Preparation and characterization of crosslinked poly(vinylimidazolium) anion exchange membranes for artificial photosynthesis

Article abstract of DOI:10.1039/c9ta00498j

The interrelated nature of material properties in ion exchange membranes, such as ion exchange capacity and water uptake, frustrates the systematic study of how membrane chemistry and structure affect the transport of water, ions, and uncharged solutes in the membrane. Herein, we describe the preparation of a series of crosslinked poly(vinylimidazolium) anion exchange membranes by UV-photopolymerization of difunctional (i.e., crosslinking) and monofunctional (i.e., non-crosslinking) monomers, in which the ion exchange capacity and crosslink density may be independently controlled. Ion exchange membranes used in artificial photosynthesis (solar-driven CO₂ reduction) devices must permit the transport of electrolyte ions and minimize the crossover of CO₂ reduction products (e.g., alcohols) between electrodes. The water content, methanol (CO₂ reduction product) permeability, and ionic conductivity of the membranes were evaluated. Ionic conductivity and methanol permeability were increased by reducing crosslink density or increasing solvent content in the prepolymerization solvent mixture. For all prepared membranes, methanol permeability was directly correlated with water volume fraction in the membrane. Minimizing the water volume fraction is critical to the design of membranes with low permeability to CO₂ reduction products.

Full text of DOI:10.1039/c9ta00498j

View Article Online
View Journal
Journal of
Materials Chemistry A
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: B. Carter, L. Keller,
M. Wessling and D. Miller, J. Mater. Chem. A, 2019, DOI: 10.1039/C9TA00498J.
This is an Accepted Manuscript, which has been through the
Volume
4 Number 1 7 January 2016 Pages 1–330
Royal Society of Chemistry peer review process and has been
accepted for publication.
Journal of
Materials Chemistry A
Accepted Manuscripts are published online shortly after
Materials for energy and sustainability
www.rsc.org/MaterialsA
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
ISSN 2050-7488
PAPER
Kun Chang, Zhaorong Chang et al.
Bubble-template-assisted synthesis of hollow fullerene-like
MoS2 nanocages as a lithium ion battery anode material
rsc.li/materials-a

Page 1 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
Preparation and characterization of crosslinked poly(vinylimidazolium)
anion exchange membranes for artificial photosynthesis
Blaine M. Carter1, Laura Keller2, Matthias Wessling2,3, and Daniel J. Miller1†
†Corresponding Author, Tel: +1 (510) 495-2353 E-mail: danieljmiller@lbl.gov
1Joint Center for Artificial Photosynthesis, Lawrence Berkeley National Laboratory, Berkeley,
CA 97420, United States
2RWTH Aachen University, Chemical Process Engineering, Forckenbeckstr. 51, 52074 Aachen,
Germany
³DWI – Leibniz Institute for Interactive Materials, Forckenbeckstr. 50, 52074 Aachen, Germany
Abstract:
The interrelated nature of material properties in ion exchange membranes, such as ion
exchange capacity and water uptake, frustrates the systematic study of how membrane chemistry
and structure affect the transport of water, ions, and uncharged solutes in the membrane. Herein,
we describe the preparation of a series of crosslinked poly(vinylimidazolium) anion exchange
membranes by UV-photopolymerization of difunctional (i.e., crosslinking) and monofunctional
(i.e., non-crosslinking) monomers, in which the ion exchange capacity and crosslink density may
be independently controlled. Ion exchange membranes used in artificial photosynthesis (solar-
driven CO2 reduction) devices must permit the transport of electrolyte ions and minimize the
crossover of CO2 reduction products (e.g., alcohols) between electrodes. The water content,
methanol (CO2 reduction product) permeability, and ionic conductivity of the membranes were
evaluated. Ionic conductivity and methanol permeability were increased by reducing crosslink
density or increasing solvent in the prepolymerization solvent mixture. For all prepared
membranes, methanol permeability was directly correlated with water volume fraction in the

Journal of Materials Chemistry A
Page 2 of 40
View Article Online
DOI: 10.1039/C9TA00498J
membrane. Minimizing the water volume fraction is critical to the design of membranes with
low permeability to CO2 reduction products.
Keywords:
Ion exchange membrane, permeability, methanol, ionic conductivity, artificial photosynthesis
1. Introduction
Ion exchange membranes are broadly used in diverse applications, including energy
storage, water treatment, and energy generation.1–3 These membranes are often comprised of
polymers with charged moieties covalently attached to the polymer backbone. Ion exchange
membranes are critical components of many electrochemical devices and have consequently
been the subject of study for many years, yet a complete understanding of how membrane
chemical and structural characteristics affect transport properties has not been realized.2
Understandably, most experimental effort has concentrated on understanding water and ion
transport,1 and recently, fundamental studies of uncharged solute (e.g., alcohol) transport in ion
exchange membranes have been reported.4
To understand how membrane chemistry and structure affect water, ion, and neutral
solute transport, model polymer systems whose properties can be systematically varied through
careful synthesis are desirable. A significant challenge in designing ion exchange polymer
systems is that several material properties may be interrelated.2 For example, the ion exchange
capacity (IEC, moles of ions/g dry polymer) affects both the ionic conductivity and the water
uptake of the polymer.2,3,5 The substantial water uptake associated with large IEC values can lead
to excessive swelling and a loss of mechanical stability.3,5 Covalent crosslinking is an effective

Page 3 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
tool to limit swelling and has been employed in numerous anion exchange materials.5–7
Permeability to ions (i.e., ionic conductivity), water, and neutral solutes typically decreases with
an increase in crosslink density, while excessive crosslinking can result in a brittle membrane.5,8
A model system that provides independent control over the IEC and crosslink density would
enable decoupling of many of the aforementioned properties, but can be synthetically
challenging because multiple chemical functionalities (e.g., crosslinkable and non-crosslinkable
moieties, charged moieties), must be incorporated.
An emerging technology that utilizes ion exchange membranes is artificial
photosynthesis, the photoelectrochemical reduction of CO2 to carbon-neutral fuels and chemicals
by which fluctuating solar energy can be converted into chemical energy available on demand.9
While CO2 reduction to CO by, for example, the Fischer-Tropsch process for the purpose of
syngas production is well known,10 CO is not readily useful as a fuel.11 Instead, the target of
much ongoing artificial photosynthesis research is the direct production of hydrocarbon fuels or
chemical feedstocks.12,13 The simplest artificial photosynthesis device configuration consists of
two half-cells, each containing a planar electrode and aqueous electrolyte, that are separated by
an ion exchange membrane.14 Carbon dioxide is reduced to alcohol or other products11 at the
cathode, while water is oxidized to O2 at the anode. The membrane in these devices serves two
functions: (a) permit the transport of electrolyte charge carriers between electrodes so that
current can flow, and (b) minimize the crossover of CO2 reduction products from the cathode to
the anode, where CO2 reduction products could be re-oxidized to CO2. Commonly the electrolyte
is near-neutral pH, but an alkaline environment is reported to promote carbon-carbon coupling
and, therefore, the formation of C2+ products.15,16 Major electrolyte charge carriers include
bicarbonate, carbonate, and hydroxide ions. Consequently, anion exchange membranes that

Journal of Materials Chemistry A
Page 4 of 40
View Article Online
DOI: 10.1039/C9TA00498J
contain covalently bound cationic groups are favorable, since these membranes readily transport
these negatively charged ions.16
Selemion AMV is a commercial anion exchange membrane that is commonly employed
in (photo)electrochemical CO2 reduction studies.16,17,39 Selemion AMV is fabricated by the
“paste method,” where a blend of poly(vinyl chloride) and functionalized poly(styrene) are
applied to a poly(vinyl chloride) nonwoven support.18 The poly(styrene) is thought to be
functionalized with quaternary ammonium and pyridine moieties, although some uncertainty
remains in the literature.19,20 While Selemion AMV sees wide use in electrochemical studies,11,21–
23 this proprietary composite membrane does not serve as a tunable system with which structure–
property relationships may be established. However, model polymers with chemical features
inspired by Selemion AMV, such as charged aromatic heterocycles covalently bound to
hydrocarbon backbones, could serve as useful analogs that enable systematic study of membrane
properties governing transport in artificial photosynthesis devices. Imidazolium-based styrenic
polymers for CO2 reduction applications have recently been commercialized and employed in
gas diffusion electrode type devices,24–26 but systematic membrane structure-property studies
have not been reported.
Difunctional (i.e., crosslinkable) vinylimidazolium monomers with alkyl spacers (Figure
1a) have been investigated for a range of applications.27–30 With a halide (e.g., Cl–, Br–)
counterion, these monomers are soluble in water or methanol and the resulting solution can be
cured into crosslinked networks using an appropriate initiator. Monofunctional analogs of these
materials (Figure 1b) have also been synthesized.29,31,32 Mixing difunctional and monofunctional
monomers of the same spacer length n allows variation of the crosslink density without changing
the IEC. Because both crosslink density and IEC affect the water uptake of the polymer, we

Page 5 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
propose that this system permits decoupling of the crosslink density and the IEC and, therefore,
independent control of water uptake by crosslink density without changing the IEC.
Furthermore, adjusting the spacer length n permits variation of both the crosslink density and
IEC. To our knowledge, no investigations have been reported on combinations of these
difunctional and monofunctional monomers to afford networks with control of both crosslink
density and IEC.
Figure 1. Structure of (a) difunctional vinylimidazolium monomers and (b) monofunctional
vinylimidazolium monomers
Herein, we describe the systematic preparation and characterization of a series of anion
exchange membranes based on the difunctional and monofunctional vinylimidazolium
monomers shown in Figure 1. These membranes were prepared by UV-photopolymerization of
aqueous or methanolic solutions of the vinylimidazolium monomers. By varying the monomer
spacer length, crosslinker (i.e., difunctional monomer) content, and prepolymerization solvent
content, membranes with a variety of crosslinked network structures were prepared at fixed
IECs. The membranes were characterized by measuring the equilibrium water content, methanol
permeability, and ionic conductivity. Because methanol is among the smallest reported CO2
reduction products, it is a challenging solute of which to limit transport across a membrane.
Furthermore, straightforward techniques for measurement of methanol permeation have been
established using in situ FTIR spectroscopy.33,34 Trends in the prepared membrane properties are

Journal of Materials Chemistry A
Page 6 of 40
View Article Online
DOI: 10.1039/C9TA00498J
discussed, as well as the relationship between water volume fraction in the membrane and
permeability to ions and solutes.
2. Experimental
2.1 Materials
N-Vinylimidazole (98%), N-methylimidazole (99%), 1,12-dibromododecane (98%),
eicosanedioic acid (98%), carbon tetrabromide (99%), 2-hydroxy-4'-(2-hydroxyethoxy)-2-
methylpropiophenone (Irgacure 2959, 98%), were purchased from TCI America and used as
received. Borane tetrahydrofuran complex (1 M) and triphenylphosphine (99%) were purchased
from Alfa Aesar and used as received. Methanol (reagent grade, ≥ 99.5% purity) was purchased
from Sigma-Aldrich and used as received. Ultrapure water was supplied by an EMD Millipore
Milli-Q Integral 3 water purification system (18.2 MΩ•cm at 25 °C, 1.2 ppb TOC). All other
reagents and solvents were ACS Reagent Grade or higher quality.
2.2 Synthesis of vinylimidazolium monomers
Monomers 1, 2, 1a, and 2a were synthesized according to Scheme 1. Details of the
synthesis are described in the subsequent sections. NMR spectra (1H and 13C) for all compounds
are provided in the Supporting Information.

Page 7 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
Scheme 1. Synthesis of monomers 1, 2, 1a, and 2a
2.2.1 Synthesis of 1,20-eicosanediol (i)
Anhydrous THF (500 mL) and borane-THF complex (100 mL, 100 mmol) were added to
a 1000 mL round-bottom flask equipped with a stir bar and reflux condenser under flowing
nitrogen. Eicosanedioc acid (11.41 g, 33.3 mmol) powder was added slowly to the reaction
mixture to minimize bubbling and the resultant slurry was heated to 45 °C for 16 hours. After
cooling to room temperature, the reaction was quenched with H2O (15 mL) dropwise. Diethyl
ether (300 mL) was added, followed by dropwise addition of 4 M NaOH (20 mL). The organic
layer was washed with H2O (3 × 50 mL), then with brine (3 × 50 mL), and then dried over
anhydrous MgSO4. The MgSO4 was rinsed with additional warm (ca. 30 °C) THF (500 mL). The
resulting organic solution was reduced under rotary vacuum to afford a white solid. The solids
were stirred in a H2O/MeOH solution (200 mL, 60/40 (v/v)) for 20 minutes, filtered, and dried in
a vacuum oven (< 1 Torr, 40 °C) to afford 1,20-eicsoanediol as a white powder (10.25 g, 32.6
mmol, 98 % yield). NMR spectra for this compound matched published data.35

Journal of Materials Chemistry A
Page 8 of 40
View Article Online
DOI: 10.1039/C9TA00498J
2.2.2 Synthesis of 1,20-dibromoeicosane (ii)
The procedure to synthesize 1,20-dibromoeicosane was adapted from an analogous
reaction described in the literature.36 Anhydrous THF (100 mL) was added to a 250 mL round-
bottom flask equipped with a stir bar and reflux condenser under flowing nitrogen. After heating
the solvent to 45 °C, carbon tetrabromide (12.65 g, 38.1 mmol) and 1,20-eicosanediol (3.00 g,
9.5 mmol) were added. Triphenylphosphine (14.15 g, 53.9 mmol) was added slowly over the
course of 30 minutes. After stirring for 4 hours at 45 °C, the resultant mixture was cooled to
room temperature and the solids were removed via filtration. Residual THF was removed under
rotary vacuum to afford an orange solid. The solids were stirred in a H2O/EtOH solution (200
mL, 40/60 (v/v)) overnight. The solids were collected by filtration and rinsed with H2O/EtOH
solution (200 mL, 40/60 (v/v)). After drying the solids (3.43 g) in a vacuum oven (< 1 Torr,
1
40 °C), H NMR typically indicated that a small amount of unreacted alcohol remained. A
similar procedure was used to convert the remaining alcohols. Anhydrous THF (50 mL) was
added to a 100 mL round-bottom flask equipped with a stir bar under flowing nitrogen. The
product mixture (3.43 g) and carbon tetrabromide (6.32 g, 19.1 mmol) were added, followed by
slow addition of triphenyphosphine (7.03 g, 26.8 mmol). After 1 hour, solids were filtered and
THF was removed under rotary vacuum. The solids were stirred in a H2O/EtOH solution (200
mL, 40/60 (v/v)) three times, filtered, and dried in a vacuum oven (< 1 Torr, 40 °C) to afford the
product as a white powder (3.16 g, 7.2 mmol, 76 % yield).
2.2.3 Synthesis of monomer 1
1,12-Dibromododecane (4.98 g, 15.2 mmol), N-vinylimidazole (3.03 g, 32.2 mmol), and
acetonitrile (7 mL) were combined in a 50 mL round-bottom flask equipped with a stir bar and

Page 9 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
reflux condenser. The reaction solution was heated to 70 °C for 16 hours. The solidified reaction
mixture was cooled to room temperature and mixed with acetone (100 mL). Filtering the solution
afforded the product as a white solid (6.75 g, 13.1 mmol, 86 % yield). NMR spectra for this
compound matched published data.37
2.2.4 Synthesis of monomer 2
1,20-Dibromoeicosane (2.49 g, 5.7 mmol) was mixed with acetonitrile (11 mL) and
toluene (3.5 mL) in a 25 mL round-bottom flask equipped with a stir bar and reflux condenser.
The mixture was heated to 55 °C and N-vinylimidazole (2.32 g, 24.7 mmol) was added. The
homogenous solution was heated to 80 °C for 16 hours. Upon cooling to room temperature, a
white precipitate formed and the slurry was poured into diethyl ether (300 mL). The solids were
filtered, rinsed with diethyl ether (100 mL), and dried under vacuum (< 0.1 Torr) to afford the
product as a white powder (3.37 g, 5.4 mmol, 94 % yield).
2.2.5 Synthesis of monomer 1a
1,12-Dibromododecane (30.05 g, 91.6 mmol) was mixed with acetonitrile (30 mL) in a
100 mL round-bottom flask equipped with a stir bar and reflux condenser. The solution was
heated to 55 °C to produce a homogenous solution. N-Vinylimidazole (1.54 g, 16.4 mmol) was
added dropwise and then the mixture was heated to 80 °C for 16 hours. The reaction mixture was
cooled to room temperature and hexanes (15 mL) was added. The product was extracted into
methanol (50 mL) and washed with hexanes (5 × 100 mL). Methanol was removed by rotary
vacuum to afford a viscous brown liquid. The product was washed with diethyl ether (2 × 150
mL) and hexanes (2 × 100 mL). After sitting in hexanes for approximately 30 minutes, the

Journal of Materials Chemistry A
Page 10 of 40
View Article Online
DOI: 10.1039/C9TA00498J
product solidified. The solids were washed with fresh hexanes (100 mL), filtered, and dried
under vaccum (< 0.1 Torr) to afford the intermediate product as a white powder (5.96 g).
The intermediate product (5.96 g) was mixed with acetonitrile (16 mL) and heated to
50 °C. N-Methylimidazole (1.5 g, 18.3 mmol) was added and the reaction mixture heated to
80 °C for 16 hours. After cooling to room temperature, acetonitrile was removed by rotary
vacuum, yielding a light brown viscous liquid. The product was washed with ethyl acetate (2 ×
100 mL) and diethyl ether (2 × 100 mL). After sitting in diethyl ether for approximately 30
minutes, the product solidified. After decanting, the solids were washed with fresh diethyl ether
(100 mL). The product was hygroscopic, so the diethyl ether slurry was transferred to a Schlenk
flask. Excess diethyl ether was decanted and the solids were dried under vacuum (< 0.1 Torr) to
afford the product as a white powder (6.75 g). This procedure results in a mixture of monomers 1
1
and 1a. The ratio of these two compounds was calculated from H NMR as described in the
Supporting Information. With this procedure, a typical concentration of monomer 1 was 6 mol
%.
2.2.6 Synthesis of monomer 2a
1,20-Dibromoeicosane (3.25 g, 7.4 mmol) was mixed with acetonitrile (6 mL) and
toluene (4 mL) in a 25 mL round-bottom flask equipped with a stir bar and reflux condenser. The
mixture was heated to 55 °C and N-vinylimidazole (0.41 g, 4.5 mmol) was added dropwise. The
homogenous solution was heated to 75 °C for 16 hours. Upon cooling to room temperature,
methanol (10 mL) and hexanes (10 mL) were added. The methanol layer was washed with
hexanes (5 × 50 mL), and then the methanol was removed by rotary vacuum. The resulting white

Page 11 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
solid was stirred in hexanes (100 mL), filtered, rinsed with hexanes (2 × 100 mL), and dried
under vacuum (< 0.1 Torr) to afford the intermediate product as a white powder (1.52 g).
The intermediate product (1.52 g) was mixed with acetonitrile (4 mL) and heated to 70
°C. N-Methylimidazole (0.32 g, 3.8 mmol) was added and the reaction mixture heated to 80 °C
for 16 h. On cooling to room temperature, a white precipitate formed and diethyl ether was added
(200 mL). The solids were filtered, rinsed with diethyl ether (100 mL), and dried under vacuum
(< 0.1 Torr) to afford the product as a white powder (1.62 g). This procedure results in a mixture
1
of monomers 2 and 2a. The ratio of these two compounds was calculated from H NMR as
described in the Supporting Information. With this procedure, a typical concentration of
monomer 2 was 25 mol %.
2.3 Membrane preparation
Appropriate amounts of monomer, solvent, and photoinitiator (0.1 wt. % relative to
monomer) were added to a small vial. The contents were gently heated to dissolve the monomer,
thoroughly mixed, and the resultant solution was filtered through a 0.45 µm PTFE filter. Two
quartz plates were treated with Rain-X® according to manufacturer’s instructions and wiped
clean with a Kimwipe to remove excess Rain-X®. The plates were then pre-heated on a
temperature-controlled hot plate to the desired temperature. Temperatures for all mixtures are
specified in the Supporting Information. Steel spacers (50 or 100 µm) in between the quartz
plates were used to control the film thickness.
The prepolymerization mixture was heated to 80 °C and transferred by pipette onto a pre-
heated (80 °C) quartz plate and quickly covered with a second pre-heated (80 °C) quartz plate.
The sample was irradiated with a 365 nm UV lamp (3 mW cm-2 at the sample surface) for 20

Journal of Materials Chemistry A
Page 12 of 40
View Article Online
DOI: 10.1039/C9TA00498J
minutes. The films were removed from the quartz plates and soaked in ultrapure water. Films
were equilibrated in fresh ultrapure water for 24 hours before characterization.
A high degree of conversion (> 90 %) for all samples was confirmed using attenuated
total reflectance Fourier transform infrared (ATR FTIR) spectroscopy. Films used for ATR FTIR
analysis were not soaked in water after curing, but immediately transferred to an empty glass jar
and dried in a vacuum oven (< 1 Torr, 40 °C) for 24 hours. The dried film spectra were
compared with the neat monomer mixture. Disappearance of the peaks at 1648 cm-1 and 945 cm-
1, corresponding to the vinyl moiety participating in crosslinking, demonstrated high conversion.
ATR FTIR spectra are provided in the Supporting Information.
2.4 Conversion of Selemion AMV membrane to bromide form
Selemion AMV membrane (AGC Engineering Co., Ltd., Chiba, Japan) is supplied in the
chloride form (i.e., the counter ions to the cationic moieties in the membrane are chloride ions).19
The as-delivered membrane was converted to the bromide form by soaking in 1 M NaBr solution
for 48 h with refreshment of the soaking solution every 12 hours. The Selemion AMV
membranes were then soaked in ultrapure water for 24 hours, refreshing the ultrapure water after
12 hours of soaking.
2.5 Water content
The water content of equilibrated membranes was measured by liquid sorption.38,39
Hydrated samples were quickly blotted between two pieces of Kimwipe and weighed, Wh. The
films were then dried under vacuum (< 1 Torr, 40 °C) to constant weight (minimum 24 h) and

Page 13 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
weighed again to determine the mass of the dry film, Wd. The percent water content, ωW, was
calculated as follows:
푊ℎ − 푊푑
휔푊
=
× 100%
푊ℎ
2.6 Polymer density
The density of the dry polymer, ρd, was measured using a gas pycnometer (AccuPyc II
1340, Micromeritics, Norcross, GA). The sample holder was dried at 120 °C and ambient
pressure and then quickly weighed. All samples were dried under vacuum (< 1 Torr, 40 °C)
overnight. Subsequently, they were cut into small squares approximately 5 mm x 5 mm in size,
added to the sample holder, and then dried under vacuum (< 1 Torr, 40 °C) for a minimum of
24 h. Upon removal from the oven, the sample was weighed and quickly transferred to the
pycnometer to minimize water uptake from the atmosphere. The density measurement was
carried out using ten purge cycles and ten measuring cycles with helium. The pressure change for
each measurement step was ≤ 0.005 psi/min before a value was taken and the next measurement
step started. For each monomer, the polymer density was determined for three membranes
samples prepared at three different prepolymerization solvent contents and the average was used
for subsequent calculations.
2.7 Calculation of water volume fraction
Assuming volume additivity, which has been shown to be reasonable for a number of
charged polymers,40,41 the water volume fraction in the fully hydrated film, ϕw, was calculated
using the measured polymer density, ρP, and both the dry and hydrated weight of the films:

Journal of Materials Chemistry A
Page 14 of 40
View Article Online
DOI: 10.1039/C9TA00498J
(푊ℎ − 푊푑)/휌푊
휙푤 =
(푊ℎ − 푊푑)/휌푊 + 푊푑/휌푝
The density of water, ρW, was taken as 1.0 g/mL.
2.8 Methanol permeability measurements
Methanol permeability was measured using a standard diffusion cell (Adams and
Chittenden Scientific Glassware, Berkeley, CA) equipped with an ATR FTIR probe (Mettler-
Toledo ReactIRTM 15 with shallow tip 9.5 mm DSub AgX DiComp probe) in the receiver cell. A
full description of this experimental apparatus and method is reported elsewhere.33,34 All
experiments were performed at 25 °C with an initial donor cell concentration of 1 M methanol.
Spectra were collected until the receiver cell concentration reached a minimum concentration of
75 mM. Permeability measurements were conducted on three individually prepared membranes
and the reported error is the standard deviation of these replicates.
2.9 Ionic conductivity measurements
In-plane conductivity was measured using a four point conductivity cell (BekkTech BT-
110) interfaced with a Biologic VSP-300 potentiostat. A rectangular section of membrane
(length: > 2.5 cm, width: 0.8 - 1.3 cm) was cut and mounted in the cell, and then the cell was
immersed in ultrapure water (300 mL). Linear sweep voltammetry (-0.1 V to +0.1 V, 10 mV/s)
was performed. The resistance, R (Ω), is the slope of the voltage vs. current curve. Ionic
conductivity was calculated from σ = L/RWT, where σ is the ionic conductivity (S/cm), L is the
distance between the sensing electrodes (0.45 cm), and W and T are the width and thickness of
the membrane, respectively. Conductivity measurements were made on three individually
prepared membranes and the reported error is the standard deviation of these replicates. Prior

Page 15 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
publications have reported similar in-plane and through-plane ionic conductivity values for
solution-cast isotropic polymer electrolyte membranes;42 future contributions will report the
performance of the membranes described here in electrochemical cells.
3. Results and Discussion
3.1 Equilibrium water content of membranes
The diffusion of small molecules in hydrated polymers is described by free volume
theory, by which solutes diffuse by executing jumps among interstitial voids that open as a result
of polymer segmental motion.43,44 In water-swollen polymers, Yasuda et al.44 proposed and
showed to a first approximation that the water in a hydrated polymer fills the available free
volume. Measurement of the equilibrium water content, therefore, provides insight into the free
volume in the membrane and forms a basis for understanding differences in transport properties
among membranes.
A series of free-standing, unsupported membranes were prepared with monomers 1 and 2
by varying solvent content in the prepolymerization mixture. At mild temperatures, monomer 1
is soluble in both methanol and water, while monomer 2 is soluble in methanol, but only slightly
soluble in water. As a result, no membranes of monomer 2 were prepared in water. After
preparation, membranes were equilibrated in water.
Figure 2(a) shows how the equilibrium water content varied in these membranes with
prepolymerization solvent content (reported here in terms of solvent weight percent for congruity
with prior publications45). In general, water content in the membrane increased with increasing
prepolymerization solvent content. For membranes prepared in water, it is useful to compare the
water content with the parity line (the dotted line in Figure 2(a)), which indicates the case where

Journal of Materials Chemistry A
Page 16 of 40
View Article Online
DOI: 10.1039/C9TA00498J
the equilibrium water content in the prepared membrane was identical to the prepolymerization
water content. Only one of the membranes of monomer 1 prepared in water at 12.5 wt. % fell on
the parity line. To ensure the solubility of monomer 1 at this low water content, the solution and
the quartz plates used for the membrane preparation were heated to 80°C (see Table S1). The
preparation of membranes with lower water contents necessitated higher temperatures (near the
solvent boiling point) and did not yield reproducible results.
All other membranes of monomer 1 prepared in water fell below the parity line, which
means that the membrane contained less water than the initial prepolymerization mixture. The
decrease in water content could be due to polymerization-induced phase separation.46,47 As the
polymer network formed and the crosslink density increased, swelling of the network was
restricted and excess solvent forced out. Even though excess solvent is forced out during
polymerization, only two films were cloudy after polymerization, suggesting a heterogeneous
structure. Membranes prepared with monomer 1 using 50 wt. % water and monomer 2 using 50
wt % methanol transitioned from transparent to translucent as the polymerization progressed. All
other prepared membranes were optically transparent and uniform after polymerization.

Page 17 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
(a)
(b)
Figure 2. (a) Equilibrium water content of membranes prepared with monomer 1 and 2 as a
function of prepolymerization solution solvent content. The dashed line is the parity line, where
the equilibrium water content is equal to the prepolymerization solvent content. (b) Equilibrium
water content of membranes prepared with monomers 1 and 2 as a function of prepolymerization
solution monomer concentration.

Journal of Materials Chemistry A
Page 18 of 40
View Article Online
DOI: 10.1039/C9TA00498J
For membranes prepared in methanol, the solvent was exchanged for water after
polymerization. Since the densities of methanol and water differ, it is useful to compare the
equilibrium water contents of the membranes as a function of prepolymerization monomer
concentration (Figure 2(b)). The equilibrium water content was similar for membranes prepared
with monomer 1 in water and in methanol for a given monomer concentration (in g monomer per
mL prepolymerization solution). This result suggests that the volume fraction of solvent in the
prepolymerization solution (rather than the weight fraction) plays an important role in
determining the equilibrium water content. Correlation of transport properties with sorbed water
volume fraction in crosslinked films will be discussed later.
For similar monomer concentrations, membranes prepared with monomer 2 had lower
equilibrium water content than membranes prepared with monomer 1. One simple explanation
for this result is that the longer hydrocarbon spacer in monomer 2 contributes to greater
hydrophobicity, leading to lower water uptake. However, the crosslink density and IEC in these
systems are also expected to affect the equilibrium water uptake.5 In difunctional monomers 1
and 2, an ionic imidazole moiety is associated with each reactive vinyl moiety. At a fixed pre-
polymerizaton solvent content and degree of polymerization (essentially 100% in all of the
polymers described here), polymers synthesized from monomer 1 were expected to have a higher
crosslink density than those synthesized from monomer 2, since the spacer in monomer 1 was
shorter than that in monomer 2. However, polymers prepared from monomer 1 were also
expected to have a higher IEC than polymers prepared from monomer 2 due to the lower
molecular weight of monomer 1. Based on the molecular weight and number of charged groups,
the IEC of membranes prepared with monomer 1 was 3.87 mmol/g, whereas the IEC was 3.18

Page 19 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
mmol/g for membranes prepared with monomer 2. An increase in IEC is expected to increase
swelling,5 but in this system, an increase in IEC was accompanied by an increase in crosslink
density, which can limit swelling.5 Despite these competing effects, the longer hydrocarbon
spacer in monomer 2 resulted in lower equilibrium water content in polymers prepared from
monomer 2 than in polymers prepared form monomer 1.
The effect of crosslinker content on membrane water content was examined by mixing
difunctional monomers (1 and 2) with monofunctional monomers (1a and 2b). Difunctional
monomer content was varied from 25 mol % to 100 mol % and prepolymerization solvent
content was the same for each pair (1/1a or 2/2a) of monomers. For monomers 1 and 1a,
membranes were prepared with a prepolymerization mixture containing 25 wt. % water. For
monomers 2 and 2a, membranes were prepared with a prepolymerization mixture containing
37.5 wt. % methanol. These prepolymerization solvent concentrations were chosen because the
methanol permeabilities of membranes prepared using only difunctional mononomers at these
concentrations were comparable to commercial Selemion AMV (see section 3.2).
Figure 3 shows that the equilibrium water content of membranes prepared from both
monomer pairs decreased as the crosslinker (i.e., difunctional monomer) content increased. This
result suggests that the amount of water sorbed by the polymer network was limited by the
degree of crosslinking. For monomers 1 and 1a, reducing the crosslinker content to 25 mol %
resulted in membranes that sorbed more water than was present in the prepolymerization mixture
(i.e., this membrane would be above the parity line in Figure 2a). While membrane pliability
qualitatively improved somewhat by decreasing crosslinker content, all of the membranes were
easy to handle.

Journal of Materials Chemistry A
Page 20 of 40
View Article Online
DOI: 10.1039/C9TA00498J
Figure 3. Equilibrium water content of membranes prepared with varying crosslinker content.
Membranes of monomer 1 and 1a were prepared with a prepolymerization solvent content of 25
wt. % water. Membranes of monomer 2 and 2a were prepared with a prepolymerization solvent
content of 37.5 wt. % methanol.
3.2 Methanol permeability of membranes
Methanol is an attractive target product for CO2 reduction, as it could serve as a substitute
for petroleum-derived liquid fuels, or as a feedstock for chemical synthesis.48,49 While longer
C2+ hydrocarbons are desirable end products of CO2 reduction, methanol has been reported as a
product of CO2 reduction on copper and other catalysts.11,49 Methanol is a useful “worst case”
permeant for the present study because it is the smallest of the alcohols, making it an especially

Page 21 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
challenging solute of which to limit the transport in a polymeric membrane. Furthermore, there is
considerable interest in controlling methanol permeation through polymer electrolyte membranes
in direct methanol fuel cells.50–52 Straightforward permeation measurements of methanol
through hydrated membranes have previously been established using in situ FTIR
spectroscopy.33,34
Methanol permeability measurements were performed on free-standing, unsupported
membranes prepared from monomers 1 and 2. As shown in Figure 4(a), increasing the
prepolymerization solvent content significantly increased the methanol permeability. For a given
prepolymerization solvent content, membranes prepared with monomer 2 exhibited much lower
methanol permeabilities than membranes with monomer 1. Figure 4(b) shows the methanol
permeability as a function of monomer concentration. Re-casting the data in terms of monomer
concentration on the basis of prepolymerization solvent volume causes the methanol
permeability values for membranes prepared using monomer 1 in both water and methanol
solvents to converge.
This result reinforces the assertion made in section 3.1 that
prepolymerization solvent volume influences the total water uptake in these membranes and
ultimately governs solute transport (cf., Figures 2(a) and 2(b)).

Journal of Materials Chemistry A
Page 22 of 40
View Article Online
DOI: 10.1039/C9TA00498J
(a)
(b)
Figure 4. (a) Methanol permeability of membranes made with monomers 1 and 2 as a function
of prepolymerization solution solvent content. (b) Methanol permeability of membranes made
with monomers 1 and 2 as a function of prepolymerization solution monomer concentration.

Page 23 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
The effect of crosslinker content on methanol permeability was explored by varying the
difunctional monomer content at a fixed prepolymerization solvent content. As shown in Figure
5, increasing the crosslinker content resulted in a decrease in methanol permeability for both
monomer systems. The relative decrease in methanol permeability was similar for both monomer
systems.
Figure 5. Methanol permeability of membranes with varying crosslinker content. Membranes of
monomer 1 and 1a were prepared with a prepolymerization solvent content of 25 wt. % water.
Membranes of monomer 2 and 2a were prepared with a prepolymerization solvent content of
37.5 wt. % methanol.

Journal of Materials Chemistry A
Page 24 of 40
View Article Online
DOI: 10.1039/C9TA00498J
For all of the prepared membranes, trends in methanol permeability correlated with
equilibrium water content measurements. Yasuda et al.44 proposed that hydration of a swollen
polymer is proportional to its free volume, and showed to a first approximation that polymer
hydration is correlated with diffusive solute transport. Specifically, the diffusion coefficient of a
̅
solute in a swollen polymer 퐷푖 can be expressed as a function of water volume fraction:
1
퐷ꢀ = 퐷0,ꢀexp [−퐴 (
− 1)]
휙푤
where D0,i is the solute diffusivity in pure water, ϕw is the volume fraction of water in the
polymer, and A is an empirical constant.38 This model is most applicable to small solutes (salt
ions and small organic molecules) in dilute and semi-dilute (i.e., highly hydrated) polymer
systems.43
The solution-diffusion model teaches that the permeability of a hydrated polymer to a
solute is the product of the solute diffusivity and solubility in the polymer.53 Yasuda et al.
showed that variations in solute diffusivity that arise with changes in polymer water volume
fraction often strongly affect solute permeation.38,44 As a result, permeability is often highly
dependent on polymer water volume fraction.
In order to understand the methanol permeability measurements in the context of free
volume theory, it was necessary to convert equilibrium water content (Section 3.1) to water
volume fraction. Assuming volume additivity, which has been shown to be reasonable for
several charged polymers,40,41 water volume fraction is easily calculated using water weight
fraction and polymer density. The density of polymers prepared with monomer 1 was 1.42 g/mL,
whereas it was 1.35 g/mL for polymers prepared with monomer 2. The measured density did not
vary significantly with prepolymerization solvent content.

Page 25 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
As shown in Figure 6, the methanol permeability of all prepared membranes varied
exponentially as a function of inverse water volume fraction (1/ϕw). Interestingly, this correlation
holds for all of the prepared membranes, regardless of the monomer spacer length or crosslinker
content. This result suggests that overall polymer free volume (as indicated by the water volume
fraction) is the primary governor of methanol permeability, and that the permeability is otherwise
insensitive to the details of network structure or polymer chemistry. Assuming this relationship
extends outside the studied range of water volume fractions, further decreases in methanol
permeability could be achieved with additional small decreases in water volume fraction. The
lowest water volume fraction of the prepared membranes was ϕw = 0.15. Decreasing the water
volume fraction to approximately ϕw = 0.10 could decrease the methanol permeability by another
order of magnitude.
For a benchmarking comparison, the dotted line in Figure 6 is the measured methanol
permeability of commercial Selemion AMV in the bromide form. The water volume fraction of
Selemion AMV has not been measured and the methanol permeability is therefore represented as
a horizontal line rather than a single point in Figure 6. Several of the prepared membranes had
lower methanol permeabilities than to Selemion AMV. In principal, these membranes could be
more effective at limiting the crossover of CO2 reduction products in a device.

Journal of Materials Chemistry A
Page 26 of 40
View Article Online
DOI: 10.1039/C9TA00498J
Figure 6. Methanol permeability of prepared membranes as a function of inverse water volume
fraction. The thick dashed line is a line of best fit for all of the prepared membranes, which
demonstrates a correlation of methanol permeability with inverse water volume fraction. For
comparison, the thin dotted line is the methanol permeability of commercial Selemion AMV
anion exchange membrane, exchanged to the bromide form.
3.3 Ionic conductivity of membranes
Ionic conductivity measurements were made on membranes prepared under the same
conditions as described in Sections 3.1 and 3.2. As shown in Figure 7(a), ionic conductivity
generally increased with increasing prepolymerization solvent content and, therefore, water

Page 27 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
content. However, membranes prepared with monomer 1 at 50 wt. % water and monomer 2 at 50
wt. % methanol showed a plateau in ionic conductivity between 40 and 50 wt. %
prepolymerization solvent content. Interestingly, the membrane prepared with monomer 1 using
50 wt % water and the membrane prepared with monomer 2 using 50 wt % methanol were both
cloudy after polymerization. The heterogeneous structure of these two films may explain the
deviation of these two membranes from the conductivity trend exhibited by the other
membranes. Other reports on anion exchange membranes have shown ionic conductivity
commonly increases with water content, but the structure and nature of the polymer and charged
group are also critical factors.5,41,54,55
Ionic conductivities of membranes prepared with monomer 1 were moderately higher
than those of membranes prepared with monomer 2. This result is not surprising, since
membranes prepared from monomer 2 had a lower IEC (3.18 mmol/g for monomer 2 vs. 3.87
mmol/g for monomer 1) and water content than membranes prepared from monomer 1.
Figure 7(b) shows the ionic conductivity as a function of monomer concentration.
Similar to the results shown in Figures 2(b) and 4(b), ionic conductivities of membranes
prepared using monomer 1 in water and methanol converged when plotted as a function of
monomer concentration, again highlighting the importance of prepolymerization solvent volume
in controlling membrane transport properties.

Journal of Materials Chemistry A
Page 28 of 40
View Article Online
DOI: 10.1039/C9TA00498J
(a)
(b)
Figure 7. (a) Ionic conductivity of membranes prepared with monomers 1 and 2 as a function of
prepolymerization solution solvent content. (b) Ionic conductivity of membranes prepared with
monomers 1 and 2 as a function of prepolymerization solution monomer concentration.

Page 29 of 40
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C9TA00498J
The effect of crosslinker content on ionic conductivity is shown in Figure 8. As the
amount of difunctional crosslinking monomer was increased, the ionic conductivity decreased.
Relative decreases in conductivity were similar for both monomer systems. This decrease in
conductivity with increasing crosslinker content is likely attributable to the decrease in water
content observed with increasing crosslinker content (Figure 3).
Figure 8. Ionic conductivity of membranes as a function of crosslinker content. Membranes of
monomer 1 and 1a were prepared with a prepolymerization solvent content of 25 wt. % water.
Membranes of monomer 2 and 2a were prepared with a prepolymerization solvent content of
37.5 wt. % methanol.