Journal of Membrane Science 520 (2016) 790–800

Contents lists available at ScienceDirect

Journal of Membrane Science

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

Modeling the water permeability and water/salt selectivity tradeoff in

polymer membranes
Huan Zhang, Geoffrey M. Geise n
Department of Chemical Engineering, University of Virginia, 102 Engineers' Way, P.O. Box 400741, Charlottesville, VA 22904, USA

art ic l e i nf o a b s t r a c t

Article history: A theoretical framework is proposed to model the observed water/salt permeability, diffusivity, and
Received 18 May 2016 sorption selectivity tradeoff relationships with water permeability, diffusion, and sorption coefficients.
Received in revised form Salt sorption is modeled using an electrostatic (or dielectric) exclusion approach. Water and salt diffusion
19 August 2016
are modeled using hindered diffusion and free volume theories. The modeling approach suggested here
Accepted 20 August 2016
Available online 24 August 2016
is compared to a set of water and salt transport property data measured using a wide variety of polymers.
The models describe the general trends observed in the tradeoff relationships but also situations where
Keywords: penetrant-polymer interactions and/or other system specific effects are likely significant. The approach
Upper bound also highlights the importance of convective frame of reference and thermodynamic correction factors in
Desalination
the analysis of water diffusion in hydrated polymers. These corrections are discussed within the fra-
Diffusion
mework of experimental data from a variety of hydrated polymer systems. The expanded tradeoff re-
Sorption
Transport lationship data set and theoretical framework help to expand insight into the fundamental phenomena
that govern water and salt transport in polymer membranes for desalination applications.
& 2016 Elsevier B.V. All rights reserved.

1. Introduction to seawater desalination by RO.

Growing demand for purified water resources and clean energy
coupled with interest in the nexus of food, energy, and water re-
sources has driven significant interest in membrane-based tech-
nologies that could help address global need for water and energy
[1–10]. Membrane-based technologies, such as reverse osmosis
(RO), currently dominate the desalination market, and increases in
energy efficiency over several decades have positioned mem-
branes to play an increasingly important role in addressing global
water needs [9,11,12]. Furthermore, emerging energy technolo-
gies, such as pressure-retarded osmosis (PRO), are of interest for
recovering energy, stored as chemical potential, in solutions of
different salt concentration [13–18]. Accordingly, there is sig-
nificant interest in optimizing, preferably inexpensive, polymer
membranes for these applications.
Seawater desalination by reverse osmosis is a relatively energy
efficient process [9,19–21]. The energy required to desalinate
seawater is roughly within a factor of two of the thermodynamic
minimum [9]. This efficiency is largely a result of improvements in
membranes over several decades and pressure recovery devices
that can be incorporated into the process [9,11,22]. On the other
hand, brackish water RO is typically less energy efficient compared
n
Corresponding author.
E-mail address: geise@virginia.edu (G.M. Geise).
Some significant challenges still face membrane desalination
technologies. The polyamide-based materials that are commonly
used as the active layers of reverse osmosis and nanofiltration
membranes are inherently susceptible to degradation by chlorine-
containing compounds that are widely used to disinfect water and
control membrane biofouling [1,23–27]. Additionally, zero-liquid
discharge requirements are subjecting membranes to concentra-
tions of salt that exceed traditional design parameters [19,28,29].
Increasingly contaminated waters that contain a variety of ions
and other pollutants also challenge membrane materials that have
traditionally been designed to reject sodium chloride [1,24,30–36].
Finally, interest in designing membrane polymers that are both
highly selective and simultaneously resistant to fouling presents a
significant engineering challenge to the membrane field [29]. In
many cases, the literature provides insufficient guidance on the
impact of polymer membrane chemistry and structure on the
fundamental transport properties that are critical for desalination
system performance.
One approach to benchmarking material properties and the
influence of polymer chemical and/or structural modification on
transport properties is to consider the impact of those modifica-
tions on established water/salt permeability, sorption, and/or dif-
fusivity selectivity tradeoff relationships [37]. For a wide range of
polymers, a tradeoff relationship is observed whereby materials
with higher water/salt permeability selectivity (i.e., the ratio of
water permeability to salt permeability) tend to have lower water

http://dx.doi.org/10.1016/j.memsci.2016.08.035
0376-7388/& 2016 Elsevier B.V. All rights reserved.
 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800
Fig. 1. Water/salt permeability selectivity plotted versus diffusive water perme-
ability for a series of reported polymer membrane transport property data, char-
acterized using sodium chloride [23,37,46–58].
permeability and vice versa (Fig. 1). The water/salt permeability
selectivity tradeoff with water permeability can be split into
sorption and diffusivity tradeoff relationships because perme-
ability is defined, in these materials, as the product of a sorption
coefficient and an average diffusion coefficient [1,24,37–40]. These
empirical tradeoff relationships allow for comparison of a variety
of materials based on fundamental transport properties (i.e., per-
meability and selectivity as opposed to flux and rejection [1]). As
has been recognized in other areas of the membrane field [41–45],
they can be used to illustrate the current state-of-the-art and/or
emphasize needed property improvements [10].
One limitation of the fundamental tradeoff relationships for
desalination membranes is that the intrinsic water and salt
transport properties, e.g., permeability, sorption, and diffusion
coefficients, must be known. Accurate measurement of these
properties is difficult for heterogeneous thin film composite
membranes such as the polyamide materials that are used in
commercial desalination processes [37,59], and this situation, in
many cases, frustrates efforts to establish fundamental structure-
property relationships in such materials. Tradeoff relationships,
however, can also be developed based on water and salt per-
meance data that are available for such materials [10,37,60].
While the empirical water permeability and water/salt se-
lectivity tradeoff relationship, introduced in 2011 [37], provides a
starting point for benchmarking polymers that are of interest for
desalination applications, a theoretical framework would be
helpful in further understanding the fundamental origins of the
tradeoff relationship. Here we incorporate additional data into the
tradeoff relationship (relative to the initial report). Additionally,
we incorporate theory to identify the contributions of electrostatic
interactions, hindered diffusion, and free volume transport in
polymers to the tradeoff relationships. Finally, we seek to gain
insight into the fundamental phenomena that result in the tradeoff
relationship.
2. Theoretical background
Transport of water and salt through polymeric materials that
are of interest for desalination applications is typically described
using the solution-diffusion model where penetrant sorption at
791
the upstream face of the membrane precedes diffusion down a
concentration gradient and is followed by desorption at the
downstream face of the membrane [39,40]. The solution-diffusion
model describes the overall permeability of a polymer to a given
penetrant in terms of sorption and diffusive contributions to
transport:
Pi = Ki × Di (1)
where Pi is permeability, Ki is the sorption coefficient, and Di is the
diffusion coefficient of component i in the polymer [39,40]. Eq. (1)
follows from analysis of the governing flux equations [1,3,24,40].
While the separation of permeability into sorption and diffusion
components using Eq. (1) appears relatively straightforward, de-
scribing the fundamental underpinnings of the sorption and dif-
fusive properties and their connections to polymer chemistry and/
or structure is often much more challenging [1,24,37].
The intrinsic permeability selectivity of a material can be de-
fined as a ratio of permeability values. For desalination, selectivity
is often defined as the ratio of the water permeability to the salt
permeability, and as shown in Eq. (1), permeability can be split
into sorption and diffusion components [37]. The water/salt per-
meability selectivity and corresponding sorption and diffusivity
selectivity expressions are related as:
Pw K D
= w × w
Ps Ks Ds (2)
Permeability selectivity can be determined by independently
measuring water and salt permeability values (i.e., the so-called
ideal selectivity), or the selectivity can be determined from si-
multaneous water and salt transport experiments. The latter can
be particularly important when osmotic de-swelling or other ion-
polymer interactions influence the water content and, thus, the
water permeability of the polymer [61]. Presently, the tradeoff
relationships do not distinguish between these experimental
methods.
2.1. The sorption component of transport
The sorption coefficient can be defined, at equilibrium, as the
ratio of the concentration of a component i in the polymer
membrane, cim , to the concentration of component i in the external
solution, cis [39,40]:
cim
Ki ≡
cis (3)
For desalination applications, sorption coefficients are typically
defined for water, Kw, and either individual ions (e.g., K þ or K  ) or
overall salt, Ks [1,24,37]. The water sorption coefficient is often
approximately equal to the volume fraction of water sorbed in the
polymer, and it is convenient, and often appropriate, to use Kw and
the volume fraction of water interchangeably [1,62]. For salt
sorption in hydrated polymers, the concentrations in Eq. (3) are
typically defined as mol of i per kg or L of water sorbed in the
polymer [63].
The concentrations used to define the salt sorption coefficient,
Ks, depend on the nature of the electrolyte and whether the
polymer is uncharged (e.g., poly(ethylene oxide)) or charged (e.g.,
cation or anion exchange materials) [1,3,24,63–65]. The salt sorp-
tion coefficient for binary monovalent electrolyte sorption in an
uncharged polymer can be calculated using either the cation or
anion concentrations [1,3,65]. In charged polymers, however, the
co-ion concentrations (i.e., anions in cation exchange materials
and cations in anion exchange materials) are used to define the
salt sorption coefficient, Ks [1,3,65]. Here we will focus on the
overall salt sorption and water sorption coefficients, Ks and Kw,
792 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800

respectively.
The sorption coefficient describes the thermodynamic parti-
tioning of a compound from the external solution into the mem-
brane phase. Therefore, the sorption coefficient can be related to
the free energy change associated with moving the component
from one state (e.g., solution) to another (e.g., the membrane
phase):

cim ⎡ ΔGi, sorption ⎤

Ki ≡ ⎢
s = exp −
⎥
ci ⎣ RT ⎦ (4)

where ΔGi, sorption is the free energy change associated with moving
component i from the external solution into the membrane phase,
R is the gas constant, and T is the absolute temperature [66]. The
relationship between the sorption coefficient and the right-hand
side of Eq. (4) is predicated on the assumption of solution ideality
as the activity coefficients have been set equal to unity. Ad-
ditionally, Eq. (4) is analogous to expressions for the Donnan po-
tential that is often discussed within the framework of ion sorp-
tion in charged polymers [63]. Conversion of Eq. (4) into the
equation typically used to define the Donnan potential is accom-
plished by setting ΔGi, sorption = ziFEDonnan where zi is the charge Fig. 2. Static dielectric constants, measured at 298 K, for solutions of water and
either dimethyl sulfoxide (DMSO, ) [76] or 1-methylimidazole (MeIM, ) [75] as
number of ion i, F is Faraday's constant, and EDonnan is the Donnan
a function of the mole fraction of water in the solution. The static dielectric con-
potential (note that the activity coefficients are still assumed to be stants of pure DMSO, MeIM, and water were reported as 47.29, 39.78, and 78.89,
taken as unity) [63,67]. In desalination applications, Ki is typically respectively [75,76].
less than unity [1] suggesting that ΔGi, sorption is positive and re-
presents the free energy barrier to sorption.
The ΔGi, sorption term in Eq. (4) should include contributions from is not necessarily clear. The relative permittivity of solutions of
all of the phenomena that influence the sorption process. For the organic molecules and water can exhibit linear correlations be-
purpose of describing ion sorption, we have chosen to consider a tween the static dielectric constant (relative permittivity) and
simple case where ΔGi, sorption is taken as the solvation energy
water composition (Fig. 2) [75,76]. While the DMSO/water system
barrier, ΔWi (i.e., the change in free energy associated with the data exhibit a linear transition from the static dielectric constant of
change in the solvation environment of the ion between the so-
DMSO to that of water, the data for the 1-methylimidazole
lution and membrane phases). This approach is often used to de-
(MeIm)/water system suggest that a highly non-linear region ex-
scribe dielectric exclusion of ions from hydrated polymers [68–70].
ists between 0.9 and 1.0 mol fraction of water. Other binary solu-
This choice neglects other phenomena that may contribute (in
tions of organic solvents exhibit non-linear static dielectric con-
some cases perhaps significantly) to the sorption free energy, but
stant correlations with composition, suggesting that different in-
as will be discussed later, this simple approach appears to be ef-
fective for describing the ion sorption in a collection of different termolecular interactions can influence the relationship between
polymers. composition and the static dielectric constant of the solution
The solvation energy barrier, ΔWi, can be calculated using the [77,78].
Born model: Relative permittivity data as a function of polymer water con-
tent are not common in the literature, but data were reported for
zi2e2 ⎛ 1 1 ⎞ Nafion™ 117, a perfluorosulfonic acid polymer (Fig. 3) [79]. As it is
ΔWi = ⎜ − ⎟
8πε0ai ⎝ εm εsol ⎠ (5) very difficult to dry Nafion™ completely, the relative permittivity
of Teflon™ was reported as the zero water content material be-
where e is the elementary charge, ε0 is the permittivity of free
cause Nafion™ and Teflon™ share a common backbone [79]. The
space, εm is the relative permittivity of the membrane, εsol is the
Nafion™ 117 relative permittivity data (Fig. 3) correlate linearly
dielectric constant of the external solution (taken equal to 80 for
aqueous electrolytes [71]), and ai is the bare ion radius [72]. with water content in a manner that is bounded by the dry and
The salt sorption coefficient can be written in terms of ΔWi as: pure water limits similar to the DMSO/water system (Fig. 2).
Based on the linear data observed for the DMSO/water and
csm ⎡ ΔWs ⎤ ⎡ zi2e2 ⎛ 1 1 ⎞⎤ Nafion™ 117/water systems, we suggest that a linear relationship
Ks ≡ s = exp⎢− ⎥ = exp⎢− ⎜ − ⎟⎥
cs ⎣ kT ⎦ ⎣ 8πkTε0as ⎝ εm εsol ⎠⎦ (6) between εm and the volume fraction of water sorbed in the
polymer (effectively equal to Kw) may be reasonable in the absence
where k is Boltzmann's constant. In order to apply Eq. (6), values of more general data or specific guidance from models. The re-
are needed for as and εm. To describe overall salt sorption in
lative permittivity of many dry hydrocarbon polymers fall in the
polymers, as was taken as 0.194 nm, which is the average of the
range of 2-6 (the value for Teflon™ is 2 [79]), so using the linear
cavity radii for sodium (0.151 nm) and chloride (0.236 nm) ions
regression from Fig. 3 may be a reasonable assumption to arrive at
[73]. The cavity radius is defined as the distance from the center of
a general empirical relationship between εm and Kw:
the ion to the point where the dielectric permittivity changes from
the vacuum value to that of the solvent (water) in continuum εm = 1.04 + ( 78.3 × K w ) (7)
models [73,74].
Determining values for εm can be challenging given a lack of Using Eq. (7) to link the relative permittivity of hydrated
permittivity data in the literature for hydrated polymers. One polymers to water sorption, it is possible to construct a model for
would expect the value of εm to increase as the water content of the variation of the salt sorption coefficient with the water sorp-
the polymer increases [1], but the functional form of this increase tion coefficient:
 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800
Fig. 3. Static dielectric constant data measured at 303 K for Teflon™ and Nafion™
117 as a function of the volume fraction of water sorbed in the polymer [79]. The
static dielectric constant of pure water is also included [71]. The dashed line was
determined by linear regression of the data.
⎡ The version of the water flux equation shown in Eq. (11) in-
csm zi2e2 ⎛ 1 1 ⎞⎤⎥
Ks ≡ = exp⎢− ⎜ − ⎟ cludes two key assumptions that distinguish it from Eq. (10). First,
css ⎢⎣ 8πkTε0as ⎝ 1.04 + (78.3 × K w ) εsol ⎠⎥⎦
Since all of the parameters other than Kw in Eq. (8) are fixed,
the model can be used to predict the water/sorption selectivity
(Kw/Ks) as a function of Kw, which can then be compared to ex-
perimental data on the sorption selectivity tradeoff plot.
2.2. Constitutive equations for diffusion
In order to determine water and salt diffusivities for hydrated
polymer membranes, constitutive equations are needed to relate
the experimentally measured fluxes (typically salt flux and water
flux) to the external driving forces for transport. These constitutive
equations have been developed for water and salt transport in
hydrated membranes [24,40].
Salt flux, ns, is related to an external salt concentration driving
force by:
Δcss Δc s
ns = KsDs = Ps s = BΔcss
L L
where L is the membrane thickness, Δcss
is the salt concentration
difference between the external salt solutions on either side of the
membrane, Ps is the salt permeability, and B is the salt permeance
[1,24,40]. Eq. (9) can be derived from ion flux expressions using
the Nernst-Planck equation and is equivalent to the expression
obtained from Fick's Law [24]. It is important to note that the salt
diffusivity, Ds, in Eq. (9) is an average over the thickness of the
membrane, and this average is defined during the integration of
the differential salt flux equation that results during analysis
[1,24]. Additionally, the dependence of Ds on ion diffusivities (i.e.,
D þ and D  ) and ion concentrations, which is important for
charged polymers, can be described using the Nernst-Planck
analysis [1,24].
Water flux, nw, is related to an applied pressure difference,
using a thermodynamic approach, as:
793
⎛ ⎞¯
Dw K w ⎜ ∂ ln K w ⎟ Vw Δp − Δπ
⎟ RT ( )
nw =
L( 1 − K w ) ⎜⎝ ∂ ln a wm m
aw ⎠
0
Pw V¯
= δ w ( Δp − Δπ ) = A( Δp − Δπ )
L( 1 − K w ) RT (10)
where a wm
is the activity of water in the membrane, a wm0is the
activity of water in the membrane at the upstream (i.e., high
pressure) face, V̄w is the partial molar volume of water, R is the gas
constant, T is absolute temperature, Δp is the applied pressure
difference, Δπ is the osmotic pressure difference across the
membrane, Pw is the diffusive water permeability, and A is the
water permeance [1,37]. The thermodynamic correction factor, δ,
relates the activity of water in the membrane to the water sorption
coefficient and is evaluated at the upstream face of the membrane
[1,80,81]. Eq. (10) can be derived from a Maxwell-Stefan multi-
component transport analysis, and much like the salt diffusivity,
the water diffusivity, Dw, in Eq. (10) is averaged over the thickness
of the membrane during integration of the differential water flux
equation [1].
Eq. (10) has been simplified for use with low water content
polymers, and much of the early desalination literature uses this
simplification [1,39,40]:
Pw V¯w
nw =
L RT
( Δp − Δπ ) = A( Δp − Δπ ) (11)
(8) the water content of the polymer (taken as Kw) is assumed to be
small, such that Kw { 1, and this assumption removes the con-
vective frame of reference term, (1 Kw), in Eq. (10). Second, the
thermodynamic factor, δ, is assumed to be equal to unity. While
these assumptions may be reasonable for some low water content
polymers, they generally fail for polymers where Kw is not very
small compared to unity [1].
Including the convective frame of reference term, (1 Kw) in Eq.
(10), is relatively straightforward when relating a measured water
flux to the average water diffusivity because the water content of
the polymer is usually known or easily obtained. The thermo-
dynamic correction factor, however, requires additional informa-
tion. Either water sorption data are needed as a function of water
activity (i.e., a sorption isotherm) or a suitable equation of state
(e.g., Flory-Huggins or Flory-Rehner theories [82,83]) is needed to
evaluate δ [1]. Both the convective frame of reference and ther-
modynamic correction factors are often significant for highly
swollen polymers, but the thermodynamic correction factor is
more significant than the convective frame of reference term in
some polymers [1].
(9)
While rigorous treatment of δ often requires water sorption
data measured as a function of water activity, which is often not
available for many materials reported in the literature, one can
approximate the significance of the thermodynamic correction
factor in Eq. (10) by performing a one point fit of the Flory-Huggins
model using pure water sorption data that is often reported to-
gether with water flux measurements [84]. In this case, the Flory-
Huggins interaction parameter is calculated from the one point fit,
and then, this interaction parameter is used in the model to cal-
culate δ [1,84]. While this approach is limited in rigor, it can be
used to highlight the influence of the thermodynamic correction
factor on the interpretation of the average water diffusivity.
Reported water diffusivity data for a variety of hydrated poly-
mers are shown in Fig. 4 [37,48–50,52,54,57,85–95]. The influence
of the convective frame of reference term and the thermodynamic
correction factor (in most cases estimated using the one point
Flory-Huggins fit described above) is highlighted. The dashed line
794 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800
Fig. 4. Reported water transport data for a variety of polymers [37,48–
50,52,54,57,85–95]. Values of Dw calculated using Eq. (11) ( ) are compared to
those values calculated using Eq. (10) with δ ¼1 ( ) and those values calculated
using Eq. (10) including thermodynamic non-ideality estimated, where necessary,
by a one point Flory-Huggins fit (◆). The thermodynamic correction factor appears
to be more significant (i.e., compare ◆ and ) than the convective frame of re-
ference term (i.e., compare and ), particularly at high values of Kw. The dashed
line represents the self-diffusion coefficient of water, Dw°
¼ 2.8  10  5 cm2/s [96].
in Fig. 4 represents the self-diffusion coefficient of water
( Dw° ¼ 2.8  10  5 cm2/s [96]). Values of Dw calculated using Eq. (11)
from data measured on highly swollen polymers often exceed the
self-diffusion coefficient of water. This situation, in some cases, has
been ascribed to hydraulic (e.g., viscous) flow in the membrane.
Properly accounting for the convective frame of reference term
and approximating the thermodynamic correction term (i.e., using
Eq. (10)), however, results in values of Dw that fall satisfactorily
below the self-diffusion coefficient of water suggesting that, as
expected, the polymer membrane slows water diffusion relative to
the rate of water diffusion in bulk water.
2.3. Diffusivity selectivity
To model the water/salt diffusivity selectivity as a function of
water diffusivity, expressions are needed to relate Dw and Ds to
material properties. Diffusion in polymers is often modeled as
either an activated or hindered transport process, and several re-
lationships have been reported with varying degrees of generality
and/or assumptions. Here we seek to avoid specific structural or
morphological assumptions in an effort to capture the transport
properties of a variety of different polymers.
2.3.1. Water diffusivity
Based on the experimental data shown in Fig. 4, one would
expect the water diffusivity to increase with increasing water
sorption. This relationship can be captured using free volume
theory to describe diffusive transport in hydrated polymers. An
approach used by Yasuda et al., assumes that the free volume
available for transport in the swollen polymer, VF, can be de-
termined using volume additivity (recall that Kw is effectively
equivalent to the volume fraction of water sorbed in the polymer):
VF = K wVF , w + ( 1 − K w )VF , p (12)
where VF,w is the free volume of pure water and VF,p is the free
volume of the dry polymer [57]. This approach can be combined
with an activated state model for diffusion using the Cohen and
Turnbull approach to arrive at a relationship between the diffusion
coefficient of water and the water content of the polymer:
⎡
Dw = Dw° exp⎢−
⎢ β ( 1
Kw)(1 − α) ⎤⎥⎥
−1
1 + α( −1) ⎥⎦
1
⎢⎣ Kw (13)
α = VF , p/VF , w (14)
β = V */VF , w (15)
where Dw°is the self-diffusion coefficient of water (taken as
5
2.8  10 cm2/s [96]) and V* is a characteristic volume parameter
that is related to the cross-section and the diffusional jump length
of the diffusing penetrant [57,97,98].
2.3.2. Salt diffusivity
Salt diffusivity, like water diffusivity, is expected to generally
increase as the water content of the polymer increases
[1,24,46,58,99–101]. Two relatively simple theories explain this
phenomenological observation. The theory of Mackie and Meares
accounts for the effect of area and tortuosity on diffusive transport
(i.e., the polymer occupies space and hinders transport) using a
random walk approach [99,102]. Alternatively, the theory of Ya-
suda is based on a free volume approach where the free volume is
taken to be proportional to the water content of the polymer (si-
milar to that described above for the water diffusion coefficient)
[58]. In both cases, the diffusion coefficient of salt in the polymer is
expected to approach the diffusion coefficient of salt in bulk
aqueous solution as the volume fraction of water approaches unity
(i.e., the polymer acts to slow the diffusion process relative to
diffusion in solution). Here we consider sodium chloride transport,
so the diffusion coefficient of NaCl in bulk aqueous solution is
taken as 1.5  10  5 cm2/s at 25 °C [103].
The Mackie and Meares model assumes that the swollen
polymer is generally homogeneous [99]. Additionally, the model
assumes that the swollen polymer contains an internal solution
that is intermixed with a polymer matrix and that the character-
istic length that separates polymer junctions (e.g., cross-links or
entanglements) is large compared to the size of the diffusing ions.
In this theory, the presence of the polymer acts to reduce the
cross-sectional area available for transport and introduces tortu-
osity [99]. Both factors slow diffusion relative to bulk solution
properties, and increases in polymer water content reduce the
effect of this transport resistance. The model depends only on the
volume fraction of water sorbed in the polymer and the bulk so-
lution diffusion coefficient:
⎛ K ⎞2
w
Ds = Ds°⎜ ⎟
⎝ 2 − Kw ⎠ (16)
where Ds° ¼ 1.5  10  5 2
cm /s [103]. The model ensures that the salt
diffusion coefficient in the membrane approaches the bulk solu-
tion salt diffusion coefficient as Kw approaches unity.
The Yasuda et al. theory is based on free volume analysis of
diffusion and the assumption that the polymer free volume is
proportional to water content, taken in this case as Kw [58]:
⎡ b ⎤
Ds = a × exp⎢ − ⎥
⎣ Kw ⎦ (17)
where a and b are adjustable constants. The constant a can be
constrained such that the model ensures that Ds will approach Ds°
as Kw approaches unity, which is a physically desirable result:
 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800
a = Ds° × exp⎡⎣ b⎤⎦
such that the model reduces to a single adjustable parameter
model:
⎡ ⎛ 1 ⎞⎤
Ds = Ds° × exp⎢ b⎜ 1 − ⎟⎥
⎣ ⎝ K w ⎠⎦
The constant b is related to the size of the diffusing penetrant
[58]. The constant b (and a if considering Eq. (17)) can be regressed
from salt diffusion coefficient data as a function of water content.
3. Results and discussion
Water and sodium chloride transport data [23,37,46–58] were
compiled and used to evaluate the effectiveness of the models
described in the previous section to describe the water and salt
transport property tradeoff relationships. The data set contains a
variety of different polymers including materials based on aro-
matic, acrylic, ethoxy, cellulosic, and hydrocarbon backbones.
Some of the materials contain fixed charge groups (e.g., cation or
anion exchange materials) while other materials are hydrophilic
uncharged polymers. Also, while most studies were conducted at
ambient temperature, the salt solution concentrations used to
measure salt permeability and the use of salt solution or pure
water to characterize water permeability vary to some extent
across the different studies reported in the literature. The goal of
this analysis is to identify the general models that capture trends
for a diverse set of materials and highlight opportunities and need
for developing specific structure property relationships for differ-
ent classes and categories of polymers. The model predictions for
Dw and Ds as a function of Kw are discussed below within the
context of the literature data set, and the models are combined
and compared to the tradeoff relationships.
3.1. Water diffusion
Yasuda et al. determined values for the parameters α ¼0.5 and
β ¼ 4.5 from water transport data measured using a series of ac-
rylic, methacrylic, and cellulosic polymers [57]. The model de-
scribed in Eqs. (13)–(15) qualitatively describes the increase of Dw
with Kw (Fig. 5) for the literature data set (where Dw was analyzed
using Eq. (10) and the thermodynamic correction procedure de-
scribed above). The scatter in Fig. 5 should be expected given that
data from a wide range of polymers are considered and that the
thermodynamic correction factor was approximated for many of
the data points using a one-point Flory-Huggins model fit.
It is instructive to consider the implications of the parameter
values reported by Yasuda, et al. [57]. It is reasonable that β 4 1
given that a characteristic volume greater than that of the free
volume of water might be needed for water transport. The value of
α ¼0.5, however, suggests that generally the free volume of water
is twice that of the dry polymer. Free volume data collected using
positron annihilation lifetime spectroscopy (PALS) data suggest
that this situation might be true for some polymers but not for
others.
PALS data (measured ortho-positronium lifetime and intensity)
can be used to determine the fractional free volume in a material
using the Tao-Eldrup approach [104,105]. The fractional free vo-
lume of bulk water is reported to be 0.041 [106]. PALS data mea-
sured using dry cellulose ester films and analyzed using the same
approach result in fractional free volumes that range from 0.013 to
0.020 [88]. These results align relatively well with the α ¼0.5
parameter discussed above as they suggest that the free volume of
bulk water may be approximately a factor of two greater than that
795
(18)
(19)
Fig. 5. Water diffusion coefficient data (determined using Eq. (10) and the ther-
modynamic correction factor approximation described above) versus water sorp-
tion coefficient [23,37,46–58]. The solid line corresponds to the model described in
Eq. (13) with α ¼0.5 and β ¼ 4.5 [57].
of the dry polymer. Alternatively, PALS measurements made on
dry sulfonated polysulfone materials result in fractional free vo-
lumes ranging from 0.030 to 0.034 [84]. Considering that ortho-
positronium inhibition processes in sulfonated polymers are be-
lieved to result in under-predictions of fractional free volume
[107,108], the sulfonated polysulfone results may suggest that the
dry polymer fractional free volume of some materials is close to
the fractional free volume of water, which would suggest a value of
α closer to unity. For the purpose of this work, the Yasuda et al.
parameters were used as reported [57].
3.2. Salt diffusion
The models in Eqs. (16), (17), and (19) are compared to ex-
perimental salt diffusion and water content data obtained from
the literature (Fig. 6). As both models predict Ds-0 as Kw-0, it is
expected that the models will break down at low water content,
and the increased scatter in the data at low water content further
suggest that other phenomena, such as specific polymer-penetrant
interactions may become more significant at low water content.
The data are reasonably well bounded by the Mackie and Meares
model (Eq. (16)), which tends to over predict Ds at moderate water
content, and the single parameter free volume model (Eq. (19)),
which tends to under predict Ds at moderate water content.
The fit of the 2-parameter free volume-based model (Eq. (17))
appears to capture the moderate water content Ds data better than
the other two models. The model, however, fails to capture the
physically significant result where the salt diffusion coefficient is
expected to approach its bulk solution value as Kw-1. Therefore,
the Mackie and Meares model (Eq. (16)), which contains no ad-
justable parameters, and the single parameter free volume model
(Eq. (19)) are preferable to Eq. (17) based on agreement with the
high water content limit. The apparent improved agreement be-
tween the data and the free volume model compared to the
Mackie and Meares model (at least at moderate to low water
content) is consistent with a recent report suggesting that free
volume transport theory may be more applicable compared to
hindered transport theory for describing diffusion in some com-
mercially available desalination membranes [109].
796 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800
Fig. 6. Salt diffusion coefficient data plotted versus water sorption coefficient for
several polymers [23,37,46–58]. Model results are shown for the Mackie and
Meares model (Eq. (16)), the 2-parameter Yasuda model (Eq. (17), a¼ 2.0( 70.5) 
10  5 cm2/s and b¼ 1.17 70.13), and the single parameter Yasuda model (Eq. (19),
b ¼2.39 7 0.15). The parameters and their uncertainties for the Yasuda model were
regressed using OriginPro software.
3.3. Modeling the water/salt tradeoff relationships
The modeling framework described in the previous sections
was applied to the water/salt sorption, diffusivity, and perme-
ability selectivity tradeoff relationships with water sorption, dif-
fusion, and permeability coefficients. The application of these
models to the tradeoff plots introduces a theoretical framework for
understanding the phenomena that result in the tradeoff re-
lationships for sorption, diffusion, and permeability. The approach
employed here is a refinement of the empirical relationships used
previously [37].
The Born model-based approach to modeling the water/salt
sorption selectivity versus water sorption coefficient tradeoff re-
lationship describes the high selectivity-high Kw frontier of the
experimental data remarkably well given the relative simplicity of
the model and the nature of the dielectric constant treatment that
was used in the model (Fig. 7). It is reasonable to expect that the
Born model might predict water/salt selectivity values that are
high compared to the majority of the data as the Born model has
been suggested to over-predict the energy barrier for sorption
[72]. Such an over-prediction would result in lower salt sorption
coefficients and ultimately higher water/salt sorption selectivity.
While a few experimental data points meet or exceed the model
predictions, the majority of the polymers considered exhibit wa-
ter/salt sorption selectivity values that are below the result pre-
dicted based strictly on electrostatics.
The potential influence of polymer chemistry on ion sorption
selectivity is highlighted by the wide range of water/salt sorption
selectivity values that can be experimentally accessed at a given
polymer water content. The data set in Fig. 7 contains a mix of
uncharged hydrophilic polymers and charged (i.e., ion exchange)
materials where additional ion exclusion mechanisms, such as
Donnan exclusion and/or specific ion-polymer interactions, may
be quite significant. While considerable work has been done to
understand fundamental ion exclusion mechanisms such as di-
electric and Donnan exclusion [53,64,65,68,70,101,112–130], many
questions still remain about the interplay of these mechanisms in
a wide range of polymers and polymer water content. The mod-
eling approach used here also assumed that the hydrated polymer
Fig. 7. Water/salt sorption selectivity versus water sorption coefficient for a series
of reported data [37,46,50–56,110,111]. The single circle point represents an aqu-
eous sodium chloride solution, and the line was calculated using Eq. (8) to de-
termine Ks (and, thus, Kw/Ks) as a function of Kw.
membrane could be considered to be a homogeneous dielectric,
and this assumption may not be appropriate for all materials
[69,70]. Additionally, all of the data presented here were measured
using sodium chloride as the electrolyte. While sorption data are
reported for other electrolytes, additional systematic sorption
studies with other electrolytes would likely highlight the balance
of the importance of different exclusion mechanisms and may give
rise to additional insight into the interplay of water content and
sorption selectivity.
The diffusion selectivity data reported in 2011 [37] were ana-
lyzed using Eq. (11), and this method of analysis resulted in some
data where the calculated value of Dw exceeded the self-diffusion
coefficient of water. In Fig. 8, those data (shown as diamonds)
were adjusted for the frame of reference and thermodynamic
correction factor terms discussed in relation to Eq. (10), and the
resulting points are shown as squares. Satisfactorily, the values of
Dw are now found to be less than the self-diffusion coefficient of
water (indicated by the single circle symbol). Perhaps due to the
nature of the approximation used to determine the thermo-
dynamic correction factor, some diffusion selectivity values fall
below unity, which generally is not expected for transport of hy-
drated ions in hydrated polymers as the hydrated ions are ex-
pected to be larger and, thus, diffuse slower than water.
The use of Eq. (10) to correct for the convective frame of re-
ference and thermodynamic terms in the salt flux equation,
however, allows for comparison of experimental data and the
diffusion models described above. The free volume-based model
for water transport (Eqs. (13)–(15)) was used to calculate Dw as a
function of Kw, and either the Mackie and Meares model (Eq. (16))
or the free volume model (Eq. (19)) were used to calculate Ds as a
function of Kw. The model for Dw was combined with each model
for Ds to produce the curves shown in Fig. 8.
Using the free volume model proposed by Yasuda et al. for Ds
describes the frontier of the diffusivity tradeoff relationship while
using the model proposed by Mackie and Meares for Ds brackets
much of the data on the low-selectivity side. This result is con-
sistent with the observation that the Mackie and Meares model
predicted higher values of Ds, at a given water content, compared
to the free volume model as shown in Fig. 6. The observation that
the data tend to lie between the two models suggests that diffu-
sive transport in many polymers may fall somewhere between the
 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800
Fig. 8. Water/salt diffusion selectivity versus water diffusion coefficient for a series
of reported data [23,37,46–58]. The short dashed line represents the empirical
tradeoff relationship based on diffusion coefficient data calculated using Eq. (11)
( ) [37]. Data calculated using Eq. (10) and the thermodynamic correction factor
procedure described above ( ) are compared to the model results where water
diffusion was modeled using the Yasuda approach (Eqs. (13)–(15)) and salt diffusion
was modeled using either the Mackie and Meares approach (long dashed line, Eq.
(16)) or the single parameter free volume model (solid line, Eq. (19)). The single
circle point represents the ratio of the self-diffusion coefficient of water and the salt
diffusion coefficient in aqueous solution plotted at the value of the self-diffusion
coefficient of water.
free volume description of transport and the obstruction (area/
tortuosity) description of transport used by Mackie and Meares.
The result is also consistent with a recent report suggesting that
diffusive transport in highly selective commercially available de-
salination membranes may be more accurately modeled using free
volume theory compared to hindered transport theory [109].
The diffusion and sorption data and the modeling analysis
shown in Figs. 6 and 7 can be combined to describe the water/salt
permeability selectivity tradeoff with the diffusive water perme-
ability (Fig. 9). To combine the data, the permeability and se-
lectivity values were combined according to Eqs. (1) and (2) [39].
As the method used to determine the diffusion coefficient ulti-
mately influences the permeability, the analysis described above
for the diffusion coefficients carries over into the analysis of the
water/salt permeability and diffusive water permeability tradeoff
relationship.
As discussed in regard to Fig. 8, correcting the water diffusion
coefficient data for the convective frame of reference and ther-
modynamic factors (i.e., using Eq. (10) as opposed to Eq. (11))
impacts the diffusive water permeability. The data shown in the
water/salt permeability selectivity tradeoff relationship with water
permeability (Figs. 1 and 9 as diamonds) were calculated using Eq.
(11). When Eq. (10) is used during the analysis, the data shift
generally to lower selectivity and diffusive water permeability
values upon applying the frame of reference and thermodynamic
correction factors. When combined, the sorption selectivity and
diffusion selectivity models (both based on the Mackie and Meares
model and the free volume model for Ds) describe the perme-
ability tradeoff relationship frontier relatively well. This result
suggests that the contribution of the sorption selectivity tradeoff
helps to offset some of the differences between the two ap-
proaches used to model salt diffusion coefficients. The observation,
in Fig. 9, that the free volume model predicts higher selectivity
values compared to the Mackie and Meares model is consistent
797
Fig. 9. Water/salt permeability selectivity versus diffusive water permeability for a
series of reported data [23,37,46–58]. The short dashed line represents the em-
pirical tradeoff relationship based on diffusion coefficient (and, thus diffusive
permeability) data calculated using Eq. (11) and also shown in Fig. 1 ( ) [37].
Diffusion coefficient (and, thus diffusive permeability) data calculated using Eq.
(10) and the thermodynamic correction factor procedure described above ( ) are
compared to the model results: where water diffusion was modeled using the
Yasuda approach (Eqs. (13)–(15)) and salt diffusion was modeled using either the
Mackie and Meares approach (long dashed line, Eq. (16)) or the single parameter
free volume model (solid line, Eq. (19)). The single circle point represents an
aqueous salt solution.
with the discussion (earlier in this section) of the applicability of
the models to the salt diffusion coefficients and the diffusivity
tradeoff relationship.
4. Conclusions
The reported modeling results can be used to better under-
stand the fundamental underpinnings of the water/salt perme-
ability, sorption, and diffusion selectivity tradeoff relationship
with water permeability, sorption, and diffusivity. The effective-
ness of the tradeoff relationship modeling approaches were
gauged using a transport property data set based on reported data
for a broad range of polymers, and the previously reported tradeoff
relationship plots were updated with additional data from the
recent literature. The analysis presented here also refines the in-
itial empirical tradeoff relationship frontiers.
The water/salt sorption tradeoff relationship can be described
by electrostatic (or dielectric exclusion) forces using the Born
model, though specific interactions between penetrants and the
hydrated polymer are clearly important for some polymers. The
water/salt diffusivity tradeoff relationship can be described using a
model for water and salt diffusivity that is based on free volume
theory, but salt diffusivity can also be modeled using the para-
meter-free Mackie and Meares model, which accounts for the
space filling and tortuosity contributions to salt diffusion. Suc-
cessful water diffusion modeling relies on properly accounting for
the convective frame of reference and thermodynamic correction
factors in the water flux equation (particularly for high water
content polymers). These approaches can be combined to describe
the water/salt permeability tradeoff relationship with diffusive
water permeability, and the influence of the sorption tradeoff
seems to mitigate differences introduced in the selection of the
salt diffusion coefficient model.
The tradeoff relationships and modeling analysis reported here
798 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800
only considers sodium chloride transport, and the set of experi-
mental data contains properties measured using a broad range of
polymers. Specific polymer-penetrant interactions are expected to
be significant in some charged polymers and in polymers that sorb
relatively little water. While the analysis reported here is generally
applicable for the wide range of materials considered, this ap-
proach is not expected to capture specific details of particular
polymer categories or material chemistry.
Appendix A. Supporting information
Supplementary data associated with this article can be found in
the online version at doi:10.1016/j.memsci.2016.08.035.
References
[1] G.M. Geise, D.R. Paul, B.D. Freeman, Fundamental water and salt transport
properties of polymeric materials, Prog. Polym. Sci. 39 (2014) 1–42.
[2] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G. Georgiadis, B.J. Marinas, A.
M. Mayes, Science and technology for water purification in the coming
decades, Nature 452 (2008) 301–310.
[3] J. Kamcev, B.D. Freeman, Charged polymer membranes for environmental/
energy applications, Annu. Rev. Chem. Biomol. Eng. 7 (2016), null.
[4] M.E. Webber, Catch-22: Water vs. Energy, Sci. Am. Sp. Ed., 18, 2008, pp. 34–
41.
[5] The Water-Energy Nexus: Challenges and Opportunities, U.S. Department of
Energy, June 2014.
[6] Energy demands on water resources: Report to Congress on the inter-
dependency of energy and water, U.S. Department of Energy, 2006 (accessed
12.09.13).
[7] Water for Energy: is energy becoming a thirstier resource? in: World Energy
Outlook 2012, International Energy Agency, 2012.
[8] B. Cai, B. Zhang, J. Bi, W. Zhang, Energy's thirst for water in China, Environ.
Sci. Technol. 48 (2014) 11760–11768.
[9] M. Elimelech, W.A. Phillip, The future of seawater desalination: energy,
technology, and the environment, Science 333 (2011) 712–717.
[10] J.R. Werber, A. Deshmukh, M. Elimelech, The critical need for increased se-
lectivity, not increased water permeability, for desalination membranes,
Environ. Sci. Technol. Lett. 3 (2016) 112–120.
[11] C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin, State-of-the-art of reverse
osmosis desalination, Desalination 216 (2007) 1–76.
[12] K.P. Lee, T.C. Arnot, D. Mattia, A review of reverse osmosis membrane ma-
terials for desalination – development to date and future potential, J. Membr.
Sci. 370 (2011) 1–22.
[13] B.E. Logan, M. Elimelech, Membrane-based processes for sustainable power
generation using water, Nature 488 (2012) 313–319.
[14] T. Thorsen, T. Holt, The potential for power production from salinity gra-
dients by pressure retarded osmosis, J. Membr. Sci. 335 (2009) 103–110.
[15] K.K. Reimund, J.R. McCutcheon, A.D. Wilson, Thermodynamic analysis of
energy density in pressure retarded osmosis: the impact of solution volumes
and costs, J. Membr. Sci. 487 (2015) 240–248.
[16] F. Helfer, C. Lemckert, Y.G. Anissimov, Osmotic power with pressure retarded
osmosis: theory, performance and trends – a review, J. Membr. Sci. 453
(2014) 337–358.
[17] C. Klaysom, T.Y. Cath, T. Depuydt, I.F.J. Vankelecom, Forward and pressure
retarded osmosis: potential solutions for global challenges in energy and
water supply, Chem. Soc. Rev. 42 (2013) 6959–6989.
[18] Y.C. Kim, M. Elimelech, Potential of osmotic power generation by pressure
retarded osmosis using seawater as feed solution: analysis and experiments,
J. Membr. Sci. 429 (2013) 330–337.
[19] S. Miller, H. Shemer, R. Semiat, Energy and environmental issues in desali-
nation, Desalination 366 (2015) 2–8.
[20] R. Semiat, Energy issues in desalination processes, Environ. Sci. Technol. 42
(2008) 8193–8201.
[21] K.S. Spiegler, Y.M. El-Sayed, The energetics of desalination processes, Desa-
lination 134 (2001) 109–128.
[22] R.L. Stover, Seawater reverse osmosis with isobaric energy recovery devices,
Desalination 203 (2007) 168–175.
[23] H.B. Park, B.D. Freeman, Z.-B. Zhang, M. Sankir, J.E. McGrath, Highly chlorine-
tolerant polymers for desalination, Angew. Chem. 120 (2008) 6108–6113.
[24] 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. Part B:
Polym. Phys. 48 (2010) 1685–1718.
[25] T. Knoell, Municipal wastewater: chlorine's impact on the performance and
properties of polyamide membranes, Ultrapure Water 23 (2006) 24–31.
[26] A.C. Anderson, R.S. Reimers, P. deKernion, A brief review of the current status
of alternatives to chlorine disinfection of water, Am. J. Public Health 72
(1982) 1290–1293.
[27] A.E. Allegrezza Jr., B.S. Parekh, P.L. Parise, E.J. Swiniarski, J.L. White, Chlorine
resistant polysulfone reverse osmosis modules, Desalination 64 (1987)
285–304.
[28] S.G.J. Heijman, H. Guo, S. Li, J.C. van Dijk, L.P. Wessels, Zero liquid discharge:
heading for 99% recovery in nanofiltration and reverse osmosis, Desalination
236 (2009) 357–362.
[29] A.G. Fane, R. Wang, M.X. Hu, Synthetic membranes for water purification:
status and future, Angew. Chem. Int. Ed. 54 (2015) 3368–3386.
[30] R. Bernstein, S. Belfer, V. Freger, Toward improved boron removal in RO by
membrane modification: feasibility and challenges, Environ. Sci. Technol. 45
(2011) 3613–3620.
[31] A. Ben-David, Y. Oren, V. Freger, Thermodynamic factors in partitioning and
rejection of organic compounds by polyamide composite membranes, En-
viron. Sci. Technol. 40 (2006) 7023–7028.
[32] L.J. Banasiak, A.I. Schäfer, Removal of boron, fluoride and nitrate by elec-
trodialysis in the presence of organic matter, J. Membr. Sci. 334 (2009)
101–109.
[33] N. Hilal, G.J. Kim, C. Somerfield, Boron removal from saline water: a com-
prehensive review, Desalination 273 (2011) 23–35.
[34] X. Jin, Q.H. She, X.L. Ang, C.Y.Y. Tang, Removal of boron and arsenic by for-
ward osmosis membrane: influence of membrane orientation and organic
fouling, J. Membr. Sci. 389 (2012) 182–187.
[35] P. Dydo, M. Turek, Boron transport and removal using ion-exchange mem-
branes: a critical review, Desalination 310 (2013) 2–8.
[36] B. Teychene, G. Collet, H. Gallard, J.P. Croue, A comparative study of boron
and arsenic (III) rejection from brackish water by reverse osmosis mem-
branes, Desalination 310 (2013) 109–114.
[37] G.M. Geise, H.B. Park, A.C. Sagle, B.D. Freeman, J.E. McGrath, Water perme-
ability and water/salt selectivity tradeoff in polymers for desalination, J.
Membr. Sci. 369 (2011) 130–138.
[38] J.G. Wijmans, R.W. Baker, The solution-diffusion model: a review, J. Membr.
Sci. 107 (1995) 1–21.
[39] R.W. Baker, Membrane Technology and Applications, 3rd ed., Wiley, New
York, 2012.
[40] D.R. Paul, Reformulation of the solution-diffusion theory of reverse osmosis,
J. Membr. Sci. 241 (2004) 371–386.
[41] L.M. Robeson, B.D. Freeman, D.R. Paul, B.W. Rowe, An empirical correlation of
gas permeability and permselectivity in polymers and its theoretical basis, J.
Membr. Sci. 341 (2009) 178–185.
[42] L.M. Robeson, Correlation of separation factor versus permeability for poly-
meric membranes, J. Membr. Sci. 62 (1991) 165–185.
[43] D.R. Paul, L.M. Robeson, Polymer nanotechnology: nanocomposites, Polymer
49 (2008) 3187–3204.
[44] G.M. Geise, M.A. Hickner, B.E. Logan, Ionic resistance and permselectivity
tradeoffs in anion exchange membranes, ACS Appl. Mater. Interfaces 5 (2013)
10294–10301.
[45] A. Mehta, A.L. Zydney, Permeability and selectivity analysis for ultrafiltration
membranes, J. Membr. Sci. 249 (2005) 245–249.
[46] H. Strathmann, A.S. Michaels, Polymer-water interaction and its relation to
reverse osmosis desalination efficiency, Desalination 21 (1977) 195–202.
[47] B.J. Sundell, E.-S. Jang, J.R. Cook, B.D. Freeman, J.S. Riffle, J.E. McGrath, Cross-
linked disulfonated poly(arylene ether sulfone) telechelic oligomers. 2. Ele-
vated transport performance with increasing hydrophilicity, Ind. Eng. Chem.
Res. 55 (2016) 1419–1426.
[48] B.J. Sundell, K.-s Lee, A. Nebipasagil, A. Shaver, J.R. Cook, E.-S. Jang, B.
D. Freeman, J.E. McGrath, Cross-linking disulfonated poly(arylene ether
sulfone) telechelic oligomers. 1. Synthesis, characterization, and membrane
preparation, Ind. Eng. Chem. Res. 53 (2014) 2583–2593.
[49] W. Xie, G.M. Geise, B.D. Freeman, C.H. Lee, J.E. McGrath, Influence of pro-
cessing history on water and salt transport properties of disulfonated poly-
sulfone random copolymers, Polymer 53 (2012) 1581–1592.
[50] W. Xie, J. Cook, H.B. Park, B.D. Freeman, C.H. Lee, J.E. McGrath, Fundamental
salt and water transport properties in directly copolymerized disulfonated
poly(arylene ether sulfone) random copolymers, Polymer 52 (2011)
2032–2043.
[51] H.B. Park, W. Xie, B.D. Freeman, J.E. McGrath, Unpublished Data, The Uni-
versity of Texas at Austin, Austin, TX, 2007.
[52] H. Ju, A.C. Sagle, B.D. Freeman, J.I. Mardel, A.J. Hill, Characterization of sodium
chloride and water transport in poly(ethylene oxide) hydrogels, J. Membr.
Sci. 358 (2010) 131–141.
[53] G.M. Geise, C.L. Willis, C.M. Doherty, A.J. Hill, T.J. Bastow, J. Ford, K.I. Winey, B.
D. Freeman, D.R. Paul, Characterization of aluminum-neutralized sulfonated
styrenic pentablock copolymer films, Ind. Eng. Chem. Res. 52 (2013)
1056–1068.
[54] H.K. Lonsdale, U. Merten, R.L. Riley, Transport properties of cellulose acetate
osmotic membranes, J. Appl. Polym. Sci. 9 (1965) 1341–1362.
[55] M.A. Frommer, J.S. Murday, R.M. Messalem, Solubility and diffusivity of
water and of salts in an aromatic polyamide film, Eur. Polym. J. 9 (1973)
367–373.
[56] J.R. Cook, Fundamental Water and Ion Transport Characterization of Sulfo-
nated Polysulfone Desalination Materials, Department of Chemical En-
gineering, University of Texas, Austin, TX, 2014.
[57] H. Yasuda, C.E. Lamaze, A. Peterlin, Diffusive and hydraulic permeabilities of
water in water-swollen polymer membranes, J. Polym. Sci.: Part A-2. 9 (1971)
1117–1131.
[58] H. Yasuda, C.E. Lamaze, L.D. Ikenberry, Permeability of solutes through hy-
drated polymer membranes Part I. Diffusion of sodium chloride, Makromol.
Chem. 118 (1968) 19–35.
[59] V. Freger, Nanoscale heterogeneity of polyamide membranes formed by in-
terfacial polymerization, Langmuir 19 (2003) 4791–4797.
[60] N.Y. Yip, M. Elimelech, Performance limiting effects in power generation
from salinity gradients by pressure retarded osmosis, Environ. Sci. Technol.
45 (2011) 10273–10282.
 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800
[61] G.M. Geise, Water and salt separation characteristics of a sulfonated styrenic
pentablock copolymer for membrane applications, in: Water and Salt
Transport Structure/property Relationships in Polymer Membranes for De-
salination and Power Generation Applications, Department of Chemical
Engineering, University of Texas, Austin, TX, 2012.
[62] 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.
[63] F. Helfferich, Ion Exchange, Dover Publications, New York, 1995.
[64] 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 (2016) 6021–6031.
[65] 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.
[66] J.M. Prausnitz, R.N. Lichtenthaler, E. Gomes de Azevedo, Molecular Ther-
modynamics Of Fluid-Phase Equilibria, 3rd ed., Prentice Hall PTR, Upper
Saddle River, NJ, 1999.
[67] F.G. Donnan, The theory of membrane equilibria, Chem. Rev. 1 (1924)
73–90.
[68] E. Glueckauf, On the mechanism of osmotic desalting with porous mem-
branes, in: Proceedings of the First International Symposium on Water De-
salination, Oct. 3–9, 1965, U.S. Department of the Interior, Washington, D.C.,
1967, pp. 143–150.
[69] J.E. Anderson, W. Pusch, The membrane/water partition coefficients of ions:
electrostatic calculations of dielectric heterogeneity, Ber. Bunsen Ges. 80
(1976) 846–849.
[70] A. Yaroshchuk, Dielectric exclusion of ions from membranes, Adv. Colloid
Interface Sci. 85 (2000) 193–230.
[71] W.M. Haynes, CRC Handbook of Chemistry and Physics, 95th ed., CRC Press/
Taylor and Francis, Boca Raton, FL, 2014, Internet Version 2014.
[72] W.R. Bowen, J.S. Welfoot, Modelling the performance of membrane nano-
filtration–critical assessment and model development, Chem. Eng. Sci. 57
(2002) 1121–1137.
[73] T.T. Duignan, D.F. Parsons, B.W. Ninham, A continuum solvent model of the
multipolar dispersion solvation energy, J. Phys. Chem. B 117 (2013)
9412–9420.
[74] T.T. Duignan, D.F. Parsons, B.W. Ninham, A continuum solvent model of the
partial molar volumes and entropies of ionic solvation, J. Phys. Chem. B 118
(2014) 3122–3132.
[75] J.-c Liu, G.-z Jia, Dielectric relaxation of 1-methylimidazole-ethanol mixtures
at the microwave frequency, Colloid Polym. Sci. 293 (2015) 2053–2059.
[76] Z. Lu, E. Manias, D.D. Macdonald, M. Lanagan, Dielectric relaxation in di-
methyl sulfoxide/water mixtures studied by microwave dielectric relaxation
spectroscopy, J. Phys. Chem. A 113 (2009) 12207–12214.
[77] P.B. Undre, P.W. Khirade, V.S. Rajenimbalkar, S.N. Helambe, S.C. Mehrotra,
Dielectric relaxation in ethylene glycol – dimethyl sulfoxide mixtures as a
function of composition and temperature, J. Korean Chem. Soc. 56 (2012)
416–423.
[78] P.W. Khirade, A. Chaudhari, J.B. Shinde, S.N. Helambe, S.C. Mehrotra, Static
dielectric constant and relaxation time measurements on binary mixtures of
dimethyl sulfoxide with ethanol, 2-ethoxyethanol, and propan-1-ol at 293,
303, 313, and 323 K, J. Chem. Eng. Data 44 (1999) 879–881.
[79] S.J. Paddison, D.W. Reagor, T.A. Zawodzinski, High frequency dielectric stu-
s
dies of hydrated Nafion , J. Electroanal. Chem. 459 (1998) 91–97.
[80] D.R. Paul, Diffusive transport in swollen polymer membranes, in: H.
B. Hopfenberg (Ed.), Permeability of Plastic Films and Coatings: To Gases,
Vapors, and Liquids, Plenum Press, New York, 1974, pp. 35–48.
[81] D.R. Paul, Relation between hydraulic permeability and diffusion in homo-
geneous swollen membranes, J. Polym. Sci. Polym. Phys. Ed. 11 (1973)
289–296.
[82] P.J. Flory, J.J. Rehner, Statistical mechanics of cross-linked polymer networks
II. Swelling, J. Chem. Phys. 11 (1943) 521–526.
[83] P.J. Flory, Thermodynamics of high polymer solutions, J. Chem. Phys. 10
(1942) 51–61.
[84] W. Xie, H. Ju, G.M. Geise, B.D. Freeman, J.I. Mardel, A.J. Hill, J.E. McGrath,
Effect of free volume on water and salt transport properties in directly co-
polymerized disulfonated poly(arylene ether sulfone) random copolymers,
Macromolecules 44 (2011) 4428–4438.
[85] Y.-J. Kim, K.-S. Lee, M.-H. Jeong, J.-S. Lee, Highly chlorine-resistant end-group
crosslinked sulfonated-fluorinated poly(arylene ether) for reverse osmosis
membrane, J. Membr. Sci. 378 (2011) 512–519.
[86] C.H. Lee, D. VanHouten, O. Lane, J.E. McGrath, J. Hou, L.A. Madsen, J. Spano,
S. Wi, J. Cook, W. Xie, H.J. Oh, G.M. Geise, B.D. Freeman, Disulfonated poly
(arylene ether sulfone) random copolymer blends tuned for rapid water
permeation via cation complexation with poly(ethylene glycol) oligomers,
Chem. Mater. 23 (2011) 1039–1049.
[87] M.R. Hibbs, M.A. Hickner, T.M. Alam, S.K. McIntyre, C.H. Fujimoto, C.
J. Cornelius, Transport properties of hydroxide and proton conducting
membranes, Chem. Mater. 20 (2008) 2566–2573.
[88] S. Zhang, R. Zhang, Y.C. Jean, D.R. Paul, T.-S. Chung, Cellulose esters for for-
ward osmosis: characterization of water and salt transport properties and
free volume, Polymer 53 (2012) 2664–2672.
[89] A.C. Sagle, PEG hydrogels as anti-fouling coatings for reverse osmosis
membranes, in: Chemical Engineering, The University of Texas, Austin, TX,
2009, pp. 168.
[90] L. Ni, J. Meng, G.M. Geise, Y. Zhang, J. Zhou, Water and salt transport prop-
erties of zwitterionic polymers film, J. Membr. Sci. 491 (2015) 73–81.
[91] Q. Duan, H. Wang, J. Benziger, Transport of liquid water through Nafion
membranes, J. Membr. Sci. 392–393 (2012) 88–94.
799
[92] C.E. Evans, R.D. Noble, S. Nazeri-Thompson, B. Nazeri, C.A. Koval, Role of
conditioning on water uptake and hydraulic permeability of Nafion mem-
branes, J. Membr. Sci. 279 (2006) 521–528.
[93] P. Meares, The mechanism of water transport in membranes, Philos. Trans. R.
Soc. B 278 (1977) 113–150.
[94] C.E. Reid, E.J. Breton, Water and ion flow across cellulosic membranes, J. Appl.
Polym. Sci. 1 (1959) 133–143.
[95] S.G. Kimura, Reverse osmosis performance of sulfonated poly(2,6-di-
methylphenylene ether) ion exchange membranes, Ind. Eng. Chem. Res. 10
(1971) 335–339.
[96] J.H. Wang, C.V. Robinson, I.S. Edelman, Self-diffusion and structure of liquid
water. III. Measurement of the self-diffusion of liquid water with H2, H3 and
O18 as tracers, J. Am. Chem. Soc. 75 (1953) 466–470.
[97] M.H. Cohen, D. Turnbull, Molecular transport in liquids and glasses, J. Chem.
Phys. 31 (1959) 1164–1169.
[98] A.T. DiBenedetto, D.R. Paul, An interpretation of gaseous diffusion through
polymers using fluctuation theory, J. Polym. Sci. Part A: General Papers. 2
(1964) 1001–1015.
[99] J.S. Mackie, P. Meares, The diffusion of electrolytes in a cation-exchange resin
membrane. I. Theoretical, Proceedings of the Royal Society of London. Series
A, Mathematical and Physical Sciences, 232 (1955) 498–509.
[100] G.M. Geise, B.D. Freeman, D.R. Paul, Characterization of a novel sulfonated
pentablock copolymer for desalination applications, Polymer 51 (2010)
5815–5822.
[101] E. Dražević, K. Košutić, V. Freger, Permeability and selectivity of reverse os-
mosis membranes: correlation to swelling revisited, Water Res. 49 (2014)
444–452.
[102] J.S. Mackie, P. Meares, The diffusion of electrolytes in a cation-exchange resin
membrane. II. Experimental, Proceedings of the Royal Society of London.
Series A, Mathematical and Physical Sciences, 232 (1955) 510–518.
[103] C.J.D. Fell, H.P. Hutchison, Diffusion coefficients for sodium and potassium
chlorides in water at elevated temperatures, J. Chem. Eng. Data 16 (1971)
427–429.
[104] S.J. Tao, Positronium annihilation in molecular substances, J. Chem. Phys. 56
(1972) 5499–5510.
[105] M. Eldrup, D. Lightbody, J.N. Sherwood, The temperature dependence of
positron lifetimes in solid pivalic acid, Chem. Phys. 63 (1981) 51–58.
[106] M. Eldrup, O. Mogensen, Positron lifetimes in pure and doped ice and in
water, J. Chem. Phys. 57 (1972) 495–504.
[107] G.M. Geise, C.M. Doherty, A.J. Hill, B.D. Freeman, D.R. Paul, Free volume
characterization of sulfonated styrenic pentablock copolymers using positron
annihilation lifetime spectroscopy, J. Membr. Sci. 453 (2014) 425–434.
[108] Y. Kobayashi, H.F.M. Mohamed, A. Ohira, Positronium formation in aromatic
polymer electrolytes for fuel cells, J. Phys. Chem. Lett. 113 (2009) 5698–5701.
[109] E. Dražević, K. Košutić, V. Kolev, V. Freger, Does hindered transport theory
apply to desalination membranes? Environ. Sci. Technol. 48 (2014)
11471–11478.
[110] A.C. Sagle, H. Ju, B.D. Freeman, M.M. Sharma, PEG-based hydrogel membrane
coatings, Polymer 50 (2009) 756–766.
[111] G.E. Zaikov, A.P. Iordanskii, V.S. Markin, Diffusion of electrolytes in polymers,
in, VSP, Utrecht, The Netherlands, 1988, pp. 321.
[112] A.E. Yaroshchuk, Non-steric mechanisms of nanofiltration: superposition of
Donnan and dielectric exclusion, Sep. Purif. Technol. 22–23 (2001) 143–158.
[113] E. Glueckauf, R.E. Watts, The Donnan law and its application to ion-exchanger
polymers, Proceedings of the Royal Society of London, Series A: Mathema-
tical, Physical and Engineering Sciences, 268 (1962) 339–349.
[114] E. Glueckauf, The distribution of electrolytes between cellulose acetate
membranes and aqueous solutions, Desalination 18 (1976) 155–172.
[115] E. Glueckauf, A new approach to ion-exchange polymers, Proceedings of the
Royal Society of London, Series A: Mathematical, Physical and Engineering
Sciences, 268 (1962) 350–370.
[116] M.E. Heyde, C.R. Peters, J.E. Anderson, Factors influencing reverse osmosis
rejection of inorganic solutes from aqueous solution, J. Colloid Interface Sci.
50 (1975) 467–487.
[117] M.E. Heyde, J.E. Anderson, Ion sorption by cellulose acetate membranes from
binary salt solutions, J. Phys. Chem. 79 (1975) 1659–1664.
[118] J.E. Anderson, H.W. Jackson, Membrane-water partition coefficients of ions.
Calculated effects of membrane thickness, J. Phys. Chem. 78 (1974)
2259–2262.
[119] J. Kamcev, D.R. Paul, B.D. Freeman, Ion activity coefficients in ion exchange
polymers: applicability of manning's counterion condensation theory, Mac-
romolecules 48 (2015) 8011–8024.
[120] A. Szymczyk, P. Fievet, Investigating transport properties of nanofiltration
membranes by means of a steric, electric and dielectric exclusion model, J.
Membr. Sci. 252 (2005) 77–88.
[121] S. Bandini, D. Vezzani, Nanofiltration modeling: the role of dielectric exclu-
sion in membrane characterization, Chem. Eng. Sci. 58 (2003) 3303.
[122] A. Seidel, J.J. Waypa, M. Elimelech, Role of charge (Donnan) exclusion in re-
moval of arsenic from water by a negatively charged porous nanofiltration
membrane, Environ. Eng. Sci. 18 (2001) 105–113.
[123] S. Mafe, P. Ramirez, J. Pellicer, Activity coefficients and Donnan coion ex-
clusion in charged membranes with weak-acid fixed charge groups, J.
Membr. Sci. 138 (1998) 269–277.
[124] S. Mafe, P. Ramirez, J.A. Manzanares, Activity coefficients and coion exclusion
in charged polymeric membranes with macroscopic inhomogeneities, Col-
loid Polym. Sci. 275 (1997) 599–603.
[125] L. Jones, P.N. Pintauro, H. Tang, Coion exclusion properties of polypho-
sphazene ion-exchange membranes, J. Membr. Sci. 162 (1999) 135–143.
[126] R. Tandon, P.N. Pintauro, Solvent effects during multicomponent ion uptake
into a Nafion cation-exchange membrane, J. Membr. Sci. 341 (2009) 21–29.
800 H. Zhang, G.M. Geise / Journal of Membrane Science 520 (2016) 790–800

[127] R. Tandon, P.N. Pintauro, Divalent/monovalent cation uptake selectivity in a
Nafion cation-exchange membrane: experimental and modeling studies, J.
Membr. Sci. 136 (1997) 207–219.
[128] P.N. Pintauro, D.N. Bennion, Mass transport of electrolytes in membranes. 2.
Determination of sodium chloride equilibrium and transport parameters for
Nafion, Ind. Eng. Chem. Fundam. 23 (1984) 234–243.
[129] J. Palomo, P.N. Pintauro, Competitive absorption of quaternary ammonium
and alkali metal cations into a Nafion cation-exchange membrane, J. Membr.
Sci. 215 (2003) 103–114.
[130] J.R. Bontha, P.N. Pintauro, Water orientation and ion solvation effects during
multicomponent salt partitioning in a Nafion cation exchange membrane,
Chem. Eng. Sci. 49 (1994) 3835–3851.