Nanoscale halide perovskites for photocatalytic CO₂ reduction: product selectivity, strategies implemented, and charge-carrier separation

Abstract

The over-use of fossil fuels leads to a sharp increase in atmospheric concentrations of carbon dioxide (CO₂), which seriously contributes to the energy crisis and climate problems. The direct transformation of CO₂ into high-value chemicals through photocatalysis offers an effective way to mitigate these problems. The key to achieving this goal is to discover a cost-effective, highly efficient, and durable photocatalyst. Due to their straightforward synthesis, high light absorption capacity, rapid exciton production efficiency, and long carrier diffusion length, nanoscale halide perovskites (NHPs) have great potential for solar photocatalysis. However, several crucial problems, like poor long-term stability, low product selectivity, and severe charge recombination, have become bottlenecks in the development of NHP photocatalysts. Therefore, this review aims to summarize the principles of the CO₂ reduction reaction (CO₂RR), the structural features of halide perovskites nanocrystals (NCs) in the system, and the principal approaches to enhancing their photocatalytic activity. Factors that influence the selectivity of CO₂RR final products are also discussed. Moreover, this review pays special attention to the techniques for studying photogenerated carrier transport processes and photocatalytic intermediates, which make a significant contribution to the insight into the reaction mechanism of photocatalytic CO₂ reduction. Finally, the main challenges and prospects for NHP's further development are also presented. This review will offer instructions for the design of NHP photocatalysts to further enhance the photocatalytic performance and product selectivity for CO₂ reduction. It will also offer insights into studying the charge transport process and mechanism for the CO₂ photocatalytic reduction reaction.

---

1. Introduction

The consumption of fossil fuels has drastically increased during the past few decades, resulting in energy shortages as the global population and industry have grown exponentially.^1–3 The atmospheric CO₂ concentration was 280 parts per million volume (ppmv) before the first industrial revolution, and it has reached 410 ppmv by 2022. Notably, 600 ppmv is expected to be reached by 2100, implying a global temperature increase of 1.9 °C by 2100, which will lead to global warming and other serious climate problems.⁴ However, fossil fuels will remain the main source of energy for the energy infrastructure for a long time to come. As a result, “carbon neutrality” has gained popularity in recent years. Many countries and regions have committed to reaching this goal by mid-century, such as the European Union, China, and the United Kingdom. The conversion of CO₂ and water into carbohydrates at ambient temperature while exposed to light and oxygen is known as photosynthesis, and it is widely regarded as nature's most significant mechanism for preserving the Earth's relative CO₂ and O₂ balance. This has led to the development of solar-driven photocatalytic conversion of CO₂ to useful chemicals as an appealing, practicable, and potentially long-term solution to the problems of energy and the environment.⁵

With a bond energy of up to 750 kJ mol⁻¹, CO₂ is a linear, thermodynamically stable molecule.⁶ Due to the high activation potential, conventional thermocatalytic CO₂ conversion usually requires high temperature and high-pressure conditions.^7–9 Furthermore, unlike electrocatalysis, which requires additional electrical energy, photocatalytic CO₂ conversion can be a green process using solar energy under ambient temperature and pressure conditions. Therefore, it is believed to be a prospective approach in terms of energy efficiency, sustainability, economy and eco-friendliness.¹⁰ Photogenerated electrons are stimulated to the conduction band (CB) when a photocatalyst absorbs light with greater or equivalent energy compared to its band gap, leaving holes at the valence band (VB). Subsequently, the electrons and holes are engaged in the reduction reaction of CO₂ and the oxidation reaction of sacrificial agents, respectively. Depending on the CB and VB positions of the photocatalyst, CO₂ molecules can be reduced to various high value-added chemicals such as CO, CH₄,¹¹ CH₃OH,¹² HCHO,¹³ HCOOH,¹⁴ C₂H₄,¹⁵etc. Therefore, it is crucial to explore photocatalysts with excellent performance and cost effectiveness. Ideal photocatalytic materials should have a wide light absorption range, long carrier lifetime, effective charge separation, excellent stability, abundant active sites, and superb CO₂ adsorption.¹⁶ The earliest research on CO₂ photocatalysis dates back to 1978, when Halmann et al. achieved the photochemical reduction of CO₂ by using p-GaAs at a high cathode applied bias.¹⁷ Inoue et al. then used a variety of semiconductor particles suspended in aqueous solution for selective photoreduction of CO₂.¹⁸ Several types of catalyst materials have been used for CO₂ photocatalytic reduction, including nitrides (CoN, C₃N₄, etc.), sulfides (e.g., CdS, CdSe, etc.), metal oxides (TiO₂, ZnO, NaNbO₃, etc.), MXenes, metal organic frameworks (MOFs), and perovskites (NaNbO₃, MAPbI₃, CsPbBr₃, etc.).^19,20 However, most semiconductors are far from satisfactory as photocatalysts due to several constraints such as rapid compounding of photogenerated charge carriers, insufficient charge potential for oxidation and reduction reactions, or activity only under UV irradiation.²¹ Therefore, the exploration of affordable, reliable, and efficient photocatalysts for CO₂ fixation remains a great need.

Nanoscale halide perovskites (NHPs) have recently gained popularity among scientists because of their unique characteristics, including their low cost, simple synthesis, tunable band gap, high defect tolerance, long charge diffusion length, etc.²² Thus, NHPs have been widely used in photovoltaic conversion, H₂/O₂ evolution, organic degradation, etc.^23–28 They have also recently become among the most exciting choices for the photocatalytic reduction of CO₂.²⁹ Compared with bulk perovskites, NHPs exhibit fascinating quantum confinement effects and superior microstructural features due to their size being comparable to the exciton Bohr radius, endowing them with a tunable band gap, enhanced optical properties and exciton characteristics, and a larger exposed active surface.³⁰ Besides, uniform size, shape, and surface properties of NHPs hold promise for achieving surface-facet-controlled catalysis for specific reaction substrates. These unique size and quantum effects endow NHPs over bulk perovskites with great application potential in the field of photocatalytic CO₂ reduction.³¹ The practical implementation of NHP in CO₂ photoreduction becomes possible under the continuous improvement of synthesis techniques, including ball milling, thermal injection, solvothermal, ultrasonication, ligand-assisted reprecipitation, and microwave assistance.^32–36 Currently, extensive research on photocatalytic CO₂ reduction for NHPs is focused on solving key problems, like low photocatalytic conversion efficiency, poor product selectivity, and weak photocatalyst stability.^37–40 However, review papers related to NHPs for CO₂ photoreduction mainly introduced the photocatalysis principle, structure and optical properties of NHPs, synthesis methods, and modification approaches. Noticeably, there is a great demand for researchers to pay attention to the product selectivity and charge transport process of CO₂ reduction photocatalysis, which will deepen our understanding of the CO₂RR mechanism and guide us to design more efficient and practical photocatalysts. Therefore, this review not only introduces the basic knowledge and modification strategies of NHPs for the CO₂RR, but also summarizes the factors affecting the product selectivity and the characterization techniques to identify the photogenerated charge separation and transfer mechanisms. Among the modification strategies, surface engineering, component engineering, heterojunction engineering, and encapsulation engineering are systematically introduced. As for the reduction products of CO₂, their selectivity relies on four main factors: photoexcitation properties, band structure, charge carrier separation and catalytically active sites. In general, the production of CO is usually due to rapid charge separation and transfer, while CH₄ is mainly produced by changing the B-site perovskites or elements of co-catalysts to provide catalytically active sites. In addition, the characterization techniques for identifying the photogenerated charge transport process and mechanism are presented in four categories according to their functions: charge separation efficiency, charge transfer direction, charge carrier lifetime, and surface reaction intermediate identification. Finally, this review concludes several challenges and prospects on the further study of NHPs for CO₂ photocatalytic reduction in terms of environmental impact, stability, photocatalytic efficiency and selectivity, and the catalytic mechanism. This review will offer an overview for the development of efficient NHP NCs in CO₂ photocatalytic reduction and contribute to an in-depth understanding of the charge transport processes and reaction mechanism of photocatalytic reactions, which will help to address key questions encountered by the scientific community in the energy and environmental fields including but not restricted to CO₂ photoreduction, H₂O splitting, and photoelectrochemical reactions.

2. Photocatalytic CO₂ reduction

2.1 CO₂ photoreduction principle

Understanding the principles of the CO₂RR is essential for the development of suitable photocatalysts. Typically, the whole process of the CO₂RR mainly involves three key steps: (1) absorption of light at suitable wavelengths by the photocatalyst followed by formation of electron–hole pairs; (2) charge carrier separation and migration from the interior to the surface of the catalyst; and (3) oxidation and reduction reactions between the charge carrier and the reactants on the catalyst surface. When a photocatalyst is irradiated by light with energy equivalent to or higher than its band gap energy (E_g), some of the electrons will be excited to leap to the CB, leaving the holes in the VB, and thus electron–hole pairs are formed. Afterwards, the generated electron–hole pairs are separated and transferred to the surface of the perovskites for reduction and oxidation reactions. Simultaneously, some electrons and holes will inevitably recombine through radiation and non-radiation modes that produce light or heat, respectively. In addition, some photogenerated electrons may also be captured in traps. Fig. 1 shows that the recombination of surface electrons and holes would produce energy through light or heat.

Fig. 1 Schematic diagram of the mechanism of the photocatalytic CO₂ reduction process and the corresponding timescales.^41,42

Carbon dioxide is a thermodynamically stable molecule composed of two linearly arranged C=O double bonds, which have a greater bond strength than a variety of chemical bonds including C–H and C–C.⁶ During the catalytic process, inert CO₂ molecules are chemisorbed on the surface of the photocatalyst in a curved structure, through which the carbon dioxide molecules exhibit a lower unoccupied molecular orbital (LUMO) energy level, which makes it easier to receive electrons from the photocatalyst. After accepting the electrons, carbon dioxide will transform into carbonate anion radicals (CO₂⁻) which are more reactive, contributing to further reactions to generate other products.⁴³ As shown in Table 1, by combining different amounts of electrons and hydrogen ions, CO₂ can produce a variety of products through different reaction routes.

Table 1 The major products of photocatalytic CO₂ reduction and the corresponding redox potentials, E⁰ (V vs. NHE at pH = 7)⁴⁴

Equation	Products	E ^0 [V]
CO2 + e^− → CO2^−	Carbonate anion radical	−1.90
CO2 + 2H^+ + 2e^− → HCOOH(aq)	Formic acid	−0.61
CO2 + 2H^+ + 2e^− → CO(g) + H2O	Carbon monoxide	−0.53
CO2 + 4H^+ + 4e^− → HCHO(aq) + H2O	Formaldehyde	−0.48
CO2 + 6H^+ + 6e^− → CH3OH(aq) + H2O	Methanol	−0.38
2CO2 + 12H^+ + 12e^− → C2H4(g) + 4H2O	Ethene	−0.34
2CO2 + 12H^+ + 12e^− → C2H5OH(aq) + 3H2O	Ethanol	−0.33
2CO2 + 8H^+ + 8e^− → CH3COOH(aq) + 2H2O	Acetic acid	−0.31
2CO2 + 14H^+ + 14e^− → C2H6(g) + 4H2O	Ethane	−0.27
CO2 + 8H^+ + 8e^− → CH4(g) + 2H2O	Methane	−0.24

---

2.2 Reaction systems

CO₂ photocatalytic reduction is usually carried out in an air-tight cell, in which the CO₂ molecules mixed with water vapor or dissolved in a solvent are reduced by photocatalysts under light. There are two main reaction modes for this reaction, namely, a “solid–liquid system” and “solid–gas system”, as shown in Fig. 2.

Fig. 2 Schematic diagram of the solid–gas system and solid–liquid system for photocatalytic CO₂ reduction.

The solid–liquid system is typically used in the majority of current research, because it only needs to disperse the photocatalyst particles in water or organic solvents to bring the photocatalyst into full contact with the dissolved CO₂. In contrast, solid–gas photoreaction systems require uniform loading of the photocatalyst onto the solid substrate by methods such as drip and spin coating to prevent its accumulation or agglomeration into clumps.⁴⁵ In addition, the solid–gas system requires additional steps to introduce water vapor, such as by heating the water injected in advance or passing CO₂ into the reactor along with water vapor. Therefore, the solid–liquid system is simpler and less expensive to operate. However, because of their intrinsic ionic nature, NHPs are particularly vulnerable to moisture and are very unstable in polar solvents.⁴⁶ CsPbBr₃ nanocrystals (NCs), for instance, will rapidly undergo phase changes to CsPb₂Br₅ in pure water, leading to a significant loss of light absorption capacity.⁴⁷ Therefore, low-polarity solvents are widely used in solid–liquid photocatalytic systems to mitigate the destruction of NHP structures. By far, the best performing low-polarity solvent is ethyl acetate (EA).⁴⁸ Some other solvents, such as acetonitrile (ACN), toluene, and benzene, have also been tried for the photocatalytic reduction of CO₂.⁴⁹ Nevertheless, in all these purely low-polar organic solvents, the organic molecules could not provide enough protons/electrons, resulting in very limited photocatalytic activity of the photocatalyst for CO₂ reduction. Therefore, researchers subsequently developed a mixture of solvents containing low-polar organic solvents and water that can provide sufficient protons/electrons, such as EA/H₂O and ACN/H₂O.⁵⁰ It was found that adding the right amount of water could significantly increase the productivity of photocatalytic CO₂ reduction.⁵¹ In addition to water, some other alcohol solvents, like methanol, isopropanol, etc., can also be used as electron sacrificial agents to be oxidized to value-added products during the photocatalytic transformation of CO₂.⁵² To avoid the use of expensive and toxic organic reagents, researchers have also started to try to synthesize water-stabilized NHPs and use pure H₂O as a dispersant.^49,53 Besides, solid–liquid systems have the problems of limited solubility of CO₂ in the solvent and hydrogen production, which can limit the yield and selectivity of photocatalytic CO₂ reduction, but these two problems are well avoided in solid–gas systems. In terms of product identification, the solid–liquid system can detect liquid-phase products such as CH₃OH, HCOOH and gas-phase products (CH₄ and CO), while the solid–gas system can only detect gas-phase products. Therefore, it is important to choose the appropriate reaction system according to the need.

2.3 The structure and optical properties of NHPs

2.3.1 Structures of NHPs. Nanoscale halide perovskites (NHPs) are derivatives of calcium titanate (CaTiO₃)⁵⁴ and therefore adhere to the fundamental structural features of the ABX₃ chemical formula. In general, the A-sites of the NHPs are usually inorganic/organic monovalent cations (Cs⁺, Rb⁺, methylammonium (MA⁺), formamidinium (FA⁺), etc.), the B-sites are divalent metal cations (Pb²⁺, Sb²⁺ Sn²⁺, Bi³⁺, etc.), and the X-sites are halogen or halogen-like monovalent anions (I⁻, Br⁻, Cl⁻, tetrafluoroborate (BF₄⁻), thiocyanate (SCN⁻), etc.). The crystal structure of halide perovskites is usually composed of eight adjacent [BX₆] octahedra.⁵⁵ The center of the octahedron is the metal cation (B) and the apex is the six halogen anions. The A-site cation is placed at the center of the eight octahedra and coordinates with the anion X on the octahedron to maintain the overall charge balance. The nature of perovskites is determined by the cations occupying the A and B sites in its lattice, with the A cation usually being larger than the B cation. Two essential factors, the Goldschmidt tolerance coefficient [t] and the octahedral coefficient [μ], affect the stability of perovskites.⁵⁶

R _A, R_B and R_X in eqn (1) and (2) correspond to the radius of the A, B, and X-site ions in the ABX₃ structure. Their changes directly lead to changes in the lattice parameter and thus affect the crystal structure. As shown in Fig. 3h, theoretically stable NHPs should have a value of [t] from 0.8 to 1.0 and a [μ] value between 0.44 and 0.9, respectively.⁵⁸ In general, NHPs will exhibit a cubic structure (α-phase) when the [t] value is in the range of 0.9 and 1.0,⁵⁹ while a perfect cube is formed when [t] = 1.0. α-Phase NHPs are desirable crystal structures for the solar light absorption because of their excellent optical properties. The divalent B-site metal ion radius tends to be large in practice. Therefore, using A-site cations with a greater ionic radius is essential to get the [t] value as close to 1.0 as possible. However, too large A-site cations may also result in serious deformation of the α-phase lattice, which finally leads to crystal structural change.⁶⁰ This may result in the formation of low-dimensional NHP structures such as two-dimensional sheets, one-dimensional chains, or zero-dimensional clusters.⁶¹ Thus, as shown in Fig. 3a–g, in addition to the conventional ABX₃ structure, NHPs are also considered to exhibit other structures such as A₄BX₆, AB₂X₅, A₂BX₄, A₂BX₆, A₂B₁⁺B₂³⁺X₆, and A₃B₂X₉. Among them, the octahedra in the A₄BX₆ structure dissociate in all ranges and no longer share halide ions between them.^62,63 AB₂X₅ is a type of two-dimensional NHP having a tetragonal phase, which contains Cs⁺ and [Pb₂X₅]⁻ layers with alternating Cs⁺ and [Pb₂X₅]⁻ layers. A₂PbX₄ are also 2-dimensional NHPs which have alternating co-angular [PbX₆]₄⁻octahedral layers.⁶⁴ A₂BX₆ is similar to ABX₃ in that the positive 2-valent cation is replaced by a positive 4-valent B-site cation. A₂B⁺B³⁺X₆ is a 3-D structure consisting of two different kinds of B-site cations. Cs₃M₂X₉ is made of isolated clusters, in which each consists of two coplanar octahedra with Cs⁺ acting as a bridge ion between the clusters.

Fig. 3 Structure schematic diagram of different NHPs. (a) 3D pseudocubic ABX₃, (b) 0D A₄BX₆, (c) 2D AB₂X₅, (d) 2D A₂BX₄, (e) 0D A₂BX₆, (f) 3D A₂B⁺B³⁺X₆, and (g) 2D A₃B₂X₉, and (h) tolerance factors of commonly used ABX₃ NHPs. Reproduced from ref. 57 with permission from Royal Society of Chemistry, Copyright (2021).

NHPs with different dimensionalities have different characteristics for photocatalytic applications. Regarding the dimensionality of the NHPs, it can be 3D bulk, 2D nanostructures (nanosheets), 1D nanostructures (nanotubes, nanowires and nanorods) and 0D nanostructures (nanocrystals and quantum dots). In general, the 3D NHP structure usually has a more compact structure and wider light absorption bandwidth, which covers a wider range of the solar spectrum. However, NHPs with low-dimensional shapes (2D, 1D, and 0D) possess optical and electrical properties that are absent in their 3D counterparts as a result of quantum confinement and steric anisotropy effects.⁶⁵ Since the photogenerated carriers are confined in the three dimensions for the 0D NHPs, they usually possess a high quantum yield and defect-tolerant band gap and, thus, they have great potential for photocatalytic applications. In terms of 1D NHPs, they have one microscopic and two nanoscopic dimensions. The motion of the carrier (electrons and holes) is quantized in two dimensions, which makes them possess a high length-to-width ratio and a well-defined size and morphology. This feature benefits 1D NHPs with efficient transport and propagation of charge carriers and photons along the longitudinal direction. 2D NHPs, with a layered structure, have one nanoscopic and two microscopic dimensions.⁶⁶ The introduction of large-sized organic molecules as isolation layers between the layers significantly improves the stability of the material. Their band gap can be adjusted by regulating the inter-layer distances, which allows for the fine tuning of the light-absorbing properties. In addition, the 2D structure increases the specific surface area, which is conducive to the separation and transfer of photogenerated carriers at the interface, thus enhancing the photocatalytic activity.⁶⁷ It is not reasonable to specify which dimension of NHPs is the best in terms of photocatalytic performance, as they all have a disadvantage of their own. For example, 3D NHPs are restricted by their limited surface-to-volume ratio and ineffective carrier transport across whole crystals, 2D NHPs have wider band gaps and cannot fully absorb the visible spectrum, 1D structures are susceptible to stresses and defects and have poorer structural stability, and 0D NHP quantum dots are more sensitive to environmental factors (e.g., water, oxygen, etc.) due to their extremely small scale. Therefore, the more promising approach in reality is to combine various dimensions of NHPs to form different types of heterojunctions, which complement each other to further enhance the photocatalytic performance.

2.3.2 Classification of NHPs. Depending on the type of cation on the A and B sites, metal halide perovskites can be classified into hybrid organic–inorganic NHPs, all-inorganic NHPs, lead-free NHPs, and double NHPs. When the A-site cations are organic ions such as MA⁺ and FA⁺, the constituted perovskites are called organic–inorganic hybrid NHPs. However, all-inorganic NHPs, with metal ions like Cs⁺ and Rb⁺ as the A-site, have better thermal stability compared to hybrid NHPs due to their more stable physical properties.^68–75 Among them, significant progress has been made in the field of research on cesium (Cs)-based halide perovskites. They are regarded as the most common perovskite catalyst for CO₂ photoreduction. CsPbBr₃ has been extensively studied because of its suitable energy band structure and excellent catalytic performance. The highest yield of CsPbBr₃ NHPs to date was described by Chen et al. who reported CO/CH₄ production with an amount of up to 1724 μmol g⁻¹ under UV light.^76–82 For the B site element, most of the studies have been focused on Pb-based NHPs,⁸³ but the toxicity of Pb limits their further development. When interacting with water, lead (Pb) in the perovskites can leach into the soil or groundwater and enter the human food chain by being absorbed by plants, posing a serious risk to human health. Excessive Pb absorption can cause nausea, muscle weakness, and mental confusion in humans.⁸⁴ Therefore, it is challenging for NHPs containing the Pb element to be commercialized and widely used. It is critical to create NHPs that are non-toxic while maintaining the superior properties of Pb-based NHPs. Construction of perovskites using lead-free metal ions is an effective way to avoid this problem, for which there are two principal strategies: one is to replace lead with other less toxic divalent metal ions of the same group IV (e.g., Ge or Sn), and the other is to substitute two lead ions with a monovalent and a trivalent metal ion (e.g., Ag⁺ and Bi³⁺) to generate A₂B⁺B³⁺X₆ double NHPs.⁸⁵

Although using elements like Ge, Sn, Ag and Bi overcomes the problem of high toxicity compared to Pb, their price is over 10 times that of Pb, which still limits their large-scale application. Therefore, researchers started to utilize much cheaper alternatives like Sb, Cu, Ti, and Mg.^86,87 For instance, Mu et al. prepared Cs₃Sb₂Br₉ hollow nanospheres (H-Cs₃Sb₂Br₉) using a simple antisolvent method. This structure increases light absorption performance, promotes the separation of photogenerated carriers, and exposes a large number of catalytically active sites.⁸⁸ Zhou et al. prepared a series of titanium-based halide perovskite materials (Cs₂TiX₆, X = Cl, Cl_0.5Br_0.5, Br). These titanium-based perovskite materials have very high stability in harsh environments such as light irradiation and heating. Under 3 hours of light irradiation, Cs₂Ti(Cl_0.5Br_0.5)₆ microcrystals catalyzed the production of 176 μmol g⁻¹ of CO and 78.9 μmol g⁻¹ of CH₄.⁸⁹ Zhao et al. investigated the performance of copper-based halide perovskites CsCuCl₃ and CsCuCl₂Br as photocatalysts for CO₂ reduction. Theoretical and experimental analyses revealed that CsCuCl₃ has a suitable bandgap (1.92 eV) and conduction band minimum, allowing it to effectively utilize sunlight and drive the reduction of CO₂ to CH₄ and CO. Partial substitution of bromine ions narrows the bandgap of CsCuCl₂Br and facilitates charge carrier transport, resulting in even better photocatalytic performance.⁹⁰ However, their current photocatalytic performance is still very much lower than that of Ge, Sn, Ag and Bi NHPs. Therefore, some literature combines the low-price metals with Bi, like Cs₂NaBiCl₆ and Cs₃Bi_2xSb_2−2xI₉, which points out a new path for the development of lead-free NHPs.^91,92 It is expected that the performance of inexpensive lead-free NHPs could be improved with continuous research in the future.

2.3.3 Band structure of NHPs. The two important components of the energy band structure of a semiconductor material are the band gap and band positions. The band gap determines the wavelength of light that can be absorbed by the photocatalyst, while the band positions of the conduction band (CB) and valence band (VB) determine the redox potential of this semiconductor material and thus affect its photocatalytic performance.⁹³ Typically, a photocatalyst with a band gap of less than 3 eV indicates that it can harvest solar energy efficiently because it can not only absorb UV light but also visible light.⁹⁴ As for the band position, in order to effectively reduce CO₂, NHPs need to have a CB that is more negative than the CO₂ reduction potentials and a VB that is more positive than the oxidation potential of an electron donor. Moreover, NHPs have high defect tolerance, which is one of the most remarkable features of this material. This is because the VB maximum in their electronic energy band structure consists mainly of antibonding orbitals, while the CB minimum is stabilized by robust spin–orbit coupling.⁹⁵ Therefore, their defective energy levels are located entirely in the VB or CB and not in the bandgap itself, so that they do not affect light absorption. Fig. 4 shows the CB and VB positions of some commonly used NHPs, and the redox potential of CO₂ reduction half-reactions.

Fig. 4 Conduction and valence band positions of some commonly used NHP nanocrystals. Reprinted with permission from ref. 43, Copyright 2020, American Chemical Society.

2.4 Selectivity of NHPs for CO₂ photoreduction

As mentioned above, CO₂ has the highest oxidation state of carbon (+4), so it can be reduced to various lower state products (+2, 0, −2 and −4) by accepting different amounts of protons and electrons. Presently, the principal photoreduction products of CO₂ using NHP photocatalysts are CO and CH₄, occasionally with some other products like HCOOH.⁹⁶ However, mixing of multiple products will lead to high cost for separation. Therefore, it is important to improve product selectivity for practical production. Photoexcitation properties, band structure, charge carrier separation and catalytically active sites are four main factors influencing the product selectivity during the photocatalytic CO₂ reduction reaction.⁹⁷

Light excitation of NHPs to generate efficient photogenerated electrons and holes is the key to the CO₂RR, which is mainly influenced by photon energy and light intensity. Photon energy determines whether the semiconductor can absorb photons to be excited and can thermodynamically affect the product selectivity of the reaction, while light intensity determines the number of photogenerated electrons and holes produced by excitation, which then kinetically affects the reaction rate and product selectivity of a reaction with multiple electrons involved.⁹⁷ For instance, Wang et al.⁹⁸ investigated the impact of light intensity on CO₂ reduction reactions using Cs₃Sb₂I₉ photocatalysts (Fig. 5a). They found that by varying the light intensity of the full spectrum from 0.5 W cm⁻² to 1.5 W cm⁻², the production rate of CO increased rapidly with increasing light intensity and the yield of CH₄ remained almost constant, leading to a higher CO selectivity.

Fig. 5 (a) CO and CH₄ production rates of the Cs₃Sb₂I₉ photocatalyst at different light intensities. Reproduced from ref. 98 with permission from Elsevier, Copyright (2021). (b) Photocatalytic CO₂ reduction performance using FAPbBr₃ QDs and CsPbBr₃ QDs. Reproduced from ref. 99 with permission from Elsevier, Copyright (2020). (c) CO yields of Cs₃Bi₂Br₉, Cs₃Bi₂Cl₉, and Cs₃Bi₂I₉ with time. Reprinted with permission from ref. 100, Copyright 2020, American Chemical Society. (d) Product yields of CsCuCl₃ and CsCuCl₂Br. Reprinted with permission from ref. 90, Copyright 2022, American Chemical Society. (e) Comparison of photocatalytic performance of g-C₃N₄, CABB, CABB@C₃N₄-10% and CABB@C₃N₄-82%. Reproduced from ref. 101 with permission from Elsevier, Copyright (2020). (f) Effect of rGO in Cs₄PbBr₆/rGO for photocatalytic CO₂ reduction under visible light (>420 nm) irradiation. Reproduced from ref. 102 with permission from Elsevier, Copyright (2020). (g) Photocatalytic CO₂ reduction performances of CsPbBr₃-Re(x) after 3 h of reaction. Reproduced from ref. 103 with permission from Wiley-VCH, Copyright (2019). (h) Photocatalytic CO₂ reduction performance of CABB and the HWO/CABB photocatalyst. Reproduced from ref. 104 with permission from Elsevier, Copyright (2023). (i) Production of gases using various photocatalysts based on pristine CsPbBr₃, pristine Ni(tpy), and CsPbBr₃-Ni(tpy). Reprinted with permission from ref. 82, Copyright 2020, American Chemical Society. (j) Bar diagram showing relative ratios of CH₄ and CO produced using addition of Fe(II) acetate to CsPbBr₃, CsPbBr₃, and Fe(II)-doped CsPbBr₃, only Fe(II) acetate without CsPbBr₃ and the control reaction without any additive. Reprinted with permission from ref. 105, Copyright 2019, American Chemical Society.

Additionally, the reduction of CO₂ needs photocatalysts with high enough reduction potentials, which are strongly related to their CB positions. The CB positions can be adjusted through changing different cations and anions in NHPs. For instance, Que et al.⁹⁹ compared the CO₂ reduction performance both in FAPbBr₃ quantum dots (QDs) and CsPbBr₃ QDs (Fig. 5b). They found that in all three reaction mediums (DI, DI + EA, EA), the photocatalytic CO yield rates of FAPbBr₃ QDs are higher than that of CsPbBr₃ QDs. Specifically, in the DI/EA system, the amount of CO can reach up to 181.25 μmol g⁻¹ h⁻¹ for the FAPbBr₃ QDs, while that for CsPbBr₃ QDs is 11.23 μmol g⁻¹ h⁻¹. Both QDs exhibited inferior CH₄ reduction rates compared to CO. In addition to the A-site, element changes in the X-site can also affect product selectivity. Sheng et al.¹⁰⁰ studied the effect of halide elements on the CO₂ to CO productivity. As shown in Fig. 5c, Cs₃Bi₂Br₉ exhibits the highest photoreduction performance with 134.76 μmol g⁻¹ of CO yield and a selectivity of 98.7%, which is much better than those of Cs₃Bi₂Cl₉ and Cs₃Bi₂I₉, 83.06 μmol g⁻¹ and 5.78 μmol g⁻¹, respectively. Zhao et al.⁹⁰ reported that Br-substituted samples (CsCuCl₂Br) show higher CH₄ selectivity than CsCuCl₃ samples (Fig. 5d). CsCuCl₂Br microcrystals (MCs) possess the best performance with CO and CH₄ yields of 5.61 and 15.36 μmol g⁻¹, respectively, corresponding to a CH₄ selectivity of 73.2%. They also found that the morphology of catalysts affects the CO₂ reduction performance and that CsCuCl₂Br MCs perform better than CsCuCl₂Br powders.

Combining NHPs with other materials to construct heterojunctions can change the separation efficiency of photogenerated electrons and holes based on the different energy band structures of the two materials, which also has an impact on the selectivity of the products. Kong et al.¹⁰³ coupled Re(CO)₃Br(dcbpy) with CsPbBr₃ NC to enhance photocatalytic performance of CO₂ reduction (Fig. 5g). For pure CsPbBr₃ NCs, CO and CH₄ were both observed, but on coupling with Re(CO)₃Br(dcbpy) molecules, CO became the predominant product. The CsPbBr₃-Re(600) sample demonstrated the highest performance, yielding 104.37 mol g⁻¹ of CO and 5.64 mol g⁻¹ of H₂, with a selectivity for CO of up to 95%. They suggested that the elimination of the CH₄ product is because the photogenerated electrons of CsPbBr₃ can be transported to Re(CO)₃Br(dcbpy) molecules timely and effectively. Wang et al.¹⁰² hybridized reduced graphene oxide (rGO) to Cs₄PbBr₆ for photocatalytic CO₂ reduction (Fig. 5f). They discovered that the yield of CO continued to increase while the yield of CH₄ decreased as the quantity of rGO in the Cs₄PbBr₆/rGO composite grew, resulting in better CO selectivity. When the ratio of rGO in the composite of Cs₄PbBr₆/rGO reaches 10 wt%, the selectivity of CO can be increased to 94.6%. The superior catalytic ability of Cs₄PbBr₆/rGO is attributed to the fast electron transfer from Cs₄PbBr₆ to rGO, and the presence of defects on rGO could trap electrons and increase the local charge density. Compared to the Schottky junctions of Re(CO)₃Br(dcbpy) and rGO, the photocatalysts in type II heterojunctions promote charge separation more efficiently. Wang et al.¹⁰¹ composed Cs₂AgBiBr₆@g-C₃N₄ (CABB@C₃N₄) for CO₂ photoreduction (Fig. 5e). They found that the surface coverage of g-C₃N₄ on CABB plays an important role in the product selectivity, due to the formation of type-II heterojunctions. The results show that in the CABB@C₃N₄ composite containing 10% C₃N₄, electron transfer from the conduction band of g-C₃N₄ to that of CABB leads to a highly selective conversion of CO₂ to CO.

Furthermore, since CO₂ reduction is a dynamic multi-step surface catalytic reaction, the types and amounts of catalytically active sites on surface of photocatalysts can significantly affect the adsorption/desorption characteristics of reactants/intermediates, thereby altering the selectivity of the final product. Zhou et al. reported a novel 2D/2D heterojunction H₂WO₄/Cs₂AgBiBr₆ (HWO/CABB) photocatalyst, which significantly improves its efficiency and selectivity for photocatalytic CO₂ reduction to CH₄ by combining the synergistic effects of the 2D heterojunction and surface bromine vacancy defects.¹⁰⁴ Specifically, this heterojunction structure facilitates charge transfer and separation, while the surface defects can promote the adsorption and activation of CO₂, thereby reducing the formation energy barrier of the CHO* intermediate, resulting in a CH₄ selectivity of 86.1%, which is 30 times higher than that of the pure CABB photocatalyst. In addition to vacancies in NHPs, some metals such as Fe, Co, Ni, Cu, etc. can also contribute to product selectivity as catalytically active sites. Shyamal et al.¹⁰⁵ found differences in product selectivity between doped and undoped Fe(II) in photocatalysts for the reduction of CO₂ (Fig. 5j). For undoped CsPbBr₃ NCs, the major product was CO (4.6 μmol g⁻¹ h⁻¹). However, after doping 25% iron ions into the CsPbBr₃ NCs, it formed mainly CH₄ (6.1 μmol g⁻¹ h⁻¹). They concluded that the active CH₄ molecules were desorbed from the catalyst surface faster due to Fe(II) doping, which enhanced the production of CH₄. Chen et al.⁸² coupled CsPbBr₃ NCs with the metal complex [Ni(terpy)₂]²⁺ (Ni(tpy)) for photocatalytic CO₂ reduction. As seen in Fig. 5i, as the Ni(tpy) content increased, the catalytic activity of CsPbBr₃-Ni(tpy) showed remarkable selectivity for CO. The researchers came to the conclusion that Ni(tpy) can serve as an electron sink and can offer catalytic sites, resulting in increased photocatalytic activity for CO₂ reduction.

As summarized in Table 2, the selectivity of photocatalytic CO₂ reduction products (CO and CH₄) using NHPs is mainly influenced by light-excitation, band structure, charge carrier separation and catalytically active sites. In order to enhance the selectivity of CH₄, several methods can be taken into consideration: first, by adjusting the types of cations and anions in the perovskite, the position of the conduction band can be tuned to provide the necessary reduction potential for CH₄ generation. Selecting perovskites with Bi, Sn or In as the B-site elements can lead to a bandgap structure more favorable for improving CH₄ selectivity. Furthermore, constructing heterojunctions between the perovskite and reduced graphene oxide or metal complexes can help enhance the separation and transfer efficiency of photogenerated charge carriers, thereby promoting the multi-electron reduction pathway. Finally, doping the perovskite with metals such as Fe, Co, Ni, Cu, and Zn or coupling it with metal complexes, can introduce catalytically active sites that are beneficial for CH₄ formation. In summary, systematically controlling the band structure, charge carrier dynamics, and catalytically active sites can effectively enhance the selectivity towards CH₄ in the CO₂ reduction reaction using perovskite-based photocatalysts.

Table 2 Photocatalyst information for CO/CH₄ selectivity

Photocatalysts	System^a	Main product (μmol g^−1 h^−1)	Selectivity (%)	Influencing factor	Ref.
a S–L: solid–liquid system and S–V: solid–vapor system.
CO selective priority
Cs3Bi2I9	S–V	CO 7.76	83.9	Band structure (Cs^+ > Rb^+ > MA^+)	106
		CH4 1.49
CsPbBr3-Ni(tpy)	S–L	CO 1464	84.9	(a) Charge separation (Schottky scheme)	82
		CH4 260		(b) Active sites (Ni)
Cs3Sb2I9	S–V	CO 95.7	87.8	(a) Light intensity	98
		CH4 2.9
		H2 10.4		(b) Active sites (defects)
FAPbBr3 QDs	S–L	CO 181.25	89.9	Band structure (FA^+ > Cs^+)	99
		CH4 16.9
		H2 3.56
Cs4PbBr6/rGO	S–L	CO 11.4	94.6	(a) Charge separation (Schottky scheme)	102
				(b) Active sites (defects)
CsPbBr3-Re(CO)3Br(dcbpy)	S–L	CO 34.79	95%	(a) Charge separation (Schottky scheme)	103
		H2 1.88		(b) Active sites (Re)
Mn:CsPb(Br/Cl)3	S–L	CO 213	95.9	Active sites (Mn)	107
		CH4 9.1
CsPbBr3 QDs/UiO-66(NH2)	S–L	CO 8.21	97	Charge separation (type II scheme)	108
		CH4 0.26
Cs3Bi2Br9 NCs	S–V	CO 26.95	98.7	Band structure (Br > Cl > I)	100
CsPbBr3 QDs/Cu-TCPP-20	S–L	CO 71.77	99	Charge separation (type II scheme)	109
		CH4 0.81
Cs2AgBiBr6/Bi2WO6	S–L	CO 42.19	99	Charge separation (direct Z-scheme)	110
		CH4 0.41
CsPbBr3 NCs/USGO/α-Fe2O3	S–L	CO 73.8	100	Charge separation (all-solid-state Z-scheme)	111
FAPbBr3/Bi2WO6	S–L	CO 170	100	Charge separation (direct Z-scheme)	112
CH 4 selective priority
Cs2AgInCl6@Ag-2	S–L	CO 6.59	52.3	Band structure (In^3+)	113
		CH4 7.23
Amorphous-TiO2/CsPbBr3 NCs	S–L	CO 11.71	55.6	Charge separation (Schottky scheme)	114
		CH4 20.15
		H2 4.38
Fe(II)-doped CsPbBr3	S–L	CO 3.2	66	Active sites (Fe^2+)	105
		CH4 6.1
MAPbI3@PCN221(Fex)	S–L	CO 6.63	66	Active sites (Fe^2+)	115
		CH4 12.86
Cs2AgBiBr6@g-C3N4	S–L	—	∼70	Charge separation (direct Z-scheme)	101
CsCuCl2Br MCs	S–L	CO 1.87	73.2	Band structure (Br + Cl > Cl)	90
		CH4 5.12
α-Fe2O3/amine-RGO/CsPbBr3 NCs	S–V	CO 2.36	78.1	Charge separation (all-solid-state Z-scheme)	116
		CH4 9.45
		H2 0.29
CsPbBr3 @ZIF-67	S–V	ZIF-67	82.1, 78.2	(a) Charge separation (Schottky scheme)	117
		CH4 3.51
		CO 0.77
CsPbBr3 @ZIF-8		ZIF-8		(b) Active sites (Co^2+, Zn^2+)
		CH4 1.81
		CO 0.51
H2WO4/Cs2AgBiBr6	S–L	CH4 22.6	86.1	(a) Charge separation (direct Z-scheme)	104
				(b) Br vacancy
Cs3Bi2Br9 and Cs2AgBiBr6/mesoporous TiO2	S–L	CH4 32.9, 24.2	88.7, 84.2	Charge separation (Schottky scheme)	118
CsPbBr3/ZnPc	S–L	CH4 168	89	(a) Charge separation (direct Z-scheme)	119
				(b) Active sites (Zn)
CsCuCl3/Cu	S–L	CH4 58.77	92.7	(a) Charge separation (Schottky scheme)	120
				(b) Active sites (Cu)
Cs2AgBiBr6-Cu-RGO	S–V	CO 1.9	93	Active sites (Cu)	121
		CH4 10.7
Cu-RGO-CsPbBr3 NS	S–L	CH4 12.7	94.6	Active sites (Cu)	122
Cs2SnI6 NCs/SnS2 NS	S–V	CH4 0.61	100	Charge separation (type II scheme)	123

---

3. Retrofit strategies

Previous research has demonstrated that four essential variables influence the photocatalytic CO₂RR performance using NHP as photocatalysts: light absorption, charge separation and transfer, catalytically active sites, and catalyst stability. The light absorption and catalytically active sites depend on the band structure and composition of NHPs, which can be tuned by using a series of methods in surface and component engineering. After absorbing light energy, the catalyst excites photogenerated carriers, which then undergo inside-to-surface transfer. However, large amounts of photogenerated electron–hole pairs are recombined during this complex process, which greatly reduces the efficiency of photogenerated carrier usage. For example, the most common method currently used is to combine NHPs with another catalyst as a photocatalyst, that is to inhibit carrier recombination by forming a heterojunction. Besides, the stability problem is usually solved by performing encapsulation. Herein, the retrofit strategies are divided into four parts as shown in Fig. 6: surface engineering, component engineering, heterojunction engineering and encapsulation engineering. Table 3 summarizes recent literature according to the type of retrofit strategy.

Fig. 6 Four main categories of retrofit strategies for NHPs.

Table 3 NHPs modified by different retrofit strategies

Type	Photocatalysts	System	Light source	Products	Stability	Ref.
Surface engineering
Size control	CsPbBr3 QDs	Solid–liquid (EA/H2O)	300 W Xe-lamp with a standard AM 1.5 filter	CO: 34.1 μmol g^−1 (8h)	8 h	124
				CH4: 12.2 μmol g^−1 (8h)
				H2: 0.8 μmol g^−1 (8h)
	Cu-RGO-CsPbBr3	Solid–liquid (H2O)	Xe-lamp irradiation 400 nm filter	CH4: 12.7 μmol g^−1 h^−1	12 h	122
				CO: 0.46 μmol g^−1 h^−1
				H2: 0.27 μmol g^−1 h^−1
	CsPbBr3 nanosheets	Solid–vapor (H2O vapor)	300 W Xe lamp (100 mW cm^−2)	(a) CsPbBr3 NCs	30 h	125
				CO: 5.7 μmol g^−1 h^−1 (b) 4 nm CsPbBr3 nanosheets
				CO: 21.6 μmol g^−1 h^−1
Morphology control	3DOM Au-CPB	Solid–liquid (IPA–EA)	300 W Xe lamp 420 nm filter	R electron: 38.0 μmol g ^−1 h ^−1	—	126
	NMF/CsPbBr3 nanowires	Solid–liquid (EA/H2O)	300 W Xe lamp with a 420 nm filter (100 mW cm^−2)	CO: 81.0 μmol g ^−1 h ^−1	—	127
	Cs2AgBiX6 (X = Cl, Br, I)	Solid–liquid (EA)	405 nm laser diode	(a) Cs2AgBiBr6 NPLs	9 h	85
				Total photocatalytic electron consumption = 255.4 μmol g^−1
				(b) Cs2AgBiBr6 NCs
				Total photocatalytic electron consumption = 30.8 μmol g^−1
	CsPbBr3 nanorods	Solid–liquid (EA/H2O)	450 W Xe lamp (150 mW cm^−2)	CO: 40.81 μmol g^−1 (2h)	—	128
				CH4: 74.45 μmol g−1 (2h)
Facet regulation	Cs3Sb2Br9 NCs	Solid–liquid (ODE/H2O)	300 W Xe-lamp with a AM 1.5 G filter (100 mW cm^−2)	(a) Cs3Sb2Br9 NCs	9 h	129
				CO: 510 μmol g^−1 (4 h)
				(b) Pristine CsPbBr3 NCs
				CO: 50 μmol g^−1 (4 h)
	CsPbBr3 NCs	Solid–liquid (EA/H2O)	450 W Xe-lamp (150 mW cm^−2)	(a) Cube-shaped CsPbBr3 NCs	6 h	130
				CO: 16.4 μmol g^−1 (4h)
				CH4: 7.6 μmol g^−1 (4h)
				(b) Hexapod CsPbBr3 NCs
				CO: 79.5 μmol g^−1 (4h)
				CH4: 38.4 μmol g^−1 (4h)
				(c) Polyhedrons CsPbBr3 NCs
				CO: 130.7 μmol g^−1 (4h)
				CH4: 58.8 μmol g^−1 (4h)
Ligand regulation	Surface Pb-rich CsPbCl3 QDs	Solid–liquid (H2O)	AM 1.5 G solar light (150 mW cm^−2)	(a) Ni: CsPbCl3 NCs	3 h	49
				CO: 8.55 μmol g^−1 h^−1
				(b) Pb-rich Ni: CsPbCl3NCs
				CO: 169.37 μmol g^−1 h^−1
				(c) Pb-rich Mn: CsPbCl3 NCs
				CO: 152.49 μmol g^−1 h^−1
	Cs2AgBiBr6 NCs	Solid–liquid (EA)	AM 1.5 G 100 W Xe-lamp (150 mW cm^−2)	(a) Washed Cs2AgBiBr6 NCs	—	131
				CO: 14.1 μmol g^−1 (6h)
				CH4: 9.6 μmol g^−1 (6h)
				(b) As-prepared Cs2AgBiBr6 NCs
				CO: 5.5 μmol g^−1 (6h)
				CH4: 0.65 μmol g^−1 (6h)
	CsPbBr3-Ni(tpy)	Solid–liquid (EA/H2O)	300 W Xe-lamp (100 mW cm^−2)	CO: 1464 μmol g^−1	4 h	82
				CH4: 260 μmol g^−1
	Cs2AgInCl6@Ag-2	Solid–liquid (EA)	300 W Xe-lamp	CO: 26.4 μmol g^−1 (4 h)	9 h	113
				CH4: 28.9 μmol g^−1 (4 h)
Defect regulation	Cs3Bi2X9 NCs (X = Cl, Br, I)	Solid–vapor (H2O vapor)	300 W Xe-lamp with a AM 1.5 filter	(a) Cs3Bi2Br9 NCs	20 h	100
				CO: 134.76 μmol g^−1 (5 h)
				(b) Cs3Bi2Cl9 NCs
				CO: 83.06 μmol g^−1 (5 h)
				(c) Cs3Bi2I9 NCs
				CO: 5.78 μmol g^−1 (5 h)
	CsPbBr3-BF4/Co	Solid–liquid (EA)	λ > 400 nm	CO: 83.8 μmol g^−1 h^−1	8 h	132
	CsPbBr3 NCs	Solid–liquid (EA/H2O)	450 W Xe-lamp (150 mW cm^−2)	(a) Cube-shaped CsPbBr3 NCs	6 h	130
				CO: 16.4 μmol g^−1 (4h)
				CH4: 7.6 μmol g^−1 (4h)
				(b) Hexapod CsPbBr3 NCs
				CO: 79.5 μmol g^−1 (4h)
				CH4: 38.4 μmol g^−1 (4h)
				(c) Polyhedrons CsPbBr3 NCs
				CO: 130.7 μmol g^−1 (4h)
				CH4: 58.8 μmol g^−1 (4h)
Compositional engineering
A-site	A3Bi2I9 (A = Rb+, Cs+ or, MA+)	Solid–vapor (H2O vapor)	UV lamp (32 W, 305 nm, 80.4 μW cm^−2)	(a) Cs3Bi2I9 NCs	12 h	106
				CO: 77.6 μmol g^−1 (10 h)
				CH4: 14.9 ± 0.8 μmol g^−1 (10 h)
				(b) Rb3Bi2I9 NCs
				CO: 18.2 μmol g^−1 (10 h)
				CH4: 17.0 ± 1.6 μmol g^−1 (10 h)
				(c) MA3Bi2I9 NCs
				CO: 7.2 μmol g^−1 (10 h)
				CH4: 9.8 ± 0.6 μmol g^−1 (10 h)
	FAPbBr3 and CsPbBr3 QDs	Solid–liquid (EA/H2O)	Xenon arc lamp	(a) FAPbBr3 QDs	—	99
				CO: 181.25 μmol g^−1 h^−1
				CH4: 16.9 μmol g^−1 h^−1
				H2: 3.56 μmol g^−1 h^−1
				(b) CsPbBr3 QDs
				CO: 11.23 μmol g^−1 h^−1
				CH4: 0.45 μmol g^−1 h^−1
				H2: 0.13 μmol g^−1 h^−1
B-site	Cs3Sb2I9	Solid–vapor (H2O vapor)	Xe-lamp (200 mW cm^−2)	CO: 13.2 μmol g^−1 h^−1	—	98
	Cs2CuBr4 QDs	Solid–vapor (H2O vapor)	300 W Xe lamp (1.5 G filter)	Cs2CuBr4 QDs	20 h	133
				CH4: 74.81 μmol g^−1 (5 h)
				CO: 148.98 μmol g^−1 (5 h)
	Cs2XCl6 (X = Hf, Zr, Te) MCs	Solid–vapor (H2O vapor)	300 W Xe lamp	(a) Cs2TeCl6 MCs	—	134
				CO: 284.4 μmol g^−1 (3 h)
				CH4: 48.96 μmol g^−1 (3 h)
				(b) Cs2HfCl6 MCs
				CO: 204.3 μmol g^−1 (3 h)
				CH4: 48.66 μmol g^−1 (3 h)
				(c) Cs2ZrCl6 MCs
				CO: 219.2 μmol g^−1 (3 h)
				CH4: 48.36 μmol g^−1 (3 h)
X-site	CsPb(Br0.5/Cl0.5)3	Solid–liquid (EA)	300 W Xe-lamp with a AM 1.5 filter	CO: 767 μmol g^−1 (9 h)	—	135
				CH4: 108 μmol g^−1 (9 h)
	Cs2AgBiX6 NCs	Solid–vapor (H2O vapor)	300 W Xe-lamp with a 420 nm filter	(a) Cs2AgBiI6 NCs	—	136
				CO: 18.9 μmol g^−1 (6 h)
				(b) Cs2AgBiCl6 NCs
				CO: 13.62 μmol g^−1 (6 h)
				(c) Cs2AgBi(Br0.5I0.5)6 NCs
				CO: 11.88 μmol g^−1 (6 h)
	Cs3Bi2X9 (X = Cl, Cl0.5Br0.5, Br, Br0.5I0.5, I)	Solid–vapor (H2O vapor)	300 W Xe-lamp with a 420 nm filter	CO: 54 μmol/g (3h)	10 h	137
	CsCuCl2Br MCs	Solid–liquid (EA/IPA)	AM 1.5 G (100 mW cm^−2)	(a) CsCuCl3 MCs	20 h	90
				CO: 2.79 μmol g^−1 (3 h)
				CH4: 11.55 μmol g^−1 (3 h)
				(b) CsCuCl2Br MCs
				CO: 5.61 μmol g^−1 (3 h)
				CH4: 15.36 μmol g^−1 (3 h)
Metal doping	Mn:CsPb(Br/Cl)3	Solid–liquid (EA)	300 W Xe-lamp with a AM 1.5 filter	(a) Mn:CsPb(Br/Cl)3	9 h	107
				CO: 1917 μmol g^−1 (9 h)
				CH4: 82 μmol g^−1 (9 h)
				(b) Pristine CsPbBr3
				CO: 135 μmol g^−1 (9 h)
				CH4: 58.6 μmol g^−1 (9 h)
	Co-CsPbBr3/Cs4PbBr6	Solid–liquid (ACN/H2O/MeOH)	300 W Xe-lamp (100 mW cm^−2)	(a) Co 1%@CsPbBr3/Cs4PbBr6	15 h	138
				CO: 1835 μmol g^−1 (15 h)
				(b) CsPbBr3/Cs4PbBr6
				CO: 678 μmol g^−1 (15 h)
	Fe(II):CsPbBr3	Solid–liquid (EA/H2O)	450 W Xe-lamp (150 mW cm^−2)	(a) 3% Fe: CsPbBr3	3 h	105
				CO: 3.2 μmol g^−1 h^−1
				CH4: 6.1 μmol g^−1 h^−1
				(b) Pristine CsPbBr3
				CO: 4.6 μmol g^−1 h^−1
				CH4: 1.9 μmol g^−1 h^−1
	Pt/CsPbBr3	Solid–liquid (EA)	150 W Xe-lamp with a 380 nm cut off filter	CO: 5.6 μmol g^−1 h^−1	—	139
	Ni:CsPbBr2.77Ac0.23	Solid–vapor (H2O vapor)	300 W Xe-lamp (100 mW cm^−2)	(a) Ni: CsPbBr2.77Ac0.23	18 h	140
				CO: 44.09 μmol g^−1 h^−1
				(b) Ni: CsPbBr3
				CO: 14.49 μmol g^−1 h^−1
Heterojunction engineering
Schottky junction	CsPbBr3 QDs/graphene oxide (GO)	Solid–liquid (EA)	100 W Xe-lamp with a AM 1.5 filter	CO: 58.7 μmol g^−1 (12 h)	12 h	48
				CH4: 29.6 μmol g^−1 (12 h)
				H2: 1.58 μmol g^−1 (12 h)
	CsPbBr3 NC/Pd NS	Solid–vapor (H2O vapor)	150 W Xe-lamp with a 420 nm filter (150 mW cm^−2)	CO:12.63 μmol g^−1 (3 h)	9 h	141
				CH4: 10.41 μmol g^−1 (3 h)
	CsPbBr3 NCs/MXene nanosheets	Solid–liquid (EA)	300 W Xe-lamp with a 420 nm filter	CO: 133.05 μmol g^−1 (5 h)	5 h	142
				CH4: 33.83 μmol g^−1 (5 h)
	Cs4PbBr6/rGO	Solid–liquid (EA/H2O)	300 W Xe lamp with a 420 nm filter (100 mW cm^−2)	CO: 11.4 μmol g^−1 h^−1	60 h	102
	CsPbBr3 NC/BZNW/MRGO	Solid–vapor (H2O vapor)	150 W Xe-lamp AM 1.5 G with a 420 nm filter (150 mW cm^−2)	R electron = 52.02 μmol g^−1 h^−1	12 h	143
	CsPbBr3 QDs/g-C3N4	Solid–liquid (ACN/H2O)	300 W Xe-lamp with a 420 nm cut-off filter	CO: 149 μmol g^−1 h^−1	—	51
	CsPbBr3/g-C3N4 containing TiO species	Solid–liquid (EA/H2O)	Xe-lamp with a 400 nm cut off filter (100 mW cm^−2)	CO: 129 μmol g^−1 (10 h)	—	144
	Cs2AgBiBr6/MXene	Solid–vapor (H2O vapor)	Xenon lamp (400 nm)	CO: 11 μmol g^−1 h^−1	—	145
				CH4: 1 μmol g^−1 h^−1
				H2: 9 μmol g^−1 h^−1
	CsPbBr3-Re(CO)3Br(dcbpy)	Solid–liquid (toluene/IPA)	AM 1.5 G with a 420 nm filter (150 mW cm^−2)	R electron = 73.34 μmol g^−1 h^−1	15 h	103
	Cs2AgBiBr6-Cu-RGO	Solid–vapor (H2O vapor)	1 Sun	CO: 1.9 μmol g^−1 h^−1	—	121
				CH4: 10.7 μmol g^−1 h^−1
Type II heterojunction	Cs2SnI6 NCs/SnS2 nanosheet	Solid–vapor (H2O/MeOH)	Xe-lamp with a 400 nm filter (150 mW cm^−2)	CH4: 6.09 μmol g^−1 (10 h)	9 h	123
	CsPbBr3 NCs/MoS2 NS	Solid–liquid (EA/H2O)	300 W Xe lamp with a 420 nm filter (200 mW cm^−2)	CO: 25.0 μmol g^−1 h^−1 (3 h)	30 h	146
				CH4: 12.8 μmol g^−1 h^−1 (3 h)
	CsPbBr3 QDs/Cu-TCPP-20	Solid–liquid (ACN)	300 W Xe-lamp with a 420 nm filter	CO: 287.08 μmol g^−1 (4 h)	16 h	109
				CH4: 3.25 μmol g^−1 (4 h)
	CsPbBr3 QDs-PCN	Solid–liquid (ACN/H2O)	300 W Xe lamp with a 420 nm filter	CO: 148.9 μmol g^−1 h^−1	6 h	147
All-solid-state Z-scheme heterojunction	CsPbBr3 NCs/USGO/α-Fe2O3	Solid–liquid (ACN/H2O)	300 W Xe-lamp with a 420 nm filter (100 mW cm^−2)	CO: 73.8 μmol g^−1 h^−1	16 h	111
	α-Fe2O3/amine-RGO/CsPbBr3 NCs	Solid–vapor (H2O vapor)	AM 1.5 G with a 420 nm filter (150 mW cm^−2)	CO: 35.47 μmol g^−1 (15 h)	40 h	116
				CH4: 141.81 μmol g^−1 (15 h)
				H2: 4.36 μmol g^−1 (15 h)
Direct Z-scheme heterojunction (S-scheme)	CsPbBr3 QDs/Bi2WO6 NS	Solid–liquid (EA/H2O)	300 W Xe-lamp with a 420 nm filter (100 mW cm^−2)	CO + CH4: 503 μmol g^−1 (10 h)	10 h	148
	FAPbBr3/Bi2WO6	Solid–liquid (trifluorotoluene)	150 W Xe-lamp with a AM 1.5 G filter (100 mW cm^−2)	CO: 170 μmol g^−1 h^−1	20 h	112
	TiO2/CsPbBr3	Solid–liquid (ACN/H2O)	300 W Xe-arc lamp	CO: 9.02 μmol g^−1 h^−1	—	149
	Cs2AgBiBr6@g-C3N4	Solid–liquid (EA/MeOH)	Xe-lamp (80 mW cm^−2)	CO + CH4: 2 μmol g^−1 h^−1	—	101
	Cs2AgBiBr6−xGCN	Solid–liquid (IPA)	250 W mercury vapor lamp with a wavelength range of 285–700 nm	CO: 12.14 μmol g^−1 h^−1	18 h	150
				CH4: 8.85 μmol g^−1 h^−1
	Cs3Bi2I9/Bi2WO6	Solid–vapor (H2O vapor)	300 W Xe lamp with a 400 nm filter (100 mW cm^−2)	CO: 66 μmol g^−1 (9h)	18 h	151
	Cs2AgBiBr6/Bi2WO6	Solid–liquid (EA/IPA)	300 W Xe lamp with an AM 1.5 G filter (100 mW cm^−2)	CO: 42.19 μmol g^−1 h^−1	—	110
				CH4: 0.41 μmol g^−1 h^−1
	Cs3Bi2I9/CeO2	Solid–vapor (H2O vapor)	300 W Xe lamp	CO: 170 μmol g^−1	—	152
				CH4: 65 μmol g^−1
	CsPbBr3/Bi3O4Br	Solid–liquid (H2O)	300 W Xe lamp with a 420 nm filter	CO: 387.57 μmol g^−1 (4h)	20 h	153
				CH4: 8.53 μmol g^−1 (4h)
	MF/WO/CsPbBr3	Solid–vapor (H2O vapor)	300 W Xe-lamp	CO: 514.06 μmol g^−1 h^−1	64 h	154
				CH4: 86.56 μmol g^−1 h^−1
	CsPbBr3@MTB	Solid–liquid (EA/H2O)	300 W Xe-lamp AM 1.5 filter	CO: 145.28 μmol g^−1 h^−1	16 h	155
Encapsulation engineering
Metal oxide	Amorphous-TiO2/CsPbBr3 NCs	Solid–liquid (EA/IPN)	150 W Xe-lamp with an AM 1.5 G filter	CO: 11.71 μmol g^−1	3 h	114
				CH4: 20.15 μmol g^−1
				H2: 4.38 μmol g^−1
	Cs3Bi2Br9 and Cs2AgBiBr6/mesoporous TiO2	Solid–liquid (IPA)	300 W Xe-lamp (70 mW cm^−2)	CH4: 32.9 and 24.2 μmol g^−1 h^−1	—	118
Nonmetallic materials	CsPbBr3 @GDY0.3-Co	Solid–liquid (ACN/H2O)	300 W Xe lamp with a 400 nm filter (100 mW cm^−2)	CO: 27.7 μmol g^−1 h^−1	—	156
	C60/CsPbBr3	Solid–liquid (ACN/H2O)	300 W Xe lamp with a 420 nm filter (150 mW cm^−2)	CO: 71.3 μmol g^−1	—	157
				CH4: 27.3 μmol g^−1
	P3HT/CsPbBr3	Solid–liquid (ACN/H2O)	300 W Xe lamp with a 420 nm cutoff filter	CO: 145.45 μmol g^−1 h^−1	—	158
				CH4: 23.05 μmol g^−1 h^−1
	Cs3Bi2Br9 in MCM-41	Solid–liquid (H2O)	300 W Xe-lamp with a 420 nm cut-off filter (350 mW cm^−2)	CO: 17.24 μmol g^−1 h^−1	—	159
				H2: 0.527 μmol g^−1 h^−1
	CsPbIxBr3−x/polyethersulfone (PES)	Solid–vapor (H2O vapor)	AM 1.5 G 1 sun condition solar simulator	(a) CsPbBr3 QDs/PES	2 h	160
				CO: 27.22 μmol g^−1 h^−1
				(b) CsPbI2.6Br0.39 QDs/PES
				CO: 32.45 μmol g^−1 h^−1
	MF/CsPbBr3	Solid–vapor (H2O vapor)	300 W Xe-lamp AM 1.5 G	CO: 29.13 μmol g^−1 h^−1	104 h	161
				CH4: 12.95 μmol g^−1 h^−1
	MF/CsPbBr3/g-C3N4	Solid–vapor (H2O vapor)	300 W Xe-lamp (without a filter)	CO: 872.22 μmol g^−1 h^−1	76 h	162
				CH4: 103.35 μmol g^−1 h^−1
Metal organic frameworks	CsPbBr3 @ZIF-67 CsPbBr3 @ZIF-8	Solid–vapor (H2O vapor)	100 W Xe-lamp with a AM 1.5 G filter (150 mW cm^−2)	(a) CsPbBr3 @ZIF-67	18 h	117
				CH4: 10.537 μmol g^−1 (3h)
				CO: 2.301 μmol g^−1 (3h)
				(b) CsPbBr3 @ZIF-8
				CH4: 5.434 μmol g^−1 (3h)
				CO: 1.515 μmol g^−1 (3h)
	CsPbBr3 QDs/UiO-66(NH2)	Solid–liquid (EA/H2O)	300 W Xe-lamp with a 420 nm filter	CO: 98.57 μmol g^−1 (12 h)	36 h	108
				CH4: 3.08 μmol g^−1 (12 h)
	Cs2AgBiBr6/Ce-UiO-66-H	Solid–liquid (H2O)	300 W Xe lamp	CO: 309.01 μmol g^−1 h^−1	10 h	163
	Cs3Bi2Br9/MOF 525 Co	Solid–vapor (H2O vapor)	300 W Xe lamp with a AM 1.5 G filter (100 mW cm^−2)	CO: 61.2 μmol g^−1 h^−1	20 h	164
				CH4: 0.3 μmol g^−1 h^−1
	MAPbI3@PCN221(Fex)	Solid–liquid (EA/H2O)	300 W Xe lamp with a 400 nm filter (100 mW cm^−2)	CO: 530.06 μmol g^−1 (80 h)	80 h	115
				CH4: 1028.94 μmol g^−1 (80 h)

---

3.1 Surface engineering

Extensive research has been reported to modify the surface structure or properties of NHP photocatalysts to improve their photocatalytic activity, which can further be divided into 5 categories: (1) size control, (2) morphology control, (3) facet regulation, (4) ligand regulation, and (5) defect regulation. Different particle sizes and morphologies often result in different surface areas and charge transfer efficiencies of photocatalysts.¹²⁸ Particle size often affects the clustering, surface area/sites, and charge diffusion pathways of NHP NCs, leading to variations in catalytic activity. Hou's team prepared colloidal CsPbBr₃ QDs of different sizes from 3.8 nm to 11.6 nm by adjusting the temperature of the solution phase during synthesis (Fig. 7a).¹²⁴ Their study showed that the increase in CsPbBr₃ QD size is efficient in narrowing the band gap and increasing the utilization of the solar spectrum, where the effect mainly stems from the intrinsic quantum size effect that effectively modulates the band gap and the light absorption of CsPbBr₃ NCs in the visible spectral region. Along with NHP QD size, changing the shape of nanoparticles can also enhance mass transfer at the photocatalytic interface and promote CO₂ adsorption.¹⁶⁶ Wu et al.¹²⁵ successfully synthesized two-dimensional CsPbBr₃ (CPB) nanosheets with different thicknesses (2, 3, 4, and 4.6 nm) and used them for the first time as photocatalysts for CO₂ reduction with H₂O as the electron donor (Fig. 7b). The nanosheets exhibited significantly higher activity and longer stability in photocatalytic CO₂ reduction compared to CsPbBr₃ NCs. Besides, Pradhan's group found that the adsorption capacity and product release rate depend on the helical nature of the nanorods (Fig. 7c).¹²⁸ They controlled the helicity of CsPbBr₃ nanorods by manipulating the composition of alkylammonium ions in the reaction system and found that the evolution of CH₄ depends on the depth of the helical nanorods.

Fig. 7 (a) Quantum-size effects in UV/vis absorption spectra and band gap structures of CsPbBr₃ QDs with different particle sizes. Reproduced from ref. 124 with permission from Wiley-VCH, Copyright (2017). (b) Schematic illustration of band structures of CsPbBr₃ NCs and CsPbBr₃ nanosheets derived from the results of the UV-vis absorption spectroscopy and flat-band potential measurements. Reproduced from ref. 125 with permission from Wiley-VCH, Copyright (2021). (c) Comparison of photocatalytic CO₂ reduction activity of four CsPbBr₃ nanorods. Reprinted with permission from ref. 128, Copyright 2021, American Chemical Society. (d) Histograms showing the formation of CO and CH₄ from CO₂ reduction reactions after 4 h using noncubes or polyhedra, hexapods, or armed and cube-shaped CsPbBr₃ nanostructures as photocatalysts. Reprinted with permission from ref. 165, Copyright 2020, American Chemical Society. (e) Schematic illustration of the regulation of the surface ligand density of the CAIC QDs@Ag-2 composite. Reproduced from ref. 113 with permission from Royal Society of Chemistry, Copyright (2021). (f) Schematic illustration for the surface modification of the CsPbBr₃ NCs. Reproduced from ref. 132 with permission from Wiley-VCH, Copyright (2021).

Additionally, the crystalline facets of NHPs influence their photocatalytic performance.¹⁶⁷ Different charge separation and transport efficiencies can be induced by the crystal's internal electric field (IEF) along different crystal orientations.¹⁶⁸ Photogenerated electrons and holes often cluster in small distinct planes, where those perpendicular to the IEF usually exhibit higher activity in photocatalysis.¹⁶⁹ Crystal facets with lower activation potentials and higher adsorption energies are more advantageous for the photocatalytic reduction of CO₂.¹⁷⁰ For example, Shyamal et al.¹⁶⁵ synthesized CsPbBr₃ NCs in different crystal shapes, such as conventional six-sided cubic and unique polyhedral non-cubic shapes and six-armed hexapod-shaped CsPbBr₃ nanocrystals (Fig. 7d). Assuming comparable surface areas, they claimed that compared to halide rich cubic CsPbBr₃ NCs, halide deficient polyhedral shaped CsPbBr₃ for CO₂ reduction shows better catalytic activity due to the presence of many surface defects, which form lead-rich reaction sites. Moreover, these surface trap states prevent quick charge carrier recombination and offer a chance for effective charge transfer to trigger the photocatalytic reaction.

Organic ligands can also control the shape and size of NCs because they can passivate surface dangling bonds and inhibit particle agglomeration, thus enhancing the stable crystal structure of NHP structures. However, they can also hinder charge transfer between the catalyst and reactants as well as inhibit CO₂ uptake at the active site.¹⁷¹ Therefore, an optimal surface ligand density is essential for high performance and good stability. Chen et al.¹¹³ controlled the ligand density on Cs₂AgInCl₆ QDs@Ag-2 (CAIC QDs@Ag-2) by adjusting the volume ratio of hexane/EC washing solvents (1 : 1, 1 : 2, 1 : 3, and1 : 4) in the purification process, as shown in Fig. 7e. Their results show that the PL intensity of CAIC QDs@Ag-2 composites purified by process II was the lowest compared to the other three purification methods, which implies that reducing the amount of ligands on the catalyst surface is beneficial for charge transfer, but too few ligands will lead to agglomeration of the catalyst and thus reduce the effective separation of the carriers.

In general, the synthesized NHPs inevitably have surface defects, such as the lack of X-site halogen anions or B-site metal cations. These defects usually become charge recombination centers and hinder the effective separation of photogenerated carriers, thus reducing the catalytic performance of photocatalysts. Therefore, it is demanded for researchers to modify the surface defects of photocatalysts. For instance, Wang et al.¹³² modified the surface defects of CsPbBr₃ NCs, i.e., uncoordinated Pb atoms and nanoparticles, by a two-step electrostatic self-assembly method (Fig. 7f). They used NH₄BF₄ salt as a defect treatment agent to peel off the defective layer on the surface of CsPbBr₃. Their experimental results showed that the surface-modified samples (CPB–BF₄) exhibited a more than 3 times higher CO₂ reduction rate compared with the pristine CsPbBr₃ samples, which are 29.8 μmol g⁻¹ h⁻¹ and 8 μmol g⁻¹ h⁻¹, respectively.

In conclusion, the photocatalytic performance of NHPs can be significantly influenced by adjusting their size, morphology, crystal facet orientation, surface ligand density, and defect states. First, tuning the nanoparticle size can substantially modify the quantum confinement effect, thereby regulating the bandgap and light absorption characteristics. Different nanostructures, such as two-dimensional nanosheets and one-dimensional nanorods, can alter the mass transfer process of reactants and the migration channels of photogenerated charge carriers. Additionally, rationally designing the structural morphology can help optimize the entire photocatalytic process. Different crystal facets have distinct internal electric field distributions, which can influence the separation and transfer of photogenerated electron–hole pairs. Meanwhile, appropriate surface ligands can passivate defect sites and suppress charge carrier recombination, but excessive ligands may hinder reactant adsorption and charge transfer. Therefore, the ligand density needs to be controlled within an optimal range. The common surface defects of perovskite nanoparticles, such as halide vacancies and metal ion vacancies, often become centers for charge carrier recombination. Through chemical modification methods, the surface defects can be effectively reduced, enhancing the separation efficiency of photogenerated charge carriers. By comprehensively applying these methods, the performance of perovskite photocatalysts can be significantly enhanced, providing important technical support for realizing efficient photochemical CO₂ reduction conversion.

3.2 Component engineering

In addition to the modifications on the NHP surface, changing components of NHPs is another effective approach to improve the photocatalytic performance of NHPs. Specifically, the compositional adjustment mainly includes the changing of organic/inorganic cations and halogen anions, and the doping of metal ions. For example, Bhosale et al.¹⁰⁶ synthesized three bismuth-based NHPs with the chemical formulae of A₃Bi₂I₉, where A-cations are Rb⁺, Cs⁺ and MA⁺ (Fig. 8a). In general, the photocatalytic activity of these three photocatalysts for the reduction of CO₂ shows a trend Cs₃Bi₂I₉ > Rb₃Bi₂I₉ > MA₃Bi₂I₉. Among them, Cs₃Bi₂I₉ and Rb₃Bi₂I₉ have the greatest yield of CO and CH₄, 77.6 and 17.0 μmol g⁻¹, respectively. They concluded that the differences in photocatalytic performance are due to changes in charge transfer properties and catalytic pathways resulting from the use of different cations. Apart from the cation at the A-site, the change in the cation at the B-site also affects the photocatalytic performance of NHPs for CO₂. As Fig. 8b shows, with the change of the B-site element from Hf to Zr and finally Te, the CO yield of Cs₂BCl₆ is enhanced obviously. Fig. 8c illustrates that compared to Pb when the B-site is Cu, the NHP has a stronger adsorption of CO₂ molecules due to the synergistic effect with other atoms (Cs and Br). Beyond these, X-site halide replacement can modify the band gap and band edge positions, influencing photon absorption and the quantity of photogenerated carriers, thus significantly affecting the catalytic behavior. Guo et al.¹³⁵ reported low-cost cubic-phase CsPb(Br_0.5/Cl_0.5)₃ perovskites with different Br and Cl ratios and investigated the effect on photocatalytic CO₂ reduction (Fig. 8c). They found that the mixed halide perovskites observed a gradual shift and rapid increase of pristine CsPbBr₃ from 517 nm to 413 nm with increasing amount of Cl in the emission spectra, which was attributed to the improved electron separation. With increasing Cl concentration, the formation of CO and CH₄ increases. Wu et al.¹²⁵ changed the I content in NHP 2D nanosheets and synthesized CsPbBr_3−xI_x (x = 0.3, 0.6, 0.9) (Fig. 8d). The synthesized CsPbBr_3−xI_x nanosheets exhibited reduced band gaps and red-shifted absorption peaks from 452 nm to 466 nm with increasing I concentration. In comparison to pristine CsPbBr₃, all CsPbBr_3−xI_x nanosheets displayed greater photocatalytic CO₂ reduction activity. Additionally, the CsPbBr_3−xI_x nanosheet surface trapping was boosted by the anion exchange reaction, which decreased the recombination of photogenerated carriers. Hence, this facilitated the CO₂ reduction reaction, and enhanced the separation and transfer efficiency of the change carriers.

Fig. 8 (a) Comparison of production of methane and carbon monoxide during the photochemical reaction for 10 h. Reprinted with permission from ref. 106, Copyright 2019, American Chemical Society. (b) Performance of the photocatalytic CO₂ reduction of the as prepared Cs₂HfCl₆, Cs₂ZrCl₆ and Cs₂TeCl₆ microcrystals. Reproduced from ref. 134 with permission from Elsevier, Copyright (2022). (c) Yields of CO₂ reduction using CsPb(Br_x/Cl_1−x)₃ (x = 1, 0.7, 0.5, 0.3, 0) nanocrystals as catalysts under 9 h Xe lamp irradiation with an AM 1.5 filter. Reproduced from ref. 135 with permission from Elsevier, Copyright (2018). (d) The schematic illustration of the band structures of CsPbBr_3−xI_x nanosheets derived from the results of flat-band potential and UV-vis absorption spectroscopy measurements. Reproduced from ref. 125 with permission from Wiley-VCH, Copyright (2021). (e) CO and CH₄ evolution rate in EA/IPA over CsPbBr₃, a physical mixture of CsPbBr₃, and Co²⁺ cations (CPB/Co), CPB-BF₄, and CPB-BF₄/Co, respectively. Reproduced from ref. 132 with permission from Wiley-VCH, Copyright (2021). (f) Schematic presentation of Fe-doped CsPbBr₃ and undoped CsPbBr₃ nanocrystal photocatalysts. Reprinted with permission from ref. 105, Copyright 2019, American Chemical Society. (g) Schematic illustration of the electron spin polarization induced longer photoexcited carrier lifetime under an external magnetic field in Mn-CsPbBr₃ NPLs. Reprinted with permission from ref. 172, Copyright 2022, American Chemical Society.

Doping metals can also be used to enhance the charge separation in NHPs by increasing the catalytically active sites and facilitating the adsorption and desorption of intermediates, thus improving the photocatalytic performance of NHPs.¹⁷³ By using density functional theory (DFT) computations, Tang et al.¹⁷⁴ modelled the photocatalytic CO₂ RR of CsPbBr₃ doped with various metal components, including Co, Fe, Ni, Cu, Ag, Mg, Mn, and Bi in benzene. This study shows that the doping of Co and Fe atoms significantly increased the catalytic activity of CsPbBr₃. This conclusion was later confirmed by a couple of teams in their experiments. For example, Wang et al.¹³² synthesized CsPbBr₃-BF₄/Co with optimal Co loading by the electrostatic self-assembly methodology. They showed that the photocatalytic activity in EA/isopropanol (IPA) under 100 mW cm⁻² light was 83.8 μmol g⁻¹ h⁻¹ at 2.0 μmol (Fig. 8e). Not only did the doped Co metal change the selectivity of the CO₂RR product, but it also reduced the formation of photogenerated electron–hole pair complexes, increased photoresponsiveness, and improved the stability of NHPs. According to the report by Shyamal et al.,¹⁰⁵ doping Fe(II) accelerates the CsPbBr₃ QD photocatalytic reduction of CO₂ and modifies the selectivity of the reaction products (Fig. 8f). Besides these, another study by Lin et al. showed that they successfully manipulated the spin-polarized electrons in CsPbBr₃ perovskite nanoplates by doping magnetic Mn²⁺ ions and applying an external magnetic field, which significantly improved the conversion efficiency of photocatalytic CO₂ reduction. Specifically, Mn doping increased the number of spin-polarized photoexcited carriers, extended the carrier lifetime, and suppressed carrier recombination, thereby substantially enhancing the photocatalytic performance of Mn-CsPbBr₃ nanoplates.¹⁷² However, these studies usually use in situ doping techniques to directly inject metal ions inside NHPs, and these metal ions can also serve as carrier recombination sites, thus inhibiting the carrier diffusion. Therefore, post-synthesis cation exchange is being studied as a potential solution to solve this problem by attaching active metal ions to the surface of NHP nanocrystals.¹⁷⁵

In summary, modifying the compositional components of NHPs is another effective approach for improving their photocatalytic performance. Substituting different A-site cations or B-site cations can significantly influence the charge transfer characteristics, catalytic pathways, and CO₂ adsorption of NHPs, thereby affecting their photocatalytic performance. Replacing the X-site halogens (such as Br, Cl, and I) can modify the bandgap and band-edge positions, impacting light absorption, photogenerated carrier density, and catalytic behavior. Doping with metal dopants (such as Co, Fe, and Mn) can increase the catalytically active sites, promote the adsorption/desorption of intermediates, and enhance carrier separation, thereby improving the photocatalytic activity and selectivity of CO₂ reduction products. By adjusting the composition of organic/inorganic cations and halide anions, as well as appropriately adding metal dopants, the charge transfer characteristics, light absorption, and charge carrier separation properties of NHPs can be effectively tuned, thereby significantly enhancing the photocatalytic CO₂ reduction activity and selectivity of NHPs, providing a valuable strategy for the development of high efficiency photocatalysts.

3.3 Heterojunction engineering

Numerous studies have demonstrated that one of the major obstacles for improving the photocatalytic performance of NHP materials is the photogenerated carrier spatial separation efficiency. To address the problem of severe carrier recombination leading to low photocatalytic activity, scientists have attempted to integrate pristine NHPs with various functional materials by constructing heterojunctions, such as Schottky junctions, type-II heterojunctions and direct/indirect Z-scheme heterojunctions.¹⁷⁷

Schottky junctions are usually formed by combining a semiconductor (e.g., NHP NCs) and a conductor (metal, graphene, etc.). The free electrons generated by the NHPs upon photoexcitation are transferred to the conductor, thus suppressing the recombination of electrons and holes. The energy band bending effect at the interface of the two materials leads to the formation of a Schottky barrier, which helps prevent electron migration from the coupled conductor back to the NHP NCs. In 2017, Xu et al.⁴⁸ prepared CsPbBr₃/GO Schottky junctions for photocatalytic CO₂ reduction by room temperature antisolvent precipitation (Fig. 9a). The photogenerated electrons in CsPbBr₃ can be easily transported to GO because CsPbBr₃ has a negative CB edge compared to the Fermi energy level of GO. The electron consumption rate (R_electron) increased by 25.5% for the CsPbBr₃/GO composite compared to that of pristine CsPbBr₃ NCs in ethyl acetate (EA) solution. Later, Xu's group¹⁴¹ successfully developed another Schottky junction composite by attaching CsPbBr₃ NCs to the 2D Pd nanosheet (NS), which can rapidly transfer photogenerated electrons from CsPbBr₃ NCs to the Pd NS (Fig. 9b). The Pd NS also offers a number of active sites for CO₂ photocatalytic reduction. Thus, the CsPbBr₃ NCs/Pd nanosheet composites achieve a high electron consumption rate of 33.79 μmol g⁻¹h⁻¹, which is 2.43 times higher than that of pure CsPbBr₃ NCs (9.86 μmol g⁻¹h⁻¹). In addition to these, metal complexes (CsPbBr₃ NCs-Ni(tpy))⁸² and MXenes¹⁴² are also regarded as promising candidates for decorating NHPs due to their suitable catalytically active sites and favorable electron receiving capability (Fig. 9c and d).

Fig. 9 (a) Schematic diagram of CO₂ photoreduction over the CsPbBr₃ QD/GO photocatalyst. Reprinted with permission from ref. 48, Copyright 2017, American Chemical Society. (b) Schematic diagram of CO₂ photoreduction over the CsPbBr₃ NCs/Pd nanosheet. Reprinted with permission from ref. 141, Copyright 2018, American Chemical Society. (c) Schematic diagram of CO₂ photoreduction over CsPbBr₃ NCs/MXene. Reprinted with permission from ref. 142, Copyright 2019, American Chemical Society. (d) Schematic diagram of CO₂ photoreduction over the CsPbBr₃ NCs-Ni(tpy) photocatalyst. Reprinted with permission from ref. 82, Copyright 2020, American Chemical Society. (e) Band structure of the 20 CPB–PCN composite photocatalyst. Reproduced from ref. 51 with permission from Wiley-VCH, Copyright (2018). (f) Schematic illustration of the possible mechanism of photocatalytic CO₂ reduction using CsPbBr₃ QDs/UiO-66(NH₂). Reproduced from ref. 176 with permission from Elsevier, Copyright (2018). (g) Schematic diagram of CO₂ photoreduction over the α-Fe₂O₃/amine-RGO/CsPbBr₃ photocatalyst. Reproduced from ref. 116 with permission from Elsevier, Copyright (2020). (h) Schematic illustrations for the CsPbBr₃/USGO/α-Fe₂O₃ photocatalyst. Reproduced from ref. 111 with permission from Wiley-VCH, Copyright (2020). (i) Schematic illustration of perovskite-based heterostructures via type-II to Z-scheme transformation for photocatalytic CO₂ reduction through an ohmic contact. Reprinted with permission from ref. 39, Copyright 2024, American Chemical Society. (j) Z-scheme photocatalytic system for CO₂ reduction. Reprinted with permission from ref. 148, Copyright 2020, American Chemical Society. (k) Schematic illustration of the TiO₂/CsPbBr₃ heterojunction. Reproduced from ref. 149 with permission from Springer Nature, Copyright (2020).

Type II heterojunctions are generally considered to exhibit higher photon utilization and catalytic performance compared to Schottky junctions because both NHPs and coupled semiconductors are capable of generating photoexcited carriers. Moreover, the energy bands of the two semiconductors forming the type II heterojunction are interleaved. The photogenerated electrons from the semiconductor with a higher CB will migrate and accumulate on the semiconductor with a lower CB; in contrast, the photogenerated holes will move to the semiconductor with a higher VB. Thus, the oxidation and reduction reactions of the type II heterojunction occur separately on both semiconductors, thus improving the utilization of electrons and holes. Currently, coupled semiconductors with suitable band edge positions (e.g., g-C₃N₄ (PCN), metal–organic frameworks (MOFs), oxides, and sulfides) have been used to construct type II heterojunctions. For example, Ou et al.⁵¹ immobilized CsPbBr₃ quantum dots (QDs) onto NH_x-rich porous PCN via strong interactions between Br atoms on the surface of CsPbBr₃ and N atoms on PCN (Fig. 9e). The separation of broad-generated charges between CsPbBr₃ and PCN was facilitated by the formation of type-II heterojunctions, in which photogenerated electrons were transported from CsPbBr₃ QDs to PCN for CO₂ reduction, while holes are transferred from PCN to CsPbBr₃ QDs for H₂O oxidation via valence band shifts. Thus, the composite catalyst after forming heterojunctions has a 3-fold higher CO generation rate than that of CsPbBr₃ QDs alone. In addition, metal–organic frameworks (MOFs) are a group of crystalline porous materials made up of metal nodes and organic linkers, which are characterized by large surface area and structural tunability.¹⁷⁸ By directly combining pre-synthesized MOFs UiO-66(NH₂) and CsPbBr₃ QDs, Wan et al.¹⁷⁶ created composites and obtained high CO₂ conversion (CO generation of 98.57 mol g⁻¹) in the EA/water system, which was significantly higher than that of pristine CsPbBr₃ QDs and UiO-66(NH₂) (Fig. 9f). The superior CO₂ reduction photoactivity is attributable to the wide reachable surface area, the better visible light collection capacity, and the improved electron extraction and transfer between two materials owing to the development of type II heterojunctions.

Although it has been demonstrated that type II heterojunctions can effectively encourage space charge separation and subsequently enhance the catalytic performance of CO₂ reduction, this improvement comes at the expense of weakening the reduction/oxidation of photogenerated electrons/holes, which is detrimental to the performance of the CO₂RR. Therefore, a novel heterojunction structure, Z-scheme heterojunction, has been heavily investigated. Similar to type-II heterojunctions, Z-scheme heterojunctions are usually composed of two semiconductors with interleaved energy bands. However, the difference is that the photogenerated electrons in the semiconductor with a lower CB in the Z-scheme heterojunction combine with the photogenerated holes in the semiconductor with a higher VB. Therefore, the photogenerated electrons left in the semiconductor with a higher CB will reduce CO₂ while the holes left in the semiconductor with a lower VB will participate in the oxidation reaction, resulting in a strong redox capability. According to a combination method of photogenerated electrons and holes, Z-scheme heterojunctions are further divided into all-solid-state Z-scheme heterojunctions and direct Z-scheme heterojunctions (S-scheme). An all-solid-state Z-type heterojunction consists of two semiconductors and a conductor that acts as an electron mediator. For effective photocatalytic CO₂ reduction, Jiang et al.¹¹⁶ originally presented an all-solid-state Z-scheme composite based on NHPs, α-Fe₂O₃/amine-rGO/CsPbBr₃, where amine-rGO operates as the electron mediator (Fig. 9g). Charge transfer from α-Fe₂O₃ to CsPbBr₃ QD is facilitated by the matching energy band structure, which also lessens the interfacial recombination of photogenerated carriers. With a yield of 181.68 mol g⁻¹ and a selectivity of 93.4%, CH₄ was identified as the primary reduction product after 15 hours of nonstop illumination. Within 5 photocatalytic reaction cycles, this Z-composite also demonstrated remarkable stability and nearly consistent photocatalytic performance. Another all-solid-state Z scheme was subsequently introduced by Mu et al.¹¹¹ CsPbBr₃ NCs, rod-like α-Fe₂O₃, and ultrathin small-sized graphene oxide (USGO) nanosheets were used to construct Z-scheme heterojunctions (Fig. 9h). When combined with α-Fe₂O₃ and CsPbBr₃, USGO functions as an electronic mediator by creating C–O–Fe and Br–O–C bonds, respectively. According to the results, the CO yield of CsPbBr₃/USGO/α-Fe₂O₃ is 19 times higher compared to that of pure CsPbBr₃, at 73.8 mol g⁻¹ h⁻¹. Recently, Song et al. explored the transformation from type-II into Z-scheme heterostructures based on perovskite materials to enhance photocatalytic CO₂ reduction as shown in Fig. 9i.³⁹ The research showed that by incorporating Au as an electron transport mediator in the CsPbBr₃/TiO₂ heterostructure, a low-resistance ohmic contact can be formed, thereby transforming the type-II structure into a Z-scheme heterojunction. Compared to pristine CsPbBr₃ and CsPbBr₃/TiO₂, the CsPbBr₃/Au/TiO₂ Z-scheme heterostructure exhibited a 5.4-fold and 3.0-fold enhancement, respectively, in the electron consumption rate for photocatalytic CO₂ reduction.

On the other hand, the presence of electron mediators in an all-solid-state Z-scheme heterojunction may hinder the efficient absorption of light by semiconductor materials. Therefore, a direct Z-scheme heterojunction also called an S-scheme heterojunction without an electron mediator is proposed. It consists of two semiconductors in which one photocatalyst should have both a higher CB position and Fermi energy level than the other one.¹⁷⁹ Therefore, at the contact interface of the two semiconductors, the alignment of the Fermi energy levels leads to the formation of an internal electric field (IEF), which not only offers an additional driving force to combine the photogenerated electrons and holes at low CB and VB levels but also prevents the recombination of the carriers at high redox potentials. Wang et al.¹⁴⁸ fabricated S-scheme CsPbBr₃ QD/Bi₂WO₆ nanosheet (CPB/BWO) photocatalysts for CO₂ reduction (Fig. 9j). The close contact between two semiconductors promotes charge separation and transfer, as the photogenerated electrons on the CB of Bi₂WO₆ can be transferred and recombined with the photogenerated holes on the VB of CsPbBr₃, thus maintaining spatially separated reduced electrons and oxidized holes on the CsPbBr₃ QD and Bi₂WO₆ NS sides, respectively. Compared with pristine CsPbBr₃, the CO yield was increased by a factor of 9.5 for a mass ratio of 1 : 5 of the CsPbBr₃ : Bi₂WO₆ heterostructure. Besides, the TiO₂/CsPbBr₃ S-scheme heterojunction was created by Xu et al. for CO₂ photoreduction (Fig. 9k).¹⁴⁹ Due to the tight contact between TiO₂ and CsPbBr₃ in heterojunctions, an IEF pointing from CsPbBr₃ to TiO₂ is created. The efficiency of photocatalytic CO₂ reduction can be increased by preventing charge recombination and the inverse reaction in TiO₂/CsPbBr₃ heterojunctions using band bending in NHP heterojunctions and related IEFs. Recently, some novel heterojunctions have been fabricated for CO₂ photoreduction, such as Cs₂AgBiBr₆/bismuthine, CsPbBr₃@Pd, 3D/2D NiTiO₃/Cs₃Sb₂I₉, Cs₃Bi₂Br₉/BiVO₄, 0D/2D Cs₃Bi₂Br₉/Bi₂WO₆, etc.^180–184

Overall, the main challenge of NHP photocatalysts is the serious recombination of photogenerated charge carriers, which limits their photocatalytic activity. To solve this problem, researchers have enhanced the separation and transfer of photogenerated charge carriers by constructing heterojunction structures of NHP with conductors, semiconductors, and highly porous materials, such as MOFs. Among them, Schottky junctions and type-II heterojunctions can effectively separate electron–hole pairs and improve the photocatalytic performance of CO₂ reduction. However, this enhancement is at the cost of sacrificing reduction/oxidation capability. To maintain strong redox ability, Z-scheme heterojunctions have attracted widespread attention. Both all-solid-state Z-scheme heterojunctions and direct Z-scheme (S-scheme) heterojunctions can better separate and utilize photogenerated charge carriers through optimized band structures and interfacial electric fields. Therefore, Z-scheme heterojunctions are more promising compared to other types for the enhancement of the photocatalytic efficiency of CO₂ reduction. The experimental findings do not specify which NHPs are suitable for which type of heterojunction. The application scenarios of different types of heterojunctions always depend on reaction redox potentials that need to be satisfied, the band positions of selected co-catalysts (metals, carbon materials, halide perovskites, etc.), and on the matching of the lattice and Fermi energy levels between the two catalyst materials. Furthermore, when designing MHP heterojunctions, other key factors should also be considered, including interface properties, energy levels, lattice parameters, and the interactions between MHP and its counterparts. In situ growth and atomic sharing strategies can greatly eliminate interface defects to form atomic-level intimate contact interfaces, thereby improving the charge transfer efficiency between the two materials.²⁰

3.4 Encapsulation engineering

Due to their ionic structure, most NHPs have serious stability problems under moisture, oxygen, light and at high temperature, which is not favorable for their practical applications in CO₂ photocatalysis.⁷⁹ Therefore, finding approaches to enhance the stability of NHPs is the main direction for further development. Encapsulates are beneficial for the stability of NHPs by reducing the direct contact between the NHPs and the aqueous solution, and can also be regarded as electron/hole acceptors to promote charge carrier transfer. Currently, many kinds of stable functional materials can be used for NHP encapsulation, such as metal oxides, nonmetallic materials, and MOFs.¹⁸⁶ However, considerable research is still needed to reasonably control the type and thickness of the encapsulation layer without significantly weakening the light trapping and reactivity of NHPs.

Amorphous TiO₂ has been considered as a good encapsulation material for the protection of NHPs, and it can also improve the efficiency of photogenerated charge separation to achieve excellent photocatalytic performance. Xu et al.¹¹⁴ reported novel amorphous TiO₂-encapsulated CsPbBr₃ NCs for photocatalytic CO₂ reduction (Fig. 10a). This effect ultimately increases the photoelectron consumption from 25.7 to 193.3 μmol g⁻¹. In addition, CO₂ adsorption and activation are promoted. Based on these synergistic effects, the photoelectron consumption was improved by nearly 6.5 times through the photocatalytic CO₂ reduction reaction. Nonmetallic materials have also been applied to encapsulate NHPs. Su et al.¹⁵⁶ successfully coated thin graphitic diyne (GDY) in situ on CsPbBr₃ NCs and then attached Co²⁺ on it (Fig. 10b). The coated GDY boosts the stability of CsPbBr₃ NCs in an aqueous system and serves as a hole transport layer to speed up the transfer of photogenerated holes into CsPbBr₃. Additionally, it offers a platform that is suitable for Co²⁺ working as catalytically active sites, enabling a photocatalytic activity of up to 27.7 mol g⁻¹ h⁻¹ for the reduction of CO₂ to CO. Another attractive possibility for NHP encapsulation is highly conductive C₆₀. In Zhang's work,¹⁵⁷ the highly conductive C₆₀ acts as an effective electron acceptor, enhancing the electron transfer process, the photocatalytic cycle stability, and solvent stability of the photocatalyst (Fig. 10c). For the end products, CO and CH₄, the C₆₀/CsPbBr₃ composites achieved an electron consumption rate of 90.2 μmol g⁻¹ h⁻¹ in the CAN/H₂O system. In addition to carbon materials, some organics have been used for the encapsulation of NHPs. Chen et al.¹⁶¹ combined CsPbX₃ into a three-dimensional melamine foam (MF), which allowed the composite to exhibit long-term stability in water for 104 h and a CO₂ reduction rate of up to 42.08 mol g⁻¹h⁻¹ (Fig. 10d). Moreover, Li et al.¹⁵⁸ constructed poly(3-hexylthiophene-2,5-diyl) (P3HT)/CsPbBr₃ composites using the P3HT conductive polymer as a protective layer and discovered that it promoted charge separation and improved the stability of CO₂ conversion to CO with CH₄ yields of 145.45 and 23.05 mol g⁻¹h⁻¹ (Fig. 10e).

Fig. 10 (a) Schematic illustration for the structure of CsPbBr₃ encapsulated in amorphous-TiO₂. Reproduced from ref. 114 with permission from Wiley-VCH, Copyright (2018). (b) Schematic illustrations for the structure of CsPbBr₃ encapsulated in GDY-Co. Reprinted with permission from ref. 156, Copyright 2020, American Chemical Society. (c) Schematic illustration of the CO₂ photoreduction process of the C₆₀/CsPbBr₃ composite. Reproduced from ref. 157 with permission from Elsevier, Copyright (2021). (d) Schematics of CsPbBr₃ attached on MF. Reproduced from ref. 161 with permission from Wiley-VCH, Copyright (2021). (e) Schematic illustration of photoreduction of CO₂ over the P3HT/CsPbBr₃ composite. Reproduced from ref. 158 with permission from Elsevier, Copyright (2021). (f) Schematic illustrations for the structure of MAPbI₃ QDs (large spheres) encapsulated in the pores of PCN-221(Fe_x). Reproduced from ref. 185 with permission from Wiley-VCH, Copyright (2019). (g) Schematic illustration of the CO₂ photoreduction process of CsPbBr₃/ZIFs. Reprinted with permission from ref. 117, Copyright 2020, American Chemical Society.

Novel materials such as metal organic frameworks (MOFs) have also been used for NHP encapsulation. MOFs with unique structures and admirable chemical/physical properties have recently attracted much attention in the encapsulation of halide perovskites because of their unique properties such as tunable structure, high surface area, and flexibility. Wu et al.¹⁸⁵ encapsulated MAPbI₃ QDs in an Fe-based MOF (PCN-211(Fe_x)) and successfully used it for CO₂ photoreduction (Fig. 10f). The tight contact between the MAPbI₃ QDs and the Fe catalytic site of PCN-211(Fe_x) facilitated the transfer efficiency of photogenerated electrons from the NHP QDs to the catalytic sites. The best MAPbI₃@PCN-221(Fe_0.2) achieved an excellent total yield (CO and CH₄) of 1559 μmol g⁻¹ with a CH₄ selectivity of 66%. PCN-221(Fe_x) also prevents MAPbI₃ from being hydrolyzed, resulting in extremely high photocatalytic durability (over 80 h). Kong et al.¹¹⁷ reported an in situ growth strategy for the synthesis of zinc/cobalt-based zeolite imidazolyl ester skeletons (ZIF) encapsulated CsPbBr₃ QDs (Fig. 10g). The Co²⁺ center in ZIF can be activated by accepting electrons under light, and further acts as an active center for CO₂ activation and reduction. The synergistic effect of CsPbBr₃ and ZIF coating not only promotes CO₂ capture/activation capacity and charge separation efficiency, but also improves water stability and photocatalytic cycle stability. The ZIF-67 coated NHP with photoabsorption ability has an electron consumption rate of 29.630 μmol g⁻¹ h⁻¹, which is 2.66 times higher than that of the pure CsPbBr₃ photocatalyst.

In brief, effective encapsulation of inorganic halide perovskites (NHPs) is a critical stabilization strategy. Various encapsulation materials, such as amorphous TiO₂, graphitic diyne, C₆₀, melamine foam, and metal–organic frameworks (MOFs), can not only protect the NHPs but also enhance charge separation efficiency and improve the photocatalytic CO₂ reduction performance. The synergistic effects between these encapsulation materials and the NHPs in terms of gas adsorption, charge separation, catalytic activity, and stability are also of great importance. Additionally, these encapsulation materials should also have good electrical conductivity to ensure that the photogenerated charge carriers can reach the surface and react with the substrates. Meanwhile, the challenge of rationally designing the encapsulation layer is to stabilize the NHPs without significantly compromising their light absorption and reactivity, which is crucial for improving the NHPs' photocatalytic CO₂ reduction capability.

4. Charge transport process and surface reaction mechanism

In addition to understanding how to improve the performance of halide perovskites on photocatalytic CO₂ reduction, it is equally important to explore the separation and charge transfer of photogenerated carriers as a prerequisite to promote the efficiency of the final catalytic reaction, because effectively improving the separation efficiency of charge carriers as well as extending their diffusion distance and lifetime are the keys to enhancing photocatalytic efficiency. A deeper understanding of how photoexcited charge carriers are transported, trapped, and recombined will provide insight into the photocatalytic mechanism and inspire us to design high-performance photocatalysts. Here, we divide the mostly used charge carrier characterization techniques into four categories (Fig. 11), namely, charge separation efficiency, charge transfer direction, charge carrier lifetime, and surface reaction intermediate identification. Finally, techniques in each category together with their features are summarized in Table 4.

Fig. 11 Characterization techniques of the charge transport process and surface reaction mechanism.

Table 4 Summary of characterization techniques

Category	Techniques	Features
Charge separation efficiency	Temperature-related photoluminescence	Exciton binding energy
	Steady-state photoluminescence (PL)	Light intensity
	Transient photo-current responses	Photocurrent
	Linear scanning voltammetry (LSV)	Photocurrent
	Electrochemical impedance spectroscopy (EIS)	Radius of Nyquist plots
	Surface photovoltage (SPV)	SPV intensity
Charge transfer direction	In situ irradiation X-ray photoelectron spectroscopy (ISI-XPS)	Electron density changes of various atoms
	Electron spin resonance/electron paramagnetic resonance (ESR/EPR)	Intensity of DMPO-·O2^−/DMPO-·OH signals
	Kelvin probe force microscopy (KPFM)	Contact potential difference (CPD) values
	Density functional theory (DFT)	Work functions (ϕ) and planar-averaged charge density
Charge carrier lifetime	Time-resolved photoluminescence (TRPL)	Photoluminescence decay
	Transient absorption spectroscopy (TAS)	Recovery time
	Fluorescence imaging (FLIM)	Fluorescence decay
	Open circuit photovoltage decay (OCVD)	Photovoltage decay
	Transient state photovoltage (TPV)	Photovoltage decay
Intermediate identification	In situ Fourier Transform Infrared (in situ FTIR)	IR peaks
	Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS)	IR peaks
	In situ attenuated total reflectance (in situ ATR-IR)	IR peaks
	Density functional theory (DFT)	Gibbs free energy (ΔG)

---

4.1 Charge separation efficiency

The absorption of light by the photocatalyst results in the production of excitons, which subsequently separate into electron and hole pairs. The lower binding energy of the exciton means that the carriers can be produced more easily, which increases the carrier separation efficiency. Zhang et al.¹⁸⁷ investigated the activation energies of exciton dissociation for pure Cs₂AgBiBr₆ NCs, and Cs₂AgBiBr₆/Ti₃C₂T_X heterostructures were investigated using temperature-related photoluminescence (temperature-related PL). Their results (Fig. 12a) showed that as the temperature increases, the PL intensity of both the pristine and heterostructured structures become weaker, leading to activation energies of roughly 55 and 39 meV, respectively. This means that the existence of Ti₃C₂T_X in the heterostructure contributes to a lower exciton binding energy and facilitates the formation of free carriers in the Cs₂AgBiBr₆ NC. Compared to temperature-related PL, steady-state photoluminescence (PL) measurement has been more commonly utilized to examine the recombination dynamics of photogenerated charge carriers. The weakening of PL intensity indicates a lower recombination rate of the photogenerated holes and electrons. Wu et al.¹¹⁵ used steady-state PL to investigate the photoexcited electron transfer process in the MAPbI₃@PCN-221 (Fe_x) hybrid catalyst. As shown in Fig. 12b, the PL peak intensities of both pristine PCN-221 and composite MAPbI₃@PCN-221 decreased dramatically upon the addition of Fe ions. This phenomenon indicates that photoexcited electrons are efficiently transferred from PCN-221 and MAPbI₃ QDs to the Fe catalytic sites, which improves the carrier separation efficiency and photocatalytic performance.

Fig. 12 (a) Pseudocolor map of temperature-dependent PL spectra of Cs₂AgBiBr₆/Ti₃C₂T_x heterostructures. Reproduced from ref. 187 with permission from Elsevier, Copyright (2022). (b) Steady-state photoluminescence spectra of PCN-221, PCN-221(Fe_0.2), MAPbI₃@PCN-221, and MAPbI₃@PCN-221(Fe_0.2). Reproduced from ref. 115 with permission from Wiley-VCH, Copyright (2019). (c) LSV curves plots of five catalysts. Reproduced from ref. 188 with permission from Elsevier, Copyright (2021). (d) Transient photo-current responses of CBB, ISS and ISS/CBB. Reproduced from ref. 189 with permission from Elsevier, Copyright (2022). (e) EIS Nyquist plots at a bias of 0.25 V Ag/AgCl under irradiation in the EA containing 0.1 M tetrabutylammonium hexafluorophosphate (TBAPF₆). Reproduced from ref. 190 with permission from Elsevier, Copyright (2022). (f) SPV spectra of CsPbBr₃ and CPB/MS samples. Reproduced from ref. 191 with permission from Elsevier, Copyright (2020).

In addition to PL, different photoelectrochemical tests were also carried out to monitor the charge separation properties of the photocatalytic materials. For instance, Zhang et al.¹⁸⁸ used linear scanning voltammetry (LSV) curves to study the charge separation in CsPbBr₃ (CPB)@Cu-TCPP-x. As shown in Fig. 12c, CPB@Cu-TCPP-x has a much lower overpotential than that of the pristine CPB QD and Cu-TCPP nanosheets, and the CPB@Cu-TCPP-20 composite further reduces the overpotential over the other CPB@TCPP-x composites. The increase in photocurrent density reflects the increased photoinduced charge separation efficiency. In Zhang's work, they investigated the carrier separation behavior of the Cs₃Bi₂Br₉/In₄SnS₈ (CBB/ISS) heterojunction using photocurrent response under visible light irradiation.¹⁸⁹ According to their results (Fig. 12d), the pure CBB and ISS had poor photocurrent responses, but the intensity of the photogenerated current increased greatly for their composite, demonstrating the efficient separation of the photogenerated electron–hole pairs in ISS/CBB. Electrochemical impedance spectroscopy (EIS) is another photoelectrochemical method to evaluate the charge separation efficiency by showing the photogenerated carrier transfer resistance, which can be derived from a Nyquist plot. In general, the small radius of the semicircle in the EIS implies that the electron transport resistance is lower, or that carrier separation is easier. As demonstrated in Fig. 12e, in the Nyquist plots produced by Tian et al.,¹⁹⁰ the diameter of the semicircle for the Cs₂TeCl₆ microcrystals (MCs) was clearly smaller than those for Cs₂HfCl₆ and Cs₂ZrCl₆ MCs, which means that the Cs₂ZrCl₆ MCs possessed a reduced charge-transfer resistance compared to the other two materials.

Another compelling technique to identify charge separation in NHPs is surface photovoltage (SPV) spectroscopy, in which the signal results from variations in surface potential barriers both before and after being exposed to light. The resulting electron–hole pairs can be instantly separated by the built-in electric field, causing a quick SPV response. Normally, greater charge separation efficiency is evidenced by a stronger SPV signal. Fig. 12f displays the SPV spectra of pristine CsPbBr₃ and CsPbBr₃/MoS₂ (CPB/MS) samples measured by Wang et al.¹⁹¹ In contrast to pure CsPbBr₃, CPB/MS manifested as a significantly increased SPV response in the 300–550 nm range. It is clearly supported by the strikingly different SPV signals between the CsPbBr₃ and CPB/MS composites that the latter's visible light-induced charge separation was more effective and that there was spatial charge accumulation/depletion at the interface.

4.2 Charge transfer direction

The separation and transfer of photogenerated charges is the key point that determines the photocatalytic activity of NHPs during the CO₂RR process. Its in-depth study can facilitate the understanding of the charge transfer direction in the semiconductors and the design of better performing photocatalysts. The in situ irradiated X-ray photoelectron spectroscopy (ISI-XPS) technique has been regarded as a competent technique for this because it can detect the density change in electron clouds around different atoms in the ground and excited states, which can directly reveal the direction of charge separation and transfer in heterojunction photocatalysts. Wang et al.¹⁹² applied ISI-XPS to verify the type of heterojunction scheme for CABB (Cs₂AgBiBr₆)@C₃N₄-82% and CABB@C₃N₄-10% (Fig. 13a). Under illumination, there is a slight positive shift in the binding energy of Cs 3d and Ag 3d for CABB@C₃N₄-82% (0.1 eV) and a negative shift in the peak of N 1s (−0.7 eV), indicating a decrease in the electron density on CABB and an increase in the electron density on g-C₃N₄ in the excited state. It can be concluded that after the binding of CABB to g-C₃N₄, the electrons on the conduction band of CABB are transferred to the valence band of g-C₃N₄, which provides direct evidence for the Z-scheme mechanism. The type-II mechanism for CABB@C₃N₄-10% is supported by the opposing trend of the binding energy shifts.

Fig. 13 (a) High-resolution XPS of CABB@C₃N₄-10% and CABB@C₃N₄-82% in the dark or under 365 nm LED irradiation. Reproduced from ref. 192 with permission from Elsevier, Copyright (2020). (b) KPFM image of CsPbBr₃-Au in the dark and under light irradiation at 630 nm. Reproduced from ref. 193 with permission from Elsevier, Copyright (2021). (c) Calculated work functions of CABB and BWO, and the planar-averaged electron density difference Δρ and the charge density difference of CABB/BWO. Reproduced from ref. 110 with permission from Elsevier, Copyright (2022). (d) ESR spectra of DMPO-·O₂⁻ and DMPO-·OH in the presence of CsPbBr₃ QDs, MTB and the CsPbBr₃@MTB hybrid. Reproduced from ref. 155 with permission from Elsevier, Copyright (2021).

Another convincing technique to confirm the electron transfer direction in the heterostructure is electron spin resonance (ESR), also called electron paramagnetic resonance (EPR), which usually uses 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) as a radical trap to capture the oxygen radical (·O₂⁻) and hydroxyl radical (·OH). Dong et al.¹⁵⁵ confirmed the carrier transfer direction in CsPbBr₃@MTB materials by ESR. As shown on the left of Fig. 13d, the DMPO⁻·O₂⁻ signal intensity of the CsPbBr₃@MTB composite is much stronger than that of the pristine CsPbBr₃ and MTB, indicating that more photogenerated electrons are accumulated in the CB of CsPbBr₃. The right side of Fig. 13d shows that the DMPO⁻·OH signal intensity of the CsPbBr₃@MTB composite is much stronger than that of pure MTB, which implies that the light-generated holes are accumulated on the VB of MTB. Thus, the above two conclusions demonstrate that the photogenerated electrons are transferred to CsPbBr₃ and the photogenerated holes are transferred to MTB, thus proving that the CsPbBr₃@MTB composite is an S-scheme heterojunction.

Besides, Kelvin probe force microscopy (KPFM) is an interesting improvement to Atomic Force Microscopy (AFM) and SPV, which is capable of imaging surface morphology and surface potential simultaneously with nanoscale spatial resolution and sub-millivolt electrical resolution. Therefore, it can be used to confirm the electron transfer direction. In Liao's work, they compared the contact potential difference (CPD) images between pure CsPbBr₃ and the CsPbBr₃-Au composite.¹⁹³ As shown in Fig. 13b, pure CsPbBr₃ cannot be excited by light, since there is no significant difference between CPD images collected in the dark and under light. In contrast, the CPD values of the CsPbBr₃-Au composites decrease upon light illumination, which suggests the accumulation of electrons on the surface. Through this, they inferred that the hot electrons generated in the Au nanoparticles can be effectively transferred to CsPbBr₃.

Aiming to obtain further evidence of the charge separation direction, Wu et al. used density functional theory (DFT) to calculate the work functions (ϕ) of Cs₂AgBiBr₆ (CABB) and Bi₂WO₆ (BWO).¹¹⁰ The calculated outcomes (Fig. 13c) demonstrate that CABB has a higher E_f than BWO, which suggests that some electrons will automatically migrate from CABB to BWO at the heterojunction interface until their E_f coincides. Additionally, the generation of heterojunctions will result in the reallocation of electrons at the interface, according to the predicted planar average charge density difference of CABB/BWO. Green represents the electron reduction on the CABB side, whereas yellow represents the electron buildup on the BWO side. As a result, a solid IEF and band bending will form because CABB and BWO will be positively and negatively charged, respectively.

4.3 Charge carrier lifetime

Exploring the interfacial charge transfer of heterojunction photocatalysts is important because it is a prerequisite to show the final catalytic reaction capability, which will provide insight into the photocatalytic mechanism and facilitate the design of high-performance photocatalysts. Since the time scale of charge kinetics in photocatalytic processes varies from femtoseconds to seconds, a series of techniques are applied to investigate the charge transfer kinetics and carrier lifetime inside the photocatalyst as well as at the interface.¹⁹⁵

Time-resolved photoluminescence (TRPL) spectroscopy is the most widely used technique to investigate excited charge carrier recombination in photocatalysts. After a photocatalyst is excited, electrons and holes will be recombined, leading to photogenerated carrier decay. Carrier decay kinetics can be obtained by extrapolating the carrier density changes directly from the time-resolved PL intensity measured using laser excitation pulses or fast electron techniques. TRPL decay is commonly used in multi-exponential models to extract the lifetimes of different processes. Zhang et al.¹⁵⁷ tested the TRPL of pristine CsPbBr₃ and the C₆₀/CsPbBr₃ composite. The normalized TRPL decay plots confirmed the C₆₀/CsPbBr₃ composite's swift charge separation, as seen in Fig. 14a. The decay curves were fitted using triexponential decay kinetics, and the average lifetime for the C₆₀/CsPbBr₃ composite was 6.1 ns, a substantial drop from the 15.8 ns of pure CsPbBr₃. The C₆₀/CsPbBr₃ composite's noticeably faster decay can be attributed to C₆₀'s ability to receive electrons, which acted as an electron reservoir and promoted the charge transfer from CsPbBr₃ to C₆₀. Different from TRPL, Fluorescence Lifetime Imaging Microscopy (FLIM) is a fluorescence imaging technique. Instead of using the emission spectra of the fluorophores, it bases its contrast on their lifetime. The average time a molecule spends in the excited state before it returns to its ground state and emits a photon is known as the fluorescence lifespan. To determine the PL lifetimes of the excited species, Laishram et al.¹⁹⁴ measured FLIM lifetimes on g-C₃N₄ (CN) and CsPbBr₃ NCs with monolayer sheet (CNM) samples. The tri-exponential fitting of the FLIM lifespan curves resulted in the average lifetimes (τ_ave) of the CN and CNM samples being determined to be 1.14 and 1.72 ns, respectively, as shown in Fig. 14b. They attributed the increase in the average lifetime value of CNM to the increase in intrasheet order and the apparent suppression of intrasheet recombination after the conversion of bulk CN into monolayers. It is evident from the increased average lifetime value that transformation of bulk sheets into monolayer sheets increases charge transport on the conjugated CNM sheets, resulting in a more efficient charge separation.

Fig. 14 (a) Time-resolved PL decay spectra of CsPbBr₃ NCs and the C₆₀/CsPbBr₃ composite. Reproduced from ref. 157 with permission from Elsevier, Copyright (2021). (b) Fluorescence lifetime imaging (FLIM) lifetime decay curve of CN (black) and CNM (blue). Reproduced from ref. 194 with permission from Elsevier, Copyright (2022). (c and d) V_oc transient rise/decay obtained during excitation/termination of visible light irradiation and average electron lifetimes (τ_n) obtained from transient OCVD measurements. Reproduced from ref. 127 with permission from Elsevier, Copyright (2021). (e–g) TA recovery dynamic plots of CsPbBr₃ and CsPbBr₃@ZIFs. Excitation wavelength: 400 nm. Reprinted with permission from ref. 117, Copyright 2018, American Chemical Society. (h) TPV spectra of CsPbBr₃ and CPB/MS samples. Reproduced from ref. 146 with permission from Elsevier, Copyright (2020).

In general, a TRPL instrument is only capable of monitoring emitting materials, which usually does not provide a complete picture of the charge carriers. In contrast, transient absorption spectroscopy (TAS) can monitor both emitting and non-emitting (i.e., trapped) substances, providing more information to model the operation of the photocatalytic system by measuring the change in absorbance or transmittance over time before and after sample excitation. The TAS of CsPbBr₃@ZIF-8 and CsPbBr₃@ZIF-67 in comparison to pure CsPbBr₃ was measured by Kong et al.¹¹⁷ Absorption change (ΔA) curves plotted against wavelength and delay time are shown in Fig. 14e–g. ZIF-8 and ZIF-67 exhibit very little absorption at a pump pulse of 400 nm, which allows CsPbBr₃ to be stimulated selectively. The ground state bleaching (GSB) signal of CsPbBr₃ is believed to have a unique negative absorption peak at 513 nm that is present in unmodified CsPbBr₃. In CsPbBr₃@ZIFs, bleaching behavior was more quickly seen, indicating an efficient interfacial electron transport between CsPbBr₃ and ZIF shells.

Additionally, photovoltage decay can be used for determining the lifetime of photogenerated electrons. For example, in Xi's work, open circuit photovoltage decay (OCVD) measurement was performed with an electrochemical station under 300 W Xe lamp irradiation.¹²⁷ The average electron lifetime of CsPbBr₃ is noticeably shorter than that of a corresponding Ni-based metal–organic framework (NMF)/CPB, as can be seen in Fig. 14c and d. This suggests that the addition of NMF reduces the recombination of photogenerated charge carriers and lengthens the lifetimes of electrons. In transient photovoltage (TPV), the semiconductor under testing initially remains disconnected from the circuit under steady-state illumination. With the application of an additional short light pulse, the semiconductor is photoexcited to produce carriers. The free electrons migrating to the material surface are collected at the electrode to obtain the photovoltage. The lifetime of the photocarriers can be determined by detecting the decay of the photovoltage. Fig. 14h shows the TPV spectra of Wang's work.¹⁴⁶ Both pure CsPbBr₃ and CsPbBr₃/MoS₂ (CPB/MS) samples exhibit a positive TPV response, but the larger response and longer lifetime of CPB/MS suggests that CsPbBr₃ is the source of the photogenerated holes that accumulate on the surface of the test electrode. The larger TPV response of CPB/MS suggests a broader space charge zone for the charge diffusion process.

4.4 Surface reaction intermediate identification

Designing effective photocatalysts will benefit from a thorough understanding of the unique active sites and associated reaction mechanisms of NHP NCs. Even though the photocatalytic process of CO₂ reduction has been extensively studied, the CO₂ reduction reaction pathways on NHP NCs remain unclear. A deeper understanding of the photocatalytic mechanism and the intermediate species involved in the CO₂ photoreduction process are required to resolve these challenges. Fourier transform infrared (FTIR) spectroscopy is considered a semi-quantitative tool for studying the molecular structure of materials using infrared radiation. This is due to the different vibrational frequencies of the covalent bonds within molecules, resulting in their ability to absorb different wavelengths of light and thus create different infrared spectra.

Several derived characterization methods, such as in situ FTIR, diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS), and in situ Attenuated Total Reflection Infrared (in situ ATR-IR) spectroscopy, have been used to identify intermediates involved in photocatalytic CO₂ reduction. DFT theoretical calculations in conjunction with advanced in situ characterization techniques will be more useful in determining the precise photocatalytic mechanism and offer helpful information for designing NHPs for CO₂ photocatalytic reduction. For example, Ding et al.¹⁶³ performed in situ FTIR measurements on the Cs₂AgBiBr₆/UIO-66 composite in order to detect reaction intermediates in the CO₂ photoreduction process. As shown in Fig. 15a, no intermediate peaks were seen prior to the illumination, but as soon as the illumination began, a new set of IR peaks emerged, including CH₃O* (1140 cm⁻¹), , COOH* (1650 cm⁻¹), and , and their intensity increased with time. The results suggest that the primary reaction intermediates in the photocatalytic conversion of CO₂ to CO are formate species. In order to detect and monitor adsorbed materials and reaction intermediates on the surface of 10% Bi₃O₄Br–3% Ni-doped CsPbBr₃ (10BC-3N) and CsPbBr₃ (CPB) in the dark and under full-spectrum light irradiation, Wang et al. used DRIFTS measurements (Fig. 15b) to identify the surface reactant intermediates.¹⁵³ They discovered that the –CO₂⁻ species matches the wavenumbers that appear at 1308 and 1693 cm⁻¹. The essential intermediate in the multi-step reduction of CO₂ to CO or CH₄ and the cause of the peak at 1510 cm⁻¹ is COOH*. As indicated by the wave number at 1745 cm⁻¹, some CO may be created during the irradiation process. Additionally, numerous chemical intermediates were discovered in the high wave number region CH₃O* (2853 cm⁻¹), , and . For in situ-ATR analysis, Choi et al.¹⁹⁶ reported that different characteristics of intermediate formation were evaluated for each catalyst. The photocatalytic reduction process of gaseous CO₂ employing perovskite photocatalysts with copper carriers results in IR distinctive peaks (Fig. 15c). At 1532 cm⁻¹ (asymmetric) and 1500 cm⁻¹ (symmetric), significant C₂ pathway intermediates *OCCO and *OCCOH were both visible at 3080 cm⁻¹ (olefins), 2987 cm⁻¹ (asymmetric –CH₃ alkanes), 2924 cm⁻¹ (asymmetric –CH₂ alkanes), and 2876 cm⁻¹ (symmetric –CH₃ alkanes). For the synthesis of C₂H₄, the olefin peak at 3080 cm⁻¹ is particularly significant evidence. At 1495 cm⁻¹, 1414 cm⁻¹, and 1340 cm⁻¹, adsorbed carbonate species derived from pure CO₂ gas were noted. In addition to characterization, DFT calculations can also be applied to predict the CO₂ photoreduction mechanism. By computing the associated Gibbs free energy (G) at each stage of the CO₂ photoreduction process, Sheng's study on Cs₂CuBr₄ corroborated the hypothesized CO₂ photoreduction pathway (Fig. 15d).¹³³ The figure depicts the creation of COOH* by a high energy input internal energy process on the surface of Cs₂CuBr₄ and CsPbBr₃ QDs. As a result, the rate-limiting step is the creation of COOH* intermediates from ˙CO₂⁻. However, Cs₂CuBr₄ has a stronger photoreduction capacity than CsPbBr₃, as evidenced by the fact that it requires less reaction energy (0.48 eV) to convert ˙CO₂⁻ to COOH* than CsPbBr₃ (1.61 eV) does. The simulations also show that the CsPbBr₃-produced CO* intermediate prefers to undergo energetically advantageous desorption because of CO rather than additional hydrogenation to CH₄.

Fig. 15 (a) In situ FTIR spectra of 20CABB/UIO-66 under CO₂ and H₂O vapor after photo-irradiation for 0, 5, 10, 15, 20, and 25 min. Reproduced from ref. 163 with permission from Elsevier, Copyright (2022). (b) In situ DRIFTS measurements for CO₂ and H₂O interaction with 10BC-3N and CPB. Reproduced from ref. 153 with permission from Elsevier, Copyright (2022). (c) In situ ATR-IR spectra of photocatalytic CO₂ reduction on the CsPbBr₃ perovskite photocatalyst embedded in a porous copper scaffold. Reproduced from ref. 196 with permission from Elsevier, Copyright (2021). (d) Calculated free energy of the main reactions in photocatalytic CO₂ reduction for Cs₂CuBr₄ and CsPbBr₃ PQDs. Reprinted with permission from ref. 133, Copyright 2018, American Chemical Society.

5. Challenges and prospects

During the past few decades, many scientists have made great progress in the field of photocatalytic CO₂ reduction using NHPs. However, it should be realized that the practical applications of NHPs still face several challenges, such as heavy metal (Pb) contamination, stability issues, low catalytic activity and selectivity, and unclear mechanism of photocatalysis. The design of novel efficient NHP photocatalysts is still in progress, and researchers are required to focus on addressing these obstacles.

5.1 Environmental impact

Even modest amounts of Pb can cause major environmental and health issues; therefore its toxicity cannot be disregarded. Currently, lead halide perovskites are the most widely used photocatalysts. Its toxicity unquestionably prevents its commercial use and the mass production of NHP photocatalysts. To address this issue, some progress has been made in exploring lead-free NHPs for photocatalytic CO₂ reduction, such as various Pb-free transition metals including Ge, Ag, In, Sn, Sb, and Bi.^197–199 In addition, bimetallic repetition has been developed due to the synergistic effect of bimetals. Noticeably, most of the current Pb-free NHPs have lower photocatalytic efficiency than Pb-based ones.²⁰⁰ In future studies, lead-free perovskite materials remain a promising strategy to reduce lead leakage by using different optimization strategies as mentioned above such as surface engineering (size, morphology, defects, etc.), component engineering (trying new metal cations or combining different non-lead metal cations) and heterojunction engineering (2D/3D structures) to improve the photocatalytic performance of lead-free perovskite materials. In addition, the physical encapsulation of lead-containing perovskites can slow down the leakage of lead, for example, TiO₂, organic materials, and MOFs, but the lead compounds will still leak into the environment after long-term operation. Therefore, lead capture after photocatalytic reactions is an effective way to reduce the leakage of lead into the environment. One of the effective methods is chemisorption, in which some functional materials with multiple functional groups, including thiols, phosphate groups, etc., are bonded to capture the mobile Pb²⁺. This approach can lead to better protection against Pb leakage because it has superior ability to reduce Pb leakage compared to physical encapsulation.

5.2 Stability

The industrialisation of NHP photocatalysts is additionally hampered by stability issues brought on by heat, humidity, UV radiation, self-separation, or repetitive catalytic cycles. The general strategy to increase the stability of NHP is based on effects of chemical or physical bonding. Using encapsulation techniques with a core–shell structure that is almost entirely covered can lessen NHP's direct interaction with the environment. The stability of NHPs can be increased by efficiently preventing polar solvents from coming into direct contact with the NHP core through a densely crystalline protective shell or replacing polar solvents with low-polarity solvents such as ethyl acetate, acetonitrile, and toluene. Besides using encapsulation and low-polarity solvents, the stability of NHPs can also be enhanced by changing the elements, surface ligands and morphology of the NHPs. For example, the intrinsic stability of NHPs can be significantly increased by using mixed cation methods, such as Cs_xFA_1−xPbI₃ (ref. 201). The lattice–ion interactions can be improved by B-site metal ion doping, leading to a more stable crystal structure. NHPs can also be made more stable by partially substituting smaller cations (Mn²⁺ or Zn²⁺) for Pb²⁺. Recently, researchers have started using thiocyanate (SCN⁻) and formate (HCOO⁻) ions to synthesize stable pseudo-halide perovskite materials for solar cell applications. Compared to halide perovskites, these pseudo-halide perovskites exhibit better stability at room temperature. This is because the SCN⁻ or HCOO⁻ anions occupy the axial positions of the lead halide octahedra, leading to an asymmetric electronic structure that helps maintain the desired structural framework. Furthermore, two-dimensional NHPs can be constructed by introducing large organic cations into the A-site. The hydrophobicity of these organic cations can be leveraged to effectively block external environmental factors, significantly improving the stability of the NHPs.²⁰² In addition, employing specific ligands can also enhance the stability of NHPs. For instance, alkyl phosphinic acid and alkylammonium halide ligands can passivate the damaged surface of NHPs and form a protective layer. These ligands can modify the NHPs and improve their stability during photocatalytic reactions.²⁰ In conclusion, comprehensive material design strategies are key to improving the stability of NHPs. Through the combination of these approaches, the long-term stability of NHP photocatalysts can be achieved, which is crucial for their practical development.

5.3 Catalytic efficiency and selectivity

The photocatalytic performance of current NHPs for the photocatalytic reduction of CO₂ is not sufficient for future practical applications, so further improvements in photocatalytic efficiency and product selectivity are needed. Numerous features have been thoroughly explored, including how to increase the specific surface area, add additional active sites, encourage the separation of photogenerated electron–hole pairs, and prevent charge recombination. For instance, co-catalysts (such as metals, MXenes, carbon materials, etc.) can be added to speed up the rate of the surface catalytic reaction, and new heterojunctions (such as Schottky junctions, and type II, all-solid-state Z-scheme, and S-scheme heterojunctions) can be built to improve charge separation. Perovskites can also be modified in size and morphology to increase the surface reaction of active sites. Additionally, to enhance catalytic activity, synergistic external field catalysis (such as thermal, electronic, magnetic, and biological enzymes) can be used. In terms of product selectivity, the ideal products for photocatalytic CO₂ reduction are chemical feedstocks and fuels (e.g., unsaturated hydrocarbons or alcohols). However, the selectivity of CO₂ reduction at present is usually directed towards CO and CH₄. Therefore, the design of NHPs with additional economic benefits for multi-carbon hydrocarbon products is the key to addressing the above challenges in the future. As for NHPs, the selectivity of photocatalytic CO₂ to CH₄ can be improved by tuning the photoexcitation properties, band structure (B-site element: Bi, Sn or In), charge carrier separation, and catalytically active sites (Fe, Co, Zn, and Cu). To improve the selectivity of photocatalytic CO₂ reduction towards C₂₊ products, the following strategies can be adopted to address the limitations of photocatalytic systems: (1) introduce metal co-catalysts such as Au, Ag, and Cu on the surface of NHP photocatalysts to promote C–C coupling reactions; (2) prepare NHP photocatalysts with suitable internal voids or hollow structures to increase the spatial sites for C–C coupling; (3) optimize the crystal phase and facet orientation of the photocatalysts since certain facets are more favorable for C–C bond formation compared to others; (4) optimize reaction conditions like temperature, pH, and light intensity (high photon flux) to influence C–C coupling; (5) construct more effective heterojunctions to regulate the distribution of photogenerated electrons on co-catalysts and enhance the driving force for C–C coupling by minimizing charge recombination;⁴¹ (6) doping or surface modification to modulate the electronic structure and surface properties of photocatalysts, e.g., introducing nitrogen-containing functional groups to enhance CO₂ adsorption, can also benefit C–C bond formation; (7) integrate NHP photocatalysis with electrochemical processes, such as photoelectrochemical, photothermal, photomagnetic, and photobiosynthetic catalysis, to provide stronger reducing potentials for the generation of C₂₊ products.²⁰³ By comprehensively applying these innovative strategies, the selectivity and efficiency of photocatalytic CO₂ reduction towards C₂₊ products can be further improved.

5.4 Catalytic mechanism

A deeper comprehension of the photocatalytic mechanism is necessary for a more logical design of NHP photocatalysts. The chemical pathway of CO₂ reduction on NHPs is still unknown, despite substantial research into the photocatalytic mechanism of CO₂ reduction on a variety of metals and semiconductors. More fundamental research on the photocatalytic mechanism is required to identify the intermediate species involved in CO₂ reduction and provide answers to these problems. DFT calculations allow the calculation of various properties such as crystal structures, band structures, effective masses of electrons and holes, and intermediate products. Therefore, more studies combining theoretical DFT calculations with experimental validation are needed. Moreover, more advanced in situ characterization techniques need to be developed to gain an in-depth understanding of the detailed photocatalytic mechanism and charge carrier separation, as well as provide useful information for the design of NHP photocatalysts for CO₂ reduction.

In conclusion, using NHPs for photocatalytic CO₂RR is a highly effective technique to deal with the problems of the greenhouse effect and the energy crisis at the same time. However, the CO₂RR performance of NHP-based catalysts is severely constrained by strong carrier recombination, poor product selectivity, and instability in humid environments. Therefore, our review summarized critical factors that influence the final product selectivity, strategies that have been implemented to enhance the photocatalytic performance of NHPs, and techniques to characterize the charge transport process and CO₂RR mechanism. Finally, some challenges and prospects are introduced to indicate the direction of future research. With the rapid development of this fascinating field, we deem that this review could provide crucial guidance for the rational design of efficient photocatalytic systems and conquer the decisive problems encountered in the field of NHP photocatalysts.

Data availability

The data that support the findings of this study are available on request from the corresponding authors, D.-J. Lee and H.-Y. Hsu.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors acknowledge financial support from the Research Grants Council of Hong Kong (grant no. CityU 21203518 and F-CityU106/18), Innovation and Technology Commission (grant no. MHP/104/21), Shenzhen Science Technology and Innovation Commission (grant no. JCYJ20210324125612035, R-IND12303 and R-IND12304), City University of Hong Kong (grant no. 9360140, 7005289, 7005580, 7005720, 9667213, 9667229, 9680331 and 9678291), National Natural Science Foundation of China (51901119, 22071070, 61874165, and 21833009), and Major State Basic Research Development Program of China (2019YFB1503401).

References

1. S. C. Roy, O. K. Varghese, M. Paulose and C. A. Grimes, ACS Nano, 2010, 4, 1259–1278.
2. X. Fu, T. He, S. Zhang, X. Lei, Y. Jiang, D. Wang, P. Sun, D. Zhao, H.-Y. Hsu and X. Li, Chem, 2021, 7, 3131–3143.
3. C. Hepburn, E. Adlen, J. Beddington, E. A. Carter, S. Fuss, N. Mac Dowell, J. C. Minx, P. Smith and C. K. Williams, Nature, 2019, 575, 87–97.
4. A. Tabish, A. M. Varghese, M. A. Wahab and G. N. Karanikolos, Catalysts, 2020, 10, 95.
5. A. A. Khan and M. Tahir, J. CO2 Util., 2019, 29, 205–239.
6. S. Xie, Q. Zhang, G. Liu and Y. Wang, Chem. Commun., 2016, 52, 35–59.
7. G. Zhang, G. Liu, L. Wang and J. T. Irvine, Chem. Soc. Rev., 2016, 45, 5951–5984.
8. S. Rao, X. Zou, S. Wang, T. Shi, Y. Lu, L. Ji, H.-Y. Hsu, Q. Xu and X. Lu, J. Electrochem. Soc., 2019, 166, D427.
9. H. Wu, T. H. Tan, R. Liu, H.-Y. Hsu and Y. H. Ng, Sol. RRL, 2021, 5, 2000423.
10. A. Saravanan, D.-V. N. Vo, S. Jeevanantham, V. Bhuvaneswari, V. A. Narayanan, P. Yaashikaa, S. Swetha and B. Reshma, Chem. Eng. Sci., 2021, 236, 116515.
11. B. S. Kwak, K. Vignesh, N.-K. Park, H.-J. Ryu, J.-I. Baek and M. Kang, Fuel, 2015, 143, 570–576.
12. F. Iqbal, B. Abdullah, H. Oladipo, M. Yusuf, F. Alenazey, T. D. Nguyen and M. Ayoub, Nanostructured Photocatalysts, 2021, pp. 519–540.
13. J. Mora-Hernandez, A. M. Huerta-Flores and L. M. Torres-Martinez, J. CO2 Util., 2018, 27, 179–187.
14. G. Yin, H. Abe, R. Kodiyath, S. Ueda, N. Srinivasan, A. Yamaguchi and M. Miyauchi, J. Mater. Chem. A, 2017, 5, 12113–12119.
15. W. Gao, S. Li, H. He, X. Li, Z. Cheng, Y. Yang, J. Wang, Q. Shen, X. Wang and Y. Xiong, Nat. Commun., 2021, 12, 4747.
16. X. Chen, C. Li, M. Grätzel, R. Kostecki and S. S. Mao, Chem. Soc. Rev., 2012, 41, 7909–7937.
17. M. Halmann, Nature, 1978, 275, 115–116.
18. T. Inoue, A. Fujishima, S. Konishi and K. Honda, Nature, 1979, 277, 637–638.
19. J. S. Zhao, Y. F. Mu, L. Y. Wu, Z. M. Luo, L. Velasco, M. Sauvan, D. Moonshiram, J. W. Wang, M. Zhang and T. B. Lu, Angew. Chem., 2024, e202401344.
20. J. Wang, Y. Shi, Y. Wang and Z. Li, ACS Energy Lett., 2022, 7, 2043–2059.
21. J. L. White, M. F. Baruch, J. E. Pander III, Y. Hu, I. C. Fortmeyer, J. E. Park, T. Zhang, K. Liao, J. Gu and Y. Yan, Chem. Rev., 2015, 115, 12888–12935.
22. Q. A. Akkerman, G. Rainò, M. V. Kovalenko and L. Manna, Nat. Mater., 2018, 17, 394–405.
23. X. Song, G. Wei, J. Sun, C. Peng, J. Yin, X. Zhang, Y. Jiang and H. Fei, Nat. Catal., 2020, 3, 1027–1033.
24. Y. Wu, P. Wang, X. Zhu, Q. Zhang, Z. Wang, Y. Liu, G. Zou, Y. Dai, M. H. Whangbo and B. Huang, Adv. Mater., 2018, 30, 1704342.
25. Y. Tang, C. H. Mak, R. Liu, Z. Wang, L. Ji, H. Song, C. Tan, F. Barrière and H. Y. Hsu, Adv. Funct. Mater., 2020, 2006919.
26. C. A. Richard, Z. Pan, H.-Y. Hsu, S. Cekli, K. S. Schanze and J. R. Reynolds, ACS Appl. Mater. Interfaces, 2014, 6, 5221–5227.
27. L. Fu, K. Fu, X. Gao, S. Dong, B. Zhang, S. Fu, H.-Y. Hsu and G. Zou, Anal. Chem., 2021, 93, 2160–2165.
28. Y. Tang, C. H. Mak, C. Wang, Y. Fu, F.-F. Li, G. Jia, C.-W. Hsieh, H.-H. Shen, J. C. Colmenares, H. Song, M. Yuan, Y. Chen and H.-Y. Hsu, Small Methods, 2022, 6, 2200326.
29. L. Liu, H. Huang, Z. Chen, H. Yu, K. Wang, J. Huang, H. Yu and Y. Zhang, Angew. Chem., Int. Ed., 2021, 60, 18303–18308.
30. Y. Lin, J. Guo, J. San Martin, C. Han, R. Martinez and Y. Yan, Chem.–Eur. J., 2020, 26, 13118–13136.
31. F. Zhong, Y. He, Y. Sun, F. Dong and J. Sheng, J. Mater. Chem. A, 2022, 10, 22915–22928.
32. E. A. Tsiwah, Y. Ding, Z. Li, Z. Zhao, M. Wang, C. Hu, X. Liu, C. Sun, X. Zhao and Y. Xie, CrystEngComm, 2017, 19, 7041–7049.
33. D. M. Jang, D. H. Kim, K. Park, J. Park, J. W. Lee and J. K. Song, J. Mater. Chem. C, 2016, 4, 10625–10629.
34. Z. Long, H. Ren, J. Sun, J. Ouyang and N. Na, Chem. Commun., 2017, 53, 9914–9917.
35. S. Javaid, X. Xu, W. Chen, J. Chen, H.-Y. Hsu, S. Wang, X. Yang, Y. Li, Z. Shao, F. Jones and G. Jia, Nano Energy, 2021, 89, 106463.
36. L. Protesescu, S. Yakunin, O. Nazarenko, D. N. Dirin and M. V. Kovalenko, ACS Appl. Nano Mater., 2018, 1, 1300–1308.
37. X. Fan, L. Zhang, M. Wang, W. Huang, Y. Zhou, M. Li, R. Cheng and J. Shi, Appl. Catal., B, 2016, 182, 68–73.
38. Y. Wang, J. Wang, M. Zhang, S. Zheng, J. Wu, T. Zheng, G. Jiang and Z. Li, Small, 2023, 19, 2300841.
39. W. Song, K. C. Chong, G. Qi, Y. Xiao, G. Chen, B. Li, Y. Tang, X. Zhang, Y. Yao and Z. Lin, J. Am. Chem. Soc., 2024, 146(5), 3303–3314.
40. N. Li, X. Chen, J. Wang, X. Liang, L. Ma, X. Jing, D.-L. Chen and Z. Li, ACS Nano, 2022, 16, 3332–3340.
41. J. Albero, Y. Peng and H. García, ACS Catal., 2020, 10, 5734–5749.
42. Z. Wu, H. Tüysüz, F. Besenbacher, Y. Dai and Y. Xiong, Nanoscale, 2023, 15, 5598–5622.
43. S. Shyamal and N. Pradhan, J. Phys. Chem. Lett., 2020, 11, 6921–6934.
44. M. Tahir and N. S. Amin, Energy Convers. Manage., 2013, 76, 194–214.
45. R. Cheng, C.-C. Chung, S. Wang, B. Cao, M. Zhang, C. Chen, Z. Wang, M. Chen, S. Shen and S.-P. Feng, Mater. Today Phys., 2021, 17, 100358.
46. Y. Wei, Z. Cheng and J. Lin, Chem. Soc. Rev., 2019, 48, 310–350.
47. Q. Chen, Y. Ma, L. Wang, X. Lan and J. Shi, Sol. RRL, 2021, 5, 2000755.
48. Y.-F. Xu, M.-Z. Yang, B.-X. Chen, X.-D. Wang, H.-Y. Chen, D.-B. Kuang and C.-Y. Su, J. Am. Chem. Soc., 2017, 139, 5660–5663.
49. J. Zhu, Y. Zhu, J. Huang, L. Hou, J. Shen and C. Li, Nanoscale, 2020, 12, 11842–11846.
50. Z.-C. Kong, H.-H. Zhang, J.-F. Liao, Y.-J. Dong, Y. Jiang, H.-Y. Chen and D.-B. Kuang, Sol. RRL, 2020, 4, 1900365.
51. M. Ou, W. Tu, S. Yin, W. Xing, S. Wu, H. Wang, S. Wan, Q. Zhong and R. Xu, Angew. Chem., 2018, 130, 13758–13762.
52. H. Huang, J. Zhao, Y. Du, C. Zhou, M. Zhang, Z. Wang, Y. Weng, J. Long, J. Hofkens and J. A. Steele, ACS Nano, 2020, 14, 16689–16697.
53. J. Wang, M. Zhang, Z. Chen, L. Li, G. Jiang and Z. Li, ACS Energy Lett., 2024, 9, 653–661.
54. M. A. Peña and J. L. G. Fierro, Chem. Rev., 2001, 101, 1981–2018.
55. Y. Zhao and K. Zhu, Chem. Soc. Rev., 2016, 45, 655–689.
56. Z. Yi, N. H. Ladi, X. Shai, H. Li, Y. Shen and M. Wang, Nanoscale Adv., 2019, 1, 1276–1289.
57. M. A. Raza, F. Li, M. Que, L. Zhu and X. Chen, Mater. Adv., 2021, 2, 7187–7209.
58. J. Tian, Q. Xue, Q. Yao, N. Li, C. J. Brabec and H. L. Yip, Adv. Energy Mater., 2020, 10(23), 2000183.
59. F. Igbari, Z.-K. Wang and L.-S. Liao, Adv. Energy Mater., 2019, 9, 1803150.
60. F. Zhou, Z. Li, H. Chen, Q. Wang, L. Ding and Z. Jin, Nano Energy, 2020, 73, 104757.
61. M. I. Saidaminov, O. F. Mohammed and O. M. Bakr, ACS Energy Lett., 2017, 2, 889–896.
62. Q. Akkerman, A. Abdelhady and L. Manna, J. Phys. Chem. Lett., 2018, 9(9), 2326–2337.
63. R. Liu, C. H. Mak, X. Han, Y. Tang, G. Jia, K.-C. Cheng, H. Qi, X. Zou, G. Zou and H.-Y. Hsu, J. Mater. Chem. A, 2020, 8, 23803–23811.
64. K. A. Huynh, D. L. T. Nguyen, V.-H. Nguyen, D.-V. N. Vo, Q. T. Trinh, T. P. Nguyen, S. Y. Kim and Q. V. Le, J. Chem. Technol. Biotechnol., 2020, 95, 2579–2596.
65. A. Feng, X. Jiang, X. Zhang, X. Zheng, W. Zheng, O. F. Mohammed, Z. Chen and O. M. Bakr, Chem. Mater., 2020, 32, 7602–7617.
66. S. González-Carrero, R. E. Galian and J. Pérez-Prieto, Part. Part. Syst. Charact., 2015, 32, 709–720.
67. F. Temerov, Y. Baghdadi, E. Rattner and S. Eslava, ACS Appl. Energy Mater., 2022, 5, 14605–14637.
68. X. Wang, T. Zhang, Y. Lou and Y. Zhao, Mater. Chem. Front., 2019, 3, 365–375.
69. S. P. Santoso, S.-P. Lin, T.-Y. Wang, Y. Ting, C.-W. Hsieh, R.-C. Yu, A. E. Angkawijaya, F. E. Soetaredjo, H.-Y. Hsu and K.-C. Cheng, Int. J. Biol. Macromol., 2021, 175, 526–534.
70. C. H. Mak, Y. Peng, M. H. Chong, L. Yu, M. Du, L. Ji, X. Zou, G. Zou, H.-H. Shen and S. P. Santoso, J. Mater. Chem. C, 2023, 11, 11303–11311.
71. H.-Y. Hsu, H.-H. Hsieh, H.-Y. Tuan and J.-L. Hwang, Sol. Energy Mater. Sol. Cells, 2010, 94, 955–959.
72. C.-N. Wu, L.-C. Sun, Y.-L. Chu, R.-C. Yu, C.-W. Hsieh, H.-Y. Hsu, F.-C. Hsu and K.-C. Cheng, Food Chem., 2020, 330, 127244.
73. A. Andreas, Z. G. Winata, S. P. Santoso, A. E. Angkawijaya, M. Yuliana, F. E. Soetaredjo, S. Ismadji, H.-Y. Hsu, A. W. Go and Y.-H. Ju, J. Mol. Liq., 2021, 329, 115579.
74. X. Yang, J. Yang, M. I. Ullah, Y. Xia, G. Liang, S. Wang, J. Zhang, H.-Y. Hsu, H. Song and J. Tang, ACS Appl. Mater. Interfaces, 2020, 12, 42217–42225.
75. C. H. Mak, X. Han, M. Du, J.-J. Kai, K. F. Tsang, G. Jia, K.-C. Cheng, H.-H. Shen and H.-Y. Hsu, J. Mater. Chem. A, 2021, 9, 4454–4504.
76. E. Luévano-Hipólito, O. L. Quintero-Lizárraga and L. M. Torres-Martínez, Catalysts, 2022, 12(11), 1410.
77. C. H. Mak, R. Liu, X. Han, Y. Tang, X. Zou, H. H. Shen, Y. Meng, G. Zou and H. Y. Hsu, Adv. Opt. Mater., 2020, 8, 2001023.
78. H.-Y. Hsu, L. Ji, M. Du, J. Zhao, T. Y. Edward and A. J. Bard, Electrochim. Acta, 2016, 220, 205–210.
79. H.-Y. Hsu, L. Ji, C. Zhang, C. H. Mak, R. Liu, T. Wang, X. Zou, S.-Y. Leu and E. T. Yu, J. Mater. Chem. C, 2018, 6, 11552–11560.
80. C. H. Mak, X. Huang, R. Liu, Y. Tang, X. Han, L. Ji, X. Zou, G. Zou and H.-Y. Hsu, Nano Energy, 2020, 73, 104752.
81. S. Rao, X. Zou, S. Wang, Y. Lu, T. Shi, H.-Y. Hsu, Q. Xu and X. Lu, Mater. Chem. Phys., 2019, 232, 6–15.
82. Z. Chen, Y. Hu, J. Wang, Q. Shen, Y. Zhang, C. Ding, Y. Bai, G. Jiang, Z. Li and N. Gaponik, Chem. Mater., 2020, 32, 1517–1525.
83. M. A. Raza, F. Li, M. Que, L. Zhu and X. Chen, Mater. Adv., 2021, 2, 7187–7209.
84. C. B. Hiragond, N. S. Powar and S.-I. In, Nanomaterials, 2020, 10, 2569.
85. Z. Liu, H. Yang, J. Wang, Y. Yuan, K. Hills-Kimball, T. Cai, P. Wang, A. Tang and O. Chen, Nano Lett., 2021, 21, 1620–1627.
86. E. Luévano-Hipólito, M. G. Fabela-Cedillo and L. M. Torres-Martínez, Mater. Lett., 2024, 136066.
87. K. Zhang, Y. Zhang, D. Zhou, Y. Yang, Z. Yang, Z. Song, J. Zhang, Q. Wang and J. Qiu, J. Alloys Compd., 2024, 976, 173283.
88. Y.-F. Mu, J.-S. Zhao, L.-Y. Wu, K.-Y. Tao, Z.-L. Liu, F.-Q. Bai, D.-C. Zhong, M. Zhang and T.-B. Lu, Appl. Catal., B, 2023, 338, 123024.
89. J. Zhou, D. Wu, C. Tian, Z. Liang, H. Ran, B. Gao, Z. Luo, Q. Huang and X. Tang, Small, 2023, 19, 2207915.
90. H.-B. Zhao, J.-F. Liao, Y. Teng, H.-Y. Chen and D.-B. Kuang, ACS Appl. Mater. Interfaces, 2022, 14, 43354–43361.
91. G. Chen, P. Wang, Y. Wu, Q. Zhang, Q. Wu, Z. Wang, Z. Zheng, Y. Liu, Y. Dai and B. Huang, Adv. Mater., 2020, 32, 2001344.
92. J. Pi, X. Jia, Z. Long, S. Yang, H. Wu, D. Zhou, Q. Wang, H. Zheng, Y. Yang and J. Zhang, Adv. Energy Mater., 2022, 12, 2202074.
93. Y. Tang, C. H. Mak, J. Zhang, G. Jia, K.-C. Cheng, H. Song, M. Yuan, S. Zhao, J.-J. Kai, J. C. Colmenares and H.-Y. Hsu, Adv. Mater., 2023, 35, 2207835.
94. C. A. Gueymard, Sol. Energy, 2004, 76, 423–453.
95. V. K. Ravi, G. B. Markad and A. Nag, ACS Energy Lett., 2016, 1, 665–671.
96. P. Wang, X. Ba, X. Zhang, H. Gao, M. Han, Z. Zhao, X. Chen, L. Wang, X. Diao and G. Wang, Chem. Eng. J., 2023, 457, 141248.
97. J. Fu, K. Jiang, X. Qiu, J. Yu and M. Liu, Mater. Today, 2020, 32, 222–243.
98. Y. Wang, Q. Zhou, Y. Zhu and D. Xu, Appl. Catal., B, 2021, 294, 120236.
99. M. Que, Y. Zhao, L. Pan, Y. Yang, Z. He, H. Yuan, J. Chen and G. Zhu, Mater. Lett., 2021, 282, 128695.
100. J. Sheng, Y. He, J. Li, C. Yuan, H. Huang, S. Wang, Y. Sun, Z. Wang and F. Dong, ACS Nano, 2020, 14, 13103–13114.
101. Y. Wang, H. Huang, Z. Zhang, C. Wang, Y. Yang, Q. Li and D. Xu, Appl. Catal., B, 2021, 282, 119570.
102. X. Wang, K. Li, J. He, J. Yang, F. Dong, W. Mai and M. Zhu, Nano Energy, 2020, 78, 105388.
103. Z.-C. Kong, H.-H. Zhang, J.-F. Liao, Y.-J. Dong, Y. Jiang, H.-Y. Chen and D.-B. Kuang, Sol. RRL, 2020, 4, 1900365.
104. B. Zhou, X. Xu, M. Li, L. Wu, S. Xu, L. Yuan, Y. Chong, W. Xie, P. Liu, D. Ye, G. I. N. Waterhouse, Y. Qiu, G. Chen, T. Shi and K. Yan, Chem. Eng. J., 2023, 468, 143754.
105. S. Shyamal, S. K. Dutta and N. Pradhan, J. Phys. Chem. Lett., 2019, 10, 7965–7969.
106. S. S. Bhosale, A. K. Kharade, E. Jokar, A. Fathi, S.-m. Chang and E. W.-G. Diau, J. Am. Chem. Soc., 2019, 141, 20434–20442.
107. Y.-W. Liu, S.-H. Guo, S.-Q. You, C.-Y. Sun, X.-L. Wang, L. Zhao and Z.-M. Su, Nanotechnology, 2020, 31, 215605.
108. S. Wan, M. Ou, Q. Zhong and X. Wang, Chem. Eng. J., 2019, 358, 1287–1295.
109. N. Zhang, J.-J. Li, Y. Li, H. Wang, J.-Y. Zhang, Y. Liu, Y.-Z. Fang, Z. Liu and M. Zhou, J. Colloid Interface Sci., 2022, 608, 3192–3203.
110. L. Wu, S. Zheng, H. Lin, S. Zhou, A. Mahmoud Idris, J. Wang, S. Li and Z. Li, J. Colloid Interface Sci., 2023, 629, 233–242.
111. Y.-F. Mu, W. Zhang, G.-X. Dong, K. Su, M. Zhang and T.-B. Lu, Small, 2020, 16, 2002140.
112. H. Huang, J. Zhao, Y. Du, C. Zhou, M. Zhang, Z. Wang, Y. Weng, J. Long, J. Hofkens, J. A. Steele and M. B. J. Roeffaers, ACS Nano, 2020, 14, 16689–16697.
113. T. Chen, M. Zhou, W. Chen, Y. Zhang, S. Ou and Y. Liu, Sustainable Energy Fuels, 2021, 5, 3598–3605.
114. Y.-F. Xu, X.-D. Wang, J.-F. Liao, B.-X. Chen, H.-Y. Chen and D.-B. Kuang, Adv. Mater. Interfaces, 2018, 5, 1801015.
115. L.-Y. Wu, Y.-F. Mu, X.-X. Guo, W. Zhang, Z.-M. Zhang, M. Zhang and T.-B. Lu, Angew. Chem., Int. Ed., 2019, 58, 9491–9495.
116. Y. Jiang, J.-F. Liao, H.-Y. Chen, H.-H. Zhang, J.-Y. Li, X.-D. Wang and D.-B. Kuang, Chem, 2020, 6, 766–780.
117. Z.-C. Kong, J.-F. Liao, Y.-J. Dong, Y.-F. Xu, H.-Y. Chen, D.-B. Kuang and C.-Y. Su, ACS Energy Lett., 2018, 3, 2656–2662.
118. Q.-M. Sun, J.-J. Xu, F.-F. Tao, W. Ye, C. Zhou, J.-H. He and J.-M. Lu, Angew. Chem., Int. Ed., 2022, 61, e202200872.
119. N. Li, J. Wang, G. Zhao, J. Du, Y. Li, Y. Bai, Z. Li and Y. Xiong, ACS Mater. Lett., 2024, 6, 999–1006.
120. H.-B. Zhao, J.-N. Huang, Q. Qin, H.-Y. Chen and D.-B. Kuang, Small, 2023, 19, 2302022.
121. S. Kumar, I. Hassan, M. Regue, S. Gonzalez-Carrero, E. Rattner, M. A. Isaacs and S. Eslava, J. Mater. Chem. A, 2021, 9, 12179–12187.
122. S. Kumar, M. Regue, M. A. Isaacs, E. Freeman and S. Eslava, ACS Appl. Energy Mater., 2020, 3, 4509–4522.
123. X.-D. Wang, Y.-H. Huang, J.-F. Liao, Y. Jiang, L. Zhou, X.-Y. Zhang, H.-Y. Chen and D.-B. Kuang, J. Am. Chem. Soc., 2019, 141, 13434–13441.
124. J. Hou, S. Cao, Y. Wu, Z. Gao, F. Liang, Y. Sun, Z. Lin and L. Sun, Chem.–Eur. J., 2017, 23, 9481–9485.
125. L.-Y. Wu, M.-R. Zhang, Y.-X. Feng, W. Zhang, M. Zhang and T.-B. Lu, Sol. RRL, 2021, 5, 2100263.
126. R. Tang, H. Sun, Z. Zhang, L. Liu, F. Meng, X. Zhang, W. Yang, Z. Li, Z. Zhao, R. Zheng and J. Huang, Chem. Eng. J., 2022, 429, 132137.
127. Y. Xi, X. Zhang, Y. Shen, W. Dong, Z. Fan, K. Wang, S. Zhong and S. Bai, Appl. Catal., B, 2021, 297, 120411.
128. S. Bera, S. Shyamal and N. Pradhan, J. Am. Chem. Soc., 2021, 143, 14895–14906.
129. C. Lu, D. S. Itanze, A. G. Aragon, X. Ma, H. Li, K. B. Ucer, C. Hewitt, D. L. Carroll, R. T. Williams, Y. Qiu and S. M. Geyer, Nanoscale, 2020, 12, 2987–2991.
130. S. Shyamal, S. K. Dutta, T. Das, S. Sen, S. Chakraborty and N. Pradhan, J. Phys. Chem. Lett., 2020, 11, 3608–3614.
131. L. Zhou, Y.-F. Xu, B.-X. Chen, D.-B. Kuang and C.-Y. Su, Small, 2018, 14, 1703762.
132. J.-C. Wang, N. Li, A. M. Idris, J. Wang, X. Du, Z. Pan and Z. Li, Sol. RRL, 2021, 5, 2100154.
133. J. Sheng, Y. He, M. Huang, C. Yuan, S. Wang and F. Dong, ACS Catal., 2022, 12, 2915–2926.
134. C. Tian, Q. Huang, D. Wu, J. Lai, F. Qi, N. Zhang, W. Zhang and X. Tang, Mater. Today Energy, 2022, 28, 101067.
135. S.-H. Guo, J. Zhou, X. Zhao, C.-Y. Sun, S.-Q. You, X.-L. Wang and Z.-M. Su, J. Catal., 2019, 369, 201–208.
136. D. Wu, X. Zhao, Y. Huang, J. Lai, H. Li, J. Yang, C. Tian, P. He, Q. Huang and X. Tang, Chem. Mater., 2021, 33, 4971–4976.
137. D. Wu, X. Zhao, Y. Huang, J. Lai, J. Yang, C. Tian, P. He, Q. Huang and X. Tang, J. Phys. Chem. C, 2021, 125, 18328–18333.
138. G.-X. Dong, W. Zhang, Y.-F. Mu, K. Su, M. Zhang and T.-B. Lu, Chem. Commun., 2020, 56, 4664–4667.
139. Y.-X. Chen, Y.-F. Xu, X.-D. Wang, H.-Y. Chen and D.-B. Kuang, Sustainable Energy Fuels, 2020, 4, 2249–2255.
140. J. Cheng, Y. Mu, L. Wu, Z. Liu, K. Su, G. Dong, M. Zhang and T. Lu, Nano Res., 2022, 15, 1845–1852.
141. Y.-F. Xu, M.-Z. Yang, H.-Y. Chen, J.-F. Liao, X.-D. Wang and D.-B. Kuang, ACS Appl. Energy Mater., 2018, 1, 5083–5089.
142. A. Pan, X. Ma, S. Huang, Y. Wu, M. Jia, Y. Shi, Y. Liu, P. Wangyang, L. He and Y. Liu, J. Phys. Chem. Lett., 2019, 10, 6590–6597.
143. Y. Jiang, J.-F. Liao, Y.-F. Xu, H.-Y. Chen, X.-D. Wang and D.-B. Kuang, J. Mater. Chem. A, 2019, 7, 13762–13769.
144. X.-X. Guo, S.-F. Tang, Y.-F. Mu, L.-Y. Wu, G.-X. Dong and M. Zhang, RSC Adv., 2019, 9, 34342–34348.
145. Z. Zhang, B. Wang, H.-B. Zhao, J.-F. Liao, Z.-C. Zhou, T. Liu, B. He, Q. Wei, S. Chen, H.-Y. Chen, D.-B. Kuang, Y. Li and G. Xing, Appl. Catal., B, 2022, 312, 121358.
146. X. Wang, J. He, L. Mao, X. Cai, C. Sun and M. Zhu, Chem. Eng. J., 2021, 416, 128077.
147. M. Ou, W. Tu, S. Yin, W. Xing, S. Wu, H. Wang, S. Wan, Q. Zhong and R. Xu, Angew. Chem., Int. Ed., 2018, 57, 13570–13574.
148. J. Wang, J. Wang, N. Li, X. Du, J. Ma, C. He and Z. Li, ACS Appl. Mater. Interfaces, 2020, 12, 31477–31485.
149. F. Xu, K. Meng, B. Cheng, S. Wang, J. Xu and J. Yu, Nat. Commun., 2020, 11, 4613.
150. S. Purohit, S. Singh, K. L. Yadav, K. K. Pant and S. Satapathi, ACS Appl. Energy Mater., 2023, 6, 5580–5587.
151. Z.-L. Liu, R.-R. Liu, Y.-F. Mu, Y.-X. Feng, G.-X. Dong, M. Zhang and T.-B. Lu, Sol. RRL, 2021, 5, 2000691.
152. Y.-X. Feng, G.-X. Dong, K. Su, Z.-L. Liu, W. Zhang, M. Zhang and T.-B. Lu, J. Energy Chem., 2022, 69, 348–355.
153. X. Wang, Z. Wang, Y. Li, J. Wang and G. Zhang, Appl. Catal., B, 2022, 319, 121895.
154. Y. Zhang, L. Shi, H. Yuan, X. Sun, X. Li, L. Duan, Q. Li, Z. Huang, X. Ban and D. Zhang, Chem. Eng. J., 2022, 430, 132820.
155. Z. Dong, Z. Zhang, Y. Jiang, Y. Chu and J. Xu, Chem. Eng. J., 2022, 433, 133762.
156. K. Su, G.-X. Dong, W. Zhang, Z.-L. Liu, M. Zhang and T.-B. Lu, ACS Appl. Mater. Interfaces, 2020, 12, 50464–50471.
157. Z. Zhang, M. Shu, Y. Jiang and J. Xu, Chem. Eng. J., 2021, 414, 128889.
158. L. Li, Z. Zhang, C. Ding and J. Xu, Chem. Eng. J., 2021, 419, 129543.
159. Z. Cui, P. Wang, Y. Wu, X. Liu, G. Chen, P. Gao, Q. Zhang, Z. Wang, Z. Zheng, H. Cheng, Y. Liu, Y. Dai and B. Huang, Appl. Catal., B, 2022, 310, 121375.
160. R. Cheng, C. C. Chung, S. Wang, B. Cao, M. Zhang, C. Chen, Z. Wang, M. Chen, S. Shen and S. P. Feng, Mater. Today Phys., 2021, 17, 100358.
161. Q. Chen, Y. Ma, L. Wang, X. Lan and J. Shi, Sol. RRL, 2021, 5, 2000755.
162. Q. Chen, X. Lan, Y. Ma, P. Lu, Z. Yuan and J. Shi, Sol. RRL, 2021, 5, 2100186.
163. L. Ding, F. Bai, B. Borjigin, Y. Li, H. Li and X. Wang, Chem. Eng. J., 2022, 446, 137102.
164. N. Li, X.-P. Zhai, B. Ma, H.-J. Zhang, M.-J. Xiao, Q. Wang and H.-L. Zhang, J. Mater. Chem. A, 2023, 11, 4020–4029.
165. S. Shyamal, S. K. Dutta, T. Das, S. Sen, S. Chakraborty and N. Pradhan, J. Phys. Chem. Lett., 2020, 11(9), 3608–3614.
166. Y. Wang, S. Wang and X. W. Lou, Angew. Chem., Int. Ed., 2019, 58, 17236–17240.
167. J. Yu, J. Low, W. Xiao, P. Zhou and M. Jaroniec, J. Am. Chem. Soc., 2014, 136, 8839–8842.
168. L. Li, P. A. Salvador and G. S. Rohrer, Nanoscale, 2014, 6, 24–42.
169. S. Selcuk and A. Selloni, Nat. Mater., 2016, 15, 1107–1112.
170. S. Bai, J. Jiang, Q. Zhang and Y. Xiong, Chem. Soc. Rev., 2015, 44, 2893–2939.
171. J. Yuan, A. Hazarika, Q. Zhao, X. Ling, T. Moot, W. Ma and J. M. Luther, Joule, 2020, 4, 1160–1185.
172. C.-C. Lin, T.-R. Liu, S.-R. Lin, K. M. Boopathi, C.-H. Chiang, W.-Y. Tzeng, W.-H. C. Chien, H.-S. Hsu, C.-W. Luo and H.-Y. Tsai, J. Am. Chem. Soc., 2022, 144, 15718–15726.
173. C. Tang, C. Chen, W. Xu and L. Xu, J. Mater. Chem. A, 2019, 7, 6911–6919.
174. C. Tang, C. Chen, W. Xu and L. Xu, J. Mater. Chem. A, 2019, 7, 6911–6919.
175. G. Huang, C. Wang, S. Xu, S. Zong, J. Lu, Z. Wang, C. Lu and Y. Cui, Adv. Mater., 2017, 29, 1700095.
176. S. Wan, M. Ou, Q. Zhong and X. Wang, Chem. Eng. J., 2019, 358, 1287–1295.
177. Q. Xu, L. Zhang, B. Cheng, J. Fan and J. Yu, Chem, 2020, 6, 1543–1559.
178. J. W. Maina, C. Pozo-Gonzalo, L. Kong, J. Schütz, M. Hill and L. F. Dumée, Mater. Horiz., 2017, 4, 345–361.
179. L. Zhang, J. Zhang, H. Yu and J. Yu, Adv. Mater., 2022, 34, 2107668.
180. M. S. Sena, J. Cui, Y. Baghdadi, E. Rattner, M. Daboczi, A. L. Lopes-Moriyama, A. G. dos Santos and S. Eslava, ACS Appl. Energy Mater., 2023, 6, 10193–10204.
181. H. Xiao, Q. Qian and Z. Zang, Sci. China Mater., 2023, 66, 1810–1819.
182. S. Fan, Q. Yang, G. Yin, X. Qi, Y. Feng, J. Ding, Q. Peng, Y. Qu, Q. Wang, Y. Shen, M. Wang and X. Gong, Small, 2024, 2311978.
183. B. Zhou, S. Xu, L. Wu, M. Li, Y. Chong, Y. Qiu, G. Chen, Y. Zhao, C. Feng, D. Ye and K. Yan, Small, 2023, 19, 2302058.
184. Y. Feng, D. Chen, Y. Zhong, Z. He, S. Ma, H. Ding, W. Ao, X. Wu and M. Niu, ACS Appl. Mater. Interfaces, 2023, 15, 9221–9230.
185. L. Y. Wu, Y. F. Mu, X. X. Guo, W. Zhang, Z. M. Zhang, M. Zhang and T. B. Lu, Angew. Chem., Int. Ed., 2019, 58, 9491–9495.
186. A. Dey, J. Ye, A. De, E. Debroye, S. K. Ha, E. Bladt, A. S. Kshirsagar, Z. Wang, J. Yin, Y. Wang, L. N. Quan, F. Yan, M. Gao, X. Li, J. Shamsi, T. Debnath, M. Cao, M. A. Scheel, S. Kumar, J. A. Steele, M. Gerhard, L. Chouhan, K. Xu, X.-g. Wu, Y. Li, Y. Zhang, A. Dutta, C. Han, I. Vincon, A. L. Rogach, A. Nag, A. Samanta, B. A. Korgel, C.-J. Shih, D. R. Gamelin, D. H. Son, H. Zeng, H. Zhong, H. Sun, H. V. Demir, I. G. Scheblykin, I. Mora-Seró, J. K. Stolarczyk, J. Z. Zhang, J. Feldmann, J. Hofkens, J. M. Luther, J. Pérez-Prieto, L. Li, L. Manna, M. I. Bodnarchuk, M. V. Kovalenko, M. B. J. Roeffaers, N. Pradhan, O. F. Mohammed, O. M. Bakr, P. Yang, P. Müller-Buschbaum, P. V. Kamat, Q. Bao, Q. Zhang, R. Krahne, R. E. Galian, S. D. Stranks, S. Bals, V. Biju, W. A. Tisdale, Y. Yan, R. L. Z. Hoye and L. Polavarapu, ACS Nano, 2021, 15, 10775–10981.
187. Z. Zhang, B. Wang, H.-B. Zhao, J.-F. Liao, Z.-C. Zhou, T. Liu, B. He, Q. Wei, S. Chen, H.-Y. Chen, D.-B. Kuang, Y. Li and G. Xing, Appl. Catal., B, 2022, 312, 121358.
188. N. Zhang, J.-J. Li, Y. Li, H. Wang, J.-Y. Zhang, Y. Liu, Y.-Z. Fang, Z. Liu and M. Zhou, J. Colloid Interface Sci., 2022, 608, 3192–3203.
189. Z. Zhang, M. Wang, Z. Chi, W. Li, H. Yu, N. Yang and H. Yu, Appl. Catal., B, 2022, 313, 121426.
190. C. Tian, Q. Huang, D. Wu, J. Lai, F. Qi, N. Zhang, W. Zhang and X. Tang, Mater. Today Energy, 2022, 28, 101067.
191. X. Wang, J. He, L. Mao, X. Cai, C. Sun and M. Zhu, Chem. Eng. J., 2021, 416, 128077.
192. Y. Wang, H. Huang, Z. Zhang, C. Wang, Y. Yang, Q. Li and D. Xu, Appl. Catal., 2021, 282, 119570.
193. J.-F. Liao, Y.-T. Cai, J.-Y. Li, Y. Jiang, X.-D. Wang, H.-Y. Chen and D.-B. Kuang, J. Energy Chem., 2021, 53, 309–315.
194. D. Laishram, S. Zeng, K. M. Alam, A. P. Kalra, K. Cui, P. Kumar, R. K. Sharma and K. Shankar, Appl. Surf. Sci., 2022, 592, 153276.
195. J. Wang, L. Xiong, Y. Bai, Z. Chen, Q. Zheng, Y. Shi, C. Zhang, G. Jiang and Z. Li, Sol. RRL, 2022, 6, 2200294.
196. B. N. Choi, J. Y. Seo, Z. An, P. J. Yoo and C.-H. Chung, Chem. Eng. J., 2022, 430, 132807.
197. Y. Tang, C. H. Mak, G. Jia, K.-C. Cheng, J.-J. Kai, C.-W. Hsieh, F. Meng, W. Niu, F.-F. Li, H.-H. Shen, X. Zhu, H. M. Chen and H.-Y. Hsu, J. Mater. Chem. A, 2022, 10, 12296–12316.
198. Z. Chen, H.-Y. Hsu, M. Arca and K. S. Schanze, J. Phys. Chem. B, 2015, 119, 7198–7209.
199. F. Meng, Y. Jia, J. Wang, X. Huang, Z. Gui, L. Huang, R. Li, R. Chen, J. Xu, W. Chen, Z. He, H.-Y. Hsu, E. Zhu, G. Che and H.-L. Wang, Sol. RRL, 2019, 3, 1900319.
200. S. Ghosh and B. Pradhan, ChemNanoMat, 2019, 5, 300–312.
201. L. Protesescu, S. Yakunin, S. Kumar, J. Bär, F. Bertolotti, N. Masciocchi, A. Guagliardi, M. Grotevent, I. Shorubalko, M. I. Bodnarchuk, C.-J. Shih and M. V. Kovalenko, ACS Nano, 2017, 11, 3119–3134.
202. W. Song, G. Qi and B. Liu, J. Mater. Chem. A, 2023, 11, 12482–12498.
203. Y. Peng, C. H. Mak, J.-J. Kai, M. Du, L. Ji, M. Yuan, X. Zou, H.-H. Shen, S. P. Santoso, J. C. Colmenares and H.-Y. Hsu, J. Mater. Chem. A, 2021, 9, 26628–26649.

---