Towards the Sustainable Production of Ultra-Low-Sulfur Fuels through Photocatalytic Oxidation

Abstract

Nowadays, the sulfur-containing compounds are removed from motor fuels through the traditional hydrodesulfurization technology, which takes place under harsh reaction conditions (temperature of 350–450 °C and pressure of 30–60 atm) in the presence of catalysts based on alumina with impregnated cobalt and molybdenum. According to the principles of green chemistry, energy requirements should be recognized for their environmental and economic impacts and should be minimized, i.e., the chemical processes should be carried out at ambient temperature and atmospheric pressure. This approach could be implemented using photocatalysts that are sensitive to visible light. The creation of highly active photocatalytic systems for the deep purification of fuels from sulfur compounds becomes an important task of modern catalysis science. The present critical review reports recent progress over the last 5 years in heterogeneous photocatalytic desulfurization under visible light irradiation. Specific attention is paid to the methods for boosting the photocatalytic activity of materials, with a focus on the creation of heterojunctions as the most promising approach. This review also discusses the influence of operating parameters (nature of oxidant, molar ratio of oxidant/sulfur-containing compounds, photocatalyst loading, etc.) on the reaction efficiency. Some perspectives and future research directions on photocatalytic desulfurization are also provided.

1. Introduction

Desulfurization of fuels such as diesel, gasoline, kerosene, and jet fuel has been a challenging operation and remains critical to the petrochemical industry [1]. The main naturally occurring sulfur-containing organic compounds (SCCs) are sulfides, disulfides, mercaptans, thiophene (Th) and its derivatives (benzothiophene (BT), dibenzothiophenes (DBTs), 4-methylbenzothiophene (4-MBT), 4,6-dimethyldibenzothiophene (4,6-DMDBT), 3,7-dimethyldibenzothiophene (3,7-DMDBT), and 2,8-dimethyldibenzothiophene (2,8-DMDBT)) [2]. The presence of these SCCs in fuels is undesirable since they create problems during refining, namely deactivation of some catalysts and corrosion of equipment. Moreover, sulfur compounds release toxic SOx and cause severe environmental problems: water and air pollution, global warming, ecological instability, as well as the harmful impact on living organisms [2,3]. As a result, many countries (USA, European Union, Japan, China, etc.) have introduced strict standards to limit the content of sulfur in fuels to 10 ppm [4].

The conventional technology for the removal of SCCs from fuels is hydrodesulphurization (HDS) through catalytic hydrogenation at high temperatures (350–450 °C) and high pressures (30–60 atm). This technique is characterized by a high efficiency in eliminating sulfides, disulfides, and mercaptans by converting them to hydrogen sulfide (Scheme 1) [5]. The main disadvantages of HDS are high energy consumption due to the harsh reaction conditions and high consumption of costly hydrogen. Moreover, the HDS catalysts (Co or Ni promoted by Mo or W sulfide and supported onto γ-Al₂O₃) are not efficient in the converting of large sulfur-containing molecules (e.g., DBTs and DMDBTs) due to the steric hindrance [6]. To overcome the disadvantages of the conventional HDS technology, alternative approaches, including adsorptive desulfurization (ADS), extractive desulfurization (EDS), biodesulfurization (BDS), and oxidative desulfurization (ODS), have been actively developed.

ADS involves the removal of SCCs by a physicochemical adsorption process that proceeds at a low temperature and pressure and does not require hydrogen. A wide range of adsorbents based on metal oxides, zeolites, and metal-organic frameworks (MOFs) have been proposed for ADS [4,7]. ADS makes it possible to obtain an ultra-high desulfurization efficiency (<10 ppm of S), but the regeneration of adsorbents is rather limited [8,9]. EDS is also investigated under mild reaction conditions in the presence of various extractants, including acetonitrile, methanol, N,N-dimethylformamide, dimethylsulfoxide, and pyrrolidone [10]. The use of these solvents is associated with emerging environmental and safety issues like wastewater emission and fire hazards. Environmental problems could be solved by using ionic liquids as extractants in EDS [11]. However, their application in large-scale processes is limited by high cost. BDS involves a simple installation process, low energy consumption and operating costs, and mild reaction conditions [12]. The main limitation of this approach is that a noteworthy amount of carbon is mineralized, which reduces the fuel value [5].

Much attention in recent years has been paid to the ODS technique as an efficient technology for deep desulfurization [13,14]. The ODS reaction includes the oxidation of SCCs to corresponding sulfoxides and sulfones in the presence of an oxidizing agent and a catalyst under mild reaction conditions (Scheme 2). ODS is often combined with EDS to separate the oxidized compounds from the mixture using pure polar solvents (e.g., acetonitrile, methanol, etc.) [15]. The use of polar solvents makes it possible to obtain a high desulfurization efficiency by the migration of the obtained sulfoxides and sulfones into the polar phase, but it is associated with the environmental issues. The most preferred route is complete mineralization of SCCs to CO₂ and SO₄²⁻ without the adding of any extractants into the reaction mixture.

As one of the ODS methods, photocatalytic oxidation desulfurization (PODS), where reactive oxygen species (ROS) [16] to reduce/oxidize the C−S−C bond are generated, has boosted a surge of scientific interest as one of the most attractive alternative routes to transform naturally abundant, clean, and sustainable solar energy into chemical energy [17,18]. The interest in this approach is also proven by the significant increase in the number of publications where PODS is used for deep desulfurization (Figure 1).

In this critical review, the recent achievements in the field of heterogeneous PODS are highlighted, and the conversions of SCCs as well as the mechanisms are discussed. Section 2 is devoted to a brief description of the general principles of the photocatalytic desulfurization, including the main ROS formed during the reaction, the most common photocatalysts for PODS, and the methods for boosting their activity. In the following Section, the current status in the development of highly active photocatalysts for PODS and the main strategies for boosting the activity of semiconductor materials, with their advantages and disadvantages, are described. The influence of operating parameters on the desulfurization efficiency is discussed in Section 4. The authors believe that this review will help colleagues in the scientific community who are actively involved in the development of photocatalytic systems for desulfurization.

2. Photocatalysis and Desulfurization

2.1. General Principles

According to the International Union of Pure and Applied Chemistry (IUPAC), photocatalysis is ‘a change in the rate of a chemical reaction or its initiation under the action of ultraviolet, visible, or infrared radiation in the presence of a substance, i.e., a photocatalyst, which absorbs light and is involved in the chemical transformation of the reaction partners’ [19]. The photocatalytic stimulation of a reaction (A → B) in the presence of a catalyst (K) in the general form can be written as follows:

A + K + hν → B + K

(1)

when the energy of a photon is higher than the band gap energy (E_g) of a semiconductor photocatalyst, it can be absorbed, resulting in the promotion of an electron from the valence band (VB) to the conduction band (CB), and leading to the formation of a hole in the VB. After entering the excitation zone, the electron becomes mobile and has a significant reduction potential. Generally, this electron (in the case of its transfer to the semiconductor surface) can be considered a strong one-electron reductant. The formed hole is also very mobile and exhibits the properties of a one-electron oxidizer [20,21].

Several ROS can be formed by interaction with the electrons and holes and then reduce/oxidize the C−S−C bond in SCCs [16]. Specifically, O₂ may react with photogenerated electrons to form superoxide radicals (^•O₂⁻):

O₂ + e⁻ → ^•O₂⁻

(2)

The generated ^•O₂⁻ could be reduced by photoinduced CB electrons to form hydrogen peroxide (H₂O₂):

^•O₂⁻ + 2H⁺ + e⁻ → H₂O₂

(3)

Hydroxyl radicals (^•OH) with the highest reactivity among ROS could be formed in several ways: (i) the oxidation of water (Equation (4)), (ii) the reaction of the photogenerated electrons with H₂O₂ (Equation (5)), and (iii) the oxidation of surface hydroxyls by the photogenerated holes (Equation (6)):

H₂O + h⁺ → ^•OH + H⁺

(4)

H₂O₂ + e⁻ → ^•OH + OH⁻

(5)

OH⁻ + h⁺ → ^•OH

(6)

It should be noted that in many cases superoxide radicals play a crucial role in the oxidation of SCCs when molecular oxygen or air are used as oxidants [22,23], while hydroxyl radicals are involved in PODS when H₂O₂ is added to the reaction mixture for the oxidation of sulfur-containing compounds [24,25].

2.2. The Most Common Photocatalysts for Desulfurization

The publication of Fujishima and Honda is considered to be a real breakthrough in photocatalysis [26]. Since then, photocatalysis has been mainly studied in the research fields of pollutant degradation [27], air purification [28], water splitting [29], organic transformations [30], and carbon dioxide reduction [31]. Nowadays, the most common types of semiconductor catalysts used in various photocatalytic reactions are inexpensive and naturally abundant transition metal oxides, especially titanium dioxide (TiO₂).

In 2002, Matsuzawa et al. [32] demonstrated that the conversion of DBT and 4,6-DMDBT in acetonitrile over TiO₂ was about 40% after 10 h under ultraviolet (UV) irradiation. The authors concluded that this method is not applicable for desulfurization of fuels due to low conversion of the substrate. However, the main disadvantage of this method is associated with the use of titanium dioxide as a photocatalyst. It is well-known that TiO₂ crystalline phases (anatase, rutile, and brookite) can only absorb UV light due to their large band gap of 3.0–3.4 eV. In order to ensure efficient solar energy utilization, the development of new photocatalytic materials, which are sensitive to visible light, and methods for shifting the photosensitivity of TiO₂-based catalysts to the visible region (E_g < 3.0 eV) are important directions in PODS.

In this regard, various ranges of photocatalysts have been proposed for the desulfurization reaction, including visible light-responsive metal oxides [33,34], perovskites [25,35], graphitic carbon nitride (g-C₃N₄) [36,37], and metal-organic frameworks (MOFs) [38,39]. However, pure semiconductor photocatalysts are characterized by low activity in PODS due to the rapid recombination of photogenerated electron–hole pairs. To improve their photocatalytic activity under visible light irradiation, various methods have been investigated: (i) band gap engineering, i.e., non-metal doping [40], (ii) deposition of a metal co-catalysts [41,42], and (iii) construction of heterojunctions [23,43]. Another interesting approach is the combination of PODS with the ADS technique. For this purpose, a photocatalytic material is supported onto mesoporous materials like MCM-41 [44] or Al-SBA-15 [24].

3. Advances in Photocatalytic Desulfurization under Visible Light

This Section is devoted to the current status of the development of highly active photocatalysts for PODS, along with the main advantages and disadvantages of strategies for boosting the activity of semiconductor materials.

Despite the fact that unmodified photocatalysts exhibit relatively low activity in PODS due to the rapid recombination of electron–hole pairs, there have been several reports showing a high efficiency of pure materials (Table 1). For instance, Dedual and colleagues [45] investigated the effect of operating parameters on the photocatalytic performance in desulfurization of BT and DBT. The authors demonstrated that optimal sulfur removal efficiency (91%) after 3 h occurred with an operating temperature of 40 °C, 0.7% vol% H₂O₂ in a methanol solvent, 6 g·L⁻¹ TiO₂ loading, a methanol-to-fuel molar ratio of 1, and an initial pH of 4. Although the desulfurization efficiency was relatively high, the use of UV irradiation is not practical in terms of solar light utilization.

It was shown that the Fe₂O₃ photocatalyst is an interesting candidate for the application in PODS [33]. Comparison of the activity of α- and β-Fe₂O₃ showed that the material containing 36.6% β-Fe₂O₃ and 63.4% α-Fe₂O₃ (E_g = 1.82 eV) exhibited the highest photocatalytic activity (92.3% after 90 min) under visible light. While the stability of this photocatalyst remains unknown, the leaching of iron (ca. 0.49 at% of the used Fe₂O₃) was detected. Thus, it can be assumed that the proposed photocatalyst will lose its activity after each PODS cycle.

It was reported that complex oxides NiCo₂O₄ [46], LaVO₄ [47,48], and Ag₃VO₄ [49] were active in the photocatalytic desulfurization under visible light. For instance, a mesoporous Ag₃VO₄ semiconductor, which was synthesized by a hydrothermal approach, exhibited excellent photocatalytic oxidative desulfurization activity up to 92% after 3 h (500 W Xe lamp as a visible light source) [49]. The photocatalyst demonstrated good stability and reusability. The desulfurization activity performance of Ag₃VO₄ was decreased slightly to 91% after six cycles. A comparable SCC removal rate (88%) was obtained using mesoporous lanthanum vanadate LaVO₄ [47]. An important advantage of the Ag₃VO₄ and LaVO₄ photocatalysts is the ability to oxidize sulfur-containing compounds in the presence of air.

Table 1. Activity of unmodified materials in PODS.

No.	Photocatalyst	SCC (C (ppm)) 1	Oxidant	Visible Light Source	Extractant	t (min) 2	X (%) 3	Main Products 4	Ref.
1	TiO2-P25	DBT (950)	H2O2	8 W UV lamp	MeOH	180	91	DBTO2	[45]
2	Fe2O3	DBT (500)	Air	350 W Xe lamp	H2O	90	92	CO2	[33]
3	LaVO4	SCCs of real diesel (410)	Air	500 W Xe lamp	MeOH	180	88	Sulfones	[47]
4	Ag3VO4	SCCs of real diesel (130)	Air	500 W Xe lamp	2-Ethoxyethanol	180	92	Sulfones	[49]

¹ The sulfur concentration in the model fuel is indicated in parentheses; ² reaction time; ³ desulfurization degree; ⁴ DBTO₂—dibenzothiophene sulfone.

Table 1. Activity of unmodified materials in PODS.

No.	Photocatalyst	SCC (C (ppm)) 1	Oxidant	Visible Light Source	Extractant	t (min) 2	X (%) 3	Main Products 4	Ref.
1	TiO2-P25	DBT (950)	H2O2	8 W UV lamp	MeOH	180	91	DBTO2	[45]
2	Fe2O3	DBT (500)	Air	350 W Xe lamp	H2O	90	92	CO2	[33]
3	LaVO4	SCCs of real diesel (410)	Air	500 W Xe lamp	MeOH	180	88	Sulfones	[47]
4	Ag3VO4	SCCs of real diesel (130)	Air	500 W Xe lamp	2-Ethoxyethanol	180	92	Sulfones	[49]

¹ The sulfur concentration in the model fuel is indicated in parentheses; ² reaction time; ³ desulfurization degree; ⁴ DBTO₂—dibenzothiophene sulfone.

3.1. Band Gap Engineering (Non-Metal Doping)

Band gap engineering is of great importance for the creation of materials sensitive to visible light and their subsequent application in the photocatalysis. There are three ways of reducing the band gap: shifting the CB minimum, shifting the VB maximum, and introducing impurity levels in the band gap [50,51]. The most common method for band gap engineering is the doping of metal oxides, especially TiO₂, into anionic positions with nitrogen, carbon, sulfur, and phosphorus. This approach leads to the formation of a new isolated impurity level, i.e., the N 2p band above the O 2p valence band, which eventually decreases the E_g of the material and shifts the optical absorption to the visible light region (Figure 2a). The main disadvantage of the non-metal doping is associated with the leaching of the dopant [52].

Following this methodology, Kalantari et al. [40] employed an N–TiO₂ photocatalyst for the photocatalytic desulfurization under ambient conditions without any solvents using air as an oxidant. The light absorption extended to the visible region with a red shift in absorption edge for the N-TiO₂ nanoparticles (Figure 2b), which resulted in the enhancement of the visible light absorption and photocatalytic activity. As a result, the conversion of DBT using N-TiO₂ was higher than that of the TiO₂-P25 by a factor of 4.7 times. Unfortunately, the desulfurization degree in the presence of the N-doped photocatalyst was low and did not exceed 40% after 4 h. The doping with nitrogen of 2D CeO₂-TiO₂ nanosheets allowed obtaining a much higher conversion of DBT of 94% after 3 h [53]. The enhancement of photocatalytic activity could be attributed to the unique nano thin layer structure of the material, self-doping of biological nitrogen, porous structure, and high surface area. However, a comparison of the photocatalytic activity of the N–TiO₂ nanoparticles [40] and N–CeO₂–TiO₂ nanosheets [53] is quite difficult, since the PODS processes were performed in different reaction conditions, namely sulfur content, nature of oxidant, and light source.

There are several investigations in which graphene was used as a visible light-responsive photocatalyst for PODS [54,55]. Graphene is a very promising nanomaterial that has piqued the curiosity of scientists due to its unique optical, electrical, and physicochemical properties [56,57]. Ma and colleagues prepared core-shell N-doped graphene nanosphere-anchored bimetallic single atoms and investigated them in PODS using H₂O₂ as an oxidant, methanol as a solvent, and a 500 W Xe lamp as a visible light source [55]. The authors found that the hollow core-shell N-doped graphene decorated with Ni/Cu had a high photocatalytic removal ratio (>99.1%). The photocatalyst demonstrated high activity even after ten cycles because of the synergism of single Ni/Cu atoms and the hollow core-shell N-doping-graphene under visible light. The main disadvantage of the N-doped graphene nanospheres decorated with Ni/Cu is the multi-stage and complex nature of the preparation method, which makes it difficult to use the photocatalyst in a large-scale process.

Despite the fact that the incorporation of non-metals is one of the common approaches for the enhancement of the photocatalytic activity of various materials, this method is often accompanied with leaching of the dopant. The most promising method for obtaining stable and active photocatalysts for PODS is considered to be doping with metals.

3.2. Metal Doping

In the case of the doping strategy, many metals have been employed for improving the photocatalytic activity, including transition and inner transition metals (Er, Zn, Mn, Ce, Nd, Pr, Sm, Sn, Al, Ti, Ni, Fe) as well as noble metals (Pt, Au, Ag) [58]. The use of a noble metal co-catalyst is the most followed approach, which has been widely used for the modification of TiO₂ [59,60], ZnO [61], g-C₃N₄ [62], CdS [63], etc. In the presence of contact with noble metal nanoparticles, the charge distribution in composite materials is determined by the position of the Fermi level of the metal relative to the band edges in the semiconductor (Figure 3). The Fermi levels in the contacting metal and semiconductor particles are aligned due to the thermionic emission currents, which leads to the formation of a space charge at the metal–semiconductor contact. If the work function of the metal is greater than the Fermi level of the semiconductor, the Schottky barrier is formed. In this case, the positive charges (h⁺) are localized on the semiconductor particle, and the negative charges (e⁻) are located on the metal particle. This spatial separation of electrons and holes reduces the recombination of electron–hole pairs, leading to an increase in the photocatalytic activity. Metal nanoparticles also introduce changes in the absorption spectrum of hybrid materials due to the localized surface plasmon resonance (LSPR) [64,65].

In the case of the doping strategy, many metals have been employed for improving the activity materials in PODS. The most common approach is doping with noble metals [6,24,41,42,66,67,68]. For instance, Wang et al. [66] demonstrated that the activity of Ag-loaded Bi₂WO₆ in PODS is greatly enhanced compared with the pure Bi₂WO₆. The authors concluded that the incorporation of the Ag nanoparticles led to the efficient separation and preventing of the recombination of electron–hole pairs. Similar observations were found for the Ag/Fe₃O₄/graphene ternary nanocomposite [68] (Figure 4). In the Ag/Fe₃O₄/graphene ternary nanocomposite, Fe₃O₄ adsorbs the visible light photons and generates the electrons and holes on its CB and VB, respectively. The Ag nanoparticles and graphene can act as acceptors for the induced electrons and holes of Fe₃O₄, respectively, to facilitate the separation of electron–hole pairs and interfacial charge transfer. In turn, the LSPR phenomena of the Ag nanoparticles promotes the visible-light absorption of the ternary nanocomposite. Thus, the synergism arising from the separation of electron–hole pairs and LSPR led to a significant increase in the photocatalytic activity of Ag/Fe₃O₄/graphene (95%) compared to Ag/graphene, Fe₃O₄/graphene, and other samples.

A green approach was proposed by Chen and colleagues [42], who synthesized the Ag–TiO₂ photocatalyst supported on porous glass and used it in the desulfurization of model fuel containing 50 ppm of DBT or BT without adding external oxidants such as O₂ and H₂O₂. It is well known that the coexistence of fuel and O₂ or H₂O₂ could trigger an explosion accident when applied in industrial applications. The authors proposed to use ethanol, which serves as an electron acceptor by consuming valance band holes to suppress the recombination of photogenerated electron–hole pairs [69] and acts as a source of highly active hydroxyl radicals. As a result, the desulfurization efficiency was about 84% after 80 min. The prepared photocatalyst was also active in the photodegradation of Rhodamine B, methylene blue, and methyl orange, with the rate constant of 0.14, 0.18, and 0.055 min⁻¹, respectively. The development of efficient photocatalysts for the removal of dyes is also at the forefront of catalysis science, because organic dyes, which remain in the effluents of the textile industry, are usually persistent and difficult to degrade by conventional wastewater treatment techniques [70,71].

There are several recent reports where it is suggested to use rare metals as dopants [72,73,74]. Despite the fact the Ir/Pr–N–CQDs–TiO₂ [72], Pr/Ce–N–TiO₂ [73], and Er/W–N–TiO₂ [74] photocatalysts provide a high degree of SCC removal, rare metals are characterized by a high price. Noble metals also have a high price; therefore, investigations should be shifted towards using Earth-abundant metals such as Cu [75], Ni [76], Na [77], and Mo [25] as co-catalysts. For instance, Zhang et al. [77] prepared Na-doped g-C₃N₄ for the photocatalytic denitrogenation and desulfurization for fuels under visible light irradiation using molecular O₂ as an oxidant to substitute for the expensive H₂O₂. They demonstrated that a moderate amount of Na in g-C₃N₄ generated the highly dispersed and porous nanosheets, which further improved the surface energy and reduced the recombination rate of electron–hole pairs. Moreover, the synthesized photocatalysts were characterized by the ability to absorb more visible light since the calculated band gaps of g-C₃N₄ were 2.70 eV, while Na-doped g-C₃N₄ were in the range of 2.02–2.44 eV, depending on the amount of the dopant. An important advantage of the proposed method is the complete conversion of N- and S-containing organic compounds into CO₂ without the formation of intermediate oxygenate products (Figure 5). The complete mineralization of SCCs was also detected by Belousov and colleagues [25], who employed nanosized Bi₂W_xMo_1−xO₆ solid solutions with various compositions as photocatalysts for PODS. The implementation of complete oxidation will not lead to the additional separation of the oxidation product from the hydrocarbon mixture by extraction methods with toxic solvents.

As can be seen from

Table 2. Activity of non-metal and metal doped materials in PODS.

No.	Photocatalyst	SCC (C (ppm))	Oxidant	Visible Light Source	Extractant	t (min)	X (%)	Main Products	Ref.
Non-metal doping
1	N–TiO2	DBT (70)	Air	55 W Xe lamp	None	240	40	DBTO2	[40]
2	N–CeO2–TiO2	DBT (100)	H2O2	300 W Xe lamp	None	180	94	DBTO2	[53]
3	N-doped graphene	DBT (175)	H2O2	500 W Xe lamp	MeOH	120	99	DBTO2	[55]
Metal doping
4	Ag–Bi2WO6	Th (500)	Air	500 W Xe lamp	None	180	88	Not specified	[66]
5	Ag–BiVO4	Th (500)	Air	400 W halide lamp	MeCN	210	95	CO2	[6]
6	Pt–RuO2/BiVO4	Th (600)	O2	300 W Xe lamp	None	180	99	CO2	[67]
7	Ag–TiO2/porous glass	DBT (10)	EtOH	50 W LED lamp	EtOH	80	84	Not specified	[42]
8	Ag/Fe3O4/graphene	Th (700)	Air	400 W Osram lamp	None	120	95	Sulfone	[68]
10	Pr/Ce–N–TiO2	DBT (200)	H2O2	300 W Xe lamp	MeOH	40	100	Not specified	[73]
11	Er/W–N–TiO2	DBT (200)	H2O2	300 W Xe lamp	MeOH	60	100	DBTO2	[74]
12	N,Ni/(Na0.5Bi0.5)TiO3−BaTiO3	DBT (100)	TBHP	300 W Xe lamp	None	150	90	Not specified	[76]
13	Na–g-C3N4	Th (200)	O2	300 W Xe lamp	None	360	89	CO2	[77]
14	Bi2W0.5Mo0.5O6	DBT (200)	H2O2	30 W LED lamp	None	120	100	CO2	[25]

, the doping strategy has been an interesting approach to improve the photocatalytic desulfurization reaction. The deposition of a noble metal co-catalyst is the most followed route. However, the use of noble metals is the main disadvantage of this approach because of their high price. Some investigations involve the application of Earth-abundant metals, but the problem of using dangerous oxidizing agents to initiate the reaction or toxic extractants has not been solved.

3.3. Creation of Heterojunctions

As mentioned earlier, doping with various metals allows the separation of the photogenerated charge carriers to be obtained. However, the most common dopants, namely noble metals, are characterized by a high price. Moreover, in some cases, the leaching of a dopant was observed. Thus, the stability of a photocatalyst remains unclear.

For the spatial separation of electron–hole pairs in photocatalysts, the creation of heterojunctions has been proven to be one of the most promising ways for the preparation of advanced materials [78,79,80,81]. There are three conventional types of heterojunction photocatalysts (Figure 6a), namely with a straddling gap (type-I), with a staggered gap (type-II), and with a broken gap (type-III). Among them, the type-II heterostructures, which can be constructed using various semiconductors, are the most useful in the field of photocatalysis because type-I and -III heterostructures do not provide an effective separation of the electron–hole pairs [71,78]. On the other hand, the main limitation for the type-II heterojunction is that the oxidation and reduction abilities of the transferred e⁻ and h⁺ decrease because of the electrons migrating to the CB of the semiconductor 1 (SC1), with lower reduction potential and holes accumulating on the VB of the semiconductor 2 (SC2) with lower oxidation potential [82]. Therefore, in addition to the type-II, the construction of Z- and S-scheme heterostructures (Figure 6b) is a promising direction in photocatalysis, since their development should address the disadvantages of the type-II heterojunctions. In the Z-scheme heterostructures, the photogenerated electrons from the CB of the SC1 with a lower reduction potential migrate to the VB of the SC2 with a lower oxidation potential owing to the electrostatic attraction between the electrons and holes. The S-scheme (Step-scheme) heterojunctions are characterized by a superior redox ability, and their efficiency has been proven in environmental remediation [83,84,85,86], water splitting [87,88,89], and CO₂ reduction [90,91,92]. The heterojunction approach may become the most promising route for the preparation of highly active materials for PODS, since a wide range of semiconductors (metal oxides, g-C₃N₄, complex oxides, MOFs, etc.) can be coupled and then used in the reaction. This allows for combining the different favorable properties of each compound, extending their absorption range, improving their chemical stability towards photocorrosion, and decreasing the recombination of electron–hole pairs [52,71,93].

There are several examples where the heterojunction strategy was used for the preparation of photocatalysts for desulfurization. The main efforts are aimed to obtain type-II and Z-scheme heterostructures.

3.3.1. Type-II Heterojunctions

Among a huge number of semiconductors, g-C₃N₄ has attracted much attention for the construction of type-II heterojunctions due to its visible light response, low cost, ease of preparation, chemical stability, non-toxicity, and appropriate energy levels and band structure [94,95]. For instance, Li et al. [36] prepared Ti₃C₂/g–C₃N₄ composite for the photocatalytic oxidative denitrogenation and desulfurization under visible light irradiation. The photocatalyst exhibited an enhanced photocatalytic performance, high mineralization efficiency, and good recyclability in the removal of small molecules of pyridine and thiophene. The authors demonstrated that the holes acted as the major active species to form the reactant intermediates, while the electrons and superoxide radicals provided a promotion effect on the final conversions. A similar mechanism of PODS was detected by Zhou et al. [23], who investigated the photocatalytic activity of the Ag₂O/Na–g-C₃N₄ type-I heterojunction photocatalyst. Another example of the type-II heterojunction was proposed by Lu and colleagues [96], who synthesized a novel CeF₃/g-C₃N₄ composite for PODS by a microwave hydrothermal method. It was found that the CeF₃ nanoparticles play an important role in the photocatalytic process because of (i) the upconversion fluorescence effect of CeF₃ could extend the light absorption range of the photocatalyst and (ii) CeF₃ coupled with g-C₃N₄ makes an efficient charge separation. On the other hand, the conversion of DBT decreased from 84 to 60% after the third photocatalytic cycle, and further experiments are needed to optimize the composition of the CeF₃/g-C₃N₄ photocatalyst and enhance its stability.

The main disadvantages of g-C₃N₄ are the limited electron migration rate caused by poor conductivity and the fewer active sites owing to the low specific surface area and poor hydrophilicity [71]. Thus, it is necessary to explore other materials to obtain type-II heterojunctions such as metal oxides, perovskites, MOFs, etc.

Recently, the photocatalytic activity of Ag₂O/ZrO₂ heterostructures in the oxidative desulfurization of thiophen was investigated using molecular oxygen as an oxidant and a 300 W Xe lamp as the light source [97]. The prepared photocatalysts significantly enhanced the photocatalytic desulfurization performance compared to the pure ZrO₂ nanoparticles. The authors declared that the sulfur-containing compound completely degraded to CO₂ under the reaction conditions due to the formation of highly active hydroxyl radicals (Figure 7). The first stage of the reaction is the formation of electrons and holes under visible light irradiation. Then, the photogenerated holes migrate to the VB of ZrO₂, while the electrons can be transferred to the CB of Ag₂O. After that, ^•OH radicals are created as a result of the photoinduced holes trapped by OH⁻ and the conversion of H₂O₂ formed from superoxide. The complete decomposition of thiophene was also observed on the Cu₂O-CeO₂ photocatalyst [98].

Metal oxides could be also coupled with perovskites and mixed oxides to obtain type-II heterojunctions [22]. Cui and colleagues [22,99,100,101,102] investigated the photocatalytic activity of bismuth vanadate (BiVO₄) modified by CuO in the PODS of model oil. The results showed that the photocatalytic activity of CuO modified BiVO₄ for organic sulfur compounds under visible light was significantly higher than that of the pristine BiVO₄. It should be noted that the conversion of DBT slightly decreased from 91 to 82% after ten reaction cycles, indicating a relatively good recyclability of the proposed photocatalyst. It should be noted that the photocatalysts based on BiVO₄ also demonstrated a relatively good stability in other studies. For instance, the Cu/Cu₂O/BiVO₄ nanoparticles with a rectangular cube morphology and a size of about 50 nm exhibited 92% photocatalytic desulfurization efficiency at 150 min under visible light irradiation without a significant loss in activity after five cycles [99] (Figure 8a). The study of the mechanism of the photocatalytic oxidative desulfurization over Cu/Cu₂O/BiVO₄ showed that holes and superoxide radicals were the active species involved in the reaction (Figure 8b). The disadvantage of the developed photocatalyst could be attributed to the multi-stage and complex nature of its preparation method.

It is also proposed to use bismuth tungstate Bi₂WO₆ [100,101] and silver tungstate Ag₂WO₄ [102] as components of type-II heterojunctions. Shawky et al. [102] prepared Mn₃O₄-coupled Ag₂WO₄ nanocomposite photocatalysts for enhanced photooxidative desulfurization of thiophene under visible light irradiation. The authors found that the heterostructure efficiency depends on the loading of Mn₃O₄. The use of the nanocomposite photocatalysts composed of 5, 10, and 15 wt% of Mn₃O₄ to Ag₂WO₄ lead to the conversions of thiophen of 75, 92, and 100% after 150 min, respectively. This observation could be explained by the enhancement of visible light absorption (band gap values of 5, 10, and 15 wt%Mn₃O₄/Ag₂WO₄ were 2.70, 2.57, and 2.40 eV, respectively).

Recently, MOFs have gained much attention for selective removal of refractory sulfur and nitrogen compounds from fuels by adsorptive and chemical methods [2,103]. MOF is a class of materials with metal cations or metal-based clusters linked by organic molecules forming a crystalline network that can result in 3D structures with permanent porosity after the removal of the guest species [104,105,106,107]. In recent years, there have also been studies showing the attractiveness of MOFs for PODS. This fact can be explained by the following advantages of MOFs as photocatalysts [71]: (i) MOFs have a porous structure and a large number of active sites; (ii) they have a unique crystal structure whose change significantly affects the photocatalytic properties [108]; (iii) optical absorption of MOFs may be changed by the functionalization of organic linkers [109] and substitution of metal nodes [110]. It should be noted that MOFs are promising materials for selective adsorption of specific guest molecules because of the easy modification of pore surfaces [111]. An outstanding adsorption capacity towards various compounds has great importance in photocatalysis. It is well known that a strong adsorption ability towards organic compounds often leads to a high photocatalytic activity of materials [112,113] and vice versa, since the photogenerated radicals are not able to migrate far from the centers of their formation to the bulk.

The effect of adsorption capacity of MOF-based photocatalysts on the PODS performance was proved by Flihh and Ammar [39], who used ZIF-67/CoWO₄ photocatalysts with various CoWO₄ loading (from 0 to 30 wt%). An increase of the CoWO₄ weight ratio from 1 to 30 wt% led to a decrease of the conversion of DBT from 85.3 to 76.2% in the presence of H₂O₂ as an oxidant. Reducing the photocatalytic activity of the composites with a high CoWO₄ content could be attributed to the pore blockage and a decrease in the surface area, thus forming heterojunctions with weak adsorption properties. The authors also proposed a charge transfer mechanism in the ZIF-67/CoWO₄ hybrid and studied the main active species involved in the oxidation of DBT (Equations (7)–(13)):

ZIF-67 + hν → ZIF-67 (e⁻ + h⁺)

(7)

CoWO₄ + hν → CoWO₄ (e⁻ + h⁺)

(8)

ZIF-67 (e⁻ + h⁺) + CoWO₄ (e⁻ + h⁺) → ZIF-67 (h⁺) + CoWO₄ (e⁻)

(9)

H₂O₂ + e⁻ → ^•OH + OH⁻

(10)

O₂ + e⁻ → ^•O₂⁻

(11)

OH⁻ + h⁺ → ^•OH

(12)

DBT + ^•OH, ^•O₂⁻, h⁺ → DBTO₂

(13)

Recently, CeO₂/MIL-101(Fe) catalysts for the PODS coupled with extraction were designed by introducing different amounts of CeO₂ into MIL-101(Fe) [114]. The obtained results illustrated that 90% of DBT in the oil phase was able to be removed within 2 h, as the volume ratio of model fuel to extractant (acetonitrile) and the H₂O₂/DBT molar ratio were 2:1 and 3:1, respectively. The trapping experiments verified that ^•OH and ^•O₂⁻ were the main reactive species in the photocatalytic reaction. The high photocatalytic activity originated from a synergic effect between active sites of CeO₂ and the surface of MIL-101(Fe) and formed a heterojunction in the photocatalyst, which broadened optical response ranges and facilitated photoinduced charge transfer and separation.

To summarize, various materials, such as g-C₃N₄, metal oxides, perovskites, and MOFs, have been utilized to create type-II heterojunctions (

Table 3. Activity of various type-II heterojunctions in PODS ¹.

No.	Photocatalyst	SCC (C (ppm))	Oxidant	Visible Light Source	Extractant	t (min)	X (%)	Main Products	Ref.
1	Ti3C2/g-C3N4	Th (140)	Air	300 W Xe lamp	None	180	70	CO2	[36]
	Ti3C2	Th (140)	Air	300 W Xe lamp	None	180	31	CO2	[36]
	g-C3N4	Th (140)	Air	300 W Xe lamp	None	180	29	CO2	[36]
2	CeF3/g-C3N4	DBT (100)	H2O2	300 W Xe lamp	None	180	84	DBTO2	[96]
	CeF3	DBT (100)	H2O2	300 W Xe lamp	None	180	25	DBTO2	[96]
	g-C3N4	DBT (100)	H2O2	300 W Xe lamp	None	180	40	DBTO2	[96]
3	Ag2O/ZrO2	Th (600)	O2	300 W Xe lamp	MeCN	150	100	CO2	[97]
	ZrO2	Th (600)	O2	300 W Xe lamp	MeCN	150	38	CO2	[97]
4	Cu2O–CeO2	Th (300)	H2O2	400 W Osram lamp	DMF 2	180	84	CO2	[98]
5	CuO–BiVO4	DBT (400)	O2	300 W halogen lamp	None	180	91	DBTO2	[22]
	BiVO4	DBT (400)	O2	300 W halogen lamp	None	180	49	DBTO2	[22]
6	Cu/Cu2O/BiVO4	Th (64)	H2O2	400 W Osram lamp	DMF	150	92	CO2	[99]
7	Nb2O5/Bi2WO6	DBT (200)	H2O2	Two 5 W LED lamp	MeOH	120	99	DBTO2	[100]
	Nb2O5	DBT (200)	H2O2	Two 5 W LED lamp	MeOH	120	65	DBTO2	[100]
8	Fe3O4@SiO2/Bi2WO6/Bi2S3	Th (500)	Air	400 W halogen lamp	None	90	100	CO2	[101]
9	Mn3O4/Ag2WO4	Th (600)	O2	125 W Hg lamp	None	150	100	CO2	[102]
10	BiOI/BiPO4	Th (500)	Air	400 W Osram lamp	None	120	95	Not specified	[43]
11	ZIF-67/CoWO4	DBT (100)	H2O2	Two 50 W LED lamp	None	120	85	DBTO2	[39]
	ZIF-67	DBT (100)	H2O2	Two 50 W LED lamp	None	120	70	DBTO2	[39]
12	CeO2/MIL-101(Fe)	DBT (500)	H2O2	500 W Xe lamp	MeCN	120	90	DBTO2	[114]

¹ In some cases, the activity of individual components of hybrids is presented to emphasize the effectiveness of heterojunctions; ² N,N–dimethylformamide.

). All investigations have demonstrated that composite materials are more efficient in PODS than the pristine photocatalysts due to the separation of photogenerated charge carriers. A promising direction seems to be the use of MOFs to obtain type-II heterojunctions due to their outstanding properties. On the other hand, MOFs have several disadvantages, which limit their application as catalysts in large-scale processes. Only a small number of MOFs are produced commercially by international companies such as BASF (Germany), MOF Technologies Ltd. (UK), and Strem Chemicals Inc. (USA) [115,116]. In addition, MOFs are characterized by a high price. As an example, the MIL-53(Al) MOF is commercially produced with a price of approximately USD $2440–3455 per 500 g [116].

Overall, the main limitation of the type-II heterojunction is that the oxidation and reduction abilities of the transferred e⁻ and h⁺ decrease because of the electrons migrating to the CB of a semiconductor, with lower reduction potential and holes accumulating on the VB of a semiconductor with lower oxidation potential. Thus, the development of Z-scheme and S-scheme heterojunctions may solve this problem.

3.3.2. Z-Scheme Heterojunctions

There are three concepts of the Z-scheme photocatalytic system:

(1)

the first-generation systems, i.e., the liquid-phase Z-scheme photocatalysts (or redox-mediator) [117];

(2)

the second-generation systems, i.e., the all-solid-state (ASS) Z-scheme photocatalysts, where electron mediators are used to the facilitate charge carrier [118];

(3)

the third-generation systems, i.e., the direct Z-scheme photocatalysts (or mediator-free), where a direct contact between two semiconductors is ensured without any charge carrier transfer mediator [119,120].

The application of first-generation Z-scheme photocatalysts is limited by backward reactions existing due to the use of reversible redox mediators as well as the possibility of their use only in the liquid phase [121,122]. Thus, for practical application in the gas or solid phases, both ASS and direct Z-scheme systems should be developed.

In the ASS Z-scheme photocatalysts, noble metal nanoparticles are the most common electron mediators. The efficiency of these photocatalytic systems in PODS was demonstrated using the Ag/AgI/α-MoO₃ [123] and Ag/AgCl/PbMoO₄ [124] heterojunctions. A series of novel ASS Z-scheme photocatalysts Ag/AgI/α-MoO₃ were investigated in the photocatalytic oxidation of thiophene [123]. It was found that the composite containing 35 wt% of AgI allows oxidation of 98% of thiophene to CO₂ for 2 h. It should be acknowledged that the Ag/AgI/α-MoO₃ photocatalyst exhibited a relatively good stability during four cycles. The authors also suggested two different mechanisms to elucidate the separation of photogenerated electron–hole pairs for binary and ternary composites (Figure 9), revealing a key role of noble metal nanoparticles for the creation of the ASS Z-scheme heterojunctions and the effectiveness of these heterostructures compared to type-II systems.

The use of noble metals as electron mediators in Z-scheme photocatalysts, which are rare and expensive, limits the wide practical application of the ASS systems [121]. In addition, noble metal nanoparticles are characterized by strong light-absorption properties [125,126], which leads to the reduction of the light-absorption ability of the ASS Z-scheme photocatalysts. To solve these problems, low-cost and Earth-abundant materials with excellent conductivity should be developed as electron mediators in the ASS Z-scheme photocatalytic systems [127]. It has been demonstrated that MoS₂ [128], carbon film [129], Cu [130], Co [131], and heteropolyacids [132] can be used as electron mediators for the construction of Z-scheme photocatalysts for various applications. Following this line of research, Fakhri and colleagues [133] investigated tungstophosphoric acid (PW₁₂)/Ce-doped NH₂-UiO-66 Z-scheme photocatalysts in the photocatalytic oxidation of DBT and quinoline using molecular oxygen as an oxidant. The use of Ce as an electron mediator facilitated the separation of charge carriers, while NH₂-UiO-66 MOF remarkably enhanced the surface area with plentiful adsorption sites and shifted the adsorption edge of PW₁₂ into the visible region. This synergetic effect resulted in superior photocatalytic ability and efficiency of PODS of 89% for 90 min. The stability tests showed that the conversion of DBT decreased partially after four consecutive runs and reached about 86%. Moreover, the XRD pattern of the reused PW₁₂/Ce-doped NH₂-UiO-66 Z-scheme photocatalyst was similar to that of the fresh photocatalyst, indicating its good stability.

Inspiring results in PODS have also been revealed using direct Z-scheme photocatalytic systems. It was demonstrated that the ZnM-layered double hydroxide (LDHs)/g-C₃N₄ (M = Al, Cr) composites can be successfully utilized as visible light-responsive photocatalysts for the photocatalytic oxidation/extraction of DBT with air using acetonitrile as an extractant [37]. The ZnAl-LDHs/g-C₃N₄ and ZnCr-LDHs/g-C₃N₄ photocatalysts exhibited an excellent desulfurization performance of 100% within 3 h. The materials showed photostability with little activity loss after five consecutive cycles. The high activity of photocatalysts can be attributed to the formation of the direct Z-scheme heterojunction, which efficiently promotes the transfer of charge carrier and suppresses the recombination of electron−hole pairs. The porous flower-like AgI/Bi₂O₃ direct Z-scheme photocatalyst with the band gap value of 2.81 eV also demonstrated superior photocatalytic desulfurization efficiency of about 93% after 2 h visible light irradiation, even after 5 cycles [134]. It was observed that the electrons on the CB of AgI and the holes on the VB of Bi₂O₃ in this heterostructure efficiently separated, and ^•O₂⁻, H₂O₂, and ^•OH were the main ROS involved in the oxidation of DBT molecules. The superior photocatalytic performance could be ascribed not only to the formation of the direct Z-scheme system, but also to the 3D morphology of the photocatalyst. Hierarchical materials with 3D structure (flower-like, 3D hollow, etc.) have a larger interface, availability of various tunnels, and more active sites for the interaction between photocatalysts and SCCs [52,135].

In recent years, dual Z-scheme heterojunction photocatalysts (ternary composites) have been widely studied due to their high efficiency in the separation of photogenerated charge carriers, strong redox ability, and abundant reaction sites [136,137]. For instance, the attractiveness of these photocatalytic systems has been proven for environmental remediation [138,139], water splitting [140,141], and CO₂ reduction [142]. Yaghoot-Nezhad et al. [143] prepared the CuO–ZnO@g-C₃N₄ dual Z-scheme heterojunction nanocomposites via a multistep ultrasound-assisted hydrothermal procedure and studied them in the ultrasound-assisted photocatalytic oxidation of DBT in the presence of H₂O₂ as an oxidant. It was found that a combined approach of ultrasound and visible light results a higher efficiency in the removal of DBT compared to the single use of ultrasound and visible light irradiation. These results could be explained by the cavitation effects of the ultrasound approach [144], leading to enhanced ^•OH radical generation. Under the optimal reaction conditions, almost complete sulfur removal was achieved using CuO–ZnO@g-C₃N₄ at a photocatalyst dose of 0.2 g·L⁻¹, initial DBT concentration of 250 ppm, and H₂O₂ loading of 250 ppm within 60 min of treatment in the presence of ultrasonic waves (80 W·m⁻²) and light irradiation (150 W). Despite the fact that the proposed photocatalyst exhibited an excellent activity and stability, the implementation of the ultrasound-assisted photocatalytic oxidation technique can lead to a significant increase in energy consumption due to the additional use of ultrasound. Moreover, there has been report that the ternary composite TiO₂(rutile)/Sb-SnO₂/C-TiO₂(anatase) exhibited a comparative activity in PODS (94% within 60 min) without ultrasound irradiation [145].

As has been demonstrated in this subsection, various semiconductors have been proposed to prepare the ASS and direct Z-scheme heterojunctions for PODS (

Table 4. Activity of ASS and direct Z-scheme as well S-scheme heterojunctions in PODS.

No.	Photocatalyst	SCC (C (ppm))	Oxidant	Visible Light Source	Extractant	t (min)	X (%)	Main Products	Ref.
ASS Z-scheme heterojunctions
1	Ag/AgI/α-MoO3	Th (500)	Air	400 W halide lamp	None	120	98	CO2	[123]
	AgI	Th (500)	Air	400 W halide lamp	None	120	75	CO2	[123]
	MoO3	Th (500)	Air	400 W halide lamp	None	120	35	CO2	[123]
2	Ag/AgCl/PbMoO4	DBT (200)	O2	Xe lamp	None	120	97	Not specified	[124]
3	PW12/Ce-doped NH2-UiO-66	DBT (250)	Air	400 W Hg lamp	None	90	89	DBTO2	[133]
	NH2-UiO-66	DBT (250)	Air	400 W Hg lamp	None	90	30	DBTO2	[133]
Direct Z-scheme heterojunctions
4	ZnAl-LDHs/g-C3N4	DBT	Air	500 W Hg lamp	MeCN	180	100	DBTO2	[37]
5	AgI/Bi2O3	DBT (800)	Air	400 W Osram lamp	DMF	120	93	DBTO2	[134]
6	CuO–ZnO@g-C3N4	DBT (250)	H2O2	Xe lamp	None	60	99	DBTO2	[143]
7	TiO2(R)/Sb-SnO2/C-TiO2(A)	DBT (200)	Air	500 W Xe lamp	None	60	94	DBTO2	[145]
S-scheme heterojunctions
8	Ag3PO4/d-C3N4	DBT (200)	H2O2	300 W Xe lamp	MeCN	180	93	DBTO2	[146]
	Ag3PO4	DBT (200)	H2O2	300 W Xe lamp	MeCN	180	76	DBTO2	[146]

). The use of the ASS Z-scheme photocatalytic materials may be industrially feasible when an effective alternative to expensive noble metals is developed. Nowadays, the most promising seems to be the use of the mediator-free direct Z-scheme photocatalysts, whose activity can be further improved by the creation of dual systems. However, developing the dual Z-scheme photocatalysts should consider the cost of the obtained materials.

In addition, it is very important to develop S-scheme heterojunctions for PODS, especially since successful attempts have been made [146]. The experimental results showed that the conversion of DBT in the presence of the Ag₃PO₄/d-C₃N₄ photocatalyst under visible light was 93% after 3 h (Table 4), mainly due to the S-scheme transfer and separation of photogenerated charge carriers.

3.4. Supported Photocatalysts

Many materials used for PODS have a tendency toward severe particle agglomeration, which reduces the photocatalytic activity as well as the number of surface active sites. To overcome these problems, support materials can be used to disperse and immobilize nanoparticles. The performance of supported photocatalysts has been recently reviewed in detail by Hitam et al. [5] and Zhou et al. [9]. On the other hand, since 2020, there have been several reports related to the supported photocatalytic materials, which should be mentioned.

It was shown that the most popular photocatalyst, TiO₂, can be immobilized onto SBA-15 [147] and porous glass [148]. For the photocatalytic oxidation of DBT, Guo and colleagues [147] prepared TiO₂@SBA-15 composites and sensitized these with organic dye (2,9-dichloroquinacridone, DCQ) to extend the spectral response range of the photocatalyst from UV light to visible light. The authors found that the DCQ-TiO₂@SBA-15 photocatalyst has a better performance than the unsensitized TiO₂@SBA-15, and the desulfurization rate can reach up to 96% within 90 min. Although photocatalyst sensitization has been a research hotspot in PODS, there is still a need to reduce the cost of this technology and improve the stability of the process. From the point of view of the practical application of the PODS technology, the most attractive is the use of materials that are sensitive to visible light without sensitization with organic dyes.

For instance, the core-shell MoS₂–g-C₃N₄–BiOBr@MCM-41 photocatalyst adsorbents with different percentages of MoS₂ (1, 3, and 5 wt%) were investigated in a one-step photooxidative-adsorptive desulfurization under simulated solar light [44]. Among the different samples, the sample with 5 wt% of MoS₂ had the highest catalytic activity in 75 min due to strong interaction between components, highest coverage of MCM-41 with active phases, high capability of light absorption, and low recombination of charge carriers (Figure 10a). The authors revealed the direct Z-scheme transfer of the photogenerated charges in this dual heterojunction (Figure 10b).

It could be assumed that a suitable selection of support materials with a high surface area might play crucial roles in enhancing the PODS performance. The support materials can not only disperse and immobilize photocatalyst nanoparticles, preventing their agglomeration, but are also capable of improving the adsorption capacity towards SCCs.

4. Influence of Operating Parameters on PODS

The activity of a semiconductor in PODS could be tuned by reaction conditions such as temperature, nature of SCC used as a model compound, nature of oxidant, molar ration of oxidant/SCC, nature of solvent/extractant, and photocatalyst loading. In this Section, the effect of various operating parameters on PODS performance is briefly discussed.

4.1. Temperature

The reaction temperature is one of the main driving forces in PODS and is necessary to achieve exciting results in the field of heterogeneous photocatalysis. It was observed that the maximum efficiency in the oxidation of DBT using TiO₂ as a photocatalyst was realized at 40 °C [45], in which the scavenging of hydroxyl radicals (self-oxidation) was minimized. Further increasing of temperature led to decreasing the amount of DBT adsorbed on the TiO₂ surface and, therefore, decreasing the photocatalytic activity. Wang et al. [74] found that when the temperature varied from 30 to 70 °C, the desulfurization ratio of Er/W–N–TiO₂ remarkably increased from 79 to 100%. However, the activity was not significantly improved when the reaction temperature was higher than 80 °C, which could be explained by the nonproductive decomposition of H₂O₂ at higher temperatures. A similar effect was detected in the case of the Pr–N–CQDs–TiO₂ photocatalyst [72].

On the contrary, Li and colleagues [36] observed that low temperature is beneficial to the degradation of SCCs over the Ti₃C₂/g-C₃N₄ composites. The photoluminescence (PL) measurements showed that the PL intensity of a hot photocatalyst (40 °C) is stronger than that at ambient temperature, indicating a faster recombination of the photogenerated charge carriers under heating conditions. Moreover, the dissolved oxygen in reaction systems may inevitably release under heating, which reduces the oxidant source to generate ROS. Decreasing of the rate of SCC removal with increasing temperature was also noted for the TiO₂/porous glass photocatalyst [148].

It could be assumed that the implementation of PODS at ambient temperature is more advantageous for large-scale application of the technology, since heating requires energy costs.

4.2. Photocatalyst Loading

The effect of photocatalyst loading on the rate of SCCs removal may be very significant. With increasing photocatalyst dosage, the percentage of sulfur removal could increase due to increasing the number of accessible active sites for the efficient production of ROS [45,149]. It was reported that with increasing of the photocatalyst dosage to a certain loading, the desulfurization ratio increased and then decreased [74]. Increasing the amount of a photocatalyst led to more active centers available for the contact of SCCs with the active cites. However, after the optimum photocatalyst loading, the trend of the desulfurization ratio was opposite because the redundant photocatalyst hindered absorption of light [97,102,123]. In addition, a catalyst may become oxidized in the reaction system by the active species and lose its activity [150].

4.3. Nature of Oxidant and Molar Ratio of Oxidant/SCC

Many studies have concentrated on the use of hydrogen peroxide as an oxidant for PODS, which could be associated with the high activity of ^•OH radicals generated from H₂O₂. In recent years, there has been a tendency to oxidize SCCs with cheaper and safer air (Table 1, Table 2, Table 3 and Table 4). At the same time, only one group carried out a comparative assessment of the effectiveness of various types of oxidants (hydrogen peroxide, molecular oxygen, air) in PODS [36]. It was demonstrated that the removal of sulfur- and nitrogen-containing compounds decreased when molecular O₂ or H₂O₂ was used as oxidants, in comparison to that in the air atmosphere (Figure 11). It is well known that molecular O₂ and H₂O₂ could form ^•O₂⁻ and ^•OH, respectively, by combining with e⁻ (Equations (14) and (15)). However, it was shown that the efficiency of PODS depends more on h⁺ active species than ^•OH. As a result, PODS is inhibited due to the consumption of h⁺ by generated OH⁻, despite the fact that the amount of radical ^•OH is increased (Equations (15) and (16)). On the other hand, the authors conducted the PODS experiments without the addition of polar solvents/extractants into the reaction mixture. Thus, there could be an inefficient interaction between polar H₂O₂ and SCC dissolved in non-polar hydrocarbons.

O₂ + e⁻ → ^•O₂⁻

(14)

H₂O₂ + e⁻ → ^•OH + OH⁻

(15)

OH⁻ + h⁺ → ^•OH

(16)

As a rule, increasing the amount of H₂O₂ in the reaction mixture causes enhanced desulfurization efficiencies [143]. However, there are studies showing an optimum molar ratio of H₂O₂/SCC [41,151]. Zhao et al. [41] found that the that the desulfurization rate over the Ag/ALa₄Ti₄O₁₅ photocatalysts (A = Ca, Sr, and Ba) increased significantly with increasing the volume of H₂O₂ from 10 to 25 mL and then decreased when more hydrogen peroxide was added. They concluded that the excess of hydrogen peroxide in the reaction mixture may hinder the adsorption of the reactants on the surfaces of the perovskite oxides. Another explanation could be associated with the poisoning of a photocatalyst surface due to the high concentration of H₂O₂ [151,152].

The presence of maxima corresponding to the optimal ratio of oxidant/SCC is also typical when air is used as an oxidant [6,23]. For instance, the sulfur removal in the presence of the Ag–BiVO₄ photocatalyst dramatically increased with increasing the air flow from 10 to 30 mL·min⁻¹, while a lower PODS efficiency was achieved at an air flow of 40 mL·min⁻¹ [6]. It was supposed that excess O₂ may act as a hole or ^•OH scavenger, or react with the photocatalyst to form peroxocompounds, which are detrimental to the photocatalytic oxidation process [153]. A similar trend was found by Zhou and colleagues [23] who explained it by the hindering of the adsorption of thiophene on the Ag₂O/g-C₃N₄ photocatalyst surface with increasing air flow. In addition, increasing air flow disrupts the formation of the excited state of sulfide to a certain extent, which in turn hinders the desulphurization reaction.

4.4. Nature of SCC Used a Model Compound

The main SCCs, which are used for the preparation of model fuels for PODS, are thiophene and its derivatives (BT, DBT, and 4,6-DMDBT). The type of SCC used determines not only the products formed during its oxidation but also the rate of its removal from the model fuel.

Many studies have tried to compare the rate of photocatalytic oxidation of different sulfur-containing model compounds [22,37,73,124,154]. Generally, the order of removal efficiency of various model compounds was DBT > 4,6-DMDBT > BT > Th. This observation may be because of the influence of the electron cloud density of the S atom. The electron cloud densities of the S atom in Th, BT, 4,6-DMDBT, and DBT are 5.739, 5.696, 5.760, and 5.758, respectively [155,156]. The S atom electron cloud of Th had the lowest density and was the most difficult to oxidize; therefore, its degradation rate was far lower than those of DBT, 4,6-DMDBT, and BT. Huang et al. [37] investigated the removal efficiency of different sulfur species during PODS and found that the performance of the ZnM-LDHs/g-C₃N₄ (M = Al, Cr) photocatalysts followed the order of BT > DBT > 4,6-DMDBT. This may arise from increasing steric hinderance from BT to 4,6-DMDBT, which reduces the chance of contact between active species and S atoms [54,124].

It should be noted that the use of thiophene as a model SCC in almost all cases leads to its complete mineralization during photocatalytic oxidation. At the same time, DBT is oxidized in most cases to DBTO₂. This behavior of DBT could be associated with a very strong S–O bond with the dissociation energy of 475 kJ·mol⁻¹ [157]. On the other hand, it is preferable to oxidize completely the SCCs to CO₂ due to fact that the implementation of incomplete oxidation will lead to the need for an additional of separation of the oxidation product from the hydrocarbon mixture by extraction methods. Currently, only two research groups have reported complete mineralization of DBT during photocatalytic oxidation [25,38].

4.5. Nature of Solvent/Extractant

It is believed that extraction using pure solvents is a crucial step in the oxidative desulfurization process. Technology involving PODS reaction is mostly biphasic, comprising polar and non-polar phases, as shown in Figure 12 [114]. This procedure involves two steps: sulfur-containing molecules are (i) extracted from the oil phase into the solvent phase and (ii) oxidized in the solvent phase by ROS. Therefore, there should be certain requirements (low cost, non-toxicity, high ability to extract oxidized products, etc.) for solvents used for extraction–photooxidation technology.

Amiri et al. [43] studied the effect of using different solvents on the desulfurization rate of DBT under visible light irradiation. The use of MeCN, DMF, and ethanol allowed removal of 95, 87, and 69% of thiophene from model fuel. However, for industrial implementation, it is necessary to find an optimal replacement for toxic acetonitrile. Recently, ionic liquids have been proposed as extractants in ODS. They are beneficial owing to thermal stability, low volatility, and high solubility for numerous inorganic and organic compounds. On the other hand, the use of ionic liquids has certain limits, for instance, low efficiency, possible side reaction in ODS, and high cost. Thus, the use of ionic liquids in the industry will require further study [3,158].

5. Conclusions

Heterogeneous photocatalysis has been considered as one of the most attractive alternative routes to transform naturally abundant, clean, and sustainable solar energy into chemical energy. One of the main advantages of photocatalysis is that reactions can be conducted under mild reaction conditions under solar irradiation.

As has been clearly shown here, many efforts have been made in the context of the development of photocatalysts for the desulfurization reaction. The interest in this topic is attributed to the possibility of obtaining ultra-low-sulfur fuels under mild reaction conditions due to traditional hydrodesulfurization technology used under harsh reaction conditions (temperature of 350–450 °C and pressure of 30–60 atm).

Nowadays, the most popular photocatalysts include systems based on titanium dioxide, which can only absorb UV light due to its large band gap value of 3.0–3.4 eV. In order to achieve efficient solar energy utilization, the development of new photocatalytic materials, which are sensitive to visible light, and methods for shifting the photosensitivity of TiO₂-based catalysts to the visible region (E_g < 3.0 eV) are important directions in PODS. In this regard, various ranges of photocatalysts have been proposed for the desulfurization reaction, including visible light-responsive metal oxides, perovskites, graphitic carbon nitride, and MOFs. Pure semiconductor photocatalysts are characterized by low activity in PODS due to the rapid recombination of photogenerated electron–hole pairs. To improve their photocatalytic activity under visible light irradiation, various methods have been investigated: (i) band gap engineering, i.e., non-metal doping, (ii) deposition of a metal co-catalyst, and (iii) construction of heterojunctions.

Despite the fact that the incorporation of non-metals is one of the common approaches for the enhancement of the photocatalytic activity of various materials, this method is often accompanied with leaching of the dopant. The doping with metals has been an interesting approach to improve the photocatalytic desulfurization reaction. The deposition of a noble metal co-catalyst is the most followed route. However, the use of noble metals is the main disadvantage of this approach because of their high price.

For the spatial separation of electron–hole pairs in photocatalysts and the preparation of highly active photocatalysts for PODS, the creation of heterojunctions has been proven to be one of the most promising ways. Nowadays, most attention regarding the synthesis of composites has been paid to type-II heterojunctions. Recent studies have demonstrated that type-II heterostructures are more efficient in PODS than the pristine photocatalysts due to the separation of photogenerated charge carriers. However, the main limitation of the type-II heterojunction is that the oxidation and reduction abilities of the transferred electrons and holes decrease because of the electrons migrating to the CB of a semiconductor with lower reduction potential and holes accumulating on the VB of a semiconductor with lower oxidation potential. Therefore, researchers should develop new types of heterojunctions to solve this challenge. Z-scheme and S-scheme heterojunction photocatalysts may become promising alternatives. Key aspects in the development of new Z-scheme and S-scheme systems are: (i) the physicochemical properties of the components for the construction of heterojunctions should be optimized, but many studies have not focused on this aspect; (ii) the photocatalytic activity of Z- and S-scheme photocatalysts could be further improved by the creation of dual systems; however, this approach can only be justified if these composites are many times more active in PODS than heterojunctions composed of two components.