Epitaxial Design of Complex Nickelates as Electrocatalysts for the Oxygen Evolution Reaction

The oxygen evolution reaction (OER) is a crucial process in electrochemical water splitting, a promising technology to renewably yield hydrogen gas from water. Designing and developing earth‐abundant, efficient, and stable OER electrocatalysts to replace the most widely used but scarce RuO2 and IrO2 are thus of critical interest. Recently, ABO3‐structured perovskite oxides, especially rare‐earth nickelates, are extensively studied for their potential use as OER electrocatalysts. In particular, the epitaxial synthesis of complex oxide thin films allows flexible and precise control over the materials so that their structure–stability–property relationships can be established. Using nickelate thin films as model systems, this review illustrates how epitaxial design allows researchers to test different hypotheses and proposed descriptors, as well as formulate new design principles. Following a brief introduction to the background of OER mechanisms, proposed activity descriptors, and synthesis methods, various epitaxial design strategies are surveyed including strain tuning, composition control, surface termination/orientation selection, defect engineering, and interface design. These have led to precise control over the atomic structures and electronic properties of nickelates which in turn determine their electrochemical performance. Finally, the remaining challenges and perspectives toward a deeper understanding and use of complex oxides as OER catalysts are discussed.


Introduction
Seeking viable, sustainable clean energy sources is urgently needed on a global scale to achieve the goal of carbon neutrality by 2050 or sooner. Hydrogen is one of the most attractive options since it has a higher energy density (142 kJg −1 compared to fossil fuels at 10-50 kJg −1 ) and has no carbon emission from its combustion or electrochemical oxidation, yielding only H 2 O. Thus, green hydrogen generation holds the promise of enabling a global transition to sustainable energy storage. While most hydrogen comes from fossil fuels and produces significant CO 2 emissions, electrochemical water splitting to yield hydrogen has been considered a sustainable approach to facilitate the storage of renewable energy. [1] However, the sluggish kinetics of oxygen evolution reaction (OER) limits the efficiency of electrochemical water splitting. While precious metal oxides (e.g., IrO 2 and RuO 2 ) exhibit superior OER activities, the high cost prevents their large-scale commercialization. [2] Developing low-cost, efficient, and stable OER electrocatalysts is necessary to increase the efficiency of water splitting for green hydrogen generation. Earthabundant 3d transition metals (TMs, i.e., Fe, Co, and Ni) and their compounds are promising alternatives for OER in alkaline solutions. [3] Of TM-based compounds, perovskite oxides (ABO 3 ) are particularly intriguing due to their high compositional flexibility and electronic tunability. [4] One prevailing mechanism for OER on ABO 3 oxides is the adsorbate evolution mechanism (AEM). In the AEM, the reaction follows a consecutive pathway via oxygen-containing intermediates (such as OH*, O*, and OOH*, where * denotes surface adsorbate sites) with surface B-site TM cations (undergoing associated redox events during charge transfer) representing the active sites. [5] The adsorption, dissociation, and desorption of oxygen-containing intermediates play a decisive role in determining OER activity. Several descriptors linked to the electronic properties of perovskite-based electrocatalysts have been proposed to deepen our understanding of their OER activity and to be able to guide materials design. The inner circle in Figure 1 shows six main descriptors for OER activity. ΔG O * − ΔG OH * (an adsorption energy difference between O * and OH * intermediates) was proposed as an OER activity descriptor. [6] Experimental overpotentials for a wide variety of metal oxide surfaces superimpose well on the volcano-shaped theoretical overpotential plots from this simple descriptor. [7] Based on this, catalysts with moderate binding strength show the best OER performance. [8] The number of electrons in 3d orbitals of B-site TMs was also proposed to be correlated with OER activity. Otagawa and co-workers [9] reported that the number of d electrons influences the OH* bond strength and that OER performance improves as the number of d-electrons in the perovskites LaB 3+ O 3 (where B is Cr, Mn, Fe, Co, or Ni) increases. In addition, a few other OER activity descriptors have been proposed based on molecular orbital (MO) and band theories. As is well known, the B-site TM 3d atomic orbitals (AOs) are often hybridized with oxygen 2p AOs, leading to the formation of and MOs. [10] The TM d AOs in MOs (also called e g orbitals) are typically more strongly hybridized with O 2p orbitals than those in orbitals (also called t 2g orbitals). [10] Therefore, the e g d electron count on the B-site cation can significantly affect the bond strength between the electrocatalyst and oxygen-containing absorbate and thus impact OER activity. Specifically, by studying a wide range of complex oxides, Suntivich et al. [11] concluded that an e g filling slightly greater than one is optimal for OER. It should be noted that e g filling does not satisfactorily demonstrate the rea-son of significantly different OER activities of LaMnO 3 , LaCoO 3 , and LaNiO 3 (LNO), which have the same e g occupancies. This result is mainly due to the fact that e g orbital filling of TM cations is based on an ionic model and thus does not reflect the electron occupancy within the TM-oxygen bond, thereby underscoring the importance of TM 3d-O 2p hybridization in descriptions of OER. Strong TM 3d-O 2p hybridization promotes the interaction between O anions and TM cations at perovskite B-site, decreasing the energetic cost associated with accepting and donating electrons at the adsorbate-catalyst interface which is considered to be important for OER. [12] Moreover, Grimaud et al. [13] reported that electrocatalysts with their O 2p-band center position near the Fermi level (E F ) display greater OER activities, but that being too close to E F reduces material stability by promoting surface amorphization. Recently, the energy difference between the TM 3d and O 2p orbitals, called as the charge transfer energy (∆), was also proposed as a key OER activity descriptor. [14] Electrocatalysts with smaller ∆ show higher OER activity.
Recently, a lattice oxygen activation mechanism (LOM) has been proposed by several groups based on theoretical and experimental work on perovskite oxides. [15] In the LOM, electron transfer involves substantial lattice oxygen redox and invokes these species (and vacancies thereof) in the reaction mechanism. www.advancedsciencenews.com www.advenergymat.de In LOM, O 2 is generated by direct coupling of O-O without the formation of OOH*, leading to a more efficient OER process. [16] Lowering V O formation energy and strengthening the metaloxygen covalency of perovskite oxide electrocatalysts can trigger the LOM. However, oxygen redox has been recognized as one of the major causes of the surface amorphization of oxide catalysts during the OER, strongly affecting long-term stability. Moreover, the degree to which the lattice oxygen participates and promotes OER activity for perovskite oxide electrocatalysts remains unclear. Besides AEM and LOM, an oxide path mechanism (OPM) was also proposed recently. [17] In OPM, OH* adsorbed on adjacent metals undergoes deprotonation to yield two metal-oxo species, which finally combine to release O 2 . Since the deprotonation of the oxyhydroxide group to form peroxide is skipped, the OER process is accelerated, and the activity is enhanced. However, the OER via OPM can only occur when the O-O distance between two neighboring metal ions is close enough (e.g., the CaCu 3 Fe 4 O 12 system). [17] Theoretical calculations demonstrated that LNO is very close to the top of the overall OER activity volcano plot of La-based perovskite oxides that takes into account both AEM and LOM. [18] Moreover, the unique characteristics of the rare earth nickelates RNiO 3 (RNO, where R is a lanthanide), including relatively weaker bond strength between Ni and OH* compared to other first-row TMs, the e g = 1 orbital occupancy, strong Ni 3d-O 2p hybridization, and small or negative ∆, make these materials promising OER electrocatalysts. [10,11,19] A chemical substitution approach has been explored to enhance the catalytic activity of bulk nickelates for OER. [20] However, the presence of secondary phases (such as NiO and SrCO 3 ) and the lack of control and reproducibility of the surface structures of bulk materials make it hard to clarify the OER mechanism of nickelate-based catalysts. [20b] In contrast, epitaxial thin films are ideally suited to generate fundamental insights as they are characterized by well-defined composition, phase, thickness, strain state, crystallographic orientation, and surface area. [21] Various structurally, compositionally, and even isotopically defined RNO thin films have been synthesized and tested, making them good model systems for fundamental investigations of structure-activity relationships. Here, we summarize recent progress on design strategies (outer circle in Figure 1) applied to epitaxial RNO films to generate a richer understanding of fundamental OER mechanisms and develop superior electrocatalysts. These design strategies include strain tuning, composition control (such as A-and B-site substitution), surface termination/orientation control, defect engineering, and interface control. The important factors affecting the OER activity are discussed in detail within these different strategies. It should be noted that these epitaxial design strategies applied to the nickelates are of general applicability to other materials systems as well. At the end of this review, the remaining challenges and future perspectives are outlined to expedite the study and use of complex oxides as OER catalysts.

Epitaxial Growth and Characterizations of Nickelate Thin Films
In recent decades, significant advances have been made in synthesizing epitaxial RNO thin films using various deposition techniques including pulsed laser deposition (PLD), molecular beam epitaxy (MBE), chemical vapor deposition, and radio frequency magnetron sputtering. RNO thin films prepared by these methods have been useful for properties determination and exploration of potential applications ranging from metalto-insulator transitions, [22] fuel cells, [23] photovoltaics, [24] information storage, [25] superconductivity, [26] and catalysis. [27] Highly crystalline RNO thin films that grow in a layer-by-layer fashion have been synthesized, providing a number of ways to further control their functionalities. The properties of RNO films can be readily varied and controlled by synthesis and processing conditions including deposition parameters, choice of substrate, and crystallographic orientation. Both PLD and MBE techniques readily take advantage of RHEED as an in situ diagnostic tool for monitoring the epitaxial growth in real-time, offering flexible and precise control for RNO thin films. Therefore, in the next section, we will briefly describe PLD and MBE as two representative epitaxial film deposition techniques that have been widely used to make high-quality RNO thin films and heterostructures, together with some of the frequently used in situ and ex situ characterization techniques.

Epitaxial Growth
The left side of Figure 2a displays a schematic of a PLD system. PLD is one kind of physical vapor deposition (PVD) technique. During deposition, a high-power pulsed ultraviolet laser ablates a target and forms a plasma-like plume that, in most cases, deposits the target material nearly congruently on the heated substrate. This process is performed in a high vacuum environment or with background gases such as O 2 and N 2 . The number of laser pulses during growth dictates the film thickness. Sub-optimal conditions, such as high growth temperature or low oxygen partial pressure, can lead to non-stoichiometry and secondary phase formation, all of which impact the structural, optical, electronic, and electrical properties. [28] In addition, parameters, such as the plume angle, target history, laser fluence, and laser spot size, can be important factors in obtaining high-quality and singlephase films. [29] Therefore, the stoichiometry of the films should always be confirmed by suitable analytical techniques, such as Xray photoemission spectroscopy (XPS) and Rutherford backscattering spectrometry (RBS), before proceeding with other characterization and property testing to avoid drawing erroneous conclusions. The target holder in PLD systems is usually designed to support multiple targets so that rotating the target assembly allows for multi-layer formation and compositional variation. For instance, the composition La 1-x Nd x NiO 3 can be achieved by programming the laser pulse sequence using LNO and NdNiO 3 (NNO) targets. [27b] MBE, which uses an ultra-high vacuum for thin film deposition, has also been used to deposit RNO thin films. A typical MBE setup consists of several Knudsen effusion cells (K cells), which point toward the center of the chamber at which a substrate holder is located (see schematic diagram shown on the right side of Figure 2a). Pneumatically actuated shutters are positioned directly in front of the crucibles to allow programmable start/stop control of the different sources. Prior to each growth, a quartz crystal microbalance is typically used to calibrate evaporation . Reproduced with permission. [30] Copyright 2017, APS. Epitaxial growth yields well-defined model systems which can be characterized by c) in situ XPS, d) AFM, e) XRD, f) STEM, etc. c) In situ Ni 3p and Fe 3p XPS spectra for LaNi 1-x Fe x O 3 thin films. d) A typical AFM image for a 30 u.c.-thick NdNiO 3 thin film grown on NdGaO 3 (110) shows clearly spaced surface steps and terraces. Reproduced with permission. [31] Copyright 2016, Springer Nature. e) High-resolution XRD -2 scans for LaNi 1-x Fe x O 3 epitaxial films. f) Cross-sectional high-angle annular dark-field (HAADF) STEM image for a LaNi 1-x Fe x O 3 film with x = 0.375. c,e,f) Adapted with permission. [32] Copyright 2021, American Chemical Society. rates for the different K cells. Atomic layer-by-layer deposition of RNO thin films can be achieved with reasonably accurate control of the fluxes (to within ≈1 at%), substrate temperature, and oxygen partial pressure. [33] For instance, by controlling the evaporation rates of K cells containing Nd, La, and Ni, films of composition Nd 1-y La y NiO 3 (0 ≤ y ≤ 1) can be grown by using MBE. [34] Due to its high conductivity and relative synthetic ease, LNO is the most widely studied nickelate for application as an OER catalyst. [27a,35] Diminishing the A-site ionic size in RNO reduces the tolerance factor, bends the Ni-O-Ni bond angle, and increases the lattice distortion. This reduces the overlap between the Ni 3d and the O 2p orbitals, which is, in principle, unfavorable for OER. [36] According to the e g filling descriptor, LNO sits on the left side of the OER activity volcano-shaped plot. [11] As a result, recent work has been carried out to investigate the potential of RNO (R = Nd or Sm) thin films in OER as V O may form in these nickelates during growth, elevating the average e g orbital occupancy to slightly above one which is more favorable for OER. [27b] However, Chang et al. [37] reported that difficulties synthesizing highquality, stoichiometric RNO thin films increase strongly as the A-site cation radius decreases. On the other hand, hole doping in RNO may push Ni to an even higher oxidation state (>3+), which can increase the Ni 3d-O 2p hybridization, thereby boosting OER performance. However, the synthesis of Sr-or Ca-doped nickelate thin films is particularly challenging because of the instability of high-valence Ni and competition between the perovskite and Ruddlesden-Popper (RP) phases. [29a,38] Thus, the deposition of nickelate thin films in an activated oxygen environment is essential for the stabilization of the Ni in an oxidation state greater than 3+. Although alternating film growth with in situ annealing in activated oxygen can yield high-quality La 1-x Sr x NiO 3 (LSNO) (where 0 ≤ x ≤ 0.5) thin films, [39] phase segregation leading to Sr 2 NiO 3 and SrNi 2 O 3 was clearly observed when attempting to deposit phase pure SrNiO 3-using oxygen-plasma-assisted MBE (OPA-MBE). [40] However, by using interface engineering, superlattices involving single unit cells of SrNi 4+ O 3 with LaFeO 3 [41] and SrTiO 3 (STO) [42] can be stabilized which may open new avenues for exploring the impact of high-valence Ni on OER performance.

In Situ and Ex Situ Characterizations
Reflection high-energy electron diffraction (RHEED) is a tool used for in situ monitoring the surface structure, growth rate, and growth mechanism in real-time. For both PLD and MBE techniques, RHEED is typically used to monitor the overall growth process of nickelate thin films. When layer-by-layer growth occurs, the oscillation of specular beam intensity is observed to oscillate with a period that is related to the deposition rate. For instance, Figure 2b shows representative RHEED intensity oscillations for an LNO thin film (15 u.c.) grown on an STO (111) substrate. [30] After growth, the RHEED patterns (inset in Figure 2b) show bright, sharp, unmodulated Bragg streaks characteristic of the desired perovskite structure, revealing a smooth surface, excellent crystallinity, and no secondary phases. More indepth RHEED analyses can provide crystallographic information on film surfaces, [43] which can be useful in understanding the OER properties of these films.
Optical spectroscopic ellipsometry (SE) is another in situ characterization tool which can monitor the electronic structure of epitaxial thin films in real-time. It should be noted that appropriate modeling and additional supporting parameters (such as thin-film thickness and complex optical constants) are required to analyze the SE data. The simultaneous use of both RHEED and SE can yield insight into the growth mechanisms by probing both ionic and electronic structures of the films in real-time. [44] In situ synchrotron X-ray diffraction (SXRD) can also be used for realtime monitoring of epitaxial film growth. [45] In SXRD, a smooth surface leads to the formation of crystal rods or spots in reciprocal space. The associated scattered intensity of a spot oscillates with film thickness during the layer-by-layer growth. Compared to RHEED which is only sensitive to the topmost atomic planes, the X-ray intensity is sensitive to the structure of the film and interface beneath, [45b,46] giving a larger penetration depth. However, SXRD is typically performed at a synchrotron beamline, [46,47] and thus is not used as frequently as RHEED or SE.
XPS is a crucial characterization tool for refining the synthesis of nickelates thin films. Especially using XPS as an in situ probe makes it even more powerful. [48] For example, measuring Ni 3p and Fe 3p core levels in situ after growth helped determine the B-site cation valence for epitaxial LaNi 1-x Fe x O 3 (LNFO) solid solutions synthesized by OPA-MBE. [32] As shown in Figure 2c, both sets of spectra show clear trends toward lower binding energy as the Fe content (x) increases. Based on the XPS spectra of Fe 3+ from LaFeO 3 (x = 1) and Ni 3+ from LNO (x = 0), the Ni 3p peak of the LNFO (x = 0.125) film shifts toward a lower binding energy, indicating a slight decrease in Ni valence. On the other hand, the Fe 3p peak obtained from the same sample shifts toward higher binding energy relative to the Fe 3+ peak in the LaFeO 3 spectrum, implying systematic increase of the Fe valence. The trend in the Ni 3p and Fe 3p spectra from LNFO with increasing Fe mole fraction (x) reveals a continuous extent of charge transfer from Fe to Ni. In addition to providing useful information on composition and cation valences, the information about the surface termination, film structure, band offsets, and built-in electric potentials can be figured out through more in-depth XPS analyses. [48,49] This information is important for understanding the OER behavior of nickelate films. Moreover, ambient pressure XPS (AP-XPS) can be utilized to experimentally measure the coverage of OER-related species in thin films exposed to oxygen and water vapor, [50] offering insight into the rate-limiting step and reaction mechanism.
Atomic force microscopy (AFM) is a powerful tool routinely used to evaluate surface topography. For instance, Figure 2d shows an AFM image and related line profile for a 30 u.c. thick NNO thin film grown on NdGaO 3 (NGO) (110). [31] An atomically flat surface with obvious steps and terrace features is observed. XRD can provide valuable crystallographic information on epitaxial nickelate thin films . Data from normal -2 scans enable the extraction of the out-of-plane lattice parameter (OOP) of nickelate films. For example, Figure 2e shows -2 scans near the (003) peak of LNFO films with a range of Fe content (x). [32] Clear Kiessig fringes indicate the high quality of the films. The increase in the OOP with x can be confirmed by the shift of the LNFO (003) diffraction peak (indicated by arrows) toward a lower angle. Reciprocal space maps and scans are often used to evaluate the strain state and epitaxial relationship of thin films, respectively. Scanning transmission electron microscopy (STEM) is a powerful tool that can confirm detailed structural information of the films at an atomic scale. For instance, the cross-sectional highangle annular dark field (HAADF) STEM image shown in Figure 2f shows a well-ordered interface and also reveals an epitaxial relationship between the film and substrate. [32] In addition, electron-energy-loss spectroscopy (EELS) can provide insight into whether the film is of uniform composition and detect the presence of phase segregation. EELS line profile analysis yields spatially resolved information about cation valence. Raman spectroscopy is a widely-available and powerful technique to observe the rotational, vibration, and other low-frequency modes of material systems. [51] In situ electrochemical Raman spectroscopy can probe real-time information about the surface reconstruction of electrocatalysts and the transformation of oxygen intermediates during OER. [52] Synchrotron-based X-ray absorption spectroscopy (XAS) can elucidate the local electronic and atomic structure of samples, such as the d orbital density of states and the Ni-O bond length of nickelates thin films. Finally, X-ray linear dichroism (XLD) can be used to investigate the orbital splitting and orbital occupancy using X-ray electric field polarizations parallel and perpendicular to the thin film surface.

Design Strategies for Nickelate Thin Films for OER
An essential research frontier for advancing water splitting is the design and development of more efficient OER electrocatalysts. Outstanding OER electrocatalysts need excellent intrinsic activity and a sufficient density of exposed active sites. There are generally two primary strategies for designing OER electrocatalysts. One is promoting intrinsic activity (e.g., increasing the e g occupancy in the d z 2 orbital, optimizing hybridization and Δ) of a material, and the other focuses on increasing the amount of exposed active sites (e.g., decreasing particle size and optimizing the structure/morphology). Next, we discuss design strategies recently used to tune nickelate thin films as more effective OER electrocatalysts.

Strain Tuning
Depositing nickelate thin films on single-crystal substrates with distinct lattice parameters can yield different strain states. By tuning the interfacial strain, the shape, size, and connectivity of NiO 6 octahedral structures can be changed. These structural distortions affect the overlap between the Ni 3d and O 2p orbital, modify the e g orbital occupancy, and tune the electronic states near E F ,  (Figure 3a). Peak shifts in the XRD -2 patterns indicate that biaxial strain is introduced to LNO films from −1.2% (compressive strain) to +2.7% (tensile strain). AFM and electrical transport measurements show that all the films are well-ordered and conductive. www.advancedsciencenews.com www.advenergymat.de XRD RSM measurements verified that all films are coherently strained, except for those grown on SLAO. Additional strain states between −1.2% and 0% are also obtained by changing the LNO thickness between 10 and 100 nm on the LAO substrate. How strain affects the OER activity of each film is characterized in an O 2 -saturated 0.1 m KOH solution. OER linear sweep voltammetry measurements (Figure 3b) indicate that the LNO film with biaxial compressive strain showed higher OER activity, even outperforming the Pt catalyst. The onset potential of LNO under compressive strain clearly shifts to a lower overpotential ( ) compared to that of LNO under tensile strain. As shown in Figure 3c, Ni L 2 -edge XAS was used to analyze the electronic basis behind the changes in OER activity with strain. Probing the absorption of X-rays polarized parallel (E//ab) and perpendicular (E//c) to the film surface allowed to resolve the orbital splitting energies as well as the d z 2 and d x 2 −y 2 orbital occupancy. As shown in Figure 3d, they found that the orbital splitting induced by compressive (tensile) strain leads to a high-occupancy of d z 2 (d x 2 −y 2 ) orbital at lower energy. DFT calculations showed that the e g -center shifts toward lower energies under compressive strain, leading to weaker bond strength between TM and O and a subsequent increase in OER activity. Choi et al. [35b] also found that compressive strain can enhance the OER activity of A-site-deficient La 1-NiO 3 epitaxial films due to an increase in the d z 2 orbital occupancy.
Beyond the structural distortion, epitaxial strain can also cause oxygen deficiency, especially for the tensile strain case, thereby offering a way to modify the e g orbital occupancy. Structural distortion and oxygen deficiency are usually strongly coupled at heterojunctions of nickelates.  Figure 3g) and weaker Ni-O chemisorption, consistent with previous work. [27a,35b] Compressive strain is favorable for OER within the range where V O is negligible, whereas a tensile strain is unfavorable to OER. However, tensile strain large enough to induce V O formation can partially reduce Ni 3+ to Ni 2+ , making the e g occupancy slightly larger than one and enhancing OER. [11]

Composition Control
Varying the composition of perovskite nickelates can alter their crystal structure, the character of the electron charge distribution and deformation fields, TM 3d occupancies, TM 3d-O 2p covalency, and active sites on the surfaces, which in turn, impact the electron transfer between TMs, reaction intermediate adsorbates via oxygen, and the OER activity. RNO allows for great flexibility in composition control, namely the rare earth (A-) site, [27b,35b,39] nickel (B-) site, [32,53] and dual-sites. [54]

A-Site Substitution
Generally, A-site rare earth cations in RNO are not considered active sites in OER. However, A-site cation can indirectly affect OER activity by altering the lattice parameter, B-site cation valence, surface V O concentration, and band structure. For instance, Wang et al. [27b] studied the OER performance of well-defined perovskite nickelate RNO (R = La, La 0.5 Nd 0.5 , La 0.2 Nd 0.8 , Nd, Nd 0.5 Sm 0.5 , Sm, and Gd) epitaxial thin films grown on STO substrates by PLD. The overall sheet resistance increases continuously as the A-site cation size decreases from La to Gd due to the decrease in Ni-O bond covalency. [36] Moreover, LNO and La 0.5 Nd 0.5 NiO 3 exhibit metallic behavior, whereas GdNiO 3 exhibits insulating behavior. The metallic behaviors of La 0.2 Nd 0.8 NiO 3 , NNO, and Nd 0.5 Sm 0.5 NiO 3 were observed at room temperature (RT) and the metal-to-insulator transition temperatures (T MI ) are 148, 190, and 283 K, respectively. SmNiO 3 film shows the insulating behavior at RT and its T MI is about 388 K. Performing intrinsic catalytic activity measurements on RNO films (R = Sm or Gd) with a T MI greater than RT is impractical due to voltage loss in the film plane. Thus, only the RNO films with lower T MI than RT were discussed for the OER activity (Figure 4a). In an early stage, ionic radius reduction increases the OER performance and La 0.2 Nd 0.8 NiO 3 shows the highest OER current density (Figure 4b). RNO with a smaller average A-site cation size synthesized from a mixture of Nd and Sm shows a comparable OER current density comparable to NNO. As can be seen from the changes in OOP and XAS spectra on the Ni L 2 edge (Figure 4c), the smaller A-site cation radius resulting from the mixing of Nd and Sm incorporates more V O under the same synthesis conditions. The corresponding reduction from Ni 3+ to Ni 2+ increases the average e g orbital filling to a favorable level for enhancing OER activity. Moreover, ultraviolet photoelectron spectroscopy was adopted to probe the work function of RNO films. It was found that the E F of RNO films approaches the electrolyte redox level (4.90 eV) at pH 13 by shifting down as the A-site cation radius decreases, which is also favorable for increasing the OER activity. Thus, a delicate balance between electronic conductivity and E F position needs to be achieved when using A-site rare earth doping.
A-site aliovalent substitution has also been applied to enhance the OER activity of nickelate systems. For instance, Liu et al. [39] systematically investigated the effect of Sr alloying on OER activity using epitaxial LSNO (where 0 ≤ x ≤ 0.5) films deposited on LAO (001) substrates via OPA-MBE. Figure 4d shows that the OER performance of LSNO increases as x increases, with x = 0.5 being the most active. For further quantitative comparison, they [39] defined the OER onset potential as the potential needed to reach a current density of 50 μA cm −2 oxide and calculated the overpotential ( ) using the formula = E OER − 1.23 V. The of pure LNO is about 0.36 V, similar to those obtained in previous studies. [11,27a] The continuously decreases with increasing x from 0 to 0.5, with La 0.5 Sr 0.5 NiO 3 having the smallest (0.29 V). This value is comparable to that of Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3− , a state-of-the-art electrocatalyst. [11] O K-edge XAS measurements show that the width of the O K pre-edge peak increases as x increases, indicating an increase in Ni 3d-O 2p hybridization. [55] The increase in Ni 3d-O 2p hybridization can enhance OER activity by facilitating electron extraction from oxygen adsorbates. [56] The evolution of the valence band (VB) reveals this enhanced hybridization. XPS VB spectra show a clear trend where the O 2p band shifts upward with increasing x in LSNO. DFT calculations reveal an upward shift in O 2p band when Sr is doped in LNO, bringing it closer to E F . The upward shift of the O 2p bands improves hybridization (covalency) between the Ni 3d and O 2p band due to a decrease in Coulomb interaction with the A site cation as the latter valence drops from 3+ to 2+. Taking another perspective, hole doping by replacing La 3+ with Sr 2+ in LNO drives a downshift of the O 2p band in E F compared to pure LNO. The O 2p band center upshifting and approaching E F induces an increase in the oxygen character of the antibonding states below E F . This in turn causes Ni 3d to mix more strongly with O 2p, promoting electron transfer to and from oxygen, leading to higher OER activity. Figure 4e shows a schematic energy band diagram for pure LNO and LSNO (x > 0). Sr substitution in LNO induces hole doping in the O 2p band, decreasing Δ and the charge transfer barrier, thus boosting OER. film with a homogeneous Fe distribution over the entire film. c) Comparison of OER current densities for pristine, Fe-doped, pre-oxidized, and pre-reduced LaNiO 3 films. Reproduced with permission. [53] Copyright 2019, Springer Nature. d) Comparison of overpotential ( ) and current density measured at 1.6 V versus RHE for LaNi 1-x Fe x O 3 (LNFO). e) Energy band diagrams for LNFO films with various x values. Adapted with permission. [32] Copyright 2020, American Chemical Society.
The formation of A-site vacancies also significantly modifies material properties. For example, Choi et al. [35b] found that La deficiency in LNO films could change the Ni-O-Ni bond angle and Δ, altering OER performance. They used PLD to deposit 50 nm thick La 1-x NiO 3 (where 0 ≤ x ≤ 0.15) films on NGO substrates. The electrocatalytic activities of the La 1-x NiO 3 films were measured in a 1 m NaOH electrolyte. Figure 4f shows that La 0.95 NiO 3 films exhibit a higher OER current density and smaller onset potential than pure LNO, while La 1-x NiO 3 films with x > 0.05 show degraded OER activity. Ni L-edge XAS analysis indicates that La 0.95 NiO 3 films exhibit a narrower Δ compared with LNO films, leading to enhanced OER activity.

B-Site Substitution
B-site Ni cations in RNiO 3 are usually considered active centers for OER. Replacing Ni with other TM cations (such as Mn, Fe, and Co) can modify the catalytic activity by adjusting the electron con-figuration, band structure, and surface defect density in nickelate thin films. For example, Bak et al. [53] synthesized RNiO 3 (where R is La, Pr, and Nd) thin films and then used a simple electrochemical method to exchange surface Ni with Fe. They demonstrated that the structural perturbation of the NiO 6 octahedra induced by Fe substitution boosts the OER activity by up to an order of magnitude. [53] The authors also prepared Fe-doped LNO films with Fe homogeneously distributed throughout the entire film. The Fe distribution for both kinds of films was verified via STEM images and energy dispersive X-ray spectroscopy (EDX) maps, shown in Figure 5a,b. The OER measurements were conducted in a 0.1 m KOH electrolyte (pH = 12.9). Figure 5c compares OER current densities for the different types of LNO films including pristine, 5%-Fe-doped, and Fe-surface-exchanged. The 5%-Fe-doped LNO displayed higher OER activity than pristine LNO thin films, while the OER current densities of the Fe-surface-exchanged films were almost ten times higher than that of Fe-doped LNO films with Fe homogeneously distributed. The higher OER activity was also found in Fe-surface-exchanged PrNiO 3 and NNO films, suggesting that there is another crucially important factor beyond Fe addition. STEM measurements were performed to investigate the structural variation during the Feexchange reaction. Strong distortion of oxygen octahedra near the surface of the LNO film, resulting from Ni extraction during the electrochemical oxidation and reduction reactions, was observed. This structural perturbation leads to a variation of the O 2p and TM (Ni/Fe) 3d states, resulting in facile charge transfer between the Ni/Fe and oxygen and enhancing the OER performance.
To further connect the fundamental structural and electronic properties of LNFO solid solutions with their OER performance, Wang et al. [32] systemically investigated a series of high-quality epitaxial LNFO thin films (where 0 ≤ x ≤ 1) deposited on LSAT substrates by OPA-MBE. XRD patterns show that OOP increases with increasing x (Figure 2e)

Surface Termination/Orientation Control
Previous studies have demonstrated that the surface termination of a complex oxide can significantly affect the surface electronic structure, [57] the surface adsorption energy, [58] and ultimately the mechanisms and kinetics of surface chemical reactions. A similar approach has been used to alter the surface termination of nickelate film and investigate the effect on OER. Recently, Baeumer et al. [35d] showed the effect of surface termination on the surface reconstruction and OER activity of LNO thin films. To tune the surface composition of LNO films, the authors used two independent approaches: 1) depositing a single NiO x layer on a LaO-terminated surface and 2) varying the deposition temperature during the PLD growth of LNO films. Standing wave XPS (SW-XPS) and STEM/EDX were measured to determine the surface termination of the LNO film. STEM/EDX results (Figure 6a) show that the surface of LNO film deposited at 650°C is a mixture of LaO termination and a double LaO x layer. SW-XPS analysis further indicates that the surface is La-rich when the LNO films were deposited at higher temperatures (650 or 750°C) but is Ni-rich when grown at lower temperatures (450 or 550°C). The Ni-terminated film was found to exhibit higher OER than the La-terminated film (Figure 6b). The overpotential difference between La-and Ni-terminated LNO films is up to 150 mV. Based on surface-sensitive information from UV-vis spectroelectrochemistry, these authors found that Ni hydroxide-like surface monolayers, which are thermodynamically stable, generate on the Ni-terminated LNO surfaces, lowering the OER overpotential, consistent with DFT predictions. For the La-terminated film, this electrochemically driven transformation is inaccessible. The authors also found a significant difference in cyclic voltammetry (CV) scans of the pre-OER regime (Figure 6c). A characteristic redox wave at about 1.4 V versus RHE is observed for Ni-terminated LNO films (T growth = 550°C). However, this redox wave disappears for La-terminated LNO films (T growth = 750°C). The magnitude of the redox wave decreases as La coverage increases. Moreover, LNO films with Ni-termination exhibit stable OER for at least 40 h at 1 mA cm −2 (Figure 6d), while the La-terminated LNO films show significant failure after 7 to 10 h. This is because of other decomposition reactions at the high overpotential required to reach the selected current density.
Surface orientation control of nickelate thin films via changing the substrate orientation is another way to tune the surface environment and OER activity. For example, Peng et al. [59] deposited epitaxial NNO thin films as model catalysts and established the relationship between the structural anisotropy and the OER mechanism. (100)-, (110)-, and (111)-oriented NNO films display similar oxidation states and TM-O covalency but exhibit different OER performance with (100) > (110) > (111) (Figure 6e). DFT simulation suggests that the OER proceeds through the lattice-oxygen-mediated mechanism on the NNO (100) surface but through the AEM on both the NNO (110) and (111) surfaces (Figure 6f). This is because the generation of highly stable oxygen reactive sites is promoted on the (100) surface than on the (110) and (111). Interestingly, Wohlgemuth et al. [60] found that LNO(111) thin films revealed lower (≈433 mV) at 3 mA cm −2 compared to LNO(110) ( = ≈620 mV) and LNO(100) ( = ≈620 mV) thin films (Figure 6g). The different surface orientation-OER activity relationships exhibited by NNO and LNO could be due to their intrinsic material behaviors and warrant further study.

Defect Engineering
The introduction of defects (e.g., RP faults, point defects, and V O ) is also a simple and effective method to influence OER activity by inducing structural distortion of the BO 6 octahedra, tuning the e g orbital occupancy and Ni 3d-O 2p hybridization, and shifting the E F position and changing Δ. For example, Bak et al. [61] demonstrated that distorted NiO 6 octahedra in the RP faults of LNO thin films could provide more active sites for OER. These authors found that 2D RP homologous faults can be generated by controlling the cation nonstoichiometry in LNO films. HAADF-STEM images in Figure 7a show RP-type LaO-LaO faults throughout the entirety of 5%-La-excess LNO films and not in the 5%-Niexcess films. A significant elongation of the NiO 6 octahedra was observed in the RP faults compared to the normal bulk LNO lattice. Annular bright-field images (the inset of Figure 7a) show that an over 20% elongation of the Ni-O bonds in the RP faults leads to the strong tetragonal distortion of the NiO 6 octahedra. DFT calculations support that the tetragonal distortion of NiO 6 octahedra is energetically favorable. The protrusion of the oxygen anions toward the RP fault plane can screen the electrostatic repulsion between two positively charged [LaO] + layers, reducing the local repulsive instability. According to Jahn-Teller theory, tetragonal distortion along the z-axis induces an e g orbital split with a lower d z 2 level. The improved linearity of the Ni-O bond ( Figure 7b) and lowered d z 2 level increases the overlap between Ni d z 2 and O 2p orbitals, leading to facile charge transfer between the electrocatalyst and adsorbates. Therefore, the distorted NiO 6 octahedra in the La-excess thin film with RP faults can be much more active for OER, leading to significantly enhanced OER current densities (Figure 7c).
Hu et al. [62] systematically investigated the contributions of V O and B-site cation orbital occupancy to the OER performance of NNO films. They used PLD to deposit epitaxial NNO films on STO single crystalline substrates at different oxygen pressures. As shown in Figure 7d, the (002) diffraction peak of NNO films shifted to lower angle with decreasing oxygen pressure due to an expansion of the OOP lattice constant. XPS spectra show how the oxygen pressure during deposition affects the V O concentration and nickel valence state. The Ni 3+ /Ni 2+ ratio increases from 0.48 to 2.02 as the oxygen pressure increases from 0.2 to 60 Pa, suggesting a higher Ni 3+ fraction in the NNO films. Figure 7e shows a comparison of OER current densities measured at 1.6 V versus RHE for all samples. The OER performance of the NNO films ex-hibits a volcano-like trend as a function of oxygen pressure. DFT simulations indicated that the change in the Ni 3+ /Ni 2+ ratio could affect the e g orbital filling whereas the V O concentration can affect the Ni 3d-O 2p hybridization (see inset in Figure 7e). Optimizing the Ni 3d-O 2p hybridization can enhance OER activity in the 0.2 to 10 Pa of oxygen pressure range. Varying the V O concentration can enhance OER catalytic activity by providing an appropriate bond strength to intermediate hydroxyl OH*. The lack of sufficient V O concentration at high oxygen partial pressure and the excessively strong hybridization strength in NNO grown at low oxygen partial pressure can both restrict the OER catalytic activity. Therefore, the OER catalytic performance shows a volcanolike trend according to the oxygen pressure, and the highest OER catalytic activity is shown in the film grown at 10 Pa with the appropriate V O concentration and Ni 3+ /Ni 2+ ratio.
Guo et al. [63] argue that introducing V O and aliovalent dopants in NNO has a detrimental effect on OER performance. As shown in Figure 7f, Ni K-edge XANES measurements exhibit adsorption edge shifts toward lower energy and a pre-edge hump decrease when V O or H + point defects are introduced in NNO thin films, indicating a decrease in the Ni valence from Ni 3+ to Ni 2+ . The authors suggest that the phase transition triggered by point defects leads to a more insulating state with a significantly weaker Ni-O bond covalency and higher Δ, leading to reduced OER performance.
In the work of Hu et al., [62] the vacancies were promoted by reducing the oxygen partial pressure during the dynamic growth process. On the other hand, Guo et al. [63] introduced  6 . c) OER activity comparison between 5%-La-excess and 5%-Ni-excess LNO films. The inset compares OER current densities measured at 1.63 V versus RHE. a-c) Adapted with permission. [61] Copyright 2017, American Chemical Society. d) XRD patterns around the (002) peak of NdNiO 3 (NNO) films grown on SrTiO 3 (001) substrates under a range of oxygen partial pressures. e) Growth oxygen partial pressure dependent OER activity of NNO films measured at 1.6 V versus RHE. The inset shows a schematic energy band diagram for different states of NNO on SrTiO 3 . Reproduced with permission. [62] Copyright 2019, Wiley-VCH. f) Normalized Ni K-edge near-edge X-ray absorption fine structure (NEXAFS) spectra of pristine, oxygen deficient, and proton-doped NNO thin films. The inset shows the pre-edge hump comparison. f) Adapted with permission. [63] Copyright 2021, American Chemical Society.
the vacancies through annealing as-grown films in Ar. The apparent differences in OER performance between these two studies may result from the different sample synthesis and processing approaches used, suggesting that a significantly deeper understanding of these materials is required to pinpoint the exact role(s) of defects. Using a similar post-growth annealing approach, Wang et al. [24] found that the E F of NNO films shifts up with vacuum annealing, moving away from the redox level (4.90 eV) of electrolyte (pH 13). This shift is unfavorable for OER [14,27b] and consistent with the observations of Guo et al.

Interface Engineering
As discussed in the above section, Baeumer et al. [35d] reported that surface termination is a crucial factor for determining the transformation pathway and consequent OER activity. Specifically, NiO 2 -terminated LNO thin films are more active than LaOterminated LNO thin films. Therefore, finding a system for finetuning the electronic properties of NiO 2 may provide new opportunities to further improve OER activity. Song et al. [19] have shown that the layered perovskite RP series La n+1 Ni n O 3n+1 (synthesized by oxide MBE) can be an ideal material system to elucidate the critical factors that affect the OER activity of nickelates. As shown in Figure 8a, the LaO-LaO layers break the NiO 6 octahedral network, disrupting the hopping of charge carriers between layers of perovskite blocks. The dimensionality of the electronic structure of this layered structure can be controlled, from quasi-2D La 2 NiO 4 (n = 1) to 3D LNO (n = ∞). These drastic changes induce the distinct electronic ground states, ranging from insulator of La 2 NiO 4 (n = 1) to metal of LNO (n = ∞). The nominal Ni valence and the e g orbital filling can also be changed  10 thin films. c) Current density at 1.8 V versus RHE and Tafel slope as a function of e g occupancy. Adapted with permission. [19] Copyright 2021, American Chemical Society. from 2 (Ni 2+ ) to 1 (Ni 3+ ) as n is increased from 1 to ∞. The larger resistances for n = 1 and 2 make them unsuitable for OER measurements. Thus, Figure 8b only displays the OER current densities measured in O 2 -saturated 0.1 m KOH electrolyte for n = 3-6 and ∞. Of all the samples, LNO (n = ∞) shows the lowest OER activity even though it is the most conductive. Compared to LNO, all layered RP nickelates show improved OER performance. La 6 Ni 5 O 16 (n = 5) shows the highest OER activity. These authors noticed that the e g filling is 1.2 for La 6 Ni 5 O 16 (Figure 8c), which is consistent with the optimal e g occupation at the top of the volcano-like curve from various TM perovskite oxides. [11] To investigate whether the e g filling is the sole factor or just the dominant driving force for enhancement of OER activity, band filling induced by cation substitution for doped charge carriers was explored. By substituting 9.75% La 3+ with Sr 2+ in the La 4 Ni 3 O 13 (n = 3), the OER activity is only slightly increased although they tuned the nominal e g filling to 1.2. Since the crystal structure and dimensionality remain the same, a minor cation substitution should have little effect on Ni-O hybridization and band alignment. This suggests that both e g filling and Ni-O hybridization are key to determining OER performance and that the combination of these two factors leads to maximum OER activity at n = 5.

Summary and Perspectives
Epitaxial growth of complex nickelate thin films has allowed researchers to realize precise control over film structure, composition, strain state, crystallographic orientation, and active sites on the surface. A detailed materials understanding of such well-defined model systems is a prerequisite to test the proposed descriptors and establishing defensible structure-OER activity relationships. The design strategies discussed here have proven effective for tuning and optimizing the OER activity of various RNOs and are applicable to other materials systems. It should be noted that different tunable strategies can bring the modulation in different key parameters in nickelates, which in turn, affect the OER activity from different pathways. Thus, it is important to combine different approaches to test various hypotheses and pinpoint the exact controlling factor.
Nevertheless, there are still several challenges to be met in order to put these materials to practical use. First, the OER activity of most RNOs is still insufficient and only a few examples show activities comparable to those of precious metal oxide-based electrocatalysts. Therefore, the rational design of new catalysts based on perovskite nickelates with superior OER performance remains a challenge. As discussed in Section 3, a single design strategy, such as strain tuning or composition control (A-or B-site substitution), leads only to limited performance improvement in most cases. Utilizing the synergistic effects of multiple design strategies offers new opportunities to further improve OER activity. For example, taking advantage of strain engineering and dual-site cation substitution is a promising direction. Besides cation substitution, anion doping (with species such as F [64] and N) [65] in the perovskite lattice can decrease the V O formation energy and tailor the surface electronic structure and this approach could be another future design strategy for improving OER activity. Second, the OER mechanisms for RNOs with or without A-(B-, or dual-) site substitution, particularly the role of lattice oxygen, are still not fully established. Isotope labeling of oxygen has been used to track oxygen exchange between liquid/oxide interfaces during OER. [35c,66] Liu et al. [35c] studied the role of lattice oxygen using well-defined LNO epitaxial thin films through 18 O-isotope labeling and found that lattice oxygen exchange occurs on LNO during OER. Further experimental and theoretical efforts are needed to elucidate the OER mechanisms of RNOs with or without A-(B-, or dual-) site substitution, enabling the predictive design of robust OER electrocatalysts. Third, the relationship between the composition and the RNO film stability during OER remains poorly understood. The important mechanisms affecting OER stability include elemental leaching, lattice oxygen evolution, surface reconstruction, and dissolution processes. [67] For example, STEM/EDX measurements after OER testing demonstrated that Ni leaching or segregation contributes to the amorphization at the surface during the OER process. [53,68] Moreover, many Ni-based oxides commonly exhibit surface reconstruction, leading to the formation of a more active nickel oxy-hydroxides (NiOOH) layer at positive potentials. [35d,69] For example, by using UV-vis spectroelectrochemistry, Baeumer et al. found that a NiOOH-like surface monolayer forms on Ni-terminated LNO surfaces, but it is not observed on La-terminated films. [35d] Therefore, advanced in situ/operando electrochemical characterization tools, such as Raman spectroscopy, [70] XAS, [71] AFM, [72] and TEM, [73] need to be performed in the near future to generate insight into the dynamic surface changes of RNO electrocatalysts at the atomic scale and establish new principles for the effective design and use of these materials.
Over the past two years, high entropy perovskite oxides (HEPO, composed of five equimolar TMs in cation sublattices) have emerged as potential high-activity electrocatalysts for water oxidation. [74] Compared with materials having relatively simple compositions, high cation disorder design in HEPOs can broaden the compositional tunability toward optimizing the electronic structures to achieve nearly continuous binding energy distributions for further promoting OER performance. [75] Moreover, due to the size mismatch between different cations, the high entropy configuration design can result in larger lattice distortion and increase the diffusion barriers, preventing the collapse of the oxygen close-packed structural framework and cation dissolution during OER, thereby enhancing the structural stability of the electrocatalysts. [76] Therefore, HEPO nickelates may offer a practical platform for designing more efficient and stable OER electrocatalysts.
www.advancedsciencenews.com www.advenergymat.de Le Wang is a materials scientist in the team of Atomically Precise Materials at PNNL. He received his Ph.D. degree from the Institute of Physics Chinese Academy of Sciences in 2014. After that, he worked at Nanyang Technological University, Singapore as a research fellow. He joined PNNL as a Postdoc in 2017 and became a staff scientist in March 2020. His research interest focuses on the synthesis of epitaxial oxide thin films and heterostructures and exploring their potential in electronics, ion diffusion/transport, and electrocatalysis.
Yingge Du is a senior staff scientist and team lead of Atomically Precise Materials within the Materials Group at PNNL. He received his Ph.D. degree from the University of Virginia in 2007. His current research interests include the predictive synthesis and processing of epitaxial complex oxide heterostructures and their applications in microelectronics and clean energy.