Degradation of Methyl 2-Aminobenzoate (Methyl Anthranilate) by H2O2/UV: Effect of Inorganic Anions and Derived Radicals

This study shows that methyl 2-aminobenzoate (also known as methyl anthranilate, hereafter MA) undergoes direct photolysis under UVC and UVB irradiation and that its photodegradation is further accelerated in the presence of H2O2. Hydrogen peroxide acts as a source of hydroxyl radicals (·OH) under photochemical conditions and yields MA hydroxyderivatives. The trend of MA photodegradation rate vs. H2O2 concentration reaches a plateau because of the combined effects of H2O2 absorption saturation and ·OH scavenging by H2O2. The addition of chloride ions causes scavenging of ·OH, yielding Cl2·− as the most likely reactive species, and it increases the MA photodegradation rate at high H2O2 concentration values. The reaction between Cl2·− and MA, which has second-order rate constant kCl2•−+MA = (4.0 ± 0.3) × 108 M−1·s−1 (determined by laser flash photolysis), appears to be more selective than the ·OH process in the presence of H2O2, because Cl2·− undergoes more limited scavenging by H2O2 compared to ·OH. While the addition of carbonate causes ·OH scavenging to produce CO3·− (kCO3•−+MA = (3.1 ± 0.2) × 108 M−1·s−1), carbonate considerably inhibits the photodegradation of MA. A possible explanation is that the elevated pH values of the carbonate solutions make H2O2 to partially occur as HO2−, which reacts very quickly with either ·OH or CO3·− to produce O2·−. The superoxide anion could reduce partially oxidised MA back to the initial substrate, with consequent inhibition of MA photodegradation. Fast MA photodegradation is also observed in the presence of persulphate/UV, which yields SO4·− that reacts effectively with MA (kSO4•−+MA = (5.6 ± 0.4) × 109 M−1·s−1). Irradiated H2O2 is effective in photodegrading MA, but the resulting MA hydroxyderivatives are predicted to be about as toxic as the parent compound for aquatic organisms (most notably, fish and crustaceans).


Introduction
Methyl 2-aminobenzoate (MA, C 8 H 9 NO 2 ) is a clear liquid that occurs in many essential oils. It has a melting point of 24 • C, a boiling point of 256 • C, and a density of 1.17 g·mL −1 [1]. MA can be found in Concord grapes, jasmine, bergamot, lemon, orange, and strawberries, and it is used as as well as to assess the effect on the process of common inorganic anions such as chloride and carbonate. To better assess the effect of the added anions, the reactivity of CO 3 · − and Cl 2 · − with MA was studied by using the nanosecond laser flash photolysis technique. Because MA is not harmless to aquatic environments, this study investigates the following: (i) whether and to what extent MA could be photodegraded under AOP conditions, also in the presence of inorganic anions such as chloride and carbonate; and (ii) the potential of MA photodegradation to produce intermediates that might have higher impact than the parent compound, and that could be formed during the AOP removal of MA and/or other contaminants.

MA Photodegradation by UV and H 2 O 2 /UV
The photoinduced degradation of 0.1 mM MA was first studied under UVC irradiation alone (lamp maximum emission at 254 nm) and under UVC irradiation in the presence of different concentration values of H 2 O 2 (see Figure 1 for the absorption spectra of MA and H 2 O 2 ). The MA time evolution under these conditions is reported in Figure 2A, while Figure 2B reports the trend of the photodegradation rate of MA (R MA ) as a function of the H 2 O 2 concentration. Table 1   by using the nanosecond laser flash photolysis technique. Because MA is not harmless to aquatic environments, this study investigates the following: (i) whether and to what extent MA could be photodegraded under AOP conditions, also in the presence of inorganic anions such as chloride and carbonate; and (ii) the potential of MA photodegradation to produce intermediates that might have higher impact than the parent compound, and that could be formed during the AOP removal of MA and/or other contaminants.

MA Photodegradation by UV and H2O2/UV
The photoinduced degradation of 0.1 mM MA was first studied under UVC irradiation alone (lamp maximum emission at 254 nm) and under UVC irradiation in the presence of different concentration values of H2O2 (see Figure 1 for the absorption spectra of MA and H2O2). The MA time evolution under these conditions is reported in Figure 2A, while Figure 2B reports the trend of the photodegradation rate of MA (RMA) as a function of the H2O2 concentration. Table 1      The error bars shown in panel (B) represent the uncertainty associated to the calculation of the photodegradation rates by fitting the MA time trend data with a pseudo-first order kinetic model (intra-series variability). In several cases the error bars were smaller than the data points. The reproducibility between experimental replicas (inter-series variability) was in the range of 15-20%.
Some MA photodegradation with a half-life time of approximately 10 min took place in the absence of H2O2, due to MA direct photolysis. The direct photolysis quantum yield of MA was calculated as follows [25]: is the incident spectral photon flux density of the lamp and Φ as the irradiation wavelength increased. Therefore, when applying artificial irradiation, the UVC spectral range and in particular the radiation at 254 nm (very near the UVC absorption maximum of MA, see Figure 1) appears to be the most suitable option to induce MA direct photolysis. The pH values of the studied systems were~neutral, with the exception of the systems containing Na 2 CO 3 . The error bars shown in panel (B) represent the uncertainty associated to the calculation of the photodegradation rates by fitting the MA time trend data with a pseudo-first order kinetic model (intra-series variability). In several cases the error bars were smaller than the data points. The reproducibility between experimental replicas (inter-series variability) was in the range of 15-20%. Some MA photodegradation with a half-life time of approximately 10 min took place in the absence of H 2 O 2 , due to MA direct photolysis. The direct photolysis quantum yield of MA was calculated as follows [25]: is the most relevant to our steady irradiation experiments, and a decrease was observed in the values of Φ MA as the irradiation wavelength increased. Therefore, when applying artificial irradiation, the UVC spectral range and in particular the radiation at 254 nm (very near the UVC absorption maximum of MA, see Figure 1) appears to be the most suitable option to induce MA direct photolysis.  Figure 2B, and in principle it might be accounted for by two different phenomena: (i) saturation of H 2 O 2 absorption with increasing H 2 O 2 concentration; and (ii) offset between photoinduced ·OH generation, and ·OH scavenging by H 2 O 2 itself. The first effect depends on the absorbance of H 2 O 2 . Considering ε H 2 O 2 ,254nm~1 5 L·mol −1 ·cm −1 and assuming b = 2 cm as the optical path length inside the irradiated solutions, the absorbance of the studied H 2 O 2 solutions was approximately 0.15 (5 mM H 2 O 2 ), 0.3 (10 mM), and 0.6 (20 mM). The absorbance of 0.1 mM MA at 254 nm is A MA,254nm~0 .2, and the fraction of radiation absorbed by H 2 O 2 in the irradiated systems can be calculated as follows [25]: ). The photogenerated ·OH can react with either MA or H 2 O 2 , and in the latter case the second-order reaction rate constant is k• OH+H 2 O 2 = 2.7 × 10 7 M −1 ·s −1 [26]. By assuming k• OH+MA as the (unknown) second-order reaction rate constant between ·OH and MA, the competition kinetics between MA and H 2 O 2 yields the following results for the MA photodegradation rate (R MA ):  Figure 2B (see dashed curve in the figure). The fit results suggested that k• OH+MA would be about two orders of magnitude higher than k• OH+H 2 O 2 . This means that the reaction of ·OH with H 2 O 2 is expected to prevail over that with 0.1 mM MA for [H 2 O 2 ] > 10 mM, which is right within the investigated range of H 2 O 2 concentrations.

Effect of Inorganic Anions on MA Photodegradation
The effect of anions commonly occurring in surface waters, and most notably of chloride and carbonate, on the photodegradation of MA induced by H 2 O 2 /UV was studied upon UVC irradiation of MA, H 2 O 2 , and, where relevant, NaCl or Na 2 CO 3 .
The time evolution of 0.1 mM MA in the presence of 0.1 M NaCl and different concentration values of H 2 O 2 is reported in Figure 2C, and the corresponding photodegradation rates are reported in Figure 2B. The figure shows that MA photodegradation became progressively faster as the H 2 O 2 concentration increased and, differently from the previous case (MA + H 2 O 2 + UV, without chloride), there was no obvious plateau trend. The experimental rate data could be fitted well with an equation of the form , where β is a constant proportionality factor (see the dashed curve in Figure 2B). In this case it seems that the observed trend just mirrored the photon absorption by H 2 O 2 , with no need to invoke an additional competition kinetics between MA, H 2 O 2 and the reactive transient species. Moreover, at elevated H 2 O 2 concentration the photodegradation of MA was considerably faster in the presence of 0.1 M NaCl than in the absence of chloride. These pieces of evidence suggest that the prevailing reactive species in the MA/H 2 O 2 /Cl − /UV system is very unlikely to be ·OH, which is expected to produce a plateau trend as per the above discussion. A different transient species should rather be involved, inducing competition kinetics between MA and H 2 O 2 to a far lesser extent than ·OH. This reactive transient, provisionally indicated here as X, should react with MA and H 2 O 2 in such a way that (4), which differs from Equation (3) in that the ·OH-based terms are replaced by X-based ones: In the presence of ·OH + Cl − , the following reactions may take place [26][27][28]: Based on the above reactions, potential X-species in the system are HOCl· − , Cl·, and Cl 2 · − . The reactivity of Cl 2 · − can be studied by laser flash photolysis, thus one can check the possible involvement of Cl 2 · − in MA photodegradation by measuring k Cl •− 2 +MA . In the H 2 O 2 /Na 2 CO 3 /UV system with 0.1 M Na 2 CO 3 , the photodegradation of MA did not accelerate when increasing [H 2 O 2 ] above 5 mM (see Figure 2D for the MA time trends, and Figure 2B for the corresponding photodegradation rates). The ·OH reactions with carbonate and bicarbonate are more straightforward than in the case of chloride and they lead to the unequivocal formation of CO 3 · − as additional reactive species [26,29]: A comparison of the MA photodegradation rates in the systems "H 2 O 2 alone" and "H 2 O 2 + Na 2 CO 3 " in Figure 2B shows that the rates were lower in the presence of carbonate, coherently with the replacement of ·OH with the less reactive species CO 3 · − .
. A potential explanation for this phenomenon is that H 2 O 2 competes more effectively with MA, for reaction with CO 3 · − , than for reaction with ·OH. In other words, this hypothesis leads to the assumption that [30], the measurement of k CO •− 3 +MA by laser flash photolysis is an appropriate test for this hypothesis.

MA Photodegradation by Persulphate/UV
The UV irradiation of persulphate yields the sulphate radical, SO 4 · − [31][32][33]. This radical has similar if not higher reduction potential compared to ·OH, but it tends to be preferentially involved in charge-transfer reactions while ·OH often triggers hydrogen-transfer or addition processes in comparable conditions [17,34].
The time trend of 0.1 mM MA upon UVC irradiation in the presence of varying concentration values of sodium persulphate (PS) is reported in Figure 3. The figure shows that PS above 1 mM concentration could considerably accelerate the photodegradation of MA, and that the photodegradation became considerably faster as the PS concentration was higher. Moreover, while there was limited difference between the MA time trends with 5 or 10 mM H 2 O 2 , the photodegradation of MA with 10 mM PS was considerably faster compared to 5 mM PS. This result suggests that the reaction between SO 4 · − and PS interferes with MA photodegradation to a lesser extent than the reaction between ·OH and H 2 O 2 .
charge-transfer reactions while ·OH often triggers hydrogen-transfer or addition processes in comparable conditions [17,34].
The time trend of 0.1 mM MA upon UVC irradiation in the presence of varying concentration values of sodium persulphate (PS) is reported in Figure 3. The figure shows that PS above 1 mM concentration could considerably accelerate the photodegradation of MA, and that the photodegradation became considerably faster as the PS concentration was higher. Moreover, while there was limited difference between the MA time trends with 5 or 10 mM H2O2, the photodegradation of MA with 10 mM PS was considerably faster compared to 5 mM PS. This result suggests that the reaction between SO4· − and PS interferes with MA photodegradation to a lesser extent than the reaction between ·OH and H2O2.

Second-Order Reaction Rate Constants of MA with Cl2· − , CO3· − and SO4· −
The second-order reaction rate constants between MA and three reactive transient species (Cl2· − , CO3· − , and SO4· − ) were measured by means of the laser flash photolysis technique. The radical Cl2· − was produced by laser irradiation of H2O2 + NaCl (0.01 M chloride) at pH 3 by HClO4, under which conditions the equilibria of reactions (4-6) are shifted towards the products and there is a consequent enhancement of the formation of Cl2· − [26][27][28]. As far as the other transient species are concerned, CO3· − was produced by laser irradiation of H2O2 + Na2CO3, and SO4· − was produced by laser irradiation of Na2S2O8. The actual occurrence of these radicals as the main transient species in the laser-irradiated solutions has been demonstrated in previous studies [35,36]. Figure 4A reports the absorption spectra of the studied solutions undergoing laser flash photolysis, obtained just after the laser pulse. Based on these results, in successive experiments the radical Cl2· − was monitored at 350 nm, CO3· − at 550 nm, and SO4· − at 450 nm. Figure   The second-order reaction rate constants between MA and three reactive transient species (Cl 2 · − , CO 3 · − , and SO 4 · − ) were measured by means of the laser flash photolysis technique. The radical Cl 2 · − was produced by laser irradiation of H 2 O 2 + NaCl (0.01 M chloride) at pH 3 by HClO 4 , under which conditions the equilibria of reactions (4-6) are shifted towards the products and there is a consequent enhancement of the formation of Cl 2 · − [26][27][28]. As far as the other transient species are concerned, CO 3 · − was produced by laser irradiation of H 2 O 2 + Na 2 CO 3 , and SO 4 · − was produced by laser irradiation of Na 2 S 2 O 8 . The actual occurrence of these radicals as the main transient species in the laser-irradiated solutions has been demonstrated in previous studies [35,36]. Figure 4A reports the absorption spectra of the studied solutions undergoing laser flash photolysis, obtained just after the laser pulse. Based on these results, in successive experiments the radical Cl 2 · − was monitored at 350 nm, CO 3 · − at 550 nm, and SO 4 · − at 450 nm. Figure 4B   The formation of CO3· − and SO4· − upon either laser-based or steady-state irradiation of, respectively, H2O2 + Na2CO3 and Na2S2O8 is rather straightforward [35,36]. In the case of H2O2 + NaCl, the laser irradiation took place at pH 3 to ensure the formation of Cl2· − . In contrast, the corresponding steady irradiation experiments took place at the natural pH, where the involvement of Cl2· − in MA photodegradation is less obvious.
To assess the actual involvement of Cl2· − in the steady irradiation process one can check (see Figure 2B). Table 2 summarises the second-order reaction rate constants of Cl2· − , CO3· − and ·OH with MA, derived in this study, and those with H2O2 and HO2 − , obtained from the literature [26,30].  The formation of CO 3 · − and SO 4 · − upon either laser-based or steady-state irradiation of, respectively, H 2 O 2 + Na 2 CO 3 and Na 2 S 2 O 8 is rather straightforward [35,36]. In the case of H 2 O 2 + NaCl, the laser irradiation took place at pH 3 to ensure the formation of Cl 2 · − . In contrast, the corresponding steady irradiation experiments took place at the natural pH, where the involvement of Cl 2 · − in MA photodegradation is less obvious.
To assess the actual involvement of Cl 2 · − in the steady irradiation process one can check  Figure 2B). Table 2 summarises the second-order reaction rate constants of Cl 2 · − , CO 3 · − and ·OH with MA, derived in this study, and those with H 2 O 2 and HO 2 − , obtained from the literature [26,30].  Figure 2B) are consistent with k• OH+H 2 O 2 (k• OH+MA ) −1~0 .01.
From this value and the condition   [30], from the condition which is again not consistent with the laser flash photolysis results. A more reasonable explanation is that the reactions of HO 2 − with ·OH and CO 3 · − are much faster than those of H 2 O 2 (see Table 2), thereby causing a considerable production of HO 2 ·/O 2 · − (reactions 11, 12; [26,30]). The superoxide radical anion that prevails at the pH conditions of the studied systems [37] is an effective reductant [38], and it could reduce the oxidised MA transients back to the initial compound (see e.g., reaction 13).
< 2 × 10 7 M −1 ·s −1 , which is again not consistent with the laser flash photolysis results. A more reasonable explanation is that the reactions of HO2 − with ·OH and CO3· − are much faster than those of H2O2 (see Table 2), thereby causing a considerable production of HO2·/O2· − (reactions 11, 12; [26,30]). The superoxide radical anion that prevails at the pH conditions of the studied systems [37] is an effective reductant [38], and it could reduce the oxidised MA transients back to the initial compound (see e.g., reaction 13).
The above reactions, ending up in an inhibition of MA photodegradation, might explain the trend of ] [ . 2 2 O H vs R MA in the presence of carbonate, reported in Figure 2B.

MA Photodegradation Intermediates
The LC-MS analysis of the MA solutions irradiated in the presence of H2O2, with a conversion percentage of 32%, allowed the detection of MA at the retention time of 12.5 min and of several photodegradation intermediates, namely P1 (10.6 min), P2 (11.0 min), P3 (11.4 min), and P4 (13.1 min). Useful information was initially obtained from the MS spectrum of MA itself. A pattern of MA fragmentation, based on the information obtained in its MS 2 spectrum at 20 eV, is shown in Figure  5a. The spectrum shows the formation of a fragment ion with an accurate mass of m/z = 120.0499, which corresponds to an elemental composition of C7H6ON + (error = −17 ppm) and is formed by the loss of a CH3OH group. This fragmentation is a peculiar behaviour of ortho-substituted esters [39]. Two additional fragment ions are also observed at m/z 92 and 65. The former with an accurate mass of 92.0500 (C6H6N + , error = −1.3 ppm) arises from the loss of HCO2CH3 from the molecular ion, which is a common fragmentation process in the methyl esters of carboxylic acids [40]. The same fragment could also be produced by CO loss from the fragment ion at m/z 120.0499. The fragment with m/z = 65.0391 (C5H5 + , error = −3.1 ppm) is obtained from m/z = 92.0500 by loss of HCN.
As far as the intermediates P1, P2, and P3 are concerned, they were characterised by the molecular ion m/z = 168.0655. This is consistent with the elemental composition C8H10O3N + (error = −3.4 ppm), corresponding to MA monohydroxy derivatives. Remarkably, despite the possibility to hydroxylate MA in four different positions, only three isomers were actually detected with P2 as the major one. The MS 2 product ions of these compounds are listed in Table 3, together with the LC retention times of the parent molecules.
In the case of P1, the most abundant product ion is 109.0515 m/z (C6H7ON + , error = −11.6 ppm), which arises from the loss of CH3COO· and is consistent with the presence of the -OH group in position 4 or 6 with respect to the ester functionality of MA. The fragment at 81.0590 m/z (C5H7N + , error = +14.2 ppm) can be explained with the further loss of another CO group. The formation of the 141.0569 m/z fragment (C7H9O3 + , error = +12.3 ppm) can be justified with the loss of HCN, whereas the detachment of a CH3O· radical group would yield the fragment at 137.0470 m/z (C7H7O2N + , error = −5.0 ppm). Unfortunately, no further information is present in the spectrum that allows for the (13) The above reactions, ending up in an inhibition of MA photodegradation, might explain the trend of R MA vs. [H 2 O 2 ] in the presence of carbonate, reported in Figure 2B.

MA Photodegradation Intermediates
The LC-MS analysis of the MA solutions irradiated in the presence of H 2 O 2 , with a conversion percentage of 32%, allowed the detection of MA at the retention time of 12.5 min and of several photodegradation intermediates, namely P1 (10.6 min), P2 (11.0 min), P3 (11.4 min), and P4 (13.1 min). Useful information was initially obtained from the MS spectrum of MA itself. A pattern of MA fragmentation, based on the information obtained in its MS 2 spectrum at 20 eV, is shown in Figure 5a. The spectrum shows the formation of a fragment ion with an accurate mass of m/z = 120.0499, which corresponds to an elemental composition of C 7 H 6 ON + (error = −17 ppm) and is formed by the loss of a CH 3 OH group. This fragmentation is a peculiar behaviour of ortho-substituted esters [39]. Two additional fragment ions are also observed at m/z 92 and 65. The former with an accurate mass of 92.0500 (C 6 H 6 N + , error = −1.3 ppm) arises from the loss of HCO 2 CH 3 from the molecular ion, which is a common fragmentation process in the methyl esters of carboxylic acids [40]. The same fragment could also be produced by CO loss from the fragment ion at m/z 120.0499. The fragment with m/z = 65.0391 (C 5 H 5 + , error = −3.1 ppm) is obtained from m/z = 92.0500 by loss of HCN. As far as the intermediates P1, P2, and P3 are concerned, they were characterised by the molecular ion m/z = 168.0655. This is consistent with the elemental composition C 8 H 10 O 3 N + (error = −3.4 ppm), corresponding to MA monohydroxy derivatives. Remarkably, despite the possibility to hydroxylate MA in four different positions, only three isomers were actually detected with P2 as the major one. The MS 2 product ions of these compounds are listed in Table 3, together with the LC retention times of the parent molecules.
In the case of P1, the most abundant product ion is 109.0515 m/z (C 6 H 7 ON + , error = −11.6 ppm), which arises from the loss of CH 3   As far as P2 and P3 are concerned, the most abundant signal occurs at 136.0375 m/z (C7H6O2N + , error = −17.3 ppm) and, in analogy with the fragmentation of MA, it could arise from CH3OH loss. As already seen for P1, one also observes the product ion at 137.0465 m/z. The occurrence of the product ion at 107.0358 m/z (H2CO loss) suggests the presence of an OH group in ortho or para position  As far as P2 and P3 are concerned, the most abundant signal occurs at 136.0375 m/z (C 7 H 6 O 2 N + , error = −17.3 ppm) and, in analogy with the fragmentation of MA, it could arise from CH 3 OH loss. As already seen for P1, one also observes the product ion at 137.0465 m/z. The occurrence of the product ion at 107.0358 m/z (H 2 CO loss) suggests the presence of an OH group in ortho or para position with respect to the amino group (i.e., in position 3 or 5 with respect to the ester functionality). A possible fragmentation pathway for the 3-hydroxyderivative is shown in Figure 5c, but a fully similar pathway could be proposed for the 5-hydroxyderivative. From the available MS data it was unfortunately not possible to attribute uniquely each isomer to the corresponding signal. However, by assuming that P2 and P3 are the 3-and 5-hydroxyderivatives of MA (irrespective of which is which), one can tentatively conclude that both of them are anyway formed. In contrast, P1 may be either the 4-or the 6-hydroxyderivative. Therefore, one could tentatively assume that hydroxylation takes place in the 3 and 5 positions, plus 4 or 6 (in other words, either the 3-, 4-, and 5-or the 3-, 5-, and 6-hydroxyderivatives would be formed).
In the case of P4, the accurate mass of the molecular ion (m/z = 331.0915) corresponds to the elemental composition C 16 H 15 N 2 O 6 + , with an error of −4.6 ppm. This indicates the possible presence of an oxidised dimeric structure. Unfortunately, based on the available MS data it was not possible to propose a univocal structure for this compound.
Based on ECOSAR predictions, the MA hydroxyderivatives would show comparable toxicity as the parent molecule [7]. In all the cases the major effects are predicted to be the acute and, most notably, the chronic toxicity towards fish and crustaceans.

Irradiation Experiments
The absorption spectra of the studied compounds (see Figure 1 for MA and H 2 O 2 ) were taken with a Varian (Palo Alto, CA, USA) Cary 3 UV-vis spectrophotometer, using 1 cm quartz cuvettes. The solution pH was measured with a combined glass electrode connected to a Meterlab pH meter (Hach Lange, Loveland, CO, USA). Solutions containing 0.1 mM MA, and other components where relevant, were inserted inside a quartz tube (100 mL total volume), which was placed in the centre of an irradiation set-up consisting of six TUV Philips (Amsterdam, Netherlands) 15 W lamps with emission maximum at 254 nm. The lamp intensity was 7.6 × 10 −9 Einstein cm −2 ·s −1 . The water solutions were magnetically stirred during irradiation. At scheduled irradiation times, 1.5 mL sample aliquots were withdrawn from the tube, placed into HPLC vials, and kept refrigerated until HPLC analysis. The time trend of MA was monitored by means of a high-performance liquid chromatograph interfaced to a photodiode-array detector (HPLC-PDA, model Nexera XR by Shimadzu, Kyoto, Japan), equipped with SIL20-AC autosampler, SIL-20AD pump module for low-pressure gradients, CT 0-10AS column oven (set at 40 • C), reverse-phase column Kinetex RP-C18 packed with Core Shell particles (100 mm × 2.10 mm × 2.6 µm) by Phenomenex (Torrance, CA, USA), and SPDM 20A photodiode array detector. The isocratic eluent was a A/B = 60/40 mixture of A = (0.5% formic acid in water, pH 2.3) and B = methanol, at a flow rate of 0.2 mL min −1 . In these conditions, the MA retention time was 7.3 min. The detection wavelength was set at 218 nm. A schematic of the experimental procedure is reported in Figure 6.

Identification of Photodegradation Intermediates
The photodegradation intermediates of MA were identified by liquid chromatography interfaced with mass spectrometry (LC-MS). A Waters Alliance (Milford, MA, USA) instrument equipped with an electrospray (ESI) interface (used in ESI+ mode) and a Q-TOF mass spectrometer (Micromass, Manchester, UK) were used. Samples were eluted on a column Phenomenex Kinetex C18 (100 mm × 2.10 mm × 2.6 µm) with a mixture of acetonitrile (A) and 0.1% formic acid in water (B) at 0.2 mL·min −1 flow rate, with the following gradient: start at 5% A, then up to 95% A in 15 min, keep for 10 min, back to 5% A in 1 min, and keep for 5 min (post-run equilibration). The capillary needle voltage was 3 kV and the source temperature 100 °C. The cone voltage was set to 35 V. Data acquisition was carried out with a Micromass MassLynx 4.1 data system. Both MS and MS/MS experiments were carried out by using this chromatographic set-up.

Laser Flash Photolysis Experiments
The reactivity of the radicals Cl2· − , CO3· − , and SO4· − was studied by means of the nanosecond laser flash photolysis technique. Flash photolysis runs were carried out using the third harmonic (266 nm) of a Quanta Ray GCR 130-01 Nd:YAG laser system instrument, used in a right-angle geometry with respect to the monitoring light beam. The single pulses energy was set to 35 mJ unless otherwise stated. A 3 mL solution volume was placed in a quartz cuvette (path length of 1 cm) and used for a maximum of three consecutive laser shots. The transient absorbance at the pre-selected wavelength was monitored by a detection system consisting of a pulsed xenon lamp (150 W), monochromator, and a photomultiplier (1P28). A spectrometer control unit was used for synchronising the pulsed light source and programmable shutters with the laser output. The signal from the photomultiplier was digitised by a programmable digital oscilloscope (HP54522A). A 32 bits RISC-processor kinetic spectrometer workstation was used to analyse the digitised signal.

Model Assessment of Toxicity
The potential acute and chronic toxicity of the detected MA intermediates was assessed with the ECOSAR software (US-EPA, Washington DC, USA). ECOSAR uses a quantitative structure-activity relationship approach to predict the toxicity of a molecule of given structure. The relevant endpoints are the acute and chronic toxicity thresholds (LC50, EC50, chronic values ChV) for freshwater fish, daphnid, and algae. The values predicted by ECOSAR are apparently very precise but, as far as accuracy is concerned, a compound can be said to be more toxic than another only when the predicted values differ by at least an order of magnitude [7,8].

Identification of Photodegradation Intermediates
The photodegradation intermediates of MA were identified by liquid chromatography interfaced with mass spectrometry (LC-MS). A Waters Alliance (Milford, MA, USA) instrument equipped with an electrospray (ESI) interface (used in ESI+ mode) and a Q-TOF mass spectrometer (Micromass, Manchester, UK) were used. Samples were eluted on a column Phenomenex Kinetex C18 (100 mm × 2.10 mm × 2.6 µm) with a mixture of acetonitrile (A) and 0.1% formic acid in water (B) at 0.2 mL·min −1 flow rate, with the following gradient: start at 5% A, then up to 95% A in 15 min, keep for 10 min, back to 5% A in 1 min, and keep for 5 min (post-run equilibration). The capillary needle voltage was 3 kV and the source temperature 100 • C. The cone voltage was set to 35 V. Data acquisition was carried out with a Micromass MassLynx 4.1 data system. Both MS and MS/MS experiments were carried out by using this chromatographic set-up.

Laser Flash Photolysis Experiments
The reactivity of the radicals Cl 2 · − , CO 3 · − , and SO 4 · − was studied by means of the nanosecond laser flash photolysis technique. Flash photolysis runs were carried out using the third harmonic (266 nm) of a Quanta Ray GCR 130-01 Nd:YAG laser system instrument, used in a right-angle geometry with respect to the monitoring light beam. The single pulses energy was set to 35 mJ unless otherwise stated. A 3 mL solution volume was placed in a quartz cuvette (path length of 1 cm) and used for a maximum of three consecutive laser shots. The transient absorbance at the pre-selected wavelength was monitored by a detection system consisting of a pulsed xenon lamp (150 W), monochromator, and a photomultiplier (1P28). A spectrometer control unit was used for synchronising the pulsed light source and programmable shutters with the laser output. The signal from the photomultiplier was digitised by a programmable digital oscilloscope (HP54522A). A 32 bits RISC-processor kinetic spectrometer workstation was used to analyse the digitised signal.

Model Assessment of Toxicity
The potential acute and chronic toxicity of the detected MA intermediates was assessed with the ECOSAR software (US-EPA, Washington DC, USA). ECOSAR uses a quantitative structure-activity relationship approach to predict the toxicity of a molecule of given structure. The relevant endpoints are the acute and chronic toxicity thresholds (LC50, EC50, chronic values ChV) for freshwater fish, daphnid, and algae. The values predicted by ECOSAR are apparently very precise but, as far as accuracy is concerned, a compound can be said to be more toxic than another only when the predicted values differ by at least an order of magnitude [7,8].

Conclusions
The H 2 O 2 /UV technique as photochemical ·OH source is a potentially effective tool to achieve MA photodegradation, and in fact the addition of hydrogen peroxide considerably accelerated the photodegradation of MA compared to UV irradiation alone. The addition of inorganic anions that act as ·OH scavengers, such as chloride and carbonate, did not necessarily quench MA photodegradation. The reason is the reactivity with MA itself of the generated radical species, i.e., Cl 2 · − produced from Cl − +·OH and CO 3 · − produced from CO 3 2− +·OH. In the case of chloride, there was even an acceleration of MA photodegradation at elevated [H 2 O 2 ], because Cl 2 · − competes more successfully than ·OH for reaction with MA in the presence of H 2 O 2 (H 2 O 2 behaves as a scavenger of ·OH and, to a lesser extent, of Cl 2 · − as well). The same effect was not observed with carbonate, possibly because the basic pH caused a considerable production of superoxide (O 2 · − ) upon oxidation of the H 2 O 2 conjugated base, HO 2 − . The radical O 2 · − is a well-known reductant that could reduce the partially oxidised MA back to the starting compound. Effective MA photodegradation was also observed with persulphate/UV, probably because of the fast reaction between MA and photogenerated SO 4 · − , and because of limited scavenging of SO 4 · − by persulphate itself. Among the MA photodegradation intermediates detected in the H 2 O 2 /UV process, the hydroxyderivatives could be about as toxic as the parent compound. Therefore, decontamination is not yet achieved once MA has disappeared, and the H 2 O 2 /UV treatment of MA should at least ensure the photodegradation of the MA hydroxylated derivatives as well. Usually, the photodegradation of both the primary compound and its intermediates takes more time than the photodegradation of the starting compound alone.