Electrocatalyst Microenvironment Engineering for Enhanced Product Selectivity in Carbon Dioxide and Nitrogen Reduction Reactions

Carbon and nitrogen fixation strategies are regarded as alternative routes to produce valuable chemicals used as energy carriers and fertilizers that are traditionally obtained from unsustainable and energy-intensive coal gasification (CO and CH4), Fischer–Tropsch (C2H4), and Haber–Bosch (NH3) processes. Recently, the electrocatalytic CO2 reduction reaction (CO2RR) and N2 reduction reaction (NRR) have received tremendous attention, with the merits of being both efficient strategies to store renewable electricity while providing alternative preparation routes to fossil-fuel-driven reactions. To date, the development of the CO2RR and NRR processes is primarily hindered by the competitive hydrogen evolution reaction (HER); however, the corresponding strategies for inhibiting this undesired side reaction are still quite limited. Considering such complex reactions involve three gas–liquid–solid phases and successive proton-coupled electron transfers, it appears meaningful to review the current strategies for improving product selectivity in light of their respective reaction mechanisms, kinetics, and thermodynamics. By examining the developments and understanding in catalyst design, electrolyte engineering, and three-phase interface modulation, we discuss three key strategies for improving product selectivity for the CO2RR and NRR: (i) targeting molecularly defined active sites, (ii) increasing the local reactant concentration at the active sites, and (iii) stabilizing and confining product intermediates.


INTRODUCTION
Many of today's environmental, economic, and societal issues are related to the transformation of two inert gases, N 2 and CO 2 . The transformation of N 2 via the Haber−Bosch process accounts for over 1% of the world's energy consumption, 1 providing nitrogen fertilizers required to sustain the current global food production. Meanwhile, the amount of CO 2 released into the atmosphere from the combustion of fossil fuels has reached unprecedented levels, further accelerating climate change. 2−6 Both CO 2 and nitrogen undergo complex environmental cycles (Figure 1a and b), increasing the challenges associated with their capture and conversion. Implementing sustainable cycles for CO 2 and N 2 and minimizing their environmental impact is critical, as recently highlighted in the latest Intergovernmental Panel on Climate Change (IPCC) report or in Europe in the EU green deal and Fit for 55 packages. 7−9 The electrochemical conversion of CO 2 and N 2 into valueadded products or net zero commodities such as materials, renewable fuels, and energy vectors appears as an appealing solution in this context, as it can utilize sustainable sources of electricity powered by solar, wind, wave, and hydro energy to promote reactions currently carried out using fossil fuels. This approach would provide a carbon-neutral route to C-and Ncontaining products while enabling the efficient storage of intermittent renewable sources of electricity as chemical bonds, largely overperforming the battery storage energy efficiency. 10 The main hurdle to developing energy efficient processes for converting nitrogen to ammonia and carbon dioxide to energy dense products such as hydrocarbons is selectivity. The chemical inertness of these reactants disadvantages their transformation compared to more kinetically facile reactions such as the hydrogen evolution reaction (HER). Furthermore, selectivity is one of the most challenging aspects to address when developing electrocatalysts to mediate the CO 2 reduction reaction (CO 2 RR) and the nitrogen reduction reaction (NRR). In both cases, multiple reaction products are typically observed. These result from the reduction of CO 2 and N 2 themselves as well as from the proton sources used to mediate these reduction reactions, which involve successive coupled electron−proton transfers. In this regard, the electrocatalyst microenvironment plays a vital role and can be engineered to improve selectivity through three key strategies: (i) targeting a narrow distribution of molecularly defined active sites, (ii) increasing the reactant/proton ratio at the three-phase interface where the reaction takes place to lower the undesired formation of H 2 , and (iii) the stabilization and confinement of reaction intermediates in the electrode vicinity to favor the formation of multielectron reduction products. While there exists an extensive amount of literature in both the CO 2 RR and NRR fields, including several recent reviews of specific subtopics, 11−14 we aim in this Review to illustrate through a handful of selected examples the key strategies for increasing selectivity toward value-added products.
After a brief explanation of the kinetic and thermodynamic origins of multiple product generation in the CO 2 RR and NRR, we discuss the key factors in catalyst design in steering product selectivity, namely nanostructuring, surface functionalization, control of crystal size and facets, and single-site engineering. We then explore the impact of the electrolyte on the activity at the electrode surface, including aspects such as pH, the alkali metal cation, and the use of novel electrolytes. The final section focuses on the implementation and optimization of triple-phase interfaces to improve the local reactant concentration and mass transport. We conclude with our perspectives on this rapidly growing topic and where we envisage future challenges and opportunities to lie.
It is important to note that the NRR field has been strongly affected by a series of false positives, and a standardized set of experiments has been outlined to identify, quantify, and eliminate experimental artifacts. 15 Ammonia contamination may arise from sources such as the air, chemicals ,and the experimental setup, which is particularly significant when the quantity of ammonia produced in the NRR is very low. Additionally, labile nitrogen-containing compounds such as nitrates, nitrites, nitrogen oxides, and amines are often present in the N 2 gas stream, the air, and the catalyst itself. To reliably attribute ammonia production to the NRR, quantitative isotope measurements with 15 N 2 gas and the removal of impurities from the gas stream are imperative. To preserve a fair comparison of performance between catalytic materials, in this Review we present only examples that follow the guidelines provided in the above ref 14, unless clearly stated otherwise.

MECHANISTIC AND THERMODYNAMIC ORIGIN OF MULTIPLE PRODUCT GENERATION IN CO 2 RR AND NRR
Both the CO 2 RR and NRR to value-added products involve multiple successive proton-coupled electron transfers (Table  1), which represent a significant kinetic challenge to overcome to achieve high selectivity, particularly compared to the more kinetically facile two-electron hydrogen generation reaction. 16−18 This kinetic challenge is further complexified by the low availability of the reactants, as both CO 2 (∼33 mM at P CO2 = 1 atm) and N 2 (∼0.7 mM at P N2 = 1 atm) have typically poor solubility in water. 19 In the context of the CO 2 RR to multicarbon products, the low solubility of the primary reaction products such as CO also decreases the overall catalyst selectivity for multicarbon products, which result from the subsequent reduction of these primary products.
In addition, a thermodynamic challenge is associated with the CO 2 RR, since proton reduction (HER) is more thermodynamically favorable than the reduction of CO 2 to most products (Figure 2a and eqs 3−6). 20−22 Although less critical in the case of NRR, the standard electrochemical potential for the proton reduction reaction is yet close to that of the nitrogen reduction reaction (NRR) at 0.057 V vs. SHE (eq 2). 23 The intrinsic stronger binding of H atoms over N 2 on most metal surfaces, highlighted in Figure 2b, further illustrates the challenge to increase NRR selectivity vs HER. 24 This illustrates the three main challenges (thermodynamic, kinetic, or related to the mass transport of the reactants and primary reaction products) that must be overcome to reach high selectivity in the CO 2 RR and NRR. We will review in the next sections the three main axes currently explored toward that goal, focusing on catalyst design, electrolyte engineering, and three-phase interface modulation.

INCREASING SELECTIVITY VIA CATALYST DESIGN
3.1. Catalyst Nanostructuring for Improved Mass Transport. Advancements in nanotechnology and characterization techniques have enabled a plethora of morphologies to be explored to improve catalytic activity and product selectivity. Porous materials have attracted particular attention due to their effect on the local chemical environment, including local pH and the mass transport of the reactant and intermediates. 25,26 The ability to increase the number of effective active sites, both by maximizing surface area and facilitating the accessibility of such sites, makes porosity useful and interesting across a broad range of fields. 27 Such effects are especially crucial when considering the poor solubility of CO 2 and N 2 in aqueous electrolytes, which causes mass transport limitations and barriers to high activity and selectivity. Hierarchical porous networks are found commonly in biological organisms as a strategy to mitigate mass transport limitations in the utilization of nutrients. 28 The threedimensional networks were replicated in early work by Huan et al., who used gold nanodendrites for electrochemical sensing. 29 Their application in catalysis has recently appeared as an efficient strategy to increase current densities and catalyst selectivity in small-molecule electroreduction and oxidation.
The dynamic hydrogen bubble templating (DHBT) method has been the most prominent technique to create such hierarchical porosity, which was recently comprehensively reviewed by the Bhargava group 30 and specifically for CO 2 RR materials by the Broekmann group. 31 The process involves the electrodeposition of a metal from aqueous solutions of the respective cations, while cogenerated hydrogen bubbles act as a dynamic template to create a metal foam. As the bubbles nucleate, grow, and detach, a hierarchical pore structure forms with layers of pores of increasing diameter (Figure 3a), including micropores in the submicrometer range and macropores in the range of 10−100 μm. 30 The DHBT technique is relatively simple, requiring aqueous solutions and no need for organic or inorganic templates (as in traditional metal foam synthesis), 32 high temperatures, high pressures, or uncommon equipment. Nonetheless, additives such as citrate are common to influence crystal growth. 33−35 The formation of Bi and multimetallic catalysts are also possible by coelec-trodeposition, galvanic replacement, stepwise electrodeposition, or spontaneous decoration. 30 For example, many studies for CO 2 RR have coupled copper with one other metal such as Ag, Sn, In, or Zn. 36−40 By fine-tuning parameters such as the proton source and concentration, the applied overpotential or current density, the substrate material, and the metal source and concentration, the nanostructure can be carefully controlled and optimized. Broekmann and co-workers produced a dendritic Cu-based DHBT foam and demonstrated a strong dependence of the C 2product selectivity on the surface pore size diameter, with the optimal size being between 50 and 100 μm. 41 They identified the temporal trapping of gaseous intermediates inside these pores as the key to product selectivity. Intermediates such as CO and C 2 H 4 , which would otherwise be released into the bulk electrolyte, were entrapped in the pores of the foam catalyst, causing them to further react to form C 2 H 6 ( Figure  3b). At −0.8 V vs RHE, the authors achieved a 55% Faradaic efficiency for C 2-products.
Such dendritic structures with large surface areas are common in this synthesis due to the deposition taking place at high current densities and therefore in the diffusion-limited regime. Copper-and oxide-derived copper dendrites have attracted particular interest due to their apparent selectivity for multicarbon products. 42−45 Huan et al. produced a dendritic CuO material from DHBT that could be used as both CO 2 R and OER catalysts. 46,47 It consisted of a triple-layer structure with a metallic Cu core covered by layers of Cu 2 O and CuO ( Figure 3c). In electrocatalytic conditions, the CuO material is reduced to metallic Cu, generating nano-Kirkendall voids within the dendrite structures. Such voids, which appear at the copper−copper oxide interface upon reduction, are termed nano-Kirkendall voids, as they appear as a consequence of the very different diffusivities of Cu and O atoms. The overall external shape of the material is maintained upon reduction, but cavities are generated under its external layer due to the lower density of Cu with respect to the original copper oxides. 48 These gas-accessible voids were proposed to enhance the confinement of secondary CO 2 RR products, such as CO, resulting in FE C2+ over 50%. By using the catalyst in a continuous flow electrolyzer, they were able to reach a stable current of 25 mA/cm 2 with 2.95 V, equating to 21% energy efficiency for hydrocarbon production. By coupling the cell to a photovoltaic cell, they achieved a 2.3% solar-to-hydrocarbon efficiency.
DHBT foams for single-carbon products such as CO and formate have also been reported. A silver foam with needleshaped features in the mesopores was produced by using a citrate additive to control growth on the nanometer scale. 33 Between −0.3 to −1.2 V vs RHE, 90% Faradaic efficiency for CO was observed; however, at higher overpotentials the foam produced C 2 products, with 51% CH 4 at −1.5 V (Figure 3d). This unusual activity for Ag was attributed to the catalyst morphology and nanostructure increasing the *CO surface concentration and residence time. Recent work by Mayer and co-workers exemplifies the advantages of the simplicity of the DHBT method. In a one-step synthesis they used waste industrial Cu−Sn bronze as a material precursor to deposit a mesoporous Cu 10 Sn foam. 49 They achieved over 85% Faradaic efficiency for CO at −0.8 V vs RHE, over double that of the plain Cu−Sn bronze, with partial current densities three times higher. Du et al. prepared a nanoporous tin DHBT foam on a tin substrate and achieved a Faradaic efficiency for formate of 90% with current densities of 23 mA/cm 2 . 50 Other morphology-based strategies have been utilized to modulate mass transport in CO 2 reduction, including the application of nanostructures such as nanowires, sheets, needles, cones, or tubes. Burdyny et al. explored the effect of the nanomorphology of a silver catalyst on gas evolution and subsequently bubble-induced mass transport. 51 By combing mathematical modeling and experimental observations using a dark-field microscope, they compared bubble formation on nanoparticles, nanorods, and nanoneedles and found mean bubble diameters of 97, 31, and 23 μm, respectively. They illustrated that the generation of smaller bubbles improved long-range mass transport of CO 2 , resulting in a small diffusion thickness and a fourfold increase in the limiting current density of CO production ( Figure 4a). Surendranath and co-workers synthesized gold inverse opal thin films and found that changing the mesostructure by increasing the porous film thickness could diminish HER 10-fold while maintaining activity for CO 2 to CO, increasing the faradaic efficiency for CO from less than 5% to over 80%. 52 They attributed this to the formation of diffusional gradients. Studies of nanocavities and their performance and mechanism of action have emerged in recent years. Yang et al. utilized finite-element method simulations and experimental measurements on a multihollow cuprous oxide catalyst. 53 Analysis from X-ray absorption studies and operando Raman spectra indicated that the pore cavities confined *CO intermediates, which bound to Cu + sites and locally protected them against reduction during CO 2 RR (Figure 4b), as well as promoted C−C coupling. The authors achieved a C 2+ product Faradaic efficiency of 75% and a partial current density of 267 mA/cm 2 .
As N 2 electroreduction is a comparatively less mature field with its own unique challenges, studies into morphological effects on catalytic activity and selectivity are less extensive. Although a range of nanostructures exist among the literature, 54 specific insight into the role morphology plays in catalysis is limited. Wei et al. loaded ruthenium nanoparticles onto carbon nanotubes, which were also applied as the gas diffusion electrode. 55 Despite using a typical H-cell setup, the GDE structure allowed N 2 gas to flow through the GDE and porous catalyst instead of being solely solubilized in the electrolyte, as illustrated in Figure 4c. They achieved a NH 3 yield rate of 2.1 nmol/cm 2 · and Faradaic efficiency of 13.5%.
A great range of nanostructures have been applied to the CO 2 RR and NRR to regulate mass transport, and although strong correlations between structure and performance have been made, their mechanisms of action are often highly complex and difficult to define. Most theories focus on the mass transport of reactants and intermediates either through improved diffusion and convection or through their physical confinement in the catalyst pores. Considerable progress has been made by combining computational and experimental research to define and improve catalyst nanomorphology, especially in the CO 2 RR field; however, their application to new materials and fields such as NRR is still an open area of research.

Surface Functionalization.
Functionalization of the electrode or catalyst surface with organic or inorganic ligands has been explored as a strategy to tune the interaction between adsorbed intermediates and catalysts, inhibiting HER and improving product selectivity. In addition to the decoration of the surface of a catalytic material with surface-bound ligands, the covalent grafting of molecular cocatalysts onto the surface of a catalytic material has also been explored as a strategy to further tune the catalyst selectivity. 56 In this section, we will outline some key examples in the diverse field of catalyst surface functionalization, which has been comprehensively reviewed for the CO 2 RR by Reisner and co-workers. 26 To date, many organic additives such as amino acids, 57 amines, 58,59 aminothiols, 60 pyridiniums, 61,62 N-heterocyclic carbenes (NHCs), 63,64 imidazolium ligands, 65 porphyrin-based metallic complexes, 66,67 polymers, 68,69 and inorganic additives, 70,71 have been proposed to control the binding energies of CO 2 RR reaction intermediates (Figure 5a). For instance, Kim et al. demonstrated a 94.2% FE for the production of CO from amine-capped Ag supported on carbon, thanks to the effective suppression of the HER and the intrinsic high selectivity toward the CO 2 RR from Ag ( Figure  5b). 58 DFT calculations suggested that the amine-capped Ag nanoparticles stabilize the *COOH intermediate while destabilizing *H. Conversely, thiol-capped Ag nanoparticles exhibited superior reaction rates toward both the HER and CO 2 reduction by indiscriminately increasing ΔG *H and ΔG *COOH .
As presented in Figure 5c, Zhao et al. developed a simple modification strategy using amines to depress the hydrogen evolution reaction on ultrasmall Au NPs and enhance CO 2 -to-CO conversion. 59 The amine groups, as well as the molecular configuration, were found to play important roles in tuning the electrocatalytic activity of low-coordinated sites of the nanoparticles. The authors claimed that strong interactions between the Au surface and the amine ligands combined with the peculiar configuration were responsible for the improved CO 2 RR performance. Remarkably, linear amines promoted the formation of CO, an effect that was enhanced by increasing the length of the alkyl chain, whereas the branched polyamine greatly depressed it. Wang et al. demonstrated 55% and 77% selectivities for ethylene and C 2+ products, respectively, using a tricomponent copolymer to modify the surface of Cu electrodes ( Figure 5d). 68 Systematic studies indicated that the three components of the copolymer control electrostatic interactions, gas diffusion, and hydrophilicity, which were found to be necessary to improve selectivity. The copolymer was obtained by ring-opening metathesis polymerization, thereby offering a new degree of freedom for tuning the selectivity.
Applying a molecular design approach to tune heterogeneous catalysts has also proved effective in the functionalization of palladium foil with chelating N-heterocyclic carbene (NHC) ligands, demonstrating a 32-fold increase in activity for CO 2 to C 1 products. 63 N-Aryl-pyridinium salts have also proved effective in tuning electronic properties to stabilize intermediates for CO 2 RR to ethylene. 62 Porphyrin-based metallic complexes have been used to functionalize copper surfaces to increase the concentration of CO intermediates and promote C−C coupling; a Faradaic efficiency of 41% for ethanol was achieved at 124 mA/cm 2 at −0.82 V vs RHE. 67 Modifying the catalyst surface indirectly has also been implemented by Varela et al. through the addition of halides to the electrolyte. 71 They hypothesized that the adsorption of halides onto copper increased the negative charge of the catalyst surface, altering the selectivity. In the case of iodide, the induced negative charge favored the protonation of CO, enhancing CH 4 production.
Applying well-defined molecular approaches to heterogeneous systems can give important insights into catalytic mechanisms and help to fine-tune active sites and product selectivity. Some functionalization strategies operate through molecular coordination and can therefore be carefully controlled by altering functional and side groups so that specific CO 2 RR intermediates can be stabilized. Other strategies, such as the addition of halides or ionic liquids, affect the charge on the catalyst surface, increasing CO ads coverage for example. 70,72 Both have proven effective in improving product selectivity in the CO 2 RR, and similar approaches could be applied to the NRR to help overcome the dominance of HER (Figure 5e). 73 The exact surface binding motifs of ligands and the mechanism for altered selectivity are still unclear. Understanding the precise nature of the interface remains a key challenge for attaining the desired catalytic properties. 56 3.3. Crystal Size and Facet Control. Tremendous advances have recently been made to engineer catalysts in order to limit the HER during the CO 2 RR and NRR processes. 72 Compared with their bulk counterparts, nanostructured catalysts show original and often enhanced activities due to their unique surface electronic and chemical properties. These properties can be finely adjusted to tune the activity and selectivity of electrocatalytic reactions. The surface of a nanomaterial catalyst typically consists of planar areas with single-crystalline orientations separated by steps and kink sites with lower coordination numbers. Complex atomic structures are therefore present at the interface between different grains in polycrystalline and/or nanostructured surfaces. Buonsanti et al. investigated the catalytic properties of exposed facets of Cu nanocatalysts at commercially relevant current densities (Figure 6a). 74 The study revealed that facet-dependent selectivity is retained in a gas-fed flow cell, showing greater HER suppression than in a conventional H-cell. The (100) facets of Cu nanocubes have been identified to be selective for the evolution of C 2 H 4 , whereas the (111) facets of Cu octahedra are selective toward CH 4 . Conversely, Cu spheres do not exhibit any specific product selectivity, suggesting that randomly mixed facets cannot depress the HER during the CO 2 RR. Chorkendorff et al. systematically investigated the structure−selectivity relationship of Au single crystals for electrocatalytic CO 2 reduction (Figure 6b). 75 Remarkably, they found that the kinetics for the formation of CO strongly depend on the surface structure. Under-coordinated sites, for instance, those on the surface of Au (110) or at the step edges of Au(211), show at least 20-fold higher activities than more coordinated configurations, such as Au (100). By selectively poisoning under-coordinated sites with Pb, the authors identified the selectivity of these active sites toward the reduction of CO 2 , effectively suppressing the HER.
Roldan Cuenya, Strasser, and co-workers investigated the role of particle size in CO 2 electroreduction using sizecontrolled Cu nanoparticles (NPs). 76 A dramatic increase in the catalytic activity and selectivity of CO against H 2 was observed once the particle size was decreased, particularly for NPs smaller than 5 nm, as shown in Figure 6c. Changes in the population of low-coordinated surface sites and their stronger chemisorption were linked to H 2 and CO selectivity. As shown in the inset of Figure 6c, a drastic increase in the number of undercoordinated atoms is observed below a particle size of 2 nm, with a coordination number lower than 8. These peculiar sites accelerate both hydrogen evolution and CO 2 reduction to CO via an increase in binding energy. However, the undercoordinated sites are unfavorable for the subsequent hydrogenation of CO, which lowers the hydrocarbon selectivity of the NPs. A plausible explanation for the observed trend is the reduced mobility of intermediate reaction species (CO and H) on the small NPs due to stronger bonding, which decreases the possibility of further recombination to form hydrocarbons. At intermediate particle sizes, the spherical particle model predicts low and constant populations of (100) and (111) facets, which is consistent with the reduced yet constant hydrocarbon selectivities observed for Cu NPs between 5 and 15 nm compared to Cu bulk surfaces. For these larger NPs, weaker binding of CO and H is expected, favoring hydrocarbon formation.
Another critical parameter for suppressing the HER with metal NP catalysts is the interparticle spacing. Mesoscale phenomena, such as interparticle reactant diffusion and readsorption of intermediates, can play an important role in the product selectivity for multistep reactions. 77,78 In this context, Mistry et al. showed that, for CO 2 electroreduction, decreasing the interparticle spacing for a constant nanoparticle size can suppress the HER, which further increases the selectivity for CH 4 and C 2 H 4 due to the increased possibility of the *CO intermediate readsorbing on a neighboring particle and being further reduced (Figure 6d and e). 79 More importantly, this study uncovers general principles of tailoring NP activity and selectivity by carefully engineering the size and distance. These principles guide the rational design of mesoscopic catalyst architectures to enhance the production of the desired reaction products. 80 3.4. Single-Site Engineering. One of the main hurdles to the rational improvement of selectivity using metallic or metal oxide/sulfide catalysts is the large distribution of accessible sites that may result in different favored reaction products and decreased selectivities. Single-atom catalysts (SACs) hence represent an attractive strategy to increase selectivity via a narrower distribution of active sites and improved control of the first coordination sphere of the active site, bridging the gap between well-defined molecular catalysts and complex heterogeneous materials. The catalytic properties of SACs hence result from the combination between the molecular tuning of the coordination environment of the active sites and its interaction with the support. 81 Different types of supports for SACs have been explored to date and include metals, carbon-based materials, and metal (hydr)oxides, nitrides, and carbides. Metal-supported SACs, also called single-atom alloys (SAAs) have also been explored. They generally yield thermodynamically more stable interactions than other atomsupported interactions due to strong metal−metal interactions. 82 Advantageously, SAAs can offer different active sites on the host metal (i.e., the support) and the individual atoms, providing further opportunities to modulate reaction pathways. 83 86 They reported that the use of Cu 1 Sn 1 comprising a core−shell structure doped with a small amount of Cu using CuSn and SnO alloys as the core and shell, respectively, leads to the preferential formation of formate with an FE greater than 95% at −1.2 V. In contrast, single atoms of Sn supported on Cu:Cu 20 Sn 1 show a high selectivity for CO with a maximum FE CO of 95.3% at −1.0 V.
Carbon substrates have been widely explored in the form of graphite, graphdiyne, and graphene and its derivatives, including heteroatom (N, O, S, and P)-doped sp 2 carbon materials. Carbon supports indeed offer several advantages such as high surface area, high electronic conductivity, and strong thermal stability, and they also possess numerous coordination environments to stabilize the single atom ACS Catalysis pubs.acs.org/acscatalysis Review sites. 87,88 The different behaviors of transition metals in the form of nanoclusters or metal−nitrogen-doped carbon catalysts (MNCs) were examined by the Chan and Strasser group. 89 The results of their calculations revealed that *CO 2 adsorption is the limiting step on metals, whereas for nitrogencoordinated SACs the reaction can be limited either by *CO 2 adsorption or by the formation of *COOH via a proton− electron transfer (Figure 7a). Pan and coauthors reported the design of MNC SACs with atomically dispersed Co sites anchored on polymer-derived hollow N-doped porous carbon spheres. 90 The single-atom Co−N 5 sites were identified as the main active centers for CO 2 activation, and the rapid formation of *COOH as a critical reaction intermediate was followed by a rapid desorption of CO. A similar behavior has also been reported on carbon nanosheet-supported Ni−N 4 sites, which resulted in near-utility selectivity for CO and a single-pass conversion of 2.6% cm −2 when implemented in a flow cell. 91 Huan et al. investigated a series of iron-based catalysts synthesized by pyrolysis of Fe-, N-, and C-containing precursors for the electroreduction of CO 2 to CO in an aqueous medium and demonstrated that the selectivity of these materials for CO 2 reduction is governed by the proportion of isolated FeN 4 sites compared to Fe-based nanoparticles. 92 They demonstrated that the nature of the metal species modulates the selectivity of the reaction pathways and suggested that FeN 4 sites are responsible for CO 2 RR, whereas the Fe cluster are responsible for HER. In a following work, they demonstrated the strong influence of the electrode support on the catalyst selectivity, highlighting the importance of reducing mass transport limitation to promote a higher selectivity toward CO 2 reduction. 93 In recent years, metal (hydr)oxides, nitrides, carbides, and sulfides have become very popular supports of SACs thanks to their high specific surface areas, abundant vacancies, and surface functional groups. 94 Thanks to their strong corrosion resistance, metal nitrides/carbides with metal centers exposed on their surface are good supports to stabilize isolated metal atoms via strong metal−support interactions. In this context, electrically conducting MXenes such as Mo 2 C have been explored as supports for the CO 2 RR. 95 Zhang et al. demonstrated an efficient approach to produce single atom copper immobilized on MXene for the electrosynthesis of methanol from CO 2 . The SACs were obtained via selective etching of hybrid A layers (Al and Cu) in quaternary MAX phases (Ti 3 (Al 1−x Cu x )C 2 ). 96 Combining X-ray absorption spectroscopy analysis and density functional theory calculations, they proposed that the Cu single atoms in the form of Cu δ+ with 0 < δ < 2 have a low energy barrier for the ratedetermining step corresponding to the conversion of HCOOH* to CHO*, a key reaction intermediate for the reduction of CO 2 to CH 3 OH (Figure 7b).

THE ELECTROLYTE: AN ACTIVE COMPONENT TO DRIVE REACTIVITY AND ENHANCE SELECTIVITY
4.1. Adjusting the Local pH at the Electrode− Electrolyte Interface. The pH value of the electrolyte greatly influences the equilibrium potential of the CO 2 RR and NRR, as highlighted in the partial Pourbaix diagrams for the CO 2 RR and NRR provided in Figure 8a and b, respectively. 97−100 A high local pH typically disfavor the HER, thus enabling higher Faradaic efficiencies for multicarbon products in the context of the CO 2 RR and for ammonia in the context of the NRR. 101,102 The groups of Sinton and Sargent have achieved remarkable results for the CO 2 RR in highly alkaline media; using 7 M KOH, they achieved a 1.3 A/cm 2 partial current density for ethylene in a flow cell. 103 Engineering of the triple-phase interface was key to these results and will be discussed further in section 5. Unfortunately for CO 2 electrolysis, the use of an alkaline electrolyte is complicated by the fatal exergonic formation of carbonate (CO 2 + 2OH − → CO 3 2− + H 2 O/ CO 2 + OH − → HCO 3 − ), which is detrimental to both energy and carbon efficiency. 104 Neutral bicarbonate electrolytes have been applied to reduce electrolyte consumption and to buffer the local pH, although at high currents CO 3 2− is still formed from CO 2 and electrogenerated OH − . Several studies have explored the dependence of the product distribution on the local pH at the electrode−electrolyte interface, as well as the concentration and buffer capability of the electrolyte. In that line, a fine-tuning of the product selectivity for CO 2 RR on Cu electrodes was achieved via the modulation of the local pH upon the variation of the electrolyte buffer capacity, CO 2 pressure, and current density. 105 Varela et al. proposed that electrolytes with a high buffer capacity could facilitate the transfer of coupled electrons/protons, thus being beneficial for the evolution of hydrogen. 106 By comparison, they found that electrolytes with a low buffer capacity could suppress the formation of H 2 due to the low concentration of protons near the electrode surface, favoring selectivity toward the formation of C 2 H 4 ( Figure 8c). Conversely, applying a higher current density can also lead to a higher local pH. This is due to a high consumption rate of local protons compared to the rate of mass transport of protons from the bulk electrolyte. Huang et al. modeled an electrode surface and found that, even in highly acidic electrolytes (pH = 1), local neutrality and alkalinity could be created above 200 mA/cm 2 . 107 They required at least 400 mA/cm 2 to produce multicarbon products. This improved the carbon efficiency considerably, although energy efficiency remains problematic. While a higher CO 2 pressure could result in a lower local pH at a constant electrolyte concentration, they demonstrated that it also favored ethylene formation by increasing the local *CO concentration and the corresponding *CO surface coverage. 108 Recently, Chen et al. reported that adjusting the thickness of a highly porous Au film allows the control of the mass transfer resistance and increases the local pH at the electrolyte−electrode interface of CO 2 reduction, which results in the promotion of the CO 2 RR while inhibiting the HER. 109 For the nitrogen reduction reaction, Xu et al. summarized the dependence of the formation of nitrogen-reduction intermediates on pH for aqueous media. 110 Due to the large overpotentials needed to activate N 2 and the low solubility of N 2 in aqueous electrolytes, when the applied overpotential is sufficient to trigger the electrochemical synthesis of NH 3 , the reaction at the active sites quickly becomes controlled by the mass transport of N 2 molecules. Consequently, the presence of protons near the electrode surface leads to the undesired production of hydrogen. As illustrated in Figure 8d, Wang et al. gauged the NRR performance of commercial Pd/C in electrolytes with different pH values. Their observations revealed that the effective suppression of the HER activity in the neutral electrolyte was attributed to a higher barrier for mass and charge transfer. 111,112 4.2. Optimizing the Components of the Electrolyte: Alkali Metal Cation Effects. Bicarbonate or carbonate are the most investigated electrolyte salts employed for the CO 2 RR, as they provide a near-neutral pH but most importantly allow a stable and high dissolved CO 2 concentration to be maintained upon operation. 113,114 Hence, while the nature of the anions are rarely explored in electrochemical studies, a wide range of studies have investigated the variation of the alkali cations. In the CO 2 RR, while the influence of alkali cations on product selectivity and catalyst efficiency are commonly accepted, 74,115,116 the origin of this effect is still largely debated in the literature. The influence of the used alkali metal cations on the CO 2 RR activity and selectivity is generally attributed to the relatively high concentration of alkali cations in the outer Helmholtz plane (OHP). Early work from Monteiro et al. proposed that large cations are specifically adsorbed more easily on the catalyst surface because of the fewer coordinated water molecules. 117 Adsorbed cations can also elevate the potential at the OHP and decrease the local proton concentration, suppressing the HER. 118 Alternatively, it was suggested that the cation size can significantly affect the rate of water hydrolysis by tuning the hydration energy. 119 For instance, the pK a value of Li + was calculated to be three times higher than that of Cs + . The hydrated Cs + acts as a buffer, maintaining a locally low pH near the electrode and increasing  (Figure 9b). 120 By applying a single set of cation sizes derived from experimental data, the model showed quantitative agreement with the experiments for the catalyst system on both Ag and Cu. Their theoretical model and experimental results indicate that the repulsive interactions derived from the hydrated cations in the Helmholtz layer should be responsible for the change of the surface charge and their electric field. The use of high-valent cations with a small hydration radius also increases the potential of zero charges or capacitance, which maximizes the surface charge density and the corresponding interfacial electric fields. 121 Bell's group provided insights regarding the beneficial effect of cations, particularly at relatively low overpotentials for which the reaction rate does not perturb the local pH. 122,123 Notably, the hydrogen and CH 4 partial currents remained steady, while formate, C 2 H 4 , and C 2 H 5 OH formation rates increased when using large alkali cations. The cation size-independent production of H 2 and CH 4 was attributed to the zero dipole moment of *H and *CHO, which are the corresponding reaction intermediates of the reactions (Figure 9c).
Alkali metal cations have also been used to promote the CO 2 RR in strongly acidic media. A key advantage to operating at a low pH is the improved carbon utilization efficiency, which is limited in neutral and alkaline media due to the formation of carbonate. Sargent and co-workers utilized a cation-augmenting layer to sustain a high K + concentration at the copper catalyst surface. 107 They achieved a Faradaic efficiency of 61% for CO 2 RR products and that of 40% for C 2+ products at 1.2 A/cm 2 , and by lowering the CO 2 flow they reached a single pass conversion efficiency of 77%. Gu et al. explored the effect of alkali cations on the CO 2 RR in acid with tin oxide, gold, and copper catalysts, achieving 90% Faradaic efficiencies for formic acid and CO. 124 Using a simulation based on the Poisson− Nernst−Planck (PNP) model, they predicted that the origin of such striking effects was the modulation of electric fields, which inhibited the migration of hydrononium ions.

The Search for Novel Electrolytes: Ionic Liquids and Nonaqueous Electrolytes.
Ionic liquids (ILs), which are defined as salts that remain liquid below 100°C, have been proven to be a promising new class of environmentally benign solvents. 125 By tuning the molecular structure and polarity of the IL, the CO 2 and N 2 absorption capacity and the ability to stabilize charged CO 2 and N 2 species can be tuned and optimized. ILs also possess several advantages, such as a wide electrochemical windows, thermal and chemical stability, negligible volatility, and electron transfer mediation for redox catalysis, which make them interesting alternatives for promoting the CO 2 RR and NRR. 126 As they are nonaqueous by nature, ILs allow control of the aqueous content to an optimum level to provide protons for hydrocarbon formation while suppressing the HER. 127−131 ILs have been extensively investigated for the CO 2 RR because the cations of ILs can form a complex with CO 2 and further activate it. Rosen et al. reported the use of 1-ethyl-3methylimidazolium tetrafluoroborate (EMIM-BF 4 ) as an IL electrolyte for the electrochemical conversion of CO 2 to CO on silver (Figure 10a). 132 The IL system lowers the energy of the *CO 2 intermediate via the formation of a complex intermediate, which lowers the energy associated with the initial step of the reduction reaction. 133 The formation of CO occurred at a very low onset overpotential, and the IL system demonstrated sustained production of CO for 7 h with a FE CO of more than 96%. ILs have also been applied with transition metal dichalcogenides, which are known to be more prone to promote the HER over other reduction reactions. Remarkably, Asadi et al. exfoliated WSe 2 nanoflakes to perform the electroreduction of CO 2 to CO using a 50 vol % [Emim]-   134 The current density, FE, and TOF in the production of CO were all superior at lower overpotentials, suggesting a high selectivity for the CO 2 RR (Figure 10b). Copper selenide nanocatalysts have been identified to convert CO 2 to CH 3 OH at low overpotentials in a [Bmim]PF 6 / acetonitrile/H 2 O mixed electrolyte. 135 In addition, in a [Bmim]BF 4 /H 2 O electrolyte, MoTe 2 could also be used as a catalyst for CO 2 reduction to CH 4 with a high FE of 83% at a relatively low overpotential. 136 Atifi et al. demonstrated that protic ionic liquids (PILs) derived from 1,8diazabicyclo [5.4.0]undec-7-ene (DBU) effectively promote the electrochemical reduction of CO 2 to formate (HCOO − ) with high selectivity (Figure 10c). 137 The use of PILs composed of the conjugate acid of DBU, [DBU-H] + , efficiently catalyzed the reduction of CO 2 to HCOO − (FE HCOOH ≈ 80%) with significant suppression of CO and H 2 production (FE CO + FE H2 ≈ 20%) in either acetonitrile or an acetonitrile/H 2 O mixed electrolyte. Ionic liquids and nonaqueous electrolytes with high N 2 solubility under ambient conditions can also increase the local concentration of N 2 near the catalyst surface by as much as 20 times compared to water on a volumetric basis. 138 MacFarlane and co-workers reported the use of ionic liquids with high N 2 solubility for the electroreduction of N 2 to ammonia at room temperature and atmospheric pressure. 139 As presented in Figure 10d, a FE NH3 as high as 60% was achieved in [P6, 6,6,14][eFAP]. Ortunõ et al. used DFT calculations to explore the nature of N 2 adsorption on different ions and found that a stronger interaction accompanied by charge delocalization will result in stronger adsorption of N 2 . 140 As shown in Figure 10e, they found that on a Ru surface the presence of ILs reduces the relative electronic energy of the N 2 RR intermediate N 2 H* more significantly than that of the HER intermediate, H 2 *, lowering the energy by 0.34 and 0.11 eV, respectively. Suryanto et al. identified the impact of the IL molar fraction (X IL ) on the physicochemical properties of the electrolyte mixture and the NRR performance. 141 A FE as high as 23.8 ± 0.8% with an NH 3 yield rate of 1.58 ± 0.05 × 10 −11 mol/s·m 2 was achieved for X IL = 0.23 at an optimal potential of −0.65 V vs. NHE (Figure 10f). Note that in this study, which predates the publication of standard NRR protocols mentioned in the above text, no 15 N labeling studies were provided, but an extensive purification of the N 2 reactant was carried out. The significant drop in the NRR performance when further increasing X IL highlights the role of 1H,1H,5H-octafluoropentyl 1,1,2,2-tetrafluoroethylene ether (FPEE) in facilitating the mass transport of N 2 in the electrolyte. The authors suggested that other factors correlating FE and X IL could play a role, such as the presence of complex molecular interactions and the different diffusion behaviors of neutral N 2 molecules and polar H 2 O within the mixed electrolyte system, a known phenomenon with ionic liquids. 142 4.4. Solid-State Electrolyte Designs. Conventional liquid electrolytes used in the CO 2 RR and NRR, such as KHCO 3 , Na 2 SO 4 , or KOH, mainly have three purposes: (i) to transport ions between the cathode and anode for efficient current flow, (ii) to provide protons for successive PCET, and (iii) to solvate liquid products. The mixture of liquid products and ion impurities requires energy-and cost-intensive downstream separation steps to obtain pure products, which complicate the infrastructure for delocalized production. 143 To tackle this problem, the concept of solid-state electrolytes was proposed, inspired by the progress in solid-state electrolytes for batteries. 144 A solid-state electrolyte is typically placed between ion-exchange membranes with close contact to efficiently transport the generated ions and minimize the ohmic loss of the device. 145 Remarkably, solid-state electrolytes were found to be very effective in suppressing the HER by limiting the flow of protons to the catalyst active sites during the electrochemical CO 2 RR. 146 The Wang group reported the continuous electrocatalyic conversion of CO 2 to pure liquid fuels using two-electrode systems with solid electrolytes. 147,148 They applied a porous solid electrolyte (PSE) layer composed of styrenedivinylbenzene copolymer microspheres with sulfonic acid functional groups for proton conduction. Using a formic acid-selective bismuth catalyst (FE HCOOH ∼ 97%), the electrochemically generated protons and formate anions could combine at the PSL to produce formic acid (Figure 11a). By directly flowing a carrier gas instead of deionized water through the PSL, the authors were able to collect product vapors that could be condensed to form the pure product (almost 100 wt % formic acid), alongside an impressive current density and stability (Figure 11b).
Sheets et al. proposed a novel polymer gel approach to convert N 2 to NH 3 at mild temperatures (30−60°C) and pressures (20 psig). 149 As illustrated in Figure 11c, the polymer gel electrolyte helped to control the rate of the HER by limiting water transport and boosting N 2 transport, thus improving the selectivity toward the NRR.

THREE-PHASE INTERFACE ENGINEERING
The abundance of protons near the catalyst active sites constitutes a significant challenge for the catalyst selectivity vs the competing HER in aqueous electrolytes, resulting in low selectivity and activity for the CO 2 RR and NRR. A mitigation strategy resides in facilitating the accessibility of the catalyst to (c) Schematic illustration of CO 2 mass transport inside the catalyst layer with added PTFE, including gas-phase diffusion (solid red arrows) and aqueous-phase diffusion (dashed blue arrows). The dashed rectangles indicate catalyst areas that are only exposed to the electrolyte, thoseexposed to both electrolyte and gaseous CO 2 , and those only exposed to gaseous CO 2  high concentrations of CO 2 or N 2 molecules. While protons (H + ) are readily available in aqueous solutions via water ionization, the supply of CO 2 and N 2 molecules to the catalyst surface is limited by their low concentration and slow diffusibility. In saturated aqueous electrolytes, the solubility of CO 2 in H 2 O is 33 mmol/L at 298 K and 1 atm pressure, whereas the value for N 2 in H 2 O remains as low as 0.7 mmol/ L. 19 By comparison, the concentration of protons in a neutral aqueous electrolyte is typically 2.7-fold and 132-fold higher than the concentrations of CO 2 and N 2 , respectively.
In the context of CO 2 RR, Raciti et al. demonstrated that the local concentration of CO 2 at the catalyst surface can reach zero under strong reaction driving force conditions, hence lowering the selectivity by limiting the supply of the reactant. 150 Significant advances to minimize this reactant supply issue at the electrode have been made thanks to the implementation of efficient three-phase interfaces between gaseous CO 2 , the liquid electrolyte, and the solid catalyst. The most typical realization of such a three-phase interface involve porous gas diffusion layer (GDL) electrodes, which allow the delivery of gas-phase CO 2 directly to the catalyst active sites. Such a strategy, resulting in higher CO 2 and lower H + surface concentrations, has the potential to improve CO 2 RR performances while significantly lowering competitive HER. The properties of the GDL can affect CO 2 and water transport heavily, and main advances in this field have been recently reviewed 151,152 and will not be extensively reviewed here in the context of CO 2 RR. Thinner GDL/catalyst layers shorten the CO 2 diffusion distance, raising the relative CO 2 concentration; however, excessively high concentrations can decrease multicarbon product formation by competing with intermediates such as CO for binding sites. Tan et al. found that by adjusting the catalyst layer structure and the CO 2 feed concentration and flow rate they could establish a moderate local CO 2 concentration that was optimal for multicarbon product selectivity. 153 Alternatively to requiring GDL-based electrodes, the catalyst support itself can be modulated to modulate the three-phase interface via a fine-tuning of the local microenvironment near the catalyst surface through nanostructuring and surface functionalization. Inspired by biological strategies to entrap a gas layer at the surface of a solid, and in particular by the plastron effect enabling the diving bell spider to breathe underwater, Wakerley et al. functionalized porous dendritic Cu electrodes generated via the DHBT strategy mentioned above in section 3 with long-chain alkanethiols. The resulting superhydrophobic Cu electrodes demonstrated a sixfold decrease of HER upon treatment with the alkanethiol and a subsequent drastic increase in CO 2 reduction selectivity. 22 They proposed that the hydrophobicity establishes triple-phase interfaces at the electrode where CO 2 mass transport is omnidirectional and H + mass transport is unilateral ( Figure  12a). This increases the local CO 2 concentration and thereby the surface concentration of Cu-COOH* and Cu-CO*, enhancing C−C coupling. This study led to the identification of the role of hydrophobicity and the formation of gaseous voids as effective levers to orient the reaction pathway toward the formation of multicarbon products. Khan and co-workers explored the idea of gas-trapping further using a gasphilic silicon substrate in proximity to the catalyst layer. Creating a CO 2 plastron adjacent to the catalyst improved mass transfer, enriching and maintaining the local CO 2 concentration. Using a smooth copper catalyst, they recorded improved activity and a decrease in FE H2 (13% compared to 29% with bulk CO 2 bubbling). These trends were replicable using nanostructured copper, demonstrating the transferability of such an approach to different catalysts. 154 Moreover, Xing et al. showed that a hydrophobic microenvironment can significantly enhance CO 2 electrolysis by facilitating reactant diffusion (Figure 12b). 155 Using commercial copper nanoparticles dispersed with hydrophobic polytetrafluoroethylene (PTFE) nanoparticles, they reported improved activity and Faradaic efficiency for CO 2 reduction with a partial current density of >250 mA/cm 2 and a single-pass conversion of 14% at moderate potentials. Importantly, this performance was approximately twice as large as that of regular electrodes without added PTFE. Similar findings were also observed from a Bibased catalyst modified with PTFE nanoparticles in the catalyst layer to demonstrate a partial current density of 677 mA/cm 2 for formate and 35% single-pass CO 2 conversion at −0.7 V vs. RHE (Figure 12c). 156 Pham et al. compared various ionomeric binders on a Cu catalyst and achieved a 77% Faradaic efficiency and 600 mA/ cm 2 partial current density for C 2+ products at −0.76 V vs RHE using a fluorinated ethylene propylene (FEP) binder. 157 They attributed these results to the hydrophobic properties of FEP. The Sinton and Sargent groups have also done notable work on modulating the three-phase interface in continuous flow and membrane electrode assembly (MEA) electrolyzers, enabling high current densities (e.g., > 1 A/cm 2 ) to be achieved. 103,146 For example, they presented a catalyst:ionomer bulk heterojunction (CIBH) architecture that had both hydrophilic and hydrophobic functionalities. By having different domains that favored gas and ion transport routes, they were able to decouple gas, ion, and electron transport, extending the reaction interface from the submicrometer range to the several micrometer range. 103 These examples illustrate that the moderate hydrophobicity of the catalyst layer can establish a microenvironment with a balance between gaseous CO 2 and liquid electrolytes inside the catalyst layer. Such microenvironments�equivalent to microreactors� reduce the thickness of the diffusion layer, accelerate CO 2 mass transport, and link highly active reaction zones at the interfaces between the three phases involved in the reaction. 158 The triple-phase interface can also be further tuned by applying ionomers to control pH and CO 2 /H 2 O concentrations. Bell and co-workers postulated that anion-exchange ionomers (e.g., sustainion) increase CO 2 solubility, cationexchange ionomers (e.g., nafion) increase local pH by trapping OH − ions, and both types increase water concentration. 159 By optimizing a bilayer ionomer coating and coupling to pulsed electrolysis, they achieved a Faradaic efficiency of 90% for C 2+ products and that of just 4% for H 2 .
In the case of NRR, when applying large potentials at the electrodes, the kinetically facile HER becomes preferable to the reduction of N 2 due to the relatively low energy barrier associated with the reaction. It was suggested that the HER should always dominate at normal proton concentrations near the metal electrode surface. However, when few protons or electrons are provided, the NRR may preferentially occur, as recently observed experimentally. Designing a triple-phase interface for NRR can increase the local N 2 concentration and improve *N 2 adsorption while limiting the availability of protons by reducing contact with the electrolyte. 160 Using this strategy, Zhang et al. realized triple-phase electrolysis via in situ fabrication of Au nanoparticles located on hydrophobic carbon fiber paper (Au/CFP) (Figure 12d). 161 The hydrophobic  (Figure 12e). A sharp decrease in the local concentration of protons does not benefit the NRR process, as protons are necessary for the successive PCET steps associated with the formation of ammonia. These investigations point out that although the release of hydrogen is a competitive reaction, protons are paradoxically essential to increase the ammonia yield rates. 162

CONCLUSIONS AND PERSPECTIVES
The industrial development of the CO 2 RR and NRR is currently plagued by low Faradaic and energy efficiencies. The successive PCET steps associated with the corresponding reaction intermediates increase the complexity and complicate the search for an ideal catalyst. In contrast, the simplicity of the HER mechanism and the abundant presence of protons in traditional electrolytes make the production of hydrogen a competitive and parasitic reaction that consumes a significant amount of electrons to the detriment of the fixation of CO 2 and N 2 . Additionally, the selectivity toward a single product, particularly important in the context of CO 2 RR is a central point to be considered. Multiple strategies have shown promise but still require the elaboration of a robust and rational framework; they have demonstrated that optimal activity and selectivity can be obtained upon modulating thermodynamics and kinetics of the reaction. By engineering the catalyst, electrolyte, and reaction interface, three main strategies have been applied toward that goal ( Table 2): (i) targeting a narrow distribution of molecularly defined active sites, (ii) increasing the reactant/proton ratio at the three-phase interface where the reaction takes place to lower the undesired formation of H 2 , and (iii) the stabilization and confinement of reaction intermediates in the electrode vicinity to favor the formation of multielectron reduction products. The complexity of the parameters involved to address these challenges simultaneously further highlights the interest in combining experimental and theoretical approaches to guide the design of both catalysts and electrolyzers for the CO 2 RR and NRR. From this perspective, machine learning will help rapid screening of catalysts with high selectivity based on massive data in the silico database by focusing on near-optimal bond energy with adsorbates, such as *CO and *N 2 H. In addition, enabling a better understanding and control of the PCET steps, notably via the elaboration of a robust framework to link the relative contributions of charge transfer and protonation steps to overpotential and the distribution of surface species, will be key to further rationally improve electrocatalysts.
This Review illustrated several examples displaying industryrelevant performances in terms of selectivity and current densities, highlighting the potential of electrochemical approaches for the preparation of carbon-and nitrogencontaining molecules. However, for the CO 2 RR, many studies have been performed in alkaline or neutral media, resulting in carbonate formation in the electrolyte. This is detrimental to carbon utilization and energy efficiency, especially considering the energy that would be required to regenerate spent electrolyte. This problem has been considerably underestimated and overlooked for some time; however, an increasing number of studies over recent years have attempted to tackle this issue. Using acidic electrolyte prevents carbonate crossover to the anode and regenerates CO 2 close to the cathode surface, improving carbon utilization efficiency. Naturally, this media poses challenges regarding hydrogen evolution, and the application of strategies covered in this review will be pivotal in overcoming this.
Moreover, most of the presented strategies introduced in the present Review enable the improvement of catalyst selectivity for a relatively short period of time but have not been investigated over industrially relevant time scales. Maintaining high selectivity for the CO 2 RR and NRR over long operation times remains the largest challenge to date, as rapid loss in activity and selectivity is observed for most of the systems reported. This notably results from the fact that in operation