Over the past several decades, a great deal of work has been performed in the field of semiconductor photocatalysis due to its potential applications in both environmental remediation and solar energy conversion [1-5]. TiO₂ has proven to be one of the most satisfactory photocatalytic materials, as it is strongly oxidizable, chemically and biologically inert, highly biocompatible, and affordable [6-8]. When illuminated, TiO₂ first produces electrons (e^‒) and holes (h⁺) in its conduction band and valence band, respectively. Subsequently, these carriers may recombine within TiO₂ or be transferred to the surface of the TiO₂ to initiate direct redox reactions or indirect redox reactions via the formation of reactive oxygen species (ROS) such as superoxide radicals (·O₂^‒) and hydroxyl radicals (·OH). However, e^‒/h⁺ recombination is very fast both in the bulk and on the surface of TiO₂. As a result, the quantum efficiency of TiO₂ photocatalysis is unsatisfactory for practical applications [9-12].

To improve the photocatalytic behavior of TiO₂, the following strategies have been developed: (1) Textural and crystal design, such as the fabrication of mesoporous TiO₂ to improve the adsorption of organic pollutants [13], hollow-structure TiO₂ to extend the photo-absorption range [12, 14, 15], and high-energy TiO₂ nanocrystals [16, 17] to impede the recombination of the photo-generated carriers by stimulating the separation of the photo-induced e^‒ and h⁺ to opposite facets; (2) doping TiO₂ with metal or nonmetal elements to reduce the band gap by modifying its band structure [18]; (3) surface modification, such as modification of TiO₂ with carbon materials [19] and deposition of noble metals on TiO₂ [20-22] to improve its light harvesting properties and drive interfacial charge separation; and (4) the formation of homojunctions [23] or heterojunctions [24] to achieve spatial segregation of the carriers.

The simple adsorption of pollutants on the surface of a photocatalyst is known to be highly important, as it can retard the recombination of the photo-excited carriers by facilitating the migration of the photo-induced e^‒ and/or h⁺ [25-27]. Therefore, the surface chemistry and structure of TiO₂ are vital factors that can affect its photocatalytic performance by influencing the selective adsorption and photocatalytic oxidation of substrates [28, 29]. Surface fluorination has been successfully used to tune the surface chemistry and structure of TiO₂ to improve its photocatalytic activity [30]. The adsorption of fluoride ions on the surface of TiO₂ not only greatly alters its surface adsorption properties, but also significantly promotes surface hole transfer and the preferential formation of mobile hydroxyl radicals (·OH_free) with superior reactivity. Doping TiO₂ with fluorine is also an efficient way to improve its photoreactivity by tuning its band structure and altering localized electronic structures [31]. The strong complexation between fluoride ions and TiO₂ has also been used to tailor the shape of TiO₂. For example, hollow-structured TiO₂ can be obtained via fluoride-induced self-transformation [32, 33]. In addition, the surface addition of fluoride ions during the growth and crystallization of TiO₂ favors the stabilization of high-energy {001} facets [17], resulting in unique facet junction effects that can promote charge separation efficiency [34]. In short, fluorine has unique and versatile effects on the photocatalytic properties and performance of TiO₂.

In this review, we briefly introduce fluorine effects on the structure, properties, and photocatalytic performance of TiO₂ (i.e. the effects of fluorine on photocatalysis) due to (1) fluoride surface modification (F-TiO₂), (2) fluorine doping (F-doped TiO₂), and (3) fluorine-mediated morphology tailoring. In addition, we also summarize recent progress in the use of fluorine effects for various photocatalytic applications, such as selective degradation/oxidation, photocatalytic H₂ production, and CO₂ reduction. In the last part, we analyze the advantages and disadvantages of fluorine effects in photocatalysis and provide an outlook on the future of fluorine effects.

Since Pelizzetti et al. [30, 35] found that the degradation of phenol over TiO₂ was greatly accelerated after the addition of NaF, surface modification of TiO₂ using fluorine has attracted a great deal of attention [9, 36-41]. The positive influence of fluoride surface modification on the degradation of organic pollutants over UV-illuminated TiO₂ was subsequently confirmed not only in aqueous solutions of benzoic acid [42], cyanuric acid [43], benzene [44], 4-chlorophenol [45], nitrosodimethylamine [46], tetramethylammonium [47], and azo dyes [48, 49], but also in gaseous photocatalytic reaction systems such as acetaldehyde [50] and acetone [50, 51]. The adsorption of fluoride ions was established to be positively related to the photoreactivity of TiO₂ [30]. With increasing solution pH, both the adsorption of fluoride on TiO₂ (Fig. 1(a)) and the removal ratios of organic pollutants (Fig. 1(b)) first increase and then decrease, with an optimal pH value of approximately 3.

(1)

(2)

(3)

(4)

When TiO₂ nanoparticles are immersed in water, various hydrates are formed by the coordination unsaturated surface Ti^VI ions. Surface OH^‒ groups are produced by dissociation of the chemisorbed water molecules and then give rise to ≡Ti-OH. Accordingly, positively charged TiO₂ is present in an acidic medium (Eqs. (1) and (2)). For example, surface specification modeling indicates that 89% ≡Ti-OH₂⁺ is formed at pH = 3 [30]. The adsorption of fluoride on the surface of TiO₂ can be considered as a ligand exchange between F^‒ and ≡Ti-OH (Eqs. (3) and (4)), and acidic media appears to be most conducive to the formation of this complex, with an equilibrium constant of 10^7.8 being observed [9]. According to surface speciation modeling, the dominant species at pH = 3 is ≡Ti-F. A maximum adsorbed fluoride concentration of 0.27 mmol g^‒1 was obtained at pH = 3; this value is considered to be surface site density of TiO₂ (Degussa P25) [52].

Although many investigations into the intrinsic relationship between surface fluoride modification and the performance of TiO₂ have been reported [48, 52, 53], the mechanism of the fluorine effect is still under debate.

Alcohol was employed as an exploratory tool to analyze the photocatalytic mechanism. The experimental results indicated that 90% of phenol was degraded via reaction with surface-bonded hydroxyl species (·OH_bonded), while the remaining of 10% could be degraded by direct reaction with the h⁺. However, the phenol-oxidation reaction on surface-fluorinated TiO₂ occurred almost completely through homogeneous hydroxyl radicals due to the inapplicability of surface-restrict hydroxyls when fluoride ions were used (Eqs. (5) and (6)) [30].

(5)

(6)

To provide direct experimental evidence, Mrowetz et al. [49] investigated the generation of hydroxyl radicals in a surface-fluorinated TiO₂ system via electron paramagnetic resonance (EPR) techniques. Compared to unmodified TiO₂, a stronger DMPO-·OH adduct signal was observed for F-TiO₂. As can be seen from Fig. 2, the kinetic profile of the DMPO-·OH adduct demonstrates that a much higher content of this species could be formed under light illumination for surface-fluorinated TiO₂.

The mobile nature of the ·OH_free species produced on an F-TiO₂ film under UV irradiation was well demonstrated in a study by Choi et al. [54]. Using the photo-oxidation of stearic acids in ambient air as a model reaction, they observed an accelerated photo-oxidation rate over illuminated TiO₂ film after surface fluorination. This observation confirmed that the formation of airborne oxidants was enhanced on the surface of F-TiO₂, and that the surface fluorination of TiO₂ promoted the desorption of ·OH radicals (forming ·OH_free). The free ·OH radicals were a transient species with an average lifetime of approximately 0.5 s.

Maurino et al. [55] reported that 1.3 mM of H₂O₂ was formed in a UV-irradiated F-TiO₂ suspension in the presence of oxygen and a hole scavenger. Importantly, the generation rate of H₂O₂ followed the F-TiO₂ surface speciation, and was greatest when the surface was thoroughly covered with ≡Ti–F groups, which demonstrated the importance of the surface speciation on the photocatalytic reaction. The experimental results indicated competition between fluoride and superoxide/peroxide species on TiO₂, which suppressed the degradation of H₂O₂.

Interestingly, Macyk et al. [43] found that bulk TiO₂ did not degrade cyanuric acid. However, this compound could be efficiently degraded in F-TiO₂, and its photodegradation rate did not decline after the addition of the ·OH scavenger ethanol. They proposed that singlet oxygen (¹O₂), rather than · OH, was responsible for the destruction of cyanuric acid (Fig. 3). Surface modification of TiO₂ by fluoride ions not only promotes energy transfer, but also restricts interfacial electron migration (Eq. (7)). However, the presence of the proposed ¹O₂ was not confirmed by electron spin resonance (ESR).

(7)

Yu et al. [56] reported that the photoreactivity of TiO₂ thin films toward acetone degradation was greatly improved after modification with trifluoroacetic acid (TFA). The -CF₃ group originating from the complex between TFA and TiO₂ was proposed to act as an electron scavenger to reduce the recombination rate of photo-generated carriers (Fig. 4). A similar model was then used by Choi et al. [48] and Yu et al. [31] to interpret the effect of fluoride. They asserted that the high electronegativity of fluorine is important, as it could increase the electron-withdrawing capacity of the surficial ≡Ti-F bands and therefore retard the recombination of photo-produced carriers (Fig. 5).

Xu et al. [52] proposed a different model based on Helmholtz layers to illustrate the effect of surface fluorination (Fig. 6), in which fluoride in the Helmholtz layers facilitates the desorption of surface-bound ·OH species from illuminated TiO₂ via hydrogen bonding with fluorine to form free ·OH radicals in the diffuse layer.

Interestingly, Jin et al. suggested that residual fluoride ions in the form ≡Ti:F-H on the surface of TiO₂ were responsible for greatly enhancing the adsorption of O₂, which in turn would greatly improve the photocatalytic activity by quickly scavenging photo-generated electrons [57].

Fluorine effects are not only sensitive to the solution pH, but also to the phase structure of TiO₂. We observed a positive fluorine effect in the degradation of Brilliant Red X-3B dye (X3B) over anatase TiO₂, while rutile TiO₂ exhibited a negative fluorine effect under identical conditions [58]. For mixed-phase anatase-rutile TiO₂, fluoride only played an active role in the photocatalytic behavior when the amount of anatase was greater than 40%. The negative effect of the surface fluorination of rutile TiO₂ was attributed to the unattainable generation of free ·OH species due to the quicker recombination of photo-produced e^‒-h⁺ pairs.

It should be noted that fluoride ions are inert; i.e. fluoride ions alone show little effect on the degradation of organic pollutants in the absence of the photocatalyst TiO₂ (Fig. 7(A)). Xu et al. [52] showed that the photoreactivity of TiO₂ depended on the amount of adsorbed fluoride ions (Fig. 7(B)).

The fluorine effect of TiO₂ can be further enhanced with the addition of an electron scavenger or electron sink. For example, Wang et al. [38] reported that Cu²⁺ and F^‒ had a synergistic effect on the photo-degradation of phenol in TiO₂ suspensions, which was ascribed to the shielding effect of fluoride ions on the charge segregation efficiency and the accessible trapping of the photo-excited electrons via the Cu²⁺ adsorbed on the TiO₂ surface. Similarly, Kim et al. [59] reported the synergistic effects of surface fluorination and Pd deposition in enhancing the photoreactivity of TiO₂ (F-TiO₂/Pd) toward urea degradation.

Surface fluorination has a negative effect on the photocatalytic activity of rutile TiO₂ due to the quick recombination of the charge carriers. However, surface fluorination was observed to have a positive effect on a rutile TiO₂ photocatalyst in the presence of Ag⁺, which is a typical electron scavenger (Table 1). The efficient removal of photo-generated electrons by Ag⁺ extends the hole lifetime, allowing them enough time to oxidize solvent water to produce ·OH_free radicals [52].

Although the addition of NaF to TiO₂ suspensions can speed the photo-degradation of organic contaminants, environmental regulations do not permit the disposal of such fluoride-containing wastewater. To solve this problem, Xu et al. [60] modified TiO₂ using barely water-soluble salts, namely, fluorite (CaF₂) and fluorapatite (Ca₁₀(PO₄)₆(OH)_2‒xF_x). The samples with small loading amounts exhibited improved photocatalytic reactivity compared to pristine TiO₂ toward the sorption and degradation of phenol and 2, 4-dichlorophenol. Ionic chromatography analysis indicated that the concentrations of dissolved fluoride and phosphate anions in the filtrate after a five-cycle test were approximately 26 and 30 μM, respectively; these values were well below the discharge standards for fluoride-containing industrial wastewater.

The strong affinity of fluoride ions to TiO₂ and high redox potential of F/F^‒ (E^ɵ_F/F- = 3.6 V vs NHE) [48] make surface fluoride effects very stable. For example, the photoreactivity of surface-fluorinated TiO₂ nanosheets with exposed (001) facets remains almost unchanged even after repeated use for more than five cycles in both the degradation of organic dye [61] and the oxidation of acetone gas [62].

Doping TiO₂ with fluorine is also an efficient way to improve its photoreactivity. Unlike surface fluorination, which affects the photocatalytic activity of TiO₂ by changing its surface chemistry/structure, doping TiO₂ with fluorine improves its photoreactivity through tailoring the band structure. To further enhance the effect of fluorine doping, strategies such as co-doping TiO₂ with metal or nonmetal elements and coupling F-doped TiO₂ to form heterojunctions have been used.

To inhibit the recombination of photo-driven e^‒ and h⁺, and/or to extend its light-response to the visible range, doping TiO₂ with metal or non-metal elements has been extensively studied [63-68]. Yu et al. [69] prepared highly photo-responsive F-doped nanocrystalline TiO₂ via the hydrolysis of titanium tetraisopropoxide (TTIP) in NH₄F solution. High-resolution X-ray photoelectron spectroscopy (XPS, Fig. 8(A)) showed two contributions to the F 1s spectrum corresponding to doped fluorine (TiO_2-xF_x) and surface-adsorbed fluoride ions (≡Ti-F). Charge compensation, therefore, resulted in the conversion of some Ti⁴⁺ into Ti³⁺ due to F-doping. When such TiO₂ is subjected to UV illumination, the photo-induced electrons on the Ti³⁺ surface states can migrate to adsorbed molecular oxygen, while the holes can accumulate at the valence band. In this way, the recombination of the e^‒-h⁺ pairs is impeded. Therefore, TiO₂ with high photoreactivity can be achieved by F-doping (Fig. 8(B)).

To clarify the recombination of photo-generated charges between two different energy states, photoluminescence (PL) spectroscopy was carried out. Li et al. [64] showed that the PL peak of an F-doped TiO₂ sample (FTO-800) could be deconvoluted into three peaks (Fig. 9(a)), where peak 1, which was centered at approximately 465 nm, originated from oxygen vacancies with two trapped electrons (F center), peak 2 centered at approximately 525 nm was due to oxygen vacancies with one trapped electron (F⁺ center), and peak 3 centered at approximately 627 nm was derived from an interference triggered by F-doping. Based on the PL characterization results, they proposed a structural model of the energy states between the VB and CB of F-doped TiO₂ (Fig. 9(b)).

Illas et al. [70] showed that F-doping could improve the thermodynamic stability of all polymorphs of TiO₂ (rutile, anatase, and brookite). For anatase, F-doping led to the formation of Ti³⁺ gap states that were well positioned in the middle of the band gap. However, F-doping had little effect on the band gaps of rutile and brookite.

Although TiO₂ nanosheets with high-energy facets show high photocatalytic activity, they do not respond to visible light. Nonmetal doping has shown great ability to endow titania with visible-light absorption; however, the introduction of dopants into anatase TiO₂ sheets with prominent {001} facets is challenging, but desirable. The incorporation of dopants into well-faceted anatase TiO₂ sheets with relatively strong crystallinity via facile post-treatment methods has proven to be unfeasible. Additionally, the introduction of dopant feedstocks in the reaction medium could affect the nucleation and growth of anatase TiO₂ sheets in an inevitable manner, preventing the desired TiO₂ sheets from being attained [71]. By subjecting TiN powder to hydrothermal treatment in HF solution, Liu et al. [72, 73] successfully synthesized visible-light-active N-doped anatase TiO₂ sheets with {001} as the main facet (Fig. 10). Similarly, Yu et al. [74] prepared C-doped TiO₂ nanosheets through the hydrothermal treatment of TiC in HF solution. The calcination of TiOF₂ cubes in a H₂S atmosphere was used by Wen et al. [75] to successfully fabricate S-doped TiO₂ nanobox assemblies from nanosheets.

Many recent works have focused on the co-doping of TiO₂ with fluorine and metal ions such as Ca [76], Cu [77], Fe [78], Sn [79], W [80], and Y [81]. For example, Wu et al. [79] synthesized F and Sn co-doped rutile TiO₂-based fluorine-doped tin oxide (FTO) films, and found that the photocurrent and the photoelectrochemical water oxidation of the FTO film were greatly improved (Fig. 11). They concluded that the presence of Sn stimulated the segregation of the photo-triggered e⁻ and h⁺, while the addition of F could retard the recombination of the carriers.

The formation of ·OH radicals is greatly improved after co-doping Cu²⁺ and F⁻ into TiO₂ nanotubes, as the doped F creates new active sites for the production of free ·OH species. Furthermore, the surface acidity of Cu²⁺ and F⁻ co-doped TiO₂ nanotubes promotes the adsorption and therefore the photocatalytic degradation of methyl orange (MO) [77].

Coupling TiO₂ with carbon materials or other semiconductors can facilitate the separation of the photo-generated carriers, which is beneficial to further improve the photoreactivity of fluorine-doped TiO₂ by retarding carrier recombination.

Giannakas et al. [82] showed that (1) co-doping TiO₂ with N and F created Ti³⁺ lattice states below the conduction band of TiO₂, and (2) coupling N and F co-doped TiO₂ with V₂O₅ greatly extended its light-response range, which in turn improved the photoreactivity of TiO₂ in the reduction of Cr(Ⅲ) and the oxidation of organic pollutants (Fig. 12). Similarly, the photoreactivity of F-doped TiO₂ was improved after coupling with Bi₂O₃ to form a Bi₂O₃/F-doped TiO₂ heterojunction [83]. Carbon nanotubes (CNTs) have also been used to modify F-doped TiO₂, improving its photoreactivity by reducing the band gap from 3.02 to 2.7 eV and facilitating quick removal of the photo-generated electrons [84].

The complexation between fluoride ions and TiO₂ has also been used to control the morphology of TiO₂ to create structures including (1) hollow-structured TiO₂ such as hollow TiO₂ microspheres (TiO₂-HMSs), (2) TiO₂ nanocrystals with high-energy facets, and (3) TiO₂ mesocrystals. Since acidic solution favors the complexation of fluoride ions and TiO₂ (Fig. 1), morphology control experiments using fluoride ions as a shape-directing agent have usually been carried out in acidic solution (mainly HF solution) [85].

Although many studies related to engineering the morphology of TiO₂ have been reported, the development of TiO₂-HMSs still merits substantial interest because of their unique advantages, such as low density, large surface area, high surface permeability, and good photo-absorption [12-15, 33, 86-90]. Recently, many photocatalysts with hollow frameworks have been obtained via a one-step template-free approach based on the novel inside-out Ostwald ripening mechanism. However, the Ostwald process is very slow. For example, 14 days were required to synthesize TiO₂-HMSs via a solvothermal route from a mixture of ethyl ether and TiOSO₄ (Fig. 13) [91].

Mann et al. [32, 86, 92-96] reported that the Ostwald ripening process can be greatly accelerated in the presence of fluoride ions (F^-). This effect is termed fluoride induced self-transformation (FIST), and can give rise to the dissolution of amorphous TiO₂ to form the soluble complex TiF₆^2‒ owing to the strong affinity of F^- to Ti⁴⁺ (Eqs. (8) and (9)). The resultant TiF₆^2‒ complex then preferentially recrystallizes on amorphous TiO₂ microspheres (Eq. (10)), producing a layer of crystalline anatase TiO₂ (shell). FIST can be applied to obtain TiO₂-HMSs within 10‒24 h.

(8)

(9)

(10)

Pan et al. [97] developed a versatile targeted etching strategy for the large-scale preparation of urchin-like mesoporous TiO₂ hollow spheres (UMTHS) with tunable particle size. The primary feature of the method was the use of a low-temperature hydrothermal reaction of surface-fluorinated, amorphous, hydrous TiO₂ solid spheres (AHTSS) assisted by a polyvinylpyrrolidone (PVP) protective coating. Due to confinement by PVP and the penetration of water, the highly porous AHTSS are selectively etched and hollowed by fluoride (FIST) without destroying their spherical morphology (Fig. 14).

Recently, we found that TiO₂-HMSs could be rapidly prepared within 3 h in the presence of H₂O₂ via hydrothermal treatment of a mixture of (NH₄)₂TiF₆ and (NH₄)₂CO. The presence of H₂O₂ not only promoted the crystallization of anatase TiO₂ nanocrystals, but also induced the hollowing of TiO₂ (Fig. 15) [89, 98]. We also found that TiO₂-HMS assemblies could be fabricated from TiO₂ hollow nanoparticles by a simple hydrothermal route in a Ti(SO₄)₂-NH₄F-H₂O₂ mixed solution at 180 ℃ after only 3 h. A possible growth mechanism for the hierarchical TiO₂-HMSs based on a twice H₂O₂-assisted FIST process was proposed (Fig. 16) [99]. The proposed route for the synthesis of TiO₂-HMSs had the virtues of being facile, repeatable, and easily scaled-up.

The estimated surface energies of the (001) and (101) facets of anatase TiO₂ are 0.90 and 0.44 J m^‒2, respectively[100]. As a result, anatase TiO₂ nanocrystals generally have a truncated bipyramidal shape dominated by the energetically stable (101) facets. By reversing the relative stability of the {101} and {001} facets using HF as a shape-directing agent, Lu et al. [17] obtained high-performance, responsive anatase TiO₂ microcrystals with 47% {001} facets on their surface by hydrothermal treatment at 180 ℃ for 20 h (Fig. 17).

Inspired by these promising findings, many reports of the controllable fabrication of micro- or nanocrystalline TiO₂ with high percentages of reactive facets, such as {001}(Fig. 18(a)) [62, 101-104], {010} [105], {100} [106-109], and {110} (Fig. 18(b)) [110], have recently been published. Highly photo-responsive anatase TiO₂ single crystals with exposed {001} and {110} facets have been successfully prepared from Ti powder using solutions containing both H₂O₂ and HF [110]. Zhang et al. [111, 112] fabricated anatase TiO₂ microspheres with exposed mirror-like plane {001} facets via the hydrothermal treatment of titanium foil in HF solution (Fig. 18(c)). Using anodized TiO₂ nanotubes, H₂O₂, and HF solution as starting materials, anatase TiO₂ microsphere assemblies were obtained with dou-like single-crystalline structures via a hydrothermal route (Fig. 18(d)) [113].

Layered TiO₂ nanosheets with exposed {001} facets were prepared by a direct hydrothermal route in the presence of (NH₄)₂TiF₆, H₃BO₃, and PrOH (Fig. 19) [114]. The superior photo-activity of the layered TiO₂ was attributed to the synergetic effect of the layered structure and the {001} facets.

TiOF₂ nanocubes can be transformed into hollow TiO₂ nanoboxes (TiO₂-HNBs) by direct calcination (Fig. 20) [115-117]. The formation of these hollow structures also occurs through a FIST process. The hollow TiO₂-HNBs are assembled from high-energy TiO₂ nanosheets in the presence of a high concentration of fluorine (Eq. (11)).

(11)

TiOF₂ cubes can also be transformed into TiO₂-HNBs via a solvothermal process in alcohol solutions (tert-butanol and ethanol) (Fig. 21) [33]. The dehydration of the alcohols results in the formation of a low concentration of water, which is very important in controlling the hydrolysis of TiOF₂ for the production of TiO₂ nanosheets with high-energy facets (Eq. (12)).

(12)

HF can act not only as a capping agent, but also an etching agent. Thermal treatment of TiO₂ single crystals with 35% {001} facets in the presence of HF-containing solution produced anatase TiO₂ single crystals with nano-carving of the {001} facet (Fig. 22) [118]. According to simulated calculations, HF stabilizes the growth of {001} facets at low concentrations, while selectively hampering the growth of {001} facets at high concentrations.

Similarly, Pan et al. [119] fabricated quantum dot (anatase TiO₂ nanodot) self-decorated TiO₂ nanosheets that contained Ti³⁺ defects due to HF etching via long-time hydrothermal treatment. However, these Ti³⁺ defects could be completely removed by further hydrothermal treatment with deionized water (Fig. 23).

Since the exposure of the high-energy {001} facet can be attained using fluoride ions as a capping reagent, the elimination of adsorbed surficial fluoride ions from TiO₂ nanocrystals with high-energy facets by heat post-treatment usually leads to the fusion of TiO₂ nanocrystals along the (001) direction [51, 120-123]. Using this strategy, Wang et al. [120] successfully fabricated one-dimensional single-crystalline TiO₂ nanochains by a two-step solvothermal approach (Fig. 24) The as-prepared TiO₂ nanochains showed high conversion efficiency in a dye-sensitized solar cell. This approach is significant for the design of novel nanoarchitectures.

TiO₂ mesocrystals have a highly ordered superstructure consisting of mesoscopic (1‒1000 nm) crystals, which exhibit the same scattering fashion and behavior as TiO₂ single crystals in polarized light, and can be considered as an intermediate species between TiO₂ nanopolycrystals and TiO₂ single crystals [124]. TiO₂ mesocrystals were first achieved by the topotactic conversion of NH₄TiOF₃ mesocrystals in the presence of nonionic surfactants [125, 126]. Currently, attempts to directly synthesize mesocrystals in the absence of surfactants are being made [124, 127-130]. Ye et al. [129] obtained anatase TiO₂ mesocrystals with spindle-shaped nanopores via mesoscale assembly in an additive-free tetrabutyl titanate-acetic acid system under solvothermal conditions (Fig. 25).

In the presence of fluoride ions (HF [128] or NH₄F [129]), anatase TiO₂ mesocrystal assemblies can be obtained from nanocrystals with exposed {001} facets. According to a study by Chen et al. [127], the generation of TiO₂ mesocrystals consists of a four-stage process (Fig. 26): (1) Production of a mixture of NH₄TiOF₃ and anatase TiO₂ (0‒0.5 h, step (A); (2) dissolution and redistribution of NH₄TiOF₃ to form anatase TiO₂ (0.5‒3.0 h, steps (B–E); (3) growth of square-shaped anatase TiO₂ crystals (3.0‒24 h, step (F), and (4) subsequent heating to obtain TiO₂ mesocrystals (step (G)).

Encouragingly, fluoride effects have also been demonstrated in many non-TiO₂ photocatalysts, such as Bi₂WO₄ [131], BiPO₄ [132], Fe₂O₃ [133], SrTiO₃ [134] and even metal-free graphitic carbon nitride (g-C₃N₄) [135].

Shi et al. [131] showed that fluoride ions could be doped into the lattice of Bi₂WO₆, resulting in a wider valence bandwidth and lower valence band position (Fig. 27(a)), which in turn improved its photoreactivity toward Methylene Blue (MB) dye degradation. The positive effect of fluoride ions on the photoreactivity of different iron (hydr)oxides was reported by the group of Xu (Fig. 27(b)) [133]. In the presence of NaF, ·OH radicals were detected using an ethanol scavenger, whereas such radicals were not found in the absence of NaF. In addition, similarly to in the case of rutile TiO₂, the photoreactivity of surface-fluorinated iron (hydr)oxides could be greatly enhanced in the presence of AgNO₃, which acts as an electron scavenger.

Shevlin et al. [136] prepared F-doped g-C₃N₄ by the polymerization of urea in the presence of NH₄F. The synthesized F-doped g-C₃N₄ exhibited improved optical absorption and enhanced photocatalytic activity in water-splitting for hydrogen production. Density functional theory (DFT) and time-dependent DFT (TDDFT) calculations showed that doping g-C₃N₄ could significantly reduce its bandgap and thus its onset energy for optical absorption, extending its light absorption from 420 to 515 nm. In addition, fluorine-doping could retard the recombination of the charge carriers, improving the photoreactivity of g-C₃N_4.

Fluoride ions can also be used as hydrophobic modifiers to synthesize super-hydrophobic SiO₂ supports to be loaded with nano-sized TiO₂ photocatalysts. The generation of surface Si–F improves the super-hydrophobicity of the TiO₂/SiO₂ photocatalysts, which in turn changes the adsorption and photocatalytic activity of TiO₂ in the degradation of organic pollutants [137] and the reduction of CO₂ [5].

Universally, the simple adsorption of pollutants on the surface of a photocatalyst is of great importance for efficient oxidation [25]. If an organic substrate adsorbs strongly on the surface of TiO₂, fluoride modification would obstruct the surface sites for organic adsorption; as a result, the oxidation mechanism of the substrate may shift to ·OH-mediated oxidation rather than direct hole transfer [9].

We compared the effects of ROS such as holes (TiO₂/tert-butyl alcohol/UV), surface-bonded ·OH radicals (TiO₂/UV), and mobile ·OH radicals (F-TiO₂/UV) on the photo-oxidation of phenol. The attack of ·OH radicals, whether surface-bound or mobile, was found to favor the production of catechol (hydroquinone), while hole oxidation tended to form resorcinol (Fig. 28) [138].

Ye et al. [139] investigated the photocatalytic oxidation of phenol in the presence of surface-fluorinated high-energy TiO₂ nanosheets. Experimental analyses demonstrated that the percentage of exposed {001} facets of the nanocrystal increased with the nominal atomic ratio of fluorine to titanium, and that both phenol photo-oxidation and catechol selectivity (yield) were positively correlated with the percentage of exposed {001} facets (Table 2). The enhanced phenol conversion and catechol selectivity (yield) were attributed to the synergistic functions of the exposed high-energy {001} facets and surface fluorination.

In a study of the selective photo-oxidation of hexane into hexanone and hexanol by O₂ in UV-irradiated TiO₂ suspensions, Xu et al. [140] found that the contents of hexanone and hexanol, as well as the molar ratio of hexanol to hexanone, both increased toward saturation with increasing fluoride concentration.

Fluoride surface modification can change the surface chemistry properties and the surface framework of TiO₂, which in turn will affect the adsorption and degradation of organic pollutants. Zhao et al. [141] reported that surface fluorination changed not only the adsorption mode, but also the degradation pathway of Rhodamine B (RhB) dye over TiO₂. RhB is preferentially anchored on pristine TiO₂ via its carboxylic (‒COOH) group (Fig. 29(A)), while it is instead adsorbed via its cationic moiety (‒NEt₂ group) on F-TiO₂ (Fig. 29(B)). Due to their positively charged nitrogen-alkyl groups, the dye molecules underwent fast N-dealkylation on F-TiO₂, while on bulk TiO₂ direct cleavage of the chromophore ring structures of the dye predominated.

Yu et al. [28] found that TiO₂-HMSs exhibited tunable photo-selectivity in the decomposition of azo dyes in water. Fluorinated TiO₂-HMSs exhibited preferential degradation of MO as compared to MB. In contrast, TiO₂-HMSs without surface fluorine showed greater degradation of MB than MO (Fig. 30). The surface chemistry and the surface structure at the atomic level are pivotal factors that can steer the adsorption selectivity, and therefore, the photo-selectivity of TiO₂-HMSs toward azo dyes. An analogous finding was also documented by Beegam et al. [142].

Typically, photocatalytic hydrogen production and the photocatalytic degradation of organics have been studied in separate systems, because photocatalytic reduction and photocatalytic oxidation require different catalyst properties and reaction conditions. Choi et al. [143] studied the performance of illuminated Pt/TiO₂ for simultaneous H₂ production and oxidation of 4-chlorophenol (4-CP). The combination of both energy and environmental applications into a single photocatalytic system is known as "dual-functional photocatalysis". They found that the H₂ production increased 20-fold after the surface fluorination of Pt/TiO₂ due to the synergistic effect of the surface-adsorbed fluoride ions and deposited Pt nanoparticles. The deposition of Pt enhanced interfacial electron transfer, while modification with fluoride ions changed the oxidation pathway of organics from direct-hole or adsorbed hydroxyl radical (·OH_bonded) oxidation to free hydroxyl radical (·OH_free) oxidation (Fig. 31). Similar dual-functional photocatalysis was also observed over surface-fluorinated Pt/GO/TiO₂ for H₂ production and oxidation of 4-CP [144]. Iervolino et al. [145] reported the positive effect of the surface fluorination of Pt/TiO₂ for photocatalytic H₂ production with simultaneous glucose degradation; the optimal conditions for glucose degradation and H₂ production were found to be pH 6 and pH 2, respectively.

Surface fluorination plays an important role in the photoreactivity of high-energy TiO₂ nanosheets. Ruan et al. [146] showed that only 50% of the photocatalytic activity of TiO₂ nanosheets for H₂ production was retained when the surface-adsorbed fluoride ions were removed by alkali washing. The work functions of TiO₂, Na-TiO₂, and F-TiO₂ were calculated to be 5.97, 6.82, and 7.93 eV, respectively (Fig. 32). As the value of the work function reflects the electron-binding activity of surfaces, it is understandable that the photocatalytic H₂ production was greatly reduced after washing the surface of the TiO₂ nanosheets with NaOH solution.

In addition to surface fluorination, the doping of TiO₂ with fluoride ions can also greatly enhance its photocatalytic H₂ production rate. For example, using NH₄F as a dopant and capping agent, Fang et al. [147] synthesized F-doped porous single-crystal rutile TiO₂ nanorods. The solar-to-hydrogen conversion efficiency of the F-doped TiO₂ nanorods was found to increase 10-fold compared to that of the pristine TiO₂ single-crystals.

The excessive depletion of fossil resources has led to ever-increasing CO₂ emissions, resulting in adverse consequences such as the greenhouse effect and energy crisis [5, 148, 149]. Photocatalysis provides a sustainable way to transform CO₂ into high-value-added solar fuels, including methanol (CH₃OH), formic acid (HCOOH), methane (CH₄), and carbon monoxide (CO)[5]. Yu et al. [34] systematically studied the effect of the exposed facets on the photocatalytic performance of high-energy TiO₂ nanocrystals in CO₂ reduction by using HF as a shape-directing agent. They found that the ratio of exposed {101} and {001} facets of the TiO₂ nanocrystals was strongly related to their ability to photocatalytically reduce CO₂ to CH₄. The high-energy TiO₂ nanocrystals exhibited the highest photocatalytic CO₂ reduction performance at a 45:55 ratio of exposed {101} and {001} facets due to the formation of surface heterojunctions.

To improve their light-harvesting ability, Fang et al. [150] prepared Ti³⁺-containing TiO₂ nanocrystals with exposed (001) facets using TiCl₃ as titanium source and HF as a capping agent. The blue-colored TiO₂ nanocrystals exhibited enhanced photocatalytic activity toward CO₂ reduction and high CH₄ selectivity without any noble-metal cocatalyst.

Similarly, Xing et al. [149] synthesized mesoporous single crystals (MSCs) of reduced TiO₂ (TiO_2−x) using silica colloids as templates. The found that fluorination of the MSCs (F-MSCs) reduced the potential of the photo-excited electrons, resulting in a 13-fold increase in the CH₄ production yield and improvement of the CH₄ selectivity from 25.7% to 85.8%. The interaction between Ti³⁺ and the substituted fluoride ions resulted in the formation of internal electric fields in TiO_2−x, causing an increase in the Ti³⁺ impurity level, which thermodynamically favored the reduction of CO₂ to produce CH₄ (Fig. 33).

Generally, heat treatment increases the photoreactivity of anatase TiO₂ by increasing its crystallinity and reducing the lattice defects, which act as recombination traps for e^‒-h⁺ pairs. However, under normal conditions, when the calcination temperature is higher than 500–600 ℃, anatase TiO₂ will be irreversibly converted to less-reactive rutile phase TiO₂; this phenomenon restricts its applicability in high-temperature applications. For example, bathroom tiles, sanitary equipment, and self-cleaning glass for the control of organic pollutants require high processing temperatures and thus high-temperature stability [151, 152]. Hence, the development of strategies to fabricate highly thermostable anatase TiO₂ is very significant, but challenging [151, 153-155].

Padmanabhan et al. [156] explored the fabrication of high-temperature-stable anatase TiO₂ that could withstand temperatures as high as 900 ℃ by fluorine doping, using titanium tetraisopropoxide and TFA as starting materials. They found that the thermal stability of TiO₂ was correlated to the presence of low concentrations of fluorine in the lattice, and that the substitution of fluorine into the oxygen atomic sites hindered the extensive Ti-O-Ti bridging.

Our group found that surface-fluorinated high-energy TiO₂ nanosheets exhibited outstanding thermal stability against phase transition up to 1000 ℃ [51, 62]. As the calcination temperature is increased, oxygen vacancies (Ov) can be formed by the desorption of surface-bonded fluorine. These Ov also have a negative effect on the extension of Ti-O-Ti bridging and crystal growth, and can be removed either by the diffusion of lattice oxygen ion or oxygen from air [62]. The diffusion of lattice oxygen ions requires additional energy, which therefore shifts the anatase to rutile phase transformation to higher temperature (Fig. 34).

In conclusion, due to the strong complexation between fluoride ions and TiO₂ and its high electronegativity, fluorine plays an important role in photocatalysis, allowing the tailoring of the bulk and/or surface structures and control of the morphology of photocatalysts.

The advantageous effects of fluorine can be summarized as follows:

(1) Surface fluorination provides a facile way to enhance the photoreactivity of TiO₂ by changing its surface chemistry, and can be achieved by simply adding a small amount of NaF to suspensions of TiO₂. The strong adsorption of fluoride ions onto the surface of TiO₂ occurs through ligand exchange between F^‒ and ≡Ti-OH (Eqs. 1‒4), and the reaction is pH sensitive (Fig. 1). According to surface speciation modeling, the dominant species generated in acidic solution is ≡Ti-F. The strong affinity of TiO₂ to fluoride ions can affect the surface chemistry of TiO₂, and therefore the adsorption and degradation of organic pollutants (Figs. 29 and 30) [48].

(2) Because the redox potential of F/F^‒ is very high (E^ɵ_F/F- = 3.6 V vs NHE), the possibility of direct hole oxidation of surface-adsorbed fluoride ions to fluorine radicals can be excluded. The holes must therefore oxidize solvent water to produce hydroxyl radicals. Because water is not chemically adsorbed on the surface of TiO₂, the produced hydroxyl radicals are mobile (·OH_free), and can diffuse into the solution to attack the organic pollutants, greatly improving the efficiency of TiO₂ photocatalysis.

The strong affinity of fluoride ions to TiO₂ and the high F/F^‒ redox potential make fluorine effects very stable. For example, the photoreactivity of surface-fluorinated TiO₂ nanosheets with exposed (001) facets remains almost unchanged even after eight repeated uses for the degradation of organic dyes [61].

(3) The high electronegativity of fluorine can increase the electron-withdrawing capacity of the surficial ≡Ti-F bands, therefore retarding the recombination of photo-produced carriers, which is beneficial to the photoreactivity of TiO₂ (Fig. 5) [31]. The addition of electron scavengers or the formation of heterojunctions can further improve the fluoride effects.

(4) Doping TiO₂ with fluorine results in the formation of Ti³⁺ due to charge compensation, which can capture photo-induced electrons and thus improve photoreactivity by retarding the recombination of e^‒-h⁺ pairs.

(5) Fluorine effects have been used to control the morphology of TiO₂. TiO₂-HMSs [32], TiO₂ nanosheets with exposed (001) facets [17], and TiO₂ mesocrystals [127] have been successfully synthesized using the strong complexation between F^- and Ti⁴⁺. The dissolution-recrystallization process is of great importance (Eqs. (9) and (10)) not only for morphology control, but also for the enhanced crystallization of TiO₂ crystals.

Although rapid progress has been made in the use of fluorine effects in photocatalysis, the following issues remain to be addressed.

(1) Fluorine usually exhibits multiple effects in photocatalysis, including surface fluorination, doping, preferential exposure of certain facets, enhanced crystallization, and others. For example, the high photoreactivity of TiO₂ nanosheets with exposed (001) facets may result from both surface fluorination and the exposure of high-energy facets, which makes the analysis of fluorine effects very complex [157].

(2) The adsorption model is very important in the photocatalytic degradation of organic pollutants [141]. Because surface fluorination can affect the adsorption of substrates, the evaluation of fluorine effects is therefore sensitive to the molecular structures of the organics [48].

Although the effects of fluorine in semiconductor photocatalysis have been studied since 2000 [30, 35], the mechanism is still under debate. Further research into fluorine effects should focus on the following points.

(1) Further study of the underlying mechanism of fluoride effects should be conducted. With the development of modern instruments and devices, especially in situ techniques such as in situ diffuse reflectance infrared Fourier transform spectroscopy, greater basic understanding of fluorine effects could be achieved to eventually unveil the mysteries of fluorine effects in photocatalysis.

(2) The application of fluorine effects in photocatalysis should shift from aqueous solution to solid-state reactions. Although fluorine effects have proven to be an efficient way to dramatically improve the efficiency of photocatalysis, applications have mainly focused on water treatment. However, environmental regulations do not permit the disposal of such fluoride-containing wastewater [60]. The application of fluorine effects to air purification via the photocatalytic oxidation of volatile organic compounds (VOCs) is promising [62].

(3) Further improvement of the efficiency of semiconductor photocatalysis should be possible by combining fluorine effects with other factors to improve their light-harvesting ability, stimulate the separation of photo-generated carriers, and introduce co-catalysts. For example, the surface plasmon resonance effect, defect engineering, and single atomic catalysis could be used.