Water Oxidation Catalysts: The Quest for New Oxide-Based Materials

Abstract

The expected shortage of fossil fuels as well as the accompanying climate change are among the major challenges of the 21st century. A global shift to a sustainable energy landscape is, therefore, of utmost importance. Over the past few years, solar technologies have entered the energy market and have paved the way to replace fossil-based energy sources, in the long term. In particular, electrochemical solar-to-hydrogen technologies have attracted a lot of interest—not only in academia, but also in industry. Solar water splitting (artificial photosynthesis) is one of the most active areas in contemporary materials and catalysis research. The development of low-cost, efficient, and stable water oxidation catalysts (WOCs) remains crucial for artificial photosynthesis applications, because WOCs still represent a major economical and efficient bottleneck. In the following, we summarize recent advances in water oxidation catalysts development, with selected examples from 2016 onwards. This condensed survey demonstrates that the ongoing quest for new materials and informed catalyst design is a dynamic and rapidly developing research area.

1. Introduction

Solar water splitting, namely the solar-driven splitting of water into molecular hydrogen and oxygen is one of the most promising routes to sustainably solve the issue of energy storage and transport, in the long term [1]. The overall water splitting process is schematically given as follows:

2H₂O + energy → 2H₂ + O₂

(1)

Typical device setups for water splitting are based on electrolysis or electrochemical water splitting cells. The involved half-cell reactions in the latter are referred to as the oxygen evolution reaction (OER) at the anode of the cell and the hydrogen evolution reaction (HER) at the cathode, which, for alkaline media, are defined as:

HER: 4H₂O + 4e⁻ → 2H₂ + 4OH⁻

(2)

OER: 4OH⁻ → O₂ + 2H₂O + 4e⁻

(3)

The minimum operational requirement for a water splitting system is the equilibrium voltage of 1.23 V at room temperature [2,3]. Nevertheless, in real systems a higher voltage, compared to the thermodynamic value, must be applied to expedite the electrochemical OER. This extra-potential is called overpotential (η), and its minimization is a prerequisite for the development of highly active and cost-effective electrocatalysts.

In principle, there are two main technological pathways for energy capture to drive the water splitting reaction. The first is the coupling of photovoltaics (PV) or any other grid-scale renewable energy source with electrochemical water splitting devices, i.e., electrolyzers [4]. An alternative route of solar to energy conversion are photoelectrochemical (PEC) devices [5,6,7]. Typically, the latter technology integrates semiconductor materials into the electrode design to directly harvest solar energy. Recently, research and development comparisons for both pathways have been subject to intense scientific debates. The first category of devices is definitely closer to large-scale technical implementations, whereas the PEC devices are at an early stage technological maturity, but provide plenty of room for flexible and off-grid setups [8,9].

Given the demanding four electron transfer process involved in the water oxidation reaction, the development of efficient WOCs remains a major step towards artificial photosynthesis [10]. The past decade has witnessed great interest in the design of single and multi-metal materials and of their underlying substrates as an efficient strategy to improve the overall catalytic activity and stability of such OER catalysts [11]. Along these lines, the synergistic effects involving layered hydroxides, spinels, amorphous oxides, and other tunable oxide-based materials have resulted in an exceptional enhancement of the catalytic performance [11]. Currently, some noble metals and their oxides, such as Ru, Ir, RuO₂, and IrO₂ still remain the best catalysts for the OER, in both aqueous and acidic media. However, their high cost and low abundance are major roadblocks for their industrial application [12,13].

This review does not intend to be exhaustive, but rather aims for an update on the recent advancements in metal oxide-based OER catalysts. Hence, selected articles from 2016 to 2018 are summarized in the following sections and are rounded-off with a brief outline of the unresolved bottlenecks in this research area. The final section of the review further summarizes new research directions towards innovative composite materials.

High performance metallic WOCs are sometimes scarce, costly, and sensitive towards chemical changes, whilst many metal oxides are earth-abundant and exhibit superior durability. Metal oxides are involved in a wide range of technical processes and represent the major part of industrially applied catalyst families. Their rapid ascent traces back to the mid-1950s when they were found to catalyze a variety of oxidation and acid-base reactions. The manifold properties of oxide materials, such as electrical conductivity, lattice oxygen ion mobility, atomic or ionic diffusion among interfaces, catalytic activity, and various optical and electronic properties, are frequently associated with the presence of defects, in contrast to an ideal ionic crystal. Oxides contain different types of defects, including electron-hole pairs, excitons, vacant lattice sites, interstitial atoms, impurity atoms at interstitial or substitution locations, dislocations, and stacking faults [14]. Consequently, catalytic reactions on metal oxides are favored, due to their surface flexibility under catalytic conditions. In the following, we outline the quest for understanding and constructing new complex oxide materials and related catalyst and composite types, which still remain to be fully explored in their infinite structural variety.

2. Recent Developments of Metal Oxide-Based WOCs

2.1. Ru-, Ir-, and Cu-based Metal Oxides

In this section, the selected articles on Ru-, Ir-, and Cu-based metal oxides received immediate recognition in the field and covered research conducted over the last three years [15,16,17,18,19,20,21,22]. Emphasis has been placed on mechanistic insights and operando techniques, as well as on the proposed Cu-based catalysts, as a possible alternative to the more costly Ru- and Ir-based WOCs. The selected studies applied the stability and onset potential required during the OER as benchmarking parameters. Generally, this focused review article places selective emphasis on progress in oxide WOC performance, as a result of new synthetic strategies (for examples of reviews containing information on operando methods [10,23,24]).

All catalysts in this section are discussed and summarized below in

Table 1. Comparison of the electrocatalytic activity of recently reported Ru-, Ir-, and Cu-based catalysts (“durability” refers to maintenance of the catalytic performance at constant current conditions, while “stability” indicates the resistivity of a given catalyst towards dissolution/leaching of the active metal sites).

Catalyst	Preparation Method	Onset Potential (V)	η (at 10 mV/cm2) (V)	Tafel Slope (mV dec−1)	Durability (h)	Stability	Electrolyte	Ref.
Ru and Ir	Sputtering & evaporation	Ru: 1.20 (both media), Ir: 1.27 & 1.43 (acidic & alkaline media, respectively)	-	30–40	0.25	Low	0.1 M H2SO4 & 0.05 M NaOH	[19]
RuO2 and IrO2	Sputtering & evaporation	RuO2: 1.35 & 1.47 (acidic & alkaline media, respectively), IrO2: 1.47 (both media)	-	30–40	0.25	Good *	0.1 M H2SO4 & 0.05 M NaOH	[19]
IrOx/SrIrO3	Pulsed laser deposition (PLD)	1.44	0.27	-	30	Reasonable	0.5 M H2SO4	[16]
Cu, Cu2O, Cu(OH)2 and CuO	Electrodeposition	-	0.417	60–66	22	Cu & Cu2O: very good *, CuO & Cu(OH)2: poor *	0.1 M KOH	[17]
IrOOH nanosheets	Exfoliation in TBAOH	-	0.344	58	14	Extraordinary	0.1 M HClO	[15]
Cu2O/ITO	Electrodeposition	<0.8	0.4	72	5	High	K2B4O8, C2H3KO2 & KH2PO4	[20]
Cu(OH)2@CoCO3(OH)2· nH2O	Hydrothermal	-	0.27	78	20	Superior	1 M KOH	[18]
RuO2 (001)	Commercially available	-	-	-	2	-	0.05 M H2SO4	[21]
IrxO/ATO	Electrodeposition	-	0.28	61.7	15	High *	0.05 M H2SO4	[22]
TiO2/CuO	Spin-coating & pulsed laser deposition	-	-	-	0.166	Excellent	0.1 M NaOH	[26]

. In the following survey tables, “durability” refers to maintenance of the catalytic performance, at constant current conditions, and “stability” refers to the resistivity of a given catalyst towards dissolution/leaching of the active metal sites. Whenever possible, the descriptions of the catalysts’ stability, in the following tables, represent the exact wording as given by the authors. Exceptions are stability assessments marked with an asterisk sign (*); these were added by us after going through the data provided in the corresponding publications. In this review, the various metal oxide families reported in the next sections are summarized in tables according to the information on the durability and stability of catalyst performance. Unless stated otherwise, the symbol “-“ indicates that no such parameters were reported in the cited study. The Tafel values in this review are generally cited as efficiency indicators of the electrocatalytic performance. This criterion is particularly important in water electrolyzers, for instance, where high current densities, at the lowest possible operational voltages, are vital for the economic viability. Furthermore, Conway et al. clarified in a representative study, three decades ago, that Tafel slope values can provide both quantitative and mechanistic insights into electrocatalytic processes [25]. On a very fundamental level, the Tafel slope is characterized by the rate of increase of the electrode potential or polarization with the log of the current density. From a more practical point of view, along with the current density obtained at a certain overpotential, the rate of change of this current density with the overpotential is also important. Hence, low Tafel slope values are generally recognized as a key parameter indicating efficient electrocatalytic materials.

Weber et al. proposed that an exfoliation route in tetrabutylammonium hydroxide (TBAOH) solution to convert bulk IrOOH to IrOOH nanosheets could open the doors to fundamental studies and mechanistic insight. This approach might as well be of great interest for industrial applications in heterogeneous catalysis [15]. This material outperformed the state-of-the-art rutile-IrO₂, under the same conditions. The IrOOH nanosheets were deposited on Ti electrodes and displayed an overpotential of 344 mV and Tafel slope of 58 mV dec⁻¹, in 0.1 M HClO₄. Chronoamperometric and XPS measurements revealed the long-lasting properties of this material. In particular, XPS showed that the trivalent species of IrOOH remained, after the OER process. Techniques for heterostructured and high quality growth of epitaxial complex oxide thin films have received increasing attention with regard to their application as water oxidation catalysts.

Seitz et al. reported on the significance of leaching in the enhancement of stability and catalytic activity [16]. In particular, they demonstrated that Sr leaching from SrTiO₃ thin films caused surface rearrangement of IrO_x films, which led to in situ formation of an IrO_x/SrIrO₃ catalyst with only 0.27 V_RHE overpotential and high stability for a long operational time of 30 h. Figure 1 displays atomic force microscopy (AFM) maps of a smooth as-prepared surface, before electrocatalytic activity, and uniform small features after electrocatalytic tests, thus illustrating the surface rearrangement. Further evidence is provided by the XPS data in Figure 2. The signal of the Sr 3d peak was attenuated, after 30 min of electrocatalytic activity. The high stability and low overpotential values of 60 and 100 nm IrO_x/SrIrO₃ electrocatalysts, respectively, after 0 and 10 min, as well as 2 and 30 h of catalytic tests are displayed in Figure 3.

Cu and Cu-oxide surfaces are widely known as catalysts of CO₂ reduction. However, Deng et al. identified that Cu materials can be very promising, earth-abundant, efficient, and low-cost catalysts for the OER, when subjected to careful designing of their morphology [17]. Their work provided new progress on both Cu surfaces, with various oxidation states, as well as on operando monitoring techniques. In situ Raman spectroscopy of Cu, Cu₂O, Cu(OH)₂, and CuO on metallic copper disks was employed to study the OER in alkaline media. Different morphologies were observed, from small-pyramid-like structures to needle-shaped structures, as well as nanoparticles <20 nm, as shown in SEM images (Figure 4).

The corresponding Tafel slope values were in the range of 60–66 mV dec⁻¹ and are comparable to those of other metal oxides in the literature. Along with the morphology of the Cu samples, the role of Cu(III) species was considered to play an important role in OER.

Cyclic voltammograms and Raman experiments were performed, showing a cathodic peak at around 1.55 V_RHE (cf. Figure 5), which was attributed to Cu (III) species. A Raman peak at 603 cm⁻¹ appeared under the oxygen evolution potential and Cu(III) oxide formation was further demonstrated by in situ XANES of the Cu K-edge. Given the aforementioned information, the authors drew the conclusion that Cu(III) oxide species were responsible for the electrochemical water oxidation.

Alternatively, a Cu(OH)₂@CoCO₃(OH)₂⋅nH₂O core-shell structure nanowire array was proposed by Xie et al., as a catalyst for the oxygen evolution reaction [18]. The catalyst showed strong stability and could be a promising candidate for other applications as well, such as methanol electro-oxidation.

A study of Amiri et al. significantly improved the onset potential for water oxidation at a value below 0.8 V_RHE, by using a Cu₂O thin film on indium tin oxide (ITO) electrodes [20]. Another aspect of their work is the facile electrodeposition method using copper perchlorate, various diamines, and 1,4-bis(2-carboxyaldehyde phenoxy)butane precursors. The stability of the catalyst was high. The electrode could be used for multiple OER experiments, without loss of its activity in various electrolytes from potassium borate to potassium acetate and potassium phosphate, as observed by the authors.

The work of Roy et al. showed the importance of different facets of a catalyst, not only for the OER activity, but also for corrosion [21]. The authors investigated how oxygen evolution and corrosion levels arising from different sites influence future catalyst engineering. Evidence was provided that there is no direct correlation between OER activity and stability of the various facets of the catalyst and commercial nanoparticles. Well-defined, oriented, thin films of rutile RuO₂ (i.e., 001, 101, and 111) were studied in acidic solution and were found to be more active towards OER, compared to (110) single crystalline and commercial RuO₂ particles. More precisely, potentiostatic experiments over 2 h on RuO₂ thin film (001) orientations showed steady state currents, at around 150 μA cm⁻², at the applied potential of 1.6 V_RHE, whilst the (101) and (111) orientations provided currents around 90 and 40 μA cm⁻², respectively. Moreover, the RuO₂ (110) surface gave a higher steady state current density at around 50 μA cm⁻², compared to the commercial RuO₂ particles, which showed the lowest steady state current density, at approximately 20 μA cm⁻². Spectrometric measurements revealed that higher amounts of Ru on all of the aforementioned oriented thin films dissolved in the acidic electrolyte, compared to the (110) single crystal and particle references, where the amounts of dissolved Ru were found to be much smaller after 2 h of electrocatalytic tests. More precisely, 225 ng Ru cm⁻² of the RuO₂ (001) oriented thin films was dissolved after 2 h of electrocatalytic experiments, compared to 190, 150, 13, and 10 ng Ru cm⁻² of the RuO₂ (101), (111), (110), and commercial particles, respectively.

In related lines, Suk Oh et al. studied the role of various electrode supports towards the OER activity and electrochemical stability of pure IrO_x nanoparticles, IrO_x-supported on carbon (IrO_x/C), and IrO_x-modified commercial and non-commercial antimony-doped tin oxide (IrO_x/comm. ATO and IrO_x/ATO, respectively) [22]. The motivation of their study was based on the ability of catalyst-supports to reduce the overall metal-loading, without lowering the durability of the electrode. Outcomes of their study showed a much higher stability and corrosion-resistance of the IrO_x nanoparticles on the metal/metal oxide supports, compared to carbon, as well as high charge-transfer kinetics. The faradaic efficiency of the IrO_x/ATO catalysts was found to be 100%, for OER, which showed their high selectivity towards water oxidation. Furthermore, the oxidation state of the IrO_x/ATO was reduced to 3.2+, compared to the unsupported IrO_x nanoparticles (4+), after long-term electrochemical experiments revealing the metal/metal oxide support effects. Furthermore, voltammetric characterizations were performed and it was found that the electroactive species of IrO_x-supported ATO nanoparticles were decreased by only 30% after electrocatalytic activity of 15 h, whereas IrO_x loading on C decreased by 99.7%, under the same conditions.

2.2. Noble Metal-Free Oxide-Related WOCs

2.2.1. Layered Hydroxides and Related Ni–Fe-Based Materials

Layer double hydroxides (LDHs) are a class of layered materials, which are also known as anionic clays. Their characteristic structural feature is the generic layer sequence [AcB Z AcB]_n, where c represents layers of metal cations, A and B are layers of hydroxide (HO⁻) anions, and Z are layers of other anions and neutral molecules [27]. Nowadays, LDHs find widespread application in dye-sensitized solar cells, photocatalytic water oxidation, chemical sensors, and supercapacitors, thanks to their open structures and chemical versatility [28]. Recently, several layer double hydroxide catalysts have been developed for OER catalysis [29,30,31,32,33].

Presently, NiFe-LDH catalysts are among the leading LDH-based WOC materials. Even though water oxidation on nickel hydroxide has been studied since the 1960s, the clear-cut assignment of Ni as the active site remains under debate for mixed metal LDHs. Some recent studies proposed that the incorporation of Fe into Ni-LDH structures provides more active sites as a potential for the enhanced OER activity [34,35]. This behavior has been further supported by Nocera et al., who argued that the enhanced OER activity is promoted by the Lewis acidity of Fe(III) [36]. The group of Duan et al. demonstrated that a charge imbalance in the M(OH)₆ layers is caused by the redox-active Fe(III) ions, which is then compensated by interlayer anions [37]. All of the representative layered hydroxide OER catalysts discussed below and summarized in

Table 2. Comparison of the electrocatalytic activity of noble metal-free LDH and related Ni–Fe catalysts (cf. Table 1 for definitions of “durability” and “stability”).

Catalyst	Preparation Method	Onset Potential (V)	η (at 10 mV/cm2) (V)	Tafel Slope (mV dec−1)	Durability (h)	Stability	Electrolyte	Ref.
CoNi-LDH/CoO nanosheets	In situ reduction & interface-directed assembly in air	1.48	0.3	123	10	Good	1 M KOH	[29]
CoAl & CoFe LDH	Co-precipitation	1.58	0.36 & 0.37	39 & 49	50	Excellent	0.1 M KOH	[30]
Co-Fe LDH nanosheets	Electrodeposition	-	0.42	131	50	Superior	0.1 M K2B4O8	[31]
Ni0.75–V0.25 LDH narrow nanosheets	Hydrothermal	-	0.35	50	25	Considerable	1 M KOH	[32]
CoAl-LDH on 3D graphene network	Hydrothermal	-	0.28	36	30	Excellent	1 M KOH	[33]
NiFe LDHs-MoOx	Hydrothermal	-	0.276	55	12	Good	1 M KOH	[38]
NiFe-LDH array	Electrodeposition	1.43	0.224	52.5	50	Superior	1 M KOH	[37]
Fe-doped Ni(OH)2 films	Precipitation	-	0.25 to 0.35	-	2	Good *	1 M KOH	[36]
Ni–FeO(OH)	Microwave-assisted autoclave synthesis under solvothermal conditions	-	0.29	40	>20	Good	0.1 M KOH	[39]
Fe2O3–Ni(OH)2 to NiFe2O4–NiOOH	Hydrothermal & anodization	-	0.24	55	1	High	1 M KOH	[40]
Ni–Fe-OH@Ni3S2	Two-step synthesis	-	0.479	124	50	Excellent	1 M KOH	[41]
Ni/NiFe-OH	Multi-step electrodeposition	-	0.205	53	12	Excellent	1 M KOH	[42]
NiFe-LDH nanosheets	Pulsed laser ablation in liquids	-	0.27-0.45		3.5	Remarkable	1 M KOH	[43]
NiFe LDH-NS@DG	Co-precipitation	1.41	0.21	52	10	Robust	1 M KOH	[44]

covered research of the last three years. The maximum performing NiFe-LDH-based catalyst to date is a NiFe-LDH array, which is also specified in Table 2 [37]. The following selected studies contributed to new catalyst architectures and designs with improved stability. Furthermore, considerable efforts were made over the past few years to replace Fe with other transition metals.

Wu et al. reported on ultrathin CoNi double hydroxide/CoO nanosheets, via an in situ reduction and interface-directed assembly method in air, as shown in the schematic illustration of Figure 6 [29]. Interfacial tension resulted in the strong extrusion of hydrated metal-oxide clusters, which led to the formation of the CoNi LDH/CoO nanosheets. The current density for OER was determined as 150 mA cm⁻² at 0.3 V_RHE, i.e., 1.5 times greater than that for RuO₂. The durability of the catalyst was also tested and the current density remained stable for 10 h. The obtained CoNi LDH/CoO nanosheets exhibited high turnover frequencies (TOFs) of 1.4 s⁻¹ at the overpotential of 0.4 V_RHE. The authors argued that the performance of their catalysts was mainly related to the higher +3 valence of the Co and Ni centers, as well as to the low crystallinity, which gave rise to a number of exposed defects and active sites. Furthermore, the functionalization of the surface with OH⁻ groups and the OH⁻/O₂ adsorption and transport in the porous structure contributed to the high OER activity.

Ping et al. introduced a CoAl-LDH nanosheets catalyst on a 3D graphene network [33]. In their study, a porous catalyst was prepared by self-assembly of the exfoliated single-layer CoAl nanosheets onto 3D graphene, via electrostatic interaction as illustrated in Figure 7. Prior to the exfoliation, the CoAl-LDH (CO₃²⁻) was fabricated through a hydrothermal process, based on urea hydrolysis, and subjected to anion exchange. The resulting catalyst outperformed an IrO₂ reference, by exhibiting a very low overpotential of 0.28 V_RHE and a Tafel slope value of 36 mV/dec. The TOF values of the 3DGN/CoAl-NS were found to be ~0.6 s⁻¹ and ~1.2 s⁻¹ at overpotentials of 0.3 and 0.35 V_RHE, respectively. Furthermore, about 22.5 μmol of O₂ was produced at a constant current of 1 mA cm⁻². Chronoamperometry and chronopotentiometry experiments were carried out and indicated excellent electrochemical stability for 30 h (Figure 8). The electrocatalytic efficiency of the catalyst was ascribed to the fast electron/charge transfer, during the OER, and acceleration of the reaction kinetics thanks to the uniform coating of the single layer CoAl nanosheets onto 3D graphene.

Yang et al. demonstrated the effect of Fe and Al contents in Co–Fe LDH catalysts on the OER activity [30]. A co-precipitation method with Al³⁺ or Fe³⁺ ions in the range of 15 to 45 at % was applied. The optimal Fe content was reported at 35 at %, which resulted in a synergistic effect on the Co–Fe LDH catalyst. The role of both ions as trivalent species was essential for the stabilization of the LDH structure. The OER activity was enhanced when the optimal Fe content was incorporated, whereas the oxygen evolution was suppressed in the presence of Al³⁺ species. The onset potential and overpotential of the optimized Co_0.65Fe_0.35(OH)₂ catalyst were observed at 1.58 V_RHE and 0.36 V_RHE, respectively. The catalyst showed a very good stability in aqueous alkaline electrolyte, for 48 h, with a small Tafel slope of 49 mV dec⁻¹.

An alternative LDH catalyst was introduced by You et al. [31]. An electrodeposition method was employed to fabricate a CoFe-LDH nanosheet array coated with an ultrathin CoFe-borate layer supported on Ti mesh. The electrocatalytic performance of the 3D catalyst electrode was evaluated in near neutral pH in 0.1 M K₂B₄O₇ solution. The CoFe-B_i@CoFe-LDH NA/TM displayed a high catalytic OER activity, with a small overpotential of 0.42 V_RHE. HRTEM images depicted the formation of an ultrathin amorphous layer (CoFe-B_i), with a thickness between 5 and 8 nm and a nanoarray feature of the catalyst was, furthermore, observed during morphological characterizations (Figure 9) and retained after electrocatalytic activity for 50 h. The TOF was calculated at 0.482 mol O₂ s⁻¹ at the overpotential of 0.6 V_RHE.

Xie et al. reduced MoO₄²⁻ intercalated nickel-iron LDHs and obtained a NiFe-MoO_x nanosheets catalyst [38]. In terms of water oxidation, the authors obtained comparable results to RuO₂. These results were ascribed to the high surface area, porosity, and surface electronic structure of the new LDH material. Their strategy led to an increase of active sites and an electronic modification of the NiFe alloy surface. The TOF for this catalyst was estimated at 0.19 s⁻¹ at the overpotential of 0.3 V_RHE and a faradaic efficiency around 95% for OER was determined.

Furthermore, a Ni–V monolayer double hydroxide was introduced by Fan and co-workers [32], applying a hydrothermal method to synthesize the bulk LDHs. The performance of this material is comparable to the best-performing NiFe-LDHs to date in the literature. In particular, the Ni_0.75–V_0.25 LDH gave a current density of 57 mA cm⁻², at an overpotential of 0.35 V_RHE, for the OER. Experimental O₂ evolution was evaluated by gas chromatography, where the Ni_0.75–V_0.25 LDH exhibited a TOF of 0.05 s⁻¹ at 0.35 V_RHE overpotential. The abundance of the active sites in the LDH structure was identified as the main reason for the high performance of Ni_0.75–V_0.25-LDH. Typical TEM images of α-Ni(OH)₂ and Ni_0.75–V_0.25 LDH show the 2D nanosheet nature of the former (Figure 10a,b), whereas a three-dimensional morphology assembly of ultrathin nanosheets was observed upon incorporation of vanadium (Figure 10c,d).

Li et al. reported a new view of the significance of iron doping in a Ni hydroxide catalyst, which promoted the formation of Ni⁴⁺ and, therefore, played an important role in the water oxidation [36]. The strong Lewis acidity of Fe³⁺ was identified as a crucial factor to increase the Ni valence and, thereby, to promote the oxyl character of the intermediates during the O–O bond formation.

Görlin et al. discussed the importance of high-surface area conductive supports and of the electrolyte pH [39]. A microwave-assisted autoclave synthesis under solvothermal conditions was applied for the synthesis of a NiFeO(OH) catalyst. Catalyst support and pH > 13 resulted in enhanced catalytic activity, and higher particle dispersion allowed more metal centers to be accessible and to be electrochemically active. The highest TOFs were determined at 0.1 and 0.2 s⁻¹, respectively, in 0.1 M KOH. The O₂ levels were found to be close to 100% of the faradaic efficiency after online gas chromatography-mass spectrometry (GC-MS) measurements. The role of Fe centers in the metal redox activity was, furthermore, discussed in new detail, based on the UV-vis monitoring data.

A NiFe₂O₄–NiOOH catalyst was prepared through a simple anodization method by Zhang et al. [40]. The catalyst afforded currents up to 30 mA cm⁻² at an overpotential of 0.24 V_RHE and only required 0.41 V_RHE to achieve a very high current density of 3000 mA cm⁻². The faradaic efficiency of O₂ evolution was close to 100%. In this study, it was stated that the excellent OER performance arose from the integration of amorphous NiFe₂O₄ and NiOOH. Finally, an ultrafast synthetic approach of only 5 s was reported by Zou and co-workers for the fabrication of a Ni–Fe-OH@Ni₃S₂/NF electrode material [41]. A schematic illustration of this ultrafast method is displayed in Figure 11. The catalyst gave currents up to 1000 mA cm⁻² at 0.469 V_RHE and displayed excellent catalytic stability for 50 h at the applied potential of 1.6 V_RHE, at the current density of 1000 mA cm⁻² in alkaline media. Furthermore, the faradaic yield of O₂ was found to be up to 95%.

Wu et al. highlighted the importance of designing well-controlled and aligned structures to enhance the catalytic capabilities of electrocatalysts for high-performance electrodes. Starting from this main incentive, for the first time, they fabricated unique hierarchical hollow Ni/NiFe (oxy)hydroxide heterostructured nanotubes radially aligned on a Ni foam, as seen in Figure 12 [42].

The material was used in a home-made electrochemical cell driven by a single battery of 1.5 V and showed excellent stability and overpotential at around 0.2 V_RHE, outperforming the state-of-the-art noble metals Ru and Pt. Hunter et al. employed a pulsed laser ablation technique in liquid media and prepared NiFe-LDH nanosheets with different interlayer anions (i.e., CO₃²⁻, SO₄²⁻, OH⁻, F⁻, Cl⁻ and I⁻) to probe their role in water oxidation catalysis [43]. The anion binding was evaluated by XPS and DFT calculations. Outcomes of their study revealed a relationship of water oxidation activity with the pK_a value of the interlayer anions and suggested that Bronsted or Lewis basicity of the anions was a key parameter for the water oxidation mechanism.

Furthermore, the authors found that nitrite species bound to edge-site Fe centers were correlated with high water oxidation activity, thus, pointing out the active role of the latter in the water oxidation process. The NiFe-LDH nanosheets showed remarkable stability for 3.5 h in 1 M KOH. The overpotential values varied between 0.27 V_RHE and 0.45 V_RHE, depending on the various interlayer anions. Jia et al. synthesized a highly active heterostructured NiFe-LDH nanosheet on defective graphene, via a co-precipitation method [44]. DFT calculations suggested that the high electrocatalytic activity of the NiFe LDH-NS@DG catalyst towards water oxidation arose from the interaction of exposed 3d-transition metal atoms with carbon defects. This synergism was also found to cause the robust stability of the catalyst for 10 h, in 1 M KOH. The catalyst showed current densities above 50 mA cm⁻² at the applied potential of 1.5 V_RHE. The overpotential at the current density of 10 mA cm⁻² was found to be 0.21 V_RHE. A small Tafel slope value of 52 mV dec⁻¹ was also found between the range of 0.1 and 0.15 V_RHE of applied potential. The onset potential of the NiFe LDH-NS@DG catalyst was observed at 1.41 V_RHE and the faradaic efficiency of O₂ levels was 97.5%, after 5 h of chronoamperometric tests, at the current density of 100 mA cm⁻².

2.2.2. Spinels

Complex spinel-type oxides have attracted a lot of research interest as water oxidation catalysts, over the last years, due to their favorable structural properties, giving rise to tunability, high conductivity, and robust operational stability. Spinel oxides adopt the general formula AB₂X₄, where A and B are cations occupying the octahedral and tetrahedral sites in the anionic sublattice (with X representing anions, typically oxygen or less often chalcogens/halogenides). Co₃O₄ is a well-known example of a binary, spinel-type WOC. Ternary spinel-type cobaltites exhibit similar structural and electronic properties to Co₃O₄ with the general composition MCo₂O₄ (M = Zn, Ni, Cu, Mn, etc.).

The gold standard of Co-oxide WOCs, over the past years, has been the amorphous CoP_i catalyst introduced by Nocera et al. [5,45]. The selection of articles below was taken from the most cited and hot papers in the field, reflecting the progress in WOC understanding and performance, over the last three years. A comparison of the electrocatalytic performance of the selected WOC materials is given in Table 3.

Recently, various synthetic strategies were employed to access Co₃O₄ spinels [46,47,48,49]. Wang and co-workers used a sol–gel method to synthesize regular cobalt spinels, comprised of Co²⁺ on the tetrahedral site and two Co³⁺ ions on the octahedral site [47]. Their work was focused on the investigation of site-dependent OER activity, by substituting Co²⁺ and Co³⁺ with inactive Zn²⁺ (ZnCo₂O₄, Co³⁺ octahedral) and Al³⁺ (CoAl₂O₄, Co²⁺ tetrahedral), respectively. The electrochemical performance was tested by cyclic voltammetry, and interestingly the redox reaction was found to occur only at the Co²⁺ tetrahedral sites, as shown in the voltammogram in Figure 13. The Tafel slope value was found to be smaller in the case of Co₃O₄, which suggested a different rate-determining step. This study also pointed out the importance of operando investigations in such systems. From operando extended X-ray absorption fine structure (EXAFS) and electrochemical impedance spectroscopy (EIS) studies, the conclusion was drawn that cobalt oxyhydroxide (CoOOH) was formed and acted as a reactive center for the oxygen evolution reaction. In contrast, the ZnCo₂O₄ catalyst was negatively affected by the strongly bonded –OH groups on the surface, and so, its electrocatalytic activity was limited compared to its counterparts. The larger Tafel slope supported the latter observation, where the OER process is rate limited due to these strong –OH bonds.

Table 3. Comparison of the electrocatalytic activity of spinel catalysts (cf. Table 1 for definitions of “durability” and “stability”).

Catalyst	Preparation Method	Onset Potential (V)	η (at 10 mV/cm2) (V)	Tafel Slope (mV dec−1)	Durability (h)	Stability	Electrolyte	Ref.
Co3O4/N-graphene	Hydrothermal	1.49	0.96	121.8	4	Good *	0.1 M KOH	[50]
Co3O4 assembled hollow spheres	Solvothermal & calcination	1.43	0.29	86	10	Prominent	1 M KOH	[46]
Co3O4, ZnCo2O4 & CoAl2O4	Sol-gel	-	-	69, 113 & 56	-	-	0.1 M KOH	[47]
Co@Co3O4/N-CNT	Pyrolysis in H2 atm. & oxidative calcination	1.62	0.39	54.3	45	Good *	0.1 M KOH	[51]
Hollow Co3O4 microtube arrays	Hydrothermal & electrochemical treatment	1.52	0.151	84	12	High	1 M KOH	[33]
ZnCo2O4 QDs/N-CNT	Hydrothermal	1.56	0.43	70.6	10	High	0.1 M KOH	[52]
NiFe/Ni Co2O4	Hydrothermal & electrodeposition	1.47	0.34	38.8	10	Excellent	1 M KOH	[53]
NiCo2O4 NPs	Two-step solution method	1.49	0.157	75	20	Excellent	1 M KOH	[54]
CoFeOx, CoFeNiOx & FeNiOx	Electrodeposition	1.43	0.24	37	-	-	1 M KOH	[55]
CoFe2O4@Co-Fe-Bi	Hydrothermal	-	0.46	127	20	Good	0.1 M K2B4O8	[56]
Co3O4	Hydrothermal	143	0.35	69	-	-	1 M KOH	[57]

Table 3. Comparison of the electrocatalytic activity of spinel catalysts (cf. Table 1 for definitions of “durability” and “stability”).

Catalyst	Preparation Method	Onset Potential (V)	η (at 10 mV/cm2) (V)	Tafel Slope (mV dec−1)	Durability (h)	Stability	Electrolyte	Ref.
Co3O4/N-graphene	Hydrothermal	1.49	0.96	121.8	4	Good *	0.1 M KOH	[50]
Co3O4 assembled hollow spheres	Solvothermal & calcination	1.43	0.29	86	10	Prominent	1 M KOH	[46]
Co3O4, ZnCo2O4 & CoAl2O4	Sol-gel	-	-	69, 113 & 56	-	-	0.1 M KOH	[47]
Co@Co3O4/N-CNT	Pyrolysis in H2 atm. & oxidative calcination	1.62	0.39	54.3	45	Good *	0.1 M KOH	[51]
Hollow Co3O4 microtube arrays	Hydrothermal & electrochemical treatment	1.52	0.151	84	12	High	1 M KOH	[33]
ZnCo2O4 QDs/N-CNT	Hydrothermal	1.56	0.43	70.6	10	High	0.1 M KOH	[52]
NiFe/Ni Co2O4	Hydrothermal & electrodeposition	1.47	0.34	38.8	10	Excellent	1 M KOH	[53]
NiCo2O4 NPs	Two-step solution method	1.49	0.157	75	20	Excellent	1 M KOH	[54]
CoFeOx, CoFeNiOx & FeNiOx	Electrodeposition	1.43	0.24	37	-	-	1 M KOH	[55]
CoFe2O4@Co-Fe-Bi	Hydrothermal	-	0.46	127	20	Good	0.1 M K2B4O8	[56]
Co3O4	Hydrothermal	143	0.35	69	-	-	1 M KOH	[57]

An electrochemical sacrificial strategy was applied by Zhu et al. for the fabrication of hollow Co₃O₄ microtube arrays [48], which is schematically represented in Figure 14. More precisely, the CoHPO₄ microrods were immersed in 1 M NaOH and a constant potential was applied. Upon the anodic bias application, the microrods’ surface became rough and nanoplates were formed. The protonated CoHPO₄ was dissolved in the NaOH solution and Co²⁺ species acted as source of cobalt oxides to be generated. The nanoporous nature of the cobalt oxide shells provided transport pathways for the OH⁻ and Co²⁺ diffusion. The fully converted CoHPO₄ resulted in Co₃O₄ spinel. These micro/nanostructures were found to be competitive with IrO₂ and Pt references as OER catalysts. The authors proposed that such design and engineering of micro/nanostructures (Figure 15) can be an efficient pathway for the enhancement of catalytic activity through the facilitation of mass transport by exposing more electroactive sites. During the water oxidation reaction, the catalyst showed impressive performance with current densities at around 300 mA cm⁻², almost two times greater, compared to IrO₂. The Tafel slope value was found to be 84 mV dec⁻¹, and the catalyst showed high stability during an operational time of 12 h. The corresponding voltammograms and Tafel slopes are displayed in Figure 16. The catalyst, furthermore, exhibited a high faradaic efficiency of 96.3% for OER.

Hollow spheres assembled by α-Co(OH)₂ nanosheets were developed by Xie and co-workers via a solvothermal method followed by further calcinations, which led to mesoporous Co₃O₄ nanostructures [46]. By tuning the amount of polyvinylpyrrolidone (PVP) different samples were synthesized as shown in Figure 17. The obtained monodisperse hierarchical structure exhibited a low onset potential at around 1.44 V_RHE, with low overpotential (0.29 V_RHE) and relatively small Tafel slopes, in the range of 86 and 97 mV dec⁻¹. The samples indicated an excellent stability in alkaline media which renders them promising for further energy conversion as well as storage applications.

A subcategory of spinels are the above-mentioned cobaltites, which have been used as OER catalysts due to their increased conductivity and versatile electrochemically active sites compared to monometallic oxide catalysts. A two-step solution method was reported by Guan et al. as a way of utilizing metal–organic frameworks (MOFs), for the development of hollow NiCo₂O₄ nanowall arrays on carbon cloth, as shown in Figure 18 [54]. A uniform coverage of the Co-MOF on carbon cloth and the Co-based nanowalls was first obtained, and after annealing in air in the final step, the nanoarray structure of NiCo₂O₄ was maintained. The catalyst displayed high stability for 20 h and low overpotential of around 0.16 V_RHE in alkaline media, which was ascribed to its hollow nature. The material was proposed for other applications besides water oxidation, such as in supercapacitors.

Xiao et al. proposed a 3D hierarchical porous catalyst architecture consisting of NiFe/NiCo₂O₄/Ni foam [53]. The synthesis procedure of this material is displayed in Figure 19. The catalyst reached very high current densities in the range of 1200 mA cm⁻² at an overpotential of 0.34 V_RHE and onset potential at 1.47 V_RHE in 1 M KOH at pH 14. The stability and durability of this material was evaluated through cyclic voltammetry and the current density remained stable, over an operational time of 10 h.

Morales-Guio et al. classified a variety of oxidatively electrodeposited thin films, according to their electrocatalytic activity into the following three categories [55]. Single-metal oxides falling into the first category, such as NiO_x, MnO_x, and FeO_x, showed the lowest performance. The second category included CoO_x and CoNiO_x, which showed a medium activity. Finally, the third category consisted of the most active catalysts, namely FeNiO_x, CoFeO_x, and CoFeNiO_x, which revealed high electrocatalytic performance, due to their synergistic effects. Their efficiency towards water oxidation reached currents up to 100 mA cm⁻², at overpotentials of around 0.24 V_RHE in 1 M KOH, at pH 14. The TOF values as a function of catalyst loading were also evaluated at the overpotential of 0.35 V_RHE. In particular, at the relatively low loading of 5 μg cm⁻², the TOF values of the more active FeNiO_x, CoFeO_x, and CoFeNiO_x catalysts were found to be around 0.5, 5, and 4 s⁻¹, respectively. The calculated faradaic efficiency of O₂ was also close to 100%.

Ji et al., on the other hand, used a complex core-shell CoFe₂O₄@Co-Fe-B_i nanoarray architecture in a potassium borate (pH 9.2) electrolyte and the catalyst showed a superior long-term stability, for at least 20 h, along with a high faradaic efficiency of 96% of O₂ evolution [56].

Regarding the operational behavior of cobalt oxides, Zhang et al. monitored the chemical transformation of multiwall carbon nanotubes supported with hydrothermally prepared Co₃O₄ nanoparticles (Co₃O₄-MWCNTs) to CoO(OH), via ambient pressure X-ray photoelectron spectroscopy (APXPS) [57]. The aforementioned transformation was only observed upon a potential application, during operando conditions. When the potential was removed, the catalyst was converted to its as-synthesized form, namely Co₃O₄. It was suggested that this rapid operando conversion was caused by proton concentration gradients in the presence of water vapor in the local environment, close to the solid/liquid interface. The electrocatalyst afforded current densities up to 100 mA cm⁻², at an applied potential of 1.65 V_RHE. The overpotential at the current density of 10 mA cm⁻² was observed at 0.35 V_RHE in 1 M KOH. The onset potential was observed at 1.43 V_RHE and the Tafel slope value was calculated at 69 mV dec⁻¹.

Recently, in one of our studies, we have synthesized cobalt oxide spinel (Co₃O₄) nanocrystals via a hydrothermal method. Our work provided deeper insight into the control parameters of hydrothermal WOC formation processes, through a combination of mechanistic in situ and ex situ analytical methods. This comprehensive study provided guidelines about how the WOC performance of binary oxides can be improved through an empirical understanding of their growth mechanisms [58].

2.2.3. Perovskite-Based WOCs

Perovskite oxides display the general formula ABO₃, where the A-site ion on the corners of the elemental cell is usually an alkaline earth or rare-earth element, with a coordination number of 12. On the other hand, the octahedrally coordinated B-site ions in the center of the elemental cell are mainly 3d, 4d, and 5d transition metals. Ever since pioneering works of the 1970s [59], numerous studies on the structure and properties of perovskites have been reported in the literature. More recently, perovskite oxides were proposed as efficient electrocatalysts for the oxygen evolution reaction [59,60,61,62,63,64,65]. The following studies contributed to the development of new perovskite representatives, which significantly improved their activity and stability as water oxidation catalysts. A comparison of the electrocatalytic performance of the materials of this category is summarized in

Table 4. Comparison of the electrocatalytic activity of perovskite oxide catalysts (cf. Table 1 for definitions of “durability” and “stability”).

Catalyst	Preparation Method	Onset Potential (V)	η (at 10 mV/cm2) (V)	Tafel Slope (mV dec−1)	Durability (h)	Stability	Electrolyte	Ref.
PrBa0.5Sr0.5Co1.5Fe0.5O5+δ	Electrospinning	-	0.37	52	12	High	0.1 M KOH	[61]
SrNb0.1Co0.7Fe0.2O3−δ nanorods	Electrospinning	1.52	0.37	61	30	Remarkable	1 M KOH	[60]
BaCo0.7Sn0.3O3−δ & BaCo0.7Fe0.1Sn0.2O3−δ	Ball milling-assisted solid-state reaction	1.53	0.45	68	2	Good	0.1 M KOH	[63]
La0.95FeO3−δ	Sol–gel	1.55	0.41	82	-	-	0.1 M KOH	[64]
La1−xSrxCoO3−δ	Reverse-phase hydrolysis	-	0.4	31	24	High	0.1 M KOH	[65]
BaNiO3 & BaNi0.83O2.5	Flux mediated crystal growth	-	-	-	20	Very good *	0.1 M KOH	[66]

and covers research of the last three years.

An electrospinning method was employed by Zhao et al. for the development of PrBa_0.5Sr_0.5Co_1.5Fe_0.5O_5+δ nanofibers [61]. The intrinsic activity and mass activity of this material were increased 4.7 and 20 times, respectively, compared to IrO₂, while the diameter of the particles was reduced to 20 nm. The aforementioned characteristics led to a high stability for 12 h, at a lower overpotential of 0.37 V_RHE, compared to IrO₂. The importance of the morphology and electronic properties of the structure resulted in an enhanced water oxidation, showing the optimization potential of these parameters. Furthermore, Zhu et al. used an electrospinning method to fabricate SrNb_0.1Co_0.7Fe_0.2O_3−δ perovskite nanorods, which displayed an excellent stability and activity in alkaline media [64]. A representative scheme of the synthesis protocol is shown in Figure 20. The results were compared with the-state-of the-art IrO₂ and showed seven times higher current densities at 0.4 V_RHE, combined with a lower Tafel slope. The material displayed an excellent stability for 30 h. The structural properties led to an increase of both charge transfer as well as of the electrochemically active surface, which boosted the water oxidation performance.

A ball-milling-assisted solid-state reaction was used by Xu and co-workers for the development of BaCo_0.7Sn_0.3O_3−δ and BaCo_0.7Fe_0.1Sn_0.2O_3−δ perovskite oxides, respectively [63]. They demonstrated that the electrocatalytic activity of the catalysts can be easily tuned by modifying the dopant concentration. This example highlights the importance of doping strategies to enhance the conductivity in a wide range of materials, which in turn, increases their electrocatalytic activity towards OER. In parallel, the work of Zhu et al. focused on a sol–gel preparation method and by tuning the cation deficiency enhanced OER activity was accomplished [60]. In particular, a La_0.95FeO_3−δ perovskite was developed; Figure 21 shows how the significant change in OER activity is related to the formation of oxygen vacancies and Fe⁴⁺, in the A-site of the perovskite.

Mefford et al. proposed a reverse-phase hydrolysis method for the synthesis of La_1−xSr_xCoO_3−δ [65]. Their mechanistic study provided further evidence of how oxygen vacancies become a crucial parameter for improving the electrocatalytic activity of metal oxides. Briefly, it was stated that the controlled substitution of Sr²⁺ for La³⁺, across the full-phase width, while maintaining the perovskite structure, allowed the effects of covalence, vacancy defects, and oxygen exchange to be probed. Lee et al. reported for the first time a hexagonal perovskite, BaNiO₃, as water oxidation catalyst in alkaline media and demonstrated that this new family of perovskites outperformed IrO₂, by at least 19 mA cm⁻² during the first cycle of OER [66]. The study of Black et al., first investigated quaternary hafnium oxynitrides with a reduced band gap and showed the potential of these compounds for photoelectrochemical water splitting devices [67]. The new RHfO₂N perovskites, with R = La, Nd, and Sm, display structures resembling the GdFeO₃-type. The LaHfO₂N and NdHfO₂N compounds, in particular, showed increased resistance towards photocorrosion, compared to the SmHfO₂N photocatalyst.

2.2.4. Mixed Oxides

Mixed oxides frequently excel through their increased stability and synergistic effects, in a wide range of properties and applications [55,56,68,69,70,71,72,73]. Table 5 summarizes progress in the performance of mixed oxide water oxidation catalysts, over the last three years.

Gholamrezaei and co-workers proposed a new SrMnO₃ nanostructure for chemical water oxidation [74]. Various preparation methods were chosen and compared by the authors, and sonochemical synthesis was found to afford the best results, with respect to morphology and nanoparticle size. Electrocatalytic water oxidation was carried out in a (NH₄)₂Ce(NO₃)₆ solution, where the effect of the Ce(IV) concentration was studied. O₂ evolution was increased with higher amounts of cerium. The results suggested that changes in the synthesis method and reaction conditions altered the morphology of the catalyst, as well as the size and the uniformity of the particles. The overall efficiency of the catalyst was enhanced along with the high uniformity of the particles.

Table 5. Comparison of the electrocatalytic activity of mixed oxide catalysts (cf. Table 1 for definitions of “durability” and “stability”).

Catalyst	Preparation Method	Onset Potential (V)	η (at 10 mV/cm2) (V)	Tafel Slope (mV dec−1)	Durability (h)	Stability	Electrolyte	Ref.
NiCeOx	Electrodeposition	-	0.27	-	>24	Excellent	1 M KOH	[68]
Sn-Fe2O3 NWs	Hydrothermal	0.1	-	-	10	Excellent	1 M KOH	[69]
Ni-FeOx & Ni-Fe-CoOx NPs	Impregnation in the presence of aniline	-	0.2	-	1	Good *	0.1 M KOH	[70]
CoNiOx/rGO	Precipitation	-	0.28	45	20	Advanced	0.1 M & 1 M KOH	[71]
FexCoyO	Precipitation	-	0.35	36.8	2.5	Good	0.1 M KOH	[72]
Fe3O4-Co9S8 NPs/rGO	Two solvothermal steps	1.48	0.34	82.8	6	High	0.1 M KOH	[75]
Na1−xNiyFe1−yO2	Hydrothermal	1.35	0.26	44	12	Excellent	1 M KOH	[76]
CoNiO, MnNiO & FeNiO	Precipitation reaction with urea	CoNiO: 1.60, MnNiO: 1.61 & FeNiO: 1.54	-	CoNiO: 39, MnNiO: 43 & FeNiO: 18	-	-	0.1 M KOH	[77]
Fe2−xCrxO3	Coprecipitation	<1.3	0.45	-	1.5	Good *	100 mM PBS	[78]
MnVOx@N-rGO	Hydrothermal	1.32	0.39	271	4	Excellent	0.1 M KOH	[79]
MnOx/N-CNT	Hydrothermal	ca. 1.4	0.36	75	2	Catalyst detachment	1 M KOH	[80]
MnxSb1−xOx	Physical Vapor Deposition (PVD)	1.5	0.3	75	25	High *	1 M H2SO4	[81]
Ho2O3/MnOx	Hydrothermal	Varied according to calcination	-	-	0.5	High	0.25 M LiClO4	[82]

Table 5. Comparison of the electrocatalytic activity of mixed oxide catalysts (cf. Table 1 for definitions of “durability” and “stability”).

Catalyst	Preparation Method	Onset Potential (V)	η (at 10 mV/cm2) (V)	Tafel Slope (mV dec−1)	Durability (h)	Stability	Electrolyte	Ref.
NiCeOx	Electrodeposition	-	0.27	-	>24	Excellent	1 M KOH	[68]
Sn-Fe2O3 NWs	Hydrothermal	0.1	-	-	10	Excellent	1 M KOH	[69]
Ni-FeOx & Ni-Fe-CoOx NPs	Impregnation in the presence of aniline	-	0.2	-	1	Good *	0.1 M KOH	[70]
CoNiOx/rGO	Precipitation	-	0.28	45	20	Advanced	0.1 M & 1 M KOH	[71]
FexCoyO	Precipitation	-	0.35	36.8	2.5	Good	0.1 M KOH	[72]
Fe3O4-Co9S8 NPs/rGO	Two solvothermal steps	1.48	0.34	82.8	6	High	0.1 M KOH	[75]
Na1−xNiyFe1−yO2	Hydrothermal	1.35	0.26	44	12	Excellent	1 M KOH	[76]
CoNiO, MnNiO & FeNiO	Precipitation reaction with urea	CoNiO: 1.60, MnNiO: 1.61 & FeNiO: 1.54	-	CoNiO: 39, MnNiO: 43 & FeNiO: 18	-	-	0.1 M KOH	[77]
Fe2−xCrxO3	Coprecipitation	<1.3	0.45	-	1.5	Good *	100 mM PBS	[78]
MnVOx@N-rGO	Hydrothermal	1.32	0.39	271	4	Excellent	0.1 M KOH	[79]
MnOx/N-CNT	Hydrothermal	ca. 1.4	0.36	75	2	Catalyst detachment	1 M KOH	[80]
MnxSb1−xOx	Physical Vapor Deposition (PVD)	1.5	0.3	75	25	High *	1 M H2SO4	[81]
Ho2O3/MnOx	Hydrothermal	Varied according to calcination	-	-	0.5	High	0.25 M LiClO4	[82]

The study of Desmond-Ng et al. was focused on a theoretical model and experimental considerations aiming for the engineering of catalytic systems that are competitive with the best water oxidation electrocatalysts to date [68]. To this end, the authors synthesized a NiCeO_x-Au catalyst, with an outstanding OER activity in alkaline media. A highlight of this approach was the new combination of electronic, geometric, and catalyst support effects. The TOF of NiCeO_x-Au catalyst at 0.28 V_RHE overpotential was determined as 0.08 s⁻¹.

Li et al. proposed Sn as a dopant in hematite nanowire photoanodes [69]. The morphology was considered the key factor leading to the high performance of the Sn-doped hematite, which was furthermore coated with a cobalt oxide catalyst for enhanced photocurrent densities. A schematic illustration of the catalyst preparation is shown in Figure 22. In this paper, the importance of retaining the concentration and uniformity of dopants along the nanowire growth was highlighted as another crucial parameter that is essential for high catalyst activity. The authors applied a silica encapsulation method, which retained the morphology of the hematite nanowire and further provided tuned nanowire lengths for maximum light absorption.

Bates et al. proposed a different way of increasing the electrocatalytic efficiency of a Ni–Fe–Co mixed metal oxide (MMO) catalyst. The morphology was optimized by addition of aniline as a capping agent, in order to generate a high surface area [70]. Charge-transfer effects contributed to a higher conductivity among the Ni–Co sites and, therefore, increased their catalytic activity in water oxidation.

A promising strategy towards the development of OER electrocatalysts was proposed by Li et al., where a CoNiO_x nanocomposite was deposited onto a reduced graphene oxide [71]. The exceptional performance of this nanocomposite in water oxidation was ascribed to the hierarchical sheet-on-sheet structure, affording a high surface area and high porosity. The catalyst showed a low overpotential at 0.28 V_RHE and a Tafel slope of 45 mV dec⁻¹. Proposing that all metal sites are active during the electrochemical reaction, the TOF value was estimated at 0.03 s⁻¹, at the overpotential of 0.35 V_RHE.

Zhuang et al. fabricated a high surface area catalyst consisting of Fe_xCo_y-O nanosheets (261.1 m² g⁻¹), with an ultrathin thickness of 1.2 nm [72]. These features led to a superior activity, compared to commercial RuO₂, with an overpotential at 0.35 V_RHE and a Tafel slope at 36.8 mV dec⁻¹. The TOF value at the overpotential of 0.35 V_RHE was calculated at 0.02 s⁻¹. This outstanding performance of the catalyst was ascribed to the large number of oxygen vacancies in the sheets, which further acted as active centers. The proposed OER mechanism of a Co³⁺ active site, near an oxygen vacancy, is illustrated in Figure 23.

Applying a two-step solvothermal method, Yang et al. fabricated Fe₃O₄-decorated Co₉S₈ nanoparticles on reduced graphene oxide [75]. The composite catalyst gave an excellent stability and catalytic activity. This was ascribed to electron transfer from the Fe species to Co₉S₈, which promoted the cleavage of the Co–O bond in the stable configuration of the Co–O–O superoxo group.

A layered Na_1−xNi_yFe_1−yO₂ double oxide was reported by Weng and co-workers with excellent stability and electrocatalytic activity, compared to the state-of-the-art RuO₂ and IrO_{2 [76]}. As a result of Na extraction, the valence states of Ni and Fe were raised and the catalyst exhibited excellent OER activity. This catalyst was also integrated in a perovskite solar cell, which delivered a solar-to-hydrogen efficiency of around 11.2%.

A new catalytic material comprising Fe_2−xCr_xO₃ particles is under investigation by Kanazawa et al. and is proposed for both photoelectrochemical and electrochemical water oxidation [78]. Findings of their study revealed that Cr-substitution resulted in a five-fold enhancement of the electrochemical water oxidation performance, compared to α-Fe₂O₃, as well as lowered charge transfer resistance, giving rise to an improved OER activity, with nearly a 100% faradaic efficiency.

Mn oxides are cost-effective, environment-friendly, and redox active. The {CaMn₄O₅} oxygen evolving center of photosystem II is a constant source for bio-inspired approaches, in many fields of catalysis. Due to their versatility, Mn oxides have even been referred to as “Swiss army knife” [83]. In the following, selected recent studies on mixed oxide catalysts, based on manganese are summarized.

Xing et al. proposed a new class of composite catalysts comprising MnVO_x on N-doped reduced graphene oxide [79]. A hydrothermal method was applied, starting from a [Mn₄V₄O₁₇(OAc)₃]³⁻ polyoxometalate precursor. The authors showed that the precursor was successfully deposited on carbon and then converted to WOCs, with recorded OER current densities of 80 mA cm⁻², at the applied potential of 1.67 V_RHE. The overpotential of the MnVO_x/N-rGO electrocatalyst was determined as 0.39 V_RHE, at 10 mA cm⁻² current density, with a Tafel slope of 271 mV dec⁻¹. The synthesis method provided nanoparticulate Mn–V oxides with excellent stability, which can be further used for a variety of technological applications.

Antoni and co-workers employed high surface area N-functionalized CNTs for the oxidative deposition of MnO_x under reflux and mild hydrothermal conditions [80]. The synthesis was carried out in the presence of CsMnO₄ and KMnO₄ as manganese sources, which gave mainly rise to the formation of crystalline birnessite manganese oxides. The MnO_x-modified N-CNTs catalysts afforded higher current densities, at around 20 mA cm⁻², compared to the unmodified N-CNTs in alkaline electrolyte. The overpotential of the catalyst was recorded as 0.36 V_RHE at 10 mA cm⁻² current density and the reported Tafel slope was 75 mV dec⁻¹. The faradaic efficiency was also increased from 75% to 90% towards the OER, after 10 min of electrocatalytic tests, and this was attributed to the oxidation of manganese.

A physical vapor deposition approach via reactive radio frequency magnetron co-sputtering of Mn and Sb was employed by Zhou et al., for the synthesis of rutile-type solid solutions, Mn_xSb_1−xO_x, with OER activity in strong acidic conditions [81]. The mixed oxides showed current densities above 50 mA cm⁻², at the applied potential of 1.79 V_RHE. The overpotential at the current density of 10 mA cm⁻² was found to be 0.58 V_RHE, with a Tafel slope value of 75 mV dec⁻¹. Notably, the fraction of Mn³⁺ increased with the overall amount of Mn in the mixed oxides. This was corroborated by XAS, AP-XPS, and computational investigations and linked to the overall increase in OER activity. Likewise, Morgan Chan et al. showed via spectroscopic and computational studies that the enhancement of MnO₂ oxygen evolution catalyst is related to the introduction of Mn³⁺ species, and the suppression of Mn³⁺ oxidation to Mn⁴⁺, due to structural restrictions, was considered crucial for higher activity [84].

Najafpour and co-workers proposed and hydrothermally fabricated a new manganese oxide supported on holmium oxide for OER [82]. Their catalyst was very stable, recoverable, and it could be used for long operational OER. The observed self-healing properties were explained with the presence of Ce(IV) species. O₂ was produced from water in the presence of Mn oxides and Ce(IV) and water, and MnO₄⁻ was further formed from the oxidation of intermediate Mn(II) ions by Ce(IV). These MnO₄⁻ ions were eventually converted back to Mn oxide, resulting in self repair of the catalyst.

2.2.5. Other Hydroxides and (Oxy)hydroxides

Commercial water electrolyzers require efficient electrocatalysts with high current densities, usually above 500 mA cm⁻², with long-term stability, at overpotentials below 0.3 V_RHE. To this end, metal hydroxides and (oxy)hydroxides have attracted a lot of research attention as target structures, to fabricate catalysts with the aforementioned characteristics [18,36,39,40,41,85,86,87,88,89,90,91,92]. Table 6 summarizes progress in the electrocatalytic activity of hydroxide and (oxy)hydroxide catalysts, over the last three years. In addition to noble metal oxides, such as RuO₂ and IrO₂, which have been established as the best OER catalysts to date, Fe-doped NiOOH ranks among the best, noble metal-free catalysts.

Recently, Shin et al. investigated new dopants using a density functional theory approach, in order to determine the atomistic mechanism for the OER of Ni_1−xFe_xOOH [93]. In their study, 17 transition metals were considered to replace Fe. They found out that the most promising materials were Co, Rh, and Ir, and the resulting overpotentials were estimated to be 0.27, 0.15, and 0.02 V_RHE, respectively. The authors concluded that Fe⁴⁺ and Ni⁴⁺ species played an essential role in the OER. The promotion of a radical environment on the metal-oxo bond was also found to be a key prerequisite for high OER activity.

Table 6. Comparison of the electrocatalytic activity of hydroxide and (oxy)hydroxide catalysts (cf. Table 1 for definitions of “durability” and “stability”).

Catalyst	Preparation Method	Onset Potential (V)	η (at 10 mV/cm2) (V)	Tafel Slope (mV dec−1)	Durability (h)	Stability	Electrolyte	Ref.
NiFeOH/NiFeP	Hydrothermal & PH3 plasma treatment & electrodeposition	-	0.258	39	24	High *	1 M KOH	[85]
NiFe-OH	Stepwise electrodeposition	-	0.24	38.9	5	Good *	1 M KOH	[86]
W0.5Co0.4Fe0.1 oxyhydroxide coralloids	Hydrothermal	1.53	0.25	32	>500	Superior	1 M KOH	[87]
CoFe-H hydroxide & CoFe-H/BiVO4	Electrodeposition	0.23	0.28	28	45	High *	1 M KOH	[88]
CeO2/FeOOH	Electrodeposition	0.21	0.25	32	50	High	1 M NaOH	[89]
FeOOH/Co/FeOOH	Electrodeposition	0.22	0.25	32	50	High	1 M NaOH	[90]
FeCoW oxyhydroxides	Modified sol-gel	-	0.191	37	550	High	1 M KOH	[91]
NiFe(OH)/NiFe:Pi	Electrodeposition & phosphorylation	1.43	0.29	38	10	Excellent	1 M KOH	[92]
Ag-decorated Co(OH)2	Selective reduction-oxidation	-	0.27	67–109 (depending on the amount of Ag)	10	Good	1 M KOH	[94]
NiFeOOH/TiO2 core-shell	Hydrothermal		0.273	86.9	24	High *	0.01 M Na2SO4	[95]
α-NiCo(OH)2	Electrodeposition	1.43	0.26	25-30	6	High *	1 M KOH	[96]
NiPS3@NiOOH	Chemical vapor transport & liquid exfoliation	1.48	0.35	80	>160	Excellent	0.1 M KOH	[97]

Table 6. Comparison of the electrocatalytic activity of hydroxide and (oxy)hydroxide catalysts (cf. Table 1 for definitions of “durability” and “stability”).

Catalyst	Preparation Method	Onset Potential (V)	η (at 10 mV/cm2) (V)	Tafel Slope (mV dec−1)	Durability (h)	Stability	Electrolyte	Ref.
NiFeOH/NiFeP	Hydrothermal & PH3 plasma treatment & electrodeposition	-	0.258	39	24	High *	1 M KOH	[85]
NiFe-OH	Stepwise electrodeposition	-	0.24	38.9	5	Good *	1 M KOH	[86]
W0.5Co0.4Fe0.1 oxyhydroxide coralloids	Hydrothermal	1.53	0.25	32	>500	Superior	1 M KOH	[87]
CoFe-H hydroxide & CoFe-H/BiVO4	Electrodeposition	0.23	0.28	28	45	High *	1 M KOH	[88]
CeO2/FeOOH	Electrodeposition	0.21	0.25	32	50	High	1 M NaOH	[89]
FeOOH/Co/FeOOH	Electrodeposition	0.22	0.25	32	50	High	1 M NaOH	[90]
FeCoW oxyhydroxides	Modified sol-gel	-	0.191	37	550	High	1 M KOH	[91]
NiFe(OH)/NiFe:Pi	Electrodeposition & phosphorylation	1.43	0.29	38	10	Excellent	1 M KOH	[92]
Ag-decorated Co(OH)2	Selective reduction-oxidation	-	0.27	67–109 (depending on the amount of Ag)	10	Good	1 M KOH	[94]
NiFeOOH/TiO2 core-shell	Hydrothermal		0.273	86.9	24	High *	0.01 M Na2SO4	[95]
α-NiCo(OH)2	Electrodeposition	1.43	0.26	25-30	6	High *	1 M KOH	[96]
NiPS3@NiOOH	Chemical vapor transport & liquid exfoliation	1.48	0.35	80	>160	Excellent	0.1 M KOH	[97]

Liang et al. used a PH₃ plasma-assisted method developed by their group, to fully convert a NiFe hydroxide into porous NiFeP, as shown in Figure 24 [85]. This method facilitated the rapid synthesis of phosphides, at around 15 min, at low temperatures of approximately 200 °C. The catalyst showed high geometric current densities of 300 mA cm⁻² at the corresponding overpotential of 0.2 V_RHE, a low overpotential of 0.258 V_RHE at 10 mA cm⁻², and a small Tafel slope of 39 mV dec⁻¹. The low overpotential of 0.25 V_RHE corresponded to 0.036 s⁻¹ TOF, as calculated by the authors. The high activity was ascribed to the morphological architecture and the electronic interaction between the α-NiFe-OH and NiFeP, respectively.

Electrodeposition was employed by many groups, in order to synthesize catalysts using FeOH or FeOOH-coupled with other co-catalyst metals or metal oxides, such as Ni, CeO₂, and BiVO₄ [86,87,89,90,92]. Of note is the work by Li et al., which dealt with the surface wettability [92]; the authors employed electrodeposition and phosphorylation methods to deposit an α-NiFe-OH nanosheet layer onto a NiFe:P_i 3D hierarchical nanostructure. From the results, the authors concluded that surface wettability is a key parameter for controlling the morphology, while it is also capable of modifying the chemical properties of a catalyst. The catalyst showed a low overpotential of 0.29 V_RHE at 10 mA cm⁻² and an excellent stability in the aqueous media. The faradaic efficiency was found to be 98.2%, which suggested that the main product evolved on the catalyst was O₂. Furthermore, the phosphorylation assisted the enhancement of the surface area, which provided more catalytic active sites for water adsorption and oxidation.

A trimetallic W_0.5Co_0.4Fe_0.1 oxyhydroxide corraloid was introduced by Pi et al.; this exhibited a superior stability and durability in alkaline media, for 21 days [87]. Their catalyst outperformed most of the reported Co-based nanomaterials and showed continuous electrolysis currents with imperceptible decay after more than 500 h. The authors followed a wet-chemical method to synthesize corraloid-like trimetallic oxyhydroxides on nickel foam and carbon nanotubes. The preparative route and the electron microscopy characterization of the catalyst are displayed in Figure 25. The images showed disintegrated and twisted nanosheets and the selected area electron diffraction data revealed the high crystallinity of the oxyhydroxides. The special structure of this catalyst went hand-in-hand with a low overpotential of 0.25 V_RHE, at the current density of 10 mA cm⁻² and a small Tafel slope of 32 mV dec⁻¹. XPS data showed that an optimized amount of Fe went hand in hand with the optimization of the binding energy of oxygen intermediates, which led to a high OER activity.

A similar trimetallic catalyst consisting of FeCoW oxyhydroxides was synthesized via a modified sol–gel method by Zhang and co-workers [91]. These gelled FeCoW oxyhydroxides exhibited superior durability for 550 h, i.e., approximately 23 days, with a very low overpotential of 0.191 V_RHE at 10 mA cm⁻², in alkaline media. The intrinsic activity of gelled-FeCoW oxyhydroxides was quantified by determining the TOF value, which was found to be 0.46 s⁻¹, at an overpotential of 0.3 V_RHE. The faradaic efficiency of O₂ evolution recorded on the catalyst was calculated to be close to 100%. The synergistic effects between the three metals produced a favorable local coordination environment, which, along with their electronic interaction, significantly enhanced the OER activity.

Electrodeposited amorphous α-phase Ni–Co(OH)₂ nanodendrite forests were synthesized on a stainless steel foil by Balram and co-workers [96]. The key parameter for the formation of nanodendritic structures was the addition of small amounts of water to the deposition bath. A limited water electrolysis and hydrogen bubble production occurred near the electrode interface with the electrolyte, which caused highly localized pH gradients, resulting in the aforementioned three-dimensional structure. The electrocatalyst afforded current densities higher than 50 mA cm⁻², at the applied potential of 1.6 V_RHE. The overpotential at the current density of 10 mA cm⁻² was found at 0.25 V_RHE and the Ni–Co(OH)₂ material showed an excellent stability during the operational time of 6 h.

Konkena and co-workers employed a chemical vapor transport and liquid exfoliation method, in order to fabricate an NiPS₃@NiOOH core-shell structure [97]. Ni(III)-based (NiOOH) species of the shell structure enhanced the overall OER, as shown by a combination of spectroscopic and computational analyses. Additionally, the high OER activity was attributed to the highly active metal-edge sites. The catalyst excelled through a low-onset potential at 1.48 V_RHE, as well as an overpotential value of 0.35 V_RHE at the current density of 10 mA cm⁻². The Tafel slope was recorded as 80 mV dec⁻¹ and the catalyst showed a high stability after 6 h of electrocatalytic tests.

3. Brief Discussion of the Unresolved Bottlenecks

The development of a highly efficient and low-cost water oxidation catalysts remains of utmost importance for commercial artificial photosynthesis applications. Notwithstanding the recent advances in this field, fundamental improvement strategies and issues need to be further addressed. Focusing efforts on mechanistic studies is essential for a deeper understanding of how a more efficient and long-lasting water oxidation catalyst can be engineered. Furthermore, these advanced engineering efforts for a promising oxidation catalyst need to be coupled with convenient and applicable benchmarking protocols. These protocols will allow us to easily compare our systems on all levels and to unambiguously identify candidates for a technological scale-up. Given that such generally accepted standard test procedures are still pending, here, we refrained from further performance comparisons between the different material types. The wide variation of preparative and experimental parameters between the selected studies still renders such comparisons far from unambiguous. Another unresolved bottleneck in the field is the need for widely available in situ/operando monitoring techniques, under real catalytic conditions. Monitoring structural changes during catalysis, under various concentrations, pH, and temperature parameters, provides indispensable first-hand information for selecting the most efficient optimization parameters.

4. Directions to New Composite Materials

In this section, an outlook is given on new composite materials for OER catalysis. Kuang et al. introduced a superior carbon material as a booster additive for water oxidation catalysts, namely graphdiyne (GDY), which comprises assembled layers of sp and sp² hybridized carbon atoms [98]. Graphdiyne has been proposed as a new and the most stable carbon allotrope, with a high π-connectivity, uniformly distributed triangular pores, and tunable electronic properties [99]. The geometrical structure of this material is illustrated in Figure 26. More specifically, Kuang and co-workers fabricated a GDY/NiFe-LDH electrocatalysts, via a hydrothermal method. Morphological characterizations of the material, as well as elemental mapping data are displayed in Figure 27. The electrocatalyst exhibited outstanding catalytic activity towards water oxidation, at a low overpotential of 260 mV in an alkaline electrolyte. Density functional theory calculations were also performed and demonstrated strong interactions between –C≡C–C≡C– and the transition metal atoms, which further contributed to the fast electron transfer and long operational stability.

Chen and co-workers synthesized an emerging type of carbon, the giant carbon tubes (GCTs), with outstanding water oxidation properties, and a unique physicochemical structure [100]. Regarding the formation protocol, quartz wool was employed as a template for the CVD growth of the giant carbon tubes. The material was placed in a quartz tube and heated to different temperatures—between 950 and 1050 °C—in a mixed H₂ and CH₄ gas stream.

The nitridation of the GNTs (N-GNTs) was also carried out by further heating of the sample in a furnace, under an NH₃ flow. SEM and TEM images of the material are shown in Figure 28. The GNTs and N-GNTs showed an overpotential of 0.44 V_RHE, at the current density of 10 mA cm⁻² and Tafel slope values of 46 and 48 mV dec⁻¹, respectively, which were significantly lower than those of other carbon-based WOCs. The authors also ran chronoamperometric studies of the catalyst for almost 30 min, and the current densities showed a time-dependent upward trend.

Heterojunction structures are of current interest for solar water splitting, as proposed recently by Wei et al. [101]. In particular, the authors employed a hydrothermal method to synthesize SrTiO₃/TiO₂ heterogeneous hollow multi-shelled structures (HoMSs), via a hydrothermal crystallization of SrTiO₃ on the surface of the TiO₂. The preparation route towards the SrTiO₃/TiO₂ heterojunctions is illustrated in Figure 29. Morphological characterizations (cf. Figure 30) showed the full coverage of the TiO₂ surface with SrTiO₃. The heterogeneous hollow multi-shelled structures increased the ability of the SrTiO₃/TiO₂ photocatalyst to absorb light and subsequently augmented their charge carrier separation efficiency. The SrTiO₃/TiO₂ photocatalyst retained its catalytic activity, without any evident decrease, after 20 h of overall water splitting reaction.

5. Conclusions and Perspectives

Water splitting as means of a clean and sustainable hydrogen energy production is of utmost importance, in order to meet our worldwide present and future energy needs. Electrochemical processes involving the oxygen evolution reaction are an essential part of this technology. As a result, the development of efficient electrocatalysts has become critical and, as a consequence, this research field has attracted increasing attention over the past years. To this end, an electrocatalyst for commercial electrolyzers should generally reach current densities of at least 500 mA cm⁻², with long-term stability, as well as overpotentials below 0.3 V_RHE.

It is important to point out that different lines of research emerge from electrolyzer optimization for photovoltaic-driven electrolysis devices (PV-PEC) and from photoelectrochemical systems (PEC). Concerning PV-PEC, increased production levels in the electrolyzer industry require comparable cost levels to the PV sector, to couple these technologies both economically and operationally. Therefore, the optimization of porous transport layers in electrolyzers is an essential task, with regard to costs, stability, and efficiency. Furthermore, improvements of the performance and stability of catalyst layers and membranes are indispensable. On the mass production scale, the use of earth-abundant electrocatalysts with similar characteristics to that of noble-metal electrocatalysts, will become inevitable. In addition, membrane-free systems are highly desirable, due to their low upfront capital costs. Another low-cost strategy will be the reduction of the ionic resistance of the membrane to significantly enhance the electrocatalyst activity and to minimize the mass transport limitations. In addition to the need for earth-abundant and low cost materials, the interfacial interactions between the absorbers, electrocatalysts, and electrolytes need to be further and more deeply understood.

While the technology readiness for the PV-PEC devices is at an advanced level, the technological maturity of PEC devices is at a rather early stage, with an estimated time horizon for applications of approximately 10 years. Consequently, progress in PEC technology is of high scientific significance to (and complements) PV-driven electrolysis devices with flexible setups.

Keeping the present situation in mind, this review presents an overview on the recent advances in water oxidation catalysts, focusing on highly visible studies in the field, between 2016 and 2018, to set research directions for the future. The selected works identify a wide range of influential parameters, such as morphology control, choice of operational media, catalyst stability and mechanisms, as well as synthesis methods, including doping strategies, which are just beginning to be fully explored for performance optimization. The complexity of these parameter interactions, furthermore, increases the need for fundamental mechanistic understanding, going hand in hand with the development of new in situ/operando monitoring techniques. In addition, computational studies provide valuable guidelines for the informed use of such new analytical and synthetic techniques. The quest for true catalysts is more important than ever, given the increasing number of yet unexplored new oxide materials, let alone the manifold emerging oxide clusters. The need for new and more efficient materials is pressing to both enhance the performance and commercialization of electrolyzers and to leverage the full potential of complementary and direct, off-grid photoelectrochemical devices.