Developments on carbon dioxide reduction: Their promise, achievements, and challenges

CO2 reduction processes continue to be developed for electrosynthesis, energy storage applications, and environmental remediation. A number of promising materials have shown high activity and selectivity to target reduction products. However, the progress has been mainly at a small laboratory scale, and the technical challenges of large scale CO2 reduction have not been considered adequately. This review covers recent advancements in catalyst materials and cell designs. The leading materials for CO2 reduction to a number of useful products are presentedwith their corresponding cell and reactor designs. The latest efforts to progress to industrially relevant scales are discussed, alongwith the challenges thatmust bemet for carbon dioxide reduction to be a viable route for mass scale production. Addresses 1 Electrochemical Engineering Laboratory, Energy Technology Research Group, Faculty of Engineering and Physical Sciences, University of Southampton, Highfield, Southampton, SO17 1BJ, UK 2 National Centre of Advanced Tribology at Southampton (NCATS), Faculty of Engineering and Physical Sciences, University of Southampton, University Rd., Southampton, SO17 1BJ, UK Corresponding author: Ponce de León, Carlos (capla@soton.ac.uk) Current Opinion in Electrochemistry 2020, 20:88–98 This review comes from a themed issue on Electrochemical Engineering Edited by Mark E. Orazem For a complete overview see the Issue and the Editorial Available online 6 May 2020 https://doi.org/10.1016/j.coelec.2020.04.014 2451-9103/© 2020 Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
CO 2 reduction is a key component in the global plan to offset carbon emissions and reduce human impact on the environment. The CO 2 reduction reaction (CO 2 RR) offers a viable synthetic route to a number of industrially important materials that are usually sourced from fossil fuels, such as methane, ethylene, formate, and CO [1]. Feedstock CO 2 could be captured from industrial processes that would otherwise release CO 2 into the atmosphere, reducing total CO 2 emissions and fossil fuel dependence in one process. Additionally, CO 2 reduction reactors can be integrated into intermittent green power sources such as wind or solar, where CO 2 is converted into a fuel such as methanol during peak energy production, which can then supplement energy production during periods of low energy generation.
The feasibility of CO 2 RR as a long-term solution to the synthesis of useful materials or in fuel production requires the optimization of both the catalyst materials and the reactor design to maximize the product production with minimal energy input. This is especially complex for the CO 2 RR, as the wide range of potential products means that even the most selective systems currently produce a mixture of multiple products. Such mixtures are costly to separate, and so achieving a system that is selective to a single CO 2 RR reduction product is a vital step for this to be economically viable. CO 2 þ 8e e þ 8H þ / CH 4 þ 2H 2 O E 0 = 0.03 V vs. RHE (4) 2CO 2 þ 8e e þ 7H þ / CH 3 COO e þ 2H 2 O E 0 = À0.26 V vs. RHE (5) 2CO 2 þ 12e e þ 12H þ / C 2 H 4 þ 4H 2 O E 0 = 0.08 V vs. RHE(8) Equations (1)e(10) highlight the challenge facing systems for selectively producing one CO 2 RR product, as there is significant overlap between the standard potentials for all of these products. This is further complicated by the hydrogen evolution reaction (HER), which is present over this same potential range.
The CO 2 RR begins with an initial adsorption of CO 2 onto a vacant catalyst site and subsequent electron transfer [3]. The CO 2 RR then proceeds via a series of intermediates that determine the product [4]. Proposed routes to a number of CO 2 RR products are given in Figure 1.
The selection of the correct catalyst material allows for reasonable selectivity to a specific product, by favoring certain reaction intermediates or by hindering or facilitating the formation of carbonecarbon bonds to give C 1 or C 2 products, respectively.

Catalyst designs
A number of metal alloys [11] and doped carbon materials [12] have been investigated for CO 2 reduction, giving reasonable selectivity for a range of C 1 and C 2 products. Some of the leading materials are given in Table 1.

Catalyst micro-structuring and nanostructuring
Modifications to the CO 2 RR catalyst structure may impact the overall activity or selectivity toward a certain product. Simple materials are enhanced by maximizing the electrochemically active surface area via

Catalyst coatings
The reaction environment at the catalyst surface has a sizeable impact on both the reactant rate and product selectivity for the CO 2 RR. Often, works refer to the 'triphasic interface' meaning gaseous CO 2 at the interface between the solid catalyst and liquid electrolyte. Various coatings can trap CO 2 and facilitate reduction, such as polymers with intrinsic microporosity for ethylene [60], and metal organic frameworks (MOFs) for methane [19]. Others use the polymer layer to hinder hydrogen evolution, where hydrophobic coatings hinder water reduction to increase the CO 2 RR yield for ethylene and ethanol [61]. Recent works have shown a new function of polymer coatings that can drive the formation of active nanostructures during CO 2 RR itself. CO 2 RR at Cu electrodes with an N-aryl pyrdinium drove the formation of Cu nanocubes, leading to enhanced ethanol and ethylene production [26,62].

Catalyst supports
Gas diffusion electrodes (GDEs) provide a rapid rate of CO 2 mass transport and negate poor CO 2 solubility in aqueous electrolytes [63]. Carbon GDEs can enhance the HER, so choosing a material that is less active toward HER gives an increased Faradaic efficiency toward CO 2 RR products [64]. One of the leading GDEs for ethylene production replaced the carbon GDE with a hydrophobic polymer, giving Faradaic efficiencies exceeding 60% up to 750 mA cm À2 , with consistent performance over 150 h [24].
Carbon-based materials can still be useful for CO 2 RR, particularly when surface and structural changes can be made to enhance the hydrophobicity [65]. Incorporating catalytic metal sites into the carbon structure gives promising catalysts for a number of CO 2 RR products, such as Au/C for CO [66] or CuN 4 /C for ethanol [28]. A further route is to start with materials that are known to have high CO 2 storage capacities, which enhances catalyst performance [67]. Catalytic sites have also been encapsulated in MOFs. Although MOF based catalysts themselves are limited by poor stability and selectivity [68], MOF-derived catalysts are an interesting route to catalytic sites in highly N-doped carbon environments, which have been achieved for formate [69] and CO [70] selective catalysts.
It is worth noting that removing the HER is not always desirable for the CO 2 RR. Syn-gas in an industrially important mixture of CO and H 2 , so a reactor capable of producing CO from the CO 2 RR and H 2 from the HER is of interest [71]. In these cases, the ratio of CO to H 2 becomes important, because the HER tends to outpace the CO 2 RR at larger current densities, although careful reactor design can give a fair degree of control over the ratio [72].

Electrolytes
Certain electrolytes can hinder the HER, improving the efficiency toward CO 2 RR products. High pH electrolytes slow the kinetics of the first water reduction step and strongly adsorbing OH À blocks H 2 evolution sites [73]. The same surface-blocking effect has been demonstrated with strongly adsorbing halide electrolytes [24]. Larger cations increase CO and formate yields on Ag, Sn, and Bi by stabilizing adsorbed CO 2 and increase formate, ethylene, and ethanol on Cu by stabilizing important precursors for CeC bond formation [18,74].
Other groups have moved to ionic liquids because of their wide solvent window and ability to stabilize charged CO 2 intermediates [75]. The protons required for the CO 2 RR have been supplied through humidifying the CO 2 [76], diluting the ionic liquid with water [77], or attaching functionalized ionic liquid components to the catalyst surface [78]. It is worth noting that, thus far, all published examples of the CO 2 RR in ionic liquids have produced C 1 products [75].

Reactor designs Cathodic compartment
Most CO 2 RR reactors are based on an electrolyser or fuel cell designs with a GDE cathode and oxygen evolving anode separated by an ionic membrane. CO 2 feed can be either in the gas or liquid phase [79].
Most high-throughput systems use gas phase supply, as aqueous-fed systems are restricted to w35 mA cm À2 because of low solubility of CO 2, [80] whereas GDEs enable current densities of two orders of magnitude higher by facilitating the transport of CO 2 to the catalyst [81]. Products can be collected in the gas or liquid phase, or both can be done simultaneously as has been achieved for the production of ethanol and ethylene [82]. The total concentration of liquid products can be increased by continually cycling the catholyte via a peristaltic pump [83]. Careful controls are needed over the gas and liquid phase flows as these can influence product selectivity, reaction rate, and lifetime. Using CO 2 /N 2 mixtures can impact C 1 vs C 2þ selectivity, where lower CO 2 ratios decrease CeC bond formation to favor C 1 products [84] and increased gas phase pressure increases CO selectivity [85].

Membrane
Most reactors use a membrane between anodic and cathodic compartments to prevent product crossover [86]. Typically, the membrane can separate liquid catholyte and anolyte (Fig 3c). Fewer works have reported zero-gap electrolysers using membrane electrode assemblies (MEAs), where the catalyst is in direct contact with the membrane (Fig 3b). MEA systems have no liquid catholyte and bring the cathode and anode into close proximity to minimize the iR drop.
CO 2 RR studies at MEAs have been limited to formate and carbon monoxide [87,88], with anion exchange membranes outperforming proton exchange membranes (PEMs) by increasing the local pH at the catalyst surface to suppress the HER [89e91]. Further advancements have been made by taking ionic liquid components that enhance the CO 2 RR, such as imidazole, and incorporating them into the membrane structure [92].
Accessing higher order CO 2 RR products like methanol or ethylene requires a liquid catholyte, so zero-gap MEAs are not feasible. Steps can be taken to keep the iR drop low, such as having a zero-gap anodic chamber and minimizing the catholyte channel depth [93]. These designs have allowed the production of ethylene with Faradaic efficiencies up to 25% at 1.3 A cm À2 [94].
One popular alternative to membrane-separated reactors are microfluidic CO 2 RR reactors (Fig 3d), where the laminar flow profile can separate anolyte and catholyte without a membrane [95]. This has been exploited to introduce different pH electrolytes to enhance the anode and cathode separately, giving an overall improvement in cell performance [96]. Turnover rates can be increased by increasing the path length or incorporating multiple channels in parallel, providing a viable means to up-scaling [97].

Anodic compartment
Most published CO 2 RR reactors use IrO 2 or Pt-based anodes to facilitate the oxygen evolution reaction in the anodic compartment. There exists a wealth of published catalysts for the oxygen evolution reaction [98,99], yet novel anodic materials are rarely incorporated into CO 2 RR reactor reports. This becomes increasingly important for up-scaling, where the high current densities can cause delamination at the anode or O 2 bubbles may block active sites if they are not removed by anolyte flow [93]. Realizing the industrial potential of the CO 2 RR will require a unified optimization of the anode, cathode, membrane, and reactor design.
Some investigations suggest the use of the hydrogen oxidation reaction to decrease the cell potential during the CO 2 RR and allow the use of GDEs on both sides of the cell [100]. Another strategy is to use the oxidation of organic materials at the anode to give additional valueadded products for the same charge passed [101]. In either case, a successful CO 2 RR reactor must consider the anodic environment as carefully as that of the cathode.

Materials manufacturing scale research
Compared with batch-type or static electrolysers, continuous-flow reactors are more suitable for scale-up applications, because of their improved CO 2 mass transport, product removal, electrolyte mixing, and temperature control [102]. The development of large or multiple-stack electrolysers have been limited, although a few start-ups (OPUS12, CERT, Dioxide Materials) and established companies (Siemens) have evidenced ongoing up-scaling activities [80].
A few research groups have demonstrated up-scaled systems using continuous flow reactors. Cycling CO 2saturated electrolyte through a flow cell achieved 63e 91% formate selectivity in a 300 cm 2 single cell without using a GDE [103]. A semipilot scale reactor capable of converting 1 kg CO 2 per day to formate at 78% efficiency is one of the largest scale designs reported thus far [87].
Up-scaling GDEs and MEAs requires regulation of the CO 2 back-pressure to prevent GDE flooding while permitting electrolyte contact with the catalyst [104]. It also becomes increasingly important to consider noncatalytic limitations to the GDE design, such as in-plane resistance and local pH changes [105]. Recent upscaled GDEs have been reported, including a 250 cm 2 silver/imidazolium GDE, which gave 95e99% carbon monoxide over a 200e600 mA cm À2 range [106].
Most up-scaling reports have used MEAs in zero-gap electrolysers, as they can be readily integrated into existing PEM fuel cell and electrolyser technologies, giving reactors with low cell resistance and a straightforward approach to raising the pressure in the reaction chamber [102]. Careful consideration is needed regarding the catholyte, anolyte, and membrane because ion crossover can significantly alter the pH and ionic strength after extended operations [107,108]. In particular, PEMs give poor stability at high reaction rates because of cation depletion [109]. Anion exchange membranes offer improved stability, though suffer from bicarbonate crossover, which can lead to substantial CO 2 evolution at the anode [50]. Bipolar membranes are promising route to improve stability at high CO 2 RR rates, hindering ionic crossover maintaining a high cathodic pH [110].
Further enhanced outputs have been demonstrated through a stacked zero-gap electrolyser, where multiple 61 cm 2 silver cells gave 95% carbon monoxide at 300 mA cm À2 [111]. Importantly, when connecting the cells in parallel, the operation of the stack was identical to the sum of each single-cell, whereas connecting the cells in series significantly increased the conversion rate. It is worth mentioning that there is still interest in efficient materials that operate in the 10e50 mA cm À2 range, because this corresponds to the photocurrent density of a solar CO 2 RR reactor operating without solar concentration [109].
Looking at long-term viability, it is also important to consider separation and purification alongside operating costs. This is particularly pertinent to formic acid/ formate production; formate production in alkali media proceeds with higher Faradaic efficiency than formic acid production acidic media but increased separation costs for formate separation make formic acid the more economically desirable option, despite the lower efficiency [112]. Some technologies have been developed to mitigate this, where the liquid formate phase passes into an electrolytic acidification reactor to facilitate separation [113].
Other works aim to mitigate costs by introducing additional value added components into their designs. Electrolysis in a membrane-free cell in bromide electrolyte has been shown to produce 2-bromoethanol, which presents a route to higher order value-added products [114]. Alternatively, CO 2 RR reactors can be hyphenated with secondary reactors, as has been achieved for a formate feed into a thermal reactor to produce oxalic acid [115].
There is little information regarding large CO 2 RR systems and the challenges and technical difficulties associated with up-scaling. Catalytic performances are sensitive toward operating conditions and surface impurities and tend to change drastically when operated in less ideal conditions. In up-scaled devices, deactivation, transport, resistance, and stability losses can result in reduced performances over prolonged operations [116]. Furthermore, managing the gaseliquid interface within the GDE, that is, avoiding flooding with electrolyte, is often crucial to maintaining product selectivity [117]. It appears that up-scaled CO 2 electrolysers have been unable to address all of these issues and delivered either low energy efficiencies and/or poor activities [111]. Regardless of the size, continued CO 2 reduction longer than several days are rarely reported in the literature possibly because of catalyst and system degradations.

Conclusions
The CO 2 RR community have produced a number of high-quality catalyst materials and reactor designs to drive the production of value-added chemicals from a CO 2 feedstock. Initial selectivity toward a specific product is achieved through the selection of an appropriate catalyst material, with further enhancements made possible by engineering the catalyst surface through alloying and micro-structuring or nanostructuring.
The next advancements in CO 2 RR field will come from meeting the requirements to up-scale the reactor operations to industrially relevant standards. Faradaic efficiencies toward a specific product tend to fall as the current density increases because of the increased HER at larger current densities. Designing catalyst materials, innovative supports and reactor designs that disfavor water reduction and enhance the CO 2 RR will facilitate improved operations on larger scales.
In addition to cathode materials, further developments of anode catalysts, cell configurations, membranes, and other interfacial contacts (i.e. within the GDE) are also necessary. Reactors need to be optimized for cathodic and anodic efficiencies, energy requirements, and stability to catalyst loss/deactivation and gas phase flooding. There is a need for computational simulations for parameters that are typically used for electrochemical reactors, including current and potential distribution, fluid flow, pressure drop, and mechanical integrity of GDEs.
Typical figures of merit for electrochemical reactors, such as mass transport, space-time velocity, or pressure drop suitable for up-scaling, have not been reported in larger systems. Technoeconomic feasibility models for CO 2 electrolysers for different conversion products will be useful for identifying the technical feasibility and financial viability of processes using particular reactions, catalysts, and cell designs.

Conflicts of interest
The authors declare that there is no conflict of interests.

.
Handoko AD, Wei F, Jenndy, Yeo BS, Seh ZW: Understanding heterogeneous electrocatalytic carbon dioxide reduction through operando techniques. Nat Catal 2018, 1:922-934. A detailed summary of the latest mechanistic understandings of the CO 2 RR with direct implications on catalyst design for favouring specific products

9.
Paris AR, Bocarsly AB: Mechanistic insights into C 2 and C 3 product generation using Ni 3 Al and Ni 3 Ga electrocatalysts for CO 2 reduction.  Polymers with intrinsic microporosity (PIMs) for targeted CO 2 reduction to ethylene. Chemosphere 2020, 248:125993. The first example of polymers with intrinsic microporosity applied to CO 2 RR. This work highlights both the positive and negative impacts of using organic polymer layers to modify catalyst surfaces for CO 2 RR to ethylene.