Electrolyte potentials and impedance measurement of polymer electrolyte membrane CO2 reduction electrolyzer

In electrochemical CO2 reduction reactors, polymer electrolyte membrane (PEM) reactors, also known as zero-gap cells, have great potential for achieving significant CO2 reduction. Because these cells have a thin reactor core with a thickness of several hundred micrometers, it is difficult to determine their internal voltage distribution. To determine the anode voltage, ohmic loss in the membrane, and cathode voltage in the PEM reactors, we set three reference electrodes in the reactor and investigated the voltage values obtained from each reference electrode. We demonstrated that the reference electrode in contact with the anion exchange membrane extending to the outside of the cell provides the most reliable voltage. The voltage measured by this reference, combined with the resistance of the exchange membrane obtained through electrochemical impedance spectroscopy, provides a breakdown of the voltage inside the cell.


Introduction
To prevent global warming, the development of CO 2 reduction technologies is being actively pursued worldwide. Zeroemission technologies, including equipment fueled by hydrogen generated from clean energy sources such as solar power, have been put to practical use in small-scale applications. 1,2) In addition to zero-emission energy technologies, research on carbon-negative technologies that capture and fix CO 2 from the atmosphere or reduce the captured CO 2 to obtain fuels or chemical materials is also essential. In particular, the products reduced from CO 2 including formic acid, ethanol, and ethylene glycol, which are also derived from ethylene, are liquids at room temperature and can be handled without high-pressure tanks. 3,4) By incorporating these chemical generation technologies, a CO 2 capture and reduced facility that is as small as an air conditioning system in a building is obtained.
Various CO 2 reduction reactors have been proposed. The electrochemical CO 2 reduction reactors can be operated at room temperature and are suitable for small facilities. Various types of electrochemical reactors exist, such as H-type cells, electrolyte flow cells, and polymer electrolyte membrane (PEM) cells. These three types of reactors are the most popular for CO 2 electrochemical reduction. H-type cells have a significantly low current density and are often used in research applications such as catalyst development. 5,6) This is because of the existence of large amounts of electrolyte (long distance from the electrolyte between the anode and cathode). Electrolyte flow cells use a porous layer known as a gas diffusion layer (GDL) for the cathode to achieve a high current density operation. 7,8) These cells are used in research applications in which only the cathode side is developed as a half-cell because the electrolyte layer has large ohmic losses.
PEM reactor cells exhibit the highest energy efficiency and are suitable for practical applications. 9,10) The cells are also a type of flow cell with infinitely thin liquid chambers; thus, they are sometimes referred to as zero-gap cells. In most cases, the current path is configured to flow perpendicularly from the backside of the cathode to the backside of the anode and it has no in-plane current path. This current path is scalable to increase the reaction area for practical applications.
PEM reactors for CO 2 reduction have not yet reached a practical level, particularly in terms of lifetime and product selectivity. [11][12][13] In addition, the operating voltage is at least 1 V greater than the thermodynamically calculated reaction energy; therefore, there is still significant room for development in terms of energy efficiency. There are various possible factors that affect the reactor voltage: (1) A significant applied bias in hydrogen generation due to CO 2 reduced CO adsorption on the Cu catalyst surface. 14) (2) An increase in membrane resistance due to the effect of ionic species carrying charges from OH − to HCO 3 − in the ion exchange membrane. 15) (3) An increase in thermodynamic potential due to local pH change in the electrolyte. 16) Some of these factors are particularly likely to occur in PEM reactors. Therefore, it is necessary to measure the voltage distribution in a PEM reactor for CO 2 reduction.
One of the difficulties in measuring the voltage distribution in developing PEM reactors is the absence of an electrolyte chamber between the cathode and anode catalysts. This makes it difficult to use internal measurement devices such as reference electrodes. 12,17,18) In this study, we used a PEMtype CO 2 reduction reactor with a Cu catalyst and added three reference electrodes at the electrolyte inlet for the anode, backside of the anode GDL, and outer edge of the ion exchange membrane. We simultaneously measured the potential difference between the cathode and the reference electrode. We also studied the obtained voltage differences and their stabilities. A significant potential difference between the front and back of the anode GDL was expected, particularly at low electrolyte concentrations. At high current densities of several hundred mA cm −2 , ohmic losses occurred even in ion exchange membranes that were several tens to hundreds of micrometers thick. We identified the anode, cathode, and ohmic voltages in the ion-exchange membrane through electrochemical impedance spectroscopy (EIS). These EIS measurements, in addition to the voltage distribution, were used to compare the conditions of the CO 2 supplied to the cathode with that of the Ar supplied.

PEM reactor for CO 2 reduction
The structure of a typical PEM reactor for CO 2 reduction is shown in Fig. 1. The anode catalyst layer was IrOx-coated on mesh-like porous Ti. The cathode catalyst of the Cu nanoparticles was in contact with an anion exchange membrane. The anolyte was supplied from the side of the anode and CO 2 gas was supplied from behind the cathode. The main part of the CO 2 reduction reaction occurred at the point of contact between the CO 2 gas and the electrolyte on the cathode catalyst.
The lifetime of CO 2 reduction is often limited by either increased cell resistance or decreased Faradaic efficiency of the product caused by changes in the reaction conditions within the reactor. Although the main CO 2 -reduced products are CO, CH 4 and C 2 H 4, in addition to the parasitic reaction of H 2 generation from H 2 O, CO 2 reduction is observed to decrease as the reduction conditions change. This is believed to be due to electrolyte leakage, 13,19,20) or salt deposition at the cathode GDL, 10,21,22) which prevents CO 2 and electrolytes from reaching the cathode catalyst properly. As a sign of such material supply problems in the reactor, cell voltage oscillations in the case of the constant current operation and current oscillations in the case of constant voltage operation are often observed. 23) This shows that it is possible to operate the CO 2 reduction reactors for longer periods of time and at higher Faradaic efficiency by optimizing the reactor drive conditions. Therefore, the observation of the reactor condition, particularly for the voltage distribution, is practically important.
The potential difference between the anolyte and anode electrode is the anode voltage, and that between the catholyte and cathode electrode is the cathode voltage. Notably, the electrolyte potential depends on the measurement conditions and electrode used. In PEM reactors for CO 2 reduction, the potential of the anolyte can be measured externally through the anolyte. These externally measured anolyte potentials can be regarded as the correct electrolyte potentials when the voltage drop owing to the current being small.
Correct electrolyte potential measurement is important, as discussed. The ion-exchange membrane itself is regarded as a conductive probe for the development of fuel cells and water electrolyzers. In this case, a small hydrogen electrode that serves as a reference electrode is placed around the periphery of the cell. 24) A convenient electrolyte-type reference electrode can be used instead of the hydrogen electrode. Because the CO 2 reduction reactor used in this study has a membrane supplied with electrolyte from the anode side, the same electrolyte should be filled between the reference electrode and membrane to avoid affecting the experiment. Some studies have incorporated such a reference electrode in a CO 2 reduction cell. 25,26) However, the electrolyte potential far from the catalyst layer may be affected by small leakage currents from the cell housing. A significant potential fluctuation in the cell can exist when a low-concentration electrolyte is used owing to its very high impedance. It is difficult to evaluate the accuracy of electrolyte potential results. It is also important to note that the potential difference due to the electric double layer on the membrane surface must be removed and measured to accurately determine the ohmic component of the membrane and electrolyte. 27)  Three reference electrodes were used to determine the best position for measuring the electrolyte potential. From these potential analyses and the ion exchange membrane resistance obtained through EIS measurements, we were able to breakdown the anode voltage, cathode voltage, and ohmic losses in the PEM reactor, as shown in Fig. 2.

Experimental methods
A schematic of the homemade PEM reactor used in this study is shown in Fig. 3. This type of cell is known as an exchange cell in which the liquid flows only from the anode. Three Ag/ AgCl reference electrodes were set, as shown in Fig. 3. The membrane electrode was set in an electrolyte pool made of  The Japan Society of Applied Physics by IOP Publishing Ltd PTFE sheets around a membrane that extended outside the housing. The anode back electrode was set at the extended silicon tubing through a hole drilled through the serpentine channel for the anode electrolyte. The inlet electrode was set in the serpentine channel with a tee connector on the tubing to supply electrolytes to the serpentine channel. All these electrodes can be used simultaneously. To avoid insulation by oxygen bubbles at high currents, the tubes containing the anode back and inlet electrodes were operated as the electrolyte inlet. The anode back and membrane electrodes were carefully insulated apart from the metal part of the cell to avoid being affected by the cell housing surface. The two inlet tubes of the anode back and inlet electrodes were driven using different pumps to prevent potential mixing in the solution flow path.
The structure of the membrane electrode is as follows. To prevent the membrane from being damaged by mechanical pressure, a thin channel is provided in the PTFE insulating gasket, which leads to the exterior of the cell. In this section, the ion exchange membrane is 180 μm thick with three folds, which is almost the same as the channel thickness of 190 μm. The electrolyte pool, which connects the membrane and Ag/ AgCl electrode, was filled with the same electrolyte as the anolyte. There was some electrolyte leakage from the inside of the cell through the channels; therefore, the pool did not dry out during the experiment.
The Ag/AgCl reference electrodes (ALS RE-1B) contain internal electrolyte with a concentration of 3 M NaCl, and the junction is made of ion-permeable porous glass. All the reference electrode potentials were measured as potential differences from the cathode. A multichannel data logger (Graphtec GL-240) was used for the measurements. The input resistance of the data logger was approximately 1 MΩ, which is not suitable for precise electrochemical measurements. Thus, we inserted an impedance isolation circuit using the voltage follower of a J-FET operational amplifier (TI TL072CP) between the data logger and reference electrode. The input bias current of this amplifier was approximately 1 pA and the effective input resistance was approximately 1 TΩ.
The cathode GDL comprised commercially available hydrophobic carbon paper (Sigracet 28BC) spray-coated with Cu nanoparticles (Taiyo Nissan, 100 nm diameter) at 0.2 g cm −2 . The anode GDL and catalyst were commercial IrO 2-coated Ti fiber meshes (TanakaT211124R). The area of both GDLs was 5 cm 2, and the ion exchange membrane between them was an anion exchange membrane (Sustainion 37-50).
A homemade cell was used in this study. Serpentine channels with a cross-sectional area of 1 mm square were carved on the current-collecting plates of both electrodes, and there was a hole in the center of the anode channel to connect the anode back electrode.
The reactor operation in the experiment was first initialized by cyclic voltammetry repeated four times from 0 to 3.5 V cell voltage, followed by constant current operation at current densities of 20, 100, 200 and 300 mA cm −2 for 5 min each to measure the cell voltage and potential of the reference electrode. EIS measurements were subsequently performed by superimposing an alternating current of 4% amplitude on each of the current densities used in the constant-current operation. The above experiments were performed for all   Figures 4(d)-4(f) illustrate the results of the same measurements with Ar instead of CO 2 as cathode gas. Because no CO 2 is supplied, almost 100% hydrogen evolution occurs as the reactor operates. There is a significant difference in cathode voltage at current densities higher than 100 mA cm −2 between CO 2 reduction and hydrogen generation. This is believed to be the effect of the intermediate product, CO, being captured on the Cu surface and suppressing H2 generation. 14) Since the thermodynamic potential of the CO generation reaction is more negative than H 2 generation by about 0.1 V, the effect of the CO 2 feed is uncertain in our experiments at a current density of 20 mA cm −2 .

Results and discussion
The potentials of the three reference electrodes were relatively similar when the electrolyte concentration was as high as 1 M. There was a difference in the potentials of the exchange membrane electrode and the other electrodes when the anolyte concentration decreased to 0.01 M. The potential  of the membrane electrode is the lowest among the three reference electrodes, which reflects the fact that the membrane electrodes are physically closer to the cathode electrode as the ground than the inlet and anode back electrodes. In several cases, the inlet electrode exhibited the highest potential among the three reference electrodes. This was an effect of the small amount of current flowing from the cell housing. While the anode back electrode reliably measures the potential of the backside of the anode GDL, the inlet electrode is affected, on average, by the potential of the entire anode GDL. As a result, the anode back and inlet electrodes exhibited similar values, as shown in Fig. 4(c).
The values of the inlet and anode back electrodes also vary with time for an electrolyte concentration of 0.01 M. This probably indicates that the potential of the electrolyte itself varies owing to the nonuniform current distribution in the cell because the current flowing through the measuring device is sufficiently small. In contrast, the potential of the membrane electrode was stable and reliable because of the relatively small impedance of the membrane. Figure 5(a) shows the Cole-Cole plot of the cell (between the anode and cathode electrodes) obtained through EIS at a current density of 200 mA cm −2 . Table I lists the resistances of the electrolytes obtained from the EIS results. To separate the series resistance from the resistance in the electric double layer, we used the general interpretation that the series resistance is the value at which the impedance on the highfrequency side crosses the real axis in the Cole-Cole plot, where the effect of the electric double layer disappears. The series resistance, except for that of the electrolyte, was evaluated separately and observed to be less than 10 mΩ. Therefore, the resistances summarized in the table are attributed to the electrolyte membrane.
The ohmic component increases significantly as the electrolyte concentration decreases. The supply of CO 2 also significantly increases the resistance of the membrane. This probably corresponds to an increase in the ratio of HCO 3 − , CO 3 2− and OH − , which are the charge-carrying anions in the membrane. There is no clear distinction in the Cole-Cole plot between the resistive component of charge transfer at the catalyst surface (half circle) and the diffusive component at lower frequencies. It is likely that several components are mixed owing to the complexity of the reaction.
The resistance of the electrolyte membrane obtained by EIS multiplied by the cell current gives the ohmic drop in the   cell. Considering that the value of the membrane electrode, which is the most appropriate reference electrode for the three electrodes, is the average value of the potential distribution uniformly distributed in the thickness direction of the membrane, the voltage breakdown of the cell was determined, as shown in Fig. 6. The anode voltage, cathode voltage, and ohmic drop illustrated in this figure were obtained from the cell voltage and potential of the membrane electrode, averaged over the last 20 s in each current step shown in Fig. 4. We calculated the difference between the cathode and membrane potentials for the cathode voltage and the difference between the anode and membrane potentials for the anode voltage, and subtracted half of the ohmic drop from each. The effect of low electrolyte concentration on the voltage distribution was more drastic under the CO 2 feed. The cell voltage in the case of 0.01 M electrolyte reaches 1.5 times that in the case of the 1 M electrolyte. Regardless of the electrolyte concentration, the cell voltage and reference electrode potential at low current densities were almost the same in all the experiments. This means that the reaction occurring in the lowest-overpotential region was independent of the ionic composition of the electrolyte.
In the case of a lower electrolyte concentration, the pH gradient inside the cell was stronger with increasing current because the buffering capacity of the supporting electrolyte was reduced. This increased (the component of pH difference) the anode and cathode overpotentials and the cell voltage. 28) In fact, in all experimental cases, the lower the electrolyte concentration and the higher the current density, the greater the anode and cathode voltages.
A summary plot of cyclic voltammetry during the initialization process for each condition is shown in Fig. 7. The cell current at the same cell voltage decreases with decreasing electrolyte concentration or CO 2 feed. The breakdown results obtained in Fig. 6 show that the main factor causing this difference is not the membrane resistance, but the cathode overpotential. The anode overpotential does not vary considerably with the conditions; however, it is significantly increased for an electrolyte concentration of 0.01 M. We could not clearly distinguish from the EIS results illustrated in Fig. 5 whether the change in cathode overpotential was due to charge transfer resistance or mass transfer limitation.

Conclusions
We added three reference electrodes to the PEM reactor for CO 2 reduction to determine the potential of the electrolyte and examined the voltage values. Three reference electrodes were placed at the membrane, inlet, and anode back.
When the electrolyte concentration was higher than 0.1 M, the three electrodes exhibited stable voltage values. When the current density in the cell was low, the three electrodes exhibited almost identical values. However, at high current densities, the voltages of the reference electrode for the membrane and the inlet and anode back exhibited significantly different values, even at high electrolyte concentrations. These differences occurred not only with the ohmic losses in the ion-exchange membrane but also with the potential distribution in the electrolyte. The difference in the potential between the membrane electrode and the other two was approximately 0.2-0.5 V, which is unacceptable in electrochemical measurements. Therefore, a reference electrode in contact with the ion-exchange membrane is a suitable method for breaking down and observing the voltage inside the cell in a PEM reactor.
The amount of ohmic loss in the ion-exchange membrane was estimated through EIS measurements. The resistance of the membrane was significantly affected by the electrolyte concentration and gas feed. We demonstrated that using the cell voltage, the voltage of the membrane reference electrode, and EIS measurements, we can breakdown the cell voltage into the anode voltage, ohmic loss in the membrane, and cathode voltage.