ArticlePDF Available

Abstract and Figures

Capture of CO2, both from fossil origin like coal combustion and from renewable origin like biogas appears to be one of the greatest technological challenges of this century. In this study, we show that Membrane Capacitive DeIonization (MCDI) can be used to capture CO2 as bicarbonate and carbonate ions produced from the reaction of CO2 with water. This novel approach allows capturing CO2 without chemicals usage, at room temperature and atmospheric pressure. In this process, the adsorption and desorption of bicarbonate ions from the deionized water solution drives the CO2(g) absorption/desorption from a gas phase. In this work, the effect of the current density and the CO2 partial pressure were studied. We found that between 55-75% of the electrical charge of the capacitive electrodes can be directly used to absorb CO2 gas. The energy requirement of such a system was found ≈40 kJe mol-1 at 15% CO2 and could be further improved by reducing the ohmic and non-ohmic energy losses of the MCDI cell.
Content may be subject to copyright.
Pleas cite as: Environ. Sci. Technol. 2018, 52, 9478-9485
Solvent-Free CO2 Capture using Membrane Capacitive Deionization
L. Legrand, †‡ O. Schaetzle, R.C.F. de Kler, H.V.M. Hamelers †,*
Wetsus, European Centre of Excellence for Sustainable Water Technology, Oostergoweg 7, 8911 MA
Leeuwarden, The Netherlands.
Department of Environmental Technology, Wageningen University, Bornse Weilanden 9, 6708 WG
Wageningen, The Netherlands.
* Corresponding author: Phone: +31-58-2543000.
The capture of CO2, originating both from fossil fuels, such as coal combustion, and from
renewables, such as biogas, appears to be one of the most significant technological challenges
of this century. In this study, we show that Membrane Capacitive Deionization (MCDI) can be used
to capture CO2 as bicarbonate and carbonate ions produced from the reaction of CO2 with water.
This novel approach allows capturing CO2 at room temperature and atmospheric pressure without
the use of chemicals. In this process, the adsorption and desorption of bicarbonate ions from the
deionized water solution drives the CO2(g) absorption/desorption from a gas phase. In this work,
the effect of the current density and the CO2 partial pressure were studied. We found that between
55-75% of the electrical charge of the capacitive electrodes can be directly used to absorb CO2
gas. The energy requirement of such a system was found 40 kJ mol-1 at 15% CO2 and could be
further improved by reducing the ohmic and non-ohmic energy losses of the MCDI cell.
Keywords: MCDI, CO2 capture, Adsorption, activated carbon
Projections made by the International Energy Agency predict that achieving a net-zero CO2
emission by the mid-century (≈2050) is critical to limit the temperature increase to 2°C[1]. While
renewable energy sources like solar and wind already replaced fossil fuels for power production
in part, the production of fuels and chemicals from renewable sources is slow. It seems that some
form of CO2 capture from different sources such as flue gas (coal, gas or biomass power plants),
biogas or even from ambient air is required to achieve net-zero emissions in 2050. The captured
CO2 can be either stored or utilized as a chemical building block by electrochemical CO2
reduction[24] or thermocatalytic CO2 conversion[5,6].
To capture CO2 from a gas mixture, various concepts have been developed including
adsorption[79], absorption[1012], membrane separation[13,14], cryogenic separation[15],
electrochemical methods[1618] and biochemical methods[19,20]. Among these concepts, amine
scrubbing (chemical absorption) is the most developed and applied technology. This technology
is based on the chemical interaction between CO2 and an amine group, which drives the
absorption process of CO2 into the amine solvent. Despite its wide usage, this process shows
several disadvantages, such as high amount of heat energy needed to regenerate the
solvent[21,22] and solvent degradation[23], which leads to high costs and some other toxic
emissions[24], solvent loss[25] and corrosion effects[26]. Thus, more energy-efficient and
environmentally friendly CO2 capture methods are still of importance to investigate.
Alternative approaches based on electrochemistry attract more and more attention. This
approach uses fewer chemicals and can potentially minimize energy consumption. Moreover,
electrochemical systems, based on electrical power, are more suitable for CO2 removal from
emission points where insufficient waste heat is available for solvent regeneration. Various
concepts have been explored such as Molten carbonate fuel cell[27,28], pH swing with ion-
exchange membranes[16,17], electrochemical generation of nucleophile[29,30] and
supercapacitive swing adsorption[18]. In this study, we propose an alternative concept to capture
CO2 based on Membrane Capacitive Deionization (MCDI)[3133].
MCDI cells are composed of activated carbon electrodes and ion-exchange membranes and are
mainly used to desalinate water. By applying a current through the MCDI cell, ions are removed
from the electrolyte into the pores of the electrodes and are stored in the electrical double layer
(EDL). During this step, energy is temporarily stored in the electrodes due to its capacitive
behavior. By reversing the current, the ions are desorbed from the pores of the electrodes to the
electrolyte, and the energy previously stored is released. An anion-exchange membrane covers
one electrode, and a cation-exchange membrane covers the other electrode. Due to the selectivity
of ion-exchange membranes, MCDI shows higher ions adsorption capacity[34] for monovalent salt
(NaCl, KCl) compared to CDI.
In this study, we show that MCDI technology can be used to capture CO2(g) in the form of HCO3-
and CO32-. These ions are produced by the reaction between CO2(g) and deionized water,
producing ions as described in Eqs 1-3. Here, H2CO3* stands for the combined concentrations of
CO2(aq) and H2CO3. These two species are usually added up as they are difficult to distinguish.
The value for Hcc (defined as aqueous concentration over gaseous concentration) was calculated
from ref[35,36], and the K1 and K2 values were taken from ref[37].
with Hcc=0.83
with K1=10-6.35 M
with K2=10-10.33 M
Depending on the CO2 content in the gas, the reactions between CO2(g) and deionized water lead
to an equilibrium composition. Fig. 1b shows the concentration of CO2(g) and H2CO3* as well as
the pH, based on the Eqs. 1-3 as a function of HCO3- concentration in deionized water. The
concentration of CO32- can be neglected as it is lower than 10-11 mM, due to the low pH of the
electrolyte (pH<5). Applying a current would lead to the adsorption of HCO3- and H+ into the porous
electrodes. Fig. 1a shows that when HCO3- and H+ are adsorbed into the porous electrodes, the
chemical equilibrium is displaced, leading to CO2(g) absorption in deionized water. Upon reversing
the current direction, desorption of HCO3- and CO32- ions from the carbon electrodes takes place,
which drives the chemical equilibrium in the opposite direction toward CO2(g) desorption into the
gas phase. Overall, the CO2(g) absorption or desorption can be controlled via the current direction,
whereas the amount of CO2(g) absorbed or desorbed is determined by the electrode charge, which
is controlled by the current and charging time.
Fig. 1: (a) Scheme of an MCDI cell during charging. AEM stands for anion exchange membrane
and CEM stands for cation exchange membrane. (b) The concentration of CO2(g), H2CO3*and
the pH depending on the HCO3- concentration in solution based on the Eqs. 1-3.
To give a proof-of-concept using MCDI, CO2 gas was alternately absorbed from and desorbed to
a constant gas volume by applying a constant electrical current through an MCDI cell. Different
CO2:N2 gas mixtures at different current densities were tested; 15% CO2(g) as artificial flue gas
from coal-fired power plants, 30% CO2(g) as artificial biogas and 100% CO2(g), to further
understand the effect of CO2 pressure. The amount of energy required to absorb and desorb
CO2(g) from the water phase was calculated at each applied current density and gas mixture used.
This energy loss was split into the ohmic and non-ohmic losses. Directions for further improvement
of the technology are proposed.
Materials and Methods
Experimental set-up
The activated carbon electrodes were prepared following a previously reported
method[38,39]. In short, a slurry containing activated carbon powder (DLC, super 30, Norit,
Amersfoort the Netherlands, BET=1600 m2 g-1), polyvinylidene fluoride (KYNAR HSV 900, Arkema
Inc., Philadelphia) and N-Methyl-2-pyrrolidone (NMP), was cast onto a graphite foil (300 µm) as a
current collector. After evaporation of the NMP for 24 hours at room temperature, a resulting
carbon layer made of 90 wt% of activated carbon and 10 wt% of PVDF was obtained. The final
layer thickness, surface area, and weight of each activated carbon electrode were around 250 µm,
50 cm2 and 1 g. One electrode was covered by a cation-exchange membrane (CMX, Neosepta,
Japan, 50 cm2) and the other electrode was covered by an anion-exchange membrane (AMX,
Neosepta, Japan, 50 cm2). In the MCDI cell, the electrodes were separated by a polymer spacer
(PA 6.6 fabric, Nitex 03-300/51, Sefar, Heiden, Switzerland, 200 µm) to create a flow-channel
between the electrodes.
The electrolyte solution fed to the MCDI cell was prepared by flushing a CO2:N2 gas mixture
through deionized water in a tank. The composition of the gas mixture was controlled by two mass
flow meters (mass view, MV-104, Bronkhorst). Three different CO2 gas mixtures were prepared:
15% CO2, 30% CO2, and 100% CO2.
The MCDI cell was operated in batch mode. A flow of the CO2-flushed deionized water
solution (33 ml ± 5 ml) was continuously re-circulated between the MCDI cell and a gas-liquid
contactor at a flow-rate of 20 mL min-1. The gas-liquid contactor ensures the CO2 exchange
between the solution and a gas volume, secures that the outgoing water is equilibrated with the
gas phase. The gas-liquid contactor was made of a spiral-shaped glass tube, in which the effluent
of the MCDI cell flowed from top to bottom in contact with a volume of gas contained in the
headspace of the spiral-shaped tube. As the solution flows, the exchange of CO2 occurs via the
interphase between the gas and liquid phases all along the tube length. Fig. 2 depicts the research
set-up during operation, and a photo of the research set-up is available in supporting information.
The relative pressure (relative to atmospheric pressure) of the gas in the tube headspace was
monitored with a manometer (Cerabar T PMP131, Endress+Hauser). The pressure measurement
is a direct measurement of the CO2 partial pressure as N2 is considered an inert gas. The gas-
liquid contactor contains 233 ml of CO2:N2 gas mixture and the water residence time inside the
gas-liquid contactor was around 45 seconds. The conductivity of the CO2-dissolved solution was
measured with a conductivity meter (pH/Cond 340i, Mettler Toledo). A Galvanostat (Ivium, the
Netherlands) was used for Galvanostatic Charge and Discharge (GCD) experiments at three
different current densities, 1 mA (0.2 A m-2, 0.5 mA g-1), 2 mA (0.4 A m-2, 1 mA g-1) and 3 mA (0.6
A m-2, 1.5 mA g-1). Multiple consecutive GCD cycles were applied between two different cell
voltage limits, 0 V and 1 V. Simultaneously, an impedance measurement at 5 kHz with an
amplitude of 1 mA was applied to the MCDI cell in order to estimate the ohmic resistance of the
MCDI cell. Triplicates were obtained for each current and each gas mixture. At the 5 kHz
frequency, the major contribution to the resistance is the ohmic resistance, mainly related to the
solution resistivity in the spacer channel, the membranes resistivity, the electrodes ionic resistance
and the External Electronic Resistance (EER)[40]. EER is defined as the electronic resistances
generated in the cables, current collectors and current collector-electrode contacts. A study[41]
showed that the current collector contact resistance can be observed at lower frequencies than 5
kHz and could represent up to 50% of the total internal resistance. Here, we consider the
resistance of the current collector-electrode contacts negligible. By following the procedure
developed in ref[40], we estimate EER to be around 0.6 in the MCDI cell, which is negligible
compared to the ohmic resistance measured (20-50 Ω).
Before the experiment, the MCDI cell was short-circuited to reach 0 V cell voltage, while
CO2-flushed deionized water was constantly fed into the MCDI cell. After this first step, a volume
of the equilibrated CO2-flushed deionized water solution (33 ± 5 mL) was constantly recirculated
between the MCDI cell and the gas-liquid contactor. Simultaneously, the CO2:N2 gas mixture was
continuously flushed through the gas-liquid contactor (0.9 L min-1). This step ensures that the liquid
and the gas phases are equilibrated with the same CO2 gas partial pressure. Afterward, gas valves
were closed to contain a volume of gas into the gas-liquid contactor, and the batch experiment
was started as described above, at 0 V.
Fig. 2: Research set-up scheme. The blue line represents the CO2-flushed deionized
water while the black dashed line represents the CO2(g) volume.
The molar amount of CO2(g) either absorbed or desorbed () during the experiments
was derived from the changes in the gas pressure using the ideal gas law (=(P·V)/(R·T)).
The gas is adsorbed from and desorbed to a fixed gas volume. As the amount of CO2 exchanged
is small compared to the amount of CO2 present in the gas phase (<9%), the effect of the changing
gas pressure can be neglected. The average between nCO2(g) obtained during the charging step
and the discharging step is reported as CO2 exchanged. The difference between 
exchanged during the charge and discharge is in average 2.2% for most of the experiments. This
difference might be caused by minor leakages in the system, especially in the gas phase and not
yet fully equilibrated conditions in the system. For instance, the initial CO2 partial pressure, the
volume of water and the volume gas might differ between the different experiments. The values
of  during charge and discharge are reported in the supplementary information. The molar
amount of charge exchanged during the experiments () was calculated based on the
Faraday number (F) and the electrical charge (Q), as =Q/F, where Q is the electrical charge
expressed in Coulombs.
CDI and MCDI systems are commonly characterized by charge efficiency ()[31,34,42],
which for a 1:1 salt solution, such as NaCl, is defined as the ratio of the desired salt ions removed
from the electrolyte, divided by the total charge transferred between the electrodes. In the case of
CO2-flushed deionized water, there are not only monovalent bicarbonate ions (HCO3-) but also
divalent carbonate ions (CO32-) present. The charge efficiency can for such a situation be
evaluated as given in Eq. 4.
  
Ideally, this metric has value =1, but due to co-ion expulsion and faradaic reactions, is
lower. This parameter has been widely studied in (M)CDI[31,34,42].
However, in this study, we are specifically interested in how much CO2 is adsorbed in the
form of carbonate and bicarbonate, as this drives the CO2(g) absorption. Thus, we define a related
metric, the electrode carbon adsorption efficiency (), which defines the amount of CO2 adsorbed
in the electrodes independently of its ionic form (HCO3- and CO32-).
While is determined by processes such as co-ion expulsion and faradaic reactions, is
also influenced by the distribution of CO32- and HCO3- in the EDL. A shift in the distribution towards
CO32- will lead to a decreased . The distribution of HCO3- and CO32- in the EDL is determined
by the pH in the EDL, as described by Eq 3. Different studies on (M)CDI showed pH effects in the
EDL and electrolyte, due to faradaic reactions[4346] or the presence of amphoteric
groups[45,47]. This might leads to a shift in the carbonate ions distribution in the EDL. The effect
of pH in the electrodes is discussed in more detail later in this study.
describes the efficiency of the CO2 adsorbed in the electrodes, in the form of HCO3- and
CO32-. For the proposed technology, the key parameter of interest is the amount of CO2(g)
removed from the gas phase =/ncharge. However, the source of CO2 adsorbed in the
electrodes can originate either from the gas phase, as CO2(g), or from the liquid phase (H2CO3*,
HCO3- and CO32-). The fraction of CO2 adsorbed in the electrode, which is derived from the liquid
phase, is denoted as fw. Thus, to describe the efficiency of the CO2(g) absorption process, the
CO2 absorption efficiency () is defined as:
The parameter fw is mainly determined by the volume of gas (Vg) and liquid (Vw) in the system
and the CO2 solubility in deionized water at 298 K (Hcc). We estimated fw=0.11 for our system.
Calculations are shown in the supporting information.
 
Overall we can relate and in the following equation:
 
Eq. 8 shows that the absorption efficiency is determined by the fraction of CO2 adsorbed
from the aqueous phase (fw), the distribution of bicarbonate/carbonate in the EDL, and the factors
influencing the charge efficiency (faradaic reactions and co-ion expulsion).
The energy needed for gas separation has two components, the separation work (Wmin)
required to separate CO2 gas to a higher concentration and the work lost as heat (irreversible loss)
during the process due to mainly resistances. In the case of CO2 gas, the separation work can be
calculated based on Eq. 9, in which R represents the ideal gas constant, T the temperature,
the proportion of N2 in the mixed gas and  the proportion of CO2 in the mixed gas
     
Eq. 9 implies that the separation work needed is around 7 kJ molCO2-1 for 15% CO2 at 298K.
This number is small compared to irreversible losses, which is generally true for all separation
processes. In our experimental set-up, the CO2 pressure only slightly changes and thus the energy
used up is actually the irreversible energy losses during the process. The key parameter in this
respect is the specific energy per mole of CO2(g) (
). This energy is defined as
=Wnet/and is calculating from the amount of CO2 absorbed () and the energy
lost during a GCD cycle of the MCDI cell (). The amount of CO2(g) can be further expressed
as =Q·F/. Eq. 10 demonstrates that 
is mainly depending on two key parameters,
which are and .
In MCDI, the energy losses ( is derived as the difference between the energy input
stored during the charge and the energy recovered during the discharge, which can be calculated
by integrating the cell voltage (Ecell) by the electrical charge shown in Eq. 11. The superscript C
stands for charging, and the superscript D stands for discharging.
   
 
In this study, the specific energy losses (
) is differentiated into two categories: (i) the
specific ohmic losses (
) and (ii) the specific non-ohmic losses (
), which is
defined in Eq. 12.
 
 
and 
are calculated based on the ohmic energy losses of the MCDI cell
() and the remaining non-ohmic energy losses of the MCDI cell (). Values for
, , 
are reported in the supplementary information. 
defined as the specific energy losses related to the ohmic resistance () of the MCDI cell
over the electrical charge. The non-ohmic losses are defined as the remaining losses
). The specific non-ohmic losses include different possible effects such
as concentration polarization50 and other phenomena such as faradaic reactions. We can write
Wohmic based on an averaged (constant) internal resistance, which can be written as:
     
The current is defined as I, and Rohmic is the ohmic resistance. We can write the specific
ohmic energy losses in our system as:
is determined by and the internal resistance (Rohmic). The study, therefore, aims at
quantifying the effect of Rohmic and on 
and the specific energy losses due to other
phenomena (
Results and Discussion
Fig. 3a shows the measured gas pressure and the electrodes charge for several GCD cycles for
an experiment at 0.4 A m-2 and 15% CO2. During charging, the pressure decreases with increasing
the electrode charge while during discharging, the pressure increases with decreasing the charge
stored inside the electrodes. Data for different currents densities and CO2 partial pressures
(supporting information) show similar behavior. The gas pressure decrease can be explained as
the absorption of CO2(g) by the water, resulting from the HCO3-/CO32- adsorption in the electrodes.
In contrast, when the electrode charge decreases, the pressure increases due to the H2CO3*
desorption into the gas phase, driven by the HCO3-/CO32- desorption. The amount of CO2 gas
exchanged, electrode charge and obtained for different current densities and different CO2
partial pressures were stable for several GCD cycles as shown for 0.4 A m-2 at 15% CO2 in Fig.
3b. For instance, at 0.4 A m-2 was between 65%-70% for several cycles during the charging
and the discharging steps. This demonstrates for the first time that an MCDI cell can be operated
to absorb CO2 from a gas stream and that its performance is stable.
Surprisingly, Fig. 3c shows that is depending on the current density. With increasing the current
density, the electrical charge decreases (Fig. 3c). The upper charging voltage (1 V) is the same
for all experiments. The ohmic voltage loss (RohmicI) is higher at higher current density, which
reduces the voltage available to charge the MCDI cell. Consequently, the charge stored in the
electrodes is lower at higher current density.  is higher at higher current densities, and reaches
a maximum of 0.76 at 0.6 A m-2. This value corresponds to =0.86 (based on Eq. 6), which
suggests that a minimum of 86% of the electrical charge has been used to adsorb some carbonate
and bicarbonate ions. Although the respective amount of HCO3- and CO32- adsorbed in the
electrode is not known, we can safely assume that 0.86 at 15% CO2 and 0.6 A m-2. This value
of is in good agreement with MCDI work using monovalent solutions, where high values,
around 0.9, are expected[31].
In contrast, lower is obtained at lower current density, ≈0.58 (0.64) for 0.2 A m-2 and
≈0.70 (0.78) for 0.4 A m-2 were not expected. Based on MCDI research with NaCl, it is not
expected that at lower current density will decrease. In the case of NaCl, a lowering of is
attributed to two main phenomena, which are the co-ion expulsion and faradaic reactions. Co-ion
expulsion is negligible in our system as the concentration of the ions is low, below 1 mM, and the
ion-exchange membranes blocked the transport of co-ions from the electrodes to the spacer.
Also, could be affected by faradaic reactions inside the electrodes, as shown in many studies in
MCDI and CDI[4346]. However, most studies[4345] suggested that the faradaic reactions
involved in MCDI are associated with O2 and Cl-, which are both negligible in our electrolyte
solution. Furthermore, charge leakages, determined as the difference between the electrode
accumulated charge of the charging step and the discharging steps, were on average 4.4% in all
our experiments. The charge leakages are reported in Supporting information. In (M)CDI, the
charge leakages are expected to be caused by faradaic reactions. Thus, faradaic reactions are
not expected to play a major role in the lowering of the adsorption process.
The lowering of would thus point to a change in the distribution of HCO3- and CO32- in
the EDL, accompanied by a local high pH in the electrodes. With capacitive electrodes, preferential
adsorption of divalent ions over monovalent has been shown to occur over time[48], which
suggests the preferential adsorption of CO32- rather than HCO3- into the pores of the electrode.
The pH of the electrolyte cannot explain the preferential adsorption of CO32- in the electrode as
the pH of the electrolyte is acidic (pH≈3.9-4.4 between 15% and 100% CO2) and thus CO32- is not
present at this condition. A higher local pH in the anode would favor the conversion of HCO3- into
CO32- (Eq. 3). During our experiments, the local pH in the pores of both electrodes is unknown.
Changes in pH in the electrodes or in the electrolyte in (M)CDI have been mostly related to faradaic
reactions[45,46], consuming H+ in the cathode (high pH) and producing H+ in the anode (low pH).
As described above, this seems unlikely under the condition employed in this experiment (no O2
and Cl- present). This leaves the question open on what causes a possible shift in HCO3-/CO32-
adsorption. A possible explanation might be related to the amphoteric property of HCO3-. Recent
studies in (M)CDI suggest that amphoteric groups could have a significant effect on both the pH
in the EDL and the electrolyte[47]. Amphoteric ions have potentially similar properties than
amphoteric groups and could also influence the pH of the electrodes. Thus, more efforts and
studies are needed to characterize the pH inside the pores of the electrode with amphoteric ions
in order to investigate and with a CO2-flushed deionized water solution electrolyte in MCDI.
Fig. 3d shows that for different CO2 partial pressures at 0.6 A m-2 is stable (≈75-77%).
With higher CO2 partial pressures, the electrode charge increases. This is a result of the fact that
the ohmic losses (I·Rohmic) decrease at higher CO2 pressure. In this case, the ohmic losses
decrease because of the lower Rohmic. An increase of CO2 partial pressure increases the solution
conductivity, resulting in a decrease of the cell ohmic resistance (2.0-2.4 kΩ cm2 for 15% CO2,
1.45-1.6 kΩ cm2 for 30% CO2 and 0.95-1.0 kΩ cm2 for 100% CO2).
Fig. 3: (a) Gas pressure in the gas-liquid contactor and electrode charge at 0.4 A m-2 in
15% CO2(g). Electrode charge, CO2(g) absorbed and the absorption efficiency () (b) for 0.4
A m-2 and 15%, (c) for different current densities at 15% CO2(g) and (d) for different CO2 partial
pressures at 0.6 A m-2.
To estimate the minimum amount of energy needed to capture CO2 using MCDI technology, the
energy losses to separate the CO2(g) from the gas phase was estimated. In MCDI, the amount of
energy required to desalinate salty streams is obtained by subtracting from the energy input,
during the charging step of the GCD, the amount of energy recovered during the discharging step
of the GCD[32]. Following the same procedure, energy to capture CO2(g) was estimated based
on the energy input during the CO2(g) absorption step and the energy recovered during the CO2(g)
desorption step during GCD cycles. The energies input and recovered are shown in supporting
information. Two different types of energy losses were differentiated in this study, which are (i) the
ohmic losses (
) related to the ohmic resistance (Rohmic) and (ii) the non-ohmic losses
), which are not related to ohmic origins, such as concentration polarization[49] and
other non-ideal cell behaviors, such as faradaic reactions.
At 15% CO2, the total energy consumption to capture CO2 was found to be between 39.5 and 49
kJ molCO2-1 (≈0.89-1.13 GJ TCO2-1) for all different current densities tested at 15% CO2 (Fig. 4a).
Fig. 4a shows that at higher current densities, the energy losses are mainly determined by 
At 0.6 A m-2, 
accounts for 70% of the energy losses, whereas at 0.2 A m-2,
for 38% of the energy losses. 
is directly dependent on the Rohmic of the capacitive cell and
the current density applied. The higher the current, the higher is 
. Rohmic is mainly determined
by the low conductivity of the electrolyte separating both electrodes, measured around 1.7 mS m-
1 at 15% CO2. Reducing Rohmic, e.g., via reducing the electrode distance could significantly improve
the energy efficiency and reduce 
. Lower current densities are mainly limited by
. Further investigation studies are required to understand the cause of 
improve the energy efficiency of the MCDI cell. Fig. 4b shows that the energy needed to absorb
CO2 decreases at higher CO2 partial pressures (from 49 kJ molCO2-1 at 15% CO2 to 27 kJ molCO2-
1 at 100% CO2). This is mainly caused by a lower 
due to a decrease of Rohmic at higher CO2
partial pressure.
Fig. 4: Ohmic losses (
) and non-ohmic losses (
) for (a) different current
densities tested at 15% CO2(g) and (b) different CO2 partial pressures tested at 0.6 A m-2.
Meaningful comparison of this technology with conventional methods should be done in realistic
and similar conditions i.e., gas composition, scale, design, and operations mode. This study shows
a proof of principle of a new concept, where data are lacking for such a comparison. Thus, we
compare the energy performance of the MCDI cell with a similar technology, which used similar
conditions, the supercapacitive swing absorption (SAA)[18]. MCDI clearly shows a lower energy
consumption than SAA (41-50 kJ molCO2-1 against 103 kJ molCO2-1)[18]. This difference is mainly
due to the much lower of SAA. We estimated, based on the data shown on SWA[18], =0.2-
0.3 at 2 mA, which is 3 times lower than obtained in MCDI. The difference in seems to point
to a difference in adsorption mechanisms of both technologies. Both SWA and MCDI use
capacitive electrodes to drive the sorption or desorption of CO2(g) in an electrolyte. In SAA,
however, a 1M NaCl electrolyte is used, while in MCDI deionized water is used containing only
dissolved carbonate species. In SAA, the authors claim that the CO2 gas is directly adsorbed as
a neutral species inside the EDL made of Na+ and Cl-. In MCDI, we claim that the CO2(g) is
absorbed into the electrolyte due to the formation of an EDL consisting of carbonate species
(HCO3- and CO32-) as no other anions are present in the deionized water. Thus, SWA is expected
to be dependent on both the CO2 partial pressure, as adsorption process is knowns to be pressure
dependent, and the electrode charge. In MCDI, the CO2 sorption is expected to be only dependent
on the electrical charge, since the CO2 sorption is driven only by the adsorption of ions. The
difference between both sorption principles is well demonstrated at different CO2 partial pressures
experiments. In SWA, the amount of CO2 adsorbed in the electrolyte increases significantly (3.5
times) between 15% and 100% CO2 gas mixtures (based on data in ref[18]), whereas, the amount
of CO2 absorbed was not significantly affected by the CO2 partial pressure in MCDI. The small
increase of CO2 absorption observed in MCDI at higher CO2 partial pressure was mainly caused
by a higher amount of electrical charge stored in the electrodes.
In applications, CO2(g) should be absorbed from a diluted gas mixture, for instance, 15% CO2(g),
when the MCDI is charged, and desorbed into a concentrated stream, ideally 100% CO2(g), when
the MCDI is discharged. A membrane contactor could be used to separate the gas from the liquid
phase, respectively after the charging and the discharging steps. Experiments at different CO2
gas mixtures show that the system can be operated with different gas mixtures. In this study, the
CO2 capture rate was still low (6·10-9 mol s-1 gcarbon-1), which is 50 times lower than the state of the
art of CO2 capture based on adsorption materials like zeolite[50]. The lower capture rate is mainly
caused by the low current densities used and the non-optimal . In our experimental set-up, the
current density is mainly limited by Rohmic, as increasing the current density increases 
Reducing the internal resistance is thus not only of paramount interest for the energetic
performance but would be even more for improving the capture rate.
In conclusion, we proposed a new concept to capture CO2 using capacitive electrodes and
deionized water. This system could potentially capture CO2 at lower energy consumption and
without any addition of chemical solvent or heat. In this study, we showed the first proof of principle
of this concept and reported energy consumption comparable to or lower than similar technology,
such as swing supercapacitive adsorption. This energy consumption can be potentially lowered
by reducing ohmic losses and non-ohmic losses. This could be based on well-known strategies
from the field of MCDI or new ones based on insight to be developed on the amphoteric behavior
of bicarbonate in the double layer. Moreover, a continuous process to capture CO2 including an
MCDI should be developed and tested to compare this new concept to other electrochemical
Author Information
Corresponding Author
*E-mail: Phone: +31-58-2843000.
Associated Content
The Supporting Information is available free of charge on the ACS Publications website at DOI:
Calculations on the External Electronic Resistance, the absorption efficiency, electrode carbon
adsorption efficiency, the net energy loss, the ohmic losses and the non-ohmic losses, seven
figures and two tables.
This work was performed in the TTIW-cooperation framework of Wetsus, Centre of Excellence
for Sustainable Water Technology. Wetsus is funded by the Dutch Ministry of Economic Affairs,
the European Union Regional Development Fund, the Province of Friesland, the City of
Leeuwarden, and the EZ/Kompas program of the ‘‘Samenwerkingsverband Noord-Nederland’’.
The authors would like to thank the participants of the research theme ‘‘CO2 Energy’’ at Wetsus,
Alliander and Engie, for the support and fruitful discussions.
[1] I.E.A. International, E. Agency, Energy Technology Perspectives 2016, (2017).
[2] C.E. Tornow, M.R. Thorson, S. Ma, A.A. Gewirth, P.J.A. Kenis, Nitrogen-Based Catalysts
for the Electrochemical Reduction of CO2 to CO, J. Am. Chem. Soc. (2012) 19520–19523.
[3] R. Kas, R. Kortlever, H. Yilmaz, M.T.M. Koper, G. Mul, Manipulating the Hydrocarbon
Selectivity of Copper Nanoparticles in CO2 Electroreduction by Process Conditions,
ChemElectroChem. 2 (2015) 354–358. doi:10.1002/celc.201402373.
[4] S. Ma, M. Sadakiyo, R. Luo, M. Heima, M. Yamauchi, P.J.A. Kenis, One-step
electrosynthesis of ethylene and ethanol from CO2 in an alkaline electrolyzer, J. Power
Sources. 301 (2016) 219–228. doi:10.1016/j.jpowsour.2015.09.124.
[5] A. Bansode, A. Urakawa, Towards full one-pass conversion of carbon dioxide to methanol
and methanol-derived products, J. Catal. 309 (2014) 66–70.
[6] S. Zander, E.L. Kunkes, M.E. Schuster, J. Schumann, G. Weinberg, D. Teschner, N.
Jacobsen, R. Schlögl, M. Behrens, The role of the oxide component in the development of
copper composite catalysts for methanol synthesis, Angew. Chemie - Int. Ed. 52 (2013)
6536–6540. doi:10.1002/anie.201301419.
[7] Z. Liu, L. Wang, X. Kong, P. Li, J. Yu, A.E. Rodrigues, Onsite CO 2 capture from flue gas
by an adsorption process in a coal-fired power plant, Ind. Eng. Chem. Res. 51 (2012) 7355–
7363. doi:10.1021/ie3005308.
[8] M.T. Ho, G.W. Allinson, D.E. Wiley, Reducing the cost of CO2 capture from flue gases using
pressure swing adsorption, Ind. Eng. Chem. Res. 47 (2008) 4883–4890.
[9] C. Lu, H. Bai, B. Wu, F. Su, J.F. Hwang, Comparative study of CO2 capture by carbon
nanotubes, activated carbons, and zeolites, Energy and Fuels. 22 (2008) 3050–3056.
[10] W. Conway, X. Wang, D. Fernandes, R. Burns, G. Puxty, M. Maeder, Comprehensive
Kinetic and Thermodynamic Study of the Reactions of CO 2 ( aq ) and HCO 3 À with
Monoethanolamine ( MEA ) in Aqueous Solution, 2 (2011) 14340–14349.
[11] R. Dugas, G. Rochelle, Absorption and desorption rates of carbon dioxide with
monoethanolamine and piperazine, Energy Procedia. 1 (2009) 1163–1169.
[12] G.T. Rochelle, Amine Scrubbing for CO2 Capture, Science (80-. ). 325 (2009) 1652–1654.
[13] K. Ramasubramanian, H. Verweij, W.S. Winston Ho, Membrane processes for carbon
capture from coal-fired power plant flue gas: A modeling and cost study, J. Memb. Sci. 421–
422 (2012) 299–310. doi:10.1016/j.memsci.2012.07.029.
[14] M. Pera-Titus, Porous inorganic membranes for CO2 capture: Present and prospects,
Chem. Rev. 114 (2014) 1413–1492. doi:10.1021/cr400237k.
[15] M.J. Tuinier, M.V.S. Annaland, G.J. Kramer, J.A.M. Kuipers, Cryogenic CO 2 capture using
dynamically operated packed beds, Chem. Eng. Sci. 65 (2010) 114–119.
[16] S. Datta, M.P. Henry, Y.J. Lin, A.T. Fracaro, C.S. Millard, S.W. Snyder, R.L. Stiles, J. Shah,
J. Yuan, L. Wesoloski, R.W. Dorner, W.M. Carlson, Electrochemical CO 2 Capture Using
Resin-Wafer Electrodeionization, (2013).
[17] M.D. Eisaman, L. Alvarado, D. Larner, P. Wang, B. Garg, K.A. Littau, Environmental
Science, 4 (2011). doi:10.1039/c0ee00303d.
[18] B. Kokoszka, N.K. Jarrah, C. Liu, D.T. Moore, K. Landskron, Supercapacitive swing
adsorption of carbon dioxide, Angew. Chemie - Int. Ed. 53 (2014) 3698–3701.
[19] R. Yadav, S. Wanjari, C. Prabhu, V. Kumar, N. Labhsetwar, T. Satyanarayanan, S. Kotwal,
S. Rayalu, Immobilized carbonic anhydrase for the biomimetic carbonation reaction, Energy
and Fuels. 24 (2010) 6198–6207. doi:10.1021/ef100750y.
[20] E. Ozdemir, Biomimetic CO2 sequestration: 1. Immobilization of carbonic anhydrase within
polyurethane foam, Energy and Fuels. 23 (2009) 5725–5730. doi:10.1021/ef9005725.
[21] H. Lepaumier, E.F. Da Silva, A. Einbu, A. Grimstvedt, J.N. Knudsen, K. Zahlsen, H.F.
Svendsen, Comparison of MEA degradation in pilot-scale with lab-scale experiments,
Energy Procedia. 4 (2011) 1652–1659. doi:10.1016/j.egypro.2011.02.037.
[22] N. Dai, A.D. Shah, L. Hu, M.J. Plewa, B. McKague, W.A. Mitch, Measurement of nitrosamine
and nitramine formation from NOx reactions with amines during amine-based carbon
dioxide capture for postcombustion carbon sequestration, Environ. Sci. Technol. 46 (2012)
9793–9801. doi:10.1021/es301867b.
[23] E.F. Da Silva, H. Lepaumier, A. Grimstvedt, S.J. Vevelstad, A. Einbu, K. Vernstad, H.F.
Svendsen, K. Zahlsen, Understanding 2-ethanolamine degradation in postcombustion CO
2 capture, Ind. Eng. Chem. Res. 51 (2012) 13329–13338. doi:10.1021/ie300718a.
[24] E.D. Wagner, K.M. Hsu, A. Lagunas, W.A. Mitch, M.J. Plewa, Comparative genotoxicity of
nitrosamine drinking water disinfection byproducts in Salmonella and mammalian cells,
Mutat. Res. - Genet. Toxicol. Environ. Mutagen. 741 (2012) 109–115.
[25] H. Lepaumier, D. Picq, P.L. Carrette, New amines for CO2 Capture. II. oxidative degradation
mechanisms, Ind. Eng. Chem. Res. 48 (2009) 9068–9075. doi:10.1021/ie9004749.
[26] S. Martin, H. Lepaumier, D. Picq, J. Kittel, T. De Bruin, A. Faraj, P.L. Carrette, New amines
for CO 2 capture. IV. Degradation, corrosion, and quantitative structure property relationship
model, Ind. Eng. Chem. Res. 51 (2012) 6283–6289. doi:10.1021/ie2029877.
[27] HUEBSCHER RG, BABINSKY AD, Electrochemical Concentration and Separation of
Carbon Dioxide for Advanced Life Support Systems. Carbonation Cell System, SAE-Paper
690640. (1969) 2164–2170.
[28] K. Li, N. Li, Removal of Carbon Dioxide from Breathing Gas Mixtures Using an
Electrochemical Membrane Cell, Sep. Sci. Technol. 28 (1993) 1085–1090.
[29] D.H. Apaydin, E.D. Głowacki, E. Portenkirchner, N.S. Sariciftci, Direct electrochemical
capture and release of carbon dioxide using an industrial organic pigment: Quinacridone,
Angew. Chemie - Int. Ed. 53 (2014) 6819–6822. doi:10.1002/anie.201403618.
[30] M.B. Mizen, M.S. Wrighton, Reductive Addition of C02 to 9 ,10-Phenanthrenequinone, J.
Electrochem. Soc. 136 (1989) 941–946.
[31] R. Zhao, P.M. Biesheuvel, A. Van Der Wal, Energy consumption and constant current
operation in membrane capacitive deionization, Energy Environ. Sci. 5 (2012) 9520–9527.
[32] P. Długołecki, A. Van Der Wal, Energy recovery in membrane capacitive deionization,
Environ. Sci. Technol. 47 (2013) 4904–4910. doi:10.1021/es3053202.
[33] S. Porada, R. Zhao, A. Van Der Wal, V. Presser, P.M. Biesheuvel, Review on the science
and technology of water desalination by capacitive deionization, Prog. Mater. Sci. 58 (2013)
1388–1442. doi:10.1016/j.pmatsci.2013.03.005.
[34] P.M. Biesheuvel, R. Zhao, S. Porada, A. van der Wal, Theory of membrane capacitive
deionization including the effect of the electrode pore space, J. Colloid Interface Sci. 360
(2011) 239–248. doi:10.1016/j.jcis.2011.04.049.
[35] E. Wilhelm, R. Battino, R.J. Wilcock, Low-Pressure Solubility of Gases in Liquid Water,
Chem. Rev. 77 (1977) 219–262. doi:10.1021/cr60306a003.
[36] R. Sander, Compilation of Henry’s law constants (version 4.0) for water as solvent, Atmos.
Chem. Phys. 15 (2015) 4399–4981. doi:10.5194/acp-15-4399-2015.
[37] X. Wang, W. Conway, R. Burns, N. McCann, M. Maeder, Comprehensive study of the
hydration and dehydration reactions of carbon dioxide in aqueous solution, J. Phys. Chem.
A. 114 (2010) 1734–1740. doi:10.1021/jp909019u.
[38] F. Liu, O. Schaetzle, B.B. Sales, M. Saakes, C.J.N. Buisman, H.V.M. Hamelers, Effect of
additional charging and current density on the performance of Capacitive energy extraction
based on Donnan Potential, Energy Environ. Sci. 5 (2012) 8642–8650.
[39] H.V.M. Hamelers, O. Schaetzle, J.M. Paz-García, P.M. Biesheuvel, C.J.N. Buisman,
Harvesting Energy from CO 2 Emissions, Environ. Sci. Technol. Lett. 1 (2014) 31–35.
[40] J.E. Dykstra, R. Zhao, P.M. Biesheuvel, A. Van der Wal, Resistance identification and
rational process design in Capacitive Deionization, Water Res. 88 (2016) 358–370.
[41] Y. Qu, T.F. Baumann, J.G. Santiago, M. Stadermann, Characterization of internal resistance
of a capacitive deionization system, Environ. Sci. Technol. (2015) 1–15.
[42] R. Zhao, P.M. Biesheuvel, H. Miedema, H. Bruning, A. van der Wal, Charge efficiency: A
functional tool to probe the double-layer structure inside of porous electrodes and
application in the modeling of capacitive deionization, J. Phys. Chem. Lett. 1 (2010) 205
210. doi:10.1021/jz900154h.
[43] D. He, C.E. Wong, W. Tang, P. Kovalsky, T. David Waite, Faradaic Reactions in Water
Desalination by Batch-Mode Capacitive Deionization, Environ. Sci. Technol. Lett. 3 (2016)
222–226. doi:10.1021/acs.estlett.6b00124.
[44] W. Tang, D. He, C. Zhang, P. Kovalsky, T.D. Waite, Comparison of Faradaic reactions in
capacitive deionization (CDI) and membrane capacitive deionization (MCDI) water
treatment processes, Water Res. 120 (2017) 229–237. doi:10.1016/j.watres.2017.05.009.
[45] J.E. Dykstra, K.J. Keesman, P.M. Biesheuvel, A. van der Wal, Theory of pH changes in
water desalination by capacitive deionization, Water Res. 119 (2017) 178–186.
[46] P. Nativ, Y. Badash, Y. Gendel, New insights into the mechanism of flow-electrode capacitive
deionization, Electrochem. Commun. 76 (2017) 24–28. doi:10.1016/j.elecom.2017.01.008.
[47] A. Hemmatifar, D.I. Oyarzun, J.W. Palko, S.A. Hawks, M. Stadermann, J.G. Santiago,
Equilibria model for pH variations and ion adsorption in capacitive deionization electrodes,
Water Res. 122 (2017) 387–397. doi:10.1016/j.watres.2017.05.036.
[48] R. Zhao, M. Van Soestbergen, H.H.M. Rijnaarts, A. Van Der Wal, M.Z. Bazant, P.M.
Biesheuvel, Journal of Colloid and Interface Science Time-dependent ion selectivity in
capacitive charging of porous electrodes, J. Colloid Interface Sci. 384 (2012) 38–44.
[49] D.A. Vermaas, M. Saakes, K. Nijmeijer, Doubled power density from salinity gradients at
reduced intermembrane distance, Environ. Sci. Technol. 45 (2011) 7089–7095.
[50] J. Merel, M. Clausse, F. Meunier, Experimental investigation on CO2 post-combustion
capture by indirect thermal swing adsorption using 13X and 5A zeolites, Ind. Eng. Chem.
Res. 47 (2008) 209–215. doi:10.1021/ie071012x.
Supporting information
Solvent Free CO2 Capture using Membrane Capacitive Deionization
L. Legrand, †‡ O. Schaetzle, R.C.F. de Kler, H V.M. Hamelers †,*
Wetsus, centre of excellence for sustainable water technology, Oostergoweg 7, 8911 MA Leeuwarden, The
Department of Environmental Technology, Wageningen University, Bornse Weilanden 9, 6708 WG
Wageningen, The Netherlands
* Corresponding author at Phone: +31-58-2543000.
1. Estimation of the External Electronic Resistance (EER)
The resistance in an (M)CDI can be divided into the electronic resistance and the ionic resistance.
The electronic resistance was previously described as the External Electronic Resistance (EER)
and was defined as the resistance generated by the cables wires connecting the galvanostat to
the MCDI cell, the resistance of the current collector and the different contact resistances (cables-
current collector and current collector-electrodes). In contrary, the ionic resistance is related to the
ionic conductivity of the ions presents in the electrodes, the membrane and the spacer
Recently, Dykstra et al. proposed a simple method to estimate the EER50 by measuring the internal
resistance of the MCDI cell at different NaCl ionic conductivity solutions. By subtracting the
contribution of the ionic resistivity of the solution on the total internal resistance measured, it is
then possible to isolate the EER. We performed this test in our cell with different concentrations
of NaCl (from 1 M NaCl to 1 mM NaCl) and CO2 dissolved deionized water at different CO2 partial
pressures. Figure S1 shows that EER was found around 0.6 Ω. This value is between 40 and 75
times lower than the resistance measured in all experiments. Since the resistance is dominated
by the low ionic conductivity of CO2-flushed deionized water, we assume EER negligible.
Figure S1: Internal resistance measurement at different salt concentrations (NaCl) and CO2-flushed
deionized water at different CO2 partial pressure. The intersection with the y-axis represents the external
electronic resistance in our set-up.
2. Relationship between the absorption efficiency and the carbon adsorption efficiency.
CDI and MCDI systems are commonly characterized by charge efficiency () for monovalent salt,
which defined the ratio of salt ions removed from the electrolyte, divided by the total electrical
charge transferred. In case of CO2-flushed deionized water, monovalent and divalent ions are in
the electrolyte. Thus, the charge efficiency is defined as:
  
In this study, we are interested in the removal of the total carbon (nT), contained into two different
phases: a liquid and gas phase. Thus, next to , we define two related metrics named electrode
carbon adsorption efficiency () and absorption efficiency (). The different metrics are
illustrated in Figure S2.
 is defined as the ratio of total carbon (H2CO3* and CO2(g)) exchanged from the system, caused
by the adsorption of HCO3-/CO32- by the molar electrode charge.
Since H+ is the only cation in the bulk solution, [H+]>>[OH-], hence the acidic pH in solution. Given
the high pKa of pK2 (10.33 M) and the low pH of the bulk solution, we can assume that CO32- is
negligible. Moreover, pKa of pK1 and pK2 suggest that H2CO3* is the dominant carbon molecule
in solution at low pH. Hence, [H2CO3*]>>[HCO3-]>>[CO32-]. Thus, we assume the amount of carbon
in deionized water only depending on [H2CO3*].
with pK1=6.35 M
with pK2=10.33 M
Thus, can be defined in eq 5.
 
is defined as the ratio between the molar amount of CO2 from the gas phase exchanged divided
by the molar electrical charge.
Figure S2: illustration of the different metrics used in the system, which are , and .
For low dilute solutions, the physical equilibrium for the dissolution of CO2(g) into deionized water
is given by the CO2 solubility constant:
  
where HCC represents the solubility constant for CO2 in water at 298 K.
Vw stands for the water volume in the system. From eqs 5 and 8, we can derive the absorption
efficiency according to the carbon adsorption efficiency (eq 9).
By integrating eqs 6 and 10, we obtain:
   with 
By substituting VW=0.033 L, Vg=0.233 L, Hcc=0.83, we obtained:
 
Figure S3: Photo of the research set-up during operation. The peristaltic pump is located
outside the temperature controlled chamber.
Figure S4: Cell potential during Galvanostatic charge-discharge at different current densities
with (a-c) 15% CO2, (d) 30% CO2 and (e) 100% CO2-flushed deionized water. Three replicates
are shown on each graph.
Figure S5: Gas-liquid contactor pressure for (a-c) 15% CO2 (d) 30% CO2 and (e) 100% CO2 for
different current densities.
Figure S6: Gas-liquid contactor pressures and electrodes charge at (a,b) 15% CO2, (c) 30%
CO2 and (d) 100% CO2 for different current densities.
Figure S7: Specific energy losses obtained during charge and discharge (a) for different current
densities at 15% CO2 and (b) for different CO2 partial pressures at 0.6 A m-2.
Table S1: Electrode charge, CO2(g) absorbed, , , Wnet, 
, 
for different current densities and different CO2 partial
Electrode charge (Coulomb)
CO2(g) absorbed (μmol)
0.2 A m-2 15% CO2
20.1 (±0.3)
18.8 (±0.1)
1.26 (±0.4)
0.59 (±0.03)
0.55 (±0.04)
0.61 (±0.05)
0.4 A m-2 15% CO2
15.2 (±0.4)
14.6 (±0.2)
0.6 (±0.5)
103 (±3.9)
0.72 (±0.02)
0.67 (±0.03)
0.75 (±0.04)
0.6 A m-2 15% CO2
12.3 (±0.3)
11.9 (±0.2)
0.4 (±0.1)
0.77 (±0.02)
0.75 (±0.01)
0.84 (±0.02)
0.6 A m-2 30% CO2
16.2 (±0.04)
15.5 (±0.1)
0.7 (±0.1)
129 (±1.6)
124 (±0.09)
0.80 (±0.01)
0.74 (±0.01)
0.83 (±0.01)
0.6 A m-2100% CO2
19.8 (±0.2)
19.2 (±0.2)
0.7 (±0.1)
178 (±5.1)
144 (±2.5)
0.89 (±0.02)
0.70 (±0.02)
0.78 (±0.02)
Energy (J cycle-1)
Energy (kJ molCO2-1)
Energy (kJ molCO2-1)
0.2 A m-2 15% CO2
13.0 (±0.3)
4.5 (±0.3)
113 (±0.2)
74.0 (±2.1)
39.42 (±2.1)
15.3 (±0.2)
24.1 (±2.0)
0.4 A m-2 15% CO2
10.6 (±0.3)
4.5 (±0.3)
98.4 (±1.7)
56.7 (±1.4)
41.7 (±1.9)
25.2 (±0.3)
16.5 (±1.7)
0.6 A m-2 15% CO2
4.4 (±0.03)
4.7 (±0.1)
95.5 (±0.7)
46.5 (±0.6)
49.0 (±1.0)
33.8 (±0.8)
15.18 (±0.4)
0.6 A m-2 30% CO2
11.3 (±0.1)
4.9 (±0.2)
89.1 (±1.4)
49.9 (±0.4)
39.2 (±1.7)
23.2 (±0.2)
16.0 (±1.4)
0.6 A m-2 100% CO2
12.7 (±0.1)
4.3 (±0.1)
78.8 (±2.0)
52.02 (±0.8)
14.5 (±0.4)
12.3 (±0.9)
Dykstra, J.E.; Zhao, R.; Biesheuvel, P.M. ; van der Wal, A. Resistance identification and rational
process desing in Capacitive Deionization, Water Res. 2016, 88, 358-370.
... Quinones and amines tend to be oxygen-sensitive. Electrochemical methods based on pHswings, e.g., membrane-capacitive deionization (MCDI), 22 electrolysis, 23 and bipolar membrane electrodialysis (BPMED), 24 have also been reported. However, MCDI require expensive materials such as cation and anionexchange membranes. ...
... Co-ion expulsion, competition of H + /HCO3 -with electrolyte ions in the double layer, and unwanted faradaic reactions (e.g., carbon oxidation, oxygen reduction) may hinder the CE to reach achieve its maximum value. 22,[43][44][45][46] In addition, the conversion of HCO3 -ions to form CO3 2-that have a greater affinity to the anode due to their higher negative charge may reduce the charge efficiency. CE ranging from 0.5 to 0.7 has been achieved by using ion-selective membrane, thinner electrodes (0.25 mm vs 0.70 mm used in this study), and pure DI water instead of salt solutions in a related membrane-capacitive deionization process (MCDI). ...
... CE ranging from 0.5 to 0.7 has been achieved by using ion-selective membrane, thinner electrodes (0.25 mm vs 0.70 mm used in this study), and pure DI water instead of salt solutions in a related membrane-capacitive deionization process (MCDI). 22 However, the added cost of membranes along with their mass-transfer limitations, mechanically weak electrodes with less absolute capacitance, and low ionic conductivity of the CO2 sparged DI water are disadvantageous for capturing CO2, especially at low concentrations. Separate studies beyond the scope of this work are needed to gain insights about the adsorbed species at the double-layer, and possible faradaic surface reactions. ...
Full-text available
Global warming due to anthropogenic CO2 emissions demands the rapid development of efficient carbon capture technologies. Supercapacitive swing adsorption (SSA) is a technology that relies on the reversible charge and discharge of supercapacitor electrodes to selectively adsorb and desorb CO2. A current limitation of SSA is the low sorption capacity. Here, we investigate a series of activated carbons derived from biomass, coke, coal, and carbide as electrode materials for SSA. The energetic and adsorptive performance metrics are quantitatively analyzed and their relationship with double layer capacitance, real capacitance, imaginary capacitance, and diffusion resistance is studied. The results show that there is a strong positive correlation between specific capacitance and CO2 adsorption capacity. The highest gravimetric sorption capacity was measured for garlic-root derived activated carbons valuing 273 for 4 cm2 electrodes having a specific capacitance of 257 F.g-1. This is a ~4-fold increase compared to BPL 4x6 carbon having a specific capacitance of 86 F.g-1 and an adsorption capacity of 70 mmol.Kg-1. Volumetric specific capacitances and volumetric sorption capacities are also correlated. The highest volumetric sorption capacity was found for garlic powder-derived activated carbon valuing 65.3 mol.m3 at a volumetric specific capacitance of 62.3 F.cm3 which is a two-fold increase compared to BPL 4x6 carbon which has a volumetric specific capacitance of 41.2 F/cm3 and a volumetric sorption capacity of 27.1 mol.m3. In addition, higher specific capacitance tends to improve the overall adsorption rate and productivity.
The build-up of carbon dioxide in the atmosphere is one of the grand challenges facing society. Addressing this challenge by removing CO2 from the atmosphere or mitigating point source emissions through the separation and concentration of CO2 from these dilute sources requires reductions in energetic and monetary cost relative to traditional thermal and pressure swing methods. Electrochemical methods of CO2 separation have drawn increasing attention in recent years as potentially cheap, low-energy, scalable carbon capture technologies. In this Primer, we provide an overview of the experimentation and analysis needed for the study of electrochemical methods for CO2 separation, including a discussion of the considerations necessary for targeting the application of such techniques. This Primer focuses on ambient temperature techniques such as pH swing and direct redox processes, which utilize similar experimental set-ups. We include considerations on the choice of redox agent and an outlook on this growing body of research. Experimentation to address real-world conditions, particularly at practical oxygen concentrations, and novel system designs that overcome transport limitations or, potentially, couple capture and CO2 utilization are emerging areas in the field. Electrochemical methods of CO2 separation offer potentially cheap, low-energy, scalable carbon capture technologies. In this Primer, Diederichsen et al. provide an overview of the experimentation and analysis needed for the study of electrochemical methods for CO2 separation.
Supercapacitive swing adsorption (SSA) is a recently discovered electrochemically driven CO2 capture technology that promises significant efficiency improvements over traditional methods. A limitation of this approach is the relatively low CO2 adsorption capacity, and the underlying molecular mechanisms of SSA remain poorly understood, hindering optimization. Here we present a new device architecture for simultaneous electrochemical and gas-adsorption measurements, and use it to investigate the effects of charging protocols on SSA performance. We show that altering the voltage applied to charge the SSA device can significantly improve performance. Charging the gas-exposed electrode positively rather than negatively increases CO2 adsorption capacity and causes CO2 desorption rather than adsorption with charging. We also show that switching the voltage between positive and negative values further increases CO2 capacity. Previously proposed mechanisms of the SSA effect fail to explain these phenomena, so we present a new mechanism based on movement of CO2-derived species into and out of electrode micropores. Overall, this work advances our knowledge of electrochemical CO2 adsorption by supercapacitors, potentially leading to devices with increased uptake capacity and efficiency.
Membrane capacitive deionization (MCDI) is a promising technique to achieve desalination of low-salinity water resources. The primary requirements for developing and designing materials for MCDI applications are large surface area, high wettability to water, high conductivity, and efficient ion-transport pathways. Herein, we synthesized ionic covalent organic nanosheets (iCONs) containing guanidinium units that carry a positive charge. A series of quaternized polybenzimidazole (QPBI)/iCON ([email protected]) nanocomposite membranes was fabricated using solution casting. The surface, thermal, wettability, and electrochemical properties of the [email protected] nanocomposite membranes were evaluated. The [email protected] anion-exchange membranes achieved a salt adsorption capacity as high as 15.6 mg g⁻¹ and charge efficiency of up to 90%, which are 50% and 20% higher than those of the pristine QPBI membrane, respectively. The performance improvement was attributed to the increased ion-exchange capacity (2.4 mmol g⁻¹), reduced area resistance (5.4 Ω cm²), and enhanced hydrophilicity (water uptake = 32%) of the [email protected] nanocomposite membranes. This was due to the additional quaternary ammonium groups and conductive ion transport networks donated by the iCON materials. The excellent desalination performance of the [email protected] nanocomposite membranes demonstrated their potential for use in MCDI applications and alternative electromembrane processes.
Electrochemical membrane technology for environmental remediation repository of basic knowledge and recent progress is reviewed in this chapter. The chapter summarizes the key processes in the use of electrochemical membranes, focusing on their need in electrodialytic and the electrocatalytic remediation of contaminated media. The fundamentals (including materials and reactor design) of the electrodialytic remediation technology are presented with consideration given to the critical operating parameters and performance indicators, mathematical simulation/modeling methods, and recent advances in electrodialytic remediation research. Meanwhile, two membrane technologies (i.e., the 3D electrochemical system and the proton-conducting membrane cell) for electrocatalytic remediation of contaminated media under different scenarios are briefly described. Finally, the chapter ends with a critical discussion on the challenges of and the perspectives for future study in electrochemical membrane technology for the purpose of environmental remediation.
This chapter introduces membrane capacitive deionization (MCDI) and flow-electrode capacitive deionization (FCDI) as membrane-based electrochemical water treatment technologies based on the capacitive deionization principles. MCDI involves the use of a large number of static electrode pairs (anode and cathode) typically arranged in parallel, whilst FCDI involves the use of a number of pairs of electrodes constructed by flowing slurries of conducting particles between the anodic and cathodic current collectors. In both cases, the ions are removed from the feed stream by the application of an electrical potential across the electrodes, and are stored in the macropores and micropores (and retained in the electrical double layers) of the electrodes. In MCDI, an ion exchange membrane (IEM) is placed in front of each electrode to improve the desalination performance; and, in FCDI, IEMs are used to separate the water and flow electrode channels. By altering the operational configuration or IEM selection, specific ions of interest can be targeted for removal. MCDI and FCDI are low energy water treatment methods and can be applied to water desalination for potable water or water reuse applications, resource recovery, and contaminant abatement.
Separation processes are employed intensively for purification purposes in chemical production, as well as potable water generation and environmental remediation. Electrochemical separation techniques have long been studied and have been subsequently implemented for commercial application in various industries, with their modularity and capability for precise control of operating conditions serving as defining positive attributes. As global resource demand continues to rise, there is an urgency to further enhance separation process efficiency. Electrochemically mediated sustainable separations are designed to address this challenge by providing greater energy efficiency and engineered targeting of specific analytes while maintaining the inherent compact and portable nature of electrochemical systems. Recent advances in the field have leveraged redox‐active species as heterogeneously functionalized composites on electrode surfaces to perform separations at the solid–solution interface via methods such as electrochemical conversion, electrochemically mediated binding, and electrochemically modulated hydrophilicity tuning. Electroactive materials help to improve removal capacity through their heightened pseudocapacitance and can also impart molecular recognition through redox‐enhanced complex formation, intercalation, ion‐imprinting, hydrophilic interactions, and conversion processes for the selective removal of various charged and neutral compounds. The complementary combination of two redox‐active species into an asymmetric framework provides even greater functionality to electrochemical separation cells by increasing adsorption amounts, suppressing changes in water chemistry, reducing energetic requirements, inhibiting unwanted side‐reactions of target analytes, and enabling the conversion of toxic substances via reactive separation. There is a broad scope for the continued development of electrochemical concepts into next‐generation technologies for sustainable water‐based electrochemical separations and other related areas.
Full-text available
Climate change mitigation scenarios that meet the Paris Agreement's objective of limiting global warming usually assume an important role for carbon dioxide removal and negative emissions technologies. Direct air capture (DAC) is a carbon dioxide removal technology which separates CO2 directly from the air using an engineered system. DAC can therefore be used alongside other negative emissions technologies, in principle, to mitigate CO2 emissions from a wide variety of sources, including those that are mobile and dispersed. The ultimate fate of the CO2 , whether it is stored, reused, or utilised, along with choices related to the energy and materials inputs for a DAC process, dictates whether or not the overall process results in negative emissions. In recent years, DAC has undergone significant technical development, with commercial entities now operating in the market and prospects for significant upscale. Here we review the state-of-the-art to provide clear research challenges across the process technology, techno-economic and socio-political domains.
Full-text available
In electrochemical water desalination, a large difference in pH can develop between feed and effluent water. These pH changes can affect the long-term stability of membranes and electrodes. Often Faradaic reactions are implicated to explain these pH changes. However, quantitative theory has not been developed yet to underpin these considerations. We develop a theory for electrochemical water desalination which includes not only Faradaic reactions but also the fact that all ions in the water have different mobilities (diffusion coefficients). We quantify the latter effect by microscopic physics-based modeling of pH changes in Membrane Capacitive Deionization (MCDI), a water desalination technology employing porous carbon electrodes and ion-exchange membranes. We derive a dynamic model and include the following phenomena: I) different mobilities of various ions, combined with acid-base equilibrium reactions; II) chemical surface charge groups in the micropores of the porous carbon electrodes, where electrical double layers are formed; and III) Faradaic reactions in the micropores. The theory predicts small pH changes during desalination cycles in MCDI if we only consider phenomena I) and II), but predicts that these pH changes can be much stronger if we consider phenomenon III) as well, which is in line with earlier statements in the literature on the relevance of Faradaic reactions to explain pH fluctuations.
The 2017 edition of the International Energy Agency (IEA)’s comprehensive publication on energy technology focuses on the opportunities and challenges of scaling and accelerating the deployment of clean energy technologies. This includes looking at more ambitious scenarios than the IEA has produced before. Improvements in technology continue to modify the outlook for the energy sector, driving changes in business models, energy demand and supply patterns as well as regulatory approaches. Energy security, air quality, climate change and economic competitiveness are increasingly being factored in by decision makers. Energy Technology Perspectives 2017 (ETP 2017) details these trends as well as the technological advances that will shape energy security and environmental sustainability for decades to come. For the first time, ETP 2017 looks at how far clean energy technologies could move the energy sector towards higher climate change ambitions if technological innovations were pushed to their maximum practical limits. The analysis shows that, while policy support would be needed beyond anything seen to date, such a push could result in greenhouse gas emission levels that are consistent with the mid-point of the target temperature range of the global Paris Agreement on climate change. The analysis also indicates that regardless of the pathway chosen for the energy sector transformation, policy action is needed to ensure that multiple economic, security and other benefits to the accelerated deployment of clean energy technologies are realised through a systematic and co-ordinated approach. ETP 2017 also features the annual IEA Tracking Clean Energy Progress report, which shows that the current progress in clean energy technology development and deployment remains sub-optimal. It highlights that progress has been substantial where policies have provided clear signals on the value of technology innovation. But many technology areas still suffer from a lack of financial and policy support.
Ion adsorption and equilibrium between electrolyte and microstructure of porous electrodes are at the heart of capacitive deionization (CDI) research. Surface functional groups are among the factors which fundamentally affect adsorption characteristics of the material and hence CDI system performance in general. Current CDI-based models for surface charge are mainly based on a fixed (constant) charge density, and do not treat acid-base equilibria of electrode microstructure including so-called micropores. We here expand current models by coupling the modified Donnan (mD) model with weak electrolyte acid-base equilibria theory. In our model, surface charge density can vary based on equilibrium constants ( 's) of individual surface groups as well as micropore and electrolyte pH environments. In this initial paper, we consider this equilibrium in the absence of Faradaic reactions. The model shows the preferential adsorption of cations versus anions to surfaces with respectively acidic or basic surface functional groups. We introduce a new parameter we term “chemical charge efficiency” to quantify efficiency of salt removal due to surface functional groups. We validate our model using well controlled titration experiments for an activated carbon cloth (ACC), and quantify initial and final pH of solution after adding the ACC sample. We also leverage inductively coupled plasma mass spectrometry (ICP-MS) and ion chromatography (IC) to quantify the final background concentrations of individual ionic species. Our results show a very good agreement between experiments and model. The model is extendable to a wide variety of porous electrode systems and CDI systems with applied potential.
Capacitive deionization (CDI) and membrane capacitive deionization (MCDI) are the most common cell architectures in the use of CDI for water treatment. In this work, the Faradaic reactions occurring in batch-mode CDI and MCDI processes were compared by investigating the variation of H2O2 and dissolved oxygen (DO) concentrations, pH, conductivity and current during charging and discharging under different charging voltages. During charging, the H2O2 concentration in CDI increased rapidly and then decreased while almost no H2O2 was generated in MCDI due to the inability of oxygen to penetrate the ion exchange membrane. Chemical kinetic models were developed to quantitatively describe the variation of H2O2 concentration and found to present satisfactory descriptions of the experimental data. The pH drop during charging could be partially explained by Faradaic reactions with proton generation associated with oxidation of the carbon electrodes considered to be the major contributor. The electrode potentials required for the induction of Faradaic reactions were analyzed with this analysis providing robust thermodynamic explanations for the occurrence of carbon oxidation at the anode and H2O2 generation at the cathode during the ion adsorption process. Finally, electrochemically-induced ageing of the carbon electrodes and the resulting performance stability were investigated. The findings in this study contribute to a better understanding of Faradaic reactions in CDI and MCDI and should be of value in optimizing CDI-based technologies for particular practical applications.
Understanding the chemistry of carbon dioxide is key to effecting changes in atmospheric concentrations. One area of intense interest is CO2 capture in chemically reversible cycles relevant to carbon capture technologies. Most CO2 capture methods involve thermal cycles in which a nucleophilic agent captures CO2 from impure gas streams (e.g. flue gas), followed by a thermal process in which pure CO2 is released. Several reviews have detailed progress in these approaches. A less explored strategy uses electrochemical cycles to capture CO2 and release it in pure form. These cycles typically rely on electrochemical generation of nucleophiles that attack CO2 at the electrophilic carbon atom, forming a CO2 adduct. Then, CO2 is released in pure form via a subsequent electrochemical step. In this Perspective, we describe electrochemical cycles for CO2 capture and release, emphasizing electrogenerated nucleophiles. We also discuss some advantages and disadvantages inherent in this general approach.
We report on Faradaic reactions producing H⁺ (anode) and OH– (cathode) in flow-electrode capacitive deionization (FCDI) operated at 1.2 V. These reactions underline an additional electrodialytical desalination mechanism within capacitive deionization, which proceeds in parallel to the known electrosorption mechanism. Examination of flow-electrodes (100 ml each, 5% (wt) activated carbon) during FCDI (121 cm² effective membrane area) of 150 ml, 4 g/l NaCl solution revealed that significant amounts of Na⁺ and Cl⁻ ions (up to 50% and 30% of Cl⁻ and Na⁺, respectively) were not adsorbed in the activated carbon particles but were rather dissolved in the aqueous phase of the flow-electrodes. Production of acid (resulting in pH ≈ 1.5) and base (pH ≈ 12.5) in the flow-anode and -cathode solutions was observed during the operation. Reverse pH behaviors were obtained during the regeneration of the flow-electrodes by potential reversal. pH neutralization of the flow-electrode solutions resulted in a sharp increase in both the desalination rate and the electric current of the FCDI cell. Reacting NaOH and HCl in a short-circuited FCDI cell resulted in NaCl production in the water compartment and pH neutralization of both flow-electrodes.
Non-Faradaic (ion electrosorption) and Faradaic (oxidation-reduction) effects in a batch-mode capacitive deionization (CDI) system were investigated, with results showing that both effects were enhanced with an increase in charging voltage (0.5-1.5 V). Significant concentrations of hydrogen peroxide (H2O2) were observed with the generation of H2O2 initiated by cathodic reduction of O2 with subsequent consumption occurring as a result of cathodic reduction of H2O2. A kinetic model of the Faradaic processes was developed and found to satisfactorily describe the variation in the steady-state concentration of H2O2 generated over a range of CDI operating conditions. Significant pH fluctuations were observed at higher charging voltages. While the occurrence of Faradaic reactions may well contribute to pH fluctuations and deterioration of electrode stability and performance, the presence of H2O2 could provide the means of inducing disinfection or trace contaminant degradation provided H2O2 could be effectively activated to more powerful oxidants (by, for example, ultraviolet irradiation).
Electroreduction of CO2 has potential for storing otherwise wasted intermittent renewable energy, while reducing emission of CO2 into the atmosphere. Identifying robust and efficient electrocatalysts and associated optimum operating conditions to produce hydrocarbons at high energetic efficiency (low overpotential) remains a challenge. In this study, four Cu nanoparticle catalysts of different morphology and composition (amount of surface oxide) are synthesized and their activities towards CO2 reduction are characterized in an alkaline electrolyzer. Use of catalysts with large surface roughness results in a combined Faradaic efficiency (46%) for the electroreduction of CO2 to ethylene and ethanol in combination with current densities of ∼200 mA cm-2, a 10-fold increase in performance achieved at much lower overpotential (only < 0.7 V) compared to prior work. Compared to prior work, the high production levels of ethylene and ethanol can be attributed mainly to the use of alkaline electrolyte to improve kinetics and the suppressed evolution of H2, as well as the application of gas diffusion electrodes covered with active and rough Cu nanoparticles in the electrolyzer. These high performance levels and the gained fundamental understanding on Cu-based catalysts bring electrochemical reduction processes such as presented here closer to practical application.
Capacitive Deionization (CDI) is an electrochemical method for water desalination employing porous carbon electrodes. To enhance the performance of CDI, identification of electronic and ionic resistances in the CDI cell is important. In this work, we outline a method to identify these resistances. We illustrate our method by calculating the resistances in a CDI cell with membranes (MCDI) and by using this knowledge to improve the cell design. To identify the resistances, we derive a full-scale MCDI model. This model is validated against experimental data and used to calculate the ionic resistances across the MCDI cell. We present a novel way to measure the electronic resistances in a CDI cell, as well as the spacer channel thickness and porosity after assembly of the MCDI cell. We identify that for inflow salt concentrations of 20 mM the resistance is mainly located in the spacer channel and the external electrical circuit, not in the electrodes. Based on these findings, we show that the carbon electrode thickness can be increased without significantly increasing the energy consumption per mol salt removed, which has the advantage that the desalination time can be lengthened significantly.