Environ. Sci. Technol. 2008, 42, 6656–6662
Oxidation of Diclofenac with Ozone in Aqueous Solution M Y I N T M Y I N T S E I N , * ,† M A R C O Z E D D A , †,‡ JOCHEN TUERK,‡ TORSTEN C. SCHMIDT,† ALFRED GOLLOCH,† AND CLEMENS VON SONNTAG§ Instrumental Analytical Chemistry, University of Duisburg-Essen, Lotharstr. 1, 47048 Duisburg, Germany, Institute of Energy and Environmental Technology (IUTA), Bliersheimerstr. 60, 47229 Duisburg, Germany, and Max Planck Institute for Bioinorganic Chemistry, Stiftstr. 34-36, 45413 Mu ¨ lheim an der Ruhr, Germany
Received March 27, 2008. Revised manuscript received May 30, 2008. Accepted June 5, 2008.
Ozonation of diclofenac in aqueous solution in the presence and absence of an •OH scavenger, tertiary butanol (t-BuOH), was studied, and the most important reaction intermediates and products were identified. The second-order O3 rate constant was determined by competition with buten-3-ol and was found to be 6.8 × 105 M-1 s-1 at 20 °C. From this high rate constant, it has been concluded that O3 must initially add on the amino nitrogen. Decomposition of the adduct results in the formation of O3•- (f •OH) and aminyl radical precursors. A free •OH yield of 30% was estimated based on the HCHO yields generated upon reaction of •OH with 0.01 M t-BuOH. Almost all diclofenac reacted when the molar ratio of O3/diclofenac was ∼5:1 in the presence of t-BuOH and ∼8:1 in its absence. As primary reaction products (maximum yield), diclofenac-2,5-iminoquinone (32%), 5-hydroxydiclofenac (7%), and 2,6-dichloroaniline (19%) were detected with respect to reacted diclofenac in the presence of t-BuOH. These primary products degraded into secondary ones when the O3 dose was increased. In the •OHmediated reaction (absence of t-BuOH) small yields of 5-hydroxydiclofenac (4.5%), diclofenac-2,5-iminoquinone (2.7%), and 2,6-dichloroaniline (6%) resulted. Practically all Cl- (95%) was released in the absence of t-BuOH but only about 45% in the presence of t-BuOH at an O3/diclofenac molar ratio of 10: 1. Based on the reaction products, mechanisms that may account for the high O3 consumption during ozonation of diclofenac are suggested. For technical applications, adequate supply of O3 is needed not only to eliminate diclofenac, but also for the degradation of its potentially toxic products like diclofenac2,5-iminoquinone and 5-hydroxydiclofenac.
Introduction Diclofenac, a nonsteroidal antiinflammatory drug, is one of the most commonly used pain killers, clinically largely used as sodium salt. About 86 t/year are prescribed in Germany (1). It is metabolized mainly in the liver (2), and only a low percentage is excreted (3). Biological treatment of wastewater * Corresponding author e-mail:
[email protected]. † University of Duisburg-Essen. ‡ Institute of Energy and Environmental Technology (IUTA). § Max Planck Institute for Bioinorganic Chemistry. 6656
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 17, 2008
in conventional municipal wastewater treatment plants (WWTPs) barely eliminates this drug (4, 5). It is therefore generally detected in their effluents in concentrations up to 10 µg/L (6–9). A maximum concentration of 28.4 µg/L has been reported in one study (10). It is also detected in groundwater at concentrations up to 0.59 µg/L (11), and in surface water, despite its relatively high photolability (12, 13), at concentrations up to 15 µg/L (14). Diclofenac causes renal lesions in the kidneys and alterations of the gills in rainbow trout with a lowest observed effect concentration (LOEC) of 5 µg/L and a no observed effect concentration (NOEC) of 1 µg/L (15). Diclofenac is readily removed from wastewater by activated carbon (16), and by ozonation (17, 18). There are also some studies using advanced oxidation processes (AOPs) (17–24) and the reactions of diclofenac with O3 in aqueous solution (19, 20). However, a clear distinction as to the primary and potentially secondary products, their yields, and, precursor radicals is still missing. Furthermore, there is a major discrepancy between the two reported O3 rate constants for diclofenac, ∼106 M-1s-1 (18) and 1.8 × 104 M-1 s-1 (20). These two values differ by a factor of approximately 50, and the suggested reaction sites of O3 attack are also different, i.e., the nitrogen (18) vs the aromatic ring (20). The main objectives of this study were (a) to determine the reaction kinetics for the diclofenac-O3 reaction by a competition reaction to facilitate identification of the preferred site of O3 attack, (b) to identify the major intermediates, primary and secondary products to enable postulation of the reaction mechanisms for the reactions in the presence and absence of a radical scavenger, and (c) to evaluate the minimum O3 demand for complete removal of diclofenac and its potentially toxic products.
Materials and Methods Diclofenac (sodium salt) from Heumann PCS (Feucht, Germany), 5-hydroxydiclofenac from Ciba-Geigy (Basel, Switzerland), tertiary butanol (t-BuOH) (99%, Merck), buten3-ol (Merck), and 2,6-dichloroaniline (98%, Aldrich) were used as received. Stock solutions were made up in Milli-Q Plus water. Ozone was generated by an O2-fed ozonator (BMT 802X, BMT Messtechnik, Berlin). O3 stock solution was produced by sparging O3 containing O2 through Milli-Q water, and its concentration was determined spectrophotometrically (PerkinElmer Lambda 25) taking (260 nm) ) 3300 M-1 cm-1 (25). The pH of experimental solutions was measured by a Metrohm 780 pH meter. Oxidation of Diclofenac by Ozone. One mL of diclofenac stock solution (1.3 mM), with and without 5 mL of t-BuOH solution (0.1 M), were mixed with varying volumes of water and of O3 stock solution (∼0.36 mM) in the absence of an added buffer. The initial concentration of diclofenac was set at 26 µM (8 mg/L). Final molar ratios of O3/diclofenac varied between 1:10 and 10:1. All samples were prepared in triplicate. Before product determination, completion of the O3 reaction was verified by the indigo method (26), and the change in pH was measured. Determination of Diclofenac Consumption and Product Formation. Diclofenac consumption and product formation were determined by HPLC with diode array detection (DAD) (Knauer K-1001) on a Nucleodur 100-5 C18 end-capped (250 × 3 mm) column (Macherey-Nagel) fitted with a ChromCart (8 mm × 3 mm) precolumn using acetonitrile (eluent A) and water pH 4.6, adjusted with dilute formic acid (eluent B) (above the pKa value of 4.2 for the carboxyl moiety of 10.1021/es8008612 CCC: $40.75
2008 American Chemical Society
Published on Web 07/29/2008
diclofenac) as mobile phases at 30 °C. The DAD was regularly set at 204 nm. In addition, UV spectra were recorded between 200 and 500 nm. A good separation of diclofenac and reaction products was achieved with a gradient program: 5-43% A within 10 min with a flow rate of 0.3 mL min-1. Run time was 55 min. Calibration was made using diclofenac-Na standard solutions of 1-26 µM. Standard stock solutions of the reference compounds 5-hydroxydiclofenac and 2,6-dichloroaniline were prepared in acetonitrile and diluted with water (0.2-10 µM). Ascorbic acid of approximately equimolar concentration was added to the 5-hydroxydiclofenac solutions to prevent oxidation. At an injection volume of 20 µL, the detection limit of diclofenac and the reference compounds was 0.1 µM. Identification of the products was made by HPLC (Shimadzu LC20AD) coupled with MS (3200 Q Trap, Applied Biosystems, Darmstadt, Germany) in the negative mode of turbo-electrospray ionization using the same chromatographic conditions as described above. The UV detector was also set at 204 nm. The ion source temperature was 550 °C and ion spray voltage -4500 V. Nitrogen was used as curtain and nebulizer gas. For characterization of •OH production in the reaction of diclofenac with ozone, formaldehyde yield generated upon reaction of •OH with 0.01 M t-BuOH was quantified as 2,4dinitrophenylhydrazone (27) by HPLC (Shimadzu LC-20AT) at 350 nm using a recently reported procedure (28). Chloride ion was quantified by ion chromatography (ICS90, Dionex) on an Ion Pac AS4A-SC 250 × 4 mm column with 1.8 mM Na2CO3/1.7 mM NaHCO3 as eluent at 1 mL/min. A 10 µL aliquot of the sample solution was injected onto the column. Calibration was made using 10-50 µM standard NaCl solutions. Cl- was eluted after 4.3 min. The detection limit was 0.5 µM. Isolation and Characterization of Diclofenac-2,5-iminoquinone. Diclofenac-2,5-iminoquinone generated from O3/diclofenac (4:1) in the presence of t-BuOH was extracted on AccuBond SPE ODS-C18 (Agilent, 200 mg/3 mL, conditioned with 5 mL methanol followed by 5 mL of pH 2 water). After adjusting to pH 2 with 0.1 M HCl for suppressing diclofenac ionization, the sample (250 mL) was introduced onto the SPE cartridge at a flow rate of 10 mL min-1. After washing with water (5 mL), the dark-red substance was extracted with methanol (2 × 3 mL). The eluent was brought to dryness under a stream of N2, and the residue was redissolved in 1 mL of the mobile phase for HPLC-MS. The UV spectrum of the extract was recorded in methanol. The IR spectrum (ATR) was run on a FT-IR spectrophotometer (PerkinElmer 2000) in the range of 4000-600 cm-1. 1H NMR spectrum in CDCl3 was recorded on a Bruker DRX-500 MHz NMR spectrometer. Competition Kinetics. The rate constant of O3 with diclofenac was determined by competition with buten-3-ol at pH 7.0 (0.01 M phosphate buffer, 20 °C). The concentrations of diclofenac (0.25 mM) and O3 (0.1 mM) were kept constant, and the buten-3-ol concentration was varied in order to have final diclofenac/buten-3-ol ratios of 0.05-1.7. Formaldehyde formed from the reactions of buten-3-ol with O3 (CH2dCHsCHOHsCH3 + O3 + H2O f CH2O + H2O2 + HC(O)sCHOHsCH3), was determined spectrophotometrically by the Hantzsch method (29), using (412 nm) ) 7700 M-1 cm-1, as described in ref 30.
Results and Discussion Determination of the Rate Constant for the Direct Reaction of Ozone with Diclofenac. Because of the large discrepancy of the reported values of the O3 rate constant with diclofenac and mechanistic implications resulting from the magnitude of this value, its redetermination was made by an independent method. For this, the competitor of choice was buten-3-ol
(kbuten-3-ol ) 7.9 × 104 M-1 s-1 (31)). In the reaction of O3 with buten-3-ol, one mol of formaldehyde is formed per mol of O3 (see Materials and Methods) whereas no formaldehyde is generated in the reaction of O3 with diclofenac. The secondorder rate constant of O3 with diclofenac (kdiclofenac) can be calculated by determining the formaldehyde concentrations at a constant O3 concentration in the absence of diclofenac, [CH2O]0, and in its presence, [CH2O], based on eq 1 (32). At a constant O3 concentration, the formaldehyde yield decreases with increasing diclofenac/buten-3-ol ratio. [CH2O]0/[CH2O] ) 1 + kdiclofenac[diclofenac]/kbuten-3-ol[buten-3-ol] (1) The slope of the straight line in the plot of [CH2O]0/[CH2O] - 1 vs [diclofenac]/[buten-3-ol] gives 8.6 (0.15 standard error) for kdiclofenac/kbuten-3-ol, and based on the known kbuten-3-ol value, kdiclofenac is calculated as 6.8 × 105 M-1 s-1 at 20 °C (for data, see the Supporting Information. There, a determination based on second-order kinetics is also shown). This value agrees well with the reported estimated value of 106 M-1 s-1 at 20 °C and pH 7 (18), whereas the other reported value of 1.8 × 104 M-1 s-1 at 25 °C and pH 6.0 (20) is markedly lower. Since the pKa value of the amino group is CdO absorption at 1700 cm-1 which is not found in the IR spectrum of diclofenac (>CdO of the sCOOH group appears at 1600 cm-1). The negative ion LC-MS spectra show m/z at 308/ 310/312 [M-H]-, 290/292/294 [M-H-H2O]-, 262/264/266 [M-H-H2OsCO]- with the expected intensities for compounds containing two chlorine atoms, and 228/230/232 (100%) (additional loss of sCl). This fragmentation indicates the presence of hydroxyl, carbonyl, carboxyl groups, and chlorine atoms in the molecule. The presence of the nonchlorine containing ion at m/z 166 strongly suggested a quinone iminium ion. The chemical shifts (δ) in ppm and multiplicity from 1H NMR spectra in CDCl3: CH2 (3.8, s), H4 (6.5, d), H6 (6.7, dd), H3 (7.05, t), H4′ (7.25, s), H3′ and H5′ (7.35, d) (numbering of H atoms, see (I)) are identical to the reported data of diclofenac-2,5-iminoquinone (22, 44–46), which may have a role in diclofenac hepatotoxicity in humans and rats (44, 45). The reported molar absorption coefficient (270 nm) ) 13 500 M-1 cm-1 for the pure diclofenac-2,5-iminoquinone extract (46) was used for quantification (MS, IR, and NMR spectra, see the Supporting Information). Formation and degradation profiles of diclofenac-2,5iminoquinone (III), 2,6-dichloroaniline (IV), and 5-hydroxydiclofenac (V) as a function of O3/diclofenac molar ratio are depicted in Figure 1 (B). In the presence of t-BuOH, iminoquinone (III) (32% with respect to reacted diclofenac) was formed as a main product and 2,6-dichloroaniline (IV) (19%) and 5-hydroxydiclofenac (V) (7%) were formed as minor ones at a 3:1 molar ratio of O3/diclofenac. In the absence of t-BuOH, the highest yields of iminoquinone (III) (2.7%), 2,6dichloroaniline (IV) (6%), and 5-hydroxydiclofenac (V) (4.5%) were formed at a molar ratio of 5:1 (Figure 1 (B)). These primary products are subsequently degraded as the molar ratio increased. Besides these products, some unidentified byproduct, peaks 1-3, were also formed. ESI (--) from peak 1 (RT 8.0 min), peak 2 (RT 10.1 min), and peak 3 (RT 12.7 min) extracted from EMS-scan show m/z ) 160/162/164, which indicates the presence of two chlorine atoms in the molecules according to their relative intensities. Peaks 2 and 3 were detected only in the presence of t-BuOH. The
FIGURE 2. Chloride release in the reaction of diclofenac (26 µM) with ozone as a function of O3/diclofenac molar ratio in the presence of 0.01 M t-BuOH (closed squares) and absence of t-BuOH (open squares). concentrations of these compounds increase as the molar ratio increases, that is, they are resistant to O3 attack. Formation of Chloride Ion. Upon degradation of diclofenac and its primary reaction products, Cl- is released. Clyield increases as the molar ratio of O3 to diclofenac was increased (Figure 2). Up to 95% diclofenac was dechlorinated (49 µM Cl- was formed) at a molar ratio of O3/diclofenac 10:1 in the absence of t-BuOH, whereas only 45% of initial chlorine was liberated, yielding free Cl- in solution (23 µM) in the presence of t-BuOH. In the latter case, there is practically no 2,6-dichloronaniline left, that is, there must be other, more persistent (O3-refractory), chlorine-containing compounds, such as peaks 1-3 in Figure 1 (A), present. This was confirmed by the investigation of the oxidation products as discussed above. Proposed Mechanisms of Reactions Involving the Diclofenac Aminyl Radical. Concomitantly with the •OH radical, the diclofenac aminyl radical (II) is formed (reaction 8). One of its decay routes is given in the cage reaction 11. Addition of O3•- to the para-position of the activated ring of the aminyl radical (II) would give rise to 5-hydroxydiclofenac (V) as the minor product in the O3-based reaction that is depicted in reactions 11-13. The formation of (V) was also reported in diclofenac toxicity studies (44, 45). Aminyl radicals are reminiscent of other heteroatom-centered radicals, such as the guanyl and adenyl radicals, formed in •OH radical
VOL. 42, NO. 17, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6659
reactions of the nucleobases guanine and adenine (47). The latter do not react with O2 and apparently largely form dimers or even oligomers. Although these have never been characterized, a similar dimer formation in the case of diclofenac aminyl radicals could well explain the incomplete mass balance in the absence of t-BuOH since only the fraction of the aminyl radicals that does not dimerize but is rather trapped by other radicals can be accounted for. Here, the major detected product is diclofenac-2,5iminoquinone (III). As it is formed in comparatively high yields only in the presence of t-BuOH, the t-BuOH-derived peroxyl radicals (reactions 14 and 15) must be involved under such conditions. A possible mechanism is shown in reactions 16-18. ·
OH + (CH3)3COH f H2O + ·CH2C(CH3)2OH
(14)
·CH2C(CH3)2OH + O2 f ·OOCH2(CH3)2OH
(15)
The formation of diclofenac-2,5-iminoquinone (III) found in the absence of t-BuOH may be explained by a reaction of the aminyl radical (II) with O3 (reactions 19-22). The aminyl
radical (II) formed reacts with O3 especially at the paraposition to nitrogen on the more electron-rich aromatic ring having the electron-donating group (CH2COO-). A 1,2-H shift gives rise to a hydroxycyclohexadienyl radical, and subsequent oxidation by O2 leads to the formation of III. The detailed mechanism of the oxidation of hydroxycyclohexadienyl radicals and competing reactions has been reviewed (47). Proposed Mechanisms of Diclofenac Oxidation by •OH. The •OH radical, on account of its electrophilic nature, will preferentially attack at activated (i.e., electron-dense) sites of aromatic rings often resulting in the formation of stable hydroxylated compounds via transition through an unstable carbon-centered radical. Reactions 23- 25 show the formation of 5-hydroxydiclofenac (V) as an expected major product in the •OH-mediated reaction (absence of t-BuOH). Upon
6660
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 17, 2008
addition of •OH, a cyclohexadienyl-type radical is formed (reaction 23). Such radicals react reversibly with O2 (note that O2 is always present in excess over O3 in O3 reactions), but with electron-donating substituents as prevail here the forward reaction (reaction 24) is close to diffusion controlled, and the reverse reaction may be neglected (48). In competition with other reactions, HO2• is eliminated (reaction 25) (for a detailed study on the parent, benzene, see ref 49), and 5-hydroxydiclofenac (V) is formed. Diclofenac must also become hydroxylated at other positions, and evidence for this has been obtained previously by GC/MS (20). Scavenging of •OH by t-BuOH must reduce the yields of these key products, and this is indeed observed. Another important •OH-related product is 2,6-dichloroaniline (IV). Its formation can be accounted for by an •OH addition at the ipso-position of the amino group (reaction 26). Such adducts are known to release the substituent rapidly (reaction 27). The phenoxyl-type radical which is formed besides 2,6dichloroaniline in this reaction does not react with O2, but will add to other radicals present (50).
In addition to the chlorine-containing positions of diclofenac, 2,6-dichloroaniline, or other chlorinated target compounds, will cause Cl- release. Its release may be taken as an approximate measure of mineralization of diclofenac. It is seen in Figure 2 that Cl- release is substantially suppressed in the presence of t-BuOH, and this is a strong indication that •OH must be involved to a large extent in this process. Reactions 28 and 29 may account for this. As with other geminal chlorohydrines, the HCl elimination reaction 29 must be very fast (k29 > 105 s-1) (51). This type of reaction
may also take part as a cage reaction with aminyl and O3•radicals as partners (reactions 30 and 31). Ortho-quinones
are considerably less stable than para-quinones, and it is conceivable that the second chloride is hydrolytically cleaved in subsequent reactions. As in another similar case, •OH with pentachlorophenol (52), no products were detected by HPLC. Environmental Implications. From the data obtained from the present study, one can predict that diclofenac can be efficiently removed from drinking water and wastewater by O3, in agreement with an experimental study (17), due to its relatively high second-order rate constant of 6.8 × 105 M-1 s-1. Compounds reacting with O3 at k > 104 M-1 s-1 are reported to be eliminated from drinking water at doses of ca. 1 mg/L (53) and from wastewater at about 5 times that dose (17). According to Figure 1 (A), a minimum molar O3 to diclofenac ratio of ∼10:1 is necessary to completely degrade diclofenac, as well as its potentially toxic primary products, in the presence or absence of an •OH scavenger. Inadequate supply of O3 may not fully eliminate these products in the treated water. For the maximum predicted amount of diclofenac (10 µg/L) in wastewater effluents (10), the calculated minimum amount of O3 needed is 0.016 mg/L for
the 10:1 molar ratio of O3/diclofenac to completely eliminate diclofenac and its reactive potentially toxic products. However, in wastewater, the O3 dose must be substantially higher because the water matrix constituents, mainly its DOC content, will consume the major part of O3/•OH. The formation of the main product diclofenac-2,5-iminoquinone, III, is enhanced in the presence of organic •OH scavenger (t-BuOH). In wastewater, the constituents such as DOM and other micropollutants will act similarly. From the data in Figure 1, one can roughly estimate that the rate constant of O3 with III is about 5% of that of diclofenac (see the Supporting Information). This rate constant would be still high enough to eliminate III from the wastewater by treatment with O3. Due to the high rate constant of O3 with diclofenac, •OH, generated in the reaction of O3 with the water matrix, contribute
