Anion exchanged polymerized ionic liquids: High free volume single ion conductors

Article abstract of DOI:10.1016/j.polymer.2011.01.031

In this study, we investigate the isolated effect of anion type on the chemical, thermal, and conductive properties of imidazolium-based polymerized ionic liquids (PILs). PILs with various anions at constant average chain length were prepared by ion exchange with a water-soluble PIL precursor, (poly(1-[(2-methacryloyloxy)ethyl]-3-butylimidazolium bromide) (poly(MEBIm-Br)). NMR, IR, and elemental analysis confirm that anion exchange of ploy(MEBIm-Br) with bis(trifluoromethanesulfonyl) imide (TFSI), tetrafluoroborate (BF ₄), trifluoromethanesulfonate (Tf), and hexafluorophosphate (PF ₆) in water resulted in nearly fully exchanged PILs. As a function of anion type, the glass transition temperature plays a dominant role, but not the sole role in determining ion conductivity. Other factors affecting ionic conductivity include the size and symmetry of the anion and dissociation energy of the ion pair. Both the Vogel-Fulcher-Tammann (VFT) and Williams-Landel-Ferry (WLF) equations were employed to investigate the temperature dependent ionic conductivities. The C1g (9.03) and C2g (168 K) values obtained from the WLF regression of these PILs greatly deviate from the classical WLF values originally obtained from the mechanical relaxation of uncharged polymers (C1g = 17.44, C2g = 51.6 K) and the WLF values obtained from the conductive properties of other polymer electrolytes. This suggests that the fractional free volume (f (T_g) = B/(2.303C1g)) and Vogel temperature (T₀ = T _g - C2g) are strong functions of ion concentration, where high free volume allows for ion mobility at temperatures farther below the glass transition temperature of the polymer.

Full text of DOI:10.1016/j.polymer.2011.01.031

Polymer 52 (2011) 1309e1317
Contents lists available at ScienceDirect
Polymer
journal homepage: www.elsevier.com/locate/polymer
Anion exchanged polymerized ionic liquids: High free volume
single ion conductors
*
Yuesheng Ye, Yossef A. Elabd
Department of Chemical and Biological Engineering, Drexel University, Philadelphia, PA 19104, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
In this study, we investigate the isolated effect of anion type on the chemical, thermal, and conductive
properties of imidazolium-based polymerized ionic liquids (PILs). PILs with various anions at constant
average chain length were prepared by ion exchange with a water-soluble PIL precursor, (poly(1-
[(2-methacryloyloxy)ethyl]-3-butylimidazolium bromide) (poly(MEBImeBr)). NMR, IR, and elemental
analysis confirm that anion exchange of ploy(MEBImeBr) with bis(trifluoromethanesulfonyl) imide
(TFSI), tetrafluoroborate (BF4), trifluoromethanesulfonate (Tf), and hexafluorophosphate (PF6) in water
resulted in nearly fully exchanged PILs. As a function of anion type, the glass transition temperature plays
a dominant role, but not the sole role in determining ion conductivity. Other factors affecting ionic
conductivity include the size and symmetry of the anion and dissociation energy of the ion pair. Both the
Vogel-Fulcher-Tammann (VFT) and Williams-Landel-Ferry (WLF) equations were employed to investi-
gate the temperature dependent ionic conductivities. The C1g (9.03) and C2g (168 K) values obtained from
the WLF regression of these PILs greatly deviate from the classical WLF values originally obtained from
the mechanical relaxation of uncharged polymers (C1g ¼ 17.44, C2g ¼ 51.6 K) and the WLF values obtained
from the conductive properties of other polymer electrolytes. This suggests that the fractional free
Received 12 October 2010
Received in revised form
11 January 2011
Accepted 14 January 2011
Available online 22 January 2011
Keywords:
Ionic liquid
Conductivity
Polymer electrolyte
volume (f (Tg) ¼ B/(2.303Cg)) and Vogel temperature (T0 ¼ Tg ꢀ C2g) are strong functions of ion
1
concentration, where high free volume allows for ion mobility at temperatures farther below the glass
transition temperature of the polymer.
Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction
are mobile. The former alleviates shortcomings of liquid electrolytes
in electrochemical devices (e.g., leakage, stability), but usually
Polymerized ionic liquids (PILs), or polymers synthesized by
polymerizing ionic liquid monomers, have attracted recent atten-
tion and their application as polymer electrolytes, sorbents,
dispersing agents, and nanomaterials has recently been reviewed
[1,2]. Overall, the interests in PILs are many fold, but in general, PILs
are of interest due to the unique properties of ionic liquids (ILs) (e.g.,
negligible vapor pressure, nonflammability, high ionic conductivity,
a wide electrochemical window, and good chemical and thermal
stability) and the broad range of new polymer electrolytes that can
be realized with a highly tunable and facile synthesis [3e7]. Unlike
ionic liquid (IL)/polymer mixtures, the IL moiety is covalently
attached to the macromolecule in a PIL (see Scheme 1). Therefore,
PILs differ in that the organic cation or anion is restricted in mobility
compared to its more mobile counterion (i.e., single ion conductor),
unlike ILs or IL/polymer mixtures where both the cation and anion
results in lower ionic conductivity when compared to the latter.
Although PILs are of great interest as solid-state polymer electro-
lytes, surprisingly, there are relatively few reports on the ionic
conductivity of PILs [6e12]. It is evident based on these few inves-
tigations that a number of factors can impact ionic conductivity in
PILs, such as polymer chemistry, glass transition temperature,
molecular weight, and polymer morphology. However, there is still
a limited fundamental understanding of ion transport in PILs.
One method for increasing the ionic conductivity in a PIL is to
replace the mobile ion with another via ion exchange. Chen et al. [6]
observed enhancements in ionic conductivity in a PIL when the
tetrafluoroborate (BF4) anion was replaced with the bis(tri-
fluoromethanesulfonyl) imide (TFSI) anion. This was the result of
a significant reduction in the glass transition temperature of the
polymer, which facilitated higher segmental motion of the polymer
and therefore higher anion mobility. This is unique to PILs, where
these anion differences typically do not have the same effect in ILs.
Therefore, with a wide variety of anions available, understanding
ion conduction as a function of anion type in PILs is of great interest.
* Corresponding author. Tel.: þ1 215 895 0986; fax: þ1 215 895 5837.
E-mail address: elabd@drexel.edu (Y.A. Elabd).
0032-3861/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.polymer.2011.01.031

1310
Y. Ye, Y.A. Elabd / Polymer 52 (2011) 1309e1317
[
[
[
[
(1)
(2)
(4)
(3)
n
n
O
O
O
O
O
O
O
O
O
Cl
X
Br
Br
N
N
N
Br
N
N
N
O
O
O
F
B
F
F
F
F
O
F
N
F
F
F
F
S
S
S
F
F
P
F
C
C
C
F
F
F
O
O
O
F
F
F
X = BF4
X= PF6
X = TFSI
X = Tf
Scheme 1. Synthesis of anion exchanged PILs (poly(MEBIm-X)). (1) 2-bromoethanol, triethylamine, dichloromethane, room temperature, 18 h. (2) 1-butylimidazole, 40 ꢂC, 24 h.
(3) AIBN, DMF, 60 ꢂC, 6 h. (4) salt, DI water, room temperature, 48 h.
However, there are only a few investigations that have focused on
the effects of anion type on ionic conductivity in PILs.
oxide) (PEO)), which suggest that fractional free volume at the glass
transition temperature and Vogel temperature are significantly
different in single ion conductor PILs.
In several previous studies [9,13e15], PILs with different anions
were prepared by anion exchange of the IL monomer followed by
polymerization under different reaction conditions. This resulted in
PILs with a variety of molecular weights or different average chain
lengths, which also can affect the glass transition temperature and
ionic conductivity. Therefore, anion type and molecular weight
were convoluted parameters in these studies. These two parame-
ters can be deconvoluted if the PILs undergo anion exchange at the
polymer stage. Only a few investigations have produced PILs with
this method [16]. However, these investigations did not explore the
effect of anion type on ion conduction.
The purpose of this study was to synthesize PILs with different
anions at constant average polymer chain length and investigate the
effect of anion type on ion conductivity as well as other properties
related with these anion exchanged single ion conductors. The
effectiveness of exchanging anions at the polymer stage in PILs is
largely dependant on the anion type, PIL, and the experimental
procedure. In order to achieve a high purity exchange, solubility of
the PIL before and after exchange in the solvents used can be
a critical issue. In general, imidazolium-based ILs and PILs neutral-
ized with halide anions are water soluble, while these same ILs and
PILs neutralized with fluorine-containing anions are water insoluble
[17]. We utilize this difference in solubilities to achieve high purity
anion exchanged PILs.
2. Experimental
2.1. Materials
2-Bromoethanol (95%), methacryloyl chloride (97%, stabilized
with 200 ppm monomethyl ether hydroquinone (MEHQ)), triethyl-
amine (ꢁ99.5%), dichloromethane (ACS reagent, ꢁ99.5%, contains
50 ppm amylene stabilizer), magnesium sulfate (anhydrous,
ReagentPlusÒ, 99%), 1-butylimidazole (98%), 2,6-di-tert-butyl-4-
methylphenol (99%), diethyl ether (anhydrous, ꢁ99.7%, contains
1 ppm BHT inhibitor), azobisisobutyronitrile (AIBN, 97%), acetone
(HPLC grade, ꢁ99.9%), sodiumtetrafluoroborate (NaBF4, 98%), lithium
bis(trifluoromethanesulfonyl) imide (LiTFSI, 97%), lithium tri-
fluoromethanesulfonate (LiTf, 96%), lithium hexafluorophosphate
(LiPF6, 98%), lithium bromide (LiBr, 99.995% metals basis), acetonitrile
(anhydrous, 99.8%), N,N-dimethylformamide (DMF, ACS reagent,
ꢁ99.8% and HPLC grade, ꢁ99.9%), and methyl sulfoxide-d6 (DMSO-d6,
99.9%, contains 0.03% v/v TMS) were purchased from Aldrich. Ultra-
pure deionized (DI) water (resistivity w 16 MU cm) was used as
appropriate.
Specifically, in this study, an imidazolium-based IL monomer,
1-[(2-methacryloyloxy)ethyl]-3-butylimidazolium bromide (MEBIme
Br), was synthesized and subsequently polymerized using conven-
tional free radical polymerization. This PIL precursor, poly
(MEBImeBr), was subsequently exchanged with other fluoride-
containing anions, TFSI, BF4, trifluoromethanesulfonate (Tf), and
hexafluorophosphate (PF6), in water, where the resulting PIL
precipitated out of solution. For comparison, the same procedure
was also performed on the IL monomer. The chemistry, molecular
weight, glass transition temperature, thermal decomposition
temperature, and ion conductivity were characterized in all of these
PILs, where anion type was the sole variable. The effect of temper-
ature on ion conductivity was exploredwith the use of regressions to
both the VFTand WLF models. Interestingly, the WLF parameters (C1
and C2) for PILs are uniquely different than neutral polymers and
other salt doped polymer electrolytes (e.g., Li salt-poly(ethylene
2.2. Synthesis of ionic liquid (IL) monomer
The synthesis method for the imidazolium-containing mono-
mer, 1-[2-(methacryloyloxy)ethyl]-3-butylimidazolium bromide
(MEBImeBr), is shown in Scheme 1. A typical reaction included
adding a mixture of 39.37 g (0.315 mol) of 2-bromoethanol and
50 mL of dichloromethane to a three-neck 500 mL flask in an ice
bath. Under nitrogen, a mixture of 33.39 g (0.33 mol) of triethyl-
amine and 50 mL of dichloromethane was slowly added to the flask,
followed by a slow addition of a mixture of 31.36 g (0.3 mol) of
methacryloyl chloride and 50 mL of dichloromethane using an
addition funnel. The reaction mixture was stirred at room temper-
ature for 18 h and then filtered. The liquid filtrate was washed with
300 mL DI water four times. The water layer was removed using
a separation funnel and the residual water in the organic layer was
further removed with anhydrous magnesium sulfate. The organic

Y. Ye, Y.A. Elabd / Polymer 52 (2011) 1309e1317
1311
solvent was then removed by vacuum, whichyielded a clear liquid of
2-bromoethyl methacrylate (70% yield).
CHeN), 4.20e4.48 (m, 6H, NeCH2eCH2eO, NeCH2eCH2eCH2),,1.79
(s, 4H, CH2eC(CH3), NeCH2eCH2eCH2eCH3), 1.28 (s, 5H, NeCH2e
CH2eCH2eCH3, CH2eC(CH3)), 0.90 (s, 6H, NeCH2eCH2eCH2eCH3,
CH2eC(CH3)), 0.40e0.66 (m, 3H, CH2eC(CH3)). Anal. Found (Anal.
Calcd.): F, 21.15 (20.76); Br, 0.51 (0). poly(MEBImeTFSI): 9.18 (s, 1H,
NeCH]N), 7.66e7.78 (d, 2H, NeCH]CHeN), 4.19e4.47 (m, 6H,
NeCH2eCH2eO, NeCH2eCH2eCH2), 1.78 (s, 4H, CH2eC(CH3),
NeCH2eCH2eCH2eCH3), 1.27 (s, 5H, NeCH2eCH2eCH2eCH3,
CH2eC(CH3)), 0.90 (s, 6H, NeCH2eCH2eCH2eCH3, CH2eC(CH3)),
0.44e0.71 (m, 3H, CH2eC(CH3)). Anal. Found (Anal. Calcd.): F, 21.09
(22.02); Br, < 0.25 (0). poly(MEBImeTf): 9.19 (s, 1H, NeCH]N),
7.74e7.81 (d, 2H, NeCH]CHeN), 4.20e4.50 (m, 6H, NeCH2e
CH2eO, NeCH2eCH2eCH2), 1.79 (s, 4H, CH2eC(CH3), NeCH2eCH2e
CH2eCH3), 1.28 (s, 5H, NeCH2eCH2eCH2eCH3, CH2eC(CH3)), 0.89
(s, 6H, NeCH2eCH2eCH2eCH3, CH2eC(CH3)), 0.42e0.69 (m, 3H,
CH2eC(CH3)). Anal. Found (Anal. Calcd.): F,14.47 (14.75); Br, 0.81 (0).
poly(MEBImePF6): 9.11 (s, 1H, NeCH]N), 7.64e7.75 (d, 2H,
NeCH]CHeN), 4.18e4.43 (m, 6H, NeCH2eCH2eO, NeCH2e
CH2eCH2), 1.78 (s, 4H, CH2eC(CH3), NeCH2eCH2eCH2eCH3), 1.28
(s, 5H, NeCH2eCH2eCH2eCH3, CH2eC(CH3)), 0.90 (s, 6H,
NeCH2eCH2eCH2eCH3, CH2eC(CH3)), 0.43e0.66 (m, 3H, CH2eC
(CH3)). Anal. Found (Anal. Calcd.): F, 28.47 (29.81); Br, 0.57 (0).
A typical quaternization reaction consisted of adding 40.46 g
(0.21 mol) of 2-bromoethyl methacrylate, 26.03 g (0.21 mol, equi-
molar) of 1-butylimidazole, and a small amount of 2,6-di-tert-
butyl-4-methylphenol (inhibitor) to a 250 mL flask. The mixture
was stirred in an oil bath at 40 ꢂC for 24 h and yielded a viscous
liquid. The resulting MEBImeBr monomer was dissolved in 30 mL
dichloromethane and then re-precipitated three times in 200 mL
diethyl ether in an ice bath. The purified MEBImeBr monomer was
a clear viscous liquid (82% yield). 1H NMR (UNITYINOVA 500 MHz,
DMSO-d6, ppm) and elemental analysis: 9.37 (s, 1H, NeCH]N),
7.86e7.88 (d, 2H, NeCH]CHeN), 6.03 (s, 1H, HCH]C(CH3)), 5.76 (s,
1H, HCH]C(CH3)), 4.53 (m, 2H, NeCH2eCH2eO), 4.48 (m, 2H,
NeCH2eCH2eO), 4.21 (t, 2H, NeCH2eCH2eCH2eCH3), 1.84 (s, 3H,
CH2]C(CH3)), 1.76 (m, 2H, NeCH2eCH2eCH2eCH3), 1.22 (m, 2H,
NeCH2eCH2eCH2eCH3), 0.89 (t, 3H, NeCH2eCH2eCH2eCH3).
Anal. Found (Anal. Calcd.): C, 46.35 (49.21); H, 7.11 (6.69); O, 14.98
(10.09); N, 8.21 (8.83); Br 23.62 (25.18).
2.3. Synthesis of polymerized ionic liquid (PIL)
The imidazolium-containing polymer, poly(MEBImeBr), was
synthesized by the conventional free radical polymerization as
shown in Scheme 1. A typical polymerization included adding
a mixture of 30.04 g (9.27 ꢃ 10ꢀ2 mol) MEBImeBr monomer and
63.0 g DMF to a 250 mL flask. The reaction mixture was then purged
with nitrogen for 30 min followed by an addition of 31.4 mg
(1.91 ꢃ10ꢀ4 mol) AIBN initiator to initiate polymerization, which was
carried out at 60 ꢂC for 6 h. The reaction mixture was then diluted
with DMF and precipitated into acetone followed by washing with
fresh acetone several times. The precipitate was then dried under
vacuum at 60 ꢂC for 72 h yielding 16.80 g of poly(MEBImeBr) (56%
yield). 1H NMR (UNITYINOVA 500 MHz, DMSO-d6, ppm) and
elemental analysis: 9.97 (s,1H, NeCH]N), 8.10e8.17 (d, 2H, NeCH]
CHeN), 4.32e4.67 (m, 6H, NeCH2eCH2eO, NeCH2eCH2eCH2),
1.84 (s, 4H, CH2eC(CH3), NeCH2eCH2eCH2eCH3), 1.30 (s, 5H,
NeCH2eCH2eCH2eCH3, CH2eC(CH3)), 0.91 (s, 6H, NeCH2eCH2e
CH2eCH3, CH2eC(CH3)), 0.32e0.64 (m, 3H, CH2eC(CH3)). Anal.
Found (Anal. Calcd.): C, 47.69 (49.21); H, 6.89 (6.69); O,13.08 (10.09);
N, 8.10 (8.83); Br 23.64 (25.18).
2.5. Characterization
1H NMR spectra were collected using a UNITYINOVA 500 MHz
spectrometer at room temperature with DMSO-d6 as the solvent.
Elemental analysis was conducted at Atlantic Microlab, Inc. in Nor-
cross, GA. The infrared spectra were collected at room temperature
using a Fourier transform infrared spectroscopy (FTIR) spectrometer
(Nicolet 6700 Series; Thermo Electron) equipped with a single-
reflection ATR attachment (Specac, Inc., Silver GateÔ, zinc selenide
crystal). The molecular weights of PILs were determined by gel
permeation chromatography (GPC) at 40 ꢂC using a Waters GPC
system (breeze 2) equipped with two Styragel columns (Styr-
agel@HR 3 and Styragel@HR 4) and a 2414 reflective index (RI)
detector. PILs were dissolved in a mixture of DMF and 0.05 M LiBr.
GPC measurements were performed at a flowrate of 1.0 mL/min at
40 ꢂC using polyethylene glycol/polyethylene oxide (PEG/PEO) as
standards. Thermal degradation temperatures (Td) were measured
by thermal gravimetric analysis (TGA, TA Instruments, Q50) at
a heating rate of 10 ꢂC/min under nitrogen environment. Glass
transition temperatures (Tg) were measured with a differential
scanning calorimeter (DSC) (TA Instruments, Q2000) under nitrogen
environment (50 mL/min) using the method of heat/cool/heat at the
same rate of 10 ꢂC/min over a temperature range of ꢀ40 to 180 ꢂC.
The Tg was determined using the mid-point method on the second
heating cycle thermogram. Ionic conductivity of PIL films and IL
monomers was measured with an AC impedance system (Solartron,
1260 impedance analyzer, 1287 electrochemical interface, Zplot
software) between 102 Hz and 106 Hz. PIL films with thickness
2.4. Anion exchange reactions
The anion exchange reactions for PILs (Scheme 1) and IL mono-
mers (see Supplementary data Scheme S1) were performed inwater.
A general procedure for the preparation of PILs by anion exchange
with poly(MEBImeBr) is given as follows: 3.16 g (2.87 ꢃ 10ꢀ2 mol)
NaBF4 was dissolved in 10 mL of DI water and this solution was then
added dropwise into 1.01 g poly(MEBImeBr) aqueous solution with
an anion mole ratio of BF4/Br ¼ 9/1 mol/mol. The resulting water-
insoluble polymer, poly(MEBImeBF4), precipitated out of the water
phase immediately. The reaction mixture was stirred for 48 h, fol-
lowed by re-precipitating and washing in fresh DI water for 72 h.
Silver nitrate testing showed that the Br anion was present in the
aqueous phase immediately after the exchange reaction, but no
AgBr precipitated after the washing step. The resulting anion
exchanged PIL was dried in vacuum oven for at least 24 h. A similar
procedure was carried out with other salts, LiTFSI, LiTf, and LiPF6, to
ranging between 100 and 400 mm were prepared by solution casting
from acetonitrile on Teflon Petri dishes at ambient conditions. The
films were dried in a fume hood at room temperature and films
were cut to w30 mm ꢃ 5 mm before they were completely dried.
The polymer films were further dried under vacuum at 120 ꢂC for at
least 24 h, and then stored in a desiccator until use. The conductivity
of IL monomers and PILs was measured in a cell with four-parallel
electrodes (four-point method) in an environmental chamber
(Tenney, BTRS model) with controlled temperature at the fixed
relative humidity of 10%. An alternating current was applied to
the outer electrodes and the real impedance or resistance, R, was
measured between the two inner reference electrodes. Resistance
was determined from the x-intercept of the imaginary versus real
impedance data over a high frequency range. The conductivity
produce
poly(MEBImeTFSI),
poly(MEBImeTf),
and
poly
(MEBImePF6), respectively. 1H NMR and elemental analysis results
for all four PIL samples are shown below, where a residual amount of
Br was present in each sample (<4 mol%). 1H NMR (UNITYINOVA
500 MHz, DMSO-d6, ppm) and elemental analysis: poly
(MEBImeBF4): 9.11 (s, 1H, NeCH]N), 7.70e7.78 (d, 2H, NeCH]

1312
Y. Ye, Y.A. Elabd / Polymer 52 (2011) 1309e1317
values were calculated from the equation
s
¼ L/AR, where
s
is
of the PILs in DMF, where there was a monomodal distribution of
hydrodynamic size at 0.05 M LiBr salt concentration (see Supple-
mentary data Fig. S2a). A bimodal distribution in hydrodynamic size
was observed without the addition of salt (see Supplementary data
Fig. S2b). Fig. 3 shows the GPC chromatograms for PILs measured in
this study, where monomodal peaks were observed indicating good
separation with 0.05 M LiBr salt concentration. In literature, both PS
[24] and PEG/PEO [25] have been used as standards for molecular
weight measurements on polyelectrolytes. Table 2 lists the molec-
ular weight average and distribution of PILs against both PS and
PEG/PEO standards, respectively. Interestingly, the molecular weight
average and distribution against PS standards are much higher and
narrower than those against PEG/PEO standards. The poly-
dispersities against PS standards ranged from 1.21 to 1.31 and are
much smaller than what is typically observed for polymers
synthesized by free radical polymerization. The reason for this is
that there is a significant interaction between PS and the Styragel
packing materials (crosslinked poly(styrene-divinylbenzene) parti-
cles) in the DMF solvent [26,27]. As a result, the elution time for
polystyrene becomes longer (see Supplementary data Fig. S3). Thus,
the molecular weight averages and polydispersities against PS
standards are often significantly larger than their true values. In
contrast, polydispersity values against the PEG/PEO standards give
more reasonable values, ranging from 2.45 to 2.77. This indicates
that molecular weights of PILs measured against PEG/PEO standards
are closer to their true molecular weights when compared to PS
standards, which interact with column packing materials.
conductivity (S/cm) and L and A are the distance between two inner
electrodes and cross section area of the polymer film (A ¼ Wl, W is
the film strip width and l is the film thickness), respectively. For
conductivity experiments, samples were equilibrated for 2 h at each
temperature followed by 12 measurements over 3 h taken over
15 min increments. The values reported are an average of these
steady-state measurements. Measurements for PIL films and IL
monomers were carried out in a Teflon-coated stainless steel cell
and a solid Teflon well-like cell, respectively. More details on these
two cells and experimental procedures can be found elsewhere [18].
A small amount of di-tert-butyl-4-methylphenol (inhibitor) was
added into the cell with the IL monomers to prevent polymerization
during conductivity experiments.
3. Results and discussion
3.1. Synthesis and chemical characterization of anion exchanged
PILs
The synthesis of the IL monomer, MEBImeBr, is shown in
Scheme 1. The resulting IL monomer was a water-soluble clear
viscous liquid. The homopolymer, poly(MEBImeBr), was synthe-
sized by conventional free radical polymerization (also shown in
Scheme 1). This resulting polymer was soluble in water or meth-
anol, but not soluble in acetone or diethyl ether. Thus, in this work,
poly(MEBImeBr) was chosen as the PIL precursor to exchange with
other anions in water. A general anion exchange reaction scheme is
also shown in Scheme 1. The chemical structures of the anion
exchanged PILs were confirmed by chemical shifts in the 1H NMR
spectra as listed in Table 1, e.g., from 9.97 ppm (NeCH]N of poly
(MEBImeBr)) to 9.11 ppm (NeCH]N of poly(MEBImeBF4)). The 1H
NMR spectra shown in Fig. 1 also indicate that the C(2)-proton peak
(9.97 ppm) associated with Br anion is not present in exchanged
PILs, suggesting that Br has been replaced with a new anion and the
amount of Br was reduced to a low level that was not detectable by
1H NMR. Further, elemental analysis shows that the halide residue
was negligibly small in the resulting anion exchanged PILs (<4 mol
%). This data provides evidence that anion exchange reactions in
water resulted in nearly fully exchanged PILs.
3.2. Thermal properties of PILs
Fig. 4 shows the thermal stability of PILs characterized by TGA
under nitrogen environment. The thermal decomposition temper-
atures (Td) of PILs were measured at 5% loss and are listed in Table 3.
The precursor PIL with Br anion has the lowest decomposition
temperature of all PILs (Td ¼ 534 K), while the TFSI-exchanged PIL
has the highest decomposition temperature (Td ¼ 636 K). The
overall thermal stability of these PILs scales with anion type as:
Br < PF6 < BF4 < Tf < TFSI. The order of relative stability follows
a similar trend reported in literature for ILs [28e30] suggesting that
the decomposition temperature of the PIL is strongly dependant on
the anion type. With respect to the decomposition mechanism, it is
generally accepted that the nucleophilicity of an anion affects the
thermal stability and a low decomposition temperature associated
with a halide anion is attributed to the attack of highly nucleophilic
halide on the primary alkyl group via an SN2 reaction with an alkyl
halide as a byproduct [31]. As compared with Br anion, it is well
known that other anions are more thermally stable than halides.
However, the detailed decomposition mechanism for these are still
not well understood. For example, the TFSI anion is a weak or non-
nucleophilic anion and one possible pathway for the decomposi-
tion of the TFSI anion is that it undergoes degradation via sulfur
dioxide release instead of dealkylation or proton transfer [32],
which makes it more difficult to thermally decompose. Also, from
Fig. 4, it is clear that PILs paired with PF6, BF4, Tf, and TFSI anions
undergo a one-step thermal decomposition, while the precursor PIL
paired Br appears to undergo a two-step thermal decomposition.
The two-step decomposition agrees with the work of Vygodskii
et al. [33] For the poly(MEBImeBr), the first thermal decomposition
step (Td ¼ 261 ꢂC) can be attributed to the degradation of the Br
anion, while the second decomposition step at w 340 ꢂC is in good
agreement with the thermal decomposition temperature of poly
(vinylimidazole) (Td w 335 ꢂC) [34].
Fig. 2 shows the IR spectra for the exchanged PILs at ambient
conditions. The characteristic infrared bands found in this study
(BF4, 1046, 1011 cmꢀ1; TFSI, 1346, 1130 cmꢀ1; Tf, 1025, 755 cmꢀ1
;
PF6, 810 cmꢀ1) are consistent with literature (BF4, 1045, 1015 cmꢀ1
[19]; TFSI, 1348, 1136 cmꢀ1 [20]; Tf, 1032, 755 cmꢀ1 [21,22]; PF6,
808 cmꢀ1 [20]). Note that at the ambient conditions, there is
significant amount of water (OeH stretching band at 3407 cmꢀ1) in
the precursor, poly(MEBImeBr), while there is a negligible amount
of water in the exchanged PILs.
Molecular weight and distribution of polyelectrolytes is often
measured by GPC with a small amount of salt added to screen
electrostatic repulsions and therefore minimize polymer chain
aggregation [23]. In this work, results from dynamic light scattering
(DLS) indicate that the addition of LiBr salt reduced the aggregation
Table 1
Chemical shifts in 1H NMR spectra for PILs.
PIL
NeCH]N (ppm)
NeCH]CHeN (ppm)
Poly(MEBImeBr)
Poly(MEBImeBF4)
Poly(MEBImeTFSI)
Poly(MEBImeTf)
Poly(MEBImePF6)
9.97
9.11
9.18
9.19
9.11
8.10, 8.17
7.70, 7.78
7.66, 7.78
7.74, 7.81
7.64, 7.75
The glass transition temperatures (Tg) of the PILs as a function of
anion type are listed in Table 3. For the same polymer chain length,
the Tg values follow the following order: Br > PF6 > BF4 > Tf > TFSI.

Y. Ye, Y.A. Elabd / Polymer 52 (2011) 1309e1317
1313
Fig. 1. 1H NMR spectra of the PIL precursor and anion exchanged PILs. X on figure represents the corresponding anion.
Overall, there is a 95 K depression in Tg from poly(MEBImeBr)
(Tg ¼ 375 K) to poly(MEBImeTFSI) (Tg ¼ 280 K). Therefore by
removing the effect of polymer chain length (each PIL has the same
degree of polymerization), the value of Tg here is largely dependent
on the plasticizing effect of the anion. In particular, the TFSI anion
has a more significant impact on Tg depression compared to other
anions, such as Tf and BF4. This can be attributed to not only the
difference in size, but also lower symmetry, extensive charge
delocalization [35], and the flexibility of the TFSI anion [36]. Also, in
the temperature range from 233 K to 573 K, no melting behavior
0.5
0.4
0.3
0.2
0.1
0
-0.1
10
12
14
16
18
Elution volume (ml)
Fig. 3. GPC chromatograms for PILs using DMF columns: poly(MEBImeBr) (black), poly
(MEBImeBF4) (blue), poly(MEBImePF6) (green), poly(MEBImeTf) (magenta), poly
(MEBImeTFSI) (red). (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article).
Fig. 2. Infrared spectra of PILs at ambient conditions. Spectra offset for clarity.

1314
Y. Ye, Y.A. Elabd / Polymer 52 (2011) 1309e1317
Table 2
Table 3
Molecular weights and polydispersities of PILs.
Thermal degradation temperatures (Td) and glass transition temperatures (Tg) of
PILs.
PIL
PEG/PEO standard
PS standard
PIL
Td (K)
Tg (K)
Mw
PDI
Mw
PDI
Poly(MEBImeBr)
534
586
636
608
579
375
358
280
337
367
Poly(MEBImeBr)
Poly(MEBImeBF4)
Poly(MEBImeTFSI)
Poly(MEBImeTf)
Poly(MEBImePF6)
58,700
61,100
43,080
56,680
54,970
2.56
2.77
2.54
2.45
2.53
367,000
427,000
312,000
356,000
340,000
1.30
1.27
1.21
1.32
1.32
Poly(MEBImeBF4)
Poly(MEBImeTFSI)
Poly(MEBImeTf)
Poly(MEBImePF6)
was observed for all of these PILs, which suggests that these PILs are
amorphous in nature.
PILs. Also, the order in conductivity differs from the PILs, where the
differences in Tg (see Supplementary data Table S4) are minor
compared to the PILs; maximum difference of 11 K for the IL
monomers compared to 95 K for the PILs. Unlike the glass transi-
tion-dependant PIL system, the IL monomer conductivity differ-
ences are dictated by other factors; most probably liquid viscosity,
which has been shown by many other investigators for common ILs
[37,38].
The glass transition dependence for the PIL system can be seen
more clearly in Fig. 5b, where conductivity is plotted versus
a normalized temperature, Tg/T. However, the data does not
collapse onto one single curve. This suggests that other factors in
addition to Tg can impact ion transport. After normalizing for Tg,
poly(MEBImeTFSI) has the lowest conductivity, where it shows the
highest conductivity in Fig. 5a (before normalization). The differ-
ences in conductivity may be attributed to the symmetry and size of
the anions and the dissociation energy of the ion pairs, where the
PILs with BF4 and PF6 have the highest conductivities when the
effects of Tg are removed [39].
3.3. Ionic conductivity
Fig. 5a shows the measured ionic conductivities of PILs as
a function of temperature. Note that the precursor polymer, (poly
(MEBImeBr)), is water soluble and thus only the hydrophobic PILs
are compared here under dry conditions (10% RH). The conductiv-
ities of the PILs follow the order: TFSI > Tf > BF4 > PF6, which is the
reverse order of Tg shown in Table 3. These results suggest that
polymer chain relaxation plays a dominant role in ion transport.
Specifically, the PIL with TFSI is significantly higher in ionic
conductivity compared to the other PILs and this coincides with the
significantly lower Tg (Table 3). Also, with increasing temperature,
the difference in ionic conductivity between PILs with different
anions decreases, suggesting that at a high temperature the
importance of Tg becomes less pronounced.
Compared to the PILs, the IL monomers provide an upper limit of
ionic conductivity since the restriction of polymer chain relaxation
is eliminated and a large number of ions and their counter ions are
available for ion transport. Fig. S4 (see Supplementary data) shows
that the ionic conductivity of the IL monomers is approximately
two to three orders of magnitude higher than their corresponding
PILs. The conductivities of the IL monomers follow the order:
TFSI > BF4>Tf > PF6, where the difference in conductivity for TFSI
compared to the other anions is not as significant compared to the
3.3.1. VFT regression
For
a solid polymer electrolyte system, the temperature
dependent ionic conductivity is often regressed to the Vogel-
Fulcher-Tammann (VFT) equation [40e42]:
ꢀ
ꢁ
b
s
ðTÞ ¼ s0exp ꢀ
(1)
T ꢀ T0
where s0 (S cmꢀ1) is the infinite temperature conductivity, b (K) is
a constant that can be related to the Arrhenius activation energy,
and T0 (K) is the Vogel temperature. T0 has been interpreted as the
temperature at which the configurational entropy vanishes, [43] the
polymer relaxation time becomes infinite, [44] or the mobility of
ions goes to zero [45] and is typically w 50 K below the measured Tg
of the polymer [46]. The physical meaning of the VFT equation is not
straightforward, but has been explained in terms of the free volume
model [47] and the configurational entropy model [43]. Fig. 6a
shows the temperature dependant ionic conductivity for the PILs
regressed to the VFT equation (solid lines), which requires three
fitting parameters (regressed values listed in Table 4).
Alternatively, Eq. (1) can be recast into the following form:
ꢀ
ꢀ
ꢁꢁ
1
1
s
ðTÞ ¼
s
ðTrÞexp ꢀ b
ꢀ
(2)
T ꢀ T0 Tr ꢀ T0
where
s
(Tr) is the conductivity at a reference temperature, Tr. In
this study, we choose an experimentally measured value for the
reference temperature, Tr ¼ Tg þ 50 K. The introduction of Tr in Eq.
(2) reduces the number of fitting parameters in the VFT equation
from three to two, b and T0. Fig. 6b shows the PIL conductivity data
regressed to Eq. (2), where the resulting fitting constants are also
listed in Table 4. Interestingly, the three-parameter regression
results in Tg ꢀ T0 values close to the expected 50 K, whereas the
two-parameter regression results in Tg ꢀ T0 values that are much
larger (ranging from 116 to 146 K).
Fig. 4. Thermal stability of PILs: poly(MEBImeBr) (black), poly(MEBImeBF4) (blue),
poly(MEBImePF6) (green), poly(MEBImeTf) (magenta), poly(MEBImeTFSI) (red). (For
interpretation of the references to color in this figure legend, the reader is referred to
the web version of this article).

Y. Ye, Y.A. Elabd / Polymer 52 (2011) 1309e1317
1315
-2
-2
-3
-4
-5
-6
-7
b
a
-3
-4
-5
-6
-7
0.5
0.6
0.7
0.8
0.9
1
2.2
2.4
2.6
2.8
3
3.2
3.4
1000/T (K-1)
Tg/T
Fig. 5. Ionic conductivity versus (a) 1000/T (Kꢀ1) and (b) Tg/T of PILs: poly(MEBImeBF4) (blue diamonds), poly(MEBImePF6) (green squares), poly(MEBImeTf) (magenta triangles),
poly(MEBImeTFSI) (red circles). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
3.3.2. WLF regression
Fig. 7 shows a master regression of all the PIL conductivity data
to the modified WLF equation (Eq. (4)) and Table 5 lists the
regressed values, C1r and C2r , from this master regression and also
each individual PIL (individual regressions not shown here). Table 5
also lists the C1g and C2g values calculated from Eqs. (5) and (6). Note
that although the WLF (Eq. (3)) and VFT (Eq. (1)) equations have
different origins, they are actually mathematically equivalent,
where b ¼ 2.303C1gC2g and T0 ¼ Tg ꢀ C2g. Therefore, the C2g and
2.303C1gC2g values listed in Table 5 determined from the WLF
regression are in a good agreement with the Tg ꢀ T0 and b values
listed in Table 4 determined from the two-parameter VFT regres-
sion. This analysis confirms the high Tg ꢀ T0 (C2g) value observed
from the two-parameter VFT regression. Actually, the master WLF
regression shows that C1g and C2g values are 9.03 and 168 K,
respectively, which greatly deviate from the classical values origi-
nally obtained from mechanical relaxation of uncharged polymers
(C1g ¼ 17.44, C2g ¼ 51.6 K).
The temperature dependence of ionic conductivity for solid
polymer electrolytes (e.g., Li salt-poly(ethylene oxide) (PEO)
mixtures) can also be regressed to the Williams-Landel-Ferry (WLF)
equation:
À
Á
ꢀ
ꢁ
C1g T ꢀ Tg
2
s
ðTÞ
À
Á
log
¼
À
Á
(3)
Cg þ T ꢀ Tg
s
Tg
where
s
(T) and
s
(Tg) are the conductivities at a given temperature,
T, and the glass transition temperature, Tg, and Cg and C2g (K) are
1
fitting parameters. Similarly, the WLF equation can be recast in
terms of an experimentally measurable reference temperature, Tr:
ꢀ
ꢁ
C1r ðT ꢀ TrÞ
s
ðTÞ
log
¼
(4)
s
ðTrÞ
C2r þ ðT ꢀ TrÞ
where C1g and C2g in Eq. (3) can be related to C1r and C2r with the
following equations:
Based on the free volume theory developed by Doolittle [48],
Williams et al. [46] interpreted the physical meaning of C1g and C2g
by relating them to the fractional free volume, f, and the thermal
C1r C2r
C1g
¼
(5)
(6)
À
Á
expansion coefficient, a, at Tg.
C2r þ Tg ꢀ Tr
À
Á
B
f Tg
¼
(7)
2:303 C1g
C2g ¼ Cr þ Tg ꢀ Tr
2
-2
-3
-4
-5
-6
-7
-2
-3
-4
-5
-6
-7
a
b
2.2
2.4
2.6
2.8
3
3.2
3.4
2.2
2.4
2.6
2.8
3
3.2
3.4
1000/T (K-1)
1000/T (K-1)
Fig. 6. VFT regression of the temperature dependent ionic conductivity of PILs: (a) three-parameter regression (Eq. (1)) and (b) two-parameter regression (Eq. (2)). PILs: poly
(MEBImeBF4) (blue diamonds), poly(MEBImePF6) (green squares), poly(MEBImeTf) (magenta trianlges), poly(MEBImeTFSI) (red circles). (For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this article).

1316
Y. Ye, Y.A. Elabd / Polymer 52 (2011) 1309e1317
Table 4
Table 5
VFT equation regression values of temperature dependent conductivity data for PILs.
WLF equation regression values of temperature dependent conductivity data for
PILs.
PIL
Three-parameter regression (Eq.
(1))
Two-parameter
regression (Eq. (2))
PIL
C1r
C2r
C1g
C2g
2.303 C1gC2g f(Tg)
a(Tg)
Poly(MEBImeBF4) 5.79 188 7.89 138 2508
Poly(MEBImeTFSI) 5.72 166 8.18 116 2185
0.055 4.0 ꢃ 10ꢀ4
0.053 4.5 ꢃ 10ꢀ4
0.054 4.3 ꢃ 10ꢀ4
0.057 4.0 ꢃ 10ꢀ4
0.049 2.9 ꢃ 10ꢀ4
s0
b
(K)
T0
(K)
TgeT0
(K)
b
(K)
T0
(K)
TgeT0
(K)
(S cmꢀ1
)
Poly(MEBImeTf)
Poly(MEBImePF6) 5.61 196 7.53 146 2532
All PILs 6.97 218 9.03 168 3494
5.75 175 8.05 125 2318
Poly(MEBImeBF4)
Poly(MEBImeTFSI)
Poly(MEBImeTf)
Poly(MEBImePF6)
0.029
0.152
0.061
0.009
302
466
373
232
303
215
282
318
55
65
55
49
2507
2187
2323
2550
220
164
216
221
138
116
121
146
Understanding the considerable deviation of C1g (9.03) and C2g
(168 K) for the PILs studied here compared to classical values
determined from uncharged polymers (C1g ¼ 17.44 and C2g ¼ 51.6 K)
is of great interest. WLF constants obtained from charged polymer
systems (solid polymer electrolytes) would serve as a better
comparison. There are few reports of WLF regressions to PIL
conductivity data. However, there are a number of studies that have
examined WLF behavior of Li salt-PEO polymer electrolyte systems.
The C1g and C2g values were examined as a function of salt type [52],
salt concentration [53,54], as well as, polymer chain length [55].
However, there is no universal agreement on the C1g and C2g values
for these systems. Although some studies showed that C1g values are
close to 17.44 [56] or C2g values are close to 51.6 [57], most studies
showed deviations of C1g or C2g from these classical values. Specifi-
cally, numerous studies report values for C1g at w7e12 for the Li
salt-PEO system (8.5e10.6 [58], 7.8e12.5 [52], 9.6e11.8 [59], 8.0
[60], 8.1e12.0 [50]). In particular, Watanabe et al. [53] studied the
effect of Li salt concentration in PEO on C1g and C2g values, where C1g
decreased from 13.5 to 10.1 and C2g increased from 39.4 to 97.9 with
increasing salt concentration from 0.015 to 0.062 ([salt]/[EO unit]).
However, a detailed explanation for the reason behind these
deviations has not yet been presented.
À
Á
B
a
Tg
¼
(8)
2:303 C1gC2g
In Eqs. (7) and (8), we assume B ¼ 1. Originally, Doolittle also
assumed B ¼ 1 [48]. Others have calculated B from volume-
temperature data of polymers (without pressure dependence) and
estimated B ¼ 0.9 ꢄ 0.3 or 1.6 ꢄ 0.6 depending on the estimation
method [49]. Most reports, however, have adopted B equal to unity,
including reports on charged polymers [50]. The values for f(Tg) and
a
(Tg), calculated from the WLF regressed Cg and C2g values, are listed
1
in Table 5, where the fractional free volumes for the PILs are on
average double that of what has been classically observed for
uncharged polymers (f(Tg) ¼ 0.025).
One should note the universal parameters for uncharged poly-
mers (C1g ¼ 17.44 and C2g ¼ 51.6 K, f(Tg) ¼ 0.025) are based on
a master regression of data from many polymers. If one were to
determine these parameters from regressing data for each polymer
individually (data in Ref. [51]), the average and standard deviation
of all of these individual regressed values are C1g ¼ 16.79 ꢄ 5.43;
C2g ¼ 63.7 ꢄ 28.1 K; f (Tg) ¼ 0.028 ꢄ 0.01. For our PILs, regressions
from both individual polymers and a master regression are listed in
Table 5. The average and standard deviation from our individual
In this work, we report C1g values that are far from uncharged
polymers [46]. This suggests a higher fractional free volume at Tg for
PILs. Also, the C2g value for the PILs in this study deviates signifi-
cantly from both uncharged polymers and the Li salt-PEO charged
polymer system, suggesting a lower Vogel temperature (the
temperature at which free volume becomes zero). At temperatures
below Tg, the interaction of the bulky cation and anion in imida-
zolium-based PILs results in increased free volume due to the
impact of steric hindrance on polymer chain packing [61]. It is
interesting to note that the typical Li salt-PEO systems investigated
has a salt concentration on the order of 0.1 mol Li salt/mol EO
repeat unit, whereas the PIL system in this study has an ionic
concentration on the order of 1 mol IL salt/mol MMA repeat unit. It
is also important to note that Li salt-PEO systems is PEO doped with
a Li salt, where both cation (Li) and anion (e.g., TFSI) are both
mobile compared to the PIL in this study, where the cation is
tethered to the polymer backbone and therefore is a single anion
conductor. Therefore, these results suggest that the higher ion
concentration and tethered cations (single anion conductor) in PILs
may be key factors that result in lower Vogel temperatures
compared to salt doped polymer electrolytes.
regressions are C1g
¼
7.92
ꢄ
0.28; C2g
¼
131 ꢄ 13 K;
f
(Tg) ¼ 0.055 ꢄ 0.001. Therefore, despite the high deviation of the
WLF regressed values for uncharged polymers, there is still
a significant difference compared to the WLF values obtained in this
study on PILs.
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
4. Conclusions
In this study, imidazolium-based PIL single ion conductors of
constant average chain length were paired with different anions,
BF4, TFSI, Tf, and PF6, at the polymer stage to solely explore the
effect of anion type on the chemical structures, thermal properties,
and ion conduction. Chemical analysis confirms that the anion
exchange reactions for both PILs and IL monomers in water resulted
in nearly fully exchanged PILs and IL monomers. GPC results show
that the molecular weights of PILs measured against PEG/PEO
-40 -20
0
20
40
60
80 100
T-Tr (K)
Fig. 7. WLF master regression (Eq. (4)) of the temperature dependent ionic conduc-
tivity of all PILs: poly(MEBImeBF4) (blue diamonds), poly(MEBImePF6) (green
squares), poly(MEBImeTf) (magenta triangles), poly(MEBImeTFSI) (red circles). (For
interpretation of the references to color in this figure legend, the reader is referred to
the web version of this article).

Y. Ye, Y.A. Elabd / Polymer 52 (2011) 1309e1317
[8] Yoshizawa M, Ohno H. Chem Lett 1999;28:889e90.
1317
standards are closer to their true molecular weights when
compared to PS standards, which interact with column packing
materials. Anion type had a significant effect on the degradation
and glass transition temperatures, where poly(MEBImeTFSI) had
the highest degradation temperature and the lowest glass transi-
tion temperature, indicating a much wider operating temperature
window for practical applications. The ionic conductivity of PILs
scales with the glass transition temperature, where poly
(MEBImeTFSI) had a significantly lower glass transition tempera-
ture and a significantly higher ionic conductivity compared with
PILs exchanged with other anions. Ion transport was also influ-
enced by other factors, including the size and symmetry of the
anion and dissociation energy of the ion pair.
[9] Yoshizawa M, Ohno H. Electrochim Acta 2001;46:1723e8.
[10] Ohno H, Ito K. Chem Lett 1998;27:751e2.
[11] Ohno H, Nakai Y, Ito K. Chem Lett 1998;27:15e6.
[12] Matsumi N, Sugai K, Miyake M, Ohno H. Macromolecules 2006;39:6924e7.
[13] Brylev O, Alloin F, Duclot M, Souquet JL, Sanchez JY. Electrochim Acta 2003;
48:1953e9.
[14] Vygodskii YS, Shaplov AS, Lozinskaya EI, Filippov OA, Shubina ES, Bandari R,
et al. Macromolecules 2006;39:7821e30.
[15] Sato T, Marukane S, Narutomi T, Akao T. J Power Sources 2007;164:390e6.
[16] Marcilla R, Blazquez JA, Rodriguez J, Pomposo JA, Mecerreyes D. J Polym Sci
Part A Polym Chem 2004;42:208e12.
[17] Collins KD. Biophys J 1997;72:65e76.
[18] Gwee L. Ph.D. dissertation. Drexel University; June 2010.
[19] Holomb R, Martinelli A, Albinsson I, Lassègues JC, Johansson P, Jacobsson P.
J Raman Spectrosc 2008;39:793e805.
[20] Seki T, Grunwaldt JD, Baiker A. J Phys Chem B 2009;113:114e22.
[21] Sanders RA, Frech R, Khan MA. J Phys Chem B 2003;107:8310e5.
[22] Iwahashi T, Miyamae T, Kanai K, Seki K, Kim D, Ouchi Y. J Phys Chem B 2008;
112:11936e41.
[23] Domard A, Rinaudo M, Rochas C. J Polym Sci Polym Phys Ed 1979;17:673e81.
[24] Schuster M, de Araujo CC, Atanasov V, Andersen HT, Kreuer KD, Maier J.
Macromolecules 2009;42:3129e37.
[25] Tang H, Tang J, Ding S, Radosz M, Shen Y. J Polym Sci Part A Polym Chem
2005;43:1432e43.
The temperature dependent ionic conductivity of all PILs studied
follows VFT behavior. Interestingly, the results obtained from a
three-parameter VFT regression were different from those obtained
from a two-parameter VFT regression, where the latter was
consistent with the results obtained from a mathematically equiv-
alent WLF regression. The C1g (9.03) and C2g (168 K) values obtained
from the WLF regression for these PILs deviate significantly from the
classical values obtained from the mechanical relaxation of
uncharged polymers (C1g ¼ 17.44, C2g ¼ 51.6 K). The fractional free
volume for the PILs at the glass transition temperature, calculated
from C1g value, was on average double that of what has been clas-
sically observed for uncharged polymers. While C1g value obtained
here for PIL single ion conductors was close to some Li salt-PEO
charged polymer systems, the C2g value was significantly higher than
any other charged polymer systems reported in the literature. This
suggests that the salt or ionic concentration in charged polymers can
[26] Dubin P, Koontz S, Wright III KL.
J Polym Sci Polym Chem Ed 1977;
15:2047e57.
[27] Tokuda T, Mori M, Yamada T. J Chromatogr A 1996;722:123e7.
[28] McEwen AB, Ngo HL, LeCompte K, Goldman JL. J Electrochem Soc 1999;146:
1687e95.
[29] Ngo HL, LeCompte K, Hargens L, McEwen AB. Thermochim Acta 2000;
357-358:97e102.
[30] Fredlake CP, Crosthwaite JM, Hert DG, Aki SNVK, Brennecke JF. J Chem Eng
Data 2004;49:954e64.
[31] Chowdhury A, Thynell ST. Thermochim Acta 2006;443:159e72.
[32] Kroon MC, Buijs W, Peters CJ, Witkamp G. Thermochim Acta 2007;465:40e7.
[33] Vygodskii YS, Mel’nik OA, Shaplov AS, Lozinskaya EI, Malyshkina IA,
Gavrilova ND. Polym Sci Ser A 2007;49:256e61.
affect free volume and the Vogel temperature (T0 ¼ Tg ꢀ Cg), where
[34] Chang TC, Wang YT, Hong YS, Chen HB, Yang JC. Polym Degrad Stabil
2000;69:317e22.
2
ion mobility at temperatures farther below the glass transition
temperature can be achieved at higher salt or ionic concentrations
in single ion conductors. Therefore, anion type and salt or ion
concentration are two key parameters that can be tuned to affect the
temperature dependent ionic conductivity in PILs.
[35] Benrabah D, Arnaud R, Sanchez JY. Electrochim Acta 1995;40:2437e43.
[36] Arnaud R, Benrabah D, Sanchez JY. J Phys Chem 1996;100:10882e91.
[37] Okoturo OO, VanderNoot TJ. J Electroanal Chem 2004;568:167e81.
[38] Tokuda H, Tsuzuki S, Susan MABH, Hayamizu K, Watanabe M. J Phys Chem B
2006;110:19593e600.
[39] Fragiadakis D, Dou S, Colby RH, Runt J. Macromolecules 2008;41:5723e8.
[40] Vogel HZ. Phys 1921;22:645e6.
[41] Fulcher GS. J Am Ceram Soc 1925;8:339e55.
Acknowledgments
[42] Tammann G, Hesse GZ. Anorg Allg Chem 1926;156:245e57.
[43] Adam G, Gibbs JH. J Chem Phys 1965;43:139e46.
[44] Rault J. J Non-Cryst Solids 2000;271:177e217.
The authors gratefully acknowledge the financial support of the
U.S. Army Research Office under Grant W911NF-07-1 0452; Ionic
Liquids in Electro-Active Devices (ILEAD) MURI.
[45] Klein RJ, Zhang S, Dou S, Jones BH, Colby RH, Runt J.
J Chem Phys
2006;124:144903e8.
[46] Williams ML, Landel RT, Ferry JD. J Am Chem Soc 1955;77:3701e7.
[47] Cohen MH, Turnbull D. J Chem Phys 1959;31:1164e9.
[48] Doolittle AK. J Appl Phys 1951;22:1471e5.
Appendix. Supplementary data
[49] Berry GC, Fox TG. Adv Polym Sci 1967;5:261.
[50] Ikeda Y, Masui H, Matoba Y. J Appl Polym Sci 2005;95:178e84.
[51] Ferry JD. Viscoelastic properties of polymers. 3rd ed. Wiley; 1980. pp. 277e89.
[52] Watanabe M, Itoh M, Sanui K, Ogata N. Macromolecules 1987;20:569e73.
[53] Watanabe M, Oohashi S, Sanui K, Ogata N, Kobayashi T, Ohtaki Z. Macro-
molecules 1985;18:1945e50.
Supplementary data associated with this article can be found on-
line at doi:10.1016/j.polymer.2011.01.031. http://www.elsevier.com/.
References
[54] Carvalho LM, Guegan P, Cheradameb H, Gomesa AS. Eur Polym J 2000;36:
401e9.
[55] Saito M, Ikuta H, Uchimoto Y, Wakihara M, Yokoyama S, Yabe T, et al.
J Electrochem Soc 2003;150:A726e31.
[1] Green O, Grubjesic S, Lee S, Firestone MA. Polym Rev 2009;49:339e60.
[2] Armand M, Endres F, MacFarlane DR, Ohno H, Scrosati B. Nat Mater
2009;8:621e9.
[3] Hirao M, Ito K, Ohno H. Electrochim Acta 2000;45:1291e4.
[4] Ohno H, Yoshizawa M, Ogihara W. Electrochim Acta 2004;50:255e61.
[5] Shovsky A, Varga I, Makuska R, Claesson PM. Langmuir 2009;25:6113e21.
[6] Chen H, Choi JH, Salas-de la Cruz D, Winey KI, Elabd YA. Macromolecules
2009;42:4809e16.
[56] Qi L, Lin Y, Jing X, Wang F. Solid State Ionics 2001;139:293e301.
[57] Ikeda Y, Masui H, Matoba Y. J Appl Polym Sci 2005;95:178e84.
[58] Watanabe M, Sanui K, Ogata N. Macromolecules 1986;19:815e9.
[59] Tabata S, Hirakimoto T, Tokuda H, Susan MABH, Watanabe M. J Phys Chem B
2004;108:19518e26.
[60] Wang SJ, Wang B, Li SQ, Peng ZL. J Radioanal Nucl Chem 1996;211:127e35.
[61] Bamford D, Dlubek G, Reiche A, Alam MA, Meyer W, Galvosas P, et al. J Chem
Phys 2001;115:7260e70.
[7] Bhowmik PK, Burchett RA, Han H, Cebe JJ. J Polym Sci Part A Polym Chem
2001;39:2710e5.