Complexation Facilitated Reduction of Aromatic N-Oxides by Aqueous

Article pubs.acs.org/est

Complexation Facilitated Reduction of Aromatic N‑Oxides by Aqueous FeII−Tiron Complex: Reaction Kinetics and Mechanisms Yiling Chen and Huichun Zhang* Department of Civil and Environmental Engineering, Temple University, Philadelphia, Pennsylvania 19122, United States S Supporting Information *

ABSTRACT: Rapid reduction of carbadox (CDX), olaquindox and several other aromatic N-oxides were investigated in aqueous solution containing FeII and tiron. Consistent with previous work, the 1:2 FeII−tiron complex, FeL26‑, is the dominant reactive species as its concentration linearly correlates with the observed rate constant kobs under various conditions. The N-oxides without any side chains were much less reactive, suggesting direct reduction of the N-oxides is slow. UV−vis spectra suggest FeL26‑ likely forms 5- or 7-membered rings with CDX and olaquindox through the N and O atoms on the side chain. The formed inner-sphere complexes significantly facilitated electron transfer from FeL26‑ to the N-oxides. Reduction products of the N-oxides were identified by HPLC/QToF-MS to be the deoxygenated analogs. QSAR analysis indicated neither the first electron transfer nor N−O bond cleavage is the rate-limiting step. Calculations of the atomic spin densities of the anionic N-oxides confirmed the extensive delocalization between the aromatic ring and the side chain, suggesting complex formation can significantly affect the reduction kinetics. Our results suggest the complexation facilitated N-oxide reduction by FeII−tiron involves a free radical mechanism, and the subsequent deoxygenation might also benefit from the weak complexation of FeII with the N-oxide O atom.

■

reductants under lab conditions.18 Typically upon reduction, deoxygenation of N-oxides occurs. For example, selective reduction of the two N−O bonds was observed in carbadox, where trimethyl phosphite in 1,1,2,2-tetrachloroethane and sodium dithionite in base solution selectively removed the N1or N4-O oxygen respectively (see Table 1 for the numbering of the N atoms).19 This information will provide a knowledge base when we examine reduction of aromatic N-oxides in aqueous environments. One of the notable properties of aromatic N-oxides is their ability to complex with various transition metals. Typically, two types of inner-sphere complexes can form. First, the N-oxide O atom can directly complex with the transition metal.20−22 For example, simple aromatic N-oxides such as pyridine N-oxide has been reported to form complex with FeII.23 CoII and NiII complexes with a series of substituted quinoline N-oxides have been investigated.24,25 IR spectra showed that the N−O stretching frequencies for the N-oxides have shifted to lower values after complexation, indicating their bonding through the N-oxide O atom.20,23,25 Second, 6-membered chelate rings form between metals and N-oxides via the N-oxide O atom and a neighboring O or N atom, such as the bidentate chelates of 2-

INTRODUCTION Various applications have been developed for aromatic Noxides (example structures shown in Table 1) in the fields of agriculture, synthetic chemistry, and industrial chemistry.1,2 The N-oxide structure can easily form during chemical synthesis or in metabolic processes of many drugs and xenobiotics.3 Through feed wastes or from wastewater treatment plants, N-oxides will likely enter soils, sediments, and aquatic environments as emerging contaminants. For example, carbadox (CDX) is a common growth promoter and antibiotic for farm animals, and was considered to be toxic and a potential carcinogen.4 Olaquindox (ODX), a feed additive for growth promotion, has been found to induce genotoxicity and oxidative DNA damage in human cells.5 Following intensive investigation of N-oxides in the 60s and 70s on their chemical behaviors,2,6 there is a renewed interest in these compounds due to their wide usage and potential toxic effects on humans and the environment. So far, most published work on environmental transformation of aromatic N-oxides focused on their adsorption to soils7 or surfaces,8,9 biodegradation,10,11 hydrolysis,12,13 photodegradation14−16 and abiotic oxidation.17 Despite the above knowledge, no information is available regarding reductive transformation of N-oxides in aqueous solution under environmentally relevant conditions. On the other hand, a large number of studies have been reported, mostly for synthesis purposes, on reduction of N-oxides by various © 2013 American Chemical Society

Received: Revised: Accepted: Published: 11023

June 14, 2013 August 16, 2013 August 19, 2013 August 19, 2013 dx.doi.org/10.1021/es402655a | Environ. Sci. Technol. 2013, 47, 11023−11031

Environmental Science & Technology

Article

Table 1. Structures, Physical-Chemical Properties, and kobs of the Selected Aromatic N-Oxides

Reactions contained 0.5 mM FeII, 5 mM tiron, and 20 μM N-oxides initially at pH 6.50. bReactions contained 2.0 mM FeII, 5 mM tiron, and 20 μM N-oxides initially at pH 7.80. Reaction for many of these compounds was too slow at Condition “a” to accurately obtain kobs. cReaction condition was the same as Condition “b”. No reduction was observed for two weeks. a

quinoxalinecarboxylic acid with CuII,21 2-picolinate N-oxide with FeIII, CuII and ZnII,26 2-pyridylcarbinol N-oxide with a number of metal ions,27 and 2-picolinic acid N-oxide with lanthanides.28 In addition, 5- or 7-membered rings are able to form with metals through the N or O atoms on the non-Noxide side chains, such as complexes between CuII and N-(3hydroxypropyl)glycinamide, and between CuII and N-(3hydroxypropyl)-DL-alaninamide.29 Studies on the reduction of oxime carbamates by FeII have demonstrated the importance of inner-sphere complex formation between the contaminants and the FeII in facilitating electron transfer.30 Based on the above information, inner-sphere complex formation between metal ions and aromatic N-oxides might facilitate their reduction. Previous studies reported that FeII complexes with organic catechol- and thiol- ligands can quickly reduce a number of organic contaminants due to their low reduction potentials and high redox reactivity.31,32 Tiron (4,5-dihydroxy-1,3-benzenedisulfonic acid) was typically selected as a model ligand because of its high solubility and a better understanding of its reactivity (than of other ligands) once complexed with FeII.31,32 Therefore, FeII−tiron complex was examined in this study as a model environmental reductant in aqueous reduction of aromatic N-oxides. In this study, the reduction kinetics and product formation of CDX, ODX, and several other model N-oxides (structures shown in Table 1) in the presence of FeII−tiron complex were examined. Impacts on the observed rate constants by varying concentrations of CDX, FeII and tiron, as well as varying pH conditions were detected. UV−visible spectra of six compounds (five N-oxides and one analog) as either free ligands or in the presence of FeII were compared to determine the effect of inner-sphere complex formation on the reaction kinetics. Computational calculations were performed to determine molecular descriptors that can serve as predictors of the

reduction reactivity. Based on the above results, the reduction mechanisms of various aromatic N-oxides by FeII−tiron complex were proposed.

■

MATERIALS AND METHODS Chemicals. CDX (1), ODX (2), N-(2-methoxyphenyl)-3methyl-2-quinoxalinecarboxamide 1,4-dioxide (3), 1-(3,6,7trimedyl-1,4-dioxide-2-quinoxalinyl)-ethanone (5), 2-methylquinoxaline 1,4-dioxide (6), 2-methylquinoxaline 1-oxide (7), picolinic acid N-oxide (8), quinoline N-oxide (9), pyridine Noxide (10), pyrazine N-oxide (11), and acetic benzylidenehydrazide (12) were purchased from Sigma-Aldrich at greater than 97% purity. Other employed chemicals were obtained from Sigma-Aldrich or Fisher Scientific. Carbadox-N4 (4) was synthesized following the method from Massy et al.19 (details in the Supporting Information (SI)). All solutions were prepared inside an oxygen-free glovebox (95% N2, 2−5% H2, Pd catalyst) using reagent-grade water (18 MΩ cm resistivity) that was deoxygenated by boiling under vacuum prior to being transferred into the glovebox. All glassware having prior contact with iron solutions was soaked in 5 N HNO3 and was rinsed several times with reagent-grade water prior to use. FeII stock solution was prepared by filtering 1.0 M FeSO4·7H2O solution through a 0.2 μm filter into 0.2 M HCl. Acetate buffer (1.25 M) was used to maintain pH in the range of 5.1−6.0, while MOPS buffer (1 M) was used in the range of 6.5−8.0. Stocks of N-oxides and related compounds were prepared in a 1:1 mixture of acetonitrile and water. Reactor Setup. All experiments were conducted in 50 mL amber glass serum bottles under constant stirring. Reactions were initiated by adding a certain amount of N-oxide stock to solutions containing buffer (0.025 M acetate or 0.02 M MOPS buffers), tiron (5 mM) and FeII (0.5 mM). Aliquots of solution were then periodically collected to analyze, within the same 11024

dx.doi.org/10.1021/es402655a | Environ. Sci. Technol. 2013, 47, 11023−11031


Article

Figure 1. Time course of the reduction products of (a) compound 6 and (b) CDX. Reactions contained (a) 1.0 mM FeII, 5 mM tiron, and 200 μM 6 initially at pH 7.82; and (b) 0.5 mM FeII, 5 mM tiron, and 20 μM CDX initially at pH 5.97.

day, the concentration of N-oxides and the expected reduction products. Reactions were quenched immediately by excessive EDTA addition.32 Ligand free and FeII free control experiments were simultaneously conducted to account for nonreduction processes. Kinetic measurements were made over a wide range of conditions (pH, tiron concentration, FeII concentration, and N-oxide concentration). Analysis of N-Oxides and Products. Decrease in the concentrations of N-oxides and structurally related compounds as well as formation of selected reaction products were monitored from 220 to 308 nm by an Agilent 1200 reversedphase high performance liquid chromatography (HPLC) system with a Zorbax XDB-C18 column (4.6 × 250 mm, 5 μm) or a Zorbax RX-C18 column (4.6 × 250 mm, 5 μm), and a diode array detector. The flow rate was set at 1 mL/min. The mobile phase consisted of a solution containing 60% ammonia acetate (20 mM) and 40% methanol for compounds CDX, 3, 4, 5, 9, and 12; 95% acetic acid (0.012 M) and 5% methanol for ODX, 6, 7, 10, and 11; while 80% H2SO4 (pH 0.8) and 20% methanol for 8. Since CDX-N1 and CDX-N4 have very similar structures, we calculated the concentration of CDX-N1 based on the standard curve for CDX-N4. Reaction products were identified by an HPLC/quadrupole time-of-flight mass spectrometry (HPLC/QToF-MS) system with a XDB-C18 column (4.6 × 250 mm, 5 μm). Prior to analysis, products were fractionated after multiple injections of the reaction mixture into HPLC, dried under a mild N2 flow, and reconstituted by methanol. The flow rate was set at 1 mL/ min, and the mobile phase was the same as in the HPLC system. The MS analysis was conducted using an Agilent 6520 high-resolution mass spectrometer with Q-TOF-HRMS electrospray positive ionization at the fragmentor voltage of 150− 200 V and mass scan range 100−1700 m/z. The nebulizer pressure was set at 45 psig and the capillary was at 4000 V. Analysis of Iron Concentration and FeII-N-Oxide Complex Formation. FeII and total iron concentrations were monitored using the standard ferrozine method33 (see the SI for details). An Agilent 8670 UV−visible spectrophotometer was used to obtain UV−vis spectra of N-oxides and FeII in aqueous solution. Samples contained 100 μM FeII and 100 μM ligand (CDX, ODX, 5, 6, 12) in the pH range of 3.00−6.00. Immediately after preparation, the UV−vis spectra of the samples were obtained. Computational Calculation. The electron affinity (EA), N−O bond length and atomic spin density of the N-oxide radical anions were calculated in the presence of water solvent, performed by the Gaussian 03 software package (Revision

C.02, Gaussian, Inc., Wallingford CT, 2004). Structural optimizations were carried out by the density functional (B3LYP) theory, and the vibrational frequencies were calculated for all the stationary structures to make sure that they are the equilibrium ones. The 6-31++G(d) basis sets were used for all the calculations. The N−O bond lengths and the B3LYP energies were obtained for the neutral molecules and the anions. EA was calculated by subtracting the energy of the anion from that of the neutral molecule.

■

RESULTS Reaction Kinetics. Degradation was not observed in solutions containing only FeII (ligand free control) or tiron (FeII free control). In contrast, significant degradation of Noxides occurred when FeII and tiron were added. A typical time course for CDX degradation in solutions containing FeII and tiron is shown in SI Figure S1. The disappearance of CDX followed pseudofirst-order reaction kinetics. The observed pseudofirst order rate constants kobs for CDX reduction by FeII−tiron complex were measured in a series of reaction solutions by varying concentrations of CDX, FeII and tiron, as well as varying pH conditions. In general, kobs increased substantially with increasing FeII concentration, tiron concentration and pH (SI Table S2). On the other hand, kobs decreased with increasing CDX concentration. For example, at pH 5.90, an increase in FeII concentration from 0.2 to 0.5 mM resulted in an increase in kobs from 0.67 h−1 to 3.45 h−1. Values of kobs increased 3 orders of magnitude when pH increased from 5.05 to 6.00. At pH >7.00, reaction was too fast to be monitored. In addition, CDX reacted much faster when its concentration decreased from 200 to 100 μM. Similar kinetic behaviors were observed for all other Noxides. Values of kobs for all the N-oxides were obtained and are provided in Table 1. The order of the observed reaction rate constants is: CDX ≫ 4 > ODX > 5 > 3 > 6 > 7 ≫ 11 > 8 > 9 > 10. Reduction Products. Based on HPLC analysis of the authentic standards, products of 6, 8, and 9 have been identified to be the corresponding deoxygenated analogs: 2-methylquinoxaline (i.e., desoxy-6), picolinic acid and quinoline. Figure 1a showed the time course of 6 and its reduction intermediates and products. The dominant intermediate of 6 was identified to be 7 (6-N1). The concentration of desoxy-6 increased and stayed stable for up to four days, confirming it to be the final product. Though not observed, 2-methylquinoxaline 4-oxide (6-N4) is likely a minor intermediate, whose presence can explain the lower mass balance at the reaction time of around 3 11025



Article

Figure 2. HPLC/QToF-MS analyses of CDX and its reaction products in the reaction mixture: (a) LC/MS chromatogram of the reaction mixture, (b) spectrum of CDX-N1, (c) spectrum of CDX-N4, and (d) spectrum of DCDX.

were intermediates, with their concentrations increasing in the beginning and then decreasing at different rates. CDX-N4 was observed as the dominant intermediate. Kinetic modeling results also showed that CDX-N1 forms at a much slower rate yet degrades at a much faster rate than CDX-N4 (details in the SI), indicating N1−O being much more reactive than N4−O. The amount of DCDX formed consistently increased throughout the reaction period for up to 2 weeks, suggesting it to be the stable final product. Similar to the case of CDX, LC/QToF-MS analyses of ODX reaction mixtures indicate three major products: two ODX mono-N-oxides (ODX-N1 and ODX-N4 as the minor and major intermediates, respectively), and bis-desoxyolaquindox as the stable final product (Figures S4 and S5, details in the SI). UV−vis Spectra. Earlier studies have reported that comparison between the UV spectra of metal−ligand mixtures and those of the free metal ion and the free ligand can indicate the types of complexes formed between the metal ion and the ligand.35 In this study, UV−vis spectra were obtained for FeII in the presence of two compounds without any side chain (5, 6) and three compounds with side chains (CDX, ODX, 12) (SI Figure S6). Compared with the UV−vis spectra of the free ligands, no shifts were observed in FeII−5 and FeII−6; significant shifts were observed in FeII−CDX and FeII−ODX; and only a minor shift was observed in FeII−12. QSAR Analysis. To determine molecular descriptors that can serve as predictors for the reactivity of aromatic N-oxides, several molecular descriptors were calculated in water solvent, and their correlation with the rate constant of reduction was investigated through quantitative structure−activity relation-

h (see more in the Discussion section below). The reason that 6-N4 was not observed can be explained based on the kinetic analysis of the time course, that is, 6-N4 forms at a much slower rate than 6-N1 (details in the SI), which leads to its low abundance. The kinetic study of 7 also confirmed that 7 was completely transformed to desoxy-6 (SI Figure S2). LC/QToF-MS analyses of CDX reaction mixtures revealed the existence of three products based on the observation of [M + Na]+ and [M + H]+ ions, with the molecular ion of 247, 247, and 231, respectively (Figure 2). The molecular ion of CDX is m/z 263, and its major fragment (m/z 231) corresponds to the loss of −OCH3 (i.e., -m/z 32) from the side chain, consistent with our previous work.17 For the two products of m/z 247, similar mass spectra were obtained, and the major fragment m/ z 215 corresponds to the loss of −OCH3 (i.e., -m/z 32) from the side chain. As a result, the two products of m/z 247 were proposed to be CDX mono-N-oxides (CDX-N1 and CDX-N4), corresponding to the loss of one O atom (i.e., -m/z 16) from the N4 or N1 position of CDX. For the product of m/z 231, the major fragment m/z 199 corresponds to the loss of -OCH3 (i.e., -m/z 32) from the side chain. Based on our previous work,17 the product of m/z 231 should be DCDX, formed by the loss of both O atoms from the N1 and N4 positions of CDX. CDX-N4 (4) was synthesized through deoxygenation of CDX by sodium dithionite19 and was analyzed on HPLC. The product peak of m/z 247 with a later retention time was confirmed to be CDXN4. As a result, CDX-N1 peak was identified as the one with the earlier retention time. Figure 1b shows the time course of CDX and its reduction products. The results revealed that CDX-N1 and CDX-N4 11026



Article

similar trend we observed in SI Figure S8 also shows that FeL26‑ is the dominant reductant in our case. FeII/FeIII Analysis. Although not a focus of this work, we monitored the concentration of FeII and total iron in systems with various initial conditions (0.5 mM FeII, 20 μM~200 μM CDX, pH 5.70−6.50) using the ferrozine method to illustrate the role of the central metal atom during the reduction reactions. All the time courses for FeII and total iron in CDX reaction solution showed that FeII and total iron concentrations remained unchanged, whereas significant degradation occurred for CDX within 3 h (one example shown in SI Figure S9). More details of the Fe analysis are in the SI.

ships (QSARs). The reduction process of N-oxides might involve complex formation on N-oxides, electron transfer from the FeII−tiron complex to the N-oxides, or cleavage of the N− O bond as the rate-limiting step(s). So we calculated the electron affinity (EA) of the neutral N-oxides, the length of the N−O bond, and the atomic spin density of the anionic Noxides. Values for these descriptors are listed in Tables 1, SI S1 and S3−S4. EA and E1/2 describe the ability of a compound to accept electrons. These descriptors were chosen because the first electron transfer is likely the rate-determining step. Both descriptors correlate poorly with log kobs for all the N-oxides (Tables 1 and SI S1 and Figure S7), suggesting the first electron transfer is not the rate-determining step for the reduction of N-oxides. A closer examination revealed that the N-oxides with side chains (CDX, ODX, 4) have higher values of EA and were reduced much faster than the N-oxides without side chains. In addition, as compared with the mono-N-oxides (4, 7, 8, 9, 10, 11), the respective di-N-oxides (CDX, 5, 6) have higher values of EA and also react faster. These results suggest that the first electron transfer has an indirect effect on the ratelimiting step. Knowing that the N−O bond must break after reduction, we calculated its bond length for all the neutral and anionic Noxides. If bond cleavage is rate-determining, a longer (and hence a weaker) N−O bond should lead to a larger value of log k. Similar values of the calculated bond lengths (SI Table S3) for all the N-oxides indicate that bond cleavage is most likely not involved in the rate-determining step. Previous studies have shown the existence of seven resonance structures of 10 (Scheme 1).36 Electron delocalization on the aromatic ring due to resonance lowers the potential energy of the N-oxide and thus makes it more stable. The calculated atomic spin densities of all the N-oxides in their anionic forms (example shown in SI Table S4) showed that the negative charge distributed among all the atoms on the aromatic ring and on the side chain, if present, confirming the extensive delocalization within the N-oxide structures. Because of this extensive delocalization, electrons, once accepted by the side chain, can delocalize to the aromatic ring leading to further reaction (see details later). As a result, complex formation between the N-oxides and FeII−tiron complex prior to electron transfer can be either rate-limiting or at least significantly affect the rate-limiting step. FeII Speciation and Reactivity. After conducting kinetic experiments with varying pH, FeII concentration and tiron concentration and calculating the concentration of FeL26‑ at each condition using MINEQL+ 4.6 (SI Table S2), we observed kobs of CDX is highly dependent on the concentration of FeL26‑ (SI Figure S8). In solutions containing both FeII ion and tiron ligand, concentrations of various FeII species (FeII, FeOH+, Fe(OH)20, Fe(OH)3−, FeHL−, FeL2‑, and FeL26‑) (L4‑ represents the fully deprotonated tiron) can be calculated as a function of medium composition, based upon the logK values (SI Table S5) reported by Naka et al.32 When examining the influence of FeII species on the observed rate constants for the reduction of nitroaromatic compounds, Naka et al.32 reported a strong correlation between the measured values of kobs and the concentration of FeL26‑, indicating FeL26‑ is the dominant reductant. The high reactivity of FeL26‑ has been explained based on its low standard one-electron reduction potential (EH0) as compared to other FeII−tiron complexes.32 The

■

DISCUSSION Reaction Sites and Complex Formation. Based on the kinetic results from Table 1, the higher reactivity of the Noxides without side chains (5−11) versus the non-N-oxide (12) indicates the reduction occurred at the N-oxide site. Although hydrazones (>CN−N< ) can be reduced electrochemically,37 they (e.g., the side chain in CDX and 12) seem to be stable in our system. The much higher reactivity of the aromatic Noxides with side chains (CDX, ODX, 3, 4) indicates that the presence of the side chains is critical in the overall reduction by FeII-tiron, where FeL26‑ forms a stable 5- or 7-membered ring complex with the N and O atoms on the chain (Scheme 2c and d) (see below). The higher reactivity of 5 than 6 and 8 than 10 suggests that the formation of bidentate complexes between FeL26‑ and the N-oxide O atom and the O atom on the side chain (Scheme 2b) is also necessary. However, the higher reactivity of ODX than 5 suggests that the complex on the side chain (Scheme 2d) is much stronger than the one formed in 5 (Scheme 2b). Direct complexation of the N-oxides with FeL26‑ is likely weak (Scheme 2a), leading to a much slower reaction. Previous studies have reported the synthesis of solid complexes of N-oxides with metal ions;38 however, due to the weak binding between the Fe and the O, either a high metal ion concentration or 6−10 fold excess of ligand was needed to form a soluble complex or to retard the dissociation of the complex.24,39−41 CDX, ODX, and 3 contain side chains that can form 5- or 7membered ring complexes with FeL26‑; however, CDX has the highest reactivity, whereas 3 has the lowest reactivity. Similar strong complexes as shown in Scheme 2c for CDX have been observed.42 Compared to CDX, ODX forms a less stable 7membered ring complex (Scheme 2d), whose carbonyl-O and hydroxyl-O also are less electron dense than the imine-N and carbonyl-O in CDX (Scheme 2c). We believe ODX prefers to form the 7-membered ring than a 5-membered ring through the amide-N (and the hydroxyl-O) because the carbonyl-O is much more basic.43 Previous work has showed that FeIII prefers O donors to amide-N donors;44 a similar 7-membered ring was also reported to form when complexing Ni(II) with the carbonyl-O (not a 5-membered ring with the amide-N) and a hydroxyl-O.45 3 reacts the slowest likely due to the steric hindrance from the phenyl group on the side chain. The monoN-oxides (4, 7, 8, 9) are much less reactive compared with the respective di-N-oxides (CDX, 5, 6) most likely due to their lower reduction potentials.34 For example, E1/2 of 11, quinoxaline mono-N-oxide and phenazine mono-N-oxide was reported to be −1.809, −1.419, and −0.972 V vs SCE as compared with their di-N-oxides analogs (−1.616, −1.241, and −0.833 V vs SCE). 11027



Article

Scheme 1. Resonance Structures of Substituted Pyridine N-Oxide, Where X = Substituents, Adapted from ref 36.

Scheme 2. Proposed Mechanisms for the Reduction of (a) Compound 9 (b) Compound 5 (c) CDX and (d) ODX

The similar UV−vis spectra observed for FeII−5 and FeII−6 complexes, compared with those of the free ligands (SI Figure

S6), suggest a very weak complexation (if any) of FeII through the N-oxide O atoms (Scheme 2a). The significant shifts 11028



Article

observed in the UV−vis spectra of FeII−CDX and FeII−ODX as compared with those of CDX and ODX confirmed the formation of the 5- or 7-membered ring complexes. The minor shift in the UV−vis spectra of FeII-12 is most likely due to the fact that there is much less resonance on the aromatic ring of 12 as compared with those of CDX and ODX. Reaction Mechanisms. Based on the analysis of reaction products, deoxygenation is considered as the main process. It has been reported that the reduction of organic compounds by one-electron reductants, such as FeII, often yields transient free radicals.46,47 The reduction of N-oxides in organic solvents also leads to the formation of radical N-oxide anions as the intermediate,48−51 followed by deoxygenation.51,52 On the basis of the reported information and our findings, reaction schemes for the reduction of aromatic N-oxides are proposed below. For all N-oxides, complex formation between them and FeL26‑ directly affects the overall reaction kinetics. For mono-Noxides, FeL26‑ might form a complex with 9 through its N-oxide O atom (scheme 2a), or with 5 through its N-oxide O atom and the O atom on the side chain (scheme 2b). One electron is then transferred from FeL26‑ to the N-oxide and delocalized within the aromatic ring to form a radical anion. Following the stepwise transfer of a second electron, deoxygenation occurs to form the products (see more below). The reduction of CDX is initiated with the formation of a 5-membered ring by FeL26‑ complexation on the side chain; meanwhile three resonance structures exist (Va, Vb, and Vc in Scheme 2c).36 The formed inner-sphere complex facilitated one electron transfer from FeL26‑ to CDX to form a radical with three similar resonance structures (VIa, VIb, and VIc).48 Similar to 5 and 9, transfer of another electron leads to deoxygenation from the N1 or N4 position to yield two CDX mono-N-oxides. The mono-Noxides will be reduced in a similar way leading to the formation of the final stable product DCDX. The mechanism of how deoxygenation occurred following radical formation is not clear yet. In deoxygenation of quinolone N-oxide by sodium alkoxide,51 an anion was believed to form after nucleophilic addition of the alkoxide to the Noxide, which led to a single electron transfer from the alkoxide to the N-oxides. The electron was delocalized on the ring and a radical formed on the N-oxide O atom. The radical then bonded with a second alkoxide and caused the N−O bond cleavage to form quinolone and aldehyde as the products. Haddadin et al.53 reported two different deoxygenation mechanisms for quinoxaline N-oxides. For alkali induced deoxygenation, N-oxides underwent prototropic shift followed by dehydration. For sodium dithionite induced deoxygenation, the carbon at the position 2 or 3 on the ring was first nucleophilicly attacked by dithionite, followed by bond rearrangement on the intermediate leading to the deoxygenated products. Ganley et al.54 assumed that the N-oxide O atom would be protonated when the quinoxaline ring accepted one electron, causing the N−O bond to cleave to form quinoxaline and hydroxyl radicals. Deoxygenation of pyridine N-oxides can also occur when nucleophilic reagents attacked the N-oxide O atom, offering an pair of electrons to the O atom and leading to deoxygenation.55 In some cases, selective monodeoxygenation of N-oxides can occur. Sodium dithionite56 and trimethyl phosphate57 prefer to attack the more electrophilic C2 carrying electron withdrawing groups and thus remove the N1−O oxygen; strong intramolecular H-bonding forms between the N1−O oxygen and the 2-amide N so that the N4−O oxygen was removed by trimethyl

phosphite.56 In addition, previous studies have reported that the reduction of protonated N-oxides was much easier than that of nonprotonated N-oxides, due to the formation of a better leaving group (−OH than −O−) and the more electron deficient nature of the N-oxides once protonated.58,59 The values of pKa for N-oxides are around 0.79−1.29,60 therefore, the free N-oxides are all nonprotonated in our system. Weak complexation of FeII species (not necessarily FeL26‑) with the oxygen atom on the N-oxides might facilitate the reduction of the N-oxides the same way as protonation does. Indeed, formation of bonds between the metal ion and the O in Noxides decreased the NO double bond character as well as electron density on the ring,61 and facilitated the deoxygenation.48,62 In our case, we hypothesize that the strength of the N−O bond in the N-oxides is weakened in the anion radicals,34 due to the extensive delocalization between the aromatic ring and the side chain and the decreased contribution of the resonance structures E−G as shown in Scheme 1. The weak complexation of FeII species with the N-oxide O atom might also facilitate the cleavage of the N−O bond. Based on the kinetic results, the N4−O oxygen in CDX is less reactive than the N1−O oxygen, most likely due to the electron withdrawing impact of the carboxylate side chain at the C2 position. Similarly for 6, the major intermediate of 6 was identified to be 6-N1, suggesting that the oxygen was removed more favorably from the N4 position due to the electron donating effect of the C2-methyl group. A similar reduction mechanism is also proposed for ODX (Scheme 2d), where the major difference is the amide tautomerization on the side chain before electron transfer occurs.35 Redox Noninnocent Ligands (RNILs). A ligand is innocent when it allows unambiguous determination of the oxidation state of the central metal atom. A number of ligands such as o-catecholate have been considered as noninnocent.63,64 It is difficult to determine the oxidation state of the central metal atom in metal-RNIL complexes because of the difficulty deciding whether the electron is removed from or added to an orbital that is largely metal or ligand in nature. Currently there is little information available regarding redox activity of metal− RNIL complexes in aqueous transformation of organic contaminants. The unchanged concentrations of FeII and total iron in various CDX reaction solutions suggest that FeII−tiron complex is likely redox noninnocent, that is, it is difficult to determine whether the electron is removed from FeII or tiron during the reduction. Similar results have been observed in an earlier study when FeII−thiol complexes were used during the reduction of nitroaromatic contaminants.65 Here, the number of electrons donated by the complex is much more than the number of electrons FeII can give. FeII was proposed to act as an electron shuttle to transfer electrons between the contaminant and the thiol complex while by itself remained catalytic. However, the proposed mechanism has not been verified yet. As reported earlier, noninnocent ligands typically exist in three different oxidation levels in metal complexes.64 For example, o-catecholates exist as catecholato dianion, osemiquinonato monoanion, or o-benzoquinone. As a result, one-electron oxidation of a metalII-catecholato complex may yield either metalIII-catecholato or metalII-semiquinonato.66 In our case, the following two forms of Fe−tiron (Scheme 3) might coexist after reducing the N-oxides but cannot be distinguished by the Ferrozine method. Further research is 11029



Article

N-oxide) and 4,4′-azobis(pyridine N-oxide) adsorbed on silver colloidal nanoparticles. Faraday Dis. 2006, 132, 111. (9) Poptoshev, E.; Claesson, P. M. Adsorption of dimethyldodecylamine-N-oxide at the mica-solution interface studied by ellipsometry. Colloid. Surface. A 2006, 291, 45. (10) Ingerslev, F.; Toräng, L.; Loke, M. L.; Halling-Sørensen, B.; Nyholm, N. Primary biodegradation of veterinary antibiotics in aerobic and anaerobic surface water simulation systems. Chemosphere 2001, 44, 865. (11) Meister, G.; Wechsler, M. Biodegradation of N-methylmorpholine-N-oxide. Biodegradation 1998, 9, 91. (12) Brycki, B.; Brzezinski, B. Kinetics and mechanism of acid hydrolysis of 2-carboethoxypyridine N-oxides. J. Phys. Org. Chem. 1990, 3, 489. (13) Rose, M. D.; Bygrave, J.; Tarbin, J. A. Determination of quinoxaline carboxylic acid (metabolite of carbadox) in animal tissue by HPLC. Food Addit. Contam. 1995, 12, 177. (14) MacIntosh, A. I.; Neville, G. A. Liquid chromatographic determination of carbadox, desoxycarbadox, and nitrofurazones in pork tissues. J. Assoc. off. Anal. Chem. 1984, 67, 958. (15) Kurasawa, Y.; Takada, A.; Kim, H. S. Progress in the chemistry of quinoxaline N-oxides and N,N′-dioxides. J. Heterocyclic Chem. 1995, 32, 1085. (16) Shaforost, K. V.; Ryzhakov, A. V.; Rodina, L. L. 4-azidoquinoline N-oxide: Synthesis and photolysis. Russ. J. Org. Chem. 2008, 44, 728. (17) Zhang, H.; Huang, C. H. Reactivity and transformation of antibacterial N-oxides in the presence of manganese oxide. Environ. Sci. Technol. 2005, 39, 593. (18) Chandrasekhar, S.; Raji Reddy, C.; Jagadeeshwar Rao, R.; Madhusudana Rao, J. Efficient and chemoselective deoxygenation of amine N-oxides using polymethylhydrosiloxane. Synlett. 2002, 349. (19) Massy, D. J. R.; McKillop, A. Selective monodeoxygenation of methyl 3-(2-quinoxalinylmethylene)carbazate N1,N4-dioxide (carbadox). Synthesis 1996, 1477. (20) Garvey, R. G.; Nelson, J. H.; Ragsdale, R. O. The coordination chemistry of aromatic amine N-oxides. Coordin. Chem. Rev. 1968, 3, 375. (21) Karayannis, N. M.; Speca, A. N.; Chasan, D. E.; Pytlewski, L. L. Coordination complexes of the N-oxides of aromatic diimines and diazines. Coord. Chem. Rev. 1976, 20, 37. (22) Torre, M. H.; Gambino, D.; Araujo, J.; Cerecetto, H.; González, M.; Lavaggi, M. L.; Azqueta, A.; López de Cerain, A.; Vega, A. M.; Abram, U.; Costa-Filho, A. J. Novel Cu(II) quinoxaline N1,N4-dioxide complexes as selective hypoxic cytotoxins. Eur. J. Med. Chem. 2005, 40, 473. (23) Prabhakaran, C. P.; Patel, C. C. Antipyrine and pyridine N-oxide complexes of iron(II). J. Inorg. Nucl. Chem. 1972, 34, 3485. (24) Nelson, J. H.; Ragsdale, R. O. Complexes of aromatic amine oxides: Nickel(II) complexes of 6-substituted quinoline N-oxides. Inorg. Chim. Acta 1968, 2, 230. (25) Nelson, J. H.; Nathan, L. C.; Ragsdale, R. O. Complexes of aromatic amine oxides. 4-Substituted quinoline 1-oxide complexes of cobalt(II) and nickel(II) perchlorates. Inorg. Chem. 1968, 7, 1840. (26) Boyd, S. A.; Kohrman, R. E.; West, D. X. Transition metal ion complexes of 2-picolinate N-oxide. J. Inorg. Nucl. Chem. 1976, 38, 607. (27) Boyd, S. A.; Kohrman, R. E.; West, D. X. Metal ion complexes of 2-pyridylcarbinol N-oxide. J. Inorg. Nucl. Chem. 1976, 38, 1605. (28) Boyd, S. A.; Kohrman, R. E.; West, D. X. Cationic lanthanide complexes of 2-picolinic acid N-oxide. Inorg. Nucl. Chem. Lett. 1976, 12, 603. (29) Ojima, H.; Yamada, K. Syntheses and characterization of μBis{N-(oxopropyl)-glycinamidato}-dicopper(II) and μ-Bis{n-(oxopropyl)-DL-alaninamidato}-dicopper(II). Z. Anorg. Allg. Chem. 1970, 379, 322. (30) Strathmann, T. J.; Stone, A. T. Reduction of oxamyl and related pesticides by FeII: Influence of organic ligands and natural organic matter. Environ. Sci. Technol. 2002, 36, 5172.

Scheme 3. Resonance of Fe−tiron Complex As a Redox Noninnocent Ligand

warranted to elucidate the noninnocent behavior of Fe−tiron complexes in aqueous solution. Overall, the above mechanistic understanding of the aqueous reduction of aromatic N-oxides will provide basis for developing QSARs that can be used to predict reductive transformation of various N-oxides. In addition, this is the first study to briefly apply the concept of noninnocent ligands to the reductive transformation of organic contaminants in the environment. Future work is warranted to further explore the behavior of RNILs in aqueous redox-active environments.

■

ASSOCIATED CONTENT

■

AUTHOR INFORMATION

S Supporting Information *

Additional figures, tables, cyclic voltammetry results, as well as kinetic modeling results. This material is available free of charge via the Internet at http://pubs.acs.org. Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This material is based upon work supported by the National Science Foundation under Grant CBET-1125713. We are thankful to Drs. William Wuest and Hongwen Zhou at the Temple University for assistance in analyzing HPLC/QToFMS samples, to Dr. Spiridoula Matsika for assistance in computational calculations, and to Drs. Eric Borguet and Zhihai Li for assistance in cyclic voltammetry measurement.

■

REFERENCES

(1) Albini, A.; Pietra, S. Heterocyclic N-oxides; CRC Press: Boston, 1991. (2) Katritzky, A. R.; Lagowski, J. M. Chemistry of the heterocyclic Noxides; Academic Press: New York, 1971. (3) Gorrod, J. W.; Beckett, A. H. Drug Metabolism in Man; Taylor & Francis Ltd.: London, 1978. (4) JECFA 60th meeting of the joint FAO/WHO expert committee on food additives. WHO Food Addit. Ser. 2003, No. 51, 49. (5) Zou, J.; Chen, Q.; Tang, S.; Jin, X.; Chen, K.; Zhang, T.; Xiao, X. Olaquindox-induced genotoxicity and oxidative DNA damage in human hepatoma G2 (HepG2) cells. Mutat. Res. 2009, 676, 27. (6) Bickel, M. H. The pharmacology and biochemistry of N-oxides. Pharmacol. Rev. 1969, 21, 325. (7) Strock, T. J.; Sassman, S. A.; Lee, L. S. Sorption and related properties of the swine antibiotic carbadox and associated N-oxide reduced metabolites. Environ. Sci. Technol. 2005, 39, 3134. (8) Pergolese, B.; Muniz-Miranda, M.; Sbrana, G.; Bigotto, A. Surface-enhanced Raman scattering investigations of 4-nitro(pyridine 11030



Article

(31) Bussan, A. L.; Strathmann, T. J. Influence of organic ligands on the reduction of polyhalogenated alkanes by iron(II). Environ. Sci. Technol. 2007, 41, 6740. (32) Naka, D.; Kim, D.; Strathmann, T. J. Abiotic reduction of nitroaromatic compounds by aqueous iron(II)-catechol complexes. Environ. Sci. Technol. 2006, 40, 3006. (33) Washington, J. W.; Endale, D. M.; Samarkina, L. P.; Chappell, K. E. Kinetic control of oxidation state at thermodynamically buffered potentials in subsurface waters. Geochim. Cosmochim. Acta. 2004, 68, 4831. (34) Kubota, T.; Nishikida, K.; Miyazaki, H.; Iwatani, K.; Oishi, Y. Electron spin resonance and polarographic studies of the anion radicals of heterocyclic amine N-oxides. J. Am. Chem. Soc. 1968, 90, 5080. (35) Huang, C. H.; Stone, A. T. Transformation of the plant growth regulator daminozide (Alar) and structurally related compounds with CuII ions: Oxidation versus hydrolysis. Environ. Sci. Technol. 2003, 37, 1829. (36) Ochiai, E. Recent Japanese work on the chemistry of pyridine 1oxide and related compounds. J. Org. Chem. 1953, 18, 534. (37) Soucaze-Guillous, B.; Lund, H. Reduction of hydrazones of aromatic carbonyl compounds in aprotic media. J. Electroanal. Chem. 1997, 423, 109. (38) Byers, W.; Fa-Chun Chou, B.; Lever, A. B. P.; Parish, R. V. Charge-transfer spectra of pyridine N-oxide metal complexes. Determination of optical electronegativities. J. Am. Chem. Soc. 1969, 91, 1329. (39) Reddy, K. V.; Jacobson, A. R.; Kung, J. I.; Sayre, L. M. Divalent metal Ion catalyzed hydrolysis of picolinanilides. Inorg. Chem. 1991, 30, 3520. (40) Nathan, L. C.; Ragsdale, R. O. Chromium(III), manganese(II), iron(III), and zinc(II) complexes of some substituted aromatic amine N-oxides. Inorg. Chim. Acta 1969, 3, 473. (41) Ryzhakov, A. V.; Andreev, V. P. Coordination of pyridine Noxides with alkali and alkaline-earth metal cations in aqueous solution. Russ. J. Gen. Chem. 2005, 75, 121. (42) Chen, W. R.; Liu, C.; Boyd, S. A.; Teppen, B. J.; Li, H. Reduction of carbadox mediated by reaction of Mn(III) with oxalic acid. Environ. Sci. Technol. 2013, 47, 1357. (43) Angus, P. M.; Fairlie, D. P.; Jackson, W. G. Synthesis, solution structure, and reactivity of oxygen-bound amides on cobalt(III). Inorg. Chem. 1993, 32, 450. (44) Farkas, E.; Csóka, H. Solution equilibrium studies on metal complexes of 2,3-dihydroxy-phenylalanine-hydroxamic acid (Dopaha) and models: Catecholate versus hydroxamate coordination in iron(III)-, aluminium(III)- and molybdenum(VI)-dopaha complexes. J. Inorg. Biochem 2002, 89, 219. (45) Blakeley, R. L.; Treston, A.; Andrews, R. K.; Zerner, B. Nickel(II)-promoted ethanolysis and hydrolysis of N-(2pyridylmethyl)urea. A Model for Urease. J. Am. Chem. Soc. 1982, 104, 612. (46) Strathmann, T. J.; Stone, A. T. Reduction of the carbamate pesticides oxamyl and methomyl by dissolved FeII and CuI. Environ. Sci. Technol. 2001, 35, 2461. (47) Bakac, A.; Butković, V.; Espenson, J. H.; Lovrić, J.; Orhanović, M. Electron transfer from aliphatic radicals to ring-substitied pyridine N-oxides and their cobalt-ammine complexes in aqueous solution. Inorg. Chem. 1996, 35, 5168. (48) Kurbatova, A. S.; Kurbatov, Y. V.; Niyazova, D. A. Chemistry of heterocyclic N-oxides and related compounds. 11. Reaction of pyridine N-oxide with metal ketyls and benzophenone dianions. Chem. Heterocycl. Compd. 1979, 15, 531. (49) Bruni, P.; Colonna, M. Nitroxide and anion radicals derived from isatogen and related indole derivatives. Tetrahedron 1973, 29, 2425. (50) Janzen, E. G.; Happ, J. W. Electron spin resonance of substituted pyridine N-oxide radical anions. Aromatic nitroxides. J. Phys. Chem. 1969, 73, 2335.

(51) Bjørsvik, H. R.; Gambarotti, C.; Jensen, V. R.; González, R. R. A novel efficient deoxygenation process for N-heteroarene N-oxides. J. Org. Chem. 2005, 70, 3218. (52) Yadav, J. S.; Reddy, B. V. S.; Reddy, m. M. Indium-mediated deoxygenation of amine-N-oxides in aqueous media. Tetrahedron Lett. 2000, 41, 2663. (53) Haddadin, M. J.; Zahr, G. E.; Rawdah, T. N.; Chelhot, N. C.; Issidorides, C. H. Deoxygenation of quinoxaline N-oxides and related compounds. Tetrahedron 1974, 30, 659. (54) Ganley, B.; Chowdhury, G.; Bhansali, J.; Daniels, J. S.; Gates, K. S. Redox-activated, hypoxia-selective DNA cleavage by quinoxaline 1,4-di-N-oxide. Bioorg. Med. Chem. 2001, 9, 2395. (55) Jr., E. H.; Olszewski, W. F. The reaction of triphenylphosphine with some aromatic amine oxides. J. Am. Chem. Soc. 1959, 81, 1483. (56) El Abadelah, M. M.; Sabri, S. S.; Tashtoush, H. I. Selective monodeoxygenation of quinoxaline-amino acid and ester dioxides. Tetrahedron Lett. 1979, 35, 2571. (57) Dirlam, J. P.; McFarland, J. W. Selective monodeoxygenation of certain quinoxaline 1,4-dioxides with trimethyl phosphite. J. Org. Chem. 1977, 42, 1360. (58) Kubota, T.; Miyazaki, H. The effect of substituents on the halfwave potential of the polarographic reduction of pyridine N-oxide derivatives. B. Chem. Soc. Jpn. 1966, 39, 2057. (59) Baymak, M. S.; Celik, H.; Lund, H.; Zuman, P. Reduction of diprotonated form of aryl hydrazones. J. Electroanal. Chem. 2006, 589, 7. (60) Kubota, T.; Miyazaki, H. Polarography of pyridine N-oxide and its alkyl derivatives. B. Chem. Soc. Jpn. 1962, 35, 1549. (61) Kida, S.; Quagliano, J. V.; Walmsley, J. A.; Tyree, S. Y. Infra-red studies of pyridine N-oxide and its Co(II), Ni(II), Cu(II), Zn(II), Al(III), Cr(III), Fe(II), Fe(III) and Sn(IV) complexes in the 3−15 μ region. Spectrochim. Acta 1963, 19, 189. (62) Yoo, B. W.; Choi, J. W.; Yoon, C. M. A facile and efficient deoxygenation of amine-N-oxides with Mo(CO)6. Tetrahedron Lett. 2006, 47, 125. (63) Pierpont, C. G.; Lange, C. W. The chemistry of transition metal complexes containing catechol and semiquinone ligands. In Prog. Inorg. Chem.; Karlin, K. D., Ed.; John Wiley & Sons: New York, 1994; p 331. (64) Butin, K. P.; Beloglazkina, E. K.; Zyk, N. V. Metal complexes with non-innocent ligands. Russ. Chem. Rev. 2005, 74, 531. (65) Naka, D.; Kim, D.; Carbonaro, R. F.; Strathmann, T. J. Abiotic reduction of nitroaromatic contaminants by iron(II) complexes with organothiol ligands. Environ. Toxicol. Chem. 2008, 27, 1257. (66) Ghosh, P.; Begum, A.; Bill, E.; Weyhermuller, T.; Wieghardt, K. Molecular and electronic structures of iron(II)/(III) complexes containing N,S-coordinated, closed-shell o-aminothiophenolato(1-) and o-iminothiophenolato(2-) ligands. Inorg. Chem. 2003, 42, 3208.

11031


Complexation Facilitated Reduction of Aromatic N-Oxides by Aqueous

Recommend Documents