Recent Progress in Design Strategy of Anode for Seawater Electrolysis

Direct electrolysis of inexhaustible seawater to generate green hydrogen represents a more environment‐friendly technology relative to freshwater electrolysis. However, current seawater splitting suffers from a low catalytic efficiency and poor operation stability caused by anodic competition between the oxygen evolution reaction and the chlorine oxidation reaction as well as the severe chloride corrosion. This article provides a comprehensive overview of the latest achievements in promoting the selectivity and stability in seawater electrolysis. Beginning with the fundamentals of the anode reactions during seawater splitting, various strategies to design advanced anodic electrocatalyst including surface selective layer engineering, structural regulation by heteroatoms doping and vacancies, and heterostructure construction are discussed in detail. Finally, the conclusion and the challenges in developing seawater electrolysis technology are highlighted.


Introduction
High-purity hydrogen (H 2 ) produced by water electrolysis represents one of the most promising alternatives to traditional fossil fuels due to its high energy density and zero-emission characteristics.3] Meanwhile, about 97% of the water resource on earth is seawater (3.0-5.0%salts) with 1.0% being brackish groundwater (0.05-3.0%salts). [4]Considering the scarcity of freshwater resources, hydrogen production by seawater electrolysis may be more promising in the long run, especially at freshwaterdeficient regions.The integration of seawater electrolysis technology and fuel cells provides not only efficient energy conversion and storage but also clean drinking water for these regions.In addition to this superiority, seawater electrolysis can also be coupled with offshore wind power or solar generation systems.This running mode is an ideal choice for intermittent energy storage, it will not compete with agricultural or other industrial land space, reduces renewable electricity costs, and has been actively explored by various countries. [5]owever, similar to freshwater electrolysis, the anodic oxygen evolution reaction (OER) is a bottleneck due to the complex and sluggish four-electron oxidation process involving the O─H bond breaking and the O─O bond formation.Moreover, the seawater electrolysis faces additional challenges due to the complex composition of seawater.On the one hand, insoluble solids inherent in seawater like bacteria and microbes will block the active sites and interfere with the electrolysis.On the other hand, there are various dissolved ions such as Na þ (%0.4862M), Mg 2þ (%0.0548M), Ca 2þ (%0.0107M), and Cl À (%0.5658M) that may cause detrimental side reaction. [6]or direct natural seawater electrolysis without pretreatments, Mg(OH) 2 and Ca(OH) 2 precipitation may be generated when pH value of electrolyte is higher than 9.5, blocking active sites and poisoning the electrocatalysts, thus degrading the catalytic performance.The Ca 2þ and Mg 2þ ions can be removed by adding alkaline solution to form precipitates and then filtrating, in which the microorganisms and bacteria can also be removed. [7]However, Cl À ions are hard to be removed by pretreatments, leading to a chloride corrosion of electrode, especially at high potentials during OER electrolysis.10] Given all these factors, it is of significance to exploit the OER catalysts with outstanding activity, high selectivity, and high chlorine corrosion resistance.
Over the past decades, a variety of highly-efficient OER electrocatalysts have been reported to exhibit a comparable performance to noble metal-based catalysts, including but not limited to transition metal (oxy)hydroxides, [11][12][13][14] transition metal sulfides, [15][16][17][18] selenides, [19][20][21][22] tellurides, [23][24][25] transition metal phosphides, [26,27] transition metal nitrides, [28,29] and their hierarchical composites. [30,31]Among them, some Ni-based compounds such as NiFe-based layered double hydroxides, [32] Fe x -Ni&Ni 0.2 Mo 0.8 N, [33] NiFeP, [34] and Ni 3 N-Ni 3 S 2 , [35] could operate steadily in alkaline saline electrolyte. [36,37]espite the prominent progress that has been made in the development of OER catalyst for seawater electrolysis, a timely summary of these new encouraging achievements is seldom.Additionally, the available reviews are generally only from a view of improved OER activity by tuning of the electronic structure, increasing electrical conductivity, and active surface area.In this respect, it would be very valuable to provide a comprehensive review from the perspective of selectivity and chlorine corrosion resistance of OER catalysts.In light of this, here, a systematic review concentrating on the design strategies toward improving the OER selectivity and chlorine resistance for seawater electrolysis is provided.We first overview the fundamental mechanisms of OER and the descriptors for OER selectivity in seawater electrolysis.Subsequently, various strategies including surface selective layer engineering, structural regulation by heteroatoms doping, vacancies, and heterostructure construction are thoroughly summarized (Figure 1) and discussed in detail.Finally, a conclusion is given followed by an outlook of the remaining challenges and perspective on the practical application of seawater electrolysis.Hopefully, this review can inspire researchers to further promote the development of seawater electrolysis.

Fundamentals of Seawater Electrolysis 2.1. Mechanisms of OER
Up to now, the mechanism of OER has been widely investigated but is still under debate in alkaline freshwater electrolytes due to the multistep electron-proton coupled reactions and the occurrence of surface reconstruction during this process.A generally accepted OER pathway is the adsorbate evolution mechanism (AEM).Under this mechanism, the surface structure in the OER pathway is regarded as stable with only valence state changes of the catalytic active sites (usually metal sites) during oxygen intermediate evolution.As shown in Figure 2a, the AEM of OER starts with the formation of M-OH and M-O, followed by the generation of oxygen molecules through two different routes: 1) Another OH À ion is absorbed on the surface of *O to produce *OOH intermediate, then *OOH combines with OH À by proton-coupled electron transfer to produce oxygen molecule and 2) Two *O species directly coupled together to form an oxygen molecule.Regardless of the pathway, the bonding energies within the intermediates (M-OH, M-O, and M-OOH) are crucial for the overall electrocatalytic ability of OER. [38]In general, the first pathway has a lower thermodynamic barrier than the second one, thus the pathway via the steps of M !M-OH !M-O! M-OOH is considered the prevailing OER process.The step with the maximum energy barrier is the rate-determining step (RDS), which dominates the catalytic efficiency.In the ideal case, the above-mentioned four steps proceed with the same reaction-free energy of 1.23 eV, which means the overpotential is zero.However, the density functional theory (DFT)-based calculated binding energies between the metal sites and the adsorbed *OH, *O, and *OOH intermediates are linearly related, among which the (ΔG *OOH -ΔG *OH ) is a constant of 3.2 AE 0.2 eV for many catalysts, indicating that the lowest overpotential in AEM mechanism is 0.37 V. [39] According to the Sabatier principle, a volcano-type relation obtained on (ΔG *O -ΔG *OH ) with the OER overpotential (a typical example in Figure 2b) is a widely accepted descriptor for predicting OER activity. [40]In fact, the adsorption energy of O atoms has a close relationship with the electronic structure of the catalysts due to the orbital interaction between oxygen 2p orbital and metal nd orbital.According to the d-band theory raised by Nørskov et al., the binding energy of an adsorbate to the active metal site is determined by the d-band center of the metal site.The higher energy of the d-band center relative to the Fermi level is, the antibonding states are less occupied and the adsorption is stronger, and vice versa (Figure 2c). [41]The d-band center can be regulated by many factors, including the ligand effects, [42,43] the strain effects, [44] heterostructure effects, [45] alloying, [46,47] doping, [48] etc.An optimal d-band center leads the optimized *OH adsorption energy, corresponding to the top of volcano plot.The d-band theory provides a practical criterion to optimize the adsorption energy of O-containing intermediates and thus a guidance to improve the OER performance.In addition to d-band center, the number of d electrons, [49] e g orbital occupancy, [50] the position of the O-p band center, [51] and metal-oxygen covalency [52,53] are also popular descriptors that correlate the structure and the activity of the catalyst.
With the exploration of more efficient OER catalysts, especially metal oxide/hydroxides with high metal-oxygen bond properties like perovskites, [54] it was found that the lattice oxygen atoms within the catalysts are often oxidized and further involved in the OER process to produce oxygen molecule, which is called lattice oxygen-mediated mechanism (LOM). [55]Compared with AEM, LOM can break the scaling relationship and better interpret the catalytic activity origin of the catalysts, while still using the binding energy of intermediates as a valid activity descriptor. [56,57]Under this mechanism, oxygen ligands are electrochemically activated and released from the lattice matrix.As shown in Figure 2d-f, three prevailing OER pathways via LOM have been proposed based on different catalytic center sites. [58]On the one hand, the activated lattice oxygen atoms  [40] Copyright 2021, Wiley-VCH.c) Schematic of the formation of a chemical bond between an adsorbate valence level and the s and d states of a transition-metal surface.Reproduced with permission. [41]Copyright 2005, Springer Nature.d-f ) Schematics of three alternative pathways of LOM in alkaline media with different catalytic centers, the chemically active lattice oxygen involving OER and oxygen from the electrolyte are marked in blue and golden colors, respectively, and □ represents lattice O vacancy: d) Oxygen-vacancy-site mechanism (OVSM), e) single-metal-site mechanism (SMSM), and f ) dual-metal-site mechanism (DMSM).Reproduced with permission. [58]Copyright 2021, The Royal Society of Chemistry.may act as the active sites, which directly adsorb OH À by the nucleophilic attack to form *OOH species.Then, the O 2 molecule is generated by the combination of one lattice oxygen atom and one oxygen atom in OH À .The release of an O 2 molecule forms an oxygen vacancy site, which is refilled by OH À from the electrolytes.Afterward, the M-OH on the surface is regenerated through a subsequently concerted proton-electron transfer step.This mechanism is called the oxygen-vacancy-site mechanism (OVSM, Figure 2d).On the other hand, the singlemetal-site mechanism (SMSM) (Figure 2e) and the dualmetal-site mechanism (DMSM) (Figure 2f ) are more similar to AEM, the metal sites still serve as the catalytic center to adsorb OH À and proceed the following concerted proton-electron transfer step.The main difference between SMSM and AEM is that surface reconstruction enables the direct coupling of the absorbed *O intermediate and activated lattice oxygen, forming an oxygen molecule and leaving an O vacancy.SMSM is energetically favorable for the high-valence metal cations. [59]As for DMSM, the adjacent activated lattice oxygen may interact and form the M-OO-M motif, and the *OO* moiety further generates an O 2 molecule.This mechanism is mainly reported in metal (oxy)hydroxides. [60]The element type, crystallinity degree, defects, and vacancies in the catalyst are critical for the OER pathway. [61]Therefore, the OER pathway is complicated and should be specifically analyzed.

The Competition between OER and COR
The electrochemical reaction competition between the OER and chloride oxidation reaction (COR) commonly occurs at the anode for seawater electrolysis.The COR mechanism may proceed through the following three steps (Table 1), in which * is an active site (surface oxygen or a metal atom). [62]The binding energy between active sites and chlorine species plays a key role in COR activity.
Catalysts exhibiting the ability to oxidize water also tend to catalyze the oxidation of chloride.A scaling relation between the binding energetics of the OER and COR intermediates was observed on RuO 2 and IrO 2 surface (Figure 3a), indicating that catalysts that interact with oxygen-bound intermediates strongly may also bind chloride-bound intermediates strongly. [63]s well, many transition metal oxides including Co 3 O 4 , [64] Fe 3 O 4 , [65] and PbO 2 , [66] can cause COR to predominate in Cl Àcontaining solutions without being alkalized.
Although the OER is dominated by thermodynamics, the kinetics are faster for the 2e À chloride reactions than the sluggish 4e À process of OER.Dionigi et al. pointed that the chlorine evolution (2Cl À !Cl 2 þ 2e À , E 0 = 1.36 V vs standard hydrogen electrode (SHE), pH = 0) at low pH and hypochlorite formation ( ) at high pH solution are the main OER competing reactions (Figure 3b). [67]Maximizing the thermodynamic potential difference between the two reactions will result in a relatively high potential window for the selective OER.In this potential window, ideal electrocatalysts can yield high OER currents but without hypochlorite formation.The potential difference (ΔE) enlarges with the increase of pH, and the ΔE is maximized as 480 mV at pH > 7.5, which means higher pH facilitates the selective splitting of seawater.Nevertheless, at a sufficiently high electrolysis rate (100-1000 mA cm À2 ), the pH at the electrodes surface may dramatically change, leading to the decrease of potential window for selective OER. [68]Additionally, in the potential window, the competitive adsorption between Cl À ion and OH À still occurs at the active sites, which has a negative effect on seawater electrolysis.In addition, the competition between COR and OER restricts the upper limit of the applied potential at anode, hindering the development of high-currentdensity electrolyzers.

Mechanism of the Chloride Corrosion
Chloride corrosion is also a serious problem in seawater electrolysis.Pt-, Ir-, Fe-, Ni-, and Co-based OER electrocatalysts are all known to be susceptible to dissolution in the presence of [71] The main reason for the high corrosiveness of Cl À is that it can directly react with these electron-deficient transition metals, resulting in a change of composition.The mechanism of chloride ion corrosion on metal surfaces is shown in Figure 2c.First, the metal surface gets polarized under an external electric field and results in Cl À adsorption (M þ Cl À !MCl ads þ e À ).Cl À ions are then adsorbed on the metal surfaces and dissolved by further coordination (MCl ads þ Cl À !MCl x À ).Finally, metal chlorine disintegrates to form the metal hydroxide corrosion product ). [72] The corrosion process gradually consumes hydroxide ions, resulting in a reduction of local pH, further accelerating the corrosion of the metal substrate.Consequently, Cl À -triggered corrosion will gradually dissolve transition-metal-based catalysts, leading to catalyst poisoning and poor catalytic durability.

The Evaluation of OER Selectivity
Given that the content of Cl À in seawater is about 0.5 M, and the Na þ is the highest-containing cation in seawater, 0.5 M NaCl solution has become a commonly used additive for simulating seawater.To quantify chloride oxidation, several techniques are usually applied as follows.Faradaic efficiency (FE) of O 2 is an important descriptor of the OER selectivity, which is usually obtained by in-line mass spectrometry (MS) test, gas chromatography (GC) test, or drainage method in a gas-tight H-cell conducted during a galvanostatic test protocol.The first two methods have the advantage of high measurement accuracy but complex equipment requirements while the last one is convenient but has low accuracy. [73,74]The FE of O 2 can be calculated using the theoretical gas concentrations (C gastheor ) which can be obtained based on Faraday law and the ideal gas equation and the nominal measured gas concentrations (C gasnominal ) as follows: FE (%) = C gasnominal /C gastheor * 100. [75]The FE can also be obtained by rotating ring-disk electrode (RRDE) test, under specific potentials (e.g., 0.5 V vs SHE), the Pt ring electrode is only sensitive to the reduction of ClO À back to Cl À (not sensitive to oxygen reduction reaction), while OER and possible COR occurs at the disk electrode.Accordingly, the FE of COR can be calculated by FE (COR) (%) = [(i COR /2)/(i COR /2 þ i OER /4)] * 100%, in which i COR can be calculated from the ring current density and the collection efficiency, and i OER can be obtained by subtracting the i COR from the disk current.The FE of OER corresponds to: FE (OER) (%) = 100% À FE (COR) (%) . [76]Notably, RRDE method is not applicable for large current densities due to the limitation of the reduction reaction rate at the ring. [77]n addition, it is necessary to detect the possibly generated chlorine species in the electrolytes after a long test.The colorimetric method combined with the ultraviolet-visible spectra (UV-vis) using some reducing reagents like N, N-dialkyl-1,4phenylenediamine (DPD) [78] and o-Tolidine [79] can be used to detect the ClO À ions in the solution after electrolysis.The existence of hypochlorite can also be directly probed by UV-vis spectroscopy, from which a characteristic absorption peak at %292 nm is observed.The iodometric titration method can also be applied to detect the amount of ClO À ions.Once KI is added to the solution, the generated ClO À ions will react with I À to form À along with color changes.Afterward, with the addition of Na 2 S 2 O 3 solution, the yellow-brown color gradually faded and transformed to faint yellow.After adding a starch indicator and up to the transparent endpoint, the volume of thiosulfate consumed in this titration was recorded and the total amount of ClO À ions can be calculated through the total amount of the added thiosulfate. [80]The conventional titration method provides a convenient and effective detection, furthermore, the concentration of I 2 can be estimated more accurately using absorption spectrometry. [81]In addition, the content of ClO À can also be tested by a portable residual chlorine detector under a certain voltage using the potentiostatic method for several minutes. [82]Considering the sodium hypochlorite is easy to be decomposed, the above-mentioned measurements should be performed immediately after the stability test.

Surface Selective Layer Engineering
Constructing a chlorine ion block and corrosion-resistant layer on the surface of the anode is an effective approach to improve the selectivity and stability of OER.This layer should satisfy the property of OH À selective permeability and Cl À impermeability, and thus H 2 O, OH À , and O 2 can diffuse through the layer while Cl À is blocked and unable to penetrate.In general, selective permeability can originate from the size-sieving effect, the electrostatic interaction, and the energy barrier difference in removing water from hydrated ions.Accordingly, three kinds of selective layers, including the microporous inert layer, electrostatic repulsive layer, and amorphous (oxy)hydroxides layer were classified and summarized.

Coating Microporous Inert Layer
The coated materials with Cl À impermeability are generally rich in micropores.They exhibit a high intrinsic stability and do not produce electrochemical reactions or degrade into soluble substances.In this part, four kinds of coating materials: manganese-based oxides, cerium oxides, silicon oxides, and carbon-based materials were reviewed.
Manganese-Based Oxides: Manganese oxides (MnO 2 ) present an outstanding OER selectivity due to the weak Cl À adsorption by surface polarization. [83,84][87] Among them, α-MnO 2 with a tunnel size of 4.6 Å can accommodate most metal cations (such as K þ , Ca 2þ , Na þ , and Mg 2þ ) and water molecules while β-MnO 2 with a size of 1.89 Å can only allow small ions like H þ or Li þ and is not conducive to ion diffusion.δ-MnO 2 with 2D layered structure has a large interlayer distance of 7.0 Å, which can hold number of water molecules, metal cations, and other substances.These tunable porosity structures varied with different phases provide the ion-selective permeability property of MnO 2 .In 1980, Bennett et al. initially reported that amorphous MnO 2 on RuO 2 /TiO 2 substrate by electrodepositing showed an effective Cl À corrosion resistance and high OER selectivity, achieving an FE of 95% and 99% in 1 M NaCl solution and seawater, respectively.The coated MnO 2 acted as a diaphragm, enabling the dominant anodic OER with no significant chlorine evolution. [88]Later, a series of Mn-based metal oxides deposited on different substrates of IrO x /Ti, Sn 1Àx Ir x O 2 /Ti, and Sn 1ÀxÀy Ir x Sb y O 2þ0.5y /Ti was developed.91][92] In 2018, Vos et al. elucidated that MnO x film with a porous and amorphous network facilitated the selective formation of O 2 over Cl 2 on the surface IrO x in strongly acidic media. [93]he MnO x coating functions as a porous overlayer, preventing the transport of Cl À ions but allowing H 2 O and H þ onto IrO x (Figure 4b,c).Additionally, it is inert for either CER or OER, and little electronic interaction between MnO x and IrO x was detected (Figure 4d).The selectivity was also verified by X-ray photoelectron spectroscopy (XPS) investigations (Figure 4e), from which the presence of an alkali metal chloride on IrO x / GC electrode could be detected while no Cl 2p peaks from MnO x /IrO x /GC displays could be observed, indicating that chloride is unable to penetrate the MnO x film.The slow transport of chlorine leads to a significant increase of the diffusion layer thickness and effective surpasses the CER activity.
Recently, a thin film of Na-intercalated layered MnO 2 þ (δ-MnO 2 , noted as Na|MnO 2 ) was coated on fluorine-doped tin oxide (FTO) glass substrate by an electrodeposition process Reproduced with permission.Reproduced with permission. [87]Copyright 2021, Wiley-VCH.b,c) Sketch of the IrO x /GC and MnO x /IrO x /GC catalysts.d) Top panel: currents for OER and COR.Lower panel: corresponding selectivities toward OER and COR.Data are plotted as a function of Q MnOx , the charge ascribed to the corresponding MnO x layer, which is an approximate indication of its thickness.e) XPS scans of Cl 2p spectral peaks.Reproduced with permission. [93]Copyright 2018, American Chemical Society.f ) Sketch of the Na|MnO x catalysts.g) Visible spectra of the electrolytes with KI, taken after electrolysis with a delivered charge of 40 C cm À2 at the indicated current densities.h) Plots of the OER efficiency.Reproduced with permission. [81]Copyright 2021, American Chemical Society.
followed by heat treatment in air.It was found that the heating temperature would change the structure of thin film, thus significantly affecting the selectivity toward OER and CER (Figure 4f ). [81]With the increase of annealing temperature, oxygen vacancies would be produced.Specifically, an O atom was removed from the edge-shared MnO 6 octahedra, resulting in the formation of corner-shared MnO 5 pyramids accompanied by a decrease in the valence state of Mn.When annealing at 200 °C, the multilayered structure of Na|MnO 2 was maintained but the interlayer distance shrank from 0.73 to 0.67 nm.The Cl À ions adsorbed more rapidly at *O (* indicates the catalytically active site , indicating the preference of COR.While annealing at 300%500 °C, the multilayered structure of Na|MnO 2 became disorderly and yielded "oxygen-deficient" MnO x nanolayers.On this layer, OER may proceed through recombination of two adjacent oxygen inter- instead of the general peroxide route, thus increasing the selectivity toward OER.Na|MnO x -500 exhibited 87% OER selectivity in 0.5 M NaCl solution and no ClO À was detected after electrolysis at constant current densities of 3, 5, and 10 mA cm À2 (Figure 4g,h).Nevertheless, more solid evidence is required to verify the above-mentioned mechanism conversion.In addition, ternary manganese-based oxide coatings such as MnMoW oxides and Mn 0.929 Mo 0.067 Sn 0.004 O 2.067 were also investigated.In particular, hierarchical Mn 0.929 Mo 0.067 Sn 0.004 O 2.067 /Ir 0.84 Sn 0.16 O 2 /Ti electrode exhibited more than 99.9% oxygen evolution efficiency for more than 4,200 h in the electrolysis of 0.5 M NaCl at the current density of 1,000 A m À2 . [94,95]erium Oxides: Although the MnO x layers can effectively obstruct Cl À ions, they may be oxidized and dissolved as MnO 4 À at a high overpotential, leading to a decrease in electrolyzer efficiency.In this respect, CeO x was investigated as permselective layers in alkaline saline solutions. [96]Obata et al. deposited CeO x layer with a thickness of 100-200 nm onto NiFeO x at Au substrates.NiFeO x is regarded as a highly active electrocatalyst for OER but suffers from poor stability due to Fe species dissolution.Once the CeO x was coated onto NiFeO x , it could effectively prevent the loss of Fe species. [97]Moreover, it allowed the permeation of OH À and O 2 while inhibiting the diffusion of Cl À ions through the layer (Figure 5a).As a result, the introduction of an additional CeO x layer greatly improved the stability of NiFeO x in 1 M KOH þ 1 M KCl solution (Figure 5b,c).Although the origin of permselectivity has not been verified, it is observed that the diffusion of different reducing agents through CeO x layer is influenced by their sizes and charges.Depending on the change of Stokes radii, the CeO x layer showed diverse permselectivity to the reducing agents (Figure 5d).In addition, a significant improvement of OER selectivity was observed in the solution with redox anions rather than neutral alcohols with similar Stokes radii, indicating that charges of reactants play a significant role in the permeation through the CeO x layer.
Silicon Oxides: Microporous SiO x nanothick films have also been investigated as selective permeable layer.Esposito et al. deposited catalytically inert SiO x overlayers on a planar Pt using a room-temperature photochemical process.(Figure 5e).The as-prepared SiO x film with 4.8-8.7 nm thickness exhibited an excellent anti-permeability of Cl À , which is 3 orders of magnitude less than that in a conventional salt-selective membrane used in reverse osmosis desalination (Figure 5f ).SiO x |Pt could effectively inhibit the occurrence of COR even at the 0.6 M chloride concentration (Figure 5g).The improved OER selectivity can be explained as follows: for the SiO x overlayers, partial dehydration is the possible rate-limiting step for Cl À transport, H 2 O as the reactant for the OER is expected to have a much smaller activation barrier than the partial dehydration energy of Cl À for crossing from the bulk electrolyte into the hydrophilic SiO x overlayer, thus improving OER selectivity (Figure 5h,i). [98,99]arbon-Based Materials: Carbon-based materials, such as carbon nanotube (CNT), graphene, and their derivatives are also investigated as protective layers due to their ion-selective transmissivity property.[100] Grossman et al. performed molecular dynamics simulations to demonstrate that single-layer graphene with nanopores could effectively separate NaCl from water, and the desalination properties mainly depend on the aperture, chemical functionalization, and applied pressure.[101] Inspired by this finding, Zhu et al. prepared graphene oxide (GO) membrane by a simple drop-casting method and investigated its selective ion-penetration properties (Figure 6a).[102] As shown in Figure 6b, the oxygen-containing functional groups are easy to form aggregates, enabling other non-oxidized regions to form a two-dimensional network of graphene nanocapillaries, which provide high capillary pressures and facilitate the low-friction flow of water.When the distance between GO interlayers is large enough, the hydrated ions would penetrate.In addition, a large number of oxygen-containing functional groups decorate the surfaces and edges of the GO sheets, leading to the electrostatic attractions and chemical interactions between these functional groups and hydrated ions.As a result, the GO membrane achieves the ion-selective permeability.The sodium salts with different anion radicals exhibited different penetration rates in the order NaOH > NaHSO 4 > NaCl > NaHCO 3 , and GO can prevent NaCl penetration under low concentrations of ions (Figure 6c,d).
On the basis of these findings, Lee et al. developed an effective OER catalyst composed of FeOOH deposited onto β-Ni-Co hydroxide with an outer GO layer (GO@Fe@Ni-Co@NF) for selective alkaline seawater splitting (Figure 7a). [103]O@Fe@Ni-Co@NF is more stable than the bare Fe@Ni-Co@NF during the 12 h chronopotentiometry (CP) test (Figure 7b).After CP test, the dissolved concentrations of Co and Ni ions in the electrolyte for GO@Fe@Ni-Co@NF are 1-2 orders of magnitude lower than that of Fe@Ni-Co@NF, indicating that the GO layer played an important role in alleviating the dissolution of Co and Ni.Benefiting from the protection of GO layer, the electrolyzer using GO@Fe@Ni-Co@NF as both the anode and the cathode showed outstanding stability for overall seawater splitting with no noticeable degradation over 378 h operation at a current density of 1,000 mA cm À2 in 1 M KOH þ 0.5 M NaCl.In addition, GO@Fe@Ni-Co@NF showed the FE close to 100% in 1 M NaClþ 1M KOH at a current density of 200 mA cm À2 , indicating its high OER selectivity (Figure 7c).Here, the GO layer serves as a sieve and effectively blocks Cl À ions while allows OH À ions and O 2 gas penetration.GO layer is also favorable to improving the chlorine corrosion resistance of NiFe layered double hydroxide (LDH).As shown in Figure 7d, GO-NiFe-LDH@NF electrode maintained an original 3D foam network while NiFe-LDH@NF electrode was seriously damaged and corroded after the CP test, indicating the improved corrosion resistance by the introduction of GO coating.
In addition to GO, amorphous carbon and carbon nanotubes coatings on the active sites also contribute to a long-term stability.Song et al. fabricated carbon-coated sodium cobalt-iron pyrophosphate (Na 2 Co 1-x Fe x P 2 O 7 /C, 0 ≤ x ≤ 1) supported on carbon cloth (NCFPO/C@CC), (Figure 7e). [104]When employed as a OER electrocatalyst for alkaline seawater electrolysis, it exhibited remarkable stability up to 100 h without obvious decay at 50 mA cm À2 in the NaCl þ KOH electrolyte (Figure 7f ).In addition, no reactive Cl 2 or ClO À species were detected after the CP test, and the faradaic efficiency of NCFPO/C@CC was measured to be %100% at a current density of 50 mA cm À2 , further confirming the high OER selectivity in the alkaline saline solution (Figure 7g,h).Suh et al. fabricated graphene-carbon nanotubecobalt hybrid (S-rGO-CNT-Co) as an efficient electrocatalyst, in which Co nanoparticles were encapsulated by CNT. [105]e CNT layer prevents the adsorption of Cl À on the inner Co and thus significantly enhancing the stability.
Notably, carbon coating not only reduces the penetration of chloride ions but also effectively improves the conductivity of the catalysts and promotes charge transfer efficiency.For example, Qi et al. synthesized NiFe phosphide arrays on P-doped graphene oxide supported on nickel foam (NiFeP/P-rGO/NF) for efficient and stable OER in seawater electrolysis. [106]The hydrophobic P-rGO layer intimately contacts with the nickel foam substrate not only effectively prevented the corrosion of substrate by chloride ions but also enhanced the electrical conductivity, thus achieving a faster kinetic process.Wang et al. developed carbon-coated nickel-iron nitride microsheet arrays (Ni x Fe y N@C/NF), which delivered current densities of 100 and 500 mA cm À2 at low overpotentials of 314 and 394 mV in alkaline seawater and operated steadily at 100 mA cm À2 for 100 h. [107]The coated carbon inhibited structural collapse during the nitridation process and protected the interior nitrides from being poisoned and etched in seawater.In addition, the  [97] Copyright 2018, Wiley-VCH.e) Schematic of the SiO x |Pt catalyst.f ) Permeability of Cl À in SiO x overlayers as a function of overlayer thickness, calculated from LSVs for SiO x |Pt in 0.5 M KHSO 4 þ 0.6 M KCl.g) LSVs for 4.8 nm SiO x |Pt.Reproduced with permission. [99]Copyright 2021, American Chemical Society.h) The model of a hydrated ion.i) The hydrated ion diameters and enthalpy of hydration of anions.Reproduced with permission. [98]Copyright 2020, The Royal Society of Chemistry.
interfacial charge transfer from Ni x Fe y N to the carbon coating facilitated the initial water adsorption and splitting.Ma et al. also found that the coated few-layer N-doped carbon shell could not only protect the CoMoP core from corrosion but also optimize the absorption behavior of intermediates during the seawater splitting. [108]t should be emphasized that the electrochemical corrosion of carbon layers at high potentials is a vital challenge and needs to be suppressed.This could be achieved through the following aspects: designing the carbon layer with higher graphitization and lower porosity to reduce the easily-corroded carbon sites [109] ; increasing the pH of electrolytes and improving the loading of OER catalyst on carbon to ensure the kinetic superiority of OER [110] ; choosing proper additives in carbon-based materials like P-containing groups and hydrophobic agents to slow down the electrochemical corrosion of carbon. [111]lthough the above-mentioned surface selective layers can prevent the transmission of Cl À ions and achieve a high selectivity, they inevitably introduce some additional problems to be solved.For example, these layers will hinder the escape of bubbles especially under high voltages, which may block the surface of active sites and cause the current density instability during the electrolysis process.Moreover, surface coatings may change the surface wettability of the catalysts, which is not conducive to the penetration of electrolyte, reducing mass-diffusion-limited current densities at high potential. [112]In addition, the mechanisms of enhanced selectivity by these coatings are still not clear, and it is deficient in determining which characteristics of ions dominate the selective permeability.More studies are urgently required to reveal these fundamental issues, as they are important for stabilizing interactions between ions and the microporous layers to further improve the selective permeability.

Electrostatic Repulsive Layer Construction
Transition metal sulfides-, selenides-, phosphates-, borides-, and nitrides-based catalysts usually suffer from reconstruction under strong electrochemical oxidation environment.They are easy to convert into corresponding oxides or hydroxides at the initial stage of the reaction, and the corresponding polyatomic anions sulfates, selenate, phosphates, borates, and nitrates would be formed under the applied bias during the OER process. [113]he existence of this electronegative layer of polyatomic anion intercalated hydrotalcite inhibits the adsorption of chloride ions on the surface of the electrodes by the same charge repulsion. [114]gure 6.a) Schematic of the penetration processes of different ions through GO membrane.b) Schematic of GO membrane and the interaction with different ions.c) The penetration processes of different ionic compounds through GO membranes.d) The penetration processes of different concentrations of NaCl.Reproduced with permission. [102]Copyright 2013, American Chemical Society.
Kuang et al. demonstrated that a multilayer catalyst consisting of nickel-iron hydroxide layer coated on nickel sulfide on Ni foam (NiFe/NiS x -Ni, denoted as Ni 3 ) could operate steadily at the current densities from 0.4 to 1 A cm À2 over 1,000 h (Figure 8a). [72]he Ni 3 electrode underwent an activation/ passivation process after 3-4 h of electrolysis (corresponding to the obvious voltage dip in Figure 8b), and after that, the voltage remained stable and the OER efficiency increased to a stable value of %100%.In the activation step, anodic etching of the NiS x layer led to the formation of sulfate ions (Figure 8c), the sulfate along with the carbonate ions in KOH solution migrated and intercalated into the near amorphous NiFe LDH layer derived from the anodization of surface NiFe.This in situ generated sulfate and carbonate co-intercalated NiFe hydroxide catalyst layer together with the underlying sulfate-rich anodized NiS x layer afforded the repulsion to block of chloride anions, improving the OER selectivity and corrosion resistance to chloride anions in seawater.
Hung et al. reported a Se-NiFe-LDH electrode for OER, which was prepared by the selenidation of NiFe foam and the following electrodeposition of NiFe-LDH layer (Figure 8d). [115]The Se-NiFe-LDH electrode exhibited an extraordinary stability and activity in an aqueous sea-like solution.As shown in Figure 8e, it demonstrated much longer lifespan than pristine NiFe foam, Se-NiFe, and NiFe-LDH at 100 mA cm À2 in a Figure 7. a) Structure of the Fe@Ni-Co@NF and the GO@Fe@Ni-Co@NF catalyst.b) CP curves of Fe@Ni-Co@NF and GO@Fe@Ni-Co@NF electrodes in 1 M KOH þ 0.5 M NaCl at 1,000 mA cm À2 .c) FE measurement of GO@Fe@Ni-Co@NF for OER.d) Digital image and FE-SEM image of the GO@NiFe-LDH and NiFe-LDH after 12 h CP test.Reproduced with permission. [103]Copyright 2020, The Royal Society of Chemistry.e) FESEM images of NCFPO/C@CC.f ) CP test of NCFPO/C@CC at 10 and 50 mA cm À2 in NaCl þ KOH electrolyte.g) Time-dependent reactive chlorine concentration profiles of NCFPO/C in NaCl and NaClþ KOH.h) Mass spectroscopy of evolved gas after OER test at 10 and 50 mA cm À2 in NaClþ KOH.Reproduced with permission. [104]Copyright 2020, American Chemical Society.Reproduced with permission. [72]Copyright 2019, National Academy of Sciences.d) The fabrication process of the Se_NiFe_LDH electrode.e) Long-term stability measurements.f ) Mechanism of the stability improvement of the Se_NiFe_LDH electrode.Reproduced with permission. [115]Copyright 2020, Elsevier.g) Differential charge densities of NiOOH þ SeO 4 when an O atom is adsorbed on the adjacent Ni site.Yellow and green contours represent electron accumulation and depletion, respectively.h) The density of states of NiOOH and NiOOH þ SeO 4 regarding the Ni 3d orbitals.Reproduced with permission. [116]Copyright 2020, Wiley-VCH.) into the LDH layer after OER tests, and the inner layer remains selenide (MSe x ).It is believed that the intercalation of the larger SeO x À ions can repel Cl À ions from the surface of anode and thus improving the stability and suppression of corrosion, as illustrated in Figure 8f.
In addition to generating electrostatic repulsion with chloride ions, the SeO 4 2À and SO 4 2À ions at the surface can also interact with the active sites, thus enhancing the OER catalytic activity.

Shi et al. used investigated OER process on SeO 4
2À absorbed-NiOOH surface by density function theory (DFT) calculations to demonstrate this effect. [116]In the optimized model (Figure 8g), SeO 4 unit connects with a Ni atom on the surface through an oxygen atom bridge, meanwhile, two hydrogen bonds are formed between two oxygen atoms in SeO 4 and two hydrogen atoms in NiOOH.As shown in Figure 8h, the calculated d-band center of NiOOH þ SeO 4 is closer to the Fermi level than that of pristine NiOOH, indicating a stronger bonding strength with the OER intermediates.The electron accumulation on the Ni site also indicates the enhanced adsorption of O*, facilitating the adsorption of the OER intermediates.Sulfates also exhibited a similar effect and it was reported that SO 4 2À ions leaching in the SO 4 2À decorated-NiFe (oxy)hydroxide catalyst could accelerate the electrochemical reconstruction of forming active NiFeOOH species. [117]The residual SO 4 2À adsorbed on the surface of NiFe (oxy)hydroxide could adjust the adsorption energy of OER intermediates and stabilize the OOH*, thus enhancing the OER performances.
The phosphate can also provide the electrostatic force to repel Cl À ions.Sun et al. prepared NiCoFe phosphide as an OER catalyst.They found that a large amount of phosphate formed on the surface of anode after long-term electrolysis, which could effectively block chloride anions to avoid corrosion. [118]Lin et al. also found that the electrodeposited iron phosphate layer on the surface of CaFeO x could attenuate the production of corrosive hypochlorite from chloride oxidation. [119]In particular, Ren et al. constructed a core-shell CoP-CoP 2 @FeOOH composite for efficient seawater electrolysis. [120]The negatively charged CoP-CoP 2 core was highly conductive and thermodynamically stable, which enhanced chloride corrosion resistance and chemical stability.Meanwhile, the FeOOH shell was highly active for OER, and the combination of the core and shell led to outstanding performance.CoP-CoP 2 @FeOOH only required overpotentials of 283 and 337 mV to attain current densities of 100 and 500 mA cm À2 in 1 M KOH seawater electrolyte, respectively, and can sustain 80 h of continuous testing at current densities of 100 and 500 mA cm À2 in alkaline seawater without forming any hypochlorite.Phosphate has also been reported as a typical inorganic coating for metal corrosion prevention.123] The presence of a negative-charge layer significantly improves the chlorine resistance of the catalysts.Nevertheless, there are also some deficiencies in the electrostatic repulsive theory in explaining the corrosion resistance.First, the cations (e.g., Na þ and K þ ) in the electrolyte may be attached to the surface of the electrode through the attractive force, leading to a shield effect and greatly weakening the electrostatic force repulsive between the Cl À and the anions.At the same time, the issue of whether the OH À ions are simultaneously blocked by the electrostatic repulsive force and whether the OER activity would be influenced by are lack of research.In addition, further research on the formation mechanism and structural stability of polyatomic anion intercalated structures is needed.

Amorphous (Oxy)Hydroxides Construction
Transition metal hydroxides generally show superior catalytic activity for OER due to their layered structure, high specific surface area, and tunable electronic structures.126] During OER process, amorphous electrocatalysts can flexibly self-regulate the exposed surface according to the electrocatalytic conditions. [127]In addition, due to the disordered structure, foreign atoms can be easily incorporated into the amorphous phase to form hybrid catalysts. [128]Benefiting from the above-mentioned features, amorphous (oxy)hydroxide layers are reported to show excellent structural stability and resistance to chlorine corrosion at the electrochemical environment in seawater.
Amorphous LDH Layers: Layered double hydroxide (LDH), a type of anionic clay mineral with a layered structure, has demonstrated an extraordinary performance for OER in alkaline saline electrolytes. [129,130]For example, Dresp et al. reported crystalline NiFe-LDH as anode materials in an anion exchange membrane-based electrolyzer with an alkaline NaCl-containing electrolyte.The system could achieve stable operation at a current density of 200 mA cm À2 at 1.6 V for 100 h in 0.5 M KOH þ 0.5 M NaCl.In addition, no Cl 2 was detected by online selectivity measurements during the electrolysis. [131][134] Zhou et al. synthesized amorphous and highly crystalline NiFe-LDHs on nickel foam, respectively, with the former one containing many nanometer-sized crystalline facets and the latter one featuring much larger crystalline planes with fewer boundaries.They found that the adsorption behaviors of Cl À and OH À at these two types of NiFe-LDHs were notably different.(Figure 9a). [135]The XPS peak-fitting of Ni 2p 3/2 and 2p 1/2 orbitals (Figure 9b) revealed that the Ni 3þ /Ni 2þ ratio increased in partially crystalline NiFe-LDH, suggesting that amorphization converts some Ni 2þ sites to higher-valence state.Anion chromatography was further used to probe the adsorption of Cl À and OH À at the two types of NiFe-LDH.As shown in Figure 9c, the mass density of the adsorbed Cl À at the partially crystalline NiFe-LDH is lower than the highly crystalline one in both saline and alkaline saline solutions.Presumably, OH À preferentially adsorbed onto the Ni 3þ sites which were abundant at the boundaries/defects, whereas Cl À adsorbed more extensively onto the Ni 2þ at crystalline planes.These adsorption behaviors could be rationalized by Pearson's hard-soft acid-base (HSAB) principle.Ni 3þ sites showed stronger alkalinity thus preferring to adsorb OH À .Correspondingly, the partially crystalline NiFe-LDH showed an overpotential of 0.257 V at 500 mA cm À2 in alkaline saline solution, much lower than that of the highly crystalline counterpart (0.354 V) (Figure 9d).
Ren et al. synthesized partially amorphous boron-modified cobalt iron LDH (B-Co 2 Fe LDH) using a two-step process (Figure 9e,f ). [73]The as-prepared B-Co 2 Fe LDH demonstrated excellent OER activity with only overpotentials of 310 and 376 mV to drive the current densities of 100 and 500 mA cm À2 , respectively, and durability in 1 M KOH seawater (Figure 9g).Moreover, after electrolysis at 100 and 500 mA cm À2 for 100 h, no hypochlorite was detected, indicating its high OER selectivity (Figure 9h).Moreover, as shown in Figure 8i, the corrosion current density of B-Co 2 Fe LDH is merely 1.38 μA cm À2 , around half of that of Co 2 Fe LDH (2.4 μA cm À2 ).This higher corrosion resistance was attributed to the boron modification that  c) Ion chromatograms of Cl À before (black lines) and after (red lines) immersing partially and highly crystalline (blue lines) NiFe-LDHs in saline and alkaline saline solutions.The right panel is the corresponding mass density of adsorbed Cl À on the partially and highly crystalline NiFe-LDHs upon immersion and after 1 h of electrolysis in the alkaline saline solution.d) Polarization curves recorded at partially and highly crystalline NiFe-LDHs.Reproduced with permission. [135]Copyright 2021, American Chemical Society.e) Schematic of the formation of partially amorphous B-Co 2 Fe LDH.f ) HRTEM images of partially amorphous B-Co 2 Fe LDH.g) Polarization curves of B-Co 2 Fe LDH before and after 2000 CV scans.h) Colorimetric reagent testing result for hypochlorite production in the 1 M KOH seawater electrolyte after OER stability testing at 500 mA cm À2 for 100 h.i) Corrosion potentials and current densities of Co 2 Fe LDH and B-Co 2 Fe LDH catalysts in natural seawater.Reproduced with permission. [73]Copyright 2021, Elsevier.
extracted oxygen to create oxygen defects and increase the number of defective crystalline-amorphous interfaces.
The in situ Generated Oxyhydroxide Layer: Transition metal oxyhydroxides (TMOOHs) in situ evolved on the surface of anode not only improve the OER activity but also endow the resistance to chloride corrosion. [136,137]Yin et al. developed a catalyst of Ni 3 S 2 /Co 3 S 4 (NiCoS) nanosheets grown on Ni foam for alkaline seawater electrolysis. [138]It was found that a thin layer of NiCoS would transform into amorphous Ni/Co (oxy)hydroxide species (NiCoOOH), while the S residues in situ doped into the surface of NiCoOOH during OER process (Figure 10a).This layer provided more active sites for OER and protected the anode from Cl À corrosion, resulting in high OER activity and selectivity.The catalyst could maintain a stable current density during  [138] Copyright 2021, Elsevier.d) HRTEM images of Ni 3 S 2 /Fe-NiP x /NF after OER test.e) Atomic ratio variation of Ni 3 S 2 /Fe-NiP x /NF before and after reconstruction measured by XPS.f ) Measured and calculated volume of O 2 for Ni 3 S 2 /Fe-NiP x /NF in 1 M KOH þ seawater.Reproduced with permission. [139]Copyright 2022, Wiley-VCH.g) HRTEM image of the NiMoN@NiFeN catalyst after OER tests.h) Optical images of a post-OER NiMoN@NiFeN sample (left), and a fresh NiMoN@NiFeN sample before and after 1-day soaking in natural seawater (right).i) Measured and calculated volume of H 2 and O 2 for NiMoN@NiFeN.Reproduced with permission. [140]Copyright 2019, Springer Nature.j) HRTEM image of MOEE.k) CP curves with NiFe and MOEE as the electrode in 1 M KOH þ 0.5 M NaCl.l) Electrocatalytic efficiency of MOEE toward the OER in 30 wt% KOH with 0.5 M NaCl electrolytes at 500 mA cm À2 .Reproduced with permission. [141]Copyright 2021, Wiley-VCH.100 h chronopotentiometry measurement in 1 M KOH þ 0.5 M NaCl and 1 M KOH þ seawater.Moreover, the electrolyzer constructed by NiCoS anode and NiMoS cathode exhibited an excellent overall seawater splitting performance with the voltages of 2.04 and 2.11 V to drive a current density of 1,000 mA cm À2 in the alkaline simulated and alkaline natural seawater (Figure 10b), respectively, and an OER Faradaic efficiency close to 100% (Figure 10c).
The existence of a thin oxyhydroxide layer on the surface of metal sulfides is also beneficial to reducing the leaching of sulfur elements, thereby slowing the structural changes of the catalyst and improving the corrosion resistance.For instance, Luo et al. reported that Ni 3 S 2 /Fe-NiP x /NF catalyst displayed outstanding OER long-term stability for nearly 225 h in simulated alkaline seawater. [139]This could be attributed to the in situ formation of Ni(OH) 2 /Ni(Fe)OOH surface layer during the rapid reconstruction process, consisting of numerous low-crystalline nanoparticles with abundant grain boundaries, which can effectively suppress the leaching of elemental S and contributes to the remarkable durability (Figure 10d,e).Additionally, Ni 3 S 2 /Fe-NiP x /NF maintained a high OER selectivity during its electrolysis, demonstrating an OER Faradic Efficiency of 95.7% in 1 M KOH þ seawater (Figure 10f ).
In addition to transition metal sulfides, transition metal oxides and nitrides showed similar effects.Haik et al. reported a hierarchical nanostructure of Gd-doped Mn 3 O 4 nanosheets supported on CuO-Cu(OH) 2 arrays (Gd-Mn 3 O 4 @CuO-Cu(OH) 2 ). [79]The in situ developed Cu hydroxide layer not only increased active sites by creating the electrophilic sites on the exposed surface but also hindered the adsorption and diffusion of Cl À ions toward the anode, as well as prevented the metal dissolution from the substrate.Ren et al. developed a 3D core-shell structure assembled by uniformly decorating NiFeN nanoparticles on NiMoN nanorods supported on Ni foam (NiMoN@NiFeN). [140]The inner NiMoN nanorods mainly served as a highly conductive substrate and provided large surface area.The outer NiFeN nanoparticles were partially oxidized to form amorphous NiFe oxide and NiFe oxy(hydroxide) layers during OER process (Figure 10g).These layers effectively slowed down chloride corrosion and improved electron transfer from the NiFeN core to the oxidized species (Figure 10h).Coupled with NiMoN cathode in alkaline natural seawater electrolyte, the electrolyzer could achieve current densities of 500 and 1000 mA cm À2 at low voltages of 1.608 and 1.709 V and operated stably at 500 mA cm À2 for 100 h.The corresponding Faradaic Efficiency on the anode is about 97.8%, indicating the high selectivity of NiMoN@NiFeN for OER (Figure 10i).
Amorphous oxyhydroxides generated from transition metal borides have also been proved to be effective for chlorine corrosion resistance.Zou et al. constructed a multilayered OER electrode (MOEE) composed of the surface oxidized NiFeB x layer, the NiFeB x interlayer, and the NiFe alloy substrate. [141]The oxidized NiFeB x layer was amorphous with the insertion of metaborate, promoting the generation and stabilization of catalytic active phase γ-(Ni,Fe)OOH, which was inert to the adsorption of Cl À ions according to DFT calculations (Figure 10i).The weak Cl À adsorption effectively hindered the subsequent elementary reaction step of COR, thus improving the OER selectivity, as well as improved Cl À corrosion resistance.In the stability tests, the applied voltage drops rapidly in 10 h at NiFe electrode accompanied by the appearance of a number of black solid powders on its surface, while MOEE worked stably for 70 h and well maintained its initial shape with %100% oxygen production efficiency (Figure 10k,l).

Structural Regulation by Heteroatoms Doping and Vacancies
Doping control is an effective method for engineering the inherent properties of OER catalysts such as electronic structure, atomic coordination number, carrier concentration, conductivity, etc.In addition, the interaction between dopants and active sites is conducive to adjusting the adsorption energy for different reaction intermediates.As a result, the active sites can selectively bind with the oxygen-containing intermediates, thus regulating the reaction pathway.Moreover, the introduction of vacancies through doping and the synergy between defects and dopants can enhance the electrocatalytic performance.
Noble metal-based (Ru and Ir) oxides are highly catalytic active for both COR and OER.The coordination environment of the active center dictates the different reaction path.][144] For instance, Krtil et al. reported that Zndoped RuO 2 (Ru 0.8 Zn 0.2 O 2 ) exhibited higher OER selectivity than RuO 2 in Cl À -containing solutions due to the tailored structure of active sites (Figure 11a). [144]The electrocatalytic activity of the rutile-type RuO 2 on (110) facet is related to two types of surface oxygen atoms along [001] direction.The first one formed a bridge between two Ru atoms, which was inactive, and the second type was only involved in one Ru─O bond, which was catalytically active both for the formation of OER and COR intermediates.The introduction of Zn 2þ into the RuO 2 lattice altered the perfect sequence of Ru cation stacking along [001] direction and led to the atom re-arrangement.As shown in Figure 11b, the peak at r % 3.5 Å significantly decreases in Zn spectrum (line a) than in the Ru spectra (lines b and c), indicating a significant deviation from the ideal rutile structure around Zn atoms, resulting in the rearrangement of the neighboring metal atoms in [111] direction.Figure 11c,d shows the electrocatalytic activity of the nanocrystalline RuO 2 and Ru─Zn─O in 0.1 M HClO 4 þ 0.15 M NaCl, revealing the change of the catalysis selectivity from the CERdominated to OER-dominated process.This change may be explained as follows: Zn 2þ ions in the rutile lattice can prevent the formation of peroxo bridges between two O cus sites, which represent the rate-limiting intermediates for both CER and OER.As a result, the chlorine-evolution process was suppressed.
Doping generally introduces vacancies as well as increases the valence states of the active metal sites, which also significantly inhibits the Cl À ions adsorption. [145]To improve OER selectivity, stabilizing -OOH formation or destabilizing -OCl formation on the surface can facilitate OER over active chlorine species formation reactions (ACSFR).Ramani et al. reported that the -OOH intermediates on the surface of RuO 2 could be stabilized by metal doping with lower d-electrons than the host cation. [146]ccordingly, the as-synthesized Pb 2 Ru 2 O 7-x electrocatalyst displayed much higher OER selectivity than RuO 2 .The lower ACSFR selectivity for Pb 2 Ru 2 O 7-x at 1.7-1.85V versus RHE compared to RuO 2 could be observed during neutral seawater electrolysis (Figure 12a).Additionally, with the increase of pH, the OER activity and selectivity of Pb 2 Ru 2 O 7-x further improved (Figure 12b), indicating a better OER-selective performance.XPS analysis indicated that Pb 2 Ru 2 O 7-x had a higher oxidation state of Ru and a large number of oxygen vacancies generated at the surface, originating from the charge imbalance between Pb (II/IV) and Ru (IV/V) (Figure 12c,d).The higher oxidation states of surface Ru atoms would stabilize the OER intermediates through the formation of stronger metal-oxygen bonds, and the oxygen vacancies would promote water dissociation with lattice oxygen vacancy quenching and enhanced bulk conductivity, thus leading to the excellent OER-selective electrocatalytic activity of Pb 2 Ru 2 O 7-x .
[149] For instance, Liu et al. prepared the Ag-doped NiFe LDH (Ag/NiFe LDH) toward seawater oxidation and investigated the influence of Ag doping on the electrochemical performance. [150]Remarkably, Ag/NiFe LDH could deliver a current density of 1,000 mA cm À2 at overpotentials of 293 and 303 mV in alkaline simulated seawater and in alkaline natural seawater, respectively, far better than the pristine NiFe LDH.Ag/NiFe LDH also exhibited an impressive stability in alkaline natural seawater with a stable current density of 1,000 mA cm À2 for 1,000 h, while the pristine NiFe LDH obviously decayed after 40 h.It was found that Ag doping promoted the formation of an amorphous structure of the NiFe LDH and increased the oxidation state of nickel, which made it more conducive to adsorbing hydroxide ions.In addition, the introduced Ag lowered the corrosion current and facilitated the phase stability of NiFe LDH, which prevented a significant material structure change and reconstruction during long-term operation, further ensuring the OER stability.Mu et al. constructed a vanadium-doped CoP/Ni 2 P heterostructure coupled with ultralow Ru (RuV-CoNiP/NF) and found that V exhibited three kinds of valence states of þ3, þ4, and þ5. [151]The multivalence state was more favorable to the OER, and the electronic synergistic effect of V doping with other metal atoms further promoted the charge transfer.Benefiting from this advantage, the cell composed of RuV-CoNiP/NF cathode and anode required a small voltage of 1.809 V to achieve 100 mA cm À2 in alkaline seawater.Chang et al. reported a catalyst of the iron and phosphor dual-doped nickel selenide nanoporous films (Fe, P-NiSe 2 NFs) on carbon cloth. [152]They found that Fe-doping could improve the electrical conductivity and increase the OER selectivity at the anode by forming higher valence Ni.P-doping plays a dual role in increasing electrical conductivity and preventing the dissolution of selenide by forming a thin amorphous passivation layer containing P-O species.Remarkably, the Fe, P-NiSe 2 NFs catalyst showed excellent activity, selectivity, and stability toward natural seawater electrolysis with a current density of 0.8 A cm À2 at 1.8 V for over 200 h and an O 2 FE of 92%.
Xia et al. verified that Fe doping in CoSe 2 could regulate the local spin state of Co species, leading to a more, e.g., orbital electron filling state and a larger degree of overlap between, e.g., Reproduced with permission. [144]Copyright 2010, Wiley-VCH.
orbital and OH À adsorbate, which was beneficial for promoting charge transfer and formation of oxygenated species. [153]Luo et al. proved that the synergistic effect of sulfur vacancies and Ni doping in CuS x could reduce the energy barrier of OER rate-determining step. [154]In detail, the dissociation-release process of *OH must overcome this strong adsorption of *OH on the ideal CuS (103) slab, and the introduction of S vacancy and the neighboring Ni atom could attract the *OOH intermediates to promote the dissociation, which synergistically improved the overall performance.
In addition, the electrostatic potential barrier formed by the electrical double layer greatly affects the charge transfer.The OER catalytic activity as well as the Cl À corrosion rate can be controlled by the electrostatic potential barrier at the surface, which is a function of chemical potential. [155]By introducing low valence-state ions as dopants, the Fermi level can be controlled and a high concentration of negatively charged space near the surface of OER catalysts would be formed, which largely affects the adsorption energy of Cl À ion.Kim et al. reported the zinc-doped nickel-iron (oxy)hydroxide (NFZ-PBA-H) for seawater oxidation, exhibiting outstanding catalytic activity, stability, and selectivity. [156]The work function (ϕ) of NF-PBA-H (6.90 eV) was substantially increased when Zn-dopants were introduced (8.30 eV) and the Fermi level was down-shifted accordingly (Figure 13a,b).Therefore, Zn served as a p-type dopant for nickel-iron (oxy)hydroxides and created a negatively charged electrocatalyst before hold-test.Reproduced with permission. [146]Copyright 2020, American Chemical Society.
surface for NFZ-PBA-H.The built-in electric field by the surface charge depletion induced the electrostatic repulsive force on Cl À and hindered the migration of Cl À ions toward the surface of NFZ-LDH, which was expected to lower the corrosion reaction rate by a factor of exp (Àϕ BB /kT) in seawater (Figure 13c).The doping atoms can also synergize with the original active sites and affect the OER pathway.Wang et al. incorporated strong proton adsorption (SPA) cations into the Co 3 O 4 framework and found that Co 3-x Pd x O 4 showed an excellent OER activity with a current density of 10 mA cm À2 at the overpotential of 370 mV and a high OER selectivity in pH-neutral seawater. [157]he superior performance can be attributed to the synergetic effects between Co active sites and SPA cations (Figure 13d).Co atoms not only served as active sites and adsorbed OER intermediates but also cooperated with SPA cations to favor the dissociation of water molecules, thus enhancing OER performance and suppressing the CER in neutral seawater conditions.

Heterostructures and the Interface Engineering
Heterostructures with abundant interfaces can provide highly exposed active sites, facilitate the electron/mass transfer, and promote gas release from the surface. [158,159]In addition, it was reported that heterostructures may prefer to adsorb OH À rather than Cl À . [114]Therefore, the construction of heterostructure is beneficial to achieving the high intrinsic activity and high OER selectivity. [160,161]ee et al. synthesized cobalt-cobalt oxide heterostructures incorporated with ultralow loading of Rh heteroatoms on 1D Cu nanowires grown on Cu foam (1D-Cu@Co-CoO/Rh), exhibiting an improved catalytic performance for OER in simulated seawater environments in comparison with single-phase counterpart. [162]The improved performance of 1D-Cu@Co-CoO/Rh originated from its enriched reactive sites and enhanced charge transfer ability due to the synergistic effects derived from the The work function for NFZ-PBA-H.c) The scheme of surface space charge effects on the concentration of Cl À ion near the surface of NFZ-LDH.Reproduced with permission. [156]Copyright 2023, Elsevier.d) Proton-adsorption-promoting strategy in pH-neutral seawater oxidation reaction.Reproduced with permission. [157]Copyright 2023, Wiley-VCH.
uniform Rh atoms doping and the Co-CoO heterostructures.Maccato et al. constructed a series of heterostructures using MnO 2 or Mn 2 O 3 and Co 3 O 4 or Fe 2 O 3 as components. [163]mong them, Co 3 O 4 /MnO 2 showed the best OER performance in simulated alkaline seawater, which might be related to the different electronic band levels at the interface.Specifically, Co Charge transfer between different metal sites at the interface would lead to charge redistribution and modify the adsorptionfree energies of the reactants or intermediates. [164]Li et al. constructed a Cr doped-FeOOH/Fe 3 O 4 heterostructure (Fe(Cr) OOH/Fe 3 O 4 ), in which the nanosized Fe(Cr)OOH and Fe 3 O 4 crystalline grains combine closely and form a large number of the heterogeneous interface (Figure 14a,b). [165]The charge redistribution at the interface and the coupling of FeOOH and Fe 3 O 4 decreased the overpotentials of OER by reducing the energy barrier at FeOOH/Fe 3 O 4 (Figure 14c).Although the Cr site was relatively adverse to the formation of *OH, it could efficiently promote this process on the neighboring Fe atom, and thus Fe(Cr)OOH/Fe 3 O 4 with Fe sites presented the smallest η value of 0.55 V (Figure 14d).The calculated density of states (DOS) Reproduced with permission. [165]Copyright 2021, Elsevier.
suggested that Fe(Cr)OOH/Fe 3 O 4 catalyst presented a higher density of electronic states in the conduction band, indicating an enhanced electrical conductivity (Figure 14e).Therefore, Fe(Cr)OOH/Fe 3 O 4 could achieve a large current density of 500 mA cm À2 at the overpotential of only 278 mV in alkaline seawater and work stably at 100 mA cm À2 for 100 h.
Zhu et al. constructed the Ru-CoMoS x heterostructure as a highly active OER catalyst in alkaline media, in which the crystalline Ru nanoparticles with strong oxygen-adsorption were incorporated in the amorphous CoMoS x with weak adsorption. [166]The strong electronic interaction between Ru and CoMoS x resulted in a positive shift of d-band center for the Co site, enhancing the *OOH adsorption at Co, thereby reducing the required overpotential for OER rate-determining step.Similarly, Zhu et al. incorporated Ru clusters in Ni 3 N substrate (cRu-Ni 3 N) for seawater electrolysis. [167]They found that the cRu-Ni 3 N heterostructure greatly induces the charge density redistribution with electrons assembled at the interface between cRu and Ni 3 N, thus leading to the change of rate-determining steps and reduction of theoretical overpotential.
In addition, the interaction can generate abundant oxygen vacancies at the interface, facilitating the catalytic activity and ameliorating the corrosion resistance.Cong et al. developed a Co 0.4 Ni 1.6 P-CeO 2 heterostructure and found that it could drive a current density of 100 mA cm À2 at an overpotential of 345 mV in solution of 1 M KOH þ 0.5 M NaCl. [168]The experimental and theoretical investigations revealed that the heterostructure could optimize the electronic structure of Co 0.4 Ni 1.6 P-CeO 2 , leading to a charge redistribution and a generation of a large number of oxygen vacancies.Furthermore, according to ultraviolet photoelectron spectroscopy (UPS) measurements, the d-band center of Co 0.4 Ni 1.6 P was uplifted after interaction with CeO 2 , and the work function (W f ) of Co 0.4 Ni 1.6 P-CeO 2 was lower than that of Co 0.4 Ni 1.6 P, indicating its higher interfacial charge transfer efficiency.Jiang et al. also constructed the NiFe LDH/FeOOH heterostructure for overall alkaline simulated seawater splitting. [169]It required a voltage of only 1.55 V to achieve a current density of 10 mA cm À2 and worked steadily for 105 h at 100 mA cm À2 .It was believed that the heterostructure could not only facilitate the formation of active NiOOH species for OER but also create more oxygen vacancies.Additionally, the existence of FeOOH contributed to preventing the occurrence of chlorine oxidation reactions on NiFe LDH.
Yang et al. developed an interface-rich NiO-Fe 2 O 3 layer on Ni foam (noted as FNE-300), which only required an overpotential of 291 mV to yield a current density of 1,000 mA cm À2 . [170]The activation energy (E a ) of FNE-300 was much lower than RuO 2 , indicating the thermodynamical superiority (Figure 15a), which was attributed to its unique electronic structure.As shown in Figure 15b, the unpaired electrons in the π symmetry (t 2g ) d-orbitals of Fe 3þ can interact with oxygen intermediates through π-donation, while the fully occupied t 2g d-orbitals of Ni 2þ result in e À -e À repulsion between the oxygen intermediates and Ni 2þ .The optimized interaction between Fe and oxygen intermediates would induce a decrease in the adsorption barrier of oxygen intermediates (Figure 15c).Bader charge analysis (Figure 15d) revealed the accumulation and depletion of the electrons near the heterogeneous interface, while the DOS highlighted the enhancement of carrier concentration at the Fermi level, promoting faster charge transfer dynamics in OER (Figure 15e).
Incorporating active sites into substrates with high resistance to corrosion can also construct heterostructures, even if their lattice parameters may not match.This approach could effectively improve the OER activity and the resistance to chlorine corrosion. [171,172]Strong interactions between the metal and the substrates prevent the aggregation of metal sites, increase the exposure of active sites, and stabilize metal atoms, thereby enhancing their catalytic activity and stability.The construction of heterostructure also reduces the concentration of Cl À at OER active sites by introducing phases that are inert to Cl À adsorption, suppressing the corrosion and dissolution.
Yang et al. utilized a pre-embedding and sequential reduction process to load highly dispersed Pt and Ru nanoparticles into corrosion-resistant Ni-Mo alloys (Figure 16a). [173]Highly dispersed Pt and Ru nanoparticles provided rich active sites exposed to the electrolyte, while Ni-Mo alloy improved the corrosion resistance of the catalyst (Figure 16b).As a result, the as-obtained Pt/Ni-Mo and Ru/Ni-Mo catalysts showed overpotentials of 42 mV for HER and 420 mV for OER at the current density of 2,000 mA cm À2 in 1 M KOH solution, respectively.More importantly, Pt/Ni-Mo catalyst could operate well in saline-alkaline water at 2,000 mA cm À2 and maintain a long durability up tp140 h.
Haq et al. fabricated an interface-rich catalyst composed of Au nanocrystals (NCs) decorated Gd-Co 2 B nanoflakes embedded in TiO 2 nanosheets on Ti foil for efficient seawater electrolysis (Au-Gd-Co 2 B@TiO 2 , Figure 16c). [174]In this multiphase catalyst, TiO 2 nanosheets provided a strong skeleton with a large surface area and high corrosion resistance, while the outer layer Gd-Co 2 B nanoflakes acted as a protective armor, contributing to high stability.The strong interaction between Au NCs and Gd-Co 2 B@TiO 2 prevented structural rearrangements and leaching of Au to the electrolyte during OER.Moreover, the Au NCs triggered the oxidation of Co and enhanced its binding affinity with OH intermediates, thus leading to a high geometric activity of 2 A at an overpotential of 370 mV and an improved OER selectivity in 1 M KOH seawater.Furthermore, the as-fabricated interface-rich catalyst also operated steadily for 200 h without hypochlorite formation.Zhou et al. developed a Ru/NiFeOOH on NiFe foam catalyst for OER (Figure 16d), which only required 285 mV to achieve the current density of 500 mA cm À2 and maintained good stability at 100 mA cm À2 for 100 h. [175]The homogeneous distribution of Ru nanoparticles in NiFeOOH was conducive to exposing active sites while the charge transfer from Ni/Fe to Ru due to the difference in electronegativity stabilized the insertion of Ru and further optimized the OER performance.Furthermore, the NiFeOOH nanosheets could effectively prevent the severe Cl À corrosion of NiFe substrate.
Enkhtuvshin et al. constructed a heterostructure composed of Ni 3 Fe alloy and amorphous Ni-Fe oxyhydroxide phases (c-NF// a-NF-LDH NS, Figure 16e). [176]Nickel-iron (oxy)hydroxide layer and Ni 3 Fe formed a metal and n-type semiconductor Schottky barrier, tuning the surface potential and repelling negatively charged Cl À ions.The calculated Cl À adsorption energy (ΔG Cl* ) on the pristine NiFe-LDH was À3.148 eV, leading to inevitable chloride corrosion at the surface of NiFe-LDH (Figure 16f ).In contrast, the ΔG Cl* at Ni-Fe alloy was significantly lower (À1.7%À1.0 eV, Figure 16g), indicating that adsorption of Cl À was inhibited on the surface of Ni 3 Fe to some extent.Therefore, the NiFe alloy served as a Cl À -blocking material that prevented chloride corrosion at the NiFe-LDH phase.As a result, the c-NF//a-NF-LDH NS exhibited a high Cl À resistance and high selectivity for OER, with an OER FE of 97.4%, and no reactive chloride species were detected after 24 h of electrolysis at 100 mA cm À2 .
Charge redistribution in heterojunctions can also regulate the local pH environment and inhibit chloride ion adsorption.Guo et al. created a stable Cr 2 O 3 -CoO x catalyst for direct electrolysis of real seawater (Figure 17a), in which a local alkalized microenvironment was created by adding a hard Lewis acid layer (Cr 2 O 3 ) onto CoO x . [177]As shown in Figure 17b,c, the normalized OER activity and the surface pH on Cr 2 O 3 -CoO x were much higher than those of CoO x in natural seawater.Figure 17d illustrates the estimated concentration of excess OH À required to "resist" Cl À and the actually measured concentration of OH À on electrode surface.For CoO x , the concentration of surface OH À is lower than the required.In contrast, for Cr 2 O 3 -CoO x , the concentration of surface OH À is nearly two orders of magnitude higher than the required concentration.This significant difference could be attributed to the gap of water dissociation energy barrier (G a ).As shown in Figure 17e,f, G  It can be seen from the above cases that the proper construction of the heterojunction can promote the charge separation, increase the carrier concentration, adjust the interaction between active sites and OER intermediates, and regulate local charge distribution and pH values, thus improving OER activity and selectivity.Reproduced with permission. [169]Copyright 2021,American Chemical Society.

Ingenious Anode Design beyond Catalyst Materials
In addition to the modification of the anode catalysts, the investigations on the effects of electrolyte and overall electrolyzer structural engineering on the anode stability have also been reported.Ma et al. reported that the addition of sulfates in the alkaline saline electrolyte could effectively alleviate the chloride corrosion on the anode (Figure 18a). [178]They first selected pure NF as the anode in highly corrosive electrolytes (1 M NaOH þ 2.5 M NaCl) with and without additional proportional SO 4 2À .The NF could only work within 50 min in the electrolyte without Na 2 SO 4 while it could stably operate up to 130 min in electrolytes with different proportional Na 2 SO 4 (Figure 18b).The Tafel plots of pure NF in electrolytes with different proportions of Na 2 SO 4 were presented in Figure 18c.NF showed higher E corr and lower I corr with the existence of Na 2 SO 4 , indicating weaker corrosion on the surface.Theoretical calculations further revealed that the additive SO 4 2À in electrolyte would be preferentially adsorbed on the anode surface, which repelled Cl À in bulk phase by electrostatic repulsive force.In contrast, the OH À concentration near the electrode surface showed no obvious difference, ensuring that the addition of SO 4 2À would not affect the OER activity, which might be attributed to the strong hydrogen bonding force between the OH À and NF surfaces (Figure 18d).
The repulsive effect of additive SO 4 2À in electrolyte was also observed in nickel-iron layered double hydroxide (NiFe-LDH) nanoarrays/NF anode, which could work stably at 400 mA cm À2 for 500-1,000 h in simulated and real seawater, 3-5 times higher than that works in electrolyte without SO 4 2À Figure 16.a) Reduction potential-dependent sequential reduction of PGMs in matrix.b) HRTEM image of Pt/Ni-Mo.Reproduced with permission. [173]opyright 2020, Wiley-VCH.c) The fabrication of Au-Gd-Co 2 B@TiO 2 .Reproduced with permission. [174]Copyright 2022, Elsevier.d) Schematic illustration for BSI procedure conducted in seawater for the synthesis of Ru/NiFeOOH/NFF.Reproduced with permission. [175]Copyright 2023, Elsevier.e) Schematic of the surface reconstruction process of pristine NF-LDH catalyst transforming to a mixture of nickel-iron (oxy)hydroxide and metallic alloy phases.f,g) (Figure 18e). [178]Li et al. prepared an OER catalyst of NiCoFe phosphide arrays on Ni foam and found that increasing NaOH concentration in the electrolyte greatly reduced the concentration of Cl À , thus eliminating anode corrosion and chloride oxidation, as well as generating NaCl crystal during electrolysis.This may be due to a common-ion (Na þ ) effect in the mixed solution.Namely, the product of the maximum concentrations of two dissociated ions (one positive and one negative) is a constant, increasing the concentration of one ion would decrease the maximum solubility of the other ion.The seawater electrolysis system worked stably in the 6 M NaOH þ NaCl (Sat.%2.8 M) electrolyte, whereas the potential constantly surged during the 60 min electrolysis in the 1 M NaOH þ NaCl (Sat.%5.0 M) electrolyte, indicating that the presence of high alkalinity could slow down the activity attenuation. [118]nion exchange membrane electrolyzer could be directly used for seawater splitting, which is composed of anode, cathode, anion exchange membrane, and porous transport layer. [136,179,180]This device can be operated in an alkaline solution and prevent the mixing of O 2 and H 2 , but is deficient in suppressing the COR and maintaining the safe and long-term operation of seawater electrolysis.The OER selectivity and high-power electrolysis performance can be improved through the innovative structural design of electrolyzers on the basis of traditional anion exchange membrane devices.
To improve the Cl À resistance of electrolyzer, asymmetric electrolyte feed devices for direct seawater electrolysis were explored.Dresp et al. proposed an asymmetric device, in which the cathode chamber was filled with a solution of 0.5 M NaCl, while the anode chamber was filled with 0.5 M KOH (Figure 19a). [181]An anion exchange membrane coated with crystalline Ni 0.66 Fe 0.34 -LDH that allowed the penetration of OH À and Cl À was utilized as anode material coupled with a commercially available cathode.This device allowed for direct neutral seawater electrolysis at the cathode while circulating pure KOH electrolyte at the anode.Compared with symmetric mixed electrolytes, the asymmetric feeds on the cathode and anode showed superior cell performances, even the saline seawater-only feed at the cathode outperformed the mixed electrolyte feed.In addition, at varied potentials for the asymmetric device, the FE of OER remained almost identical and close to 100%, indicating no significant Cl À oxidation.Zhang et al. investigated the influence of different feeding modes (Figure 19b). [75]Compared to the symmetric electrolyzers (III and IV), the device I and II showed higher FE for OER.Frisch et al. also designed an asymmetric-feed electrolyzer, in which the alkaline seawater and dry N 2 are introduced into the cobalt chalcogenides-based anode chamber and nickel phosphide-based cathode chamber, respectively. [182]It could achieve a current density of 1.0 A cm À2 below the cell voltage of 2.0 V.More importantly, the asymmetric feed exhibited a  [177] Copyright 2023, Springer Nature. .Reproduced with permission. [178]Copyright 2021, Wiley-VCH.
high corrosion resistance with a stable current density of 400 mA cm À2 up to 100 h.
Marin et al. designed a bipolar membrane water electrolyzer (BPMWE) device with asymmetric saline electrolyte conditions to generate H 2 and O 2 at high current densities. [183]Different from monopolar proton exchange membrane water electrolyzers (PEMWE) or anion exchange membrane devices, the BPMWE is comprised of a cation exchange layer (CEL) and an anion exchange layer (AEL) (Figure 19c).The CEL restricts the Cl À crossover to the anode due to cation transport selectivity while the AEL provides a local alkaline pH at anode, resulting in ultra-low Cl À transport rates and high OER selectivity (Figure 19d).As a result, the FE of Cl À oxidation to corrosive OCl À at the anode was limited to only 0.005%, much lower than that of PEMWE (%10%).In addition, the lifetimes of BPMWE significantly improved compared with the PEM devices under real seawater conditions.Therefore, the asymmetric electrolyzer design represents a new trend in seawater electrolysis devices, which is worthy of further research.
Xie et al. developed a novel membrane-based seawater electrolyzer based on a liquid-gas-liquid phase transition water migration mechanism, which could achieve an in situ self-driven water purification and water electrolysis in a single system. [184]As shown in Figure 20a, the lab-scale electrolyzer was assembled by 10 layers.The electrocatalyst layers were fixed to the electrode plates, and the hydrophobic porous polytetrafluoroethylene (PTFE) membrane could transport water vapor and prevent liquid penetration as well as block the mixing of H 2 and O 2 .Concentrated KOH solution served as a self-dampening electrolyte (SDE).In this electrolyzer, liquid water evaporated on the seawater side and migrated across the membrane in the state of water vapor, then liquefied again by absorption in the SDE.This phase transition migration mechanism allows the in situ generation of pure water for electrolysis from seawater source with 100% ion-blocking efficiency, and the phase transition of water provides the driving force for migration without additional energy consumption.At proper conditions, the water migration amount induced by the interface pressure difference and the water consumption amount of electrolysis reaches a dynamic balance.The system could reach a stable "in situ water purificationelectrolysis" process, enabling continuous and efficient H 2 production from seawater.In addition, a scaled-up demo-type device at a 386 L h À1 H 2 generation scale was also fabricated (Figure 20b,  c), which consists of 11 cells with a total effective geometric surface area of 3696 cm 2 for each electrode.The scaled-up electrolyzer was operated stably for over 3200 h at 250 mA cm À2 for seawater electrolysis with an energy consumption of approximately 5.0 kWh Nm À3 H 2 (Figure 20d).No obvious ClO À , SO 4 2À and Figure 19.a) Schemes for AEM electrolyzers using independent electrolyte feeds with different electrolyte compositions.Reproduced with permission. [181]Copyright 2020, The Royal Society of Chemistry.b) Schematics of the electrolyzers using different electrolyte feeding modes.Reproduced with permission. [75]Copyright 2021, Wiley-VCH.c) BPMWE and PEMWE device schematics.d) Proposed key ion-transport effects for PEMWE (left), BPMWE (middle), and AEMWE (right), including diffusion (solid arrows), migration (dashed arrows), and Donnan exclusion effects (curved arrows) that dictate the transport of ions across the ion-exchange membranes.Reproduced with permission. [183]Copyright 2023, Elsevier.
Mg 2þ ions were detected in the SDE after long-term operation, and the concentrations of all three ions were four orders of magnitude lower than those in seawater, suggesting the 100% ion-blocking efficiency.Furthermore, the catalyst layer also maintained its original morphology after long-term electrolysis.The production of green hydrogen through the electrolysis of seawater shows many advantages.Through surface selective layer engineering, structural regulation by heteroatoms doping and vacancies, heterostructures and the interface engineering, the corrosion alleviation influence of the electrolyte, and ingenious structural design of the electrolyzer, the OER selectivity, activity, and chloride corrosion resistance could be obviously improved, especially at a large current density.More reported OER catalysts in terms of overpotential at different current densities, FE, and stability in alkaline (simulated) seawater were listed and compared in Table 2.  [184] Copyright 2022, Springer Nature.
costs and greater demands on the ionic corrosion resistance of electrolyzers; 3) The practical seawater electrolysis required a large current density (over 1,000 mA cm À2 ), and large amounts of bubbles would be involved.Accordingly, it is important to develop catalysts that exhibit strong interfacial binding force with the support and weak adhesion force with bubbles.Furthermore, more attention should be paid to developing the electrolyzer, especially cell designs to avoid COR at the anode and the development of durable anion exchange membranes and bipolar plates; and 4) The economic feasibility of industrial seawater electrolysis requires comprehensive consideration.For example, combining conventional freshwater electrolyzer with a matured desalination system, such as seawater reverse osmosis (SWRO) or solar-thermal seawater evaporation is cost-effective for seawater splitting, because the costs of desalination are insignificant in the whole system. [5,190]While the quality of water from SWRO may be not sufficient, the residual trace ions will poison the Pt-or Ir-based electrocatalysts in conventional proton exchange membrane or anion exchange membrane devices.Therefore, the factors of technical, environmental, and economic should be synthetically considered to identify the most suitable technology using multicriteria decision-making methods.

Figure 1 .
Figure 1.A brief description for the design of advanced anodes in seawater electrolysis.

Figure 2 .
Figure2.a) Schematic of the AEM pathway in alkaline media, the oxygen from the electrolyte is marked in golden color.b) Typical volcano curves plotted against ΔG O* À ΔG OH* for oxides.Reproduced with permission.[40]Copyright 2021, Wiley-VCH.c) Schematic of the formation of a chemical bond between an adsorbate valence level and the s and d states of a transition-metal surface.Reproduced with permission.[41]Copyright 2005, Springer Nature.d-f ) Schematics of three alternative pathways of LOM in alkaline media with different catalytic centers, the chemically active lattice oxygen involving OER and oxygen from the electrolyte are marked in blue and golden colors, respectively, and □ represents lattice O vacancy: d) Oxygen-vacancy-site mechanism (OVSM), e) single-metal-site mechanism (SMSM), and f ) dual-metal-site mechanism (DMSM).Reproduced with permission.[58]Copyright 2021, The Royal Society of Chemistry.

Figure 4 .
Figure 4. a) Crystallographic characteristics of α-, β-, γ-, ε-, δ-MnO 2 .Reproduced with permission.Reproduced with permission.[87]Copyright 2021, Wiley-VCH.b,c) Sketch of the IrO x /GC and MnO x /IrO x /GC catalysts.d) Top panel: currents for OER and COR.Lower panel: corresponding selectivities toward OER and COR.Data are plotted as a function of Q MnOx , the charge ascribed to the corresponding MnO x layer, which is an approximate indication of its thickness.e) XPS scans of Cl 2p spectral peaks.Reproduced with permission.[93]Copyright 2018, American Chemical Society.f ) Sketch of the Na|MnO x catalysts.g) Visible spectra of the electrolytes with KI, taken after electrolysis with a delivered charge of 40 C cm À2 at the indicated current densities.h) Plots of the OER efficiency.Reproduced with permission.[81]Copyright 2021, American Chemical Society.

Figure 5 .
Figure 5. a) Schematic of the improved stability and selectivity achieved by the deposition of CeO x .b) CV tests for NiFeO x and CeO x -coated NiFeO x electrodes.c) Electrocatalytic stability tests of the bare and CeO x -coated NiFeO x electrodes.d) Faradaic efficiency of O 2 during controlled current electrolysis at 10 mA cm À2 for the bare and CeO x -coated NiFeO x electrode in 1 M KOH solution in the presence of different kinds of reducing agents with the corresponding Stokes radii.Reproduced with permission.[97]Copyright 2018, Wiley-VCH.e) Schematic of the SiO x |Pt catalyst.f ) Permeability of Cl À in SiO x overlayers as a function of overlayer thickness, calculated from LSVs for SiO x |Pt in 0.5 M KHSO 4 þ 0.6 M KCl.g) LSVs for 4.8 nm SiO x |Pt.Reproduced with permission.[99]Copyright 2021, American Chemical Society.h) The model of a hydrated ion.i) The hydrated ion diameters and enthalpy of hydration of anions.Reproduced with permission.[98]Copyright 2020, The Royal Society of Chemistry.

Figure 8 .
Figure 8. a) Schematic of the Ni 3 catalyst.b) OER constant current activation of Ni 3 in 1 M KOH þ 0.5 M NaCl.The decrease in voltage that occurred between 3 and 4 h was due to the etching-passivation process.c) Raman spectra of Ni 3 and NiS x /Ni after 12 h activation in 1 M KOH þ 0.5 M NaCl.Reproduced with permission.[72]Copyright 2019, National Academy of Sciences.d) The fabrication process of the Se_NiFe_LDH electrode.e) Long-term stability measurements.f ) Mechanism of the stability improvement of the Se_NiFe_LDH electrode.Reproduced with permission.[115]Copyright 2020, Elsevier.g) Differential charge densities of NiOOH þ SeO 4 when an O atom is adsorbed on the adjacent Ni site.Yellow and green contours represent electron accumulation and depletion, respectively.h) The density of states of NiOOH and NiOOH þ SeO 4 regarding the Ni 3d orbitals.Reproduced with permission.[116]Copyright 2020, Wiley-VCH.

solution of 1 M
KOH þ 1 M NaCl.The lifespan even extended to approximately 600 h in 5 M KOH þ 2 M NaCl, indicating that the synergistic effect between the layers of Se_NiFe and NiFe_LDH plays an important role in improving long-term stability.The high-resolution XPS and secondary ion mass spectrometer investigation results revealed the intercalation of generated SeO x 2À (SeO 3 2À and SeO 4 2À

Figure 9 .
Figure9.a) The comparison of partially and highly crystalline NiFe-LDHs.b) Ni 2p XPS spectra of partially and highly crystalline NiFe-LDHs.c) Ion chromatograms of Cl À before (black lines) and after (red lines) immersing partially and highly crystalline (blue lines) NiFe-LDHs in saline and alkaline saline solutions.The right panel is the corresponding mass density of adsorbed Cl À on the partially and highly crystalline NiFe-LDHs upon immersion and after 1 h of electrolysis in the alkaline saline solution.d) Polarization curves recorded at partially and highly crystalline NiFe-LDHs.Reproduced with permission.[135]Copyright 2021, American Chemical Society.e) Schematic of the formation of partially amorphous B-Co 2 Fe LDH.f ) HRTEM images of partially amorphous B-Co 2 Fe LDH.g) Polarization curves of B-Co 2 Fe LDH before and after 2000 CV scans.h) Colorimetric reagent testing result for hypochlorite production in the 1 M KOH seawater electrolyte after OER stability testing at 500 mA cm À2 for 100 h.i) Corrosion potentials and current densities of Co 2 Fe LDH and B-Co 2 Fe LDH catalysts in natural seawater.Reproduced with permission.[73]Copyright 2021, Elsevier.

Figure 10 .
Figure 10.a) HRTEM images of the NiCoS after OER.b) LSV curves.c) Measured and calculated volume of H 2 and O 2 for NiCoS.Reproduced with permission.[138]Copyright 2021, Elsevier.d) HRTEM images of Ni 3 S 2 /Fe-NiP x /NF after OER test.e) Atomic ratio variation of Ni 3 S 2 /Fe-NiP x /NF before and after reconstruction measured by XPS.f ) Measured and calculated volume of O 2 for Ni 3 S 2 /Fe-NiP x /NF in 1 M KOH þ seawater.Reproduced with permission.[139]Copyright 2022, Wiley-VCH.g) HRTEM image of the NiMoN@NiFeN catalyst after OER tests.h) Optical images of a post-OER NiMoN@NiFeN sample (left), and a fresh NiMoN@NiFeN sample before and after 1-day soaking in natural seawater (right).i) Measured and calculated volume of H 2 and O 2 for NiMoN@NiFeN.Reproduced with permission.[140]Copyright 2019, Springer Nature.j) HRTEM image of MOEE.k) CP curves with NiFe and MOEE as the electrode in 1 M KOH þ 0.5 M NaCl.l) Electrocatalytic efficiency of MOEE toward the OER in 30 wt% KOH with 0.5 M NaCl electrolytes at 500 mA cm À2 .Reproduced with permission.[141]Copyright 2021, Wiley-VCH.

Figure 11 .
Figure 11.a) Versatile control of the selectivity was demonstrated by Zn substitution in RuO 2 (O red, Cl green, Zn blue, Ru white).b) k 2 -normalized EXAFS functions of Ru 0.8 Zn 0.2 O 2 material at (a) Zn K and (b) Ru K edges, c) RuO 2 at Ru K edge, and d) ZnO at Zn K edge.(c-d) Measured current density (black) and the corresponding chlorine (red) and oxygen (blue) evolution current density during anodic polarization of (c) RuO 2 and (d) Ru 0.8 Zn 0.2 O 2 electrodes in 0.1 M HClO 4 /0.15M NaCl solution.Reproduced with permission.[144]Copyright 2010, Wiley-VCH.

Figure 12 .
Figure 12. a) OER current efficiencies on Pb 2 Ru 2 O 7-x and RuO 2 at various potentials in 0.6 M NaCl.b) LSV curves of Pb 2 Ru 2 O 7-x and RuO 2 in neutral and alkaline SSW.c) XPS spectra of Ru-3p in RuO 2 and Pb 2 Ru 2 O 7-x electrocatalyst before hold-test.d) XPS spectra of O-1s in RuO 2 and Pb 2 Ru 2 O 7-xelectrocatalyst before hold-test.Reproduced with permission.[146]Copyright 2020, American Chemical Society.

Figure 13 .
Figure 13.a) Work function measurements using UPS for NF-PBA-H.b)The work function for NFZ-PBA-H.c) The scheme of surface space charge effects on the concentration of Cl À ion near the surface of NFZ-LDH.Reproduced with permission.[156]Copyright 2023, Elsevier.d) Proton-adsorption-promoting strategy in pH-neutral seawater oxidation reaction.Reproduced with permission.[157]Copyright 2023, Wiley-VCH.
3 O 4 / MnO 2 and Co 3 O 4 /Mn 2 O 3 can form the p-n heterojunctions, and electrons in MnO 2 or Mn 2 O 3 (n-type semiconductor) would flow to the Co 3 O 4 (p-type semiconductor).In contrast, Fe 2 O 3 /MnO 2 and Fe 2 O 3 /Mn 2 O 3 form n-n heterojunctions, and electrons would flow from the higher energy conduction band of MnO 2 or Mn 2 O 3 to the lower-energy Fe 2 O 3 .There is a larger energy difference (ΔE B ) between the valence band of Co 3 O 4 and the conduction band of manganese oxides than the ΔE B between the conduction bands of Fe 2 O 3 and manganese oxides, indicating that the separation of charge carriers could be promoted at the Co 3 O 4 and manganese oxides interfaces, resulting in higher OER activity in simulated alkaline seawater.
a on Cr 2 O 3 and CoO 2 is exothermic and endothermic, respectively.G a on Cr 2 O 3 is only 0.08 eV at 1.60 V RHE , facilitating water dissociation and leaving the surface covered with numerous OH* to form OH À in the electrical double layer (EDL).Meanwhile, the generated H* desorbed from the Cr 2 O 3 surface, oxidized to H, and combined with a water molecule nearby in the EDL to form H 3 O þ , which can be rapidly transferred out of the EDL under the drive of the external electric field.Therefore, the interface of Cr 2 O 3 -CoO x splits water molecules and captures hydroxyl anions, generating local alkalinity, facilitating OER process, and avoiding chloride attack on the electrodes.

Figure 15 .
Figure 15.a) The activation energy at an equilibrium state (overpotential = 0 V) of FNE300, Ni foam, and commercial RuO 2 .b) Schematic diagram of the interaction between Fe 3þ /Ni 2þ and oxygen intermediates.c) Bader charge analysis of Fe 2 O 3 and FNE300.d) Differential charge density distributions of FNE300 (blue and yellow regions represent electron depletion and accumulation, respectively).e) The calculated total DOS of NiO, Fe 2 O 3, and FNE300.Reproduced with permission.[169]Copyright 2021,American Chemical Society.

Figure 17 .
Figure 17.a) Overall seawater splitting performance of Cr 2 O 3 -CoO x in natural seawater, the inset: HAADF-STEM image of Cr 2 O 3 -CoO x .b) OER polarization curves of 6 (at%) Cr 2 O 3 -CoO x , CoO x , Fe-doped NiOOH, and RuO 2 catalysts in natural seawater.c) Measured pH values on CoO x and Cr 2 O 3 -CoO x anode surfaces at different potentials and current densities.d) Measured OH À concentration and theoretical concentration of excess OH À required to resist Cl À .e) Schematic of local alkaline microenvironment generation on Lewis acid-modified anode, which facilitates OER and inhibits chlorine chemistry.f ) DFT calculated energy barrier of water dissociation on CoO 2 and Cr 2 O 3 at 1.60 V RHE .Reproduced with permission.[177]Copyright 2023, Springer Nature.

Figure 18 .
Figure 18.a) Catalysts optimization (left) and electrolyte optimization (right) to protect the metal substrate from Cl À corrosion.b) Stability tests recorded at a constant current of 100 mA cm À2 for pure NF in electrolytes with different proportions of Na 2 SO 4 .c) Tafel plots of NF in different electrolytes.d) The amounts of various anions versus the distance above the electrode surface obtained through classical molecular dynamics simulations.e) Long-term stability tests at a constant current of 400 mA cm À2.Reproduced with permission.[178]Copyright 2021, Wiley-VCH.

Figure 20 .
Figure 20.a) The liquid-gas-liquid phase transition-based migration mechanism of the water purification and migration process and the driving force.b,c)Schematic of the scaled-up SES.d) Electrolysis durability test for the scaled-up SES.Reproduced with permission.[184]Copyright 2022, Springer Nature.