Recent Achievements in Development of TiO2-Based Composite Photocatalytic Materials for Solar Driven Water Purification and Water Splitting

Clean water and the increased use of renewable energy are considered to be two of the main goals in the effort to achieve a sustainable living environment. The fulfillment of these goals may include the use of solar-driven photocatalytic processes that are found to be quite effective in water purification, as well as hydrogen generation. H2 production by water splitting and photocatalytic degradation of organic pollutants in water both rely on the formation of electron/hole (e−/h+) pairs at a semiconducting material upon its excitation by light with sufficient photon energy. Most of the photocatalytic studies involve the use of TiO2 and well-suited model compounds, either as sacrificial agents or pollutants. However, the wider application of this technology requires the harvesting of a broader spectrum of solar irradiation and the suppression of the recombination of photogenerated charge carriers. These limitations can be overcome by the use of different strategies, among which the focus is put on the creation of heterojunctions with another narrow bandgap semiconductor, which can provide high response in the visible light region. In this review paper, we report the most recent advances in the application of TiO2 based heterojunction (semiconductor-semiconductor) composites for photocatalytic water treatment and water splitting. This review article is subdivided into two major parts, namely Photocatalytic water treatment and Photocatalytic water splitting, to give a thorough examination of all achieved progress. The first part provides an overview on photocatalytic degradation mechanism principles, followed by the most recent applications for photocatalytic degradation and mineralization of contaminants of emerging concern (CEC), such as pharmaceuticals and pesticides with a critical insight into removal mechanism, while the second part focuses on fabrication of TiO2-based heterojunctions with carbon-based materials, transition metal oxides, transition metal chalcogenides, and multiple composites that were made of three or more semiconductor materials for photocatalytic water splitting.


Photocatalytic Water Treatment
The general mechanism of semiconductor photocatalysis (Figure 1) is composed of three main steps: 1. e − /h + pairs are generated on the surface of the semiconductor under illumination with the required wavelength or energy; then, 2.) photogenerated charges (i.e. e − /h + ) migrate to the surface of the semiconductor; and lastly, 3.) e − and h + induce redox reactions on the surface that facilitate destruction of organic pollutants [19,20]. As stressed above, TiO2 is still the most studied and widely used material for photocatalytic degradation reactions. However, TiO2 suffers from fast e − /h + pair recombination and large band gap (Eg = 3.1-3.2 eV), which can only be excited under UV light irradiation. The strategies for improving these issues are provided above, while, among them, semiconductor-coupling presents a viable structure-properties engineering solution for the enhancement of TiO2 photocatalytic activity due to the simultaneously reduced e − /h + recombination rate and enhanced visible light absorption [21]. Three main types of heterojunction architectures are reported for TiO2/semiconductor composites [22]. In Type I heterojunction, the conduction band (CB) of TiO2 is higher in energy (more negative potential) when compared to the CB of semiconductor 2 and the valence band (VB) of TiO2 is lower in energy (more positive potential) as compared to the VB of semiconductor 2 [23,24]. This leads to the accumulation of photogenerated h + and e − in semiconductor 2. In Type II heterojunction (where TiO2 can be semiconductor 1 or 2), the CB of semiconductor 2 is higher than the CB position of semiconductor 1 leading to facile transfer of photogenerated e − from CB of semiconductor 2 to CB of semiconductor 1 [25]. Meanwhile, photogenerated h + in VB of semiconductor 1 can travel to the VB of semiconductor 2, which facilitates efficient charge separation. Type III heterojunction (also known as broken gap situations) [26] shares the same charge transfer mechanism, like Type II heterojunction. In this case, the CB and VB of semiconductor 2 are higher than CB and VB of TiO2 [27,28]. These heterojunction types are explained in detail in the context of particular material combinations in the further text and graphically represented through Figures 2,3,5,6,7,8,9,and 10. Three main types of heterojunction architectures are reported for TiO 2 /semiconductor composites [22]. In Type I heterojunction, the conduction band (CB) of TiO 2 is higher in energy (more negative potential) when compared to the CB of semiconductor 2 and the valence band (VB) of TiO 2 is lower in energy (more positive potential) as compared to the VB of semiconductor 2 [23,24]. This leads to the accumulation of photogenerated h + and e − in semiconductor 2. In Type II heterojunction (where TiO 2 can be semiconductor 1 or 2), the CB of semiconductor 2 is higher than the CB position of semiconductor 1 leading to facile transfer of photogenerated e − from CB of semiconductor 2 to CB of semiconductor 1 [25]. Meanwhile, photogenerated h + in VB of semiconductor 1 can travel to the VB of semiconductor 2, which facilitates efficient charge separation. Type III heterojunction (also known as broken gap situations) [26] shares the same charge transfer mechanism, like Type II heterojunction. In this case, the CB and VB of semiconductor 2 are higher than CB and VB of TiO 2 [27,28]. These heterojunction types are explained in detail in the context of particular material combinations in the further text and graphically represented through Figures 2, 3 and 5-10.  3 ), which is a visible light active photocatalyst with band gap of 2.4-2.8 eV, is a promising candidate for photocatalytic applications, due to its oxidative properties, nontoxicity, low cost, and stability in acidic solutions. In addition, WO 3 directly matches the band positions of TiO 2 to form a heterojunction (Type II Heterojunction) [29][30][31][32]. Several authors studied the application of TiO 2 /WO 3 composites for the degradation of various CECs; either pesticides or pharmaceuticals (Table 1). Hence, Macias et al. [24] studied the photocatalytic degradation of herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) while using TiO 2 /WO 3 composites under natural sunlight. They reported the rather high effectiveness of the studied system: 94.6% degradation of 2,4-D and 88.6% mineralization of overall organic content under two and four hours of natural sunlight irradiation, respectively.
Besides, they studied the mechanisms that are responsible for forming reactive species within the system and, based on their findings, proposed that, upon forming e − /h + pairs under solar irradiation, photogenerated e − from CB of TiO 2 are transferred to CB of WO 3 . Consequently, W 6+ was first reduced to W 5+ on WO 3 surface, while the W 5+ ions are then oxidized to W 6+ by adsorbed O 2 producing superoxide anion radical (O 2 •-). The photogenerated h + in VB of WO 3 are transferred to VB of TiO 2 where they reacted with water (or hydroxyl ions, HO − ) forming hydroxyl radicals (•OH) ( Figure 2). The generated reactive oxygen species (ROS) promoted the degradation of 2,4-D and its intermediates, eventually yielding rather high mineralization extents, while their occurrence in the system was confirmed through tests with common scavenging agents (e.g., tert-butanol (TB) for •OH, formic acid (FA) for h + , and p-benzoquinone (BQ) for O 2 •-) [24]. Tungsten oxide (WO3), which is a visible light active photocatalyst with band gap of 2.4-2.8 eV, is a promising candidate for photocatalytic applications, due to its oxidative properties, nontoxicity, low cost, and stability in acidic solutions. In addition, WO3 directly matches the band positions of TiO2 to form a heterojunction (Type II Heterojunction) [29][30][31][32]. Several authors studied the application of TiO2/WO3 composites for the degradation of various CECs; either pesticides or pharmaceuticals (Table 1). Hence, Macias et al. [24] studied the photocatalytic degradation of herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) while using TiO2/WO3 composites under natural sunlight. They reported the rather high effectiveness of the studied system: 94.6% degradation of 2,4-D and 88.6% mineralization of overall organic content under two and four hours of natural sunlight irradiation, respectively.
Besides, they studied the mechanisms that are responsible for forming reactive species within the system and, based on their findings, proposed that, upon forming e − /h + pairs under solar irradiation, photogenerated e − from CB of TiO2 are transferred to CB of WO3. Consequently, W 6+ was first reduced to W 5+ on WO3 surface, while the W 5+ ions are then oxidized to W 6+ by adsorbed O2 producing superoxide anion radical (O2 •-). The photogenerated h + in VB of WO3 are transferred to VB of TiO2 where they reacted with water (or hydroxyl ions, HO − ) forming hydroxyl radicals (•OH) ( Figure 2). The generated reactive oxygen species (ROS) promoted the degradation of 2,4-D and its intermediates, eventually yielding rather high mineralization extents, while their occurrence in the system was confirmed through tests with common scavenging agents (e.g., tert-butanol (TB) for •OH, formic acid (FA) for h + , and p-benzoquinone (BQ) for O2 •-) [24]. The same composite type was used in the degradation of pharmaceuticals. Hence, Mugunthan et al. [30] treated diclofenac (DCF) while using TiO2/WO3 composites under 4 hrs of visible light irradiation and reported a maximum of 92% mineralization of overall organic content. They also elucidated the DCF degradation pathway by LC/MS measurements, which included C-N cleavage in the DCF molecule forming benzene-ring based intermediates at the first stage, and open-ring intermediates at the later stage, which were eventually mineralized. Such findings were quite similar to other studies employing •OH based processes in the degradation of DCF ( [33,34]), thus implying the important role of formed ROS, primarily •OH, in the case of TiO2/WO3 solar driven photocatalysis as well. Arce-Sarria et al. [35] studied the performance of TiO2/WO3 composite for the degradation of another pharmaceutical, Amoxicillin (AMX), in pilot scale reactor, where they achieved 64.4% degradation.
Besides "pure" TiO2/WO3 composite, several authors studied the performance of its enriched analogues (Table 1). Hence, Cordero-García et al. [32] studied DCF degradation by WO3/C-doped TiO2 and reported 100% DCF degradation and 82.4% mineralization of the overall organic content under 250 kJ/m 2 and 400 kJ/m 2 of solar-accumulated energy, respectively. They also stated that the The same composite type was used in the degradation of pharmaceuticals.
Hence, Mugunthan et al. [30] treated diclofenac (DCF) while using TiO 2 /WO 3 composites under 4 hrs of visible light irradiation and reported a maximum of 92% mineralization of overall organic content. They also elucidated the DCF degradation pathway by LC/MS measurements, which included C-N cleavage in the DCF molecule forming benzene-ring based intermediates at the first stage, and open-ring intermediates at the later stage, which were eventually mineralized. Such findings were quite similar to other studies employing •OH based processes in the degradation of DCF ( [33,34]), thus implying the important role of formed ROS, primarily •OH, in the case of TiO 2 /WO 3 solar driven photocatalysis as well. Arce-Sarria et al. [35] studied the performance of TiO 2 /WO 3 composite for the degradation of another pharmaceutical, Amoxicillin (AMX), in pilot scale reactor, where they achieved 64.4% degradation. Besides "pure" TiO 2 /WO 3 composite, several authors studied the performance of its enriched analogues (Table 1). Hence, Cordero-García et al. [32] studied DCF degradation by WO 3 /C-doped TiO 2 and reported 100% DCF degradation and 82.4% mineralization of the overall organic content under 250 kJ/m 2 and 400 kJ/m 2 of solar-accumulated energy, respectively. They also stated that the WO 3 /C-doped TiO 2 composite showed superior photocatalytic activity for the complete degradation and mineralization of DCF when comparing to the pristine TiO 2 , used as benchmark material. Based on the findings on elucidated mechanisms within the studied composite and DCF degradation pathway provided, the authors concluded that the incorporation of elemental carbon to TiO 2 crystal structure promoted the formation of a C2p-hybridized valence band that lowered the band gap of TiO 2 by mixing with O2p orbitals. As a result, upon visible light irradiation, TiO 2 generates e − /h + pairs, where the photogenerated e − are promoted to the Ti 3d states (VB), thus reducing Ti 4+ to Ti 3+ . Ti 3+ can be easily oxidized by WO 3 due to the differences in the reduction potential between TiO 2 (−0.70 V vs NHE) and WO 3 (−0.03 V vs NHE). Subsequently, W 6+ traps photogenerated e − to form its reduced state W 5+ , while the redox reaction occurs further by returning to its original oxidation state in reaction with adsorbed O 2 on the composite catalyst surface (similarly as discussed above in the case of "pure" TiO 2 /WO 3 ), thus leading to improved charge separation and the formation of ROS, which contributed in DCF degradation and mineralization of formed intermediates. The same authors studied the degradation of DCF with another enhanced WO 3 /TiO 2 composite (N-doped TiO 2 ), and again reported high degradation and mineralization rates; 100% according to both indicators under 250 kJ/m 2 and 400 kJ/m 2 of solar-accumulated energy, respectively [31]. They stressed that the same mechanism that was responsible for the enhancement of photocatalyst activity in C-doped WO 3  Iron oxide (α-Fe 2 O 3 ) is a promising candidate for photocatalytic applications, due to its abundance, nontoxicity, low cost, stability in aqueous solutions (pH > 3), and narrow band gap (2.0-2.2 eV), which directly matches the band positions of TiO 2 to form heterojunction (Type I Heterojunction) [23,36].
Several authors report the photocatalytic degradation of CECs using TiO 2 /Fe 2 O 3 composites (Table 2). Hence, Mirmasoomi et al. [37] used TiO 2 /Fe 2 O 3 as a catalyst for photocatalytic degradation of Diazinon (DZ). The authors reported an optimized system with maximum degradation of DZ equal to 95.07% within 45 min. under visible light irradiation. In another study by Moniz et al. [23], photocatalytic degradation of 2,4-D while using TiO 2 /Fe 2 O 3 composites was investigated, reporting 100% 2,4-D degradation and 100% mineralization of overall organic content within 2 h and 3 h, respectively, using irradiation from a 300 W Xenon Lamp. The authors found out that, when compared to the benchmark TiO 2 (P25), the TiO 2 /Fe 2 O 3 composite shows superior photocatalytic activity. Based on photoluminescence and photocurrent studies, the TiO 2 /Fe 2 O 3 composite exhibits enhanced separation of e − /h + pairs due to the formed heterojunction. The proposed mechanism was supported with DFT studies, which firstly involved the transfer of photogenerated e − from TiO 2 CB to Fe 2 O 3 CB. In addition, Fe 2 O 3 binds strongly with (dissolved) oxygen, thus aiding the photoelectron transfer. This in-situ second stage mechanism facilitates the facile migration of h + from the VB of TiO 2 [23]. Macías et al. [24] studied the same system, TiO 2 /Fe 2 O 3 composites for photocatalytic degradation of 2,4-D, but while using natural sunlight. The authors reported 96.8% 2,4-D degradation and 90.0% mineralization of overall organic content under two hours and four hours, respectively. Contrary to the presented mechanism of Moniz et al. [23], Macias et al. [24] (Figure 3). The formation of mentioned ROS and their involvement in degradation of targeted pollutant was confirmed through common scavenging tests using TB, FA, and BQ.  [41] photogenerated e − /h + pairs. Photogenerated e − in CB of Fe2O3 can be transferred to O2 dissolved in water to form O2 •-, while photogenerated h + in VB of Fe2O3 can facilitate generation of •OH eventually contributing to the degradation of present organics [24] (Figure 3). The formation of mentioned ROS and their involvement in degradation of targeted pollutant was confirmed through common scavenging tests using TB, FA, and BQ. The photocatalytic degradation of the pharmaceutical tetracycline (TC) and its derivatives, such as oxytetracycline (OTC), using TiO2/Fe2O3 materials has also been reported. Hence, it was found out that, under visible light irradiation (λ = 400-750 nm), α-Fe2O3 was activated and generated e − /h + pairs, and then photogenerated e − from CB of α-Fe2O3 moved to TiO2 trapping sites for atmospheric O2 to form O2 •-, which was proven to largely contribute to the degradation of OTC. On the other hand, the photogenerated h + from VB of α-Fe2O3 primarily reacted with OH − , resulting in the generation of •OH, which also contributed to the degradation of OTC. When compared to TiO2 reference material, the TiO2/Fe2O3 composite exhibited an enhanced photocatalytic activity under visible light due to efficient e − /h + separation, as stated above [38]. The same authors [38,39] also studied the degradation mechanism of OTC while using LC/MS TOF analysis and, based on the formed intermediates, established the OTC degradation pathway, and concluded that •OH mainly mediated degradation. Besides, "pure" TiO2/Fe2O3, enriched analogue with carbon nanotubes (CNTs) was also studied ( Table 2). Hence, TiO2/Fe2O3/CNTs was used as the catalyst for photocatalytic degradation of TC, under visible light illumination [40]. It was found that the effectiveness of photocatalytic degradation of TC within 90 min. treatment using TiO2/Fe2O3/CNTs was almost twice higher when comparing to that achieved by benchmark TiO2; 89.41% and 47.64%, respectively. The authors attributed the improved photocatalytic efficiency to the presence of the CNT, which acted as a photogenerated e − acceptor, thereby suppressing e − /h + recombination [40]. In another study, the core-shell structured α-Fe2O3 (with TiO2 shell of around 15 nm) exhibited 100% TC removal in 90 min. [41]. The degradation improvement was ascribed to the addition of H2O2 in the system, which generated more ROS than by the common photocatalytic mechanisms described above [41]. Hence, The photocatalytic degradation of the pharmaceutical tetracycline (TC) and its derivatives, such as oxytetracycline (OTC), using TiO 2 /Fe 2 O 3 materials has also been reported. Hence, it was found out that, under visible light irradiation (λ = 400-750 nm), α-Fe 2 O 3 was activated and generated e − /h + pairs, and then photogenerated e − from CB of α-Fe 2 O 3 moved to TiO 2 trapping sites for atmospheric O 2 to form O 2 •-, which was proven to largely contribute to the degradation of OTC. On the other hand, the photogenerated h + from VB of α-Fe 2 O 3 primarily reacted with OH − , resulting in the generation of •OH, which also contributed to the degradation of OTC. When compared to TiO 2 reference material, the TiO 2 /Fe 2 O 3 composite exhibited an enhanced photocatalytic activity under visible light due to efficient e − /h + separation, as stated above [38]. The same authors [38,39] also studied the degradation mechanism of OTC while using LC/MS TOF analysis and, based on the formed intermediates, established the OTC degradation pathway, and concluded that •OH mainly mediated degradation. Besides, "pure" TiO 2 /Fe 2 O 3 , enriched analogue with carbon nanotubes (CNTs) was also studied ( Table 2). Hence, TiO 2 /Fe 2 O 3 /CNTs was used as the catalyst for photocatalytic degradation of TC, under visible light illumination [40]. It was found that the effectiveness of photocatalytic degradation of TC within 90 min. treatment using TiO 2 /Fe 2 O 3 /CNTs was almost twice higher when comparing to that achieved by benchmark TiO 2 ; 89.41% and 47.64%, respectively. The authors attributed the improved photocatalytic efficiency to the presence of the CNT, which acted as a photogenerated e − acceptor, thereby suppressing e − /h + recombination [40]. In another study, the core-shell structured α-Fe 2 O 3 (with TiO 2 shell of around 15 nm) exhibited 100% TC removal in 90 min. [41]. The degradation improvement was ascribed to the addition of H 2 O 2 in the system, which generated more ROS than by the common photocatalytic mechanisms described above [41]. Hence, the contribution of H 2 O 2 in such a system can be described through restraining e − /h + recombination and increasing HO• generation in the system, as in Equation (1) [15]:  [42,43]. Spinel ferrites band positions match TiO 2 , thus possessing compatibility to form a heterojunction (Type II Heterojunction) [44][45][46][47].
The literature provides applications of MFe 2 O 4 /TiO 2 materials as photocatalysts in treatment of CECs, as in the case of previously discussed TiO 2 -based composites, however, it should be noted that authors within such composites used modified TiO 2 (Table 3). Hence, Chen et al. [44] studied photocatalytic degradation of TC and its derivatives using N-doped TiO 2 /CaFe 2 O 4 /diatomite, and reported 91.7% removal of TC within 150 min. under visible light irradiation. The authors studied the composite stability and reusability; the results obtained after five cycles indicates that employed composite is rather stable, enabling 89.2% removal of TC. They also proposed the photocatalytic mechanism occurring within the composite; the excitation of both N-TiO 2 and CaFe 2 O 4 by visible light leads to the formation of e − /h + pairs ( Figure 4) (Figure 4). Such a transfer of charge carriers between the two semiconductors hinders the recombination process and enhances the photocatalytic activity of the composite, thus leading to more efficient generation of ROS (O 2 •and •OH) [44]; the existence of formed ROS was confirmed through scavenging tests while using isopropyl alcohol (IPA), ammonium oxalate (AO), and BQ for •OH, h + and O 2 •-, respectively. Such behavior is confirmed by studying the degradation pathway of TC; it was found that the TC intermediates match those that formed through radical driven reactions undergoing in the first step demethylation and hydroxylation. The second step considered the removal of functional groups (amino, hydroxyl, and methyl) and further ring opening reactions that are mainly mediated by photogenerated h + , yielding small fragments that were eventually mineralized [44]. ammonium oxalate (AO), and BQ for •OH, h + and O2 •-, respectively. Such behavior is confirmed by studying the degradation pathway of TC; it was found that the TC intermediates match those that formed through radical driven reactions undergoing in the first step demethylation and hydroxylation. The second step considered the removal of functional groups (amino, hydroxyl, and methyl) and further ring opening reactions that are mainly mediated by photogenerated h + , yielding small fragments that were eventually mineralized [44]. Such a pathway confirmed the dual role of photogenerated h + , as a promotor •OH generation and as sites for the direct oxidation of adsorbed organics. There are several other studies investigating the application of different MFe2O4/TiO2 materials (N-doped TiO2/SrFe2O4 diatomite [46]; Ce/N-co-doped TiO2/NiFe2O4/ diatomite and ZnFe2O4/TiO2 [47]) for the photocatalytic degradation of CECs, such as TC, OTC, and bisphenol A (BPA). Interestingly, the same mechanisms responsible for charge transfer and consequent generation of ROS were reported, regardless of the different M type within the spinel ferrite part of composite and/or TiO2 (non-doped or doped with metal and/or non-metal ions).

TiO2/Cu2O
Cu2O, a p-type semiconductor (Eg = 2.0-2.2 eV), is also a good candidate for making heterojunctions with TiO2 (Type II Heterojunction). Hence, the photocatalytic degradation of various CECs (TC [48], and tetrabromodiphenyl ethers [49]) using TiO2/Cu2O composite materials was reported under solar light irradiation (Table 4). Based on the findings in the mentioned studies, the photocatalytic mechanism of TiO2/Cu2O under solar light illumination involves the activation of both Cu2O and TiO2 to generate e − /h + pairs ( Figure 5). Photogenerated e − in CB of Cu2O then can migrate to CB of TiO2 and, along with photogenerated e − in CB of TiO2, react with O2 to form O2• _ . Simultaneously, photogenerated h + in VB of Cu2O can be directly involved in the oxidation of adsorbed organics, while photogenerated h + in VB of TiO2 can directly oxidize adsorbed organics or react with H2O (i.e. OH − ) and generate •OH. Besides, these h + can also directly migrate to the VB of Cu2O, thus leading to effective charge separation that improves the overall photocatalytic activity of the composite [48].   [48], and tetrabromodiphenyl ethers [49]) using TiO 2 /Cu 2 O composite materials was reported under solar light irradiation (Table 4). Based on the findings in the mentioned studies, the photocatalytic mechanism of TiO 2 /Cu 2 O under solar light illumination involves the activation of both Cu 2 O and TiO 2 to generate e − /h + pairs ( Figure 5). Photogenerated e − in CB of Cu 2 O then can migrate to CB of TiO 2 and, along with photogenerated e − in CB of TiO 2 , react with O 2 to form O 2 • _ . Simultaneously, photogenerated h + in VB of Cu 2 O can be directly involved in the oxidation of adsorbed organics, while photogenerated h + in VB of TiO 2 can directly oxidize adsorbed organics or react with H 2 O (i.e., OH − ) and generate •OH. Besides, these h + can also directly migrate to the VB of Cu 2 O, thus leading to effective charge separation that improves the overall photocatalytic activity of the composite [48].  Bi2O3, a semiconductor with band gap range in the visible region (2.5-2.8 eV), is also a good candidate for making heterojunctions with TiO2 (Type II Heterojunction). Studies including its application in photocatalytic degradation of CECs (quinalphos [50] and ofloxacin [51]) under solar light irradiation (Table 5) revealed the occurring photocatalytic mechanism. Both of the composite phases can be activated under solar irradiation generating e − /h + pairs ( Figure 6). Accordingly, photogenerated h + in VB of TiO2 are involved in the production of •OH (through reactions with H2O, i.e. OH − ) as of e − /h + pairs. In addition, h + in VB of Bi2O3 can be transferred to VB of TiO2 that contributes to the direct oxidation of adsorbed organics or the generation of •OH [51]. , is also a good candidate for making heterojunctions with TiO 2 (Type II Heterojunction). Studies including its application in photocatalytic degradation of CECs (quinalphos [50] and ofloxacin [51]) under solar light irradiation (Table 5) revealed the occurring photocatalytic mechanism. Both of the composite phases can be activated under solar irradiation generating e − /h + pairs ( Figure 6). Accordingly, photogenerated h + in VB of TiO 2 are involved in the production of •OH (through reactions with H 2 O, i.e., OH − ) as of e − /h + pairs. In addition, h + in VB of Bi 2 O 3 can be transferred to VB of TiO 2 that contributes to the direct oxidation of adsorbed organics or the generation of •OH [51].

Coupling of TiO 2 with Metal Sulfides
Cadmium sulfide (CdS), a metal sulfide semiconductor with a visible light range band gap (E g = 2.1-2.4 eV), has been proven to be compatible with TiO 2 , due to its higher position of CB than that of TiO 2 (Type II Heterojunction) ( Figure 7) [25,52]. However, one should be aware that its application can lead to adverse effects due to its instability, resulting in the leaching of toxic Cd 2+ during treatment [53]. Although its CB and VB positions are thermodynamically favorable for photocatalytic application, CdS as a photocatalytic material faces serious problems. Next to the above-mentioned promotion of toxic effects, issues, like poor stability due to photocorrosion and limited separation efficiency of photogenerated charge carriers, do not speak in favor of CdS application [54,55]. Photocorrosion is not only related to the photogenerated h + in semiconductor itself that oxidizes S 2and release Cd 2+ to the solution, but also with newly formed O 2 , where higher solubility in water leads to more dramatic levels of photocorrosion of CdS [54,56]. However, CdS was widely investigated in photocatalytic purposes, even in recent studies that focused on the degradation of CECs (ofloxacin, ciprofloxacin, tetracycline, and 17α-ethynylestradiol), where it was used in various forms (nano-rods, nano-belts) [25,52,57,58] (Table 6)  Copper sulfide (CuS), which is another metal sulfide semiconductor with narrow band gap of 2.0 eV, has also been reported to be coupled with TiO2 (Type II Heterojunction) [59,60]. Jiang et al. [59] reported a 85.5% degradation of enrofloxacin and 27.7% mineralization of overall organic content using immobilized CuS/TiO2 nanobelts (Table 6). They elucidated the mechanisms occurring in the composite upon excitation by solar irradiation. Hence, such broad wavelengths excited both composite phases (CuS and TiO2) and resulted in e − /h + pairs, while the transfer of charges was analogous, as in the case of the CdS/TiO2 composite. Photogenerated e − in CB (−0.33 eV) of CuS underwent transfer to CB (−0.19 eV) of TiO2 and were consumed in reactions with O2 forming O2 •-. Photogenerated h + in VB of CuS remained there and present potential active sites for direct degradation of organics that were adsorbed at the CuS surface, since they cannot be involved in generation of •OH due to too high energy band positioning. On the other hand, photogenerated h + in VB of TiO2 can directly react with adsorbed organics and OH − generating •OH. Chen et.al [60], incorporated Au nanoparticles to CuS/TiO2 nanobelts structure to enhance the photocatalytic degradation ability of the composite by capturing e − and, consequently, suppressing the recombination of photogenerated charges. As a result, they obtained 96% degradation of OTC and 68% mineralization of the overall organic content within one hour under artificial sunlight illumination. Accordingly, the mechanism of such enriched CuS/TiO2 composite involves, besides the above discussed mechanism, the path considering the transfer of e − to Au, which leads to enhanced charge separation, thus delaying recombination. In such a case scenario, photogenerated h + would have higher probability to react either with adsorbed organics or with HOin order to generate •OH (exclusively those in VB of TiO2), thus contributing to the overall system efficiency. The involvement of formed ROS into reaction mechanisms for OTC degradation was confirmed by scavenging tests using TB, AO, and BQ. Copper sulfide (CuS), which is another metal sulfide semiconductor with narrow band gap of 2.0 eV, has also been reported to be coupled with TiO 2 (Type II Heterojunction) [59,60]. Jiang et al. [59] reported a 85.5% degradation of enrofloxacin and 27.7% mineralization of overall organic content using immobilized CuS/TiO 2 nanobelts (Table 6). They elucidated the mechanisms occurring in the composite upon excitation by solar irradiation. Hence, such broad wavelengths excited both composite phases (CuS and TiO 2 ) and resulted in e − /h + pairs, while the transfer of charges was analogous, as in the case of the CdS/TiO 2 composite. Photogenerated e − in CB (−0.33 eV) of CuS underwent transfer to CB (−0.19 eV) of TiO 2 and were consumed in reactions with O 2 forming O 2 •-. Photogenerated h + in VB of CuS remained there and present potential active sites for direct degradation of organics that were adsorbed at the CuS surface, since they cannot be involved in generation of •OH due to too high energy band positioning. On the other hand, photogenerated h + in VB of TiO 2 can directly react with adsorbed organics and OH − generating •OH. Chen et.al [60], incorporated Au nanoparticles to CuS/TiO 2 nanobelts structure to enhance the photocatalytic degradation ability of the composite by capturing e − and, consequently, suppressing the recombination of photogenerated charges. As a result, they obtained 96% degradation of OTC and 68% mineralization of the overall organic content within one hour under artificial sunlight illumination. Accordingly, the mechanism of such enriched CuS/TiO 2 composite involves, besides the above discussed mechanism, the path considering the transfer of e − to Au, which leads to enhanced charge separation, thus delaying recombination. In such a case scenario, photogenerated h + would have higher probability to react either with adsorbed organics or with HOin order to generate •OH (exclusively those in VB of TiO 2 ), thus contributing to the overall system efficiency. The involvement of formed ROS into reaction mechanisms for OTC degradation was confirmed by scavenging tests using TB, AO, and BQ. Molybdenum disulfide (MoS 2 ), a two-dimensional (2D) layered metal chalcogenide with an indirect band gap of 1.1 eV and 1.9 eV direct band gap in monolayered form, with unique structure, low-cost, high thermal stability, and electrostatic integrity, is also a suitable candidate for forming heterojunction with TiO 2 (Type II Heterojunction) [61][62][63]. Hence, Kumar et al. [64] reported its application in the photocatalytic degradation of paracetamol. Furthermore, Irandost et al. [61] applied the modified MoS 2 /TiO 2 composite (they used N,S-co-doped TiO 2 ) in the photocatalytic degradation of DCF under visible LED lamp irradiation (Table 6). Hence, the synergistic effect of dopants in TiO 2 promoted its visible light activity, yielding the formation of e − /h + pairs in both composite phases. The mechanism of charge formation and consequent transfer was similar, as described above for CuS/TiO 2 , which was excited by solar irradiation. Hence, photogenerated e − in CB of N,S-co doped TiO 2 and CB of MoS 2 were able to undergo reactions with O 2 forming O 2 •-, while h + in VB of TiO 2 promoted •OH formation in reactions with HOand provide the direct oxidation of adsorbed organics, while, again, h + in MoS 2 were able to do only the latter. The importance of •OH and h + in DCF degradation was confirmed by trapping agents used in scavenging tests: TB and potassium iodide (KI), respectively. Tin sulfide (SnS 2 ), which is a metal sulfide semiconductor with band gap of 2.2 eV [65], has also been reported to be coupled with TiO 2 (Type II Heterojunction) [66,67]. Hence, Kovačić et al. [66] reported improved the degradation of 17β-estradiol (E2), for 51%, using SnS 2 /TiO 2 when comparing to the benchmark material (P25) TiO 2 under solar irradiation. A similar improvement was obtained by comparing performances of the same materials in the case of DCF degradation [67] ( Table 6). The reason for such improvement relies on the potential of photogenerated e − in CB of SnS 2 to migrate to CB of TiO 2 , while h + remained at the VB of SnS 2 . In such case, the efficient separation of charges is achieved, thus facilitating the improved redox reactions, enabling effective degradation of adsorbed organics directly on the surface by h + , in spite of the limited ability of such a composite to generate •OH. Accordingly, the adsorption has been shown as an important step in the effectiveness of the SnS 2 /TiO 2 composite. Kovačić et al. [67] utilized DFT calculations to study the surface interaction of polar compounds (DCF) and non-polar compounds (memantine) at SnS 2 /TiO 2 composite and found that DCF was more efficiently degraded due to much higher adsorption ability in comparison to memantine, which is one of its structure feature limitations (amine functionality).

Coupling of TiO 2 with Silver-Based Semiconductors
Silver Phosphate (Ag 3 PO 4 ), a promising semiconductor with narrow band gap (E g ≥ 2.4 eV), showed good photocatalytic performance in the degradation of organic pollutants under visible light irradiation [68,69]. Namely, Ag 3 PO 4 exhibits a quantum efficiency of up to 90% [68] and it can absorb wavelengths that are shorter than~530 nm [69]. Despite the qualities of Ag 3 PO 4 as a potential photocatalyst, it still suffers from limitations, such as photocorrosion, small but not negligible solubility in water (K sp = 1.6 × 10 −16 ), and particle agglomeration upon synthesis [70]. To overcome these limitations, constructing a heterojunction between Ag 3 PO 4 and a compatible semiconductor has attracted attention due to the increase in charge separation and production of more ROS [71]. The positions of VB and CB in TiO 2 directly match the Ag 3 PO 4 band positions, thus providing the compatibility to form a heterojunction.
Hence, Wang et al. [72] investigated the performance of TiO 2 nanotubes/Ag 3 PO 4 quantum dots for the degradation of TC under visible light illumination, and reported a high removal rate within a short treatment period; 90% TC removal within 8 min (Table 7).   [28] The conventional heterojunction transfer mechanism (Figure 8a) explains that the photogenerated h + in the composite would be promoted from the VB of Ag 3 (Figure 9). They also investigated the role of these species in the degradation of BPA and found, based on monitoring BPA degradation pathway by LC/MS analysis, that intermediates are formed through two pathways: 1) hydroxylation, through reactions with O 2 •yielding BPA-o-catechol; and, 2) direct oxidation by h + forming isopropenylphenol and phenol, which was further oxidized by h + yielding hydroquinone and its dehydrated form benzoquinone.
The positions of VB and CB in TiO2 directly match the Ag3PO4 band positions, thus providing the compatibility to form a heterojunction. Hence, Wang et al. [72] investigated the performance of TiO2 nanotubes/Ag3PO4 quantum dots for the degradation of TC under visible light illumination, and reported a high removal rate within a short treatment period; 90% TC removal within 8 min ( Table 7).
The conventional heterojunction transfer mechanism ( Figure 8a) explains that the photogenerated h + in the composite would be promoted from the VB of Ag3PO4 to VB of Ti 3+ -doped TiO2 nanotubes, where can react with H2O or HO − forming •OH. Simultaneously, photogenerated e − from the Ti 3+ -doped TiO2 nanotubes CB can react with O2 forming O2 •-or may transfer to the CB of Ag3PO4. However, O2 •-are not formed in Ag3PO4, due to the fact that the position of its CB is lower than the standard reduction potential of O2 •-/O2. Wang et al. [72] concluded that TC was primarily degraded by O2 •− and photogenerated h + based on the results of the conducted electron trapping experiments. Accordingly, they have extended the study by proposing a Z-scheme heterojunction transfer mechanism (Figure 8b). Under this mechanism, Ag(0) acts a recombination center, "collecting" photogenerated e − from CB of Ag3PO4, where they undergo recombination with the photogenerated h + from VB of Ti 3+ -doped TiO2 nanotubes. In such case, photogenerated h + on VB of Ag3PO4 might participate in the direct oxidation reactions with adsorbed organics, while the photogenerated e − in the CB of Ti 3+ -doped TiO2 nanotubes can be involved in forming desired ROS, O2 •-, thus contributing to the enhanced performance of composite photocatalyst. Du et al. [73] applied analogue TiO2/Ag3PO4 composite employing TiO2 nanotube arrays for the degradation of ciprofloxacin (CIP) under solar irradiation and reported that 85.3% removal of CIP within 60 min. was facilitated through the above-mentioned mechanisms. Furthermore, Liu et al. [74] reported 95% degradation of BPA in 16 min. using TiO2−X/Ag3PO4 under visible light irradiation (Table 7). They reported that both composite phases, TiO2−X and Ag3PO4, were excited and generated e − /h + pairs. Hence, photogenerated h + in VB of TiO2−X are promoted to VB of Ag3PO4 and contributed to the direct oxidation of adsorbed organics, similarly as reported in the study by Wang et al. [72]. Photogenerated e − from the CB of Ag3PO4 are transferred to oxygen vacancies (Vo) of TiO2 and contributed in reactions with adsorbed O2 generating O2 •- (Figure 9). They also investigated the role of these species in the degradation of BPA and found, based on monitoring BPA degradation pathway by LC/MS analysis, that intermediates are formed through two pathways: 1) hydroxylation, through reactions with O2 •-yielding BPA-o-catechol; and, 2) direct oxidation by h + forming isopropenylphenol and phenol, which was further oxidized by h + yielding hydroquinone and its dehydrated form benzoquinone.  Silver oxide (Ag2O), a visible light active photocatalyst with band gap of 1.2 eV, is another silver-based compound with semiconducting properties. Based on the band positions (VB and CB), it represents a promising matching candidate to form heterojunctions with TiO2 (Type III Heterojunction). Hence, photocatalytic degradation of levofloxacin (LEV) using Ag2O/TiO2 quantum Silver oxide (Ag 2 O), a visible light active photocatalyst with band gap of 1.2 eV, is another silver-based compound with semiconducting properties. Based on the band positions (VB and CB), it represents a promising matching candidate to form heterojunctions with TiO 2 (Type III Heterojunction). Hence, photocatalytic degradation of levofloxacin (LEV) using Ag 2 O/TiO 2 quantum dots is reported with the maximum of 81% LEV degradation within 90 min. of visible light irradiation [27]. Based on the proposed mechanism under visible light illumination (Figure 10  Silver oxide (Ag2O), a visible light active photocatalyst with band gap of 1.2 eV, is another silver-based compound with semiconducting properties. Based on the band positions (VB and CB), it represents a promising matching candidate to form heterojunctions with TiO2 (Type III Heterojunction). Hence, photocatalytic degradation of levofloxacin (LEV) using Ag2O/TiO2 quantum dots is reported with the maximum of 81% LEV degradation within 90 min. of visible light irradiation [27]. Based on the proposed mechanism under visible light illumination (Figure 10), upon excitation of Ag2O, e − /h + pairs are formed, whereas TiO2 is not activated due to its wide band gap. Photogenerated e − in the CB of Ag2O were transferred to CB of TiO2 and involved in reactions with adsorbed O2 forming O2 •-that participated in LEV degradation. In addition, photogenerated h + in VB of Ag2O yielded the formation of •OH, through reactions with OH − , and participated in LEV degradation as well. The authors employed LC-MS analysis to elucidate LEV degradation pathway and, as such, establish the role of formed ROS. Hence, parent compound LEV underwent decarboxylation of the acetyl group; hydroxylation resulting in the formation of quinolone moieties; demethylation and the subsequent addition of hydrogen atom generating modifications at piperazine ring; while successive •OH attack resulted in multi-hydroxylated intermediates. Such findings confirmed the dominant role of •OH in LEV degradation. In another study, Gou et al. [28] investigated the application of Ag2O/TiO2/zeolite composite for solar-driven degradation of norfloxacin (NOR) ( Table 7). Besides high effectiveness (98.7% NOR degradation and 83.1% mineralization of organic content within 60 min. treatment), they elucidated the NOR degradation pathway, involving in the initial stage decarboxylation, defluorination or In another study, Gou et al. [28] investigated the application of Ag 2 O/TiO 2 /zeolite composite for solar-driven degradation of norfloxacin (NOR) ( Table 7). Besides high effectiveness (98.7% NOR degradation and 83.1% mineralization of organic content within 60 min. treatment), they elucidated the NOR degradation pathway, involving in the initial stage decarboxylation, defluorination or hydroxylation of parent compound (NOR), which confirmed the involvement of both formed ROS (O 2 •and •OH).

TiO 2 /Graphene Composites
Graphene is a zero bandgap semiconductor with a sheet-like structure (i.e., it is considered as a 2D monolayer material) consisting of sp 2 hybridized carbon atoms with excellent thermal conductivity, optical transmittance, high mechanical strength, large surface area (2600 m 2 /g), and appreciable charge carrier transport [75]. Under light illumination, it can achieve a reverse saturation state with high density (˜10 13 cm 2 ) of hot electrons above the Fermi level, which can be used as a powerful agent in redox reactions [76]. It was also found that the incorporation of graphene-based materials (i.e., graphene oxide and its reduced form; GO and rGO, respectively) with TiO 2 might suppress e − /h + pairs recombination. As such, TiO 2 /graphene-based composites were employed in the photocatalytic degradation of CECs (Table 8).

TiO 2 /Semiconductor/Graphene Composites
As described in above sections, the coupling of TiO 2 with other semiconductors promotes efficient charge transfer, eventually yielding improved photocatalytic activity. However, in most cases, the recombination is still an existing issue that needs to be suppressed. Such a double effect can be obtained by combining composite concept involving two semiconductors (even "pure" TiO 2 , which cannot be active under visible light) with graphene-based materials. For instance, Hao et al. [77] reported 93.2% degradation of BPA in seven hours of sunlight irradiation while using the TiO 2 /WO 3 /GO composite. The mechanism occurring in such combined composite involved the excitation of both TiO 2 and WO 3 under solar light irradiation (TiO 2 utilized UV-A fraction), yielding the generation e − /h + in both semiconductors. Hence, photogenerated e − in CB of TiO 2 can directly react with absorbed O 2 , producing O 2 •− , or it can be transferred to CB of WO 3 , and then further migrate to GO enhancing charge separation. Since the amount of adsorbed O 2 is quite limited, the tendency of e − to recombine with h + is rather favored;~90% of pairs recombine rapidly after excitation [14]. Hence, the charge separation represents an important factor in the evaluation of photocatalyst performance. Accordingly, in the case of effective separation and recombination suppression, as in the case with GO, photogenerated h + in VB of activated composite components, e.g., of TiO 2 and WO 3 in the case of TiO 2 /WO 3 /GO, can be involved in a larger amount, either directly or indirectly (through formation of •OH) in the degradation of present organics. It should be noted that, in composites with two semiconductors, GO could also act as redox site, attracting photogenerated e − and h + , thus promoting improved surface migration of charges [77]. Table 8 summarizes several works regarding TiO 2 /semiconductor/GO composites employed for the degradation of CECs with analogous mechanism, as mentioned above.

TiO 2 /g-C 3 N 4
Graphitic carbon nitride (g-C 3 N 4 ), a two-dimensional, metal-free polymeric π-conjugated semiconductor material, which has attracted a lot of attention [83][84][85][86][87][88][89][90][91] since the pioneering work of Wang et al. [92] in 2009, due to its high stability, visible light response with the bandgap of 2.7 eV and non-toxicity [93], thus representing a viable candidate to be applied in photocatalytic water treatment [80], has certainly been one of the most investigated photocatalysts inside carbon-based nanomaterials. It can be easily synthesized through the direct pyrolysis of nitrogen-rich precursors, such as melamine, cyanamide, dicyandiamide, and urea, but its practical application and principle drawback is low specific surface area and high rate of electron-hole recombination [83,94,95]. Therefore, g-C 3 N 4 modification to address shortcomings are needed, e.g., as an excellent candidate to form heterojunction with TiO 2 (Type II Heterojunction), due to their matched band positions (VB and CB). Hence, several studies employing g-C 3 N 4 /TiO 2 were focused on photocatalytic degradation of CECs (Table 9). For instance, Yang et al. [96] reported 88.1% degradation of CIP within 180 min. under visible light irradiation. The authors ascribed the improved photocatalytic activity to multiple effects: (i) an increase in the surface area of the composite; (ii) good dispersity of TiO 2 in g-C 3 N 4 enabling the intimately coupling of composite phases; and, (iii) extension of light absorption of the composite due to low band gap of g-C 3 N 4 . Trapping experiments that were conducted revealed that photogenerated h + were the major reactive site involved in CIP degradation.
In another study, Li et al. [97] reported the 100% degradation of Acyclovir in 90 min. using g-C 3 N 4 /TiO 2 under visible light irradiation. However, after seven hours of continuous irradiation, any TOC removal was not noticed, implying the formation of rather recalcitrant intermediates with high resistance to degradation by ROS that formed within the studied system. Trapping experiments for formed reactive species elucidated that g-C 3 N 4 /TiO 2 under visible light irradiation only produced h + and O 2 •-, and not the most reactive •OH, explaining limited oxidation capability and none TOC removal in the case of acyclovir degradation. This significant contribution proves that the use of g-C 3 N 4 /TiO 2 under visible light irradiation must undergo careful laboratory tests regarding the susceptibility of targeted organics and their intermediates to degradation by h + and O 2 •prior to considering real scale application [97].
Several studies also showed that the tailoring of composite morphology promotes improved photocatalytic efficiency. For instance, Yu et al. [98] prepared a mesoporous g-C 3 N 4 /TiO 2 that was applied to polysulfone ultrafiltration membranes for sulfamethoxazole (SMX) removal. It was found that mpg-C 3 N 4 /TiO 2 exhibit 69% SMX degradation within 30 hours of sunlight irradiation. On the other hand, TiO 2 nanosheets with exposed facets (001) (core)-g-C 3 N 4 (shell) composite exhibit a higher degradation rate of 2.2 mg/min., which is 36% faster when compared to TiO 2 and g-C 3 N 4 physically-mixed composite. The improved effect is ascribed to the close interaction of TiO 2 and g-C 3 N 4 core-shell structure, whereas, in physically mixed composite the formed heterojunction is random and non-uniform [99].
The use of support materials, such as clays [100] and polymers [101], has been also utilized for improved adsorption capacity and the stability of g-C 3 N 4 /TiO 2 composites. For instance, Chen et al. [101] used g-C 3 N 4 -shielding polyester fiber (PET)/TiO 2 for photocatalytic degradation of sulfaquinoxaline and thiamethoxam. Interestingly, the composite removal efficiency for sulfaquinoxaline reached 97%, after 10 consequent cycles. Furthermore, the introduction of kaolinite with g-C 3 N 4 /TiO 2 improved the surface area and adsorption capacity of the composite, leading to 92% degradation of CIP in 240 min. of visible light irradiation [100].
An additional approach considers doping of metals and non-metals in TiO 2 , enhancing its light absorption capacity from UV absorption to visible light absorption. Thus, incorporating doped TiO 2 with g-C 3 N 4 structures has also attracted great attention for the degradation of CECs. For instance, S-Ag/TiO 2 @g-C 3 N 4 [102] was employed for the degradation of Triclosan (TS) and yielded 92.3% degradation of TS within 60 min. under visible light irradiation. Song et al. [103] fabricated a nanofibrous Co-TiO 2 coated with g-C 3 N 4 , which was applied to TC removal; the authors reported a consistent stability of composite photocatalyst during five consecutive cycles.
Besides doping, sensitization with dyes [104] and carbon dots [105] was also found to enhance the light absorption capacity of g-C 3 N 4 /TiO 2 composite. For example, D35 organic dye was applied next to g-C 3 N 4 /TiO 2 and it was found that the light absorption range was enhanced up to 675 nm [104]. On the other hand, Su et al. [105] studied the application of C dots decorated/g-C 3 N 4 /TiO 2 for the degradation of enrofloxacin under visible light and assigned the observed enhancement to the upconversion photoluminescence properties of C dots, which convert near-infrared light wavelength into visible light wavelength [106]. As effective solutions for improving g-C 3 N 4 /TiO 2 performance, the incorporation of graphene quantum dots [107] and another semiconductor (i.e., MoS 2 [108], WO 3 [109]) is also reported; such systems resulted in enhanced separation of charges and the suppression of their recombination, thus leading to improved photocatalytic activity in the degradation of CECs.

Photocatalytic Water Splitting
Photocatalytic water splitting implies a non-spontaneous process, where the light photons are used to break the water molecules assisted by a photocatalyst, which generates photoexcited charge carriers, i.e., e − /h + pairs, delivering them to the solid-liquid interface, where the redox half-reactions of water oxidation and reduction are catalyzed [110,111], analogously, as described above for photocatalytic water treatment. The difference in water splitting is that photogenerated charges (i.e., e − and h + ) need to react with H + as the electron acceptor adsorbed on the photocatalyst surface or within the surrounding electrical double layer of the charged particles in order to generate H 2 [112], instead of O 2 generating O 2 •− , as in photocatalytic water treatment (Figure 1). The donors are the same; H 2 O, however, desired the product of such reaction is O 2 . Figure 11 shows the principal mechanism of photocatalytic water splitting with the use of TiO 2 semiconductor nanoparticle. The VB and CB of semiconductor or their composites have to have favorable positions in order to enable occurrence of such reactions.
Materials 2020, 13, x FOR PEER REVIEW 19 of 39 e − and h + ) need to react with H + as the electron acceptor adsorbed on the photocatalyst surface or within the surrounding electrical double layer of the charged particles in order to generate H2 [112], instead of O2 generating O2 •− , as in photocatalytic water treatment (Figure 1). The donors are the same; H2O, however, desired the product of such reaction is O2. Figure 11 shows the principal mechanism of photocatalytic water splitting with the use of TiO2 semiconductor nanoparticle. The VB and CB of semiconductor or their composites have to have favorable positions in order to enable occurrence of such reactions. Figure 11. The principle photocatalytic water splitting mechanism over illuminated TiO2 nanoparticle.
Generally speaking, there are two competitive processes that occur inside the photocatalyst and affect H2 evolution. Similarly as in the case of water purification, first is the charge recombination process. Such a process reduces the excited charges for >90%, as mentioned above [14]; according to some authors, even less than 1% of photoexcited charge carriers are able to participate in the photo-redox reactions forming H2 [111]. Such a negative tendency can be improved by controlling the recombination rate [75], as also described in detail in the case of the composite materials used for water purification. The second process is the separation of photogenerated charge carriers that favor H2 evolution, also mentioned above in the case of water purification, but here with more important role [111].
The positions of CB and VB define the redox potential of photogenerated charge carriers. A CB minimum (CBmin) that is smaller than 0 V vs. standard hydrogen electrode (SHE) is required for H2 generation, while the maximum of VB (VBmax) has to be higher than O2/H2O reduction potential, by definition, in order to enable O2 evolution [112]. As mentioned above, H2 generation through this process is non-spontaneous, needing the standard Gibbs free energy change of +237 kJ/mol or 1.23 eV, and to accomplish water splitting under visible light irradiation, the bandgap of the photocatalyst should be more than 1.23 eV and less than 3.0 eV [111]. The electronic structures of diverse semiconductors fulfill the necessary conditions for the water splitting reaction, as can be seen from Figure 12. Generally speaking, there are two competitive processes that occur inside the photocatalyst and affect H 2 evolution. Similarly as in the case of water purification, first is the charge recombination process. Such a process reduces the excited charges for >90%, as mentioned above [14]; according to some authors, even less than 1% of photoexcited charge carriers are able to participate in the photo-redox reactions forming H 2 [111]. Such a negative tendency can be improved by controlling the recombination rate [75], as also described in detail in the case of the composite materials used for water purification. The second process is the separation of photogenerated charge carriers that favor H 2 evolution, also mentioned above in the case of water purification, but here with more important role [111].
The positions of CB and VB define the redox potential of photogenerated charge carriers. A CB minimum (CB min ) that is smaller than 0 V vs. standard hydrogen electrode (SHE) is required for H 2 generation, while the maximum of VB (VB max ) has to be higher than O 2 /H 2 O reduction potential, by definition, in order to enable O 2 evolution [112]. As mentioned above, H 2 generation through this process is non-spontaneous, needing the standard Gibbs free energy change of +237 kJ/mol or 1.23 eV, and to accomplish water splitting under visible light irradiation, the bandgap of the photocatalyst should be more than 1.23 eV and less than 3.0 eV [111]. The electronic structures of diverse semiconductors fulfill the necessary conditions for the water splitting reaction, as can be seen from Figure 12. Within the scope of this review are recent achievements in TiO2-heterojunction systems for photocatalytic H2 generation. It is important to explain the separation mechanisms of charge carriers that occur in such hybrid materials: (i) Schottky junctions-photogenerated e − migration from semiconductor to metal surface due to a higher work function of metal than those of semiconductor, thus forming a Schottky junction ( Figure 13a); (ii) Type II heterojunction (represented in details in the case of water purification) ( Figure 13b); and, (iii) p-n Heterojunction-supply of an additional electric field to accelerate the charge carrier transfer ( Figure 13c); and, (iv) Direct Z-scheme heterojunction-e − in the CB of second semiconductor recombined with the photogenerated h + in the VB of the first semiconductor, leaving the photogenerated e − in first semiconductor and the photogenerated h + in second semiconductor for photocatalysis (Figure 13d) [112]. Within the scope of this review are recent achievements in TiO 2 -heterojunction systems for photocatalytic H 2 generation. It is important to explain the separation mechanisms of charge carriers that occur in such hybrid materials: (i) Schottky junctions-photogenerated e − migration from semiconductor to metal surface due to a higher work function of metal than those of semiconductor, thus forming a Schottky junction ( Figure 13a); (ii) Type II heterojunction (represented in details in the case of water purification) ( Figure 13b); and, (iii) p-n Heterojunction-supply of an additional electric field to accelerate the charge carrier transfer ( Figure 13c); and, (iv) Direct Z-scheme heterojunction-e − in the CB of second semiconductor recombined with the photogenerated h + in the VB of the first semiconductor, leaving the photogenerated e − in first semiconductor and the photogenerated h + in second semiconductor for photocatalysis (Figure 13d) [112].
The process efficiency is determined through the Quantum yield (QY) and Apparent Quantum Yield (AQY), as described with Equations (2) and (3) [93]. The overall quantum yield is predicted to be higher than the apparent one since the number of absorbed photons is usually less than that of incident photons [111]. H 2 generation can also be realized in the presence of sacrificial agents, which, in this case, serve as electron donors that accept photogenerated h + of the VB, thus enhancing the separation of photogenerated charge carriers, which results in higher quantum efficiency [113]. Alcohols are generally used as a h + scavenger, and the more α-H atoms the alcohol has, the higher H 2 production rate is achieved due to more efficient consumption of h + in the photoreaction. The number of α-H atoms in the alcohols can serve as the reference when selecting an appropriate scavenger for photocatalytic reaction [112]. The process efficiency is determined through the Quantum yield (QY) and Apparent Quantum Yield (AQY), as described with Equations (2) and (3) [93]. The overall quantum yield is predicted to be higher than the apparent one since the number of absorbed photons is usually less than that of incident photons [111].
H2 generation can also be realized in the presence of sacrificial agents, which, in this case, serve as electron donors that accept photogenerated h + of the VB, thus enhancing the separation of photogenerated charge carriers, which results in higher quantum efficiency [113]. Alcohols are generally used as a h + scavenger, and the more α-H atoms the alcohol has, the higher H2 production rate is achieved due to more efficient consumption of h + in the photoreaction. The number of α-H atoms in the alcohols can serve as the reference when selecting an appropriate scavenger for photocatalytic reaction [112].
After briefly providing the basic principles to fully understand the H2 evolution through photocatalytic water splitting, following sections are more focused on the recent achievements in fabrication and the evaluation of different TiO2-based heterojunctions with different families of materials, including carbon-based, transition metal oxides and chalcogenides, and multiple-based composites consisting of three or more semiconductor materials for H2 generation.

Carbon-Based/TiO2
Among a variety of materials that are selected for the preparation of TiO2-based nanocomposites to increase their photocatalytic efficiency, nanostructured carbon materials, such as carbon nanotubes and graphene family nanomaterials (e.g., GO, rGO, g-C3N4), are of particular interest [114]. The advantages, such as chemical stability, structural diversity with prominent light-absorptive, and electron transport properties, make them promising materials for use in photocatalytic H2 generation by the water splitting processes [115]. After briefly providing the basic principles to fully understand the H 2 evolution through photocatalytic water splitting, following sections are more focused on the recent achievements in fabrication and the evaluation of different TiO 2 -based heterojunctions with different families of materials, including carbon-based, transition metal oxides and chalcogenides, and multiple-based composites consisting of three or more semiconductor materials for H 2 generation.

Carbon-Based/TiO 2
Among a variety of materials that are selected for the preparation of TiO 2 -based nanocomposites to increase their photocatalytic efficiency, nanostructured carbon materials, such as carbon nanotubes and graphene family nanomaterials (e.g., GO, rGO, g-C 3 N 4 ), are of particular interest [114]. The advantages, such as chemical stability, structural diversity with prominent light-absorptive, and electron transport properties, make them promising materials for use in photocatalytic H 2 generation by the water splitting processes [115].
3.1.1. TiO 2 /g-C 3 N 4 The advantages and limitations of g-C 3 N 4 are already mentioned above in the case of water treatment. The limitations referring to low light utilization efficiency and insufficient surface area can be easily broken by the preparation of 2D nanomaterials, especially g-C 3 N 4 nanosheets (CNNS) [84]. The self-assembly method of construction 2D/2D TiO 2 /CNNS heterojunction composites achieved a hydrogen evolution rate (HER) of 350 µmol/h/g under visible light, in comparison with the produced H 2 with the use of pure TiO 2 nanosheets (20 µmol/h/g) and g-C 3 N 4 nanosheets (130 µmol/h/g) [85].
Liu et al. recorded another use of CNNS [84], who synthesized partially reduced TiO 2−x through NaBH 4 treatments with the formation of an additional mid-gap band state (Ti 3+ and oxygen vacancies-O vs ) to extend absorption edge. The implementation of novel design tactic in the form of a protective carbon layer that was coated onto TiO 2−x /CNNS hetero-junction photocatalyst enhanced the photocatalytic efficiency. The H 2 evolution was tested under visible and simulated solar light with the use of triethanolamine (TEOA) as a sacrificial agent and Pt as a co-catalyst. In the case of visible light irradiation, the highest HER was 417.24 µmol/h/g, while, under AM 1.5 irradiation, the obtained amount was 1830.93 µmol/h/g. The enhanced photocatalytic activity that was ascribed to the formation of Ti 3+ defects was also noticed with the use of g-C 3 N 4 /Ti 3+ -doped TiO 2 Z-scheme system that was synthesized via the polycondensation of urea with TiO 2 , followed by hydrogenation treatment [86]. UV-Vis diffuse reflectance spectroscopy, X-ray photoelectron spectroscopy (XPS), and electron paramagnetic resonance (EPR) have shown that hydrogenation treatment conferred Ti 3+ defect states that were below the CB min of TiO 2 and improved the visible light absorption of the composite with the obtained HER of 1938 µmol/h/g under solar light.
Although special efforts are being made to synthesize noble-metal free nanocomposites, there is still widespread use of Pt as a co-catalyst in H 2 evolution reactions. Except for the already mentioned TiO 2−x /CNNS photocatalyst [84], TiO 2 /g-C 3 N 4 composites with the use of photodeposited Pt as co-catalyst reached HER of 4128 µmol/h/g [87] and 1041 µmol/h/g [83] under solar and visible light irradiation, respectively. Pan et al. [88] also exhibited a high HER of 13800 µmol/h/m 2 by the use of Pt as a co-catalyst with g-C 3 N 4 /TiO 2 nanofilm. Enhanced activity is also attributed to the use of a magnetic-driven rotating frame, which was developed to enhance the mass transfer process during the photocatalytic reaction.
The charge transfer efficiency between TiO 2 /g-C 3 N 4 composite can be enhanced by the doping of different heteroatoms, like C and K atoms. Hence, Zou et al. [89] synthesized C-doped TiO 2 @g-C 3 N 4 core-shell hollow nanospheres with enhanced visible-light photocatalytic activity for H 2 evolution of 35.6 µmol/h/g, which was 22.7 and 10.5 times higher than that of C-TiO 2 and g-C 3 N 4 . The structure of TiO 2 hollow spheres resulted in the reflection of light within the interior cavity, thus increasing the utilization of the light energy. Ma et al. [90] prepared a series of K intercalated g-C 3 N 4 modified TiO 2 nanobelts with enhanced light absorption, transfer efficiency, and H 2 evolution efficiency of 50 µmol/h, which is 6.4 times greater than that of pristine g-C 3 N 4 . The use of carbon atoms in the form of carbon quantum dots (CQDs) as electron reservoirs improves the efficiency of separating the photogenerated charge carriers. CQDs present an important class of carbon materials since their discovery in 2004 by Xu et al. [116], with varying sizes in the range of 1-10 nm. They are good materials for photocatalytic applications due to features, like superiority in chemical stability and low toxicity [117]. Pan et al. synthesized he 2D carbon quantum dots modified porous g-C 3 N 4 /TiO 2 nano-heterojunction [91] and reached 6.497 µmol/h/g of produced H 2 with the full spectrum absorption.
Yi et al. [118] synthesized a composite in which TiO 2 nanobelts were supported by N-doped graphene (NG) coordinated with a single Co atom to replace noble metals with a cost-effective photocatalyst. Under simulated solar irradiation (Figure 14), e − /h + pairs are formed. The transfer of photogenerated e − from the CB of TiO 2 to Co-NG was energetically favorable since the Fermi energy level of graphene (−0.08V vs. NHE) is lower than the CB of TiO 2 (−0.39 V vs. NHE). NG, with a large specific surface area, acted as "freeway" for e − transportation, delivering e − from TiO 2 to Co single-atom, where they were trapped catalyzing H + reduction to form H 2 due to lower the overpotential needed for Co-NG when comparing to that of NG. Co-NG/TiO 2 showed HER of 677.44 µmol/h/g under the illumination of AM 1.5 G simulated sunlight. graphene (NG) coordinated with a single Co atom to replace noble metals with a cost-effective photocatalyst. Under simulated solar irradiation (Figure 14), e − /h + pairs are formed. The transfer of photogenerated e − from the CB of TiO2 to Co-NG was energetically favorable since the Fermi energy level of graphene (−0.08V vs. NHE) is lower than the CB of TiO2 (−0.39 V vs. NHE). NG, with a large specific surface area, acted as "freeway" for e − transportation, delivering e − from TiO2 to Co single-atom, where they were trapped catalyzing H + reduction to form H2 due to lower the overpotential needed for Co-NG when comparing to that of NG. Co-NG/TiO2 showed HER of 677.44 Figure 14. Schematic diagram of proposed photocatalytic mechanism in the CO-NG/TiO2 system. GO/TiO2 nanocomposites have recently been used for H2 production via photocatalytic water splitting under visible light through the formation of Ti-O-C bonding by unpaired π electron of GO with TiO2 surface [114,119]. GO acts as an e − acceptor, promoting the separation of the photogenerated e − /h + pairs in TiO2. These nanocomposites can be synthesized by photo assisted reduction via mixing or sonication and by sol-gel [114]. Hernández-Majalca et al. [114] enhanced the synthesis for the GO-TiO2 nanocomposite using photoassisted anchoring and modifying GO oxidation method through the use of microwaves. The obtained nonporous product had a specific surface area of 45 m 2 /g and absorption onset of 477 nm, which made it active under visible light. Finally, the photocatalytic activity of the nanocomposite was enhanced towards the production of H2, reaching 6500 mol/g in 8 h, which was much higher amount when comparing to that obtained by TiO2-P25 (460 mol/g) at the same irradiation time.
The reduced form of GO, rGO, is a two-dimensional carbon material with the role of an electron mediator that is much superior in chemical stability and morphological diversity than GO [120]. Iwase et al. published the very first report using rGO as a e − mediator in 2011 [121]. Since then, a number of published works were recorded for the use of rGO-based composites in photocatalytic H2 production [75,122,123]. Recent achievements in the synthesis of TiO2/rGO composites for the purpose of H2 generation include work from Reedy et al. [75] and Samal et al. [122]. They obtained rather high HERs while using TiO2/rGO composites under solar and visible light: 24880 µmol/h/g Figure 14. Schematic diagram of proposed photocatalytic mechanism in the CO-NG/TiO 2 system. GO/TiO 2 nanocomposites have recently been used for H 2 production via photocatalytic water splitting under visible light through the formation of Ti-O-C bonding by unpaired π electron of GO with TiO 2 surface [114,119]. GO acts as an e − acceptor, promoting the separation of the photogenerated e − /h + pairs in TiO 2 . These nanocomposites can be synthesized by photo assisted reduction via mixing or sonication and by sol-gel [114]. Hernández-Majalca et al. [114] enhanced the synthesis for the GO-TiO 2 nanocomposite using photoassisted anchoring and modifying GO oxidation method through the use of microwaves. The obtained nonporous product had a specific surface area of 45 m 2 /g and absorption onset of 477 nm, which made it active under visible light. Finally, the photocatalytic activity of the nanocomposite was enhanced towards the production of H 2 , reaching 6500 mol/g in 8 h, which was much higher amount when comparing to that obtained by TiO 2 -P25 (460 mol/g) at the same irradiation time.
The reduced form of GO, rGO, is a two-dimensional carbon material with the role of an electron mediator that is much superior in chemical stability and morphological diversity than GO [120]. Iwase et al. published the very first report using rGO as a e − mediator in 2011 [121]. Since then, a number of published works were recorded for the use of rGO-based composites in photocatalytic H 2 production [75,122,123]. Recent achievements in the synthesis of TiO 2 /rGO composites for the purpose of H 2 generation include work from Reedy et al. [75] and Samal et al. [122]. They obtained rather high HERs while using TiO 2 /rGO composites under solar and visible light: 24880 µmol/h/g and 2700 µmol/h/g of produced H 2 , respectively. Ida et al. undertook further investigation on TiO 2 /rGO composites [123], managing to enhance the photocatalytic activity of the obtained composite by the simultaneous doping of nitrogen on TiO 2 and rGO. The following values for the HER are obtained: TiO 2 (1585 µmol/h/g) < N-TiO 2 (6179 µmol/h/g) < TiO 2 /RGO (12244 µmol/h/g) < N-TiO 2 /N-RGO (15,028 µmol/h/g).

TiO 2 /CNT
Recently, TiO 2 /carbon nanotubes (CNT) have been of great interest due to their high-quality active sites, large specific surface area, and retention of charge recombination, where CNTs can act as a p-type semiconductor, having a role as a powerful electron sink [119]. The coupling of CNT with TiO 2 forms an advanced nanocomposite with enhanced quantum efficiency that forms heterojunction acting as an impurity by forming Ti-O-C or Ti-C defects that enable visible light absorption and, consequently, the creation of e − /h + pairs and hindering e − /h + recombination [124]. CNTs, such as single-wall carbon nanotubes (SWCNTs) and multiwall carbon nanotubes (MWCNTs), have attracted much interest due to their unique chemical, electrical, and optical properties [125]. Olowoyo et al. [124] prepared a series of TiO 2 nanoparticles that were modified with MWCNTs by a combined sonothermal-hydrothermal method. The synthesized photocatalysts were examined for water splitting under batch conditions at different pH ranges. The highest rate of H 2 yield, amounting 69.41 µmol/h/g, was obtained using 2 wt.% CNT-TiO 2 under visible light at pH 2. Hence, the acidic medium improved the photocatalytic feasibility of the system due to a higher concentration of H + ions, serving as the reactants, thus increasing the reaction rate.
Bellamkonda et al. [126] used a different approach in synthesizing CNT-G-TiO 2 composites and prepared nanocomposites via the solution-based method, in which nanocrystalline anatase TiO 2 was grown onto graphene nanosheets and carbon nanotubes. Spectroscopic and photocatalytic studies revealed that graphene acts as an electron reservoir, while the role of CNTs is to prevent the restacking of graphene nanosheets and provide additional electron transport channels, thereby suppressing the recombination rate of e − /h + pairs in the obtained composite. The combination of all these factors resulted in increasing the HER from 19000 µmol/h/g (obtained by anatase TiO 2 ) to 22000 µmol/h/g (obtained by G-TiO 2 ), and finally to 29000 µmol/h/g (obtained by CNT-G-TiO 2 ), which is 8-fold higher than obtained by the commercial TiO 2 (Degussa P25).
The photocatalytic performance of TiO 2 under visible light can be promoted by coupling both MWCNTs and SWCNTs, as presented by Umer et al. [125]. Such an effect occurs due to their dual natural behavior, such as reducing rapid recombination of e − /h + pairs and providing support in harvesting visible light. The maximum H 2 evolution rate of 5486 µmol/h/g was achieved over MWCNT/TiO 2 /SWCNT, which is 1.24-and 1.42-fold higher than using single CTN-TiO 2 composites (SWCNT/TiO 2 and MWCNT/TiO 2 , respectively).

Transition-Metal Oxides/TiO 2
Excellent chemical stability has opened the possibility of the application of transition metal oxides (TMO) in the field of clean energy production. The above displayed Figure 12 contains main TMOs, like p-type (CuO, V 2 O 5 ) and n-type (TiO 2 , WO 3 , MoO 3 , ZnO, Fe 2 O 3 ) semiconductors that are used in photocatalytic H 2 production with pertaining VB and CB energy levels. Visible-light driven TMOs with narrow band gaps are highly desired. The most used materials within this group, such as CuO, Fe 2 O 3 , and WO 3 , have the bandgap energies that allow for them to be active in the visible light region, but the low energy levels of CB position disable them from consuming photoinduced electrons in reactions yielding H 2 . By changing the morphologies of desired components and co-doping with different elements, their CB and VB edges can be shifted toward a H 2 reduction and O 2 oxidation potential [127].
Some TMOs, specifically WO 3 , have been loaded with a different co-catalyst, like Rh, to effectively produce H 2 from water, to control the desired morphology in the form of nanorods, nanotubes, and nanowires. Camposeco et al. [128] focused on the use of Rh-WO 3 photocatalyst that was supported on TiO 2 nanotubes (Rh-WO 3 /NT) for H 2 production via the water splitting process. WO 3 alone cannot take part in H 2 production since the CB energy level of WO 3 is lower than H 2 reduction potential. However, by loading with Rh nanoparticles, the enhancement in H 2 production was noticed. An analysis of energy band levels for the VB and CB that were determined by UV-Vis results and XPS spectra showed that the presence of WO 3 and Rh in the titanate nanotubes simultaneously shift the VB max and CB min , thus reducing the bandgap of titanate nanotubes. 0.5 wt.% Rh-3 wt.% WO 3 /NT nanocomposite under visible light irradiation yielded HER of 87 µmol/h, while 3 wt.% WO 3 /NT showed much lower effectiveness (only 13 µmol/h).
In another work, Ren et al. [129] constructed cooperative Schottky and p-n (SPN) heterojunction by forming a NiO/Ni/TiO 2 heterostructure that showed a narrower band gap, higher photocurrent density, and ability to absorb light in the visible region. High HER is obtained for the observed composite, amounting 4653 µmol/h/g, which is approx. 2.3 times higher than obtained with NiO/TiO 2 with a p-n junction (2059 µmol/h/g), under visible light irradiation and by the use of Pt as a co-catalyst.
A representative example of TMOs is also hematite (Fe 2 O 3 ), already referred above to as promising visible light active photocatalyst for water treatment. The coupling of Fe 2 O 3 with TiO 2 efficiently inhibits the recombination of photogenerated charge carriers and enhances the absorption of solar light [130]. By the use of thermal decomposition of FeCl 3 and TiCl 4 as precursors, Bhagya et al. [113] synthesized the Fe 2 O 3 -TiO 2 composite and investigated its photocatalytic activity for H 2 production under the influence of different proton sources. Besides focusing on the use of TMOs for improving photocatalytic efficiency, this work also highlights the influence of different sacrificial agents as electron donors that consume photogenerated h + , yielding H 2 production. Under simulated solar irradiation, a very high H 2 rate of 880 µmol/h, with an apparent quantum efficiency of 19.39%, is achieved while using Fe 2 O 3 -TiO 2 and diethylamine hydrogen chloride (DAH), which is much higher than that obtained in the case without DAH (323 µmol/h). Madhumitha et al. [130] also explored the influence of different sacrificial reagents on H 2 production under a visible light source. With the optimization of parameters, i.e., catalyst dosage, flow rate, incident light irradiation, and type of sacrificial agent, they achieved the maximum of HER, amounting 2700 µmol/h. The increase in photoactivity was attributed to the effective charge transfer from TiO 2 to Fe 2 O 3 and the use of EDTA, which suppressed the recombination of photogenerated charge carriers.
Easy preparation, environmental friendliness, and good re-utilization enable the wide use of ZnO/TiO 2 -based composites. Xie et al. [131] achieved high rates of H 2 evolution while using ZnO/TiO 2 composites with Pt as a co-catalyst. Under visible light irradiation, 2150 µmol/h/g of H 2 is achieved. Additionally, high carrier mobility can be achieved by the use of ZnO in the form of quantum dots (QDs), which presents ideal heavy-metal free "green" modification of TiO 2 composites. Chen et al. obtained the fabrication of ZnO QDs decorated TiO 2 nanowires via a facile calcination method [132]. They used the obtained composite under solar irradiation next to Pt as a co-catalyst and achieved HER of 313.5 µmol/h.
The use of TMO QDs is also recorded in the work of Liu et al. [133], who decorated two-dimensional TiO 2 nanobelts with zero-dimensional Co 3 O 4 quantum dots. When compared with bulk materials, 0D Co 3 O 4 QDs have attracted considerable attention due to their small size (<10 nm), providing a large specific surface area with more active sites and shorter charges transport paths. Due to the decoration of Co 3 O 4 QDs, the bandgap of the obtained composite was also narrowed, and in application next to Pt as co-catalyst, they obtained a rather high HER of 1735.1 µmol/h/g. In comparison, Zhang et al. [134], with the use of bulk p-Co 3 O 4 /n-TiO 2 , achieved a smaller HER of 8.16 µmol/h/g.
Among the TMOs that appear as promising candidates for coupling with TiO 2 , CuO is one of such, especially due to its narrow bandgap (1.4-1.6 eV) and the promotion of effective charge separation [5,135], as already reported as promising water treatment photocatalyst. Hasan et al. [5] conducted solar H 2 production using a TiO 2 /CuO nanofiber composite that was synthesized by electrospinning technique. Fabricated nanofibers were annealed in different atmospheres to determine the crystalline phase and photocatalytic performance. For the nanofibers that were crystallized in the anatase phase, EPR and XRD analysis referred to the substitution of some Ti 4+ ions by Cu 2+ ions, leading to the formation of some defects below the CB of TiO 2 , which led to a narrow band gap, yielding enhanced HER in the amount of 2715 µmol/h/g. For comparison, without annealing in a different atmosphere, Wang et al. [135] only achieved 47 µmol/h/g of produced H 2 while using the TiO 2 /CuO composite that was irradiated by solar light. Other oxidation states of Cu inside the oxides are investigated for coupling with TiO 2 , such as Cu 2 O [136]. With all of the benefits, such as environmental compatibility, high visible light activity and earth abundance, wider applications of Cu 2 O in water splitting are still limiting, since the redox potential of monovalent copper lies within its band gap, thus photogenerated charge carriers thermodynamically prefer the transformation of Cu 2 O into CuO and Cu, rather than to be used in redox reactions with water constituents forming H 2 [136]. Wei et al. [136] stabilized the Cu 2 O by modulating the defects in faceted Cu 2 O/TiO 2 heterostructures to suppress this disproportionation process. Hence, Cu 2 O was arranged onto 101-faceted TiO 2 and it was found that oxygen vacancies in {101}-faceted TiO 2 can create a unique channel for Z-scheme charge transfer in Cu 2 O/TiO 2 heterostructures. Composite that was obtained by such an approach showed the maximum HER of 32600 µmol/h/g, with a quantum efficiency of 53.5% at an irradiation wavelength of 350 nm.

Transition Metal Chalcogenides-TiO 2
As already mentioned, various techniques have been applied to modify the TiO 2 photocatalysts with a purpose of wider application in the field of solar-driven H 2 production through water splitting. Sensitization with narrow bandgap semiconductors was found to be an effective method for enhancing all of the deficiencies that occur during the sole use of TiO 2 . Regarding an appropriate band gaps, another group of semiconductors with great application potential in photocatalytic H 2 generation are transition metal chalcogenides. Recently, CdS is one of the most studied materials [54,55,[137][138][139][140]; however, the toxic effects due to potential leaching of Cd 2+ have to be strongly considered, followed by others (MoS 2 , ZnS, ZnSe, CdSe) indicated in Figure 12, as mentioned in part related to water treatment [127].

TiO 2 /CdS
CdS is the most important chalcogenides semiconductor as a hydrogen production catalyst due to its narrow bandgap (2.4 eV), which enables its visible light response [56]. Its drawbacks described above in Section 2.2. can be alleviated; susceptibility to photocorrosion can be suppressed by the use of sacrificial agents (sodium sulfite/sulfide) that effectively consume photogenerated h + , while the limited separation efficiency of photogenerated charge carriers can be solved either by using CdS in the form of QDs due to a shorter transportation path or by incorporating CdS onto support materials, such as TiO 2 [55,56].
Rao et al. [137] synthesized CdS/TiO 2 core/shell nanorods with tunable shell thickness to minimize charge carriers recombination and limit photocorrosion. The investigation of photocatalytic activity performed under UV-vis light irradiation confirmed that optimized concentration of sacrificial agents (0.3 M Na 2 S and Na 2 SO 4 aqueous solution), shell thickness of 6.3 nm, and solution pH of 8.0 enhance the H 2 production rate of 5791 mL/h/g. Du et al. [138], who fabricated pyramid-like CdS nanoparticles that were grown on porous TiO 2 , obtained the same type of composite, but with the different morphology. Under UV-vis irradiation and without noble-metal co-catalysts, 5 mol% CdS-TiO 2 achieved an H 2 production rate of 1048.7 µmol/h/g, which is almost six times and 1.5 times higher than that of pure TiO 2 and CdS, respectively. Table 10 provides further examples of TiO 2 /CdS-based composites with their respective photocatalytic activities for H 2 production.

TiO 2 /CuS
CuS has emerged as an alternative co-catalyst for H 2 production, which is abundant, cheap, and nonhazardous. With its VB and CB at positions of −1.56 eV vs. NHE, and −0.09 eV, CuS shows low reflectance in the visible and relatively high reflectance in the near-infrared region, which makes it a good candidate for solar energy absorption [144]. Chandra et al. [144] investigated the photocatalytic activity for H 2 generation of synthesized CuS/TiO 2 (CT) heterostructured nanocomposite under UV-vis and only visible light. Under irradiation, the highest HER of 12362 µmol/h/g was achieved while using 0.4 mol.% CuS/TiO 2 , while only 155 µmol/h/g of H 2 was produced under visible light and comparable conditions.
In comparison, Dang et al. [145] produced a series of Cu x S (x = 1 or 2) co-modified TiO 2 nanocomposites while using a one-step precipitation approach. The EDS, XPS, and XRF results confirmed the existence of three different phases: CuS, Cu 2 S, and TiO 2 . The results have shown that CuS and Cu 2 S dual co-catalysts under simulated solar light exhibited high H 2 production of 5620 µmol/h/g, which is about 58 times higher than that of the unloaded TiO 2 . The enhancement in H 2 production can be contributed to the co-deposition of CuS and Cu 2 S onto the TiO 2 nanoparticle surface that efficiently extended visible light absorption and facilitated the separation of charge carriers. The Cu x S/TiO 2 composite showed high stability; after three consecutive cycles the photocatalytic efficiency for H 2 production decreased for only 11.7%.
3.3.3. TiO 2 /MoS 2 MoS 2 represents two-dimensional transition metal dichalcogenide (TMD) that can be prepared into ultrathin-layered structures and, thus, participates in the H 2 evolution reaction as an effective non-noble metal alternative [117,146]. Without the use of any sacrificial agent and co-catalyst, Huang et al. [146] observed photocatalytic H 2 generation while using MoS 2 quantum dots@TiO 2 nanotube arrays nanocomposite in pure water under visible light. The photocatalytic activity was influenced by the amount of MoS 2 QDs coated on TiO 2 NTAs. Hence, the maximum of 53.9 µmol/cm 2 /h of H 2 was produced, which was ascribed to the decreased bandgap and the surface plasmonic properties of the obtained composite promoting charge carrier separation and the absorption capacity to visible light. Du et al. [63] grew in situ two transition metal chalcogenides-MoS 2 and CdS-on porous TiO 2 by using the sol-gel method, followed by the calcination and hydrothermal method. Under visible light irradiation and without the use of noble metals as the co-catalyst, 3% MoS 2 -CdS-TiO 2 produced 4146 µmol/h/g of H 2 . In this ternary composite (Figure 15), the porous structure of TiO 2 accepts generated e − from CdS and provides surface area for H 2 production, while MoS 2 , as a conductive medium, enabled the transfer of e − between CB of CdS and TiO 2 , simultaneously inhibiting the photocorrosion of CdS as h + collector.

Multiple TiO2-Based Composites
This section is dedicated to multiple ternary and quaternary TiO2-based composites. Recently, significant emphasis was laid on the formation of Z-scheme structured photocatalysts. In such materials, redox reactions occur in each semiconductor, allowing for the combination of a semiconductor with strong reduction power with another semiconductor with strong oxidation power [93].
All of the solid Z-scheme heterojunctions are usually composed of two photocatalysts and an

Multiple TiO 2 -Based Composites
This section is dedicated to multiple ternary and quaternary TiO 2 -based composites. Recently, significant emphasis was laid on the formation of Z-scheme structured photocatalysts. In such materials, redox reactions occur in each semiconductor, allowing for the combination of a semiconductor with strong reduction power with another semiconductor with strong oxidation power [93].
All of the solid Z-scheme heterojunctions are usually composed of two photocatalysts and an electron mediator, which enable efficient H 2 production through the synergistic action between two isolated photosystems and electron mediator cleverly arranged in a nano-platform [147]. Reversible redox couples (e.g., IO 3 − /I − , Fe 3+ /Fe 2+ ) are usually applied as electron mediators in the Z-scheme system. Solid electron mediators are more suitable for the application. Noble metals (Au, Ag) and carbon-based materials (MWCNT, rGO, CQDs) are commonly used as solid electron mediators for photocatalytic H 2 generation [148]. Ng et al. [147], who synthesized Zn 0.5 Cd 0.5 S-MWCNT-TiO 2 ternary nanocomposite, where MWCNT acted as an electron mediator, fabricated a solid Z-scheme system. The obtained material efficiently suppressed charge recombination and promoted water reduction, achieving HER of 21.9 µmol/h. Furthermore, Liu et al. [149] have used carbon quantum dots (QDs) as an electron mediator between TiO 2 and Zn 0.5 Cd 0.5 S film and achieved 38740 µmol/h/m 2 of produced H 2 under solar light. Lv et al. investigated the use of rGO as an electron mediator [120], synthesizing sandwich-like TiO 2 /rGO/LaFeO 3 ternary heterostructure, which, under solar light, obtained HER of 893 µmol/h/g, which is almost 3.2, 14.4, and 11.4 times superior to the direct Z-scheme components TiO 2 /LaFeO 3 composite, pure TiO 2 , and LaFeO 3 , respectively. The photocatalytic mechanism of H 2 production is the same for all three above mentioned solid electron mediators: MWCNT, rGO, and CQDs ( Figure 16). Upon excitation by light, e − /h + are formed in Z-scheme semiconducting components, depending on the ability of each component to absorb emitted light. Hence, photogenerated e − from semiconductor I can be easily recombined with h + from semiconductor II through the electron mediator, leaving more oxidative holes and reductive electrons to participate in the redox reactions in the corresponding active sites [148].
In Z-type heterojunction, noble metals can also perform the role of electron mediators. Zou et al. performed the construction of g-C 3 N 4 /Au/C-TiO 2 hollow spheres with Au nanoparticles (NPs) as the electron mediator [150]. The as-prepared composite showed high HER of 129 µmol/h/g under visible light, which can be attributed to the efficient charge separation in the constructed Z-scheme system, the broadened visible-light response range, owing to the surface plasmon resonance (SPR) effects on Au nanoparticles, and the hollow structure of C-TiO 2 that gives photocatalyst unique properties of low density and high light-harvesting efficiency. In another work presented by Yang et al. [151], Au NPs were applied as a solid electron mediator in the ternary urchin-like ZnIn 2 S 4 -Au-TiO 2 nanocomposite, which, at an optimal ratio of 24 wt.% Au NPs and 60 wt.% ZnIn 2 S 4 , achieved HER of 186.3 µmol/h/g under solar light irradiation. Table 11 lists other multiple composites, except ones with the Z-scheme heterojunction that are already mentioned, as well as reaction conditions in the photocatalytic system and the obtained hydrogen evolution rates.  N-TiO 2 /g-C 3 N 4 @Ni x P 50 mg photocatalyst was suspended in a 100 mL solution containing 10 vol.% triethanolamine (TEOA) under 300 W Xe lamp irradiation. 5438 µmol/h/g [158] TiO 2 /CdS/CNT 0.1 g of photocatalyst was dispersed in solution containing 70 mL of distilled water or seawater and 30 mL of sacrificial agent. The photocatalyst were irradiated using three visible-light sunlamps of each 100 W and UV lamp of 8 W.
3502 µmol/h from pure water and 1373 µmol/h from seawater [159] WS 2 / C-TiO 2 / g-C 3 N 4 50 mg of photocatalyst was added in to 80 mL TEOA aqueous solution with the use of Pt as a cocatalyst.
17726 for DI water and 29978 µmol/g for seawater [148]  semiconductor I can be easily recombined with h + from semiconductor II through the electron mediator, leaving more oxidative holes and reductive electrons to participate in the redox reactions in the corresponding active sites [148]. In Z-type heterojunction, noble metals can also perform the role of electron mediators. Zou et al. performed the construction of g-C3N4/Au/C-TiO2 hollow spheres with Au nanoparticles (NPs) as the electron mediator [150]. The as-prepared composite showed high HER of 129 µmol/h/g under visible light, which can be attributed to the efficient charge separation in the constructed Z-scheme system, the broadened visible-light response range, owing to the surface plasmon resonance (SPR) effects on Au nanoparticles, and the hollow structure of C-TiO2 that gives photocatalyst unique properties of low density and high light-harvesting efficiency. In another work presented by Yang et al. [151], Au NPs were applied as a solid electron mediator in the ternary urchin-like ZnIn2S4-Au-TiO2 nanocomposite, which, at an optimal ratio of 24 wt.% Au NPs and 60 wt.% ZnIn2S4, achieved HER of 186.3 µmol/h/g under solar light irradiation. Table 11 lists other multiple composites, except ones with the Z-scheme heterojunction that are already mentioned, as well as reaction conditions in the photocatalytic system and the obtained hydrogen evolution rates.

Conclusions
TiO 2 -based photocatalytic technology represents a promising up-coming technology for both renewable energy generation and water purification applications. It is necessary to work on developing TiO 2 -based semiconductor materials, which are active under broader spectrum of solar light, overcoming the indicated disadvantages of the solely TiO 2 utilization, since TiO 2 as a photocatalytic material cannot be used alone due to its limitations such as activity only in UV light region and rapid recombination of photogenerated charge carriers.
TiO 2 -semiconductor coupling offers promising results in water purification, particularly for the degradation and mineralization of CECs. However, it is necessary to evaluate the toxicities of degradation intermediates of CEC to check the real efficiency of such composites. Furthermore, the immobilization of TiO 2 -Semiconductor composites to photocatalytic reaction membranes must be envisaged for further upscale opportunities.
Great progress has also been recorded in the development of TiO 2 -based heterojunction for the application in solar-driven photocatalytic hydrogen generation. By morphological control of the obtained composites, higher H 2 generation, as well as better light harvesting can be achieved. It is noticed that still a great deal of research is being conducted by the use of different noble-metal co-catalysts and sacrificial agents, which, although increasing the efficiency of the process, reduces its environmental friendliness and increases the performance cost. Accordingly, to allow for the practical deployment of such units, as well as commercialization, it is necessary to produce cost-effective systems that will consider economic impact assessment, including operation cost and energy consumption, and that will not require the use of costly co-catalysts and other substances that promote system performance.