Accelerat ing t he world's research.

The development of an all copper

hybrid redox flow battery using deep
eutectic solvents
David Lloyd
Electrochimica Acta

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

Preparat ion of a cost -effect ive, scalable and energy efficient all-copper redox flow bat t ery
Lasse Murt omaki, laura sanz

Descript ion and performance of a novel aqueous all-copper redox flow bat t ery
laura sanz

Cyclable membraneless redox flow bat t eries based on immiscible liquid elect rolyt es: Demonst rat ion …
Saif Almheiri
 Electrochimica Acta 100 (2013) 18–23

Contents lists available at SciVerse ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

The development of an all copper hybrid redox flow battery using

deep eutectic solvents
David Lloyd ∗ , Tuomas Vainikka, Kyösti Kontturi
Aalto University, Department of Chemistry, Kemistintie 1, PO Box 16100, 00076 Aalto, Finland

a r t i c l e i n f o a b s t r a c t

Article history: The performance of a redox flow battery based on chlorocuprates dissolved in an ionic liquid analogue
Received 28 November 2012 is reported at 50 ◦ C. The kinetics of the positive electrode reaction at a graphite electrode are favourable
Received in revised form 21 March 2013 with a heterogeneous rate constant, k0 , of 9.5 × 10−4 cm s−1 . Coulombic efficiency was typically 94% and
Accepted 23 March 2013
independent of current density. The small cell potential of 0.75 V and slow mass transport result in energy
Available online xxx
efficiencies of only 52% and 62% at current densities of 10 and 7.5 mA/cm2 respectively. The successful
development of a separator by jellifying the electrolyte using polyvinyl alcohol is reported.
Keywords:
© 2013 Published by Elsevier Ltd.
Redox flow battery
Ionic liquid
Deep eutectic solvent
All copper
Chlorocuprate

1. Introduction on transition-metal ethanolamine cation complexes and discrete

organic anions concentrations of the electroactive species of up to
Redox flow batteries (RFBs) are undergoing significant research 6.5 M can be achieved [6].
for large scale stationary energy storage applications. A typical RFB Major challenges to the development of IL based RFB systems
has two heterogeneous redox reactions occurring at a positive and are their high cost and limited availability at even the tonne scale.
negative electrode respectively [1]. RFBs employing the deposi- Compared to aqueous systems they are also viscous and resistive
tion/stripping reaction of a metal are known as hybrid redox flow [7]. Additionally, the preparation of a transition-metal-salt solution
batteries. may require a complicated metathesis of the desired metal salt from
The RFB chemistry of greatest commercial significance is based the protonated anion [8].
on the multiple redox states exhibited by vanadium. However, as Work has begun on the development of RFBs that utilise ILs [9].
recently highlighted by Wadia et al. [2], there are fundamental Additionally, papers describing the application of ILs as supporting
limitations to the availability of some transition metals used in electrolytes in molecular solvents used in RFBs also exist [10]. To
numerous RFB systems currently undergoing research, for instance achieve a liquid state at room temperature, ILs typically require rel-
vanadium and cerium [3]. For this reason, it is logical to develop atively bulky cations and anions, making them average supporting
simple, cost effective systems using materials exhibiting fewer sup- electrolytes when dissolved in molecular solvents compared to a
ply constraints, for instance base metals and bulk chemicals. typical choice, such as an alkali-metal perchlorate [11].
Some ionic liquids (ILs) can be based on widely available mate- Deep eutectic solvents (DES) are formed when an organic halide
rial streams. ILs are defined as highly dissociated molten salts that salt, typically choline chloride, is combined with a material capa-
are liquid below 100 ◦ C [4]. ILs offer the scope to achieve signif- ble of forming a complex with the halide, such as urea, to form a
icant increases in energy density. Firstly, they can be engineered material that is liquid at ambient conditions [12]. DES exhibit simi-
to have extremely wide windows of electrochemical stability. This lar properties to chloride rich chloroaluminate ionic liquids, but can
allows, for instance, the deposition of metallic lithium [5], hence be prepared and used under ambient conditions [13]. They are often
large cell potentials are possible. Secondly, an electroactive species non-toxic and tend to be an order of magnitude cheaper than ionic
may actually be part of the ionic liquid. For instance, in ILs based liquids. To the best of our knowledge the use of DES as electrolyte
media in a RFB has not yet been reported.
Abbott et al. were the first to report the electrochemistry of cop-
∗ Corresponding author. Tel.: +358 4578402492. per complexes in a DES [12]. We have recently expanded this work
E-mail address: [email protected] (D. Lloyd). and found that copper can exist in three different stable states:

0013-4686/$ – see front matter © 2013 Published by Elsevier Ltd.

http://dx.doi.org/10.1016/j.electacta.2013.03.130
 D. Lloyd et al. / Electrochimica Acta 100 (2013) 18–23 19

Cu(s), Cu(I)Cl3 2− and Cu(II)Cl4 2− [14]. These states are connected a hotplate with stirring. After 90 min all the PVA dissolved to form
by the redox reactions shown in Eqs. (1) and (2). a clear, colourless liquid. Upon cooling, a translucent gel is formed,
which can be returned to the liquid state an indefinite number of
charge
Negative electrode: CuCl3 2− + e− ⇋ Cu(S) + 3Cl− (1) times by reheating briefly to 150 ◦ C. The composite separators used
discharge in this work were prepared by applying the hot liquid sequentially
to both sides of a vertically suspended filter paper (Schleicher &
discharge Schuell, 589/1) and removing any excess using a flat spatula.
Positive electrode: CuCl4 2− + e− ⇋ CuCl3 2− + Cl− (2)
charge
2.2. Description of redox flow battery
The DES used in both that work and here is a mixture of choline
chloride and ethylene glycol, combined in a 1:2 molar ratio, which A simple RFB was constructed by sandwiching a separator
is known as ethaline. Ethaline has the lowest viscosity and highest between two polycarbonate plates. Sealing was achieved using
conductivity of this class of materials and has been studied by a silicone gaskets, which also served to hold a carbon cloth (E-tek
number of groups for over five years [15–17]. elat GDL microporous layer on non-woven web, thin configura-
Since ethylene glycol is a liquid at ambient conditions and no tion) against each polycarbonate plate. Electrical connection to
studies detailing the liquid–solid phase transition for this DES have each cloth was made by compressing a short strip of platinum foil
been published, there is some ambiguity as to whether the stoi- between the polycarbonate plate and the unwetted edge of the car-
chiometry commonly used corresponds to a eutectic composition. bon cloth. The macroscopic area of exposed carbon cloth in each half
Additionally, at higher copper chloride concentrations the solute cell was 5.5 cm2 , the separation between the composite separator
significantly competes with ethylene glycol in the complexation of and carbon cloth was 0.25 mm.
chloride. At this point the system can become a ternary eutectic. The two half-cell electrolytes were stored in two separate glass
For this reason in this paper we examine the reproducibility of the reservoirs fitted with thermostatic jackets. Pumping was achieved
experimental results across a range of solute concentrations. using a peristaltic pump, at a rate of 44 ml/min. The electrolyte from
In ethaline the reaction shown in Eq. (2) is relatively facile at a each half cell was returned to the same reservoir from which it
platinum electrode with a k0 of 10−3 cm s−1 at 25 ◦ C and a charge originated.
transfer coefficient, ˛, of 0.39. With a formal potential, E 0′ , of 0.43 V The electrolyte used in the RFB was prepared by dissolving anhy-
versus a Ag/AgCl reference electrode it is separated from the upper drous CuCl2 (Riedel de Häen, purum) in the DES at 50 ◦ C in a closed
window of stability of the electrolyte by around 0.5 V. It is sepa- vessel with stirring to form a 1 M solution. A length of single-strand
rated from the reaction shown in 1, for which the kinetics have not copper wire (99.9% ASTM designation C100140) was introduced
yet been characterised, by around 0.7 V [18]. This results in the two and the electrolyte left stirring for 12 h at 50 ◦ C to allow the compro-
redox reactions being located centrally within the electrolytes’ win- portionation reaction to occur. During comproportionation Cu(II)
dow of stability, which should ensure simple operation and good in solution oxidises Cu0 to form Cu(I), Cu(II) is reduced to Cu(I) in
Coulombic efficiency if employed in an RFB. the process. The solution still had a dark brown colour after the
The existence of three stable states theoretically allows the comproportionation reaction was stopped, at which point the cop-
development of an all copper RFB. This was already suggested, in per wire was removed. Exposure of the freshly comproportionated
the form of a secondary battery, by Porterfield and Yoke for the solution to ambient conditions rapidly leads to the development
chlorocuprate ionic liquids they investigated in the seventies [19]. of a green colour, which we speculate is due to the formation of a
Their system required the use of a protective atmosphere, exhib- Cu(I) complex with oxygen.
ited a plethora of cuprate anions with various irreversible redox
reactions, short circuiting due to dendrite formation and poor sol- 2.3. Description of miniature redox battery
ubility of the Cu(II) species formed. In this work we present the
first application of DES as electrolyte media for the development of To determine the performance of the battery with minimal
what is, to the best of our knowledge, the first all-copper RFB. oxygen present and a well-defined, uniform cell temperature a bat-
Preliminary testing with commercially available separator tery was constructed using a pair of thermostated diffusion cells
materials has not been successful, as they show a rapid decrease in (PermeGear Horizontal Cell (Side-Bi-Side) 5G-00-00-09, chamber
conductivity on contact with the DES. This may be due to dehydra- volume 2.4 ml). In the negative half-cell the electrolyte was plain
tion of the membrane by the hygroscopic ionic liquid. Hence in this ethaline and the electrode was a copper wire (99.9% ASTM designa-
work we also report the development of jellified deep eutectic sol- tion C100140, diameter 0.5 mm, wound in a 3 mm spiral, effective
vents. Jellification of ionic liquids using a polymer to form a rubbery electrode area of 3 cm2 ). In the positive half-cell the electrolyte was
solid electrolyte is common practice, for instance Angell jellified a a 0.5 M solution of anhydrous CuCl2 (Riedel de Häen, purum) in
range of lithium salts using polypropylene oxide [20]. Very recently, ethaline and the electrode was a platinum wire (diameter 1 mm,
Abbot developed a biodegradable composite with good mechanical wound in a 3 mm spiral, effective electrode area of 1.7 cm2 ).
properties based on a mixture of a DES and starch [21]. The distance between the two electrodes was 35 mm. The same
separator used in the preparation of the RFB was used to pre-
2. Experimental vent mixing of the two electrolytes, the diameter of the opening
connecting the two cells was 9 mm, resulting in 0.636 cm2 of the
All starting materials were used as delivered and all work was separator being available for ionic transport.
performed under ambient conditions. The DES was prepared as pre- Since mass transport limitations at the electrodes appear to be
viously described [14]. Unless otherwise stated all measurements the major limiting factor in this system, the two electrodes were
were performed at 50 ◦ C. purposefully dimensioned to give a geometric area in excess of the
separator area. This provides a closer approximation to a typical
2.1. Preparation of separator RFB, which would utilise high surface area porous electrodes. Con-
tamination of the electrolytes during operation was minimised by
The jellified electrolyte was prepared by adding 5 wt.% polyvinyl passing the electrodes through plastic stoppers, which minimised
alcohol (PVA, Mowiol 4-98, Mw ∼27, 000) to plain ethaline. The sus- contact of the electrolytes with air. Only the electrolyte initially
pension of PVA flakes was heated in a covered beaker to 150 ◦ C on present at the time of loading the cells is available for conversion
20 D. Lloyd et al. / Electrochimica Acta 100 (2013) 18–23

during cycling. Convection in each half-cell is maintained using a 20

pair of 7.5 mm magnetic stirrers.
−2
15 j = 17.2 mA cm
l,a
2.4. Characterisation of the Cu(II)/Cu(I) reaction on glassy carbon
and graphite using the RDE 10

−2
j/mA cm
The glassy carbon (GC) rotating disc electrode (RDE) was sup- [Cu+ ] j l,a −2/3
5 = ·2 = 1.55
plied by Pine Instruments (AFE3T050GC) and was prepared as [Cu2+] j l,c
previously described [18]. The graphite RDE was machined from
graphite rod (Alfa Aesar 99.9995%), fitted in a Pine Instruments 0
ChangeDisk tip assembly (AFE6R1PT) and polished using standard
−2
note paper. −5 j = −7.0 mA cm
l,c
The redox behaviour of the Cu(I)/Cu(II) chloro complexes were
investigated at the GC electrode using a 20 mM CuCl2 electrolyte −10
prepared using anhydrous CuCl2 (Riedel de Häen, purum). Mea- −0.4 −0.3 −0.2 −0.1 0 0.1 0.2 0.3 0.4
surements on the graphite electrode using this electrolyte were E vs. Pt quasi ref./V
complicated by a slow background process, which may be related
Fig. 1. Staircase voltammogram of a freshly comproportionated electrolyte mea-
to the porous nature of the electrode. For this reason the kinetics
sured at a GC RDE, prepared as described in Section 2.4. The CuCl2 concentration
at the graphite electrode were studied using the same 2 M Cu(I) was 1 M prior to introduction of the copper wire. The voltammogram was measured
electrolyte used for the RFB measurements. at 30 ◦ C and at a rotation speed of 2000 rpm.
100% IR correction was applied during RDE measurements.
Steady-state voltammetry measurements were acquired using a
Levich equation indicates that for each of the two soluble elec-
staircase waveform. Provided the step size is sufficiently large this
troactive species the resulting limiting current is proportional to
allows full relaxation of the Nernst diffusion layer at higher scan
the bulk concentration and the diffusion coefficient raised to the
rates than are possible with cyclic voltammetry. This can be a
power of two-thirds. As we have previously reported, the diffu-
matter of concern in viscous liquids at slow rotation speeds, if
sion coefficient of Cu(I) is twice that of Cu(II) [14]. Hence we can
accurate determination of the heterogeneous kinetics is the objec-
estimate the ratio of the bulk concentrations from the ratio of the
tive. Relaxation of the Nernst layer was checked by performing the
limiting currents. Typically the concentration of Cu(II) complex in
voltammetric measurements in both scan directions and validating
the comproportionated electrolyte is two-thirds of the Cu(I) com-
that the current observed at each potential was independent of the
plex concentration.
scan direction.
The comproportionated electrolyte has a strong green colour,
Measurements performed using the RDE were analysed by
which becomes faint yellow when diluted to concentrations
preparing Koutecký-Levich plots to determine the kinetic current,
suitable for UV–vis spectroscopy (∼2 mM). UV–vis spectroscopy
Ik , at each potential. This was then used to quantify the heteroge-
showed the presence of the same Cu(II) and Cu(I) species we have
neous rate constant, k0 , by preparing the Tafel plot. This method is
previously reported [14]. The identity of the high concentration
described in detail in standard electrochemistry textbooks [22].
species is not yet known. We have observed the same transition
from a colourless complex when the comproportionation reac-
2.5. Instrumentation and electrochemical techniques tion is performed under argon to a green complex under ambient
conditions for a chloride rich Cu(I) electrolyte based on 1-butyl-1-
Measurements were performed using a Solartron 1286 poten- methylpyrrolidinium bis(trifluoromethylsulfonyl)imide [23].
tiostat. Waveform generation and acquisition was performed using During charge cycling of the miniature redox battery no green
a National instruments 6251 DAQ controlled with custom Mat- complex was observed at any point during operation. However,
lab software. All electrochemical measurements were preceded by when the experiment was completed and the cell decommissioned
measuring electrochemical impedance spectra (eis) at the open cir- the Cu(I) rich electrolyte from the negative half-cell was exposed
cuit potential (ocp) between 100 kHz. and 1 Hz. The intercept with to ambient conditions and also underwent a rapid transition from
the real axis in the complex plane plot was taken as the solution a colourless to a green complex.
resistance.
The conductivity of ethaline was determined using a DC method 3.2. Durability and permeability of the jellified membrane
we have previously reported [18]. The conductivity of jellified etha-
line was determined by eis measurements of a gel that had been Testing of the jellified membrane showed only minor perme-
cast in a PTFE cylinder, of known geometry, which was sandwiched ation of the Cu(II) complex at 50 ◦ C. During prolonged RFB testing
between two thermostated aluminium electrodes. The conduc- at 50 ◦ C, over periods of weeks, the jellified membranes maintained
tivity of the separator was determined in a similar manner, by effective separation of the two electrolytes. The long term stability
sandwiching a sample between two copper discs inside the PTFE has not yet been proven. Since the battery chemistry reported here
cylinder. utilises only one element, permeation of the electroactive species
through the membrane only impacts on the Coulombic efficiency
3. Results and discussion of the system and should not cause a deterioration in long term
performance.
3.1. Composition of electrolytes
3.3. Conductivity of the plain, gellified and supported electrolyte
The distribution of Cu(II) and Cu(I) species in the compro-
portionated electrolyte was estimated based on the cathodic and Fig. 2 shows that the plain and jellified electrolyte have virtually
anodic limiting currents, il,c and il,a , respectively, observed during indistinguishable conductivity over a range of temperatures. The
staircase voltammetry measurements at an RDE, as shown in Fig. 1. conductivity of the composite separator, also shown in Fig. 2, is
These show that a portion of the Cu(II) remains unreacted. The around four times lower than the plain electrolyte or gel. As the
 D. Lloyd et al. / Electrochimica Acta 100 (2013) 18–23 21

−1 3.5
10

2.5

−1
−10 (1/I)/A
2
κ/S cm−1

−2
10

4
1.5

0.5

−3 0
10 0 0.02 0.04 0.06 0.08 0.1
2.6 2.8 3 3.2 3.4 3.6 −1/2 1/2
−1 ω /s
(1000/T)/K
Fig. 4. Koutecký-Levich plots prepared from the voltammetric data shown in Fig. 3
Fig. 2. Temperature dependence of conductivity for pure ethaline (), ethaline gel
for  values between −0.05 and −0.4 V. The arrow indicates increasingly negative
containing 5 wt.% PVA (䊉) and ethaline gel supported in a filter paper (♦).
overpotentials.

thickness of the composite separator is 300 ␮m it should contribute

30% of the ohmic losses in the RFB reported here. The conductivity The difference between glassy carbon and graphite may be
of the gel prepared here is around three orders of magnitude higher attributable to the greater surface roughness of graphite, since I0
than the DES-starch composite reported by Abbott [21]. was normalised by the macroscopic surface area to determine k0 .
The higher rate on a platinum electrode may be attributable to
3.4. Characterisation of the Cu(I)/Cu(II) redox reaction at a mediation of the redox reaction by chloride ions adsorbed on the
carbon electrodes platinum surface, as has been reported in aqueous electrolytes [25].

The reaction shown in Eq. (2) was characterised at glassy carbon

and graphite rotating disc electrodes at rotation speeds of 1000, 3.5. Performance of an all copper RFB
1250, 1500, 1750 and 2000 rpm. Normalised voltammetry data for
the glassy carbon electrode is shown in Fig. 3. This data was used Fig. 6 shows the staircase voltammogram of the all copper RFB
to determine the kinetic current, Ik , at each potential by evaluation described in Section 2.2 after two weeks operation under ambient
of the Koutecký-Levich equation. Fig. 4 shows this for the cathodic conditions. The transition from charging to discharging occurs at
branch of the voltammetric data in Fig. 3. The resulting Tafel plots 0.72 V, which is indicative of the expected cell potential. The win-
are shown in Fig. 5. By extrapolating the linear part of the cathodic dow of stability of the electrolyte was 2.5 V when using graphite
branch of Ik to zero overpotential ( = 0) the exchange current, I0 , electrodes. The ohmic impedance of the RFB was 20.7 cm2 , which
can be determined and this can be used to estimate k0 . On GC and is around four times higher than would be expected based on the
graphite k0 is 1.5 × 10−4 and 9.5 × 10−4 cm s−1 , respectively, com- conductivity results shown in Fig. 2 and may indicate either addi-
pared to a value of 30 × 10−4 cm s−1 which we previously reported tional ohmic losses, for instance in the carbon cloth, or poor control
at a platinum electrode under the same conditions. These values of temperature. After running a deposition cycle at limiting current
appear favourable compared to the rate constant associated with for 2 h the RFB was opened and the copper deposit on the negative
the positive electrode reaction in conventional aqueous vanadium electrode was found to be matte, adherent and free of dendrites.
sulphate based RFBs, which is reported to be 2 × 10−6 cm s−1 [24].

2 0
10
0
−1
−2 10
106(I/ω1/2)/A s1/2

−4 −2
10
−6
Ik/A

−3
−8 10

−10 −4
10
−12

−14
−0.5 −0.4 −0.3 −0.2 −0.1 0 0.1 0.2 0.3 −0.4 −0.3 −0.2 −0.1 0 0.1 0.2 0.3
η/V η/V

Fig. 3. Normalised staircase voltammetry data for a 20 mM CuCl2 electrolyte at a Fig. 5. Tafel plot for a 20 mM CuCl2 electrolyte at a glassy carbon electrode () and
5 mm GC electrode at rotation rates of 1000, 1250, 1500, 1750 and 2000 rpm. The a 2 M CuCl electrolyte at a graphite electrode (+) The Ik values for the GC electrode
arrow indicates decreasing rotation speed. 100% IR correction was applied during measured at negative  values are determined from the intercept of the Koutecký-
acquisition of the voltammogram. Levich plots shown in Fig. 4.
22 D. Lloyd et al. / Electrochimica Acta 100 (2013) 18–23

6 125% 125%

4
100% 100%
2

Qdischarge/Qtheory
0 75% 75%
−2

ηQ, ηE
j/mA cm

−2
50% 50%
−4

−6
25% 25%
−8

−10 0% 0%
0 0.2 0.4 0.6 0.8 1 1.2 0 10 20 30 40
E/V cycle number

Fig. 6. Staircase voltammogram of the all copper RFB described in Section 2.2. Fig. 7. The evolution of Coulombic efficiency, Q (), energy efficiency, E (+), and
the discharge capacity, Qdischarge (䊉), normalised by the theoretical capacity, Qtheory ,
during 38 charge–discharge cycles for the redox battery described in Section 2.3.
Cycling was performed at a current density of 10 mA/cm2 , with the exception
3.6. Performance of a miniature redox battery
of cycles 8–9 and 24–27 which were performed at current densities of 4 and
7.5 mA/cm2 respectively. The pronounced outliers in Q correlate with changes in
The miniature redox battery described in Section 2.3 was also the applied current density and resolve within a few cycles.
used to check the applicability of the ex-situ conductivity mea-
surements. Based on the approximate geometry of the diffusion
1.2
cells and ex situ conductivity measurements a total cell impedance
of 318 Ohm was expected at 50 ◦ C, of which 9 Ohm would be
attributable to the separator. In practise, measurements with 1
this battery in the presence and absence of a separator showed
impedances of 267 Ohm and 256 Ohm, respectively, with some of 0.8
Cell potential/V

the 20% difference from the ex-situ measurements possibly being

attributable to the slightly larger aperture of the cell in the vicinity 0.6
of the stirrer. This proves that the ex situ conductivity measure- cycle 1 cycle 2 cycle 3
ments can provide a good indication of the performance that can ηQ = 95.0% ηQ = 97.7% ηQ = 96.9%
0.4
be expected in a complete cell, provided all other ohmic losses
ηE = 61.5% ηE = 62.7% ηE = 61.8%
are minimised and the temperature is well defined. It also demon-
strates that the conductivity of the electrolyte is largely unchanged 0.2
at the solute concentrations used in this work and still corresponds
to that of ethaline. 0
The long-term performance of the redox chemistry presented 0 2 4 6 8 10 12 14 16
time/hour
in this work was tested by performing charge cycling in the minia-
ture redox battery over a period of seven days. A current density of
Fig. 8. Three charge/discharge cycles performed in the miniature redox battery
10 mA/cm2 (normalised used the separator area) was used for this described in Section 2.3 at a current density of 7.5 mA/cm2 , post experimental cor-
operation. During this time the battery was cycled 38 times. The rection of the ohmic losses in the cell were applied as described previously [18].
coulombic and energy efficiencies, Q and E , respectively, showed
little deviation over time, averaging 94.3±6.6% and 52.1±9.1%
respectively. Charge–discharge cycles 8 and 9 of Fig. 7 were performed at a
During charging and discharging the typical amount of charge current density of only 4 mA/cm2 . This appears to improve energy
passed were Qcharge = 23.0 ± 4.6 and Qdischarge = −21.8 ± 5.5 C respec- efficiency further, with both cycles displaying an energy efficiency
tively. This corresponds to only a fifth of the theoretical capacity, of around 75%. However, since a single charge–discharge cycle
Qtheory of the battery based on the amount of CuCl2 used. The vari- required 14 h this current density was not used for further mea-
ation in Q , E and Qcharge over time is shown in Fig. 7. surements and hence insufficient cycles were acquired to validate
The ocp values measured after each charge or discharge opera- the reproducibility of the observed energy efficiency.
tion were equally invariant over the time of operation with values
of 0.773 ± 0.006 and 0.683 ± 0.004 V respectively. These potentials 4. Conclusions
agree well with the voltammetric data for the higher concentration
electrolyte shown in Fig. 6. The redox reactions selected exhibit reasonable electrochemical
Fig. 8 shows charge–discharge cycles 24–27. These were per- kinetics. However, the small ocp suggests an all copper RFB is not a
formed at a lower current density of 7.5 mA/cm2 , this appeared to useful battery system, particularly in poorly conductive electrolytes
benefit the performance of the battery significantly, resulting in such as ILs where ohmic losses are pronounced.
a doubling of the amount of current passed to 46.8 ± 2 C, which An appreciable current density was not achievable in the RFB,
corresponds to a 40% depletion of the theoretical battery capacity. despite the high metal concentration in solution, due to the poor
Energy efficiency also improved, with an average value of 62 ± 1%. mass transport properties of the electrolyte. Even had a current
The coulombic efficiency was unchanged. Ocp values after charg- density of, for example, 50 mA/cm2 been achieved, this would have
ing and discharging shifted to 0.806 ± 0.006 V and 0.672 ± 0.002 V, resulted in ohmic losses of around 0.5 V and negligible efficiency.
respectively, with the larger shift in the ocp after charging explain- Elevating the operating temperature will reduce this problem and
ing some of the increase in efficiency. should be possible based on the separator presented here.
 D. Lloyd et al. / Electrochimica Acta 100 (2013) 18–23 23

This work has demonstrated that conceptually a DES based elec- [8] T. Katase, K. Murase, T. Hirato, Y. Awakura, Redox and transport behaviors of
trolyte system can be employed in a RFB and shown the simple Cu(I) ions in TMHA-Tf2N ionic liquid solution, Journal of Applied Electrochem-
istry 37 (2007) 339.
preparation of a suitable separator. None of the problems associ- [9] T.M. Anderson, D. Ingersoll, A.J. Rose, C.L. Staiger, J.C. Leonard, Synthesis of an
ated with chlorocuprate electrolytes reported by Porterfield and ionic liquid with an iron coordination cation, Dalton Transactions 39 (2010)
Yoke [19] were apparent, provided the electrolyte temperature was 8609.
[10] D. Zhang, Q. Liu, X. Shi, Y. Li, Tetrabutylammonium hexafluorophosphate and 1-
kept at 50 ◦ C or higher, suggesting that DES are suited for further ethyl-3-methyl imidazolium hexafluorophosphate ionic liquids as supporting
investigation as RFB electrolyte media. The behaviour of the system, electrolytes for non-aqueous vanadium redox flow batteries, Journal of Power
for instance the conductivity, cell potentials and electrode kinet- Sources 203 (2012) 201.
[11] R.L. Kay, B.J. Hales, G.P. Cunningham, Conductance and association behavior of
ics, showed good reproducibility across the range of concentrations
the alkali metal perchlorates and tetraphenylborides in anhydrous acetonitrile,
reported here. We will explore this further utilising an improved Journal of Physical Chemistry 71 (1967) 3925.
stack design that facilitates higher temperatures and alternative [12] A.P. Abbott, K.J. McKenzie, Application of ionic liquids to the electrodeposition
of metals, Physical Chemistry Chemical Physics 8 (2006) 4265.
redox couples that allow larger cell potentials.
[13] A.P. Abbott, G. Frisch, K.S. Ryder, Metal complexation in ionic liquids, Annual
For ethaline the viscosity, conductivity and speciation of metal Reports on the Progress of Chemistry, Section A: Inorganic Chemistry 104
complexes are similar to chloride rich ionic liquids, while the ease (2008) 21.
of handling and cost are improved. This makes it, at the very least, [14] D. Lloyd, T. Vainikka, L. Murtomäki, K. Kontturi, E. Ahlberg, The kinetics of
the Cu(II)/Cu(I) redox couple in deep eutectic solvents, Electrochimica Acta
a useful platform to develop the application of ILs as electrolyte 56 (2011) 4942.
media for RFBs. In this work we have shown the preparation of a [15] A.P. Abbott, G. Capper, K.J. McKenzie, A. Glidle, K.S. Ryder, Electropolishing of
simple separator which should be widely applicable for a range of stainless steels in a choline chloride based ionic liquid: an electrochemical
study with surface characterisation using sem and atomic force microscopy,
ionic liquids. Due to the small electrolyte volumes, excellent tem- Physical Chemistry Chemical Physics 8 (2006) 4214.
perature control, corrosion resistance and relative simplicity, the [16] A.P. Abbott, K. El Ttaib, G. Frisch, K.J. McKenzie, K.S. Ryder, Electrodeposition
miniature redox battery reported here is also particularly suitable of copper composites from deep eutectic solvents based on choline chloride,
Physical Chemistry Chemical Physics 11 (2009) 4269.
for initial measurements of novel redox systems in ionic liquids or [17] K. Haerens, E. Matthijs, K. Binnemans, B. Van der Bruggen, Electrochemical
molten salts. decomposition of choline chloride based ionic liquid analogues, Green Chem-
istry 11 (2009) 1357.
[18] D. Lloyd, T. Vainikka, S. Schmachtel, L. Murtomäki, K. Kontturi, Simultaneous
References characterisation of electrode kinetics and electrolyte properties in ionic liquids
using a rotating disc electrode, Electrochimica Acta 69 (2012) 139.
[1] M. Skyllas-Kazacos, M.H. Chakrabarti, S.A. Hajimolana, F.S. Mjalli, M. Saleem, [19] W.W. Porterfield, J.T. Yoke, Inorganic Compounds with Unusual Properties, ACS
Progress in flow battery research and development, Journal of The Electro- Publications, Washington, DC, p. 104, 1976.
chemical Society 158 (2011) R55. [20] C. Angell, C. Liu, E. Sanchez, Rubbery solid electrolytes with dominant cationic
[2] C. Wadia, P. Albertus, V. Srinivasan, Resource constraints on the battery energy transport and high ambient conductivity, Nature (1993) 137.
storage potential for grid and transportation applications, Journal of Power [21] A.P. Abbott, A.D. Ballantyne, J.P. Conde, K.S. Ryder, W.R. Wise, Salt modified
Sources 196 (2011) 1593. starch: sustainable, recyclable plastics, Green Chemistry 14 (2012) 1302.
[3] G. Nikiforidis, L. Berlouis, D. Hall, D. Hodgson, Evaluation of carbon composite [22] A. Bard, L. Faulkner, Electrochemical Methods: Fundamentals and Applications,
materials for the negative electrode in the zinc-cerium redox flow cell, Journal Wiley, 2000.
of Power Sources 206 (2012) 497. [23] T. Vainikka, D. Lloyd, L. Murtomäki, J.A. Manzanares, K. Kontturi, Elec-
[4] T. Tsuda, C.L. Hussey, electrochemistry of room-temperature ionic liquids and trochemical study of copper chloride complexes in the RTIL 1-butyl-1-
melts Modern Aspects of Electrochemistry, vol. 45, Springer, New York, 2009. methylpyrrolidinium bis(trifluoromethylsulfonyl)imide, Electrochimica Acta
[5] M. Lipsztajn, R.A. Osteryoung, Electrochemistry in neutral ambient- 87 (2013) 739.
temperature ionic liquids. 1. Studies of iron(III), neodymium(III), and [24] A.Z. Weber, M.M. Mench, J.P. Meyers, P.N. Ross, J.T. Gostick, Q. Liu, Redox
lithium(I), Inorganic Chemistry 24 (1985) 716. flow batteries: a review, Journal of Applied Electrochemistry 41 (2011)
[6] H.D. Pratt III, A.J. Rose, C.L. Staiger, D. Ingersoll, T.M. Anderson, Synthesis and 1137.
characterization of ionic liquids containing copper, manganese, or zinc coordi- [25] N.G. Hung, Z. Nagy, Kinetics of the ferrous/ferric electrode reaction in the
nation cations, Dalton Transactions 40 (2011) 11396. absence of chloride catalysis, Journal of the Electrochemical Society 134 (1987)
[7] C. Austen Angell, Y. Ansari, Z. Zhao, Ionic liquids: past, present and future, 2215.
Faraday Discussions 154 (2012) 9.