CO2 electrolysis to form CO with the use of electricity from renewable sources is a promising technique in closing the carbon cycle. Via the electrochemical reduction CO2 can be converted into different hydrocarbon products. These value-added products are for example CO, HCOOH, CH3OH or C2H4. The direct electrochemical route of CO2 reduction is often tested on three metrics: the faradaic efficiency, the current density and the energy efficiency. The implementation of CO2 reduction has some technical challenges. One of which is controlling the local pH and carbonate formation inside the electrochemical cell producing hydrocarbon products. Controlling the pH inside the electrochemical cell becomes an issue because of the production of hydroxide anions that increase the pH. A high faradaic efficiency is affected by the local pH. With an increasing pH CO2 is consumed to form carbonate and Hydrogen Evolution Reaction (HER) is favored. Mass transfer limitations result in the pH increasing near the electrode. The local pH is dependent on the electrochemical reaction, buffering reactions inside the electrolyte and mass transfer within the electrolyte flow. Previous studies have used numerical modelling to obtain a 2D transport model to present concentration gradient along the cathode or inside the cathode. Other studies that show the pH experimentally do this using scanning probe techniques or measure the intensity of dye sensitive to pH.
In this project the effects on the local pH are studied with the use of Fluorescence Lifetime Imaging Microscopy (FLIM). This method is able to image the local pH with the use of a fluorescent dye that has a lifetime dependent on the local surroundings. This thesis researched the effects of three process parameters: electrolyte anion type and concentration, the catholyte flow rate and the current density. The electrolytes studied are: 0.1 M KHCO3, 1 M KHCO3 and 0.4 M K2SO4. The catholyte flow rates studied are corresponding to Reynolds number 0.8, 8 and 47. The different current densities are −1, −5, −10 and −50 mA cm-2 . The effect of these parameters was studied with performing electrochemical tests and studying the cell potential.
This study obtains a fluorescence lifetime-pH calibration curve inside an electrochemical cell. From this calibration curve could be concluded that we have a clear trend above pH 9 to the phase-shift fluorescence lifetime. We also concluded that the used salt KHCO3 is likely to have an effect on the performance of the alpha dye. An unexpected effect of using FLIM onto the spatial resolution in an electrolyser was due to the presence of bubbles. We suggest a more elaborate study into the effects of the settings used in the FLIM method.
The results show that the characteristics affected by the anion type and concentration of electrolytes can be described in the buffer capacity and conductivity of the electrolytes. With increasing buffer capacity and bicarbonate concentration the pH difference between the bulk and near the cathode decreased. The study also found that the overall cell potential was increasing with increasing conductivity of the electrolyte. We cannot distinguish a clear effect of the concentration overpotential as an effect of the buffer capacity onto the overall cell potential.
Additionally, the study found that a higher Reynolds number leads to a decrease in potential due to lower concentration overpotential and better gas removal. From a study on the current density and its effects could be concluded that even though we expect laminar flow with these Reynolds number mass transfer is occurring perpendicular to the flow of the electrolyte. This is an effect of the formation of hydrogen or carbon monoxide bubbles inside the electrolyte that improve mixing. This was also shown in the decrease of pH increase near the cathode with increasing Reynolds number. For the highest current density an overall increase of the bulk pH was linked to the improved mixing due to gas bubbles. The effect of gas bubbles onto the local pH and the cell potential should not be underestimated. Improvement of gas removal with a higher Reynolds number of the electrolyte flow rate or in other ways remains a recommended field of research. We suggest the use of Particle Image Velocimetry alongside of FLIM to obtain more insights into the flow profile inside the catholyte affected by the formation and removal of gas bubbles. Finally, we recommend to research the system with the implementation of the gas channel and different types of membranes.

Electrochemical CO₂ reduction may present a solution to close the carbon cycle and to utilise CO₂ emissions. However, for this technology to have a significant impact, it has to be successfully implemented on an industrial scale. Numerical simulations can aid with the study of process parameters and reactor design.

The overall aim of this project is to develop and utilise a numerical model that can describe phenomena arising in the CO₂ electrolyser inside the flooded catalyst layer (CL). First, the general operation of the electrolyser is addressed, and this is extended for the effect of liquid flow rate, electrolyser length, and operating pressure. To investigate the performance and limitations arising at the large-scale, the model is scaled-up to describe a one meter long electrolyser. The study is concluded with two considerations that could improve the electrolyser performance. These points were addressed by developing a 2D numerical model of a gas diffusion-based CO₂ electrolyser in COMSOL Multiphysics. We assessed the performance of the electrolyser in terms of current density, reflecting rate of species formation, and of faradaic efficiency for CO (FE), reflecting selectivity towards the desired product.

Investigating the small-scale electrolyser we find that at high current density (200 mA cm^-2), the pH in the CL immediately increases by 3 units and further diagonally increases from pH 10.1the inlet to 12.2 around the outlet. When operating the electrolyser with excess of CO₂ supply, we find the CL to perform the best near the gas phase boundary (311 mA cm^-2, 95% FE), while the regions close to the electrolyte are underperforming (250 mA cm^-2, 89% FE). This shows that the performance in certain regions of the CL needs to be improved.

When scaling-up the electrolyser to a length of one meter we find that the performance does not change dramatically when operating at excess of gas supply. However, if a high CO₂ conversion should be achieved, the long electrolyser shows a 10% decrease in FE, and CO₂ conversion compared to a small-scale electrolyser, at the same level of current density (115 mA cm^-2).  Analysing the current density locally, we find that difference between the inlet and outlet can be as large as 100 mA cm^-2. Next, we find that FE can fall to almost 50% around the outlet. This shows that when adding extra length to the long electrolyser, this extra length only adds a fraction of its potential performance.

The uneven utilization of the catalyst can be improved by varying the catalyst loading along the electrolyser length. This improves the FE by around 5% while using 40% less catalyst. We also find that while the current density is slightly lower, the amount of product generated per mass of catalyst has significantly increased. This shows that carefully engineering the catalyst loading can save the amount of catalyst needed and could potentially improve the cost-effectiveness of the CO₂ electrolyser.

In all cases the performance over the CL is unevenly distributed. To achieve a higher performance, research needs to find ways how to enhance the performance also in the poorly utilised regions of the CL. Scaling-up the electrolyser just by extending its length proves inefficient and inevitably leads to a lower performance. The beneficial buffering effect provided by the electrolyte at a small-scale does not translate to a large-scale. At this moment, performance of large-scale CO₂ electrolysers seems satisfactory only when operating at very low CO₂ conversion. From the investigated parameters that address the performance issues, higher operating pressure and smart catalyst loading seem only promising options, however, other options should be found to speed up the development.