Mechanistic Insights into Photodegradation of Organic Dyes Using Heterostructure Photocatalysts

Due to its low cost, environmentally friendly process, and lack of secondary contamination, the photodegradation of dyes is regarded as a promising technology for industrial wastewater treatment. This technology demonstrates the light-enhanced generation of charge carriers and reactive radicals that non-selectively degrade various organic dyes into water, CO2, and other organic compounds via direct photodegradation or a sensitization-mediated degradation process. The overall efficiency of the photocatalysis system is closely dependent upon operational parameters that govern the adsorption and photodegradation of dye molecules, including the initial dye concentration, pH of the solution, temperature of the reaction medium, and light intensity. Additionally, the charge-carrier properties of the photocatalyst strongly affect the generation of reactive species in the heterogeneous photodegradation and thereby dictate the photodegradation efficiency. Herein, this comprehensive review discusses the pseudo kinetics and mechanisms of the photodegradation reactions. The operational factors affecting the photodegradation of either cationic or anionic dye molecules, as well as the charge-carrier properties of the photocatalyst, are also fully explored. By further analyzing past works to clarify key active species for photodegradation reactions and optimal conditions, this review provides helpful guidelines that can be applied to foster the development of efficient photodegradation systems.


Introduction
The widespread presence of organic dyes in industrial wastewaters from the textile, apparel, and paper industries results in significant environmental contamination. These dye-polluted effluents contain highly hazardous, carcinogenic, non-biodegradable, and colored pigments that can cause damage to humans [1,2]. Even at very low concentrations (below 1 ppm), dyes are clearly visible in water and seriously deteriorate aqueous environments [3][4][5]. Therefore, the removal of colored organic dyes from wastes is imperative and important. For conventional treatment on industrial wastewater, adsorption [6] and coagulation [7] are common methods used to remove the organic dyes. However, these processes cause secondary hazardous pollution because dyes are only changed from a liquid phase into a solid phase. Thus, further treatments are necessary to resolve the problem of secondary pollution [8,9]. Over the past few years, photocatalysis was regarded as a promising

Classification of organic dyes
Basically, the chemical structure of dye molecules determines their color and properties. Therefore, they can be classified according to their chemical structure (functional groups), color, or aspects of usage [26]. In the textile industry, commonly used dyes include acid, basic, direct, azo, naphtha, reactive, mordant, vat, disperse, and sulfur dyes [27], with azo dyes being the most used at present. To study their properties with regard to photodegradation reactions, dyes are usually classified using their molecular charge upon dissociation in aqueous-based applications. Table 1 presents the chemical properties of several representative dyes that are frequently used in photodegradation applications. According to the chemical structure, they are divided into cationic and anionic dyes. The cationic dyes, including methylene blue (MB), rhodamine B (RhB), malachite green (MG), rhodamine 6G (Rh6G), crystal violet (CV), and safranin O (SO), contain cationic functional groups that can dissociate into positively charged ions [28] in an aqueous solution. The most common cationic functional group is the onium group and, thus, most of the cations are N + ions. On the other hand, the anionic dyes cover direct, acid, and reactive dyes [29], such as acid orange 7 (AO7), eosin Y (EY), methyl orange (MO), acid red 14 (AR14), alizarin red S (ARS), rose bengal (RB), and phenol red (PR). All anionic dyes contain anionic functional groups, e.g., sulfonic or carboxylic acid groups [5]. These functional groups are water-soluble and can effectively interact with photocatalysts with hydrophilic surfaces. Hence, cationic and anionic dyes are also known as basic and acidic dyes, respectively.

Classification of organic dyes
Basically, the chemical structure of dye molecules determines their color and properties. Therefore, they can be classified according to their chemical structure (functional groups), color, or aspects of usage [26]. In the textile industry, commonly used dyes include acid, basic, direct, azo, naphtha, reactive, mordant, vat, disperse, and sulfur dyes [27], with azo dyes being the most used at present. To study their properties with regard to photodegradation reactions, dyes are usually classified using their molecular charge upon dissociation in aqueous-based applications. Table 1 presents the chemical properties of several representative dyes that are frequently used in photodegradation applications. According to the chemical structure, they are divided into cationic and anionic dyes. The cationic dyes, including methylene blue (MB), rhodamine B (RhB), malachite green (MG), rhodamine 6G (Rh6G), crystal violet (CV), and safranin O (SO), contain cationic functional groups that can dissociate into positively charged ions [28] in an aqueous solution. The most common cationic functional group is the onium group and, thus, most of the cations are N + ions. On the other hand, the anionic dyes cover direct, acid, and reactive dyes [29], such as acid orange 7 (AO7), eosin Y (EY), methyl orange (MO), acid red 14 (AR14), alizarin red S (ARS), rose bengal (RB), and phenol red (PR). All anionic dyes contain anionic functional groups, e.g., sulfonic or carboxylic acid groups [5]. These functional groups are water-soluble and can effectively interact with photocatalysts with hydrophilic surfaces. Hence, cationic and anionic dyes are also known as basic and acidic dyes, respectively.

Classification of organic dyes
Basically, the chemical structure of dye molecules determines their color and properties. Therefore, they can be classified according to their chemical structure (functional groups), color, or aspects of usage [26]. In the textile industry, commonly used dyes include acid, basic, direct, azo, naphtha, reactive, mordant, vat, disperse, and sulfur dyes [27], with azo dyes being the most used at present. To study their properties with regard to photodegradation reactions, dyes are usually classified using their molecular charge upon dissociation in aqueous-based applications. Table 1 presents the chemical properties of several representative dyes that are frequently used in photodegradation applications. According to the chemical structure, they are divided into cationic and anionic dyes. The cationic dyes, including methylene blue (MB), rhodamine B (RhB), malachite green (MG), rhodamine 6G (Rh6G), crystal violet (CV), and safranin O (SO), contain cationic functional groups that can dissociate into positively charged ions [28] in an aqueous solution. The most common cationic functional group is the onium group and, thus, most of the cations are N + ions. On the other hand, the anionic dyes cover direct, acid, and reactive dyes [29], such as acid orange 7 (AO7), eosin Y (EY), methyl orange (MO), acid red 14 (AR14), alizarin red S (ARS), rose bengal (RB), and phenol red (PR). All anionic dyes contain anionic functional groups, e.g., sulfonic or carboxylic acid groups [5]. These functional groups are water-soluble and can effectively interact with photocatalysts with hydrophilic surfaces. Hence, cationic and anionic dyes are also known as basic and acidic dyes, respectively.

Classification of organic dyes
Basically, the chemical structure of dye molecules determines their color and properties. Therefore, they can be classified according to their chemical structure (functional groups), color, or aspects of usage [26]. In the textile industry, commonly used dyes include acid, basic, direct, azo, naphtha, reactive, mordant, vat, disperse, and sulfur dyes [27], with azo dyes being the most used at present. To study their properties with regard to photodegradation reactions, dyes are usually classified using their molecular charge upon dissociation in aqueous-based applications. Table 1 presents the chemical properties of several representative dyes that are frequently used in photodegradation applications. According to the chemical structure, they are divided into cationic and anionic dyes. The cationic dyes, including methylene blue (MB), rhodamine B (RhB), malachite green (MG), rhodamine 6G (Rh6G), crystal violet (CV), and safranin O (SO), contain cationic functional groups that can dissociate into positively charged ions [28] in an aqueous solution. The most common cationic functional group is the onium group and, thus, most of the cations are N + ions. On the other hand, the anionic dyes cover direct, acid, and reactive dyes [29], such as acid orange 7 (AO7), eosin Y (EY), methyl orange (MO), acid red 14 (AR14), alizarin red S (ARS), rose bengal (RB), and phenol red (PR). All anionic dyes contain anionic functional groups, e.g., sulfonic or carboxylic acid groups [5]. These functional groups are water-soluble and can effectively interact with photocatalysts with hydrophilic surfaces. Hence, cationic and anionic dyes are also known as basic and acidic dyes, respectively.

Kinetics study for photodegradation reactions
The kinetics for photodegradation reactions are examined based on the dye concentration change by measuring the characteristic absorbance peak at different irradiation times. All the common dyes have their specific characteristic absorptions in the visible range (400-700 nm), as shown in Table 1. Herein, the efficiency of photodegradation (also known as the decolorizing ratio) is determined using the following equation: where C0 and C are the solution concentration at t = 0 and after some irradiation time. However, the absorption peak shift or excitation of the dye molecules may cause the inaccurate estimation of the photodegradation efficiency, which is further discussed in Sections 3.1 and 3.2.

Absorption peak shift of dye molecules
During the photodegradation reaction, a redshift or blueshift can be sometimes seen in the

Kinetics study for photodegradation reactions
The kinetics for photodegradation reactions are examined based on the dye concentration change by measuring the characteristic absorbance peak at different irradiation times. All the common dyes have their specific characteristic absorptions in the visible range (400-700 nm), as shown in Table 1. Herein, the efficiency of photodegradation (also known as the decolorizing ratio) is determined using the following equation: where C0 and C are the solution concentration at t = 0 and after some irradiation time. However, the absorption peak shift or excitation of the dye molecules may cause the inaccurate estimation of the photodegradation efficiency, which is further discussed in

Kinetics study for photodegradation reactions
The kinetics for photodegradation reactions are examined based on the dye concentration change by measuring the characteristic absorbance peak at different irradiation times. All the common dyes have their specific characteristic absorptions in the visible range (400-700 nm), as shown in Table 1. Herein, the efficiency of photodegradation (also known as the decolorizing ratio) is determined using the following equation: where C0 and C are the solution concentration at t = 0 and after some irradiation time. However, the

Kinetics Study for Photodegradation Reactions
The kinetics for photodegradation reactions are examined based on the dye concentration change by measuring the characteristic absorbance peak at different irradiation times. All the common dyes have their specific characteristic absorptions in the visible range (400-700 nm), as shown in Table 1. Herein, the efficiency of photodegradation (also known as the decolorizing ratio) is determined using the following equation: where C 0 and C are the solution concentration at t = 0 and after some irradiation time. However, the absorption peak shift or excitation of the dye molecules may cause the inaccurate estimation of the photodegradation efficiency, which is further discussed in Sections 3.1 and 3.2.

Absorption Peak Shift of Dye Molecules
During the photodegradation reaction, a redshift or blueshift can be sometimes seen in the characteristic absorption of dye molecules, possibly caused by the aggregation of organic dyes [13,30]. Furthermore, the decomposition process of dye molecules could also cause an absorption peak shift. These spectral shifts bring difficulty in determining the concentration of the remaining dyes from the absorbance of the characteristic peak. Special care is, therefore, necessary in order to represent accurate photodegradation efficiency. As shown in Figure 1a, a typical example can be found for RhB photodegradation on Ag 3 PO 4 nanoparticles [31]. The blueshift of the absorption band from 554 to 530 nm was caused by the de-ethylation of RhB molecules. The generation of intermediates can be further observed by steady-state photoluminescence (PL) spectroscopy in Figure 1b. The emission at 575 nm decreased, while a new emission appeared after 7 min of light illumination, confirming the formation of the de-ethylated intermediate. Figure 1c shows the de-ethylation process of RhB molecules. Upon light illumination, the four ethyl groups of RhB can be sequentially removed until it is totally converted into rhodamine without any ethyl groups. This process causes a large blueshift in absorption from 553 nm to 498 nm [32,33], as shown in Figure 2d. The further decomposition of rhodamine with its conjugated ring structure causes a further decrease of the absorption peak without a corresponding peak shift [34]. In the presence of benzoquinone (BQ) as an ·O 2− scavenger (see Section 5.1), the peak intensity decreases without shifting, while the peak shows a blueshift in the presence of 2-propanol (IPA) as an ·OH scavenger, further confirming that the blueshift is caused by the attack of active oxygen species on the N-ethyl groups. Moreover, it was found that the formation of N-de-methylated MG products through the attack of active oxygen species accounts for the observed absorption blueshift of the MG absorption during the MG photodegradation over TiO 2 [35].
(BQ) as an ·O 2− scavenger (see Section 5.1), the peak intensity decreases without shifting, while the peak shows a blueshift in the presence of 2-propanol (IPA) as an ·OH scavenger, further confirming that the blueshift is caused by the attack of active oxygen species on the N-ethyl groups. Moreover, it was found that the formation of N-de-methylated MG products through the attack of active oxygen species accounts for the observed absorption blueshift of the MG absorption during the MG photodegradation over TiO2 [35].

Photobleaching of dye molecules
The direct excitation of dye molecules may induce the formation of colorless and unstable transition forms, instead of complete mineralization, especially in the presence of dissolved oxygen, which also causes the inaccurate estimation of the photocatalytic activity. Taking the TiO2/MB system as an example [36], the photoexcited electrons within TiO2 can transform blue MB molecules

Photobleaching of Dye Molecules
The direct excitation of dye molecules may induce the formation of colorless and unstable transition forms, instead of complete mineralization, especially in the presence of dissolved oxygen, which also causes the inaccurate estimation of the photocatalytic activity. Taking the TiO 2 /MB system as an example [36], the photoexcited electrons within TiO 2 can transform blue MB molecules into their colorless leuco form (LMB) upon UV irradiation, resulting in the photobleaching of MB (step i), as shown in Figure 2a. In an oxygen-free atmosphere (N 2 ) without UV irradiation, the bleached condition persists (step ii), with the formation of the stable LMB. In contrast, recoloration takes place if the system is exposed to air (step iii), leading to back electron transfer from LMB to electron acceptors, and causing a reversion to the blue oxidized form of MB. As shown in Figure 2b, the recovery process becomes faster in an oxygen atmosphere (step iv), revealing the recovery rate is proportional to the level of O 2 [37]. A similar observation can be found in the carbon-doped TiO 2 /MB system [38]. As Figure 2c shows, under UV irradiation, the excitation of TiO 2 makes the photobleaching reaction (step 1) become dominant. On the other hand, visible light irradiation can drive self-catalyzed LMB oxidation to MB, thus dramatically enhancing the recoloration rate (step 3). A competing reaction (step 4) usually exists due to the visible photoactivity of carbon-doped TiO 2 , whereas this reaction is drastically suppressed in the presence of O 2 ; thus, the oxidative LMB to MB transition is predominant under visible irradiation in O 2 atmosphere. Thus, the transformation of MB and LMB can be achieved and repeated by changing the irradiation from UV to visible light, as shown in Figure 2d. This photobleaching phenomenon and recovery process was also found for RhB. The recoloration of RhB from leuco RhB (LRhB) can be observed in 55 min of visible irradiation, and the decoloration-recoloration process can be repeated under sequential UV and visible light irradiations (Figure 2e). takes place if the system is exposed to air (step iii), leading to back electron transfer from LMB to electron acceptors, and causing a reversion to the blue oxidized form of MB. As shown in Figure 2b, the recovery process becomes faster in an oxygen atmosphere (step iv), revealing the recovery rate is proportional to the level of O2 [37]. A similar observation can be found in the carbon-doped TiO2/MB system [38]. As Figure 2c shows, under UV irradiation, the excitation of TiO2 makes the photobleaching reaction (step 1) become dominant. On the other hand, visible light irradiation can drive self-catalyzed LMB oxidation to MB, thus dramatically enhancing the recoloration rate (step 3). A competing reaction (step 4) usually exists due to the visible photoactivity of carbon-doped TiO2, whereas this reaction is drastically suppressed in the presence of O2; thus, the oxidative LMB to MB transition is predominant under visible irradiation in O2 atmosphere. Thus, the transformation of MB and LMB can be achieved and repeated by changing the irradiation from UV to visible light, as shown in Figure 2d. This photobleaching phenomenon and recovery process was also found for RhB. The recoloration of RhB from leuco RhB (LRhB) can be observed in 55 min of visible irradiation, and the decoloration-recoloration process can be repeated under sequential UV and visible light irradiations ( Figure 2e).

Chemical oxygen demand (COD) and total organic carbon (TOC) analysis
To precisely evaluate the extent of decomposition of organic dye molecules during photodegradation, COD and TOC analyses are usually utilized. COD is an indicative measure of oxygen amount that is consumed by oxidation reactions in the solution, which can be used to deduce the number of organic molecules in water. TOC on the other hand is an indicative measure of carbon amount in organic compounds, which can more accurately reflect the total amount of organic compounds in the solution. As shown in Figure 3, the photodegradation of bisphenol A (BPA) was

Chemical Oxygen Demand (COD) and Total Organic Carbon (TOC) Analysis
To precisely evaluate the extent of decomposition of organic dye molecules during photodegradation, COD and TOC analyses are usually utilized. COD is an indicative measure of oxygen amount that is consumed by oxidation reactions in the solution, which can be used to deduce the number of organic molecules in water. TOC on the other hand is an indicative measure of carbon amount in organic compounds, which can more accurately reflect the total amount of organic compounds in the solution. As shown in Figure 3, the photodegradation of bisphenol A (BPA) was demonstrated using BiOI as photocatalysts [39]. The morphology of BiOI with two specific facets ( (110) and (001)) is shown in Figure 3a. Under visible irradiation, BiOI-110 exhibited better photodegradation efficiency than BiOI-001 did because it has a higher capability of adsorbing O 2 , thereby facilitating the generation of reactive radicals. The TOC value of the BiOI-110 system decreased to 5% of the initial TOC value, indicating almost complete mineralization of BPA. In contrast, the TOC value of the BiOI-001 system was reduced by only 44%. This observation is consistent with the change in the photodegradation results. Zhao  To confirm the mineralization of SRB, Figure 3e,f respectively show the temporal changes of COD and TOC during the SRB photodegradation. In the presence of TiO 2 /Pt, approximately 64% of the total COD was reduced after 210 min of irradiation. Conversely, the SRB/TiO 2 system required a much longer irradiation time (around 480 min) to attain a similar decrease. In both cases, the COD remained constant with further irradiation, indicating the total discoloration of SRB molecules. This observation was consistent with the changes in TOC. TOC's increase in the first hour of irradiation is due to photodesorption of the dye or the formation of intermediates. After a gradual decrease to the steady state, the TOC remained unchanged because the degraded fragments were not further decomposed with longer irradiation. photodegradation efficiency than BiOI-001 did because it has a higher capability of adsorbing O2, thereby facilitating the generation of reactive radicals. The TOC value of the BiOI-110 system decreased to 5% of the initial TOC value, indicating almost complete mineralization of BPA. In contrast, the TOC value of the BiOI-001 system was reduced by only 44%. This observation is consistent with the change in the photodegradation results. Zhao et al. reported the photodegradation of anionic sulforhodamine B (SRB) over TiO2-Pt photocatalysts [40] under visible irradiation, in which TiO2-Pt (0.2 wt.%) showed the fastest photodegradation rate, capable of degrading SRB molecules within 130 min (Figure 3(d)). To confirm the mineralization of SRB, Figure  3e,f respectively show the temporal changes of COD and TOC during the SRB photodegradation. In the presence of TiO2/Pt, approximately 64% of the total COD was reduced after 210 min of irradiation. Conversely, the SRB/TiO2 system required a much longer irradiation time (around 480 min) to attain a similar decrease. In both cases, the COD remained constant with further irradiation, indicating the total discoloration of SRB molecules. This observation was consistent with the changes in TOC. TOC's increase in the first hour of irradiation is due to photodesorption of the dye or the formation of intermediates. After a gradual decrease to the steady state, the TOC remained unchanged because the degraded fragments were not further decomposed with longer irradiation.

Pseudo kinetics
To quantify the heterogeneous photodegradation activity, the Langmuir-Hinshelwood (L-H) model is usually considered, as shown in Equation (2).
where K and k are the thermodynamic adsorption constant and photodegradation rate constant, respectively. Because one of the reactants acts as a photocatalyst whose concentration remains unchanged, the reaction kinetics can be simplified, and the term "pseudo" is used to prefix the

Pseudo Kinetics
To quantify the heterogeneous photodegradation activity, the Langmuir-Hinshelwood (L-H) model is usually considered, as shown in Equation (2).
where K and k are the thermodynamic adsorption constant and photodegradation rate constant, respectively. Because one of the reactants acts as a photocatalyst whose concentration remains unchanged, the reaction kinetics can be simplified, and the term "pseudo" is used to prefix the reaction rate expression. At high concentrations of dye, the photocatalyst surfaces are fully covered, leading to the approximation of (1 + KC) to KC. A pseudo zero-order reaction is, thus, observed for saturation coverage on the surface of the photocatalyst [41], since the photodegradation rate is independent of the change in the dye concentration, as shown in Equation (3).
Integrating the equation under the boundary conditions C = C 0 at t = 0 and C = 1 2 C 0 at t = t 1/2 respectively yields A plot of C 0 − C vs. irradiation time gives a slope equal to the zero-order rate constant (k 0 ). Additionally, the half-life (t 1/2 ) is the time required to photodegrade half of the initial dye concentration, which is used to quantitatively compare the photodegradation reaction. Therefore, the zero-order t 1/2 can be expressed by Equation (5), and it increases with the initial concentration of dye molecules. On the other hand, at low initial concentrations of dye molecules, i.e., (KC + 1)-1, a pseudo first-order rate expression is obtained [42], as shown in Equation (6). The equation is valid by assuming the driving force of degradation is constantly proportional to the dye concentration.
where k 1 corresponds to the first-order rate constant. Integrating the equation under the two boundary conditions yields The linear region can be obtained from the plot of ln(C 0 /C) vs. irradiation time, in which the slope gives the rate constant of photodegradation. This model is the most common one used to represent the entire photodegradation process. Here, the half-life is derived from Equation (8). Obviously, the t 1/2 of the first-order model is independent of the dye concentration.
Conversely, Kumara et al. reported that the photodegradation of AO over ZnO photocatalysts followed second-order kinetics [43]. A second-order reaction in which a single reactant is involved is characterized by the chemical reaction (2 C → products).
At equilibrium, the second-order kinetics depends on the amount of dye molecules adsorbed on the photocatalyst surface, which is calculated as follows: Similarly, by integrating the equation under the two boundary conditions, the second-order rate constant (k 2 ), as well as t 1/2 , can be obtained. The second-order t 1/2 increases as the initial concentration is decreased.

Quantum Yield of Photodegradation
The photodegradation efficiency is difficult to directly compare with other reported values, since the photodegradation rate constant is affected by various operational effects (see Section 4). In order to objectively estimate the efficiency of the photodegradation reaction, quantum yield (φ) is used and defined as the number of reacted charge carries for dye molecule decomposition per absorbed photon. The apparent quantum yield (AQY) on the other hand is calculated by dividing the number of charge carries participating in degrading dye molecules by the incident photons. Because the absorbed photons are a certain fraction of the incident photons, the φ value is basically higher than the AQY. Bora et al. estimated the AQY of MB photodegradation over Au nanoparticle-decorated ZnO nanorods (Au-ZnO NRs) [44]. The morphology of Au-ZnO is shown in Figure 4a. It was reported that the photodegradation of MB is initiated by accepting one electron to form semi-reduced MB, followed by further accepting one electron to produce LMB. As a result, two electrons are required for the complete degradation of MB molecules, and the AQY of MB photodegradation can be calculated as follows: where n MB (mol) stands for the amount of MB degraded during the irradiation period, N A (mol −1 ) is Avogadro's constant, P light (W/m 2 ) is the incident power density, S (m 2 ) is the irradiation area, t (s) is irradiation time, h is Planck's constant, and λ is the wavelength of incident light. As shown in Figure 4b, all the AQY values of Au-ZnO from 300 to 600 nm were higher than those of bare ZnO NRs and MB only, which was ascribed to the enhanced charge separation at the Au-ZnO interface. The highest AQY value obtained from Au-ZnO exceeded 30% in the UV region, while the bare ZnO NRs showed fairly low photodegradation activity in the visible region and almost no activity at wavelength longer than 450 nm. On the other hand, Au-ZnO NRs showed a six-fold increase of AQY over bare ZnO NRs in the visible region, which can be attributed to the plasmonic effect of Au. In addition, the BiOBr/Bi 2 O 3 heterostructures were demonstrated for photodegradation of an RhB/p-cresol mixture [45]. The SEM image showed that the composites comprised BiOBr platelets arranging in whorls perpendicular to the Bi 2 O 3 surface (Figure 4c). Under blue light-emitting diode (LED) light (430-470 nm) irradiation, the AQY value of p-cresol photodegradation in the mixture was substantially higher than the value of RhB due to the carrier transfer from RhB to the co-adsorbed p-cresol. Note that there is no photobleaching of RhB as the RhB is unable to absorb blue light (Figure 4d).

Factors influencing the photodegradation reaction
A heterogeneous photocatalytic reaction is composed of two consecutive steps. Firstly, dye molecules interact with the photocatalyst and are adsorbed onto its surface; then, the photodegradation of the dye commences. The overall efficiency of a photocatalysis system is closely dependent on operational parameters that dictate the adsorption and photodegradation of dye In addition, Shams-Ghahfarokhi et al. reported the calculation of the φ value from the first-order rate constant of photodegradation process [15,46] using the following equation: where k 1 (s −1 ) stands for the photodegradation first-order rate constant, I o,λ (Einstein l −1 s −1 ) is the incident intensity at wavelength λ, ε λ (cm −1 M −1 ) represents the molar absorptivity at wavelength λ, and l states the cell path length (cm).

Factors Influencing the Photodegradation Reaction
A heterogeneous photocatalytic reaction is composed of two consecutive steps. Firstly, dye molecules interact with the photocatalyst and are adsorbed onto its surface; then, the photodegradation of the dye commences. The overall efficiency of a photocatalysis system is closely dependent on operational parameters that dictate the adsorption and photodegradation of dye molecules. Herein, possible factors affecting the photodegradation reactions are discussed in detail.

Interaction between Dye Molecules and Photocatalysts
Adsorption is the initial step of the photodegradation reaction prior to the decoloration of dye molecules, which is, thus, an important process for initiating the photodegradation reaction. Both the surface of the photocatalyst and the structure of the dye molecule affect their interactions. Two of the most important interactions between the dye molecules and the photocatalysts are direct bonding and electrostatic interactions. Chemical bonding between functional groups of the dyes and the surface sites of the photocatalysts is a strong interaction to anchor dyes on to photocatalysts [47,48]. For example, RhB with a carboxylic group was demonstrated to bond onto the surface hydroxyls of TiO 2 via an esterification reaction [48]. On the other hand, dye adsorption via electrostatic interaction depends on the nature of dyes, surface properties of photocatalysts, and solution pH. Essentially, the dye adsorption is determined by the strength of the ionic interactions between photocatalysts and dye molecules. An aqueous solution containing salts has a certain value of ionic strength (I), which is defined as where z i is the charge of the salts, and C i is molar concentration. Higher ionic strength of the solution is obtained as the concentration of salts increases. Chen et al. demonstrated the influence of ionic strength on the adsorption of PR dye by tuning the concentration of NaCl [49]. The results showed a decreased adsorption capacity upon increasing the ionic strength because the adsorption of charged moieties competed with that of dye molecules or adsorbents in the solution. Similar results were also published [50][51][52]. Furthermore, the variation of pH value in the solution modifies the electrical double layer of the photocatalysts, which is composed of the charged surface of the photocatalyst and dye molecules in solution. Thus, a high adsorption capacity can be found when anionic or cationic dyes are respectively adsorbed on the photocatalyst surface at acidic or basic pH. The electrical double layer refers to the two layers between the photocatalyst and the bulk solution. Here, the region with dye molecules adsorbed onto the surface-charged photocatalyst is called the Stern layer. The surface-charged photocatalysts create an electrostatic field which affects the dye molecules in the solution, forming the first layer of the electrical double layer. The second diffusion layer is loosely associated with the photocatalyst and is composed of dye molecules that diffuse in the solution under the influence of electrostatic attraction. The electric potential at this plane is called the zeta potential, while the point of zero charge (pzc) forms where the zeta potential is 0. Depending on the solution pH, either positive or negative charges can be formed on a surface, since H + and OH − are the charge-determining ions for most surfaces.
Amphoteric characteristics were observed in many photocatalyst materials, for example, g-C 3 N 4 [53], TiO 2 [54], and most metal oxides. The formation of metal hydroxyl groups (M-OH) is attributed to the adsorption of H 2 O molecules and dissociation of OH − groups at surface metal sites. When the solution pH is higher than the pzc of photocatalysts, their surface is negatively charged; it is positively charged at pH value < pzc of the photocatalysts. The equilibrium of amphoteric metal hydroxides under acidic and alkali conditions can be considered as Therefore, at lower pH, protonation of the photocatalyst creates a surface with a positive charge, making the photocatalyst behave as a strong Lewis acid [55]. The anionic dye, with strongly ionized anionic groups, serves as a strong Lewis base and can be readily adsorbed on the positively charged photocatalyst surface, as shown in Figure 5a. This adsorption process is not favorable for negatively charged photocatalysts because of electrostatic repulsion [56], giving rise to negligible adsorption and a subsequent low degradation rate when pH > pzc of the photocatalyst. Bourikas et al. reported that the adsorption of AO7 on the surface of TiO 2 cannot occur at a pH value higher than 7 [57] due to the electrostatic repulsion arising from the negative sulfonic group of the azo dye. Similar results were also observed for other anionic dye systems [58][59][60]. On the other hand, cationic dyes preferred to adsorb on the negatively charged photocatalysts in alkaline media. Bubacz et al. observed an increased photodegradation rate of MB on anatase TiO 2 with an increase in pH [61]. Fan et al. demonstrated that polyacrylonitrile fiber-hyperbranched polyethylenimine (PANF-g-HPEIs) activated with solutions at different pH may selectively adsorb cationic or anionic dyes [62]. Figure 5b,c show that PANF-g-HPEIs pre-treated with solution at pH = 5 can selectively adsorb the anionic MO dye from an MB/MO mixture, while PANF-g-HPEIs can selectively adsorb the cationic MB molecules after being pre-treated with solution at pH = 10. Zhao et al. investigated the photodegradation efficiency for RhB-sensitized BiOCl nanostructures at pH 3.36 and pH 11.08 [63], as shown in Figure 6. Note that the self-photosensitization of cationic RhB adsorbed on negatively charged BiOCl nanostructures can degrade MO and RhB molecules at pH 3.36 and pH 11.08 (see Section 5.2). Both RhB and MO display efficient photodegradation within 5 min of visible irradiation at pH 3.36. However, the photodegradation of both RhB and MO was suddenly depressed in an alkali solution with pH 11.08, indicating the pH of solution strongly affects the electrical double layer, even though RhB can adsorb onto BiOCl nanostructures at both pH 3.36 and pH 11.08. In an acidic solution with pH 3.36, the presence of a high concentration of H + ions may push the cationic RhB molecules into the Stern Zhao et al. investigated the photodegradation efficiency for RhB-sensitized BiOCl nanostructures at pH 3.36 and pH 11.08 [63], as shown in Figure 6. Note that the self-photosensitization of cationic RhB adsorbed on negatively charged BiOCl nanostructures can degrade MO and RhB molecules at pH 3.36 and pH 11.08 (see Section 5.2). Both RhB and MO display efficient photodegradation within 5 min of visible irradiation at pH 3.36. However, the photodegradation of both RhB and MO was suddenly depressed in an alkali solution with pH 11.08, indicating the pH of solution strongly affects the electrical double layer, even though RhB can adsorb onto BiOCl nanostructures at both pH 3.36 and pH 11.08. In an acidic solution with pH 3.36, the presence of a high concentration of H + ions may push the cationic RhB molecules into the Stern layer, because of the electrostatic repulsion between H + ions and cationic RhB molecules in the diffusion layer, thus improving the electron transfer from RhB to BiOCl and the following photodegradation reaction. On the other hand, the electrostatic attraction of the increased OH − ions to the cationic RhB molecules causes more RhB to stay in the diffusion layer in an alkali solution with pH 11.08, hindering the electron transfer from RhB to BiOCl and resulting in less efficient photodegradation.
Zhao et al. investigated the photodegradation efficiency for RhB-sensitized BiOCl nanostructures at pH 3.36 and pH 11.08 [63], as shown in Figure 6. Note that the self-photosensitization of cationic RhB adsorbed on negatively charged BiOCl nanostructures can degrade MO and RhB molecules at pH 3.36 and pH 11.08 (see Section 5.2). Both RhB and MO display efficient photodegradation within 5 min of visible irradiation at pH 3.36. However, the photodegradation of both RhB and MO was suddenly depressed in an alkali solution with pH 11.08, indicating the pH of solution strongly affects the electrical double layer, even though RhB can adsorb onto BiOCl nanostructures at both pH 3.36 and pH 11.08. In an acidic solution with pH 3.36, the presence of a high concentration of H + ions may push the cationic RhB molecules into the Stern layer, because of the electrostatic repulsion between H + ions and cationic RhB molecules in the diffusion layer, thus improving the electron transfer from RhB to BiOCl and the following photodegradation reaction. On the other hand, the electrostatic attraction of the increased OH − ions to the cationic RhB molecules causes more RhB to stay in the diffusion layer in an alkali solution with pH 11.08, hindering the electron transfer from RhB to BiOCl and resulting in less efficient photodegradation. Chen et al. further utilized the amphoteric properties of organosilica nanoparticles (OSNPs) to recover anionic PR molecules [49], based on the surface charge change of the OSNPs in acidic and Chen et al. further utilized the amphoteric properties of organosilica nanoparticles (OSNPs) to recover anionic PR molecules [49], based on the surface charge change of the OSNPs in acidic and alkali solutions. The morphology of the OSNP is shown in Figure 7a,b, which displays the apparent color of OSNP changing from white to pink. Interestingly, by controlling the pH of the solvent, PR was desorbed from the surface of the OSNPs in an NaOH solution, and the color of the OSNP returned to white, due to the negatively charged surface of the OSNP in NaOH solution. As can be seen in Figure 7c, the zeta potential of OSNPs returned to its initial value after PR desorption, which confirmed that OSNPs were stable during adsorption and desorption. Additionally, Figure 7d shows that adsorption/desorption tests of anionic PR molecules can be repeated for 10 cycles, further pointing out the high stability of OSNPs.
However, Kong et al. found a decline of the MB photodegradation reaction rates at high alkali pH [64], as shown in Figure 8. Although electrostatic attraction occurs between the cationic MB molecules and negatively charged surface of Ta-doped ZnO at alkali pH, the otherwise Coulombic repulsion of the negatively charged Ta-doped ZnO surface against the OH − ions results in the breakage of hydroxylation of the ZnO surface. The Coulombic repulsion was reported for other cationic dyes (MB, RhB) at pH 9.5, which can reduce the number of ·OH radicals and thereby decrease the photodegradation rate [65]. The steric structure of dyes also affects the adsorption process. Both MB and RhB are cationic dyes; however, research shows that MB exhibits a higher photodegradation rate than RhB, which can be attributed to the steric repulsion of the carboxylate anions in RhB which inhibits the extent of adsorption [65].
was desorbed from the surface of the OSNPs in an NaOH solution, and the color of the OSNP returned to white, due to the negatively charged surface of the OSNP in NaOH solution. As can be seen in Figure 7c, the zeta potential of OSNPs returned to its initial value after PR desorption, which confirmed that OSNPs were stable during adsorption and desorption. Additionally, Figure 7d shows that adsorption/desorption tests of anionic PR molecules can be repeated for 10 cycles, further pointing out the high stability of OSNPs. However, Kong et al. found a decline of the MB photodegradation reaction rates at high alkali pH [64], as shown in Figure 8. Although electrostatic attraction occurs between the cationic MB molecules and negatively charged surface of Ta-doped ZnO at alkali pH, the otherwise Coulombic repulsion of the negatively charged Ta-doped ZnO surface against the OH − ions results in the breakage of hydroxylation of the ZnO surface. The Coulombic repulsion was reported for other cationic dyes (MB, RhB) at pH 9.5, which can reduce the number of ·OH radicals and thereby decrease the photodegradation rate [65]. The steric structure of dyes also affects the adsorption process. Both MB and RhB are cationic dyes; however, research shows that MB exhibits a higher photodegradation rate than RhB, which can be attributed to the steric repulsion of the carboxylate anions in RhB which inhibits the extent of adsorption [65].  However, Kong et al. found a decline of the MB photodegradation reaction rates at high alkali pH [64], as shown in Figure 8. Although electrostatic attraction occurs between the cationic MB molecules and negatively charged surface of Ta-doped ZnO at alkali pH, the otherwise Coulombic repulsion of the negatively charged Ta-doped ZnO surface against the OH − ions results in the breakage of hydroxylation of the ZnO surface. The Coulombic repulsion was reported for other cationic dyes (MB, RhB) at pH 9.5, which can reduce the number of ·OH radicals and thereby decrease the photodegradation rate [65]. The steric structure of dyes also affects the adsorption process. Both MB and RhB are cationic dyes; however, research shows that MB exhibits a higher photodegradation rate than RhB, which can be attributed to the steric repulsion of the carboxylate anions in RhB which inhibits the extent of adsorption [65].

Initial Dye Concentration
The initial concentration of organic dyes strongly affects the photodegradation reaction. Saquib et al. demonstrated the photodegradation of gentian violet (also known as CV) with varying concentrations from 0.18 mM to 0.5 mM [66]. The photodegradation rate and TOC increased with the concentration of gentian violet up to 0.25 mM and then declined, as shown in Figure 9a. Kumar et al. reported that the degradation efficiency of VBB over TiO 2 /polyaniline (PAni)/graphene oxide (GO) decreased with an increase in initial dye concentration [67], as shown in Figure 9b. This phenomenon was studied for other dye molecules, including the AR14/TiO 2 [3] and other systems [13,14,68,69]. Figure 9c shows the photodegradation of three different dye molecules (MG, MB, and RhB) over Fe 3 O 4 /reduced graphene oxide (rGO) photocatalysts [70]. The photodegradation efficiency also decreased with the increase in dye concentration.
reported that the degradation efficiency of VBB over TiO2/polyaniline (PAni)/graphene oxide (GO) decreased with an increase in initial dye concentration [67], as shown in Figure 9b. This phenomenon was studied for other dye molecules, including the AR14/TiO2 [3] and other systems [13,14,68,69]. Figure 9c shows the photodegradation of three different dye molecules (MG, MB, and RhB) over Fe3O4/reduced graphene oxide (rGO) photocatalysts [70]. The photodegradation efficiency also decreased with the increase in dye concentration. The adsorption of dye molecules on the photocatalyst surface affects its ability of photon absorption and the subsequent generation of reactive radicals, posing a significant impact on the photodegradation rate. As the initial dye concentration increased, several monolayers of adsorbed dye formed, resulting in more adsorbed dye molecules available for photodegradation. Until reaching the critical level, constant reaction rate is obtained because the surface is completely covered. The photodegradation rate is, however, decreased with further increases in dye concentration. A high amount of adsorbed dye may have an inhibitive effect on the reactions between dye molecules and reactive radicals [71], since the excessive dye concentration may hinder light penetration to the solution [72] and fewer photons can reach the photocatalyst surface. The adsorption of dye molecules on the photocatalyst surface affects its ability of photon absorption and the subsequent generation of reactive radicals, posing a significant impact on the photodegradation rate. As the initial dye concentration increased, several monolayers of adsorbed dye formed, resulting in more adsorbed dye molecules available for photodegradation. Until reaching the critical level, constant reaction rate is obtained because the surface is completely covered. The photodegradation rate is, however, decreased with further increases in dye concentration. A high amount of adsorbed dye may have an inhibitive effect on the reactions between dye molecules and reactive radicals [71], since the excessive dye concentration may hinder light penetration to the solution [72] and fewer photons can reach the photocatalyst surface. Therefore, the generation of charge carriers and reactive radicals is simultaneously reduced, resulting in a decrease of photodegradation efficiency.

Light Intensity
It was shown that the photodegradation rate increases linearly with increasing light intensity at low light intensity. Figure 10a shows that the photodegradation of 2,4-dichlorophenoxyacetic acid (2,4-D) increased as light intensity increased from 100 to 600 lx [73], with the efficiency proportional to the light intensity. In Figure 10b, the photodegradation of benzene on nitrogen-doped TiO 2 was still enhanced by increasing light intensity, but the efficiency was nonlinearly increased [74]. With further increases to higher light intensity, the photodegradation rate became independent of light intensity. It is proposed that, at low light intensity, the separation of photoexcited electrons and holes competes with their recombination, thus impeding the generation of reactive radicals. The electron-hole generation becomes the predominant process as light intensity increases, resulting in a higher photodegradation rate. However, the total active sites for photodegradation remains constant and, therefore, the reaction rate shows a maximum value, even though the light intensity continues to increase [75].
intensity. It is proposed that, at low light intensity, the separation of photoexcited electrons and holes competes with their recombination, thus impeding the generation of reactive radicals. The electron-hole generation becomes the predominant process as light intensity increases, resulting in a higher photodegradation rate. However, the total active sites for photodegradation remains constant and, therefore, the reaction rate shows a maximum value, even though the light intensity continues to increase [75].

Reaction temperature
Not only do initial dye concentration and light intensity possess optimal conditions, but an optimal temperature range also exists for photodegradation. Low temperature favors the adsorption of the reactant, which is a spontaneous exothermic process, whereas the apparent activation energy increases as the temperature decreases close to 0 °C. Low temperature also favors the adsorption of the final product, albeit while decreasing the number of active sites. Therefore, compared to photodegradation and the adsorption of reactants, the slower desorption of product inhibits the reaction and serves as the rate-limiting step under low reaction temperatures. In contrast, when the temperature increases up to the boiling point of the solvent (water for most of the cases), the exothermic adsorption of reactants becomes disfavored, thus limiting the photodegradation reaction [76]. Charge-carrier recombination is also substantially promoted [77] as the reaction temperature exceeds 80 °C. At higher temperatures, the enhanced kinetic energy of dye molecules might allow them escape from the the photocatalyst surface [78], leading to decreased photodegradation efficiency. Thus, the adsorption of dye molecules becomes the limiting step at high temperatures. As a result, reaction temperatures between 20 and 80 °C [15] are considered as the desired temperature for the effective photodegradation of dye molecules.

Reaction Temperature
Not only do initial dye concentration and light intensity possess optimal conditions, but an optimal temperature range also exists for photodegradation. Low temperature favors the adsorption of the reactant, which is a spontaneous exothermic process, whereas the apparent activation energy increases as the temperature decreases close to 0 • C. Low temperature also favors the adsorption of the final product, albeit while decreasing the number of active sites. Therefore, compared to photodegradation and the adsorption of reactants, the slower desorption of product inhibits the reaction and serves as the rate-limiting step under low reaction temperatures. In contrast, when the temperature increases up to the boiling point of the solvent (water for most of the cases), the exothermic adsorption of reactants becomes disfavored, thus limiting the photodegradation reaction [76]. Charge-carrier recombination is also substantially promoted [77] as the reaction temperature exceeds 80 • C. At higher temperatures, the enhanced kinetic energy of dye molecules might allow them escape from the the photocatalyst surface [78], leading to decreased photodegradation efficiency. Thus, the adsorption of dye molecules becomes the limiting step at high temperatures. As a result, reaction temperatures between 20 and 80 • C [15] are considered as the desired temperature for the effective photodegradation of dye molecules.

Intrinsic Properties of Photocatalysts
The photodegradation efficiency can be enhanced by increasing the photocatalyst amount, which is a feature of heterogeneous photocatalysis. The increased photocatalyst amount provides more active sites for the discoloration of dye solution. However, beyond a certain amount, the reaction solution turns into turbid and is subjected to limited light penetration to the photocatalyst surface, leading to the inhibition of the photodegradation reaction.
In addition, the intrinsic properties of the photocatalyst, such as light absorption range, redox potential, charge separation efficiency, and stability, strongly affect the photodegradation activity. The fast charge recombination of a single-component material limits its photodegradation efficiency. Additionally, the single-component photocatalyst cannot simultaneously satisfy the requirement of large redox potential and wide light absorption. To improve the photocatalytic efficiency, a variety of studies were devoted to exploring new heterostructure systems and using them in the photodegradation reaction. Typical strategies include element doping, metal decoration, and semiconductor modification (type II and Z-scheme heterostructures), which can broaden the light absorption range to enhance the light utilization and inhibit charge recombination. Among the aforementioned factors, charge-carrier dynamics are crucial for determining the overall photocatalytic efficiency. It is, thus, of great importance to study the underlying charge-carrier dynamics in semiconductor photocatalysts. Time-resolved PL (TRPL) techniques were demonstrated to be a powerful tool to observe the charge transfer processes of semiconductor heterostructures. In this section, the implication of charge-carrier dynamics in the photodegradation efficiency for the three most relevant heterostructure systems, i.e., metal-semiconductor, type II semiconductor-semiconductor, and Z-scheme semiconductor-metal-semiconductor heterostructures, is discussed.

Modification with Metals
Au-CdS metal-semiconductor nanocrystals were demonstrated to photodegrade RhB molecules [79]. Using TRPL to study the photoexcited charge transfer kinetics, Figure 11a,b display the TRPL spectra for two Au-CdS nanocrystals with different shell thicknesses (14.0 nm and 18.6 nm). Compared with their CdS counterpart, obtained by etching Au cores, a fast decay was found in Au-CdS samples, indicating electron transfer from CdS to Au. This difference became more noticeable for Au-CdS with shell thickness increasing to 18.6 nm, suggesting much more pronounced electronic interaction between CdS and Au. By further analysis with biexponential kinetics, the interfacial charge transfer rate constant (k et ) was estimated. In addition, the photodegradation rate constant (k RhB ) changed with CdS thickness in the core-shell nanocrystals, as shown in Figure 11c. It was enhanced with increasing shell thickness due to the raised ratio of CdS to Au, which led to greater light absorption and, thus, generation of more charge carriers. The correlation among k RhB , k et , and CdS shell thickness is shown in Figure 11d. The change in k ct with CdS shell thickness was consistent with the trend of k RhB , revealing that efficient charge separation can provide a hole-enriched CdS shell for the photodegradation reaction and further enhance the photodegradation efficiency. Moreover, the metal content and composition were tuned in a metal (Ag, Au, Pd)-decorated ZnO system for photodegradation of MB in ethanol [80]. The morphology of ZnO-Au with different Au contents is shown in Figure 12a. The Au content is obviously increasing and was measured to be 0.6 at.%, 1.0 at.%, 1.3 at.%, 2.0 at.%, and 2.3 at.%. As shown in Figure 12b,c, an optimal Au content for the photodegradation reaction was found, as excess metal loading for metal-semiconductor heterostructures may compromise the effectiveness of the overall charge separation. Figure 12d also indicates that the photodegradation rate of ZnO-Pd was higher than that of ZnO-Au, while ZnO-Ag was the worst. This is due to the most positive Fermi level potential of Pd. Larger differences between the CB of ZnO and the Fermi level (EF) of the metal result in stronger driving forces for interfacial charge transfer, giving rise to the most efficient charge separation for ZnO-Pd and the most effective photodegradation of MB molecules. Moreover, the metal content and composition were tuned in a metal (Ag, Au, Pd)-decorated ZnO system for photodegradation of MB in ethanol [80]. The morphology of ZnO-Au with different Au contents is shown in Figure 12a. The Au content is obviously increasing and was measured to be 0.6 at.%, 1.0 at.%, 1.3 at.%, 2.0 at.%, and 2.3 at.%. As shown in Figure 12b,c, an optimal Au content for the photodegradation reaction was found, as excess metal loading for metal-semiconductor heterostructures may compromise the effectiveness of the overall charge separation. Figure 12d also indicates that the photodegradation rate of ZnO-Pd was higher than that of ZnO-Au, while ZnO-Ag was the worst. This is due to the most positive Fermi level potential of Pd. Larger differences between the CB of ZnO and the Fermi level (E F ) of the metal result in stronger driving forces for interfacial charge transfer, giving rise to the most efficient charge separation for ZnO-Pd and the most effective photodegradation of MB molecules.

Modification with semiconductors
CdS-CdSnO3 type-II heterostructures were also employed to investigate the effect of CdSnO3 content for the photodegradation of RhB [81]. Figure 13a shows the direct contact of CdS and CdSnO3, and the content of CdSnO3 was precisely controlled. The surface-decorated CdSnO3 acts as an efficient electron scavenger for CdS because of its lower CB level (+0.9 V vs. NHE) than the CB level of CdS (−0.5 V vs. NHE), giving rise to the fast PL decay of CdS-CdSnO3, as shown in the TRPL analysis in Figure 13b. As a result, the photoexcited electrons of CdS nanowires preferentially transferred to CdSnO3 nanocrystals, leaving photoexcited holes in the CdS domain to react with RhB molecules. As shown in Figure 13c, the photodegradation rate was enhanced with CdSnO3 content increasing from 1.25 at.% to 2.5 at.%, and then depressed with further increases in CdSnO3 content. This observation was in accordance with their interfacial charge transfer kinetics, indicating that an excess amount of CdSnO3 would reduce the overall charge separation efficiency and result in the depressed photocatalytic efficiency, as shown in Figure 13d. The amounts affecting the subsequent photodegradation performance were also proposed in ZnSe-ZnO [82], In2O3-TiO2-Pt [83], and Cu2O-rGO [84] heterostructures. These observations indicate that the interfacial charge transfer kinetics of the photocatalysts play an important role for the photodegradation of dye molecules. Moreover, the composition and facet effects of heterostructure systems such as TiO2-Au@Cu7S4 [85] and ZnS-Cu2O [86] also influence the photodegradation efficiency.

Modification with Semiconductors
CdS-CdSnO 3 type-II heterostructures were also employed to investigate the effect of CdSnO 3 content for the photodegradation of RhB [81]. Figure 13a shows the direct contact of CdS and CdSnO 3 , and the content of CdSnO 3 was precisely controlled. The surface-decorated CdSnO 3 acts as an efficient electron scavenger for CdS because of its lower CB level (+0.9 V vs. NHE) than the CB level of CdS (−0.5 V vs. NHE), giving rise to the fast PL decay of CdS-CdSnO 3 , as shown in the TRPL analysis in Figure 13b. As a result, the photoexcited electrons of CdS nanowires preferentially transferred to CdSnO 3 nanocrystals, leaving photoexcited holes in the CdS domain to react with RhB molecules. As shown in Figure 13c, the photodegradation rate was enhanced with CdSnO 3 content increasing from 1.25 at.% to 2.5 at.%, and then depressed with further increases in CdSnO 3 content. This observation was in accordance with their interfacial charge transfer kinetics, indicating that an excess amount of CdSnO 3 would reduce the overall charge separation efficiency and result in the depressed photocatalytic efficiency, as shown in Figure 13d. The amounts affecting the subsequent photodegradation performance were also proposed in ZnSe-ZnO [82], In 2 O 3 -TiO 2 -Pt [83], and Cu 2 O-rGO [84] heterostructures. These observations indicate that the interfacial charge transfer kinetics of the photocatalysts play an important role for the photodegradation of dye molecules. Moreover, the composition and facet effects of heterostructure systems such as TiO 2 -Au@Cu 7 S 4 [85] and ZnS-Cu 2 O [86] also influence the photodegradation efficiency.

Modification with metals and semiconductors
Na2-xTi3O7-Au-Cu2O Z-scheme heterostructure nanobelts (denoted as ST-Au-Cu2O NBs) were demonstrated to photodegrade MB molecules [87]. Figure 14a,b display the successful decoration of Au nanoparticles on ST and the further growth of Cu2O on Au surface. By decreasing the volume of the Cu 2+ -citrate precursor, the shell thickness of the grown Cu2O can be tuned from 1.4 nm and 1.2 nm to 1.1 nm, resulting in the modulation of interfacial charge transfer dynamics and, thus, the enhancement of photodegradation efficiency. As shown in Figure 14c, the ST-Au-Cu2O Z-scheme NBs showed higher photodegradation efficiency than ST-Au and ST-Cu2O type-II NBs, illustrating the superiority of Z-scheme heterostructures in photodegradation applications. Meanwhile, the highest photodegradation efficiency was achieved for ST-Au-Cu2O prepared with 50 µL of Cu 2+citrate, which can be attributed to the most efficient charge transfer dynamics at the Cu2O thickness of 1.1 nm. As Figure 14d presents, the electron-scavenging rate constant (kes) for ST-Au-Cu2O was enhanced with decrease in Cu2O shell thickness, which can be ascribed to the quantum size effect of Cu2O. Since decreasing Cu2O shell simultaneously expanded the bandgap of Cu2O, its CB and VB respectively shifted toward higher and lower potential, giving a larger driving force of interfacial electron transfer for ST-Au-Cu2O and thereby improving the photodegradation efficiency.   Figure 15 displays the redox potentials for the reactive species and the band structures of common photocatalysts, along with the lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO) levels of five representative dye molecules. Based on their band positions, the generation of reactive species in semiconductor photocatalysts can directly degrade dye molecules (see Section 5.1). Alternatively, the self-photosensitization of dye may occur to improve the generation of reactive species when the CB of photocatalysts is more negative than the LUMO level of dye molecules (see Section 5.2), finally dissociating the dye molecules.

Direct photodegradation process
In principle, the direct photodegradation of dye molecules involves the excitation of semiconductor photocatalysts under light irradiation, leading to the scavenging of photoexcited  Figure 15 displays the redox potentials for the reactive species and the band structures of common photocatalysts, along with the lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO) levels of five representative dye molecules. Based on their band positions, the generation of reactive species in semiconductor photocatalysts can directly degrade dye molecules (see Section 5.1). Alternatively, the self-photosensitization of dye may occur to improve the generation of reactive species when the CB of photocatalysts is more negative than the LUMO level of dye molecules (see Section 5.2), finally dissociating the dye molecules.  Figure 15 displays the redox potentials for the reactive species and the band structures of common photocatalysts, along with the lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO) levels of five representative dye molecules. Based on their band positions, the generation of reactive species in semiconductor photocatalysts can directly degrade dye molecules (see Section 5.1). Alternatively, the self-photosensitization of dye may occur to improve the generation of reactive species when the CB of photocatalysts is more negative than the LUMO level of dye molecules (see Section 5.2), finally dissociating the dye molecules.

Direct photodegradation process
In principle, the direct photodegradation of dye molecules involves the excitation of semiconductor photocatalysts under light irradiation, leading to the scavenging of photoexcited

Direct Photodegradation Process
In principle, the direct photodegradation of dye molecules involves the excitation of semiconductor photocatalysts under light irradiation, leading to the scavenging of photoexcited electrons by dissolved O 2 in the solution, as the CB of the photocatalyst is more negative than the reduction potential Meanwhile, the photoexcited holes can oxidize the adsorbed water to generate ·OH, as the VB of photocatalysts are more positive than the oxidation potential E(H 2 O/·OH). These highly reactive ·O 2 − and ·OH can oxidize or degrade the adsorbed dye molecules. In addition, the photoexcited electrons and holes which are essentially active may also attack the dye molecules to complete the photodegradation process.
To clarify the major contributor in the photodegradation reaction, several studies investigated the change of the photodegradation rate in the presence of different scavengers, as listed in Table 2. Pu et al. developed a Cu 2 O-rGO system (Figure 16a) and explored the photodegradation mechanism of MO, with TBA used as the ·OH scavenger [84]. Figure 16b shows various control experiments for the photodegradation of MO. No obvious change on MO photodegradation was found with the addition of TBA, indicating that the photoexcited holes of Cu 2 O and the subsequent ·OH radicals were minor factors. Furthermore, the photodegradation reaction was performed with purging using O 2 and N 2 . The MO photodegradation was abated under an N 2 purge, whereas the photodegradation rate was slightly enhanced under O 2 purging, confirming that dissolved O 2 in the solution played a crucial role for the MO photodegradation. Figure 16c shows   rate in the presence of AgNO3 is due to the decreased generation of ·O2 − from photoexcited electrons. This outcome was also consistent with the g-C3N4 [91], ZnO/graphene [92], and C3N4-BiVO4 [93] systems for the photodegradation of MB. The aforementioned scavenger experiments revealed that major contributors for each dye molecule might be different. Table 3 specifies the dominating active species for other commonly reported dye molecules.   Cao et al. proposed that the photodegradation of MB on AgSiO/Ag 2 CO 3 photocatalysts in water was dominated by ·O 2 − and photoexcited holes [90]. Figure 16d shows the morphology of AgSiO/Ag 2 CO 3 photocatalysts. With addition of IPA, the photodegradation rate slightly decreased, suggesting ·OH was not the main active species (Figure 16e). Meanwhile, the addition of EDTA-2Na and N 2 purging resulted in a significant decrease of photodegradation efficiency, indicating the important roles of ·O 2 − and photoexcited holes in the photodegradation process of MB. The possible photodegradation pathway is displayed in Figure 16f. This photodegradation pathway can be supported by adding AgNO 3 into the reaction solution. The suppression of the photodegradation rate in the presence of AgNO 3 is due to the decreased generation of ·O 2 − from photoexcited electrons.
This outcome was also consistent with the g-C 3 N 4 [91], ZnO/graphene [92], and C 3 N 4 -BiVO 4 [93] systems for the photodegradation of MB. The aforementioned scavenger experiments revealed that major contributors for each dye molecule might be different. Table 3 specifies the dominating active species for other commonly reported dye molecules.

Sensitization-Mediated Degradation Process
When the photon energy is not high enough to excite photocatalysts to generate reactive charge carriers and radicals, the photodegradation might occur via photosensitization process. Under visible light illumination, a dye molecule can be excited to its excited state (LUMO level), producing abundant excited electrons at the LUMO level. Provided that the LUMO of the dyes is more negative than the CB of photocatalysts, these photoexcited electrons can then transfer from the dye molecules to the photocatalysts, facilitating the generation of reactive species for commencing photodegradation. This process is known as the photosensitization pathway. The LUMO levels for five representative dye molecules in comparison with the CB levels of common photocatalysts are illustrated in Figure 15, from which one can tell whether or not the photosensitization can occur. Zhao et al. demonstrated the self-photosensitization process of RhB and MO over BiOCl hierarchical nanostructures [63], as shown in Figure 17a. Note that the as-synthesized BiOCl exhibits a negatively charged surface in the pH range from 2 to 11, suggesting the as-synthesized BiOCl can selectively adsorb cationic RhB in RhB/MO mixtures. As shown in Figure 17b,c, RhB showed almost 40% adsorption prior to the light irradiation, while MO showed a negligible adsorption. Upon visible light irradiation, although BiOCl nanostructures cannot be excited with visible light since the bandgap of BiOCl is approximately 3.3 eV, the RhB dye was completely degraded within 15 min, whereas no photodegradation was found for the MO/BiOCl system (Figure 17d). The RhB photodegradation was attributed to the self-photosensitization of RhB, in which the photoexcited electrons are injected from the LUMO level of RhB to the CB of the BiOCl nanostructures. The electrons on the BiOCl nanostructures can subsequently reduce O 2 to ·O 2 − radicals, which further degrades the RhB molecules. The self-photosensitization of MO should also be able to degrade the MO molecules; however, its poor adsorption capacity inhibits the electron transfer from the excited MO to the CB of BiOCl. Therefore, BiOCl nanostructures exhibited poor MO photodegradation efficiency. The self-photosensitization of the RhB/BiOCl system can be further applied to the photodegradation of MO dye molecules, as shown in Figure 17e. Compared to the extremely low MO photodegradation efficiency in the MO/BiOCl system, apparently, the MO photodegradation efficiency was significantly enhanced in the MO/RhB/BiOCl system, indicating that MO photodegradation is mediated by RhB via a photosensitization pathway. Moreover, RhB still demonstrated a higher photodegradation efficiency in the MO/RhB/BiOCl system. This photodegradation of dye molecules via the photosensitization process was reported in the cationic new fuchsin/graphene quantum dots [114], RhB/Zn-doped BiOBr [115], MB and MO/Eu 3+ -doped ZnO [116],

Summary and Outlook
Most dyes have either a positive or a negative charge upon dissociation in aqueous solutions, characteristic of cationic and anionic dyes, respectively. The pH of the solution modifies the electrical double layer of the photocatalyst, affecting the interaction between dye molecules and photocatalysts, the charge transfer for the self-photosensitization of dyes, and the subsequent decoloration efficiency, as highlighted in this review. The optimization of other operational factors (i.e., initial dye concentration, light intensity, reaction temperature) and charge-carrier properties of heterostructure photocatalysts can establish the most efficient photodegradation system. In addition, research on the reactive species, as well as the photosensitization pathway of dye molecules for photodegradation, is helpful to clarify the overall decoloration mechanism, which is beneficial for further application to the degradation of multicomponent industrial wastewaters. With strong progress in photocatalytic techniques, more in-depth characterization and modeling of the photodegradation, and even mineralization processes for multicomponent dyes in practical applications will be possible in the near future. Clearly, many questions still remain unanswered or are poorly addressed, which are briefly outlined below.
Unlike other heterogeneous photocatalysis, such as photoelectrochemical water splitting and CO2 reduction in which the solar-to-hydrogen (STH) and AQY are introduced to determine the

Summary and Outlook
Most dyes have either a positive or a negative charge upon dissociation in aqueous solutions, characteristic of cationic and anionic dyes, respectively. The pH of the solution modifies the electrical double layer of the photocatalyst, affecting the interaction between dye molecules and photocatalysts, the charge transfer for the self-photosensitization of dyes, and the subsequent decoloration efficiency, as highlighted in this review. The optimization of other operational factors (i.e., initial dye concentration, light intensity, reaction temperature) and charge-carrier properties of heterostructure photocatalysts can establish the most efficient photodegradation system. In addition, research on the reactive species, as well as the photosensitization pathway of dye molecules for photodegradation, is helpful to clarify the overall decoloration mechanism, which is beneficial for further application to the degradation of multicomponent industrial wastewaters. With strong progress in photocatalytic techniques, more in-depth characterization and modeling of the photodegradation, and even mineralization processes for multicomponent dyes in practical applications will be possible in the near future. Clearly, many questions still remain unanswered or are poorly addressed, which are briefly outlined below.
Unlike other heterogeneous photocatalysis, such as photoelectrochemical water splitting and CO 2 reduction in which the solar-to-hydrogen (STH) and AQY are introduced to determine the photocatalytic efficiency for comparison, a quantitative comparison with AQY is relatively unexplored for photodegradation reactions. The introduction of AQY into the photodegradation reaction provides helpful information to understand the efficiency of semiconductor photocatalysts. However, for visible-light-responsive photocatalysts, the feasible determination of photodegradation efficiency is especially challenging due to the relatively high photoabsorption coefficient of dye molecules, as well as the complicated mechanism of dye degradation [118].
Another major concern in reactions involving self-photosensitization of dye molecules is the interactions between photocatalysts and multiple types of dye molecules, which are not always clear. Limited studies described the underlying interactions and mechanisms. By using self-photosensitization of dye molecules, both cationic and anionic dyes can be simultaneously decomposed, which has the potential to degrade multicomponent dyes in real wastewater systems. However, for multicomponent dye systems, the quantification of efficiency for each individual dye molecule is difficult to determine with simple spectrophotometric methods especially when their absorption spectra overlap. The development of a new method to easily and reliably obtain quantitative detection of dye molecules is necessary for accurate estimation.