Recent Advances of First d-Block Metal-Based Perovskite Oxide Electrocatalysts for Alkaline Water Splitting

First d-block metal-based perovskite oxides (FDMPOs) have garnered significant attention in research for their utilization in the water oxidation reaction due to their low cost, earth abundance, and promising activities. Recently, FDMPOs are being applied in electrocatalysis for the hydrogen evolution reaction (HER) and overall water splitting reaction. Numerous promising FDMPO-based water splitting electrocatalysts have been reported, along with new catalytic mechanisms. Therefore, an in-time summary of the current progress of FDMPO-based water splitting electrocatalysts is now considered imperative. However, few reviews have focused on this particular subject thus far. In this contribution, we review the most recent advances (mainly within the years 2014–2020) of FDMPO electrocatalysts for alkaline water splitting, which is widely considered to be the most promising next-generation technology for future large-scale hydrogen production. This review begins with an introduction describing the fundamentals of alkaline water electrolysis and perovskite oxides. We then carefully elaborate on the various design strategies used for the preparation of FDMPO electrocatalysts applied in the alkaline water splitting reaction, including defecting engineering, strain tuning, nanostructuring, and hybridization. Finally, we discuss the current advances of various FDMPO-based water splitting electrocatalysts, including those based on Co, Ni, Fe, Mn, and other first d-block metal-based catalysts. By conveying various methods, developments, perspectives, and challenges, this review will contribute toward the understanding and development of FDMPO electrocatalysts for alkaline water splitting.


Introduction
Water electrolysis is garnering extensive interest for the clean production of hydrogen, notably with the advantages of high efficiency, abundant sources, mild environmental impact, etc. In addition, by storing electricity in the form of chemical energy (i.e., hydrogen), water electrolysis helps to mitigate the intermittent issue of renewable energies (e.g., wind, solar). Water electrolysis is a process that decomposes water into oxygen and hydrogen using external electrical energy; it is composed of two half-reactions; namely, the anodic oxygen evolution reaction (OER) and cathodic hydrogen evolution reaction (HER). Both the OER and HER are multi-electron processes involving several reactive intermediates, such as H, OH, and O. In alkaline media, the OER generally proceeds via the following pathway on the active sites ( * ) [1]: For the HER, it takes place via the following mechanism [2]: Volmer step: Heyrovsky step: Or Tafel step: The sluggish kinetics observed for the OER and HER are among the most serious bottlenecks preventing the mass-adoption of water electrolyzers [3]. Alkaline water electrolysis renders it possible to utilize non-precious metal-based electrocatalysts, and is thereby widely considered to be the most promising method for large-scale application of water electrolyzes for hydrogen production [4].
Among the various non-precious electrocatalysts reported thus far [5,6], perovskite-type oxides have garnered much attention due to their low cost, earth abundance, and promising activities [7][8][9]. Primitive perovskite oxides have the general formula ABO 3 , where the A-site cation is usually an alkaline/rare-earth metal, while the B-site is occupied by a transition metal surrounded by six oxygen atoms. In addition to primitive perovskite oxides (ABO 3 ), double-(A-site ordered (A 2 BB'O 6 ) and B-site ordered (A 2 BB'O 6 )), Ruddlesden-Popper (A n+1 B n O 3n+1 ), and quadruple (AA 3 B 4 O 12 ) types have also been studied for water electrolysis (Scheme 1) [10][11][12]. The structure of perovskite oxides is flexible and can accommodate different levels of oxygen vacancies and various dopants, allowing their composition to be easily modified. As a result, the catalytic performance of perovskite oxides toward water electrolysis can be easily tuned and optimized.  For perovskite oxides, the adsorption energies of the reaction intermediates determine their catalytic activity to a large extent, which is typically influenced by the electronic configuration on the catalyst surface [14]. First d-block metal-based (e.g., Co, Ni, Fe, Mn, Cu, Cr, V, and Ti) perovskite oxides can utilize the unique characteristics of 3d valence electrons and have attracted much attention for water electrolysis [15]. In addition, first d-block metals are relatively abundant and cheap; their earth abundance and current price are shown in Scheme 2. Consequently, first d-block metal-based perovskite oxides (FDMPOs) have been extensively investigated for use in the water oxidation reaction [16]. Recently, extending the application of perovskite oxides toward the HER [17,18] and overall water splitting reaction [8,19] has also become popular. An in-time summary of the strategies used to rationally design perovskite oxides and their current progress will be beneficial for their further development. Although there are several excellent reviews on perovskite electrocatalysts [9,20,21], the efforts of reviews on FDMPOs used in water electrolysis are few, which motivates us to mitigate this gap. We aim to contribute to the development of perovskite oxide-based electrocatalysts that are used for the alkaline water splitting reaction by presenting design strategies and current advances.

Design Strategies
In general, there are two strategies used to design superior perovskite oxide-based water splitting electrocatalysts. One is to improve the intrinsic activity, while the other is to enhance the number of active sites. The specific methods used to achieve these strategies include defect engineering, strain tuning, nanostructuring, and hybridization, which are summarized below.

Defect Engineering
Defect engineering is commonly used to modify the electronic structure and enhance the intrinsic activity of perovskite oxide catalysts [23]. The strategies of engineering perovskite defects include the introduction of dopants [24][25][26][27] and modulating their deficiencies [28,29]. Specifically, dopants can be introduced into the A- [30,31], B- [32,33] and oxygen-sites [34,35]. One of the wellknown examples is Fe-doping in the B-site of cobalt-based perovskite oxides, which has been demonstrated to significantly boost the OER activity [36]. In order to clarify the mechanism behind the positive role of Fe-doping, operando Co K-edge X-ray absorption fine structure (XAFS) spectra were collected for Ba0.5Sr0.5Co0.8Fe0.2O3-δ (BSCF) and Ba0.5Sr0.5CoO3-δ (BSC) during the OER. From the Fourier-transformed extended X-ray absorption fine structure (FT-EXAFS) spectra shown in Figure  1a, a more noticeable development of the Co-Co coordination peak (represented by the dashed line)

Design Strategies
In general, there are two strategies used to design superior perovskite oxide-based water splitting electrocatalysts. One is to improve the intrinsic activity, while the other is to enhance the number of active sites. The specific methods used to achieve these strategies include defect engineering, strain tuning, nanostructuring, and hybridization, which are summarized below.

Defect Engineering
Defect engineering is commonly used to modify the electronic structure and enhance the intrinsic activity of perovskite oxide catalysts [23]. The strategies of engineering perovskite defects include the introduction of dopants [24][25][26][27] and modulating their deficiencies [28,29]. Specifically, dopants can be introduced into the A- [30,31], B- [32,33] and oxygen-sites [34,35]. One of the well-known examples is Fe-doping in the B-site of cobalt-based perovskite oxides, which has been demonstrated to significantly boost the OER activity [36]. In order to clarify the mechanism behind the positive role of Fe-doping, operando Co K-edge X-ray absorption fine structure (XAFS) spectra were collected for Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3-δ (BSCF) and Ba 0.5 Sr 0.5 CoO 3-δ (BSC) during the OER. From the Fourier-transformed extended X-ray absorption fine structure (FT-EXAFS) spectra shown in Figure 1a, a more noticeable development of the Co-Co coordination peak (represented by the dashed line) was observed on BSCF when compared to that on BSC, indicating that the formation of an increased number of CoO(OH) active species was responsible for the higher activity [37]. A-site dopants usually play an important role in influencing the catalytic activity of perovskite oxides by modifying the electronic configuration of the B-site cations. For example, partial substitution of La with Sr in LaCoO 3 has also been reported to optimize the oxygen adsorption energy of cobalt and boost the OER activity [38]. In addition to OER catalysis, doping in perovskite oxides has also been investigated in the HER. If the perovskite surface binds H too weakly, the overall HER rate is limited by the adsorption (Volmer) step. On the other hand, an excessively strong binding with H renders the desorption step difficult and limits the Heyrovsky/Tafel step rate. As a result, the binding of H on the perovskite oxide surface (H ad ) is critical for HER activity. In a recent study, anionic F-doping was introduced into La 0.5 Ba 0.25 Sr 0.25 CoO 3-δ (LBSC) to optimize the H ad and boost the HER activity. In comparison to pristine LBSC (−0.634 eV), La 0.5 Ba 0.25 Sr 0.25 CoO 2.9-δ F 0.1 (LBSCF) exhibited a more favorable ∆G H* value (−0.279 eV), which was closer to zero (not too strong or weak binding) and contributed to the higher HER activity [39].
Similar to the doping strategy, modulating the compositions of the deficiencies in the perovskite A-sites [40,41] and oxygen sites [42] is also a promising strategy for tuning the catalytic properties. Oxygen vacancies are generally considered beneficial for the intake of hydroxyl groups and OER activity [42,43]. However, recent works have demonstrated that there is an optimal concentration of oxygen vacancies for catalyzing the OER [44,45]. By varying the oxygen gas partial pressure, the oxygen vacancy concentration inside NdNiO 3 can be easily tuned [44]. By comparison, the OER activity of NdNiO 3 displayed a volcano-type relationship with the oxygen partial pressure, and the highest OER activity was achieved at a moderate oxygen partial pressure (i.e., 10 Pa, Figure 1c), in which the oxygen vacancy concentration and the Ni valence are balanced [44]. In another work, a large increase in the number of oxygen vacancies was found to be detrimental to the intrinsic OER activity of PrBaCo 2 O 6-δ [45]. The increased number of oxygen vacancies tend to align orderly in PrO 1−δ , which not only lowered the cobalt valence state, but also induced the cobalt transition from a high-spin to low-spin state, resulting in significant deceleration of the OER kinetics [45]. The roles of the oxygen vacancies are assumed to be multi-sided and require further investigation. Since the highly oxidizing environments during the OER may oxidize the lattice oxygen [46], the stability of highly oxygen-deficient perovskite oxides is also worth studying.

Strain Tuning
In addition to resorting to defects, engineering the strain can also modify the electronic structure of perovskite oxides and further tune the intrinsic activity [47,48]. Epitaxial growth is one of the standard methods for building strain in the lattice of perovskites. Strain is induced due to the mismatch when two planes with different lattice parameters are firmly connected to one another. By carefully choosing the substrate with a wide range of lattice spacings, the lattice mismatch can be tuned, generating tensile or comprehensive strain. The strain propagates into the internal solid film and gradually relaxes. Strain plays a role in modifying the catalytic performance of perovskites by changing the TM-O bond length and distorting the octahedral [49]. By controlling the lattice orientation of LaCoO 3 (LCO) epitaxial films, Xie and coworkers prepared a series of strained LaCoO 3 films with different degrees of distortion in the CoO 6 octahedron. They found that the strain-induced distortion causes a cobalt spin-state transition from the low spin state (t 2g 6e g 0) to an intermediate spin state (t 2g 5e g 1) in LCO(100) film, as shown in Figure 2(a1). This optimized e g orbital filling lowered the OER adsorption energy of LCO(100) (Figure 2(a2)), contributing to the excellent OER activity [50]. which not only lowered the cobalt valence state, but also induced the cobalt transition from a highspin to low-spin state, resulting in significant deceleration of the OER kinetics [45]. The roles of the oxygen vacancies are assumed to be multi-sided and require further investigation. Since the highly oxidizing environments during the OER may oxidize the lattice oxygen [46], the stability of highly oxygen-deficient perovskite oxides is also worth studying. changing the TM-O bond length and distorting the octahedral [49]. By controlling the lattice orientation of LaCoO3 (LCO) epitaxial films, Xie and coworkers prepared a series of strained LaCoO3 films with different degrees of distortion in the CoO6 octahedron. They found that the strain-induced distortion causes a cobalt spin-state transition from the low spin state (t2g6eg0) to an intermediate spin state (t2g5eg1) in LCO(100) film, as shown in Figure 2(a1). This optimized eg orbital filling lowered the OER adsorption energy of LCO(100) (Figure 2(a2)), contributing to the excellent OER activity [50]. Epitaxial strain is also a powerful tool used to manipulate the oxygen vacancy concentration inside perovskites beyond adjusting the electronic configurations [51]. Lee and coworkers demonstrated that tensile strain can assist SrCoO3-δ to maintain a certain number of oxygen vacancies during the OER, in contrast to the pristine SrCoO3-δ which was almost fully oxidized and contained Epitaxial strain is also a powerful tool used to manipulate the oxygen vacancy concentration inside perovskites beyond adjusting the electronic configurations [51]. Lee and coworkers demonstrated that tensile strain can assist SrCoO 3-δ to maintain a certain number of oxygen vacancies during the OER, in contrast to the pristine SrCoO 3-δ which was almost fully oxidized and contained few oxygen vacancies ( Figure 2(b1)). Due to the well-maintained oxygen vacancies, strained SrCoO 3-δ achieved an excellent OER activity, which was comparable to that of IrO 2 (Figure 2(b2)) [47].
Epitaxial growth-prepared strained thin films serve as an ideal platform for the fundamental study of perovskite oxides. However, investigations on the mechanisms of the changes observed in the reactivity of epitaxially-strained films should be carefully undertaken due to the convoluted effects of the defects and the film thickness-dependent strain states [52]. In addition, it is challenging to apply solid thin films for practical water electrolysis due to the lack of porosity and the high preparation cost. The introduction of strains into nanostructured perovskite particles is suggested for future studies to prepare highly-active electrocatalysts. Furthermore, the stability of strains during OER and HER is also worth investigating.

Nanostructuring
Thanks to the development of oxide-based nanomaterial synthesis methodologies, various nanostructures have been fabricated for perovskite oxides, such as nanoparticles [53,54], nanofibers [55,56], nanorods [57,58], nanotubes [59], and nanosheets [60]. A wide range of synthetic approaches have been reported, including wet chemistry methods (e.g., sol-gel, hydrothermal), deposition (e.g., physical laser deposition (PLD), chemical vapor deposition (CVD), and electrodeposition), electrospinning, exsolution, infiltration, and template-assisted synthesis. The rationale behind the nanostructuring Catalysts 2020, 10, 770 7 of 22 strategy is to increase the surface area and expose more active sites [61,62]. Shao and co-workers fabricated SrNb 0.1 Co 0.7 Fe 0.2 O 3-δ nanorods (SNCF-NR, Figure 3(a1)) using an electrospinning technique, which display promising bifunctional OER and HER activity [58]. The overall water splitting activity of SNCF-NR examined using a homemade water electrolyzer was observed to be superior to that achieved with a commercial catalyst pair (Pt/C-IrO 2 ) at a high current density (Figure 3(a2)). In addition to exposing more surface sites to facilely access the electrolyte [63], the down-sized structure altered the electronic configuration of the perovskite oxide [54]. The e g orbital filling of LaCoO 3 perovskite oxide was found to be highly dependent on the particle size ( Figure 3(b1)). Upon reducing the LaCoO 3 particle sizes from 200 to 60 nm, the corresponding e g orbital filling of cobalt increased gradually from 1.1 to 1.27, as shown in Figure 3(b2). This change was attributed to the spin-state transition from the low-spin to high-spin states observed for the cobalt ions on the nanoparticle surface. LaCoO 3 with a particle size of 80 nm exhibited an optimal e g orbital filling of 1.2 and achieved the best OER activity [54].
Thanks to the development of oxide-based nanomaterial synthesis methodologies, various nanostructures have been fabricated for perovskite oxides, such as nanoparticles [53,54], nanofibers [55,56], nanorods [57,58], nanotubes [59], and nanosheets [60]. A wide range of synthetic approaches have been reported, including wet chemistry methods (e.g., sol-gel, hydrothermal), deposition (e.g., physical laser deposition (PLD), chemical vapor deposition (CVD), and electrodeposition), electrospinning, exsolution, infiltration, and template-assisted synthesis. The rationale behind the nanostructuring strategy is to increase the surface area and expose more active sites [61,62]. Shao and co-workers fabricated SrNb0.1Co0.7Fe0.2O3-δ nanorods (SNCF-NR, Figure 3(a1)) using an electrospinning technique, which display promising bifunctional OER and HER activity [58]. The overall water splitting activity of SNCF-NR examined using a homemade water electrolyzer was observed to be superior to that achieved with a commercial catalyst pair (Pt/C-IrO2) at a high current density (Figure 3(a2)). In addition to exposing more surface sites to facilely access the electrolyte [63], the down-sized structure altered the electronic configuration of the perovskite oxide [54]. The eg orbital filling of LaCoO3 perovskite oxide was found to be highly dependent on the particle size ( Figure 3(b1)). Upon reducing the LaCoO3 particle sizes from 200 to 60 nm, the corresponding eg orbital filling of cobalt increased gradually from 1.1 to 1.27, as shown in Figure 3(b2). This change was attributed to the spin-state transition from the low-spin to high-spin states observed for the cobalt ions on the nanoparticle surface. LaCoO3 with a particle size of 80 nm exhibited an optimal eg orbital filling of 1.2 and achieved the best OER activity [54]. Distinct from conventional strategies such as infiltration, exsolution offers a unique method for growing nanoparticles in-situ from the perovskite oxide backbone [65,66]. The flexible nature of the perovskite structure can easily incorporate catalytically active elements. When exposed to a reducing environment, those elements can be partially exsolved onto the perovskite oxide surface as nanoparticles [67]. In addition, the exsolution process can be facilitated by introducing a moderate number of A-site deficiencies [68]. Perovskite oxide surfaces decorated with uniformly dispersed and catalytically active nanoparticles are favorable for excellent performance. For example, Liu and coworkers fabricated a unique surface with LSCN/Ni 2 P interfaces via exsolving Ni nanoparticles  Figure 3(c1)), which boosted the intrinsic OER activity ∼6.2-fold and the mass activity ∼10.2-fold. When used as a bifunctional electrocatalyst in a symmetrical two-electrode water electrolyzer, this catalyst delivered enhanced stability when compared to Pt/C-IrO 2 during a 500-h chronopotentiometry test (Figure 3(c2)) [64].
We should note that perovskite oxide supports are usually required for exsolution-prepared nanoparticles, while substrates are necessary for deposition-derived nanofilms, which potentially limits the number of active sites available. For the sake of fully utilizing the catalytic potential of the nanostructured moieties, the removal of the supporting substrate is highly desirable. Very recently, freestanding perovskite oxide thin films down to a few atomic layers were fabricated using a sacrificial buffer layer method [69], which may open up new opportunities for nanostructured perovskite oxide electrocatalysts.

Hybridization
In addition to the above-mentioned strategies used to boost the activity of perovskite oxide catalysts intrinsically by modifying their electronic structures and extrinsically by exposing more active sites, another exciting method involves hybridizing perovskite oxides with other active species. The motivation behind this approach is to expose a large number of active sites, [70] improve the electronic conductivity [71], or modify the electronic structure [72]. One example is presented in Figure 4(a1), which demonstrates the positive role of hybridization with MoS 2 in boosting the HER activity of LaNiO 3-δ (LNO) perovskite oxide. This enhancement in the HER activity was observed for other hybrids, including Ba 0.95 La 0.05 FeO 3-δ /MoS 2 and Pr 0.5 (Ba 0.5 Sr 0.5 ) 0.5 Co 0.8 Fe 0.2 O 3-δ /MoS 2 [73]. The increased number of heterogeneous interfaces formed upon mechanical hybridization (ball-milling) modified the surface structure of LNO, which was reflected by the positive shift observed for the Ni 3p XPS peak towards the higher binding energy region (Figure 4(a2)) [73]. The increased Ni valence state caused by hybridization contributes to the intake of hydroxyl groups due to their strong electrostatic affinity and thus, boosts the HER activity [18]. A similar improvement in the activity was observed by combining perovskite oxide with transition metal dichalcogenides in La 0.5 Sr 0.5 CoO 3-δ (LSC)/MoSe 2 , which enabled stable alkaline water splitting for 1000 h without any obvious degradation in its activity ( Figure 4b) [74].  The synthesis of crystalline perovskite oxides usually requires a high calcination temperature, The synthesis of crystalline perovskite oxides usually requires a high calcination temperature, where the particles tend to agglomerate and decrease the surface area [76,77]. In order to mitigate this shortcoming, perovskite oxides are usually hybridized with porous conductive carbon materials to expose more active sites [78]. In addition to functioning as a conductive support, carbon materials can influence the electronic configuration of perovskite oxides [72,75]. It was found that conductive acetylene black (AB) mixed with BSCF led to a negative shift in the Co K-edge when compared to bare BSCF (Figure 4(c4)), while maintaining comparable XANES (X-ray absorption near edge structure) fingerprints at the Ba L-edge (Figure 4(c1)), Sr K-edge (Figure 4(c2)), and Fe K-edge (Figure 4(c3)). The fact that AB carbon induces a lower mean oxidation state for the Co cations in BSCF/AB than that in BSCF indicates that it creates more oxygen vacancies, which may be the reason for its improved OER performance [75].
Defect engineering, strain tuning, nanostructuring, and hybridization are common strategies used to boost the catalytic performance of perovskite oxides. Due to these design strategies, perovskite oxides have been substantially advanced as OER and HER electrocatalysts. A summary of their activity is shown in Table 1. Rationally combining these strategies to work in a synergistic manner is suggested toward the design of superior perovskite-based water splitting electrocatalysts.

Advances of Perovskite Electrocatalysts
Due to the design strategies summarized beforehand, perovskite oxides have been substantially advanced as water splitting electrocatalysts. In this chapter, we group the perovskite oxides using the 3d transition metal elements in the B-site, including Co, Ni, Fe, Mn and others, and review their most recent advances accordingly.

Co-Based
In various FDMPOs, cobalt-based perovskite oxides are the most extensively investigated for water electrolysis [82]. One of the most widely investigated Co-based perovskites is Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3-δ , which has an optimal e g orbital filling close to unity and intrinsic OER activity approximately one order of magnitude higher than that of IrO 2 [83]. One particularly interesting phenomenon regarding this material is that Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3-δ experiences an in-situ surface transformation into amorphous cobalt oxyhydroxide during the OER, leading to increased OER activity upon electrochemical cycling [84]. In addition, the amorphous perovskite surface can be facilitated by heat treatment in Ar [85], Li treatment [86], photochemical decomposition [87], etc. Recently, Fabbri and coworkers found that the in-situ surface restructuring of Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3-δ was related to the lattice oxygen evolution reaction (LOER) process ( ABO 3−δ + OH − ↔ BO(OH) + A 2+ aq + 2−δ 2 O 2 + 3e − ) using operando X-ray absorption fine structure spectroscopy (XAFS) [46].
The LOER process is activated by the large oxygen vacancy content in Co-based perovskites, which improves the covalency between the Co 3d and O 2p bands (Figure 5a) and makes the oxidation of the lattice oxygen atoms thermodynamically favorable [88]. Shao-Horn and coworkers provided direct evidence for the oxidation of lattice oxygen during the OER using online electrochemical mass spectrometry (OLEMS). In their experiments, the perovskite oxides were isotopically-labeled with 18 O and the oxygen gas evolved by perovskite catalysts under normal KOH electrolyte conditions was monitored in-situ using OLEMS. In addition to 32

Ni-Based
Even though Co-based perovskite oxides have shown promising OER and HER activities, the relatively high price of cobalt is unfavorable for their large-scale application. When compared with Co, many other first d-block transition metals, such as Ni, Mn, and Fe, are significantly cheaper (Scheme 1). As a result, they are also frequently investigated for water electrolysis [89]. One popular Ni-based perovskite oxide is LaNiO3 [90], which is a well-known conducting perovskite oxide. Its high electronic conductivity (1000 Ω −1 cm −1 at room temperature) is expected to favor the charge transfer kinetics required during electrolysis, eliminating the need for a conductive carbon support, which affects the catalytic activity [91]. Bulk LaNiO3 is superior to Pt in catalyzing the OER [92], even though its OER activity is still inferior to that observed for more active NiCoOx and RuO2 [93,94]. In order to further enhance the catalytic activity of LaNiO3, strategies such as Fe-doping [89,95], strain tuning [96], and oxygen vacancy engineering [97] have been investigated to date. It is worth noting In addition to the OER, Co-based perovskite oxides have been also used in the HER. A number of cobalt-based perovskite oxides including both simple perovskites and double perovskites have been systematically studied for the alkaline HER [79]. By changing the A-site dopants with different ionic electronegativity, the cobalt valence was facilely tuned and optimized. A volcano-type relationship with thirteen different compositions was observed for the HER activity as a function of the A-site electronegativity ( Figure 5(b1)), suggesting that the A-site electronegativity may serve as an HER descriptor for perovskite oxides. The highest HER activity, which was comparable to Pt/C, was achieved at a moderate electronegativity value equal to ≈2.33, in which B a0.4 Ca 0.6 Gd 0.4 La 0.6 Co 2 O 5.5+δ and Ba 0.5 Ca 0.5 Pr 0.5 La 0.5 Co 2 O 5.5+δ achieved optimal Co valence and band structures ( Figure 5(b2)) [79].

Ni-Based
Even though Co-based perovskite oxides have shown promising OER and HER activities, the relatively high price of cobalt is unfavorable for their large-scale application. When compared with Co, many other first d-block transition metals, such as Ni, Mn, and Fe, are significantly cheaper (Scheme 1). As a result, they are also frequently investigated for water electrolysis [89]. One popular Ni-based perovskite oxide is LaNiO 3 [90], which is a well-known conducting perovskite oxide. Its high electronic conductivity (1000 Ω −1 cm −1 at room temperature) is expected to favor the charge transfer kinetics required during electrolysis, eliminating the need for a conductive carbon support, which affects the catalytic activity [91]. Bulk LaNiO 3 is superior to Pt in catalyzing the OER [92], even though its OER activity is still inferior to that observed for more active NiCoO x and RuO 2 [93,94]. In order to further enhance the catalytic activity of LaNiO 3 , strategies such as Fe-doping [89,95], strain tuning [96], and oxygen vacancy engineering [97] have been investigated to date. It is worth noting that these strategies can perturb or distort the octahedral geometry of BO 6 . Identifying this perturbation and further clarifying its role in water electrolysis are imperative. Chung and coworkers visualized the atomic-scale perturbation of the octahedral oxygen atoms in LaNiO 3 thin films using high-resolution scanning transmission electron microscopy (STEM), as shown in Figure 6(a1). In contrast to the bulk where four oxygen atoms were placed at the corners of a white rhombus for a Ni octahedron, the surface unit cells denoted by the yellow rhombus were substantially distorted. The strong distortions in the oxygen octahedron were generated by the surface exchange of Ni with Fe, which enhanced the OER activity by several orders of magnitude. Density functional theory (DFT) calculations further demonstrated that this atomic-scale structural perturbation changes the O 2p and Ni/Fe 3d states in particular near the Fermi level (from −2 to 0 eV), as shown in Figure 6(a2,a3), leading to facile charge transfer between the metal and oxygen atoms [98]. Notably, Ni-based perovskite oxides can also undergo a structural transformation during electrolysis. One representative example is BaNiO3 [99], which has a hexagonal structure with faceshared octahedral NiO6 oriented toward the chain units, as shown in Figure 6(b1). During the OER cycling process, BaNiO3 was transformed into BaNi0.83O2.5 (Ba6Ni5O15), which has mixed valence states of Ni 2+ , Ni 3+ , and Ni 4+ with distorted-prism coordinations (Figure 6(b1)). Due to these structural transformations, the OER activity of BaNiO3 was improved substantially, even far beyond the stateof-the-art IrO2 catalyst (Figure 6(b2)). From the DFT calculations shown in Figure 6(b3), BaNi0.83O2.5 Notably, Ni-based perovskite oxides can also undergo a structural transformation during electrolysis. One representative example is BaNiO 3 [99], which has a hexagonal structure with face-shared octahedral NiO 6 oriented toward the chain units, as shown in Figure 6(b1). During the OER cycling process, BaNiO 3 was transformed into BaNi 0.83 O 2.5 (Ba 6 Ni 5 O 15 ), which has mixed valence states of Ni 2+ , Ni 3+ , and Ni 4+ with distorted-prism coordinations (Figure 6(b1)). Due to these structural transformations, the OER activity of BaNiO 3 was improved substantially, even far beyond the state-of-the-art IrO 2 catalyst (Figure 6(b2)). From the DFT calculations shown in Figure 6(b3), BaNi 0.83 O 2.5 requires a smaller overpotential to make all the OER sub-steps downhill when compared to BaNiO 3 , supporting its higher OER activity [99]. Several Ni-based perovskite oxides have been introduced with promising OER activities. However, their use as HER electrocatalysts has seldom been investigated and awaits future effort.

Fe-Based
Considering the abundance of Fe in the earth's crust is far larger than that of Co and Ni, it is appealing to develop Fe-based perovskite electrocatalysts with comparable activity to their Co-and Ni-based counterparts. Fe is usually incorporated into Co-or Ni-based catalysts to boost the catalytic performance [5,94]. However, Fe-based perovskites, in which Fe acts as a single active site, have not been investigated as broadly as their Co-and Ni-based counterparts. This is mainly due to the low intrinsic activity of Fe-based perovskite oxides [100]. Recent efforts have been devoted to improving the catalytic performance of Fe-based perovskite oxides [70,101]. A series of Fe-based perovskite oxides including Ca 2 Fe 2 O 6-δ , CaSrFe 2 O 6-δ , and Sr 2 Fe 2 O 6-δ , which have different structures (Figure 7(a1)) have been compared in the OER. The tetrahedral chains in each layer of Ca 2 Fe 2 O 6-δ are oriented in an opposite direction to their neighboring layers, which is in contrast to CaSrFe 2 O 6-δ where the tetrahedral chains are orientated in the same direction. Sr 2 Fe 2 O 6-δ has a different type of defect order with alternating square pyramidal and octahedral coordination. These unique structures render distinct electronic conductivities. The room temperature conductivities determined using 4-probe measurements followed the order of: Ca 2 Fe 2 O 6−δ (1.13 × 10 −9 S cm −1 ) < CaSrFe 2 O 6−δ (7.78 × 10 −3 S cm −1 ) < Sr 2 Fe 2 O 6−δ (7.30 S cm −1 ). With the highest conductivity favorable for the charge transfer process, Sr 2 Fe 2 O 6-δ exhibited the best OER activity [81].
In addition to optimizing the conductivity, another strategy is to increase the proportion of Fe 4+ to improve the binding strength with the reaction intermediates [102,103]. Even though cationic and anionic dopants have been introduced into the perovskite framework to provoke Fe 4+ [34], stabilizing most or even all of the Fe cations in the tetravalent state remains challenging. A quadruple AA' 3 B 4 O 12 -type perovskite structure (CaCu 3 Fe 4 O 12 ) has been applied to accommodate Fe 4+ [11]. In this type of perovskite structure, the A-sites are occupied by alkaline/alkaline-earth/rare-earth metal ions, the A -sites are occupied by Jahn-Teller active ions such as Cu 2+ or Mn 3+ , and the B-sites contain d-block transition metal ions (Figure 7(b1)). The intrinsic OER activity (normalized by surface area) of CaCu 3 Fe 4 O 12 exceeds that of the state-of-the-art BSCF and RuO 2 (Figure 7(b2)), demonstrating its great potential for water electrolysis. The tetravalent Fe atoms in CaCu 3 Fe 4 O 12 improved the covalency of the transition metal-oxygen bonds and shortened the interatomic distance between the nearest neighboring OH adsorbates (Figure 7(a3)), making the formation of O-O bonds possible, which is regarded as the rate-limiting step in the OER [11]. In addition to optimizing the conductivity, another strategy is to increase the proportion of Fe 4+ to improve the binding strength with the reaction intermediates [102,103]. Even though cationic and anionic dopants have been introduced into the perovskite framework to provoke Fe 4+ [34], stabilizing most or even all of the Fe cations in the tetravalent state remains challenging. A quadruple AA'3B4O12type perovskite structure (CaCu3Fe4O12) has been applied to accommodate Fe 4+ [11]. In this type of perovskite structure, the A-sites are occupied by alkaline/alkaline-earth/rare-earth metal ions, the A′sites are occupied by Jahn-Teller active ions such as Cu 2+ or Mn 3+ , and the B-sites contain d-block transition metal ions (Figure 7(b1)). The intrinsic OER activity (normalized by surface area) of CaCu3Fe4O12 exceeds that of the state-of-the-art BSCF and RuO2 (Figure 7(b2)), demonstrating its great potential for water electrolysis. The tetravalent Fe atoms in CaCu3Fe4O12 improved the covalency of the transition metal-oxygen bonds and shortened the interatomic distance between the nearest neighboring OH adsorbates (Figure 7(a3)), making the formation of O-O bonds possible, which is regarded as the rate-limiting step in the OER [11].

Mn-Based
Mn-based oxides, especially CaMn oxides, have recently gained a significant amount of interest as water oxidation catalysts due to their similarity to the natural oxygen-evolving cluster (CaMn4O5) found in photosynthetic plant cells [104,105]. In addition, the high abundance and low cost of Mn have also helped attract interest for use as electrocatalysts. However, the catalytic activity of Mnbased perovskite oxides is typically low and almost inert [71,106]. One of the widely known methods to boost the catalytic performance of Mn-based perovskite oxides is to increase the oxygen vacancy concentration [107,108]. The motivation behind this strategy is to facilitate the intake of hydroxyl groups, which should readily adsorb on the oxygen vacant sites in the Mn-perovskite oxide ( Figure  8(a1)) [43]. One representative example is orthorhombic Ca2Mn2O5, which has five-coordinated square pyramid subunits composed of manganese and oxygen (Figure 8(a2)). In contrast, orthorhombic CaMnO3 without oxygen vacancies has six-coordinated octahedral subunits. The OER activity of Ca2Mn2O5 was significantly improved due to these oxygen vacancies [43].

Mn-Based
Mn-based oxides, especially CaMn oxides, have recently gained a significant amount of interest as water oxidation catalysts due to their similarity to the natural oxygen-evolving cluster (CaMn 4 O 5 ) found in photosynthetic plant cells [104,105]. In addition, the high abundance and low cost of Mn have also helped attract interest for use as electrocatalysts. However, the catalytic activity of Mn-based perovskite oxides is typically low and almost inert [71,106]. One of the widely known methods to boost the catalytic performance of Mn-based perovskite oxides is to increase the oxygen vacancy concentration [107,108]. The motivation behind this strategy is to facilitate the intake of hydroxyl groups, which should readily adsorb on the oxygen vacant sites in the Mn-perovskite oxide (Figure 8(a1)) [43]. One representative example is orthorhombic Ca 2 Mn 2 O 5 , which has five-coordinated square pyramid subunits composed of manganese and oxygen (Figure 8(a2)). In contrast, orthorhombic CaMnO 3 without oxygen vacancies has six-coordinated octahedral subunits. The OER activity of Ca 2 Mn 2 O 5 was significantly improved due to these oxygen vacancies [43]. Catalysts 2020, 10, x FOR PEER REVIEW 14 of 22 Society.
In addition to improving the adsorption of the hydroxyl groups, the oxygen vacancies can also cause structural transformations within Mn-based perovskite oxides and affects water electrolysis performance. In a recent study, Wang and coworkers demonstrated that perovskite oxides with ordered oxygen vacancies can achieve enhanced HER and OER activity in an alkaline electrolyte [12]. In their study, Nd0.5Ba0.5MnO3-δ (NBM3-δ) was used as the parent material and a series of perovskites with different structures and oxygen vacancy concentrations were easily obtained by extending the reductive annealing time, including NdBaMn2O5.5+δ (NBM5.5+δ, where δ < 0.5), NdBaMn2O5.5 (NBM5.5), and NdBaMn2O5.5−δ (NBM5.5−δ, where δ < 0.5) [12]. Upon reduction, the partial removal of crystalline oxygen and an increased number of oxygen vacancies (2MnMn• + OO× → 1/2O2 + VO•• + 2MnMn×) forces the A-site cations to be ordered; the layers of Nd-O and Ba-O are alternated along the c-axis, in which the oxygen vacancies were disorderly localized in the Nd-O layer. However, when the oxygen non-stoichiometry reached 0.5, MO5 square pyramids and MO6 octahedra were alternated periodically along the b-axis, generating orderly distributed oxygen vacancies. The detailed structures are displayed in Figure 8(b1). This unique arrangement of orderly distributed oxygen vacancies endowed NBM5.5 with a half-filled eg orbital, optimized O p-band, and distorted structure, which significantly improves the OER and HER activity, as shown in Figure 8(b2). However, it should be noted that Mn-based perovskite oxides exhibit far from satisfactory catalytic performance. The OER and HER activity are still inferior to those observed for their commercial counterparts and future efforts are required to further enhance their activity.

Other First d-Block Metal-Based
More than Co, Ni, Fe, Mn, other first d-block metal-based (e.g., V, Cr, Cu, Ti) perovskite oxides have also been studied for water electrolysis [109][110][111][112]. Recently, a systematic study of simple perovskite oxides used as OER catalysts has been conducted by Yagi and coworkers [113].  In addition to improving the adsorption of the hydroxyl groups, the oxygen vacancies can also cause structural transformations within Mn-based perovskite oxides and affects water electrolysis performance. In a recent study, Wang and coworkers demonstrated that perovskite oxides with ordered oxygen vacancies can achieve enhanced HER and OER activity in an alkaline electrolyte [12]. In their study, Nd 0.5 Ba 0.5 MnO 3-δ (NBM 3-δ ) was used as the parent material and a series of perovskites with different structures and oxygen vacancy concentrations were easily obtained by extending the reductive annealing time, including NdBaMn 2 O 5.5+δ (NBM 5.5+δ , where δ < 0.5), NdBaMn 2 O 5.5 (NBM 5.5 ), and NdBaMn 2 O 5.5−δ (NBM 5.5−δ , where δ < 0.5) [12]. Upon reduction, the partial removal of crystalline oxygen and an increased number of oxygen vacancies (2Mn Mn• + O O× → 1/2O 2 + V O •• + 2Mn Mn× ) forces the A-site cations to be ordered; the layers of Nd-O and Ba-O are alternated along the c-axis, in which the oxygen vacancies were disorderly localized in the Nd-O layer. However, when the oxygen non-stoichiometry reached 0.5, MO 5 square pyramids and MO 6 octahedra were alternated periodically along the b-axis, generating orderly distributed oxygen vacancies. The detailed structures are displayed in Figure 8(b1). This unique arrangement of orderly distributed oxygen vacancies endowed NBM 5.5 with a half-filled e g orbital, optimized O p-band, and distorted structure, which significantly improves the OER and HER activity, as shown in Figure 8(b2). However, it should be noted that Mn-based perovskite oxides exhibit far from satisfactory catalytic performance. The OER and HER activity are still inferior to those observed for their commercial counterparts and future efforts are required to further enhance their activity.

Other First d-Block Metal-Based
More than Co, Ni, Fe, Mn, other first d-block metal-based (e.g., V, Cr, Cu, Ti) perovskite oxides have also been studied for water electrolysis [109][110][111][112]. Recently, a systematic study of simple perovskite oxides used as OER catalysts has been conducted by Yagi and coworkers [113]. 25 kinds of perovskite oxide catalysts including CaBO 3 (B = Ti, V, Cr, Mn, Fe, and Co), SrBO 3 (B = Ti, V, Cr, Mn, Fe, and Co), YBO 3 (B = V, Cr, Mn, Fe, Co, and Ni), and LaBO 3 (B = V, Cr, Mn, Fe, Co, Ni, and Cu) were experimentally compared in the OER and their overpotentials displayed in Figure 9(a1) (except for some extremely inert compositions). From the comparison, V-, Cr-, Cu-, and Ti-based perovskite oxides are not that competitive when compared to their Co/Ni/Fe counterparts with similar structures. In addition to the activity comparison, several OER activity descriptors were examined for the 25 perovskite oxides, including e g orbital filling, oxygen 2p band center relative to the Fermi energy, and charge-transfer energy. In contrast to the e g orbital occupancy and oxygen 2p band location, only the charge-transfer energy was found to display a linear relationship with the OER overpotential and can serve as the most helpful descriptor for the OER [113].  From the above study, there is great scope to boost the catalytic activity of V-, Cr-, Cu-, and Tibased perovskite oxides. Recent studies have shown that BaTiO3-δ is a promising candidate for the OER [114,115]. BaTiO3-δ nanoparticles with a large oxygen vacancy concentration have been fabricated using a sol-gel method. A number of oxygen vacancies stabilize the hexagonal structure of BaTiO3-δ, where the TiO6 octahedra share both corners and faces (Figure 9(b1)). BaTiO3-δ outperformed the IrO2 catalyst in the OER at relatively low potentials (<1.6 V) with a reduced onset potential (1.32 V vs. RHE) and an increased current density (Figure 9(b2)). The OER activity of BaTiO3-δ can be further improved upon Fe-and Ni-doping by boosting the electronic conductivity [116].

Perspectives and Challenges
In this review, we have provided an in-time summary of the current advances in first d-block perovskite oxide-based water splitting electrocatalysts, in which the popular design strategies used to boost their activity have also been elaborated. Great strides have been made with many promising perovskite-type electrocatalysts exhibiting activities comparable or superior to those of commercial electrocatalysts have been designed and new theories involving the lattice-oxygen-evolutionmediated process have been experimentally confirmed. These will be beneficial to better understanding and further developing first d-block perovskite oxide electrocatalysts. However, many challenges still remain to be addressed: 1.
More precise control over the perovskite composition, structure, and active sites is required. Despite the fact that strategies such as defect engineering, strain tuning, and hybridization have proven effective to modulate the electronic structure and boost the intrinsic activity, precisely modulating the specific From the above study, there is great scope to boost the catalytic activity of V-, Cr-, Cu-, and Ti-based perovskite oxides. Recent studies have shown that BaTiO 3-δ is a promising candidate for the OER [114,115]. BaTiO 3-δ nanoparticles with a large oxygen vacancy concentration have been fabricated using a sol-gel method. A number of oxygen vacancies stabilize the hexagonal structure of BaTiO 3-δ , where the TiO 6 octahedra share both corners and faces (Figure 9(b1)). BaTiO 3-δ outperformed the IrO 2 catalyst in the OER at relatively low potentials (<1.6 V) with a reduced onset potential (1.32 V vs. RHE) and an increased current density (Figure 9(b2)). The OER activity of BaTiO 3-δ can be further improved upon Fe-and Ni-doping by boosting the electronic conductivity [116].

Perspectives and Challenges
In this review, we have provided an in-time summary of the current advances in first d-block perovskite oxide-based water splitting electrocatalysts, in which the popular design strategies used to boost their activity have also been elaborated. Great strides have been made with many promising perovskite-type electrocatalysts exhibiting activities comparable or superior to those of commercial electrocatalysts have been designed and new theories involving the lattice-oxygen-evolution-mediated process have been experimentally confirmed. These will be beneficial to better understanding and further developing first d-block perovskite oxide electrocatalysts. However, many challenges still remain to be addressed: 1.
More precise control over the perovskite composition, structure, and active sites is required. Despite the fact that strategies such as defect engineering, strain tuning, and hybridization have proven effective to modulate the electronic structure and boost the intrinsic activity, precisely modulating the specific doping/strain/interface sites and configuration still have major difficulties. In addition, we should note that many perovskite-type electrocatalysts undergo in-situ surface reconstruction during water electrolysis [37,99], which makes the identification of their active sites quite complicated. As a result, advanced in-situ/operando experimental tools are required to monitor the dynamic changes observed in perovskite oxide electrocatalysts during water electrolysis, elucidate their catalytic mechanism, and identify their active sites. DFT calculations provide effective theoretical tools to understand and predict the active species/configurations in perovskite oxide electrocatalysts. However, DFT calculations are subject to many approximations. In addition, it is also quite challenging for current theoretical modeling to precisely describe the working conditions of an electrocatalyst and rationally correlate the composition, structure, and morphology with the apparent activity. Further efforts are required to continuously optimize DFT methods to better model the real working conditions, improve the calculation accuracy, and reduce the computational cost.

2.
The catalytic performance of perovskite-type water splitting electrocatalysts is far from satisfactory. Although many perovskite compositions have been reported to outperform Ru-or Ir-based electrocatalysts in catalyzing the OER, the alkaline HER activity of many perovskite oxides is still far below that of Pt metal/alloys. In addition, promising OER activity has been reported for Co-based perovskite oxides, which are relatively expensive. Developing other first d-block metal perovskite oxides (e.g., Fe, Mn, Cu) with comparable activity to Co-based perovskite oxide are favorable to save the cost, which, in turn, accelerates their practical applications. In addition to the half-cell OER/HER test, it is very crucial to build the practical water electrolyzer and exam the realistic performance where other practical and yet-unsolved issues such as the durability and catalyst loading can be clearly identified. Thus, catalytic stability is another imperative factor that should be further improved besides the activity.
In conclusion, there is still a long way to go before the large-scale utilization of perovskite oxides as bifunctional electrocatalysts in alkaline water splitting. With cooperative efforts from the research community, the revolution of perovskite-type water splitting electrocatalysts will be foreseen.