Kinetics and Degradation Products for Direct Photolysis of β-Blockers

The overall results demonstrated that with fast direct photolysis half-lives, ... (e.g., U.S.), it is less practical in others (e.g., UK) because of t...
0 downloads 0 Views 248KB Size
Environ. Sci. Technol. 2007, 41, 803-810

Kinetics and Degradation Products for Direct Photolysis of β-Blockers in Water Q I N - T A O L I U * ,† A N D HELEN E. WILLIAMS‡ AstraZeneca UK Ltd. Global Safety Health and Environment, Brixham Environmental Laboratory, Brixham, Devon, UK TQ5 8BA, and Pharmaceutical and Analytical Research and Development, Macclesfield, Cheshire, UK

Related to improving persistence assessment of active pharmaceutical ingredients (APIs), direct aqueous photolysis of β-blockers: propranolol (hydrochloride salt), atenolol, and metoprolol (succinate salt) were investigated by exposing the samples (0.0003-10 mg L-1) to a solar irradiator (filtered xenon lamp: 290-800 nm) at 20-26 °C. Results suggested that direct photolysis in optically dilute solutions followed pseudo first-order kinetics. The measured halflives of propranolol, atenolol, and metoprolol were approximately 16, 350, and 630 h, respectively. These were 3-5 orders of magnitude slower than the estimated minimum half-lives. The measured half-lives were related to day light surface conditions by comparing the light intensity of the lamp and the sun at different latitudes and seasons. Major direct photolysis products were identified from propranolol that led to a proposed reaction pathway, involving ring oxidation, rearrangement, and deoxygenation. Electron paramagnetic resonance (EPR) spectroscopy results confirmed that at least one carbon-based radical intermediate was formed during the direct photolysis of propranolol in aqueous solutions. The overall results demonstrated that with fast direct photolysis halflives, propranolol is unlikely to be persistent in natural waters. Further work is needed to investigate indirect photolysis of atenolol and metoprolol in surface waters in order to understand the overall persistence of these APIs in the environment.

Introduction Phototransformation has historically been ignored as a process for the removal of organic chemicals in the environment, partly due to the difficulties in understanding the reaction kinetics and in relating them to natural environmental conditions. This is reflected by the fact that OECD photolysis test guidelines have only been in draft form since their release in 2000 (1). However, recently, phototransformation of APIs has attracted increasing attention as a potentially effective depletion mechanism in natural surface waters (2-4). Research results demonstrated that direct photolysis had a significant impact on reducing predicted environmental concentrations (PECs) of propranolol, a β-blocker, in rivers (5, 6). Seasonal measurements of some * Corresponding author phone: +44 (0)1803 884 229; fax: +44 (0)1803 882 974; e-mail: [email protected]. † Global Safety Health and Environment. ‡ Pharmaceutical and Analytical Research and Development. 10.1021/es0616130 CCC: $37.00 Published on Web 12/23/2006

 2007 American Chemical Society

pharmaceuticals showed lower concentrations in surface waters in summer than in winter, implying phototransformation could be one of the effective removal processes in the summer (7). Beta-blockers (β-adrenergic receptors) belong to a group of cardiovascular APIs and are generally used for the treatment of hypertension and cardiac arrhythmias. They are water-soluble, ionizable, and aromatic pharmaceuticals with multi-functional groups and low vapor pressures. Among the β-blockers, metoprolol, atenolol and propranolol have had long-term use in Europe and North America (8). Direct photolysis of APIs in optically dilute solutions has been found to follow pseudo first order reaction kinetics (9-11). However, some measurements were undertaken at non-environmentally relevant conditions with vacuum UV irradiation, photocatalysts and relatively high starting concentrations (10-10 000 mg L-1); therefore, there are difficulties in understanding the reaction kinetics and in relating them to natural environmental conditions (12, 13). Among the environmentally relevant studies, propranolol was found to undergo direct photolysis under solar irradiation at 40 °N (14) and under filtered xenon lamp with wavelength of 295800 nm (5, 6, 15). However, to our knowledge, no identified phototransformation products of propranolol have been reported. Although two photodegradation products of atenolol were identified when it was exposed to UVA-UVB radiations with the main degradation product being 1-(4-hydroxyphenyl) acetamide (16), there were no detailed kinetics data available. Furthermore, to our knowledge there is no previously published data for phototransformation of metoprolol in waters. Direct photolysis rates of organic chemicals usually vary in natural surface waters, corresponding to the light intensity at different latitudes (17, 18). There are different lines of research into using natural sunlight versus solar simulators, such as a filtered xenon lamp. A review of using different light sources is given by Boreen et al. (2). While exposing samples to natural sunlight may be applicable in some regions (e.g., U.S.), it is less practical in others (e.g., UK) because of the weather conditions. The authors here accept that the filtered xenon irradiation (295-800 nm) would generate reliable and reproducible kinetics due to the accurate measurement of photon fluxes and the easy control of light intensity. However, in order to apply direct photolysis rates to the varying natural environment for the refinement of environmental risk assessment (ERA), these measured rates need to be related to the near-surface sunlight conditions for the particular region of interest by considering the photon fluxes of the xenon lamp and the sun (1, 5, 6). In this paper, a tiered approach of theoretical screening, laboratory measurements, and relating kinetics to sunlight conditions was presented. Mass balance of the degradation products from propranolol was measured by exposing 14Clabeled propranolol under the same conditions. Major degradation products and free radical intermediates were identified. The long-term objective is to combine the results here with that of indirect photolysis for further understanding the persistence behavior of APIs, hence improving their overall ERA.

Materials and Methods Test Substances. The test substances were racemic mixtures supplied by AstraZeneca UK Limited (Macclesfield, UK). These included atenolol, benzeneacetamide, 4-[2-hydroxy3-[(1-methylethyl)amino]propoxy]- (99.5% w/w); metoprolol, (()-1-isopropylamino-3-[4-(2-methoxyethyl)[phenoxy]proVOL. 41, NO. 3, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

803

FIGURE 1. UV-visible spectra and estimated maximum rate constants (kd,max) of propranolol, atenolol, and metoprolol in pH 7.0 (0.05 M phosphate buffer) and 10 mg L-1 deionized water solutions. akd(max) ) ΣEλLλ. bkd (max)corr. ) OpropΣEλLλ, Oprop ) 0.002 (14). pan-2-ol succinate (100.0% w/w), and propranolol, (()-1isopropyl-amino-3-(1-naphthyloxy)propan-2-ol hydrochloride (99.8% w/w). 14C-propranolol, (()-[2-hydroxy-3-([1-14C]1-naphthyloxy)propyl] isopropylammonium chloride (99% total radio-activity), was prepared by ICI plc. on 6 November 1990. The specific activity was measured just before the experiment and was at 1635 Bq µg-1. Measurement of UV-VIS Spectra. Before the direct photolysis experiment, UV-VIS spectra of propranolol, metoprolol, and atenolol (10 mg L-1) were measured by a UV-VIS spectrometer (Perkin-Elmer, Lambda 11) in deionized water and pH 7.0 phosphate buffer (0.05 M) made with deionized water (Figure 1). The UV spectra in deionized water were similar to the spectra in the pH 7.0 buffer. The UV-VIS absorbance was used to calculate molar absorption coefficient (λ, M-1cm-1) of the β-blockers at wavelength between 295 and 800 nm (Table S1, Supporting Information). Theoretical Screening. Theoretical screening refers to the calculation of maximum rate constants (kd(max)) and minimum half-lives based on the λ of test substances at 295-800 nm and the measured midseason day average solar irradiance (Lλ, millieinsteins cm-2d-1) (1, 17, 18, 21). The measured quantum yield (φ) of propranolol was 2.22 × 10-3 under the sunlight (14) and 5.2 × 10-3 under a xenon lamp (295-800 nm) with 765 W m-2 total photon flux (15) at pH 5.5. Measured quantum yields for atenolol and metoprolol were not available; therefore, their predicted maximum rate constants and minimum half-lives were corrected by considering the quantum yield of propranolol measured under the sunlight (eqs 1 and 2).

kd (max)corr. (d-1) ) φpropΣλLλ t1/2(min)corr (d) )

ln 2 kd(max)corr

(1) (2)

Where kd (max)corr. is the corrected maximum rate constant (d-1); φprop is the quantum yield of propranolol (14); Lλ is the mid-summer daytime average solar irradiation (millieinsteins cm-2 d-1) (18); and t1/2 (min)corr is the estimated minimum half-life (d). An example for how to estimate kd(max) and 804

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 3, 2007

kd(max)corr. is given in Table S1. The calculated t1/2(max)corr values were then used for the experimental design of detailed kinetics studies of the β-blockers at different starting concentrations. Direct Photolysis Kinetics. A modified version of the OECD draft test guideline (1) was applied for the measurement of direct photolysis kinetics of the β-blockers and relating them to the near surface sunlight conditions. The method was reported previously (5). In summary, samples in deionized water (0.3 µg L-1 to 10 mg L-1) were placed in borosilicate glass reaction vessels (4.6 cm i.d. × 3.2 cm depth) with quartz glass lids and exposed to a solar irradiator (Hanau Suntest CPS) with a 1.1 kW xenon (Xe) arc lamp. The UV (800 nm) were removed through a system of filters. Light intensity of the Xe lamp was set at its maximum level and was measured at the beginning and the end of the experiment with a spectroradiometer (SpectRad). Dark control solutions at the same concentrations were kept in darkness. Temperature was maintained at 20-26 °C and was measured every hour by a digital temperature recorder connected to thermocouples over the reaction period for both the samples and the dark controls. The time course of each reaction was measured, for the different concentrations, by taking an aliquot from the sample (∼1 mL) at regular intervals and measuring the concentration of the test substance with HPLC and LC-MS. The reaction rate constants (kd(xe)) were determined by fitting a pseudo first-order regression to the data. The measured rate constants were then converted to the rate constants under near surface sunlight conditions (5, 6). In brief, three factors were applied that, when combined, addressed differences of light intensity, light penetration, and the time a chemical spends in the phototransformation layer due to river mixing. The extrapolation was based on the similarity of polychromatic irradiation under the filtered xenon lamp (295-800 nm) and the sun (Figure S1, Supporting Information). Chemical Analysis by HPLC and LC-MS(/MS). The initial and photodegraded β-blockers in samples and dark controls were measured by HPLC coupled with a 996 photodiode array detector (Waters) and LC-MS with an electrospray

ionization (ESI) interface (Finnigan TSQ Quantum Ultra, Thermo Electron Corporation). The HPLC method was used for samples of starting concentrations at 0.1, 1, and 10 mg L-1. β-Blockers were separated on a Genesis Aq RP C18 column (120 Å 4 µm, 150 × 4.6 mm). Quantification was based on the chromatograph at 214 nm for propranolol and 220 nm for metoprolol and atenolol. The linear range (through zero) of external standard curves was 10 µg L-1 to 10 mg L-1 for propranolol (R2 ) 0.9987), 10 µg L-1 to 1 mg L-1 for atenolol (R2 ) 0.9997), and 1-10 mg L-1 for metoprolol (R2 ) 0.9974). The LC-MS method was used for samples of starting concentrations at 0.3, 1, and 10 µg L-1. β-Blockers were separated on a Genesis Aq RP C18 column (120 Å 4 µm, 15 cm × 2.1 mm) and detected by positive ion MS SIM code (propranolol: retention time ) 13.0 min, [M + H]+ ) 260; metoprolol: retention time ) 11.4 min, [M + H]+ ) 268, atenolol: retention time ) 6.2 min, [M + H]+ ) 267). The method limits of detection (LOD) were determined experimentally from the lowest standard concentration in the linear range with a signal-to-noise ratio g4. LOD values of atenolol, metoprolol, and propranolol were 1.4, 10, and 10 µg L-1, respectively for the HPLC method and 0.5, 0.5, and 0.05 µg L-1, respectively for the LC-MS method. Propranolol and its direct photolysis products at 0, 18, 42, 48, and 66 h were identified by positive ion LC-MS (TSQ Quantum Instrument) operated in full scan mode. The sample at 48 h was further identified by LC-MS/MS. Daughter ions of mass m/z 260, 292, 264, 308, and 310 ([M+H]+ ion) were observed over a mass range of 50-270 amu at 0.5 s per scan. The collision energy was set at 10, 20, 30, and 40 eV. Other LC and MS conditions were the same as described previously. HPLC Fractionation and Confirmation of Direct Photolysis Products by NMR. To further confirm the degradation products, propranolol samples (10 and 100 mg L-1) were irradiated for 23 h at the conditions described in the previous section. Dark controls at the same concentration were prepared but not irradiated. Degradation products of propranolol were separated on a Genesis Aq RP C18 column (120 Å 4 µm, 250 × 4.6 mm) by a preparative HPLC (Agilent 1100 series fraction collection system) with a UV detector set at 280 nm. Fractions were collected according to the identified peaks with a signal-to-noise ratio of 4:1. Approximately 1 mL of HPLC collected fractions of five major degradation products (P1-P5) was blown down to dryness by N2 gas and then made up in 200 µL CD3CN prior to the identification by 1H-NMR (Brucker, model A600) equipped with a 3 mm PABBI probe at 300K. Mass Balance of 14C-Propranolol and its Direct Photolysis Products. Mass balance of photodegraded propranolol and its degradation products was measured by using radiolabeled propranolol. 14C-Propranolol (1 and 10 mg L-1, 1635 Bq µg-1) was irradiated for up to 46 h. Samples (1 mg L-1: at 0, 6, and 21 h; 10 mg L-1: at 0, 6, 27, and 46 h) were taken and separated by the same HPLC method as described in the previous section (Waters HPLC). Fractions for the analysis of 14C activity were collected from 0 to 30 min (F1: 0-5 min, F2: 5-10 min, F3: 10-12 min, F4-F20: every 0.5 min from 12 to 20.5 min, F21: 20.5-25 min, and F22: 25-30 min). These fractions from the irradiated samples and matching fractions from the dark control were analyzed by liquid scintillation counting (LSC) to determine the 14C radioactivity. Measurement of Radical Formation During Direct Photolysis by EPR Spectroscopy. EPR spectra were recorded at room temperature (22 ( 2 °C) on a cw-Bruker EMX X-band spectrometer, operating at the 100 kHz field modulation, using 100 mW of microwave power and equipped with an ER4104OR optical transmission cavity. A LOT Oriel 500 W xenon arc lamp, fitted with a liquid IR filter, a less than 310

FIGURE 2. Reaction kinetics for the direct photolysis of (a) propranolol at (1) 25 °C, (2) 20 °C, (3) 22 °C, (4) 23 °C; (b) metoprolol at 24 °C, and (c) atenolol at (1) 24 °C and (2) 26 °C. Points are measured data and lines are predicted data by pseudo first order fitting of the measured data. The pseudo first order reaction rate constants and half-lives are included in Table 1. Other linear regressions in (a) are 0.001 mg L-1 (4), y ) -0.052x-0.18, R2 ) 0.978; 0.1 mg L-1 (3), y ) -0.049x-0.028, R2 ) 0.996; 0.1 mg L-1 (2), y ) -0.037x-0.044, R2 ) 0.998.

nm cut off filter, and a focusing assembly was used to direct the filtered UV-visible light down a 3 mm diameter quartz rod into the EPR sample cavity and onto the sample. β-Blocker solutions in deionized water (1000 mg L-1), both in the presence and absence of the spin trap 5,5-dimethyl1-pyrroline N-oxide (DMPO, ∼1.5 mg mL-1, Sigma Aldrich), were placed in an EPR flat cell and exposed to light in situ in the spectrometer for up to 30 min. The spin trap reacted VOL. 41, NO. 3, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

805

TABLE 1. Measured and Extrapolated Direct Photolysis Kinetics of β-blockers in Water

β-blocker propranolol

starting concentration (mg L-1) 1 0.1 0.01 0.001 0.0003

average ( SD metoprolol atenolol

temperature (°C)

kd(xe) (h-1)

20 ( 0 25 ( 1 20 ( 0 22 ( 1 22 ( 1 23 ( 1 23 ( 1

0.033 0.051 0.036 0.045 0.045 0.051 0.058

t1/2(Xe) (h)

< 12.5b 10 1 0.1 0.01 200c

t1/2(solar) (d) June, C.EU 52 °N

21 14 19 15 15 14 12

1.0 0.7 1.0 0.8 0.8 0.7 0.6

16 ( 3

0.8 ( 0.2 1.7

a

t1/2(solar) (d) Dec., C.EU 52 °N 12 7.6 11 8.5 8.5 7.5 6.7 8.7 ( 1.8 6.0

t1/2(solar) (d) June, U.S. 40 °N

t1/2(solar) (d) Dec., U.S. 40 °N

1.9 1.3 1.8 1.4 1.4 1.3 1.1

6.8 4.5 6.3 5.0 5.0 4.4 3.9

1.5 ( 0.3

5.1 ( 1.0

24 ( 1 24 ( 1

0.0007 0.0011

990 630

44 28

449 286

95 61

298 190

26 ( 2 24 ( 1

0.0019 0.0020

365 347 540

16 15

166 157

35 33

110 104

a Results are the means of two measurements and four analyses. Half-lives under solar irradiation were extrapolated from the xenon results by assuming the quantum yields were independent of the wavelength (5, 6). Light intensity data for Central EU and the U.S. were from refs 21 and 18, respectively. Kd(Xe) is the first order direct photolysis rate constant under the xenon lamp (295-800 nm). t1/2(Xe) is the half-life under the xenon lamp; t1/2(solar) is the half-life for near-surface experimental conditions. b Results were measured values from ref 14. Samples were exposed to the sun in spring or summer in Italy. Starting concentration and irradiation period were not stated in the paper. The number was calculated based on personal communication with Andreozzi, R. (July 2004). c Results were calculated from data given in ref 16. Samples were exposed to a filtered xenon lamp.

with free radical species to form a stable spin adduct, as shown below, that can then be detected by EPR spectroscopy.

photon flux of the xenon lamp and the sun, it was found more efficient to undertake a theoretical screening followed by detailed kinetics studies of β-blockers.

Results and Discussion

A control sample of DMPO only in water (∼1.5 mg mL-1) was also exposed to the same light and analyzed under the same conditions, in order to investigate the photostability of DMPO itself under these conditions. All the spectra obtained were simulated to aid interpretation, using the SimEPR32 program (19). Quality Control and Rationale for the Experimental Design. The light intensity at six different vertical and horizontal positions of the reaction chamber was measured before the direct photolysis experiments were carried out. No significant difference in the photon fluxes was observed at different horizontal positions; however, total photon fluxes varied at different vertical positions (5, 10, and 15 cm from the bench) ranging from 12.8 to 40.3 W m-2. Therefore, all samples were exposed at the same vertical position of 15.9 cm from the bench to gain the maximum light intensity from the xenon lamp (approximately 41 W m-2). During the reaction periods, the change from starting concentrations of the dark controls was less than 1% for 0.1 and 1 mg L-1 samples and less than 10% for 0.0003, 0.001, and 0.01 mg L-1 samples, possibly representing the variation and sensitivity of the HPLC and LC-MS analyses. Therefore, it was demonstrated that dark controls for all samples were not photodegraded in pure water. These dark control results also represented noncatalyzed hydrolysis at environmentally relevant concentrations and non-stressed conditions (pH ∼7 and 20-25 °C), therefore, may be useful for ERA of the β-blockers. No actonimeters were used for direct photolysis of β-blockers because there are already measured values for propranolol in water (14). Although it is better to have quantum yields when the theoretical screening is undertaken, it must be noted that the determination of quantum yields can be more tedious than the kinetics studies and that they only generate an estimated rate constants. With the measured 806

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 3, 2007

UV-Visible Spectra and Theoretical Screening. The UVvisible spectra of metoprolol and atenolol in deionized water (pH 7, 0.05 M phosphate buffer) were very similar with maximum absorbance at ∼220 and 273 nm (Figure 1). However the spectrum of propranolol was different from that of metoprolol and atenolol. It showed absorbance at >295 nm range, indicating that it may undergo solar initiated direct photolysis. This is in agreement to the previous findings that the naphthalene skeleton in propranolol structure suggests the possibility for photoinitiated reactions (20). In spite of the difference in structures and UV-visible spectra of the β-blockers, the calculated maximum rate constants kd(max) of propranolol, atenolol, and metoprolol under Central Europe summer conditions were very similar at 1655, 1472, and 1609 d-1, respectively (Figure 1). This represents the best scenario assuming the quantum yields (φ) are 1. In natural waters, φ of organics is almost always less than 1. The φ of propranolol was 0.002 at pH 5.5 (14), therefore, the corrected maximum rate constant kd(max)corr. for propranolol was 3.3 d-1 and the corrected minimum half-life t1/2(min)corr was 0.21 d. The t1/2(min)corr of propranolol was approximately 8 times lower than the half-life measured under natural sunlight (14). There is no measured quantum yield of atenolol and metoprolol available; however, it is hypothesized that the φ values of atenolol and metoprolol would be smaller than propranolol because there is only one aromatic ring in their structures. Therefore, half-lives of atenolol and metoprolol would be longer than propranolol. Using the φ for propranolol, the corrected kd(max)corr. for atenolol and metoprolol were 2.9 and 3.2 d-1, respectively. Note that Lλ values used for this estimation are for clear sky conditions, for water bodies with shallow depths, and for chemicals with very weak absorbance in pure water. The variation of the radiation intensity caused by climatic factors may be higher than that caused by the geographic location (21). Direct Photolysis Kinetics. In optically dilute solutions, i.e., absorbance (A) < 0.02, direct photolysis of all three

FIGURE 3. HPLC separation of propranolol and its direct photolysis products at 10 and 100 mg L-1. Peaks P1-P5 were degradation products and P6 was propranolol. The reaction time was 23 h. measured β-blockers followed pseudo first order reaction kinetics (Figure 2). The measured and subsequently extrapolated first order rate constants and half-lives are summarized in Table 1. The extrapolation assumed that the quantum yields of the β-blockers were independent of wavelength for the xenon lamp and the sun (1, 5, 6). The difference of half-lives between Central Europe and the U.S. in the summer and winter demonstrated the effects of latitude and light intensity on the direct photolysis kinetics under clear sky conditions. It was likely that light intensity was reduced 30% due to cloud cover in Central Europe (21); however, further studies on photolysis under cloudy sky conditions may be necessary at different locations. Propranolol. Propranolol underwent relatively fast direct photolysis with measured rate constants, kd(xe) between 0.033 and 0.058 h-1 (Figure 2a, Table 1). Measured kinetics on propranolol photoreaction was significantly different at a starting concentration of 10 mg L-1 and e1 mg L-1 (Figure 2a). Two-way ANOVA was applied to further determine the impact of concentrations (1 and 0.1 mg L-1) and reaction time (1, 3, 6, 21, 24, 27, and 46 h) on the rate constants, kd(xe). Two samples were included for each concentration. The analysis showed that reaction time had an effect on the

measured rate constants (F ) 2.93, p ) 0.020), possibly due to the nonequilibrium at the beginning of the reaction ( 0.02), thus did not fit to a simple first order regression. Furthermore, the reaction kinetics was not significantly different at the temperature range of 20-26 °C, although slightly faster reactions were observed at a higher temperature (25 °C) for propranolol. The extrapolated half-lives of propranolol in Central Europe and the U.S. were between 0.6 and 1.9 d in summer and between 3.9 and 12 d in winter (Table 1). The extrapolated half-life in Central Europe was comparable to the summer and winter results of propranolol reported by Andreozzi et al. (14), which was measured in the field under solar irradiation. However, the extrapolated half-lives in California, U.S. in winter, were smaller than the values for Central Europe, representing the influence of the weather differences between the U.S. and Europe. VOL. 41, NO. 3, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

807

FIGURE 4. Proposed direct photolysis pathway of propranolol based on LC-MS/MS and 1H-NMR. Metoprolol. Direct photolysis of metoprolol showed first order kinetics at the tested concentrations of 1 and 10 mg L-1 (Figure 2b). However, the reaction was 5 orders of magnitude slower than the theoretical maximum rate constant (Figure 1) and was much slower than the direct photolysis of propranolol with measured kinetics kd(xe) at approximately 0.0007-0.0011 h-1 (Table 1). This confirmed the hypothesis that chromophore structure differences of propranolol and metoprolol had a great influence on the direct photolysis kinetics of the two compounds, i.e., propranolol has a naphthalene backbone and metoprolol has a benzoic backbone. The extrapolated half-lives of metoprolol in Central Europe and the U.S. ranged from 28 to 95 d in summer and from 190 to 449 d in winter (Table 1). The half-lives were much longer in the Central EU (52 °N) than in the U.S. (40°N), reflecting the geographic impact on the light intensity. Atenolol. Again, in optically dilute solutions (0.1 and 0.01 mg L-1), the photoreaction fit well into the pseudo first order hypothesis (Figure 2c). The rate constants (0.002 h-1) and half-lives were similar at 0.1 and 0.01 mg L-1 and were approximately 3 orders of magnitude slower than the kd(max) (Figure 1). However, at a higher concentration (10 mg L-1), the reaction curve could be fitted by a Weibull distribution (Figure S3, Supporting Information), thus the first order rate constant for this reaction could not be generated. The different kinetics at the starting concentration at 10 mg L-1 may be because the solution was not optically dilute again. In addition to the results from propranolol (Figure 2a), it further suggests that for future experimental design, measurements at optically dilute solutions would be needed and this would also be more relevant to environmental concentrations. The extrapolated half-lives of atenolol in Central Europe and the U.S. ranged from 15 to 35 d in summer and from 104 to 166 d in winter (Table 1). The half-lives were much longer in the Central EU (52°N) than in the U.S. (40 °N); again, reflecting the geographic impact on the light intensity. Direct Photolysis Products from Propranolol. Direct photolysis of propranolol produced approximately 20 small products (Figures 3 and 5). Three major direct photolysis products of propranolol were identified at 46-hour reaction time (Figures 3 and 4). The electro spray positive ionization of MS generated the protonated molecular ions [M + H]+ 808

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 3, 2007

(m/z ) 260) together with their daughter ions. The MS and MS/MS data on protonated molecular ions and their daughter ions were then used for subsequent interpretation of major degradation products of propranolol P1, P2, and P4. P1 and P2 had the same molecular ion ([M + H]+ ) 292) and the same daughter ions. P1, P2, and P3 also showed the same 1H-NMR spectra, confirming an aldehyde backbone. Therefore, it suggested P1 and P2 were structural isomers. P3 was not further identified because it was not found in samples at environmentally relevant concentrations (