Clean production of chlorine (Cl₂) and hypochlorous acid (HOCl) from photocatalytic and photoelectrochemical seawater splitting

Abstract

Seawater splitting has emerged as a promising alternative to overall water splitting because it eliminates the kinetically sluggish oxygen evolution reaction (OER), which is a bottleneck in water splitting, and avoids the low economic value of O₂. Moreover, in seawater splitting, H₂ evolution coupled with the oxidation of chloride (Cl⁻) to value-added chlorine (Cl₂) and/or hypochlorous acid (HOCl) can simultaneously benefit the energy and environmental sectors. Cl₂ and HOCl are widely used for bleaching, disinfection, sanitisation and sterilisation in the medical sector and for purifying drinking water and water in swimming pools owing to their strong oxidising and antibacterial properties. Mainstream industrial production employs the chlor-alkali electrolysis of sodium chloride (NaCl), which requires significant energy input and releases enormous amounts of CO₂. To achieve the sustainable production of Cl₂ and HOCl while reducing energy consumption and environmental impacts, photocatalytic (PC) and photoelectrochemical (PEC) technologies have been employed as green alternatives. Importantly, PC and PEC enable the on-site production of Cl₂/HOCl in remote areas, which can circumvent their instability (decomposition), storage and transport issues. This article reviews the recent progress in the PC and PEC production of Cl₂/HOCl, along with the catalytic materials used and their designs and photocatalytic performance. The applications of in situ HOCl production in anti-bacterial treatment, ammonia removal, the selective oxidation and conversion of organic compounds, and CO₂ conversion are discussed. We also address the challenges in this area and highlight prospects for future research directions. Overall, we demonstrate that the PC and PEC production of Cl₂/HOCl serves as a green and sustainable alternative to the chlor-alkali process. This research area is still in its infancy, and we hope that this review article will garner the attention of researchers to contribute to this area, leading to a step closer toward practical applications.

---

Wider impact

The chlor-alkali process has been the predominant industrial production process for chlorine gas (Cl₂) via the electrolysis of seawater since 1892. Despite its industrial success, the large energy input, high capital costs, complex separation and purification processes, and toxic cell components (mercury in mercury cells and asbestos in diaphragm cells) have led to serious energy and environmental concerns. The most straightforward path to clean and sustainable production of Cl₂/HOCl is by employing a solar-to-chemical conversion mechanism. Progress in photocatalytic (PC) and photoelectrochemical (PEC) technologies has achieved maturity, yet their application in Cl₂/HOCl production is still in its infancy. PC and PEC technologies harvesting solar energy and using seawater in the presence of suitable photocatalysts are expected to reduce energy input and environmental impacts. Moreover, these technologies enable on-site production of Cl₂/HOCl in remote areas using portable PC and PEC cells, which can overcome the instability, transport, and storage issues of Cl₂/HOCl. This review discusses the recent progress in the PC and PEC production of Cl₂/HOCl from seawater splitting in the past decade. Importantly, the oxidation of chloride (Cl⁻) at the anode (or anodic site) can be coupled with hydrogen production or CO₂ reduction at the cathode (or cathodic site) for the simultaneous production of fuels and useful chemicals as a type of artificial photosynthesis.

1. Introduction

The two most pressing long-standing global challenges are the energy crisis due to the depletion of fossil fuels and environmental pollution resulting from human activities, industrial chemical processes and power plants. Therefore, there is an urgency to implement renewable energy strategies, including wind energy, photovoltaic solar cells and H₂ production, to mitigate these issues. H₂ production via photocatalytic water splitting has been dominating the photocatalysis research for the last 50 years since the groundbreaking work by Honda and Fujishima.^1,2 Despite extensive research efforts, photocatalytic water splitting is far from practical realisation due to the challenging issues like the reverse reaction of H₂ and O₂, slow kinetics of O₂ evolution reaction (OER), and low solar-to-hydrogen (STH) efficiency (<10%).² A recently proposed strategy is to couple the H₂ evolution reaction (HER) with more kinetically favourable oxidation reactions that produce high value-added compounds to circumvent the issues of the slow kinetics of the OER and the low economic value of O₂.^3–5 Sayama has pioneered the photoelectrochemical oxidation of several high value-added chemicals, including H₂O₂, HOCl, S₂O₈²⁻ and Ce⁴⁺, paving the way towards the effective utilization of solar light energy and oxidative power of photogenerated holes.⁶ Chloride (Cl⁻) oxidation to Cl₂ and/or HOCl proceeds smoothly in a seawater solution, allowing the use of abundant and low-cost seawater/saltwater. To the best of our knowledge, photocatalytic seawater splitting has not been extensively discussed in the literature, and therefore, it is timely to present a comprehensive review on photocatalytic (PC) and photoelectrochemical (PEC) seawater splitting.

Hypochlorous acid (HOCl) and sodium hypochlorite (NaOCl) are widely used in a plethora of applications in industrial processes and commercial applications, including pulp and paper bleaching, sanitization, and wound care.⁷ For use in bleaching and disinfection, 40% of the application goes into drinking water and wastewater treatments, and the remaining 60% is used in household applications, including pool sanitization, laundry bleaching and surface disinfection, in food processing and healthcare facilities. The most common type of drinking water disinfection is chlorination, which is highly effective in killing bacteria and other microorganisms to ensure that the water is safe for drinking. The World Health Organization (WHO) has set the recommended Cl₂ concentration to 2 mg L⁻¹ for effective household water treatment and the residual free Cl₂ to 0.2–0.5 mg L⁻¹.⁸ For convenience, Cl₂, HOCl and ClO⁻ are classified as reactive chlorine species (RCS) and are used throughout this article although sometimes they are termed HOCl in other articles.^9,10 RCS are produced commercially on a large-scale using the chlor-alkali electrochemical process in a brine (NaCl) solution. However, due to the uphill thermodynamic process and formation of unwanted species, the electricity consumption in the chlor-alkaline process is huge, which leads to enormous CO₂ emissions, thus resulting in a significant reduction in energy consumption and carbon footprint using green and sustainable alternatives.¹¹ In light of energy-efficient processes and reduced environmental impacts, photocatalytic (PC) and photoelectrochemical (PEC) technologies pave the way towards a green and sustainable strategy for the production of RCS.

In water splitting, O₂ is produced along with H₂ (eqn (1)). However, the oxygen evolution reaction (OER) has become a kinetic bottleneck due to its large overpotential. Moreover, from a practical point of view, O₂ has a negligible economic value. In saltwater/seawater, Cl⁻ ions are present in the majority, and the Cl⁻ oxidation reaction (COR) via a two-electron process is more kinetically favourable than the OER due to the lower overpotential. The use of seawater can dramatically reduce the operating costs of PC or PEC water splitting, where pure water is very scarce and expensive. In the chlor-alkali process, Cl⁻ is oxidized at the anode to form Cl₂ (eqn (3)), and dissolved Cl₂ in water forms hypochlorous acid (HOCl) via a disproportion reaction (eqn (4)) or is mixed with sodium hydroxide (NaOH) to form NaOCl. The distribution of Cl₂, HOCl and OCl⁻ in aqueous solution depends on the pH of the solution. Under acidic, neutral and basic conditions, Cl⁻ is oxidized to Cl₂, HOCl and OCl⁻, respectively (Fig. 1a).¹² HOCl exists in equilibrium with molecular Cl₂ (eqn (6)). If Cl₂ is allowed to escape, the equilibrium shifts to the left, reducing the amount of HOCl. Therefore, the selectivity of RCS for Cl⁻ oxidation depends on the complexity of the pH-dependent equilibria of the solution. Fig. 1b shows the Pourbaix diagram of the chlorinated species across all pH values. Nonetheless, it is difficult to maintain the stability of HOCl because of its rapid decomposition even in the dark. The decomposition of HOCl follows two disproportionate pathways, producing either chlorate and/or oxygen (refer to Section 2 for details).¹³ Such rapid decomposition makes on-site production of HOCl via PC or PEC even more compelling to avoid the need for long-distance transport and long-term storage.

2H⁺ + 2e⁻ → H₂ E = 0 V vs. RHE (1)

O₂ + 4H⁺ + 4e⁻ → 2H₂O E = −1.23 V vs. RHE (2)

2Cl⁻(aq) → Cl₂(g) + 2e⁻ E = 1.36 V vs. RHE (3)

Cl⁻ + H₂O → HOCl + H⁺ + 2e⁻ E = 1.49 vs. RHE (4)

Cl⁻ + H₂O → OCl⁻ + 2H⁺ + 2e⁻ E = 1.72 V vs. RHE (5)

Cl₂ + H₂O ⇌ H⁺ + Cl⁻ + HOCl (6)

HOCl ⇌ H⁺ + OCl⁻ (7)

Fig. 1 (a) Distribution of reactive chlorine species (RCS) of Cl₂, HOCl, and ClO⁻ across a wide pH range from acidic to neutral and basic. Reproduced from ref. 14. Copyright 2020, Elsevier. (b) Pourbaix diagram of the reactive chlorine species at all pH values. Reproduced from ref. 15. Copyright 2025, Springer.

A photocatalytic process requires a suitable photocatalyst, a Cl-saturated solution and light to produce Cl₂ and/or HOCl. The sources of Cl⁻ include a brine solution, aqueous HCl, and seawater/saltwater. Seawater is an ideal and practical reaction solution due to its abundance, cost-effectiveness, and unlimited availability compared to pure water, which incurs an additional cost for its high-purity resource. Most ions in seawater are Cl⁻ (55%), followed by sulfate (SO₄²⁻, 7.7%) and carbonate (CO₃²⁻, 0.33%). For this reason, a seawater solution for photocatalytic HOCl or Cl₂ production is often used at 0.5–0.6 M (or 3.5 wt%) aqueous NaCl to mimic natural seawater.¹⁶ Therefore, the photocatalytic Cl⁻ oxidation reaction (COR) brings about four merits: (i) the use of renewable solar energy, (ii) a low-cost and low carbon footprint process, (iii) on-demand production for remote areas, and (iv) no requirement for facilities for storage, handling and transport of RCS. Notably, the simultaneous PC production of H₂ and HOCl avoids the use of an expensive membrane due to self-separation by phase (gaseous H₂ vs. liquid HOCl). Additionally, a seawater solution can be an ideal, low-cost electrolyte for operating PEC cells.

For Cl⁻ to be oxidized to Cl₂ by a photogenerated hole, the valence band (VB) must be more positive than 1.36 V (vs. RHE). This redox potential is not significantly different from that of water oxidation to O₂ potential (1.23 V vs. RHE); therefore, Cl⁻ oxidation usually competes with water oxidation. In seawater or seawater solution, Cl⁻ oxidation is preferred to water oxidation due to a large overpotential (∼0.25 V) and thus sluggish kinetics of the latter.¹⁷ As an example, Johnston's group recorded a current density in HClO₄ and NaCl/HClO₄ electrolytes and observed a twice higher current density for Cl⁻ oxidation compared to that of water oxidation between 0.5 and 2.0 V (Fig. 2a).¹⁸ Moreover, it has been shown that Cl⁻ can quench a photogenerated hole, giving an advantage of selective HOCl formation over water oxidation.^19,20 In some work, it has been shown that O₂ evolution in saltwater/seawater originates from the decomposition of HOCl, providing a kinetically easier route for O₂ evolution than direct 4-electron water oxidation.⁹ Notably, Huang et al. unravelled that adding a small amount of Cl⁻ in water or pre-adsorbing Cl⁻ on TiO₂ lowered the activation barrier, thereby boosting the H₂ production rate by 3-fold and achieving a stoichiometric ratio of H₂ and O₂ evolution in water splitting (Fig. 2b and c).¹⁷ Similarly, enhanced H₂ production using a p-GaN-based nanowire in seawater (cf. pure water) is attributed to Cl⁻ oxidation and better conductivity of seawater.²¹ These works also suggest that seawater splitting is a feasible strategy for improving water splitting for a high production rate of H₂. Although this article focuses exclusively on the production of Cl₂/HOCl via PC and PEC seawater splitting, readers interested in photocatalytic H₂ production from seawater splitting are encouraged to refer to related review articles elsewhere.^22–24

Fig. 2 (a) Photocurrent density of water (black) and Cl⁻ (red) oxidation. Reproduced from ref. 18. Copyright 2023, the Royal Society of Chemistry. (b) Mechanism of seawater splitting into H₂ and O₂ through the formation of an HOCl intermediate in the presence of Cl⁻. (c) Free energy diagram of seawater splitting. Reproduced from ref. 17. Copyright 2014, the Royal Society of Chemistry.

With growing attention in this research area over the past decade, it is timely to present a comprehensive and critical discussion on the development of the effective production of Cl₂/HOCl. This review article aims to discuss the recent progress in PC and PEC of Cl₂/HOCl production and review the photocatalytic materials and designs, as well as their photocatalytic performance. First, we present an overview of the challenges in the effective production of Cl₂ and HOCl. Next, we review the photocatalytic materials designed for this purpose, including metal oxides, metal chloride and oxychlorides. In addition, we highlight the applications of photocatalytic HOCl production in other related areas, including anti-bacterial treatment, degradation of ammonia–nitrogen, selective oxidation and conversion of organic compounds, and photocatalytic CO₂ reduction. Given the rapid progress in this area, it is appropriate and meaningful to summarise the developments in PC and PEC material designs for Cl₂/HOCl production. We believe that this timely review will guide catalytic researchers, materials scientists and computational chemists in catalytic materials designs and developments for industrially relevant processes.

2. Challenges in Cl₂ and HOCl production

Low Cl₂ and HOCl yields

Low yields and selectivity of Cl₂ and HOCl formation result from a few factors. First, further oxidation of Cl₂ or HOCl to higher chlorine oxo species, like ClO₂⁻, ClO₃⁻, and ClO₄⁻, under UV irradiation, which are usually toxic, is shown in eqn (8)–(10).^25,26 Formation of higher chlorine oxo species over a prolonged reaction time leads to the reduction of faradaic efficiency (FE).

Cl⁻ + 2H₂O → ClO₂⁻ + 4H⁺ + 4e⁻ E⁰ = 0.76 vs. RHE (8)

Cl⁻ + 3H₂O → ClO₃⁻ + 6H⁺ + 6e⁻ E⁰ = 1.45 vs. RHE (9)

Cl⁻ + 4H₂O → ClO₄⁻ + 8H⁺ + 8e⁻ E⁰ = 1.39 vs. RHE (10)

Second, Cl₂ dissolves in water to produce HOCl and HCl (refer to eqn (6)). HOCl has a limited stability and forms radicals, like ˙OH and ˙Cl, upon exposure to UV light irradiation (eqn (11)).^27–29 In the presence of ˙OH, it reacts with HOCl to generate ˙OCl and consequently to reduce the HOCl yield.³⁰ HOCl can also decompose to yield O₂ (eqn (12)) as discussed in the previous section. As a result, the use of wide band gap semiconductors like ZnO and TiO₂, which absorb UV light, poses a problem for Cl₂/HOCl formation. Notably, the deep valence band maximum for some semiconductor photocatalysts underscores their strong oxidising capability to decompose Cl₂ and HOCl.³¹ In addition, there is evidence of HOCl or OCl⁻ decomposition to ClO₃⁻ and Cl⁻ under UV (365 nm LED) irradiation, as reported by Okada et al., reducing the overall yield of HOCl (eqn (13)).³² As a matter of fact, Koo et al. showed that the formation of toxic ClO₃⁻ can be inhibited using visible light instead of UV light in PEC systems.³⁰

Decomposition of HOCl

For immediate applications, such as pollutant degradation or oxidative organic reactions using in situ generated Cl₂ and/or HOCl, rapid decomposition to radicals is highly favoured because they are responsible for oxidising pollutants. Similarly, in situ-generated O₂ from HOCl decomposition might serve as an oxidant in oxidation reactions. Although short-lived HOCl seems like a drawback owing to its high-yield production, its use in wastewater or sewage treatment and organic transformation is a merit. A discussion on the photodegradation of pollutants and bacteria, and the photocatalytic conversion of organic compounds, is provided in Section 7.

The need for O₂ supply

It has been found that the presence of O₂ for ORR is required to achieve a high production rate of HOCl in Cl⁻ oxidation. Molecular O₂ acts as an electron scavenger to suppress the recombination of charge carriers so that holes can be consumed by Cl⁻ ions for the chloride oxidation reaction (COR). For some photocatalysts, like BiOCl, molecular O₂ is required to prevent self-reduction of lattice Bi³⁺ to lower oxidation states, e.g. Bi metal. Nonetheless, the need for a constant supply of O₂ may restrict the feasibility of the green strategy in photocatalytic Cl₂/HOCl production. Additionally, CO₂ has also been found to quench electrons, leading to the coupling of the photoreduction of CO₂ and the photooxidation of Cl⁻ ions as a means of producing fuels and oxidants in a single photocatalytic system. Therefore, the use of natural seawater containing a decent amount of dissolved CO₂ might be an advantage for COR.

Photocorrosion and dissolution of catalytic materials

Another issue is the dissolution and/or photocorrosion of photocatalytic materials in a neutral seawater solution or seawater. The production of Cl₂ and HOCl leads to local acidification, which changes the pH of the seawater solution to become acidic (refer to eqn (6)). Jadwiszczak et al. calculated the pH of 0.5 M NaCl to be 2.5 upon Cl₂ evolution and dissolution.³³ Although some photocatalytic materials, like WO₃, TiO₂ and BiVO₄, are chemically stable in an acidic medium, other materials such as Cu₂O and CdS undergo photocorrosion upon exposure to light for a prolonged duration. Furthermore, although metal co-catalysts dramatically enhance the production rate and yield of Cl₂/HOCl, their use in acidic media has limited stability. For instance, Okada et al. observed a partial dissolution of Pt co-catalyst in acidic seawater solution at pH 1 after one hour of photocatalysis; it must be noted that the dissolution of Pt only occurs under photoirradiation.³² Therefore, modifications to photocatalysts and photoanodes are often required to preserve stability in the presence of Cl⁻ ions under highly oxidising conditions (Cl₂ and HOCl) and/or under photoirradiation. For example, a MoO_x layer is inserted between Co₂(OH)₃Cl and BiVO₄ to suppress the photocorrosion of Co₂(OH)₃Cl by photogenerated holes of BiVO₄.³⁴

Complex mixture of natural seawater

Natural seawater is an ideal reaction medium owing to its abundance and free availability. Although natural seawater contains Cl⁻ ions as a majority, the presence of other cations (Na⁺, Ca²⁺, Mg²⁺, and Ba²⁺) and anions (HCO₃⁻, CO₃²⁻, and SO₄²⁻) interferes with the surface of photocatalysts by selective adsorption and reaction, thereby reducing the overall production yield and rate of HOCl. Some of these ions, like CO₃²⁻, SO₄²⁻ and PO₄³⁻, are effective hole scavengers that compete with Cl⁻ oxidation, thereby reducing its selectivity.³⁵ Additionally, HCO₃⁻, CO₃²⁻, and PO₄³⁻ have been shown to consume Cl₂/HOCl.³⁶ Furthermore, the adsorption of these ions on the surface of photocatalysts might block the diffusion and interaction of Cl⁻ with the surface. Therefore, the deleterious effects of these ions are inevitable. As a result, designing photocatalytic materials that can perform efficiently in seawater containing various ions, as opposed to pure seawater solutions, is a major challenge in this research area. The average composition of natural seawater is summarized in Table 1.

Table 1 Percentage composition of ions in natural seawater^37,38

Ion	Composition (%)
Cl^−	55
Na^+	31
SO4^2−	7.7
Mg^2+	3.7
Ca^2+	1.2
K^+	1.1
HCO3^−	0.3
Br^−	0.2
CO3^2−	0.03
Sr^2+	0.03
Other ions	0.09

---

3. Mechanism of Cl₂ formation

Although Cl⁻ oxidation to Cl₂ looks like a simple 2-electron oxidation process, this mechanism is still under debate. Three mechanistic reaction pathways have been proposed based on theoretical calculations and experimental electrochemical studies: Volmer–Tafel, Volmer–Heyrovsky and Krishatalik.³⁹

Volmer–Tafel

2Cl⁻ + 2* → 2Cl* + 2e⁻ (14)

Cl* + Cl* → 2* + Cl₂ (15)

Volmer–Heyrovsky

2Cl⁻ + * → Cl* + Cl⁻ + e⁻ (16)

Cl⁻ + Cl* → * + Cl₂ + e⁻ (17)

Krishatalik

2Cl⁻ + * → Cl* + Cl⁻ + e⁻ (18)

Cl⁻ + Cl* → Cl*⁺ + Cl⁻ + e⁻ (19)

Cl⁻ + Cl*⁺ → * + Cl₂ (20)

where * represents the active sites, e.g. metal atoms or surface oxygen. In all the proposed pathways, a Cl⁻ ion is oxidised at the active site (*), forming adsorbed Cl* species. In the Volmer–Tafel mechanism, two Cl* species combine and are released from the surface of the electrode as Cl₂. However, in Volmer–Heyrovsky, Cl* reacts with a Cl⁻ ion to form Cl₂. The Krishatalik mechanism requires three steps for the formation of Cl₂ through the reaction between a Cl⁻ ion and Cl*⁺ species.

Three possible reaction pathways have been proposed based on theoretical computations, kinetic data and experimental verifications as plausible mechanisms. The elementary steps in COR are Volmer, Tafel, and Heyrovsky. The adsorption of Cl⁻ on the active site of an electrode is termed the Volmer step. The Tafel step includes a combination of two adsorbed Cl* species forming a Cl₂ product. The formation and desorption of a product following a reaction between an adsorbed species (Cl*) and a reactant (Cl⁻) is called the Heyrovsky step. In the Krishtalik step, a second electron transfer occurs between a reactant and an adsorbed species prior to product formation. The three mechanistic pathways are illustrated in Fig. 3.

Fig. 3 Scheme of the mechanistic pathways for chloride oxidation according to the mechanisms of Volmer–Heyrovsky, Volmer–Tafel, and Krishtalik. Reproduced from ref. 40. Copyright 2023, Wiley-VCH.

Herein, we discuss the mechanism of the chlorine evolution reaction (CER) using a RuO₂ model system, one of the best performing catalysts for the chlor-alkali process. The mechanistic reaction pathway is often proposed based on kinetic experimental data. However, kinetic data alone do not provide the nature of the active sites of a catalyst and intermediates. Moreover, the nature of the active sites has not been elucidated by in situ spectroscopy. Exner et al. performed ab initio atomistic thermodynamic calculations on the model catalyst O-covered RuO₂(110).⁴¹ Electrochemical adsorption (Volmer step) of Cl⁻ on the fully O-covered RuO₂(110) surface at electrode potential, U = 0 V, costs a Gibbs energy change of 1.49 eV. In the Volmer–Heyrovsky mechanism, a direct reaction between an adsorbed chlorine (Cl*) and a chloride (Cl⁻) from the solution to yield gaseous Cl₂ requires a Gibbs energy change of 1.23 and −0.13 eV at U = 0 V and 1.36 V, respectively. The net Gibbs energy loss in the Volmer–Heyrovsky mechanism is 0.13 eV. Meanwhile, the second Cl⁻ adsorption in the Volmer–Tafel mechanism requires a Gibbs energy change of 1.59 and 0.23 eV at U = 0 V and 1.36 V, respectively. Therefore, the Gibbs energy loss is 0.23 eV in the Volmer–Tafel mechanism. The formation of a Cl⁺ ion in the Krishtalik mechanism results in a Gibbs energy loss of 0.85 eV, which is much higher than the Volmer–Heyrovsky and Volmer–Tafel mechanisms. Hence, the Krishtalik mechanism can be ruled out for CER on the fully O-covered RuO₂(110) surface while favouring Volmer–Heyrovsky as the most plausible mechanism based on thermodynamic considerations. Additionally, doping a monolayer of PtO₂ in RuO₂(110) further lowers the Gibbs energy loss of 0.08 eV. Importantly, it should be noted that while the Volmer–Heyrovsky mechanism is the most plausible and preferred pathway, all proposed mechanistic pathways are feasible on the RuO₂(110) surface.³⁹ The simplified mechanism of Volmer-Heyrovsky on metal oxides (MOs) is given by

M–O + Cl⁻ → M–OCl + e⁻ (21)

M–OCl + Cl⁻ → M–O + Cl₂ + e⁻ (22)

In a previous study, Khrishtalik proposed a three-elementary step mechanism following a combination of electrochemical measurements and electrokinetic data on RuO₂–TiO₂.⁴² A combination of a low coverage of intermediate and a low slope of the anodic polarisation curve provides a strong indication that the transfer of the first electron is reversible and the second electron transfer is a slow step. A total of two electrons per one Cl⁻ is consumed in the slow step, suggesting that a positive Cl (Cl⁺) should react with a chloride (Cl⁻) to form Cl₂. Khristalik then proposed a mechanism with three elementary steps (eqn (18)–(20)) for CER. Although this proposed mechanism based on electrokinetic data and electrochemical measurements seems plausible, it certainly does not provide a correlation with the physical and structural characteristics of electrocatalysts, a limitation that must be addressed to gain a correct understanding of CER selectivity on the electrode surface. Furthermore, the microscopic molecular level of insights is not captured and elucidated by electrokinetic data, which often requires computational calculations, such as ab initio DFT with thermodynamic considerations/frameworks. DFT calculations by Hansen et al. on chlorine evolution disclosed the possibility of Cl⁻ adsorption on two oxygen atoms in RuO₂(110), forming ClO₂^cc (cc denotes two coordinatively unsaturated sites).⁴³ This result, in which ClO₂^cc is more positively charged than O₂^cc, is consistent with the Khrishtalik mechanism.

4. Detection and determination of Cl₂ and HOCl concentrations

The photocatalytic activity in Cl⁻ oxidation is assessed using the production rate and/or concentration of Cl₂ or HOCl. The concentration of Cl₂ in seawater splitting can be quantified using iodometric titration. The iodide-starch test is usually employed where iodide is oxidised to iodine in the presence of Cl₂ as an oxidising agent, following eqn (23). The generated iodine is titrated with sodium thiosulfate using starch as a colour indicator to determine the concentration of iodine (eqn (24)), and thus Cl₂ indirectly.⁴⁴

Cl₂ + 2I⁻ → I₂ + 2Cl⁻ (23)

I₂ + 2S₂O₃²⁻ → 2I⁻ + S₄O₆²⁻ (24)

The evolved Cl₂ can also be detected using the N,N′-bis(2,4-di-sulfobenzyl)tolidine tetrasodium method. Briefly, Cl₂ is purged from a photoreactor under an inert gas flow and trapped in a 4 M aqueous NaOH solution to form hypochlorite (ClO⁻) ions, and the amount of ClO⁻ is determined using N,N′-bis(2,4-di-sulfobenzyl)tolidine tetrasodium (SBT).³² The oxidized form of SBT following a reaction with chlorine has a characteristic absorption at 675 nm at pH 5.2, which allows a spectroscopic detection.⁴⁵ The concentration of HOCl is commonly determined spectroscopically using the N,N-diethyl-p-phenylenediamine (DPD) method. In the presence of HOCl, DPD is oxidized to a red species with characteristic absorption at 515 or 552 nm.

The selectivity for Cl⁻ oxidation to Cl₂ or HOCl in PEC systems is often evaluated in terms of faradaic efficiency (FE), which is the ratio of the measured HOCl/Cl₂ concentration to the theoretical one when passed through current. Hence, the sum of FE for HOCl and O₂ should ideally be unity (100%). The FE can be calculated based on the following formula (eqn (25)):

where F is the Faraday constant, n(Cl₂) is the number of moles of Cl₂ produced, and Q is the number of charges passed through the electrochemical system.

5. Photocatalytic Cl₂/HOCl production

Photocatalytic systems are much simpler than their photoelectrocatalytic counterparts and can be more readily scaled up. One advantage of nanoparticulate powders used in photocatalytic systems is gram-scale preparation. Furthermore, the former does not need a specialized set up and can be used in a simple test tube or beaker in a slurry reactor without an external bias (or applied potential). Photocatalytic systems run with a single-compartment reactor where both oxidation and reduction occur in the same compartment, unlike PEC systems, which have dual compartments containing an anode and a cathode for both half-reactions. This, in turn, often requires the use of an electron scavenger, e.g. molecular O₂, to enhance the formation of Cl₂ because both oxidation and reduction take place on the same photocatalytic system. Photocatalytic materials can be prepared using various wet chemistry methods, such as sol–gel, hydrothermal/solvothermal and colloidal synthesis. These photocatalytic materials can be annealed further at high temperatures to improve crystallinity and enhance photocatalytic performance.

Metal oxides are commonly employed as photocatalysts owing to their chemical stability, suitable band gap, low toxicity, low-cost and facile preparation, as well as a highly positive valence band that is suitable for chloride oxidation reaction (COR). TiO₂, WO₃, Bi₂WO₆, and BiVO₄ are the most extensively studied metal oxide photocatalysts. WO₃ exhibits a moderate band gap (2.5–2.7 eV) and low toxicity and possesses suitable band edge positions for water and Cl⁻ oxidation (Fig. 2). It is well known that photoactivity depends on the size, morphology, and defect chemistry of photocatalysts. Lu et al. synthesized various WO₃ morphologies, including nanorods, nanowires, nanotube bundles, nanoplates and hollow microspheres, loaded with 1 wt% Pt as a co-catalyst and studied the structure-activity relationships in HOCl synthesis.⁴⁶ Pt/WO₃ hollow microspheres show the highest HOCl concentrations in 0.5 M NaCl and seawater at 14.52 µM (1.24 mg L⁻¹) and 4.34 µM (0.354 mg L⁻¹), respectively, with good stability up to four recycle tests. The author attributed the highest photoactivity to the hollow structure of WO₃ that facilitates charge separation. This work achieved the recommended free chlorine (0.2–2 mg L⁻¹) set by the WHO, highlighting its practical feasibility.

A systematic study on the effects of photocatalysts (BiVO₄, TiO₂, H₂WO₄, and Bi₂WO₆) and metal co-catalysts (Au, Ag, Cu, Rh, Pd and Pt) by Pang et al. disclosed that 1 wt% Pt/WO₃ exhibited the highest photoactivity with 17.6 µM HOCl produced in 1 hour while suppressing water oxidation (Fig. 4a).¹⁰ A Pt co-catalyst enhances the catalytic activity by serving as an electron sink to capture photoexcited electrons from the conduction band (CB) of WO₃ and reduce O₂ to superoxide O₂˙⁻ or H₂O. The same group also modified Pt/WO₃ by co-loading with various other metals as a co-catalyst. The Pt–M (M = Ni, Cr, Mn, Co, and Zn) co-catalyst improves the selectivity toward Cl⁻ oxidation to HOCl and promotes the decomposition of H₂O₂ (a product of ORR) to prevent its reaction with formed HOCl (Fig. 4b); Pt–Mn/WO₃ shows the highest HOCl production (37 µmol h⁻¹), which is twice that of Pt/WO₃ in an earlier work.⁴⁷ However, the dissolution of Mn presents a problem for recyclability and long-term usage of Pt–Mn/WO₃. Moreover, precise control must be taken to avoid the formation of MnO_x, which favours the oxidation of water over Cl⁻ oxidation.⁴⁸ Interestingly, the formation rate of HOCl is suppressed severely over Pt–M (M = Fe, Cu, and Ag). To ease recyclability in heterogeneous photocatalysis, a magnetic material, like Fe₃O₄, is typically incorporated into catalytic material designs for magnetic separation. Wang et al. constructed a multishell quaternary Pt/WO₃/SiO₂/Fe₃O₄ photocatalyst that achieved 6.31 µM HOCl in 2 hours.⁴⁹ The SiO₂ intermediate layer prevents an electron transfer from WO₃ to Fe₃O₄, inhibiting the loss of activity in HOCl generation.

Fig. 4 Schematic of the Cl⁻ oxidation mechanism over (a) Pt/WO₃; reproduced from ref. 10. Copyright 2020, the American Chemical Society; and (b) Pt-Mn/WO₃ photocatalysts; reproduced from ref. 47. Copyright 2022, the Royal Society of Chemistry.

A family of metal tungstate (MWO₄, where M is a bivalent metal) has emerged as a new type of photocatalyst owing to its low toxicity, low-cost and earth-abundant starting materials.⁵⁰ Wide band gap ZnWO₄ nanorods show a hypochlorite (OCl⁻) yield of 1.01 µg mL⁻¹ in seawater.⁵¹ The production yield increases 2.5-fold when ZnWO₄ is loaded with an IrO₂ co-catalyst due to higher selectivity of Cl⁻ oxidation on IrO₂. More recently, Han et al. reported that MnWO₄ with a band gap of 2.83 eV could produce 12.6 µM h⁻¹ HOCl in 0.5 M NaCl with good stability and recyclability of up to four runs, paving the way toward promising noble metal-free photocatalysts.⁵² In another work, the same group constructed a Z-scheme WO₃/CdS photocatalyst. An excited electron in the CB of CdS serves to reduce O₂ to ˙O_2⁻ and a photogenerated hole in the VB of WO₃ oxidises Cl⁻ to free Cl₂ (3.54 mg L⁻¹),⁵³ with a much higher activity than Pt/WO₃ (1.24 mg L⁻¹ free Cl₂).⁴⁶ This work demonstrates the feasibility of designing promising noble metal-free photocatalysts for HOCl synthesis. Notably, noble metals, like Pt and Au, are not involved in Cl⁻ oxidation but are used to quench the excited electrons by capturing them from the VB of the photocatalyst. A similar quenching role can be applied through a Z-scheme mechanism or other types of photocatalyst design, thereby reducing the material cost in catalyst development.

In 2020, Li et al. investigated the origin of elevated Cl₂ in the atmosphere using KCl-coated TiO₂ to understand atmospheric Cl₂ chemistry in urban areas.⁵⁴ Although this work is not directly related to COR via seawater splitting, it shows a strong correlation between the amount of surface-bound Cl⁻ and the amount of Cl₂ evolved. TiO₂ is widely used in photocatalytic research due to its low cost, low toxicity and high chemical stability in a wide pH range.⁵⁵ Nonetheless, due to the fast charge carrier recombination and low photocatalytic activity, metal nanoparticles are often required to boost photoactivity. Okada et al. systematically studied the effects of pH on seawater splitting over Pt/TiO₂ (0.1 wt% Pt) and demonstrated a near stochiometric evolution of H₂ and Cl₂ with a minor O₂ evolution from direct water oxidation for over 14 hours (Fig. 5a and b); the selectivity toward Cl₂ formation is 81.1%, highlighting preferential oxidation of Cl⁻ over water oxidation.³² High Cl₂ evolution was detected in an acidic (pH 1) seawater solution driven by the change in the VB maximum of TiO₂, contributing to stronger Cl⁻ oxidation at low pH. In another work, Jadwiszczak et al. found that PEC Cl₂/HOCl synthesis in acidic environments can suppress their decomposition to chlorates and oxygen, thus favouring acidic conditions for promoting high Cl₂/HOCl yields.³³ Nonetheless, the Cl₂ evolution could not be detected in neutral and basic conditions in Okada's work. The reason for inactivity is attributed to the rapid decomposition of HOCl to ClO₃⁻ and Cl⁻ ions. Additionally, the competing reactions between Cl⁻ and OH⁻ oxidation at high pH, which favours the latter, could be another reason for inactivity.^56,57 More recently, the same group studied the effects of Pt loading on the production rate of H₂ and Cl₂.⁵⁸ The rate of gas evolution increases as the Pt loading increases and reaches a maximum at 1.0 wt% (Fig. 5c). Beyond 1.0 wt% of Pt loading, the photoactivity decreases most likely due to the light-shading effects of Pt. Notably, it is worth noting that Pt dissolution in an acidic seawater solution is triggered by photoirradiation, with two plausible mechanisms of dissolution. First, photogenerated electrons reduce PtO_x to Pt, and PtO_x reacts with Cl⁻ to form PtCl₆²⁻ species, which are soluble in an aqueous solution. In the second mechanism, photogenerated electrons and holes reduce PtO_x and oxidize Cl⁻ to Cl₂, respectively, where Cl₂ then reacts with Pt to form soluble PtCl₆²⁻.

Fig. 5 Photocatalytic seawater splitting (0.5 M NaCl) over Pt/TiO₂ (a) at different pH values (1, 6, 13) and (b) time course reaction over 14 hours. Reproduced from ref. 32. Copyright 2024, the Royal Society of Chemistry. (c) Time course reaction with different Pt loadings (wt%) on TiO₂. Reproduced from ref. 58. Copyright 2025, Elsevier.

Rh/Cr₂O₃ and RhCrO_x alloy are used extensively as a co-catalyst for the hydrogen evolution reaction (HER) in photocatalytic water splitting.^59–62 Importantly, CrO_x overlayers have been shown to act as a protective layer to prevent unwanted side reactions and undesirable species in water splitting.⁶³ Nonetheless, their use in seawater splitting is relatively unexplored; thus, the knowledge about their roles and performance in seawater splitting is limited. Okunaka et al. investigated the role of co-catalyst (RhO_x, CrO_x, and RhCrO_x) loaded on SrTiO₃:Al for HOCl production in aqueous NaCl solution.⁶⁴ It was found that only RhCrO_x shows a significant HOCl production along with H₂ evolution, where the Rh species promotes H⁺ reduction and CrO_x suppresses HOCl decomposition. Additionally, the role of RhCrO_x is to enhance the charge separation between the co-catalyst and the photocatalyst.

The extraction and insertion of surface lattice atoms/ions via self-oxidation of lattice Cl⁻ in Cl-containing photocatalysts (e.g. AgCl and BiOCl) to Cl₂ by a photogenerated hole provide smooth oxidation, often followed by a structural change during a catalytic cycle. The lost/oxidized lattice Cl⁻ is then replaced by Cl⁻ ions in the bulk seawater solution, creating a chlorine cycle. The participation of lattice atoms/ions as active species in chemical reactions is known as the Mars–van Krevelen (MvK) mechanism. Hirai's group reported that the compensation of Cl⁻ ions in seawater solution suppresses photocatalyst deactivation.⁶⁵ In light of this characteristic, AgCl and BiOCl have advantages for photocatalytic Cl⁻ oxidation in seawater splitting. Although wide band gap AgCl scarcely exhibits photocatalytic activity, loading Au NPs onto AgCl drastically boosts Cl⁻ oxidation with a HOCl formation yield of 21.4 µmol, as reported by Shiraishi et al.⁶⁶ The localised surface plasmon resonance (LSPR) in Au NPs enables visible light (>420 nm) absorption to create hot electrons (e_hot⁻) and holes (h_hot⁺), where the latter oxidises lattice Cl⁻ in AgCl to Cl₂ (Fig. 6). Although the redox potential of Cl⁻/Cl₂ (1.36 V vs. RHE) is more positive than the VB of Au NP, the photogenerated holes in the VB of Au NP can oxidize the lattice Cl in AgCl, circumventing the insufficient redox potential. The oxidation of lattice Cl⁻ was confirmed by XPS analysis and control experiments.

Fig. 6 Band energy diagrams of (a) AgCl and (b) Au/AgCl. Reproduced from ref. 66. Copyright 2023, the American Chemical Society.

Another promising photocatalyst that has garnered attention is the non-toxic, inexpensive, and wide band gap BiOCl (3.1–3.3 eV). The structure is made of [Bi₂O₂]²⁺ slabs interleaved between double slabs of [Cl]⁻ along the [001] direction, which creates an internal electric field (IEF) along the crystal orientation perpendicular to the [Bi₂O₂]²⁺ and [Cl]⁻ slabs that promote charge separation and migration (Fig. 7a–c).⁵² Atoms in the [Bi₂O₂]²⁺ layers are held by a covalent bond, while the interaction between the [Cl–Bi–O–Bi–Cl] stacks is held together by a non-covalent van der Waals force.⁶⁷ Therefore, the removal of Cl⁻ ions from the crystal structure of BiOCl is much easier owing to the weaker van der Waals interaction, which is beneficial for Cl⁻ lattice extraction and insertion, as demonstrated in chloride ion batteries,⁶⁸ which may have implications for photocatalytic Cl⁻ oxidation.

Fig. 7 Simplified surface of BiOCl with (a) different facets. (b) Side view of the surface along the [001] direction and (c) top view of the surface along the perpendicular [110] direction. TEM images of BiOCl (d) before reaction and (e) after reaction. DFT calculations of the free energy for the Cl₂ evolution reaction on the (f) (001) surface and (g) (110) surface. Reproduced from ref. 56. Copyright 2021, Elsevier.

Wang et al. demonstrated that the insertion/extraction of Cl⁻ proceeded on the {110} facet, as evidenced by the photocorroded {110} facet of the BiOCl microplates after several catalytic runs (Fig. 8d and e).⁵⁶ The finding is reinforced by DFT calculations, which suggest that the open channel {110} facet exhibits lower overpotential and easier lattice Cl⁻ transport than the {001} facet, giving simultaneous advantages of Cl⁻ extraction and oxidation (Fig. 8f and g). In another work, Zhou et al. systematically investigated the effects of different facets {001} and {110}, on the formation rate and yield of HOCl. The highest production rate of RCS (95.9 µg L⁻¹ min⁻¹) was achieved with the largest {001}/{110} ratio owing to the largest IEF, as verified by transient photocurrent response, EIS and PL spectroscopy; the lowest {100}/{110} ratio of BiOCl yields 75.6 µg L⁻¹ min⁻¹.³⁶ The Cl⁻ lattice oxidation and extraction contribute to the weakening of Bi–Cl vibrations, the emergence of metallic Bi (reduction of Bi³⁺ to Bi⁰), and the formation of oxygen vacancies, as confirmed by Raman spectroscopy, XRD and XPS.^36,69 The presence of molecular O₂ helps to prevent self-oxidation of lattice Bi³⁺ and regulate the electronic structure of BiOCl by adsorbing to the oxygen vacancies (V₀), usually leading to a higher photoactivity in Cl⁻ oxidation.

Fig. 8 (A) Transient photocurrent measurements for the as-deposited WO₃ electrodes at (i) 0.8 V, (ii) 0.5 V, and (iii) 0.2 V vs. SCE and (iv) annealed WO₃ electrode at 0.5 V vs. SCE. Illustration of charge behaviour (B) when irradiated with light, (C) under photostationary conditions, and (D) when light is off. Reproduced from ref. 79. Copyright 2013, the American Chemical Society.

Doping foreign elements (non-metals or metals) into a host material has been a straightforward strategy for improving charge transport properties. In 2016, Li et al. doped Bi₃O₄Cl with carbon to boost the IEF by 126 times and achieved a bulk charge separation efficiency of 80% that enabled photocatalytic water oxidation without noble metals or electron scavengers.⁷⁰ Motivated by their work, Xu et al. reported that carbon-doped BiOCl (C-BiOCl) was found to enhance IEF and improve charge separation and migration as well as preference for Cl⁻ oxidation over water oxidation in photocatalytic Cl₂ formation.⁷¹ A 2-fold enhancement in IEF in C-BiOCl leads to a 2-fold increase in the production rate of Cl₂ (0.23 mg L⁻¹ min⁻¹) compared to that of pristine BiOCl (0.11 mg L⁻¹ min⁻¹), underscoring a strong correlation between IEF and photoactivity. Nitrogen doping in BiOCl has the advantage of providing basic Lewis sites for reduction reactions. Shi et al. attempted the thermal decomposition of ammonium oxalate that delaminated bulk BiOCl into atomic-layer nanosheets and simultaneously introduced N dopants and O vacancies (N–O_V associate). The N–O_V associate reduces the exciton binding energy (from 199 to 46 meV) and lifetime (107.5 to 25.5 ms) to facilitate dissociation of an exciton into free charge carriers, leading to a HOCl production rate of up to 87.2 µmol g⁻¹ h⁻¹ (Table 2).⁷²

Table 2 Summary of the photocatalytic performance of various photocatalysts in HOCl production

Photocatalyst	Light source	Reaction solution	[HOCl]^a/µM	Time /h	AQY (%)	Ref.
a The default product measured was set to HOCl as determined using the DPD method. However, some studies reported the yield of Cl2.b SCC refers to solar-to-chemical conversion.
0.1% Pt/TiO2	LED 365 nm	0.5 M NaCl (pH 1)	Cl2: 10.3	1	∼0.65% (365 nm)	32
1% Pt/TiO2	LED 365 nm	0.5 M NaCl (pH 1)	Cl2: 70 µmol h^−1	—	4.3% (365 nm)	58
1% Pt/WO3	AM 1.5 G	0.5 M NaCl	17.6	1	2.3% (420 nm)	10
Pt–Mn/WO3	300 W Xe (>420 nm)	0.5 M NaCl	37	1	—	47
1% Pt/WO3 hollow microsphere	Visible light	0.5 M NaCl	14.52	2	—	46
		Seawater	4.34	1	—
Pt/WO3/SiO2/Fe3O4	White LED	0.5 M NaCl	6.31	2	—	49
MnWO4	Visible light	0.5 M NaCl	12.6	1	SCC^b 0.15%	52
IrO2/ZnWO4	UV 254 nm	Seawater	2.49 µg mL^−1	1	—	51
CdS/WO3	LED (380–780 nm)	0.5 M NaCl	3.54 mg L^−1	2	—	53
0.1% RhCrOx/SrTiO3:Al	AM 1.5 G	1 M NaCl	∼1 µmol	1	—	64
Au/AgCl	Xe lamp (>420 nm)	0.55 M NaCl	21.4 µmol	24	SCC^b ∼0.03%	66
BiOCl	UV lamp	6 g L^−1 NaCl	Cl2: 0.125 mg L^−1 min^−1	1	—	69
BiOCl	UV 254 nm	6 g L^−1 NaCl	9.8 µmol	0.5	—	56
BiOCl	UV 254 nm	6 g L^−1 NaCl	95.9 µg L^−1 min^−1	—	—	36
C-BiOCl	UV 254 nm	6 g L^−1 NaCl	0.023 mg L^−1 min^−1	1	—	71
BiOCl–N–Ov	300 W Xe (>400 nm)	0.5 M NaCl with CO2 saturation	87.2 µmol g^−1 h^−1	4	0.69% (400 nm)	72
BiOCl	UV lamp	0.51 M NaCl	19.8 µmol	0.5	—	73

---

The ocean absorbs around 30% of the CO₂ that is emitted to the atmosphere, which consequently decreases the pH of the ocean, also known as ocean acidification. The dissolution of CO₂ in water forms weak carbonic acid (H₂CO₃), which exists in equilibrium with (H⁺) and bicarbonate (HCO₃⁻) ions. A recent work revealed that dissolved CO₂ in a NaCl solution boosted the formation rate of Cl₂ on BiOCl by over 3-fold and 2-fold compared to N₂-rich and O₂-rich atmospheres, respectively.⁷³ SEM images of BiOCl before and after photocatalysis disclose interlaminar deformation and distortion, which are probably caused by the insertion of CO₂ into the interlayers and anion exchange. The proposed mechanism of CO₂-mediated formation of Cl₂ in synthetic seawater is given by

HCO₃⁻ → CO₃²⁻ + H⁺ (26)

CO₃²⁻ → ˙CO₃⁻ + e⁻ (27)

˙CO₃⁻ + H₂O + h⁺ → HCO₄⁻ + H⁺ (28)

HCO₄⁻ + 2Cl⁻ + 2H⁺ → HCO₃⁻ + Cl₂ + H₂O (29)

6. Photoelectrochemical (PEC) Cl₂/HOCl production

In a photoelectrocatalytic system, HOCl formation occurs at the photoanode, and other reduction reactions proceed at the photocathode, including H₂ evolution, H₂O₂ production or CO₂ reduction (in the presence of CO₂). Advantages of photoelectrocatalysis are relatively simple electrode construction with tailored properties, such as controlled thickness, porosity, defects (or vacancies) and other surface properties. Photoelectrode performance is highly dependent on the interface, morphology and thickness of the thin film electrode. The thickness affects the light absorption (up to several µm), photocurrent generation and charge carrier diffusion length (up to several hundred nm). These properties are easily optimized to maximize photocurrent responses and charge transport efficiency. Moreover, it allows the use of various techniques, like layer-by-layer deposition, electrodeposition, pulse-laser deposition (PLD), and successive ionic layer adsorption and reaction (SILAR). Additionally, catalytic materials can be prepared using common methods, such as sol–gel and hydrothermal, and successively deposited on conducting substrates (ITO, FTO) by doctor-blading, drop-coating and spin-coating.

WO₃ exhibits high chemical stability in acidic solution (pH < 4), which is beneficial for photoelectrochemical Cl₂ synthesis (eqn (4)) but prone to dissolution in neutral or basic media.⁶⁰ An acidic seawater solution due to the presence of formed HOCl is advantageous for WO₃ because it ensures the WO₃ stability. At a higher pH, WO₃ forms WO₄²⁻ by reacting with OH⁻, leading to its dissolution in alkaline media. As early as 1987, Desilvestro and Gratzel reported ∼76% FE of Cl₂ production over polycrystalline WO₃ in 1 M HCl.⁷⁴ It has also been demonstrated that porous structures of WO₃ favour Cl⁻ oxidation over water oxidation. Recently, highly porous WO₃ nanosheets with a specific surface area of 100 m² g⁻¹ and a thickness between 10 and 20 nm prepared by chemical bath deposition achieved 100% FE.⁷⁵ However, a rapid decrease in FE as the charge increased was observed most likely due to the formation of higher chlorine oxo species. Hill and Choi showed that WO₃ in PEC favoured the oxidation of Cl⁻ (with 95% consumption of photogenerated holes) and completely suppressed water oxidation to O₂ and peroxo species in aqueous NaCl solution.⁷⁶ A similar observation of suppression of water oxidation was reported by Mi et al. using a 0.1 M HCl aqueous solution.⁷⁷ These studies demonstrated WO₃ thin films as a promising photoanode for Cl₂ evolution.

Shi et al. reported a stable WO₃ film prepared by 25 cycles of deposition-annealing process that produced 1.95 mA cm⁻² at 1.23 V (vs. RHE) in natural seawater. However, the FE of Cl₂ evolution was quite low (56%).⁷⁸ Thick, nanoporous WO₃ ca. 3 µm with particle size of 20–30 nm provides a large internal surface area that allows contact with electrolytes, achieving large saturation photocurrents of 4.5 mA cm⁻² at 0.76 V (vs. Ag/AgCl), although the FE is moderate (70%) due to competing O₂ evolution.³³

Annealing of thin films is usually performed to remove excess solvent, improve the crystallinity, control the structural phase, and create a strong interface contact between the sample and a substrate (ITO and FTO). A “roll-on” process was employed to deposit commercial nano-WO₃ on ITO. The annealed WO₃ film at 500 °C shows a much higher transient photocurrent and Cl₂ concentration due to enhanced interparticle charge transport by hopping between aggregates compared to the as-made WO₃ film (Fig. 8).⁷⁹ The influences of annealing temperature on the compositional phases and PEC performance of WO₃ electrodes were reported by Jo et al. At the annealing temperature of 500 °C, the WO₃ contained predominantly monoclinic (87%) and orthorhombic (13%) phases, which gave the highest photocurrent of 2.7 mA cm⁻² and FE of 93% at 1.0 V (vs. Ag/AgCl).⁸⁰ The fraction of monoclinic increases as the annealing temperature increases, which induces more defects and increases donor density for enhanced PEC performance. However, too large a concentration of defects can be detrimental to the PEC performance of WO₃ in Cl⁻ oxidation because defects could serve as a recombination centre.⁷⁵

Recently, Johnston and co-workers employed photoinduced absorption spectroscopy (PIAS) and transient photocurrent (TC) measurements to unravel the reaction order of Cl⁻ oxidation on a WO₃ photoanode.¹⁸ It has been found that the rate of Cl₂ formation has a first-order dependence on the surface holes, [h_s⁺], albeit the faradaic efficiency is not unity (62%). The remaining 38% is speculated to be due to the formation of higher oxo species, such as ClO₄⁻ or Cl₂, trapped in the PEC cell, as no O₂ was detected in the study. Such a high selectivity of Cl⁻ oxidation over water oxidation reinforces the lower overpotential and higher exchange current density of the former reaction.^77,81

Proton intercalation can modify the electronic structure and optical properties of WO₃, rendering H_xWO₃ a promising photocatalyst due to increased electron mobility and high conductivity.⁸² Remarkably, Bartlett's group unravelled that fast electron injection from Cl⁻ ions into the VB of H_xWO₃ leads to 100% FE for Cl⁻ oxidation.⁸³ Such fast electron injection also suppresses oxidation of W⁵⁺ by holes, giving high photostability and stable photocurrent during Cl⁻ oxidation in 0.5 M NaCl at pH 4.

BiVO₄ has a moderate band gap (2.4–2.5 eV) and a deep valence band suitable for many oxidation reactions; therefore, it has been widely employed as a photoanode in water oxidation under visible light irradiation.^84,85 Nonetheless, the fast charge carrier recombination, low electron mobility, poor charge transfer and short hole diffusion length (∼70 nm) become obstacles to achieving its theoretical maximum STH (9.2%) and current efficiency (7.5 mA cm⁻²).⁸⁶ Doping BiVO₄ with other atoms or ions is a straightforward and efficient strategy that increases charge carrier density, improves charge separation and transfer efficiencies, introduces surface states, modifies the electronic band structure and enhances IPCE of BiVO₄.^87–89 Mo-doped BiVO₄ exhibits enhanced IPCE, charge accumulation and FE of 78% with ∼7.5 µmol ClO⁻ produced at ∼2.1 C compared to the undoped BiVO₄ owing to the increased donor density by Mo dopants in BiVO₄.⁹⁰ Similarly, Luo et al. reported that Mo-doped BiVO₄ showed a 6-fold increase in photocurrent and 42% IPCE in the wavelength range of 365–440 nm compared to pure BiVO₄ (10% IPCE).⁹¹ The Mo⁶⁺ substitution of V⁵⁺ in BiVO₄ enhances electron donation and creates more carrier density, and further loading with the RhO₂ layer results in improved photostability in seawater.

The use of a multilayer strategy to improve charge separation and transport efficiency, as well as catalytic activity, is ubiquitous in designing efficient photoanodes.⁸⁶ In the construction of a multilayer photoanode, WO₃ is first deposited onto an FTO glass prior to BiVO₄ deposition, in which the WO₃ layer serves as an electron transport layer (ETL) and a hole mirror to block the transfer of holes to the FTO substrate, leading to an enhanced charge transfer. It was reported that a multilayer BiVO₄/WO₃ photoanode exhibited a low applied bias (0.5 V) to achieve high current efficiency (∼80%) and HOCl concentration (660 µM) in the 2000 s under optimised conditions.⁹ Without the WO₃ underlayer, the formation yield and current efficiency of HOCl are lower and require a much higher external bias. A similar BiVO₄/WO₃ photoanode has been extended to the PEC production of HOBr.⁹²

Surface passivation with metal oxide overlayers has several benefits. The overlayer can serve as a protective layer to prevent photocorrosion or dissolution, as a co-catalyst to enhance activity and selectivity in Cl⁻ oxidation, as a hole (or electron) transport layer to boost charge transport, and as a passivating layer to quench surface states.⁹³ Okunaka et al. explored several metal oxides (MO_x; M = Mn, Co, Ni, Fe, Pt, and Rh) as a surface passivator on BiVO₄/WO₃ photoanodes to investigate the selectivity (O₂ evolution vs. HOCl production) of oxidation half-reaction.⁴⁸ All metal oxides except MnO_x favour the formation of HOCl owing to their lower overpotential for Cl⁻ oxidation (Fig. 9a). However, it should be noted that an accurate determination of each (O₂ and HOCl) overpotential in these photoanodes is difficult because both water and Cl⁻ oxidation occur with different selectivities.

Fig. 9 (a) Faradaic efficiency of the HClO and O₂ formation on various metal oxide overlayers. Reproduced from ref. 48. Copyright 2020, Elsevier. (b) Cl₂ production rate and faradaic efficiency with different CoO_x thicknesses. Reproduced from ref. 94. Copyright 2023, the American Chemical Society. (c) HOCl concentrations with varying CoO_x precursor concentrations. (d) Effects of calcination temperature on the HOCl concentration and faradaic efficiency. Reproduced from ref. 95. Copyright 2021, the Royal Society of Chemistry.

A unique feature of monoclinic BiVO₄ is spatial charge separation by different crystallographic facets. BiVO₄ is known to accumulate photogenerated holes at the {110} facet for oxidation reactions and to direct photoexcited electrons to the {010} facet for reduction reactions.⁷⁶ Hence, selective deposition of cobalt borate (CoBi) as a co-catalyst on the WO₃/BiVO₄ {110} facets has been observed to enhance the charge transfer kinetics, photocurrent response and conversion of photogenerated holes to redox species at the anodic sites for improved Cl₂ production.⁹⁶

A thin layer coating with amorphous cobalt oxide (CoO_x) prepared by pulsed laser deposition is demonstrated to act as a co-catalyst and to protect BiVO₄ against corrosion. The thickness of CoO_x overlayer from 2 to 5 nm is controlled using the number of laser pulses, with the optimum thickness of 3 nm achieving a current density of 2.5 mA cm⁻² at 1.2 V (vs. Ag/AgCl) and 60% FE in 3 M KCl solution at pH 7 (Fig. 9b).⁹⁴ By studying various CoO_x precursor concentrations, Sayama's group achieved near 100% FE of HOCl formation with 0.01 M CoO_x precursor concentration (Fig. 9c).⁹⁵ Such an enhancement in FE is attributed to improved charge separation, efficient capture of photogenerated holes by CoO_x, and possibly the presence of active sites for Cl⁻ adsorption on CoO_x.⁹⁷ These studies showed that beyond a certain CoO_x thickness (>3 nm) and loading (>0.03 M), the FE and current density decrease due to increased charge carrier recombination, thus requiring optimisation studies to correlate to reach the right conclusion.

Sayama's work also showed that the calcination temperature of CoO_x precursor should be optimized at 400 °C because a higher temperature (>500 °C) promotes Co doping in BiVO₄, which may favour recombination of charge carriers, thus lowering the FE (Fig. 9d).⁹⁵ However, this finding seems to contradict Chauhan's work, which demonstrates that Co doping in BiVO₄ reduces the charge transfer resistance, facilitates charge carrier lifetime and improves the Cl⁻ oxidation kinetics.⁹⁸ The authors investigated the effect of Co dopant concentration in BiVO₄ and found that 0.05 mol% Co gives the highest photocurrent of 0.190 mA cm⁻² at 1.1 V vs. RHE and 92% FE, which exhibited excellent stability against photocorrosion in acidic (pH 2.3) medium.⁹⁸ Other Co dopant concentrations yield much lower photocurrent and higher onset potential. The PEC performance is highly dependent on dopant concentrations; therefore, a systematic study of Co dopant concentration and its effect on the PEC performance must be optimized to achieve maximum FE and current density, as well as to resolve conflicting findings with other works.

The instability of BiVO₄ in electrolytes and acidic medium poses a problem for PEC reactions because it causes the dissolution of V⁵⁺ ions, leading to compositional and structural changes.^99,100 Chen and Wang theoretically demonstrated that BiVO₄ undergoes photocorrosion in an acidic medium (pH = 0) to form BiCl₃.¹⁰¹ To prevent photocorrosion, a 20 nm-thick amorphous WO₃ layer is applied as a coating on BiVO₄ owing to its high chemical and electrochemical stability in acidic media and high selectivity for the COR over the WOR (Fig. 10a–c).¹⁰² This strategy achieved a photocurrent density of 2.5 mA cm⁻² at 1.42 V (vs. RHE) and FE of 85% (Fig. 10d). The photocurrent is remarkably stable for up to three hours.

Fig. 10 (a) Illustration of the BiVO₄/WO₃ photoanode, (b) energy level diagram of the BiVO₄/WO₃ photoanode, (c) TEM image of BiVO₄/WO₃, and (d) photocurrent density across the applied potential. Reproduced from ref. 102. Copyright 2019, Springer Nature.

Pristine TiO₂ suffers from fast charge carrier recombination and separation. Modification strategies, such as doping, defect creation, and structural transformation are often performed to improve charge separation and transfer. 1D TiO₂ nanostructures, such as nanorods and nanotubes, are favourable photocatalysts owing to their high specific surface area and improved charge transport properties. TiO₂ nanotubes demonstrate a higher FE of 60.6% at 0.2 V (vs. Ag/AgCl) than TiO₂ nanoparticles prepared by sol–gel for Cl₂ formation.¹⁰³ Ye et al. introduced oxygen vacancies (V_o) in TiO₂ nanotubes (TiO₂-V_o) and achieved 37.1 µmol h⁻¹ cm⁻² of Cl₂ with FE of 73.9%.⁴⁴ The improved PEC performance of oxygen-deficient TiO₂ nanotubes by 1.5-fold compared to bare TiO₂ nanotubes is attributed to the enhanced utilization of photogenerated charge carriers promoted by oxygen vacancies rather than selective COR over OER. Another study also showed an improved photocurrent in rutile TiO₂ with V_o and stability of Ti³⁺ surface states over 5 hours in PEC conditions.⁵⁷ It was suggested that Ti³⁺ sites allow Cl⁻ ions to bind and react with photogenerated holes; however, no concrete spectroscopic evidence was provided in the study. Another study by Wu et al. combined DFT calculations and EPR trapping experiments to reveal that the activation of Cl⁻ on TiO₂-V_o occurs spontaneously (negative Gibbs free energy).¹⁰⁴ In summary, these findings highlight that oxygen vacancies (V_o) are highly beneficial for enhanced Cl⁻ adsorption and oxidation, leading to enhanced PEC performance in the Cl₂/HOCl formation (Table 3).^104,105

Table 3 Summary of the photoelectrochemical performance of various photoanodes in HOCl production

Photoanode	[HOCl] or [Cl2]	Photocurrent (mA cm^−2)	Potential (V)	FE (%)	Ref.
a Referenced to SCE.b Referenced to Ag/AgCl.c Referenced to RHE.d Two-electrode cell system.e The steady current was set to 2 mA on a 1.5 × 5 cm^2 electrode and potential (vs. SHE).f At 2.3 V vs. Ag/AgCl.
WO3	—	0.5–0.6	0.5–0.6^a	76	74
WO3	—	∼0.8	1.6^b	100	75
WO3	—	3–3.5	0.9^b	100	76
WO3	140 µmol	1.95	1.23^c	56	78
WO3	—	4.5	0.76^b	70	33
WO3	—	2.7	1.0^b	93	80
WO3	—	∼0.62	1.0^b	62	18
WO3	—	∼0.21	1.6^d	30	81
WO3	—	∼0.60	1.23^c	100	83
Mo:BiVO4	—	∼0.05	1.2^b	78	90
RhO2/Mo–BiVO4	∼1.25 µmol h^−1	2.16	1.0^c	—	91
BiVO4/WO3	60 µM (500 s)	∼3.0	0.22^d	81	9
	660 µM (2000 s)		0.50^d	∼80
CoOx/BiVO4/WO3	—	2	0.68^e	88	48
CoOx/BiVO4/WO3	12.5 µmol h^−1 cm^−2	2.5	1.2^b	∼60	94
CoOx/BiVO4/WO3	~160 μM	∼2.3	1^c	100	95
CoBi/BiVO4/WO3	39.2 mg L^−1^f	0.38	1.2^b	—	96
Co:BiVO	—	0.19	1.1^c	92	98
WO3/BiVO4	6 µmol	2.5	1.42^c	85	102
TiO2-NT	—	∼0.5	0.2^b	60.6	103
TiO2-NT/Ti	37.1 µmol h^−1 cm^−2	3.21	1.23^c	73.2	44
TiO2 (rutile)	—	8	1.0^a	—	57
TiO2-Vo	48.7 mg L^−1 h^−1	∼2.5	0.5^a	—	104

---

7. Scalability challenges

PC and PEC seawater splitting demonstrates the feasible production of Cl₂/HOCl. Both the systems have merits and challenges in efficiently producing Cl₂/HOCl. The advantages, limitations and challenges of both the systems are summarised in Table 4.

Table 4 Comparison of key parameters in upscaling between PC and PEC seawater splitting

System	Photocatalysis (PC)	Photoelectrochemistry (PEC)
a The recommended Cl2 concentration for water treatment is a maximum of 4 mg L^−1 for drinking water, as recommended by US EPA, and 0.2–0.5 mg L^−1 for residual disinfection of free chlorine, as recommended by WHO.
Efficiency	Produced Cl2/HOCl concentration has already reached the benchmark value for water treatment set by WHO.^a However, the AQY and SCC efficiencies are below 10% and 0.1%, respectively	Much higher Cl2/HOCl concentration obtained than the benchmark value set by WHO, giving higher SCC efficiencies than PC
		The moderate operating potential (<1.2 vs. RHE) for most photoanodes is an advantage for higher efficiency
Selectivity	High selectivity of Cl2/HOCl is usually obtained. In many works, minimal O2 evolution from water oxidation has been observed due to its sluggish kinetics	Faradaic efficiency is usually high (>90%) for most metal oxide photoanodes, and the use of applied voltage is rather moderate
Scalability	Construction of large-scale PC panels or planar reactors is straightforward (without electrical wiring and connectivity) and cost-effective	Complex configuration, including external electrical bias integration of PEC cells, makes up-scaling difficult
	Photocatalyst films usually have lower SCC efficiencies than photocatalyst nanopowders	Photoelectrode thin films often exhibit non-uniformity for large electrodes, reducing the current density and efficiencies
	The leaching of immobilised photocatalysts from the films might be a problem	High Cl2/HOCl concentrations obtained might be a merit for small-scale demonstration of water treatment using small portable devices
	Issues with oxygen flow input, mass transfer, and separation of oxidation (Cl2/HOCl) and reduction (H2) products	Product separation is readily available in two-compartment PEC cells
Cost	Construction of panels or planar reactors is relatively cost-effective	Complex configuration for large-scale PEC cells makes it less cost-effective than PC reactors
	Material cost is relatively high due to the use of noble metals (Pt, Au, and Rh) as a co-catalyst and electron sink	Material cost is low because many performing photoanodes are transition metal oxides
	The recovery of leached photocatalysts adds an extra cost	Electrical energy input is required to provide the applied potential

---

Industrial and commercial practical applications require the high-performance, cost-effective, and long-term stability of photocatalytic materials (photocatalysts and photoelectrodes) and innovative large-scale reactor designs. These factors constitute scalability challenges for the large-scale production of Cl₂/HOCl. The efficiencies of photocatalytic materials are usually expressed using apparent quantum yield (AQY) and/or solar-to-chemical conversion (SCC). Many excellent photocatalysts with high AQY are wide-band gap semiconductors, such as TiO₂, WO₃, and BiOCl, which restrict the efficient utilization of visible light in the solar spectrum. Therefore, a search for other promising photocatalysts or modifications of wide-band gap photocatalysts is highly desirable.

The choice of photocatalytic materials is essential for achieving a high production rate and yield of Cl₂/HOCl. Single-component photocatalysts are often restricted by low photon absorption, fast charge carrier recombination and slow charge transfer, which contribute to low SCC and AQY in the visible light range. A typical and convenient method to improve charge carrier separation and transfer is to load a noble metal (Pt, Au, and Rh) co-catalyst onto a photocatalyst. Metal co-catalysts serve as an electron sink and a catalytic site for reduction reactions. This, however, places an additional material cost on the system. Many works have attempted to limit the usage of such noble metal to below 1% wt. to achieve a compromise between high photocatalytic efficiency and material costs. Another promising strategy is to construct heterojunction photocatalysts, like Z-scheme or S-scheme, in photocatalysis to enhance carrier separation and transfer, as well as retain strong redox potentials for chloride oxidation.

The long-term stability of photocatalytic materials presents another challenge. The issues are corrosion in an acidic medium due to the presence of Cl₂/HOCl, and photocorrosion by charge carriers (electrons and holes) under light irradiation. Sayama's group demonstrated photocorrosion and dissolution of Pt co-catalysts, forming soluble PtCl₆²⁻ when reacting with Cl⁻ in seawater. One simple strategy is to create a core–shell Pt-MOX (MOX = metal oxide) co-catalyst where a MOX shell protects the Pt core from photocorrosion and dissolution (Fig. 11a). A few stable and robust metal oxides have been commercially employed in the chlor-alkali process. Notably, TiO₂, RuO₂, and IrO₂ are widely used as coatings on anodes owing to their excellent chemical resistance in a wide range of pH. Forming this oxide coating can suppress the photocorrosion and dissolution of the metal core. For thin film photoanodes, such metal oxides can be deposited as a thin overlayer to protect the photoactive semiconductor (Fig. 11b). The direct deposition of an overlayer can be easily performed using photodeposition or electrodeposition, which creates robust thin films. For instance, Xiao et al. constructed hollow, mesoporous TiO₂ containing Pt/Cu₂O that maintained high photoactivity in seawater for 30 h.¹⁰⁶

Fig. 11 Illustration of (a) oxide shell protecting the metal core and (b) overlayer protecting the photocatalyst from photocorrosion and dissolution.

Reactor design is the most critical aspect of large-scale practical applications. When designing reactors, considerations include the reactor geometry, amount of light passing through the solution and reaching the photocatalyst, mass transfer process, leaching of photocatalysts from films or electrodes, accumulation of sodium salt, and separation and collection of Cl₂/HOCl. Domen's group constructed a scalable panel reactor with 100 m² in size that consists of photocatalyst sheets (Fig. 12a).^107,108 The photocatalyst sheets were prepared by applying a spray method on frosted glass sheets. A similar design for photocatalytic seawater can be adopted, where seawater flows through the film panel and the gases evolved can be collected at the top. Because H₂ is the expected reduction product, there is a need for a dedicated separation of Cl₂ and H₂ gases. Fortunately, HOCl (resulting from the dissolution of Cl₂ in water) can be collected in a liquid phase, thereby easing the separation of H₂ and HOCl. Unlike pure water splitting, two further considerations must be taken into account, namely, the removal of accumulated sodium salt and oxygen flow input (where oxygen acts as an electron scavenger). The accumulated sodium salt on the film panel needs to be adsorbed, precipitated and filtered out for removal. Considering that the salinity of seawater is usually 3.5% (∼0.6 M) NaCl, which is relatively low compared to the one used in the industrial chlor-alkali process, precipitation of sodium salt is much less, and thus its removal is relatively easier. It should also be noted that the solution of aqueous NaCl is temperature dependent; therefore, the temperature of the seawater solution must be controlled to avoid salt precipitation during a photocatalytic reaction.

Fig. 12 (a) Scalable reactor design for a large-scale 100-m² panel reactor with 33 × 33 cm² films for photocatalytic water splitting. Reproduced from ref. 108. Copyright 2018, Cell Press. (b) Schematic diagram of large-scale panel experiments and (c) photograph of the 25 cm² photoreactor for the photoreforming of plastics, biomass and mixed waste. Reproduced from ref. 109. Copyright 2020, Wiley-VCH.

Another consideration is designing moderate-scale configurations for photocatalytic seawater splitting with a similar design to flow cells with a low-cost LED light source externally mounted to the cell. Uekert et al. designed a 25 cm² CN_x/Ni₂P panel for a real-world demonstration under low sunlight and seawater conditions (Fig. 12c and d).¹⁰⁹ The authors reported a loss of 50% efficiency in an upscaled (25 cm², 30 mL solution volume), custom-built flow reactor under real-world conditions. With this design, it is also possible to stack the cells into a larger configuration to boost the production rate and yield of Cl₂/HOCl. Considering that a tiny amount of Cl₂/HOCl (4 mg L⁻¹) is required for drinking water, this technology brings us a step closer toward real-world applications for on-site, on-demand production of Cl₂/HOCl for disinfection in remote areas.

Upscaling photochemical (PEC) reactors remains challenging because of several problems associated with PEC cells. However, PEC cells already have built-in compartments to collect H₂ and Cl₂/HOCl separately. The identified challenges for large-scale PEC cells include high cell efficiency (>10% STH), long life cycles and cost-effective fabrication process. Another issue is the incorporation of external bias to apply potential to the system. To overcome this issue, a research direction can be aimed at developing unassisted stand-alone PEC devices where one compartment carries out Cl⁻ oxidation using a photoanode and another carries out proton reduction using a photocathode. Large-sized photoelectrode fabrication presents another bottleneck that ensures homogeneous current density distribution. Dilger et al. successfully upscaled a PEC cell dimension from the typical 1 cm² to 40 cm² using LaTiO₂N coated with co-catalysts.¹¹⁰ The authors reported that the reduced activity and thus photocurrent density (by 54%) were attributed to the variation in the spatial distribution of potential and local photocurrent density. Overall, while fabricating homogeneous large-size photoelectrodes, controlling the optimised thickness of photoactive semiconductors and loading of co-catalysts, optimising cell geometry (including the distance of the light path), and inducing fast mass transfer dynamics remain challenging for upscaled PEC cells, notably when the electrode is larger than 10 cm by length, as remarked by Hankin et al.¹¹¹

Very recently, Rodríguez-Gutiérrez designed a 3D-printed reactor and assembled ten such reactors to form a 100 cm² module for moderate upscaling with stable photocurrents for over 120 hours (Fig. 13).¹¹² This modular PEC reactor design paves the way for scalable outdoor applications. To minimize ohmic losses when photoelectrodes are in contact with an electrolyte, and to obtain efficient product separation in individual compartments, Tsampas's group constructed a zero-gap PEC device in which an anion exchange membrane (AEM) was sandwiched directly between a photoanode and a cathode.¹¹³ The up-scaled device with an electrode area of 100 cm² preserves ∼98% of the photocurrent density exhibited by the 1 cm² photoanodes, demonstrating the practical feasibility and high performance of scalable PEC devices. More recently, a scalable PEC cell for H₂ and H₂O₂ production with a 144 cm² array panel has been reported.¹¹⁴ Precipitation of electrolyte (NaHCO₃) on the photoanode films can be removed by rinsing with deionized water to recover the initial photocurrent. A similar design and post-treatment of photoanode films can be adopted for PEC chloride oxidation from seawater splitting; sodium salts can be removed by rinsing with water to recover the photoanode films.

Fig. 13 (a) Illustration of the design of a modular PEC cell, (b) photograph of a real reactor with hematite photoanodes, and (c) outdoor demonstration for upscaled solar hydrogen production. Reproduced from ref. 112. Copyright 2025, the American Chemical Society.

8. Emerging photoactive semiconducting materials

Perovskite photocatalysts have garnered attention in water and seawater splitting. Metal halide perovskites (MHPs) adopt the crystal structure of ABX₃, where A and B are cations and X is a halide (Cl, Br, and I), with a well-known example of CsPbBr₃. The A site can also be occupied by organic cations, such as methylammonium and formamidinium in MAPbI₃ and FAPbI₃, respectively. MHPs are interesting and attractive because they exhibit high optical absorption coefficient, tunable optical properties via compositional engineering, defect-tolerant properties and large carrier diffusion lengths (∼100 nm–∼1 µm).¹¹⁵ Due to the toxicity of Pb, several other metal cations have been explored, such as CsSnI₃, CsGeI₃, Cs₃Bi₂I₉ and CsNaBiCl₆. For perovskite oxides, piezoelectric properties present an advantage in applying mechanical vibration to enhance photoactivity. Although the work on Cl₂ /HOCl production from seawater splitting using MHPs has been insignificant, the potential for Cl₂ /HOCl production is desirable because the inherent Cl⁻ ions participate in the chlorine evolution reaction (CER) through the Mars–van Krevelen mechanism.

Perovskite-type oxynitrides AB(N)₃ (A = La, Ca, Sr, and Ba; B = Ti, Nb, and Ta) have recently emerged as promising photocatalysts due to their narrow band gap, which can harness visible light for photoredox processes (Fig. 14). SrNbO₂N exhibits a narrow band gap (1.8–2.0 eV), a maximum current density of ∼20.5 mA cm⁻² and suitable band positions for seawater splitting. Trinh and Seo reported that porous SrNbO₂N/Nb photoanodes exhibited a photocurrent density of 2.6 mA cm⁻² at 1.36 V_RHE, and three-fold selectivity for CER than OER.¹¹⁶ The same group also reported that Co(OH)_x-loaded BaTaO₂N promoted CER and subsequently decomposed Cl₂ back to O₂ and Cl⁻, creating a self-sustaining steady-state Cl⁻ concentration in the bulk electrolyte.¹¹⁷ This work paves the way towards sustainable H₂ production at the cathode without replenishing the NaCl electrolyte.

Fig. 14 Band gap and redox potentials of several perovskite oxynitrides. Reproduced from ref. 118. Copyright 2024, the Royal Society of Chemistry.

Perovskite oxides are another class of emerging photocatalytic materials with unique features, such as piezoelectric and ferroelectric properties. Moreover, perovskite oxides display superior stability to metal halide perovskites and perovskite-type oxynitride.^119,120 La₂NiO₄ achieves high activity for H₂ and Cl₂ evolution in coupled piezo-photocatalytic seawater splitting without any sacrificial agent.¹²¹ Inducing mechanical vibration via ultrasonication improves the charge separation and transfer, which reduces the need for noble metal co-catalysts.¹²² A self-generated Na_0.5Bi_0.5TiO₃/Na_0.5Bi_4.5Ti₄O₁₅ heterojunction also exhibits a decent photoactivity in seawater splitting albeit the HOCl concentration was not quantified.¹²³

Porous coordination polymers, including metal–organic frameworks (MOFs) and covalent organic frameworks (COFs), are porous and crystalline materials with large surface areas, high porosity, diverse structures and topologies, and tunable electronic and optical properties through molecular engineering. These remarkable features render them promising in photocatalysis, electrocatalysis and photoelectrocatalysis. The catalytic activity and selectivity in MOFs are dictated by the unsaturated metal nodes and functional organic groups.¹²⁴ Moreover, the existence of charge transfer (linker-to-linker, metal-to-ligand, and ligand-to-metal) provides photoactive characteristics to MOFs.¹²⁵ Additionally, porous solid materials exhibiting different properties than the original MOFs can be prepared by pyrolysis of MOFs.

MOFs have been widely investigated for their promising applications in gas and liquid separation, membrane filtration and molecular sieving. The high-surface area and porosity offer a high capacity for ion adsorption, including chloride (Cl⁻). Although the direct use of MOFs for the production of Cl₂/HOCl via seawater splitting is still limited and at an early stage of development, electrocatalytic and photocatalytic H₂ and O₂ evolution in seawater splitting have been demonstrated.^126,127 Jothi et al. demonstrated that bimetallic CuV-MOF showed significant CER activity albeit the faradaic efficiency was not quantified.¹²⁸ High durability and stability of several MOFs under harsh and corrosive conditions of seawater offer potential for the design of robust co-catalysts or electrocatalysts to produce Cl₂/HOCl. For example, crystalline-amorphous ferrocene-based Ni-BDC-aNiOOH MOFs exhibit superior durability with ∼90% retention of current density for over 400 hours.¹²⁹

Bao et al. recently reported the selectivity of Ni-BDC/NM88B(Fe) MOF-on-MOF heterojunction in seawater oxidation.¹³⁰ Although the intermediate in CER generally has a lower Gibbs free energy and a lesser number of intermediates at both Fe and Ni sites, the lower adsorption free energy of OH⁻ governs the selectivity, thereby favouring OER over CER. The formation of the NiFeOOH protective layer defends Ni-BDC/NM88B(Fe) against dissolution by Cl₂, leading to an excellent stability of 200 hours. Ligand defect engineering has proven to be a viable strategy for improving durability in seawater splitting. The electronic structure of NH₂-MIL-88B(Fe) has been modified by introducing ligand vacancies using acetic acid, resulting in the suppression of Cl⁻ corrosion.¹³¹

Few studies have applied porous coordination networks to PC and PEC seawater splitting. The major focus in seawater splitting is the H₂ evolution in the presence of a sacrificial agent, notably a hole scavenger.¹³² For example. Li et al. employed Au-loaded thioether-functionalised COF in PC seawater splitting and observed no formation of Cl₂ because the presence of triethanolamine effectively scavenges photogenerated holes.¹³³ Attempts have been made to test PC seawater splitting using a variety of COFs without hole scavengers; however, little or no photoactivity was observed.¹³⁴ Crown ethers discontinuously conjugated polymers (DCPs) showed reduced photoactivity in a narrow range of NaCl concentrations (0–0.6 M) in synthetic seawater.¹³⁵ The finding implies that the presence of Na⁺ and Cl⁻ ions negatively impacts the photocatalytic performance. The crown ether moiety adsorbs ions, leaving only the proton to interact with the active sites. In contrast, 2,4,6-trihydroxy-1,3,5-benzenetricarbaldehyde (TPBD) COF shows enhanced photoactivity in seawater due to the strong polorisation effect induced by C=O groups, leading to better charge separation and transfer.¹³⁶ Similarly, thieno[3,4-c]pyrrole-4,6-dione (TPD) COFs exhibit higher photoactivity in seawater owing to their increased hydrophilicity and enhanced charge carrier separation and transfer.¹³⁷

The limited studies in photocatalytic seawater splitting particularly chloride oxidation using COFs originates from the low charge mobility and rapid charge carrier recombination that often requires a hole scavenger which precludes oxidation of chloride.¹³⁸ Additionally, certain organic functional groups in COFs readily adsorb ions, preventing Cl⁻ from reaching the actual active sites for CER. Future work should be directed to investigating and controlling the hydrophilicity/hydrophobicity, improving charge mobility and charge transfer, and creating selective active sites for Cl⁻ adsorption and oxidation.

9. PC and PEC applications of Cl₂ and HOCl

The advanced oxidation process (AOP) based on reactive chlorine species (RCS) provides a feasible strategy for the removal of organic contaminants and pollutants. HOCl undergoes photolysis under UV light irradiation to form highly reactive radical species, like Cl˙, ˙OH, and ˙OCl, which are strong oxidants. In fact, as discussed below, RCS exerts a much more powerful and selective oxidising ability than reactive oxygen species (ROS) in some cases, leading to more efficient degradation performance. As a result, in situ generated oxidants from HOCl have been utilised to drive oxidative reactions, and organic pollutant degradation and removal. A well-established chlorination process has been employed for over 100 years in drinking water treatment, swimming pool disinfection and medical disinfection/sterilization, underscoring a profound and ubiquitous use of Cl₂/HOCl in bacteria deactivation and pollutant degradation. It is therefore imperative to discuss the relevant applications of Cl₂/HOCl in various chemical processes, considering the impacts of PC and PEC production of Cl₂/HOCl.

(a) Bacteria deactivation and inhibition

Chlorination is a well-established and effective means of disinfecting drinking water, swimming pools, and wastewater treatment. Bacterial deactivation is utterly important for marine antifouling, wastewater treatment and drinking water disinfection. Photocatalytic antibacterial properties are usually tested using Gram-negative bacteria, such as Escherichia coli (E. coli) and Pseudoalteromonas, and Gram-positive bacteria, such as Staphylococcus aureus (S. aureus) and Bacillus sp. E. coli is a common Gram-negative bacterium found in contaminated water sources (lakes, rivers, and reservoirs) and food that can be pathogenic and cause poisoning, stomach pains and cramps, watery diarrhea, vomiting and fever. Thus, E. coli is commonly used as a model bacterium for bacterial deactivation.

Koo et al. showed the complete deactivation of E. coli in four hours over WO₃ photoanodes due to the in situ generation of HOCl as an active species.³⁰ In another work by Juodkazyte et al., the complete deactivation of E. coli and Bacillus sp. was demonstrated using a WO₃ photoanode and was ascribed to chlorine radicals (Cl˙ and Cl₂˙⁻).⁸¹ It is worth noting that ˙OH radicals are less effective in bacteria deactivation compared to RCS, highlighting the superiority of the latter. For example, Sun et al. disclosed that IrO₂/ZnWO₄ displayed a higher sterilisation efficiency against E. coli. owing to a higher production yield of RCS (HOCl, ClO⁻) than ZnWO₄, which generated mainly ˙OH.⁵¹ Moreover, the quenching experiment verified that scavenging of ˙OH led to a slight decrease in sterilisation efficiency, suggesting a minor role of ˙OH compared to RCS. More recently, carbon-doped BiOCl (C-BiOCl) displays impressive E. coli deactivation in 40 minutes (Fig. 15).⁷¹

Fig. 15 Inactivation of E. coli over C-BiOCl with two different RCS concentrations. Reproduced from ref. 71. Copyright 2024, Elsevier.

A quaternary Pt/Fe₃O₄@SiO₂@WO₃ photocatalyst shows good antibacterial effects on a broad spectrum of bacteria, including E. coli, Pseudoalteromonas, S. aureus, and Bacillus, with antibacterial rates of 90%, 83%, 86%, and 88%, respectively, in three hours.⁴⁹ When the same photocatalyst was applied for algae inhibition on Dicrateria inornata and Chlorella under visible light irradiation for 12 hours, the inhibition rates reached 93% and 95%, respectively. Lu et al. reported an impressive bacterial deactivation (>97%) in two hours and algae inactivation (>90%) in 8 hours using Pt/WO₃ hollow microspheres owing to high in situ generated free chlorine (1.24 mg L⁻¹).⁴⁶ More recently, a Z-scheme WO₃/CdS photocatalyst exhibits higher deactivation efficiencies (>94%) on E. coli, Pseudoalteromonas, S. aureus, and Bacillus in three hours.⁵³ The inhibition rate of Dicrateria inornata and Chlorella achieved >90% in 12 hours.

In conclusion, these works demonstrate the practical feasibility of using in situ generated RCS for bacteria deactivation and algae inhibition. A potential future direction is to design portable photoreactors for PC or PEC systems that are suitable for use in remote areas for on-site sanitization and water treatments for small water sources or storage facilities, like wells, ponds and tanks. A small amount of RCS from PC or PEC seawater splitting is beneficial for sanitizing surfaces in the kitchen and lounge in rural houses.

(b) Ammonia removal

Ammonia is a key fertilizer for plant growth and nutrition in agriculture because it supplies nitrogen to improve crop quality. Although it is crucial for agriculture, excess ammonia in soil has negative impacts on human health and the environment, behaving as a pollutant in the atmosphere that causes acid rains, depositing in the soil and being dissolved in water (lakes, rivers), and changing species composition through soil acidification and direct toxic damage. Indirect discharge into wastewater containing ammonia nitrogen (NH₄⁺–N) remains a challenging issue in agriculture and thus becomes a problem of environmental concern. Chlorine disinfectants have been used to remove NH₄⁺–N by converting it to N₂. Utilizing photocatalytic Cl₂/HOCl production may eliminate the need for external chlorine sources. For example, Xiao et al. showed that RCS (Cl₂/HOCl) plays a critical role in NH₄⁺–N degradation on TiO₂ nanotube photoanodes, where the degradation rates increase with NaCl concentration across a wide pH range.¹³⁹ Zhu's group demonstrated the practical feasibility of NH₄⁺–N removal using landfill leachate and real wastewater as examples.^36,71

The overall reaction for NH₄⁺–N degradation proceeds according to eqn (30). The mechanism of NH₄⁺–N degradation over a BiVO₄/WO₃ photoanode has been proposed to proceed via chloramine intermediates (NH₂Cl and NHCl₂).¹⁴⁰ COR produces Cl˙, which abstracts H from NH₄⁺ to form NH₂˙; this radical reacts with HOCl to yield NH₂Cl (eqn (31)–(33)). A further reaction leads to an NHCl₂ and eventually N₂ evolution as described below (eqn (34)–(36)). NH₄⁺–N can also be oxidized to NO₂⁻–N and NO₃⁻–N species (eqn (37) and (38)), thereby presenting competitive steps for selective N₂ formation in NH₄⁺–N degradation.¹⁴¹ The formation of NO₃⁻ is proposed to include reactive oxygen species (ROS), such as ˙OH, O₂˙⁻, and H₂O₂.¹⁴² Therefore, it is expected that COR in seawater solution would favour the selective formation of N₂ in the degradation of NH₄⁺–N.

2NH₄⁺ + 3HOCl → N₂ + 3HCl + 3H₂O (30)

Cl⁻ + h⁺ → Cl˙ (31)

NH₄⁺ + Cl˙ → NH₂˙ + 2H⁺ + Cl⁻ (32)

NH₂˙ + HOCl → NH₂Cl + OH˙ (33)

NH₂Cl + Cl˙ → NHCl˙ + H⁺ + Cl⁻ (34)

NHCl˙ + HOCl → NHCl₂ + OH˙ (35)

NHCl₂ + NH₂Cl → N₂ + 3H⁺ + 3Cl⁻ (36)

NH₂˙ + OH˙ → NH₂OH (37)

NH₂OH + OH˙ → NO₂⁻→NO₃⁻ (38)

2NH₄⁺ + 2ClO˙ → N₂ + 4H⁺ + 2Cl⁻ + 2H₂O (39)

An early work by Ji et al. using a WO₃ photoanode disclosed that the complete degradation of NH₄⁺–N to a major N₂ and minor NO₃⁻ occurred in 90 min and followed the pseudo first-order kinetics.¹⁴³ However, the degradation of NH₄⁺–N over BiOCl was shown to follow the zeroth-order kinetics with almost complete degradation and 97% selectivity toward N₂ over BiOCl in 80 minutes.⁶⁹ Zhu's group reported that BiOCl with dominant {001} facets exhibited the degradation efficiency of 91.9% with stoichiometric Cl⁻ (in NH₄Cl) ions and near unity with extra Cl⁻ (from NaCl) in the reaction solution within 150 minutes.³⁶ Without Cl⁻ in the reaction solution, only 25.9% degradation was observed, highlighting the crucial role of RCS (Cl₂, HOCl, and Cl˙) for promoting NH₄⁺–N degradation. The higher degradation efficiency with increasing Cl⁻ content is attributed to the higher amount of RCS formed. Using a landfill leachate from a landfill in Zhongshan City as a real wastewater for treatment, the authors showed a degradation efficiency of 85.4% in 30 hours and 95.6% in 40 hours of treatment over BiOCl and C-BiOCl, respectively.^36,71 Nonetheless, Zhu and co-workers realised that the degradation efficiency under the Xe lamp is far inferior to that of the UV lamp.⁷¹ The cause might be attributed to the less efficient photolysis of HOCl that creates radicals under the Xe lamp compared to that of the UV lamp.

Yan et al. disclosed that OCl˙ radicals are responsible for achieving the complete degradation of NH₄⁺-N in 120 min and boosting 100% selectivity towards N₂ in acidic NaCl solution (pH ≤ 6), as shown in Fig. 16a and b.¹⁴⁴ By scavenging and ESR trapping experiments, the authors disclosed that OCl˙ radicals resulted from photolysis of HOCl under UV irradiation; the mechanism NH₄⁺–N degradation is illustrated in Fig. 16c. Importantly, other works have also attributed the active species for NH₄⁺–N degradation to ClO˙ radicals.^141,145 It is worth noting that ˙OH radicals are relatively inert to NH₄⁺–N and thus are less effective in the degradation process. Similarly, the imperative role of RCS (Cl₂, HOCl, Cl˙, and ClO˙) in promoting the degradation of NH₄⁺-N was reinforced by Zhang et al. using a CoB_i/BiVO₄/WO₃ photoanode, whereas ˙OH radicals play a negligible role.⁹⁶ Notably, it has been revealed that ˙OH reacts with HOCl to form ˙OCl (and H₂O), showing an indirect participation of ˙OH to form OCl˙ active species.¹⁰⁴ Although OCl˙ radicals are regarded as the primary active species in NH₄⁺–N degradation,¹⁴⁶ HOCl plays a role in completing the cyclic chlorine species between Cl⁻ and Cl˙ (or Cl₂), indicating complex reaction pathways (refer to eqn (29)–(37), and Scheme 1).

Fig. 16 (a) Time-course degradation of NH₄⁺–N, (b) conversion of NH₄⁺–N to different oxidised species (N₂, NO₂⁻–N, and NO₃⁻–N), and (c) a proposed mechanism of NH₄⁺–N degradation in acidic water over BiOCl. Reproduced from ref. 144. Copyright 2021, Elsevier.

Scheme 1 NH₄⁺–N degradation pathways to N2 through chloramine intermediates and radicals. Reproduced from ref. 96. Copyright 2023, Elsevier.

In summary, these reports show that RCS resulting from COR plays a predominant and significant role in degrading NH₄⁺–N through chloramine intermediates and radicals, which are much more efficient than reactive oxygen species (ROS). Despite a lower oxidation potential of Cl˙ (2.47 V vs. NHE) than ˙OH (2.80 V vs. NHE), Cl˙ is more selective and reacts faster with electron-rich moieties, such as NH₄⁺.^143,147 The high selectivity towards N₂ formation renders PC and PEC degradation of NH₄⁺–N using in situ generated RCS appealing, green and sustainable.

(c) Conversion of alkenes and alcohols

Due to the strong oxidising ability of HOCl in bacterial deactivation and pollutant degradation in wastewater, several works have extended its utility in the oxidation of alkenes and alcohols using in situ generated HOCl. HOCl behaves as a redox mediator and strong oxidant to oxidize organic compounds with a lower applied potential than the direct oxidation of such organic compounds by adding molecular O₂, H₂O₂ or t-BuOOH. For example, Yang et al. reported almost complete conversion and selectivity in the epoxidation of water-soluble alkenes on a Co₂(OH)₃Cl/BiVO₄ photoanode (Fig. 17a and b).¹³ The mechanism of such epoxidation involves the formation of a secondary alcohol and terminal alkyl chloride as intermediates resulting from the reaction between in situ generated HOCl and alkenes (Fig. 17c). Such high photoactivity in epoxidation can potentially replace the use of meta-chloroperbenzoic acid (m-CPBA), which possesses a high risk of explosion and limits its large-scale utilization in alkene epoxidation.¹⁴⁸

Fig. 17 (a) Alkenes used in the epoxidation reaction, (b) conversion and selectivity in epoxidation over CoOCl/BVO, and (c) proposed mechanism of epoxidation via in situ generated HOCl. Reproduced from ref. 34. Copyright 2024, Wiley-VCH. (d) Proposed mechanism of the selective oxidation of methane over BiOCl-OV. Reproduced from ref. 149. Copyright 2023, Elsevier.

Oxidation of ethanol was demonstrated by in situ PEC HOCl generation on H_xWO₃, without the addition of external oxidants. Ethanol oxidation to acetaldehyde and acetic acid was mediated by HOCl serving as a strong oxidant – most likely through ˙OH and Cl˙ from the photolysis of HOCl – rather than direct oxidation on the H_xWO₃ electrode as confirmed by linear sweep voltammetry (LSV).¹⁴⁹ The EtOH conversion achieves 62% in 8 hours with 50% and 40% selectivity to acetaldehyde and acetic acid, respectively. The proposed reaction pathways are described by eqn (40)–(42). Such oxidation can be extended to secondary alcohols, like 2-propanol, with 92% selectivity to acetone.

Cl⁻ + H₂O → HOCl + H⁺ + 2e⁻ (40)

CH₃CH₂OH + HOCl → CH₃CHO + H⁺ + Cl⁻ + H₂O (41)

The selective photocatalytic oxidation of methane to methanol is highly desirable to simultaneously mitigate greenhouse gases and to produce fuels, but it is very challenging due to a large energy barrier of C–H activation and bond energy (439 kJ mol⁻¹).¹⁵⁰ Fortunately, active chlorine species can be used to abstract H from CH₄, which can then be oxidised to CH₃OH with 80% selectivity in the presence of H₂O₂. H₂O₂ undergoes decomposition and supplies ˙OH and O₂˙⁻ radicals. Cl⁻ oxidation to a ˙Cl radical offers an advantage in carrying out CH₄ oxidation in a seawater solution. Wang et al. showed that both Cl˙ and ˙OH are crucial for the conversion of CH₄ to CH₃OH over BiOCl with oxygen vacancies (BiOCl–O_V), where oxygen vacancies (V_o) serve as an active site for Cl⁻ adsorption (Fig. 17d).¹⁴⁹ Cl˙ is required to form a ˙CH₃ intermediate, which is oxidised by ˙OH to yield CH₃OH. Interestingly, pristine BiOCl (without oxygen vacancies) produces Cl₂, which is responsible for halogenated compounds (CH₂Cl₂ and CHCl₃), while BiOCl-V_o yields Cl˙, which is responsible for H abstraction from CH₄. In another work, the same group also reported the formation of diethyl ether (C₄H₁₀O) as the majority and other chlorinated products such as CH₂Cl₂, CHCl₃ and C₂H₄Cl₂O in synthetic seawater.⁵⁶

Overall, in situ generated RCS are key for the oxidation and conversion of organic compounds. Importantly, RCS shows high selectivity towards particular products, which renders COR advantageous and promising for selective oxidation.

(d) Oxidation and chlorination of amines and sulfides

A strong oxidising capability of reactive chlorine species (Cl₂ and HOCl) renders it suitable for various organic transformations, including the oxidation of amines and sulfides. Notably, HOCl can initiate oxidative fragmentation without incorporating Cl atoms into the final products. For example, Hull et al. reported the mechanism of triethylenediamine oxidation by HOCl to produce piperazine, which can be further oxidised to ethylenediamine and formaldehyde.¹⁵¹ The first step usually involves hydrogen abstraction and the inclusion of a chlorine atom in amines. Aldehydes seem to be a common product of amine oxidation by HOCl, as also observed in the reaction of aminoalcohols with HOCl. Interestingly, kinetic studies have shown that tertiary amines react much faster with HOCl than primary and secondary amines.¹⁵²

Reactions between organic amines and RCS often yield the corresponding chlorinated amine (N–Cl) compounds, which are significant in biochemistry and environmental chemistry. Organic amines, like amino acids, are often present in wastewater and sewage sludge, which are treated with HOCl. Reactions between amino acids and peptides with HOCl proceed readily. The importance of the reaction between RCS and amino acids in enzymes in bacteria is paramount to understanding bacterial deactivation and inactivation in wastewater treatment. A free amino group in amino acids reacts with HOCl to yield unstable chloroamine intermediates (RNHCl/RNH₂⁺Cl), which subsequently decompose to NH₃, CO₂, nitriles, and aldehydes.^153–156 Notably, N-chloro derivatives form rapidly from the reaction of amino acids and HOCl and decompose slowly in multisteps (Scheme 2). Although the products of amine oxidation by HOCl are well known and well characterised, the mechanism remains elusive and less understood. Hawkins and Davies verified the formation of radical intermediates from the decomposition of chloroamines and chloroamides by electron paramagnetic resonance (EPR).¹⁵⁷ However, several dipeptides were found to be resistant to HOCl at room temperature.

Scheme 2 Mechanistic pathways of the oxidation of amino acids by HOCl with the liberation of CO₂ and formation of aldehyde and nitrile. Reproduced from ref. 158. Copyright 2017, the American Chemical Society.

N-chlorination reactions render it suitable for the application of in situ generated RCS by photocatalytic and photoelectrochemical means. Ding et al. demonstrated enantioselective chlorofunctionalization of N-substituted amides using in situ generated HOCl.¹⁵⁹ The role of HOCl is identified as a proton donor and a hypochlorite source. It is therefore expected that in situ generated RCS from seawater splitting facilitates organic transformations of amine compounds.

Hydrogen sulfide (H₂S) is a weak acid and toxic gas, which is produced as an emission gas from chemical, oil and gas industries. Chemical scrubbing is a typical method for removing H₂S emissions using a gas–liquid contactor containing an oxidant, such as HOCl or H₂O₂.¹⁶⁰ The oxidation of H₂S is a fast reaction, follows a complex mechanism, and is strongly pH dependent. The nucleophilic characteristics of HS⁻ and S²⁻ (sulfide) make it reactive to electrophilic HOCl. Sulfide in H₂S is oxidised by HOCl to yield elemental sulfur (S) and chloride (Cl⁻).¹⁶¹ Reacting emitted H₂S with in situ generated HOCl may pave the way toward efficient H₂S removal with low capital and operating costs. Moreover, the reactivity between H₂S and HOCl helps us to understand the role of HOCl in neurodegenerative diseases associated with the H₂S concentration in the brain.¹⁶²

The oxidation products of organic sulfur compounds depend on the type of sulfur. The reaction rate for a few sulfur compounds was established in the following order: thiol > thioether > sulfoxide. Sulfides are oxidised to sulfoxides, and overoxidation in excess of chlorine yields sulfones, while thiols form disulfides and gradually transform to sulfonic acids.¹⁶³ Meanwhile, thioethers are oxidised to sulfoxides.¹⁶⁴ Sulfur-containing compounds, such as thiourea, thiosulfate and thioglycolate, are reactive towards oxidation by HOCl. For instance, Stanbury's group reported that the reaction rate of these sulfur compounds with HOCl is much more rapid than using H₂O₂ and [Pt(CN)₄Cl₂]²⁻ oxidants, highlighting the former's superior oxidative capability owing to the highly electrophilic HOCl.¹⁶⁵ Moreover, the solubility of HOCl in polar organic solvents, such as CH₂Cl₂, renders it advantageous in various organic reactions because no phase transfer catalysts are required for the oxidation process.

(e) Photocatalytic CO₂ reduction

Photocatalytic CO₂ reduction (PCR) represents a sustainable carbon cycling system and a clean route to mitigate the CO₂ level in the atmosphere and produce high value-added chemicals and fuels, like CO, CH₄, and CH₃OH. However, research in PCR generally employs water as an electron donor; thus, the non-trivial role of other photooxidation half-reactions is often overlooked. It has been shown that Cl⁻ oxidation enhances activity and selectivity toward specific products in PCR. Coupling CO₂ reduction to CO and Cl⁻ oxidation to Cl₂ (or HOCl) is kinetically favoured because both half-reactions involve a stoichiometric amount of charge carriers, i.e. two electrons and two holes. Layered-double hydroxides (LDHs) with remarkable adsorption capacities, diverse compositions and ions, and tunable optical and photoelectrochemical properties are widely used for CO₂ capture and gaining attention for photocatalytic CO₂ reduction.¹⁶⁶ Ni–Al LDHs exhibit a selective formation of CO (86%) from photoreduction of CO₂ with almost twice the CO₂ yield in aqueous NaCl (0.1 M) compared to pure water; the remaining 14% is H₂ evolution, and no other products, such as CH₄, CH₃OH and HCOOH, are detected in the reaction.¹⁹ However, the mechanism and the underlying reason the addition of Cl⁻ ions favours higher activity and selectivity are not addressed in the work, except that Cl⁻ acts as a hole scavenger to form HOCl and thus remains elusive.

Recently, surface Lewis basicity has been introduced by doping the {001} facets in BiOCl with nitrogen (N-BiOCl) to enhance CO₂ adsorption and activation for PCR in natural seawater (The Yellow Sea, China, was used as an example).⁷² The near stochiometric production rates of HOCl and CO under visible light are 83.2 and 92.8 µmol g⁻¹ h⁻¹, respectively, with product selectivity above 90% for both HOCl and CO (Fig. 18a and b). The reaction pathway attributes CO₂ reduction on the {001} facets, where photoelectrons are accumulated, and HOCl formation to be most likely on the lateral {110}/{010} facets via self-oxidation of lattice Cl⁻ ions. In another work, Wang et al. observed that the CO₃²⁻ intermediate in PCR may participate in the BiOCl interlayers, causing distortion and exchange of Cl⁻ ions in BiOCl as evidenced by XRD patterns, Raman and XPS measurements (Fig. 18c–f).⁷³ However, no reduction products (e.g. CO, CH₄, and CH₃OH) were quantified in the study.

Fig. 18 Photocatalytic formation of (a) CO and (b) HOCl over N-BiOCl. Reproduced from ref. 72. Copyright 2023, Wiley-VCH. Characterizations of the BiOCl photocatalysts before and after reactions: (c) XRD patterns, (d) Raman spectra, (e) C 1s XP spectra, and (f) Bi 4f XP spectra. Reproduced from ref. 73. Copyright 2023, Wiley-VCH.

Notably, the findings from photocatalytic COR improving the photoactivity and selectivity in PCR have led researchers to design novel Cl-modified photocatalysts through surface modification, grafting or doping with Cl⁻ ions.^167–169 Although these works are not related to PC or PEC seawater splitting or Cl₂/HOCl formation, they highlight how the progress in seawater splitting has impacted other photocatalytic reactions and photocatalyst designs.

10. Challenges and future direction

The competing reactions between oxygen evolution reaction (OER) and chlorine evolution reaction (CER) determine the selectivity in photo(electro)catalytic seawater splitting. A considerable number of factors govern the selectivity of CER, including the Cl⁻ concentration, pH of the electrolyte, electrocatalyst, and applied potential. In acidic and near neutral media, CER dominates over OER, while OER is preferred at high pH. Thus, acidification is often performed to enhance CER selectivity. In addition, the free energy difference in the –OCl and –OOH intermediates is used as the selectivity descriptor to predict the selectivity of CER and OER. For example, the stabilisation energy of OCl_ad is used as an activity descriptor to evaluate the effectiveness of CER electrocatalysts and is further employed to explain the superior activity of RuO₂ than IrO₂.¹⁷⁰ DFT calculations by Hansen et al. disclosed that Cl⁻ is adsorbed on the O-covered rutile surfaces spontaneously forming an –OCl intermediate.⁴³ Moreover, CER only involves a single intermediate compared to OER that proceeds via three intermediates, giving an advantage to the former reaction. Similarly, ab initio density functional theory (DFT) was performed using different RuO₂ crystal facets to investigate the overpotential of CER and OER.¹⁷¹ RuO₂(101) was found to promote CER, and RuO₂(001) preferentially catalysed OER.

Zeradjanin et al. proposed a dynamic catalytic descriptor based on vibrational frequencies of the crystal lattice from in situ Raman spectroscopy and electrochemical methods.¹⁷² The Raman shifts of metal–oxide bonds form a volcano plot, where the most active electrocatalysts with low overpotential for CER are RuO₂ and IrO₂, located close to the Cl–O bond (725 cm⁻¹) found in HOCl (Fig. 19a). The volcano plot rationalises the established knowledge that RuO₂ and IrO₂ are the best performing electrocatalysts in dimensionally stable anodes (DSAs) for the chlor-alkali process. The authors also concluded that CER is a resonance phenomenon in which the catalyst surface, intermediates and products exhibit vibrational bonds with a similar frequency. It is therefore possible to predict and screen the best performing catalysts based on Raman spectroscopy and electrochemical studies where the vibrational modes of the metal–oxide (M–O) bond are close to those of the Cl–O bond. Notably, the Raman frequency and the length of the M–O bond are inversely proportional, which uses X-ray absorption spectroscopy (EXAFS/XANES) in determining the bond lengths useful for predicting suitable co-catalysts for CER.¹⁷³

Fig. 19 (a) Overpotential of CER as a function of Raman shift for several metal oxides and (b) resistivity of several metal oxides. Reproduced from ref. 40. Copyright 2023, Wiley-VCH.

Another key parameter for efficient and robust electrocatalysts for CER is high conductivity to survive anodic conditions. The two best-performing and most employed electrocatalysts in the chlor-alkali process, RuO₂ and IrO₂, are metallic oxides (Fig. 19b). Rutile RuO₂ has partially filled 4d orbitals and therefore exhibits metallic conductivity.¹⁷⁴ For IrO₂, a large 5d orbitals overlap which is comparable to the Ir–O bond distance renders it metallic oxide.¹⁷⁵ Furthermore, low CER overpotential was observed for perovskite oxides with metallic conductivity, such as La_0.5Sr_0.5CoO₃ and LaNiO₃. It should be noted that mixed metal oxides (MMO) usually contain a TiO₂ semiconductor to improve the stability of DSAs while contributing little to their activity and selectivity. Metallic conductivity serves as a useful guide for exploring and designing suitable co-catalysts for photoanodes in CER.

A reasonable direction for exploring novel photocatalysts and photoanodes would involve the utilisation of in situ spectroscopy and electrochemical measurements to probe the vibrational frequencies and correlate them with the corresponding CER overpotential. This is possible for coordination polymers, such as metal–organic frameworks (MOFs) and covalent organic frameworks (COFs). The presence of metal nodes in MOFs and organic functional groups in COFs opens a new avenue for establishing similar overpotential-based volcano plots. Moreover, in situ Raman and infrared spectroscopy have proved advantageous in identifying intermediates and their binding with specific metal nodes or functional groups. Currently, the lack of such data restricts the prediction and knowledge-based design of effective novel photocatalysts for CER.

Computational studies provide a means of describing phenomena on the surface of photocatalysts at the molecular level, which can complement and help understand experimental studies. DFT with thermodynamic considerations was utilised to establish a correct selectivity descriptor for CER. Exner unravelled that the formation free-energy of oxygen with respect to hydroxide, ΔG_O–OH, provides a correct description of the activity (or selectivity) trend for CER than the free-energy of binding energy of oxygen (ΔG_O), which is typically used in the oxidation activity of metal oxides.¹⁷⁶ Importantly, the ΔG_O–OH descriptor has reproduced the activity trends of CER and OER for RuO₂(110) and IrO₂(110) accurately. Because OER and CER compete in an anodic process, both reactions must be considered to reproduce activity trends. The ΔG_O-OH descriptor is capable of describing both OER and CER activity trends correctly, whereas the ΔG_O cannot reproduce OER activity correctly. A prime advantage of using the ΔG_O–OH descriptor is that the values can be computed by ab initio calculations and verified by cyclic voltammetry.¹⁷⁷

Besides a direct Cl⁻ adsorption to an oxygen atom in rutile metal oxides, Cl⁻ can also bind to two oxygen atoms, forming ClO₂⁻ in CER.⁴³ Exner included this possibility of ClO₂⁻ in his unifying material-screening framework, which includes overpotential, kinetics and electrochemical-step symmetry index, to comprehend the selectivity issues in CER.¹⁷⁸ Based on this thermodynamic consideration, the formation of –Cl intermediate (i.e. O^cCl, where O^c is a coordinatively unsaturated oxygen atom in metal oxides) should be promoted over –OCl adsorbate (i.e. O^c₂Cl) to enhance the Cl₂/HOCl selective formation. The implication of this finding is to search for and design novel photocatalysts and photoelectrodes that can stabilise a Cl⁻ intermediate for enhanced Cl₂ selectivity.

In many scenarios, the active sites that bind to the Cl* intermediate also bind to the OH* intermediate, i.e. they both share a common active. Hence, the scaling relationship correlates with CER and OER in the chlor-alkali process. To break this scaling relationship, active sites with larger adsorption energy for OH* must be created to promote selectivity towards CER.¹⁷⁹ Doping RuO₂ that results in an increase in the free energy of the OH* formation leads to an improved selectivity toward CER. For example, DFT calculations reveal that replacing the outermost Ru atoms with Ti atoms through the growth of a TiO₂ monolayer on RuO₂ enhances the CER selectivity by several orders of magnitude.¹⁸⁰ Experimental work has also confirmed that doping is an effective strategy for enhancing CER selectivity.^39,181,182

Although most computational studies on CER are directed at rutile-based metal oxides, there is limited work on emerging materials, such as covalent-organic frameworks (COFs) and metal–organic frameworks (MOFs). Future direction should be dedicated to computing the free-energy of binding energy of Cl⁻ with N, S, and O atoms in organic functional groups and establishing the corresponding overpotential-based volcano plots because the findings will provide a knowledge-based guideline for rational design of efficient COFs and MOFs for CER. Furthermore, the findings from computational work can be validated by experimental techniques, such as infrared, Raman and nuclear magnetic resonance spectroscopy.

Photocatalytic Cl₂/HOCl production from seawater splitting is still at an early stage of development. In what historically started as a hole scavenger in photocatalytic water splitting using seawater or aqueous seawater solution, the oxidation of Cl⁻ to Cl₂/HOCl has emerged as a new direction in photocatalysis research for the past decade. Early works attempted to enhance the selectivity of oxidation half-reaction towards water oxidation, neglecting the potential of Cl⁻ oxidation to RCS (Cl₂/HOCl). The challenging issues in overall water splitting, low economic value of molecular O₂ and scarcity of pure/fresh water have urged researchers to assess other oxidation products that are kinetically favoured and have a high value-added.⁶ More importantly, PC and PEC seawater splitting have the potential to replace the energy-intensive conventional, industrial chlor-alkaline process in Cl₂/HOCl production.

PC and PEC Cl₂/HOCl production coupled with H₂ evolution at the anode is a novel strategy for simultaneously generating clean energy and commodity chemicals from widely abundant seawater or saltwater. In natural seawater, dissolved CO₂ could be an advantage in consuming photoexcited electrons, thereby suppressing the recombination of charge carriers and avoiding the continuous supply of O₂ as an electron scavenger. However, dissolved CO₂ forms carbonic acid (H₂CO₃) and HCO₃⁻, in which the carbonate anion (CO₃²⁻) is a good hole scavenger, as reported by Hiroaki Tada.¹⁸³ Therefore, there is a competing Cl⁻ and CO₃²⁻ oxidation in PC or PEC systems. Further work should explore PC or PEC Cl₂/HOCl oxidation in natural seawater/saltwater to investigate the effects of other ions and factors affecting the selectivity in Cl⁻ oxidation. Importantly, such work using natural and/or synthetic seawater will bring us closer toward a practical application.

Given that HOCl easily decomposes under exposure to UV light, the use of wide band gap photocatalysts should be minimized to achieve a high production rate and yield of HOCl. Nonetheless, good photocatalysts for COR are still dominated by wide band gap photocatalysts, like TiO₂ and BiOCl (band gap > 3.0 eV), which render visible-light-driven photocatalysts less explored. Many earth-abundant, visible light photocatalysts have appropriate valence bands (>1.40 V vs. RHE), such as g-C₃N₄, CdS, TaON, MoS₂ and Fe₂O₃, or other polymeric photocatalysts that remain relatively unexplored for COR and thus open a window of opportunity for catalyst development. Furthermore, photocatalysts for Cl⁻ oxidation should consist of few (precious) metals because the acidic medium, resulting from the formation of HOCl, tends to dissolve metals under photoirradiation. Sayama's group observed a dissolution of Pt co-catalyst, forming soluble PtCl₆²⁻ through a reaction with evolved Cl₂, during photoirradiation in seawater splitting.⁵⁸ The same group also reported a dissolution of Mn co-catalysts under photocatalytic conditions.

Except for limited photocatalytic materials, such as BiOCl and BiVO₄, the facet-dependent reactivity and selectivity of Cl/HOCl formation remain underexplored. Different crystal facets have different surface energetics and accumulate different charge carriers caused by the internal electric field (IEF).¹⁸⁴ However, the selectivity of oxidation-favoured facets is complicated by competitive water and chloride oxidation reactions. Theoretical simulations of specific facets can reveal the relative activation energy or overpotential for Cl⁻ oxidation. Moreover, the intermediates in COR have not yet been fully clarified, thus presenting an opportunity for further theoretical work.

Currently, despite numerous investigations on PC and PEC of Cl₂/HOCl formation using various photocatalysts and co-catalysts, the knowledge about the active sites remains elusive. Except for a few photocatalytic materials, like BiOCl and AgCl as well as Co₂(OH)₃Cl co-catalyst, where participation of the lattice Cl⁻ in COR via the Mars–van Krevelen mechanism is verified, the active sites in many other catalytic materials are unknown. Moreover, the postulate of the existence of Co–Cl interaction when using a CoO_x co-catalyst and/or overlayer has not been successfully verified by X-ray photoelectron spectroscopy.⁹⁵ CoO_x is a well-known electrocatalyst in oxygen evolution reaction (OER);^185–187 therefore, understanding the nature of the catalytically active sites and interaction with Cl⁻ is of paramount importance to tune the selectivity between OER and COR. Additionally, it has been reported that Cl⁻ can stabilise oxygen vacancies in TiO₂ and BiOCl, thereby promoting the formation of RCS or Cl˙ radicals for photocatalytic oxidation and degradation.^149,188 It has also been suggested that Cl⁻ interacts favourably with the TiO₂ surface although this interaction has not been verified by physical or electrochemical characterizations.¹³⁹ Future work should disclose the interaction and/or adsorption of Cl⁻ with the surface of photocatalysts. Currently, such limitations impede the knowledge-driven design and optimization of selective, active, and stable photocatalysts for COR.

We highlight that PC or PEC seawater splitting represents another type of artificial solar photosynthesis where H₂ evolution at the cathode can be coupled with Cl₂/HOCl formation at the photoanode.^33,44,64 More importantly, the separation of the evolution of H₂ and Cl₂ in PEC systems enables a direct collection of both gases for further utilisation or storage (Fig. 20). Such a strategy potentially circumvents the low-economic value of O₂ and sluggish kinetics of O₂ evolution in water-splitting. Therefore, PC and PEC seawater splitting present a feasible green alternative and economical method for renewable energy and CO₂ emission mitigation compared to the traditional chlor-alkaline process.

Fig. 20 Schematic of the simultaneous Cl₂ and H₂ evolution at a TiO₂ photoanode and a Pt cathode, respectively. Reproduced from ref. 44. Copyright 2023, Elsevier.

Finally, we offer a perspective on coupled redox reactions involving a COR half-reaction. The other half-reduction reactions are not limited to H₂ evolution or CO₂ reduction, as discussed above. Very recently, Chen and Zhang constructed In₂S₃/MnIn₂S₄-based photoanode and photocathode for simultaneous production of HOCl and H₂O₂, respectively, which highlighted a promising approach for the production of fine chemicals via PEC.¹⁸⁹ N₂ reduction to NH₃ has been another intense area of photocatalysis research for producing energy carriers. Shiraishi et al. demonstrated photocatalytic N₂ fixation to produce NH₃ in a seawater solution over oxygen-deficient BiOCl–O_v with solar-to chemical conversion of 0.05% HOCl.⁶⁵ Moreover, halogenation of alkanes into organic halides – important industrial and medical chemicals – has been reported using a V_o-rich TiO₂ photoanode.¹⁸⁸ These findings open an avenue for broad, significant impacts of Cl₂/HOCl production, photodegradation of pollutants, production of high value-added chemicals, and artificial photosynthesis involving the production of clean energy. Progress in COR via seawater splitting is still in its infancy and offers much room for development in the field.

More importantly, for remote areas that lack water treatment facilities, PC and PEC production of Cl₂/HOCl using natural seawater/saltwater is a sustainable method for on-site and rapid production of Cl₂/HOCl for treating drinking water and sanitizing purposes. It is undoubtedly that PC and PEC seawater splitting to produce Cl₂/HOCl addresses at least four SDGs’ goals: Good Health and Well-being (#3), Clean Water and Sanitization (#6), Affordable and Clean Energy (#7), and Climate Action (#13).

Conflicts of interest

There are no conflicts to declare.

Abbreviation

AQY	Apparent quantum yield
COR	Chloride (Cl^−) oxidation reaction
DFT	Density-functional theory
DPD	N,N-diethyl-p-phenylenediamine
DSA	Dimensionally stable anode
EIS	Electrochemical impedance spectroscopy
ETL	Electron transport layer
FE	Faradaic efficiency
HTL	Hole transport layer
IEF	Internal electric field
IET	Interfacial electron transfer
IPCE	Incident photon conversion efficiency
LDH	Layered double hydroxide
LSPR	Localised surface plasmon resonance
LSV	Linear sweep voltammetry
MSI	Metal-support interaction
OER	Oxygen evolution reaction
PL	Photoluminescence
P25	TiO2 with anatase (70%–80%) and rutile (20%–30%) phases
PCR	Photocatalytic CO2 reduction
PWS	Photocatalytic water splitting
RCS	Reactive chlorine species
ROS	Reactive oxygen species
SCC	Solar-to-chemical conversion
STH	Solar-to-hydrogen
Vo	Oxygen vacancies

Data availability

No primary research results, software or code have been included and no new data were generated or analysed as part of this review.

Acknowledgements

RHA thanks Greg Metha (Adelaide University) and Gunther Andersson (Flinders University) for the fruitful discussions on Sayama's paper (https://doi.org/10.1021/acsenergylett.8b00318), which motivated this work on photocatalytic production of high value-added chemicals beyond water splitting. This project was financially supported by KAUST (BAS/1/1433-01-01 and FCC/1/5937-19-01).

References

1. A. Fujishima and K. Honda, Nature, 1972, 238, 37–38.
2. R. Marschall, Eur. J. Inorg. Chem., 2021, 2435–2441.
3. C. R. Lhermitte and K. Sivula, ACS Catal., 2019, 9, 2007–2017.
4. Z. Guo, Y. Tian, J. He and H. Song, ChemCatChem, 2024, e202301595.
5. S. Cao, T. Sun, Y. Peng, X. Yu, Q. Li, F. L. Meng, F. Yang, H. Wang, Y. Xie, C.-C. Hou and Q. Xu, Small, 2024, 20, 2404285.
6. K. Sayama, ACS Energy Lett., 2018, 3, 1093–1101.
7. W. A. Rutala and D. J. Weber, Clin. Microbiol. Rev., 1997, 10, 597–610.
8. World Health Organization, Guidelines for drinking-water quality, 4th edition, incorporating the 1st addendum, World Health Organization, 2017.
9. S. Iguchi, Y. Miseki and K. Sayama, Sustainable Energy Fuels, 2018, 2, 155–162.
10. R. Pang, Y. Miseki, S. Okunaka and K. Sayama, ACS Sustainable Chem. Eng., 2020, 8, 8629–8637.
11. R. K. B. Karlsson and A. Cornell, Chem. Rev., 2016, 116, 2982–3028.
12. Z. Chen, J. J. Concepcion, N. Song and T. J. Meyer, Chem. Commun., 2014, 50, 8053–8056.
13. M. Busch, N. Simic and E. Ahlberg, Phys. Chem. Chem. Phys., 2019, 21, 19342–19348.
14. M. Luna-Trujillo, R. Palma-Goyes, J. Vazquez-Arenas and A. Manzo-Robledo, J. Electroanal. Chem., 2020, 878, 114661.
15. C. Zhao, Z. Ding, K. Zhang, Z. Du, H. Fang, L. Chen, H. Jiang, M. Wang and M. Wu, Nano-Micro Lett., 2025, 17, 113.
16. K. Mase, M. Yoneda, Y. Yamada and S. Fukuzumi, Nat. Commun., 2016, 7, 11470.
17. L. Huang, R. Li, R. Chong, G. Liu, J. Han and C. Li, Catal. Sci. Technol., 2014, 4, 2913–2918.
18. J. Johnston, C. O’Rourke and A. Mills, Phys. Chem. Chem. Phys., 2023, 25, 31825–31835.
19. S. Iguchi, K. Teramura, S. Hosokawa and T. Tanaka, Phys. Chem. Chem. Phys., 2015, 17, 17995–18003.
20. G. Hippargi, P. Mangrulkar, A. Chilkalwar, N. Labhsetwar and S. Rayalu, Int. J. Hydrogen Energy, 2018, 43, 6815–6823.
21. X. Guan, F. A. Chowdhury, N. Pant, L. Guo, L. Vayssieres and Z. Mi, J. Phys. Chem. C, 2018, 122, 13797–13802.
22. V.-H. Dang, T.-A. Nguyen, M.-V. Le, D. Q. Nguyen, Y. H. Wang and J. C.-S. Wu, Chem. Eng. J., 2024, 484, 149213.
23. F. Dingenen and S. W. Verbruggen, Renewable Sustainable Energy Rev., 2021, 142, 110866.
24. M. S. Yesupatham, A. Augustin, N. Agamendran, B. Honnappa, M. Shanmugam, P. J. J. Sagayaraj, G. Thennarasu, N. C. Sagaya Selvam and K. Sekar, Sustainable Energy Fuels, 2023, 7, 4727–4757.
25. B. Rao, N. Estrada, S. McGee, J. Mangold, B. Gu and W. A. Jackson, Environ. Sci. Technol., 2012, 46, 11635–11643.
26. S. Dresp, F. Dionigi, M. Klingenhof and P. Strasser, ACS Energy Lett., 2019, 4, 933–942.
27. R. Vogt and R. N. Schindler, J. Photochem. Photobiol., A, 1992, 66, 133–140.
28. J. Fang, Y. Fu and C. Shang, Environ. Sci. Technol., 2014, 48, 1859–1868.
29. X. Li, M. Kan, T. Wang, Z. Qin, T. Zhang, X. Qian, Y. Kuwahara, K. Mori, H. Yamashita and Y. Zhao, Appl. Catal., B, 2021, 296, 120387.
30. M. S. Koo, X. Chen, K. Cho, T. An and W. Choi, Environ. Sci. Technol., 2019, 53, 9926–9936.
31. M. Szabó, N. Lihi, N. Simic and I. Fábián, Chem. Eng. Res. Des., 2021, 169, 97–102.
32. T. Okada, M. Kodera, Y. Miseki, H. Kusama, T. Gunji and K. Sayama, Chem. Commun., 2024, 60, 3299–3302.
33. M. Jadwiszczak, K. Jakubow-Piotrowska, P. Kedzierzawski, K. Bienkowski and J. Augustynski, Adv. Energy Mater., 2020, 10, 1903213.
34. Y. Yang, X. Yuan, Q. Wang, S. Wan, C. Lin, S. Lu, Q. Zhong and K. Zhang, Angew. Chem., Int. Ed., 2024, 63, e202314383.
35. Y. Wang, Y. Wang, J. Zhao and Y. Xu, Appl. Catal., B, 2021, 299, 120676.
36. Y. Zhou, Z. Xu, L. Tang, J. Qin, G. Lu, H. Dong, Z. Bian and M. Zhu, Environ. Sci. Technol., 2024, 58, 6049–6057.
37. J. N. Hausmann, R. Schlögl, P. W. Menezes and M. Driess, Energy Environ. Sci., 2021, 14, 3679–3685.
38. J. Yu, B.-Q. Li, C.-X. Zhao and Q. Zhang, Energy Environ. Sci., 2020, 13, 3253–3268.
39. Y. Wang, Y. Liu, D. Wiley, S. Zhao and Z. Tang, J. Mater. Chem. A, 2021, 9, 18974–18993.
40. S. Choi, W. I. Choi, J.-S. Lee, C. H. Lee, M. Balamurugan, A. D. Schwarz, Z. S. Choi, H. Randriamahazaka and K. T. Nam, Adv. Mater., 2023, 35, 2300429.
41. K. S. Exner, J. Anton, T. Jacob and H. Over, Electrocatalysis, 2015, 6, 163–172.
42. L. I. Krishtalik, Electrochim. Acta, 1981, 26, 329–337.
43. H. A. Hansen, I. C. Man, F. Studt, F. Abild-Pedersen, T. Bligaard and J. Rossmeisl, Phys. Chem. Chem. Phys., 2010, 12, 283–290.
44. Y. Ye, B. Liao, M. Li, M. Mai, L. Zhang, L. Ma, D. Lin, J. Zhao, D. Chen and X. Ma, J. Alloys Compd., 2023, 947, 169480.
45. R. Sakamoto, D. Horiguchi, T. Ikegami, M. Ishiyama, M. Shiga, K. Sasamoto and Y. Katayama, Anal. Sci., 2003, 19, 1445–1447.
46. Q. Lu, Y. Wang, D. Zhang and H. Cong, Res. Chem. Intermed., 2022, 48, 29–47.
47. R. Pang, Y. Miseki and K. Sayama, Catal. Sci. Technol., 2022, 12, 2935–2942.
48. S. Okunaka, Y. Miseki and K. Sayama, iScience, 2020, 23, 101540.
49. Y. Wang, Q. Lu, S. Han, D. Zhang, H. Cong and C. Chen, Mater. Res. Bull., 2023, 166, 112347.
50. J. Ke, M. Adnan Younis, Y. Kong, H. Zhou, J. Liu, L. Lei and Y. Hou, Nano-Micro Lett., 2018, 10, 69.
51. J. Sun, Y. Chi, W. Wang, S. Zhan and F. Zhou, React. Kinet., Mech. Catal., 2024, 137, 1239–1253.
52. S. Han, Y. Wang, D. Zhang and H. Cong, Chem. Commun., 2023, 59, 6857–6860.
53. Y. Wang, S. Han, Q. Lu, D. Zhang, J. Wang and P. He, ChemSusChem, 2024, e202400795.
54. Y. Li, W. Nie, Y. Liu, D. Huang, Z. Xu, X. Peng, C. George, C. Yan, Y. J. Tham, C. Yu, M. Xia, X. Fu, X. Wang, L. Xue, Z. Wang, Z. Xu, X. Chi, T. Wang and A. Ding, Environ. Sci. Technol. Lett., 2020, 7, 70–75.
55. A. Fujishima, X. Zhang and D. A. Tryk, Surf. Sci. Rep., 2008, 63, 515–582.
56. H. Wang, Z. Han, Y. Zhou, X. Liu, D. Zeng, W. Wang, D. Sarker, L. Zhang and W. Wang, Appl. Catal., B, 2021, 297, 120436.
57. T. R. Harris-Lee, Y. Zhang, C. R. Bowen, P. J. Fletcher, Y. Zhao, Z. Guo, J. W. F. Innocent, S. A. L. Johnson and F. Marken, Electrocatalysis, 2021, 12, 65–77.
58. T. Okada, M. Kodera, Y. Miseki, T. Gunji and K. Sayama, J. Catal., 2025, 443, 115969.
59. K. Maeda, K. Teramura, D. Lu, N. Saito, Y. Inoue and K. Domen, Angew. Chem., Int. Ed., 2006, 45, 7806–7809.
60. K. Maeda, K. Teramura, D. Lu, N. Saito, Y. Inoue and K. Domen, J. Phys. Chem. C, 2007, 111, 7554–7560.
61. M. Yoshida, K. Takanabe, K. Maeda, A. Ishikawa, J. Kubota, Y. Sakata, Y. Ikezawa and K. Domen, J. Phys. Chem. C, 2009, 113, 10151–10157.
62. T. Takata, J. Jiang, Y. Sakata, M. Nakabayashi, N. Shibata, V. Nandal, K. Seki, T. Hisatomi and K. Domen, Nature, 2020, 581, 411–414.
63. M. Qureshi, T. Shinagawa, N. Tsiapis and K. Takanabe, ACS Sustain. Chem. Eng., 2017, 5, 8079–8088.
64. S. Okunaka, T. Nakamura, T. Ikeda, K. Tsuruda and H. Tokudome, Sustainable Energy Fuels, 2024, 8, 745–751.
65. Y. Shiraishi, M. Hashimoto, K. Chishiro, K. Moriyama, S. Tanaka and T. Hirai, J. Am. Chem. Soc., 2020, 142, 7574–7583.
66. Y. Shiraishi, Y. Shimabukuro, K. Shima, S. Ichikawa, S. Tanaka and T. Hirai, JACS Au, 2023, 3, 1403–1412.
67. J. Li, Y. Yu and L. Zhang, Nanoscale, 2014, 6, 8473–8488.
68. X. Zhao, Z. Zhao-Karger, D. Wang and M. Fichtner, Angew. Chem., Int. Ed., 2013, 52, 13621–13624.
69. Y. Zhou, H. Wang, X. Liu, D. Zeng, K. Wang, L. Zhang and W. Wang, Appl. Catal., B, 2021, 294, 120265.
70. J. Li, L. Cai, J. Shang, Y. Yu and L. Zhang, Adv. Mater., 2016, 28, 4059–4064.
71. Z. Xu, L. Tang, Y. Zhou, G. Lu, H. Dong and M. Zhu, Chem. Eng. J., 2024, 493, 152589.
72. Y. Shi, H. Shou, H. Li, G. Zhan, X. Liu, Z. Yang, C. Mao, J. Cheng, X. Zhang, Y. Jiang, S. Zhao, J. Wang, X. Liu, L. Song, H. Sun and L. Zhang, Angew. Chem., Int. Ed., 2023, 62, e202302286.
73. J. Wang, L. Zhang, R. Li, D. Zeng, W. Wang and W. Wang, Chem. – Eur. J., 2023, 29, e202301457.
74. J. Desilvestro and M. Grätzel, J. Electroanal. Chem. Interfacial Electrochem., 1987, 238, 129–150.
75. M. Parvin, I. Savickaja, S. Tutliene, A. Naujokaitis, R. Ramanauskas, M. Petruleviciene and J. Juodkazyte, J. Electroanal. Chem., 2024, 954, 118026.
76. J. C. Hill and K.-S. Choi, J. Phys. Chem. C, 2012, 116, 7612–7620.
77. Q. Mi, A. Zhanaidarova, B. S. Brunschwig, H. B. Gray and N. S. Lewis, Energy Environ. Sci., 2012, 5, 5694–5700.
78. Y. Shi, Y. Li, X. Wei, J. Feng, H. Li and W. Zhou, J. Electron. Mater., 2017, 46, 6878–6883.
79. S. Ahmed, I. A. I. Hassan, H. Roy and F. Marken, J. Phys. Chem. C, 2013, 117, 7005–7012.
80. S. Y. Jo, J. Y. Bae, J. Y. Park, M. K. Kim, W.-S. Chae, Y.-I. Kim and K. M. Nam, ACS Appl. Mater. Interfaces, 2024, 16, 54864–54872.
81. J. Juodkazytė, M. Petrulevičienė, M. Parvin, B. Šebeka, I. Savickaja, V. Pakštas, A. Naujokaitis, J. Virkutis and A. Gegeckas, J. Electroanal. Chem., 2020, 871, 114277.
82. S. J. Calero, P. Ortiz, A. F. Oñate and M. T. Cortés, Int. J. Hydrogen Energy, 2016, 41, 4922–4930.
83. A. G. Breuhaus-Alvarez, Q. Cheek, J. J. Cooper, S. Maldonado and B. M. Bartlett, J. Phys. Chem. C, 2021, 125, 8543–8550.
84. D. Chen, Z. Xie, Y. Tong and Y. Huang, Energy Fuels, 2022, 36, 9932–9949.
85. H. Shi, H. Guo, S. Wang, G. Zhang, Y. Hu, W. Jiang and G. Liu, Energy Fuels, 2022, 36, 11404–11427.
86. J. H. Kim and J. S. Lee, Adv. Mater., 2019, 31, 1806938.
87. X. Zhao, J. Hu, S. Chen and Z. Chen, Phys. Chem. Chem. Phys., 2018, 20, 13637–13645.
88. M. Chen, X. Chang, C. Li, H. Wang and L. Jia, J. Colloid Interface Sci., 2023, 640, 162–169.
89. S. Sriwichai, R. Irani, F. Xi, D. Friedrich, C. Höhn, I. Y. Ahmet, N. Wetchakun and F. F. Abdi, Solar RRL, 2021, 5, 2100268.
90. M. Petruleviciene, I. Savickaja, A. Griguceviciene, A. Naujokaitis, R. Ramanauskas and J. Juodkazyte, J. Photochem. Photobiol., A, 2023, 443, 114842.
91. W. Luo, Z. Yang, Z. Li, J. Zhang, J. Liu, Z. Zhao, Z. Wang, S. Yan, T. Yu and Z. Zou, Energy Environ. Sci., 2011, 4, 4046–4051.
92. S. Iguchi, H. Tateno, S. Takasugi, Y. Miseki and K. Sayama, Sustainable Energy Fuels, 2019, 3, 3441–3447.
93. P. Chen, B. Kang, P. Liu, X. Cheng, S. Zhong, X. Wang and B. Fang, J. Power Sources, 2025, 628, 235860.
94. Z. Xi, C. Zhou, K. Kisslinger, T. Nanayakkara, F. Lu, X. Tong and M. Liu, ACS Appl. Mater. Interfaces, 2023, 15, 49281–49288.
95. S. Okunaka, Y. Miseki and K. Sayama, Catal. Sci. Technol., 2021, 11, 5467–5471.
96. Y. Zhang, J. Cui and Y. Pei, Chem. Eng. J., 2023, 464, 142652.
97. Q. Xiong, X. Zhang, Q. Cheng, G. Liu, G. Xu, J. Li, X. Ye and H. Gao, Nano Res., 2021, 14, 1443–1449.
98. I. Chauhan, K. K. Patra, H. Bajpai, N. B. Mhamane, K. N. Salgaonkar and C. S. Gopinath, Dalton Trans., 2023, 52, 2051–2061.
99. S. Zhang, I. Ahmet, S.-H. Kim, O. Kasian, A. M. Mingers, P. Schnell, M. Kölbach, J. Lim, A. Fischer, K. J. J. Mayrhofer, S. Cherevko, B. Gault, R. van de Krol and C. Scheu, ACS Appl. Energy Mater., 2020, 3, 9523–9527.
100. S. Zhang, M. Rohloff, O. Kasian, A. M. Mingers, K. J. J. Mayrhofer, A. Fischer, C. Scheu and S. Cherevko, J. Phys. Chem. C, 2019, 123, 23410–23418.
101. S. Chen and L.-W. Wang, Chem. Mater., 2012, 24, 3659–3666.
102. A. M. Rassoolkhani, W. Cheng, J. Lee, A. McKee, J. Koonce, J. Coffel, A. H. Ghanim, G. A. Aurand, C. Soo Kim, W. Ik Park, H. Jung and S. Mubeen, Commun. Chem., 2019, 2, 57.
103. C. Boyle, N. Skillen, H. Q. N. Gunaratne, P. K. Sharma, J. A. Byrne and P. K. J. Robertson, Mater. Sci. Semicond. Process., 2020, 109, 104930.
104. J. Wu, Y. Tao, C. Zhang, Q. Zhu, D. Zhang and G. Li, J. Hazard. Mater., 2023, 443, 130363.
105. Z. Hu, H. Wu, S. Li, C. Zhang, R. Liang and M. Zhou, Chem. Eng. J., 2024, 498, 155308.
106. S.-T. Xiao, S.-M. Wu, L. Wu, Y. Dong, J.-W. Liu, L.-Y. Wang, X.-Y. Chen, Y.-T. Wang, G. Tian, G.-G. Chang, M. Shalom, P. Fornasiero and X.-Y. Yang, ACS Nano, 2023, 17, 18217–18226.
107. H. Nishiyama, T. Yamada, M. Nakabayashi, Y. Maehara, M. Yamaguchi, Y. Kuromiya, Y. Nagatsuma, H. Tokudome, S. Akiyama, T. Watanabe, R. Narushima, S. Okunaka, N. Shibata, T. Takata, T. Hisatomi and K. Domen, Nature, 2021, 598, 304–307.
108. Y. Goto, T. Hisatomi, Q. Wang, T. Higashi, K. Ishikiriyama, T. Maeda, Y. Sakata, S. Okunaka, H. Tokudome, M. Katayama, S. Akiyama, H. Nishiyama, Y. Inoue, T. Takewaki, T. Setoyama, T. Minegishi, T. Takata, T. Yamada and K. Domen, Joule, 2018, 2, 509–520.
109. T. Uekert, M. A. Bajada, T. Schubert, C. M. Pichler and E. Reisner, ChemSusChem, 2021, 14, 4190–4197.
110. S. Dilger, M. Trottmann and S. Pokrant, ChemSusChem, 2019, 12, 1931–1938.
111. A. Hankin, F. E. Bedoya-Lora, C. K. Ong, J. C. Alexander, F. Petter and G. H. Kelsall, Energy Environ. Sci., 2017, 10, 346–360.
112. I. Rodríguez-Gutiérrez, L. R. P. Peregrino, G. H. Morais, F. Coa, D. S. Teodoro Martinez, R. V. Gonçalves and F. L. Souza, ACS Energy Lett., 2025, 10, 4769–4776.
113. P. P. Kunturu, S. Bera, H. Johnson and M. N. Tsampas, Artif. Photosynth., 2025, 1, 106–116.
114. Y. An, C. Lin, C. Dong, R. Wang, J. Hao, J. Miao, X. Fan, Y. Min and K. Zhang, ACS Energy Lett., 2024, 9, 1415–1422.
115. K. Bienkowski, R. Solarska, L. Trinh, J. Widera-Kalinowska, B. Al-Anesi, M. Liu, G. K. Grandhi, P. Vivo, B. Oral, B. Yılmaz and R. Yıldırım, ACS Catal., 2024, 14, 6603–6622.
116. V.-H. Trinh and J. Seo, ACS Sustainable Chem. Eng., 2023, 11, 1655–1665.
117. J. Seo, S. Mun and V.-H. Trinh, Inorg. Chem. Front., 2026, 13, 51–62.
118. J. Seo and K. Domen, Mater. Chem. Front., 2024, 8, 1451–1479.
119. A. Kumar, A. Kumar and V. Krishnan, ACS Catal., 2020, 10, 10253–10315.
120. X. Li, H. Zhao, J. Liang, Y. Luo, G. Chen, X. Shi, S. Lu, S. Gao, J. Hu, Q. Liu and X. Sun, J. Mater. Chem. A, 2021, 9, 6650–6670.
121. X. Ma, J. Qu, L. Zhang, J. Ma, Y. Cao, K. Nie, Q. Ji, C. Wang, L. Ma and D. Jing, Int. J. Hydrogen Energy, 2024, 55, 441–454.
122. X. Ma, Y. Gao, B. Yang, X. Lou, J. Huang, L. Ma and D. Jing, Mater. Today Nano, 2023, 21, 100289.
123. Y. Jiang, S. Zhou, S. S. Mofarah, R. Niu, Y. Sun, A. Rawal, H. Ma, K. Xue, X. Fang, C. Y. Toe, W.-F. Chen, Y.-S. Chen, J. M. Cairney, R. Rahman, Z. Chen, P. Koshy, D. Wang and C. C. Sorrell, Nano Energy, 2023, 116, 108830.
124. S. Navalón, A. Dhakshinamoorthy, M. Álvaro, B. Ferrer and H. García, Chem. Rev., 2023, 123, 445–490.
125. K. Sun, Y. Qian and H.-L. Jiang, Angew. Chem., Int. Ed., 2023, 62, e202217565.
126. B. Singh and S. Verma, Mater. Chem. Front., 2024, 8, 3136–3149.
127. W.-Y. Chen, Y.-A. Yang and I.-W. P. Chen, Chem. Eng. J., 2025, 520, 166045.
128. M. Jothi, P. Gnanasekar, B. S. Ooi and J. Kulandaivel, J. Phys. Chem. C, 2024, 128, 12799–12807.
129. H. Liu, P. Fu, D. Cui, A. Yuan, H. Zhang, J. Du, Y. Cheng, G. Wang, X. Yang and L. Chen, J. Colloid Interface Sci., 2026, 703, 139234.
130. Y. Bao, H. Ru, Y. Wang, K. Zhang, R. Yu, Q. Wu, A. Yu, D.-S. Li, C. Sun, W. Li and J. Tu, Adv. Funct. Mater., 2024, 34, 2314611.
131. D. Xie, J. Wang, B. Huang, Y. Yang, D. Lin, C. Xu and F. Xie, CrystEngComm, 2025, 27, 678–686.
132. J. Li, S. Sun, N. Gao, H. Li, K. Liang, J. Hai, S. He, X. Mu and B. Wang, Nanoscale Adv., 2023, 5, 3527–3535.
133. L. Li, Z. Zhou, L. Li, Z. Zhuang, J. Bi, J. Chen, Y. Yu and J. Yu, ACS Sustainable Chem. Eng., 2019, 7, 18574–18581.
134. Q. Yue, G. Li, P. Fu, B. Meng, F. Ma, Y. Zhou and J. Wang, Nano Res., 2023, 16, 6251–6259.
135. C.-L. Chang, T.-F. Huang, W.-C. Lin, L.-Y. Ting, C.-H. Shih, Y.-H. Chen, J.-J. Liu, Y.-T. Lin, Y.-T. Tseng, Y.-H. Wu, Y.-E. Sun, M. H. Elsayed, C.-W. Chen, C.-H. Yu and H.-H. Chou, Adv. Energy Mater., 2023, 13, 2300986.
136. Q. Yue, Z. Zhang, X. Liu, C. Zhu, Y. Wen, P. Fu, Q. Hu, X. Qu, Y. Zhou and J. Wang, Chem. Eng. J., 2025, 507, 160481.
137. X. Zhang, M. Chang, D. Wang, L. Wang, X. Yang, Z. Ben, Q. Zhang and Y. Lu, J. Colloid Interface Sci., 2025, 682, 1151–1163.
138. J. Zhang, X. Li, J. Chang, P. She, H. Li, S. Qiu and Q. Fang, Angew. Chem., Int. Ed., 2025, e19752.
139. S. Xiao, D. Wan, K. Zhang, H. Qu and J. Peng, J. Environ. Sci., 2016, 50, 103–108.
140. Y. Zhang, Y. Ji, J. Li, J. Bai, S. Chen, L. Li, J. Wang, T. Zhou, P. Jiang, X. Guan and B. Zhou, J. Hazard. Mater., 2021, 402, 123725.
141. F. Li, L. Sun, Y. Liu, X. Fang, C. Shen, M. Huang, Z. Wang and D. D. Dionysiou, J. Hazard. Mater., 2020, 400, 123246.
142. X. Zhang, W. Li, X. Gong, W. Fan and P. Ren, Desalin. Water Treat., 2015, 54, 1003–1012.
143. Y. Ji, J. Bai, J. Li, T. Luo, L. Qiao, Q. Zeng and B. Zhou, Water Res., 2017, 125, 512–519.
144. Z. Yan, Z. Dai, W. Zheng, Z. Lei, J. Qiu, W. Kuang, W. Huang and C. Feng, Water Res., 2021, 205, 117678.
145. Y. Zhang, J. Li, J. Bai, L. Li, S. Chen, T. Zhou, J. Wang, L. Xia, Q. Xu and B. Zhou, Environ. Sci. Technol., 2019, 53, 6945–6953.
146. Z. Shen, Y. Zhang, C. Zhou, J. Bai, S. Chen, J. Li, J. Wang, X. Guan, M. Rahim and B. Zhou, Water Res., 2020, 170, 115357.
147. J. E. Grebel, J. J. Pignatello and W. A. Mitch, Environ. Sci. Technol., 2010, 44, 6822–6828.
148. X. Zhang, A. Hu, C. Pan, Q. Zhao, X. Wang and J. Lu, Org. Process Res. Dev., 2013, 17, 1591–1596.
149. J. Wang, R. Li, D. Zeng, W. Wang, Y. Zhang, L. Zhang and W. Wang, Chem. Eng. J., 2023, 452, 139505.
150. S. J. Freakley, N. Dimitratos, D. J. Willock, S. H. Taylor, C. J. Kiely and G. J. Hutchings, Acc. Chem. Res., 2021, 54, 2614–2623.
151. L. A. Hull, W. P. Giordano, D. H. Rosenblatt, G. T. Davis, C. K. Mann and S. B. Milliken, J. Phys. Chem., 1969, 73, 2147–2152.
152. W. H. Dennis, L. A. Hull and D. H. Rosenblatt, J. Org. Chem., 1967, 32, 3783–3787.
153. E. E. van Tamelen, V. B. Haarstad and R. L. Orvis, Tetrahedron, 1968, 24, 687–704.
154. W. E. Pereira, Y. Hoyano, R. E. Summons, V. A. Bacon and A. M. Duffield, Biochim. Biophys. Acta, Gen. Subj., 1973, 313, 170–180.
155. V. C. Hand, M. P. Snyder and D. W. Margerum, J. Am. Chem. Soc., 1983, 105, 4022–4025.
156. T. Stelmaszyńska and J. M. Zgliczynski, Eur. J. Biochem., 1978, 92, 301–308.
157. C. L. Hawkins and M. J. Davies, J. Chem. Soc., Perkin Trans. 2, 1998, 1937–1946.
158. Z. T. How, K. L. Linge, F. Busetti and C. A. Joll, Environ. Sci. Technol., 2017, 51, 4870–4876.
159. T. Ding, L. Long, M.-J. Yuan, W. Xu, S.-Y. Wong, H. Gao, Y.-H. Wang, Y.-Y. Yeung and X. Jiang, J. Am. Chem. Soc., 2025, 147, 16225–16236.
160. P.-F. Biard, A. Couvert, C. Renner and J.-P. Levasseur, Can. J. Chem. Eng., 2010, 88, 682–687.
161. M. Azizi, P.-F. Biard, A. Couvert and M. Ben Amor, Chem. Eng. Res. Des., 2015, 94, 141–152.
162. M. Whiteman, N. S. Cheung, Y.-Z. Zhu, S. H. Chu, J. L. Siau, B. S. Wong, J. S. Armstrong and P. K. Moore, Biochem. Biophys. Res. Commun., 2005, 326, 794–798.
163. X. L. Armesto, M. Canle, M. I. Fernandez, M. V. Garcia and J. A. Santaballa, Tetrahedron, 2000, 56, 1103–1109.
164. J. Ra, J. Tolu, D. Rentsch, T. Manasfi and U. von Gunten, Water Res., 2025, 270, 122806.
165. Y. Hu, G. Xie and D. M. Stanbury, Inorg. Chem., 2017, 56, 4047–4056.
166. G. Ding, C. Li, Y. Ni, L. Chen, L. Shuai and G. Liao, EES. Catal., 2023, 1, 369–391.
167. Y.-Y. Li, J.-S. Fan, R.-Q. Tan, H.-C. Yao, Y. Peng, Q.-C. Liu and Z.-J. Li, ACS Appl. Mater. Interfaces, 2020, 12, 54507–54516.
168. L. Hao, L. Kang, H. Huang, L. Ye, K. Han, S. Yang, H. Yu, M. Batmunkh, Y. Zhang and T. Ma, Adv. Mater., 2019, 31, 1900546.
169. W. Zhang, W. Ma, Y. Ma, P. Chen, Q. Ye, Y. Wang, Z. Jiang, Y. Ou and F. Dong, Nano Mater. Sci., 2024, 6, 235–243.
170. D.-Y. Kuo, H. Paik, J. N. Nelson, K. M. Shen, D. G. Schlom and J. Suntivich, J. Chem. Phys., 2019, 150, 041726.
171. S. Saha, P. Gayen and V. K. Ramani, ChemCatChem, 2020, 12, 4922–4929.
172. A. R. Zeradjanin, N. Menzel, P. Strasser and W. Schuhmann, ChemSusChem, 2012, 5, 1897–1904.
173. I. E. Wachs, Catal. Today, 1996, 27, 437–455.
174. Y. Krockenberger, J. Ruiters, H. Irie, Y. Taniyasu and H. Yamamoto, J. Phys. Chem. C, 2025, 129, 18157–18163.
175. Y. Ping, G. Galli and W. A. I. Goddard, J. Phys. Chem. C, 2015, 119, 11570–11577.
176. K. S. Exner, ChemElectroChem, 2020, 7, 1448–1455.
177. D.-Y. Kuo, H. Paik, J. Kloppenburg, B. Faeth, K. M. Shen, D. G. Schlom, G. Hautier and J. Suntivich, J. Am. Chem. Soc., 2018, 140, 17597–17605.
178. K. S. Exner, Phys. Chem. Chem. Phys., 2020, 22, 22451–22458.
179. K. S. Exner, J. Phys. Chem. C, 2019, 123, 16921–16928.
180. K. S. Exner, J. Anton, T. Jacob and H. Over, Angew. Chem., Int. Ed., 2014, 53, 11032–11035.
181. C. E. Finke, S. T. Omelchenko, J. T. Jasper, M. F. Lichterman, C. G. Read, N. S. Lewis and M. R. Hoffmann, Energy Environ. Sci., 2019, 12, 358–365.
182. H. Gong, B. Zhu, D. Zhang, T. Liu, P. Kuang and J. Yu, Appl. Catal., B, 2024, 349, 123892.
183. M. Teranishi, R. Hoshino, S. Naya and H. Tada, Angew. Chem., Int. Ed., 2016, 55, 12773–12777.
184. S. Sun, L. He, M. Yang, J. Cui and S. Liang, Adv. Funct. Mater., 2022, 32, 2106982.
185. D. Chen, C.-L. Dong, Y. Zou, D. Su, Y.-C. Huang, L. Tao, S. Dou, S. Shen and S. Wang, Nanoscale, 2017, 9, 11969–11975.
186. Y. Yan, C. Liu, H. Jian, X. Cheng, T. Hu, D. Wang, L. Shang, G. Chen, P. Schaaf, X. Wang, E. Kan and T. Zhang, Adv. Funct. Mater., 2021, 31, 2009610.
187. F. T. Haase, A. Bergmann, T. E. Jones, J. Timoshenko, A. Herzog, H. S. Jeon, C. Rettenmaier and B. R. Cuenya, Nat. Energy, 2022, 7, 765–773.
188. Z. Li, L. Luo, M. Li, W. Chen, Y. Liu, J. Yang, S.-M. Xu, H. Zhou, L. Ma, M. Xu, X. Kong and H. Duan, Nat. Commun., 2021, 12, 6698.
189. Y. Chen and L. Zhang, Appl. Catal., B, 2024, 347, 123768.

---