Journal of Membrane Science 653 (2022) 120549

Contents lists available at ScienceDirect

Journal of Membrane Science

journal homepage: www.elsevier.com/locate/memsci

Salt and ion transport in a series of crosslinked AMPS/PEGDA

hydrogel membranes
Ni Yan a, Rahul Sujanani a, Jovan Kamcev b, Eui-Soung Jang a, Kentaro Kobayashi a,
Donald R. Paul a, Benny D. Freeman a, *
a
McKetta Department of Chemical Engineering, The University of Texas at Austin, 200 E. Dean Keeton Street, Austin, TX, 78712, USA
b
Department of Chemical Engineering, University of Michigan, North Campus Research Complex B28, 2800 Plymouth Road, Ann Arbor, MI, 48109, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Salt permeability and ionic conductivity are critical properties for membranes in water purification and energy
Ion exchange membrane applications. Both properties depend on individual ion sorption and diffusion coefficients, which are significantly
Salt permeability influenced by polymer chemical and physical parameters such as fixed charge concentration and membrane
Ionic conductivity
water content. However, systematic studies connecting polymer structure to ion transport properties are still
Salt diffusion
lacking. In this study, a series of uncharged and charged membranes were synthesized using poly(ethylene
Ion diffusion
glycol) diacrylate (PEGDA) as a cross-linker and 2-acrylamido-2-methyl-1-propanesulfonic acid (AMPS) as a
charged monomer. Membrane fixed charge concentration and water uptake were systematically varied by
adjusting AMPS content in the pre-polymerization mixture. Salt sorption and permeability coefficients and ionic
conductivity of these membranes were measured as a function of NaCl solution concentration (0.01–1 M).
Combining the solution-diffusion model and Nernst-Planck equation, individual ion diffusion coefficients were
calculated. Experimental Na+ diffusion coefficients for all materials were well described by the Mackie and
Meares tortuosity model, highlighting the strong influence of water content on ion diffusivity in both uncharged
and charged polymers. Model predictions for Cl− diffusion coefficients agree reasonably well with experimental
values, with some deviations occurring in more highly charged membranes. This discrepancy might result from
interactions not captured by the Mackie and Meares model (e.g., fixed charge-ion interactions).

1. Introduction
Ion exchange membranes (IEMs) are actively used and explored for
applications such as reverse osmosis, forward osmosis, electrodialysis
(ED), reverse electrodialysis (RED), and fuel cells due to their ion se­
lective properties [1–9]. To enhance separation efficiency and reduce
energy costs, these technologies often have specific requirements for
membrane properties (e.g., ionic conductivity, permselectivity) [10].
Meeting these specifications requires rational control of ion transport
properties via tailoring polymer chemical and physical parameters [10,
11]. IEMs contain charged functional groups covalently bound to the
polymer backbone (i.e., fixed charge groups). These ionogenic groups
can ionize and retain substantial amounts of water in the polymer upon
exposure to aqueous electrolyte solutions [12]. The concentration of
fixed charge groups and water content of the membrane significantly
influence ion transport [13–16].
Ion transport in nonporous membranes is described by the solution-
diffusion model [17,18]. Ions from the external solution first partition
into the membrane at the upstream side, diffuse down an electro­
chemical potential gradient, and desorb into the contiguous solution at
the downstream side [12,18,19]. Ion sorption in highly charged IEMs is
strongly influenced by the fixed charge groups via electrostatic in­
teractions between the fixed charges and ions [20–22]. Typically,
counter-ions (i.e., ions bearing a charge opposite to that of the fixed
charge groups) are attracted to the membrane, sorbing in large quanti­
ties and co-ions (i.e., ions bearing the same charge as that of the fixed
charge groups) are largely excluded from the membrane (i.e., Donnan
exclusion) [12]. This exclusion effect is stronger when the fixed charge
concentration in the membrane is higher than the electrolyte concen­
tration in solution [12]. At equilibrium, sorbed counter-ions must
electrically balance both the fixed charge groups on the polymer back­
bone and any co-ions sorbed in the membrane, so there are more
counter-ions in a charged membrane than co-ions, if the counter-ions
and co-ions have the same valence.

* Corresponding author.
E-mail address: freeman@che.utexas.edu (B.D. Freeman).

https://doi.org/10.1016/j.memsci.2022.120549
Received 12 January 2022; Received in revised form 1 April 2022; Accepted 5 April 2022
Available online 8 April 2022
0376-7388/© 2022 Elsevier B.V. All rights reserved.
N. Yan et al. Journal of Membrane Science 653 (2022) 120549

Ion diffusion is strongly influenced by the water content in a mem­ where ξ is the Manning parameter and can be determined based on
brane [13,23]. Water can plasticize polymer chains, thereby facilitating polymer structure [25,33]. Condensation does not occur in the polymers
ion diffusion [24]. In addition, fixed charges may affect ion diffusion via considered in this study, and in this case Cm,w can be solved numerically
electrostatic interactions with sorbed ions [25,26]. However, the
−
by combining Eqns. (1) and (4) [33]:
methods used to determine ion transport properties in IEMs vary greatly ⎡ ⎤
in the literature, and inconsistent results across these studies make
fundamental relations between polymer structure and ion transport ( ) ⎢
C−m,w C−m,w + CAm,w exp⎢
ξ ⎥ ( s )2 ( s )2
⎥= γ Cs (5)
⎣− 1 + 2 CC−m,w ⎦
properties difficult to discern [27–31]. Moreover, studies unifying ion
m,w ±

transport in both uncharged and charged polymers remain incomplete.

A

To bridge these gaps, this study focuses on exploring ion transport

properties in a series of uncharged and negatively charged membranes where γs± values can be estimated using the Pitzer model [39].
(i.e., cation exchange membranes, CEMs). The concentration of fixed Since the entire swollen membrane volume (i.e., volume of polymer,
charge groups in the membrane was systematically controlled by vary­ water, and ions) is the relevant volume to model transport phenomena
ing the charged monomer content in the pre-polymerization mixture. [25], ion concentrations based on the volume of sorbed water were
Ion sorption, salt permeability, and ionic conductivity of the membranes converted to concentrations based on the volume of swollen membrane
were measured as a function of external NaCl concentration (i.e., 0.01–1 as follows [25]:
M). Individual ion diffusion coefficients were calculated using the Cim,p = Cim,w ϕw (6)
Nernst-Planck framework [32]. To the best of our knowledge, such in­
formation for the same set of uncharged and charged membranes is where Cm,pi is expressed as mol of ions (i = + for counter-ion, i = − for
rarely presented in the literature. Sorption results were interpreted using co-ion) per liter of swollen membrane and ϕw is the water volume
the Donnan/Manning model [33]. Diffusion and permeation results fraction, reported in our prior work [33]. The salt sorption coefficient,
were interpreted using the Mackie and Meares model and the Ks , is defined as follow [20]:
solution-diffusion model [25].
Csm,p
Ks = (7)
2. Background Css

2.1. Donnan/Manning model where Cm,p

s is the mobile salt concentration (i.e., co-ion concentration for
a CEM equilibrated with a monovalent electrolyte) in the membrane.
When thermodynamic equilibrium is established between a CEM (e.
g., a sulfonated polymer) and a contiguous 1:1 electrolyte solution, the 2.2. Mackie and Meares model
co-ion concentration in the CEM is given by [20]:
Ion diffusion in highly hydrated polymers has often been described
[ ( s )2 ( s )2 ]12
(CAm,w )
2
C γ Cm,w by simple tortuosity or obstruction based models, including the Mackie
m,w
C− = + s m m± − A (1) and Meares model, which was developed based on a rigid lattice-type
4 γ+ γ− 2
matrix [23]. In this model, water in highly swollen polymers forms a
continuous pathway where ion transport occurs [14,27]. In this limit,
where Cm,w is the co-ion concentration in the membrane (mol of ions per
−
the path traveled by the ions in the polymer is considered to be tortuous
liter of water sorbed in the membrane), Cm,w is the fixed charge con­
A because the polymer chains act as immobile roadblocks that the ions
centration in the membrane (mol of fixed charge groups per liter of must traverse around to diffuse through the membrane [23]. In this
water sorbed in the membrane), Css is the electrolyte concentration of the model, the ratio of an ion’s diffusion coefficient in the membrane, Dm i , to
solution surrounding the membrane, γm m
+ and γ − are the cation and anion that in aqueous solution, Dsi , is related only to the polymer water volume
activity coefficients in the membrane, and γs± is the mean ion activity fraction, ϕw , and requires no fitting parameters [23]:
coefficient in the external solution (γ s± )2 = γ s+ γ s− [12,20]. In ideal ( )2
Donnan theory, the ratio of ion activity coefficients in the membrane Dmi
=
ϕw
(8)
( ) Dsi 2 − ϕw
(γs )2
and solution phases γm±γm is equal to 1 [34], and Eqn. (1) becomes
Using the Mackie and Meares model, Dm i can be predicted based on
+ −

[12]: polymer water volume fraction and penetrant diffusion coefficients in

[
2
]12 aqueous solution (e.g., DsNa+ = 13.3 × 10− 6 cm2/s, DsCl− = 20.3 × 10− 6
(CAm,w ) ( )2 CAm,w
C−m,w = + Css − (2) cm2/s, and DsNaCl = 16.1 × 10− 6 cm2/s at 25 ◦ C), which are tabulated
4 2
elsewhere [25,40]. Ion and salt diffusion coefficients in aqueous NaCl
The counter-ion concentration in the CEM is calculated using the solutions are relatively insensitive to changes in salt concentration in the
condition of electroneutrality as [35]: range of NaCl concentrations considered in this study (i.e., 0.01–1 M)
[25].
C+m,w = C−m,w + CAm,w (3)

Several studies have shown that in highly charged IEMs can be

γm m 2.3. Solution-diffusion model
+ γ−
predicted using Manning’s counter-ion condensation theory, for the case
At steady state, the integral salt permeability coefficient, Ps , is given
where condensation occurs [22,35–37]. In a recent study, we observed
by Refs. [17,41]:
that Manning’s model also provides good predictions in highly charged
IEMs when condensation does not occur [33]. For this case, γm
+ γ − is given
m
< Ps >=< Ds > Ks
m*
(9)
by Ref. [38]:
⎡ ⎤ m*
where < Ds > is the concentration averaged (i.e., apparent) salt
⎢ ξ ⎥ diffusion coefficient computed from experimental salt permeability and
γ m+ γm− = exp⎢
⎣−
⎥ (4)
1 + 2 CC−m,w ⎦ salt sorption coefficient values via Eqn. (9). However, salt diffusion
m,w

coefficients determined in this manner contain inherent contributions

A

from frame of reference and non-ideal thermodynamic effects [41],

2
N. Yan et al. Journal of Membrane Science 653 (2022) 120549

while diffusion coefficients predicted using the Mackie Meares model do
not. Frame of reference effects can be accounted for by including con­
vection in Fick’s law [41]. Non-ideal thermodynamic effects can be
accounted for using ion activity coefficient gradients in the membrane,
which can be obtained from experimental ion sorption results [41,42].
After accounting for these effects, the local salt diffusion coefficient, Dm
s ,
can be obtained [41]. In this study, frame of reference and non-ideal
thermodynamic effects nearly offset each other for highly charged
m*
samples considered, such that Dm s ≈< Ds >, (cf. Supporting Informa­
tion), similar to findings from a previous report on salt diffusion in ion
exchange membranes [41]. Dm s values for the uncharged sample are
m*
slightly higher (<5%) than < Ds > values, since both the frame of
reference and non-ideal thermodynamic effects increased salt diffusion
coefficients (cf. Supporting Information). However, the difference be­
m*
tween Dm s and Ds is within the experimental uncertainty. Thus, for the
materials considered in this study, using the apparent salt diffusion co­
efficients is a reasonable approximation for the local salt diffusion co­
efficients for further data analysis and comparison to model.
2.4. Nernst-Planck equation
Ion transport in membranes can be described by the Nernst-Planck
equation, which accounts for concentration and electric potential gra­
dients across the membrane [43]:
[ m,p ] received from Sigma-Aldrich (Milwaukee, WI). De-ionized (DI) water
dCi zi FCim,p dψ
Jim = − Dmi + (10)
dx RT dx
where Jim and Dm i are the flux and diffusion coefficient of ion i in the
membrane. ψ is the electric potential, F is Faraday’s constant, R is the
gas constant, and T is absolute temperature.
Ion transport through membranes varies with the type of driving
force applied [32]. For example, when a membrane separates two so­
lutions with different salt concentrations and no electric potential
gradient is imposed across the membrane, a chemical potential (i.e.,
concentration) gradient across the membrane drives cations and anions
to diffuse in the same direction to maintain electroneutrality [32]. This
coupled diffusion of cations and anions across the membrane can be
described by the salt diffusion coefficient of the membrane [41]:
(
Dm Dm Cm,p + C+m,p
) ductivity change in the downstream solution was monitored as a func­
Dms = + m− m − m m
(11)
D+ C+ + D− C−
where Dm
s is the local salt diffusion coefficient of the membrane
mentioned above, Cm,p+ and C−
m,p
are the cation and anion concentrations
in the membrane (mol of ions per liter of swollen membrane), and Dm +
and Dm− are the cation and anion diffusion coefficients in the membrane
[41].
When a membrane separates two electrolyte solutions at the same
concentration, and an electric potential gradient is applied across the
membrane, cations and anions migrate in opposite directions towards
their respective electrodes, inducing an electric current [12,32,43]. In
this case, ion transport is described by the ionic conductivity, κ, which
for a monovalent electrolyte is given by Ref. [43]:
coefficient, Dm
m*
s (which ≈< Ds > in this study as calculated from < Ps >
and Ks ). For example, once Dm
− is calculated from Eqn. (13), Eqn. (11) or
(12) can be used to calculate Dm+.
3. Experimental methods
3.1. Materials
A detailed description of the polymer synthesis procedure is reported
elsewhere [42]. Fig. 1 presents the structures of XLPEGDA homopoly­
mers and XL(AMPS-PEGDA) copolymers, which were prepared as
free-standing films by UV irradiation of a pre-polymerization mixture
[44–47]. This mixture contained the cross-linker, poly(ethylene glycol)
diacrylate (PEGDA) [n = 13, n = 10, and n = 4, where n represents the
average number of ethylene oxide (EO) groups per cross-linker mole­
cule], the ionic monomer, 2-acrylamido-2-methyl-1-propanesulfonic
acid (AMPS), and the photoinitiator, 2,2-dimethoxy-2-phenylacetophe­
none (DMPA). Water and, in some cases, water/methanol (MeOH)
mixtures served as the solvent for the free radical polymerization.
Table 1 presents compositions of the pre-polymerization mixtures. After
UV-polymerization, the solid hydrogel films were soaked in DI water for
at least one day to extract residual components not bound to the
network. Water was changed three times daily. The as-synthesized acid
form (H+) XL(AMPS-PEGDA) films were converted to the Na+ form
using procedures described elsewhere [42]. All reagents were used as
(18.2 MΩ cm, 1.2 ppb total organic carbon) was generated by a Millipore
Milli-Q Advantage A10 water purification system (Millipore Corpora­
tion, Bedford, MA).
3.2. Salt permeability
Prior to salt permeability tests, films of XLPEGDA and Na+ form XL
(AMPS-PEGDA) were equilibrated in DI water for at least one day.
Films were then clamped between a pair of jacketed glass diffusion cells
(PermeGear Side-bi-Side, Hellertown, PA) [47,48]. The donor (up­
stream) and receiver (downstream) chambers were filled with equal
volumes of NaCl solution at a desired concentration (i.e., 0.01–1 M) and
DI water, respectively. The solutions in both chambers were vigorously
stirred, and the temperature was maintained at 25 ± 0.1 ◦ C. The con­
tion of time using a conductivity meter (WTW inoLab Cond 730,
Woburn, MA). The downstream salt (i.e., NaCl) concentration was
computed from conductivity data using a calibration curve. As described
elsewhere, at pseudo-steady state, the integral salt permeability coeffi­
cient, < Ps >, is determined as follows [47,48]:
[ ] ( )
2Cs [t] 2A < Ps >
ln 1 − s sl =− t (14)
Cs0 [0] VL

F 2 ( m m,p )
κ= D C + Dm− C−m,p (12)
RT + +
Combining Eqns. (11) and (12) yields:
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
κRT
( κRT )2 Cm,p
m,p m
− C + Ds
F2
± F2
− 4 κRT F 2 C−m,p +Cm,p
(13)
+
m
D− =
2C−m

An analogous expression for Dm + can also be derived. Thus, Di values

m

can be calculated using data for ionic conductivity, κ, the concentrations Fig. 1. Chemical structures of: (a) XL(AMPS-PEGDA) copolymers and (b)
of sorbed ions in the membrane, Cm,p
− and Cm,p
+ , and the local salt diffusion XLPEGDA homopolymers.

3
N. Yan et al. Journal of Membrane Science 653 (2022) 120549

where Css0 [0] is the initial upstream salt concentration, and Cssl [t] is the

Methanol [g]
downstream salt concentration at time t. A is the active sample area
(1.77 cm2), V is the liquid volume in both cells (35 mL), and L is the

2.20
hydrated membrane thickness measured immediately after the experi­
–
–
–
–
–
–
–

–
ment was completed.
For low upstream salt concentrations, dissolution of CO2 from air in
the solutions of the diffusion cells can artificially elevate the down­
DI water [g]

stream conductivity, causing the measured salt permeability coefficient

to be higher than the true value [49]. This effect is more pronounced

Theoretical IEC values of the acid form (H+) XL(AMPS-PEGDA) films were determined as milliequivalents (i.e., mmol) of AMPS/(g of AMPS + g of PEGDA)42 or meq/g (dry polymer).
when salt permeation tests are performed with highly charged mem­
5.50
5.77
5.00
5.50
5.00
5.50
5.00
3.30
5.00
branes at low upstream salt concentrations (i.e., 0.01 M) [49]. To
mitigate this effect for the highly charged membranes (e.g., samples
with IEC values of 0.44–1.93 meq/g) used in this study, ultra-high purity
N2 (Airgas, Austin, TX) was bubbled through the upstream (i.e., 0.01 M
Initiator (DMPA) [g]

NaCl) and downstream solutions before and during these experiments.

This strategy mitigates the influence of CO2 on salt permeability results
measured at low upstream solution concentrations, as reported else­
0.0165
0.0173
0.0200
0.0165
0.0150
0.0165
0.0150
0.0165
0.0150

where [49].

3.3. Osmotic water permeability

Water flux, driven by an osmotic pressure difference across the

Ionic monomer (AMPS) [g]

membrane during salt permeability tests, was measured using a plastic

diffusion cell, as described elsewhere [41]. Following equilibration in DI
water, films were clamped between the two chambers of the diffusion
cell. In these experiments, the upstream and downstream solutions were
1 M NaCl and DI water of equal volumes (200 mL), respectively. When
filling each chamber, care was taken to ensure that no air bubbles were
1.5
5.0

1.5
3.0
5.0
6.0
1.5
6.0

trapped between the solution and cell wall [41]. Two capillary tubes
The samples were labeled as x-y, where x the number of EO units in PEGDA, n, and y is the theoretical IEC value [33].
–

were then attached to the top of each cell, each containing the same
solution as in their respective cells. At pseudo-steady state, the decrease
of the solution level in the downstream tube (or the solution level in­
Cross-linker (PEGDA) [g]

crease in the upstream tube) was recorded as a function of time, ΔV/Δt

(e.g., ml/10 min), and the water permeability coefficient, Pw , was
calculated as follows [41]:
|ΔV| L
Pw = (15)
Δt Aρw |Δp − Δπ|
15.0
12.3
15.0
15.0
12.0
11.5
9.00
15.0
9.00

where L is the hydrated membrane thickness determined following the

test, A is the active membrane area (1.77 cm2), ρw is the density of water
(1 g/mL), and Δp is the hydrostatic pressure difference applied on the
AMPSb content [wt%]

membrane (= 0 in this experiment). Δπ is the osmotic pressure differ­

ence across the membrane. The osmotic pressure in the cell initially
containing DI water is considered to be zero (due to its relatively low salt
concentration during the entire course of the experiment) and osmotic
pressure of the cell initially containing 1 M NaCl is estimated from the
Pitzer model (46.37 bar) [39].
29

20
30
40

40
9

0
9

wt% AMPS = g of AMPS/(g of AMPS + g of PEGDA).

3.4. Membrane ionic conductivity

IECa [meq/g (dry polymer)]
Compositions of the pre-polymerization mixtures.

Membrane ionic conductivity was measured using an Electro­

chemical Impedance Spectroscopy (EIS) system (1287A, 1260A, Solar­
tron, Ameteck Scientific Instruments, Berwyn, PA). A detailed
description of this measurement has been reported previously [50].
Prior to EIS experiments, films of XLPEGDA and sodium form XL
(AMPS-PEGDA) were equilibrated in a NaCl solution (100 mL, 0.01–1
M for one day) at the concentration at which conductivity was to be
0.44
1.40

0.44
0.97
1.46
1.93
0.44
1.93
0

measured. Membrane resistance values of the uncharged and weakly

charged films (i.e., IEC of 0–0.13 meq/g) were determined using a
previously described difference method [50]. In this method, films were
sandwiched between two plastic half cells (kindly provided by Calera
Sample (n-IEC)c

Corporation, Los Gatos, CA), which were equipped with platinum

electrodes. The cells were connected to a jacketed beaker containing
13–0.44
13–1.40

10–0.44
10–0.97
10–1.46
10–1.93
4–0.44
4–1.93
Table 1

about 200 mL of feed solution (i.e., NaCl solution having the same
10–0

concentration as the equilibration solution maintained at 25 ± 0.1 ◦ C),

b
a

4
N. Yan et al. Journal of Membrane Science 653 (2022) 120549

which was then pumped through the cell chambers using a peristaltic
pump (150 mL/min, Masterflex 7528–10, 77292–50). In the measure­
ments conducted at high salt concentrations (0.1–1 M), a DC current of
5 mA and an AC current of 1 mA were used [50]. At lower salt con­
centrations (0.01–0.03 M), an AC current of 0.3 mA was used with no DC
current [50]. The frequency sweep was between 40 and 0.5 kHz, and the
step interval for frequency change was 10 Hz. The total electrolyte
resistance of the membrane loaded cell (i.e., membrane + solution
resistance), Rm+s , and a blank cell containing no membrane (i.e., solu­
tion resistance), Rs , were recorded. The membrane resistance, Rm , was
calculated as follows [50]:
Rm = Rm+s − Rs (16)
In uncharged and weakly charged membranes, Rm+s , was 2–10 times
higher than Rs over the entire range of salt concentrations considered.
This large difference ensures that the Rm value can be accurately
determined using the difference method. The ionic conductivity of the
membrane was calculated as follows [50]:
L
κ= (17)
Rm A

where L is the hydrated membrane thickness measured directly after the

experiment, and A is the area of the electrodes in contact with cell so­
lutions (3.81 cm2).
Fig. 2. Water volume fraction as a function of external NaCl concentration in
For highly charged membranes (i.e., IEC of 0.44–1.93 meq/g), the
membranes prepared using PEGDA of n = 10. The uncertainty, determined as
difference between Rm+s and Rs is small, which can cause large un­ the standard deviation from measurements made on at least six separate sam­
certainties in Rm and κ values extracted using the difference method [50, ples, was less than 5% of the average of these measurements.
51]. In these cases, a recently developed direct contact method was used
to measure the membrane resistance [50]. In this method, the contri­
homopolymers measured using the difference method and direct contact
bution of solution layer resistance to the total cell resistance was mini­
method were equivalent within experimental uncertainty (cf. Support­
mized by quickly dipping films in a highly conductive NaCl solution and
ing Information).
directly clamping the film between the electrodes, allowing the true
membrane resistance to be measured. Following equilibration in NaCl
4. Results and discussion
solution, films were quickly dipped in 3 M NaCl solution (~1 s) to rinse
off the dilute equilibration solution (i.e., 0.01–1 M) left on membrane
4.1. Water sorption
surface and then directly clamped between two platinum electrodes. The
frequency range was 40–1 kHz with a DC current of 5 mA and an AC
Water sorption and more characterization information for the sam­
current of 1 mA [50]. The resistance results from the direct contact
ples considered in this study can be found in our previous reports [33,
method changed little as the dipping concentration increased from 3 to
42]. As shown in Table 2, for a given value of n (i.e., number of EO units
5 M (cf. Supporting Information). However, above 3 M, the hydrogel
in PEGDA), water volume fraction measured in DI water increases as IEC
films used in this study can easily break due to the sudden de-swelling
increases, presumably due to enhanced polymer-water affinity as more
caused by the elevated salt concentration in the solution surrounding
hydrophilic fixed charge groups are present in the membrane [12,42,
the membrane. Membrane ionic conductivity values for XLPEGDA
52]. Fig. 2 presents water volume fraction as a function of external NaCl

Table 2
Polymer water uptake, water volume fraction, and fixed charge concentration [42].
Sample (n-IEC) wu a [g (water)/g (dry polymer)] ϕw b [L (water)/L (swollen membrane)] Cm,w
A
c
[mol/L (water sorbed)]

13–0.44 0.989 ± 0.012 0.547 ± 0.003 0.444 ± 0.005

13–1.40 1.346 ± 0.006 0.634 ± 0.002 1.036 ± 0.005
10–0 0.581 ± 0.006 0.411 ± 0.004 0
2
10–0.01 0.593 ± 0.014 0.418 ± 0.006 (1.69 ± 0.04) × 10−
2
10–0.05 0.619 ± 0.003 0.429 ± 0.003 (8.40 ± 0.05) × 10−
10–0.13 0.645 ± 0.004 0.440 ± 0.002 0.208 ± 0.005
10–0.44 0.764 ± 0.009 0.487 ± 0.003 0.575 ± 0.007
10–0.97 0.936 ± 0.014 0.546 ± 0.004 1.031 ± 0.015
10–1.46 1.110 ± 0.024 0.593 ± 0.005 1.318 ± 0.029
10–1.93 1.300 ± 0.018 0.633 ± 0.003 1.485 ± 0.021
4–0.44 0.423 ± 0.007 0.353 ± 0.004 1.037 ± 0.017
4–1.93 0.882 ± 0.010 0.544 ± 0.003 2.189 ± 0.025
a
wu is the polymer water uptake (grams of water per g of dry polymer) measured in DI water.
b
ϕw is the polymer water volume fraction (liter of water per liter of swollen membrane) measured in DI water, and is calculated assuming volume additivity via
ϕw = wu /(wu +ρw /ρp ) , where ρw and ρp are the densities of water and dry polymer [33,42].
c
Cm,w
A is the membrane fixed charge concentration (mols of fixed charges per liter of sorbed water), which was determined based on the theoretical IEC values in
Table 1 and DI water uptake values via Cm
A = IEC⋅ρw /1000⋅wu [42].

5
N. Yan et al. Journal of Membrane Science 653 (2022) 120549

Fig. 3. The dependence of: (a) sorbed Cl− concentration, Cm,p

− , and (b) NaCl sorption coefficient on external NaCl concentration in membranes prepared with PEGDA
of n = 10. The uncertainty, determined as the standard deviation from measurements made on at least six separate samples, was less than 15% of the average of these
measurements. The measured Ks value of the uncharged XLPEGDA membrane at Css = 0.1 M agrees with previously reported results [47].

function of external NaCl concentration, Css . As Css increases from 0.01 to

1 M, the sorbed Cl− concentration increases for all samples considered.
However, the rate of increase of Cm,p − vs. Css in uncharged and highly
charged membranes is significantly different. For example, the Cl−
concentration in the most highly charged membrane (IEC = 1.93 meq/g)
increases by over four orders of magnitude as external NaCl concen­
tration increases. This large increase in co-ion sorption is typical for
charged membranes, due to strong Donnan exclusion at low Css values
and weakened Donnan exclusion at high Css [20,21]. In contrast, the Cl−
concentration in the uncharged membrane (labeled “0”) only increases
by approximately two orders of magnitude, which is approximately
proportional to the increase in Css [20].
Salt sorption coefficients were computed from the measured Cl−

Fig. 4. NaCl sorption coefficients as a function of IEC (meq/g) at external NaCl

concentrations of 0.01 M and 1 M in membranes prepared with PEGDA of n =
10. The filled symbols represent experimental sorption coefficients, and the
dashed lines denote predicted salt sorption coefficients according to the
Donnan/Manning model [i.e., Eqn. (5)].

concentration in samples prepared with PEGDA of n = 10. At a given IEC

value, water volume fraction decreases by about 7–12% as external salt
concentration increases due to osmotic de-swelling, since water activity
decreases in the external solution [25,33,42,52].

4.2. Ion sorption

4.2.1. Cl- sorption Fig. 5. The dependence of sorbed Na + concentration, Cm,p

Equilibrium Cl− concentrations in uncharged and charged mem­
branes prepared with PEGDA of n = 13, 10, and 4 were measured and
reported elsewhere [33,42]. As a representative example, Fig. 3 (a)
presents Cl− concentrations in selected membranes (i.e., n = 10) as a
+ , on external NaCl
concentration in membranes prepared with PEGDA of n = 10. The dashed lines
were drawn to guide the eye. The uncertainty, determined as the standard
deviation from measurements made on at least six separate samples, was less
than 10% of the average of these measurements.

6
N. Yan et al.
concentrations in the membranes via Eqn. (7). Fig. 3 (b) presents NaCl
sorption coefficients, Ks , as a function of external NaCl concentration.
The dependence of Ks on Css in uncharged and highly charged mem­
branes is qualitatively different. Ks values in the most highly charged
membrane (IEC = 1.93 meq/g) exhibit significant variation with Css ,
while those of the uncharged membrane (IEC = 0) change relatively
little with Css [20].
Fig. 4 presents the dependence of experimental NaCl sorption co­
efficients on IEC at the lowest and highest external NaCl concentrations
considered, 0.01 M and 1 M. For comparison, theoretical NaCl sorption
coefficients calculated using the Donnan/Manning model [i.e., Eqn. (5)]
are also shown in Fig. 4. At the lowest salt concentration considered (i.e.,
Css = 0.01 M), NaCl sorption coefficients initially decrease by about one
order of magnitude as IEC increases from 0 to 0.44 meq/g, presumably
due to enhanced Donnan exclusion induced by increased fixed charge
concentration in the membrane (cf. Cm,w A in Table 2). Then, Ks values are
relatively constant at high IEC values (i.e., IEC >0.44 meq/g). These
trends are qualitatively consistent with the model predictions [53]. That
is, the effects of Donnan exclusion are very sensitive to the change in
fixed charge concentration when only low levels of fixed charges are
introduced to the polymer network [42]. Further addition of fixed
charges to an already highly charged polymer promotes co-ion exclusion
to a lesser extent [42]. At the highest salt concentration considered (i.e.,
Css = 1 M), both experimental and theoretical Ks values depend relatively
weakly on IEC and asymptotically approach that of the uncharged
membrane (IEC = 0). This phenomenon indicates the Donnan exclusion
effect in charged membranes is greatly reduced at high Css , such that the
charged membranes sorb co-ions in a similar fashion to the uncharged
membranes [20]. The dependence of experimental NaCl sorption co­
efficients on fixed charge concentration (i.e., mol of fixed charge groups
per liter of water sorbed in the membrane) can be found in the Sup­
porting Information.
The quantitative agreement between the experimental values and
theoretical predictions are good for membranes with higher IEC values
(i.e., IEC >0.44 meq/g), suggesting that thermodynamic non-idealities
in these membranes are primarily governed by strong electrostatic in­
teractions between fixed charges and ions captured by Manning’s
counter-ion condensation theory [35]. For membranes with lower IEC
values (i.e., IEC <0.44 meq/g), the experimental Ks values are below the
model predictions, likely due to additional thermodynamic
non-idealities arising from dielectric exclusion and unfavorable
Fig. 6. Na + diffusion coefficients, Dm
+ , as a function of: (a) upstream NaCl concentration and (b) polymer water volume fraction in membranes prepared with PEGDA
of n = 10. The uncertainty, determined using propagation of error analysis [54], was less than 15% of the average Dm + value. The dashed lines were drawn to guide the
eye, and the solid line represents the Mackie and Meares model predictions [cf. Eqn. (8)]. The Na+ diffusion coefficient in external solution was taken as 13.3× 10−
cm2/s [55].
7
Journal of Membrane Science 653 (2022) 120549
non-electrostatic interactions between polymer segments, water, and
ions in such materials [22,33,35,42].
4.2.2. Na + sorption
Fig. 5 presents sorbed Na+ concentrations (i.e., the counter-ion
concentration in a CEM) as a function of external NaCl concentration,
Css . For the uncharged membranes, the Css dependence of Cm,p + resembles
that of Cm,p
− in Fig. 3 (a), because the membrane contains equal numbers
of Na+ and Cl− at any external NaCl concentration to ensure electro­
neutrality in the membrane [20]. In contrast, the Na+ concentration in
the membrane of IEC = 1.93 meq/g changes little with external NaCl
concentration. This weak dependence of Cm,p s
+ on Cs is typical for highly
charged cation exchange membranes [21], where the majority of the
sorbed cations are electrically balancing fixed charges on the polymer
backbone (i.e., Cm,p m,p
+ = CA + C−
m,p
≈ Cm,p
A ), such that the concentration of
sodium ions is relatively independent of Css [33,42]. For samples in be­
tween these extremes (i.e., IEC of 0.01–1.46 meq/g), Cm,p exhibits
+
behavior intermediate to these two limiting cases (i.e., the uncharged
and most highly charged samples considered).
4.3. Ion diffusion coefficients
4.3.1. Na + diffusion coefficients
The Na+ diffusion coefficients were computed using the measured
ion concentrations, salt permeability coefficient, and membrane ionic
conductivity values using the Nernst-Planck framework [i.e., Eqn. (11)
and (12)]. Fig. 6 (a) presents Na+ diffusion coefficients as a function of
upstream NaCl concentration. For each membrane, Na+ diffusion co­
efficients are fairly constant at low NaCl concentrations (Css < 0.1 M).
Then, Dm + values decrease by about 25–35% at high NaCl concentrations
(Css > 0.1 M). The dependence of Dm s
+ on Cs is qualitatively similar to the
trend of polymer water volume fraction (ϕw ) shown in Fig. 2. Addi­
tionally, Dm
+ values increase as IEC increases, similar to the increase in ϕw
with increasing IEC (cf. Fig. 2). The similarities in the dependence of
Na+ diffusion coefficients and water volume fraction on Css and IEC can
be qualitatively rationalized by the Mackie and Meares model [cf. Eqn.
(8)], which predicts that ions diffuse faster in samples with higher water
content.
To quantatitively compare experimental and theoretical results,
Fig. 6 (b) presents the experimental Na+ diffusion coefficients as a
function of water volume fraction. The experimental results are
6
N. Yan et al. Journal of Membrane Science 653 (2022) 120549

Fig. 7. Cl− diffusion coefficients, Dm

− , as a function of: (a) external NaCl concentration and (b) polymer water volume fraction in membranes prepared with PEGDA of
n = 10. The uncertainty, determined using the propagation error analysis [54], was less than 25% of the average Dm − value. The dashed lines were drawn to guide the
eye, and solid line represents the Mackie and Meares model predictions [cf. Eqn. (8)]. The Cl− diffusion coefficient in external solution was taken as 20.3× 10− 6
cm2/s [55].

predicted remarkably well by the Mackie and Meares model over a wide
range of ϕw values, regardless of whether ϕw was increased by adding
fixed charges to the polymer backbone (i.e., varying IEC) or decreased
by osmotic de-swelling caused by the increase in external NaCl con­
centration. Thus, membrane hydration largely controls counter-ion
diffusion in both uncharged and charged polymers where counter-ion
condensation does not occur. Similar agreement was also found in
other polymers prepared with PEGDA of n = 13 and n = 4 (cf. Sup­
porting Information).
4.3.2. Cl- diffusion coefficients
The Cl− diffusion coefficients were determined via Eqn. (13) and
presented in Fig. 7 (a) as a function of upstream NaCl concentration. For
all membranes considered, decreasing IEC and increasing upstream
NaCl concentration led to decreasing Cl− diffusion coefficients, pri­
marily due to reduced water content in membranes with lower IEC
values and in membranes equilibrated with more concentrated salt
solutions. This behavior is qualitatively consistent with the Mackie and
Meares model [cf. Eqn. (8)]. As shown in Fig. 7 (a), Dm − values in the
uncharged membrane depend somewhat more strongly on Css relative to
the weakly charged samples (i.e., IEC = 0.01–0.13 meq/g). The mo­
lecular basis for this behavior is not fully understood.
Fig. 7 (b) presents experimental and theoretical Dm − values as a
function of water volume fraction. Given the relatively simple nature of
the Mackie and Meares model, which contains no adjustable parameters,
this agreement is reasonable. However, some deviation is observed in
more highly charged samples for reasons that are not fully understood at
this time. Recently, the effect of electrostatic interactions between fixed
charges and ions on ion diffusion coefficients, as described by Manning’s
model [26], was found to be small compared to the effect of tortuosity
on ion diffusion, as captured by the Mackie and Meares model [25].
Similar results were also found for the membranes in this study. Recent
simulation studies have observed reduced co-ion diffusion coefficients in
charged matrices without obstructive polymer chains relative to those in

m*
Fig. 8. Apparent NaCl diffusion coefficients, < Ds >, as a function of: (a) upstream NaCl concentration and (b) polymer water volume fraction in membranes
prepared with PEGDA of n = 10. The dashed lines were drawn to guide the eye, and the solid line represents the Mackie and Meares model predictions [cf. Eqn. (8)].
m* m*
The uncertainty of < Ds >, determined using propagation of error analysis [54], was less than 15% of the average < Ds > value.

8
N. Yan et al.
Fig. 9. Apparent NaCl diffusion coefficient as a function of IEC values (meq/g)
at the lowest and highest upstream NaCl concentrations considered (0.01 and 1
M) in membranes prepared with PEGDA of n = 10. The dashed lines denote the
Mackie and Meares model predictions [cf. Eqn. (8)].
aqueous solution [56,57]. This effect was ascribed to a so-called “elec­
trostatic excluded volume” effect, since co-ions would tend to reside
further away from fixed charges due to electrostatic repulsion. Thus, the
volume surrounding fixed charges is inaccessible to co-ions, which may
result in a more tortuous diffusion pathways for co-ions relative to
counter-ions [57]. Similar results were also obtained in polymers pre­
pared with PEGDA of n = 13 and n = 4 (cf. Supporting Information).
4.4. Salt diffusion
Salt diffusion coefficients, determined in concentration gradient
driven measurements, depend on the concentrations and diffusion co­
efficients of individual ions in the membrane [cf. Eqn. (11)], since every
transported ion, driven by a concentration gradient across the
Fig. 10. The influence of upstream NaCl concentration on salt permeability coefficients of: (a) uncharged and weakly charged membranes (i.e., IEC <0.44 meq/g)
and (b) highly charged membranes (i.e., IEC >0.44 meq/g). These membranes were prepared with PEGDA of n = 10. The uncertainty, determined as the standard
deviation from measurements made on at least six separate samples, was less than 15% of the average of these measurements. The dashed lines were drawn to guide
the eye. The measured < Ps > value at Css = 0.1 M agrees with previous results reported for XLPEGDA membranes [7, 12].
9
Journal of Membrane Science 653 (2022) 120549
membrane, must be electrically balanced by an ion of opposite charge.
m*
Apparent salt diffusion coefficients, < Ds >, were computed from
experimental salt permeability and sorption coefficients via Eqn. (9). <
m*
Ds > values are presented in Fig. 8 (a) as a function of upstream NaCl
concentration. For all samples, salt diffusion coefficients decrease with
increasing Css and decreasing IEC values, primarily due to reduced
membrane water content under such conditions, which is consistent
m*
with the Mackie and Meares model. Interestingly, the values of < Ds >
and its dependence on Css for highly charged membranes (i.e., IEC >0.44
meq/g) are nearly identical to those of the Cl− diffusion coefficients
observed in Fig. 7 (a). For highly charged CEMs equilibrated with a
relatively dilute NaCl solution, the sorbed counter-ion (Na+) concen­
tration is much higher than that of co-ions (Cl− ) due to Donnan exclu­
sion (i.e., Cm,p m,p
+ ≫C− ). In this limit, Eqn. (11) can be simplified as follows:
(18)
m*
Dms ≈ Ds ≈ Dm−
Thus, salt diffusion in highly charged membranes in contact with
dilute electrolyte solutions is primarily governed by the diffusion of the
co-ions (i.e., the minority species), which are the Cl− ions in this case.
m*
In Fig. 8 (b), the experimental < Ds > data are presented as func­
tion of polymer water volume fraction along with theoretical predictions
from the Mackie and Meares model [58]. Within experimental uncer­
tainty, the experimental salt diffusion coefficients are reasonably
described by the model with no adjustable parameters, suggesting salt
diffusion coefficients in hydrated polymers are mainly affected by water
content, similar to results reported elsewhere [25]. Similar agreement
was also observed in polymers prepared with PEGDA of n = 13 and n = 4
(cf. Supporting Information).
Fig. 9 presents the dependence of experimental and theoretical salt
diffusion coefficients on IEC at fixed external NaCl concentrations (0.01
and 1 M). Increasing IEC from 0 to 0.13 meq/g leads to a 50% increase in
salt diffusion coefficients. In contrast, as IEC increases from 0.44 to 1.93
m*
meq/g, Ds values increase by roughly a factor of two, since polymer
water content increases more significantly at high IEC values (i.e., IEC
>0.13 meq/g) than that at low IEC values (i.e., IEC <0.13 meq/g),
consistent with the Mackie and Meares model.
N. Yan et al.
Fig. 11. The influence of IEC values on salt permeability coefficients at fixed
upstream NaCl concentrations of 0.01 M and 1 M in membranes prepared with
PEGDA of n = 10. The dashed lines denote the product of theoretical Ks values
predicted by the Donnan/Manning model [cf. Eqn. (5)] and Ds values predicted
by the Mackie and Meares model [cf. Eqn. (9)].
4.5. Salt permeability
Salt permeability coefficients, < Ps >, are presented as a function of
upstream NaCl concentration in Fig. 10 (a) and (b), respectively. In
Fig. 10 (a), salt permeability coefficients of the uncharged membrane (i.
e., IEC = 0) decrease by roughly 12% as Css increases from 0.01 to 1 M,
similar to other XLPEGDA membranes [20]. The decreasing trend of <
Ps > with increasing Css can be rationalized within the framework of the
m*
solution-diffusion model [i.e., < Ps > = Ks < Ds >]. The salt sorption
coefficients (Ks ) of the uncharged membrane increase by about 28%
over the entire range of Css values considered [cf. Fig. 3 (b)]. Thus, the
dependence of < Ps > on Css is predominantly controlled by the behavior
m*
of the salt diffusion coefficient (< Ds >), which decreases by about
31% as Css increases, due to a reduction in water content caused by
Fig. 12. The influence of: (a) external NaCl concentration and (b) sorbed Na+ concentration on membrane ionic conductivity in membranes prepared with PEGDA of
n = 10. The uncertainty of κ, determined as the standard deviation from measurements made on at least six separate samples, was less than 25% of the average of
these measurements. The dashed lines were drawn to guide the eye.
10
Journal of Membrane Science 653 (2022) 120549
osmotic de-swelling at high Css [20].
In Fig. 10 (b), < Ps > values of the highly charged membranes
exhibit a strong dependence on Css , which is typical for highly charged
IEMs due to the influence of the fixed charge groups on salt sorption [cf.
Fig. 3 (b)] [49,59]. For example, Ks in the samples having an IEC value of
1.93 meq/g increases by about 2 orders of magnitude as Css increases
m*
from 0.01 to 1 M, while only a modest decrease in < Ds > was observed
over the same range of Css . Thus, the strong Css dependence of < Ps > in
this highly charged membrane is primarily governed by the dependence
of Ks on Css . In this way, the concentration dependence of salt perme­
ability in highly charged membranes is controlled by the partitioning
and diffusion behavior of co-ions [cf. Eqns. (7) and (19)]. For all the
other charged membranes (e.g., IEC = 0.01–1.46 meq/g), the depen­
dence of < Ps > on Css is in between the trends observed in the uncharged
and highest charged membranes.
The dependence of experimental salt permeability coefficients on
IEC at the lowest and highest Css values considered is shown in Fig. 11.
Theoretical salt permeability coefficients were computed from the
solution-diffusion model using salt sorption coefficients predicted from
the Donnan/Manning approach, and salt diffusion coefficients predicted
from the Mackie and Meares model. Interestingly, at the lowest Css value
considered (i.e., 0.01 M), the experimental salt permeability coefficients
decrease to a minimum followed by a slight increase (~20%) as IEC
increases. This phenomenon can be explained by the different de­
m*
pendencies of Ks and < Ds > on IEC, as shown in Figs. 4 and 9,
respectively. When Css = 0.01 M and IEC values are below about 0.44
meq/g, NaCl sorption coefficients decrease significantly as IEC increases
(cf. Fig. 4). Increases in IEC are accompanied by increases in water
content, so salt diffusion coefficients increase with increasing IEC. The
m*
increase in < Ds > is more pronounced at higher IEC values (>0.44
meq/g) than at lower IEC values (<0.44 meq/g). Moreover, when Css =
0.01 M, Ks values plateau at higher IECs (cf. Fig. 4). Thus, the initial
decrease in salt permeability coefficient with increasing IEC is primarily
caused by the initial, strong decrease in salt sorption coefficient, since
the diffusion coefficient changes little at low IEC values. Salt perme­
ability coefficients then reach a minimum and begin to increase with
increasing IEC (i.e., >0.44 meq/g), due to relatively constant Ks values
and stronger increases in the salt diffusion coefficient as water content
increases more significantly with increasing IEC. The dependence of
experimental salt permeability coefficients on fixed charge concentra­
tion (i.e., mol of fixed charge groups per liter of water sorbed in the
membrane) can be found in the Supporting Information.
N. Yan et al. Journal of Membrane Science 653 (2022) 120549

In contrast, when Css = 1 M, salt permeability coefficients increase

monotonically with increasing IEC. The IEC dependence of salt perme­
ability coefficient at high Css is mainly governed by the salt diffusion
coefficient, since salt solubility in the membrane is relatively indepen­
dent of IEC at high Css [cf. Fig. 4]. Thus, < Ps > increases essentially in
m*
the same fashion as < Ds > with increasing IEC values when Css = 1 M.
Generally, the salt permeability behavior is described reasonably well by
predictions using the Mackie/Meares and Donnan/Manning models for
higher IEC values, though substantial discrepancies are observed in the
weakly charged samples, due to a failure in the Donnan/Manning model
to describe sorption coefficients in these materials as discussed else­
where [33].

4.6. Membrane ionic conductivity

Membrane ionic conductivity, κ, depends on ion concentrations and

diffusion coefficients in the membrane, as shown in Eqn. (12). Fig. 12 (a)
presents membrane ionic conductivity values as a function of NaCl
content in the external solution. Generally, the dependence of κ on Css is
qualitatively similar to that of the sorbed Na+ concentration in the
membrane (cf. Fig. 5). For a CEM is equilibrated with a relatively dilute
NaCl solution, Cm,p m,p
+ ≫C− , so Eqn. (12) simplifies to Ref. [32]:

F2 m m,p Fig. 13. Experimental and theoretical membrane ionic conductivity in mem­
κ≈ D C (19) branes prepared with PEGDA of n = 10. The dashed line was drawn to guide the
RT + +
eye. The uncertainty of κ was determined as the standard deviation from
Thus, κ primarily depends on the counter-ion concentrations and measurements made on at least six separate samples.
diffusion coefficients. The counter-ions are more numerous in the sam­
ples than the co-ions, so they carry most of the current [12]. As Css in­
weakly charged samples (i.e., IEC <0.05), since the Cl− concentrations
creases from 0.01 to 1 M, Cm,p+ is relatively constant in the most highly in such materials cannot be accurately predicted by the Donnan/Man­
charged membrane (i.e., IEC = 1.93 meq/g), as shown in Fig. 5. Dm + of ning model, which was designed for highly charged materials [33].
this sample also changes little with Css due to relatively constant water Minor discrepancies are also found in highly charged samples (i.e., IEC
content in the membrane [cf. Fig. 7 (a)]. Consequently, κ is essentially >1.46), which may be due to the effect of polymer-ion interactions on
independent of Css in the most highly charged membrane considered, as co-ion diffusion that is not captured by the Mackie and Meares model.
suggested by Eqn. (19).
Unlike the highly charged membrane, the uncharged membrane’s κ 5. Conclusions
value increases by orders of magnitude as Css increases, which is quali­
tatively similar to the trend of Cm,p s
+ with increasing Cs in this sample (cf. Ion sorption, salt permeability, and ionic conductivity in a series of
Fig. 5). As mentioned earlier, equal numbers of Na+ and Cl− sorb in the uncharged and charged samples were characterized. Ion diffusion co­
uncharged membrane to maintain electroneutrality (i.e., Cm m
+ = C− ). efficients were calculated from these data using the solution-diffusion
Thus, Eqn. (12) can be rewritten as follows: model and Nernst-Planck framework. Diffusion coefficients were
mainly affected by the sample’s water volume fraction, which is
F 2 m,p ( m )
κ= C D+ + Dm− (20) consistent with the Mackie and Meares model. Model predictions for
RT s
Na+ diffusion coefficients agree remarkably well with the experimental
where the sorbed salt concentration (Cm,p = Cm,p m,p results, using no adjustable parameters. Reasonable agreement was also
+ = C− ) and both Na
+
s
and Cl diffusion coefficients contribute to the membrane conductivity.
− found between experimental and theoretical Cl− diffusion coefficients.
Since Cm Minor deviations were observed in highly charged membranes, pre­
s of the uncharged membrane has a relatively strong dependence
on Css , as shown in Fig. 5, and ion diffusion coefficients are weakly sumably due to interactions between polymer and ions that are not
dependent on Css [cf. Fig. 6 (a) and Fig. 7 (a)], the strong dependence of κ captured by the relatively simple Mackie and Meares model.
Salt permeability coefficients depend on salt sorption and diffusion
on Css of the uncharged membrane is largely dictated by the behavior of
coefficients. The salt concentration dependence of the salt permeability
Cm,p
s .
in uncharged membranes was mainly influenced by salt diffusivity,
Motivated by Eqn. (19) and (20), experimental κ values are presented
while the behavior in charged membranes was dominated by salt solu­
in Fig. 12 (b) as a function of Na+ (i.e., counter-ion) concentration in the
bility. Increases in polymer IEC are often accompanied by a simulta­
membrane. Fig. 12 (b) shows a strong correlation between ionic con­
neous increase in fixed charge concentration and water content. When a
ductivity and Na+ concentration in almost all the membranes studied.
membrane is equilibrated with a dilute electrolyte solution, salt
Ionic conductivity values can be predicted from Eqn. (12), where Na+
permeability initially decreases as IEC increases due to the strong
and Cl− diffusion coefficients were estimated using the Mackie and
depression in salt solubility caused by Donnan exclusion. As IEC con­
Meares model, Cl− concentrations were estimated using the Donnan/
tinues to increase, salt permeability increases slightly due to increased
Manning model, and Na+ concentrations were computed using Eqn. (3).
salt diffusivity induced by higher water content in the membrane at
Membrane fixed charge concentration at each external NaCl concen­
higher IEC values. When a membrane is equilibrated with a concen­
tration was estimated using the IEC and water uptake values, as reported
trated electrolyte solution, salt permeability increases monotonically
elsewhere [42]. Fig. 13 shows the predicted and measured ionic con­
with increasing IEC, which is mainly governed by the increase of salt
ductivity results in a parity plot. Reasonably good agreement is found
diffusion coefficient as IEC increases.
between theoretical and experimental κ values, given that no adjustable
Membrane ionic conductivity depends on sorbed ion concentrations
parameters are used. Some deviations are evident in uncharged and

11
N. Yan et al.
and diffusion coefficients in the membrane. The dependence of mem­
brane ionic conductivity on external NaCl concentration correlates well
with the sorbed Na+ concentration, both for charged and uncharged
membranes. Reasonably good agreement was found between experi­
mental and predicted ionic conductivity values. Some discrepancy was
observed in uncharged and weakly charged samples, since the Donnan/
Manning model only describes ion sorption well in highly charged
sample, but fails in samples that are uncharged or weakly charged.
Author contributions
Ni Yan: Conceptualization, Methodology, Validation, Formal Anal­
ysis, Investigation, Data Curation, Writing – Original Draft, Writing –
Review and Editing, Visualization, Project Administration, Rahul Suja­
nani: Visualization, Writing – Original Draft, Writing – Review and
Editing, Jovan Kamcev: Methodology, Writing – Original Draft, Writing
– Review and Editing, Kentaro Kobayashi: Data Curation, Writing –
Original Draft, Writing – Review and Editing, Eui-soung Jang: Investi­
gation, Writing – Original Draft, Writing – Review and Editing, Donald
R. Paul: Writing – Conceptualization, Validation, Original Draft, Writing
– Review and Editing, Supervision, Benny D. Freeman: Conceptualiza­
tion, Validation, Resources, Writing – Original Draft, Writing – Review
and Editing, Supervision, Project Administration, Funding Acquisition.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Acknowledgements
This work was supported as part of the Center for Materials for Water
and Energy Systems (M-WET), an Energy Frontier Research Center
funded by the U.S. Department of Energy, Office of Science, Basic Energy
Sciences under Award #DE-SC0019272.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.memsci.2022.120549.
References
[1] P. Długołęcki, K. Nymeijer, S.J. Metz, M. Wessling, Current status of ion exchange
membranes for power generation from salinity gradients, J. Membr. Sci. 319 (1)
(2008) 214–222.
[2] G.M. Geise, D.R. Paul, B.D. Freeman, Fundamental water and salt transport
properties of polymeric materials, Prog. Polym. Sci. 39 (1) (2013) 1–42.
[3] E. Güler, R. Elizen, D.A. Vermaas, M. Saakes, K. Nijmeijer, Performance-
determining membrane properties in reverse electrodialysis, J. Membr. Sci. 446
(2013) 266–276.
[4] S. Maurya, S.-H. Shin, Y. Kim, S.-H. Moon, A review on recent developments of
anion exchange membranes for fuel cells and redox flow batteries, RSC Adv. 5 (47)
(2015) 37206–37230.
[5] A. Galama, M. Saakes, H. Bruning, H. Rijnaarts, J. Post, Seawater predesalination
with electrodialysis, Desalination 342 (2014) 61–69.
[6] J. Veerman, R.M. De Jong, M. Saakes, S.J. Metz, G.J. Harmsen, Reverse
electrodialysis: comparison of six commercial membrane pairs on the
thermodynamic efficiency and power density, J. Membr. Sci. 343 (1) (2009) 7–15.
[7] L.F. Greenlee, D.F. Lawler, B.D. Freeman, B. Marrot, P. Moulin, Reverse osmosis
desalination: water sources, technology, and today’s challenges, Water Res. 43 (9)
(2009) 2317–2348.
[8] T.Y. Cath, A.E. Childress, M. Elimelech, Forward osmosis: principles, applications,
and recent development, J. Membr. Sci. 281 (2006) 70–87.
[9] T.S. Chung, S. Zhang, K.Y. Wang, J. Su, M.M. Ling, Forward osmosis processes:
yesterday, today and tomorrow, Desalination 287 (2012) 78–81.
[10] T. Sata, Ion Exchange Membranes: Preparation, Characterization, Modification and
Application, Tokuyama Research: Tokuyama City, Japan, 2002.
[11] J. Kamcev, B.D. Freeman, Charged polymer membranes for environmental/energy
applications, Annu. Rev. Chem. Biomol. Eng. 7 (2016) 111–133.
[12] F. Helfferich, Ion Exchange, McGraw-Hill Book Co., Inc., New York, 1962.
12
Journal of Membrane Science 653 (2022) 120549
[13] H. Yasuda, C.E. Lamaze, L.D. Ikenberry, Permeability of solutes through hydrated
polymer membranes. Part I. Diffusion of sodium chloride, Makromol. Chem. 118
(1) (1968) 19–35.
[14] P. Meares, Ion exchange membranes: principles, production and processes, in: Ion
Exchange: Science and Technology, Springer, 1986, pp. 529–558.
[15] E.Y. Safronova, D.V. Golubenko, N.V. Shevlyakova, M.G. D’yakova, V.A. Tverskoi,
L. Dammak, D. Grande, A.B. Yaroslavtsev, New cation-exchange membranes based
on cross-linked sulfonated polystyrene and polyethylene for power generation
systems, J. Membr. Sci. 515 (2016) 196–203.
[16] I.A. Prikhno, E.Y. Safronova, I.A. Stenina, P.A. Yurova, A.B. Yaroslavtsev,
Dependence of the transport properties of perfluorinated sulfonated cation-
exchange membranes on ion-exchange capacity, Membr. Membr. Technol. 2
(2020) 265–271.
[17] J.G. Wijmans, R.W. Baker, The solution-diffusion model: a review, J. Membr. Sci.
107 (1995) 1–21.
[18] G.M. Geise, H.S. Lee, D.J. Miller, B.D. Freeman, J.E. McGrath, D.R. Paul, Water
purification by membranes: the role of polymer science, J. Polym. Sci. B Polym.
Phys. 48 (2010) 1685–1718.
[19] G.S. Park, J. Crank, Diffusion in Polymers, Academic Press, New York and London,
1968.
[20] G.M. Geise, L.P. Falcon, B.D. Freeman, D.R. Paul, Sodium chloride sorption in
sulfonated polymers for membrane applications, J. Membr. Sci. 423–424 (2012)
195–208.
[21] J. Kamcev, M. Galizia, F.M. Benedetti, E.-S. Jang, D.R. Paul, B.D. Freeman, G.
S. Manning, Partitioning of mobile ions between ion exchange polymers and
aqueous salt solutions: importance of counter-ion condensation, Phys. Chem.
Chem. Phys. 18 (8) (2016) 6021–6031.
[22] M. Galizia, F.M. Benedetti, D.R. Paul, B.D. Freeman, Monovalent and divalent ion
sorption in a cation exchange membrane based on cross-linked poly (p-styrene
sulfonate-co-divinylbenzene), J. Membr. Sci. 535 (2017) 132–142.
[23] J.S. Mackie, P. Meares, The diffusion of electrolytes in a cation-exchange resin
membrane. I. Theoretical, Proc. Roy. Soc. Lond.: Math. Phys. Eng. Sci. 232 (1191)
(1955) 498–509.
[24] P. Meares, Ion-exchange membranes, in: Mass Transfer and Kinetics of Ion
Exchange, Springer, 1983, pp. 329–366.
[25] J. Kamcev, D.R. Paul, G.S. Manning, B.D. Freeman, Predicting salt permeability
coefficients in highly swollen, highly charged ion exchange membranes, ACS Appl.
Mater. Interfaces 9 (4) (2017) 4044–4056.
[26] G.S. Manning, Limiting laws and counterion condensation in polyelectrolyte
solutions II. Self-diffusion of the small ions, J. Chem. Phys. 51 (3) (1969) 934–938.
[27] J. Mackie, P. Meares, The diffusion of electrolytes in a cation-exchange resin
membrane. II. Experimental, in: Proceedings of the Royal Society of London A:
Mathematical, Physical and Engineering Sciences, The Royal Society, 1955,
pp. 510–518.
[28] P. Meares, The conductivity of a cation-exchange resin, J. Polym. Sci. 20 (96)
(1956) 507–514.
[29] P. Meares, Self-diffusion coefficients of anions and cations in a cation-exchange
resin, J. Chem. Phys. 55 (1958) 273–279.
[30] P. Meares, D. Dawson, A. Sutton, J. Thain, Diffusion, conduction and convection in
synthetic polymer membranes, Ber. Bunsen Ges. Phys. Chem. 71 (8) (1967)
765–775.
[31] W. McHardy, P. Meares, A. Sutton, J. Thain, Electrical transport phenomena in a
cation-exchange membrane II. Conductance and electroosmosis, J. Colloid
Interface Sci. 29 (1) (1969) 116–128.
[32] N. Lakshminarayanaisah, Transport Phenomena in Membranes, Academic Press,
New York and London, 1969.
[33] N. Yan, R. Sujanani, J. Kamcev, M. Galizia, E.-S. Jang, D.R. Paul, B.D. Freeman,
Influence of fixed charge concentration and water uptake on ion sorption in AMPS/
PEGDA membranes, J. Membr. Sci. 644 (2022), 120171.
[34] K.S. Pitzer, G. Mayorga, Thermodynamics of electrolytes. II. Activity and osmotic
coefficients for strong electrolytes with one or both ions univalent, J. Phys. Chem.
77 (19) (1973) 2300–2308.
[35] J. Kamcev, D.R. Paul, B.D. Freeman, Ion activity coefficients in ion exchange
polymers: applicability of Manning’s counterion condensation theory,
Macromolecules 48 (21) (2015) 8011–8024.
[36] Q. Lei, K. Li, D. Bhattacharya, J. Xiao, S. Kole, Q. Zhang, J. Strzalka, J. Lawrence,
R. Kumar, C.G. Arges, Counterion condensation or lack of solvation?
Understanding the activity of ions in thin film block copolymer electrolytes,
J. Mater. Chem. 8 (31) (2020) 15962–15975.
[37] R. Sujanani, L.E. Katz, D.R. Paul, B.D. Freeman, Aqueous ion partitioning in Nafion:
applicability of Manning’s counter-ion condensation theory, J. Membr. Sci. (2021)
638.
[38] G.S. Manning, Limiting laws and counterion condensation in polyelectrolyte
solutions I. Colligative properties, J. Chem. Phys. 51 (3) (1969) 924–933.
[39] K.S. Pitzer, Thermodynamics of electrolytes. I. Theoretical basis and general
equations, J. Phys. Chem. 77 (2) (1973) 268–277.
[40] J.H. Wang, S. Miller, Tracer-diffusion in liquids. II. The self-diffusion as sodium ion
in aqueous sodium chloride Solutions1, J. Am. Chem. Soc. 74 (6) (1952)
1611–1612.
[41] J. Kamcev, D.R. Paul, G.S. Manning, B.D. Freeman, Accounting for frame of
reference and thermodynamic non-idealities when calculating salt diffusion
coefficients in ion exchange membranes, J. Membr. Sci. 537 (2017) 396–406.
[42] N. Yan, D.R. Paul, B.D. Freeman, Water and ion sorption in a series of cross-linked
AMPS/PEGDA hydrogel membranes, Polymer 146 (2018) 196–208.
[43] A.J. Bard, L.R. Faulkner, J. Leddy, C.G. Zoski, Electrochemical Methods:
Fundamentals and Applications, vol. 2, Wiley, New York, 1980.
N. Yan et al.
[44] H. Lin, B.D. Freeman, S. Kalakkunnath, D.S. Kalika, Effect of copolymer
composition, temperature, and carbon dioxide fugacity on pure-and mixed-gas
permeability in poly (ethylene glycol)-based materials: free volume interpretation,
J. Membr. Sci. 291 (1) (2007) 131–139.
[45] H. Lin, E.V. Wagner, J.S. Swinnea, B.D. Freeman, S.J. Pas, A.J. Hill,
S. Kalakkunnath, D.S. Kalika, Transport and structural characteristics of
crosslinked poly (ethylene oxide) rubbers, J. Membr. Sci. 276 (1) (2006) 145–161.
[46] A.C. Sagle, E.M. Van Wagner, H. Ju, B.D. McCloskey, B.D. Freeman, M.M. Sharma,
PEG-coated reverse osmosis membranes: desalination properties and fouling
resistance, J. Membr. Sci. 340 (1) (2009) 92–108.
[47] H. Ju, A.C. Sagle, B.D. Freeman, J.I. Mardel, A.J. Hill, Characterization of sodium
chloride and water transport in crosslinked poly (ethylene oxide) hydrogels,
J. Membr. Sci. 358 (1) (2010) 131–141.
[48] G.M. Geise, B.D. Freeman, D.R. Paul, Characterization of a novel sulfonated
pentablock copolymer for desalination applications, J. Membr. Sci. 24 (51) (2010)
5815–5822.
[49] J. Kamcev, E.-S. Jang, N. Yan, D.R. Paul, B.D. Freeman, Effect of ambient carbon
dioxide on salt permeability and sorption measurements in ion-exchange
membranes, J. Membr. Sci. 479 (2015) 55–56.
[50] J. Kamcev, R. Sujanani, E.-S. Jang, N. Yan, N. Moe, D. Paul, B. Freeman, Salt
concentration dependence of ionic conductivity in ion exchange membranes,
J. Membr. Sci. 547 (2018) 123–133.
13
Journal of Membrane Science 653 (2022) 120549
[51] J.C. Díaz, J. Kamcev, Ionic conductivity of ion-exchange membranes: measurement
techniques and salt concentration dependence, J. Membr. Sci. (2021) 618.
[52] P.J. Flory, Principles of Polymer Chemistry, Cornell University Press, 1953.
[53] F.G. Donnan, Theory of membrane equilibria and membrane potentials in the
presence of non-dialysing electrolytes. A contribution to physical-chemical
physiology, J. Membr. Sci. 100 (1995) 11.
[54] P.R. Bevington, D.K. Robinson, J.M. Blair, A.J. Mallinckrodt, S. McKay, Data
reduction and error analysis for the physical sciences, Comput. Phys. 7 (4) (1993)
415–416.
[55] L. Yuan-Hui, S. Gregory, Diffusion of ions in sea water and in deep-sea sediments,
Geochem. Cosmochim. Acta 38 (5) (1974) 703–714.
[56] M. Jardat, B. Hribar-Lee, V. Vlachy, Self-diffusion coefficients of ions in the
presence of charged obstacles, Phys. Chem. Chem. Phys. 10 (3) (2008) 449–457.
[57] M. Jardat, B. Hribar-Lee, V. Dahirel, V. Vlachy, Self-diffusion and activity
coefficients of ions in charged disordered media, J. Chem. Phys. 137 (11) (2012),
114507.
[58] R.A. Robinson, R.H. Stokes, Electrolyte Solutions, Courier Corporation, 2002.
[59] G.M. Geise, B.D. Freeman, D.R. Paul, Sodium chloride diffusion in sulfonated
polymers for membrane applications, J. Membr. Sci. 427 (2013) 186–196.