High-Property Anode Catalyst Compositing Co-Based Perovskite and NiFe-Layered Double Hydroxide for Alkaline Seawater Splitting

: The progress of high-efﬁciency non-precious metal anode catalysts for direct seawater splitting is of great importance. However, due to the slow oxygen evolution reaction (OER) kinetics, competition of chlorine evolution reaction (ClER), and corrosion of chloride ions on the anode, the direct seawater splitting faces many challenges. Herein, we develop a perovskite@NiFe layered double hydroxide composite for anode catalyst based on Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3 (BSCF) and NiFe layered double hydroxide (NiFe-LDH) heterostructure. The optimized BSCF@CeO 2 @NiFe exhibits excellent OER activity, with the potential at 100 mA cm − 2 (E j = 100 ) being 1.62 V in the alkaline natural seawater. Moreover, the electrolytic cell composed of BSCF@CeO 2 @NiFe anode shows an excellent stability, with negligible attenuation during the long-term overall seawater splitting with the remarkable self-recovery ability in the initial operation stage, and the direct seawater splitting potential increasing by about 30 mV at 10 mA cm − 2 . Our work can give a guidance for the design and preparation of anode catalysts for the direct seawater splitting.


Introduction
Hydrogen with high energy density (142 MJ kg −1 ) and clean nature can hold great significance for "carbon neutrality" and has been widely hypothesized to be the most promising sustainable energy in the future [1][2][3][4][5]. Using intermittent clean energies such as wind energy and solar energy to split fresh water is well considered as one of the most sustainable H 2 production technologies [6]. Generally, water splitting ( 2H 2 O + energy → 2H 2 + O 2 ) consists of two half-reactions, the oxygen evolution reaction (OER) on the anode side and the hydrogen evolution reaction (HER) on the cathode side. The kinetic process of OER is much slower than that of HER due to the fact that the former involves a complicated fourelectron transfer, while the latter only involves two-electron transfer [7][8][9]. Therefore, the OER performance of the anode is critical to the water splitting efficiency.
Compared with fresh water, seawater occupying about 97% of the Earth's water resources is inexhaustible [10]. Therefore, splitting seawater to produce H 2 instead of fresh water has received more and more attention due to the following facts [11,12]. Firstly, seawater is a very abundant resource on our planet compared with the scarce freshwater resources [11]. Secondly, the direct seawater splitting without desalination can simplify electrolysis systems by eliminating the complex desalination equipment [13]. Finally, the direct seawater splitting is of great importance. However, the direct seawater splitting faces many challenges, especially for the anode sides [3]. The biggest challenges for direct seawater splitting are the chlorine evolution reaction (CER, 2Cl − → Cl 2 + 2e − , E 0 = 1.36 V vs. reversible hydrogen electrode (RHE)) and the corrosion of chloride ions on anode electrocatalysts. It is well known that the chloride ion content in seawater is about 0.5 M [14], and its oxidation to form Cl 2 involves only a two-electron/proton transfer process, which will compete with OER on the anode side. Under alkaline conditions (pH >7.5), the equilibrium potential of CER ( 2Cl − → Cl 2 + 2e − , Cl 2 + 2OH − → OCl − + H 2 O, E 0 = 1.72 V RHE ) is large, being about 490 mV higher than that of OER (1.23 V vs. RHE) [15]. The Cl 2 produced by CER in the alkaline medium will further combine with the OH − in the electrolyte to form hypochlorite, which aggravates the corrosion on the anode catalyst. In addition, during the direct splitting of seawater to produce hydrogen [15], the bacteria, microorganisms, suspended solid particles, and some metal ions (such as Mg 2+ , Ca 2+ , etc.) that can easily form hydroxide precipitation ( M 2+, x+ + OH − → M(OH) 2, x , M = Mg 2+ , Ca 2+ , etc. ) also have serious impacts on OER catalysts. Therefore, direct seawater splitting should follow several conditions: (1) adopting alkaline electrolyte to suppress CER; (2) the anode voltage below 1.72 V (vs. RHE) to avoid the generation of ClO − and ensure an approximative 100% OER; and (3) the electrodes on both sides have good activity, selectivity, corrosion resistance, and durability.
In the past few years, some Co-based perovskites with excellent OER activities and stabilities have been exploited as anode catalysts for water oxidation [37][38][39][40]. Among these Co-based perovskites, Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3 (BSCF) even shows an acceptable OER catalytic activity compared with the state-of-the-art iridium and ruthenium dioxides (IrO 2 and RuO 2 ) [27,41]. The latest work indicates that the compositing cerium oxide (CeO 2 ) can significantly improve the OER catalytic properties of some metal compounds such as LaFeO 3 [42], RuO 2 [43], and NiMn-LDH [44] . CeO 2 has a strong oxygen accumulation ability and lots of oxygen vacancies [45], and its (111) facet can promote OER catalysis. Therefore, it is easy to assume that compositing NiFe-LDH, CeO 2 with BSCF can construct a robust anode catalyst with good OER performances for seawater splitting.
Herein, we synthesized a novel compositing anode catalyst of BSCF@CeO 2 @NiFe with a heterogeneous structure by a facile precipitation method. The BSCF@CeO 2 @NiFe catalyst exhibits much higher OER activity (with E j = 100 = 1.52 V vs. RHE) than that of NiFe-LDH in alkaline medium. Moreover, our BSCF@CeO 2 @NiFe has high durability in the alkaline natural seawater, suggesting that it can be used as a promising candidate for seawater splitting.

Synthesis of Materials
All the reagents mentioned in this work were analytical grade (AR) and purchased from commercial channels without further processing unless otherwise specified. The classic sol-gel method was adopted to synthesize the Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3 (BSCF) according to our previous work [46]. The stoichiometric ratio of Co(NO 3 ) 2 ·6H 2 O, Fe(NO 3 ) 3 ·9H 2 O, Ba(NO 3 ) 2 and Sr(NO 3 ) 2 (Sinopharm Chemical Reagent Co., Ltd. Ningbo, China) were dissolved into 250 mL deionized (DI) water followed by adding citric acid (CA) and ethylenediaminetetraacetic acid (EDTA, Aladdin Ltd., Shanghai, China). The total mole of metal ions:CA:EDTA = 1:1.5:1, and the pH value of the mixture was adjusted to about Processes 2022, 10, 668 3 of 12 6 with NH 3 •H 2 O. Then, the mixture was vigorously stirred at a temperature of 80 • C. After the gel formed, it was placed at 180 • C and kept for 10 h in air. Subsequently, the dry gel was sintered at 950 • C for 5 h, and its heating rate was 100 • C h −1 . Finally, the obtained powder was further ball-milled for 1 h to obtain BSCF.
The subsequent loading of CeO 2 on BSCF are synthesized by a precipitation method. For synthesizing BSCF@CeO 2 , 1 mmol (0.218 g) BSCF was dispersed in 1 mL ethanol, then 100 mL DI water was added with continuous stirring at room temperature. Then, 100 mL 0.060 wt % cationic polyacrylamide (CPAM, 0.060 g, Baichuan Biotechnology Co., Ltd., Ningbo, China) clear solution was slowly added to pretreat the BSCF powder for 1h. Afterwards, the product was centrifuged, filtered, and then placed in 100 mL of DI water and heated at 50 • C. Next, 1 mmol Ce(NO 3 ) 2 ·6H 2 O and 3.2 mmol NaOH (Sinopharm Chemical Reagent Co., Ltd., Ningbo, China) was dissolved in 100 mL DI water, respectively. Then, two solutions were slowly dripped into the above mentioned BSCF-water mixture and the solution pH value kept at 9.5 during the whole process. Next, the precipitated product was centrifuged and filtered for three times with DI water and ethanol, and then dried at 100 • C for 10 h. Finally, the above dried product was heated to 600 • C for 2 h in order to yield the product of BSCF@CeO 2 -11. It should be noted that the former "1" represented the total moles of BSCF, and the latter "1" represented the total moles of added cerium ions. For comparison, a similar method was adopted to prepare the CeO 2 except for the use of BSCF and CPAM.
The subsequent loading of NiFe-LDH on BSCF@CeO 2 are synthesized by a precipitation method. For synthesis of BSCF@CeO 2 @NiFe, 0.1 g BSCF@CeO 2 -11 was placed into 100 mL DI water and maintained a constant stirring at ambient temperature. We dissolved a certain amount of Ni(NO 3 ) 2 ·6H 2 O and Fe(NO 3 ) 3 ·9H 2 O (Ni:Fe = 3:1) into 100 mL DI water, then dissolved a certain amount of NaOH and Na 2 CO 3 (Sinopharm Chemical Reagent Co., Ltd., Ningbo, China) into another 100 mL DI water, and kept the solutions at [OH − ] = 1.6 × [Ni 2+ + Fe 3+ ], [CO 3 2− ] = 2.0 × [Fe 3+ ] [47]. Then, we slowly dripped the above two solutions into the solution containing BSCF@CeO 2 at the same time, maintaining the pH value of 9.5, and stirring for 2 h during the whole process. Ultimately, the product was centrifuged, filtered, and evaporated at 80 • C for 10 h to yield the product of BSCF@CeO 2 @NiFe. In order to probe the loading effect on the OER properties, we loaded NiFe-LDH (Ni:Fe = 3:1) with total moles fraction of 2.0, 3.0, 4.0, and 6.0 mmol on BSCF@CeO 2 , and abbreviated as BSCF@CeO 2 @NiFe-11x (x = 2, 3, 4, 6), respectively. For comparison, a similar method was adopted to prepare the NiFe-LDH except for the use of BSCF@CeO 2 and stirring at 50 • C for 6 h [48].

Electrochemical Measurements,
The preparation method of the catalyst ink was described as the following according to our previous work [49]: 5 mg catalyst and 5 mg carbon (Vulcan-XC72, Cabot Ltd., Shanghai, China) were dispersed in 2.0 mL ethanol plus 80 µL of 5 wt% Nafion (DuPont Ltd., Bejing, China) solution and ultrasonically blended for 2h to form a well-dispersed ink. Then, 20 µL of the ink was pipetted on a polished glassy carbon (GC) electrode for preparation of working electrode, and the solvent was evaporated naturally with a catalyst loading of 0.245 mg cm −2 . For measuring the working electrode, oxygen with ultrahigh purity grade was bubbled into the electrolyte to make it O 2 -saturated during the measurements so as to guarantee the O 2 /H 2 O equilibrium at 1.23 V vs. RHE. The standard three-electrode system (the above working electrode, the counter electrode of carbon rod, and the reference electrode of Hg/HgO (1 M NaOH) electrode), rotating disc electrode (RDE, PINE) and the electrochemical performances of the above specimens were recorded by CHI 760E electrochemical workstation (Shanghai Chenhua, China) in 1 M KOH under a rotating speed of 1600 rpm. Before the electrochemical test, several cyclic voltammetry (CV) scans (0.2~1.0 V vs. Hg/HgO) were performed to remove the bubbles covered on the catalysts. Linear sweep voltammetry (LSV) was recorded from 0.2 to 1.0 V vs. Hg/HgO at a scan rate of 5 mV s −1 . Electrochemical impedance spectrum (EIS) was measured at 1.676 V vs. RHE over a frequency range from 10 6 to 1 Hz. The equation: η = E RHE −1.23 V was adopted to work out the overpotentials (η) of the working electrodes, and the Tafel equation: η = blogj + a was adopted to work out Tafel slopes, where η is the overpotential, j represents current density, and b is the Tafel slope [50]. The electrochemically active surface area (ECSA = C dl /C s ) was assessed by measuring double-layer capacitance (C dl ) from CV with different scanning rates. C S represents the specific capacitance of 1.0 M KOH (0.04 mF/cm 2 ) [50]. The C dl which was positively correlated to the ECSA was assessed by C dl = j c /V, where j c represented the current density, and V was the voltage measured by CV. The scanning potential range was from 0.15 to 0.25 V (vs. Hg/HgO). The measured potentials were transformed to potentials vs. RHE by the Nernst equation ( E (RHE) = E (Hg/HgO) + 0.098 + 0.059 × pH). In addition, the equation ( E C = E M − iR s ) was adopted to correct the compensated potential, where R s is the ohmic solution resistance measured by EIS, i is the current, E C is the correction voltage and E M is the measurement voltage. The stability tests of the catalysts were performed by the continuous LSV and CV scanning.
The anodes for overall water (alkaline water, alkaline simulated seawater, and alkaline natural seawater) splitting (OWS) were composed of the as-prepared catalyst and RuO 2 ink drip-coated on Ni foam (NF, 1 cm × 1 cm, Sinero Co., Ltd., Beijing, China) with the total load being about 4.0 mg cm −2 . The cathode was commercial 20 wt% Pt/C (Aladdin Ltd., Cardiff, UK) which was dripped onto NF with the total load of about 2.0 mg cm −2 . The stability tests of OWS were performed at 100 mA cm −2 for alkaline aqueous solution and alkaline simulated seawater [51] (seeing Table S1, Supplementary Information), but 10 mA cm −2 for alkaline natural seawater (taken from the East China Sea, 29 • 52 00.00" N, 121 • 31 00.00" E). All the OWS stability tests were uncompensated for resistance. In our work, the simulation seawater in our paper is 0.5 M NaCl solution. The alkaline nature seawater in our paper indicates the mixed solution of nature seawater with 1 M KOH. Figure 1a shows the simple precipitation synthesis process of BSCF@CeO 2 @NiFe. The micromorphology of the different catalysts is observed by SEM. As shown in Figure 1b, the BSCF powder particles with a relatively smooth surface are very large, with the diameter >1 µm. The surface of BSCF@CeO 2 (Figure 1d) is very rough due to a large number of fine CeO 2 particles with the diameter of about tens of nanometers coated on the BSCF surface. BSCF@CeO 2 @NiFe shows a similar morphology with BSCF@CeO 2 , while some NiFe-LDH nanorods entangled with the CeO 2 nanoparticles can be found on the BSCF surface ( Figure 1e). In addition, BSCF@NiFe almost shows the same micromorphology with BSCF@CeO 2 @NiFe (Figure 1c). Figure 1f shows the XRD patterns of BSCF@CeO 2 @NiFe, BSCF@NiFe, NiFe-LDH, BSCF@CeO 2 , BSCF, and CeO 2 . It is obvious that the XRD peaks of BSCF@CeO 2 @NiFe correspond with BSCF (PDF#01-075-6980, space group Pm-3m), CeO 2 (PDF#43-1002, space group Fm-3m), and NiFe-LDH (PDF#00-040-0215, space group R-3m), and no other impurity phases can be found. It is worth noting that Sr ions in A-site of BSCF is easy to segregate during sintering and electrochemical processes [52,53], and then deteriorate the stability of the BSCF crystal structure. When BSCF is directly contacted with alkaline solution, especially carbonate solution, A-site ions (Ba 2+ and Sr 2+ ) are easy to form carbonates (Sr 0.8 Ba 0.2 CO 3 ) (Figure 1g) due to the electronegative nature of BSCF surface [54]. Therefore, during the synthesizing process of BSCF@CeO 2 , it is essential to coat the organic Processes 2022, 10, 668 5 of 12 compound (CPAM) containing cationic characteristics on the BSCF surface to reduce the electronegativity [54]. By introduction of CPAM, the impurity phase of Sr 0.8 Ba 0.2 CO 3 can be effectively prevented.  Figure 1f shows the XRD patterns of BSCF@CeO2@NiFe, BSCF@NiFe, NiFe-LDH, BSCF@CeO2, BSCF, and CeO2. It is obvious that the XRD peaks of BSCF@CeO2@NiFe correspond with BSCF (PDF#01-075-6980, space group Pm-3m), CeO2 (PDF#43-1002, space group Fm-3m), and NiFe-LDH (PDF#00-040-0215, space group R-3m), and no other impurity phases can be found. It is worth noting that Sr ions in A-site of BSCF is easy to segregate during sintering and electrochemical processes [52,53], and then deteriorate the stability of the BSCF crystal structure. When BSCF is directly contacted with alkaline solution, especially carbonate solution, A-site ions (Ba 2+ and Sr 2+ ) are easy to form carbonates (Sr0.8Ba0.2CO3) (Figure 1g) due to the electronegative nature of BSCF surface [54]. Therefore, during the synthesizing process of BSCF@CeO2, it is essential to coat the organic compound (CPAM) containing cationic characteristics on the BSCF surface to reduce the electronegativity [54]. By introduction of CPAM, the impurity phase of Sr0.8Ba0.2CO3 can be effectively prevented.

Results
The OER LSV curves of the different catalysts are measured to evaluate their catalytic activities. BSCF perovskite as a classic oxygen evolution reaction (OER) catalyst, has an excellent intrinsic catalytic activity compared with other perovskites [55]. As shown in Table S2, the OER potential of BSCF at 10 and 100 mA cm −2 are much lower than those of most perovskites. However, the HER activity of BSCF is very poor, and it cannot be used as a cathode catalyst for direct seawater splitting ( Figure S1, Supplementary Information). In order to optimize the composition of BSCF@CeO2@NiFe composite catalysts, we measure the OER catalytic activities of BSCF@CeO2@NiFe-11x (x = 2, 3, 4, 6) in 1M KOH alkaline solution. Obviously, BSCF@CeO2@NiFe-113 has the highest OER performance among these samples (Figure 2a) and can be used as a follow-up research object (abbreviated as BSCF@CeO2@NiFe). Moreover, compared with BSCF@NiFe, BSCF@CeO2, and BSCF, BSCF@CeO2@NiFe has the much lower overpotential of 297 mV at 100 mA cm −2 , which is even lower than that of NiFe-LDH (315 mV) and RuO2 (344 mV) (Figure 2b,c). The OER LSV curves of the different catalysts are measured to evaluate their catalytic activities. BSCF perovskite as a classic oxygen evolution reaction (OER) catalyst, has an excellent intrinsic catalytic activity compared with other perovskites [55]. As shown in Table S2, the OER potential of BSCF at 10 and 100 mA cm −2 are much lower than those of most perovskites. However, the HER activity of BSCF is very poor, and it cannot be used as a cathode catalyst for direct seawater splitting ( Figure S1, Supplementary Information). In order to optimize the composition of BSCF@CeO 2 @NiFe composite catalysts, we measure the OER catalytic activities of BSCF@CeO 2 @NiFe-11x (x = 2, 3, 4, 6) in 1M KOH alkaline solution. Obviously, BSCF@CeO 2 @NiFe-113 has the highest OER performance among these samples (Figure 2a) and can be used as a follow-up research object (abbreviated as BSCF@CeO 2 @NiFe). Moreover, compared with BSCF@NiFe, BSCF@CeO 2 , and BSCF, BSCF@CeO 2 @NiFe has the much lower overpotential of 297 mV at 100 mA cm −2 , which is even lower than that of NiFe-LDH (315 mV) and RuO 2 (344 mV) (Figure 2b,c). Additionally, BSCF@CeO 2 @NiFe owns the smallest Tafel slope of 56.0 mV dec −1 among above four samples (Figure 2d), which indicates its superior OER catalytic kinetics. From Figure 2e and Table S3, it is obvious that the OER activity of our catalyst is almost the highest among those of the previously reported perovskites and NiFe-LDH catalysts.
Additionally, BSCF@CeO2@NiFe owns the smallest Tafel slope of 56.0 mV dec −1 among above four samples (Figure 2d), which indicates its superior OER catalytic kinetics. From Figure 2e and Table S3, it is obvious that the OER activity of our catalyst is almost the highest among those of the previously reported perovskites and NiFe-LDH catalysts. Although NiFe-LDH has an excellent OER catalytic activity, its stability needs to be further improved (seeing Figure S2, Supplementary Information) [36]. The stabilities of BSCF@CeO2@NiFe and NiFe-LDH are performed by 25-cycle CV scans (Figures 2f,g and  S3). The result shows that the potential of BSCF@CeO2@NiFe slightly increases from 1.52 to 1.55 V at 100 mA cm −2 after 25-cycle CV scanning, whereas the potential of NiFe-LDH increases from 1.53 to 1.58 V at 100 mA cm −2 after the same CV scanning cycles. Compared with NiFe-LDH, the high stability of BSCF@CeO2@NiFe can be attributed to the synergistic effect of BSCF [56] and NiFe-LDH.
Subsequently, we explore the OWS properties of BSCF@CeO2@NiFe. As shown in Figure 3a Although NiFe-LDH has an excellent OER catalytic activity, its stability needs to be further improved (seeing Figure S2, Supplementary Information) [36]. The stabilities of BSCF@CeO 2 @NiFe and NiFe-LDH are performed by 25-cycle CV scans (Figure 2f,g and Figure S3). The result shows that the potential of BSCF@CeO 2 @NiFe slightly increases from 1.52 to 1.55 V at 100 mA cm −2 after 25-cycle CV scanning, whereas the potential of NiFe-LDH increases from 1.53 to 1.58 V at 100 mA cm −2 after the same CV scanning cycles. Compared with NiFe-LDH, the high stability of BSCF@CeO 2 @NiFe can be attributed to the synergistic effect of BSCF [56] and NiFe-LDH.
Subsequently, we explore the OWS properties of BSCF@CeO 2 @NiFe. As shown in Figure 3a, the OER catalytic activities of BSCF@CeO 2 @NiFe are measured in 1 M KOH, 1 M KOH + 0.5M NaCl, and 1 M KOH + seawater on RDE. The results show that the OER catalytic activities of BSCF@CeO 2 @NiFe in 1 M KOH and 1 M KOH + 0.5 M NaCl solutions are very close, while the OER catalytic activity of BSCF@CeO 2 @NiFe is far lower in 1 M KOH + seawater, compared with that in 1 M KOH. This can correspond to the fact that the bacteria, microorganisms, suspended solid particles, and some metal ions have serious impacts on the OER catalysts. Furthermore, we assemble an electrolytic cell with BSCF@CeO 2 @NiFe anode and 20 wt% Pt/C cathode (Figure 3b) to study the OWS performances of BSCF@CeO 2 @NiFe anode in 1 M KOH, 1 M KOH + 0.5 M NaCl, and 1 M KOH + seawater. As shown in Figure 3b, the OWS potentials of the electrolytic cell with BSCF@CeO 2 @NiFe are 1.69V and 1.70V at 100 mA cm −2 in 1 M KOH and 1 M KOH + 0.5 M NaCl, respectively, which is even lower than that with RuO 2 (1.74 V in 1 M KOH). In addition, the OWS potential of the electrolytic cell at 100 mA cm −2 in + seawater. As shown in Figure 3b, the OWS potentials of the electrolytic cell with BSCF@CeO2@NiFe are 1.69V and 1.70V at 100 mA cm −2 in 1 M KOH and 1 M KOH + 0.5 M NaCl, respectively, which is even lower than that with RuO2 (1.74 V in 1 M KOH). In addition, the OWS potential of the electrolytic cell at 100 mA cm −2 in 1 M KOH + seawater is as large as 1.76 V, corresponding to the OER catalytic activity of BSCF@CeO2@NiFe in the alkaline natural seawater.  Figure 3d shows that the voltage of the electrolytic cell with BSCF@CeO2@NiFe exhibits a negligible elevation of about 10 mV after 25 h short-term OWS test, showing the much higher stability than that with RuO2 (70 mV). Moreover, the electrolytic cell with BSCF@CeO2@NiFe (Figure 3e) also shows a slight degradation after long-term 100 h test, and the voltage of this electrolytic cell almost has no attenuation. Nevertheless, it is worth noting that, in the initial 10 h of the stability test, the electrolytic cell voltage rapidly increases until it reaches the maximum value of 1.87 V, and then sharply drops to 1.81 V in the next approximate 1 h test. The quick degradation of the electrolytic cell can be related to the corrosion of BSCF@CeO2@NiFe in the Cl − -containing electrolyte and appearance of the passivation layer [3]. The subsequent performance recovery can be ascribed to be the surface amorphization of BSCF perovskite anode in the alkaline overall water splitting test [53,57] and subsequent appearance of CoFe-LDH on the BSCF surface caused by the dynamic surface self-reconstruction [52,58]. Considering the high splitting overpotential of natural seawater, the stability test of BSCF@CeO2@NiFe in the natural seawater is performed at 10 mA cm −2 to prevent the formation of the strong corrosive ClO -. Figure 3c is a corresponding seawater splitting schematic. As shown in Figure 3f, during the 200 h  Figure 3d shows that the voltage of the electrolytic cell with BSCF@CeO 2 @NiFe exhibits a negligible elevation of about 10 mV after 25 h short-term OWS test, showing the much higher stability than that with RuO 2 (70 mV). Moreover, the electrolytic cell with BSCF@CeO 2 @NiFe (Figure 3e) also shows a slight degradation after long-term 100 h test, and the voltage of this electrolytic cell almost has no attenuation. Nevertheless, it is worth noting that, in the initial 10 h of the stability test, the electrolytic cell voltage rapidly increases until it reaches the maximum value of 1.87 V, and then sharply drops to 1.81 V in the next approximate 1 h test. The quick degradation of the electrolytic cell can be related to the corrosion of BSCF@CeO 2 @NiFe in the Cl − -containing electrolyte and appearance of the passivation layer [3]. The subsequent performance recovery can be ascribed to be the surface amorphization of BSCF perovskite anode in the alkaline overall water splitting test [53,57] and subsequent appearance of CoFe-LDH on the BSCF surface caused by the dynamic surface self-reconstruction [52,58]. Considering the high splitting overpotential of natural seawater, the stability test of BSCF@CeO 2 @NiFe in the natural seawater is performed at 10 mA cm −2 to prevent the formation of the strong corrosive ClO − . Figure 3c is a corresponding seawater splitting schematic. As shown in Figure 3f the anode and cathode products are collected by the drainage method and compared with their theoretical values (seeing Figure S4, Supplementary Information). The results show that the test values are basically consistent with the theoretical values, indicating that the Faraday efficiency are close to 100%, and almost no Cl 2 is produced (Figure 3g). The result indicates an excellent OER selectivity of BSCF@CeO 2 @NiFe.

Discussion
As mentioned before, BSCF@CeO 2 @NiFe shows a superior OER catalytic activity in alkaline solutions. To clarify the mechanisms, we further perform the EIS and C dl measurements. It is known that ECSA of the electrocatalysts can be estimated by calculating C dl from the cyclic voltammetry (CV) curves (Figure 4a-d). The C dl value of BSCF@CeO 2 @NiFe is 2.70 mF cm −2 which is higher than that of NiFe-LDH (1.19 mF cm −2 ), BSCF@NiFe (1.07 mF cm −2 ) and BSCF (1.91 mF cm −2 ) (Figure 4e). The EIS Nyquist plots in Figure 4f further shows that BSCF@CeO 2 @NiFe owns a lower charge-transfer resistance (R ct , 2.1 Ω) compared with other samples [59], which demonstrates its quick charge transport capability and favorable OER dynamics [60]. The improvement of C dl and smaller R ct of BSCF@CeO 2 @NiFe can be related to the introduction of CeO 2 with strong oxygen-gathering ability and heterogeneous interface formed between CeO 2 and metal (oxy)hydroxides (NiFe-LDH) [61].  Figure S4, Supplementary Information). The results show that the test values are basically consistent with the theoretical values, indicating that the Faraday efficiency are close to 100%, and almost no Cl2 is produced (Figure 3g). The result indicates an excellent OER selectivity of BSCF@CeO2@NiFe.

Discussion
As mentioned before, BSCF@CeO2@NiFe shows a superior OER catalytic activity in alkaline solutions. To clarify the mechanisms, we further perform the EIS and Cdl measurements. It is known that ECSA of the electrocatalysts can be estimated by calculating Cdl from the cyclic voltammetry (CV) curves (Figure 4a-d). The Cdl value of BSCF@CeO2@NiFe is 2.70 mF cm −2 which is higher than that of NiFe-LDH (1.19 mF cm −2 ), BSCF@NiFe (1.07 mF cm −2 ) and BSCF (1.91 mF cm −2 ) (Figure 4e). The EIS Nyquist plots in Figure 4f further shows that BSCF@CeO2@NiFe owns a lower charge-transfer resistance (Rct, 2.1 Ω) compared with other samples [59], which demonstrates its quick charge transport capability and favorable OER dynamics [60]. The improvement of Cdl and smaller Rct of BSCF@CeO2@NiFe can be related to the introduction of CeO2 with strong oxygen-gathering ability and heterogeneous interface formed between CeO2 and metal (oxy)hydroxides (NiFe-LDH) [61].

Conclusions
In this work, we successfully synthesized heterostructure electrocatalyst of BSCF@CeO2@NiFe by compositing CeO2, NiFe-LDH with BSCF. Benefiting from its large ECSA as well as smaller charge-transfer resistance, the synthesized BSCF@CeO2@NiFe showed a superior OER catalytic activity with Ej = 100 being 1.52 V in 1 M KOH, 1.54 V in 1 M KOH + 0.5 M NaCl and 1.62 V in 1 M KOH + seawater, which were even lower than Corresponding C dl values (e) and EIS Nyquist plots (f) of BSCF@CeO 2 @NiFe, BSCF@NiFe, NiFe-LDH, BSCF@CeO 2 , BSCF, CeO 2 , and RuO 2 . The insert of (f) is the EIS plots in a smaller range.

Conclusions
In this work, we successfully synthesized heterostructure electrocatalyst of BSCF@CeO 2 @NiFe by compositing CeO 2 , NiFe-LDH with BSCF. Benefiting from its large ECSA as well as smaller charge-transfer resistance, the synthesized BSCF@CeO 2 @NiFe showed a superior OER catalytic activity with E j = 100 being 1.52 V in 1 M KOH, 1.54 V in 1 M KOH + 0.5 M NaCl and 1.62 V in 1 M KOH + seawater, which were even lower than NiFe-LDH and RuO 2 . Additionally, our BSCF@CeO 2 @NiFe had an excellent long-term durability and self-recovery capability during the stability test.