Chemosphere 117 (2014) 316–323
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Degradation pathways of lamotrigine under advanced treatment by direct UV photolysis, hydroxyl radicals, and ozone Olya S. Keen a,c,⇑, Imma Ferrer b, E. Michael Thurman b, Karl G. Linden a a
Department of Civil, Environmental and Architectural Engineering, University of Colorado, UCB 428, Boulder, CO 80309, United States Center for Environmental Mass Spectrometry, University of Colorado, UCB 428, Boulder, CO 80309, United States c Department of Civil and Environmental Engineering, University of North Carolina, 9201 University City Blvd, Charlotte, NC 28223, United States b
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Lamotrigine was exposed to ozone,
hydroxyl radical and direct photolysis. Degradation products in each process were identified. Hydroxyl radical was the most efficient degradation pathway. Quantum yield and reaction rate constants were measured. Novel mass spectrometry friendly, bench-scale ozone quenching method is proposed.
a r t i c l e
i n f o
Article history: Received 5 February 2014 Received in revised form 20 July 2014 Accepted 27 July 2014
Handling Editor: Hyunook Kim Keywords: Lamotrigine Advanced oxidation Hydroxyl radical reaction rate constant Quantum yield Ozone quenching Mass spectrometry
a b s t r a c t Lamotrigine is recently recognized as a persistent pharmaceutical in the water environment and wastewater effluents. Its degradation was studied under UV and ozone advanced oxidation treatments with reaction kinetics of lamotrigine with ozone (4 M 1 s 1), hydroxyl radical [(2.1 ± 0.3) 109 M 1 s 1] and by UV photolysis with low and medium pressure mercury vapor lamps [quantum yields 0 and (2.7 ± 0.4) 10 4 respectively] determined. All constants were measured at pH 6 and at temperature 20 °C. The results indicate that lamotrigine is slow to respond to direct photolysis or oxidation by ozone and no attenuation of the contaminant is expected in UV or ozone disinfection applications. The compound reacts rapidly with hydroxyl radicals indicating that advanced oxidation processes would be effective for its treatment. Degradation products were identified under each treatment process using accurate mass time-of-flight spectrometry and pathways of decay were proposed. The main transformation pathways in each process were: dechlorination of the benzene ring during direct photolysis; hydroxyl group addition to the benzene ring during the reaction with hydroxyl radicals; and triazine ring opening after reaction with ozone. Different products that form in each process may be to a varying degree less environmentally stable than the parent lamotrigine. In addition, a novel method of ozone quenching without addition of salts is presented. The new quenching method would allow subsequent mass spectrometry analysis without a solid phase extraction clean-up step. The method involves raising the pH of the sample to approximately 10 for a few seconds and lowering it back and is therefore limited to applications for which temporary pH change is not expected to affect the outcome of the analysis. Ó 2014 Elsevier Ltd. All rights reserved.
⇑ Corresponding author at: Department of Civil and Environmental Engineering, University of North Carolina, 9201 University City Blvd, Charlotte, NC 28223, United States. Tel.: +1 (704) 687 5048. E-mail address: [email protected] (O.S. Keen). http://dx.doi.org/10.1016/j.chemosphere.2014.07.085 0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.
O.S. Keen et al. / Chemosphere 117 (2014) 316–323
1. Introduction Lamotrigine has only recently been detected in water bodies (Ferrer and Thurman, 2010). One of the most comprehensive studies surveying pharmaceuticals and personal care products in wastewater, drinking water and water resources (Kolpin et al., 2002) did not report lamotrigine because it was completed in 2000, not long after lamotrigine was introduced as an alternative to carbamazepine (Ramsay, 1993). Several influential studies were completed even earlier including major review articles in 1998 (Halling-Sørensen et al., 1998) and 1999 (Daughton and Ternes, 1999). Subsequent studies typically selected a set of pharmaceuticals for monitoring based on what was expected to be present, many relying on the findings by Kolpin et al. (2002) and earlier studies for their choices (Ternes et al., 2003; Stackelberg et al., 2004; Bendz et al., 2005). Although introduction of new drugs into the market is a slow process, and most of the common usage medication has been around for decades, some new widely prescribed compounds are introduced every year. A recent study demonstrated that lamotrigine is present in 94% of wastewater impacted samples at mean concentration of 488 ng L 1 (Ferrer and Thurman, 2010). The same study found lamotrigine in 2 of 7 drinking water samples and in 93% of alluvial groundwater impacted by wastewater either by hydraulic connection to surface water or via irrigation with recycled water. The high concentrations and ubiquitous presence combined with relatively new introduction to the consumer world indicate that the compound may be persistent in the wastewater treatment process and the environment, and ultimately reach drinking water treatment processes. The neuroactive properties of the drug makes it a potential aquatic life hazard similar to other psychoactive pharmaceuticals that have demonstrated an effect on predator–prey relationships of aquatic organisms (Foster et al., 2010). In addition, the compound is known to have a toxic interaction in humans with the 10,11-epoxide of carbamazepine (Warner et al., 1992) – a common metabolite found in wastewater impacted streams (Miao et al., 2005). The objective of this paper is to study the fate of lamotrigine in several treatment processes typically effective for a broad range of contaminants: advanced oxidation, ozonation and UV photolysis. Advanced oxidation processes (AOPs) have been studied as a potential treatment technology for degradation of pharmaceutically active compounds in wastewater and drinking water (Ternes et al., 2003; Huber et al., 2003). Both ozone based AOPs and UV based AOPs have been shown effective for transforming pharmaceutically active compounds, in most cases into pharmaceutically inactive (Linden et al., 2007; Dodd et al., 2009) or biodegradable products (Keen et al., 2012). In some instances formation of more recalcitrant or toxic products is possible (Alvares et al., 2001; Dantas et al., 2011), therefore understanding the reaction pathways is very important. This study determined the susceptibility of lamotrigine to degradation by hydroxyl radical, ozone and direct photolysis by UVC range wavelengths (98–99%. Lamotrigine, t-butanol, ammonium hydroxide (28% solution) and phenol were purchased from Sigma–Aldrich (St. Louis, MO, USA), 30% solution of hydrogen peroxide was manufactured by J.T. Baker (Phillipsburg, NJ, USA), and para-chlorobenzoic acid (pCBA) was manufactured by ICN Biomedicals (Aurora, OH, USA). The samples were prepared in ultrapure water (arium611-VF, Sartorius Stedim, Bohemia, NY, USA). For the HPLC/MS analysis, a gradient of HPLC-grade water (Solvent A) and HPLC-grade acetonitrile (Solvent B) was used (both solvents from Honeywell Burdick & Jackson, Muskegon, MI, USA). Solvent A contained 0.1% of formic acid (Fluka, St. Louis, MO, USA). Photolysis: The UV lamp collimated beam apparatus setup similar to the one used in this study is described in Bolton and Linden (Bolton and Linden, 2003) along with the methodology used to calculate the UV fluence delivered to the samples. Two types of lamps were used in the study: low pressure mercury vapor (General Electric, Fairfield, CT, USA) and medium pressure mercury vapor (Calgon Carbon, Pittsburg, PA, USA) – both emitting a UV spectrum characteristic for mercury vapor at given pressure. A NIST calibrated radiometer was used to measure UV irradiance from each lamp (IL1700, International Light, Peabody, MA, USA). Absorbance spectra of the samples were collected with a Cary100Bio UV/VIS spectrophotometer (Agilent Technologies, Santa Clara, CA, USA). Lamotrigine was added to ultrapure unbuffered water at 5 lM concentration in this and subsequent experiments. The final pH of the sample was approximately 6. Lamp irradiance was set at 1.5 mW cm 2. Sample depth was 1.7 cm. Petri factors were 0.97 and 0.99 for low pressure and medium pressure lamps respectively. Advanced oxidation and competition kinetics: Hydrogen peroxide (H2O2) at 5 and 10 mg L 1 concentration was used to generate hydroxyl radicals (HO). The concentration of H2O2 was measured using the triiodide method (Klassen et al., 1994). The experiments were repeated twice at each H2O2 concentration. Higher concentrations of H2O2 result in higher HO exposure and may create products with a greater extent of structural transformation. The reaction rate constant between HO and lamotrigine (kOH,LMG) was measured using competition kinetics with 5 lM pCBA (Huber et al., 2003). The value of kOH,pCBA is 5.0 109 M 1 s 1 (Buxton et al., 1988). Degradation of lamotrigine or pCBA by direct photolysis in a control experiment was subtracted from the overall degradation with H2O2 to isolate the degradation by HO. Neither
318
O.S. Keen et al. / Chemosphere 117 (2014) 316–323
of the two compounds reacted directly with H2O2 at the experimental timescale. Ozonation: An Ozone Solutions (Hull, IA, USA) TG-40 ozone (O3) generator was used to prepare a stock solution of O3. The stock solution was prepared in ultrapure water kept in a jacketed reactor at 4 °C temperature maintained by a continuously recirculating chiller (VWR, Radnor, PA, USA). Ozone concentration in the stock solution was determined by measuring absorbance at 258 nm. Ozone stock concentration was 1.0 mM (50 mg L 1) upon equilibration and was used immediately. The concentration of ozone in the sample was 100 lM making ozone stock 1/10 of the sample volume. The sample was prepared with the consideration that it will be further diluted once the ozone stock is added. For example, if the final concentration of lamotrigine was expected to be 5 lM, the solution was prepared at 5.56 lM and was diluted to the goal concentration on 5 lM once ozone stock was added. A single ozone dose was applied because the reaction between lamotrigine and ozone proved to be very slow and additional investigation of different ozone doses to get a more accurate rate constant was deemed unnecessary. Lamotrigine has a molecular structure that includes a triazine ring moiety with two amine groups and a benzene ring moiety substituted with two chlorines (Fig. 1). Although amine groups on a benzene ring would indicate that the compound is highly reactive with ozone (von Gunten, 2003), it is not so for a triazine ring where the nitrogen atoms in the ring structure diminish the ability of the amine substitutes to donate electrons. Atrazine has a similar core structure with a triazine ring containing two amine groups and a chlorine, and it has a very low reaction rate with ozone (6.0–6.3 M 1 s 1) (Beltrán et al., 1998; Acero et al., 2000). The other aromatic moiety of lamotrigine was not expected to be reactive with ozone because of the electron withdrawing properties of the chlorine atoms. Lamotrigine has water solubility of 0.17 mg mL 1at 25 °C, pKa of 5.7, log Kow of 1.17 at pH 7.6, and estimated vapor pressure of 9.4 10 9 mm Hg at 25 °C (Lyman, 1985; Mosby’s GenRx, 2001; Levy, 2002). The physicochemical properties indicate that the compound is not volatile or strongly lipophilic, and will have a tendency to remain in aqueous phase in the environment. Due to expected low reactivity of lamotrigine with O3, ozone was dosed in excess of lamotrigine for the experiment (Huber et al., 2003). The final solution contained 1 lM concentration of lamotrigine, 50 mM of tert-butanol for HO control and approximately 100 lM of O3. The control solution contained the
Fig. 1. The structure of lamotrigine and its molar UV absorption coefficients (values are listed in Supplemental Information).
lamotrigine and tert-butanol only. Because O3 concentration was 100 times the lamotrigine concentration (on molar basis), minimal consumption of O3 was expected in the reaction with lamotrigine, although some minimal ozone decomposition and volatilization was also expected within the 30 min reaction time. After 30 min O3 was quenched to stop the reaction. Ozone quenching procedure: Because the samples were destined for liquid chromatography mass spectrometry analysis, traditional methods of quenching with sulfite or thiosulfate salts were not suitable because of salt accumulation in the source of mass spectrometer and suppression of the mass spectral signal. Furthermore, non-volatile compounds, such as salts, can accumulate in the instruments and cause precipitation problems. Phenol reacts with ozone at nearly diffusion controlled rate, which makes phenol a suitable volatile quenching agent. The reaction is pH dependent, and the deprotonated form reacts with 1.4 ± 0.4 109 M 1 s 1 rate, which is six orders of magnitude faster than the rate for protonated species (1.3 ± 0.2 103 M 1 s 1) (Hoigné and Bader, 1983). For the most efficient use of phenol as a quenching agent, the pH of the solution should be raised close to or above the pKa of phenol of 9.9 (Hoigné and Bader, 1983). Lamotrigine pKa is 5.7 (Mosby’s GenRx, 2001), so no major change in the compound speciation was expected as pH was increased from approximately neutral to about 10. The overall reaction rate of phenol with ozone increases by approximately an order of magnitude for every pH unit increase between 4 and 10 (Hoigné and Bader, 1983). The stoichiometric ratio of ozone to phenol is approximately 1.8–2 (Wu and Masten, 2002), so 500 lM phenol was added to quench 100 lM of O3 in excess of the stoichiometric ratio. Ammonium hydroxide (a volatile buffer suitable for mass spectrometry work) was used concurrently with phenol addition to raise the pH to 10 to increase the rate of ozone scavenging by phenol to diffusion controlled levels. Both phenol and ammonium hydroxide were added to the ozone-free control sample to maintain the consistency between the samples. The quenching with this method is almost instantaneous (for example, at pH 10, 50% of phenol is in its deprotonated form; with phenol added at 500 lM, the reaction rate constant for ozone scavenging is (1.4 109 M 1 s 1) (250 10 6 M) = 3.5 105 s 1. After the quenching is complete, pH can be lowered back by adding a volatile organic acid, e.g., formic acid. Even though the sample stays at pH 10 for a matter of seconds before it can be lowered back to neutral, this method is limited to compounds not susceptible to fast hydrolysis at pH 10 and those that do not change speciation appreciably in the pH 8–10 range (e.g., fluoxetine and erythromycin). Additionally, this method is best suited for clean water matrices for determination of reaction rate constants or products of ozonation of specific compounds. In complex environmental sample matrices many potential pH dependent interactions may limit the applicability of this method. Analysis of lamotrigine and intermediates: For routine analysis of lamotrigine in the experiments measuring the quantum yield and the reaction rate constants, an 1100 series HPLC coupled to an XCT Plus ion trap MS instrument was used (Agilent, Santa Clara, CA). The ion trap was operated in the positive electrospray ionization mode throughout the analysis. The protonated ion [M+H]+ of lamotrigine is 256. The HPLC was equipped with a diode array detector and a 4.6 50-XDB-C8 column with 1.8 lm particle size. The diode array detector was set at 234 nm to monitor pCBA in the competition kinetics experiments. A linear gradient with increase from 10% B to 100% B over 13 min with an additional 2 min flush with 100% B and 5 min for post-time column equilibration. The retention time for the parent compound was 5.5 min. For accurate mass analysis of the products formed by the studied processes, a time-of-flight HPLC/MS was used (LC/TOF-MS). The analytical method for the extraction and analysis of lamotrigine by LC/TOF-MS has been previously studied and published (Ferrer and
O.S. Keen et al. / Chemosphere 117 (2014) 316–323
Thurman, 2010). In this previous paper both the recoveries and the method detection limit (MDL) from water samples are reported. In the present paper no extraction of samples was necessary because the spiking level was high enough for direct injection, thus not requiring any pre-concentration step. The MDL for lamotrigine using LC/TOF-MS is 1 ng L 1 (Ferrer and Thurman, 2010) and these experiments were carried out at the lg L 1 range. The analysis was intended to elucidate the structure of the products that form but not to quantify the product concentrations, therefore no isotope labeled standards were used. The separation of the analytes was carried out using an HPLC system (consisting of vacuum degasser, thermostated autosampler, column compartment and a binary pump) (Agilent Series 1290, Agilent Technologies, Santa Clara, CA, USA) connected to an ultra high definition quadrupole TOFMS Model 6540 Agilent (Agilent Technologies, Santa Clara, CA, USA). The HPLC was equipped with a reversed phase C8 analytical column of 150 mm 4.6 mm and 3.5 lm particle size (Zorbax Eclipse XDB-C8). Column temperature was maintained at 25 °C. The injected sample volume was 20 lL. Mobile phases A and B were acetonitrile and water with 0.1% formic acid, respectively. The optimized chromatographic method held the initial mobile phase composition (10% A) constant for 5 min, followed by a linear gradient to 100% A after 30 min. The flow-rate used was 0.6 mL min 1. A 5-min post-run time was used after each analysis. The instrument was operated in full-spectrum mode, except on those cases where MS–MS was necessary to elucidate chemical structures, as well as for identification of selected compounds and degradation products as explained in the results. The MS was equipped with electrospray Jet Stream Technology, operating in positive ion mode. The detailed operational parameters are provided in Supplementary Information.
3. Results and discussion Direct photolysis: Chlorinated molecules like polychlorinated biphenyls and chlorophenols are often susceptible to photodechlorination with quantum yields close to unity (Bunce, 1982; Czaplicka, 2006) due to relatively low CACl bond dissociation energy (about 330 kJ mol 1). UV wavelengths carry sufficient energy to break a CACl bond. The energy corresponding to the 254 nm wavelength emitted by low pressure mercury vapor sources is 470 kJ Es 1 (as calculated by Planck’s equation). Medium pressure mercury vapor UV sources emit a range of wavelengths P200 nm, with energy of up to 570 kJ Es 1. Based on this it would be reasonable to expect some photodechlorination of lamotrigine. In addition, lamotrigine has very high molar absorption coefficients, especially in the 200–220 nm range (see Fig. 1). However, the compound was not susceptible to direct photolysis by the low pressure UV source and only slightly susceptible to the medium pressure UV source with the quantum yield = (2.7 ± 0.4) 10 4. Electron donating nitrogen atoms in the triazine ring and its amine groups may result in stabilization of the CACl bonds and explain the demonstrated photostability of lamotrigine. Few studies are available discussing the effects of substituents on photolability of CACl bonds (Bunce, 1982; Stegeman et al., 1993). Previous studies have shown low quantum yields of photolysis for chlorinated compounds with electron donating moieties, e.g., pesticides such as atrazine (Sharpless et al., 2003), terbacil (Shemer et al., 2006), diuron (Shemer et al., 2006) and metoxuron (Boulkamh et al., 2001). Another study of pesticides (commonly studied chlorinated aromatics) showed high quantum yields for organophosphate compounds with the chlorinated aromatic ring attached to an electron withdrawing phosphate group. The quantum yields were dramatically lower when in addition to the
319
phosphate group, an electron donating entity was attached to the chlorinated aromatic ring (ACH3) or was part of the ring (AN@) (Wan et al., 1994). Although the quantum yield with low pressure lamp was not measurable, a photolysis product was detected, indicating that some degree of transformation occured with 254 nm irradiation as well. The product had a measured m/z of 220.0386, which is less than 1 ppm of the calculated exact mass for the singly dechlorinated lamotrigine with a formula of C9H6ClN5. The isotope signature of the m/z 220.0386 ion also showed the correct accurate mass for the 37Cl isotope with 2 ppm accuracy, which indicates a single chlorine atom. Thus, the m/z value found was consistent with the loss of a single chlorine from the parent molecule. With medium pressure UV source, several singly dechlorinated products were detected along with the m/z 220 product, but m/z 220 was the most prominent product peak. Interesting products of photolysis of lamotrigine were detected by mass spectrometry of the solutions analyzed at m/z 238, 222, and 254. Approximate UV fluence required for each product to double ranked as follows from fastest to slowest forming products: m/z 220 > m/z 238 > m/z 222 (2 peaks) and m/z 256 (parent isomers, 2 peaks) > m/z 254. The retention time of the products from most hydrophilic to least hydrophilic is m/z 254 > m/z 238 > m/z 222 > m/z 220 > m/z 256 (parent and its isomers). Based on the structure, most likely parent isomers form when a chlorine atom detaches but cannot leave the solvent cage and reattaches at a different location on the ring. Some of the product structures were identified by accurate mass and are shown in Fig. 2. Products with m/z 220, 238 and 254 as well as the parent molecule isomeration has been confirmed for lamotrigine transformation under simulated sunlight recently (Young et al., 2014). Structures of 5 degradation products found by LC/MS ion trap mass spectrometry were confirmed by LC/QTOF-MS analysis. A dechlorinated product was found with a mass of m/z 220.0384 (calculated exact mass). Because there are two chlorine atoms, either one could have been removed by the UV light. Only one isomer was found upon extraction of the exact mass of m/z 220.0384. This suggests that only one bond is susceptible to fragmentation. The measured mass was 220.0386 (1 ppm mass accuracy). This dechlorination product was a major component of the chromatogram. The most likely loss of chlorine would be from the orthoposition, because it is activated compared to the meta-position. The loss of chlorine is most likely a radical loss (Schwarzenbach et al., 2005), where the odd electron is available on the chlorine atom to re-react with the lamotrigine molecule to form an isomer of lamotrigine, which is found at both slightly longer and slightly shorter retention time than the parent compound (11.6 and 13.7 min retention time compared to 12.0 min for the parent compound). The two new isomers are consistent with the available locations on the lamotrigine structure. Three degradation products were found with an exact calculated mass of 238.0490. The peaks occurred at retention times of 4.0, 8.0, and 9.2 min with accuracies shown in Table 1 at less than 1 ppm mass accuracy. The chromatogram and mass spectrum, which was identical for the 3 peaks, are shown in Supplementary Information and show that one chlorine atom has been removed and a hydroxyl group has been added. The fact that three isomers were observed brings up an important conclusion since lamotrigine has only two chlorines. Apparently, the new isomer of lamotrigine that formed from the radical chlorine attack has also been hydroxylated, which would account for the third isomer of the m/z 238.0490 degradation product. Two degradation products were detected at the m/z of 222.0541 at retention times of 9.1 and 10.8 min. Their formula shows that a chlorine atom has been removed but an oxidation has occurred at the site of chlorine loss with water being the source of the hydrogen atoms. Thus, the accurate mass is 2 mass units greater
320
O.S. Keen et al. / Chemosphere 117 (2014) 316–323
Fig. 2. Potential products and degradation pathways of lamotrigine during direct photolysis.
Table 1 Degradation products detected by accurate mass spectrometry. Calculated exact mass
Measured accurate mass (retention time)
ppm error
Elemental composition
256.0151 256.0151 220.0384 222.0541 222.0541 238.0490 238.0490 238.0490 259.9988 272.0100
256.0153 256.0149 220.0386 222.0539 222.0537 238.0490 238.0490 238.0491 259.9986 272.0097
0.8 0.8 0.7 0.7 1.8 0.0 0.0 0.5 0.8 1.1
C9H8Cl2N5 C9H8Cl2N5 C9H7ClN5 C9H9ClN5 C9H9ClN5 C9H9ClN5O C9H9ClN5O C9H9ClN5O C9H8Cl2N3O2 C9H8Cl2N5O
(13.7 min) (11.6 min) (11.35 min) (9.1 min) (10.8 min) (4.0 min) (8.0 min) (9.2 min) (10.8 min) (13.4 min)
than the UV oxidation product of m/z 220.0384 which was discussed earlier. Table 1 lists the accurate masses with the corresponding ppm error. Despite good LC-MS detection, these products form in minute quantities due to extremely low quantum yield of the parent
Fig. 3. Observed product of advanced oxidation of lamotrigine.
compound and only under high absorbing low wavelength UV (
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Degradation pathways of lamotrigine under advanced treatment by direct UV photolysis, hydroxyl radicals, and ozone Olya S. Keen a,c,⇑, Imma Ferrer b, E. Michael Thurman b, Karl G. Linden a a
Department of Civil, Environmental and Architectural Engineering, University of Colorado, UCB 428, Boulder, CO 80309, United States Center for Environmental Mass Spectrometry, University of Colorado, UCB 428, Boulder, CO 80309, United States c Department of Civil and Environmental Engineering, University of North Carolina, 9201 University City Blvd, Charlotte, NC 28223, United States b
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Lamotrigine was exposed to ozone,
hydroxyl radical and direct photolysis. Degradation products in each process were identified. Hydroxyl radical was the most efficient degradation pathway. Quantum yield and reaction rate constants were measured. Novel mass spectrometry friendly, bench-scale ozone quenching method is proposed.
a r t i c l e
i n f o
Article history: Received 5 February 2014 Received in revised form 20 July 2014 Accepted 27 July 2014
Handling Editor: Hyunook Kim Keywords: Lamotrigine Advanced oxidation Hydroxyl radical reaction rate constant Quantum yield Ozone quenching Mass spectrometry
a b s t r a c t Lamotrigine is recently recognized as a persistent pharmaceutical in the water environment and wastewater effluents. Its degradation was studied under UV and ozone advanced oxidation treatments with reaction kinetics of lamotrigine with ozone (4 M 1 s 1), hydroxyl radical [(2.1 ± 0.3) 109 M 1 s 1] and by UV photolysis with low and medium pressure mercury vapor lamps [quantum yields 0 and (2.7 ± 0.4) 10 4 respectively] determined. All constants were measured at pH 6 and at temperature 20 °C. The results indicate that lamotrigine is slow to respond to direct photolysis or oxidation by ozone and no attenuation of the contaminant is expected in UV or ozone disinfection applications. The compound reacts rapidly with hydroxyl radicals indicating that advanced oxidation processes would be effective for its treatment. Degradation products were identified under each treatment process using accurate mass time-of-flight spectrometry and pathways of decay were proposed. The main transformation pathways in each process were: dechlorination of the benzene ring during direct photolysis; hydroxyl group addition to the benzene ring during the reaction with hydroxyl radicals; and triazine ring opening after reaction with ozone. Different products that form in each process may be to a varying degree less environmentally stable than the parent lamotrigine. In addition, a novel method of ozone quenching without addition of salts is presented. The new quenching method would allow subsequent mass spectrometry analysis without a solid phase extraction clean-up step. The method involves raising the pH of the sample to approximately 10 for a few seconds and lowering it back and is therefore limited to applications for which temporary pH change is not expected to affect the outcome of the analysis. Ó 2014 Elsevier Ltd. All rights reserved.
⇑ Corresponding author at: Department of Civil and Environmental Engineering, University of North Carolina, 9201 University City Blvd, Charlotte, NC 28223, United States. Tel.: +1 (704) 687 5048. E-mail address: [email protected] (O.S. Keen). http://dx.doi.org/10.1016/j.chemosphere.2014.07.085 0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.
O.S. Keen et al. / Chemosphere 117 (2014) 316–323
1. Introduction Lamotrigine has only recently been detected in water bodies (Ferrer and Thurman, 2010). One of the most comprehensive studies surveying pharmaceuticals and personal care products in wastewater, drinking water and water resources (Kolpin et al., 2002) did not report lamotrigine because it was completed in 2000, not long after lamotrigine was introduced as an alternative to carbamazepine (Ramsay, 1993). Several influential studies were completed even earlier including major review articles in 1998 (Halling-Sørensen et al., 1998) and 1999 (Daughton and Ternes, 1999). Subsequent studies typically selected a set of pharmaceuticals for monitoring based on what was expected to be present, many relying on the findings by Kolpin et al. (2002) and earlier studies for their choices (Ternes et al., 2003; Stackelberg et al., 2004; Bendz et al., 2005). Although introduction of new drugs into the market is a slow process, and most of the common usage medication has been around for decades, some new widely prescribed compounds are introduced every year. A recent study demonstrated that lamotrigine is present in 94% of wastewater impacted samples at mean concentration of 488 ng L 1 (Ferrer and Thurman, 2010). The same study found lamotrigine in 2 of 7 drinking water samples and in 93% of alluvial groundwater impacted by wastewater either by hydraulic connection to surface water or via irrigation with recycled water. The high concentrations and ubiquitous presence combined with relatively new introduction to the consumer world indicate that the compound may be persistent in the wastewater treatment process and the environment, and ultimately reach drinking water treatment processes. The neuroactive properties of the drug makes it a potential aquatic life hazard similar to other psychoactive pharmaceuticals that have demonstrated an effect on predator–prey relationships of aquatic organisms (Foster et al., 2010). In addition, the compound is known to have a toxic interaction in humans with the 10,11-epoxide of carbamazepine (Warner et al., 1992) – a common metabolite found in wastewater impacted streams (Miao et al., 2005). The objective of this paper is to study the fate of lamotrigine in several treatment processes typically effective for a broad range of contaminants: advanced oxidation, ozonation and UV photolysis. Advanced oxidation processes (AOPs) have been studied as a potential treatment technology for degradation of pharmaceutically active compounds in wastewater and drinking water (Ternes et al., 2003; Huber et al., 2003). Both ozone based AOPs and UV based AOPs have been shown effective for transforming pharmaceutically active compounds, in most cases into pharmaceutically inactive (Linden et al., 2007; Dodd et al., 2009) or biodegradable products (Keen et al., 2012). In some instances formation of more recalcitrant or toxic products is possible (Alvares et al., 2001; Dantas et al., 2011), therefore understanding the reaction pathways is very important. This study determined the susceptibility of lamotrigine to degradation by hydroxyl radical, ozone and direct photolysis by UVC range wavelengths (98–99%. Lamotrigine, t-butanol, ammonium hydroxide (28% solution) and phenol were purchased from Sigma–Aldrich (St. Louis, MO, USA), 30% solution of hydrogen peroxide was manufactured by J.T. Baker (Phillipsburg, NJ, USA), and para-chlorobenzoic acid (pCBA) was manufactured by ICN Biomedicals (Aurora, OH, USA). The samples were prepared in ultrapure water (arium611-VF, Sartorius Stedim, Bohemia, NY, USA). For the HPLC/MS analysis, a gradient of HPLC-grade water (Solvent A) and HPLC-grade acetonitrile (Solvent B) was used (both solvents from Honeywell Burdick & Jackson, Muskegon, MI, USA). Solvent A contained 0.1% of formic acid (Fluka, St. Louis, MO, USA). Photolysis: The UV lamp collimated beam apparatus setup similar to the one used in this study is described in Bolton and Linden (Bolton and Linden, 2003) along with the methodology used to calculate the UV fluence delivered to the samples. Two types of lamps were used in the study: low pressure mercury vapor (General Electric, Fairfield, CT, USA) and medium pressure mercury vapor (Calgon Carbon, Pittsburg, PA, USA) – both emitting a UV spectrum characteristic for mercury vapor at given pressure. A NIST calibrated radiometer was used to measure UV irradiance from each lamp (IL1700, International Light, Peabody, MA, USA). Absorbance spectra of the samples were collected with a Cary100Bio UV/VIS spectrophotometer (Agilent Technologies, Santa Clara, CA, USA). Lamotrigine was added to ultrapure unbuffered water at 5 lM concentration in this and subsequent experiments. The final pH of the sample was approximately 6. Lamp irradiance was set at 1.5 mW cm 2. Sample depth was 1.7 cm. Petri factors were 0.97 and 0.99 for low pressure and medium pressure lamps respectively. Advanced oxidation and competition kinetics: Hydrogen peroxide (H2O2) at 5 and 10 mg L 1 concentration was used to generate hydroxyl radicals (HO). The concentration of H2O2 was measured using the triiodide method (Klassen et al., 1994). The experiments were repeated twice at each H2O2 concentration. Higher concentrations of H2O2 result in higher HO exposure and may create products with a greater extent of structural transformation. The reaction rate constant between HO and lamotrigine (kOH,LMG) was measured using competition kinetics with 5 lM pCBA (Huber et al., 2003). The value of kOH,pCBA is 5.0 109 M 1 s 1 (Buxton et al., 1988). Degradation of lamotrigine or pCBA by direct photolysis in a control experiment was subtracted from the overall degradation with H2O2 to isolate the degradation by HO. Neither
318
O.S. Keen et al. / Chemosphere 117 (2014) 316–323
of the two compounds reacted directly with H2O2 at the experimental timescale. Ozonation: An Ozone Solutions (Hull, IA, USA) TG-40 ozone (O3) generator was used to prepare a stock solution of O3. The stock solution was prepared in ultrapure water kept in a jacketed reactor at 4 °C temperature maintained by a continuously recirculating chiller (VWR, Radnor, PA, USA). Ozone concentration in the stock solution was determined by measuring absorbance at 258 nm. Ozone stock concentration was 1.0 mM (50 mg L 1) upon equilibration and was used immediately. The concentration of ozone in the sample was 100 lM making ozone stock 1/10 of the sample volume. The sample was prepared with the consideration that it will be further diluted once the ozone stock is added. For example, if the final concentration of lamotrigine was expected to be 5 lM, the solution was prepared at 5.56 lM and was diluted to the goal concentration on 5 lM once ozone stock was added. A single ozone dose was applied because the reaction between lamotrigine and ozone proved to be very slow and additional investigation of different ozone doses to get a more accurate rate constant was deemed unnecessary. Lamotrigine has a molecular structure that includes a triazine ring moiety with two amine groups and a benzene ring moiety substituted with two chlorines (Fig. 1). Although amine groups on a benzene ring would indicate that the compound is highly reactive with ozone (von Gunten, 2003), it is not so for a triazine ring where the nitrogen atoms in the ring structure diminish the ability of the amine substitutes to donate electrons. Atrazine has a similar core structure with a triazine ring containing two amine groups and a chlorine, and it has a very low reaction rate with ozone (6.0–6.3 M 1 s 1) (Beltrán et al., 1998; Acero et al., 2000). The other aromatic moiety of lamotrigine was not expected to be reactive with ozone because of the electron withdrawing properties of the chlorine atoms. Lamotrigine has water solubility of 0.17 mg mL 1at 25 °C, pKa of 5.7, log Kow of 1.17 at pH 7.6, and estimated vapor pressure of 9.4 10 9 mm Hg at 25 °C (Lyman, 1985; Mosby’s GenRx, 2001; Levy, 2002). The physicochemical properties indicate that the compound is not volatile or strongly lipophilic, and will have a tendency to remain in aqueous phase in the environment. Due to expected low reactivity of lamotrigine with O3, ozone was dosed in excess of lamotrigine for the experiment (Huber et al., 2003). The final solution contained 1 lM concentration of lamotrigine, 50 mM of tert-butanol for HO control and approximately 100 lM of O3. The control solution contained the
Fig. 1. The structure of lamotrigine and its molar UV absorption coefficients (values are listed in Supplemental Information).
lamotrigine and tert-butanol only. Because O3 concentration was 100 times the lamotrigine concentration (on molar basis), minimal consumption of O3 was expected in the reaction with lamotrigine, although some minimal ozone decomposition and volatilization was also expected within the 30 min reaction time. After 30 min O3 was quenched to stop the reaction. Ozone quenching procedure: Because the samples were destined for liquid chromatography mass spectrometry analysis, traditional methods of quenching with sulfite or thiosulfate salts were not suitable because of salt accumulation in the source of mass spectrometer and suppression of the mass spectral signal. Furthermore, non-volatile compounds, such as salts, can accumulate in the instruments and cause precipitation problems. Phenol reacts with ozone at nearly diffusion controlled rate, which makes phenol a suitable volatile quenching agent. The reaction is pH dependent, and the deprotonated form reacts with 1.4 ± 0.4 109 M 1 s 1 rate, which is six orders of magnitude faster than the rate for protonated species (1.3 ± 0.2 103 M 1 s 1) (Hoigné and Bader, 1983). For the most efficient use of phenol as a quenching agent, the pH of the solution should be raised close to or above the pKa of phenol of 9.9 (Hoigné and Bader, 1983). Lamotrigine pKa is 5.7 (Mosby’s GenRx, 2001), so no major change in the compound speciation was expected as pH was increased from approximately neutral to about 10. The overall reaction rate of phenol with ozone increases by approximately an order of magnitude for every pH unit increase between 4 and 10 (Hoigné and Bader, 1983). The stoichiometric ratio of ozone to phenol is approximately 1.8–2 (Wu and Masten, 2002), so 500 lM phenol was added to quench 100 lM of O3 in excess of the stoichiometric ratio. Ammonium hydroxide (a volatile buffer suitable for mass spectrometry work) was used concurrently with phenol addition to raise the pH to 10 to increase the rate of ozone scavenging by phenol to diffusion controlled levels. Both phenol and ammonium hydroxide were added to the ozone-free control sample to maintain the consistency between the samples. The quenching with this method is almost instantaneous (for example, at pH 10, 50% of phenol is in its deprotonated form; with phenol added at 500 lM, the reaction rate constant for ozone scavenging is (1.4 109 M 1 s 1) (250 10 6 M) = 3.5 105 s 1. After the quenching is complete, pH can be lowered back by adding a volatile organic acid, e.g., formic acid. Even though the sample stays at pH 10 for a matter of seconds before it can be lowered back to neutral, this method is limited to compounds not susceptible to fast hydrolysis at pH 10 and those that do not change speciation appreciably in the pH 8–10 range (e.g., fluoxetine and erythromycin). Additionally, this method is best suited for clean water matrices for determination of reaction rate constants or products of ozonation of specific compounds. In complex environmental sample matrices many potential pH dependent interactions may limit the applicability of this method. Analysis of lamotrigine and intermediates: For routine analysis of lamotrigine in the experiments measuring the quantum yield and the reaction rate constants, an 1100 series HPLC coupled to an XCT Plus ion trap MS instrument was used (Agilent, Santa Clara, CA). The ion trap was operated in the positive electrospray ionization mode throughout the analysis. The protonated ion [M+H]+ of lamotrigine is 256. The HPLC was equipped with a diode array detector and a 4.6 50-XDB-C8 column with 1.8 lm particle size. The diode array detector was set at 234 nm to monitor pCBA in the competition kinetics experiments. A linear gradient with increase from 10% B to 100% B over 13 min with an additional 2 min flush with 100% B and 5 min for post-time column equilibration. The retention time for the parent compound was 5.5 min. For accurate mass analysis of the products formed by the studied processes, a time-of-flight HPLC/MS was used (LC/TOF-MS). The analytical method for the extraction and analysis of lamotrigine by LC/TOF-MS has been previously studied and published (Ferrer and
O.S. Keen et al. / Chemosphere 117 (2014) 316–323
Thurman, 2010). In this previous paper both the recoveries and the method detection limit (MDL) from water samples are reported. In the present paper no extraction of samples was necessary because the spiking level was high enough for direct injection, thus not requiring any pre-concentration step. The MDL for lamotrigine using LC/TOF-MS is 1 ng L 1 (Ferrer and Thurman, 2010) and these experiments were carried out at the lg L 1 range. The analysis was intended to elucidate the structure of the products that form but not to quantify the product concentrations, therefore no isotope labeled standards were used. The separation of the analytes was carried out using an HPLC system (consisting of vacuum degasser, thermostated autosampler, column compartment and a binary pump) (Agilent Series 1290, Agilent Technologies, Santa Clara, CA, USA) connected to an ultra high definition quadrupole TOFMS Model 6540 Agilent (Agilent Technologies, Santa Clara, CA, USA). The HPLC was equipped with a reversed phase C8 analytical column of 150 mm 4.6 mm and 3.5 lm particle size (Zorbax Eclipse XDB-C8). Column temperature was maintained at 25 °C. The injected sample volume was 20 lL. Mobile phases A and B were acetonitrile and water with 0.1% formic acid, respectively. The optimized chromatographic method held the initial mobile phase composition (10% A) constant for 5 min, followed by a linear gradient to 100% A after 30 min. The flow-rate used was 0.6 mL min 1. A 5-min post-run time was used after each analysis. The instrument was operated in full-spectrum mode, except on those cases where MS–MS was necessary to elucidate chemical structures, as well as for identification of selected compounds and degradation products as explained in the results. The MS was equipped with electrospray Jet Stream Technology, operating in positive ion mode. The detailed operational parameters are provided in Supplementary Information.
3. Results and discussion Direct photolysis: Chlorinated molecules like polychlorinated biphenyls and chlorophenols are often susceptible to photodechlorination with quantum yields close to unity (Bunce, 1982; Czaplicka, 2006) due to relatively low CACl bond dissociation energy (about 330 kJ mol 1). UV wavelengths carry sufficient energy to break a CACl bond. The energy corresponding to the 254 nm wavelength emitted by low pressure mercury vapor sources is 470 kJ Es 1 (as calculated by Planck’s equation). Medium pressure mercury vapor UV sources emit a range of wavelengths P200 nm, with energy of up to 570 kJ Es 1. Based on this it would be reasonable to expect some photodechlorination of lamotrigine. In addition, lamotrigine has very high molar absorption coefficients, especially in the 200–220 nm range (see Fig. 1). However, the compound was not susceptible to direct photolysis by the low pressure UV source and only slightly susceptible to the medium pressure UV source with the quantum yield = (2.7 ± 0.4) 10 4. Electron donating nitrogen atoms in the triazine ring and its amine groups may result in stabilization of the CACl bonds and explain the demonstrated photostability of lamotrigine. Few studies are available discussing the effects of substituents on photolability of CACl bonds (Bunce, 1982; Stegeman et al., 1993). Previous studies have shown low quantum yields of photolysis for chlorinated compounds with electron donating moieties, e.g., pesticides such as atrazine (Sharpless et al., 2003), terbacil (Shemer et al., 2006), diuron (Shemer et al., 2006) and metoxuron (Boulkamh et al., 2001). Another study of pesticides (commonly studied chlorinated aromatics) showed high quantum yields for organophosphate compounds with the chlorinated aromatic ring attached to an electron withdrawing phosphate group. The quantum yields were dramatically lower when in addition to the
319
phosphate group, an electron donating entity was attached to the chlorinated aromatic ring (ACH3) or was part of the ring (AN@) (Wan et al., 1994). Although the quantum yield with low pressure lamp was not measurable, a photolysis product was detected, indicating that some degree of transformation occured with 254 nm irradiation as well. The product had a measured m/z of 220.0386, which is less than 1 ppm of the calculated exact mass for the singly dechlorinated lamotrigine with a formula of C9H6ClN5. The isotope signature of the m/z 220.0386 ion also showed the correct accurate mass for the 37Cl isotope with 2 ppm accuracy, which indicates a single chlorine atom. Thus, the m/z value found was consistent with the loss of a single chlorine from the parent molecule. With medium pressure UV source, several singly dechlorinated products were detected along with the m/z 220 product, but m/z 220 was the most prominent product peak. Interesting products of photolysis of lamotrigine were detected by mass spectrometry of the solutions analyzed at m/z 238, 222, and 254. Approximate UV fluence required for each product to double ranked as follows from fastest to slowest forming products: m/z 220 > m/z 238 > m/z 222 (2 peaks) and m/z 256 (parent isomers, 2 peaks) > m/z 254. The retention time of the products from most hydrophilic to least hydrophilic is m/z 254 > m/z 238 > m/z 222 > m/z 220 > m/z 256 (parent and its isomers). Based on the structure, most likely parent isomers form when a chlorine atom detaches but cannot leave the solvent cage and reattaches at a different location on the ring. Some of the product structures were identified by accurate mass and are shown in Fig. 2. Products with m/z 220, 238 and 254 as well as the parent molecule isomeration has been confirmed for lamotrigine transformation under simulated sunlight recently (Young et al., 2014). Structures of 5 degradation products found by LC/MS ion trap mass spectrometry were confirmed by LC/QTOF-MS analysis. A dechlorinated product was found with a mass of m/z 220.0384 (calculated exact mass). Because there are two chlorine atoms, either one could have been removed by the UV light. Only one isomer was found upon extraction of the exact mass of m/z 220.0384. This suggests that only one bond is susceptible to fragmentation. The measured mass was 220.0386 (1 ppm mass accuracy). This dechlorination product was a major component of the chromatogram. The most likely loss of chlorine would be from the orthoposition, because it is activated compared to the meta-position. The loss of chlorine is most likely a radical loss (Schwarzenbach et al., 2005), where the odd electron is available on the chlorine atom to re-react with the lamotrigine molecule to form an isomer of lamotrigine, which is found at both slightly longer and slightly shorter retention time than the parent compound (11.6 and 13.7 min retention time compared to 12.0 min for the parent compound). The two new isomers are consistent with the available locations on the lamotrigine structure. Three degradation products were found with an exact calculated mass of 238.0490. The peaks occurred at retention times of 4.0, 8.0, and 9.2 min with accuracies shown in Table 1 at less than 1 ppm mass accuracy. The chromatogram and mass spectrum, which was identical for the 3 peaks, are shown in Supplementary Information and show that one chlorine atom has been removed and a hydroxyl group has been added. The fact that three isomers were observed brings up an important conclusion since lamotrigine has only two chlorines. Apparently, the new isomer of lamotrigine that formed from the radical chlorine attack has also been hydroxylated, which would account for the third isomer of the m/z 238.0490 degradation product. Two degradation products were detected at the m/z of 222.0541 at retention times of 9.1 and 10.8 min. Their formula shows that a chlorine atom has been removed but an oxidation has occurred at the site of chlorine loss with water being the source of the hydrogen atoms. Thus, the accurate mass is 2 mass units greater
320
O.S. Keen et al. / Chemosphere 117 (2014) 316–323
Fig. 2. Potential products and degradation pathways of lamotrigine during direct photolysis.
Table 1 Degradation products detected by accurate mass spectrometry. Calculated exact mass
Measured accurate mass (retention time)
ppm error
Elemental composition
256.0151 256.0151 220.0384 222.0541 222.0541 238.0490 238.0490 238.0490 259.9988 272.0100
256.0153 256.0149 220.0386 222.0539 222.0537 238.0490 238.0490 238.0491 259.9986 272.0097
0.8 0.8 0.7 0.7 1.8 0.0 0.0 0.5 0.8 1.1
C9H8Cl2N5 C9H8Cl2N5 C9H7ClN5 C9H9ClN5 C9H9ClN5 C9H9ClN5O C9H9ClN5O C9H9ClN5O C9H8Cl2N3O2 C9H8Cl2N5O
(13.7 min) (11.6 min) (11.35 min) (9.1 min) (10.8 min) (4.0 min) (8.0 min) (9.2 min) (10.8 min) (13.4 min)
than the UV oxidation product of m/z 220.0384 which was discussed earlier. Table 1 lists the accurate masses with the corresponding ppm error. Despite good LC-MS detection, these products form in minute quantities due to extremely low quantum yield of the parent
Fig. 3. Observed product of advanced oxidation of lamotrigine.
compound and only under high absorbing low wavelength UV (
