Anion-exchange membranes with internal microchannels for water control in CO2 electrolysis

Electrochemical reduction of carbon dioxide (CO2R) poses substantial promise to convert abundant feedstocks (water and CO2) to value-added chemicals and fuels using solely renewable energy. However, recent membrane-electrode assembly (MEA) devices that have been demonstrated to achieve high rates of CO2R are limited by water management within the cell, due to both consumption of water by the CO2R reaction and electro-osmotic fluxes that transport water from the cathode to the anode. Additionally, crossover of potassium (K+) ions poses concern at high current densities where saturation and precipitation of the salt ions can degrade cell performance. Herein, a device architecture incorporating an anion-exchange membrane (AEM) with internal water channels to mitigate MEA dehydration is proposed and demonstrated. A macroscale, two-dimensional continuum model is used to assess water fluxes and local water content within the modified MEA, as well as to determine the optimal channel geometry and composition. The modified AEMs are then fabricated and tested experimentally, demonstrating that the internal channels can both reduce K+ cation crossover as well as improve AEM conductivity and therefore overall cell performance. This work demonstrates the promise of these materials, and operando water-management strategies in general, in handling some of the major hurdles in the development of MEA devices for CO2R.


Definition of Transport Properties:
In the ionomer domains, the conductivity of the electrolyte is In the above expression, is the conductivity of a liquid-equilibrated AEM, which is set to a constant value of 20.6 mS cm -1 (See Table S1 for a summary of parameter values used in the simulation). 1 is the conductivity of a vapor-equilibrated AEM, which is a function of the vapor activity (a w ) by = 0.003 (8.1432 ) S M is defined by an empirical relationship roughly related to the interior surface energies and waterphase network. 18 When S M is 1, the ionomer is fully liquid equilibrated, when S M is 0, the ionomer is fully vapor equilibrated.
of the porous electrode is defined to be 220 S m -1 for the diffusion medium, and 100 S m -1 for the catalyst-layer domains.
Figure S1 -Schematic of a single pore within the porous catalyst layer in the electrochemical model. As simulated, the porous catalyst layer is assumed to be a homogeneous continuum of CL pores with volumes defined as shown above.
Lastly, all conductivities in the porous-electrode domains (shown schematically in Figure S1) are corrected for tortuosity and porosity using the Bruggeman correlation, where is the volume fraction of the phase of interest. For the ionomer or membrane phase, where is the volume fraction of ionomer in the pore space and is the volume fraction of the solid volume of the porous electrode.
The diffusion coefficients in the gaseous phase are: where is the mole fraction of species i. The gas phase volume fraction, , is: where S is the CL or PTL liquid saturation.
Lastly, because water activity or chemical potential in the ionomer cannot be readily measured or observed, the simulated water activities are converted to membrane water content, , by the following semi-empirical expression: 2 = 30.752 3 -41.194 2 + 21.141 (8) where is the water content of vapor equilibrated AEM and is the water content of a liquid equilibrated AEM (set to a constant value of 17) 3 .

Source Term Definitions
, represents the molar source terms of species i due to charge transfer reactions, respectively, where s i,k is the stoichiometric coefficient of species i in reaction k, and n k is the number of electrons transferred in reaction k.
For water vapor, an additional phase-transfer term related to the modeled transfer of water from the liquid or ionomer to the gas phase is required, where is the mass-transfer coefficient between the vapor phase and hydrated ionomer phase, , RH is the relative humidity, and p L is the bulk pressure of the liquid phase. The first term in the above equation describes mass transfer between vapor phase and the hydrated CL ionomer. The second term describes water evaporation or condensation in both the CL and PTL. A mass transfer coefficient of and implementation of the Heaviside step function ' = 10 7 -3 -1 0 ( ) ensure that RH is always 100% when liquid water is present and that the RH never exceeds 100%.
Similarly, for liquid phase water: where is the pressure of liquid water in the membrane. Again, the first term describes transfer . between the liquid and ionomer phases, and the second term describes evaporation or condensation.
Additionally, the phase-transfer source term associated with water in the ionomer phase is given It is important to note that, while for vapor-or liquid-phase water there was no charge-transfer source term, there is a source term associated with the consumption of water by charge-transfer reactions in the ionomer phases. The phase-transfer source term associated with water in the ionomer phase is given as Q p describes the source term into or out of a given phase p. For the gas phase, the expression is For liquid phase, S7 Table S1: Parameter values for simulation.

Tortuosity Calculations
To calculate the tortuosity of the ionically conducting medium, we consider an applied potential for which the membrane is fully hydrated, and there are no variations in conductivity across the domain, . The ionomer conductivity at this potential is 20.6 . The power = 1.6 -1 loss due to ohmic losses throughout the modified ionomer domain is calculated as: 8 where is the local ionic current density vector, and is the local AEM conductivity. The calculated power loss represents the loss of power through the ionomer, accounting for the tortuous pathway of the ions around the water channel.
If there were no tortuous pathway, the power loss would be the ideal power loss: where A is the through-plane area of the AEM. We can use the above expression to determine an effective conductivity of the ionomer using the real ohmic power loss. Essentially, this value indicates the corresponding conductivity of a membrane without a channel that has the same ohmic power loss: This value is reduced compared to the bulk AEM conductivity of 20. 6 ..

S10
The calculated effective conductivities and tortuosities for the bilayer AEM with varying channel spacings (H) can be found in Table S2 below.

Custom-made Electrolyzer
Our custom-made electrolyzer chassis was assembled from two milled polyetheretherketone (PEEK) plates. One of the plates had four entrances: an inlet and outlet for the flow field, and two connections for electrodes. Gold spring contact electrodes were used to apply voltage/current to the electrolyzer. The other plate ( Figure S1) had six entrances: the same four as the first plate, and an extra inlet and outlet for the membrane internal microchannels. The cathode was a silver-coated GDE and the anode was an iridium-oxide coated GDE. Both of these GDEs were 2.25 cm 2 and were prepared by sputter coating (AJA International Sputter Machine) pure Ag and Ir, respectively, onto a Toray TGP-H-060 porous carbon paper (Alfa Aesar).

S13
To measure the membrane resistance with different electrolyte concentrations in the microchannels, two T-junctions were added in the inlet tubing close to the electrolyzer plates. In those junctions, two leakless Ag/AgCl reference electrodes were connected. This way, the voltage drop was measured between the electrolyte inlets. Since current flows only between the electrodes, and the membrane separates the electrodes, the voltage drop across the membrane is measured.
The operating conditions can be found in the main text.

Electrolysis Experiments
A humidified CO 2 stream was introduced into the cathode side of the reactor for the electrolysis experiments. The gas stream was humidified by bubbling dry CO 2 through a sparger S14 into a DI water column at room temperature, and the relative humidity was measured with a humidity sensor. A 0.1M KOH solution or a humidified N 2 stream were fed as reactants for the oxygen evolution reaction on the anode side. Linear sweeps of the voltage from 0 to 4 V, while measuring the current, were made in triplicate to assess the effect of the different electrolyte concentrations in the microchannels.
To quantify the K + crossover to the cathode side, 0.1 M KOH was fed into the anode channel, and a constant current of 5 mA/cm 2 was applied for 73 minutes. Aliquots of the anolyte and of the electrolyte in the internal membrane channels were taken before and after current was applied. At the end of the experiment, the serpentine flow channels in the cathode plate were rinsed (collecting the liquid), and the cathode GDE was placed in 30 mL of aqueous solution containing 5 mL isopropanol and approximately 1.5 mL of concentrated HCl. This solution was chosen to counter the GDE's hydrophobicity and to dissolve any potassium salts that had deposited. Five aliquots were analyzed using ion chromatography and a mass balance on the K + was made. S15 Figure S5 -Electrolysis setup.

Supplementary Results
Water Content Distribution at 2.5 V for Base Case (Spacing of 360 μm) Figure S6 -Water content distribution in the bilayer AEM with a curved channel geometry and spacing of H = 360 µm. The applied potential for this simulation is 2.5 V.  Figure S10 -(a-  That means the ratio between co-ions and counter-ions can be rewritten as:

Local Water Content and Current Density for AEMs Where the Microchannel Contacts the CL
That means the ratio increases as c increases, which is equivalent to a relatively high sorption coefficient of co-ions in the membrane at high external concentration. This has also been confirmed by experiments. 9