© The Electrochemical Society, Inc. 2010. All rights reserved. Except as provided under U.S. copyright law, this work may not be
reproduced, resold, distributed, or modified without the express permission of The Electrochemical Society (ECS). The archival version
of this work was published in Lazzari, M. et al. 2010. Role of carbon porosity and ion size in the development of ionic liquid based
supercapacitors. Journal of the Electrochemical Society 158 (1): A22-A25.
Role of carbon porosity and ion size in the development of ionic liquid based supercapacitors
M. Lazzaria, M. Mastragostino a,*, A.G. Pandolfo b, V. Ruizb, F. Soavi a
a
Dipartimento di Scienza dei Metalli, Elettrochimica e Tecniche Chimiche, Via San Donato 15 –
40127 Bologna, Italy
b
CSIRO Division of Energy Technology, Box 312, Clayton South, Vic. 3169, Australia
*Corresponding author: [email protected]
Key words: ionic liquid, porous carbon electrode, supercapacitor, capacitance.
Abstract
The role played by carbon porosity and electrolyte chemistry in the development of doublelayer supercapacitors based on solvent-free ionic liquids (ILs) of wide electrochemical stability
window is investigated. Voltammetric studies performed in N-methyl-N-butyl-pyrrolidinium
bis(trifluoromethanesulfonyl)imide
(PYR14TFSI),
N-trimethyl-N-propylammonium
bis(trifluoromethanesulfonyl)imide
(N1113TFSI),
N-methyl-N-butyl-pyrrolidinium
tris(pentafluoroethyl)trifluorophosphate (PYR14FAP) ionic liquids, and PYR14TFSI – tetraethyl
ammonium bis(trifluoromethanesulfonyl)imide (N2222TFSI) solutions demonstrate that the pore-toion size ratio and porous electrode/IL interface properties may have a higher impact on electrode
electrical response than the inherent IL bulk properties. The effect of carbon porosity on the
electrode capacitance and charge storage capability in both the positive and negative potential
domains is discussed in relation to the IL properties and an estimation of the upper limits of the
performance of IL-based supercapacitors with carbons of optimized porosity is reported .
1
Introduction
The interest in ionic liquid-based electrolytes for electrochemical energy storage and/or
conversion systems is growing world wide. The key features that trigger the interest in room
temperature ionic liquids (ILs) for different applications, ranging from proton-exchange membrane
fuel cells, to lithium ion batteries and supercapacitors, are their low vapour pressure and high
chemical stability which make them promising electrolytes for the development of safe
electrochemical systems [1]. In electrical double-layer capacitors (EDLCs) the cell potential is
limited by the redox decomposition of the electrolyte at the carbon electrodes, hence the use of ILs
with a wide electrochemical stability window (ESW) like those with substituted pyrrolidinium
cations allows cell voltages higher than 3.5 V to be achieved and specific energies exceeding those
of commercial EDLCs [2, 3]. The successful use of pyrrolidinium-based ILs was demonstrated by
asymmetric EDLCs (AEDLCs) with mesoporous positive and negative carbon electrodes of
different weight that featured maximum specific energy up to ca. 45 Wh/kg between 3.9 V and 1.95
V (referred only to electrode mass) [4].
In EDLCs based on solvent-free ILs, the absence of molecular solvents like propylene
carbonate or acetonitrile implies an intimate contact between carbon electrode surface and IL,
therefore the structure of the porous electrode/IL interface is crucial [5-7]. Here, the role played by
carbon micro-mesoporosity and IL chemistry in the development of EDLCs based on aprotic IL of
wide ESW is investigated. The voltammetric capacitance responses of micro- and mesoporous
carbon electrodes in the positive and negative voltage domains allowed by N-methyl-N-butylpyrrolidinium bis(trifluoromethanesulfonyl)imide (PYR14TFSI), N-trimethyl-N-propylammonium
bis(trifluoromethanesulfonyl)imide
(N1113TFSI),
N-methyl-N-butyl-pyrrolidinium
tris(pentafluoroethyl)trifluorophosphate (PYR14FAP) ionic liquids and by PYR14TFSI-tetraethyl
ammonium bis(trifluoromethanesulfonyl)imide (N2222TFSI) solutions (see Scheme 1) are reported
and an estimation of the upper limits of the performance of IL-based supercapacitors with carbons
of optimized porosity is given.
2
Experimental
A series of polymer–derived activated carbons (ACs) were prepared. The AC precursor
material was polyfurfuryl alcohol (PFA), which was obtained by acid catalysed polymerization [8].
Briefly, p-toluensulfonic acid monohydrate (0.5 g) was dissolved in Triton X-100 (50 ml) and
heated gently prior to the addition of 50 ml of furfuryl alcohol (FA). The mixture was stirred at an
initial temperature of 10 ºC and then allowed to rise to room temperature whilst stirring
continuously for two days. The resulting PFA polymer was then chemically activated using
potassium hydroxide. The polymer was mixed with KOH in proportions of: 1:1, 2:1, 2.5:1 or 3:1
parts KOH : PFA. The mixture was activated by slow heating (2.5 ºC min-1), under a nitrogen flow
(90 mL min-1), up to 700 ºC and held at this temperature for 1h. The system was allowed to cool to
room temperature and the product was removed and neutralized with HCl (reflux during 1h) and
thoroughly washed with distilled water until neutral pH. The resultant activated carbons were
labeled PFAKX, where X refers to the ratio of chemical agent used. Physical adsorption of nitrogen
at 77 K was performed on a Micromeritics ASAP 2010 volumetric gas sorption unit after thorough
degassing at 300°C. The N2 adsorption isotherms were analyzed by density functional theory (DFT)
assuming a slit pore geometry.
PYR14TFSI (98%, Solvent Innovation) was dried over night at 80° C under dynamic vacuum
(Büchi Glass Oven B-580) before use. N1113TFSI (99.5 % Solvionic), PYR14FAP (Merck) and
N2222TFSI (puriss., Fluka) were used as received. Water content in the ILs was checked by Karl
Fisher titration (684 KF Coulometer Metrohm) and was lower than 30 ppm.
The electrochemical stability of the ILs was evaluated at 60 °C with glassy carbon electrodes
(0.071 cm2 area) by cyclic voltammetry (CV) at 20 mV s-1; the reference electrode was a silver wire
in 0.1 M AgTFSI-PYR14TFSI solution and featured an equilibrium potential of 0.42 V vs. the
reversible redox couple ferrocene (Fc)/ferrocinium (Fc+).
3
Conductivity measurements were carried out with a Radiometer Analytical CDM 210
conductimeter and the samples thermally equilibrated in a Haake thermocryostat. Viscosity was
evaluated by a Micro-Ubbelhode viscosimeter with a Viscoclock system (SCHOTT Instruments).
The solvent polarity of the ILs was evaluated in ambient atmosphere on the basis of the 2,6diphenyl-4-(2,4,6-triphenylpyridinium-1-yl)phenolate
(Reichardt
dye,
Sigma-Aldrich)
solvatochromism [9] measured by a Perkin-Elmer Lambda 19 spectrometer; the data are reported in
terms of normalized molar transition energy ENT with ENT = 0 for TMS and 1 for water.
Carbon coated on aluminum (30 µm thick) collectors were utilized as electrodes. The carbon
coating
was
prepared
by
mixing
powdered
PFA – derived AC,
carbon
black
and
carboxymethylcellulose (CMC) binder in a proportion 1 : 0.2 : 0.1 with the minimum amount of
water to form a slurry. The carbon mixture was coated onto aluminum foil with a grooved rod
applicator. The composite electrode loadings ranged from 0.7 to 1.1 mg cm-2, the electrode area was
1.27 cm2.
“Three-electrode” Swagelok cells were used for CV tests (20 mV s-1) at 60°C where IL
conductivities are of interest for practical applications; CVs were performed in the electrode
potential ranges that allow at least 95% coulombic efficiency of the charge/discharge processes. A
double-layer carbon counterelectrode, with charge storage capability significantly higher than that
of the working electrode, was used so as not to limit the capacitive response of the latter. The
reference electrode was a silver disk; hereinafter the electrode potentials are given vs. Fc/Fc+
(EFc/Fc+ = EAg + 0.200 + 0.010 V). The cells were assembled in a dry box (MBraun Labmaster 130,
H2O and O2<1 ppm) using a fiberglass separator (Durieux, 200 μm thick when pressed).
The electrochemical tests were performed with a Perkin-Elmer VMP multichannel
potentiostat/galvanostat at temperatures controlled by a Thermoblock (FALC) or a Memmert oven.
4
Results and Discussion
Table 1 summarizes the porosity features of the investigated activated carbons. They contain
micro- and mesopores with size ranging from 1.5 to 4.5 nm as demonstrated by the pore size
distributions (PSDs) reported in Figure 1, obtained by the application of density functional theory
(DFT) to the nitrogen adsorption isotherms. The minima observed in Figure 1, at ~1nm, is attributed
to a well known artifact arising from the application of DFT to the measurement of narrow pores
[10].
Table 1 reports the specific pore volumes and surface areas from pores belonging to the two
pore size ranges: V<1 nm, S<1 nm and V>1 nm, S>1 nm. Both the total pore volume and pore size of the
carbons increase in the order PFAK1 < PFAK2 < PFAK2.5 < PFAK3. While PFAK1 and PFAK2
are microporous carbons, PFAK2.5 features large micropores and small mesopores and PFAK3 is
basically mesoporous.
The carbon electrodes were tested in PYR14TFSI, N1113TFSI and PYR14FAP that display large
ESWs, even wider than 5 V when evaluated by glassy carbon electrode (see Table 2),which allows
wide polarization and high supercapacitor cell voltage. These ILs feature quaternary nitrogen–based
cations and fluoroalkyl-based anions of differing size and conductivity, viscosity and normalized
solvent polarity that vary with ion size and chemistry (see Scheme 1 and Table 2).
Figure 2 displays the CVs in both the negative and positive potential ranges of the carbon
electrodes in PYR14TFSI. The Figure clearly shows that the capacitive response increases in the
order PFAK1<PFAK2<PFAK2.5<PFAK3. The specific capacitance of PFAK1, with pore size
lower than 1.5 nm and a total specific surface area of 685 m2 g-1, is 30 F g-1. Specific capacitance
has a 3 fold increase up to 100 F g-1 with PFAK2 that features also pores within the 1.5 - 2 nm
range with a total surface area that only doubles that of PFAK1 (Table 1). This indicates that the
portion of pores exploitable for the double-layer charge-discharge process, those wider than IL size,
is higher in PFAK2 than in PFAK1. On the other hand, on the basis of the steric hindrance of the
5
ILs reported in Scheme 1, it is reasonable to consider that only pores wider than ca. 1 nm are
accessible by the IL, with ion access being enhanced by the increase of pore size.
With PFAK2.5, that has almost the same S<1nm than PFAK2 and a S>1nm of 460 m2 g-1 from
pores narrower than 2.5 nm and mainly of 1.5 nm, the high specific capacitance of ca. 150 F g-1 is
reached. This value is only slightly increased up to 160 F g-1 with PFAK3 that has the highest S>1nm
of 1140 m2 g-1 and pores mainly of 3 nm. Hence, the gain in capacitance seems not to be directly
related to the increase of S>1nm. This can be explained by taking into account that the capacitive
response of a porous electrode in an IL is the result of a series combination of the capacitances
inside the IL (CIL) and of the space charge layer within the carbon (Cc) [11-16]. CIL depends on the
IL ion arrangement at the electrode surface and is governed by ion-to-pore size ratio as well as by
IL chemistry. For moderate polarizations and small pores, the ions facing the electrode surface
arrange in a compact layer whose chemistry affects permettivity in the double-layer and thus
electrode capacitance response. For large electrode polarizations and wide pore size, multilayer
arrangement of ions brings about a diffuse layer, i.e. to series connected capacitances that lower the
overall response of the electrode. Hence, for a given electrode surface area CIL may have a
maximum value when the carbon pore size closely matches the IL ion size as in the case of PFAK2
electrode. In PFAK3 carbons the pore size may be wide enough to allow multilayers of ions,
resulting in a diffuse layer capacitance. Therefore the thickness of the effective double layer may
differ significantly among samples and consequently affecting the capacitance response [11, 12].
The Cc space charge capacitance may also play a role in the different capacitive response of
the electrodes. Indeed, in polarized electrodes the decay of the electric potential takes place within
ca. 0.3-0.4 nm from the surface facing the electrolyte and pore walls thinner than 2-3 times such
screening length would account for charge storage limitation in the solid part of the porous
electrode/IL interface [16]. Total pore volumes increase in the order PFAK2 < PFAK2.5 < PFAK3
along with the decrease of the average pore wall thickness and, consequently, Cc may impact even
more in the global capacitive response of the PFAK3 electrode.
6
The CVs reported in Figure 2 also show that the negative scans are ca. 1 V wider than the
positive and this asymmetry, which is exploited to reach the high cell voltages of AEDLCs, implies
that the specific charge stored by negative electrodes will be higher than that of the positive
electrode. The CV profile in the negative domain of PFAK3 is almost box-shaped while the currents
of PFAK2.5 and PFAK2 decrease with increased electrode polarization. The PFAK3
charge/discharge cycles involve ca. 70 – 80 mAh g-1 which requires at least 0.8-0.9 cm3 of
PYR14TFSI per gram of carbon for charge counterbalancing. Sufficient electrolyte can be
accommodated in the PFAK3 carbon pores, with a pore volume V>1nm of 1.2 cm3 g-1 (Table 1),
which act as electrolyte reservoirs. PFAK2.5 and, in particular PFAK2, feature significantly lower
pore volume than PFAK3 and this leads to electrolyte starving effects and slow kinetics of ion
arrangements in the pores upon charge-discharge that are detrimental for a high charge storage
capacity of the supercapacitor.
The views discussed above are also supported by the results in Figure 3 which compares the
CVs of PFAK2, PFAK2.5 and PFAK3 in PYR14TFSI, N1113TFSI and PYR14FAP which have ions
with differing steric hindrance (Scheme 1). Figure 3 shows that when both carbon pore-to-ion size
ratio and carbon pore volume are reduced, capacitance and charge storage limitations arise at both
positive and negative electrode potential domains. As PFAK2 exhibits the finest porosity, an
increase in cation size moving from N1113TFSI to PYR14TFSI and in anion size from PYR14TFSI to
PYR14FAP results in a reduced current, both in the negative and positive potential domains (Figure
3a); this supports the idea that both ions, and not only the counterions, are involved in double-layer
formation. In the case of PFA-K2.5 (Figure 3b) the profiles obtained with N1113TFSI and
PYR14TFSI are very similar both for the negative and positive polarizations, whilst the current
decreases for the PYR14FAP electrolyte with the biggest anion, more pronouncedly in the positive
potential range. The CV response of PFAK3, with the widest pores, is similar in the three ILs and
seems slightly affected by ion chemistry (Figure 3c). These data indicate that the bulk properties of
the investigated ILs like conductivity, viscosity and normalized solvent polarity have a lower
7
impact on electrode electrochemical response and, hence, on AEDLC performance than the pore-toion size ratio. Indeed, while conductivity increases moving from PYR14FAP to PYR14TFSI and
N1113TFSI, along with the decrease of viscosity (see Table 2), the CV currents of the electrodes with
pore-to-ion size ratio high enough to allow easy ion access to carbon surface are similar in the
different ILs. This is the case of the CVs of PFAK-2.5 in N1113TFSI and PYR14TFSI and of PFAK3
in all the ILs. However, in the case of the latter carbon with the highest porosity it cannot be
excluded that the term affecting the capacitive response is a low Cc.
Given that the capacitance of PFAK2 was that more influenced by the nature of the IL, the
carbon was also tested in PYR14TFSI – N2222TFSI mixtures to evaluate the effect of a small,
symmetric quaternary ammonium cation. Figure 4 displays the CVs performed in neat PYR14TFSI
and in solutions of N2222TFSI – PYR14TFSI with molar percentage ratios of ca. 1 % and 30 % and
shows that the currents are significantly reduced even by the addition of a very small amount of
N2222TFSI. Ionic conductivity and normalized solvent polarity of the N2222TFSI-PYR14TFSI
solutions were the same as those of the neat IL. Therefore, the decreasing CV currents observed in
Figure 4 may be explained by a change in the electrode/IL interface properties with N2222TFSI
being involved in the double-layer. N2222+ has a high symmetry and a lower polarizability than that
of PYR14+ and this would influence the permittivity in the double layer to give a CIL value which is
lower than that in neat PYR14TFSI.
Carbon porosity is crucial also in view of the complete cell design of the IL-based AEDLCs.
Indeed, with a density of at least 1.4 g cm-3 like that of PYR14TFSI and N1113TFSI, the volume of
the ILs hosted in the carbon pores will significantly contribute to the total mass of the
supercapacitor. The 0.8-0.9 cm3 of IL per gram of carbon required to avoid electrolyte starving
effects in carbon pores would provide an amount of electrolyte in the complete cell that compares
that of carbon so that the specific capacitance and energy of the supercapacitor referred to its total
weight will be a half than that calculated considering only the electrode material loadings. IL based
on quaternary ammonium cations like those here investigated typically exhibit the highest
8
electrochemical stabilities. The ESW widths reported in Table 2 narrow down to ca 4.0 V when
high-surface area carbons are used, and this can be taken as limiting value of the maximum cell
voltage Vmax of AEDLCs. Thus, given that even by an optimized porosity with pore size of 2-3 nm
and pore volume of ~1 cm3 g-1 it may be difficult to achieve electrode specific capacitances higher
than 150-160 F g-1 at specific currents of interest for practical applications (at least 2 A g-1), the
maximum value of specific energy of an IL-based AEDLC which is deliverable between Vmax and
½ Vmax , i.e. Emax = 3/8 (C/4) Vmax2 (where C is the single electrode specific capacitance), is ca.
60 Wh kg-1 and reduces to ca. 30 Wh kg-1 when all the supercapacitor component weights are
included.
Conclusions
The capacitive response and charge storage capacity of porous carbons in solvent-free ILs is
modulated both by carbon porosity and IL properties. Only pores wider than IL ion size are easily
exploitable for the double-layer charge process. At moderate polarization, specific capacitance
increases with the accessible surface to a maximum value of ca. 150-160 F g-1 in PYR14TFSI,
N1113TFSI and PYR14FAP. When pores closely match IL size, electrolyte starving effects cause
limitations in charge storage capability upon the high electrode polarizations required for the charge
of high voltage AEDLCs. At high pore-to-ion size ratios and pore volumes, the bulk properties of
the PYR14TFSI, N1113TFSI and PYR14FAP ILs like conductivity, viscosity and solvent polarity have
a low impact on electrode responses but their density becomes a key parameter because the IL
hosted in carbon pores can significantly contribute to the overall supercapacitor weight. Hence,
carbon pore-to-ion size and pore volume have a high impact on AEDLCs performance and we
estimated that, even by an optimized carbon porosity, 30 Wh kg-1 should be the highest value of
specific energy deliverable from Vmax = 4 V to ½ Vmax for complete cells working with ammonium
or pyrrolidinium-based ILs.
9
References
1. M. Galiński, A. Lewandowski and I. Stępniak, Electrochim. Acta, 51, 5567 (2006).
2. A. Lewandowski and M Galinski, J. Power Sources, 173, 822 (2007).
3. M. Mastragostino and F. Soavi, in Encyclopedia of Electrochemical Power Sources, p. 649, J.
Garche, C. Dyer, P. Moseley, Z. Ogumi, D. Rand, B. Scrosati, Elsevier, Amsterdam (2009).
4. M. Lazzari, F. Soavi and M. Mastragostino, Fuel Cells (2010) DOI: 10.1002/fuce.200900198.
5. M. Lazzari, M. Mastragostino and F. Soavi, Electrochem. Commun., 9, 1567 (2007).
6. C. Largeot, C. Portet, J. Chmiola, P. L.Taverna, Y. Gogotsi and P. Simon, J. Am. Chem. Soc.,
130, 2730 (2008).
7. C. O. Ania, J. Pernak, F. Stefaniak, E. Raymundo-Piñero and F. Béguin, Carbon, 47, 3158
(2009).
8. C. L. Burket, R. Rajagopalan, A. P. Marencic, K. Dronvajjala, H. C. Foley, Carbon, 44,
2957(2006).
9. C. Reichardt, Green Chem., 7, 339 (2005).
10. J.P. Olivier, Carbon, 36, 1469 (1998)
11. Md. M. Islam, M. T. Alam, T. Okajima and T. Ohsaka, J. Phys. Chem. C, 113, 3386 (2009).
12. Md. M. Islam,. M. T. Alam, T. Okajima and T. Ohsaka, J. Phys. Chem. C, 112, 16568 (2008).
13. M. V. Fedorov, N. Georgi and A. A. Kornyshev, Electrochem. Commun., 12, 296 (2010).
14. M. V. Fedorov and A. A. Kornyshev, Electrochim. Acta, 53, 6835 (2008).
15. Y. Shim and H. J. Kim, ACS Nano, 4, 2345 (2010).
16. O. Barbieri, M. Hahn, A. Herzog, R. Kötz, Carbon, 43, 1303 (2005).
10
Figure Captions
Scheme 1. Ion structures and their approximate dimensions.
Figure 1. Pore size distributions (PSD) of activated carbons.
Figure 2. CVs of a) PFAK1, b) PFAK2, c) PFAK2.5, d) PFAK3 carbon electrodes in PYR14TFSI
(20 mV s-1 and 60°C).
Figure 3. CVs of: a) PFAK2, b) PFAK2.5, c) PFAK3 activated carbons in: 1) N1113TFSI, 2)
PYR14TFSI and 3) PYR14FAP ILs (20 mV s-1 and 60°C).
Figure 4. CVs of PFAK2 activated carbon in: 1) neat PYR14TFSI, 2) PYR14TFSI-1% N222TFSI, and
3) PYR14TFSI-30% N222TFSI at 20 mV s-1 and 60°C.
11
Table 1. Maximum pore size (d), specific pore volume and surface area of pores smaller (V<1 nm,
S<1nm) and wider (V>1 nm, S>1 nm) than 1 nm.
d
V<1 nm
S<1 nm
V>1 nm
S>1 nm
nm
cm3 g-1
m2 g-1
cm3 g-1
m2 g-1
PFAK1
1.5
0.23
595
0.07
90
PFAK2
2.0
0.45
1260
0.19
180
PFAK2.5
2.5
0.45
1230
0.39
460
PFAK3
4.5
0.21
650
1.16
1140
Carbon
12
Table 2. Formula weight (FW), density and viscosity (ν) at RT, ESW and ion conductivity (σ) at
60°C and normalized solvent polarity (ETN) of the ILs.
FW
IL
Density
-1
-1
ν
ESW
σ
+
mS cm-1
ETN
g mol
g mL
cP
V vs. Fc/Fc
PYR14TFSI
422
1.41
98
-3.4 / 2.4
6.0
0.59
N1113TFSI
382
1.44
79
-3.3 / 2.4
9.3
0.59
PYR14FAP
587
1.59
292
-3.2 / 2.4
2.2
0.84
13