вход по аккаунту



код для вставкиСкачать
Author’s Accepted Manuscript
Flow Battery Based on Reverse Electrodialysis
with Bipolar Membranes: Single Cell Experiments
Jiabing Xia, Gerhart Eigenberger,
Strathmann, Ulrich Nieken
To appear in: Journal of Membrane Science
Received date: 18 April 2018
Revised date: 23 July 2018
Accepted date: 25 July 2018
Cite this article as: Jiabing Xia, Gerhart Eigenberger, Heinrich Strathmann and
Ulrich Nieken, Flow Battery Based on Reverse Electrodialysis with Bipolar
Membranes: Single Cell Experiments, Journal of Membrane Science,
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
Flow Battery Based on Reverse Electrodialysis with Bipolar
Membranes: Single Cell Experiments
Jiabing Xia, Gerhart Eigenberger, Heinrich Strathmann, Ulrich Nieken*
Institute of Chemical Process Engineering, Universität Stuttgart, Boeblinger Strasse 78, D70199 Stuttgart, Germany
E-mail address:
Corresponding author.
The efficient storage of electrical energy is a key issue for a sustainable electrical
energy supply from fluctuating sources such as windmills or photovoltaic devices. In
addition to traditional electrical energy storage devices, flow batteries have become a
subject of intensive research and development activities. This publication is
concentrated on a flow battery type, based on the neutralization of an acid and a
base by reverse electrodialysis with bipolar membranes. The fundamental aspects of
the process will be discussed and experiments at laboratory scale in a single
repeating cell unit will be presented, using HCl as acid, NaOH as base and NaCl as
neutral salt solution. The components and process parameters which determine the
performance of this battery, such as the monopolar and bipolar membranes, the
concentrations of acid and base, and the operating temperature, have been
investigated and their effect on the overall efficiency of the battery will be discussed.
In a following second part, experimental results for stacks with 5 to 20 repeating cell
units will be presented [15]. Key issues for a successful implementation of a flow
battery based on acid/base neutralization are the membrane permselectivities,
depending on the fixed charge density, the ionic resistances of the mono- and bipolar
ion-exchange membranes and the water permeability of the bipolar membrane.
Graphical Abstract
AM, anion exchange membrane; BM, bipolar membrane; CM, cation exchange
membrane; DC, direct current; REDBP, reverse electrodialysis with bipolar
Electrical energy storage, acid/base neutralization flow battery, reverse electrodialysis with
bipolar membranes; bipolar membrane
The efficient storage of energy is a key issue for a sustainable electric energy supply
from fluctuating sources such as windmills or photovoltaic devices. Energy storage
systems particularly suited for this application are so-called flow batteries in which the
storage and the conversion of electrical energy are separated. This allows for the
optimization of the required conversion power, independently of the capacity of the
storage. Efficient and affordable flow battery systems are in particular required for the
decentralized use of renewable energy and could help to reduce the load fluctuations
of the electric grid. Presently, vanadium redox flow batteries are among the best
known and studied flow batteries. They are based upon the electrochemical
conversion of vanadium salt solutions between different degrees of oxidation [1] and
provide a comparatively high open cell voltage of about 1.4 V. Drawbacks are the
relatively high costs and the environmental impact of the required vanadium
A flow battery based on the neutralization of acids and bases in an electrodialysis
process with bipolar membranes is an interesting alternative to the redox battery. The
main advantage of this concept is that the electrolytes to be used in the process such
as HCl, NaOH and NaCl are abundant and inexpensive. Electrodialysis with bipolar
membranes is an established electromembrane process, used for converting a salt
solution into its respective acid and base [2-4]. Its reversal as battery, however, is still
at a very early stage of development since it has first been mentioned in [5] and [6].
Neutralization flow battery with bipolar membranes
2.1. Principle of a bipolar membrane neutralization flow battery
The process principle of the battery is illustrated in Fig. 1, which shows a schematic
drawing of an electrodialysis stack. The stack consists of two electrode
compartments at each side and several repeating cell units in between. During
charging of the battery (lower half of Fig. 1), in each repeating cell unit an acid, here
HCl, and a base, NaOH, are generated from a NaCl solution via water splitting in the
bipolar membrane (BM). The reversal of this process represents the discharge of the
battery with the generation of electricity from the electrodialytical neutralization of
acid and base in the bipolar membrane (upper half of Fig. 1).
Each repeating cell unit of the stack is composed of three compartments, separated
by two monopolar and one bipolar membrane. The NaCl, HCl and NaOH containing
compartments are connected to outside reservoirs. The sizes of the reservoirs
determine the charge capacity of the battery. In a practical application 50 to 100 cell
units could be stacked between two electrodes. In the experiments to be discussed,
electrodes of expanded titanium mesh, coated with platinum, have been used. The
electrode compartments are rinsed with a Na2SO4 solution to avoid the production of
Cl2 at the electrodes. The respective electrode reactions are displayed at the
electrodes. To maintain electroneutrality and constant Na +-concentration, the
electrode solution is circulated through both electrode compartments. This leads to
the neutralization of OH- and H+, generated at opposite electrodes, and results in an
increased electrode voltage loss. Gas consuming electrodes could reduce this loss,
but this was not in the focus of the research presented.
If a sufficiently high electric potential difference is applied between the electrodes of
the stack in Fig. 1, the voltage across the bipolar membrane exceeds the water
spitting value of about 0.8 [V]. Then water molecules are dissociated at the interface
of the bipolar membrane (BM) into H+ and OH- (lower part of Fig. 1). OH- combines
with Na+-ions, which migrate from the NaCl solution across the cation-exchange
membrane (CM) to form NaOH. Simultaneously, H+ combines with Cl--ions, migrating
from the NaCl solution across the anion-exchange membrane (AM), to form HCl. The
overall reaction is the well-established electrochemical production of HCl and NaOH
from a NaCl solution by electrodialysis with bipolar membranes. In the flow battery,
the production of acid and base represents the charging of the battery, with electrode
reactions and direction of electron(e-)-flux shown in the lower part of Fig. 1.
Fig. 1 Schematic drawing, illustrating the operating principle of a neutralization flow
battery based on electrodialysis with bipolar membranes. The diagram shows
the main fluxes of the different species during charge (bottom half) and
discharge (top half) with one repeating cell-unit of the stack, consisting of an
anion exchange membrane (AM), a bipolar membrane (BM), and a cation
exchange membrane (CM). The respective electrode reactions are also
If the voltage across the electrodes is reduced such that the voltage across the
bipolar membrane drops below the water splitting value, the process is reversed
(upper part of Fig. 1). Now H+- and OH—ions migrate from the acid and base solution
into the interface of the bipolar membrane where they are neutralized to water
according to the water dissociation equilibrium. An equivalent number of Na+- and Cl-ions migrates through the corresponding monopolar ion-exchange membranes into
the NaCl solution to preserve electroneutrality. The overall process is the
electrodialytic neutralization of HCl and NaOH with the end-products of NaCl, water
and usable electrical energy. Now electrode reactions and direction of the external
electron flux have changed as indicated in the upper part of Fig. 1.
The driving force of the reverse electrodialysis with bipolar membranes (“REDBP”) is
the neutralization reaction of H+ with OH–, with a Gibbs free energy of -80 [kJ∙mol–1]
at ambient conditions. If this energy could be completely transferred into electric
energy, the gain would be 22.2 [W·h∙mol–1]. Since HCl or NaOH remain liquid in high
concentrations in water (HCl up to 13.75 [mol L–1] and NaCl up to 20 [mol L–1] [7]) the
theoretical storage density in [Wh∙L–1] of liquid solutions of HCl and NaOH would be
much higher than that of vanadium salt solutions with a maximum solubility of about 2
[mol L–1]. Unfortunately, presently available bipolar membranes are not sufficiently
permselective and stable at acid/base concentrations above about 1 [mol L–1] (see
e. g. [14]). An acid and base concentration of 2 [mol L -1] however should lead to
about 20 [Wh] per liter of (acid + base) electrolyte. This is in the same order of
magnitude as reported for current vanadium redox flow batteries [16].
2.2. Water splitting and acid/base neutralization in the bipolar membrane
The key processes of the acid/base flow battery take place in the bipolar membrane
of each repeating cell unit [8, 9]. In the following, a simplified picture of the main
steps will be presented in Fig. 2. Fig. 2a is a sketch of a cation-exchange membrane
(CM) in contact with HCl, completely separated from an anion-exchange membrane
(AM) which is in contact with NaOH. Under these conditions the fixed charges in the
membranes are compensated by counterions, in CM by H+ and in AM by OH–. A
Donnan equilibrium has been established with a (small) negative potential difference
between CM and HCl solution, which hinders the mobile H+ counter-ions to diffuse
into the acid and the co-ions Cl- to enter the membrane pores. A similar but positive
potential gradient at the AM side hinders the OH– counter-ions from leaving and the
Na+co-ions from entering AM. As a result, only a small concentration of co-ions (Clon the CIM side and Na+ on the AM side) is present in the membrane.
The distribution of ionic concentrations cj and of the electrical potential φ is depicted
for an acid or base concentration of 0.5 [mol L–1] and a fixed charge concentration of
the membranes of XF = 1 [mol L-1]. The distribution can be calculated with the
following equations. In the solution as well as in the membrane electroneutrality must
be fulfilled, requiring:
where zj , zF represent the respective positive or negative charge numbers of the ions
j and of the fixed charge XF . The potential difference between the ion exchange
membrane and the outside solution can be determined from the Donnan equilibrium,
which requires that the electrochemical potentials ̃ of each ion j,
are equal between the outside solution and the surface of the ion-exchange
membrane. Neglecting the pressure term
and approximating activities by molar
leads to an approximation of the Donnan potential difference
between solution and ion exchange polymer, which has to be identical for all ions j:
Here prime´ and double prime´´ represent values in the solution and in the ion
exchanger, respectively. From Eq. (3) it follows that
for all ions j, k present, leading to
Eq. (1) and (3) to (5) allow to calculate the electric potential difference Δφ between
outside solution and membrane as well as the membrane concentrations of counterand co-ions, if the concentration of the outside solution is specified.
With T = 25°C = 298 K, gas constant R = 8.314 [J∙mol-1·K-1 ], Faraday constant
F = 96 485 [A·s∙mol-1],
[mol·L-1], the resulting
values in the cation exchange membrane CM are
[mol·L-1] and
[V]. As shown in Fig. 2a, similar values with
opposite sign apply for the anion exchange membrane (AM). The small potential
stabilizes the higher counter-ion concentration of 1.207 [mol·L-1] in the
pores of the ion exchanger over its outside solution concentration of 0.5 [mol·L-1].
Fig. 2 A simplified picture of ionic concentrations cj and electric potential differences
Δφ at zero current in two ion exchange membranes with concentration of
fixed-ions X =1 [mol·L-1]. Left is a cation-exchange membrane (CM) in
contact with HCl and right an anion-exchange membrane (AM) in contact
with NaOH. In a) both membranes are completely separated. In b) and c) CM
and AM are in direct contact, forming a bipolar membrane. In a) and b) the
concentration of outside HCl and NaOH is 0.5 [mol L-1], in c) it has been
raised to 1 [mol L-1].
The simplified case of a bipolar membrane with direct contact between CM and AM is
shown in Fig. 2b. The distribution of ions between the outside solution and the
membrane surface is about the same as in Fig. 2a. But in the interface between both
membranes the direct contact would lead to a rapid reaction of H+ with OH- to form
neutral water. The resulting loss of H+ at the CM side of the interface however
creates a strong positive potential
, which impedes the flow of H+, while a similar
negative potential impedes the flow of OH- at the AM side. Both potentials add to the
total potential in the interface.
The co-ions (Cl- in CM and Na+ in AM) are strongly attracted by this potential
difference across the bipolar membrane. A co-ion flux of Cl- across CM and of Na+
across AM results, with their accumulation in the adjacent membrane. This leads to
profiles of Na+ and Cl- in both membranes as depicted qualitatively in Fig. 2b. Since
the potential gradient is much larger in the interface between CM and AM compared
to the adjacent membranes, the co-ions are rapidly sucked across the interface. Their
concentrations at the interface are therefore very low, and increase again at the
accepting membrane.
The potentials
across CM and AM of the bipolar membrane can be approximated
by Eq. (3) with
[mol·L-1] and
[mol·L–1] (for neutral water)
for CM and
[mol·L–1], and
[mol·L–1] for AM, resulting in
[V] for each membrane or
[V] across the bipolar membrane. If the
outside concentration
of HCl and NaOH is doubled, the resulting potential drop is
only slightly higher, but co-ion concentrations in the membranes and fluxes are about
doubled (Fig. 2c). The presence of co-ions in the interface will reduce the abovementioned potentials, but as will be shown in Fig. 6, the difference between the
theoretical and the measured open cell voltages only amounts to about 6%.
The concentration profiles inside the membranes have to satisfy the electroneutrality
condition, Eq. (1), the Nernst-Planck equation, which specifies the ionic fluxes
across the membrane,
and Faraday´s law:
In Eq. (6), the liquid velocity v of the ionic solution can be considered zero.
represent the ionic mobilities of the different ions in the specific membranes. In
Figs. 2b, c equal ionic mobilities
of all ions have been assumed, leading to the
symmetric concentration profiles shown. In reality the ionic mobilities differ, but the
basic picture is similar [13]. In Faraday´s law i is the (positive or negative) electric
current density, which can be applied over the electrodes (Fig. 1). In Fig. 2 zero
current conditions (
) have been assumed.
At zero current, the individual ionic fluxes ̇ are obviously not zero as already
mentioned above. Instead, according to Eq. (7) with
, the crossover of
- and
-ions through the bipolar membrane has to be compensated by an equivalent flux
- and
-ions. The net result is a neutralization reaction of HCl with NaOH
without generation of electrical power. It leads to a certain self-discharge of the
battery also at zero current, which is proportional to the co-ion concentrations
gradient in CM and AM. This self-discharge reduces the Coulomb efficiency of the
process and has to be considered in addition to the voltage efficiency, which will be
discussed in the following experimental results. In the flow battery concept discussed,
the self-discharge at zero current is limited to the amount of acid and base present in
the respective compartments of the stack and does not influence the acid and base
in their storage tanks.
2.3. Influence of cross-flow of co-ions in the repeating cell unit
The influence of the flow of co-ions across the bipolar membrane on the ionic fluxes
and on the electric potential in a repeating cell unit is shown schematically in Fig. 3.
At zero current (Fig. 3, left) the co-ion fluxes of Na+ and Cl– across the bipolar
membrane are compensated by equivalent fluxes of Na+ through the cation
exchange membrane to the right of the bipolar membrane and by Cl– fluxes through
the cation exchange membrane to the left of the bipolar membrane. Because of
electroneutrality, the co-ion fluxes also require compensating fluxes of H+ and OH–
into the bipolar membrane. The co-ion fluxes are therefore responsible for a selfdischarge with formation of NaCl in the acid and base compartment also at zero
current. The profile of the electric potential, resulting in the difference
, is given in
bold broken lines (qualitative, not to scale).
In order to charge the flow battery, the potential difference Δφ has to be raised above
the zero current value via the external electrodes (Fig. 3, middle). The resulting main
ionic fluxes are given by bold arrows, leading to HCl and NaOH formation in the
respective compartments. But due to the higher potential difference
, the
(additional) ionic fluxes resulting from co-ion break-through also increase (broken
arrows). As a result, the NaCl concentration in the NaCl compartment decreases, but
formation of HCl and NaOH is also combined with some NaCl formation in both
acid/base compartments. Under discharge (Fig. 3, right), the main ionic fluxes are
reversed (solid arrows) and the potentials Δφ across the ion exchange membranes
Fig. 3 Ionic fluxes (arrows) and electric potentials (bold broken lines) across the
repeating cell unit of Fig. 1 under zero current (left), under charge (middle)
and under discharge (right). The main ionic fluxes under charge and discharge
are indicated by bold arrows. The (additional) fluxes due to co-ion leakage
through the bipolar membrane (BM) are given by broken arrows.
In addition to the indicated ionic fluxes, water has to diffuse through both sides of the
bipolar membrane into the BM interface during charge, providing the feed for H+ and
OH–-formation. And the water formed at the interface during discharge has to diffuse
back into the HCl and NaOH compartment. The influence of a restricted water
transport through BM will be discussed in the following sections.
Summarizing, Fig. 1 shows a schematic of the experimental set-up of the repeating
cell unit with the main ionic fluxes and the electrode reactions under charge and
discharge. In Fig. 2b, and Fig. 2c a simplified picture of the electric potential and of
the ionic concentration gradients inside the bipolar membrane is given at zero
current. The resulting ionic fluxes and the resulting potentials across the repeating
cell unit are given in Fig. 3. Here the focus is on the additional influence of the co-ion
flux through the bipolar membrane and its compensation under zero current, charge
and discharge.
Experimental study
The performance of a neutralization flow battery has been experimentally studied
under different charge and discharge conditions with different acid and base
concentrations in a repeating cell unit, using commercially available ion-exchange
membranes and bipolar membranes. Based on the results, the potentials and the
present shortcomings of a neutralization flow battery will be discussed below.
3.1. Test-setup of one repeating cell unit
Fig. 4 shows the simplified flow sheet of the test set-up. In addition to one repeating
cell unit with the three compartments (2, 3, 4), an additional salt compartment (5),
and two electrode compartments (1, 6) have been added. Aqueous solutions of NaCl,
HCl and NaOH as electrolytes are circulated from respective storage tanks. Na 2SO4
has been used as a fourth electrolyte in the electrode compartments to convert ionic
transport into electron transport, while avoiding chlorine formation at the electrodes.
In the following experiments, the NaCl concentration in the salt compartments (2, 5)
is always 0.5 [mol∙L-1] and the Na2SO4 concentration in the electrode compartments
(1, 6) is 0.25 [mol∙L-1]. The acid and base concentrations were kept equal and their
concentrations were changed between 0.25 [mol∙L-1] and 1.0 [mol∙L-1].
Fig. 4 The experimental setup used to study the behavior of a bipolar membrane in
an electrodialysis stack with one repeating cell unit (2, 3 4). The voltage
across the bipolar membrane is measured with two Haber–Luggin capillaries.
The DC power supply allows imposing a positive, a negative or zero current across
the stack, which is measured via the voltage drop over the depicted resistance of 2
[Ω]. The resulting voltage drop over the bipolar membrane can be measured by a
high impedance voltmeter via saturated calomel electrodes and Haber-Luggin
capillaries, placed close (~1 mm apart) to both sides of the membrane. This allows
measuring the voltage across the bipolar membrane as function of the applied
current density with high accuracy.
Similar measurements were repeated for the cation- and anion-exchange membrane
to the left and the right of the bipolar membrane. Fig. 5 shows the set-up for the
measurement of the cation exchange membrane CM to the right of the bipolar
membrane. A comparable set-up has been used for the anion exchange membrane
AM left of the bipolar membrane. The commercially available monopolar and bipolar
membranes, fumasep® FKB, fumasep® FAB and fumasep® FBM, have been
provided by Fumatech GmbH in Germany [14]. Their properties are specified in
Table 1.
Table 1: Properties of the membranes provided by Fumatech [14]
[ ]
Specific area
The cell unit has an active membrane area of 5 × 5 = 25 [cm 2]. Because of the size of
the Luggin capillaries, the respective compartments are 30 [mm] wide while the other
compartments have a width of 5 [mm]. All compartments contain spacers with a void
fraction of 80 [%] for improved flow distribution. The cell temperature is measured by
thin thermocouples, placed at the membrane surface near the voltage measuring
Fig. 5 Experimental set-up for the measurement of the current/voltage behavior of
the cation exchange membrane CM between compartment 4 and 5 in Fig. 4.
Results and discussions
With the experimental setup of Fig. 1 and Fig. 4, positive, negative or zero current
densities i can be applied over the repeating cell unit in order to measure the voltage
drop Δφ over the respective membranes. If a positive current is applied in Fig. 1, the
external voltage is higher than the open cell voltage and the battery is charged. If a
negative current is applied, the external voltage is lower than the open cell voltage
and the battery is discharged. At zero current the open cell voltage (OCV) can be
measured. The focus of the experimental studies will be on the current/voltage
behavior of the bipolar membrane. Its performance has been measured under
different operating conditions and will be discussed in detail. Similar measurements
were carried out with the cation- and anion-exchange membranes at both sides of the
bipolar membrane in order to determine their resistances at different operating
conditions. Since a specific current density will be imposed via the power supply, the
electrode reactions need not to be considered in the following.
Zero current voltage of the bipolar membrane as function of acid and
base concentration and temperature
For zero current voltage measurements, no external electric current is applied, while
the electrolytes are fed with constant flow velocity through the respective
compartments to ensure constant electrolyte concentrations. Fig. 6 shows the
measured voltages across the bipolar membrane for different (always equal) acid and
base concentrations. The dashed line with circular points are the theoretical
maximum open cell voltage, calculated according to Eq. (3). The square points are
experimentally measured data. The difference can be mainly attributed to the
crossflow of co-ions discussed in Section 2.1. The differences between the calculated
and measured open cell voltages increase with the acid/base concentrations from
about 6.2 [%] at acid and base concentration of 0.25 [mol L–1] to ca. 9 [%] at 2 [mol L–
] solutions. This is an obvious consequence of a stronger crossflow of coions at
higher acid/base concentration, as discussed when explaining the differences
between Fig. 2b and Fig. 2c.
Fig. 6 Zero-current voltage of the bipolar membrane vs. acid/base
concentration at 25 [°C] and 0.1 [MPa. The dashed line with circular points
shows the theoretical values according Eq. (3), the measured zero current
voltage is shown by the square points.
In another set of experiments the zero current voltage has been measured as
function of the temperature with 4 different acid and base concentrations, (1.00 [mol
L–1], 0.75 [mol L–1], 0.50 [mol L–1], 0.25 [mol L–1]). The test results are shown in
Fig. 7. According to Eq. (3) the zero current voltage should be a linear function of the
temperature T. The experimental measurements indicate that at high temperatures
and high acid and base concentrations the experimental results differ from this linear
dependency. This can be attributed to the temperature dependency of the ionic
. Since all experiments to be presented have been performed at
temperatures between 24 and 28 °C, the temperature influence can be neglected in
the following.
Fig. 7
Experimentally determined zero current voltages of the bipolar
membrane for 4 different acid and base concentrations as function of
the temperature at 0.1 [MPa].
4.2. Bipolar membrane charge and discharge behavior
Fig. 8 shows the measured voltages and current densities across the bipolar
membrane during charge and discharge at different acid and base concentrations.
Starting with zero current, the discharge current density is decreased stepwise in 1
minute intervals to -50 [mA cm-2], kept constant for 7 minutes and then increased
stepwise back to 0 [mA cm-2]. After 2 minutes pause, charging starts with stepwise
increasing the current density up to 50 [mA cm-2]. After 7 minutes charging at 50
[mA∙cm–2], the current density is decreased stepwise back to zero. All experiments
have been conducted at 25 [°C], 0.1 [MPa].
It is obvious that the zero current voltage increases and the voltage difference
between discharge and charge decreases, if the acid and base concentration is
increased from 0.25 [mol L–1] to 1 [mol L–1]. The voltage increase with acid/ base
concentration is in accord with Eq. (3) and Fig. 6. The decreasing voltage losses with
increasing acid/base concentration at charge and discharge are a consequence of
increasing conductivities of the membranes and electrolytes, resulting in lower
resistive losses.
Fig. 8 Charge and discharge curves of the bipolar membrane for four different HCl
and NaOH concentrations, determined at 25 [°C] and 0.1 [MPa]. The broken
lines refer to the applied current densities and the solid lines to the measured
During the extended period of 7 min with the highest discharge current density of
-50 [mA cm-2], a continuous decrease of the discharge voltage is visible for acid/base
concentrations between 0.25 [mol L–1] and 0.75 [mol L–1]. With 1 mol/l acid and base
concentration however, a self-accelerating breakdown of the voltage can be
observed in Fig. 8d. To study this effect in detail, the results of an extended
discharge and charge period of 20 min are displayed in Fig. 9 for different current
densities and acid and base concentrations. The results show that with an acid/base
concentration of 0.25 [mol L–1] there is almost no voltage drop during the discharge
process at current densities up to 40 [mA cm-2]. At 0.50 [mol L–1] acid and base
concentration a voltage decrease during the discharge period becomes noticeable
already at/above 40 [mA cm-2]. For 0.75 [mol L–1] acid and base however the
continuous voltage decrease changes into a self-accelerating voltage drop, if the
discharge current density is raised above 40 [mA cm-2] (Fig. 9c).
A plausible explanation for the self- accelerating voltage drop above a certain current
density is the excessive accumulation of water in the reaction layer of the bipolar
membrane. At zero current, water formation results from the self-discharge because
of the co-ion flux across the bipolar membrane. It requires a similar flux of H+ and
OH–, which neutralize in the interface, as explained in connection with Fig. 2b. This
flux increases with increasing acid/base concentration, as can be concluded from the
concentration slopes in Fig. 2c. At zero current and at small discharge current
densities under low acid/base concentrations, the generated water can obviously
diffuse completely through the cation- and the anion-exchange membrane into the
adjacent acid and base compartments. This is no longer the case at higher acid/base
concentrations and increasing current densities. An extensive water accumulation
may even lead to a delamination of the bipolar membrane. Improving the water
permeability of the bipolar membrane will therefore be a requirement, if high current
densities at elevated acid/base concentrations should be achieved.
Interestingly, no delamination was observed in the experiments performed. Instead,
the water accumulation seems to be completely reversible if the current density is
lowered again. This can be seen from the rapid recovery of the open cell voltages,
both in Fig. 8d, and in Fig. 9 as soon as the current density is back to zero.
After changing from the open cell voltage with no current to charging with high
current density, the resulting charge voltage passes through a small maximum,
before approaching a constant value, as shown in Fig. 9. This can be attributed to
the increased co-ion accumulation in the bipolar membrane during discharge with
high current density, which temporarily reduces the membrane conductivity. These
co-ions will be replaced by H+ and OH-, produced in the reactive layer during
subsequent charging.
Fig. 9
Voltage versus time curves of the bipolar membrane for 20 minutes
discharge and 20 minutes charge periods for 4 current densities of 20, 40,
60, and 80 [mA cm-2] and three acid and base concentrations of 0.25, 0.5
and 0.75 [mol∙L-1] at 25 [°C] and 0.1 [MPa].
Cycle stability of bipolar membranes
For operation as flow battery, the stability over an extended cycle period and over a
larger number of charge/discharge cycles is of importance. Fig. 10 shows
experimental results for charge/discharge cycles of 42 minutes cycle period with 40
[mA cm-2] at 0.75 [mol L-1] acid and base concentration over 20 cycles (860 minutes).
Each cycle starts with 1 minute open cell voltage, followed by 20 minutes discharge,
1 minute open cell voltage, 20 minutes charge. During the experiments the
temperature increased slightly from 24 to 27 °C. As mentioned in Section 4.1, this
temperature change has only negligible influence on the open cell voltage.
Fig. 10
Bipolar membrane voltage (square symbols) during 20 discharge and
charge cycles with 43 [min] cycle period and 40 [mA cm-2] current
density for 0.75 [mol L-1] acid and base concentration.
It can be seen that several critical voltages remain almost constant over the whole 20
cycles. These include the open cell voltage at 0.775 [V], the voltage at the start of
discharge at 0.66 [V] and the voltage at end of charge at 0.87 [V]. However, with
increasing cycles, the voltage dropped both over the length of the discharge period
and over the length of the charge period. The (desired) drop during charge is due to
the removal of co-ions during the charging process, already mentioned in connection
with Fig. 9. The undesired drop during discharge can be mainly attributed to an
accumulation of water in the BP interface. It is almost constant until about cycle 10
(400 min) after which it increases with constant slope.
After 20 cycles, the test stack was disassembled and several tiny holes were
detected at the anode side of the bipolar membrane. We consider this the reason for
the accelerated drop of the discharge voltage after around the 10th cycle. A probable
cause for the tiny hole formation is friction between the membrane and the rigid
spacers in the compartment during repeated cycling. It should be mentioned that no
such hole formation has been observed in the subsequent stack experiments,
described in [15]. In the above experiments it was necessary to use acid and base
compartments of 30 [mm] width, filled with many spacers, stacked upon another, to
provide sufficient space for the Haber-Luggin capillaries (see Fig. 4). In the later
stack experiments, thin acid and base compartments with only 0.5 mm thickness with
a single spacer have been used.
Summarizing, with a mean bipolar membrane voltage of 0.63 [V] at discharge and
0.87 [V] at charge, a bipolar membrane voltage efficiency of 0.63 [V] / 0.87 [V] = 72
[%] has been reached at [40 mA∙cm-2] charge/discharge current density.
It should be mentioned that this storage voltage efficiency is not equal to the
Coulomb efficiency, since it considers only the voltage differences between charge
and discharge and not the efficiency with which the acid and base reaction produces
electric energy. By definition, the storage voltage efficiency is 100 [%] at open cell
voltage, but the Coulomb efficiency will be lower, because the co-ion flux has to be
compensated by a respective flux of H+ and OH- both at zero current and even more
so during discharge. This has been discussed in Section 2.2 in connection with Fig.
2b and Fig. 2c.
Bipolar membrane behavior under rapid charge and discharge switching
As concluded in Section 4.2, increasing the water permeability of bipolar membranes,
at higher acid and base concentrations and high current density, will become an
important issue for improving the neutralization flow battery with bipolar membranes.
In order to investigate which performance can be expected if the influence of limited
water permeability is eliminated, a series of experiments with rapid switching
between positive and negative current densities (and hence without water
accumulation) has been performed. At each measuring point the respective current
density remained constant for 1 second.
Fig. 11 shows the measured voltages across the bipolar membrane between charge
and discharge for different acid and base concentrations from 0.25 to 1 [mol·L–1]. In
addition to the measured voltages (broken lines), corrected voltages (solid lines),
representing the voltages at the membrane surface, are also shown in Fig. 11. To
correct the measured values, the voltage drop between the tip of the Luggin capillary
and the membrane surface has been subtracted, assuming a distance of 1mm
between capillary tips and membrane on both sides. For this calculation, the
resistances of the different electrolytes of Table 2 have been used. The corrected
values should represent the true voltages across the bipolar membrane. For
increasing acid and base concentration, the difference between the measured and
the corrected curves is getting smaller, due to the decreasing resistance of the
Compared with the previous “steady state” measurements in Figs. 9 and 10 it is
obvious that now much higher current densities even at 1 M acid/base concentration
can be achieved, since the water transport restrictions have been avoided. However,
compared with the steady state measurements in Fig. 10, the measured
discharge/charge voltage efficiency at 40 [mA cm-2] and 0.75 [mol L–1] acid base
concentration of 0.63V/0.88V =72% is almost identical with the results in Fig. 11. This
means that the voltage/current density results obtained under rapid cycling do not
differ from the steady state results, as long as the current density does not exceed
40 [mA cm-2] and the acid/base concentration does not exceed 0.75 [mol L–1]. For
1.0 [mol L–1] acid/base concentrations, the limiting influence of reduced water
permeability already starts at current densities above 20 [mA cm-2], as can be
concluded from Fig. 8. The non-shaded area in Fig. 11 will therefore also represent
the voltage/current dependencies at steady state.
Fig. 11 Measured (broken lines) and corrected voltages (solid lines) across the
bipolar membrane for rapid switching between charge and discharge with
increasing current densities for four different acid and base concentrations.
The horizontal broken line is the open cell voltage calculated with Eq. (3).
The non-shaded areas also describe the steady state measurements of
Figs. 8 to 10.
Table 2: Resistivities of different electrolyte solutions at 20°C [10]
4.5. Calculated voltage/current behavior of the repeating cell unit
In the experimental setup it was necessary to use acid and base compartments of 30
[mm] width to allow positioning of the Haber-Luggin capillaries. As mentioned, in
reality much thinner compartments can be used to limit the internal resistances.
The current/voltage behavior of such a more realistic unit cell can be calculated if the
resistivities of the respective electrolytes as well as the resistances of all membranes
are known. The resistance of the bipolar membrane depends on the acid and base
concentration at both sides of the membrane but it is about equal for charge and
discharge and it is independent of the current density, as is obvious from Fig. 11. The
required resistance of a cation-exchange membrane between a base and a salt
compartment as well as that of an anion-exchange membrane between a salt and an
acid compartment has been studied with a setup, illustrated in Fig. 5. The resulting
area resistances for different acid/base concentrations are given in Table 3. In the
following, a single cell thickness of 1.96 [mm] (three chambers of 0.5 [mm] plus
thickness of the bipolar, the anion-exchange and the cation-exchange membrane) is
Table 3
Area resistances of a cation-exchange membrane (CM) between
different HCl concentrations and 0.5 [mol L–1] NaCl, and of an anion
exchange membrane (AM) between different NaOH concentrations
and 0.5 [mol L–1] NaCl, as determined from single membrane
0.25 [mol L ]
0.5 [mol L ]
2.34 [Ω cm ]
3.64 [Ω cm ]
4.6 [Ω cm ]
6.14 [Ω cm ]
0.75 [mol L ]
1.31 [Ω cm ]
2.21 [Ω cm ]
1.0 [mol L ]
1.06 [Ω cm ]
1.39 [Ω cm ]
In Fig. 12, the results of the calculation for one repeating cell unit are shown. As
explained above, the non-shaded areas also describe the steady state behavior. The
performance of the bipolar membrane is given by the solid lines with square points. It
is slightly better with 0.75 [mol L–1] than with 1 [mol L–1]. This is due to an increasing
co-ion crossover with 1 [mol L–1] solutions, as explained in Section 2.2. Although the
performance of the bipolar membrane is slightly decreasing above 0.75 [mol L–1], the
overall performance still improves due to increasing conductivities of the monopolar
The difference between the solid lines with round points and the thin dashed lines is
the voltage drop due to resistance of the electrolyte solution. It does not change
much with increasing concentration of acid and base, since the main resistance is
due to the poor conductivity of the salt solutions (always 0.5 [mol L–1]). The difference
between dashed lines and solid lines with square points is the voltage drop across
anion- and cation-exchange membrane, which is getting larger with decreasing
solution concentration, since the conductivity of anion or cation exchange membrane
strongly depends on acid or base concentration (Table 3). At low concentrations of
acid and base this is one of the main efficiency losses. The depth of discharge of the
neutralization flow battery with bipolar membrane should therefore be limited to about
0.5 [mol L–1], if a high current density is required.
Fig. 12 Calculated voltage-current curves for a repeating cell unit for acid and base
concentrations between 0.25 [mol L–1] and 1.0 [mol L–1] at 25 [°C] and 0.1
[MPa]. The solid lines with square marks are the corrected voltages across
the bipolar membrane from Fig. 11. The thin dashed lines show the
voltages if the drops across the monopolar membranes are added, and the
solid lines with round marks are the voltages of the whole cell unit, where
the voltage drops in the NaCl-, the HCl- and the NaOH-compartments of 0.5
[mm] width are added. The dark shaded area above 40 [mA cm-2] current
density and above 20 [mA cm-2] at 1.0 [mol L–1] acid/base is only valid for
rapid charge/discharge cycling.
For the evaluation of a flow battery, the power density under discharge and the
voltage efficiency between discharge and charge are important criteria. The power
density PD [mW cm-2] under discharge can be calculated by
Here, ( ) is the voltage difference across the repeating cell unit of thickness
a given current density i. It can be calculated with Eq. (9) from the zero current
voltage OCV if the combined resistance R of all membranes and electrolyte
chambers is known. According to Eq. (9), R can be obtained from the measured
voltage difference between OCV and φ(i) shown in Fig. 12 for each of the acid/base
concentrations. The results are displayed in Table 4. The zero current voltage OCV
required for the calculation also depends on the acid/base concentration. OCV differs
slightly between the different membranes tested in Figs. 6, 11 and 12 (although all
membranes came from the same series). In Table 4 the mean OCVs used in the
calculation are also given.
The voltage efficiency η between discharge and charge can be calculated with
Fig. 13 shows the resulting discharge power densities PD and the discharge/charge
efficiencies η over the current density i in the range of interest.
0.25 [mol L–1]
17.4 [Ω cm2]
OCV 0.705 [V]
Table. 4:
0.5 [mol L–1]
0.75 [mol L–1]
9.8 [Ω cm2]
0.735 [V]
7.3 [Ω cm2]
0.753 [V]
Total resistances R of membranes and electrolyte chambers for steady
state operation and rapid cycling and mean OCV values for acid and
base concentrations between 0.25 and 0.75 [mol L–1].
5. Conclusions
The decisive element of the flow battery concept presented is the bipolar membrane,
where a conversion between electrical and chemical energy takes place. This is the
main difference from most other flow batteries, where the chemical energy is
transformed in the electrode reactions. Compared with the presently dominating
vanadium redox flow battery, potential advantages of the neutralization flow battery
with bipolar membranes are the inexpensive electrolytes with low environmental
impact, as well as the potentially high energy density at high electrolyte
Fig. 13
Calculated discharge power density Pd (left) and discharge/charge
voltage efficiency η (right) at steady state conditions over current density
for acid/base concentrations of 0.25 [mol L–1], 0.5 [mol L–1] and 1 [mol L–
In Fig. 13 the results based on the single cell measurements and valid for steady
state conditions are summarized, showing the presently obtainable power density
and the voltage efficiency of one repeating cell unit. The results show that the power
density as well as the charge/discharge efficiency reached under the presented
experimental conditions is still well below the values reached with vanadium redox
flow batteries. In addition to the higher single cell voltage of 1.4 V, vanadium redox
flow batteries are reported to operate at current densities of about 40 - 50 [mA cm-2]
if voltage efficiencies of > 70% are required [1, 16]. To compete, the REDBP
battery should be operated with higher acid, base and salt concentrations since this
would reduce the ionic resistivities and both boost the power density as well as the
discharge/charge efficiency.
This requires ion exchange membranes with are able to operate at higher electrolyte
concentrations: As explained in Section 2.1, if the acid/base concentration could be
raised to 2 mol per liter, the neutralization of 0.5 L HCl and 0.5 L NaOH would lead to
an electric energy of about 20 Wh, which corresponds to the specific energy of 20
Wh/kg electrolyte, reported for present vanadium redox flow flow batteries [1, 16].
It is important to note that - except for the ability to operate at high electrolyte
concentrations - the requirements for the bipolar membrane and for the monopolar
membranes are quite different. In addition to sufficient water permeability, both sides
of the bipolar membrane should have a high permeability for the small counterions
(H+ and OH-) while the permeability for the larger co- and counterions (Na+ and Cl-)
should be small, in order to limit their break-through and the consequences
discussed with Fig. 2b and c. The monopolar membranes on the other hand should
have a good permeability for the larger counterions (Na + and Cl-) to limit the
respective resistive losses.
The fact that Nafion membranes are used commercially for alkali electrolysis at
concentrations up to 5 mol per liter [17] should indicate the range of achievable
Summarizing, the main goal of our contribution is to point at the somewhat forgotten
flow battery concept of acid/base neutralization with bipolar membranes and to
present experimental evidence that this concept could have potential economic and
environmental advantages, in particular for domestic applications, if some of its
present limitations could be overcome. This requires additional detailed research,
both concerning membrane development and further experimental studies, but also
concerning detailed modelling and simulation. We plan to continue our research in
these directions and would also like to welcome other researchers to join us in this
interdisciplinary effort.
Jiabing Xia gratefully acknowledges the PhD scholarship from the GREES program
at University of Stuttgart in Germany.
List of symbols
activity of ion
concentration of ion
Faraday constant
electric current density
flux of ion
gas constant
mobility of ion
liquid velocity
specific volume of ion
fixed charge concentration of the membrane
charge number of ion
charge number of fixed charge
electric potential
electrochemical potential of ion
1. C. Ponce de Leon, A. Frias-Ferrer, J. Gonzales-Garcia, D. A. Szanto, F. C. Walsh,
Review: Redox flow cells for energy conversion, Jl. of Power Sources 160 (2006),
2. K. Nagasubramanian, F.P. Chlanda, K.-J. Liu, Use of bipolar membranes for
generation of acid and base, an engineering and economic analysis, J. Membrane.
Sci. 2, 109–124 (1977)
3. H. Strathmann, H.-J. Rapp, B. Bauer, C.M. Bell, Theoretical and practical aspects
of preparing bipolar membranes, Desalination. 90, 303–323 (1993)
4. K.N. Mani, Electrodialysis water-splitting technology, J. Membrane Sci. 58, 117138 (1991)
5. J.F. Walther, Process for production of electrical energy from the neutralization of
acid and base in a bipolar membrane cell, US 4311771 (1982)
6. E. Zholkovskij, M. Müller, E. Staude, The storage battery with bipolar membranes,
J. Memb. Sci. 141, 231–243 (1998)
7. D.W. Green, R.H. Perry: Perry's Chemical Engineers' Handbook, 8th Edition.
(McGraw-Hill, 2008)
8. H. Strathmann, J. Krol, H.-J. Rapp, G. Eigenberger, Limiting current density and
water dissociation in bipolar membranes, J. Membrane. Sci. 125, 123–142 (1997)
9. S. Mafé, P. Ramires, Electrochemical characterization of polymer ion-exchange
bipolar membranes, ActaPolym. 48, 234-250 (1997)
10. W. Hamer, H. DeWane, Electrolytic Conductance and the Conductances of the
Hydrohalogen Acids in Water, Natl. Bur. Stand. 33 (1970)
11. R. Simons, Preparation of high performance bipolar membranes J. Membrane
Sci. 78, 13-23 (1993)
12. R. Fu, K.H. Ng, Electrodialysis systems and methods for energy generation and
waste treatment, US 2013/0288142 A1 (2013)
13. Jiabing Xia, Reverse electrodialysis with bipolar membranes (REDBP) as an
energy storage system, PhD thesis, Stuttgart University, Germany, 2018
14. Fumatech homepage,
15. J. Xia, G. Eigenberger, H. Strathmann, U. Nieken, Flow Battery Based on
Reverse Electrodialysis with Bipolar Membranes: Stack Experiments, to be
16. (07/2018)
17. A.T. Davis, J.D. Genders, D. Pletcher, Ion permeable membrane, 35–72,
Alresford Press Ltd, Alresford, Hants, UK
1. Flow battery based on the neutralization of an acid and a base by reverse electrodialysis with
bipolar membranes
2. Fundamental aspects and experimental results in one repeating cell unit
Без категории
Размер файла
2 090 Кб
memsci, 073, 2018
Пожаловаться на содержимое документа