Oxidation of Acetaminophen by Fluidized-bed Fenton Process: Optimization using Box-Behnken Design

Oxidation of Acetaminophen by Fluidized-bed Fenton Process: Optimization using Box-Behnken Design

• 1.
  Oxidation of Acetaminophenby Fluidized-bed Fenton Process: Optimization using Box-Behnken Design M.C. Lu*, R.M. Briones**, and M.D.G. de Luna**, *** *Department of Environmental Resources Management, Chia Nan University of Pharmacy and Science, Tainan 717, Taiwan (E-mail: [email protected]) ** Environmental Engineering Graduate Program, University of the Philippines, 1011 Diliman, Quezon City, Philippines (Email: [email protected]) *** Department of Chemical Engineering, University of the Philippines, 1011 Diliman, Quezon City, Philippines Abstract One of the most frequently used over-the-counter analgesic and antipyretic is acetaminophen (ACT). This drug finds its way into effluent wastewaters in concentrations that still pose an environmental threat even after conventional treatment. This study demonstrates the effectiveness of a fluidized-bed Fenton process in the degradation of acetaminophen in synthetic wastewater. Box-Behnken experimental design was employed to optimize initial pH, Fe2+ and H2O2 concentrations. The best ACT removal was achieved at pH 3.00 with both initial Fe2+ and H2O2 concentrations at their maximum. At this operating condition, almost 98% degradation was attained within only 20 minutes of reaction time. Optimization of operating conditions gave the best removal efficiency at pH=3.22, [Fe2+] = 0.06mM and [H2O2] = 19.87mM. Verification studies resulted in a 97.83% ACT removal with an initial rate of 0.234 mM/min. COD and TOC removals of 38.18% and 59.62%, respectively, were achieved. Parametric studies showed that a single stage slow degradation occurs at very low FH ratio. At higher FH ratio, a fast initial degradation followed by slow degradation occurs. Keywords: Fluidized-bed Fenton process; Box-Behnken design; acetaminophen INTRODUCTION Pharmaceuticals and personal care products (PPCPs) from pharmaceutical companies and human use reach surface waters usually unaltered. PPCPs are persistent and mostly resistant to microbial attack. Most of these substances are found to be endocrine disrupting compounds (EDCs), an emerging class of pollutants. Even at low concentrations, PPCPs have shown severe effects to the environment. A recent study attributed the rapid decline of vulture population in Northern India to diclofenac, a non-steroidal anti-inflammatory drug (NSAID) used as analgesic to treat arthritic and rheumatic conditions (Rizzo, et al., 2009). PPCPs were not only detected in effluents from wastewater treatment plants (WWTPs) but also in surface and ground waters (Larsen, et al., 2004). The discharge of these substances into sewers and WWTPs poses serious problems and challenges in their removal. Existing water and WWTPs are not designed to remove these unregulated micro- pollutants. Many studies have shown that the reduction of pharmaceutical compounds in conventional WWTPs is usually incomplete. Though incomplete, WWTPs with tertiary treatment have better removal of PPCPs (Zhou, et al., 2009). To mitigate environmental contamination from PPCPs, complete removal of these compounds in wastewaters is the best solution. Acetaminophen (ACT) is a non-steroidal anti-inflammatory drug (NSAID) used as an over-the- counter analgesic and antipyretic. It is one of the most frequently used drugs and ranked as the 5th most used drug in the Philippines in 2006 (IMS, 2007). When ingested, 58-68% of this drug is
• 2.
  excreted by thebody unchanged (Zhang, et al., 2008). It is present in wastewaters in concentrations exceeding 1000 ng/L (Wiegel, et al., 2004). Figure 1. Structure of acetaminophen Advanced oxidation processes (AOPs) can achieve total degradation of target pollutants producing organic acids, inorganic salts and CO2 as by-products. These processes generate hydroxyl radicals (•OH) in solution. Hydroxyl radicals are very powerful nonselective oxidizing agents which attack organic matter in wastewater thereby promoting its degradation. Fenton oxidation, one of the most important AOPs known, utilizes Fenton’s reagent, a combination of H2O2 and Fe2+, to produce hydroxyl radicals for complete mineralization of organics (Khataee, et al., 2008). (1) (2) (3) Values of rate constants suggest that ferric ions are produced (Equation 1) more rapidly than they are reduced to Fe2+ (Equation 3) resulting in the formation of excess Fe3+at the end of the process. Neutralization of the final solution leads to the formation and accumulation of Fe(OH)3 sludge -a major disadvantage of the Fenton process. Fluidized-bed (FB) Fenton minimizes sludge production as the carrier itself acts as a seeding material for crystallization of Fe3+ ions. Aside from the homogenous catalytic reaction between H2O2 and Fe2+ in solution, heterogeneous chemical reaction also occurs as the iron oxide-coated carrier also acts as a catalyst to produce •OH (Chou, Huang, & Huang, 1999). This study focused on the degradation of synthetic acetaminophen wastewater by chemical oxidation using FB-Fenton process. It is a preliminary research in advanced wastewater treatment technology particularly in sewage and pharmaceutical industries. MATERIALS AND METHODS Chemicals and analytical methods All chemicals used, including 35% H2O2, 4-hydroxy acetanilide (ACT), FeSO4•7H2O, HClO4, HCl, H2SO4, NH4C2H3O2, C12H8N2•H2O, K2TiO4, acetonitrile, NaOH, were purchased from Merck. All solutions were prepared using Millipore system deionized water with a resistivity of 18.2 M . Residual H2O2 in solution was analyzed using titanium oxalate method. Residual ferrous concentration was determined by complexation with 1,10-phenanthroline. Both methods were analyzed using a Thermo Spectronic Genesys 20 spectrophotometer at 400 nm for H2O2 and 510 nm for ferrous ions. ACT concentration was determined with SpectraSYSTEM SN4000 HPLC equipped with Asahipak ODP-50 6D using 20 mM phosphoric acid and acetonitrile at 85:15, flow rate of 1 mL/min and at 220 nm. COD was measured using closed reflux titrimetric method,
• 3.
  Standard Methods 5220C. Total iron concentration was determined by Perkin Elmer AAnalyst 200 AAS. Fluidized-bed Fenton and Fenton experiments The reactor with a working volume of 1.45 L was made of a cylindrical glass with inlet, outlet and recirculating sections. It was equipped with a Suntex portable pH meter. All batch experiments were done at room temperature. Synthetic acetaminophen wastewater at 5 mM was poured into the reactor and the pump was turned on. The initial pH was adjusted by adding concentrated HClO4 or 0.1N NaOH. The desired amount of FeSO4•H2O was added into the solution as ferrous source. Glass beads of diameters 4 mm and 2 mm were added as support followed by the addition of SiO2 carrier with a diameter of 0.5 mm. The pH of the solution was further adjusted. Samples were taken for analyses of initial conditions. Samples taken were immediately injected into tubes containing sodium hydroxide solution to quench Fenton reaction and were filtered through a 0.22 µm syringe microfilters. Hydrogen peroxide was finally added to start the reaction. The pH of the solution was not further adjusted as the reaction proceeded. Samples were taken at different time intervals of 0, 3, 5, 10, 20, 40, 60, 90 and 120 minutes. Fenton experiments were done following the same procedure as in fluidized-bed Fenton process but without the addition of glass bead support and SiO2 carriers. Design of experiment Optimization of operating conditions was done using Box-Behnken design (BBD), a three-level design used to fit second-order models. It can be expanded to estimate the combinations of third order terms, i.e. x12x2, x12x3 and x1x22(Davis & Draper, 1998). This design has the advantage of having very efficient number of required runs to fit the model. Design-Expert 7.0 software (Stat- Ease, Inc., Minneapolis, USA) was used to determine the number of experiments needed to optimize and analyze the system. Three important parameters namely: pH, Fe2+ and H2O2, were studied and optimized using BBD. A total of 17 experimental runs were conducted with five replicates at the center point. All runs were conducted at room temperature. Based on previous FB-Fenton studies (Muangthai, Ratanatamsakul, & Lu, 2010), the amount of carrier material has little significant impact on removal efficiency. Hence, its amount and size were fixed at 100g and 0.5mm respectively. Table 1 shows the levels for each factor used in the BBD. Table 1. Levels of factors used in Box-Behnken Design Levels Factors Symbol Low (-1) Center (0) High (+1) pH A 2 3 4 Fe2+ (mM) B 0.01 0.055 0.1 H2O2 (mM) C 5 15 25 RESULTS AND DISCUSSION Box-Behnken design Aside from fitting a model for ACT removal, initial rate was also included as a response to predict the efficiency of the process at optimum conditions. Using Design-Expert software, a reduced cubic model best fits ACT removal. For initial rate, a full quadratic model is sufficient and gives an
• 4.
  insignificant lack offit. Results of ANOVA gives adjusted R2 values of 0.9979 and 0.9764 for ACT removal and initial rate, respectively. The best ACT removal was achieved at pH 3 and with the highest initial Fe2+ and H2O2 concentrations of 0.01 and 25 mM, respectively. Almost 98% degradation was attained within only 20 min of reaction time. The worst removal was 20.91% at pH=2, [Fe2+]=0.01mM and [H2O2]=15mM. Correlation of each factor on ACT removal and initial rate. Table 2 shows the correlation of each parameter studied on ACT removal efficiency and initial rate. All three parameters have significant positive effects on both responses and must be considered in the analysis of the effect of each factor. Table 2. Correlation values of each factor on ACT removal and initial rate Correlation Factor ACT removal Initial rate pH 0.304 0.481 Fe2+ 0.429 0.654 H2O2 0.555 0.211 Although all parameters have positive effect on ACT removal, 3D surface plots reveal pH level and Fe2+ concentration levels where removal of ACT start to decrease.
• 5.
  (a) (a) (b) (b) (c) (c) Figure 2. 3D surface plots of the two parameter Figure 3. 3D surface plots of the two parameter interaction effects of initial pH, [Fe2+] and interaction effects of initial pH, [Fe2+] and [H2O2] on initial rates: (a) [H2O2]=25mM, (b) [H2O2] on ACT removal: (a) [H2O2]=25mM, pH=3, (c) [Fe2+]=0.055mM. (b) pH=3, (c) [Fe2+]=0.055mM.
• 6.
  Optimization using BBD.As generated by Design Expert 7.0 software, the equations for ACT removal and initial rate, respectively, by fluidized-bed Fenton process in terms of coded factors are as follows: % ACT removal= 92.99 + 3.15A + 5.51B + 16.89C – 12.68AB + 3.17AC + 4.25BC – 10.72A2 - 12.71B2 – 7.57C2 + 15.09A2B + 12.21AB2 Initial rate, mM/min = 0.22 + 0.21A + 0.29B + 0.094C + 0.28AB – 0.0079AC + 0.099BC + 0.073A2 + 0.12B2 + 0.014C2 where A, B and C are initial pH, initial Fe2+ concentration and initial H2O2 concentration, respectively with values of (-1) to (1) indicating the level. For a cost effective operation, the amount of chemicals added were kept at a minimum while ACT removal was set to a target of 95-100% for best removal efficiency. The software generated only one solution with the criteria as shown in Table 3. Table 3. Optimum condition factors and responses as predicted by Design-Expert 7.0 Factor Condition pH 3.22 [Fe2+] 0.06 mM [H2O2] 19.87 mM ACT removal 100% Initial rate 0.3185 mM/min Comparison between conventional Fenton and fluidized-bed Fenton process at the optimum condition. To validate the model generated by the software, FB-Fenton reaction was carried out at optimum conditions and the resulting efficiency was compared to that of conventional Fenton process using the same parameters. As shown in Table 4, the experimental results obtained for the model at optimum conditions were close to the predicted values indicating a good fit for the range of concentrations investigated. Table 4. Comparison between actual and predicted values Response Actual Predicted Difference ACT removal, % 97.83 100 2.17 Initial rate, mM/min 0.2343 0.3185 0.0842 Figure 4 shows the comparison between Fenton and fluidized-bed Fenton using optimum conditions of the latter. The trends for both methods were almost the same for residual H2O2, residual ACT and residual COD. The marked difference is evident in total residual iron. Total residual iron from the Fenton process was much higher than that from the FB-Fenton process. FB-Fenton process resulted to 62.92% iron removal compared to only 9.06% using Fenton process. This was expected since FB-Fenton was developed to reduce sludge formation in the form of iron precipitates. The presence of SiO2 in the reactor provides a site for crystallization of iron oxides onto the surface of these carriers thereby reducing the amount of iron in solution. SEM/EDS analysis supports this as results showed an increase in iron from 0.83% to 2.10%.
• 7.
  (a) (c) Fenton process (a) Fenton process FB-Fenton process 1.0 FB-Fenton process 1.0 0.8 0.8 Total iron (C/Co) [H2O2] (C/Co) 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0 20 40 60 80 100 120 0 20 40 60 80 100 120 Time (min) Time (min) (b) (d) Fenton process Fenton process FB-Fenton process 1.0 FB-Fenton process 1.0 0.8 0.8 [ACT] (C/Co) COD (C/Co) 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0 20 40 60 80 100 120 0 20 40 60 80 100 120 Time (min) Time (min) Figure 4. Comparison of residual (a) H2O2, (b) ACT (c) total iron and (d) COD between Fenton and Fluidized-bed Fenton processes at optimum condition: [ACT]=5mM, pH=3.22, [Fe2+]=0.06mM, [H2O2]=19.87mM Parametric Studies To be able to discuss in detail the effect of each factor on the removal efficiency, parametric studies were done. This involved changing the value of the parameter being studied while keeping other variables constant. Effect of initial pH. Previous studies have shown that the pH of the contaminated solution is a very important parameter that should be controlled in Fenton processes to achieve effective removal. The concentration of Fe2+ in solution was found to be maximum at pH=2.8 (Brillas, Sires, & Oturan, 2009). However, the operative optimum pH depends on the pollutant/s in solution. As shown in Fig. 5, only a small amount of ACT was removed at pH=2 (20.9%) as compared to the removal at pH=3 (68.0%) and pH=4 (77.0%). But if pH of the solution was not adjusted (pH=6.5), ACT removal decreased drastically to 8.65%. At a very low pH, the high H+ concentration in solution scavenges •OH as shown in Equation 4 thereby decreasing the degradation rate (Devi, et al., 2010). Also, at low pH conditions, reduction of ferric to ferrous is inhibited (Equation 5). (4) (5)
• 8.
  ACT removal 0.3 Initial rate 80 Initial rate (mM/min) ACT removal (%) 60 0.2 40 0.1 20 0 0.0 2 4 6 pH Figure 5. Effect of pH: [ACT]=5 mM, [Fe2+]=0.01 mM, [H2O2]=15 mM At pH between 3 and 5, the predominant species is Fe2+ and degradation occurs at a faster rate in this region. However, at a very high pH, Fe2+ is unstable and is easily oxidized to Fe3+ in solution. This precipitates out as Fe(OH)3 and reduces the amount of free Fe2+ in solution to catalyze Fenton reaction. Also, hydrogen peroxide is unstable at higher pH levels as it decomposes to water and oxygen. (6) Effect of initial ACT concentration. To determine the effect of acetaminophen on the degradation rate and removal, different concentrations of 2.5, 5, 7.5 and 10mM ACT were treated while fixing other factors constant at pH=3, [Fe2+]=0.01mM and [H2O2]=5mM. Results showed a 50% decrease in ACT removal after 2 hours of reaction time as the concentration was increased 4 times from 2.5 to 10mM. The decrease in the rate of degradation at higher concentrations was observed because there is lower hydroxyl radical available in the solution compared to the target organic compounds. A lower oxidant to pollutant ratio resulted to a decrease in removal efficiency. It is also important to note that the degradation rate follows the trend of H2O2 concentration in solution. This implies that the amount of H2O2 available is directly related to the removal rate. If there is no or little observed decline in H2O2 concentration, it is possible that ACT degradation has stopped and that the process has reached its maximum removal effectiveness. In Fig. 7 (a) and (b), ACT concentration is almost constant at 40 min which coincides with the slow H2O2 disappearance.
• 9.
  0.20 ACT removal Initial rate 60 0.15 Initial rate (mM/min) ACT removal (%) 45 0.10 30 0.05 15 0 0.00 3 6 9 [ACT] (mM) Figure 6. Effect of ACT: pH=3, [Fe2+]=0.01mM, [H2O2]=5mM (a) (b) 1.2 ACT=5mM ACT=5mM ACT=10mM 1.0 ACT=10mM 1.0 0.8 0.8 [H2O2] (C/Co) [ACT] (C/Co) 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0 20 40 60 80 100 120 0 20 40 60 80 100 120 Time (min) Time (min) Figure 7. Effect of ACT on FB-Fenton process: pH=3, [Fe2+]=0.1mM, [H2O2]=25mM, (a) ACT remaining and (b) residual H2O2 Effect of initial [Fe2+]:[H2O2] ratio. Removal of ACT increased as initial Fe2+ concentration increased from 0.01 to 0.1 mM as depicted in Fig. 8. However, at Fe2+ concentration of 0.1 mM, ACT degradation started to decrease. This suggests that there is a competing reaction which involves Fe2+ aside from its catalytic role shown in Equation 1. Fe2+ does not act merely as a catalyst to speed up Fenton reaction, it also reacts with •OH (Equation 7) (Kang, Lee, & Yoon, 2002).There is a certain Fe2+:H2O2 ratio wherein scavenging of •OH in the solution manifests as observed by a decrease in removal efficiency. (7)
• 10.
  ACT removal ACT removal 65 1.0 2.0 Initial rate Initial rate 100 60 0.8 95 1.5 Initial rate (mM/min) Initial rate (mM/min) ACT removal (%) ACT removal (%) 55 0.6 90 1.0 50 0.4 85 0.5 45 0.2 80 40 0.0 75 0.0 0.0 0.3 0.6 0.0 0.3 0.6 2+ 2+ [Fe ] (mM) [Fe ] (mM) Figure 8. Effect of Fe2+: pH=3, [ACT]=5 mM, Figure 9. Effect of Fe2+: pH=3, [ACT]=5mM, [H2O2]=5 mM [H2O2]=25mM ACT removal 0.10 Initial rate 80 0.08 Initial rate (mM/min) ACT removal (%) 0.06 70 0.04 60 0.02 50 0.00 0 10 20 30 [H2O2] (mM) Figure 10. Effect of H2O2: pH=3, [ACT]=5 mM, [Fe2+]=0.01 mM Aside from organic compounds, •OH can react with other species present in solution (Ting, Lu, & Huang, 2009) which depletes the amount of available •OH in solution. (8) (9) (10) For the concentration range of H2O2 studied between 5 and 25 mM at [Fe2+]=0.01 mM, there was no observed decrease in ACT removal. Figure 10 shows that an increase in H2O2 concentration leads to better removal. However, as the concentration was increased from 20 to 25mM, there was only a slight increase in removal. From previous studies made, a very high H2O2 concentration resulted in a decreased reduction because this H2O2 will scavenge the •OH produced as in Equation 8. There is a certain threshold concentration wherein the effect of increasing H2O2 does not result to an increase in removal. Beyond this optimum condition, it was observed that little or no change in degradation occurs.
• 11.
  Another observation fromthis study was the difference in degradation rates at various Fe2+ concentrations not only at low H2O2 but also at high H2O2 concentrations. It was observed that for some runs, degradation occurs in two stages (Fig. 7(a)). There is a fast initial degradation rate followed by a slow rate in ACT removal. However, for other runs, degradation of ACT occurs in a single slow rate all throughout the reaction time. The two-stage degradation was observed to occur at [Fe2+] higher than 0.02mM. This two-stage degradation is due to the very fast reaction of ferrous ions with hydrogen peroxide (Lu, Chen, & Chang, 1999). During the first minutes, there is a large amount or hydroxyl radical produced that attacks organic compounds. This results to a rapid decomposition of ACT in solution (Fig. 8 and 9) for the initial 20 minutes. This first stage is the Fe2+/H2O2 stage. However, as the reaction proceeds, ferrous ions are converted to ferric ions as shown in Equation 1. This limits the amount of ferrous that can react with H2O2 to produce •OH. The Fe3+ions produced reacts with H2O2 at a slower rate to produce hydroperoxyl radicals (•OH2) as in Equation 3, which is a weaker oxidant with E°=1.65 as compared to the oxidizing potential of •OH at 2.80. Thus, the degradation rate is much slower. The second stage is known as the Fe3+/H2O2 stage. CONCLUSION This study applied Box-Behnken design in the optimization of fluidized-bed Fenton process on the degradation of the drug, acetaminophen. The optimum condition was found to be at pH=3.22, [Fe2+]=0.06mM and [H2O2]=19.87mM. Actual ACT removal at the optimum was 97.83%. Comparison of conventional Fenton and fluidized-bed Fenton process at optimum condition showed that FB-Fenton has reduced sludge formation with total iron removal at 62.92% compared to 9.06% of Fenton process and higher TOC removal, 59.62% vs. 16.37%. Parametric studies revealed that ACT removal without pH adjustment was almost negligible at only 8.65% which proves that initial pH adjustment is crucial and necessary for this system. Increasing Fe2+ concentration to 0.1mM at [H2O2]=5mM decreased ACT removal. Increasing H2O2 concentration always lead to an increase in ACT removal for the concentration range studied. A two-stage degradation rate was observed at Fe2+ concentrations higher than 0.02mM. FB-Fenton proves to be a promising method, requiring minimal space, complete degradation and mineralization with reduced sludge production. ACKNOWLEDGEMENT This research was financially supported by the National Science Council, Taiwan (Grant: NSC 99- 2221-E-041-012-MY3), the Department of Science and Technology, Philippines and the Engineering Research and Development for Technology (ERDT), Philippines. REFERENCES Brillas, E., Sires, I., & Oturan, M. A. (2009). Electro-Fenton Processes and Related Electrochemical Technologies Based on Fenton's Reaction Chemistry. Chemical Reviews . Chou, S., Huang, C., & Huang, Y.-H. (1999). Effect of Fe2+ on catalytic oxidation in a fluidized- bed reactor. Chemosphere 39 (12) , 1997-2006. Davis, T. P., & Draper, N. R. (1998). Fitting 3rd order terms in Box-Behnken Experiments. University of Wisconsin, Department of Statistics. Devi, L. G., Raju, K. S., Kumar, S. G., & Rajashekhar, K. E. (2010). Photo-degradation of di azo dye Bismarck Brown by advanced photo-Fenton process: Influence of inorganic anions and
• 12.
  evaluation of recyclingefficiency of iron powder. Journal of the Taiwan Institute of Chemical Engineers . Kang, N., Lee, D. S., & Yoon, J. (2002). Kinetic modeling of Fenton oxidation of phenol and monochlorophenols. Chemosphere 47 (9) , 915-924. Khataee, A. R., Vatanpour, V., & Amani Ghadim, A. R. (2008). Decolorization of C.I. Acid Blue 9 solution by UV/Nano-TiO2, Fenton, Fenton-like, electro-Fenton and electrocoagulation processes: A comparatice study. Journal of Hazardous Materials 161 (2-3) , 1225-1233. Larsen, T. A., Lienert, J., Joss, A., & Siegrist, H. (2004). How to avoid pharmaceuticals in the aquatic environment. Journal of Biotechnology 113 (1-3) , 295-304. Lu, M.-C., Chen, J.-N., & Chang, C.-P. (1999). Oxidation of dichlorvos with hydrogen peroxide using ferrous ion as catalyst. Journal of Hazardous Materials 65 (3) , 277-288. Muangthai, I., Ratanatamsakul, C., & Lu, M.-C. (2010). Removal of 2,4 dichlorophenol by fluidized-bed Fenton process. Sustain. Environ. Res. 20 (5) , 325-331. Rizzo, L., Meric, S., Kassinos, D., Guida, M., Russo, F., & Belgiorno, V. (2009). Degradation of diclofenac by TiO2 photocatalysis: UV absorbance kinetics and process evaluation through a set of toxicity bioassays. Water Research 43 (4) , 979-988. Ting, W.-P., Lu, M.-C., & Huang, Y.-H. (2009). Kinetics of 2,6-dimethylaniline degradation by electro-Fenton process. Journal of Hazardous Materials 161 (2-3) , 1484-1490. Wiegel, S., Aulinger, A., Brockmeyer, R., Harms, H., Loffler, J., Reincke, H., et al. (2004). Pharmaceuticals in the river Elbe and its tributaries. Chemosphere 57 (2) , 107-126. Zhang, X., Wu, F., Wu, X., Chen, P., & Deng, N. (2008). Photodegradation of acetaminophen in TiO2 suspended solution. Journal of Hazardous Materials 157 (2-3) , 300-307. Zhou, J. L., Zhang, Z. L., Grover, D., & Jiang, J. Q. (2009). Pharmaceutical residues in wastewater treatment works effluents and their impact on receiving river water. Journal of Hazardous Materials 166 (2-3) , 655-661.