Mechanisms of removal of heavy metals and arsenic from water by TiO₂-heterogeneous photocatalysis

Basic principles of heterogeneous photocatalysis

In HP, the SC is excited in a primary process by absorption of photons of energy equal or higher than the SC bandgap (E_g), leading to the promotion of an electron from the valence band (VB) to the conduction band (CB), with creation of a hole:

The electron-hole recombination is a unproductive process, detrimental for the photocatalytic efficiency, that should be avoided. In the absence of electron transfer processes, electrons and holes recombine in <100 ns in a TiO₂ nanoparticle (size: ca. 10 nm) [2, 4]. The productive processes are electron transfer reactions between electrons or holes and organic or inorganic species adsorbed on the SC surface or close to it. Electron acceptors (A) can be reduced by e_CB^–, and donor species (D) can give electrons to holes, being oxidized. The probability of these transfer processes depend on the relative redox positions of the CB and VB in the SC in relation with the redox potential of the species to be transformed and of the rate of competitive processes. HP carried out under normal illumination (e.g., black light or xenon lamps) leads to redox processes that take place through monoelectronic step reactions (due to the low frequency of photon absorption).

TiO₂ is so far the most useful semiconductor material for photocatalytic purposes because of its exceptional optical and electronic properties, chemical stability, non-toxicity, and low cost. The energy bandgaps of the most used photocatalytic forms of TiO₂, anatase and rutile, are 3.23 eV and 3.02 eV, corresponding to 384 and 411 nm, respectively [10]. The most popular commercial TiO₂ material for HP was originally produced by the German company Degussa (now Evonik) with the denomination P-25. Some other commercial products, such as Hombikat UV100, and different products of Cristal Global (PC50, PC100, PC500), Fluka, etc. have been tested in several photocatalytic systems [11, 12]. Although the data are divergent in the literature, the redox levels of h_VB⁺ and e_CB^– photogenerated by HP in P-25 taken in this article will be +2.9 and –0.3 V,^[1] respectively [13].

In addition to the direct attack of D species adsorbed on or close to TiO₂, holes can attack also adsorbed water molecules/surface hydroxyl groups, generating HO^·. These radicals are very oxidizing species with a reduction potential of +2.8 V [3]:

HP oxidative reactions are usually performed in the presence of molecular oxygen, with the objective of enhancing the mineralization of organic compounds; e_CB^– reduce O₂ with formation of hydroperoxyl or superoxide radicals (HO₂·/O₂·^–), in a rather slow but thermodynamically feasible electron transfer reaction, decreasing the probability of electron-hole recombination [2].

The value for the reduction potential for the hydroperoxyl/superoxide radical has been reported as E⁰(O₂/HO₂·) = –0.05 V [14]; the reduction potentials onto the TiO₂ surface are probably less negative. As the following set of equations indicates, this cathodic pathway is an additional source of hydroxyl radicals:

When a metal ion is present in a HP aqueous system solution, it can undergo reduction or oxidation reactions. Figure 1 shows the reduction potential of different couples together with the energy levels for the e_CB^– and the h_VB⁺; the comparison of the redox levels leads to a first approximation of the thermodynamic feasibility of transformation of the species. All couples with redox potentials more positive than the e_CB^– level can be directly reduced and all couples with redox potentials less positive than the h_VB⁺ level can be, in theory, oxidized.

Fig. 1

Position of the reduction potentials of various metallic couples related to the energy levels of the CB and VA of P-25 adapted from reference [8].

Figure 1 shows global (multielectronic) reactions, and these can be favorable (exergonic) in principle. However, if one-electron steps are considered, only those thermodynamically allowed reactions would occur. Accordingly, the ion could be reduced by e_CB^– in a direct reduction step, a reaction that requires the e_CB^– reduction potential to be more negative than the corresponding to the Mⁿ⁺/M^(n–1)+ pair:

Alternatively, oxidation of the metal ion can occur by reaction with holes or HO·, reaching a higher oxidation state:

Some metallic species (such as Cr(VI), Hg(II) or U(VI)) cannot be transformed to a higher oxidation state, but they can be directly reduced by e_CB^– [8]. This process can be accelerated and even produced in thermodynamically unfeasible conditions, i.e., when the redox potential of the couple to be reduced is more negative than the level of e_CB^–, by the addition of sacrificial donors to the solution. This donor mediated reduction of the metal or metalloid constitutes an indirect reduction process. The donors can be separated in two different groups: direct h_VB⁺ acceptors (as in eq. (3)), e.g., EDTA, citric acid, salicylic acid, or indirect acceptors via HO· reaction, such as methanol, acetic acid or formaldehyde [e.g., eq. (11) for an alcohol].

The reaction with h_VB⁺ or HO· reduces the electron-hole recombination probability and generates very strong reducing radicals which allow formation of reducing species that are not likely to be reduced directly by e_CB^–. A synergetic effect exists generally between oxidative and reductive processes, especially in the presence of organic compounds that can undergo oxidation [3, 6, 7].

Photocatalytic based treatments can take advantage of these transformations in order to obtain a metallic solid form deposited on the photocatalyst surface; in this way, the species is removed from the solution. This procedure could also be applied for metal recovery (copper, platinum, silver, gold). Another possibility is the formation of a less toxic soluble species [Cr(VI) to Cr(III), As(III) to As(V)], which are also easier to be removed by a ulterior treatment.

Chromium

Chromium is a metal with multiple industrial and technological applications including metallurgy, electroplating, textile industry, leather tanning and wood preservation. As a consequence of its uses, Cr is frequently found in wastewaters, mainly in the Cr(III) and Cr(VI) forms. Cr(VI) presents the highest environmental threat due to its toxicity for biological organisms together with its high solubility and mobility. The World Health Organization (WHO) [15] has established the maximum contaminant level of Cr(VI) in drinking water to be 0.05 mg L^–1. Cr(III) is considered non toxic or of very low toxicity [16], and its mobility is lower than that of Cr(VI). The conventional treatment for removal of Cr(VI) dissolved in water involves its reduction to Cr(III) with the use of sodium thiosulfate, ferrous sulfate, sodium metabisulfite, sulfur dioxide or other chemicals, with the subsequent economical costs and generation of residues.

Figure 2 shows the redox potentials for chromium in acid conditions taken from reference [17].

Fig. 2

Latimer diagram connecting the different Cr species [17].

Cr(VI) reduction is enhanced in acid solutions, and the net reaction can be written as:

At neutral pH:

Although the process is thermodynamically feasible, Cr₂O₇^2– is stable at room temperature in acidic solutions due to the high overpotential of the conjugate reaction of water oxidation. Therefore, strategies for improving the reduction reaction should be implemented.

Numerous works have been published on the photocatalytic reduction of Cr(VI) employing TiO₂, Fe/TiO₂, Pt/TiO₂, and other SCs such as ZnO, CdS, ZnS and WO₃ and many examples have already been described in our previous reviews [6–9] including reactors for technological applications, microparticles used in slurries or conveniently supported, nanoparticles and nanotubes. The major conclusions that emerge from the past publications can be summarized in the following points:

• Cr(VI) HP reduction depends upon pH. Low pH favors the net reaction [eq. (13)]; however, neutral or alkaline conditions favors precipitation and immobilization of Cr(III) as the oxides or hydroxides, facilitating a further separation;

• Cr(VI) reduction can be accelerated by addition of organic compounds acting as hole or HO· scavengers;

• molecular oxygen does not affect Cr(VI) HP reduction, at least at acid pH, as will be explained in detail later.

From a thermodynamic point of view, e_CB^– have the appropriate potential to directly reduce Cr(VI), Cr(V) and Cr(IV) (see Figs 1 and 2), and the HP mechanism of Cr(VI) reduction to Cr(III) is proposed to occur through three monoelectronic steps [18–20]:

This was proven by our group through electron paramagnetic resonance spectroscopic studies (EPR), which allowed the identification of paramagnetic Cr(V) species [18–20].

Detrimental reoxidation of reduced Cr species by holes or hydroxyl radicals is however possible:

An enhancement of the reaction takes place if convenient electron donors are present, either by reducing the probability of recombination or through the indirect reduction driven by radicals formed by hole/HO· attack [eqs. (3) and (11)], which, at high concentrations, hinder reaction and contribute to the reduction of chromium species.

The role of O₂ in photocatalytic Cr(VI) systems is noteworthy due to its unique behavior compared with other metal cations, because it appears to be independent of the presence of O₂, at least at acid pH. This process was explained by a very strong association between Cr(VI) and TiO₂ through the formation of a charge-transfer complex, identified by an absorption band at 380 nm [21]. Due to the fast capture of electrons by Cr(VI) as a result of the formation of this complex, no competition of O₂ by e_CB^– [eq. (5)] takes place, and consequently, no inhibition by O₂ was observed neither in the absence of organic donors nor in the presence of EDTA, oxalic or citric acid [18–20, 22, 23]. Experiments on platinized photocatalysts supported this independency, the same temporal evolution of Cr(VI) concentration being obtained using Pt/TiO₂ or bare TiO₂ either in the presence or in the absence of O₂ [21–23]. As Pt decreases the overpotential for electron transfer from the CB to O₂, the presence of O₂ when using Pt/TiO₂ should have affected the Cr(VI) decay in two opposite ways: a) either enhancing the reaction rate, with O₂·^– being a mediator for Cr(VI) reduction or b) decreasing the rate by competition with e_CB–.

Cr(VI) photocatalytic reduction continues to be a very rich and interesting study system, and extensive work has been done in the latest years, either with the aim of water treatment or as a practical system for the evaluation of the photocatalytic activity of several semiconductor samples in aqueous phase. For example, the behavior of different commercial photocatalysts in the Cr(VI) heterogeneous photocatalytic system in the presence of EDTA was analyzed, in particular, the influence of the type of photocatalyst on the yield, the progressive inactivation of the photocatalyst during the reduction and the effect of adsorbed species. In this work, studies by TRMC (time resolved microwave conductivity) were performed, being the first time that these studies have been applied to a HP reaction of a metal system [24]. The decrease of the TRMC signal of the electrons in the presence of Cr(VI) reinforced the theory that the Cr(VI) reduction proceeds through the e_BC^- of TiO₂. A further work using P-25 [25] indicates that TiO₂ is inactivated after the photocatalysis of Cr(VI) in the absence of EDTA due to deposition of Cr(III) species on the photocatalyst surface, but the organic compound preserves the activity of the photocatalyst forming soluble Cr(III) complexes; these results were also confirmed by TRMC. Another paper of our group evaluated the activity of anatase TiO₂ films deposited by cathodic arc (CA) films on glass substrates for Cr(VI) reduction in the presence of EDTA and compared the new films with P-25 films obtained by dip-coating immersion (DC films). Despite of DC films showing higher photoactivity, its adhesion properties were very poor in contrast with CA films, making the latter more promising materials for photocatalytic applications that require immobilized catalysts [26].

Photocatalytic Cr(VI) reduction is also possible under visible irradiation and, in our previous review [9], several examples were described. Dye photosensitized TiO₂ samples were proved to be able to reduce Cr(VI) under visible light. For example, Di Iorio et al. [21] studied alizarin red chelated to TiO₂, and the results indicate a high efficiency for Cr(VI) reduction, almost independent of the photon flux and of the irradiation wavelength, and slightly dependent on the Cr(VI) concentration in the explored range (40–200 μM), with similar results in air or nitrogen atmosphere. EPR experiments confirmed Cr(V) formation, pointing out to three one-electron consecutive steps for Cr(VI) reduction. Another work [27] using TiO₂ coated with hydroxoaluminiumtricarboxymonoamide phthalocyanine (AlTCPc) irradiated under visible-light in the presence of 4-chlorophenol (4-CP) as sacrificial donor showed a very rapid Cr(VI) reduction to Cr(III). In the presence of dyes, the photocatalytic mechanism is different from that taking place under UV light, according to the scheme shown in Fig. 3; after excitation of the dye, electron injection into the CB promotes Cr(VI) reduction:

Fig. 3

Visible light HP Cr(VI) reduction with TiO₂ modified with AlTCPc at pH 2 (from reference [9], with permission).

A very good protecting effect of 4-CP preventing AlTCPc photobleaching was observed, as reactive oxygen species (ROS, e.g., O₂·^–, HO₂·), which could oxidize the dye, cannot be formed due to the fast reaction of e_CB^– with Cr(VI). Further studies on photocurrent and absorption spectra of AlTCPc/TiO₂ films, and of absorption and fluorescence of the dye in solution in function of the dye concentration demonstrated that the photoactive species is the monomeric dye electronically coupled to TiO₂ [28].

The scientific community have been successful in the generation of knowledge about fundamental mechanisms, mechanistic pathways, role of O₂ and electron donors, etc., related with the very rich reductive photocatalytic Cr(VI) system. More studies and efforts should be made to transfer the knowledge on this promising technology into practical applications.

Arsenic

Arsenic contamination in water can be anthropogenic (mining, use of biocides, wood preservers), but it mainly comes from natural sources due to dissolution of minerals in surface or groundwaters, or to volcanic processes [29, 30]. Chronic ingestion of arsenic for prolonged periods of time results in arsenicosis, a hydric disease, which causes severe skin lesions including pigmentation changes, palmoplantar keratosis and other syndromes, ending generally in cancer. Arsenic in drinking water constitutes a serious problem, affecting several million people all over the world. The World Health Organization (WHO) [31] recommends 10 μg L^–1 as the maximum allowable As concentration in drinking water, a value which is taken by most national regulatory agencies.

Conventional water treatments used to remove As are oxidation/coagulation/adsorption processes on iron or aluminum hydroxides, ionic interchange, activated alumina, lime softening, and reverse electrodialysis and osmosis. Transformation to As(V) makes easier the application of conventional technologies, e.g. ion exchange and adsorption. However, new emerging techniques have to be investigated to offer low-cost solutions to the arsenic problem, especially for low-income populations, as mentioned in some references [29].

Photocatalysis of arsenical systems was most studied for the oxidative system, and mechanisms for As(III) oxidation were reported in studies that spanned concentrations from the micro- to the millimolar range, showing in every case very fast oxidation in 10 – 100 min. Our previous reviews describe the latest advances [8, 9]. On the other hand, multielectronic reduction of As(V) to As(III) or As(0) is thermodynamically favored even with mild reductants, as can be inferred from the Latimer diagram (Fig. 4). Further reduction to arsine [As(–3)] is more difficult. All the reductive steps are less favored at higher pH values.

Fig. 4

Latimer diagram connecting the different As species [32].

Oxidative TiO₂ HP has been shown as a good alternative for As(III) oxidation. In an effort to explain the mechanism of oxidation, monoelectronic steps with formation of As(IV) that could involve either the reaction with HO· [eq. (19)], with VB holes [eq. (20)], or with O₂·^– [eq. (21)] have been suggested:

The reduction potential of the As(IV)/As(III) couple has been reported to be E⁰ ≈ +2.4 V [33]. Therefore, formation of As(IV) by attack of holes, HO· or HO₂·^/O₂· is thermodynamically possible. Regardless of the first oxidation step, As(IV) goes further to As(V), the stable form:

The mechanism of photocatalytic oxidation of As(III) has been a matter of discussion for many years, the controversy being centered on determining the major oxidant in the system, either O₂·^– HO· or h_VB⁺. However, whatever the oxidant is, it is not possible to deny the efficiency of the HP process to transform As(III) into As(V). The real problem is, however, how to remove later dissolved As(V).

Regarding low-cost applications, several works of our laboratory [29, 34–36] report results of an application to remove arsenic from natural waters of rural areas; to perform HP experiments, walls of PET plastic bottles were impregnated with TiO₂ by a very simple technique [37]. As(III) solutions ([As(III)]₀ = 1000 μg L^–1, pH 7.8) were placed in these bottles, which were put in a horizontal position and irradiated during 6 h by UV light (366 nm, 800 μW cm^–2). After irradiation, non-galvanized packing iron wire pieces (usually employed in agricultural applications) were added. After 24 h settlement in the dark, As removals ranging 80–86 % were obtained. The same bottle could be reused at least three times without loss of efficiency. The HP procedure was proposed to remove As (500–1800 μg L^–1, unidentified speciation) from well water samples of Las Hermanas (Santiago del Estero Province, Argentina). In this case, a FeCl₃ solution was added at the end of an experiment under solar irradiation, and more than 94 % As removal took place, attaining concentrations below the WHO limits.

The HP oxidation of As(III) in the presence of Hg(II) in equimolar concentration was studied by our group at pH 7. The results indicated that, in the absence of TiO₂, some oxidation of As(III) in air takes place in the dark, without changes by addition of TiO₂. Under UV irradiation in anoxic conditions in the presence of the photocatalyst, the conversion was much higher, with formation of a gray solid (Hg(0) + Hg₂Cl₂) while in the presence of O₂, HP oxidation was produced in a similar way in the presence or in the absence of Hg(II), indicating a competence between O₂ and Hg(II) [38].

In contrast, reductive photocatalytic systems for arsenic transformation are still scarce. Reduction of As(III) by TiO₂ e_CB^– is thermodynamically possible either in the presence or the absence of an electron donor in the aqueous system. However, in the case of As(V), the direct reduction seems to be not feasible because the reported reduction potential of the As(V)/As(IV) couple is highly negative (E⁰ ≈ –1.2 V [33]) and thus, the indirect pathway by using sacrificial electron donors is indispensable in this case. Our group [39, 40] studied the reaction starting from As(V) and also from As(III) in the presence of methanol; the As(III) reaction in the absence of the organic compound was also analyzed. A mechanism based on the formation of a hydroxymethyl radical produced by hole/HO· attack to methanol, was proposed. This radical can donate electrons to the CB or be itself effective as As(V) reductant, with formaldehyde generation in both cases:

CH₂O can be transformed to formic acid and finally mineralized to CO₂ with generation of CO₂^·– (E⁰_{(CO2/CO2·–)} ≈ – 2.0 V) [14], another strong reducing radical that can contribute to the reducing process:

In a further step, As(IV) is reduced by CB or trapped electrons, or by ·CH₂OH or CO₂·^– to As(III). Unlike to As(V), the direct photocatalytic As(III) reduction by e_CB^– in the absence of MeOH was also observed for the first time; this indicates that, although the value for the monoelectronic couple As(III)/As(II) is not known, it should be located below the CB level:

Therefore, after As(III) formation, monoelectronic consecutive steps would lead to the formation of stable products such as As(0) and AsH₃, which were unambiguously identified by the authors in all cases through XPS and XANES analyses of the solid residues formed onto the TiO₂ surface after the reaction. In this way, reduction to solid As(0) was achieved, which is a very convenient method for purposes of As removal from water. However, attention must be paid in this kind of reactions to formation of AsH₃, one of the most toxic forms of As. Strategies for AsH₃ treatment were proposed, like adsorption on suitable materials or oxidative treatment by a coupled gas phase photocatalytic system, what are currently under development in our research group.

In a recent paper, it was demonstrated that reduction of As(V), thermodynamically not possible by HP in normal conditions, can be produced in the dark by the action of accumulated electrons by UV irradiation of TiO₂ nanoparticles in alcohol. The accumulation of electrons and their participation in As(V) reduction was followed by UV-vis spectrophotometry and EPR, and detected as Ti(III) [41, 42].

Due to the extensive and good quality work done on photocatalytic systems for As removal and the importance of the problem of As in the environment, research on practical applications of these systems should be reinforced in the future in order to provide cost effective systems.