Electrochemical conversion of carbon dioxide to high value chemicals using gas-diffusion electrodes

Introduction

Fossil fuel, which includes coal, natural gas and oil, besides being a primary source of energy, is also the primary source of carbon atoms for making chemicals. However, its reserve is limited and localized to some regions in the world. The price of fossil fuels is also sensitive to geopolitics and wars. These limitations can be circumvented by synthesizing fuels and chemicals from new feedstocks such as CO2 and H2O [1]. Recycling of CO2 can also be a potential solution to reduce the accumulation of atmospheric CO2, brought about by burning fossil fuels. For example, we can abate anthropogenic CO2 by converting it into useful chemicals right at the emission source, rather than releasing it to the atmosphere [2].

The electrochemical reduction of CO2 is one of the promising methods to convert this greenhouse gas into chemicals namely carbon monoxide, formic acid, methanol, methane, ethylene, ethane, ethanol, acetic acid, propanol, etc. [3,4,5••]. A big advantage of an electrochemical approach is that it can utilize excess electrical energy generated from intermittent sources such as solar and wind [6]. Because of the unpredictable nature of electricity generation from these sources, electricity supply may not match demand throughout the day, and sometimes, this mismatch results in surplus electrical energy. In such a scenario, the electrochemical CO2 reduction process is a potentially efficacious way to store all excess energy as transportable chemicals and fuels.

Over the last four decades, there has been intense research to discover and tailor the selectivity and activity of catalysts for the electrochemical reduction of CO2, and to unravel its underlying operating mechanisms. Works on scaled-up systems for electrochemical CO2 reduction reaction have also been recently reported. In this review article, we summarized recent advancements, particularly focusing on large scale electrochemical CO2 reduction in low temperature (<100°C) electrolyzer systems from year 2017 and onwards. Our article is organized as follows: we first discussed the electrochemical CO2reduction reaction, its figures of merits, and its scale-up methods. We then present notable developments for the electrosynthesis of some selected products. Finally, directions for future research are suggested.

Electrochemical CO2 reduction

We focus on CO2 reduction products that have high economic values and that have been synthesized with high yields. These are, namely, carbon monoxide, formic acid, ethylene, and the alcohols (methanol, ethanol, and propanol). The electrochemical reactions for the formation of some of these products, including their electron transfer numbers n and standard potentials E0 are shown in Table 1.

Table 1. Electrochemical reactions, electron transfer numbers and standard potentials for the formation of various CO2 reduction products

Reaction	Electron transfer number n	Standard potential [3] E0 versus SHE (at pH 7.0)
CO2 + H2O + 2e− → CO + 2OH−	2	−0.52 V
CO2 + H2O + 2e− → HCOO− + OH−	2	−0.43 V
2CO2 + 9H2O + 12e− → C2H5OH + 12OH−	12	−0.33 V
2CO2 + 8H2O + 12e− → C2H4 + 12OH−	12	−0.34 V
3CO2 + 13H2O + 18e− → C3H7OH + 18OH−	18	−0.32 V

Catalysts are needed to improve the yield of all these reactions, which involve transfer of two or more electrons and protons in multiple steps on a catalyst surface [7]. Zn, Ag, and Au primarily catalyze the formation of carbon monoxide, while Sn, Pb, In, and Bi catalyze formic acid formation [7]. Copper occupies a unique position among all metals, because it is the only metal that can facilitate C single bond C coupling to give significant amounts of multi-carbon molecules, albeit with rather poor selectivity [8]. The mechanisms for these reactions are still under research. A large number of articles, elucidating the details of each of the reactions, are available. Interested readers could refer to the review articles by Hori, Koper, Yeo, and most recently, by Chorkendorff [3,7,8,9•].

Figures of merit for electrochemical CO2 reduction

In an electrochemical reaction, product yield can be estimated from the current density. Multiple reactions producing two or more products usually happen in parallel during an electrochemical CO2 reduction reaction process. In the presence of two or more concurrent reactions, ‘Faradaic efficiency’ (FE) is used to indicate the percentage of electrons that contributes to the formation of a particular product. FE indicates the selectivity of the reaction toward a product, and can be calculated from the expression: $F E = n \times F \times \frac{η}{Q}$ n, F, η, and Q, are respectively, the electron transfer number for a product, Faraday constant (96,485 C/mol), the product amount (in mol), and the total charge. The charge can be determined by integrating current with the time. Multiplying FE with total current density gives the partial current density of a product.

The ‘energy efficiency’ of a system, is another crucial figure of merit. This is related to the operational voltage of the electrolyzer, which depends on the overpotential required by the electrodes and ohmic losses in the system. The energy efficiency, ξ, is determined by: $ξ = \sum_{i} \frac{E_{i}^{n} {F E}_{i}}{E_{c}}$ $E_{i}^{n}$ , ${F E}_{i}$ , and $E_{c}$ are thermo-neutral voltage for a product i, Faradaic efficiency of the product i, and total cell voltage, respectively [10]. Working stability of the catalysts and cost of the systems are also key figures-of-merit [5••].

Scaling up of electrochemical CO2 reduction processes using gas diffusion electrodes

High performance electrodes are necessary to generate products with yields required for industrial scales. Electrodes such as single crystal metal surfaces, studied in the lab using a H-cell, are incapable of this due to mass transport limitation of CO2. This arises because of the poor solubility of the reactant CO2in aqueous electrolyte (33 mM at 298K and 1 atm). The mass transport constraint can be managed by either filling the H-cell with CO2 at higher pressures [11] or by using gas diffusion electrodes (GDE). The latter type of electrodes are widely used in fuel cells and other electrochemical processes. GDEs can circumvent the mass transport bottle-neck and improve the yield by two orders of magnitude or even more [12,13•,14].

In the simplest terms, the structure of a typical GDE (Figure 1a) consists of an electrically conducting porous support layer (I), and a micro-porous gas diffusion layer (II) (GDL) [13•,14]. The catalysts are coated on the interior wall of channels in the GDL. In an electrolyzer, the GDE stays between the gas and electrolyte compartments [15•,16]; the electrolyte partially penetrates those catalysts-coated channels and forms a three-phase boundary. During a reaction, gaseous CO2 directly approaches the catalytic centers at those three phase boundaries through tiny channels from the gas side of the electrode. Note that the bulk electrolyte and CO2 will not be considerably mixed to give carbonates. Thus, alkaline electrolytes, which have been experimentally shown to improve C2+ product, can be used [7,17,18]. In contrast, in an electrochemical system housed in a H-cell, the gaseous CO2 usually has to dissolve in the bulk electrolyte first before arriving at the catalytic centers. This arrangement gives rise to the CO2 mass transport bottle-neck, due to poor solubility of the CO2. It also rules out the use of alkaline electrolytes owing to its rapid reaction with CO2.

The electrolyte in an electrolyzer can be a liquid (Figure 1b) or a solid polymer ionic membrane (Figure 1c). The first type of electrolyzers is popularly known as flow electrolyzers or flow cell, whereas the second type are known as polymer electrolyte membrane (PEM) electrolyzers or PEM cell. In a flow electrolyzer, the ionic membrane separates the anodic and cathodic compartments to prevent liquid products from reaching the anode. Usually, two pumps feed the anolyte and catholyte, respectively, to the anodic and cathodic compartments in flow electrolyzers [19••]. In PEM electrolyzers, the feed-in constitutes of humidified gases or a mixture of liquid and gas, and a solid polymer ionic membrane is used as the electrolyte. The solid polymer electrolyte membrane is placed between the anode and cathode GDEs, and the catalyst coated side of each GDE is facing the electrolyte membrane. The reader can refer to the article by Vennekoetter et al. for more details [15•].

Modelling and simulation of GDEs and electrolyzers play an important role for the design and optimization of a system [20•]. However, such modeling and simulation works are comparatively rare. A steady-state isothermal model, for CO generation, studying the effects of applied cell potential, feed in concentration, feed flow rates, channel length, and porosity of GDE was developed by Wu et al. [21]. A modelling work on GDEs to study the effects of catalysts layer, hydrophobicity, loading, porosity, and electrolyte flow rate was also presented by Weng et al. [22].

Techno-economic analysis is also important to assess the economic feasibility and scalability of CO2 reduction [23, 24, 25]. One recent analysis [25] indicates that CO2 reduction can become competitive against fossil fuel feedstock, when electrical to chemical energy conversion factor is above 60% for carbon monoxide, ethanol and ethylene, assuming electricity cost is less than 4 US cents per kWh. However, we caution that some of these analyses may have overestimated the expected lifetimes of the catalysts and electrodes (beyond what have been experimentally proven).

There are multiple factors that can decrease the lifetimes of GDEs. This includes catalyst deactivation as a result of poisoning from contaminants. Flooding of GDE pores with electrolyte, which lead to subsequent reduction of electrode performance is another critical issue for the cathodic side of GDE. Rapid release of product gases can damage catalysts layer. Effort has been made to avoid the flooding problem by using PTFE membranes as the gas diffusion layer, instead of the traditionally used porous carbon-based gas diffusion layers [26]. From our survey of the literature, we found little efforts devoted to studying the issues related to the lifetimes of the GDEs. We suggest more systematic work could be performed in this direction.

GDE systems for various CO2 reduction products

Carbon monoxide

The electroreduction of CO2 to CO is a two electrons transfer reaction. This makes the extent of CO2 sequestration to CO per unit current higher than for products such as C2H4 which involves twelve electron transfers. Catalysts for highly selective CO synthesis with FEs close to 100% are already available. High selectivity is one of the highly desired performances because it helps to make the separation of products simpler.

Large scale CO generation has been attempted in PEM electrolyzers with Ag-based catalysts [27] as well as Au [28], Cu [29], Pd [29], Pt [29] and Co-phthalocyanine [30] catalysts. Masel et al. used imidazolium-functionalized Sustanion™ membranes with Ag-based catalysts and demonstrated continuous operation for six months with 98% FE CO at 50 mA cm−2 total current density [27].

Flow electrolyzers using Ag catalysts [26,31, 32, 33,34•,35, 36, 37] have been studied intensively by several groups. Some groups have developed GDEs in-house, while others used commercial GDEs. Noteworthy effort has been made by Kenis et al. to develop carbon GDL-based GDEs with Ag catalysts. They investigated effects of composition of the micro porous layer and the substrate on performance [31]. They have also studied the effect of electrolyte composition [32]. Their best performing system showed CO partial current density as high as 440 mAcm−2 with FE above 90% [32]. Another outstanding work on GDE development was made by Sargent et al. [26]. They developed a new type of GDE (Figure 2) by employing porous PTFE membrane as a hydrophobic GDL in place of traditional carbon GDL [26]. In such GDEs, catalysts were deposited on the PTFE membrane, and a porous carbon layer, for current collection, was deposited on top of catalyst layer. This new type of GDEs exhibited more than 100 hours of stable operation at 150 mA cm−2 total current density with >90% CO FE. It is believed that the hydrophobic PTFE membrane, which prevents electrolyte flooding, is responsible for long term performance.

Commercially available Ag-based GDEs have been tested by at least two groups in flow electrolyzers. Haas et al. [33] obtained remarkable performance using Ag-based GDE from Covestro. They demonstrated 1200 hours of stable operation at a total current density of 300 mA cm-2, with 60% average FE for CO. Using a similar Ag-based GDE from Covestro, Fleischer et al. [34•] also made significant progresses on the development of pilot scale CO generation prototype system. They have shown over 600 hours of operation on a 100 cm2electrode with almost 60% FE for CO at 150 mA cm−2 total current density.

Besides silver, Au [38], Co-phthalocyanine [30] and Ni-N-C-based [39] catalysts have been investigated for CO generation using homemade GDEs in flow electrolyzers. Among these works, Strasser et al. reported an interesting result using Ni-N-C catalysts for CO generation. They achieved 20 hours operational duration and nearly 85% FE at a total current density of 200 mA cm-2 [39].

Formic acid

The large-scale production of formic acid from CO2 has also been studied intensively. Like CO, formic acid generation is a two electrons process and catalysts with high selectivity are available.

Formic acid generation in PEM electrolyzers has been studied using Sn catalysts on carbon paper-based GDE [40], and Sn-coated membrane electrode [41]. GDEs were developed in-house in all these publications. Among these, Masel et al.[40] developed a novel three-compartment PEM electrolyzer (Figure 3) with Sustanion™ anion exchange membrane which is stable over 500 hours of operation. Their system shows up to 94% FE at a total current density of 140 mA cm−2.

Flow electrolyzers for generating formic acid using home-made GDEs have been studied by a few groups. The majority of their works involved Sn-based catalysts on carbon GDL [42, 43, 44]. A significant improvement on GDE preparation method was reported by Kopljar et al. [44]. They developed a dry pressing method for GDE preparation which gives improved mechanical stability and high reproducibility. These improved GDEs gave close to 75% FE of formic acid at 400 mA cm−2 total current density.

Indium [45], Pb [46], and Bi-based [47,48] catalysts have also been tested in flow electrolyzers. Among these, Sargent et al. reported over 60 hours of operation and 90% FE at 200 mA cm−2 total current density with Bi oxy-halide derived catalysts [47]. Their results show that Bi oxy-halide derived catalysts is superior to Bi catalysts. The preferential exposure of highly active Bi $(1 \bar{1} 0)$ facets is proposed to be responsible for the superior performance.

Considering the above findings, it appears that both PEM electrolyzer and flow electrolyzer can achieve 90% formate FE at a total current density well above 100 mA cm−2. In terms of stability, the three-compartment PEM electrolyzer based on Sustanion™ anion exchange membrane is superior [40].

Ethylene

Large scale synthesis of hydrocarbons by electrochemical reduction of CO2 has been attempted by numerous groups. It is notable that all hydrocarbon synthesis reactions are multiple proton-electron transfer processes and involve a myriad of intermediates. Thus far, copper is an essential component of the catalysis. The selectivity (FE) for these products is typically far less than 90%. This makes the separation process non-trivial.

Generation of hydrocarbons using PEM electrolyzers is relatively unexplored. In contrast, articles on ethylene generation with flow electrolyzer systems have been published. Reports are available on GDEs developed in-house using carbon GDL and Cu nano-particles catalysts [49], nano-porous copper film [50], nano-porous copper silver alloy [51], copper (I) chloride derived copper [52], metal organic framework (MOF) based Cu catalysts [53] and other Cu-based catalysts [54,55]. Among these works, Kenis et al. showed that copper-silver bimetallic wire catalysts could reduce CO2 to ethylene with 60% FE and 180 mA cm−2partial current density [51].