoporous carbon design for ionic liquid-based, double-layer supercapacitors, Fuel Cell. 10 (2010) 840e847, https://doi.org/10.1002/fuce.200900198. ~ oz-Torrero, J. Palma, M. Anderson, R. Marcilla, AllG. Ayalneh Tiruye, D. Mun solid state supercapacitors operating at 3.5 V by using ionic liquid based polymer electrolytes, J. Power Sources 279 (2015) 472e480, https://doi.org/ 10.1016/j.jpowsour.2015.01.039. e, M. Mastragostino, et al., A. Balducci, R. Dugas, P.L.L. Taberna, P. Simon, D. Ple High temperature carbonecarbon supercapacitor using ionic liquid as electrolyte, J. Power Sources 165 (2007) 922e927, https://doi.org/10.1016/ j.jpowsour.2006.12.048. M. Lazzari, F. Soavi, M. Mastragostino, High voltage, asymmetric EDLCs based on xerogel carbon and hydrophobic IL electrolytes, J. Power Sources 178 (2008) 490e496, https://doi.org/10.1016/j.jpowsour.2007.12.029. A.J.R. Rennie, V.L. Martins, R.M. Torresi, P.J. Hall, Ionic liquids containing sulfonium cations as electrolytes for electrochemical double layer capacitors, J. Phys. Chem. C 119 (2015) 23865e23874, https://doi.org/10.1021/ acs.jpcc.5b08241. T. Nishida, Y. Tashiro, M. Yamamoto, Physical and electrochemical properties of 1-alkyl-3-methylimidazolium tetrafluoroborate for electrolyte, J. Fluorine Chem. 120 (2003) 135e141, https://doi.org/10.1016/S0022-1139(02)00322-6. rez, F. Malbosc, et al., R. Lin, P.-L. Taberna, S. Fantini, V. Presser, C.R. Pe Capacitive energy storage from 50 to 100 C using an ionic liquid electrolyte, J. Phys. Chem. Lett. 2 (2011) 2396e2401, https://doi.org/10.1021/jz201065t. C. Lian, K. Liu, K.L. Van Aken, Y. Gogotsi, D.J. Wesolowski, H.L. Liu, et al., Enhancing the capacitive performance of electric double-layer capacitors with ionic liquid mixtures, ACS Energy Lett. 1 (2016) 21e26, https://doi.org/ 10.1021/acsenergylett.6b00010. K.L. Van Aken, M. Beidaghi, Y. Gogotsi, K.L. Van Aken, M. Beidaghi, Y. Gogotsi, Formulation of ionic-liquid electrolyte to expand the voltage window of supercapacitors, Angew. Chem. Int. Ed. 54 (2015) 4806e4809, https://doi.org/ 10.1002/anie.201412257. A. Balducci, D. Belanger, T. Brousse, J.W. Long, W. Sugimoto, Perspectiveda guideline for reporting performance metrics with electrochemical capacitors: from electrode materials to full devices, J. Electrochem. Soc. 164 (2017) A1487eA1488, https://doi.org/10.1149/2.0851707jes. M. Ue, M. Takeda, M. Takehara, S. Mori, Electrochemical properties of
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
quaternary ammonium salts for electrochemical capacitors, J. Electrochem. Soc. 144 (1997) 2684, https://doi.org/10.1149/1.1837882. K. Xu, M.S. Ding, T. Richard Jow, Better quantification of electrochemical stability limits for electrolytes in double layer capacitors, Electrochim. Acta 46 (2001) 1823e1827, https://doi.org/10.1016/S0013-4686(01)00358-9. K. Xu, S.P. Ding, T. Richard Jow, Toward reliable values of electrochemical stability limits for electrolytes, J. Electrochem. Soc. 146 (1999) 4172e4178, https://doi.org/10.1149/1.1392609. €ger, P. Goodrich, J. Alvarez Vicente, J. Jacquemin, S. Pohlmann, T. Olyschla A. Balducci, Azepanium-based ionic liquids as green electrolytes for high voltage supercapacitors, J. Power Sources 273 (2015) 931e936, https:// doi.org/10.1016/j.jpowsour.2014.09.167. C. Wolff, S. Jeong, E. Paillard, A. Balducci, S. Passerini, High power, solvent-free electrochemical double layer capacitors based on pyrrolidinium dicyanamide ionic liquids, J. Power Sources 293 (2015) 65e70, https://doi.org/10.1016/ j.jpowsour.2015.05.065. A.R. Neale, S. Murphy, P. Goodrich, C. Schütter, C. Hardacre, S. Passerini, et al., An ether-functionalised cyclic sulfonium based ionic liquid as an electrolyte for electrochemical double layer capacitors, J. Power Sources 326 (2016) 549e559, https://doi.org/10.1016/j.jpowsour.2016.06.085. A.J.R. Rennie, V.L. Martins, R.M. Smith, P.J. Hall, Influence of particle size distribution on the performance of ionic liquid-based electrochemical double layer capacitors, Sci. Rep. 6 (2016) 22062, https://doi.org/10.1038/srep22062. V.L. Martins, A.J.R. Rennie, R.M. Torresi, P.J. Hall, Ionic liquids containing tricyanomethanide anions: physicochemical characterisation and performance as electrochemical double-layer capacitor electrolytes, Phys. Chem. Chem. Phys. 19 (2017) 16867e16874, https://doi.org/10.1039/C7CP03377J. J.P. Randin, E. Yeager, Differential capacitance study of stress- annealed pyrolytic graphite electrodes, J. Electrochem. Soc. 118 (1971) 711, https:// doi.org/10.1149/1.2408151. V.L. Martins, A.J.R. Rennie, J. Lesowiec, R.M. Torresi, P.J. Hall, Using polymeric ionic liquids as an active binder in supercapacitors, J. Electrochem. Soc. 164 (2017) A3253eA3258, https://doi.org/10.1149/2.1411713jes. F.F.C. Bazito, Y. Kawano, R.M. Torresi, Synthesis and characterization of two ionic liquids with emphasis on their chemical stability towards metallic lithium, Electrochim. Acta 52 (2007) 6427e6437, https://doi.org/10.1016/ j.electacta.2007.04.064. L. Guo, Y. Liu, C. Zhang, J. Chen, Preparation of PVDF-based polymer inclusion membrane using ionic liquid plasticizer and Cyphos IL 104 Carrier for Cr(VI) transport, J. Membr. Sci. 372 (2011) 314e321, https://doi.org/10.1016/ j.memsci.2011.02.014. T. Kitamura, K.-I. Kurumada, M. Tanigaki, M. Ohshima, S.-I. Kanazawa, Formation mechanism of porous structure in polytetrafluoroethylene (PTFE) porous membrane through mechanical operations, Polym. Eng. Sci. 39 (1999) 2256e2263, https://doi.org/10.1002/pen.11613. Z. Sun, Z. Liu, B. Han, S. Miao, J. Du, Z. Miao, Microstructural and electrochemical characterization of RuO2/CNT composites synthesized in supercritical diethyl amine, Carbon N. Y. 44 (2006) 888e893, https://doi.org/10.1016/ j.carbon.2005.10.023. ~ ero, E. Frackowiak, F. Be guin, High-voltage V. Khomenko, E. Raymundo-Pin asymmetric supercapacitors operating in aqueous electrolyte, Appl. Phys. A Mater. Sci. Process 82 (2006) 567e573, https://doi.org/10.1007/s00339-0053397-8. S. Vaquero, J. Palma, M. Anderson, R. Marcilla, Mass-balancing of electrodes as a strategy to widen the operating voltage window of carbon/carbon supercapacitors in neutral aqueous electrolytes, Int. J. Electrochem. Sci. 8 (2013) 10293e10307. S. Vaquero, J. Palma, M. Anderson, R. Marcilla, Improving performance of electric double layer capacitors with a mixture of ionic liquid and acetonitrile as the electrolyte by using mass-balancing carbon electrodes, J. Electrochem. Soc. 160 (2013) A2064eA2069, https://doi.org/10.1149/2.063311jes. Z. Dai, C. Peng, J.H. Chae, K.C. Ng, G.Z. Chen, Cell voltage versus electrode potential range in aqueous supercapacitors, Sci. Rep. 5 (2015), https://doi.org/ 10.1038/srep09854. V.L. Martins, A.J.R. Rennie, N. Sanchez-Ramirez, R.M. Torresi, P.J. Hall, Improved performance of ionic liquid supercapacitors by using tetracyanoborate anions, ChemElectroChem 5 (2018) 598e604, https://doi.org/10.1002/ celc.201701164. A.C. Forse, J.M. Griffin, C. Merlet, P.M. Bayley, H. Wang, P. Simon, et al., NMR study of ion dynamics and charge storage in ionic liquid supercapacitors, J. Am. Chem. Soc. (2015), https://doi.org/10.1021/jacs.5b03958. J.M. Griffin, A.C. Forse, W.-Y. Tsai, P.-L. Taberna, P. Simon, C.P. Grey, In situ NMR and electrochemical quartz crystal microbalance techniques reveal the structure of the electrical double layer in supercapacitors, Nat. Mater. 14 (2015) 812e819, https://doi.org/10.1038/nmat4318.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Design considerations for ionic liquid based electrochemical double layer capacitors Vitor L. Martins a, b, *, Roberto M. Torresi b, Anthony J.R. Rennie a, ** a b
Chemical and Biological Engineering, University of Sheffield, Sir Robert Hadfield Building, Mappin Street, Sheffield S1 3JD, England, UK ~o Paulo, Av. Prof. Lineu Prestes 748, 05508-000 Sa ~o Paulo, SP, Brazil Depto. Química Fundamental, Instituto de Química, Universidade de Sa
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 December 2017 Received in revised form 12 March 2018 Accepted 15 March 2018 Available online 16 March 2018
A critical parameter in the design of EDLCs is the determination of electrochemical stability limits for individual combinations of electrode and electrolyte. Using [EMIm][BF4] as an example we demonstrate that the physical properties of the activated carbon used in the electrodes influence the permissible operating potential of an EDLC, and also that the binder material employed can have further effects. When [EMIm][BF4] electrolyte is coupled with electrodes containing PTFE binder, operating potentials as wide as 3.8 V can be utilised. Due to the differences in response under opposing polarisations, it is necessary to employ an asymmetric mass loading in EDLC cells in order to make full use of the operating window. By balancing the charge in the cell, the differences in stable potential limits and capacitative behaviour can be overcome, however we also found that this balance can be influenced by the rate at which these parameters are determined. Three-electrode measurements show that using an appropriate mass loading ratio results in each electrode operating within their determined stability limits. Stable cycling of a full cell at an operating potential of 3.8 V was demonstrated over 50,000 cycles. © 2018 Elsevier Ltd. All rights reserved.
Keywords: Ionic liquid Supercapacitor Mass balance
1. Introduction Electrochemical energy storage devices are key facilitators for the widespread adoption of renewable energy sources and electrochemical double layer capacitors (EDLCs) are a class of such devices that store energy at the interface between an electrode and electrolyte [1e4]. The electrochemical stability of this interface defines the safe operating potential of such devices and, in turn, determines its energy and power handling characteristics. Electrolytes that are usually found in EDLCs comprise a solution of a conducting salt in an organic solvent and have been demonstrated to withstand operating potentials as high as 3.5 V [5]. Ionic liquids (ILs) have attracted a lot of attention relating to their use as electrolytes due to their large electrochemical stability windows [6e8]. As they do not require any solvent to render them ionically conductive, and typically show a high level of thermal stability, ILs can also be considered as a safer electrolyte for EDLCs [9e11].
* Corresponding author. Chemical and Biological Engineering, University of Sheffield, Sir Robert Hadfield Building, Mappin Street, Sheffield S1 3JD, England, UK. ** Corresponding author. E-mail addresses: [email protected] (V.L. Martins), a.rennie@sheffield.ac.uk (A.J.R. Rennie). https://doi.org/10.1016/j.electacta.2018.03.094 0013-4686/© 2018 Elsevier Ltd. All rights reserved.
Among the most widely studied ILs used as EDLC electrolytes are 1-ethyl-3-methyl tetrafluoroborate ([EMIm][BF4]) [6,12e22] and N-butyl-N-methyl- pyrrolidinium bis(trifluoromethane sulfonyl imide) ([Pyr1,4][Tf2N]) [6,8,23e26]. In the latter case, operating potentials up to 3.9 V have been demonstrated [26], however performance is hindered by the relatively high viscosity and low ionic conductivity of the IL (78 mPa s and 3.0 mS cm-1 at 25 C [27]). [EMIm][BF4] typically operates over a smaller potential window (up to 3.5 V) but performs better at higher rates due to its more favourable physical properties (37 mPa s [16] and 14 mS cm-1 at 25 C [28]). Furthermore, it has been shown that mixtures of ILs can demonstrate improved characteristics when compared to the pure IL. For instance low temperature operation is made possible using IL mixtures [20,29,30]. Alternatively, Van Aken et al. [31] reported that mixtures of ILs can also extended the maximum operating voltage of EDLCs, prolonging the cycle life of symmetrical devices. A crucial parameter when assessing suitability of an IL as an EDLC electrolyte is the determination of a stable operating voltage, however there is no definitive method by which to determine this and as a result, a range of operating potentials have been reported for individual ILs [32]. Typically, stability limits have been identified using a specific current density [6,15,33] or value of coulombic efficiency [27,34e39], both methods that rely on the selection of
454
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
arbitrary cut-off values calibrated toward each individual system. However a more reliable technique for the determination of stable operating potentials has been developed by Weingarth et al. [17]. A modified version of this method is used in this work to investigate the effects that different carbon materials and binders can have on the electrochemical stability of [EMIm][BF4]. In this work it is shown that the electrochemical stability of [EMIm][BF4] in EDLCs is not independent of the electrodes used as the type of activated carbon as well as the nature of the polymeric binder play a major role in the determination of a suitable operating potential. We also show that operating at different rates can result in individual electrodes operating beyond their stability limit.
specific capacitance, CCV, was determined from the charge delivered R during the discharge process, i$dt, the operating potential window, U, and the active mass of both electrodes (total mass of activated carbon in cell), m, as shown in Equation (1):
Z i$dt CCV ¼
U$m
(1)
Cells were cycled galvanostatically between 0 V and the determined operating voltage at different rates from 0.1 to 10 A g1, using a Maccor 4000 M cell test system. Specific capacitance, CGC, was determined considering the current, i, the slope of the discharge curve after iR drop, dV/dt, and the active mass, m, as shown in Equation (2):
2. Experimental [EMIm][BF4] (>99%, ’for electrochemistry’) was purchased from Fluka and was dried by heating at >100 C with vigorous stirring for several hours in an argon-filled glovebox (H2O < 0.1 ppm, O2 < 0.1 ppm). Karl-Fischer titration (KF899, Metrohm) was used to determine the moisture content of the ILs before use and was found to be less than 10 ppm after the drying procedure. EDLC electrodes were prepared by mixing an activated carbon, conductive carbon black (Super C45, Imerys) and PVdF or PVdF-HFP in 80:10:10 ratio by mass using 1-methyl-2-pyrrolidone (anhydrous 99.5%, Sigma-Aldrich). Each slurry was then spread using an adjustable gap paint applicator to a wet film thickness of 150 mm on 15 mm thick aluminum foil. Sheets were dried at 80 C under vacuum prior overnight prior to being punched into individual electrodes with 12 mm diameter. A similar procedure was used to produce electrodes using sodium carboxymethylcellulose (CMC, Sigma-Aldrich) as binder, using deionized water rather than 1methyl-2-pyrrolidone. In order to obtain counter electrodes with at least 10 times the active mass compared with conventional electrodes, the same composition described previously was used with PTFE (Teflon 30-N, Alfa Aesar) as binder in ethanol. This produced self-supporting electrodes with a thickness of approximately 600 mm. (PTFE bound working electrodes were produced by calendering the above mixture to a thickness of roughly 50 mm). Cell were assembled using two electrode button cells (2016) with stainless steel spacers, electrodes, and glass fibre filter paper as separator (GF/F, Whatman). The separator was impregnated with the IL and the cell was placed under vacuum in the glovebox antechamber for at least 5 min to encourage thorough wetting of the electrodes. Cells were sealed inside the glovebox. Swagelok™ type three-electrode cells were assembled using the same materials as for coin cells with 0.5 mm diameter silver wire as a pseudoreference electrode. Electrochemical stability windows were investigated using a modified version of the technique proposed by Weingarth et al. [17], using electrodes with a mass at least 10 times greater than the working electrode, as described elsewhere [27,39,40]. Four cyclic voltammograms were performed in these asymmetric cells from the open circuit potential (OCP) to 0.5 V (vs OCP) at 5 mV s1. After that, the window was increased in increments of 0.1 V to a maximum of þ2.0 V (vs OCP). The same procedure was performed using fresh cells from OCP to 1.0 V (vs OCP), to a maximum of 3.0 V (vs OCP). Dq-values were calculated from the last cycle at each electrochemical window, and stability limits identified by a sharp rise in the value of d2(Dq)/dV2. Cyclic voltammetry was performed using a Solartron Analytical 1470E Multi-channel Potentiostat/Galvanostat. The electrochemical performance of EDLCs were evaluated by cyclic voltammetry using scan rates from 5 to 500 mV s1. The
CGC ¼
i ðdV=dtÞ$m
(2)
Specific energy, Eave and specific power, Pave, were determined from the galvanostatic experiments considering the current, i, operating voltage, U, time of discharge, td, and both electrodes active mass, m, using Equations (3) and (4):
Z Eave ¼ i$
Pave ¼
U $dt m$3:6 d
Eave $3600 td
(3)
(4)
Cycle life was investigated by cycling cells between 0 and the stated operating potential at 2.0 A g1 using a Maccor 4000 M cell test system. Cells were maintained at 25 C using an environmental chamber unless stated otherwise. 3. Results & discussion Five different carbon materials were used to produce electrodes using PVdF binder. Three of the materials (A, B and C) are commercially available carbons used widely in EDLC electrodes. Carbons D and E are described as conductive carbon blacks with high specific surface area. Electrodes containing carbons A, B and C were fabricated using 10% wt. of carbon black and 10% wt. PVdF binder, however carbons D and E required a greater mass fraction of PVdF to produce mechanically stable electrodes and no carbon black was needed. Results from carbons D and E were obtained from electrodes comprising 80% wt. carbon and 20% wt. PVdF. The carbons possess important different physical and textural properties that are considered in the following discussions; they are described and illustrated in Supplementary Material, in Table S1, Figs. S1eS3. 3.1. Electrochemical stability of [EMIm][BF4] with different carbon materials Fig. 1 shows the results of stability determinations for [EMIm] [BF4] conducted in a manner similar to that described by Weingarth et al. [17]. In this method, cells with a highly asymmetrical mass loading are cycled potentiodynamically to increasing vertex potentials, with the quantity of charge passed in each step being recorded. These measurements are then expressed in terms of a stability factor, or ‘S- value’ which is equivalent to 1- (Qdischarge/ Qcharge). However, during the course of this work it was necessary to adjust the criteria to allow for the use of different activated carbon materials and alternative binders, as well as different electrolytes. For this reason, the results are presented in a slightly
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
Fig. 1. Operating potential determination for [EMIm][BF4] using electrodes containing different carbon materials and PVdF binder. (a) Dq vs. vertex potential measured in asymmetric cells at 5 mV s1. Positive/negative polarisations were performed in separate cells at 25 C. (b) Corresponding cyclic voltammograms at the determined stability limits.
different format where the quantity on the y-axis in Fig. 1 represents the difference in the specific quantity of charge associated with the charging and discharging steps of each cyclic voltammogram (i.e. Dq ¼ jqchargejjqdischargej). The plot in Fig. 1 can be thought of to represent the efficiency, and therefore stability, of a given electrode/electrolyte system with all irreversible processes resulting in a deviation in Dq from the ideal of zero. The physical and chemical properties of the carbon material used to produce electrodes are known to influence their electrochemical behaviour, especially when used in EDLCs. For instance, pore characteristics have been shown to have a large impact not only the values of specific capacitance attained with IL electrolytes, but also contributes to their impedance and defines their rate dependent performance [15]. Considering Fig. 1 it is also clear that the characteristics of the carbon used in the electrode can affect the stability of individual ILs. The clearest distinction in Fig. 1 occurs as the potential increases
455
above þ1.0 V (vs OCP) where the behaviour of carbons D and E diverge substantially from the other carbons. This increase in instability may be due to the larger fraction of binder used, however direct comparisons between these carbon black based electrodes and carbons A, B and C should be treated with caution. When polarized negatively all of the carbons appear to behave similarly, with instability increasing after 2.0 V is reached, also there appears to be distinct differences between carbons A, B and C when positively polarized beyond þ1.0 V. There is a significant change in gradient for these carbons with B showing the steepest change and C showing the most gradual as the vertex potential moves away from the OCP. Stability limits determined from Fig. 1 are summarized in Table 1. The limits were defined by considering the second derivative of Dq with respect to vertex potential (similar to the method proposed by Weingarth et al. [17]) which negates the effect of background and/or leakage currents arising from electrolyte resistance or competing self-discharge processes. Normalizing Dq on the basis of electrode mass allows for comparisons between different systems to be made and by considering Dq (rather than coulombic efficiency or ‘S-value’) with respect to vertex potential, changes in behaviour and limits were easier to identify. Limits were identified as the last potential moving away from the OCP before a peak value in d2(Dq)/dV2 was observed. It is thought that such a peak indicates the potential where the current that can be attributed to electrolyte decomposition becomes significant. It is also worth noting that the behaviour of identical cells is highly reproducible until this point and a greater degree of variation between cells is observed after the limit has been exceeded. This is likely due to different products arising from electrolyte decomposition which may be electroactive or may passivate regions of the electrode in subsequent cycles. From Fig. 1 it is seen that carbons D and E produce a more drastic response to increases in potential when compared with the more microporous materials. As the electrolytic decomposition of the ILs involves the transport of species to and from the electrode surface, it is proposed that the presence of a network of mesopores in the electrode facilitates this mass transfer. Once initiated, decomposition of electrolyte within a mesoporous carbon can proceed at a faster rate than in a microporous electrode and it is possible that the relatively high adsorption potentials present in mesopores can contribute not only to electrolyte decomposition but can enhance the electrochemical stability of an electrolyte (in comparison with a planar electrode). Alternative physical differences that could account for the variation in electrochemical stability were explored. For instance, we have previously reported that the particle size distribution of the activated carbon in an electrode can influence the electrochemical stability of ILs [39]. However, this effect was not responsible for the differences in Fig. 1 as the particle size distributions do not show any significant volumes associated with submicron particles (see Fig. S2 in Supplementary Material). It is also possible that the degree of graphitization present in the electrode materials can have an influence on the ESWs determined as it has been shown that different graphitic planes present distinctly different behaviour in the formation of the double layer [41]. A greater number of edge sites may effectively catalyse the decomposition of the electrolyte. Again, this was discounted as an influencing factor in this case as the ratios of disordered to graphitic carbon in each of the carbons were found to be similar using Raman scattering (see Fig. S3 in Supplementary Material). Symmetrical EDLCs were assembled using carbon A with PVdF binder electrodes and [EMIm][BF4] as electrolyte. Cells were then tested at a rate of 2 A g1 using different operating potentials either side of the determined stability limits (i.e. 2.0 V, 2.5 V, 3.0 V and
456
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
Table 1 Operating potential limits determined for [EMIm][BF4] using electrodes containing different carbon materials and/or different binder materials. (cf. Figs. 1 and 3). Carbon/Binder
A/PVdF
B/PVdF
C/PVdF
D/PVdF
E/PVdF
A/PVdF-HFP
A/CMC
A/PTFE
Positive limit, Uþ[V (vs OCP)] Negative limit, U[V (vs OCP)] Total Stability Window [V]
0.9 1.8 2.7
1.1 1.8 2.9
1.0 1.7 2.7
1.1 1.7 2.8
1.0 1.8 2.8
0.9 1.8 2.7
0.7 1.9 2.6
1.7 2.1 3.8
3.5 V). From Table 1 the stability limits of this combination indicate a stable operating potential of 2.7 V, and it can be expected that the specific capacitance should be fairly constant when cycled below this limit and will drop quickly if cycled above this limit. Fig. 2(a) shows that in the initial cycles, higher values of specific capacitance are associated with higher operating potentials, for example at 3.5 V the initial specific capacitance is 26.3 F g1. Fig. 2(b) portrays the same results expressed as a fraction of their initial specific capacitance, highlighting the fact that the specific capacitance drops more rapidly in the initial stages as the operating potential increases. When an operating voltage of 3.5 V is used it can be seen that the specific capacitance drops drastically over the first few hundred cycles indicating that a substantial degree of electrolyte decomposition occurs in the early stages. At an operating potential of 2.5 V, 70% of the initial capacitance was retained and at 2.0 V the specific capacitance is seen to fade gradually upon
Fig. 3. Operating potential determination for [EMIm][BF4] using electrodes containing carbon A and different binder materials.
cycling, retaining 80% of the initial capacitance after 50,000 cycles. Even at the relatively low operating potential of 2.0 V, which should be safely within the ESW, a significant drop in the specific capacitance is seen during the initial stages of cycling in Fig. 2(b). This behaviour can be explained by considering the balance of charge in the cell, i.e. the amount of charge stored in the double layer of the positive electrode must be matched by the quantity of charge in the double layer of the negative electrode. As the assembled cells are symmetrical (where the mass of the positive electrode is equal to the mass of the negative electrode) this means that if the quantity of charge passed to the positive electrode results in the potential exceeding the positive limit, electrolyte decomposition will occur even if the potential across the cell is lower than the determined operating window. Under the assumption that the specific capacitance is the same when the material is charged positively and negatively, this essentially limits the operating potential to twice the value of the lesser limit, therefore in the case of the cells in Fig. 2 the stable operating potential for a symmetrical cell is as low as 1.8 V. 3.2. Electrochemical stability of [EMIm][BF4] with different binder materials
Fig. 2. Capacitance vs. cycle number for symmetrical EDLCs using [EMIm][BF4] electrolyte operating over voltage ranges of 2.0e3.5 V under galvanostatic cycling at 2 A g1. Expressed as (a) specific cell capacitance and (b) a fraction of the initial cell capacitance. (Electrode composition was 80:10:10 of carbon A, carbon black and PVdF binder.)
Clearly an operating potential of 1.8 V is far below that of conventional electrolytes and even if the whole operating windows in Table 1 were to be used, the determined operating potentials fall far below that usually reported for [EMIm][BF4]. From a previous study we reported that when a polymeric IL was used as an electrode binder it could operate reliably at a higher potential [42] by interacting with the electrolyte. In order to investigate the effect of using different inert binders on the stability of [EMIm][BF4], electrodes were produced using some commonly utilised materials; PVdFHFP, CMC and PTFE. Fig. 3 illustrates the results using electrodes containing different binders using the same method as Fig. 1. The most apparent feature
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
in Fig. 3 is that cells using PTFE binder are substantially more stable than those using other binders as these cells were associated with the smallest values of Dq over the whole potential range investigated. At a potential of around þ1.0 V (vs OCP) the behaviour of PTFE bound cells clearly diverges from that of cells using the other binders. The instability at positive potentials for PVdF-based and CMC binders may indicate that reactive species are generated when using [EMIm][BF4] as an electrolyte. The labile proton present on the cation may be responsible for such decomposition reactions as it is known to limit both electrochemical stability and the stability of the dialkylimidazolium cation with metallic lithium [7,26,43]. The generation of protons and radicals is likely to produce intermediates that can degrade the binders, or the imidazolium cations themselves may polymerize. It is also possible that [EMIm] [BF4] acts as a plasticizer when combined with PVdF and PVdF-HFP polymers [44] and, in contrast with the other binder materials studied, PTFE is known to bind particles together by forming a network of fibrils rather than by producing a film [45] which may also influence the results. As a result, the operating potential of devices coupling these binders with [EMIm][BF4] are very limited when compared with those employing PTFE. Stability limits for each of the binders coupled with carbon A were derived from the information in Fig. 3 and are presented in Table 1. It is worth noting that when expressed in terms of S-value, stability limits determined using the criteria that d2S/dV2 > 0.05, namely 2.1 V and þ1.5 V (vs OCP) correlate well with those reported by Weingarth et al. [17] for [EMIm] [BF4]. Electrodes that use PTFE binder show a substantial improvement in operating potential when compared against the other binders studied, with a total of 3.8 V arising from stability limits of 2.1 V and þ1.7 V (vs OCP). 3.3. Asymmetric mass loading However, a feature of the determined stability limits for a vast majority of electrolytes is that they are asymmetrical with respect to the OCP of the cell. Typically, the limit determined for positive deviations from the OCP is lower than that under negative polarisations. To adjust for this and permit the use of wider operating potentials, it has become common practice [26,27,36,37,39,40,46e51] to employ asymmetric mass loadings based on the balance of charge in the cell. This is typically represented by Equation (5) where mþ represents the mass of the positive electrode, cþ the specific capacitance of the positive electrode and Uþ the positive stability limit. (m, c and U represent the same quantities for the negative electrode)
mþ cþ Uþ ¼ m c U
(5)
Equation (5) is often applied under the assumption that the material exhibits the same specific capacitance when either a positive or negative potential is applied. This however is not found often in practice, especially concerning ILs, as the capacitive response from the same material polarized in opposite directions can vary substantially. For example, using the values of discharge capacity determined during cyclic voltammetry, the relationship between specific capacitance and applied potential using [EMIm][BF4] for each of the carbon materials is illustrated in Fig. 4(a). Carbon A clearly displays the highest specific capacitances over the operating potential region (i.e. 2.0e1.2 V (vs OCP)) which can be attributed to this material having a high specific surface area but also containing a significant mesopore volume that facilitates electrolyte access. It is unsurprising that carbon C has a similar specific capacitance to carbon A when positively polarized as this material displays the
457
highest specific surface area. The difference seen during negative polarization may be a result of the incompatibility of the PVdF binder with the IL electrolyte and the charge attributed to the double layer capacitance may actually arise from the decomposition process and/or generation of electroactive species. Carbon D performs poorest, which can be expected as it possesses a low surface area and lesser mesopore volume than carbon E, however it is worth reiterating that these carbons were tested in composite electrodes possessing 20% wt. binder which can be expected to adversely influence their specific capacitances. From Fig. 4(a) it can be seen that in general, as the vertex potential of the cell is swept to increasing values the specific capacitance rises. Increasing electrode potential can be seen as a driving force for the adsorption of ions on the surface and this in turn displaces a greater amount of charge within the electrode. This effect has also been reported by Grey and collaborators [52,53] where NMR and a quartz crystal microbalance showed that the populations of ions in pores are dependent on the cell voltage. The accelerated decomposition of electrolyte in carbon E may result from its relatively high mesopore volume (1.19 cm3 g1). It is likely that improved mass transfer through the mesopore network facilitates the increased rate of IL decomposition in these electrodes. This could result in the formation of a passivating layer on the electrode surface or the blocking of pores, reducing the capacitive response from the electrode. Therefore, the specific capacitance of the material at the electrochemical stability limits, in addition to the potential limits themselves are required to determine suitable mass balance ratios for EDLCs. From Fig. 4(b) it is clear that the electrode binder can also have a substantial effect on the values of specific capacitance attained at a given potential with variations of up to 65 F g1 being observed between the binders studied. It is also clear that larger values of specific capacitance are obtained when the electrode is negatively polarized. It is unsurprising that different specific capacitances are observed when the opposite polarization is applied as the size of the cations and anions present in the IL are significantly different; EMImþ is substantially larger than BF 4. In Fig. 4(b) there are clear differences in the specific capacitance obtained when negatively polarized; PVdF bound electrodes produce values as high as 205 F g1 (at 2.8 V (vs OCP)) whereas PTFE bound electrodes yield 137 F g-1 at the same voltage. As mentioned previously, it is likely that there are different adhesion mechanisms present when using PVdF based binders and PTFE, as the method of electrode production differs; in the former case the polymer is left behind after evaporation of a solvent, whereas in the latter a network of polymer fibrils is established through kneading and mixing. These different methods can be expected to result in the composite electrodes possessing different degrees of electrical conductivity and accessible porosity. Also, polymer binder was added as a constant fraction of mass of the dried composite electrode and differences in binder density is likely to influence the behaviour of composite electrodes. The asymmetry between the positive and negative stability limits is often disregarded when using [EMIm][BF4] electrolyte and symmetrical cells are typically used for EDLC testing. In order to make full use of the operating potential, the use of asymmetrical mass loadings was investigated. In Fig. 4(b) the determined stability limits of 2.1 V and þ1.7 V (vs OCP) for PTFE electrodes correspond with capacitance values of 124.7 F g1 and 111.6 F g1 respectively. However, these values of specific capacitance were determined at a rate of 5 mV s1 which is equivalent to a very low current in a full cell. Specific capacitance is typically seen to diminish with increasing rates of charge, especially when ILs are used as electrolytes, due to the relatively poor transport properties of the electrolyte, which kinetically limits the
458
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
Fig. 4. Discharge specific capacitance at each voltage step in stability determination (a) using different carbons and PVdF as binder and (b) using carbon A and different binders.
charge storage process. As EDLCs are high power devices, typically delivering all of their charge in tens of seconds, it can be expected that the specific capacitances achieved under the expected operating conditions will be substantially lower than those measured using cyclic voltammetry at low sweep rates. Asymmetry between the specific capacitances measured at low rates may be due to the long charge and discharge times that enable small ions to access a greater fraction of the electrode surface than those at higher rates where ions face increased resistance to migration. Therefore, the application of Equation (5) to define a suitable mass loading should use values of specific capacitance that are more representative of full cells operating at practical currents. For example, cells designed using the limits of 2.1 V and þ1.7 V and the corresponding capacitance values determined at 5 mV s1 results in a mass loading ratio, mþ/m, of 1.38, however this would only be suited to a discharge time of over 12 min. Asymmetric cells coupling carbon A with PVdF and PTFE binders were tested using cyclic voltammetry between the OCP and stability limit using a range of sweep rates from 5 to 500 mV s1. The mass ratios calculated from these cyclic voltammograms are illustrated in Fig. 5. In the case of PVdF binder, Fig. 5 shows that a mass loading ratio
Fig. 5. Mass ratio of positive electrode and negative electrode determined at different scan rates, using carbon A and PVdF (black square) or PTFE (green diamond) binders. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
of around 2.3 is suitable for sweep rates lower than 25 mV s1, however as the sweep rate increases towards 500 mV s1 the mass loading ratio is seen to increase. This arises not only due to the asymmetry in stability limits, but also due to the rate dependent values of specific capacitance. In a cell designed using the capacitance values obtained at low rates (i.e. mþ/m ¼ 2.3) the system is balanced when cycling at low rates, however as the rate increases the amount of charge that can be stored on the positive electrode becomes limited and to compensate for this the relative potential across the positive electrode increases. Consequently, this results in the electrode operating outside the stability limits of the electrolyte. Fig. 5 illustrates different behaviour for electrodes using PTFE binder. The required mass loading ratio is much closer to symmetry than for PVdF based electrodes and at low rates a suitable ratio is around 1.3. Also, in contrast with PVdF based electrodes, as the scan rate increases, the mass loading ratio is seen to decrease slightly. This means that if a cell with a mass loading ratio of 1.3 is cycled as higher rates the negative electrode, rather than the positive electrode, becomes limiting. This is evident in Fig. 6 where we employed three-electrode Swagelok™ type cells that enable the potential of each electrode to be measured during cycling vs a Ag quasi-reference electrode. Fig. 6(aec) shows the behaviour of electrodes in asymmetric cells with different mass loading ratios when cycled at 0.1 A g1. In the case of the symmetrical cell in Fig. 6(a) it is seen that the positive electrode operates slightly above the determined stability limit of þ1.7 V as anticipated. When using a ratio of 1.3 in Fig. 6(b) both electrodes remain within their stability limits, and at a ratio of 2.0 the negative electrode becomes limiting and operates below 2.1 V in Fig. 6(c). This effect is more apparent at the higher rate of 1.0 Ag1 shown in Fig. 6(f) which also shows a large increase in iR drop across the positive electrode resulting from the higher mass loading ratio. Although the differences between the determined stability limits appear trivial, these points deviations represent instances where electrolyte decomposition occurs and quickly become evident during charge/discharge cycling. Romann et al. [22] used in-situ IR to show that [EMIm][BF4] decomposes when an EDLC is operating outside the voltage limits, and determined that 2.8 V is stable in a symmetrical cell, and 3.5 V is permissible when using asymmetric cell. It is clear that these maximum operating voltages depend on the carbon and binder utilised. EDLCs performance using asymmetric mass balance is shown in Supplementary Material in Section EDLC Performance and Fig. S4.
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460
459
Fig. 6. Potential (vs. OCP) of positive (blue line) and negative (red line) electrodes measured in a two electrode cell with a silver quasi-reference third electrode. Galvanostatic charge-discharge at 0.1 A g1 (aec) and 1.0 A g1 (def) for cells with mþ/m of 1.0 (a, d), 1.3 (b, e) and 2.0 (c, f). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
4. Conclusions
https://doi.org/10.1016/j.electacta.2018.03.094.
In this work we have outlined some of the considerations necessary for the design of EDLCs employing IL electrolytes. Of crucial importance is the determination of electrochemical stability limits for individual combinations of electrode and electrolyte. Using [EMIm][BF4] as an example we demonstrated that the physical properties of the activated carbon used in the electrodes influence the permissible operating potential of an EDLC, and also that the binder material employed can have further effects. It was shown that in the case of [EMIm][BF4], using PTFE binder allows for operating potentials as wide as 3.8 V. Due to the differences in response under opposing polarisations, it is necessary to employ an asymmetric mass loading in EDLC cells in order to make full use of the operating window. By balancing the charge in the cell the differences in stable potential limits and capacitative behaviour can be overcome, however we also found that this balance can be influenced by the rate at which these parameters are determined. Three-electrode measurements show that using a mass loading ratio of 1.3 results in each electrode operating within their determined stability limits and stable cycling of a full cell at an operating potential of 3.8 V was demonstrated over 50,000 cycles. Premature ageing of the cell can result from only considering the capacitance at low rates.
References
Acknowledgements The authors would like to thank the EPSRC (EP/K021192/1) and FAPESP (15/26308-7, 13/22748-7 and 14/14690-1) for funding. Appendix A. Supplementary data Supplementary data related to this article can be found at
€tz, M. Carlen, Principles and applications of electrochemical capacitors, [1] R. Ko Electrochim. Acta 45 (2000) 2483e2498, https://doi.org/10.1016/S00134686(00)00354-6. [2] P.J. Hall, M. Mirzaeian, S.I. Fletcher, F.B. Sillars, A.J.R. Rennie, G.O. Shitta-Bey, et al., Energy storage in electrochemical capacitors: designing functional materials to improve performance, Energy Environ. Sci. 3 (2010) 1238e1251, https://doi.org/10.1039/c0ee00004c. [3] P. Simon, Y. Gogotsi, Capacitive energy storage in nanostructured carboneelectrolyte systems, Acc. Chem. Res. 46 (2013) 1094e1103, https:// doi.org/10.1021/ar200306b. guin, V. Presser, A. Balducci, E. Frackowiak, Carbons and electrolytes for [4] F. Be advanced supercapacitors, Adv. Mater. 26 (2014) 2219e2251, https://doi.org/ 10.1002/adma.201304137. [5] A. Krause, A. Balducci, High voltage electrochemical double layer capacitor containing mixtures of ionic liquids and organic carbonate as electrolytes, Electrochem. Commun. 13 (2011) 814e817, https://doi.org/10.1016/ j.elecom.2011.05.010. ski, A. Lewandowski, I. Stepniak, Ionic liquids as electrolytes, Elec[6] M. Galin trochim. Acta 51 (2006) 5567e5580, https://doi.org/10.1016/ j.electacta.2006.03.016. [7] M. Armand, F. Endres, D.R. MacFarlane, H. Ohno, B. Scrosati, Ionic-liquid materials for the electrochemical challenges of the future, Nat. Mater. 8 (2009) 621e629, https://doi.org/10.1038/nmat2448. [8] A. Eftekhari, Supercapacitors utilising ionic liquids, Energy Storage Mater 9 (2017) 47e69, https://doi.org/10.1016/j.ensm.2017.06.009. [9] D.R. MacFarlane, M. Forsyth, P.C. Howlett, M. Kar, S. Passerini, J.M. Pringle, et al., Ionic liquids and their solid-state analogues as materials for energy generation and storage, Nat. Rev. Mater 1 (2016) 15005, https://doi.org/10.1038/ natrevmats.2015.5. [10] M. Salanne, Ionic liquids for supercapacitor applications, Top. Curr. Chem. 375 (2017) 63, https://doi.org/10.1007/s41061-017-0150-7. [11] D.R. MacFarlane, A.L. Chong, M. Forsyth, M. Kar, R. Vijayaraghavan, A. Somers, et al., New dimensions in saltesolvent mixtures: a 4th evolution of ionic liquids, Faraday Discuss (2017), https://doi.org/10.1039/C7FD00189D. €nes, E. Lust, Electrochemical characteristics of carbide-derived [12] H. Kurig, A. Ja carbonr1-ethyl-3-methylimidazolium tetrafluoroborate supercapacitor cells, J. Electrochem. Soc. 157 (2010) A272, https://doi.org/10.1149/1.3274208. €nes, E. Lust, Electrical double layer capacitors based on [13] H. Kurig, M. Vestli, A. Ja
460
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
V.L. Martins et al. / Electrochimica Acta 270 (2018) 453e460 two 1-ethyl-3-methylimidazolium ionic liquids with different anions, Electrochem. Solid state Lett. 14 (2011) A120, https://doi.org/10.1149/1.3596722. H. Kurig, M. Vestli, K. Tonurist, A. Janes, E. Lust, Influence of room temperature ionic liquid anion chemical composition and electrical charge delocalization on the supercapacitor properties, J. Electrochem. Soc. 159 (2012) A944eA951, https://doi.org/10.1149/2.022207jes. F.B. Sillars, S.I. Fletcher, M. Mirzaeian, P.J. Hall, Effect of activated carbon xerogel pore size on the capacitance performance of ionic liquid electrolytes, Energy Environ. Sci. 4 (2011) 695e706, https://doi.org/10.1039/C0EE00337A. F.B. Sillars, S.I. Fletcher, M. Mirzaeian, P.J. Hall, Variation of electrochemical capacitor performance with room temperature ionic liquid electrolyte viscosity and ion size, Phys. Chem. Chem. Phys. 14 (2012) 6094, https://doi.org/ 10.1039/c2cp40089h. €tz, A reliable D. Weingarth, H. Noh, A. Foelske-Schmitz, A. Wokaun, R. Ko determination method of stability limits for electrochemical double layer capacitors, Electrochim. Acta 103 (2013) 119e124, https://doi.org/10.1016/ j.electacta.2013.04.057. R. Palm, H. Kurig, K. Tonurist, A. Janes, E. Lust, Influence of different organic solvent additives on 1-ethyl-3-methylimidazolium tetrafluoroborate electrolyte based electrical double layer capacitors, J. Electrochem. Soc. 160 (2013) A1741eA1745, https://doi.org/10.1149/2.046310jes. ~ nurist, A. Ja €nes, E. Lust, Electrical double layer capacR. Palm, H. Kurig, K. To itors based on 1-ethyl-3-methylimidazolium tetrafluoroborate with small addition of acetonitrile, Electrochim. Acta 85 (2012) 139e144, https://doi.org/ 10.1016/j.electacta.2012.08.030. ~ nurist, A. J€ R. Palm, H. Kurig, K. To anes, E. Lust, Is the mixture of 1-ethyl-3methylimidazolium tetrafluoroborate and 1-butyl-3-methylimidazolium tetrafluoroborate applicable as electrolyte in electrical double layer capacitors? Electrochem. Commun. 22 (2012) 203e206, https://doi.org/10.1016/ j.elecom.2012.06.029. ~ nurist, T. Thomberg, A. Ja €nes, T. Romann, V. Sammelselg, E. Lust, InfluK. To ence of separator properties on electrochemical performance of electrical double-layer capacitors, J. Electroanal. Chem. 689 (2013) 8e20, https:// doi.org/10.1016/j.jelechem.2012.11.024. T. Romann, O. Oll, P. Pikma, H. Tamme, E. Lust, Surface chemistry of carbon electrodes in 1-ethyl-3-methylimidazolium tetrafluoroborate ionic liquid e an in situ infrared study, Electrochim. Acta 125 (2014) 183e190, https:// doi.org/10.1016/j.electacta.2014.01.077. M. Lazzari, F. Soavi, M. Mastragostino, Mesoporous carbon design for ionic liquid-based, double-layer supercapacitors, Fuel Cell. 10 (2010) 840e847, https://doi.org/10.1002/fuce.200900198. ~ oz-Torrero, J. Palma, M. Anderson, R. Marcilla, AllG. Ayalneh Tiruye, D. Mun solid state supercapacitors operating at 3.5 V by using ionic liquid based polymer electrolytes, J. Power Sources 279 (2015) 472e480, https://doi.org/ 10.1016/j.jpowsour.2015.01.039. e, M. Mastragostino, et al., A. Balducci, R. Dugas, P.L.L. Taberna, P. Simon, D. Ple High temperature carbonecarbon supercapacitor using ionic liquid as electrolyte, J. Power Sources 165 (2007) 922e927, https://doi.org/10.1016/ j.jpowsour.2006.12.048. M. Lazzari, F. Soavi, M. Mastragostino, High voltage, asymmetric EDLCs based on xerogel carbon and hydrophobic IL electrolytes, J. Power Sources 178 (2008) 490e496, https://doi.org/10.1016/j.jpowsour.2007.12.029. A.J.R. Rennie, V.L. Martins, R.M. Torresi, P.J. Hall, Ionic liquids containing sulfonium cations as electrolytes for electrochemical double layer capacitors, J. Phys. Chem. C 119 (2015) 23865e23874, https://doi.org/10.1021/ acs.jpcc.5b08241. T. Nishida, Y. Tashiro, M. Yamamoto, Physical and electrochemical properties of 1-alkyl-3-methylimidazolium tetrafluoroborate for electrolyte, J. Fluorine Chem. 120 (2003) 135e141, https://doi.org/10.1016/S0022-1139(02)00322-6. rez, F. Malbosc, et al., R. Lin, P.-L. Taberna, S. Fantini, V. Presser, C.R. Pe Capacitive energy storage from 50 to 100 C using an ionic liquid electrolyte, J. Phys. Chem. Lett. 2 (2011) 2396e2401, https://doi.org/10.1021/jz201065t. C. Lian, K. Liu, K.L. Van Aken, Y. Gogotsi, D.J. Wesolowski, H.L. Liu, et al., Enhancing the capacitive performance of electric double-layer capacitors with ionic liquid mixtures, ACS Energy Lett. 1 (2016) 21e26, https://doi.org/ 10.1021/acsenergylett.6b00010. K.L. Van Aken, M. Beidaghi, Y. Gogotsi, K.L. Van Aken, M. Beidaghi, Y. Gogotsi, Formulation of ionic-liquid electrolyte to expand the voltage window of supercapacitors, Angew. Chem. Int. Ed. 54 (2015) 4806e4809, https://doi.org/ 10.1002/anie.201412257. A. Balducci, D. Belanger, T. Brousse, J.W. Long, W. Sugimoto, Perspectiveda guideline for reporting performance metrics with electrochemical capacitors: from electrode materials to full devices, J. Electrochem. Soc. 164 (2017) A1487eA1488, https://doi.org/10.1149/2.0851707jes. M. Ue, M. Takeda, M. Takehara, S. Mori, Electrochemical properties of
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
quaternary ammonium salts for electrochemical capacitors, J. Electrochem. Soc. 144 (1997) 2684, https://doi.org/10.1149/1.1837882. K. Xu, M.S. Ding, T. Richard Jow, Better quantification of electrochemical stability limits for electrolytes in double layer capacitors, Electrochim. Acta 46 (2001) 1823e1827, https://doi.org/10.1016/S0013-4686(01)00358-9. K. Xu, S.P. Ding, T. Richard Jow, Toward reliable values of electrochemical stability limits for electrolytes, J. Electrochem. Soc. 146 (1999) 4172e4178, https://doi.org/10.1149/1.1392609. €ger, P. Goodrich, J. Alvarez Vicente, J. Jacquemin, S. Pohlmann, T. Olyschla A. Balducci, Azepanium-based ionic liquids as green electrolytes for high voltage supercapacitors, J. Power Sources 273 (2015) 931e936, https:// doi.org/10.1016/j.jpowsour.2014.09.167. C. Wolff, S. Jeong, E. Paillard, A. Balducci, S. Passerini, High power, solvent-free electrochemical double layer capacitors based on pyrrolidinium dicyanamide ionic liquids, J. Power Sources 293 (2015) 65e70, https://doi.org/10.1016/ j.jpowsour.2015.05.065. A.R. Neale, S. Murphy, P. Goodrich, C. Schütter, C. Hardacre, S. Passerini, et al., An ether-functionalised cyclic sulfonium based ionic liquid as an electrolyte for electrochemical double layer capacitors, J. Power Sources 326 (2016) 549e559, https://doi.org/10.1016/j.jpowsour.2016.06.085. A.J.R. Rennie, V.L. Martins, R.M. Smith, P.J. Hall, Influence of particle size distribution on the performance of ionic liquid-based electrochemical double layer capacitors, Sci. Rep. 6 (2016) 22062, https://doi.org/10.1038/srep22062. V.L. Martins, A.J.R. Rennie, R.M. Torresi, P.J. Hall, Ionic liquids containing tricyanomethanide anions: physicochemical characterisation and performance as electrochemical double-layer capacitor electrolytes, Phys. Chem. Chem. Phys. 19 (2017) 16867e16874, https://doi.org/10.1039/C7CP03377J. J.P. Randin, E. Yeager, Differential capacitance study of stress- annealed pyrolytic graphite electrodes, J. Electrochem. Soc. 118 (1971) 711, https:// doi.org/10.1149/1.2408151. V.L. Martins, A.J.R. Rennie, J. Lesowiec, R.M. Torresi, P.J. Hall, Using polymeric ionic liquids as an active binder in supercapacitors, J. Electrochem. Soc. 164 (2017) A3253eA3258, https://doi.org/10.1149/2.1411713jes. F.F.C. Bazito, Y. Kawano, R.M. Torresi, Synthesis and characterization of two ionic liquids with emphasis on their chemical stability towards metallic lithium, Electrochim. Acta 52 (2007) 6427e6437, https://doi.org/10.1016/ j.electacta.2007.04.064. L. Guo, Y. Liu, C. Zhang, J. Chen, Preparation of PVDF-based polymer inclusion membrane using ionic liquid plasticizer and Cyphos IL 104 Carrier for Cr(VI) transport, J. Membr. Sci. 372 (2011) 314e321, https://doi.org/10.1016/ j.memsci.2011.02.014. T. Kitamura, K.-I. Kurumada, M. Tanigaki, M. Ohshima, S.-I. Kanazawa, Formation mechanism of porous structure in polytetrafluoroethylene (PTFE) porous membrane through mechanical operations, Polym. Eng. Sci. 39 (1999) 2256e2263, https://doi.org/10.1002/pen.11613. Z. Sun, Z. Liu, B. Han, S. Miao, J. Du, Z. Miao, Microstructural and electrochemical characterization of RuO2/CNT composites synthesized in supercritical diethyl amine, Carbon N. Y. 44 (2006) 888e893, https://doi.org/10.1016/ j.carbon.2005.10.023. ~ ero, E. Frackowiak, F. Be guin, High-voltage V. Khomenko, E. Raymundo-Pin asymmetric supercapacitors operating in aqueous electrolyte, Appl. Phys. A Mater. Sci. Process 82 (2006) 567e573, https://doi.org/10.1007/s00339-0053397-8. S. Vaquero, J. Palma, M. Anderson, R. Marcilla, Mass-balancing of electrodes as a strategy to widen the operating voltage window of carbon/carbon supercapacitors in neutral aqueous electrolytes, Int. J. Electrochem. Sci. 8 (2013) 10293e10307. S. Vaquero, J. Palma, M. Anderson, R. Marcilla, Improving performance of electric double layer capacitors with a mixture of ionic liquid and acetonitrile as the electrolyte by using mass-balancing carbon electrodes, J. Electrochem. Soc. 160 (2013) A2064eA2069, https://doi.org/10.1149/2.063311jes. Z. Dai, C. Peng, J.H. Chae, K.C. Ng, G.Z. Chen, Cell voltage versus electrode potential range in aqueous supercapacitors, Sci. Rep. 5 (2015), https://doi.org/ 10.1038/srep09854. V.L. Martins, A.J.R. Rennie, N. Sanchez-Ramirez, R.M. Torresi, P.J. Hall, Improved performance of ionic liquid supercapacitors by using tetracyanoborate anions, ChemElectroChem 5 (2018) 598e604, https://doi.org/10.1002/ celc.201701164. A.C. Forse, J.M. Griffin, C. Merlet, P.M. Bayley, H. Wang, P. Simon, et al., NMR study of ion dynamics and charge storage in ionic liquid supercapacitors, J. Am. Chem. Soc. (2015), https://doi.org/10.1021/jacs.5b03958. J.M. Griffin, A.C. Forse, W.-Y. Tsai, P.-L. Taberna, P. Simon, C.P. Grey, In situ NMR and electrochemical quartz crystal microbalance techniques reveal the structure of the electrical double layer in supercapacitors, Nat. Mater. 14 (2015) 812e819, https://doi.org/10.1038/nmat4318.