Electric Double Layer Structure in Electrocatalytic Carbon Dioxide Reduction

Environmental degradation and climate change caused by excessive CO2 emissions have become the most serious challenges facing humanity. The electrocatalytic CO2 reduction reaction (CO2RR) is an ideal way to produce high‐value chemicals and solve environmental pollution at the same time. The electric double layer (EDL) structure at the electrochemical interface can greatly influence the local catalytic environment, but it is not fully understood. Herein, recent advances in EDL structure for CO2RR are focused. The main components of the EDL, including adsorbates on the electrode surface, metal/water interface, cations, anions, and differential capacitance are described. The latest experimental and theoretical understanding of EDL and their impact on CO2RR are presented. Finally, the current challenges encountered and the outlook for future development are discussed. This review describes the influence of EDL structure on CO2 reduction, which is of guiding significance for the design of new and efficient CO2 electrocatalysts.

(OHP): This is where the cations in the electrolyte gather, and in response to the applied voltage, a local electric field is formed between the cations gathering here and the electrode surface.
3) Diffuse layer: Due to electrostatic repulsion and thermodynamic diffusion, ions form a diffuse layer whose concentration decreases from the diffusion layer outward.
IHP includes all adsorbates specifically adsorbed on the electrode surface. For CO 2 RR, the intermediates of CO 2 RR reduction are the adsorbates that have the greatest influence on the reaction process. The proton-coupled electron transfer (PCET) of each intermediate species at different atomic sites affects the final reduction product. The water molecule is an important background connecting IHP and OHP for the EDL. The interfacial water layer in contact with the metal electrode has many peculiar properties compared to bulk water. [14] Most importantly, charge transfer may occur between the interfacial water and the metal surface, and since water molecules are strongly polar molecules, their orientation, dipole moment, and work function may change significantly, which can have a non-negligible effect on CO 2 RR. OHP is mainly composed of cations. Under an applied negative voltage, electrons will accumulate on the metal surface. At the same time, due to electrical attraction, cations will accumulate near the metal surface leading to the formation of a local electric field between the metal and the cations. Since CO 2 RR intermediates are polar molecules, this local electric field affects the adsorption of intermediates on the active site, which in turn changes the activity and selectivity of CO 2 reduction. The diffusion layer usually has less effect on CO 2 RR due to its distance from the active site where the reaction takes place.
In this review, we focus on recent advances in EDL structure for CO 2 RR. We divided the EDL into three parts, chemisorbed adsorbates, metal/water interface, and cation. Their effects on the CO 2 RR are also introduced. In addition, the influence of some other parts on the overall structure of the EDL, such as anion and differential capacitance, is discussed. We summarize the current challenge of EDL, and provide an outlook on the further design of novel highly selective and active CO 2 RR catalysts from the EDL aspect.

Chemisorbed Adsorbates
The reaction processes at electrochemical interfaces are often influenced by the morphology, adsorption sites, and kinetic migration of various adsorbates on the electrode surface. [15] These adsorbates may interact with atomic arrangements, defects, and steps on the electrode surface. Together, these interactions affect the kinetic barriers to reduction and the final reduction products.

CO 2
The prerequisite for CO 2 reduction is its adsorption and activation on the electrode surface. CO 2 is a linear nonpolar molecule bound in the form of O─C─O. Its bond length is about 1.16 Å and it has two stable delocalized π orbitals, so it's hard to activate ( Figure 2a). In the bonding process of CO 2 , O is more electronegative than C, and electrons are transferred from C to O. Therefore, the lower unoccupied molecular orbital (LUMO) orbitals are concentrated near C, while the highest occupied molecular orbital (HOMO) orbitals are concentrated near O. Since the HOMO of CO 2 is located in the deep energy level, its electron giving ability is weak. Therefore, CO 2 is a weak electron donor and a strong electron acceptor. The main way of CO 2 activation is the injection of electrons into its LUMO by strong electron donors (e.g., Fe, Co, Ni, Cu, Pt, and Pd). [16] The mode of electron transfer also determines how CO 2 binds to the active site (Figure 2b-e). When CO 2 accepts electrons from the catalyst surface, its C-terminal is usually in contact with the active site which facilitates electron transfer to the LUMO of CO 2 (Figure 2b). In addition, some catalysts can gain electrons from the HOMO   orbital of CO 2 (e.g., TiO, CrO, MnO, and VO), [17] which are usually in contact with the active site at the O-terminal (Figure 2c,d).
When the active center contains both electron acceptor sites and electron donor sites, the C═O bond of CO 2 is in contact with the catalyst surface at the same time ( Figure 2e). [18] When CO 2 is activated, the subsequent reduction products are often closely related to the way in which CO 2 binds protons. The CO 2 reduction reaction is a complex multi-step PCET process. Different ways of binding protons can lead to the production of different intermediates. These intermediates are influenced not only by the catalyst itself but also by many external factors such as electrode potential, pH, and electrolyte.

C 1 Intermediates
In aqueous solutions, the common products of electrocatalytic reduction of CO 2 to C 1 are carbon monoxide (CO), formic acid (HCOOH), and methane (CH 4 ). It is generally believed that CO is a simple CO 2 reduction product, which only requires two steps of PCET. It is now generally accepted that the O-terminus of CO 2 is first hydrogenated to produce a *COOH intermediate, followed by hydrogenation to form the adsorbed *CO, after which *CO is desorbed to produce CO molecules. [19] The * COOH is a key intermediate that limits the reaction rate. Au is the most active metal catalyst for the electrochemical conversion of CO 2 to CO, and it can achieve high Faraday efficiencies in the potential range of À0.25 to À0.67 V versus reversible hydrogen electrode (RHE). [20] In addition, Ag, Zn, Cd, and Pd are also typical metals that can reduce CO 2 to CO. [21] Formic acid is the simplest monoacid and can be used as a potential fuel for fuel cells.  [23] Using *OCOH as a descriptor, Feaster et al. combined experimental current densities and theoretically calculated binding energies to derive the volcano plot for the formation of HCOOH from different metals at À0.9 V versus RHE. [24] It can be found that Sn is located at the top of the volcano plot and it is the most active catalyst for the production of formic acid. Au, Ag, Pt, and Cu all have low *OCOH binding energies, so the selectivity of CO 2 reduction to HCOOH is also low. Zn and Ni have too strong *OCOH binding energy, which is not conducive to the desorption of HCOOH from the catalyst surface. In addition, some other metals also show high HCOOH selectivity, such as Bi, Pd, Pb, and In. [25] Methane is an important hydrocarbon in industrial production, and it is also a raw material for the manufacture of synthesis gas and chemical products. The key to CH 4 formation is the further protonation of *CO. Its first two steps are the same as CO, but when *CO is harder to desorb from the surface, it may be further hydrogenated. Cu is a common catalyst for CH 4 production, and Peterson et al. investigated the lowest energy path for the reduction of CO 2 to C 1 products on the Cu(211) surface by free energy calculations (Figure 3a-d). [26] They proposed that *CO hydrogenation to form *CHO is the rate-determining step. The reaction path is CO 2 !*COOH!*CO!*CHO!*CH 2 O!CH 3 O!CH 4 . The rate-determining step is the hydrogenation of *CO to *CHO with a reaction energy of 0.74 eV. Furthermore, Cheng et al. studied the reaction energy barrier for CH 4 formation on Cu(100) surface by quantum molecular dynamics. [27] Considering the solvent effect, they found the adsorption of *CHOH on the Cu(100) surface is a key factor in determining the high selectivity of CH 4 relative to CH 3 OH. Recently, some new catalysts for CH 4 generation have been developed. For example, Dutta et al. prepared Ag nanofoam and found them to be effective in reducing CO 2 to CH 4 . [28] The maximum Faraday efficiency is 51% at À1.5 V versus RHE. However, the Ag nanofoam was not stable and rapid degradation of the catalyst surface was observed during the reaction, with its Faraday efficiency decreasing to 32% within 5 h. Despite all this, this work shows that rational design can lead to the development of Cu-like catalysts for the production of hydrocarbons.

C 2 Intermediates
The C 2 product has a higher energy density than the C 1 product, which is important for long-term large-scale renewable energy storage and industrial applications. In CO 2 RR, conditions for the formation of C─C bonds are very demanding, which makes the formation of C 2 products very difficult. [7] So far, Cu is the only metal that shows high selectivity for the reduction of CO 2 to C 2 products, the highest Faraday efficiency of its simplest C 2 product (C 2 H 4 ) is %60%. [29] It is generally believed that the C 2 product is formed by the coupling of two C 1 intermediates. Numerous experiments and theories have shown that the formation of C 2 intermediates does not involve the PCET process, and the coupling of *CO and *CO* or the coupling of *CO and *COH is the key.
Ethylene (C 2 H 4 ) and ethanol (C 2 H 5 OH) are the common C 2 products generated in Cu-based catalysts. Compared to C 1 , the formation of C 2 products is more complex, with more intermediates. Kortlever et al. investigated the possible pathways for the generation of C 2 H 4 and C 2 H 5 OH on the Cu surface by theoretical calculations ( Figure 4). [30] The process of C 2 H 4 generation is firstly through C─C coupling of C 1 intermediate to form *COCOH, and then hydrogenation and dehydration to form *CCO. Further hydrogenation of *CCO gradually forms *CHCO, *CHCHO, and CH 2 CHO. Then, CH 2 CHO is hydrogenated to form C 2 H 4 , and O atoms on the catalyst surface can continue to hydrogenate to form water molecule desorption. C 2 H 5 OH can be formed by the continued hydrogenation of the two C atoms of *CH 2 CHO.
The introduction of other metals to form bimetallic catalysts is an effective method to regulate the selectivity and activity of C 2 products. It can regulate the electronic structure of the active site through the interaction between different metal elements, thus reducing the reaction energy barrier of C─C coupling. Considering that the C─C coupling is affected by the coverage of *CO on the surface, the introduction of other metals in Cu to increase the Cu surface CO concentration is an effective strategy to improve the selectivity of C 2 products. Buonsanti et al. prepared Ag-Cu nanodimers with tunable domain sizes and found the C 2 H 4 Faradaic efficiency for C 2 H 4 was 3.4-fold with pure Cu nanoparticles. This is attributed to the tandem catalytic effect between Cu and Ag, which can effectively reduce the C─C coupling energy barrier. [31] Du and coworkers reported Cu-Zn alloy catalysts that achieved a Faradaic efficiency of 33% for C 2 H 4 at À1.1 V versus RHE. [32] The high C 2 H 4 production activity is due to the uniform distribution of Cu and Zn on the catalyst surface, which can reduce the CO transport distance between different active sites. Kenis et al. reported a phase-separated CuPd alloy with a Faraday efficiency of 48% for C 2 H 4 and 15% for C 2 H 5 OH at À0.8 V versus RHE. [33] In this catalyst, the proportion of Cu atoms adjacent to the surface is relatively high, which is conducive to C─C coupling.
In conclusion, CO 2 RR is a complex multi-step PCET process with a large number of reaction intermediates. The hydrogenation of each intermediate at different atomic sites is associated with the overall selectivity. The combination of experiments and theory to improve the understanding of the PCET process can help to further understand the reaction mechanism and rationalize the design of high-performance CO 2 RR catalysts.

Water/Metal Interfaces
In electrochemical systems, water is involved in every part of the electrocatalytic reaction. The structure and properties of the water/metal interface play an important role in EDL. The correct description of the interactions between water and metals is also the focus of theoretical and experimental attention in CO 2 RR. Up to now, with the development of experimental techniques and the improvement of computer performance, it has become possible to describe the interfacial behavior of water molecules at the atomic level by scanning tunneling microscopy (STM) and density functional theory (DFT).

Fundamental Structure of the Water/Metal Interface
Water takes a variety of forms on metal surfaces. There is theoretical and experimental evidence that isolated water molecules ( Figure 5a,d,g), dimers (Figure 5b,e,h), and hexamers ( Figure 5c,f,i) are all common forms of water. [34] Isolated water molecules usually adsorb in a flat geometry to the top sites of the metal surface. [35] This adsorption mode facilitates the interaction of water molecular orbitals with the metal surface sites. Michaelides et al. investigated the adsorption of isolated water molecules on different metal surfaces by first principles calculations, and found that their adsorption energies ranged from À0.1 to 0.4 eV. [36] The adsorption strengths of the different metals were ordered according to Au < Ag < Cu < Pd < Pt < Ru < Rh. The distance between the water molecule and the top atom on the metal surface is also about 2 to 3 Å. Water may be present as a dimer on metal surfaces. The formation of this water dimer results from the dissipation of the additional energy gained by the initial kinetic energy of water-metal and water-water to the surface phonons and internal water degrees of freedom, and the two are located at stable sites in the form of hydrogen bonds. [37] Driven by the initial kinetic energy, the dimer will keep moving, but the two will still approach each other due to the hydrogen bonding. [14] Compared to isolated water molecules, the adsorption energy of water dimers is much stronger on Pt, [38] Pd, [39] and Pd/Au surfaces by DFT calculation. [40] Hexamers are also a common form of small water clusters, which are thought to be the smallest structural units of ice. [41] This has been demonstrated in several STM experiments. [42] In the hexamers, the different water molecules depend mainly on the interaction of oxygen atoms with the hydrogen atoms of other water molecules to keep the structure stable. The regular hexamer is very similar to the surface atomic arrangement of dense hexagonal metal, so it is easy to stably adsorb on the surface of the dense hexagonal metal. Multiple hexamers joined on the metal surface will form ice-like water bilayers. In the water layer, one oxygen and one hydrogen atom from each water Figure 5. Experimental scanning tunneling microscopy (STM) images and computationally simulated interfacial water models. a,d,g) are images of water molecules, dimers, and hexamers for STM, respectively. b,e,h) are the geometric structures of water molecules, dimers, and hexamers respectively. c,f,i) are the side images of the geometric structure. Reproduced with permission. [34] Copyright 2012, Springer Nature.
www.advancedsciencenews.com www.advenergysustres.com molecule are arranged in parallel to form a hexamer lying flat on the surface. The other hydrogen atom can point toward the surface or away from the top of the surface, which is called an H-down or H-up structure. [43] In general, the energy difference between the H-down and H-up structures is small depending on the metal. DFT calculation shows that the H-down structure is more stable on Rh(111), Ag(111), Pt(111), and Pd(111), while Ni(111), Cu(111), and Ru(0001) prefer the H-up structure. [36] In electrochemical systems, more attention is paid to the water film on the metal surface. Understanding the liquid properties of interfacial water is critical to reveal the nature of the EDL. However, when the liquid water is in direct contact with the metal interface, the structure of the interface water can no longer be directly detected as it is under low water coverage and ultrahigh vacuum conditions.
With the rapid development of computer performance in recent years, ab initio molecular dynamics (AIMD) has shown good potential for revealing the microstructure of interfacial water. Due to the limitation of early calculation conditions, some early AMID studies mainly focused on a single water layer on the metal surface. Izvekov et al. initially explored the Cu(110) surface with 9 Cu atoms covered by 12 layers of water molecules and found that the initial water structure was very unstable. [44] After that, they determined a bilayer water structure consisting of 7 water molecules. However, it is difficult to make a direct comparison with the experimental results in the case of fewer water molecules. Using AIMD to simulate the real properties of liquid water, the vibration spectra derived from the Fourier transform of velocity autocorrelation function can be directly compared with the experimental spectral correlation data. Meng et al. investigated the vibrational spectra of Ru(0001), Rh(111), Pd(111), and Au(111) at low temperatures, and found that frequencies below 1000 cm À1 belonged to the free movement of water molecules. The vibration mode of HOH bending may be around 1600 cm À1 . Schnur et al. raised the temperature of the AIMD simulation to room temperature and found results consistent with the experiment that the vibrational spectrum of the water layer on the metal surface has many wider spectra. [45] Since the CO 2 RR process is a multi-step PCET process, the orientation of the interfacial water layer is also an important factor affecting the reaction rate. Le et al. simulated the motion of water molecules on Pt(111), Pd(111), Au(111), and Ag(111) by AIMD at 330 K. [46] At this temperature, the ice-like structure has disappeared and the interfacial water has a layered structure. Compared to Au(111) and Ag(111), Pt(111) and Pd(111) bind more strongly and reactively to water molecules. The pattern of oxygen atoms pointing toward the metal surface and the other two hydrogen atoms moving away from the surface occurs more frequently in Pt(111) and Pd(111). The different orientations may have affected the binding strength of water molecules to the surface Although AIMD has provided some understanding of the interfacial water film at the molecular atomic level and has been able to directly match some experimental spectral data, there are still some difficulties in the current theory. For example, the process of molecular bonding and breaking during chemical reactions requires a more accurate description of the potential energy surface. The corresponding statistical implementation of the reaction kinetics requires significant computational effort. In addition, the temperature, volume, and particle number determined by the simulation system cannot respond accurately to the change of the applied electrode potential. This makes simulating interfacial water a very challenging task.

Electronic Properties of Interfacial Water
The biggest difference between interfacial water and bulk water lies in the charge transfer between the metal surface and the water layer, which leads to a strong polarization of the interfacial water and generates a built-in electric field. Sakong et al. evaluated the charge polarization at the electrochemical metal-water interface using an AIMD simulation of 144 water molecules on a Pt(111) slab ( Figure 6a). [47] In the AIMD simulation of up to 40 ps, the calculated average work function is around 4.96 eV (Figure 6b), which is very close to the experimental absolute zero charge potential of 4.9 eV. In addition, the potential perturbation mainly occurs in the first water layer near the electrode Figure 6. Work function and zero charge potential of ab initio molecular dynamics (AIMD) simulation on Pt(111) surface. Reproduced with permission. [47] Copyright 2018, AIP Publishing.
www.advancedsciencenews.com www.advenergysustres.com (about 3 Å) (Figure 6c). The CO 2 RR process usually occurs at this scale, so the charge polarization between the interfacial water and the electrode will greatly affect the formation of the reaction intermediate. For metallic electrodes, there is a very important concept in electrochemical systems which is the potential of zero charge (PZC). PZC is potential without residual charge on the electrode surface, in which case there is no strong charge separation characteristic between the electrode and the electrolyte. Due to its importance in understanding the EDL structure and interfacial potential, many experimental techniques have been developed to determine the PZC of metal electrodes. However, it is difficult to prepare a single crystal electrode and exclude electrolyte ions adsorbed specifically on the metal surface, which leads to great uncertainty in many measurements. [48] Based on the AIMD method, Le et al. developed the computational standard hydrogen electrode methods to determine the zero-charge potential of metal electrodes. [49] This method uses the solvent energy of aqueous protons as a reference to avoid the expensive calculation of

Effect of Interfacial Water on CO 2 Reduction Intermediates
The CO 2 RR process contains not only the correlation between the water and metal interfaces but more importantly the influence of these interfacial waters on the intermediates. The process of understanding the effect of aqueous solvents on catalytic reactions can be mirrored in the field of theoretical calculations from ignoring solvent effects to modeling solvents. In the past years, the computational hydrogen electrode model (CHE) developed by Nørskov et al. has been applied very successfully in predicting the development of new catalysts and explaining the reaction mechanism. [50] Electrocatalytic CO 2 reduction can generally be assumed to be a multi-step process of PCET occurring at electrode surfaces. The largest free energy step is the rate-determining step of the entire reaction, which is a function of the operating electrode potential. To accommodate the effect of the external potential, 0 V versus RHE was defined as equilibrium potential. Hydrogen production was in a state of rapid equilibrium on reversible hydrogen electrodes. In this situation, it can be assumed that protons and electrons have half the energy of hydrogen. As a result, the free energy of the adsorbed intermediate can be easily calculated by using the total energy and zero energy calculated based on DFT and the experimental entropy corrections. As mentioned earlier, they successfully calculated the lowest energy path of CO 2 reduction to CO, CH 4 , and HCOOH intermediates using the CHE model. [26] They also developed activity descriptors that can be used to predict the formation of CH 4 from different transition metals. [51] Although the CHE model combining DFT theory is widely used in the field of electrocatalysis and has made many successful applications in predicting catalyst activity and explaining CO 2 reduction mechanisms, the simple model also has some shortcomings. For example, to calculate the free energy, the first step is to build a reasonable slab model to simulate the electrode surface. The slab models are usually constructed based on specific crystal faces of the catalyst and exposed to a vacuum of about 10 to 20 Å. Nevertheless, the electrode surface with a high vacuum does not exist in the experiment, and the catalytic environment is usually located at the interface between the electrode surface and the electrolyte. The electrocatalytic reaction interface is composed of the complex interaction of electrode, adsorbent, and electrolyte. Ignoring the solvent effect cannot simulate the real catalytic reaction process.
There are two ways in which solvent effects can be included. One is an implicit solvent model that treats the aqueous solvent as a polarizable dielectric, which has the disadvantage of not considering the bonding and orientation of solvent molecules. [51,52] In addition, a straightforward strategy to consider solvent effects is to include one or more explicit water molecules in the periodic slab model. Including some water molecules are helpful to accurately simulate the interaction between adsorbates, solvent, solutes, and electrode surfaces. This method is relatively more accurate, but it is also more computationally expensive. Heenen et al. compared the adsorption energies of CO 2 RR intermediates calculated by three methods: vacuum, implicit solvent, and explicit solvent (Figure 7a,b). [53] The explicit solvent model is built by the AIMD method. In fact, the disparities that result from different approaches are very obvious. The deviation between AIMD and implicit solvation is large, especially for strongly solvated adsorbents such as *OH and *OOH for metals'(111) surface. Hydrogen bonding and competitive water adsorption are the main contributors to the solvent energy in AIMD simulations, which is the main source of difference.
CO molecules have special vibrational characteristics. These characteristics are easily recognized by spectral techniques, such as surface-enhanced infrared (IR) spectroscopy, [54] frequency generation spectroscopy, [55] and attenuated total reflection infrared spectroscopy (ATR-IR). [56] Therefore, CO is often used as a molecular marker to detect changes in molecular structure and environment. In addition, CO is also an important intermediate of CO 2 RR. Based on Car-Parrinello molecular dynamics, Lan et al. investigated the CO coverage of Pt(111) surface ( Figure 8). [57] In the simulation, CO molecules are adsorbed on the surface of Pt(111) in an upright mode, where C atoms form bonds with Pt atoms. There are three possible adsorption sites for adsorbed CO at different positions on the surface: top site, hollow site, and bridge site. Different site C atoms bond to different Pt sites. Interestingly, different CO coverage significantly changes the structured water layer formed on the Pt(111) surface. At θ ¼ 0.11 (Figure 8a), CO adsorbed on Pt surface mainly in top site. When θ ¼ 0.25 (Figure 8b), the adsorption mode of CO remains unchanged, but the change of water layer is very obvious. The number of chemisorbed water molecules is reduced, and the characteristics of the interface water are also weakened. At the high coverage θ ¼ 0.5 (Figure 8c), the top adsorption and bridge adsorption exist at the same amount. The water is also more liquid on top of the CO molecules, which means that the adsorption of the CO molecules isolates the water molecules from interacting with the surface Pt. The surface aqueous solvent molecules have an important effect on the reaction energy barrier and the selectivity of the reduction products for CO 2 RR. Some theoretical calculations have found that vacuum, single water molecules, multiple water molecules, and multiple water layers have different effects on CO 2 RR. Asthagiri et al. included a definite water molecule in their computational model (Figure 9a), [58] which is significantly different from the results of the CHE vacuum model by Nørskov   et al. [26] CHE method shows that the rate-determining step is the reduction of CO to formyl (CHO) on Cu surface for CO 2 reduction to methane. However, in the Cu model with water molecules, methane is produced through the hydroxymethyl (COH) intermediates, and the CO will be converted to COH instead of CHO, which is consistent with the previous experimental results.
The results show that solvent effect greatly affects the accuracy of theoretical calculation. Furthermore, Hussain and co-workers investigated the current efficiency of transition metals for CO 2 reduction by considering multiple water molecules on the electrode surface (Figure 9b). [59] In this model, the explicit solvent is defined as a bilayer of ice with different concentrations of hydrated hydrogen ions. The corresponding electrostatic potential can be adjusted by changing the number of H atoms in the added surface in the model. Their results show that the current efficiency calculated by the multiple water molecular model is remarkably good with the experimentally measured data. Besides, Cheng et al. investigated the CO 2 reduction process on the Cu surface by using quantum molecular dynamics with six layers of explicit water molecules (Figure 9c). [27,60] The results show that the hydroxylmethyl group (CHOH) adsorbed on Cu(100) surface is the key intermediate that determines the selectivity of methane over methanol. [27] During CO formation, the physically adsorbed CO 2 to chemisorbed *CO 2 are rate-determining step, in which the chemisorbed *CO 2 is usually less stable in the CHE vacuum model. [60a] This method also simulates the effect of different applied potentials, where ethylene is the main product when the potential is less than À0.6 V (RHE). When the applied potential is less than À0.60 V (RHE), some *CO surface sites will be replaced by H*, resulting in a decrease in the ethylene formation rate. [60b] With the development of advanced spectroscopy technology, the dynamic behavior of the electrochemical interface can be understood by observing the peaks at the water/metal interface at the atomic scale. The vibration spectra calculated by AIMD can be compared with experiments to verify the accuracy of the calculation. Nevertheless, it needs to be acknowledged that far less is known about interface behavior. In addition to spectroscopy, additional experimental techniques need to be developed to understand the electronic properties of interfaces and the kinetic behavior of CO 2 RR. More reliable atomic-level modeling of EDL is also a major challenge for theoretical calculations.

Cations
In CO 2 RR, cations are important components in the EDL. Under the external negative potential, the negative charge on the surface of the metal electrode will attract the cations in the electrolyte to gather above the electrode, which forms the OHP plane. Figure 9. The explicit solvent model with different levels of water coverage on the catalyst surface. a) Schematic diagram of a model with one water molecule. Reproduced with permission. [58] Copyright 2013, Wiley-VCH GmbH. b) The model with multiple water molecules. Reproduced with permission. [59] Copyright 2018, American Chemical Society. c) The explicit solvent model with a six-layer water molecule. Reproduced with permission. [27] Copyright 2018, American Chemical Society. The specific location and distribution of cations in EDL are determined by many factors, including the size of the cations, the degree of hydration, the applied potential, and the specific adsorbent. In general, the direct effects of the experimental cation effect are reflected in the activity aspects such as Faraday efficiency and current density. Theoretical calculations can further explore the deeper effects of cations on the EDL, such as the effect of cations on the applied potential and the interfacial electric field.

Size Effect
The size of the cation has a significant effect on the activity and selectivity of CO 2 RR, and many studies have shown that Ag, Hg, Cu, and Sn are subject to this effect. [61] Recently, Singh and coworkers investigated the effect of five alkali metal cations (Li þ , Na þ , K þ , Rb þ , and Cs þ ) on the activity and selectivity of Cu and Ag (Figure 10a-d). [62] Interestingly, for both Cu and Ag, increasing the size of cations can enhance the current density and Faraday efficiency. At the same potential, the current density increases by a factor of 2.4 for Cu and 2.1 for Ag from Li þ to Cs þ . The Faraday efficiency increases by %15% for Cu and %55% for Ag. The source of activity can be attributed to the cationic effect on the local pH effects. As the cation size increases, the pKa of cation hydrolysis decreases leading to a decrease in the local pH near the cathode. This increases the local solubility of CO 2 , which affects the activity and selectivity of CO 2 RR. Ayemoba and coworkers probed the pH at the electrochemical interface using in situ attenuated total reflection surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS) and demonstrated that the large size of the cation reduced the local pH at the electrode surface. [63] The results confirm that the increase in pH at the interface follows a trend of Li þ > Na þ > K þ > Cs þ , which is consistent with the results of Singh et al. [62] The effects of cations on activity and selectivity may be derived from their regulatory effects on CO 2 RR intermediates. It is generally believed that the catalyst is the decisive factor affecting the energy barrier of CO 2 RR reaction, and the role of cation in regulating the energy of intermediate formation reaction is often ignored. Resasco and coworkers investigated the influence of CO 2 RR intermediates on Cu, Ag, and Sn surfaces by the size of the electrolyte cations. [61d] They found that HCOO À , C 2 H 4 , and C 2 H 5 OH were significantly enhanced by large cations on Cu(100) and Cu(111). CO and HCOO À on Ag (Sn) were similarly affected by the cation size. The cations have little effect on the production of H 2 . Furthermore, the enhancement effect of cations has been attributed to the fact that large-size cations can stabilize the adsorption of intermediates with significant dipole moments such as *CO 2 , *CO, and *OCCO.
In conclusion, the enhancement of the activity of CO 2 RR by large-size cations is extremely significant. Although the experimental results are very consistent, there are various explanations for them. The specific reasons for the cation effect deserve further experimental exploration.

Surface Electric Field
The cations in the electrolyte will accumulate on the OHP with the enhancement of the applied potential. These cations will generate charge transfer with the electrode surface to form the built-in electric field, which is the basic characteristic of EDL. Combining the in situ Raman and AIMD methods, Figure 10. Effect of cations on the current density and Faraday efficiency of CO 2 RR. a) The effect of cations on the current density of the Ag electrode. b) the Faradaic efficiency of CO and H 2 produced by the Ag electrode. c) The effect of cations on the current density of the Cu electrode. d) The Faradaic efficiency of CO 2 reduction products on Cu electrode. Reproduced with permission. [62] Copyright 2016, American Chemical Society.
www.advancedsciencenews.com www.advenergysustres.com Wang and coworkers regulated the number of cations in the OHP to simulate the effect of the electric field generated between the cations and the electrode surface on the orientation of water molecules ( Figure 11). [64] They found that the electric field existing between the metal surface and the electrolyte solution is caused by Na þ ions acting as counter ions to compensate for the negatively charged electrode surface. The spectra of the interface water detected by in situ Raman are in good agreement with the results calculated by AIMD (Figure 11a,b). The dynamic transition of water from disordered to ordered can be observed by adjusting the number of cations (Figure 11c). The ordered water results in promoting efficient electron transfer at the interface, thus increasing the electrochemical reaction rate. Through DFT calculations and microdynamic modeling, Chen and coworkers simulated the electric field distribution of cations approaching CO 2 molecules (Figure 12). [65] The addition of cations will enhance the electric field strength on the electrode surface. In the absence of cations, CO 2 adsorption on the Ag surface is difficult, and the addition of these cations makes CO 2 adsorption easy to the extent that *COOH ! *CO becomes a potentiation limiting step. In addition, the double-layer capacitance in the EDL is also affected by the cation. Ringe and coworkers combined impedance spectroscopy and AIMD simulations and found that the EDL capacitance increases as the cation size increases. This means that the strength of the EDL electric field from different cations can also vary considerably.
The combination of experimental and theoretical calculations has led to a gradual deepening of our understanding of cations from size to local electric fields. This also advances the understanding of the EDL structure at electrochemical interfaces. Nevertheless, some quantitative means to measure and simulate the electric field strength generated by cations in EDL are still missing. This is the goal that needs further work.

Tuning Strategy
The local ion concentration near the EDL is closely related to the surface shape of the catalyst. [66] Liu and coworkers proposed that Au nanoneedles can aggregate K þ to form a local electric field, thus enhancing the activity of CO 2 RR ( Figure 13). [66a] Comparing Figure 11. Combined in situ Raman and AIMD probing of electrochemical interfaces. a) Schematic of in situ Raman device. b) 3D-finite-difference time-domain (3D-FDTD) simulation of the electromagnetic field distribution. c) Electrochemical interfaces simulated by AIMD. Reproduced with permission. [64] Copyright 2021, Springer Nature.  (Figure 13c), while the electric field of nanoparticles is the smallest. The enhanced local electric field has also been confirmed to increase the concentration of K þ at the tip (Figure 13d,h, l). The aggregation of K þ can effectively activate CO 2 , thus improving the current density and Faraday efficiency of CO 2 RR. Furthermore, Liu et al. investigated the degree of ordering of nanoneedles and the effect of curvature on local cation concentration and reactivity. They fabricated vertically ordered Cu needle arrays and randomly disordered Cu needle arrays by template-assisted electrodeposition. [4a] The K þ ion adsorption experiment showed that the adsorption capacity of the ordered array for K þ was 30 times higher than that of the disordered array. The Faradaic efficiency of C 2 production of ordered Cu nanoneedle arrays was much higher than that of disordered Cu nanoneedle. These results suggest that ordered Cu nanoneedle arrays can effectively aggregate cations to enhance the reduction activity of CO 2 RR. [4a] In addition, they prepared ordered Ag nanoneedle arrays with different tip curvatures by vacuum thermal evaporation. [67] The results show that the higher the curvature, the stronger the aggregation effect of the tip, which leads to a higher C 2 product activity. This series of studies shows that the catalyst surface shape greatly affects the distribution of cations in the electrolyte, which in turn affects the CO 2 RR reaction.
Overall, cations are an essential part of the electrolyte and an important component of the EDL, which plays a key role in tuning the local reaction activity for the CO 2 RR. Modulating the concentration of cations near the catalyst affects both the electric field strength of the EDL and the intermediate process of the CO 2 RR reaction. The tip effect can be used to efficiently aggregate cations, which is expected to be a general strategy to optimize the catalyst activity.

Other Parts
In addition to adsorbates, water/metal interfaces, and cations, EDL also contains some other parts, such as anions and differential capacitance. They are relatively less dominant than the above three, but they also have some influence on the structure and properties of EDL.

Anions
Unlike cations which mainly gather in the OHP, anions are more dispersed and generally adsorb on the electrode surface or interact with other substances in the electrolyte. This means that anions can occur in all parts of the double layer, including on the electrode surface, near the OHP, and in the diffusion layer. The most common electrolyte used in CO 2 RR is an aqueous bicarbonate solution. The most direct role of HCO 3 À is as a Figure 13. a,e,i) Scanning electron microscope images of Au needles, rods, and particles; b,f,j) Transmission electron microscope images; c,g,k) Kelvin probe atomic force microscope image of the electric field distribution; d,h,l) SEM and K þ adsorption test images. Reproduced with permission. [66a] Copyright 2016, Springer Nature.
www.advancedsciencenews.com www.advenergysustres.com proton donor and source of carbon for the CO 2 RR reaction. [68] By in situ spectroscopy, isotope labeling, and mass spectrometry, it has been demonstrated that most of the carbon sources for the CO 2 RR reaction are derived from the dynamic equilibrium between CO 2 and bicarbonate. [68,69] Wuttig and coworkers developed a microdynamic model and compared it with the zero-order dependence in bicarbonate observed experimentally and found that bicarbonate can act as a potential proton donor to participate in CO 2 RR. Therefore, an effective strategy to improve the CO 2 RR rate is to increase the solubility of bicarbonate, which can increase the local solubility of CO 2 on the electrode surface due to the rapid equilibrium between bicarbonate and CO 2 . Halogen ions can be specifically adsorbed on the electrode surface as part of the IHP to modify the catalyst activity. Hsieh and coworkers prepared a high surface area silver nanocoral catalyst in the aqueous medium in the presence of chloride ions. [70] This catalyst has a 32-fold increase in activity at an overpotential of 0.49 V compared to silver foil for the reduction from CO 2 to CO. They attributed the enhanced activity to the adsorbed Cl À on the electrode surface. Hong et al. explored the effect of different anions on the activity of Au catalysts for the reduction of CO 2 . The Cl À could significantly improve the Faraday efficiency of Au. [71] Br À , I À , and F À have also been reported to significantly modify the catalyst surface activity. Varela prepared a Cu catalyst and added halogen ions Cl À , Br À , and I À on the surface. [72] They found that Cl À had little effect, Br À could significantly improve CO selectivity, and I À could increase methane formation rate. Besides, F À has been reported to modify the Cu surface and improve the Faradaic efficiency and current density of C 2 products. [73] The reason for the improved efficiency of C 2 product formation is thought to be that F À can improve C─C coupling and enhance water activation as well as *CO and *CHO adsorption. Through DFT calculations, Yoon and coworkers provided a theoretical explanation for the effect of different anions and cations on CO 2 reduction. [74] They calculated the adsorption energies and work functions of Cl À , Br À , and I À on Bi catalysts. The results reveal that the adsorption energies and work functions increased with decreasing anion size. The calculation of free energy also shows that different halogen ions have an important influence on the selectivity of CO 2 reduction.
Although both bicarbonate and halide ions have been reported to influence the reactivity and selectivity of CO 2 RR. However, it is still difficult to distinguish the exact role they play. For example, it is difficult to determine whether the rapid equilibrium between bicarbonate and CO 2 occurs in the diffusion layer or the IHP. The detailed process of the influence of adsorbed halide ions on the reaction intermediates needs to be urgently investigated.

Differential Capacitance
The differential capacitance characterizes the ability of EDL to store charge when the electrode potential changes slightly. The bell-shape or camel-shape curve measured in electrochemical impedance spectroscopy is a very well-known phenomenon experimentally concerning the EDL differential capacitance. Recently, Kim et al. combined the DFT of the explicit solvent model with the classical molecular dynamics of the electrolyte to simulate the camel-shape differential capacitance curve appearing on the surface of the Ag(111) electrode ( Figure 14). [75] To reliably calculate the differential capacitance, they defined the calculated value of the differential capacitance as C ¼ dσ=dE (σ is the surface charge density and E is the applied electrode potential), and the MD method is used to sample the data finely to obtain a continuous capacitance curve. Simulating the change of applied potential is realized by introducing excess electrons or ions into the electrolyte for Ag(111) electrode. They found that the EDL capacitance curves predicted by this method closely matched those corresponding to the staircase potentiostatic electrochemical impedance spectroscopy data for Ag(111) measured in a dilute 3 mM NaF electrolyte, and the camel-shape curve was successfully realized.
Abareghi et al. explored the correlation between the wall curvature and its concavity in the capacitance curve and the surface charge density using a modified fundamental measure theory. They believe that the intersection of the convex wall at the differential capacitance curve in EDL appears at a concentration greater than that of the concave wall. [76] The effects of the finite size of ions and dielectric attenuation of differential capacitance on differential capacitance in EDL were systematically investigated by Andelman et al. [77] At high surface charge densities, both mechanisms may lead to an increase in the concentration of ions with opposite charges near the charged surface. By Monte Carlo simulation of the differential capacitance curves of ionic liquids, Kornyshev concluded that the camel-shape curve is generated by the neutral counterpart of the electrolyte (EDL in ionic liquids: The nature of the camel shape of capacitance). In addition, they also investigated the temperature dependence of the differential capacitance in EDL by molecular dynamics method. [78] The results show that the differential capacitance curve changes from camel-shape to bell-shape with the increase in temperature.
The bell-shape or camel-shape differential capacitance curve is a very interesting phenomenon in the experiment. It is helpful to understand the charge characteristics of EDL theoretically to reproduce the curve. However, because it is difficult to simulate the electrode potential in the theoretical calculation, it is still difficult to accurately simulate the differential capacitance curve to be consistent with the experiment quantitatively.

Summary and Outlook
In this review, we review the recent advances in EDL structures, including adsorption intermediates on IHP, water/metal interface structures, cations, anions, and differential capacitors. Overall, advances in advanced spectroscopy and computational performance have led to a more comprehensive understanding of the various parts of the EDL structure. However, there are still many critical issues that need to be resolved.
The kinetic processes of proton transport during CO 2 reduction have been too little studied. At present, the information on adsorbed intermediates obtained experimentally is mainly based on the infrared spectrum. Most of the theoretical calculations are based on the CHE model to calculate the reaction-free energy. There is no unified model for the latest AIMD method. These methods mainly focus on identifying reaction intermediates and determining reaction-free energies. Few studies have been conducted at the atomic scale for CO 2 RR proton transport dynamics.
Obtain more information about the basic properties of liquid water layers on metal surfaces for experimental and theoretical comparisons. The properties of the liquid water layer at the metal interface are very important for understanding the EDL structure, but only the vibration spectra calculated by AIMD can be compared with the experimental results. More computational methods and experimental techniques need to be developed to understand the mechanism of CO 2 RR reaction in the electrochemical field.
A more intuitive understanding of the influence of cations in CO 2 reduction is needed. Cations are generally distributed near OHP in EDL, but how cations respond to electrode potential and the dynamic effects of cations on PCET processes need to be continued to be investigated.
The quantitative simulation and experimental measurement of the electric field strength between the electrode surface and the OHP is a very challenging and valuable study to be explored. The intermediates of the CO 2 RR reaction are polar molecules, which are easily affected by electric field polarization. Measuring and simulating the electric field strength of the EDL provides insight into the CO 2 RR mechanism. In conclusion, with the development of advanced experimental techniques and improved computational performance, the understanding of the EDL structure will become deeper, which will also help to better design excellent CO 2 RR catalysts.