Studies on iron (Fe3+/Fe2+) -Complex/bromine (Br 2 Br-) redox flow cell in sodium acetate solution

Article abstract of DOI:10.1149/1.2186040

The formal potential of the Fe(III)Fe(II) couple shifts markedly in the negative direction by complexation with ethylenediamine tetraacetate (EDTA), oxalate, and citrate. The potentials of the complexes with EDTA and oxalate are less pH-dependent than wit

Full text of DOI:10.1149/1.2186040

Journal of The Electrochemical Society, 153 ͑5͒ A929-A934 ͑2006͒
A929
0
013-4651/2006/153͑5͒/A929/6/$20.00 © The Electrochemical Society
3
+
2+
−
Studies on Iron „Fe /Fe …-Complex/Bromine „Br /Br … Redox
2
Flow Cell in Sodium Acetate Solution
a,b
a,z
a
a
a
a
Y. H. Wen, H. M. Zhang, P. Qian, H. T. Zhou, P. Zhao, B. L. Yi, and
b
Y. S. Yang
a
Full Cell R&D Center, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian,
Liaoning 116023, China
b
Research Institute of Chemical Defense, Beijing 100083, China
The formal potential of the Fe͑III͒/Fe͑II͒ couple shifts markedly in the negative direction by complexation with ethylenediamine
tetraacetate ͑EDTA͒, oxalate, and citrate. The potentials of the complexes with EDTA and oxalate are less pH-dependent than with
citrate. But, the relatively high pH of around 6.0 is favorable electrochemically due to high corresponding currents. Complexation
of Fe͑III͒/Fe͑II͒ couple can provide fast electrode kinetics except for the complex with citrate. But, the solubility of the complex
with citrate is up to 0.8 M. Charge–discharge measurements were conducted with the iron-complex/Br redox cells. The results
2
show that performance of the cells with 0.1 M Fe͑III͒/Fe͑II͒-oxalate or Fe͑III͒/Fe͑II͒-citrate is relatively poor due to slow
kinetics for the Fe͑III͒/Fe͑II͒-citrate and the unstability of the ferric form for the Fe͑III͒/Fe͑II͒-oxalate, whereas performance of
the iron-citrate/Br2 cell is improved considerably by increasing concentration of the Fe͑III͒-citrate complex. Also, energy effi-
ciencies of up to approximately 80 and 70% could be obtained for the cell with 0.1 M Fe͑III͒/Fe͑II͒-EDTA and 0.8 M
Fe͑III͒/Fe͑II͒-citrate, respectively. The preliminary study shows that novel Br /iron-complex cells are technically feasible in redox
2
flow batteries but need further investigation.
©
2006 The Electrochemical Society. ͓DOI: 10.1149/1.2186040͔ All rights reserved.
Manuscript submitted August 5, 2005; revised manuscript received January 17, 2006. Available electronically March 31, 2006.
Redox flow batteries ͑RFB͒ can be considered an electrochemi-
parison. Sodium acetate trihydrate ͑BDH͒ is used as a supporting
electrolyte as well as buffer solution. Then, the
cal system intermediate classical secondary batteries and fuel
a
1
-3
cells. Unlike conventional batteries, a redox flow battery stores
Fe͑III͒/Fe͑II͒-complex redox couples can be used in conjunction
−
energy in liquid electrolytes ͑in external tanks͒ containing different
with the Br /Br couple for a new redox flow cell combination. It is
2
4
redox couples that are pumped through the cell. Therefore, as the
preferable to use a sodium-based couple to allow the transport of
sodium to complete the circuit and minimize the cross-
contamination problem during operation of the cell.
active species, soluble redox couples have determining effects on
the performance of an RFB.
In the past few years, a number of redox couples have been
proposed and fabricated. Among the redox systems developed to
A new iron-complex/bromine redox flow cell is thus proposed
and employs the Fe͑III͒/Fe͑II͒-complex in the negative half-cell
5,6
−
date, only a polysulfide/bromine system and an all-vanadium sys-
tem have been successfully developed, showing the potential of
electrolyte and the Br /Br couple in the positive half-cell. Iron and
2
bromine species are chosen for this investigation because both of
them are abundant and inexpensive elements. Cyclic voltammetry is
used to study the electrochemical behaviors of the
Fe͑III͒/Fe͑II͒-complex couples, and the charge-discharge perfor-
mance of a small iron-complex/bromine test cell is reported.
1
-4,7-11
commercialization.
However, a long-term existing problem in
the storage cells has not been solved perfectly: the possible inter-
mixing of the components of two half-cells through an ion exchange
membrane. In addition, the long-term stability of Na S and V͑V͒
2
x
solutions in high vanadium concentrations and elevated tempera-
tures is limited in application due to the formation of insoluble pre-
cipitates.
Experimental
Reagents sodium oxalate, sodium citrate·2H O, disodium EDTA,
sodium acetate trihydrate, ferric sulfate, and sodium bromide were
all analytical grade. All solutions were prepared with distilled water.
2
An alternative approach to minimize this intermixing problem
and to improve the stability of redox solutions is the use of metal ion
coordination compounds as redox couples in flow cells. Bard and
1
2,13
co-workers
studied the possibility of Fe͑III͒/Fe͑II͒ and
Apparatus and procedure.— The supporting electrolyte was 1 M
aqueous sodium acetate ͑NaAC͒. It can also be used as a buffer
solution to maintain the pH of the solution around 6.0. When nec-
essary, 4 M NaOH or H SO solution was added to the electrolyte to
Co͑III͒/Co͑II͒ couples complexed with o-phenanthroline and bipy-
14
ridine ligands as positive half-cells. Murthy et al. further examined
the Fe͑III͒/Fe͑II͒ couple and shifted the redox potential to more
negative values by complexation with diethylenetriamine pentaace-
tic acid, nitrilotriacetic acid, and ethylenediaminetetraacetic acid
2
4
adjust the pH to the desired value. The pH of the solution was
measured with a calibrated pH meter ͑Shanghai instruments pHS-
25, China͒. Solutions were deaerated 15 min by bubbling with ni-
trogen before each measurement. The complexes were prepared by
mixing known concentrations of ferric sulfate and the ligand. In the
case of EDTA and citrate, a mole ratio of ligand/Fe͑III͒ of slightly
more than 1 was used. In the case of oxalate, a mole ratio of 3 for
ligand/Fe͑III͒ was used. The conductivity of various solutions was
measured by the DDS-12D electrical conductive meter ͑Shanghai
LIDA Instrument Factory, China͒. The curves of current vs potential
͑
EDTA͒ in 0.5 M K SO medium, making a negative half-cell pos-
2 4
sible. However, these studies are only related to electrochemical
behaviors of the half-cells and the solubility of active materials is
rather low. The redox flow battery based on Fe͑III͒/Fe͑II͒ com-
plexes has not been developed. In addition, the use of expensive
chelating agents such as o-phenanthroline and diethylenetriamine
pentaacetic acid also limited the application of the systems in some
ways.
We have developed a new electrolyte system, the iron-complex/
bromine redox system. In this system, the ligands with low cost are
employed to be capable of shifting the formal potential of
Fe͑III͒/Fe͑II͒ in the negative direction. The Fe͑III͒/Fe͑II͒-EDTA
complex couple was also tested under the same conditions for com-
were recorded in a three-compartment cell with a graphite rod ͑area
.103 cm ͒ as inert working electrodes and a large area graphite
2
0
sheet electrode as an auxiliary electrode. All potentials were ex-
pressed relative to an aqueous saturated calomel electrode ͑SCE͒,
which was connected with the electrochemical cell through a salt
bridge full of saturated potassium chloride solution. The cyclic vol-
tammogram was measured by the CHI660 electrochemical station
͑CH Corporation, USA͒.
z
E-mail: zhanghm@dicp.ac.cn
Downloaded on 2014-06-01 to IP 128.148.252.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A930
Journal of The Electrochemical Society, 153 ͑5͒ A929-A934 ͑2006͒
Table I. Redox couples used for the constant current charge–
discharge tests.
Anolyte
Catholyte
Cell 1
Cell 2
Cell 3
1 M NaBr
1 M NaBr
1 M NaBr
0.1 M Fe͑III͒-EDTA +1 M sodium acetate
0.1 M Fe͑III͒-oxalate +1 M sodium acetate
0.1 M Fe͑III͒-citrate +1 M sodium acetate
Small-redox-flow test cells were used for the constant current
charge–discharge tests with graphite felt electrodes, graphite plates
as the current collectors, and a series of different electrolytes. The
2
electrode and membrane areas were 5 and 6 cm , respectively. The
+
Nafion 117 cation-exchange membrane in the H form was soaked
in 10% NaOH solution for 1 h at 80°C and then converted into the
+
Na -form membrane. The two half-cell electrolytes were separated
+
by a sheet of Nafion 117 membrane in the Na form. Two Xishan
Figure 2. A standard curve determining the content of Br− by the UV-vis
absorption spectra.
pumps ͑China͒ were used to pump each half-cell electrolyte through
the corresponding half-cell cavity where the charge/discharge reac-
tions occurred. Each cell compartment was filled with 50 mL of
these electrolytes. Prior to charging the cell, the negative electrolytes
were purged with nitrogen gas at 0.2 L min−1 for 30 min. The gal-
0.4 mL of the buffer solution and 0.1 mL of 0.6 mM phenol red
were added. After shaking fully, 0.2 mL of 1.8 mM chloramine T
was added into the flask and maintained for 5 min. Finally, 0.1 mL
of 2 M sodium thiosulfate solution was added to reduce the exces-
vanostatic charge and discharge of the test cell were carried out
using a CT2001A multichannel generator ͑Land instruments,
China͒. The applied current density for the charge and discharge was
−
−
2
sive chloramines T. Then Br was oxidized by the chloramine T to
1
0 mA cm . A range of combinations of redox couples was used in
Br , reacting with the phenol red to the phenol blue. The blue solu-
2
the performance tests of the redox flow cells. These are shown in
Table I.
tion can be analyzed by ultraviolet-visible ͑UV-vis͒ absorption spec-
tra obtained with a double-beam spectrometer ͑Unico, UV-4802͒
The permeation of Br across the membrane was determined. A
2
referenced to Br -free aqueous solutions, as shown in Fig. 1. A
2
redox-flow-model cell was used for the measurement of the perme-
−
−
standard curve was also obtained to determine the Br ion content,
ation of Br . Br exists in the form of Br in the concentrated NaBr
2
2
3
−
as shown in Fig. 2.
solution. The Br3 solution was obtained by charging the cell em-
ploying 2.0 M NaBr as the positive electrolyte and 0.8 M
Fe͑III͒/Fe͑II͒-citrate complex as the negative electrolyte to the state
of 50% state of charge ͑SOC͒ at a current density of 10 mA cm .
Then one reservoir was filled with the above Br solution and the
Results and Discussion
According to stable constants, it is qualitatively known that a
number of ligands can form highly stable complexes with Fe͑III͒
and cause negative shift in the potential of the Fe͑III͒/Fe͑II͒ couple.
However, only a small quantity of chelating agents such as oxalate,
citrate, and EDTA can form Fe͑III͒/Fe͑II͒-complexes whose elec-
trode reactions are reversible to some extent.
−
2
−
3
other one with a solution of 0.1 M EDTA. Both solutions were
5
0 mL and circulated through the cell compartments, which were
separated by the Nafion 117 exchange membrane with effective area
of 5 cm at ambient temperature. Samples of 2 mL were taken at
2
regular intervals from the EDTA aqueous solution reservoir and ana-
Solubilities.— Table II is a list of the solubilities of the com-
plexes in aqueous 1 M sodium acetate. Uncomplexed Fe͑III͒-sulfate
salts are quite soluble and yield solutions with metal-ion concentra-
tions close to 1 M. The solubility of the complexes varies with the
nature of the ligand. Sodium citrate is quite soluble in water media
up to 1.8 M. The solubility of the Fe͑III͒-citrate complex is approxi-
mately the same as uncomplexed Fe͑III͒. However, the complex
with citrate in concentrations above 0.5 M forms relatively viscous
and deeply colored solutions. The solubilities of sodium oxalate acid
and disodium EDTA in water media are limited, at 0.4 and 0.3 M,
−
lyzed to determine the Br ion content by a method described in
1
5
detail in the literature, because the penetrated Br was reduced by
2
−
EDTA to Br ions. This method is as follows: the mixing solution of
sodium acetate and acetate acid was used as a buffer to maintain the
pH of the solution in the range 4.5–4.7. Standard solutions contain-
−
ing 1–20 ␮g Br were respectively added into a 10 mL flask, then
a
Table II. Estimated solubilities of Fe„III…-Fe„II… couples and
b
ligands.
Solubility
Concentration
Substance
͑g/100 mL͒
͑mol/L͒
Fe ͑SO ͒
4 3
Fe͑III͒-EDTA
Fe͑III͒-citrate
32
6.9
22
3.2
15
69.3
4
0.8
0.2
0.8
0.1
0.4
1.8
0.3
2
Fe͑III͒-͑oxalate͒3
Disodium EDTA
Sodium citrate·2H O
2
Sodium oxalate
a
Figure 1. A typical UV-vis absorption spectra of the Br -phenol bule solu-
tion containing 20 ug Br .
Measured in 1 M sodium acetate.
Measured in distilled water. T = 20°C ± 2°C.
2
−
b
Downloaded on 2014-06-01 to IP 128.148.252.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 153 ͑5͒ A929-A934 ͑2006͒
A931
form for quasi-reversible 1e transfer reactions with peak separations,
⌬
rates.
Ep more than 60 mV, which increases with potential scan
1
6-19
However, the Fe͑III͒-EDTA complex shows the lowest
value of ⌬Ep ͑ϳ70 mV͒ close to a reversible system, which be-
comes wider slightly with potential scan rates, indicating that the
Fe͑III͒-EDTA complex redox reaction would be very fast. The
Fe͑III͒-citrate complex shows much lower responding currents and
wider peak potential separation ͑ϳ180 mV͒ even at the low sweep
of 10 mV s− than the other two complexes. The higher the potential
scan rates, the more round its reduction peak becomes. This indi-
cates that the electrode process of Fe͑III͒/Fe͑II͒-citrate may be
slow, possessing relatively large polarization resistance. In addition,
the electrode process of the Fe͑III͒-oxalate complex is more like
that of the Fe͑III͒-EDTA complex in terms of the responding cur-
rents and peak potential separation.
1
The same reaction of the complex, i.e., the Fe͑III͒-oxalate, is
quasi-reversible, as deduced by the proportionality between the an-
1
/2
odic peak current density ͑i ͒ and v seen in Fig. 4a. Similar plots
pa
of the peak current vs the square root of the scan rate were obtained
for the complex with citrate, whereas the plot of the peak current for
the oxidation process of the Fe͑III͒-EDTA complex vs the square
root of the scan rate shown in Fig. 4b is an almost linear response,
indicative of relatively fast electrode kinetics and a diffusion-
controlled reaction.
Figure 3. CVs recorded at different potential scan rates at a graphite elec-
trode for 50 mM Fe͑III͒-complexes with EDTA ͑a͒, oxalate ͑b͒, and citrate
The effects of pH on the CV of various Fe͑III͒ complexes are
demonstrated in Fig. 5. It can seen that with an increase in pH over
͑
c͒ in 1 M sodium acetate. Scan rate: ͑1͒ 10, ͑2͒ 25, ͑3͒ 50, ͑4͒ 100, and ͑5͒
00 mV/s.
1
–7, the general appearance of the CVs improves, with higher re-
2
sponding currents and smaller peak potentials separation, indicating
a greater reversibility and improved kinetic characteristics, as well
as stabilization of Fe͑III͒ complexed species. Also, the formal po-
tentials shift in the negative direction. A similar observation was
respectively, leading to low concentration of complexes. The com-
plexes in low concentrations are transparent in the visible region,
similar to the uncomplexed Fe͑III͒ ions in aqueous solution.
14
19
made by Murthy et al. and Kolthoff et al. The negative shift of
the Fe͑III͒/Fe͑II͒ formal potential can be explained in terms of the
formation of the more stable complexes. However, the pH cannot be
too high. For example, when the pH is up to 6.60 in Fig. 5a, the
potential at which oxygen evolution takes place shifts in the nega-
tive direction. In addition, for EDTA and oxalate, there is little
change in the formal potentials of Fe͑III͒-complexes when the pH is
above 3. In the case of citrate, the formal potential of the complex
cannot remain unchanged until the pH reaches 5.11. Therefore, the
half-cell potentials of the complexes with EDTA or oxalate are less
pH-dependent. However, the preferable pH range may be in the
neighborhood of 6 for the Fe͑III͒/Fe͑II͒ complexes due to higher
responding currents.
Cyclic voltammetry tests for Fe͑III͒/Fe͑II͒-complexes redox pro-
cesses.— As shown in Fig. 3, typical cyclic voltammograms ͑CVs͒
recorded at a graphite electrode at a low scan rate of 10 mV s− in
1
5
0 mM of various Fe͑III͒-complexes dissolved in 1 M sodium ac-
etate buffer solution show well-formed oxidation and coupled reduc-
tion peaks below 0 V vs SCE. The formal potentials of the
0
Fe͑II͒/Fe͑III͒ complexes E , was estimated from the CVs by taking
the mean of the average of the anodic and cathodic peak potentials,
0
Epa and E , namely, E = ͑E + E ͒/2. The formal potentials of
pc
pa
pc
the complexes with EDTA, oxalate, and citrate are calculated to be
−
0.104, −0.190, and −0.213 V ͑vs SCE͒, respectively.
To investigate the kinetics of the Fe͑III͒-ligands/Fe͑II͒-ligands
Comparison of charge–discharge performance of the test cell
redox reaction qualitatively, the CVs recorded at a range of potential
scan rates ͑␯͒ for 50 mM Fe͑III͒-ligands in 1 M sodium acetate
solution are shown in Fig. 3. The voltammograms have the classical
battery employing a range of redox couples.— Performance of a
−
RFB employing the Br /Br couple as anolyte-active species and the
2
Fe͑III͒/Fe͑II͒-complexes with EDTA, oxalate, and citrate as
Figure 4. Plots of the peak current density
1
/2
͑
ip͒ vs the square root of scan rate ͑v ͒
for the oxidation processes of the
Fe͑III͒-complexes with oxalate ͑a͒ and
EDTA ͑b͒.
Downloaded on 2014-06-01 to IP 128.148.252.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A932
Journal of The Electrochemical Society, 153 ͑5͒ A929-A934 ͑2006͒
Figure 7. Charge–discharge curves for the iron-ligand/Br test cells employ-
2
ing different Fe͑III͒/Fe͑II͒-complexes as negative couples: ͑a͒ cell 1, with
Fe͑III͒/Fe͑II͒-/EDTA redox couples, ͑b͒ cell 2, with Fe͑III͒/Fe͑II͒-oxalate
couples, and ͑c͒ cell 3, with Fe͑III͒/Fe͑II͒-citrate couples. The initial con-
centrations of the electrolytes before charging for the three cells are given in
Table I. The electrolyte volume in each compartment was 50 mL and the
charge–discharge current was 50 mA.
Figure 5. Effect of pH on the CVs of the Fe͑III͒ complexes with different
ligands: ͑a͒ EDTA, ͑b͒ oxalate, and ͑c͒ citrate.
catholyte-active species was evaluated with constant-current
charge–discharge tests and open-circuit voltage measurements, re-
548 min due to the Donnan exclusion effect of the cation-exchange
−
membrane to Br −3 anions.
20,21
spectively. During charging, Br is oxidized to Br , which goes into
In addition, employing the thick
Nafion 117 membrane is also beneficial to prevent the penetration of
ions of two half-cells.
In Fig. 7, three charge–discharge curves are shown for the RFB
test cell with different negative Fe͑III͒/Fe͑II͒-complexes:
2
−
solution as tribromide ions, Br , which are available to reoxidize the
3
Fe͑II͒-complex to Fe͑III͒-complex during discharge. Free bromine
is not very soluble in water, but it is very soluble in bromide salt
solutions in which it forms polybromide ions without separation of a
second liquid phase. A small quantity of Br −3 ions may cross the
membrane due to a large concentration difference between the posi-
tive and negative electrolytes. At the present stage, the concern is
͑
a͒ cell 1, with Fe͑III͒/Fe͑II͒-EDTA redox couples; ͑b͒ cell 2, with
Fe͑III͒/Fe͑II͒-oxalate couples; and ͑c͒ cell 3, with
Fe͑III͒/Fe͑II͒-citrate couples. Initial concentrations of the electro-
lytes for the three cells are given in Table I. It is shown that cells 2
and 3 can be charged to higher voltage due to relatively low formal
potentials of the Fe͑III͒/Fe͑II͒-complexes with oxalate and citrate.
mainly to demonstrate that the iron-complex/Br redox cell is still
2
technically feasible. Figure 6 shows the permeability of the Nafion
1
17 cationic exchange membrane for Br . It is confirmed that in the
2
presence of bromine, the three complexing agents may be oxidized
and the bromine is reduced to Br due to no Br determined. How-
ever, though the permeation of Br ions through the membrane in-
creases with time, the permeability of Br in concentrated Br solu-
However, in the case of the Fe͑III͒/Fe͑II͒-EDTA/Br system ͑cell
2
−
2
1͒, the initial voltage charged is much lower than that of the other
−
3
systems, exhibiting rather low internal iR drop. For the
−
2
Fe͑III͒/Fe͑II͒-oxalate/Br system ͑cell 2͒, the initial voltage charged
2
tion is rather low, on the order of parts per million, for as long as
is lower than that of the Fe͑III͒/Fe͑II͒-citrate system, but a larger
internal iR drop can be observed, probably due to the limited solu-
bility of the Fe͑II͒-oxalate complex.
Data obtained from the charge–discharge test of the cells are
summarized in Table III, where various cell efficiencies are the av-
erage values over ten charge/discharge cycles. It is found that the
three cells have similar open-circuit voltages under the present con-
ditions. As expected, due to faster electrode kinetics, cell 1 with
Fe͑III͒/Fe͑II͒-EDTA has much higher voltage and energy efficien-
cies than the other two cells, while the average voltage and energy
efficiencies for cell 3 are a bit higher than that for cell 2.
Table III. Open-circuit voltage and efficiencies of the Fe-ligand/
Br2 test cells employing a series of Fe complexes with different
ligands as negative couples.
Current
efficiency
͑%͒
Voltage
efficiency
͑%͒
Energy
efficiency
͑%͒
Open circuit
voltage ͑V͒
Cells
Figure 6. Permeability of Nafion 117 cationic exchange membrane for Br
2
−
in the Br solution obtained by charging the cell employing 2.0 M NaBr as
Cell 1
Cell 2
Cell 3
1.04
1.17
1.07
92.1
92.3
92.0
86.2
65.5
63.8
79.4
60.4
58.6
3
the positive electrolyte and 0.8 M Fe͑III͒/Fe͑II͒-citrate complex as the nega-
−
2
tive one to the state of 50% SOC at the current density of 10 mA cm
.
Downloaded on 2014-06-01 to IP 128.148.252.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 153 ͑5͒ A929-A934 ͑2006͒
A933
Figure 8. Cell capacity vs cycle number for the iron-ligand/Br2 test cells
employing
Fe͑III͒/Fe͑II͒-EDTA,
Fe͑III͒/Fe͑II͒-citrate.
Fe͑III͒/Fe͑II͒-complexes
as
negative
couples:
and
͑a͒
͑c͒
͑b͒
Fe͑III͒/Fe͑II͒-oxalate,
Figure 10. Conductivity of the Fe͑III͒-citrate complex solution as a function
of concentration of the Fe͑III͒-citrate.
Variation of the cell discharging capacity with cycling number is
given in Fig. 8. The capacity of cells 1 and 3 almost keeps constant
with an increase in cycle number, indicating that the intermixing of
two half-cells is minimized, while the capacity of cell 2 continues to
decrease with rising cycle number, possibly due to the instability of
the ferrous form in the oxalate solution.
tration of the electrolyte up to 0.5 M. Afterward, the conductivity of
the electrolyte basically remains unchanged, possibly resulting from
an increase in viscosity. Since a high conductivity will minimize the
battery resistance and optimize the energy efficiency, variation of
cell performance of the Fe͑III͒/Fe͑II͒-citrate/Br system with con-
2
centration of the Fe͑III͒-citrate lower than 0.7 M is mainly attrib-
uted to a change in conductivity of the Fe͑III͒-citrate complex so-
lution. When the concentration of the Fe͑III͒-citrate is 0.8 M, the
cell efficiencies start to decay slightly. Figure 11 shows charge–
discharge curves of the Fe͑III͒/Fe͑II͒-citrate/Br2 cells employing
the 0.7 and 0.8 M Fe͑III͒-citrate complexes, respectively. It can be
seen that compared with the cell with 0.7 M Fe͑III͒-citrate, the dis-
charging potential of the cell with 0.8 M Fe͑III͒-citrate initially
drops more rapidly and the discharging reaction mainly takes place
at the lower potential indicative of a large polarization resistance,
possibly resulting from a high viscosity of the electrolyte. So the
maximal possible concentration of the Fe͑III͒/Fe͑II͒-citrate com-
plex in 1 M sodium acetate feasible in the RFB can be up to 0.8 M.
Conversely, when the concentration of the Fe͑III͒-EDTA complex is
increased to 0.2 M, the current efficiency is almost unchanged, but
the voltage efficiency decreases to 73.16%. Obviously, this also may
be attributed to an increase in viscosity of the electrolyte.
As shown above, the solubility of the Fe͑III͒-citrate complex in
M sodium acetate can be up to 0.8 M. Hence, dependence of cell
1
efficiencies of the Fe͑III͒/Fe͑II͒-citrate/Br system on concentration
2
of the Fe͑III͒-citrate complex should be discussed as seen in Fig. 9,
where 2.0 M NaBr is employed as the positive electrolyte when
concentration of the Fe͑III͒-citrate is beyond 0.5 M. The current
efficiency decreases a bit with concentration of the Fe͑III͒-citrate
complex due to an increase in the charge–discharge time. However,
a considerable increase in both voltage and energy efficiency is ob-
served for concentration of the Fe͑III͒-citrate increased to 0.3 M,
followed by an almost unchanged trend after concentration of the
Fe͑III͒-citrate continues to be raised to 0.7 M. When concentration
of the Fe͑III͒-citrate is as high as 0.8 M, the voltage and energy
efficiencies start to show a relatively apparent drop. In order to give
an explanation for the above phenomenon, the conductivity of the
Fe͑III͒-citrate electrolyte as a function of concentration was deter-
mined experimentally, as shown in Fig. 10. It is found that the
conductivity of the Fe͑III͒-citrate electrolyte increases with concen-
Figure 11. Charge–discharge curves for the iron-ligand/Br test cell employ-
2
ing the 0.7 M ͑a͒ and 0.8 M ͑b͒ Fe͑III͒-citrate complex as negative couple
and 2 M NaBr as the positive electrolyte. The electrolyte volume in each
compartment was 50 mL and the charge–discharge current was 50 mA.
Figure 9. Dependence of efficiencies of the Fe-citrate/Br2 cell on concen-
tration of the Fe͑III͒-citrate.
Downloaded on 2014-06-01 to IP 128.148.252.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A934
Journal of The Electrochemical Society, 153 ͑5͒ A929-A934 ͑2006͒
Therefore, results of the charge/discharge tests are largely in
with Fe͑III͒/Fe͑II͒-EDTA and Fe͑III͒/Fe͑II͒-citrate almost remains
unchanged with increasing cycle number, minimizing the cross-
contamination problem. Although performance of the cell with the
0.1 M Fe͑III͒/Fe͑II͒-citrate is relatively poor, cell performance is
improved considerably by increasing concentration of the
Fe͑III͒-citrate complex. The maximal possible concentration of the
Fe͑III͒/Fe͑II͒-citrate complex in 1 M sodium acetate feasible in the
RFB can be up to 0.8 M with energy efficiency of about 70%. How-
ever, further work is still needed to improve performance of the
iron-complex/bromine cells by optimizing the electrolyte composi-
tion and cell component materials and increasing the solubility of
the Fe͑III͒/Fe͑II͒-complexes.
agreement with the CV experiments. Moreover, cell performance of
the Fe͑III͒/Fe͑II͒-citrate/Br2 system has been improved consider-
ably by increasing concentration of the Fe͑III͒-citrate complex.
While the electrolyte composition and cell component materials
have yet to be optimized and the solubility of the
Fe͑III͒/Fe͑II͒-complexes is still relatively low for redox flow cell
applications, the above results have demonstrated that novel iron
3
+
2+
−
͑
Fe /Fe ͒-ligand/bromine͑Br /Br ͒ cells are technically utilizable
2
in redox flow batteries. It is also suggested that the solubility of the
Fe͑III͒-complexes might be improved by adding some additives or
elevating temperatures. Further investigations are needed.
Conclusion
Acknowledgments
In this study, the voltammetric behaviors of the
Fe͑III͒/Fe͑II͒-complexes as negative redox couples in 1 M sodium
acetate medium, influences of pH, electrode kinetics, and electro-
lytic solubility were investigated. The following conclusions can be
drawn.
The authors are grateful for financial support from the Scientific
Research and Innovation Fund of the Knowledge Innovation Pro-
gram of the Chinese Academy of Science ͑no. K2002D3͒.
Chinese Academy of Sciences assisted in meeting the publication costs of
this article.
Complexation of Fe͑III͒/Fe͑II͒ couples with EDTA, oxalate, and
citrate results in significant negative shifts in the potential of the
redox couple. The solubility of the Fe͑III͒/Fe͑II͒-citrate complex
appears more satisfactory. The potentials of the complexes with
EDTA or oxalate are less pH-dependent, but relatively high pH of
around 6.0 is electrochemically favorable. The medium employed
causes the systems to have a suitable pH value ͑ϳ6͒. The electrode
process for the Fe͑III͒/Fe͑II͒-complexes is electrochemically quasi-
reversible. Nevertheless, complexation of the Fe͑III͒/Fe͑II͒ couple
can provide fast electrode kinetics, except for the complex with
citrate.
References
1
. E. Sum and M. Skyllas-Kazacos, J. Power Sources, 15, 179 ͑1985͒.
2. M. Skyllas-Kazacos, M. Rychcik, R. G. Robins, A. G. Fane, and M. Green, J.
Electrochem. Soc., 133, 1057 ͑1986͒.
. M. Skyllas-Kazacos, D. Kasherman, R. Hong, and M. Kazacos, J. Power Sources,
5, 399 ͑1991͒.
. L. Joerissen, J. Garche, C. Fabjan, and G. Tomazic, J. Power Sources, 127, 98
͑2004͒.
. C. J. Rydh and B. A. Sanden, Energy Convers. Manage., 46, 1957 ͑2005͒.
. C. J. Rydh and B. A. Sanden, Energy Convers. Manage., 46, 1980 ͑2005͒.
. M. Skyllas-Kazacos, J. Power Sources, 124, 299 ͑2003͒.
8. R. J. Remick and P. G. P. Ang, U.S. Pat. 4,485,154 ͑1984͒.
. A. Price, S. Bartley, S. Male, and G. Cooley, Power Eng. J., 13, 122 ͑1999͒.
3
4
3
5
6
7
9
Results obtained from a comparison of the charge–discharge per-
−
10. Ch. Fabjan, J. Garche, B. Harrer, L. Jörissen, C. Kolbeck, F. Philippi, G. Tomazic,
and F. Wagner, Electrochim. Acta, 47, 825 ͑2001͒.
11. C. H. Bae, E. P. L. Roberts, and R. A. W. Dryfe, Electrochim. Acta, 48, 279 ͑2002͒.
formance of test cells employing Br /Br as anolyte-active species
2
and a variety of Fe͑III͒/Fe͑II͒-complexes as catholyte-active species
3
+
2+
1
2. Y. W. D. Chen, K. S. V. Santhanam, and A. J. Bard, J. Electrochem. Soc., 128,
ones, demonstrate that novel iron ͑Fe /Fe ͒-complex/bromine
−
1460 ͑1981͒.
͑
Br /Br ͒ cells are technically feasible for use in a RFBs. An open-
2
1
3. Y. W. D. Chen, K. S. V. Santhanam, and A. J. Bard, J. Electrochem. Soc., 129, 61
͑1982͒.
14. A. S. N. Murthy and T. Srivastava, J. Power Sources, 27, 119 ͑1989͒.
15. D. Y. Qi, C. L. Qu, and T. Z. Zhou, Marine Sciences, 4, 27 ͑1981͒.
6. R. S. Nicholson, Anal. Chem., 37, 1351 ͑1965͒.
7. W. F. Zhou, Electrochemical Measurement, 1st ed., pp. 67–70, Scientific Technol-
circuit voltage of approximately 1.1 V and an energy efficiency of
approximately 80% could be obtained for the cell with
−
2
Fe͑III͒/Fe͑II͒-EDTA at a current density of 10 mA cm , while the
1
1
cells with Fe͑III͒/Fe͑II͒-oxalate or Fe͑III͒/Fe͑II͒-citrate exhibit
relatively
low
voltage
efficiencies.
The
cell
with
ogy Press, Shanghai ͑1983͒.
1
1
2
8. J. G. Lbanez, C. S. Choi, and R. S. Becker, J. Electrochem. Soc., 134, 3083 ͑1987͒.
9. I. M. Kolthoff and C. Auerbach, J. Am. Chem. Soc., 74, 1452 ͑1952͒.
0. T. Mohammadi and M. Skyllas-Kazacos, J. Membr. Sci., 77, 98 ͑1995͒.
Fe͑III͒/Fe͑II͒-oxalate has a high initial discharge capacity, but the
instability of the ferrous form results in a rapid loss of capacity upon
a charge–discharge cycle test. Conversely, the capacity of the cells
21. A. Heintz and C. Illenberger, J. Membr. Sci., 113, 175 ͑1996͒.
Downloaded on 2014-06-01 to IP 128.148.252.35 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).