Recent progress in photocatalytic degradation of chlorinated phenols and reduction of heavy metal ions in water by TiO2-based catalysts

Among the various semiconductor photocatalysts reported so far, TiO2 is still the most promising material for real applications because of its excellent chemical and thermal stability, non-toxicity, low cost and highly oxidising photogenerated holes. This review summarises the recent progress (mainly over the last five years) in photocatalytic oxidation of non-biodegradable organic pollutants (chlorophenols) and reduction of toxic heavy metal ions in aqueous solution. The review details the recently developed strategies for improving the performance of TiO2-based photocatalysts, with particular respect to the visible light activity, charge separation efficiency, stability, separability and adsorption capacity for the remediation of the aforementioned categories of water contaminants, as these factors heavily affect the practical application of this technology. Next, the underlying semiconductor photocatalytic mechanisms have been thoroughly addressed experimentally and theoretically, together with the proposed defect engineering to improve the photocatalytic performance. Finally, the prospect of TiO2 photocatalysis was discussed.


Introduction
Access to potable water is one of the biggest challenges globally, especially in developing countries. Several persistent organic pollutants (pharmaceuticals, pesticides, personal care products, endocrine disrupters) are frequently detected in wastewater effluents [1][2][3]. Chlorophenols (CPs) and their metabolites are common and recalcitrant environmental pollutants, believed to have high bioaccumulation capability and carcinogenic effect. Figure 1 shows four major CPs (2-CP, 2,4-DCP, 2,4,6-TCP and PCP) that have been classified as first-degree toxic pollutants by the US Environmental Protection Agency (EPA 2003) [4,5].
Adverse effects of phenolic compounds on the human nervous system have been reported and linked to several health disorders, e.g. kidney failure, liver and lung damage [4][5][6]. CPs find extensive application in the chemical, forestry and wood-working industries. They are used as herbicides, insecticides, fungicides, wood preservatives and chemical intermediates [4,[7][8][9][10]. Generally, these organic pollutants are released into the environment because of several man-made activities including water disinfection, waste incineration, uncontrolled use of pesticides and herbicides and as byproducts in the bleaching of paper pulp with chlorine [11].
Various strategies have been employed to remove CPs from the environment, as shown in Table 1. Conventional methods include coagulation-flocculation, reverse osmosis, active carbon adsorption, biodegradation, air stripping and incineration. However, these techniques have some drawbacks and limitations, e.g. toxic byproduct generation, incomplete mineralisation, low efficiency, high energy and capital cost [12][13][14]16]. CPs absorb light of wavelength below 300 nm; thus, they do not undergo direct sunlight photolysis. Also, they are difficult to decompose completely to CO 2 and HCl by conventional methods due to high stability of the benzene ring [17]. Therefore, it is important to find innovative and cost-effective techniques for the safe and complete degradation of chlorinated organic pollutants such as CPs.
The current trend in wastewater treatment has moved from simple phase transfer to the destruction of pollutants such as by advanced oxidation processes (AOPs). These techniques involve reactive free radical species for non-selective mineralisation of organic compounds to harmless end products. The AOPs are generally based on the generation of the hydroxyl radicals ( . OH), which have a great oxidation power, thus can almost oxidise all organic compounds to carbon dioxide and water [15]. There are several methods for generating hydroxyl radicals, e.g. Fenton-based processes, UV-based processes, Ozone-based processes and Photocatalytic processes [16,[18][19][20].
In spite of the advantages of AOPs, there are several limitations in their use: (a) costs may be higher than competing technologies because of energy requirements, (b) harmful intermediates may be formed, (c) pre-treatment of the wastewater may be required to minimise cleaning and maintenance of UV reactor and quartz sleeves, (d) handling and storage of ozone and hydrogen peroxide require special safety precautions and (e) major challenges for the photocatalytic process are catalyst deactivation, slow kinetics, low photo efficiency and unpredictable mechanism [21]. However, AOPs are still more effective than the other techniques for wastewater treatment containing toxic and persistent pollutants [15].
Recently, application of titanium dioxide photocatalytic oxidation technology in environmental remediation has gained considerable attention as a cheap and clean alternative, though there are still various challenges to be resolved for its commercialisation. A high number of organic compounds (dyes, drug residues, pesticides and herbicides) have been eliminated from aquatic environment via TiO 2 photocatalysis [22][23][24][25].
Complementary to the reviews published before, this review only concentrates on the progress of TiO 2 photocatalysis over the past five years, thus providing a comprehensive view of the very recent progress in this field.

Basic principles of TiO 2 photocatalysis and applications
This review addressed the background in photocatalysis, and the removal of wastewater contaminants, via irradiation of semiconductors (e.g. nano-TiO 2 ) with a light of energy higher than its bandgap energy ( Figure  2). This is the first step in photocatalysis. Subsequently, the excitation of photogenerated electrons into the conduction band takes place, leaving behind energetic photogenerated holes in the valence band. Oxidation reaction is initiated in the valence band, while reduction reaction is initiated in the conduction band of the semiconductor. This requires the semiconductor to have proper band alignment required for organic pollutant decomposition (Figure 3), e.g. the conduction band minimum (CBM) should be higher (more negative in potential) than the redox potential for oxygen reduction to superoxide radicals, and its valence band maximum (VBM) should be lower (more positive in potential) than the redox potential for water or hydroxyl ion oxidation to hydroxyl radicals [13,27,28].

TiO 2 as a photocatalyst
One of the most important aspects of environmental photocatalysis is the availability of an efficient   photocatalyst, which should have several respects. For example, it is relatively inexpensive, physically and chemically stable, and non-toxic. Also, the band alignment in such a photocatalyst ( Figure 3) makes the photogenerated holes to be highly oxidising. In addition, photogenerated electrons have reduction potential enough to produce superoxide radical from oxygen [26,29,30]. TiO 2 is the typical material meeting these criteria. Zhang et al. [31] and Hirakawa and Nosaka [32], based on experimental evidence, proved that the conduction band electrons in TiO 2 have the potential to generate superoxide radicals (O 2 .− ) from dissolved oxygen in water.

Challenges with TiO 2 application in water treatment
The most studied phases of TiO 2 are anatase and rutile. The anatase phase is known to display higher photocatalytic activity than the rutile phase. The low activity observed in rutile (E g = 3.0 eV, λ = 413 nm), despite its lower bandgap energy than anatase (E g = 3.2 eV, λ = 388 nm), is attributed to the rapid rate of electron-hole recombination. Also, the CBM of anatase is about 0.2 eV higher than that of rutile ( Figure 3). This implies that the conduction band electrons in anatase are more reductive than those in rutile [33,34]. However, the inefficient utilisation of visible light and high recombination rate of photoelectrons and holes, that drastically reduces quantum efficiency, are the major drawbacks of photocatalytic application [28,35]. The challenges with TiO 2 as a potential photocatalyst for the removal of organic pollutants in wastewater are summarised in Figure 4.

Visible light activity
The ultraviolet (UV) region only accounts for <5% of the entire solar spectrum, which TiO 2 is responsive to. The strategies used in extending the optical absorption of TiO 2 into the visible region include, metal-doping, non-metal doping, dyes sensitisation and direct reduction of TiO 2 (e.g. hydrogenation, chemical reduction and electrochemical reduction). Three of the methods will be addressed here.
(a) Metal-doped TiO 2 (e.g. Ag, Pt, Au, Fe, Cu, Co, Ni, Cr, Mn and Ru). This method has been extensively investigated over the past decades for improving the photocatalytic degradation efficiency of TiO 2 for several organic pollutants under visible light (λ > 420 nm) [34,36]. Ag-doped TiO 2 and Cudoped TiO 2 are the most reported in photocatalytic water treatment. The lowest bandgap energy reported in the literature for metal-doped TiO 2 is 1.6 eV with 7.5% Cu-doped TiO 2 , prepared using a low-temperature hydrolysis reaction [37]. Khan  et al. reported the synthesis of Ag-doped TiO 2 using sol-gel method and the application in photocatalytic degradation of oxytetracycline under UVvisible light irradiation. The experimental results indicated 70% and 100% photocatalytic degradation of oxytetracycline in 60 min with pure TiO 2 and 1.9 wt.% Ag/TiO 2 , respectively, under UV-visible light irradiation. With only visible light irradiation (λ > 420 nm), the degradation efficiency of the 1.9 wt.% Ag/TiO 2 sample reduced drastically to 60% [38]. Generally, for metal-doped TiO 2 , the metal ion doping can either be substitutional or interstitial in the TiO 2 crystal lattice and can also form a mixture of oxides [38][39][40][41][42][43]. The limiting factors with this strategy include the nature of dopant species, dopant concentration, photochemical stability and thermal treatment. Several methods employed in metal-doped TiO 2 (e.g. impregnation, co-precipitation, sol-gel, hydrothermal, spray pyrolysis, ion-assisted sputtering and chemical vapour deposition) were reported [36,[44][45][46][47]. However, sol-gel method is commonly used in the preparation of nano-size-doped TiO 2 materials. The procedure is simple to carry out and does not require complicated facility [36]. The dopant dosage and calcination temperature play a significant role in photocatalytic activity of prepared TiO 2 samples [48]. The visible light activity of metal-doped TiO 2 is primarily due to the introduction of new energy levels in the bandgap of TiO 2 , and photogenerated electrons can be excited from the defect state to the conduction band in TiO 2 . Dopant ions above the optimum dosage could act as electron-hole recombination centres and will automatically reduce photocatalytic and quantum efficiencies [34]. (b) Non-metal-doped TiO 2 (e.g. S, N, C, P, F and B).
This method extends the visible light absorption of TiO 2 more than metal-doped counterparts. This is because their impurity states are close to the valence band edge of TiO 2 [34]. Compared to other non-metals, the N-doped and C-doped TiO 2 powders have been reported to show exceptional photocatalytic activity under visible light irradiation. Chen et al. [49] prepared C-TiO 2 , N-TiO 2 and C-N-TiO 2 using sol-gel method and tested their photocatalytic activities for the degradation of methylene under visible light irradiation.
In the absence of catalysts, about 5% photodegradation of methylene blue was observed in 7 h. The C-N co-doped sample exhibited the highest efficiency with about 95% methylene blue removal in 7 h, while 26, 60 and 70% degradation efficiencies were observed for unmodified TiO 2 , N-TiO 2 and C-TiO 2 samples, respectively. Co-doped TiO 2 generally exhibit higher visible light absorption than single-doped TiO 2 due to a synergistic effect between the two dopants, which increases charge carrier lifetime (quantum efficiency) [48]. The major drawbacks associated with non-metal doping include the formation of oxygen vacancies in the bulk in which the created defects act as electron-hole recombination sites, thus reducing the visible light photocatalytic efficiency. More so, instability is a major concern due to liberation of doped non-metals atoms in solution during photocatalysis [50,51]. (c) Reduced TiO 2 (TiO 2−x ). Hydrogenation of TiO 2 was reported in 2011 by Chen et al. as an effective method in shifting the bandgap of TiO 2 to around 1.5 eV (∼820 nm) with an optical onset of around 1.0 eV (black TiO 2 ) [52]. The photocatalytic activity of the coloured TiO 2 was compared with unmodified TiO 2 for methylene blue decomposition under solar light irradiation. Complete bleaching of methylene blue was achieved after 8 min for the black TiO 2 , while the unmodified white TiO 2 under the same experimental conditions, took nearly 1 h. Under similar experimental conditions, complete degradation of phenol was achieved after 40 min for the black TiO 2 , while the unmodified TiO 2 took 80 min to achieve the same. Total organic carbon (TOC) removal efficiency of the black TiO 2 was not reported by the author. Furthermore, without any form of treatment, the black TiO 2 was highly stable after eight successive photocatalytic degradation cycles. This discovery led to an extensive research into coloured TiO 2 synthesis globally. Other methods used in fabricating reduced TiO 2 include Al reduction [53], Zn reduction [54], imidazole reduction [55], NaBH 4 reduction [56], CaH 2 reduction [57] and electrochemical reduction [58]. It has been widely accepted that heating TiO 2 under vacuum or in a reducing atmosphere also leads to colour changes [59]. The coloured TiO 2 (red, yellow, blue or black) resulted from the creation of Ti 3+ or/and oxygen vacancies in the white TiO 2 . However, their activities in the visible region are still far from satisfactory [59,60].

Dispersibility
TiO 2 particles suffer from severe aggregation during photocatalytic application, which will reduce their active sites and light-harvesting capability. Since photocatalytic degradation occurs at the surface of TiO 2 , mass transfer limitations need to be reduced for its effective application in water treatment [61,62]. Two possible approaches employed are: (a) Stabilisation by support structures (glass beads, fibre glass, glass pellets, glass sheets, silica, organo-clays, stainless steel, Al 2 O 3 fibre, quartz beads, polymers, activated carbon and zeolites) [61]. However, there is still no ideal support that is suitable in terms of mechanical stability and selectivity [62]. Immobilisation on supports reduces TiO 2 surface area-to-volume ratio and the photocatalytic efficiency compared to unsupported TiO 2 particles [63]. (b) Stabilisation by surface modification (humic acid, carboxylic acids, amino acids and polylactic acid). This method tends to generate high negatively charged density (zeta potential) on the TiO 2 surface and promotes repulsive forces among particles, thus reduces aggregation rate. The main drawback is the stability of these organic modifiers during photocatalysis [64][65][66][67].

Separability
Recovery of fine particles of TiO 2 from solution is a huge challenge in suspension systems. Thus, the recovery and reusability of TiO 2 is required before the technology can proceed beyond lab scale to industrial application. Two potential approaches applied are: (a) Immobilisation on solid supports (glass fibres, glass, quartz and stainless steel). This method has been discussed under TiO 2 particles stabilisation using solid supports [63]. Although this approach overcomes the solid-liquid separation problem, slurry-type reactors are still more efficient over immobilised catalyst-type reactors, due to the availability of catalyst's surface and superior mass transfer properties [68].  [72,73]. It was reported that a direct deposition of TiO 2 onto the surface of magnetic oxide particles led to severe photo-dissolution of the magnetic core. In order to overcome the challenge, the addition of silica (SiO 2 ) layer between the magnetic core and the TiO 2 shell provided magnetic nanoparticles with a chemically inert surface, which inhibited the oxidation of the magnetic core (Fe 3 O 4 ) by the photogenerated VB holes in the outer layer (TiO 2 ) [74]. The synthesis of a multipurpose catalyst (graphene/TiO 2 /SiO 2 / Fe 3 O 4 ) was reported for the degradation of 2,4-dichlorophenoxyacetic acid (a herbicide). It took the advantages of TiO 2 photocatalysis, graphene (excellent electron transfer ability and high adsorption capability), Fe 3 O 4 (magnetic separation) and SiO 2 (suppressing photo-dissolution of the magnetic core). The TiO 2 composite achieved 100% degradation of 2,4-D under simulated solar irradiation, while commercial P25 recorded only 33% under the same conditions. More so, the composite was highly stable after eight successive cycles. The photocatalytic activity was dependent on the calcination temperature, magnetic core size and the silica layer content [75,76].

Adsorption capacity
The photocatalytic activity of TiO 2 is also dependent on its adsorption capacity for the pollutants in aqueous solution. Several approaches have been used to overcome the low adsorption capacity of TiO 2 .
Modification of TiO 2 with some chelating ligands (arginine and salicylic acid) and carbon nanomaterials (graphene, carbon nanotubes and activated carbon) have been reported to enhance the adsorption of organic water contaminants [77][78][79]. With respect to doping with carbon-based nanoparticles, the higher adsorption capability and better photocatalytic activity were reported to be related to their high surface area, high conductivity and high visible light absorption intensity [54,55].
respectively, under simulated solar light at pH 5.0. POM was reported to scavenge electrons from the conduction band of TiO 2 to decrease the charge recombination, thus enhancing the photocatalytic (quantum) efficiency of TiO 2 . Peng et al. synthesised a multipurpose photocatalyst, La-doped magnetic TiO 2 (Fe 3 O 4 /SiO 2 /La-TiO 2 ) using sol-gel method and investigated its photocatalytic activity for 2,4,6-TCP degradation [93]. The magnetic La-TiO 2 exhibited a higher photocatalytic degradation activity (90% in 60 min) compared to magnetic TiO 2 (50%). However, the La-doped magnetic TiO 2 was fairly stable after seven successive cycles. The drawback was attributed to the reduction in surface hydroxyl groups and reaction sites on TiO 2 . Sinirtas et al. [94] prepared a binary oxide catalyst (V 2 O 5 /TiO 2 ) using solid-state dispersion method and evaluated the photocatalytic activity for 2,4dichlorophenol degradation under UV-visible light irradiation. Photocatalytic degradation efficiencies of 55%, 60%, 70% and 85% were achieved in 30 min with TiO 2 (synthesised), commercial TiO 2 (P25), V 2 O 5 and V 2 O 5 /TIO 2 , respectively. The effect of surfactant additives (CTAB, HTAB and PVA) on the photocatalytic activity of the binary oxide catalyst was also investigated. Under similar operating conditions, the (V 2 O 5 /TiO 2 /CTAB) sample displayed the highest activity (complete degradation in 30 min), while the samples prepared with HTAB and PVA recorded degradation efficiencies of ∼92% and ∼55%, respectively. The enhanced photocatalytic performance of V 2 O 5 /TiO 2 /CTAB was attributed to an appropriate pore distribution and high separation efficiency of photoinduced charge carriers. However, the photochemical stability of V 2 O 5 /TiO 2 /CTAB was not reported.

Hexavalent chromium
The discharge of hexavalent chromium (Cr(VI)) from industrial processes, such as electroplating, leather tanning, metal fabrication, paint and pigment production into the aquatic environment has raised serious concerns over the potential water pollution issues [111,112]. In this regard, the reduction of toxic and carcinogenic Cr(VI) to the less harmful Cr (III) is imperative to mitigate their adverse effects towards biological and environmental systems. Over the years, photocatalytic reduction of Cr(VI) by TiO 2 photocatalysts has been regarded as one of the more promising treatment alternatives owing to its direct one-step photocatalytic reduction of Cr(VI) under UV light irradiation [113][114][115]. Table 3 summarises the recent findings (mainly over the past five years) on the application of TiO 2 -based photocatalysts for the photocatalytic reduction of aqueous-based Cr (VI).
The simplified mechanism for the photocatalytic reduction process: Cr 2 O 7 2− + 14H + + 6e − → 2Cr 3+ + 7H 2 O, indicating that the photogenerated electrons separated from the electron-hole pairs are solely responsible for the photocatalytic reduction of Cr (VI) to Cr(III). Lin et al. [131] revealed that the initial reduction rate of Cr(VI) using TiO 2 photocatalyst is following the Langmuir reaction model. They also discussed on the pH-related thermodynamic aspects of the photocatalytic reduction of Cr(VI), as well as found that the photocorrosion of TiO 2 to be relatively unfavourable in basic media. Ku and Jung [132] evaluated the photocatalytic reduction rates of Cr(VI) under various operating conditions (i.e. solution pH, initial Cr(VI) concentration, TiO 2 photocatalyst loads, light intensities and the presence of scavengers). They found a higher photocatalytic reduction rate of Cr(VI) in acidic medium than in basic medium, which is contradicting to the previous study by Lin et al. [131]. The higher photocatalytic reduction rate of Cr(VI) in acidic medium is attributed to the highly positively charged TiO 2 surfaces, as well as the presence of less competing molecules for redox reaction at the lower pH. Yoon et al. [133] immobilised TiO 2 photocatalysts onto inert substrates to overcome the photocatalysts' recovery and post-separation issues via anodising various nanotubular TiO 2 arrays under oxygen stream in the range 450-850°C for reduction of Cr(VI). Their findings demonstrated that anodised nanotubular TiO 2 arrays annealed at 450°C exhibited the highest reduction efficiency of Cr(VI) at 98% along with a hydrogen evolution rate of 38 μmol h −1 cm −2 . The presence of higher-activity anatase TiO 2 phase in the anodised nanotubular TiO 2 arrays annealed at lower temperatures (<450°C) could explain their excellent reduction efficiency of Cr(VI) [134].
In order to develop highly efficient TiO 2 -based photocatalysts with extended absorption into the visible light spectrum, the introduction of dopants into morphological-modified TiO 2 -based photocatalysts has been proven as an effective strategy. The introduction of dopants could hinder the rapid recombination rate of electron-hole pairs and narrowing the wide bandgap of TiO 2 photocatalysts and hence, improving its visible light photocatalytic activity. Lu et al. [135] optimised the amount of vanadium (V) dopant on TiO 2 nanosheets since this sheet-like structure possesses higher exposed surface areas for Cr(VI) adsorption. It is noteworthy that the introduction of V dopant resulted in the reduction of TiO 2 bandgap leading to a high visible light photoactivity, and the co-existence of RhB and Cr(VI) greatly improved photocatalytic degradation kinetics as Cr(VI) is an efficient scavenger of photogenerated electrons. Besides, the V-doped TiO 2 nanosheets could effectively trap photogenerated electrons and introduce a higher density defect (Ti 3+ ), which are essential to enhance the effective separation of photogenerated charge carriers and high quantum efficiency. They extended their works in another study by synthesising TiO 2 nanosheets loaded with Mn x O y and CoO x nanoparticles for simultaneous photocatalytic reduction of Cr(VI) and decontamination of organic compounds [136,137]. Similarly, the introduction of Mn x O y and CoO x nanoparticles resulted in a considerable enhancement of visible-light photoactivity, formation of active electron centres and the generation of Ti 3+ density defects.
Recently, Xue et al. [138] reported the rational design of an interfacial heterojunction photocatalyst between 2D BiOBr nanoplates and 1D rutile TiO 2 nanorods. Figure 5(a) shows the synthesis route of the heterojunction photocatalyst in which the 1D TiO 2 nanorods are first fabricated using the molten salt flux method followed by the bonding between positively charged alkoxide complexes, Bi-(OCH 2-CH 2 OH) 2+ and negatively charged TiO 2 nanorods via solvothermal reaction. This eventually leads to the nucleation of BiOBr nanoplates, which are later being vertically embedded with TiO 2 nanorods to synthesise the heterojunction photocatalyst (i.e. Figure 5 (b)). The HRTEM image ( Figure 5(c)) showed the interplanar spacings of 0.230 and 0.248 nm are well corresponded to the {200} and {101} facets of the rutile phase of TiO 2 , respectively. While the measured 0.277 and 0.353 nm fringes of BiOBr nanoplates can be assigned to the {110} and {101} planes of tetragonal BiOBr, respectively. Additionally, the rectangular spotty electron diffraction pattern can be clearly observed in Figure 5(d), indicating the single-crystalline rutile phase TiO 2 in this heterojunction photocatalyst. More importantly, the photocatalytic results ( Figure 5(e)) indicated that the TB-2 (i.e. the sample with a molar ratio of Ti to Bi, n(Ti/Bi) = 2:1) exhibited the highest photoreduction rate of Cr(VI) with an apparent rate constant of 0.025 min −1 under visible light illumination. This is ascribed to the formation of an interfacial junction between BiOBr and TiO 2 , which provides directional transfer nanochannels for the migration of photoinduced charge carriers and thus, enhancing quantum efficiency and reducing the charge recombination as evidenced by the PL spectra in Figure 5(f) leading to highphotoactivity as per Figure 5(g). Interestingly, similar theory on enhanced physisorption and charge transfer of PANI/MnO 2 / TiO 2 nanocomposite shows the importance of TiO 2 heterojunctions to improve the visible light-driven photoreduction of Cr(IV) [139].

Pentavalent arsenic
Owing to their carcinogenicity and mutagenicity nature, arsenic (As) compounds contamination in natural water can present a major threat to the environmental system and public health even in a very low concentration [140][141][142]. Previous studies are mainly focusing on the oxidation of acutely toxic arsenite (As(III)) to arsenate (As(V)); however, the high adsorbability of As(V) on soil could adversely influence bacteria and microscopic fungi [143,144]. Particularly, this section presents the recent findings on the photocatalytic reduction of aqueous As(V) by using TiO 2 -based photocatalysts in detail. Previously, Levy et al. [145] have applied pristine TiO 2 photocatalyst to investigate its photoreduction performance over As(V) under UV light in deoxygenated aqueous suspension. They found that the direct reduction of As(V) to As(IV) by accepting photogenerated electrons at conduction band was not thermodynamically possible. Nevertheless, the addition of sacrificial electron donors (i.e. alcohols or carboxylic acids) in the photocatalytic medium could lead to the production of strongly reductive radicals and thus, promoting the indirect reductive mechanism of As(V).
Similarly, the deposition of transition and precious metals onto TiO 2 -based photocatalysts has been widely attempted in order to extend their light absorption into the visible light spectrum as well as serving as active centres to inhibit the recombination of electron-hole pairs [34,146]. For instance, Andjelkovic et al. [147] prepared a Fe-doped TiO 2 photocatalyst via microwave-assisted hydrothermal method in order to provide large capacity for the adsorption and removal of As(V). They found that the adsorption efficiency of As(V) was proportional to the amount of Fe(III) dopants because it tends to provide a higher specific surface area which directly responsible for faster As(V) adsorption rate. Isotherm studies showed that the Langmuir model provides the best fitting for the adsorption data of As(V) better and the Fedoped TiO 2 photocatalyst attained the maximum adsorption capacity of 17.35 mg g −1 As(V). In another study, Liu et al. [148] deposited Fe onto morphological-modified titanate nanotubes (TNTs) for sequential As(III) oxidation and As(V) adsorption. The resultant Fe-TNTs exhibited a good adsorption performance for As(V) at acidic medium (i.e. adsorption capacity of ∼49.1 mg g −1 at pH 3; removal efficiency of 98.2%) due to the opposite surface charges between Fe-TNTs and As(V) anions. It was found that there are two different forms of deposited Fe that could improve the photocatalytic activity through: (1) Fe 3+ ions that locate in the interlayers to act as temporary electronor hole-trapping sites, and (2) attached α-Fe 2 O 3 that facilitates the charge separation from TNTs to reduce recombination of electron-hole pairs. A more recent study by Gao et al. investigated the performance of a 3D cake-like COOH functionalised anatase (TiO 2 @C) (Figure 6(a,b)) for simultaneous Cr(VI) reduction and As(III) oxidisation followed by the adsorption of As(V) [149]. The simultaneous conversion of Cr(VI) and As(III) was significantly enhanced from 14.5% to 61.8% and 38.5% to 92%, respectively, in the exposure system of TiO 2 @C/As (III)/Cr(VI). It is worth mentioning that As(V), which was oxidised from As (III), has also been effectively removed under acidic condition, as shown in Figure 6(c). Moreover, they also reported that the removal rate of As(V) was essentially relied upon the pH, TiO 2 @C dosage ( Figure 6(d)) and initial concentration of Cr(VI)/As(III) solution ( Figure 6(e)). Before the reaction, Figure 6(f) shows the full XPS spectrum of TiO 2 @C containing carbon groups, oxygen and titanium elements. The characteristic peaks of As (III) (45.1 eV) and As(V) (46.1 eV) were detected on TiO 2 @C with the proportion of 48.5% and 51.5%, respectively, after exposure to Cr(VI)/As(III) solution ( Figure 6(g)) [150]. As a result, the removal performance of As(III)/Cr(VI) using TiO 2 @C was improved through co-adsorption of As(III) and Cr(VI) ions, accelerated electrons transfer from As(III) to Cr(VI) initiated by carboxyl groups on TiO 2 , and larger binding sites provided by TiO 2 surface.

Outlook
After decade's development on different photocatalysts, TiO 2 is still widely recognized as the best semiconductor material for real application in photocatalytic water purification technology due to its low toxicity, high stability and availability. However, TiO 2 is confronted with some limiting factors such as poor visible light absorption, low photonic efficiency and low adsorption capacity. In this review, we have highlighted various strategies for TiO 2 modification (such as ion doping, cocatalyst decoration and heterojunction formation) not only to improve its light absorption but also to enhance charge separation. Furthermore, this review summarises the recent application of TiO 2 -based photocatalysts for the catalytic oxidation of CPs and reduction of heavy metal ions such as chromium and arsenic in aqueous solution, which are two typical while the most challenging topics.
The co-doping of TiO 2 with a metal and non-metal is receiving considerable attention and proved to be more effective than single ion doping towards efficient photocatalytic degradation of organic contaminants in water. This improvement was attributed to an extension in the visible light absorption capacity. Apart from the suppression of photogenerated electronhole recombination rate, the bandgap energy is also largely decreased due to the introduction of impurity energy levels above the valence band and below the conduction band. The solution pH is also an important operating variable in photocatalytic degradation of organic contaminants. In most TiO 2 -based studies, it was shown that acidic pH range (4-6) is the optimum for the efficient degradation of organic substrates in water. This was attributed to the positively charged surface of TiO 2 , favouring increased adsorption capacity for organic compounds. It is also worth noting that the fabrication of different bandgap alignments (i.e. Type I, II and III) and Z-scheme heterojunctions was proved to efficiently enhance charge separation of photogenerated electron-hole pairs and reduce the charge carriers recombination rate, beneficial to photocatalytic water treatment.
Reduction of a reactive, high oxidation state of metal ion to a lower oxidation state and less toxic species by TiO 2 -based photocatalysts has been regarded as an effective approach for heavy metal removal from aqueous environment. Upon UV light irradiation, the anodic and cathodic redox reactions are initiated, whereby the water oxidation by photoholes is involved and thus transfer of photoexcited electrons from the conduction band of TiO 2 photocatalysts to the adsorbed heavy metal ions takes place. Acceptance of electrons into the metal atom (reduction) could suppress its reactivity and thus, the toxicity effect. Additionally, it is also showed that the equilibrium adsorption capacity of TiO 2 photocatalysts is higher at lower pH medium, signifying a higher density of adsorption sites for heavy metals. In order to extend the functionality and applicability of TiO 2 photocatalysts in visible light range, several material modification strategies such as morphological alteration, dopants introduction and heterojunction are proved to be effective to separate the photogenerated charge carriers and T: 20 ± 2°C pH: 3.0); XPS spectra of (f) pure TiO 2 @C full spectrum before reaction; and (g) As 3d spectra of TiO 2 @C after exposed in As(III)/Cr(VI) solution. Adapted from Gao et al. [149]. Copyright 2018, Elsevier. Reprinted with permission. therefore, leveraging the photon responsiveness of pristine TiO 2 photocatalysts.
For both photocatalytic organic contaminants decomposition and heavy metal ions reduction, cocatalysts play a proved key role, which not only help charge separation but also lower the reaction barrier. The so far best cocatalyst is still noble metal, e.g. Pt.
To develop a low-cost cocatalsyt is also important for TiO 2 photocatalyst application. Based on the above discussion, we proposed that the future commercial and large-scale application of photocatalysis in water treatment is feasible if the following could be solved: . The preparation of low cost, reproducible and environmental-friendly cocatalyst/TiO 2 -based photocatalysts that can be photo-activated by solar light (not artificial UV light). . The efficient recovery and reuse of the photocatalysts during water purification via immobilisation on flexible and highly stable support materials. . The design of photocatalyst materials capable of selective photocatalytic oxidation of organic pollutants or reduction of heavy metal ions in water. . The design of efficient photocatalytic reactor with satisfactory efficiency in acidic, neutral and basic media, reflecting the real environment of the wastewater.
The fundamental understanding underlying the photochemical process is still far behind discovery of the photocatalysts.
The photocatalytic mechanism of a single photocatalyst is relatively simple and to some extent straightforward. The photochemical reaction involves the excitation of electrons to the conduction band of a photocatalyst while leaving positive holes at the valence band upon light illumination whereby the photon energy (hv) is greater than or equal to the bandgap energy of photocatalyst. Studies on charge carrier dynamics have been successfully investigated over the years using time-resolved analytical techniques (time-resolved absorption spectroscopies) to observe the generation, trapping, and transfer of electrons and holes during photocatalytic processes [151,152]. The generation of hydroxyl radicals and superoxide radicals in aqueous systems have been widely established in the literature through spin trapping ESR experiments and fluorescence spectroscopy. The next step is still unclear. The discovery of hydrogen peroxide in oxygen-free systems justifies the generation of hydroxyl radicals by valence band holes, though hydrogen peroxide can also be generated through oxygen reduction. It is still difficult to distinguish the hydroxyl radicals generated from both different pathways, that is, hole oxidation of adsorbed water/hydroxyl ions or through oxygen reduction. Different hydroxyl radical production pathway will provide a different route for improvement of the photocatalytic activity. More importantly, although the time-resolved IR has been successfully employed to other fields, e.g. homogeneous photocatalysis, it has not been used in water treatment. By using this cutting edge technology, the detailed reaction pathway and the rate-determining step could be elucidated, which can also indicate the factor to mitigate the toxic byproducts production during photocatalytic contaminants' treatment and heavy metal cation reduction. In addition, the development of highly efficient photocatalysts is growing intensively, the photocatalytic mechanism within the complex photocatalyst systems and compositing structures is also important while more challenging. These timeresolved spectroscopies should be applied to these complicate systems to understand the reaction pathway. Overall, a practical photocatalytic system for water treatment still requires substantial efforts.