Role of Hydroxyl, Superoxide, and Nitrate Radicals on the Fate of Bromide Ions in Photocatalytic TiO2 Suspensions

The influence of bromide ions on systems containing highly reactive radical species is of great interest for environmental remediation, atmospheric chemistry and green synthesis of high added value compounds. In this regard, irradiated TiO2 suspensions are simple and suitable systems to highlight some mechanisms of general validity in the above mentioned research fields. In this work the spintrapping technique has allowed to highlight the concurrent action of hydroxyl and superoxide radicals towards bromide ions. In fact, hydroxyl radicals oxidize bromide ions to bromine atoms which in turn can be reduced back to bromide ions by superoxide radicals. Results suggest that this relais mechanism, mainly based on secondary redox products (such as reactive oxygen species, ROS) rather than on direct interfacial electron trasfer, is responsible for the generally observed constant concentration of bromide ions in irradiated TiO2 suspensions. The presence of nitrate ions in the system significantly alters this equilibrium. In fact, nitrate radicals generated by hole induced oxidation of nitrate ions, selectively oxidize bromide ions to elemental bromine even in anoxic conditions, but in the presence of bromate as an opportune electron scavenger. This result not only reveals the importance of nitrate radicals towards the photocatalytic production of elemental bromine but confirms the active role of nitrate ions in irradiated photocatalytic suspensions.


Introduction
Bromide ion is a ubiquitous species in natural aqueous systems. Its concentration is quite low in rain water, ranging from 0 to 110 μg•L -1 [1], but reaches considerable values in groundwater, between 0.01 and 2 mg•L -1 [2], and in seawater, typically about 70 mg•L -1 . Moreover, in specific areas such as the Dead Sea, or the Salt lakes in USA, bromide concentration is in the order of several grams per liter. Coastal or estuarine areas where seawater could come in contact with surface water usually present higher bromide concentration with respect to inland areas [3], therefore, bromide ions concentration has been often considered as an indicator of seawater intrusions in soil water [4].
Bromide is present in almost every drinking water at concentrations ranging from ca. 0.01 up to 3 mg L -1 [5]. Notably, the presence of bromide ions per se does not represent a threat to living organisms.
However, bromide ions are easily involved in oxidation reactions of natural or anthropogenic origin, giving rise to not negligible side effects for environmental and human safety.
As far as natural oxidations are concerned, bromide ions play a key role in sea water and in the atmophere, where they are transported in seawater droplets which constitute the most important aerosol type in the atmosphere. In fact, in the troposphere seawater particles accounts for ca. 27 Tg on a dry mass basis [6]. These particles, not only are efficient centers of cloud condensation [7], but are important sites for SO2 oxidation to sulphuric acid [8,9], and constitute a source of reactive Br2 and BrCl species. The latter ones, in turn, trigger oxidation reactions involving ozone and hydrocarbons [10]. As a matter of fact, the depletion of tropospheric ozone largely depends on reactions involving halogen species [11]. In this regard, the interaction of bromine species with reactive oxygen species (ROS) such as hydroxyl ( • OH) or superoxide radicals (HO2 • ) is of paramount importance. These radicals are also known as "atmospheric cleansers" because they are mainly responsible for the self-cleaning ability of the troposphere during day time [12]. ROS can be also generated within dispersed seawater particles via direct photolysis of light absorbing species (nitrate, nitrite, hydrogen peroxide, ozone) or redox reactions induced by excited chromophores, involving water and oxygen molecules [13]. Therefore, the interplay between ROS and bromide species is a "hot topic" of atmospheric chemistry. Similarly, bromide ion is usually considered the main • OH scavenger in seawater affording bromine radical (Br • ), which in turn, mainly reacts with another bromide ion giving rise to dibromide radical, Br2 −• [14]. Bromine centered radicals are able to react with dissolved organic compounds producing bromoderivatives which are usually persistent and toxic pollutants [15].
As far as the human activities are concerned, it has been evidenced that ozonation of bromide containing water, besides the desired effects of water remediation and pollutant degradation, quantitatively induces oxidation of bromide to bromate ion. The latter has been classified by the International Agency for the Research on Cancer (IARC) as a potentially carcinogenic species, and the World Health Organization (WHO) [16] guidelines reported that a concentration of bromate of 3 μg•L -1 is associated with an upper-bound excess lifetime cancer risk of 10 -5 . By considering that the bromide ion concentration in drinking water ranges from less than 0.01 to nearly 3 mg•L -1 [17], it is necessary to avoid or control the formation of bromate ions, whose accumulation can be extremely dangerous in particular in closed recirculated systems [18]. TiO2 photocatalysis has been proposed as a tool to control bromate formation when coupled with ozonation [19][20][21]. In fact, under irradiation of suitable energy, photogenerated electrons are able to reduce bromate ions, while avoiding further oxidation of bromide ions.
Recently, production of elemental bromine through oxidation of bromide ions has been reported in irradiated aqueous TiO2 suspensions containing catalytic amounts of nitric acid [22,23]. This process, driven by light irradiation and without using strong oxidants, is a green alternative to the currently applied industrial ones which instead employ toxic and harmful gaseous chlorine as the oxidant.
Hydroxyl radicals can be also formed by hole-induced water oxidation (Eq. 6). In the presence of nitrate ions, however, a further reaction can occur that is, as recently proposed [25], the formation of nitrate radicals (Eq. 7).
H2O + h + VB → [H2O + ] → • OH + H + NO3 -+ h + VB (or H2O + ) → NO3 • (+ H2O) Therefore, it is evident that photocatalytic water suspensions are suitable platforms to evidence and simulate the complex chemistry occurring in nature involving bromide ions and oxygen active species, which can be thereby quantitatively produced and easily detected.
In this paper the spin-trapping technique (based on the detection of stable adducts of OH radicals by Electron Paramagnetic Resonance, EPR) and UV-vis spectroscopy have been employed to investigate the evolution as a function of time of hydroxyl and superoxide radicals, respectively, in irradiated suspensions of TiO2. This was done both in the presence and in the absence of bromide and nitrate ions and at different pH values. Understanding the mechanisms of the reactions occurring in the irradiated system has a twofold value since it allows to shed light into a delicate process of environmental chemistry and, at the same time, to gain further evidence on a possible, more benigne, alternative route for bromine production. Commercial titanium dioxide (TiO2 P25-Evonik, 20% Rutile and 80% Anatase, specific surface area: 50 m 2 •g -1 ) was used as the photocatalyst.

5,5-
Electron Paramagnetic Resonance (EPR) spectra were run using a X-band CW-EPR Bruker EMX spectrometer equipped with a cylindrical cavity operating at 100 kHz field modulation.
UV-vis spectra were recorded by means of a UV-vis spectrophotometer (Uvikon, Kontron Instruments). Spectra were acquired in the range 200-600 nm by using suprasil quartz cells.
"In situ" irradiation experiments for the hydroxyl radical detection were carried out directly in the EPR cavity by using a 1600W Hg lamp (Newport Instruments) with a light beam output set at 800W and equipped with a IR water filter, while "ex situ" irradiation experiments to monitor the formation of superoxide radicals were performed irradiating by means of a 500W Xe/Hg lamp (Oriel Instruments) equipped with a IR water filter.
In each experiment the light beam also passed through a filter (Oriel 320FG01-50S) aimed to exclude the deep UV components (λ < 320 nm) in order to avoid the onset of a radical chemistry caused by direct interaction between the DMPO molecule and the UV light [26]. The incident irradiation for the UV component was ca. 350 W•m -2 and 250 W•m -2 for the "in situ" and "ex situ" experiments, respectively. The incident light irradiance in the UVA region for each experiment was measured by a Deltahom instrument equipped with a detector operating in the in the UV-A range (315-400 nm).
Radical species were generated via UV irradiation of four sets of TiO2 aqueous suspensions (1 g•L -1 ) at different pH values. The first set of suspensions, characterized by the presence of nitrate ions (4 mM) at pH 2.5, 4.0, and 6.5, was prepared by mixing the appropriate amount of HNO3 and KNO3. A second group of solutions containing sulphate ions (4mM) was prepared by mixing the appropriate amount of H2SO4 and K2SO4, in order to afford a sulphate concentration equal to that of nitrate in the first set of suspensions by maintaining the same pH values. Other two sets of suspensions were prepared by adding potassium bromide (2 mM) to the above mentioned ones. In this way it was possible to evidence separately the effect of nitrate and bromide ions. Table 1 reports the various suspensions prepared and the related label. In a typical experiment 3 ml of a TiO2 suspension containing DMPO (8.8 mM) was stirred in the dark for 5 minutes. A portion of this suspension (50 μL) was then transferred into the spectrometer cavity and the EPR spectra were recorded, at regular time intervals, during "in situ" irradiation and after turning off the irradiation source. The signal intensity of the DMPO-OH adduct has been obtained via double integration of the corresponding EPR signal. The amount of trapped radicals was evaluated by spin counting comparing the intensities of the various signals with those of 2,2-diphenyl-1picrylhydrazyl radical (DPPH) standard solutions in toluene.
Superoxide radicals were detected via UV-vis spectroscopy according to a method reported by Goto et al. [28]. In a typical experiment 50.0 ml of the TiO2 suspension containing NBT (20 μM) was stirred in the dark for 15 minutes in order to establish the adsorption-desorption equilibrium. A fraction of this suspension was filtered with an INCOFAR nylon syringe filter (diameter 13mm, pore diameter 0.2 μm) and the initial NBT concentration retrieved by UV-vis spectroscopy. Thereafter, the suspension was irradiated "ex situ" for 15 minutes and a fraction was filtered and immediately analyzed by means of UV-vis spectroscopy. In order to exclude undesired radical reactions under dark, the remaining suspension was stirred in the absence of irradiation for further 15 minutes, filtered, and finally analyzed by UV-vis spectroscopy.
This method is based on Eqs. 9-10.
2O2 •-+ 2HO2 • + NBT 2+ → NBTH2 + 4O2 (10) in which the water soluble yellow NBT reacts with superoxide radicals (with both the protonated and unprotonated form) to produce NBTH2 (formazan), an insoluble purple compound. Therefore, the decrease of the absorption band of NBT, is proportional to the amount of superoxide radicals photocatalytically produced during irradiation of the TiO2 suspension.
Photocatalytic generation of bromine was performed by using an experimental set up elsewhere reported [22,23] and here briefly described. 40°C by means of a heating wire surrounding the outer wall of the reactor, in order to facilitate the stripping of the bromine produced. Notably, the Pyrex jacket surrounding the lamp cut off radiation wavelengths lower than 320 nm, so to exclude direct photolysis of HNO3 and to consider TiO2 the only relevant light absorbing species.
Prior to irradiation, the suspension was maintained in the dark and under nitrogen bubbling for 30 minutes, in order to ensure equilibrium conditions when irradiation started.
The gas flowing outside the photoreactor passed through one ice cold trap containing 50 mL of carbonium tetrachloride (CCl4 99.9%, Sigma-Aldrich). The bromine concentration in the trap was measured at fixed time intervals by means of UV-vis spectroscopy, by comparing the intensity of the characteristic bromine absorption band at 415 nm with standard bromine CCl4 solutions. Runs were repeated at least three times and the standard deviation was always lower than 5%, thus indicating good reproducibility of results.

The role of hydroxyl radicals
Irradiation of the TiO2 water suspensions in presence of the DMPO spin trap leads, according to Eq.   [29] and to the reported analysis of OH radical photocatalytic generation that also decreases varying pH from 7 to 3 [30].
The trend of the hydroxyl radical concentration shown in Figure 1 is the result of the interplay between generation and consumption phenomena. In particular, the blue curves, obtained in the absence of bromide ions, express the photocatalytic generation process of OH radicals. On the other hand, the red curves, obtained in the presence of bromide ions, represent the sum of photocatalytic OH generation and of bromide induced OH consumption. Therefore, assuming that the rate of OH photogeneration is the same in the presence and in the absence of bromide ions, the difference between the two curves reported in each quadrant of Figure 1 expresses the trend of OH radical consumption due to the presence of bromide ions. This procedure leads to the curves shown in Figure   2 which express the bromide induced consumption of OH radicals. However, the results shown in Figure 2 suggest a kinetics of OH radical consumption of the first order with respect to OH • , i.e. a direct oxidation of bromide ions by OH radicals according to Eq. 12-13 [31,32].
These results suggest that bromide ions act mainly as OH radical scavengers rather than as traps for the photogenerated hole. Notably, both the paths have been reported in the relevant literature, but bromide ions have been often used as a direct hole scavenger [33]. Therefore, it can be suggested to use more suitable hole scavengers such as melamine [34] for mechanistic studies, instead of bromide ions.
By substituting sulfate with nitrate ions under otherwise identical conditions, the EPR trapping signal under irradiation presents a completely different behaviour as shown in Figure   The reason of the behaviour observed for OH evolution in the absence of bromide ions (Figure 2, blue lines) has been proposed in a previous work [25]. Briefly, the observed decrease in the amount of photogenerated OH radicals is due to hole induced oxidation of nitrate ions to nitrate radicals (not trapped by DMPO) which efficiently competes with the formation of OH radicals by water oxidation.
This competition seems to take place efficiently after the first 50 seconds of irradiation during which, on the contrary, formation of OH radicals prevails. It can be speculated that, due to the large excess of water, that water molecules constituting the first adsorption layers are firstly oxidized, but then facilitate in a concerted way the oxidation of nitrate ions. In fact, as above mentioned below the zeropoint charge of TiO2, the density of nitrate ions close to the positively charged surface could be significantly high. This hypothesis is supported by an analogous interpretation adavanced on the basis of results on surface hopping and DFT calculations [35], that recently highlighted the importance of hydrogen bonding between adsorbed water molecules in stabilizing the hydrated hole (H2O + ). This effect could prolong the lifetime of H2O + which therefore, in the presence of nitrate ions, can promote formation of nitrate radicals (Eq. 7) instead of evolving into hydroxyl radicals. Figure 4 schematically describes this mechanism.

Ti O Ti
Ti O Ti The stabilization of the hydrated hole by hydrogen bonding has been also experimentally observed [36][37][38][39]. Scan diffuse-reflectance infrared Fourier transform spectroscopy (DRIFTS) evidenced the same effect on anatase TiO2 (101) with a consequent suppression of charges recombination at the time scale of minutes [38]. The number of water adsorbed layer influences this process as highlighted by means of infrared transient and steady-state absorption spectra of anatase TiO2 nanoparticles [39].

Ti O Ti
In fact, the interaction between first-layer water and TiO2 is lowered by the further water layers, thus reducing the hole trapping capability. As a consequence, the mobility of water onto the surface of TiO2 plays a key role in this process [40,41].
Furthermore, it must be noted that the one electron oxidation of nitrate ions (E 0 NO3•/NO3-= 2.4 V vs NHE) is thermodynamically favoured with respect to one electron water oxidation (E 0 •OH/H2O = 2.73 V vs NHE).
However, in the presence of nitrate, bromide ions did not significantly affect the EPR signal, especially at pH 6.5 and 4.0. This is possibly due to the fact that the formed nitrate radicals act as efficient scavengers of bromide ions [42] so that the presence of bromide ions does not significantly influence the primary electron transfer steps.

The role of superoxide radicals
Superoxide radicals exist in the protonated (HO2 • ) or in the deprotonated (O2 •-) form depending on the pH of the irradiated TiO2 aqueous suspension according to Eq. 14.
The pKa of this equilibrium is equal to 4.8. [43], i.e. superoxide radicals mainly exist in the protonated form at pH 2.5 and 4.0, and in the deprotonated form at pH 6.5.  Bromide ions can be oxidized by hydroxyl radicals as above mentioned according to Eqs. 12-13. Literature reports that in sea water the bromine atom thereby generated reacts with a bromide ion producing the Br2 •anion radical according to Eq. 15.
This species is considered the key intermediate for the production of elemental bromine in the atmosphere. Its chemistry with superoxide radicals has been widely investigated [44]. Briefly, Br2 •affords elemental bromine by reaction with another Br2 •radical (Eq. 16) or with HO2 • (Eq. 17).
The same authors report an apparent kinetic constant of 4.4 • 10 9 M -1 s -1 which suggests as highly probable the occurrence of Eq. 17. On the other hand, reaction of Br2 •with the superoxide radical anion (O2 •-) quenches the Br2 •radical to bromide ion according to Eq. 18 [45].
This reaction, contrarily, is reported to be of minor importance with respect to Eq. 17.
However, according to the above mentioned results, only slight differences could be observed between the protonated and the not protonated superoxide radicals in the TiO2 suspensions containing bromide ions. Moreover, at the surface of irradiated TiO2, due to the dual nature of the photogenerated excitons, generally oxygen reduction takes place simultaneously with oxidation processes. Therefore, it is more probable that superoxide radicals react with bromine atoms faster than with secondary species such as Br2 •-. Even if the presence of Br2 •could not be assessed in the present conditions, it must be considered that according to Eq. 17 and its high kinetic constant, in a photocatalytic suspension at acidic pH (in the absence of nitrate ions) considerable amount of elemental bromine should be produced, while, on the contrary, it has been experimentally proved [22] that only negligible amount of elemental bromine could be photocatalytically produced. Finally, the role of Br2 •in these systems could have been overestimated. In fact, BrOH •intermediate (Eq. 12), i.e. the primary product of the OH radical induced oxidation of bromide ions, possesses virtually the same UV-vis absorption spectrum as Br2 •-, as recently reported by Lampre et al. [32].
For these reasons, according to our results Eq. 19-20 seems to be more plausible in photocatalytic suspension: Br • + HO2 • → Br -+ H + + O2 (20) This mechanism allows to explain why, in the absence of nitrate ions, the concentration of bromide ions in oxygenated aqueous TiO2 suspensions under irradiation, does not macroscopically change. In fact, bromide ions can be firstly oxidized mainly by hydroxyl radicals giving rise to bromine atoms (Eq. 12-13) which, in turn, can be reduced back to bromide ions by superoxide radicals. This relais mechanism is schematically depicted in Figure 6.  The above presented results on the role of hydroxyl and superoxide radicals allow to clarify some mechanistic aspects related to the photocatalytic production of elemental bromine recently reported [22,23] and briefly below summarized for the benefit of the reader.

Consequences on the photocatalytic production of elemental bromine
Production of elemental bromine has been observed when a TiO2 (P25-Evonik) aqueous suspension containing bromide ions and catalytic amounts of nitrate ions at acidic pH, was irradiated under UV light in the presence of O2. In the absence of nitrate ions and/or of oxygen, only negligible amount of bromine could be produced. Therefore, it has been hypothesized that oxygen reduction and nitrate oxidation to nitrate radical were the essential primary steps for the production of Br2. However, by considering the above mentioned relais mechanism promoted by reactive oxygen species involving bromide ions, it is possible to hypothesize that oxygen behaves as a simple electron scavenger and the reaction could also proceed when substituting oxygen with a species such as bromate, similarly easy to be reduced. Notably, bromate ions are reported to compete with oxygen for the photogenerated electron in photocatalytic suspensions at acidic pH values [20] according to the following Eq. 21.
BrO3 − + 6H + + 6 eCB − → Br − + 3 H2O To this aim we substituted oxygen with bromate ions (4 mM In the absence of bromate no elemental bromine production could be observed, in agreement to what previously reported [22]. Similarly, no bromine was produced under dark conditions (Figure 7).
However, bromine production started under irradiation when bromate ions were present in the deaerated suspension. According to these results it is possible to propose the mechanism depicted in Figure 8 for the formation of elemental bromine. Bromate ions act as electron scavengers, while holes mainly oxidize nitrate ions to nitrate radicals which, in turn, oxidize bromide ions to elemental bromine. OH radical formation induced by water oxidation is also represented in Figure 8, even if of minor importance in the presence of nitrate ions, as described throughout the text.
Elemental bromine formation in deaerated TiO2 suspensions under UV light irradiation and in the presence of nitrate indirectly supports the hypotheses hereby proposed in this work. Further investigation is ongoing because of the technological consequences of this preliminary result. In fact, bromide-bromate mixtures are side products of the industrial ''cold process'' of bromine production, and green technologies exploiting these mixtures are highly desirable [46].

Conclusions
The joint use of EPR and UV-vis spectroscopy allowed to highlight the role of hydroxyl, superoxide and nitrate radicals in TiO2 aqueous suspensions containing bromide ions. In particular, it is proposed that hydroxyl radicals rather than photogenerated holes oxidize bromide ions to bromine atoms. On the other hand, superoxide radicals are able to reduce back bromine atoms to bromide ions so that the interplay between these reactive oxygen species does not macroscopically affect the observed concentration of bromide ions in irradiated TiO2 aqueous suspensions in the presence of oxygen.
These findings shine some light on relevant mechanisms in the field of atmospheric chemistry and environmental remediation. Moreover, on the basis of the above mentioned results it is evidenced that the photocatalytic synthesis of elemental bromine relies on the key role of nitrate radicals, while oxygen mainly acts as electron scavenger. The possibility of substituting oxygen with bromate ions in this reaction is proposed as a promising tool to endow with "green" features and to improve the efficiency of the industrial "cold process" for bromine production.