1. Introduction

Electrochemical conversion of air-abundant molecules (e.g., CO₂, N₂, and O₂) into value-added products is recognized as a feasible pathway to close the carbon loop while utilizing intermittent renewable energy [1-5]. Recent progress in heterogeneous catalyst design has achieved a deep understanding on the catalytic mechanism and the close relationship between catalyst structure and catalytic performance [6-8]. As with the heterogeneous electroreduction processes of CO₂, N₂, and O₂, sufficient reactants and favorable microenvironments are indispensable for efficient energy conversion under large working current density [9-11]. However, under industrially-level current density for practical application, the dramatic change of cation, solvent, pH, and reactant molecular distribution near electrode would limit the reaction kinetics.

Particularly, the low dissolubility of the air-abundant molecules in aqueous electrolyte, e.g., 34 × 10⁻³ mol/L for CO₂, 1.2 × 10⁻³ mol/L for O₂, and 0.6 × 10⁻³ mol/L for N₂ at ambient temperature and pressure [12-14], which is calculated to support low net current density < 35 mA/cm² for CO₂ to CO, 0.15 mA/cm² for N₂ to NH₃, and 5 mA/cm² for O₂ to H₂O, respectively [15]. In this case, severe concentration polarization occurs, which can compromise the energy reversion efficiency [15]. Besides, the competitive protons carried by water reach over 55 mol/L, 1.6 × 10³, 4.6 × 10⁴, and 9.2 × 10⁴-fold higher than CO₂, O₂, and N₂ at catalyst surface immersed in electrolyte. Therefore, the current density is restricted by not only the gas transfer dependence in aqueous phase but also the dominating hydrogen evolution disruption, far behind the commercially-relevant conditions (> 200 mA/cm²). Various architectures for electrolyzer and cell design are inspiring new types of gas diffusion layers and devices, including flow cells, solid-state electrolyte (SSE) electrolyzers and microfluidic electrolyzers. These new architectures can facilitate gas diffusion and strengthen the exposure of catalyst to the gas phase instead of electrolyte, differing from traditional aqueous-fed systems. In microcosmic view, mass transfer at specific catalyst surfaces and across the electric double layer (EDL) can affect local concentration of gas reactants, which deserves comprehensive understanding and rational design at molecular level to improve the reaction microenvironments and catalytic kinetics. Therefore, understanding the intricacies of mass transport at the catalyst/electrolyte interface serve as the key premise to realize globally-relevant scales (e.g., > 1 GW for CO₂ conversion to CO [15], 1.6 GW for nitrogen fixation into ammonia [16], and ∼10 GW for O₂ conversion to water in proton exchange membrane fuel cell [17]) towards efficient air molecule fixation. Breaking the solubility limitation and balancing mass transfer between reactant and electrolyte for boosting reaction current become a non-negligible but challenging issue in heterogeneous electrocatalytic fixation of air-abundant molecules.

Here, we first fundamentally discussed the involved steps of gas transport from bulk electrolyte to catalyst surface and within EDL under operating conditions. Next, we reviewed the optimizing strategy of mass transfer on the catalytic surface. Then, we systematically compared the gas-fed pathways in different cell architectures and present recent advances in regulating mass transport behaviors. In each part, critical factors are provided that affect mass transfer and electrocatalytic performance for electrochemical air molecule fixation. Finally, we proposed several innovative design strategies that may break the intrinsic gas solubility limitation by altering electrochemical gas transport pathway. We believe these thorough mechanism discussions would be helpful to identify potential bottlenecks and limitations within current systems, paving the way for innovative solutions and thereby advancing the clean energy and environmental sustainability.

2. Mass transfer from bulk electrolyte to electrolyte/catalyst interface and optimization strategies 2.1. Mass transfer in diffuse boundary region

Heterogeneous electrocatalytic CO₂ reduction reaction (CO₂RR), N₂ reduction reaction (NRR), and O₂ reduction reaction (ORR) mainly take place at the interface between electrode and electrolyte, within the diffuse boundary region. The movement of chemical species (ions or molecules) from the bulk electrolyte toward the electrode interface is an inevitable process in a liquid-fed system. The H-type electrolyzer (H-cell) is a typical liquid-fed system that is widely used in fundamental electrochemistry research for evaluating the intrinsic activity and stability of catalysts. A typical H-cell is a three-electrode system composed of two separate compartments with non-flowing electrolyte separated by anion or cation exchange membrane, where the catalyst is immersed in liquid media, generating a pseudo-heterogeneous nature of the reaction process. That is, the reacting molecules are all derived from the dissolved gas rather than directly sourced from the gas phase. Therefore, optimizing the mass transport process of the catalytic system is essential for a relatively high current density.

In general, mass transfer includes the transport of ions, intermediates, and reactant molecules. Among these species, the mass transfer of reactant molecules has the most critical impact on current density, while the mass transfer of ions and intermediates has potential and supportive effects on reaction selectivity. There are several pathways of mass transfer from the bulk electrolyte to the interface, including diffusion, convection, and electromigration [18]. Since CO₂, N₂, and O₂ are nonpolar molecules, their transport can mainly be affected by concentration gradient, fluid flow, and diffusion coefficient of molecule in certain electrolyte (Fig. 1). A model presented by Smith et al. [15] systematically describes the mass transfer of CO₂ molecule during CO₂RR operating conditions in an H-cell configuration. It is generally assumed in experimentation that the bulk electrolyte is continuously saturated with CO₂, which can be replenished from the bulk electrolyte as consumed at the cathode. The maximum CO₂ reduction current (i _max) under kinetic steady state conditions can be calculated by the Fick's law [19] as below:

$ i_{\max }=n F A D \frac{c^*-c}{\delta} $

(1)

where n, F, A, D, c*, c, and δ stand for electron transfer number, Faraday's constant, electrode surface area, diffusion coefficient, bulk electrolyte concentration, local concentration, and diffusion layer thickness. Apparently, the i _max that can be maintained are negatively related to the thickness of the diffusive region, which is influenced by fluid flow.

In an H-cell, continuous mechanical stirring is the most common strategy to weaken the fluid boundary range. Calculations and experimental works have found that a higher stirring rate serves to bring the interfacial pH closer to that in the bulk solution and boost the flux of dissolved CO₂ towards the electrode [20,21]. However, mechanical stirring is very limited to weaken the fluid boundary layer (∼µm) and often neglected for a constant stirring rate [22]. Also, these results are applicable to ORR and NRR in H-cell. Furthermore, magnetic field inducing is an another strategy to improve electrolyte convection towards the electrode surface by reducing concentration polarization and thinning the diffusion layer [23]. Bhargava et al. [24] harvested the magnetohydrodynamic (MHD) effect to facilitate the mass transfer of electroactive species, thus leading to an enhanced full cell energy efficiency for CO₂ electrolysis. Pan et al. [25] found that the magnetic field can further assist the diffusion of HCO₃ ⁻ and accelerate its migration to the electrode surface, thus enhancing the local CO₂ concentration. Moreover, the MHD effect can also release the supersaturated bubbles accumulated on the electrode surface by stirring the electrolyte, accelerating the transfer of gas molecule directly on the surface. However, it should be noted that these methods are nonspecific in nature, meaning that they rather affect the overall fluid dynamics instead of target specific ions or molecules.

Obviously, increasing concentration gradient and diffusion coefficient of the solution is essential for enhancing mass transfer in an H-cell configuration in steady state. The former requires considerable solubility of gas in electrolyte, which can be enhanced by increasing the reaction pressure, lowering the electrolyte temperature, and introducing aerophilic solvent. Though pressurized electrolytes have been reported to effectively enhance activity and product selectivity [26], the requirement for the non-ambient pressure and temperature imposes higher demands and challenges on the device complexity and energy input. For the diffusion coefficient, it represents the intrinsic transport ability of gas in the aqueous media which can be increased by elevating temperature or introducing alternative/secondary solvents with low viscosity and small size/molar mass.

Above all, one of the most effective ways to enhance the liquid-fed mass transfer in bulk electrolyte is selecting the appropriate solvents that feature low viscosity, small size/molar mass, and high gas affinity. Among vast kinds of organic molecules, alcohols (R-OHs) are promising solvents due to their low cost, low toxicity, and higher solubility of nonpolar gas. The solubility of CO₂ in methanol is about four times and more than eight times that in water at room temperature and at temperatures below 273 K, respectively [27]. The solubility of O₂ and N₂ in methanol is about 17 and over 14 times that in water at room temperature, respectively [28,29]. In this respect, Ohta et al. [30] studied CO₂RR in methanol with a Cu electrode and found relatively high Faradaic efficiency (FE) for the methane synthesis up to 62%. Methanol has been reported as one of the best non-aqueous electrolytes for obtaining hydrocarbons [31]. Besides CO₂, Kalu et al. [32] found the presence of methanol promotes O₂ dissolution and diffusion in electrolyte, where the ORR test demonstrated that the diffusion-limited current increased over 1.6 times in a 16.0 mol/L methanol containing electrolyte [33]. By introducing methanol into the electrolyte for NRR, the FeOOH-based catalyst exhibits high NH₃ yield rate of 262.5 ± 7.3 µg h^–1 mg_cat. ^–1 with a remarkable selectivity of 75.9% ± 4.1%, indicative of ∼8-fold enhancement compared with that in pure aqueous electrolytes [34]. Besides methanol, many other alcohols have also been used as co-solvents in electrolytes. Krishnamurthy et al. [35] studied the effects of different alcohols including linear aliphatic alcohols, branched aliphatic alcohols, aliphatic polyalcohols and halogenated alcohols on lithium-mediated NRR (LM-NRR). The results showed that LM-NRR exhibited higher Faradaic efficiency when most alcohols were used as co-solvents, with 1-butanol having the most significant effect.

Besides, other non-aqueous solvents that possess high dissolubility of gas for facilitating reduction of air-abundant molecules have been reported, such as methyl cyanide (MeCN) [36,37], dimethyl sulfoxide (DMSO) [38], dimethylformamide (DMF) [39], propylene carbonate (PrC) [40], tetrahydrofuran (THF) [41], and acetonitrile [42,43]. Notably, these solvents can not only concentrate gas in bulk electrolyte for enhanced gas transfer but also serve as inactive proton donor for suppressed hydrogen evolution reaction (HER). Additionally, several non-aqueous solvents can co-react with the gas reactant or formed intermediate to produce higher value-added products [44].

In addition to physical dissolution, the gas dissolution can also be boosted by the interaction with the special ions in ionic liquid. Meanwhile, the interaction can promote the electromigration of the nonpolar molecule. Most ionic liquids possess adjustable structure, high thermal stability, good conductivity, wide electrochemical window, and nearly zero vapor pressure, making them an ideal choice to accelerate air-abundant molecule reduction [27]. Moreover, ionic liquids are found to be capable of decreasing overpotential, stabilizing reaction intermediates, and enhancing the final product selectivity [45,46]. Rosen et al. [47] first reported an 18 mol/L 1-ethyl-3-methylimidazolium tetra-fluoroborate (EMIM-BF₄) electrolyte with a balanced EMIM−CO₂ interaction, which was sufficient to facilitate CO₂ reduction for the synergetic transfer of CO₂ and ions. This system attained a ∼96% of CO selectivity. Then EMIM-BF₄ was widely used in many works for CO₂RR considering its merits that it not only possesses remarkable CO₂ transport ability, but also helps to decrease the overpotential for production of the carbon dioxide radical anion intermediate (CO₂ ⁻˙). In NRR system, Zhou et al. [48] selected a series of ionic liquids that possess high solubility for N₂. Based on DFT calculations, they found [P_6,6,6,14] [eFAP] can facilitate N₂ activation due to the low binding energy and high affinity of N₂ and [eFAP]⁻ ion. Through lowering the water concentration into ppm level, the ionic liquids helped to attain a 60% ± 6% of N₂ conversion selectivity.

2.2. Mass transfer within the EDL

It is worth noting that in addition to bulk electrolyte, EDL that possesses dense charge concentration and broad electric field distribution also plays an important role in mass transfer and charge balance [49,50]. In recent years, numerous studies have shown a strong correlation between the EDL structure and electrocatalytic activity [51,52], but most work focuses on the water electrocatalysis and electron transfer instead of mass transfer issues in EDL. There are some evidences implying that mass transfer may follow unconventional mechanisms at EDL interface in nanoscale [53,54]. Specifically, the Debye length (represents the scale of the EDL) varies with different concentration of electrolytes and size of catalysts, thus adjusting the transport distance [55] and inducing extra mass transfer effect (e.g., local electric field induced mass transfer, LMT) [54]. Since the unique property of EDL, the reaction environment and mass transfer behavior are totally different from bulk electrolyte. Koper et al. [56] proposed a pH-dependent kinetics where the reorganization of interfacial water and specific adsorption of hydroxyl species (OH_ads) can affect proton and hydroxide through the EDL.

Recently, Lu et al. [57] established a bistable model to uncover the relationship between EDL rigidity and CO molecule mass transfer. Experimental evidence was provided that EDL is in a dynamic state. The rigidity and structure of EDL can be tuned by densely packed hydrated cations or less compact cations, resulting a kinetic trapping phenomena and mass transfer behavior change for CO molecule. Besides cations, Du et al. [58] found co-occurrence of alkali ions and halide ions (F⁻, Cl⁻, Br⁻, and I⁻) in EDL can enhance the adsorption of CO₂ compared to that in pure water by using ab initio molecular dynamics. Different types of halides can form hydration shells of different sizes within EDL, thus affecting the coordination number of water molecules in the hydration shell and the differential charge density between the ions and the copper surface. The halide ions were found to create localized charges at the solid-liquid interface that promotes the subsequent adsorption and activation of CO₂. Experimentally, Ren et al. [59] conducted a well-controlled oscillating potential ORR, where they found that the dynamic adsorption and desorption microenvironment can enhanced mass transport of O₂ molecule. The dynamic EDL structure under potential oscillation can increase the selectivity of H₂O₂ by 1.35 times compared with the static potential conditions.

Based on the emerging works, we can envision that the capability of probing the solvation environment and reorganization properties of EDL may uncover the speed of interfacial mass transfer, which can further provide ample research opportunities regarding modulation of electrochemical performance for fixation of air-abundant molecules. Nonetheless, studies on EDL can still be challenging due to the intricate interplay between electrostatic forces, molecular interactions, and the behavior of ions at interfaces, which necessitates accurate modeling and advanced experimental techniques.

3. Optimizing mass transfer on the catalytic surface

Surface design is a common strategy to adjust the intrinsic catalytic behavior of a catalyst [60-62]. Changing the surface composition, facet and electro-structure can optimize the binding energies of gas molecules and key intermediates on the surface, thus favoring the target reactions [63,64]. Besides, the specific shape and composition of the catalytic surface can also affect mass transfer process. According to certain physicochemical nature, novel-shaped catalytic surfaces were designed, including cavity structure [65,66], high-curvature structure [67-69], stepped-surface structure [70,71], and channel structure [72-75]. Additionally, increasing the hydrophobicity of the catalyst surface can also enhance the mass transfer of gas molecules, including loading a hydrophobic layer on the catalyst [76,77] and synthesizing intrinsically hydrophobic catalysts [78-80].

3.1. Cavity-structure catalyst

The internal space of the cavity-structure catalyst can generate a spatial confinement effect which enriches as-formed intermediates and thereby increases reaction kinetics (Fig. 2a). In CO₂ system, the confinement of key intermediates by the cavity structure can also favor the production of multi-carbon products [81]. For instance, Yang et al. [82] designed a Cu₂O catalyst with nanocavities, which achieved a C₂₊ selectivity of 75.2% ± 2.7% with a partial current density of 267 ± 13 mA/cm², where the ratio of C₂₊ to C₁ product over multi-hollow particle reaches 7.2, which is 9 times higher than that of solid structures. According to simulation results, the increased numbers and sizes of cavity also lead to the enhanced C₂₊/C₁ ratio. Zhuang et al. [83] developed an open Cu nanocavity structure with a tunable geometry for nanoconfinement of C₂ intermediates, which further promoted C₂ and C₁ coupling to form C₃ products. The cavity structure ultimately achieving 21% ± 1% of propanol selectivity with a current density of 7.8 ± 0.5 mA/cm². Cavity-structure catalysts are used for efficient ORR and NRR. Nazemi et al. [84] prepared hollow Au nanocages (AuHNCs) with different pore size/density and achieved a NH₃ selectivity of 35.9% with a NH₃ yield rate of 3.74 µg cm^–2 h^–1. Cheng et al. [85] fabricated N-doped carbon porous microspherical cavity for ORR and it exhibited a high onset potential of ∼0.92 V versus RHE, a remarkable specific activity (SA) of 213.4 µA/cm², and a good mass activity (MA) of 162.88 mA/µg.

3.2. High-curvature structure catalyst

High-curvature nanostructures have been identified to be electroactive regions since it can induce tenfold or even higher increase in electric fields (Fig. 2b), which is more susceptible to concentrate electrons, cations, and even reactants, thus remarkably boosting the activation and transfer of the gas molecules. Liu et al. [86] first reported Au needles for field-induced electrochemical CO₂RR, which exhibited 90% of CO selectivity at −0.35 V versus RHE, far higher than that of Au rods (3%) and Au particles (0%). Similarly, Gao et al. [87] prepared cadmium sulfide (CdS) nanoneedle arrays and achieved 95.5% Faradaic efficiency of CO with a current density of 212 mA/cm². Moreover, they found the gap between needles can also affect the concentration of the reactants because of the proximity effect. The unique gas inducing and transferring mechanisms also enlightened the catalyst development in NRR [68,88,89] and ORR systems [69].

3.3. Stepped-surface structure catalyst

In the field of electrochemical fixation for air-abundant molecules, facet engineering has been used to screen and identify catalyst facets with high intrinsic catalytic activity (e.g., Pt (111) for ORR to water [90], Cu (100) for CO₂RR to C₂H₄ [91]). The stepped surface has been verified to be highly active towards air-abundant molecules fixation (Fig. 2c) [92]. Choi et al. [70] reported a stepped surface [3(100) × (111)] of Cu exhibiting a high local population of 2CO* and a higher barrier for the C₁ path compared with that for Cu(100). Luo et al. [71] synthesized Cu fibers whose surface is stepped structure with preferred (111) facet and achieved 71.1% ± 3.1% Faradaic efficiency of formate.

Bao et al. [93] designed gold nanorods with stepped (730) facet which is composed of (210) and (310) facets for NRR. They achieved yields of 1.648 µg h⁻¹ cm⁻² for NH₃ and 0.102 µg h⁻¹ cm⁻² for N₂H₄·H₂O. Feliu et al. [94] evaluated the catalytic performance of Pt(S)[n(111) × (111)] (20 ≤ n ≤ 50) for ORR. They found that stepped surface can enhance the catalytic activity of Pt for ORR in acid medium. Similar effect was also found on grain boundary structure. Huang et al. [95] demonstrated that Cu₂O nanoparticles with a combination of both (100) and (111) facets has a better Faradaic efficiency for C₂H₄ than Cu₂O nanoparticles with only (100) facet or (111) facet in CO₂RR system. All these mechanisms were discussed based on thermodynamics, where the facet effect possesses better adsorption energy for the key intermediates. However, from the dynamic perspective of the gas reduction on the stepped surfaces, it can be inferred that the steps on the rough surface can also affect the mass transfer of reactants and subsequent formation of key intermediates, which is worth investigating.

3.4. Channel-structure catalyst

Channel structures are widely adopted in catalyst design of the fixation of air-abundant molecules, including layered, pipe-like, and other open-framework structures [72,96-98]. The main aim to construct the channel structures is to broaden the mass transfer mode of the gas molecules (Fig. 2d). The long channels can enrich and trap gas molecules in a highly induced and active region during the reaction. Strasser et al. [97] designed 2-dimensional (2D) Cu(Ⅱ) oxide nanosheet (CuO NS) catalysts for CO₂RR and achieved a high C₂₊ selectivity with a current density above 400 mA/cm². Yu et al. [72] prepared TiO₂ nanotubes for NRR and achieved Faradaic efficiency for NH₃ of 26% with a yield rate of 5.50 µg h⁻¹ cm⁻².

3.5. Hydrophobic layer loaded catalyst

Hydrophobic layer loading is an efficient way to change mass transfer behavior (Fig. 3a) [76,77] and some hydrophobic layer even possess functional groups affinitive to the gas molecules. Generally, the mass transfer channel is situated at the interlayer between the hydrophobic layer and catalyst layer. In an H-cell, the hydrophobic layer is exposed to the aqueous phase, protecting the catalyst from fully wetting and accelerating the gas transfer. In flow cell, the hydrophobic layer is exposed to gas phase or both side of catalyst for the fast gas transfer and high gas concentration, where the refreshing half-wetted thin film between the hydrophobic and catalyst layer can break the aqueous-fed limitations. Daasbjerg et al. [77] loaded different hydrophobic and hydrophilic polymers on CuO to explore their effects on CO₂RR. The results showed that the hydrophobic polymers can improve ethylene selectivity, while the hydrophilic polymers promoted HER. Specifically, the Faradaic efficiency and the partial current density of C₂H₄ over CuO loaded with hydrophobic poly(vinylidene fluoride) (PVDF) is twice and three times that of bare CuO, respectively. Du et al. [99] reported self-assembled monolayer of hexanethiol (HEX) on metal electrocatalysts such as Cu, Au, Pt, Pd and Ni to create a hydrophobic microenvironment for NRR. Among them, Cu-HEX exhibited a high NH₃ selectivity of 35.9% with a NH₃ formation rate of 1.2 µg h⁻¹ cm⁻². Tang et al. [100] used hydrophobic PTFE-coated Co₃O₄ nanosheets to enhance ORR on air cathode in zinc–air batteries (ZABs), which achieved a higher power density of 171 mW/cm² for ZABs.

3.6. Intrinsically hydrophobic catalyst

Notably, the addition layer may cause severe impedance and overpotential increase, which compromises energy efficiency. Moreover, electrolyte flooding caused by the decreased hydrophobicity of hydrophobic layer during long-term operation would reduce the stability of carbon-based gas diffusion layer electrode. The development of the intrinsically hydrophobic catalysts can solve these problems [78-80]. Gao et al. [79] developed a biomimetic copper dendrite catalyst with a hydrophobic copper hierarchical structure (Fig. 3b). This catalyst showed a 64% selectivity of C₂₊ products and can operate stably for 45 h at current density of 300 mA/cm². Lee et al. [78] prepared different morphologies of gold nanostructures with different hydrophobicity by electrodeposition at different potentials. The roughest gold has 6 times higher Faradaic efficiency and 327 times higher partial current density than smooth gold due to having the most hydrophobic features. Moreover, rough gold demonstrates outstanding CO₂ capture capabilities when exposed to low CO₂ concentrations. The authors inferred that the inherent hydrophobicity of the catalyst created an ideal gas-liquid-solid triple-phase interfaces, enhancing mass transfer of CO₂ and elevating the local concentration of CO₂ near the catalyst. Xu et al. [101] reported that N-doped phosphorene nanosheets with high surface hydrophobicity due to doping of N and O can achieved an excellent yield rate of 18.79 µg h⁻¹ mg_CAT ⁻¹ and Faradaic efficiency of 21.51% for NH₃. Yan et al. [102] designed intrinsically hydrophobic CoS/Fe₃S₄ nanoparticles grown on S, N co-doped carbon plate arrays with superior ORR activity and rechargeable Zn–air batteries (RZABs) assembled with this catalyst exhibited a high power density of 272 mW/cm² and cycle stability of more than 1400 h.

4. Mass transfer on gas-diffusion electrode and architecture design

Rational design of cell architecture for optimizing the electrolyte/electrode interface can boost mass transfer at the triple-phase interface. In this respect, gas-diffusion electrode and flow cell architecture have been proposed and shown enormous application potential for their remarkable promotion in high current density [103,104]. As shown in Fig. 4, the major difference between the diffusion architecture and electrolyte-fed system is the gas sources and the mass transfer directions. In a diffusion system, gas molecules diffuse and react at the back-side of the electrolyte. Cooperating with a flowing electrolyte, the gas transfer aligns with the electrolyte turbulence direction at high current densities. In comparison, the gas in an H-cell diffuses to the catalyst layer from the front-side bulk electrolyte against the direction of the electrolyte refreshing, thus fails in breaking the liquid-fed limitation, which results in a relatively high transfer resistance. It has been evaluated that the path of CO₂ diffusion to the catalyst surface is reduced by approximately three orders of magnitude in a gas diffusion layer (∼50 nm) than in an H-cell (∼50 µm) [15]. The gas-diffusion system significantly shortens the fluid boundary range of the gas-electrolyte-catalyst interface, which allows to increase the maximum current density (j _max). The flow cell architecture can be divided into Kenis-type electrolyzer and membrane electrode assembly (MEA) electrolyzer, which will be discussed in the following content.

4.1. Kenis-type electrolyzer

The liquid flow cell which known as Kenis-type electrolyzer consists of three flow houses, a gas diffusion electrode separating the catholyte and gas phase, and an ion-exchange membrane separating the catholyte and anolyte liquid streams (Fig. 5a). The catalyst film can be bind, drop-cast, or deposited at the front side facing the flowing catholyte while gas is continuously fed to the catalyst surface through the back of the gas diffusion electrode. Kenis-type electrolyzer has achieved CO₂ conversion at high current density (up to ampere level) toward CO [105], formate [106], and multi-electron products [107,108]. Notably, though Kenis-type electrolyzer has also been adopted in NRR, achieving commercially-relevant current density is still challenging due to the low intrinsic activity of the catalysts that is difficult to break the kinetic limit of N₂ transfer [109,110]. The unique advantage of Kenis-type electrolyzer is that gas and catholyte streams at both side of the cathode, which can be precisely controlled and optimized from gas molecule transfer to the reaction microenvironment, where the flowing velocity can be adjusted by changing the electrolyte characteristic (e.g., pH, concentration, and viscosity).

The flowing liquid electrolyte can also be beneficial for refreshing the wet interface of GDL to maintain the high-level of gas saturation. Moreover, the metastable equilibrium of wetting and hydrophobic effect as well as dynamic refreshing effect enable saturation and fast supply of the gas molecule during the cell operation. However, the continuous turbulence of the catholyte and the electric wetting effect can also lead to catalyst flooding. Additionally, the CO₂RR flow cell is subjected to bicarbonate salt deposition, which can reduce electrolyte conductivity, change electrolyte pH, and block gas diffusion electrode/membrane pores, jointly destabilizing the device [111]. Therefore, to achieve high current density, gas transfer must be synchronized with the catholyte turbulence, ensuring equilibrium between humidity and pH in the triple-phase environment. Otherwise, this equilibrium will be disrupted, leading to bicarbonate formation and excessive H₂ evolution.

4.2. MEA electrolyzer

Another flow cell is electrolyzer with non-flowing catholyte, also called MEA electrolyzer or zero-gap electrolyzer. As shown in Fig. 5b, this device integrates one-unit cathode house consisting of a gas-diffusion layer, catalyst and ion exchange membrane [112,113]. This configuration abandons the catholyte and provides the necessary water environment required for gas reduction by maintaining the humidity of the flowing gas, which is conductive to reduce ohmic losses and increase current density.

Compared to Kenis-type electrolyzers, MEA electrolyzers further enhance the mass transfer of gas molecules and increase the concentration of available gas molecules. Lee et al. [114] reported a comparison between the MEA system and the Kenis-type electrolyzer, where a significant improvement was obtained in formate selectivity and system stability in the MEA electrolyzer.

Many reported works have showed that non-Cu metal catalysts (e.g., Pt, Fe, Co) in the MEA electrolyzer can produce multi-carbon products such as ethanol, acetic acid, acetone, and isopropyl alcohol [115-117], while they typically exhibit limited C—C polymerization ability in aqueous-fed systems or in Kenis-type flow cells. Though the mechanisms of long-chain products formation are still in controversy, it can be inferred that the optimized CO₂ transfer process in the MEA electrolyzer can affect the reaction pathway. The selectivity change may not only be attributed to the increased mass concentrations, but also the change in transfer pathway of the key reaction intermediates. Whether the gas fixation reaction occurs at a three-phase system (gas-electrolyte-catalysts) or a two-phase media (dissolved gas-catalysts) remains elusive. Burdyny et al. reported that nanoscale thickness of water was observed in the MEA electrolyzer, which is likely to surround the catalyst even near the gas-liquid interface, forming a much shorter diffusion pathway [15]. In this scenario, the partially-wetted GDL offers a complex and porous interface, which possesses vast surface area, facilitating the gas transfer and reduction.

In the MEA electrolyzer, the ion exchange membrane can both determine the transport of specific ions and mass transfer, jointly affecting electrocatalytic performance. In CO₂RR system, cation exchange membrane (CEM) and anion exchange membrane (AEM) are two commonly used ion exchange membranes, which can transport cations (such as H⁺, K⁺ and Na⁺) and anions (such as OH⁻, CO₃ ²⁻ and HCO₃ ⁻), respectively [118]. In CEM electrolyzers, protons can be transported from the anolyte to the cathode, whereas in AEM electrolyzers, protons can only be provided through the dissociation of water molecules in the humid gas [12]. AEM can effectively inhibit HER and improve product selectivity [119], thus being the most popular membrane in MEA electrolyzers. However, it suffers from severe reactant crossover (such as CO₃ ²⁻) and product crossover (such as HCOO⁻ and CH₃COO⁻), leading to low carbon utilization efficiency and severe product loss. Salting out also blocks gas diffusion channels and affects membrane conductivity [120]. In contrast, CEM exhibits high conductivity and can prevent crossover of anionic products, but the competed HER decreases the product selectivity [119]. In this regard, Sargent et al. [121] increased the K⁺ concentration in the bulk electrolyte by adding KCl to the acidic electrolyte and concentrated K⁺ at the electrode interface by modifying cationic perfluorosulfonic acid (PFSA) on the Cu catalyst. K⁺ can enhance CO₂ activation and enabled a single-pass CO₂ utilization of 77% and a C₂₊ selectivity of nearly 50%.

4.3. Auxiliary means for architecture design

Multiscale simulations and calculations are valuable tools for comprehending mass transfer mechanisms within cell architectures and optimizing the efficiency. Integrating computational models at various length and time scales would be helpful to capture the intricate interplay of processes occurring at different levels, from molecular interactions to macroscopic phenomena. For example, Xie et al. [122] established a theoretic model of coupled mass transport and electrochemical reaction for N₂ electroreduction in the flow cell using mathematical analysis. This model illustrates that increasing N₂ feeding, decreasing catalytic layer thickness and channel width can significantly improve the mass transfer, whereas increasing electrolyte flow rate has negligible improvement in current density.

Besides the simulations and calculations, the advanced manufacturing and characterization techniques are desired. Feng et al. [123] decreased the diffusion layer thickness with polytetrafluo-roethylene (PTFE)-decorated hydrophobic electrode layer that can repelling liquid electrolyte and gathering gaseous reactants, remarkably facilitating the mass transport and electrocatalysis kinetics. Furthermore, by applying confocal laser scanning microscopy to image triple-phase interfaces during CO₂RR [124], Cassie-Wenzel coexistence wetting state is suggested as guideline to design ideal interface structure. The wettability of gas-liquid-solid interfaces has been well tuned by fabricating a typical Au/C gas diffusion electrode and in situ electrochemical spectroscopy detection. As a result, the variations of wettability over triple-phase interface can dramatically change the interfacial transportation, where the interfacial CO₂ mass-transfer coefficient can reach 0.27 cm/s and maintain 80% of the initial CO₂ concentration at the interface at high operating current density above 100 mA/cm².

5. Emerging electrolyzers with novel mass transfer pathways

Though flow cell improvement and catalyst design have gained remarkable success in increasing the operating current density and product selectivity, some critical issues still need to be solved, including stability of the cathode, high cost of the separator membrane, low energy efficiency of the total cell. Therefore, stable, low-cost, and efficient electrochemical systems with new mass transfer pathway for air-abundant molecules fixation at commercially-relevant current density deserve further development. Here, we highlight two emerging architectures in tuning the mass transfer mode, including SSE electrolyzer, and microfluidic electrolyzer.

5.1. SSE electrolyzer

The application of solid-state electrolyte can circumvent the aqueous-fed mechanism in which mass transfer is limited by the low solubility of gas, holding the fastest mass transfer process (Fig. 6a) [125,126]. SSE electrolyzers can be categorized into two types, i.e., solid polymer electrolyte (SPE) electrolyzers and solid oxide electrolyte (SOE) electrolyzers. The common types of SPE include sulfonated styrene-divinylbenzene copolymer, polyvinyl alcohol (PVA) and sulfonated polyetheretherketone (SPEEK), among which sulfonated styrene-divinylbenzene copolymer is widely used due to its high ionic conductivity. The ion conduction mechanism of SPE is primarily attributed to the segmental motion of the polymer chains and the introduction of ion-conductive additives [126]. Although SPE has high mechanical strength, good battery stability, ease of operation, and effective suppression of hydrogen evolution, its catalytic activity and selectivity is insufficient [27]. With the merit of high proton/oxygen ion conductivity and low electron conductivity, SOE can achieve reduced products at lower voltages with high current density and selectivity [127]. The commonly used types of SOE are lanthanum strontium gallium magnesium oxide, ceria-based oxides and yttria-stabilized zirconia [12,126]. A favorable gas diffusion environment in SOE helps the mass transfer of gas molecules. However, SOE electrolyzers often operate at an extremely high temperature (>600 ℃), which increases cell design complexity and cost of use [126]. In addition, without the ion promoting effect, the C₂₊ product is limited by the low C−C coupling activity. Therefore, constructing an SSE electrolyzer with friendly reaction environment, room temperature tolerance, and low building cost is essential but challenging.

5.2. Microfluidic electrolyzer

Due to the unique property of capillarity, microfluidic devices have been a research hotspot in many fields, including microfluidics lab, microfluidics cell culture, microfluidics analysis, microfluidics catalysis [128,129]. Kenis et al. [130] first reported a microfluidic electrolyzer for ORR, which consists a thin electrolyte flow-field channel (<1 mm) to separate electrodes rather than a separating membrane. A GDL is adopted for the diffusion of O₂ and gas products, and crossover of reactants and products is controlled at laminar flow conditions because of the slow diffusion of products. Due to its high surface area to volume ratio and the capillarity property, a microfluidic electrolyzer design could allow fast rates of gas mass transfer to the cathode surface and thus achieve high current density for the gas fixation. Benefiting from the superiority of the microfluidic technology, Xu et al. [131] designed an advanced bubble-based microreactor (BBMR) to orderly regulate the behavior of N₂ microbubbles at the T-junction of the microdevice. The uniform bubble flow provides abundant N₂ at the triple-phase interface, leading to improved NRR performance (twice that of the H-cell) and high universality for different catalysts. Prospectively, microfluidic devices hold significant potential in rapid gas transport and fast gas-liquid mixing for advancing, scalable, and sustainable energy conversion. Thanks to relatively compact design, a smart microfluidic chip with a high level of integration may achieve high-throughput catalysis (Fig. 6b).

6. Conclusion and outlook

The development of energy conversion and storage technologies such as fuel cell, artificial nitrogen fixation, and CO₂ conversion, are indispensable to address the energy shortage and environmental pollution. The reduction of multi-phase involved air-abundant molecules (e.g., CO₂, N₂, and O₂) is a key part of these technologies. Since molecules probably experience dissolution, diffusion, and reaction near the catalyst surface, mass transfer plays a crucial role in determining the energy conversion efficiency. In electrochemical reduction, there are several factors that can affect mass transfer, including the operating conditions (such as temperature, pressure, and agitation speed), electrolyte composition, EDL structure, catalyst morphology, and electrolyzer architecture. In this review, we focus on the mass transfer issues on CO₂, N₂, and O₂ reduction reaction and review the significant progresses. Our discussions primarily include fundamental gas transport mechanism from bulk electrolyte to catalyst surface under operating conditions, rational design of the electrolyte/catalyst interface, modification of the cell configuration, and emerging methods for breaking regular gas transport limitation. Despite significant progress in this area, several challenges persist and require further breakthrough:

(1) Multiphase-interface catalysis and emulsion electrolyte:

Drawing inspiration from the soluble additives or cosolvents in water, prospects for multiphase-interface catalysis and emulsion electrolyte formed by insoluble organic solvents or Pickering emulgators will give chance to create different reaction environment for decoupling mass/electron transfer and increasing active surface area. The two-phase interface between insoluble organic solvents and water can create two types of reaction environment for reacting molecules, leading to the vibrational, energetic, and solvation-structural change of the different parts of molecule exposed to different phase. Moreover, according to the product distribution, rational design of the two-phase system can directly separate the products. The emulsion electrolyte can efficiently enrich the active interface, thus remarkably changing the gas mass transfer pathway for the confinement effect of the micelle. Even in an aqueous-fed system, the gas can transfer with the micelle and release at the electrode surface, which can increase the surface concentration of the gas reactant, break the natural limitation of the aqueous mass transfer and boost the current density.

(2) EDL structure in molecular and dynamic view:

Manipulating the solvent and ion microenvironment of EDL may be a central focus of future research, especially the physical fields, interface property, and molecular-ion configurations. This is of great significance for not only elucidating the structure-function relationship between the EDL and catalytic activity, but also exploring the key factors influencing the interfacial mass transfer kinetics. The EDL exhibits a dynamic and intricate structure that varies with electrode potential, electrolyte composition, and surface properties. Characterizing these structural changes and elucidating their influence on electrochemical reactions require sophisticated experimental techniques and theoretical models.

(3) Oriented electrode assembly and novel device architecture:

To maintain rapid mass transport and break solubility limitations, ordered electrode will be an efficient strategy for the next-generation flow cells. Through advanced and novel electrode assembly techniques including synthetic regulation, solvent-assisted membrane swelling, ionomer transition layer casting, electrospinning, nanoimprinting, microfluid, mass transfer at the triple-phase reaction interface can be better controlled over the precise management of humidity level, flow field and gas pressure.

(4) Multiscale simulations and accurate calculations:

In mesoscopic scale, numerical simulation methods (e.g., volume of fluid, lattice Boltzmann method, smoothed particle hydrodynamics) help to solve gas transport behaviors by exploiting knowledge of fluid mechanics. In microscopic scale, deeper theoretic calculations (e.g., ab initio molecular dynamics, density functional theory, grand canonical Monte Carlo) can provide a comprehensive understanding on the molecule transport at triple-phase interfaces, thereby providing valuable guidance for optimizing electrode design, reactor structure and electrolysis condition to achieve more efficient catalytic systems. In addition, multiscale finite element simulations and data-driven artificial intelligence can be employed for optimizing fluid field and designing device architecture.

In conclusion, the electroreduction of air-abundant molecules is a promising technology to harness intermittent energy sources and significantly tackle climate change. Future research endeavors can concentrate on enhancing mass transport efficiency. It is expected that the discussion and outlook in this review can shed light on further promotion of mass transport, thus facilitating sustainable and scalable deployment in utilizing air-abundant molecules via electrochemical catalysis.

Declaration of competing interest

The authors declare that they have no competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

CRediT authorship contribution statement

Xiaoyu Du: Writing – original draft, Investigation, Conceptualization. Huan Wang: Writing – review & editing, Supervision, Funding acquisition.

Acknowledgments

We acknowledge the support from the National Natural Science Foundation of China (Nos. 22375103, 22105107), Ministry of Science and Technology of China (No. 2021YFA1201900) and the Fundamental Research Funds for the Central Universities.