You are on page 1of 7

Evaluation of a AgjAg

+
reference electrode for use in room
temperature ionic liquids
G.A. Snook
*
, A.S. Best, A.G. Pandolfo, A.F. Hollenkamp
Commonwealth Scientic and Industrial Research Organisation (CSIRO) Energy Technology, Box 312, Clayton South, Vic. 3169, Australia
Received 19 June 2006; received in revised form 5 July 2006; accepted 5 July 2006
Available online 4 August 2006
Abstract
Interest continues to grow in the use of room temperature ionic liquids (RTILs) as electrolytes in a range of electrochemical appli-
cations, such as lithium batteries, supercapacitors and dye-sensitized solar cells. Underpinning this growth, investigations into the elec-
trochemical behaviour of RTILs and RTIL-based systems rely on accurate and precise data on the potentials of redox processes. While
most researchers have continued the practice (developed with non-aqueous solvents) of reporting potentials relative to one of the metal-
organic standards (such as ferrocene), little attention has been given to the development of a reliable reference electrode, based on an
ionic liquid. Such an electrode is always valuable, especially in situations where addition of a reference material is not possible.
We report a AgjAg
+
reference electrode, incorporating a known concentration of silver triuoromethanesulfonate (AgTf) in 1-butyl-
1-methyl-pyrrolidinium bis(triuoromethanesulfonyl)imide (P
14
TFSI), which provides a stable and reproducible reference potential.
Voltammetric monitoring of the redox potentials for ferrocene and cobalticinium hexauorophosphate have shown that the electrode
AgjAg
+
(10 mM AgTf, P
14
TFSI) is stable to within a millivolt, over a period of around three weeks, when used in an argon atmosphere
at room temperature. Higher concentrations of silver ion reveal close-to-Nernstian behaviour. All AgjAg
+
congurations were signi-
cantly more stable than a silver wire quasi-reference electrode, even when the latter was separated in a salt-bridge. Voltammetric data
recorded in a range of dierent ionic liquids, against the AgjAg
+
(10 mM AgTf, P
14
TFSI) reference electrode, produced apparent junc-
tion potentials of a few tens of mV. Changes in sign of the junction potential are usefully discussed in terms of the relative mobilities of
the anions and cations present, while the magnitude can be discussed in the terms of a classic molten salt treatment.
2006 Elsevier B.V. All rights reserved.
Keywords: Reference electrode; Quasi-reference electrode; Silver; Room temperature ionic liquid; Junction potential
1. Introduction
Room temperature ionic liquids (RTILs) continue to
receive much attention, due to their ability to be tailored
for an ever-expanding range of applications. One impor-
tant research area is in the eld of electrochemistry, where
they are being considered as electrolytes in lithium batter-
ies, supercapacitors and dye-sensitized solar cells. In these
applications, accurate information is required on the posi-
tion and the magnitude of the electrochemical window, the
diusion of charged species and the reversibility of charge-
transfer processes. This in turn requires stable referencing
of the potential scale.
In probing the electrochemical properties of RTILs
(e.g., with cyclic voltammetry), the experimental congura-
tion often reported includes a quasi-reference electrode.
This practice goes back to electrochemistry in non-aqueous
solvents, where well-dened reference electrodes are di-
cult to prepare and stabilize, and can introduce extra
uncompensated resistance. From the literature on the elec-
trochemistry of ionic liquids, little can be found on the use
and understanding of reference electrodes. The majority
report the use of quasi-reference electrodes, based on silver
[110] or platinum wire [1114], which are immersed
directly into the analyte solution. Less commonly, magne-
sium [15] or aluminium wire [16] have been used in the
1388-2481/$ - see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.elecom.2006.07.004
*
Corresponding author. Tel.: +61 395458863; fax: +61 395628919.
E-mail address: Graeme.Snook@csiro.au (G.A. Snook).
www.elsevier.com/locate/elecom
Electrochemistry Communications 8 (2006) 14051411
same role. The problems that can arise when using quasi-
reference electrodes are often not appreciated and, we
argue, are also likely to be more severe when the electrolyte
is an RTIL.
As is generally known, the commonly used quasi-refer-
ence electrode (based on either silver or platinum wire)
functions through the presence of various compounds
(most likely oxides) on the metal surface. The exact identity
of the redox couple is never known with any certainty. A
second unknown is the amount of the oxidized compo-
nent(s), which, clearly, is very small when compared with
commercial reference electrodes. The problem here is that
the composition of the surface, and hence the potential of
the reference electrode, is liable to change greatly in the
event of: (i) reaction with components of the surrounding
solution; (ii) dissolution in the electrolyte; (iii) polarization,
due to lack of potentiostatic control.
1
For these reasons,
employing a quasi-reference electrode always carries the
assumption that the potential(s) of the electrochemical pro-
cesses of interest will be recorded, at some stage, against a
true reference electrode (i.e., one based on a stable, well-
dened redox couple) or that the potential of a suitable
standard compound (e.g., ferrocene) will be recorded under
identical test conditions [18].
In this study, we examine the behaviour of a AgjAg
+
ref-
erence electrode comprising a silver wire immersed in a xed
concentration of silver triuoromethanesulfonate (AgTf)
dissolved in a common pyrrolidiniumionic liquid, and sepa-
rated by a porous junction fromthe test solution. In a recent
book, Katayama [19] briey describes the usage of the
AgjAg
+
couple as a reference systemin several ionic liquids.
While some details are presented on the construction of the
reference electrodes, there is no discussion of the stability
of this system. In a later work, Katayama and co-workers
[20] report using a AgjAg
+
reference electrode, withthe silver
salt dissolved in 1-ethyl-3-methyl-imidazolium bis(trif-
luoromethanesulfonyl)imide (EMImTFSI). Here a direct
measurement of the potential of the ferrocene oxidation pro-
cess is reported, but there is no description of the character-
istics (stability, etc.) of the electrode. More recently, Saheb
et al. [21] described a AgjAg
+
(imidazoliumionic liquid) ref-
erence electrode in which the concentration of silver ion is set
by addition of a solution of silver (I) nitrate in acetonitrile. A
problemwith such a system, noted by the authors, is that the
electrodes potential varies signicantly with the volume
fraction of acetonitrile, and some loss of the latter would
be likely after several days of use at ambient temperature,
or indeed, after brief periods above ambient temperatures.
We compare a AgjAg
+
reference electrode, constituted
in an ionic liquid, with a silver-wire quasi-reference elec-
trode which is either immersed directly into the ionic liquid
in the electrochemical cell, or separated from the latter by
means of a salt-bridged compartment which contains the
ionic liquid. The performance of each reference system is
evaluated in a range of RTILs by recording the voltammet-
ric responses for ferrocene and cobalticinium hexauoro-
phosphate. Both compounds are widely used as internal
standards for referencing redox potentials [2,3,6,2224].
In this paper, we argue that because many of the RTILs
commonly being used in electrochemical investigations
are excellent solvents for a range of metal compounds, a
greater level of caution is required when employing any
type of quasi-reference electrode. Moreover, in some situa-
tions, it will be prudent to utilize a true reference electrode,
similar to those examined here. Such systems have proven
to be reassuringly stable and robust.
2. Experimental
All solution preparations and electrochemical measure-
ments were performed in an argon-lled glove box. All
chemicals were used as received (unless indicated other-
wise). The ionic liquid 1-butyl-1-methyl-pyrrolidinium
bis(triuoromethanesulfonyl)imide (P
14
TFSI) was either
purchased from Merck (high purity specication, <100
ppm water, <100 ppm chloride) or prepared and puried
according to a standard procedure [25]. All samples were
dried under vacuum for 24 h at elevated temperature prior
to use. P
14
TFSI was used as the electrolyte in the reference
electrode element (salt-bridged glass compartment) due to
its wide electrochemical window (>5.5 V), low viscosity
(85 mPa s) and good stability. Reference electrode solu-
tions were prepared by dissolving silver triuoromethane-
sulfonate (AgTf, 99.95+% purity, Aldrich) at 10 mM and
100 mM in P
14
TFSI. Each solution was covered and heated
with stirring, to ensure complete dissolution, then stored in
sealed brown bottles and covered with aluminium foil. The
ILs were pipetted into quartz glass compartments, each ter-
minated with an ultra-ne glass frit to separate the AgTf-
IL mixture from the analyte. After approximately 10 min,
the solution had permeated through the frit.
2
The assembly
of the electrode was completed by inserting a length of sil-
ver wire, which was held in place by a machined teon
plug. A diagram of the assembled components is presented
in Fig. 1.
The ionic liquids used were: P
14
TFSI; EMImTrifAc,
EMImTf, EMImEtSO
4
(1-ethyl-3-methylimidazolium salts
of triuoroacetate, triate and ethylsulfate, respectively, all
obtained as high purity samples from Merck); BMImBF
4
,
BdMImBF
4
(1-butyl-3-methylimidazolium and 1-butyl-
2,3-dimethylimidazolium salts of tetrauoroborate, respec-
tively, both from Solvent Innovations GmbH); AmmoEng
100 ((C
12
C
18
)alkylpoly(3)oxyethyldihydroxyethylmethy-
lammonium methylsulfate, 95% purity from Solvent
Innovations); P
14
BOB (1-butyl-1-methylpyrrolidinium
bis(oxalatoborate), prepared in our laboratories).
1
Bond and Lay [17] reported that the use of a Pt quasi-reference
electrode in a two-electrode cell, against a microelectrode, led to observed
shifts in redox potentials of up to 500 mV.
2
By contrast, the use of a vycor tip led to an increase in the time
required to wet completely the liquid junction (from several hours to
overnight).
1406 G.A. Snook et al. / Electrochemistry Communications 8 (2006) 14051411
Bis(cyclopentadienyl)cobalt(III) hexauorophosphate
(cobalticinium hexauorophosphate, or CcPF
6
) and
bis(cyclopentadienyl)iron(II) (ferrocene, or Fc) were both
obtained in 98% purity from Aldrich, and were dissolved
at several dierent concentrations by stirring overnight at
50 C.
All electrochemical experiments were undertaken with a
lAutolabIII potentiostat/galvanostat (Ecochemie, Nether-
lands), at 23 1 C, with a 100 lm diameter Pt working
electrode, a large area (wire-wound) Pt counter electrode
and one of the reference electrodes prepared above. Con-
ductivity measurements were carried out at 25 C (except
for AmmoEng 100, 22 C) with a HewlettPackard
4924A LCR Meter at frequencies from 20 Hz to 1 MHz.
Labview
TM
software on a PC was used to control temper-
ature (via thermocouple and heated block), sweep the fre-
quency of the meter and record the results. Samples were
prepared and kept under argon and the instrument was cal-
ibrated against aqueous 0.1 M KCl. Viscosity was mea-
sured at 25 C with an Anton Paar (AMVn) Automated
Microviscometer. The 2.5 mm diameter ball (7.85 g cm
3
)
gave a constant of 0.114 mPa cm
3
g
1
with a 3 mm wide
capillary and 20 dropping angle. Density of the ionic
liquid was measured by accurately weighing the amount
of liquid in the 1.45 cm
3
capillary tube.
3. Results and discussion
3.1. Voltammetric studies of Fc and CcPF
6
in P
14
TFSI
A typical cyclic voltammogram for a solution of
20 mM cobalticinium hexauorophosphate and 5 mM fer-
rocene in the ionic liquid P
14
TFSI is shown in Fig. 2a.
Here a quasi-reference electrode (silver wire) is immersed
directly in the ionic liquid. Responses were recorded at
scan rates between 20 mV s
1
and 1000 mV s
1
. With
the distance between the working and reference electrodes
set at 10 mm, peak-to-peak separations were recorded in
the range of 60 to 80 mV for both redox couples, with the
values trending higher at the higher scan rates. This is
most likely due to the eects of a small amount of uncom-
pensated solution resistance. According to the theory [22]
that relates this resistance to the separation between the
reference electrode and what is a relatively small
(0.1 mm dia.) working electrode, the separation would
have to be reduced to below 0.5 mm before there would
be a signicant decrease in uncompensated resistance.
Given that physical shielding of the working electrode
would then be a serious additional problem, we decided
to leave the electrodes in their initial positions. Potentials
midway between the forward and reverse peaks were aver-
aged (for scans at 20 mV s
1
and 50 mV s
1
) to determine
values of E
00
[26] for the cobaltocenecobalticinium and
the ferroceneferricinium couples. Results are presented
in Table 1. Repeating the cyclic voltammetric measure-
ments on this solution, on dierent days, produced a
range of peak positions for which the standard deviation
was 100 mV (based on ve runs).
The silver wire quasi-reference electrode was then placed
in a compartment that was lled with P
14
TFSI (but no
added Ag
+
salt) and immersed in the P
14
TFSI solution of
the two metallocenes. From a set of repeat measurements
(cyclic voltammograms) with this conguration, the varia-
tion in peak potentials (Table 1) gave a standard deviation
of 55 mV (based on four runs). In a study of the behaviour
of CcPF
6
and various derivatives of ferrocene in the ionic
liquid 1-butyl-3-methyl-imidazolium hexauorophosphate,
Hultgren et al. [3] reported that separating a silver quasi-
reference electrode (with a similar compartment-and-frit
arrangement) reduced potential drift to within 5 mV over
a 24 h period. These data show that while the behaviour
of a system that employs a quasi-reference electrode may
appear to be stable, comparison of day-to-day potential
stability indicates a relatively large variability. The addi-
tional variability associated with the directly immersed
electrode is probably because the silver wire is in contact
4.5mm
6.4mm
2.5mm
0.5mm Ag wire
Glass Tubing
Teflon Cap
4.5mm
Ultrafine Porous
Glass Frit
Ultrafine Porous
Glass Frit
Salt-bridge
b
a
Fig. 1. Diagram outlining the components and dimensions of: (a) the
AgjAg
+
reference electrode and (b) the reference electrode inserted into a
salt-bridge compartment (optional).
Fig. 2. Cyclic voltammetric responses for 20 mM cobalticinium hexau-
orophosphate (reduction) and 5 mM ferrocene (oxidation) in P
14
TFSI
with reference electrode: (a) Ag wire directly immersed [response is
positioned according to averaged data (Table 1)] and (b) AgjAg
+
(10 mM
AgTf, P
14
TFSI). Scan rate = 20 mV s
1
.
G.A. Snook et al. / Electrochemistry Communications 8 (2006) 14051411 1407
with not only the ions that constitute the ionic liquid, but
also the redox-active species (Cc
+
and Fc).
Two variants of a AgjAg
+
reference electrode were con-
structed by lling salt-bridge compartments with either
10 mM AgTf or 100 mM AgTf. Cyclic voltammograms
were then recorded with the same experimental arrange-
ment as used for Fig. 2a (viz., 20 mM CcPF
6
and 5 mM
Fc in P
14
TFSI). A typical voltammogram, with the
10 mM AgTf electrode in place, is shown in Fig. 2b. Peak
potential data (and interpolated values of E
00
) for the two
redox processes, with each of the reference electrodes in
place, are summarized in Table 1. Repetition of these mea-
surements on dierent days (with a given electrode, over a
period of one month), showed variation in the peak poten-
tials of 1 mV, up to the point when the electrode was three
weeks old. Beyond this, the potential shifted steadily, in a
direction that indicated a progressive fall in the concentra-
tion of silver (I). Given that the instrumentation sets
applied potential with an accuracy of 1 mV, the variations
probably reect a system-limit rather than one imposed by
the reference electrode. The AgjAg
+
(P
14
TFSI) reference
electrode therefore appears to satisfy the important
requirement of stability.
Table 1 includes calculated potential dierences (based
on the metallocene E
00
data) for each of the silver-based
reference systems, relative to the AgjAg
+
(100 mM AgTf)
electrode. Comparing the quasi-reference electrodes with
the AgjAg
+
electrodes, we note that the latter register
potentials some 800 mV more negative than the former.
Similar shifts were measured directly by means of a digital
voltmeter connected across two electrodes dipped in
P
14
TFSI. These shifts in potential are consistent with the
likely dierences in silver speciation for the two reference
systems.
In broad terms, the AgjAg
+
(P
14
TFSI) reference elec-
trodes register potentials similar to the standard potential
for Ag
+
(+0.799 V vs. NHE [27]) while the quasi-reference
electrodes lie in the range occupied by the AgjAgX
(X = halide, pseudohalide) redox couples (0 0.3 V vs.
NHE). The former conclusion is based on our observation
of E
00
(Fc, Fc
+
) in the region of 0.4 V vs. AgjAg
+
(100 mM
AgTf, P
14
TFSI) and the assumption that the generally sol-
vent-independent standard potential of the ferrocene couple
(+0.400 V vs. NHE [28]) also applies in the ionic liquid
medium employed here. The non-coordinating nature of
the anions present (triate and TFSI) is also consistent with
this assessment.
For the quasi-reference electrode, the observed potential
is established by a concentration of silver that is dened by
the solubility of silver (I) species present on the surface of
the wire. Given that the ionic liquid used in this study con-
tains traces of either bromide or iodide (from the prepara-
tion), it is likely that AgBr or AgI will determine the
position on the potential scale. Presumably, the variability
noted here in the potential of the quasi-reference electrode,
even when only P
14
TFSI from a single source was used, is
due to the very small amounts of silver species present,
which are therefore susceptible to change from run to
run, as well as the likelihood that other silver-containing
species are present. A further indication of variability
comes from comparing our voltammetric data for the
reduction of the cobalticinium cation (E
00
= 991 mV vs.
salt-bridge-separated Ag wire) with those of Hultgren
et al. [3] who reported a value of 1145 mV, with the same
quasi-reference system, but in the ionic liquid 1-butyl-3-
methyl-imidazolium hexauorophosphate. Their value is
150 mV more negative, i.e., shifted towards the values that
we obtained with AgjAg
+
(P
14
TFSI) reference electrodes. It
is worth noting that Hultgren et al. used an alkyl chloride
in the preparation of their imidazolium ionic liquid and
that the presence of trace levels of chloride (as opposed
to bromide or iodide) would explain the direction of the
shift in quasi-reference potential (based on the argument
proposed above). This argument is largely speculative,
though, and a rigorous explanation of the potential shifts
would rst require a complete analysis of all species present
(halides, etc.) in the ionic liquids that were used.
It is also useful to compare our results with those
obtained by Fukui et al. [20] who appear to be the rst
group to have measured the ferroceneferricinium couple
relative to a AgjAg
+
(ionic liquid electrolyte) reference
electrode. In that case, the ionic liquid was EMImTFSI
and the same AgTf salt (at 100 mM) was used. From
the voltammetry of ferrocene, they reported E
00
(Fc,
Fc
+
) = 440 mV. This is within 5 mV of our reported
value (Table 1) and indicates a small dependence of the
observed reference electrode potential on changing from
P
14
TFSI to EMImTFSI.
Table 1
Cyclic voltammetric data
a
(in mV) for the reduction of 20 mM cobalticinium hexauorophosphate and oxidation of 5 mM ferrocene in P
14
TFSI at
20 mV s
1
Reference electrode E
00
(Cc,Cc
+
) E
00
(Fc,Fc
+
) DE
00
(CcFc) Potential shift
b
Ag wire (immersed directly in electrolyte) 960
c
373
c
1333 818
Ag wire (immersed in a salt-bridge lled with P
14
TFSI) 991
d
342
d
1333 787
AgjAg
+
(10 mM AgTf, P
14
TFSI) 1723 390 1333 55
AgjAg
+
(100 mM AgTf, P
14
TFSI) 1777 445 1332 0
a
Data reported as E
00
= 1/2 [E
p(forward sweep)
+ E
p(reverse sweep)
].
b
Potential shift = [E
00
(Fc,Fc
+
) vs. reference electrode] [E
00
(Fc,Fc
+
) vs. AgjAg
+
(100 mM AgTf, P
14
TFSI)].
c
Average of ve runs with standard deviation = 100 mV.
d
Average of four runs with standard deviation = 55 mV.
1408 G.A. Snook et al. / Electrochemistry Communications 8 (2006) 14051411
The voltammetric data recorded with the AgjAg
+
elec-
trodes allow us to check the reversibility of this reference
couple in the pyrrolidinium ionic liquid by comparing the
shift of the potential on changing the concentration of
AgTf from 10 mM to 100 mM. From Table 1, the recorded
shift is 55 mV, which compares well with the expected value
of 59 mV [26].
3.2. Voltammetric studies in RTILs other than P
14
TFSI
To assess the general utility of the P
14
TFSI-based
AgjAg
+
reference electrode in studies of RTIL electro-
chemistry, cyclic voltammograms were recorded for Fc
and CcPF
6
in several dierent ionic liquids, for which vis-
cosity and conductivity spanned a signicant range. We
were particularly interested in characterizing any junction
potential [29,30] from contact between P
14
TFSI (in the ref-
erence compartment) and the ionic liquid in the electro-
chemical cell. Data describing the properties of the ionic
liquids and the results from the voltammetric studies are
collected in Table 2. Only the potential data for ferrocene
are reported as the reduction process of cobalticinium
was usually not completely resolved from the reduction
limit of the imidazolium ionic liquids.
Inspection of Table 2 reveals junction potentials (DE
j
) of
a few tens of millivolts. [Note, as implied earlier, we pro-
ceed with the assumption that the formal potential of the
FcFc
+
couple does not change appreciably with a change
in ionic liquid.] For the imidazolium ILs, there is no obvi-
ous relationship between the observed DE
j
and either vis-
cosity or conductivity, when compared with the
respective values for P
14
TFSI. To analyse this system,
though, a description similar to that used for classic molten
salt electrochemistry is required. Laity [31] has outlined a
treatment of molten salt liquid junctions in which DE
j
is
dened by the relative mobilities of the ions present.
Importantly, when the junction features either a common
cation or anion, the governing equation becomes a
straightforward relationship of the equivalent (molar)
conductivities:
DE
j

RT
F

K
13
K
23
z
1
K
13
z
2
K
23

ln
z
1
K
13
z
2
K
23

1
where K is equivalent conductivity, z is ionic charge, species
1 and 2 are the dierent ions (+ or ) and species 3 is the
ion common to both salts. From Table 2, this approach
can be taken with the junction between P
14
TFSI and
P
14
BOB. With values of K (in S cm
2
mol
1
) calculated from
data in Table 2 and Refs. [25,32], Eq. (1) gives a value of
80 mV for DE
j
. The fact that this is much greater than
the observed value is also a common nding in molten salt
studies [31]. Laity notes that the main assumption in deriv-
ing Eq. (1) is that ion mobilities do not change with con-
centration (i.e., where the two salts mix). In simple terms,
the low value of DE
j
probably means, in this case, that
the mobilities of the two anions are similar in the IL-mix-
tures that form at the junction, while being clearly dierent
in the pure phases. The relative movement of ions for
P
14
TFSIiP
14
BOB is illustrated schematically in Fig. 3a,
mainly to indicate that anion movement determines the
sign of DE
j
.
We extend the qualitative discussion to the Ammo-
Eng100iP
14
TFSI junction, where DE
j
is quite large and
has changed sign (compared with the rest of the data).
AmmoEng100 consists of a quaternary ammonium cation
with a long-chain alkyl group and methyl sulfate as the
anion. Given the relatively large size of the cation,
Table 2
Junction potentials (DE
j
) inferred from voltammetric data for ferrocene in
several ionic liquids, together with data for viscosity and conductivity
Ionic liquid Viscosity
a
(mPa s)
Conductivity
a
(mS cm
1
)
E
00
(Fc,Fc
+
)
b
(mV)
DE
j
(mV)
P
14
TFSI 85
c
2.2
c
390 0
EMImTf 39 9.9 416 26
BMImBF
4
87 4.1 418 28
EMImEtSO
4
125 4.0 432 42
BdMImBF
4
515 0.91 415 25
P
14
BOB 5000 0.11 401 11
AmmoEng 100 1665 0.042 338 +52
a
25 C except AmmoEng 100 at 22 C.
b
vs AgjAg
+
(10 mM AgTf, P
14
TFSI).
c
From Ref. [25].
C12-C18 NMeR2
+
MeSO4
-
+ -
P14
+
TFSI
-
liquid junction
Reference
Compartment
Analyte
Compartment
P14
+
BOB
-
- +
P14
+
TFSI
-
Reference
Compartment
liquid junction
Analyte
Compartment
a
b
Fig. 3. Illustration of localized accumulation of charge (d
+/
) on either
side of the liquid junction established between two dierent ionic liquids,
leading to observed junction potentials (Table 2): (a) P
14
BOB (ana-
lyte)jP
14
TFSI (reference) and (b) AmmoEng 100 (analyte)jP
14
TFSI
(reference). Length of arrow indicates relative mobility of ionic species.
G.A. Snook et al. / Electrochemistry Communications 8 (2006) 14051411 1409
compared with the pyrrolidinium species, and assuming
similar mobilities for the TFSI

and MeSO

4
anions, we
suggest that dierential cation transport now dominates
the localized accumulation of charge. A relative excess of
the AmmoEng100 cation on the working solution side of
the junction (Fig. 3b) gives the observed polarity. Without
a common ion, we are unable to utilize Eq. (1) to establish
an upper estimate of DE
j
. In summary, though, it is impor-
tant to note that in each of the cases investigated (Table 2),
stable performance of the AgjAg
+
(10 mM AgTf, P
14
TFSI)
reference electrode was maintained, thereby conrming the
utility of this reference system for use in studies that
involve a range of dierent ionic liquids.
4. Concluding remarks
In this work, we have demonstrated that a AgjAg
+
ref-
erence electrode can be congured in a commonly available
ionic liquid and that its performance, if maintained accord-
ing to the usual precautions for silver solutions, is stable
over several weeks. A concentration of 10 mM Ag
+
in
the reference element (added as the triate) is suggested,
as this is easily prepared (with the ionic liquid used here)
and gives more stable potential behaviour than with higher
concentrations (the rate of degradation of silver (I) in the
IL appears to be concentration dependent). Contamination
of the analyte solution by Ag
+
is also less likely. The pres-
ence of the triate ion has no apparent inuence on the
behaviour of the reference electrode, which is consistent
with the known non-coordinating properties of this ion
(particularly with silver (I)). In terms of reproducibility of
the reference potential, we found that repeated prepara-
tions of the AgjAg
+
(10 mM AgTf, P
14
TFSI) electrode
gave a variation in potential of 2 mV. Most of this vari-
ance should be due to the errors associated with prepara-
tion of the AgTf solution and selection and cleaning of
the silver wire. Once prepared, individual examples are sta-
ble, on day-to-day use, to within 1 mV, for up to about
three weeks. Further improvements on lifetime could be
achieved by further deoxygenation and purication of the
ionic liquid, together with more rigorous protection of
the reference electrode from light.
When used in voltammetric studies in a variety of other
ionic liquids, the AgjAg
+
(10 mM AgTf, P
14
TFSI) refer-
ence electrode exhibits stable potential behaviour. Assum-
ing that the potential of the FcFc
+
couple does not vary
signicantly between the dierent RTILs, the data indicate
the presence of junction potentials of a few tens of milli-
volts. From a qualitative view, relative ionic mobilities pro-
vide an explanation of the polarity of the junctions, while a
quantitative treatment establishes an upper limiting value
of the potential for the junction between P
14
TFSI and
P
14
BOB. More rigorous analysis will require information
on the transport properties of the mixed phases that form
at the boundary between two ionic liquids.
Finally, it seems clear that the AgjAg
+
reference system
will also work well when constituted in one of a range of
commonly used ionic liquids. As noted earlier, Fukui
et al. [20] employ the AgjAg
+
(100 mM AgTf, EMImTFSI)
reference electrode in their studies with imidazolium and
pyrrolidinium ionic liquids and report a potential for the
oxidation of ferrocene that is within 5 mV of our value.
In situations where a large number of experiments are to
be performed in the same ionic liquid, it will be sensible
to prepare a AgjAg
+
reference electrode in the ionic liquid
and evaluate its behaviour. As shown here, the latter is
readily accomplished by examining the voltammetric
responses for ferrocene and a cobalticinium compound.
Acknowledgements
We thank Dr. Thomas Ruether for synthesising samples
of the pyrrolidinium ionic liquids, and Drs. Noel Duy and
Junhua Huang for comments on the manuscript and help-
ful discussions, and nally Dr. Junhua Huang for assis-
tance with viscosity measurements.
References
[1] J. Zhang, A.M. Bond, Analyst 130 (2005) 1132.
[2] J. Zhang, A.M. Bond, H. Schumann, K. Suhring, Organometallics 24
(2005) 2188.
[3] V.M. Hultgren, A.W.A. Mariotti, A.M. Bond, A.G. Wedd, Anal.
Chem. 74 (2002) 3151.
[4] F. Endres, W. Freyland, J. Phys. Chem. B 102 (1998) 10229.
[5] C.A. Zell, F. Endres, W. Freyland, Phys. Chem. Chem. Phys. 1 (1999)
697.
[6] J. Zhang, A.M. Bond, Anal. Chem. 75 (2003) 2694.
[7] I.M. Alnashef, M.L. Leonard, M.A. Matthews, J.W. Weidner, Ind.
Eng. Chem. Res. 41 (2002) 4475.
[8] U. Schroder, J.D. Wadhawan, R.G. Compton, F. Marken, P.A.Z.
Suarez, C.S. Consorti, R.F. De Souza, J. Dupont, New J. Chem. 24
(2000) 1009.
[9] D.R. MacFarlane, J. Sun, J. Golding, P. Meakin, M. Forsyth,
Electrochim. Acta 45 (2000) 1271.
[10] E. Dickinson, M.E. Williams, S.M. Hendrickson, H. Masui, R.W.
Murray, J. Am. Chem. Soc. 121 (1999) 613.
[11] R.J.C. Brown, P.J. Dyson, D.J. Ellis, T. Welton, Chem. Commun.
(2001) 1862.
[12] F. Endres, C. Schrodt, Phys. Chem. Chem. Phys. 2 (2000) 5517.
[13] P.A.Z. Suarez, V.M. Selbach, J.E.L. Dullius, S. Einloft, C.M.S.
Piatnicki, D.S. Azambuja, R.F. Desouza, J. Dupont, Electrochim.
Acta 42 (1997) 2533.
[14] K. Caban, M. Donten, Z. Stojek, J. Phys. Chem. B 108 (2004) 1153.
[15] Y. Nuli, J. Yang, J.L. Wang, J.Q. Xu, P. Wang, Electrochem. Solid
St. 8 (2005) C166.
[16] D.M. Ryan, T.L. Riechel, T. Welton, J. Electrochem. Soc. 149 (2002)
A371.
[17] A.M. Bond, P.A. Lay, J. Electroanal. Chem. 199 (1986) 285.
[18] G. Gritzner, J. Kuta, Pure Appl. Chem. 56 (1984) 461.
[19] Y. Katayama, in: H.E. Ohno (Ed.), Electrochemical Aspects of Ionic
Liquids, John Wiley & Sons, Inc., New Jersey, 2005 (Chapter 3) and
references therein.
[20] R. Fukui, Y. Katayama, T. Miura, Electrochemistry 73 (2005) 567.
[21] A. Saheb, J. Janata, M. Josowicz, Electroanalysis 18 (2006) 405.
[22] A.M. Bond, K.B. Oldham, G.A. Snook, Anal. Chem. 72 (2000) 3492.
[23] R.S. Stojanovic, A.M. Bond, Anal. Chem. 65 (1993) 56.
[24] A.M. Bond, E.A. Mclennan, R.S. Stojanovic, F.G. Thomas, Anal.
Chem. 59 (1987) 2853.
[25] D.R. MacFarlane, P. Meakin, J. Sun, N. Amini, M. Forsyth, J. Phys.
Chem. B 103 (1999) 4164.
1410 G.A. Snook et al. / Electrochemistry Communications 8 (2006) 14051411
[26] P.T. Kissinger, W.R. Heineman, J. Chem. Educ. 60 (1983) 702.
[27] J.G.E. Speight, Langes Handbook of Chemistry, 16th ed., McGraw-
Hill, New York, 2005.
[28] H.M. Koepp, H. Wendt, H. Strehlow, Z. Elektrochem. 64 (1960)
483.
[29] W. Davison, T.R. Harbinson, Anal. Chim. Acta 187 (1986) 55.
[30] J. Newman, K.E. Thomas-Alyea, Electrochem. Syst., third ed., John
Wiley & Sons, Inc., USA, 2004.
[31] R.W. Laity, in: D.J.G. Ives, G.J. Janz (Eds.), Reference Electrodes
Theory and Practice, Academic Press, New York, 1961, p. 524.
[32] W. Xu, L.M. Wang, R.A. Nieman, C.A. Angell, J. Phys. Chem. B 107
(2003) 11749.
G.A. Snook et al. / Electrochemistry Communications 8 (2006) 14051411 1411

You might also like