Iron (III) hydroxocomplex-methyl viologen dication system as a prospective tool for determination of hydroxyl radical reaction rate constants with environmental pollutants

Reactivity of oxidative species with target pollutants is one of the crucial parameters for application of any system based on advanced oxidation processes (AOPs). This work presents new useful approach how to determine the hydroxyl radical reaction rate constants (kOH) using UVA laser flash photolysis technique. Fe (III) hydroxocomplex at pH 3 was applied as a standard source of hydroxyl radicals and methyl viologen dication (MV2+) was used as selective probe for •OH radical. Application of MV2+ allows to determine kOH values even for compounds which do not generate themselves optically detectable transient species in reaction with hydroxyl radicals. Validity of this approach was tested on a wide range of different persistent pesticides and its main advantages and drawbacks in comparison with existing steady-state and time-resolved techniques were discussed.


Introduction
Advanced oxidation processes based on generation of highly oxidative hydroxyl ( • OH) radicals are very popular nowadays due to increasing contamination of natural waters and necessity to develop effective and low-cost water treatment procedures (Deng and Zhao 2015;Giannakis et al. 2017a; Giannakis et al. 2017b;Shen et al. 2019;Villegas-Guzman et al. 2017). As the absorption band of • OH radical lies in the deep UV (Buxton et al. 1988), the direct observation of its reactions with target pollutants is complicated. Usually, unknown hydroxyl radical reaction rate constant (k OH X ) is determined by competitive method using a substance (S) with known k OH value as a standard (Haag and Yao 1992;Joseph et al. 2001;Orellana-García et al. 2015;Sánchez-Polo et al. 2013). The main advantage of aforesaid approach is the possibility to use steady-state (photo)chemical methods of generation of • OH radical and widespread and convenient optical spectroscopy or HPLC techniques for calculation of k OH X value. However, this approach has some serious drawbacks as follows: 1) This is relative method giving k OH X /k OH S ratio in result, so standard with very well determined k OH S value has to be used. 2) If photochemical excitation is used for • OH radical generation, neither X nor S should exhibit good absorption and/or own photochemical activity under excitation light.
3) If there are reactions between secondary organic radicals and initial compounds X and S, the determination of k OH X using this method will give an incorrect result.
A less popular, but more straightforward approach is based on using a time-resolved method (laser flash photolysis (LFP) or pulse radiolysis (PR), namely), which allows direct optical registration of • OH radical reaction with target compound (Peller and Kamat 2005;Rafqah et al. 2004;Terzian et al. 1995;Zona et al. 2012). The main advantage of this approach that it is absolute method which allows direct calculation of k OH X values. However, it also has some drawbacks as follows: 1) Both LFP and PR techniques are less common and more sophisticated methods in comparison with ordinary HPLC. 2) In LFP method, a photochemical system with high quantum yield of hydroxyl radical generation is needed and X should not exhibit good absorption and/or own photochemical activity under excitation light. 3) If • OH adduct with X molecule exhibits low absorption in registration range of LFP or PR setup or its absorption is borrowed by intensive absorption of initial compounds, the calculation of k OH X will be very complicated or even impossible.
Last two drawbacks could be overcome by addition of another hydroxyl radical probe which do not absorb at excitation wavelength, reacts readily with hydroxyl radical, and produce an intermediate with good absorption in convenient optical range (Catastini et al. 2004). In this work, we develop this approach using Fe(III) hydroxocomplex as a well-recommended photolytic source of hydroxyl radical and methyl viologen dication as a selective probe for this active species.

Experimental
All chemicals used in the study were of analytical reagent grade or higher purity and were used without further purifcation. Milli-Q water was used in all photochemical experiments. The initial pH of solutions was adjusted to ~3 by chemically pure perchloric acid and was controlled by the Anion-4100 ionometer (Infraspak-Analit, Russia). Optical spectra were recorded using Agilent 8453 spectrophotometer (Agilent Technologies, USA).
Time-resolved experiments were performed using a laser flash photolysis (LFP) setup described in details elsewhere (Pozdnyakov et al. 2000). Briefly, a tunable LS-2137U laser (Lotis, Belarus) with an excitation wavelength of 355 nm was used as excitation source with pulse duration of about 6 ns, and pulse energy from 1 to 15 mJ. The probe light source was a xenon arc lamp DKSSh-150 (Russia). An increase of the light intensity by a factor of about 100 was achieved by an additional current pulse (~150 A, ~1 ms). A fraction of probe light was sent to a photodiode with a quartz plate and was used as feedback signal to stabilize the light intensity.
The excitation and probe light beams were directed to the sample with a small angle (≈2°) through a diaphragm (2 mm in diameter). After the sample, the probe light passed through a monochromator MDR-23 (LOMO, Russia) equipped with a photomultiplier FEU-84 (Russia). The transient absorption signal was amplified (up to 256 times) and then directed to an 8-bit ADC with 1024 counts and a time resolution of 50 ns. With this PC-controlled setup, absorption changes as small as 5×10 −4 could be measured. Each kinetic curve was obtained by averaging of 20-30 independent laser flashes to obtain proper signal-to-noise ratio. Quartz cells with an optical path length of 1 cm and a total volume of 3 ml at normal oxygen content in solutions were applied.

Results and discussion
The set of organic pesticides used in this study as typical environmental contaminants is presented in Fig. 1. It is worth to note that all of these compounds exhibit no absorption at 355 nm, are stable in the presence of Fe(III) ions, and do not complex with them. It was demonstrated by coincidence of sum of absorption spectra of individual species with spectrum of the mixture and stability of the latter during characteristic times of LFP experiments. It is also assumed that X-• OH adducts exhibit negligible absorption at wavelengths higher than 400 nm. Such situation is typical for X-• OH adducts formed upon hydroxyl radical attack to benzene ring (Buxton et al. 1988;Terzian et al. 1995;Zona et al. 2002) and coincide with known literature data for target compounds (Table 1).
The elementary reactions of • OH radicals with target pollutants were studied with efficient sensitizer of these radicals, Fe(OH) 2+ hydroxocomplex (Benkelberg and Warneck 1995;Joseph et al. 2001;Lee and Yoon 2004;Pozdnyakov et al. 2000). In our LFP experiments, the concentration of Fe(III) perchlorate was fixed as 600 μM at pH 2.9 In these conditions, the FeOH 2+ hydroxocomplex was a main form of Fe(III) in solution (about 80%) and a sole photoactive species with absorption of about 50% of light quanta at 355 nm (Fig. 2).
In condition of our LFP experiments, typical • OH radical concentration is about 1 μM which is much less then concentration of both MV 2+ (240 μM) and target compounds (25-500 μM) used in this study. It allows taking into account only reactions of • OH radical with aforesaid  where k d is the observed rate constant of MV( • OH) 2+ adduct decay. In this scheme, we did not consider decay of X( • OH) adducts because they have not own absorption at 470 nm and their decay do not interfere with measuring of MV( • OH) 2+ signal. It is worth to note that according to Solar et al., the main pathway of MV( • OH) 2+ disappearance in aqueous solutions is disproportionation with the rate constant, 2k dis = (1.3 ± 0.2) × 10 8 M -1 s -1 . However, due to very low concentration of MV( • OH) 2+ adduct, we can successfully use firstorder reaction (4) to describe the decay of this species in our time window (Fig. 3). Solution of kinetic schemes (5 and 6) gives the dependence of MV( • OH) 2+ adduct concentration on time: It is worth to note that amplitude of the signal at 470 nm gradually decreased with increasing of 2,4-DB concentration due to competition between the herbicide and MV 2+ for • OH radicals (Eq. 9). However in all cases signal-to-noise ratio is high enough to determine k obs value with about 10-15% precision. Very good fit was obtained for all concentration of 2,4-DB as well as for other compounds (Fig. 4) which proves validity of simplified kinetic scheme (5 and 6) and allows to calculate k OH X values for studied pesticides ( Table 1). Analysis of data presenting in Table 1 allows to conclude that our approach gives the k OH X values which well coincides with result obtaining by both steady-state (stationary photolysis, ozonation) and time-resolved (LPF, PR) methods. It is worth to note that k OH X values are practically independent on pH for the studied compounds. It could be explained by the fact that studied herbicides (for exception of AMT, ATR, and DIC) do not contain an acid/ base substituent directly attached to aromatic π-system. For this reason, deprotonation of an aliphatic substituent does not change greatly red-ox properties of a target molecule and its reactivity towards hydroxyl radical. This is clearly seen in comparison of k OH X values for 2,4-DB; 2,4,5-T; and TCP which are practically the same for both neutral and anionic forms of the herbicide. In a case of ATR, at all pH higher than 2.5 these compound exists mainly in neutral form (pK a = 1.6), so k OH X values for this compound are not affected by pH (Table 1). Similar situation is expected for DIC (pK a = 1.9) which exist as monoanion in environmentally relevant pH range.
For AMT which contains protonated amino group directly attached to the triazole π-system, k OH X increases about 1.6 times for neutral form in comparison with cationic one (Table 1). So we can conclude that for aromatic amines, phenols and similar compounds with pK a values higher than 3 k OH X values obtained by our approach could be used only as tentative estimates of the reactivity of the compounds towards hydroxyl radical at higher pH. However, our approach allows to measure k OH X values for a set of target compounds (from two to four) for a day due to high analytical signal of MV( • OH) 2+ adduct at selected registration wavelength (470 nm). It is not a case of general approach using LFP or PR techniques where one first needs to find the absorption maximum of unknown adduct of • OH radical with studied contaminant and this is limiting time-consuming step. In a case of application of steady-state techniques, one needs to spend a time for finding a proper competing standard and for time-consuming HPLC measurements.

Conclusions
LPF of Fe(III) hydroxocomplex-MV 2+ system at pH 3 can be successfully used to determine k OH values for a wide range of typical contaminants of natural waters. Validity of this approach was proven on by comparison of obtained k OH values for a set of persistent pesticides with literature ones found by another approaches. The main advantages of proposed system in comparison with existing popular competitive steady-state approaches or direct LFP with target molecule are as follows: 1) Absolute method allowing direct measurement of k OH value in opposite to competitive steady-state approaches. 2) MV( • OH) 2+ adduct absorption spectrum does not overlap with spectra of • OH adducts with typical environmental contaminants. 3) "One-wavelength registration" method, allowing to avoid time-consuming determination of transient absorption spectrum of • OH adduct with a studied environmental contaminant. 4) k OH values could be obtained with good experimental precision (c.a. 10-15%) due to high absorption of MV( • OH) 2+ adduct in convenient optical range (λ max = 470 nm, ε max = 16000 ± 700 M −1 cm −1 ). 5) The method could be applied for compounds which do not generate themselves optically detectable transient species in reaction with hydroxyl radicals.
However, some limitations of aforesaid approach also should be stressed: 1) The method should not be applied for target contaminants which are not stable at acidic pH can be easily oxidized by Fe(III) ions (like catechols and hydroquinones) or formed a stable complexes with them (aromatic and aliphatic polidentate compounds like salicylate, citrate, and EDTA). 2) The approach could fail for target contaminants with intensive absorption in UVA and visible range (like dyes, pigments, and similar complex aromatic and/ or heterocyclic compounds) which can compete with Fe(III) hydroxocomplex for excitation light or exhibit own photochemical activity under laser excitation.
3) The method should not be applied for compounds which generate X( • OH) adducts with good absorption at 470 nm. In this case, direct measurement of observed rate constant of formation of transient absorption of X− • OH adduct in the maximum of its absorption is recommended. 4) k OH X values obtained by our approach at pH 3 could be used only as tentative estimates of the reactivity of Fig. 4 The dependence of k obs calculated by Eq. 9 from initial concentration of 2,4-DB (1), MSM (2), TCP (3), and AMT (4). Straight lines are the best fits by Eq. 8 aromatic amines, phenols, and similar compounds with pK a >3 towards hydroxyl radical at higher pH.
We believed that current approach will be successfully used in the studies devoted to application of AOPs based on hydroxyl radical generation for determination of k OH values for target environmental contaminants.