דוח מסכם תלת שנתי 01/12/2010עד 31/11/2013 מענק מספר בקרן0455004054 : 1 חלק א ' :פרטים כלליים על המחקר שם המחקר סילוק שאריות תרופות ממי שתייה וקולחים בעזרת תהליכי חמצון מתקדם שמות החוקרים דר' הדס ממן ,דר' דרור אבישר שם המוסד הראשי אוניברסיטת תל אביב תקופת ההתקשרות 3שנים 2 חלק ב ' :רקע מדעי כללי מספר עבודות שנעשו בארה"ב ובאירופה הראו כי ברבים ממקורות המים העיליים הגדולים (נהרות) ואף במאגרי מי תהום נמצאו ריכוזים מדאיגים של שאריות תרופות והורמונים כתוצאה מהזרמת שפכים וקולחים מטוהרים המכילים מזהמים אלו באגני ההיקוות ,בהחדרת קולחים למי-תהום ובהשקיה .מזהמים אלו ,אשר מקורם במטבוליזם לא שלם בגוף האדם ,מגיעים דרך שפכים ביתיים, חקלאיים ותעשייתיים למכון טיהור שפכים .מחקרים שנעשו בארץ ובעולם הוכיחו כי שיטות הטיפול בשפכים הנפוצות כיום ,כגון בוצה משופעלת ,לא מתוכננות להרחקה מלאה של מזהמים אורגניים אלו. תהליכי חמצון מתקדם מבוססי אוזון ידועים כבעלי פוטנציאל גבוה מאוד בפירוק ובסילוק מזהמים רבים .התהליכים כוללים אוזון בלבד וכן שילוב של אוזון ומי חמצן ( )H2O2/O3או קרינה אולטרה- סגולה ( .)UV/O3מערכת (AOP - Advanced Oxidation Processes) AOPהמשלבת מספר מחמצנים וקרינת אור נחשבת כבעלת פוטנציאל רב לפירוק מזהמים ואף לקטילת מיקרואורגניזמים במים ).(Mamane et al., 2007נמצא כי תהליך הכולל אוזון בלבד הינו יעיל בפירוק מיקרו-מזהמים רבים ,ואילו מזהמים העמידים לאוזון ניתנים לפירוק על ידי הוספת מי חמצן לתהליך (.)H2O2/O3 מורכבות התהליך בטיפול בקולחים גוברת (לעומת טיפול במי שתייה) ,כאשר בנוסף לפרוק הישיר והעקיף של המזהם ,חומרים אורגנים אחרים בקולחים מותקפים גם הם על ידי רדיקלי ההידרוקסיל ויכולים ליצור שרשרת תגובות מעגלית אשר תאיץ או תעכב את יצירת הרדיקלים .רדיקל ההידרוקסיל מיוצר בתהליך החמצון המתקדם ,מאופיין ביכולת חמצון מהגבוהות ביותר לפירוק של מיקרו-מזהמים אורגנים .התגובה הלא סלקטיבית של הרדיקלים מאפשרת את התאמת התהליך לטיפול במיקרו- מזהמים רבים ומגוונים .יצירת הרדיקלים מוצגת בעבודתינו בעזרת טיפולים המשלבים מי חמצן, קרינת UVואוזון ,וכן תהליך פוטוקטליזה העושה שימוש באור השמש ובקטליסט המבוסס על תחמוצת ביסמוט. דוח זה מסכם את עיקר תוצאותיו של מחקר בן שלוש שנים שנערך באוניברסיטת תל אביב ומדגים את יעילות שיטות הטיפול שנבדקו עבור תרופות מסוגים שונים ועבור סוגי מים שונים. בפרק הראשון ניתן לראות את הירידה בריכוז האנטיביוטיקות CiprofloxacinוTrimethoprim- והתרופה הכימותרפית Cyclophosphamideבמים ,בעקבות טיפול באוזון בלבד (מנת אוזון 1מג"ל) ובאוזון משולב עם מי חמצן (ריכוז מי חמצן 1מג"ל) .יישום נוסף אשר נבדק לתהליך AOPהינו הפחתת רמת הפחמן האורגני ( )TOCבקולחים שניוניים ,כטיפול קדם לפני ממברנות אוסמוזה הפוכה .הניסויים שנערכו הראו כי תהליכי החמצון המתקדם יכולים להביא למינרליזציה של החומר האורגני בקולחים ולירידה משמעותית בריכוז ( TOCעד 54%ירידה) ,כאשר התהליך המשלב אוזון ומי חמצן נמצא כיעיל ביותר מבחינה אנרגטית מבין חלופות רבות שנבחנו. 3 קרינת UVמשמשת כיום בעיקר לחיטוי מי שתייה וקולחים ,אולם היא נמצאה כבעלת פוטנציאל לפרוק מיקרומזהמים כגון שאריות תרופות (בתהליך פוטוליזה) .פרמטר חשוב היכול להשפיע על קצב הפוטוליזה של חומרים רבים הוא .pHהמחקר המוצג בפרק השני מצא כי למזהמים רבים קיים ערך pHבו קצב הפוטוליזה שלהם מקסימאלי .לכן ,על ידי אופטימיזציה של pHהמים ניתן להגיע לקצב פוטוליזה מקסימאלי של המזהם .הפוטנציאל של פוטוליזה ב pH-אופטימאלי ,כשיטת טיפול במים, הודגמה על ידי הסרת תערובת של התרופות סולפאמטוקסזול וטריכלוזן ממי תהום ,תוך שינוי pH המים במהלך הניסוי .התהליך המוצע שיפר בצורה דרמטית את יעילות הסרת התערובת (ואת צריכת האנרגיה) לעומת טיפול UVב pHקבוע. פוטוקטליזה הינה שיטת טיפול נוספת מבוססת קרינה שנחקרה ומוצגת בפרק השלישי ,כאשר נעשה שימוש בקטליסט חדש מבוסס ביסמוט ( )BiOCl0.875Br0.125ובקרינה סולרית ,לפרוק התרופות: .carbamazepine-CBZ, ibuprofen-IBF, bezafibrate-BZF and propranolol PPLהקטליסט נמצא יעיל בפירוק כל התרופות שנבדקו ,כאשר ,ב ,pH = 7קצב הפרוק היה על פי סדר יורדPPL : .> BZF > IBF > CBZמספר פרמטרים השפיעו על יעילות הטיפול ,כולל pHהמים ,ריכוז הקטליסט והמזהם ותוספת מי חמצן לתהליך .לדוגמא ,הורדת ה pH -מ 9ל 0-גרמה לעלייה ברמת הספיחה של CBZלקטליסט ,וכתוצאה מכך לעלייה חדה בקצב פרוק התרופה .מספר מנגנונים הוצעו כאחראים לספיחת המזהמים השונים על הקטליסט ,כולל משיכה הידרופובית ל CBZהניטרלי וחילוף יונים ל BZFו IBF-הטעונים שלילית. השפעת חלקיקים על יעילות תהליכי אוזונציה והשפעת אוזון על חלקיקים בקולחים נבדקה אף היא במחקר חדשני באוניברסיטת תל אביב ומוצגת בפרק הרביעי .חלקיקים בשפכים מיוצגים בדרך כלל על-ידי פרמטר כללי במים כחלקיקים מרחפים ( .)TSSבעוד מחקרים אחרים מצאו כי TSSמשפיע מעט על האוזונציה בשפכים ,אך פרקציות שונות של TSSעם מאפיינים ספציפיים (למשל גודל החלקיקים ,ומטען) עשויים להשפיע עדיין על האוזונציה .יתר על כן ,התגובה של האוזון עם החלקיקים עלול להשפיע על מאפייני החלקיקים .השפעתם של חלקיקים על האוזונציה באמצעות פרמטרים כלליים כמו TSSנבחנה ,ואילו אף מחקר לא בדק ההשפעה של פרקציות שונות של TSS (כלומר גודל חלקיקים ופיזורם) .מטרת מחקר זה היה לבדוק את ההשפעה של גודל החלקיקים שונים (לאחר פרקצונציה ומדידה ע"י אנליזת חלקיקים) על אוזונציה בשפכים ,דרך הפירוק של מיקרומזהמים נבחרים ושינויים בפרמטרים הכללים של שפכים (למשל בליעת .)UVבנוסף ,נבדקה ההשפעה של האוזון על חלקיקים בשפכים ועל פוטנציאל הזטה של החלקיקים. 0 CHAPTER I Ozone Degradation of Cyclophosphamide – Effect of Alkalinity and Key Effluent Organic Matter Constituents Introduction Ozone has potential in degrading and removing trace organic micropollutants in water (e.g. pharmaceuticals) (e.g. Dodd et al., 2006; Wert et al., 2009). These organic contaminants are degraded during ozone application through two main pathways: direct oxidation by molecular ozone and the indirect radical oxidation, where •OH radical is the main contributor. The degradation kinetics of the contaminant and the contribution of each of the oxidation processes are influenced by the target contaminant characteristics (i.e. its reaction rate with O3 and •OH) and by the composition of the treated water. Staehelin and Hoigne (1985) showed that the aqueous ozone is decomposed (and •OH formed) generally through a radical chain reaction initiated by reaction of O3 with OH-. This chain reaction can be promoted by solutes which transfer •OH radical into superoxide radical ion (O2•-) (e.g. humic acid), or inhibited by solutes which do not promote O2•- (e.g. carbonate). Alternatively, Pi et al. (2005) proposed that the radical chain reaction can be accelerated by different aromatic compounds, where insitu hydrogen peroxide is a main intermediate and chain carrier; while Pocostales et al. (2010) suggested that •OH generation may additionally occur through ozone adducts to aromatic compounds, the elimination of singlet oxygen and the formation of phenol. In natural waters, various researchers have identified natural organic matter (NOM) and alkalinity as the main promoters and inhibitors of the radical chain reaction in the ozonation process (e.g. Acero and Von Gunten, 2001). Huber et al. (2003) found that the ozonation efficiency of various pharmaceuticals, which react slowly with O3 and quickly with •OH, increased with increasing NOM concentration and decreased with increasing alkalinity in different natural waters. 4 In wastewater effluent, ozonation kinetics differs from drinking waters due to the relatively high concentration and complexity of the effluent organic matter (EfOM). Effluent organic matter (EfOM) is mainly composed of NOM, soluble microbial products (SMPs) and nonbiodegradable organic materials. SMPs (e.g. polysaccharides and proteins) mostly originate from the biological processes within the wastewater treatment plant (WWTP); while NOM, which includes a vast variety of organic molecules and functional groups, typically originates from the source water (i.e. drinking water) (Shon et al., 2006). The relative contribution of the different EfOM constitutes may vary with place, season and treatment. Numerous studies have demonstrated the differences in ozonation kinetics of pharmaceuticals between natural water and wastewater effluents (e.g. Benitez et al., 2009). Zimmerman et al. (2011) investigated the degradation of different micropollutants in a full-scale reactor, treating secondary wastewater effluent, for ozone doses in the range of 0.21 to 1.24 g O 3: g DOC. They found that compounds reacting fast with ozone (kP;O3 > 104 1/M s) were eliminated at almost all O3 doses, while substances with lower ozone reactivity (kP;O3 < 104 1/M s) were only fully eliminated at the high ozone doses. Buffle et al. (2006) concluded that the decomposition of ozone in wastewater and the formation of •OH radicals are initiated and controlled by direct reaction of O3 with reactive moieties in the EfOM, rather than through the radical chain reaction. Their study showed the presence of remarkably high •OH concentration during the first seconds of the O3 process (higher than in most advanced oxidation processes (AOPs) in natural waters). In wastewater effluents, it is difficult to model ozone degradation of micro-pollutants as a result of the indistinct effects of different EfOM compounds. The objective of the present study is to determine the degradation of the anticancer drug cyclophosphamide (CPD) by ozonation at, either low or high alkalinity values, in the presence of different EfOM’s constituents. Cyclophosphamide (CPD) degradation is of particular interest due to its known mutagenic properties and its resistance to most conventional and advanced wastewater treatments (Kim et al., 2008). Three model compounds were used to simulate main 5 constituents in the EfOM, separately and in a mixture: (I) Alginate, an acidic polysaccharide, extensively studied in the field of biofouling in water and wastewater systems, and a potential contributor to SMPs in wastewater effluent (Lee et al., 2006), (II) Peptone from casein, used as a source of amino acids and peptides and, (III) Suwannee River NOM, used as NOM contributor. Material and methods Standards and Reagents Cyclophosphamide (CPD) standard (>99% purity) was obtained from Sigma-Aldrich, LC-MS grade acetonitrile, methanol and water from Bio-Lab Ltd. (Jerusalem, Israel). Cyclophosphamide (CPD) stock solution was prepared by dissolving the compound in deionized (DI) water (Direct-Q3 UV system, Millipore-France; resistivity > 18 mΩ cm) at a concentration of 100 mg/L. The probe compound p-chlorobenzoic acid (pCBA) was used to determine •OH radical reaction rate constants with the EfOM model compounds (Cat. No. 13,558-5, Sigma Aldrich, Germany). Alginic acid (alginate), extracted from brown algae, was purchased from Sigma-Aldrich (CAS: 9005-32-7). Peptone from casein was obtained from Fluka (CAS: 91079-40-2). Suwannee River NOM was obtained from the International Humic Substances Society-IHSS (St. Paul, MN). Alkalinity of the water was modified using sodium bicarbonate (Sigma-Aldrich). Stock solutions of alginic acid and NOM (500 mg/L) were prepared in DI water, adjusted to pH 10 by the addition of NaOH. Peptone (500 mg/L) and sodium bicarbonate (30 gr/L) stock solutions were prepared in DI water. All stock solutions were filtered through a 0.45 μm cellulose acetate filter. Cyclophosphamide ozonation at different EfOM model compounds solutions Ozone experiments were performed in a thermostated (21 ºC) 1 L glass cylindrical batch reactor (Ace-glass, Vineland, NJ). Ozone was generated from pure oxygen (>99.9%) using an ozone generator (2-5 gr/h, OZO-1VTT, Ozomax, Canada). The ozone stream was introduced into the aqueous solution using a diffuser located at the bottom of the reactor, while another tube carried off gases from the headspace of the reactor to the ozone destructor. The tested solution was constantly mixed via magnetic stirring. Ozone in the aqueous solution was 7 continually measured using the WADIS310 dissolved O3 sensor (Walchem Corporation, Holliston, MA). Cyclophosphamide (CPD) ozonation experiments were designed to examine the influence of the following parameters: (a) different EfOM model compounds at dissolved organic carbon-DOC concentration range of 0-8 mg/L as C and, (b) the influence of alkalinity at concentrations of 25 and 200 mg/L as CaCO3 (referred in the text as low and high alkalinity). In all cases, experiments were initiated by diffusing the ozone stream into a solution composed of phosphate buffer (2.5 mM) at pH 7.6 (within the pH range of wastewater effluent) and different concentrations of sodium bicarbonate, until approximately 5.3 mg/L dissolved O3 concentration was achieved. Ozone diffusion was then stopped, followed by the injection of CPD (initial concentration of 1 mg/L) and different concentrations of the tested EfOM model compounds. Sampling began after a mixing time of ~10 s, when dissolved O3 concentration reached exactly 5 mg/L. Samples (2 mL) were taken periodically, quenched immediately with excess of sodium thiosulfate to decompose residual O3 and analyzed chromatographically for CPD concentration. Although typical mg O3:mg DOC ratio do not exceed 1, in the present experiments ozone dose was held constant in order to measure the isolated effect of different DOC and alkalinity concentrations. The pH of the solution remained steady throughout the experiment duration. Determination of the EfOM model compounds rate constants with •OH The second-order rate constants of •OH with the EfOM model compounds (kEfOM,•OH, L/(molC) s) were determined in a medium pressure (MP) UV/H2O2 collimated beam apparatus (described in details elsewhere; Lester et al., 2010), using a modified method adapted from Rosenfeldt and Linden (2004). UV based AOP was used in the present study as a simple method for obtaining the rate constants with •OH. Irradiation experiments were performed on buffered water samples (100 mL PBS 2.5 mM at pH 7.6) with added pCBA (1 µM), H2O2 (1.47 mM) and different concentrations of the tested EfOM model compound (0-8 mg/L as C). The UV incident irradiance was obtained using a calibrated spectroradiometer 5 (USB4000, Ocean Optics, Florida, USA) and the UV absorbance of the treated solution was measured via UV-Vis spectrophotometer (Varian, Cary 100BIO, Victoria, Australia). Analytical methods Cyclophosphamide (CPD) was detected and quantified by HPLC-UV Agilent, model 1100 (ACE-RP C18 column 2.5mm×250mm) and an MS detector (Finnigan LCQ). The mass spectrometer was used in positive electro-spray ionization (ESI) mode and the probe temperature was set to 220°C. The flow from the HPLC was passed through a split connector with 60 μL/min of effluent introduced into the MS interface. Ions in the range 200-300 m/z were registered in the conventional scanning mode. The mobile phase was ammonium formate 0.05M (A) and methanol (B), at pH 5. The mobile phase eluent gradient started with 50% of eluent A, followed by a 2.5-min linear gradient to 30% of eluent A, 3-min isocratic elution and a 2 min linear gradient back to 50% of eluent A, maintained for 4 min to equilibration time. Conditions for p-chlorobenzoic acid (pCBA) quantification are detailed in Lester et al. (2010). Dissolved organic carbon (DOC) of the EfOM solutions was measured using a TOC analyzer (Apollo 9000, TekmarDohrmann). Results and discussion O3 decomposition The difference in dissolved ozone decomposition in the presence of the EfOM model compounds, at low and high alkalinity values, is demonstrated in Figure 1, where Ln [O3]t/[O3]0 is plotted with time for alginic acid, NOM and peptone at DOC concentration of 1 mg/L (as C), and alkalinity concentrations of 25 and 200 mg/L. In the presence of NOM and peptone, an initial rapid O3 decomposition phase was observed (0-60 s), followed by a second slower decomposition phase. Nothe et al. (2009) observed three-phase kinetics for ozone decay in wastewater effluent, where the first phase occurred within seconds. Therefore, we suspect that the two phases measured in our study represent the latter two phases of a threephase O3 decay kinetics. In contrast, only one O3 decomposition phase was observed in the presence of alginic acid. In all cases, ozone decay exhibited an apparent first-order kinetics, 9 and generally, in the order of: peptone > NOM > alginic acid (Figure 1). O 3 first-order decay rate constants in the presence of the EfOM model compounds, for the rapid and slower phases, are presented in Table 1 (k, 1/s). Figure 1: Dissolved O3 decomposition (O3 dose 5 mg/L) in the presence of alginic acid, peptone and NOM (1 mg/L as C), at alkalinity concentrations of 25 and 200 mg/L and pH 7.6 (in the presence of CPD). The inset is a magnification of 0-60 s. The difference in ozone decomposition rate between the EfOM model compounds depends on their structure and, more specifically, on the existence of different moieties reacting directly with ozone and/or promoting the radical chain reaction, as follows: NOM structure is highly complex, consisting of a wide variety of organic molecules (e.g. humic acids) and functional groups (e.g. phenolic moieties). Numerous studies have already confirmed the reactivity of NOM and its influence on ozone decomposition (e.g. Westerhoff et al., 1999). For example, Mvula and Von sonntag (2003) showed that phenol may enhance O3 decomposition both by direct reaction and by promoting the radical chain reaction. 15 Peptone contains a mixture of small proteins, peptides and amino acids. Buffle and Von Gunten (2006) have demonstrated the high reactivity of deprotonated amino compounds in general, and of different amino acids in particular towered ozone. Moreover, Hoigne and Bader (1983b) found extremely high rate constants for the reaction of ozone with different amino acids which contain thio groups (e.g. cysteine). The peptide linkage connecting the α-amino group of the amino acids in polypeptides and proteins has a very low reactivity toward ozone (Pryor et al., 1984). Therefore, the high decomposition rate of ozone in the present study is probably due to its direct reaction with free amino acids and reactive side-chain groups in the peptides and proteins. Alginic acid is a linear polysaccharide composed of mannuronic and guluronic acid subunits. In general, saturated compounds such as polysaccharides react slowly with molecular ozone (e.g. reaction rate constant of ozone with glucose = 0.45 1/M s; Hoigne and Bader, 1983a), thus the radical chain reaction will most likely predominate O3 decay (hence the one-phase kinetic). Akhlaq et al. (1990) concluded that over 70% of the •OH radicals reacting with alginic acid during ozonation lead to the formation of O2•-, further reacting with O3 to enhance its decay. The influence of the EfOM compound’s concentration on ozone decomposition is demonstrated in figure 2, for NOM at DOC values of 1, 3 and 8 mg/L, and alkalinity concentrations of 25 and 200 mg/L, and in Table 1. Generally, as expected, ozone decay rate increases with increase in DOC concentrations for all EfOM model compounds at both low and high alkalinity values. However, in the presence of alginic acid, the increase in ozone decay rate initiates only at DOC > 1 mg/L, emphasizing the alginic acid’s low reactivity toward molecular ozone. 11 Figure 2: Dissolved O3 decomposition (O3 dose 5 mg/L) in the presence of different NOM concentrations (as DOC), at different alkalinity concentrations of 25 and 200 mg/L and pH 7.6 (in the presence of CPD). Numbers in the legends (1, 3 and 8) refer to DOC concentration in mg/L. Carbonate alkalinity is known to inhibit ozone decomposition in water by reacting with the generated •OH radicals, and forming oxidation products that do not promote the radical chain reaction. The stabilizing effect of alkalinity on ozone was demonstrated by comparing ozone decay rate at low and high alkalinity for a specific EfOM model compound concentration (Table 1). For alginic acid, the stabilizing effect of alkalinity on O3 decay rate was clearly demonstrated throughout the entire DOC range, where increasing alkalinity at a specific DOC value always decreased ozone decay rate by more than 10%. In the presence of NOM, alkalinity inhibition was less pronounced during the first phase of O3 decay, for DOC ≥ 3 mg/L. For peptone, the effect of alkalinity was minor for DOC ≥ 3 mg/L, during both first and second O3 decay phases. The different effects of alkalinity on O3 decay emphasize the diverse mechanisms responsible for O3 decay in the presence of the different EfOM model compounds. Using alginic acid, ozone decay occurred most likely through the radical chain reaction path. This 12 reaction path is relatively slow and highly sensitive to the presence of •OH scavengers (i.e. alkalinity). In solutions containing the highly reactive NOM and peptone, decomposition of ozone followed both the direct and radical chain reaction paths. Direct reaction of ozone with different moieties in the organic matter may be more rapid and independent of alkalinity concentration (Buffle et al., 2006). The relative contribution of each path differs in the first and second phase of O3 decay and, depends on the EfOM model compound’s concentration and its reactivity. For example, direct reaction of O3 with the organic matter is probably the dominant mechanism for O3 decay in the presence of NOM, during the first O3 decay phase, at DOC ≥ 3 mg/L and, in the presence of peptone, at both the first and second O 3 decay phases, at DOC ≥ 3 mg/L. In an attempt to better simulate “real” wastewater effluent, O3 decomposition experiments were conducted using a mixture of the model compounds. Since the fraction distribution of EfOM may vary substantially depending on the wastewater origin and the type of treatment (Imai et al., 2002; Jarusutthirak et al., 2002), an exemplary ratio of 2:1:2 (for alginic acid:NOM:peptone) was chosen. The experimental O3 decomposition rate (termed mixture) was compared to the calculated value (termed sum) (i.e. weighted sum of the individual rate constants) (Table 1). The experimental rate constants were higher than the calculated values in almost all cases. Rosario-Ortiz et al. (2008) found that the reaction rate constants of •OH with different non-isolated EfOM were 3-5 times higher than the rate constants of •OH with fractionated EfOM. Thus, a system containing various organic solutes is not a simple mixture with respect to its reactivity toward ozone and •OH. Possibly, interactions between the examined EfOM model compounds contributed to this phenomenon. 13 Table 1: First-order ozone decay rate constant (in the presence of CPD) k*(ksec**), s -1 TOC mg L-1 0 0.5 Alkalinity 25 mg L-1 Alkalinity 200 mg L-1 1 3 5 8 Alginic acid 0.0026 0.0027 0.0026 0.0039 0.0065 0.0081 NOM 0.0028 Peptone 0.0025 Mixture 0.0025 Sum*** 0.0026 Alginic acid 0.0016 0.0036 (0.0029) 0.0076 (0.0047) 0.0059 (0.0043) 0.0046 (0.0033) 0.0015 0.0050 (0.0034) 0.0125 (0.0068) 0.0089 (0.0053) 0.0070 (0.0044) 0.0014 0.0135 (0.0115) 0.0288 (0.0173) 0.0198 (0.0172) 0.0158 (0.0108) 0.0017 0.0210 (0.0161) 0.0357 (0.0190) 0.0280 0.0218) 0.0211 (0.0139) 0.0024 0.0380 (0.0205) 0.0407 (0.0204) 0.0373 (0.236) 0.0271 (0.0155) 0.0024 NOM 0.0014 Peptone 0.0014 Mixture 0.0014 Sum*** 0.0015 0.0025 (0.0016) 0.0049 (0.0018) 0.0037 (0.0018) 0.0031 (0.0016) 0.0037 (0.0017) 0.0103 (0.0049) 0.0058 (0.0021) 0.0054 (0.0029) 0.0125 (0.0062) 0.0282 (0.0161) 0.0155 (0.0111) 0.0145 (0.0084) 0.0193 (0.0109) 0.0351 (0.0173) 0.0254 (0.0180) 0.0187 (0.0101) 0.0362 (0.0172) 0.0391 (0.0189) 0.0344 (0.0194) 0.0234 (0.0120) * ksec– First-order rate constant for the first (rapid) phase of O3 decay ksec– First-order rate constant for the second (slower) phase O3decay *** Sum = kalginic acidx 0.4 + kNOM x 0.2 + kPeptone x 0.4 ** Cyclophosphamide (CPD) removal The removal of CPD was recorded until all applied ozone was consumed (i.e. 5 mg/L); however, under different experimental conditions, CPD concentration decreased below the HPLC-MS limit of detection (100 ng/L). Therefore, CPD removal is presented herein for an applied ozone dose of only 3 mg/L (i.e. until dissolved ozone reached 2 mg/L). Figure 3 presents the CPD removal as a function of DOC concentration, for the different EfOM model compounds and alkalinity concentrations, at an applied ozone dose of 3 mg/L. Cyclophosphamide (CPD) removal in a buffered water (without organic matter), for low and high alkalinity levels was approximately 92% and 58% respectively. Generally, CPD removal decreased with increase in DOC concentration (with an exception of alginic acid at 10 high alkalinity and DOC 0.5 mg/L), and in some cases it reached a plateau at higher DOC values. The influence of peptone on CPD removal was most pronounced, where almost no CPD removal can be seen at DOC ≥ 5 mg/L; while in the presence of alginic acid CPD removal decreased only moderately with increase in DOC concentration. Figure 3: Cyclophosphamide (CPD) removal as a function of DOC concentration for alginic acid, NOM and peptone, at alkalinity concentrations of 25 and 200 mg/L. Ozone dose 3 mg/L. Due to the low reaction rate of CPD with molecular ozone (kO3,CPD = 2.8 1/M s; Lester et al., 2011), its removal during the ozonation process occurs mainly through oxidation by •OH radicals (k•OH,CPD = 1.3 x 109 1/M s; Lester et al., 2011). Thus, the differences in the model compounds’ impact on the CPD removal can be explained by differences in the •OH production yield and different scavenging effects on •OH radicals by the EfOM model compounds (evaluated in the following section). In ozonation of pure water, the yield for •OH formation through the radical chain reaction was found to be ~55% (Staehelin and Hoigne, 1982). This relatively high yield may partially 14 explain the high CPD removal rate at low EfOM concentrations (where the radical chain reaction is the dominant •OH formation mechanism). The high removal rate of CPD at high concentrations of alginic acid may be due to the alginic acid’s promotion effect on •OH, as described earlier. Different •OH production yields are expected when direct reaction of ozone with the EfOM is the dominant mechanism (e.g. at high NOM and peptone concentrations). For example, Notch et al. (2009) found that the •OH yield during the ozonation process of wastewater effluent was ∼13%, while Mvula and von Sonntag (2003) calculated •OH yield for direct reaction of ozone with phenol to be ~22%. Higher CPD removal is obtained at low alkalinity values compared to high alkalinity values for all EfOM model compounds at DOC concentration < 5 mg/L, due to the scavenging effect of alkalinity on •OH radicals. At higher DOC levels (≥5 mg/L), a comparable CPD removal rate can be seen at both alkalinity values, indicating that the scavenging effect of the EfOM model compounds is dominant. In natural waters, alkalinity is considered a main •OH scavenger; while in water with high concentration of organic matter (i.e. wastewater effluent), its contribution to the water scavenging is less significant. Elovitz et al. (2000) estimated the •OH scavenging of alkalinity in lake water to be approximately 50% of DOC scavenging; while Nothe et al. (2009) have calculated this value to be ∼10% in wastewater effluent. Rate constant for the reaction of •OH with the EfOM model compounds UV/H2O2 degradation of pCBA, in the presence of different concentrations of the EfOM model compounds, was used to calculate k•OH,EfOM. Degradation of pCBA involves direct UV photolysis and indirect photo-oxidation by •OH radicals, as described in the following equations (Rosenfeldt and Linden, 2004): ln [ pCBA]t / t k obs k 'k pCBA,OH OH ss [ pCBA]0 (1) k ' pCBA k s , pCBA (2) 15 k s , pCBA 10 3 0p pCBA 1 10 a z 200300 a z (3) where, [pCBA]0 and [pCBA]t are initial pCBA concentration (M) and its concentration after exposure time t (s), kobs and k’ are the observed (total) and direct-photolysis time-based pseudo-first-order degradation rate constants of pCBA respectively (1/s). [•OH]ss is the steady-state •OH radical concentration (M) and kpCBA,•OH is the second-order rate constant of pCBA reaction with •OH, reported to be 5x109 1/M s (Buxton et al., 1988). ФpCBA is the quantum yield for pCBA removal (0.0182 mole/Einstein; Lester et al., 2010), ks,pCBA is the specific rate of light absorption by pCBA (Einstein /mole s). E 0p(λ) is the incident photon irradiance (Einstein/cm2 s), εpCBA(λ) is the molar (decadic) absorption coefficient of pCBA (1/M cm), a(λ) is the solution absorption coefficient (1/cm) and z is the depth of solution (cm). The steady-state concentration of •OH radical was calculated as the ratio of the formation of •OH radicals to the destruction of the radicals (Rosenfeldt and Linden, 2004): OH ss RForm OH k pCBA,OH [ pCBA] k H 2O2 ,OH [ H 2 O2 ] k EfOM ,OH [ EfOM ] (4) where, kH2O2,•OH (1/M s) and kEfOM,•OH (L/(molC) s) are the second-order rate constants of •OH with H2O2 and the EfOM model compound respectively. RForm OH is the rate of •OH formation (M/s) and is calculated using equation 5, taking into account the UV light absorbance of the irradiated solution. RForm OH k s , H 2O2 H 2O2 [ H 2 O2 ] (5) 17 where, фH2O2 is the quantum yield for •OH formation (QY = 1; Baxendale and Wilson, 1957), and ks,H2O2 is the specific rate of light absorption by H2O2, calculated using equation 3 (modified for H2O2). Substituting equation 4 into 1, inverting both sides and rearranging results in eq. 6. k H O ,OH [ H 2 O2 ] k EfOM ,OH [ EfOM ] RForm OH [ pCBA] 2 2 k obs k ' k pCBA,OH k pCBA,OH (6) Figure 4 is a plot of RForm OH /kobs-k’ vs. the EfOM model compounds’ concentration. Multiplying the slope of the linear lines by the known value of kpCBA,•OH resulted in reaction rate constants of •OH with alginic acid, NOM and peptone of 0.92, 0.95 and 1.30 x 10 8 L/(molC) s respectively. The calculated values are lower than most values presented in the literature for EfOM constitutes. Moreover, the similar values obtained for alginic acid and NOM are unexpected, in view of the differences in their characteristics (aliphatic vs. aromatic). The Suwannee River NOM used in our study contains however relatively high proportion of polar aliphatic substances in addition to humic and fulvic substances (Serkiz and Perdue, 1990), which may partially explain its relatively low reaction rate constants with •OH. Westerhoff et al. (1999), using ozone as an •OH source, found •OH reaction rate constants with Suwannee River humic and fulvic acids to be 8.1 and 3.7 x 108 L/(molC) s respectively. A latter study by Westerhoff et al. (2007), using electron pulse radiolysis, presented values of 1.6 x 108 L/(molC) s for the reaction of •OH with Suwannee River fulvic acids, and an average value of 2.23 x 108 L/(molC) s for seven DOM isolates from different sources. Myint et al. (1987) measured the rate constant for the reaction of •OH with hyaluronic acid (a carbohydrate polymer with some resemblance to alginic acid) to be 7x10 8 1/M s, expressed in terms of the disaccharide repeating sub-unit (equivalent to approximately 3x108 L/(molC) s), using pulse radiolysis. 15 Figure 4: RForm OH /kobs–k’ as a function of DOC concentration for NOM, alginic acid and peptone. Initial hydrogen peroxide concentration of 50 mg/L. The reaction rate constants for peptone and alginic acid with •OH obtained in the present study correlate well with the different CPD removal behaviors (Figure 3). Increasing the concentration of the highly •OH reactive peptone results in a sharp decrease in CPD removal rate, while addition of the less-reactive alginic acid only moderately decreases the CPD removal rate. For NOM, the low reaction rate constant with •OH may provide an explanation only if considering that a high portion of the •OH radicals reacting with NOM are scavenged (i.e. do not promote the chain reaction by producing O2•-). Conclusions Influence of the EfOM model compounds on ozone decay Peptone and NOM were highly reactive toward molecular ozone; therefore at high peptone and NOM concentration ozone decay was fast and controlled by direct reaction with the model compounds. Alginic acid was least reactive toward molecular ozone, thus, ozone decay was relatively slow and controlled by a radical chain reaction. 19 Influence of the EfOM model compounds on CPD removal Cyclophosphamide (CPD) removal occurred mainly through its reaction with •OH radicals and decreased with increasing model compounds concentration (most pronounced for peptone and least pronounced for alginic acid). The sharp decrease in CPD removal at high peptone concentration likely results from the low yield of •OH formation in the direct reaction of ozone with peptone and its scavenging of •OH. The moderate decrease in CPD removal with increasing alginic acid concentration is due to the relatively high yield of •OH formation from ozone decay through the radical chain reaction and, the role of alginic acid as an •OH promoter. Influence of alkalinity Increasing alkalinity decreased CPD removal rate at DOC concentration < 5 mg/L, due to its scavenging of •OH. At higher DOC levels (≥5 mg/L), alkalinity had only a minor effect on CPD removal, indicating that the scavenging effect of the Efom model compounds was the dominant mechanism. References Acero, J.L., and U. Von Gunten, "Characterization of oxidation processes: ozonation and the AOP O-3/H2O2", Journal American Water Works Association. 93(10): 90-100 (2001). Akhlaq, M.S., H.P. Schuchmann, and C. Vonsonntag, "Degradation of the Polysaccharide Alginic Acid - a Comparison of the Effects of Uv-Light and Ozone", Environmental Science & Technology. 24(3): 379-383 (1990). Baxendale, J.H., and J.A. Wilson, "The photolysis of hydrogen peroxide at high light intensities ", Transactions of the Faraday Society. 33(344-346 (1957). Benitez, F.J., J.L. Acero, F.J. Real, and G. Roldan, "Ozonation of pharmaceutical compounds: Rate constants and elimination in various water matrices", Chemosphere. 77(1): 53-59 (2009). Buffle, M.O., J. Schumacher, S. Meylan, M. Jekel, and U. von Gunten, "Ozonation and advanced oxidation of wastewater: Effect of O-3 dose, pH, DOM and HO center dot-scavengers on 25 ozone decomposition and HO center dot generation", Ozone-Science & Engineering. 28(4): 247-259 (2006). Buffle, M.O., and U. Von Gunten, "Phenols and amine induced HO center dot generation during the initial phase of natural water ozonation", Environmental Science & Technology. 40(9): 3057-3063 (2006). Buxton, G.V., C.L. Greenstock, W.P. Helman, and A.B. Ross, "Critical -Review of Rate Constants for Reactions of Hydrated Electrons, Hydrogen-Atoms and Hydroxyl Radicals (.Oh/.O-) in Aqueous-Solution", Journal of Physical and Chemical Reference Data. 17(2): 513-886 (1988). Dodd, M.C., M.O. Buffle, and U. Von Gunten, "Oxidation of antibacterial molecules by aqueous ozone: Moiety-specific reaction kinetics and application to ozone-based wastewater treatment", Environmental Science & Technology. 40(6): 1969-1977 (2006). Elovitz, M.S., U. von Gunten, and H.P. Kaiser, "Hydroxyl radical/ozo ne ratios during ozonation processes. II. The effect of temperature, pH, alkalinity, and DOM properties", OzoneScience & Engineering. 22(2): 123-150 (2000). Hoigne, J., and H. Bader, "Rate Constants of Reactions of Ozone with Organic and Inorganic Compounds in Water .1. Non-Dissociating Organic-Compounds", Water Research. 17(2): 173-183 (1983a). Hoigne, J., and H. Bader, "Rate Constants of Reactions of Ozone with Organic and InorganicCompounds in Water .2. Dissociating Organic-Compounds", Water Research. 17(2): 185-194 (1983b). Huber, M.M., S. Canonica, G.Y. Park, and U. Von Gunten, "Oxidation of pharmaceuticals during ozonation and advanced oxidation processes", Environmental Science & Technology. 37(5): 1016-1024 (2003). Imai, A., T. Fukushima, K. Matsushige, Y.H. Kim, and K. Choi, "Characterization of dissolved organic matter in effluents from wastewater treatment plants", Water Research. 36(4): 859-870 (2002). Jarusutthirak, C., G. Amy, and J.P. Croue, "Fouling characteristics of wastewater effluent organic matter (EfOM) isolates on NF and UF membranes", Desalination. 145(1-3): 247-255 (2002). 21 Kim, I.H., H. Tanaka, T. Iwasaki, T. Takubo, T. Morioka, and Y. Kato, "Classification of the degradability of 30 pharmaceuticals in water with ozone, UV and H2O2", Water Science and Technology. 57(2): 195-200 (2008). Lee, S., W.S. Ang, and M. Elimelech, "Fouling of reverse osmosis membranes by hydrophilic organic matter: implications for water reuse", Desalination. 187(1-3): 313-321 (2006). Lester, Y., D. Avisar, and H. Mamane, "Photodegradation of the antibiotic sulphamethoxazole in water with UV/H2O2 advanced oxidation process", Environmental Technology. 31(2): 175-183 (2010). Lester, Y., D. Avisar, and H. Mamane, "Removal of pharmaceuticals using combination of UV/H2O2/O3 advanced oxidation process", Water Science & Technology. in press, doi: 10.2166/wst.2011.079((2011). Mvula, E., and C. von Sonntag, "Ozonolysis of phenols in aqueous solution", Organic & Biomolecular Chemistry. 1(10): 1749-1756 (2003). Myint, P., D.J. Deeble, P.C. Beaumont, S.M. Blake, and G.O. Phillips, "The Reactivity of Various Free-Radicals with Hyaluronic-Acid - Steady-State and Pulse-Radiolysis Studies", Biochimica Et Biophysica Acta. 925(2): 194-202 (1987). Nothe, T., H. Fahlenkamp, and C. von Sonntag, "Ozonation of Wastewater: Rate of Ozone Consumption and Hydroxyl Radical Yield", Environmental Science & Technology. 43(15): 5990-5995 (2009). Pi, Y.Z., J. Schumacher, and M. Jekel, "Decomposition of aqueous ozone in the presence of aromatic organic solutes", Water Research. 39(1): 83-88 (2005). Pocostales, J.P., M.M. Sein, W. Knolle, C. von Sonntag, and T.C. Schmidt, "Degradation of Ozone-Refractory Organic Phosphates in Wastewater by Ozone and Ozone/Hydrogen Peroxide (Peroxone): The Role of Ozone Consumption by Dissolved Organic Matter", Environmental Science & Technology. 44(21): 8248-8253 (2010). Pryor, W.A., D.H. Giamalva, and D.F. Church, "Kinetics of Ozonation .2. Amino -Acids and Model Compounds in Water and Comparisons to Rates in Nonpolar -Solvents", Journal of the American Chemical Society. 106(23): 7094-7100 (1984). 22 Rosario-Ortiz, F.L., S.P. Mezyk, D.F.R. Doud, and S.A. Snyder, "Quantitative correlation of absolute hydroxyl radical rate constants with mon-isolated effluent organic matter bulk properties in water", Environmental Science & Technology. 42(16): 5924-5930 (2008). Rosenfeldt, E.J., and K.G. Linden, "Degradation of endocrine disrupting chemicals bisphenol A, ethinyl estradiol, and estradiol during UV photolysis and advanced oxidation processes", Environmental Science & Technology. 38(20): 5476-5483 (2004). Serkiz, S.M., and E.M. Perdue, "Isolation of Dissolved Organic-Matter from the Suwannee River Using Reverse-Osmosis", Water Research. 24(7): 911-916 (1990). Shon, H.K., S. Vigneswaran, and S.A. Snyder, "Effluent organic matter (EfOM) in wastewater: Constituents, effects, and treatment", Critical Reviews in Environmental Science and Technology. 36(4): 327-374 (2006). Staehelin, J., and J. Hoigne, "Decomposition of Ozone in Water - Rate of Initiation by Hydroxide Ions and Hydrogen-Peroxide", Environmental Science & Technology. 16(10): 676-681 (1982). Staehelin, J., and J. Hoigne, "Decomposition of Ozone in Water in the Presence of Organic Solutes Acting as Promoters and Inhibitors of Radical Chain Reactions", Environmental Science & Technology. 19(12): 1206-1213 (1985). Wert, E.C., F.L. Rosario-Ortiz, and S.A. Snyder, "Effect of ozone exposure on the oxidation of trace organic contaminants in wastewater", Water Research. 43(4): 1005-1014 (2009). Westerhoff, P., G. Aiken, G. Amy, and J. Debroux, "Relationships between the structure of natural organic matter and its reactivity towards molecular ozone and hydroxyl radicals", Water Research. 33(10): 2265-2276 (1999). Westerhoff, P., S.P. Mezyk, W.J. Cooper, and D. Minakata, "Electron pulse radiolysis determination of hydroxyl radical rate constants with Suwannee river fulvic acid and other dissolved organic matter isolates", Environmental Science & Technology. 41(13): 4640-4646 (2007). Zimmermann, S.G., M. Wittenwiler, J. Hollender, M. Krauss, C. Ort, H. Siegrist, and U. von Gunten, "Kinetic assessment and modeling of an ozonation step for full-scale municipal wastewater treatment: Micropollutant oxidation, by-product formation and disinfection", Water Research. 45(2): 605-617 (2011). 23 CHAPTER II Enhanced removal of PPCPs using pH modification coupled with UV photolysis Introduction Ultraviolet (UV) treatment is being increasingly used for disinfection of wastewater and drinking water in North America, Europe, and numerous other countries around the world. UV radiation can also be used to chemically degrade or break down organic micropollutants via photolysis (photodegradation), a process in which a chemical species undergoes a chemical change as the result of the absorption of photons (Legrini et al., 1993). If a molecule absorbs a photon, it is then in an excited state and can more readily transform. For most chemicals, direct UV photolysis alone is not a practical process for degradation, however, numerous chemical contaminants of concern absorb UV at wavelengths below 300 nm; hence can potentially undergo direct photolysis (Stefan and Bolton, 2002). For example, UV was well established as the technology of choice to remove N-nitrosodimethylamine (NDMA) from drinking water (Stefan and Bolton, 2002; Sharpless and Linden, 2003). NDMA is a water contaminant of emerging concern in North American and a potential human carcinogen. Wu et al. (2007) investigated the photodegradation of a widely used herbicide, metolachlor, applying monochromatic (254 nm) UV light. Approximately half of the metolachlor was degraded at UV fluence of 1000 mJ/cm2 (at pH 7.5) which is 25 times higher than typical UV dose at water treatment plants (WTPs) required for disinfection. Various studies found that pH of the treated solution affected direct photolysis of different organic pollutants. Shemer et al. (2005) showed that UV photolysis rate of 3,5,6trichloro-2-pyridinol (a degradation product of the insecticide chlorpyrifos) increased with increasing the solution's pH up to a constant maximum value of 6.40×10−3 cm2/mJ at pH 5, thus was highly pH dependent within the pH range 2.5–5. Other researchers showed the 20 influence of pH on the photodegradation kinetics of the antibiotics tetracycline (e.g. Werner et al., 2006), sulfadimethoxine (Lester et al., 2008) and the pesticides atrazine and bensulfuron methyl (Lam et al., 2003). Obviously, pH impacts the degradation kinetics of many micro-pollutants. However, none of the previous studies have examined the use of direct UV photolysis together with an artificial pH modification of the water to treat water polluted with PPCPs. Therefore, the main goal herein was to examine the combination of direct UV photolysis and artificial pH modification of the treated water (during the treatment itself), as a potential treatment technology for water remediation. The specific goals were to (a) determine the photodegradation kinetics of different PPCPs separately, in buffered water, using direct UV light and pH adjustment and (b) demonstrate the potential of pH induced polychromatic UV treatment to remove a mixture of PPCPs from groundwater. Material and Methods The examined compounds Twelve organic compounds were selected for the study based on their environmental relevance (Table 2.1). All compounds were purchased from Sigma-Aldrich. Stock solutions (100 mg/L) were prepared in water or methanol according to the compound’s solubility. Other high-performance liquid chromatography (HPLC)-grade solvents and chemicals (acetonitrile, methanol, formic acid, sodium hydroxide, and water) were purchased from BioLab Ltd. (Jerusalem, Israel). 24 Table 2.1 – Selected target contaminants Class Compounds Antibiotics Ciprofloxacin (CIP), Enrofloxacin (ENR), norfloxacin (NOR), trimethoprim (TMP), tetracycline (TC), oxytetracycline (OTC), amoxicillin (AMX) and sulfamethoxazole (SMX) Anti-epileptic drug Carbamazepine (CBZ) Anti-inflammatory drugs Ketorolac (KTR) and diclofenac (DCF) Anti-microbial agent Triclosan (TCS) Experimental setup All irradiation experiments were conducted using polychromatic UV light. Specifically, photolysis was carried out in a bench-scale collimated beam apparatus, using a 0.45 kW polychromatic medium-pressure (MP) Hg vapor lamp (Ace-Hanovia lamp, Ace Glass Inc., Vineland, NJ, USA). The treated solution (100 mL) was spiked with a known concentration of the tested compound and then irradiated with constant gentle stirring. Samples were withdrawn at appropriate intervals for chromatographic analysis (using HPLC-DAD). UV dose calculation UV dose (fluence) was determined by multiplying the average irradiance of the UV lamp between 200 and 300 nm by exposure time. The average irradiance was calculated using a calibrated spectroradiometer (spectral range 200-390 nm; USB4000, Ocean Optics, Dunedin FL, USA) placed in the same position as the irradiated solution, using a procedure adapted from Bolton and Linden (2003). The UV absorption coefficient of the treated solutions at different pHs was measured in a UV-Vis spectrophotometer (Varian, Cary 100BIO, Victoria, Australia) and the molar absorption coefficients for the target compounds were determined. Experimental procedure 25 To achieve the goals of the research, work was carried out in two stages: (I) Determination of optimal pH values for the photodegradation of each target contaminant, and (II) Demonstration of the potential of pH enhanced photolysis to remove a mixture of compounds from groundwater. Each of these stages is discussed in detail below. I) Optimal pH determination for the photodegradation of each compound The optimal pH values for the photodegradation of each target contaminant were determined by testing the photodegradation rates of each contaminant separately, at different pH values. Experiments were conducted in phosphate buffered saline (PBS, 5 mM), and pH was adjusted by adding formic acid or sodium hydroxide at the appropriate volume and concentration. Once the pH of the buffer containing the contaminant (at an initial concentration of 1 mg/L) was adjusted, the sample was subjected to irradiation to determine the pH value that leads to the maximal photodegradation rate. It should be noted that the initial concentration of the pharmaceuticals was higher than their concentration in aquatic environments; however, it was low enough to minimize the compound's contribution to the total solution absorbance, such that first-order photodegradation kinetics were obtained for all compounds. II) Removal of a mixture of compounds from groundwater The main goal of the second part of this study was to determine the potential of UV photodegradation combined with artificial pH modification for the removal of a mixture of two pharmaceuticals, i.e., sulfamethoxazole (SMX) and triclosan (TCS), from groundwater, using their predetermined optimal pH values. The removal of SMX and TCS from groundwater is of particular interest. Triclosan (TCS) has been identified by the EU water framework directive priority substance list (together with diclofenac, ibuprofen, and clofibric acid) as a future emerging priority candidate of particular environmental concern, due to its wide usage (as cited by Ellis, 2006). Sulfamethoxazole (SMX) is one of the most commonly detected antibiotics worldwide (e.g. Sacher et al., 2001; e.g. Barnes et al., 2008). Groundwater samples were taken from three different aquifer basins in Israel: GW-1 and GW-3 were from a local carbonate aquifer in the upper Galilee (eastern slopes) and the major carbonate aquifer in the eastern slopes of the Judea Mountain Ridge in central Israel, 27 respectively, both representative of the Judea group. GW-2 was taken from a local aquifer in the Golan Heights, which is composed of interbeded carbonate and basalt layers. The groundwater samples were spiked with the two target contaminant (initial concentration of 0.5 mg/L for each compound), adjusted to their optimal pH value, and then subjected to UV irradiation. The general chemical composition (major ions) of the groundwater samples used in this study is presented in Table 2.2 Table 2.2 - General chemical composition of the three groundwater types GW-1 GW-2 GW-3 pH Cl - HCO3- Ca ++ 7.5 7.8 7.9 19 24 80 260 198 236 70 26 62 Mg ++ mg/L 16 18 31 NO3 - Na + SO4 - 17 15 33 9 32 46 7 6 28 Analytical methods The target compounds were detected and quantified by HPLC (HPLC-Agilent 1100 series; ACE-RP C18 column 2.5 mm×250 mm) equipped with a UV diode array detector and a mass spectrometer (MS, Finnigan LCQ). The HPLC consisted of a microvacuum degasser, a binary pump and a thermostatic column compartment. The flow rate was 0.5 mL/min and the injected volume was 100 μL. The flow from the HPLC was passed through a split connector at 60 μL/min of effluent introduced into the MS interface. The MS was used in positive electrospray ionization (ESI) mode, where ions in the range of 70-500 m/z were registered in the conventional scanning mode. Data analysis The degradation kinetics were expressed as a natural logarithm of the ratio of the concentration (CH) remaining following a UV dose H (mJ/cm) to initial concentration (C0) (Equation 2.1). ln CH k H C0 (2.1) 25 The data were fitted using a linear regression approach resulting in pseudo-first-order reaction kinetics which reflected the difference in photodegradation between samples. The UV dosebased degradation rate constant k (cm2/mJ) was calculated as the negative slope obtained when the degradation was plotted logarithmically. The UV dose required for 90% degradation of the target contaminants was used as a comparative parameter between compounds and between pH values, to emphasize the improved energy consumption of the degradation process due to pH modification. The UV dose required for 90% degradation was calculated using the obtained value for k, by inserting the appropriate numbers: ln 10% 2.3 k 100% H (2.2) Results and discussion Optimal pH determination To better understand the effect of pH variations on photodegradation rate of PPCPs and other organic micro-pollutants, several photodegradation tests were conducted on the selected compounds at different pH values. The pH values were in the common environmental range of 5 to 8. Degradation rates at the different pH values are exemplified in Figure 2.1, which shows the photodegradation of TCS (in PBS) plotted against UV dose. 29 Figure 2.1- UV photodegradation of TCS (1 mg/L) in PBS (5 mM) as a function of UV dose, at different pH values (5, 6, 7 and 8), obtained by MP collimated beam apparatus. The dashed line represents 90% degradation. pH strongly affected the rate of direct TCS photodegradation: lowering the pH of the UVirradiated solution from 8 to 5 resulted in a significant decrease in the pollutant’s degradation rate (Figure 2.1); the UV dose required for 90% removal of TCS was above 400 mJ cm -2 at pH 5, and 162 mJ cm-2 at pH 8. The acid dissociation constant (pKa) of TCS is approximately 7.9 (Equation 2.3) (Wong-Wah-Chung et al., 2007), and there may therefore be a relation between the acid-base speciation of the compound and its photodegradation rate at different pH values. Cl Cl (2.3 O O ) pKa 7.9 Cl HO Cl Cl [TCS-] + [H+] [TCS0] 35 - O Cl Key parameters in evaluating the rate of a photochemical reaction are the molar absorption coefficient of the compound and its overlap with the emission spectra of the UV lamp (together with the compound’s quantum yield). These parameters quantify the light absorption of a compound at each wavelength. pH dependence of the selected compounds’ molar absorption coefficients was examined. Figure 2.2 illustrates the molar absorption coefficients of TCS at different pH values in the range of 5 to 8, as well as the emission spectra of the UV lamp. TCS absorbed light in the UV range at all examined pH values. Slight differences in absorbance peaks at around 278 nm could be seen at pH values of 5 to 7, whereas at pH 8, its absorbance peak was significantly shifted to the right, at 288 nm. Moreover, the peak at pH 8 was higher (4410 1/cm M) than those at pH 5, 6 and 7 (4050, 3480 and 3500 1/cm M, respectively). The apparent shift in the molar absorption coefficients of TSC between pH values (with a clear relation to its pKa) and consequently, the change in the overlap of the compound's absorption coefficients with the emission spectra of the UV lamp, could potentially explain the differences in the photodegradation kinetics of this compound at different pH values. Figure 2.2- Molar absorption coefficients of TCS at various pH values, demonstrated with the emission spectra of the polychromatic UV lamp. 31 Similarly, Wong-Wah-Chung et al. (2007) found that TCS is more photodegradable at high pH (i.e. 11.8) than at low pH (i.e. 5.6), using UV light at different wavelengths. Moreover, the molar absorption coefficient of TCS at pH 5.5 peaked at 280 nm with a value of approximately 4200 1/cm M, while at pH 11.8, a maximum value of 8300 1/cm M appeared at 291 nm. Other studies also confirm the pH dependence of the TCS molar absorption coefficient and photostability (e.g. Mezcua et al., 2004). The UV doses required for 90% removal, and the molar absorption coefficient peaks, for the studied compounds (in PBS) at different pH values are presented in Table 2.3. UV dose was calculated for wavelengths between 200 and 300 nm, as these are commonly used in MP UV treatments (e.g. Bolton and Linden, 2003). It should be noted however, that various compounds absorb light at wavelengths above 300 nm (e.g. ENR), and these wavelengths may influence their photolysis rate. For most of the examined compounds, the UV dose required to achieve 90% removal was strongly dependent on the pH of the treated solution (with the exception of NDMA and NDEA), and could often be related to the compound's pKa value. The extent of the pH dependency and its correlation (i.e. positive or negative) with removal rate by irradiation varied between compounds. For example, increasing the pH of the treated solution from 5 to 8 significantly reduced the UV dose required for 90% removal of NOR and ENR, from 3286 and 2091 mJ/cm2 to 319 and 852 mJ/cm2, respectively. For SMX, the opposite trend was observed: changing the pH of the buffer from 5 to 8 increased the UV dose required to achieve 90% removal (from 173 to 676 mJ/cm2). Thus, by optimizing the pH of the treated solution, markedly lower UV doses can be exploited to achieve the same removal of a target compound. In addition, changes in molar absorption coefficient, due to pH modifications, were observed for most compounds (Table 2.3), and may provide a partial explanation for the changes in their degradation rates at different pH values. However, various compounds showed significant changes in their photodegradability due to pH modification with no 32 associated changes in their molar absorption coefficient (e.g. CBZ). Therefore, various mechanisms may be involved in the observed changes in the kinetics of the UV photodegradation of each compound by pH modification (e.g. changes in quantum yield). Table 2.3 - UV dose required for 90% removal, and the molar absorption coefficient peaks for the studied compounds (in PBS), at different pH values compound Relevant pKa UV dose for 90% removal, mJ/cm2 (molar absorption coefficients peak, 1/cm M; wavelength, nm) pH 5 pH 6 pH 7 pH 8 78 (4410; 288) TCS 7.9 210 (4050; 279) 163 (3480; 280) 138 (3500; 282) CIP 5.5, 7.7 2091 (20725; 277) (6292; 315) 742 (19115; 274) (6452; 319) 404 (17676; 271) (6876; 324) NOR 6.3, 8.7 4600 (40445; 277) 2091 (38079; 275 ) 697 (36771; 272) 328 (38265; 272) ENR 5.94 3286 (38194; 277) (11512; 316) 2300 (36532; 276) (11361; 316) 1045 (32844; 271) (12615; 323) 1264 (32702; 271) (13041; 323) SMT 6.08 767 (21698; 268) 958 (23508; 268) 1045 (25990; 268) 885 (25538; 268) SMX 5.7 112 (17234; 264) 226 (19259; 259) 411 (19615; 257) 451 (16324; 256) TMP 6.6 6420 (6240; 272) 6570 (5622; 276) 1533 (5312; 278) 742 (6265; 287) KTR 3.5 1150 (7031; 249) (20284; 322) 1000 (6519; 249) (20073; 323) 657 (6678; 249) (20177; 323) 315 (6324; 246) (19230; 322) CBZ 13.9 10000 (11227; 285) 10000 (11334; 285) 5750 (11755; 285) 3594 (11939; 285) ACR 5.5 23000 (53497; 255) 11500 (435420; 249) 7667 (438824; 249) 5750 (474354; 249) AMX 6.9 1075 (9531; 229) (1414; 272) 1095 (9805; 228) (1125; 273) 710 (9720; 227) (1241; 273) 708 (9819; 227) (1696; 273) TC 4.5, 7.3 3833 (14525; 275) (14450; 358) 1045 (14898; 276) (14918; 357) 575 (14854; 275) (15131; 360) 343 (14848; 271) (15879; 367) OTC 3.3, 7.7 5750 (8187; 218) (12671; 275) 2556 (8477; 218) (12851; 276) 1643 (9320; 217) (12940; 276) 33 DCF 4.0 120 (9620; 275) 113 (10342; 276) 112 (10164; 276) 97 (10328; 276) It is important to note that even though the photolysis rates of several of the examined compounds were strongly pH-dependent, direct photolysis was still ineffective for their degradation: UV doses above 6765 and 10,474 mJ cm-2 were required for 90% removal of CBZ and ACR, respectively (at their optimal pH value). Groundwater experiments The main goal of the second part of the study was to determine the potential of UV photodegradation in combination with artificial pH modification, for the removal of a mixture of two pharmaceuticals (i.e. SMX and TCS) from groundwater, using the optimal pH values determined earlier. Figure 2.3 demonstrates the UV dose-based photo-degradation rate constants of the two pharmaceuticals, in PBS at varied pH values. SMX and TCS photolysis rates show high pH dependency with opposite directions, where, increasing pH of the solution from 5 to 8 leads to higher TCS degradation rate and lower SMX degradation rate. Accordingly, the optimal pH values to achieve the highest photo-degradation rate for TCS and SMX are 8 and 5, respectively. A treatment of groundwater containing TCS and SMX in a mixture was suggested. Due to the noticeable differences in their optimal pH values when measured separately, the treatment was conducted under the following conditions: (a) Irradiation of the mixed solution at the groundwater original pH (i.e. pH 7.5-7.9), for a span of approximately 1.5 min (UV dose of 45 mJ/cm2), in order to maximize the TCS degradation; and (b) Modification of the groundwater pH to a value of 5, followed by an additional irradiation time of approximately 4.5 min (UV dose of 135 mJ/cm2), to accelerate the degradation of SMX. A removal of the mixture by 90% was considered only when the slowest degraded compound in the mixture was removed by 90%. 30 0.04 TCS SMX k ( cm2 / mJ ) 0.03 0.02 0.01 0 4 5 6 7 8 9 pH Figure 2.3- UV dose-based photo-degradation rate constant for TCS and SMX, measured separately in PBS (5 mM), as function of pH The pH modification point was determined using linear programming (LP) which allows optimizing the treatment, to achieve maximum removal of the mixture at minimum UV dose. The linear programming goal was to minimize the objective function as follows: Minimize: H1 H 2 (2.4) Subject to constraints: 10% k1 H 1 k 2 H 2 Ln 100% (2.5) 10% k '1 H 1 k ' 2 H 2 Ln 100% (2.6) where, H1 and H2 are the UV dose applied (mJ/cm2) from time zero to pH modification point and from the pH modification point to at least 90% removal of the mixture, respectively. k1 and k2 are the dose-based photo-degradation rate constants of SMX at pH 7.5-7.9 and pH 5 respectively, measured during preliminary irradiation tests on groundwater mixture of TCS and SMX, under constant pH (e.g. Figure 2.4b). k'1 and k'2 are the degradation rate constants of TCS under the same conditions. 34 UV absorbance of the water is considered when calculating average UV dose (fluence) in a collimated beam apparatus. The UV dose is then the product of the average irradiance and the residence time. Thus, ideally the dose-based photo-degradation rateconstants should be independent of the water quality (dissolved constituents), when considered in dose calculations. Regardless, minor variations were observed in the rate constants for SMX and TCS when measured in buffer and groundwater. Figure 2.4a illustrates the removal of TCS and SMX in a mixture from GW-1, using the irradiation procedure where the water pH was modified from 7.5 to 5 after UV dose of 45 mJ/cm2. While, Figure 2.4b demonstrates the removal of a TCS and SMX mixture from GW1, at constant pH values of 7.5 and 5 (no pH modification). It is clear from Figure 2.4a that during the first irradiation period (at pH 7.5) the degradation of TCS was highly accelerated, while SMX degradation was relatively slow. In the second irradiation set, where pH was adjusted to 5, the opposite direction was observed. The amount of UV dose required for 90% removal of the mixture was only 170 mJ/cm2, while, when the pH of the irradiated groundwater was maintained constant (pHs 7.5 and 5; Figure 2.4b), a significant higher amount of UV doses of 548 and 256 mJ/cm2 were needed for obtaining 90% removal of the mixture. Therefore, by modifying the treated groundwater pH during the process, a highly improved removal efficiency of the target mixture is accomplished (in comparison to UV treatment at constant pH). 35 (a) 0 -0.5 TCS 5 TCS 7.5 SMX 5 SMX 7.5 Ln ( C H/C0 ) -1 -1.5 -2 90% removal -2.5 -3 0 50 100 150 200 UV dose (mJ/cm2) (b) 0 Ln ( CH/C0 ) -1 TCS 5 TCS 7.5 SMX 5 SMX 7.5 -2 90% removal -3 -4 0 100 200 300 400 500 600 700 2 UV dose (mJ/cm ) Figure 2.4- UV dose-based photo-degradation of TCS and SMX, measured in a mixture (0.5 mg/L each compound), in GW-1, using MP collimated beam apparatus, at (a) pH modification of the treated water from 7.5 to 5 after irradiation dose of 45 mJ/cm2; and, (b) constant pH values 5 and 7.5. The dashed lines represent 90% degradation. 37 Table 2.4 presents the UV dose required to achieve 90% removal of TCS and SMX (in a mixture), from three different groundwater types, using the aforementioned irradiation procedures. Results indicate that modifying the groundwater pH during the photodegradation process, from their original pH values i.e. pH 7.5-9.7 to pH 5, has vastly decreased the UV dose required for removal of TCS and SMX in a mixture in relation to constant pH. Consequently this pH modification during the treatment has improved the treatment efficiency. Table 2.4 - UV dose required to achieve 90% removal for TCS and SMX (in a mixture), in different groundwater, using constant pH and pH modification during irradiation UV dose for 90% removal (mJ/cm2) Constant pH pH modification during the treatment pH 7.5-8 pH 5 GW-1 548 256 170 GW-2 511 247 162 GW-3 325 509 179 Practical application The influence of pH on direct photolysis rate of PPCPs can be applied to actual practices of UV water treatment and to design of a pH modified UV based photolytic system. The first step in treating water polluted with multiple pollutants (more than one) using the proposed pH enhanced photolysis is the determination of the optimal pH values for each pollutant or group of pollutants. Then, a sequential pH optimization can be used for an optimized photolytic treatment of the water, which includes several steps as follows: (a) first pH modification of the water to optimize the degradation of one pollutant type, UV irradiation of the water and (b) a second pH modification to optimize the degradation of the second pollutant type and a second exposure to UV radiation. Figure 2.5 is a schematic diagram illustrating a ‘series’ type exemplary system for treating a mixture of pollutants in water via pH optimized photolysis. 35 Pollutants A + B in water Input 1st pH adjustment for A 1st direct photolytic UV reactor 2nd pH adjustment for B 2nd direct photolytic UV reactor Output Fig 2.5- Sequential pH optimization Photo-degradation intermediates analysis When suggesting an engineered photodegradation-based treatment to remove PPCPs from water, it is highly important to evaluate the influence of the photo-generated intermediate products on the treatment. It is commonly accepted that intermediates formed during process only marginally influence the treatment’s efficiency, due to their low environmental 39 concentration. However, active degradation intermediates may occasionally be formed, in particular if the active part of the molecule remains unmodified, thus adversely impacting the quality of the treated water (e.g. Sisson et al., 1997). The importance of this topic is demonstrated in Table 2.5, presenting 12 byproducts identified herein during direct UV photolysis of CIP. The moieties which are involved in the CIP photodegradation are the piperazine residue, the carboxylic group and the fluorine atom. The quinolone skeleton however remains unchanged. The piperazine ring is either subjected to breakdown (intermediates 1-2, 4-5 and 8-12) or is epoxidized at positions 2, 3 (of the piperazine) (intermediate 6). Decarboxylation occurs at position 3 of the quinolone skeleton (intermediate 7). The fluorine at position 6 is either substituted by hydroxyl group (intermediate 3) or by hydrogen atom (intermediates 1,2,4,5 and 12). Various researchers have already identified the formation different by products as a result of CIP photo-degradation (e.g. Vasconcelos et al., 2009). In light of the fact that the characteristic quinolone is believed to be responsible for the CIP antimicrobial activity (Dodd et al., 2006), intermediates which merely involve modification of the piperazine residue may still potentially exhibit antibacterial potencies. Table 2.5- HPLC/MS analysis of CIP direct UV degradation byproducts Nº (m/z) CIP 332.1 Rt† Proposed product structures 4.9 HN N Nº (m/z) Rt† Proposed product structures N O F O 1 2.9 (316.1) H2N OH N 7 7.0 (288.1) O N HN N OH N F O O O 2 6.1 (316.1) 8 3.4 (334.1) O NH N O H2N N O N NH OH OH O O O 05 F 3 2.8 (330.1) 9 6.8 (334.1) HN N N O NH N O NH HO O O OH O OH F 4 7 (330.1) 10 4.0 (306.1) O NH N O N H2N OH H3C N O NH OH O O F 5 5.5 (344.1) O NH N O N 11 7.5 (263.1) H2N N OH O F O O O 6 4.4 (346.1) O 12 3.2 (288.1) HN N N H2N OH N O O F NH O † OH O OH MS retention time (min) 01 CHAPTER III A highly efficient bismuth-based photocatalyst for the degradation of PPCPs under solar light irradiation Introduction The use of semiconductors in combination with sunlight irradiation (i.e. photocatalysis) for the treatment of water and wastewater has attracted growing attention and intense research interest over the last decade. TiO2 is the most widely used photocatalyst, mainly due to its high efficiency, photochemical stability, nontoxic nature and low cost (Hoffmann et al., 1995). TiO2 has been widely applied to remove trace micropollutants (e.g. PPCPs) from water (Doll and Frimmel, 2005a; Reyes et al., 2006; Yang et al., 2008) and wastewater effluent (e.g. Hapeshi et al., 2010; Miranda-Garcia et al., 2011), and as a water disinfectant (Sjogren and Sierka, 1994; Rincon and Pulgarin, 2004). The main limitation of TiO2 is a relatively wide band gap that results in about 5% spectral overlap between its absorbance and sunlight emission (λ<~390 nm). Different attempts to overcome this drawback have used physical and chemical means such as morphological modifications to increase active surface area (e.g. Wong et al., 2011), doping TiO2 with different elements, such as nitrogen (e.g. Cheng et al., 2011) and use of other semi-conductors such as WO3 and ZnO, used with TiO2 or as independent catalysts (Lin et al., 2008; Martinez et al., 2011). Bismuth-based photocatalysts showed effective solar-light activity, due to their potential absorbance in the UVA and visible range and their non-toxic nature (Rohr, 2002). Fu et al. (2005) found that Bi2WO6 was more efficient than TiO2 (Degussa P25) in the degradation of the dye molecule Rhodamine B under visible light, and Wang et al. (2010) showed its potential to remove the endocrine disrupting compound Bisphenol A under solar simulated irradiation. Other bismuth-based catalysts such as CaBi2O4 (Tang et al., 2004) and 02 NaBiO3 (Kako et al., 2007) also exhibited reactivity under visible light, degrading dyes and other organic pollutants. A recently new development of bismuth-based catalyst is the bismuth-oxyhalides (common formula BiOX, X=F, Cl, Br, I). These photocatalysts structure comprise of a layer of [Bi2O2]+2 slabs interleaved by double slabs of halogen atoms. Zhang et al. (2006) demonstrated the advantage of BiOCl over TiO2 (Degussa P25) in the decomposition of the dye molecule methyl orange under UVA irradiation. Other studies established the potential of different mono-halogen compounds such as BiOF (Su et al., 2010), BiOI (Li et al., 2011) and BiOBr (Feng et al., 2011) to degrade various pollutants in water under UV and visible light irradiation. Enhanced photoreactivity was reported with composite materials such as xBiOI(1-x)BiOCl (Wang et al., 2007) and xBiOBr-(1-x)BiOI (Wang et al., 2008), where the formed hetero-junction improved the separation of electron-hole pairs and inhibited charge recombination. Shenawi-Khalil et al. (2011) were the first to demonstrate the photoreactivity of a new family of bismuth-mixed oxyhalides catalyst with a general structure of BiOCl 1-xBrx (0 ≤ x ≤ 1). Specificaly, the bismuth catalyst (with x = 0.5) exhibited a higher reactivity than Degussa P25 (to remove the dye molecule Rhodamine B), using similar catalyst concentration, under both UV-vis (λ ≥ 385 nm) and visible light (λ ≥ 420 nm). Recently, Gnayem and Sasson (2012) have optimized the synthesis of the BiOCl1-xBrx catalyst mainly by using surface active quaternary ammonium salts as bromide and chloride sources, and performing the synthesis under acidic conditions. These modifications resulted in a far more effective photocatalyst, which totally decomposed Rhodamine B within 120 seconds (with x>0.87) under visible light irradiation. However, this newly optimized BiOCl1-xBrx photocatalyst was never examined to remove water micropollutants, such as PPCPs. The goal of the present study is to determine the photocatalytic efficiency of this catalyst to degrade frequently used PPCPs under simulated solar irradiation, and to examine the influence of different experimental parameters. 03 Material and methods Materials and reagents All PPCPs were obtained from Sigma-Aldrich (>99% purity), LC-MS grade methanol and water were from Bio-Lab Ltd. (Jerusalem, Israel). All chemicals were used as obtained and stock solutions were prepared by dissolving the compounds (separately) in deionized (DI) water (Direct-Q3 UV system, Millipore-France) at a concentrations of 50 mg/L. Degussa P25 (21 nm particle size, ~50 m2/g BET area), a commercially available TiO2, was used as a reference catalyst. Preparation and characterization of BiOCl0.875Br0.125 Synthesis of the new BiOClxBr1-x catalyst is detailed elsewhere (Gnayem and Sasson, 2012). In brief, the catalyst was prepared from bismuth nitrate and surface active quaternary ammonium salts (such as cetyltrimethylammonium halides), in the presence of acetic acid as a co-solvent. The received powder, composed of hierarchiral flower-like microspheres (Figure 3.1), was filtered and washed with water and ethanol and dried at ambient conditions. The ammonium salts served both as bromide and chloride sources as well as structure directing agents. The acidic conditions allowed the complete dissolution of the bismuth nitrate and an instant reaction with the halide anions at room temperature, yielding the desired bismuth oxyhalide product. 00 A Cl-/Br- ratio in the catalyst of 0.875/0.125 produced the highest photocatalytic activity for the degradation of Rhodamine B under visible light (Gnayem and Sasson, 2012). Therefore, the optimized formula BiOCl0.875Br0.125 was used in the present study. Figure 3.1- SEM image of the as-prepared BiOCl0.875Br0.125 . Adated from Gnayem and Sasson (2012) Photodegradation experiments Solar irradiation experiments were carried out in a 150-W xenon arc lamp solar simulator (Sciencetech Inc, SS150W, Canada). The light beam was filtered with an air mass (AM) 1.5 Global filter that simulates the total (direct and diffuse) solar spectrum equivalent to natural sunlight at 48.2° latitude at sea level. Irradiance was measured using a calibrated spectroradiometer (International Light, Model ILT 900R, USA), placed in the same position as the irradiated solution. Total incident irradiance, integrated between 280-950 nm was 710 W/m2, including UVA irradiance of 21.5 W/m2 and UVB irradiance of 0.9 W/m2. Experiments were conducted with 30 mL aqueous solution (DI water) containing the target pharmaceutical (initial concentration of 1 mg/L) and known concentrations of the tested catalyst in suspension, at an inherent initial pH of ~ 5. For specific experiments borate buffer (5 mM) was added to control pH variations during photocatalysis, where pH was adjusted using HCl or NaOH. The examined solution was stirred for 60 min in the dark to ensure adsorption/desorption equilibrium of the tested pharmaceutical on the catalyst prior to irradiation. All experiments were performed in triplicate and relative standard deviations were less than 10%. Water temperature remained approximately constant at 22 (±2) °C during all 04 photocatalytic experiments. Samples of 0.5 mL were taken periodically and the catalyst suspension was separated from the aqueous phase by centrifugation (15 min, 14,600 rpm). The supernatant was then analyzed using HPLC to quantify the tested pharmaceutical concentration. Chemical analysis Pharmaceuticals were detected and quantified by HPLC Agilent 1100 (ACE-RP Phenyl column 2.1mm×250mm) with a diode array detector. The flow rate was set to 0.5 mL/min and the volume injected was 50 µL. The mobile phase consisted of water (A) and methanol (B), at pH 3. The mobile phase eluent gradient started with 60% of eluent A, followed by a 4-min linear gradient to 10% of eluent A, 4-min isocratic elution and a 2 min linear gradient back to 60% of eluent A, maintained for 5 min to equilibration time. The concentration of formaldehyde was measured using the Nash reagent (2 M ammonium acetate, 0.05 M acetic acid, 0.02 M acetylacetone) (Nash, 1953). In this assay, the reagent is mixed with equal volume of the tested sample (after catalyst removal) and heated for 10 min at 50°C in the dark. Formaldehyde was determined from its absorption at 412 nm (7600 1/M cm). Concentration of H2O2 was determined by the molybdate-activated iodide method (Klassen et al., 1994). Absorbance measurements True absorbance measurements of BiOCl0.875Br0.125, accounting for scattering by particles of the catalyst suspended in DI water, were performed with a UV-Vis dual beam spectrophotometer (Varian, Model Cary 100BIO, Victoria, Australia) equipped with 150 mm diameter integrating sphere-IS attachment (Labsphere Diffuse Reflectance accessory (DRA)CA-30) and a center mount sample holder used to position the sample inside the IS. The turbid sample (suspension with catalyst) is placed in a 1 cm path length quartz cuvette with all four windows optically polished. The cuvette is fixed to a spring loaded holder that hangs in the center of the sphere, connected to the top sphere cover. 05 Results and discussion Parametric study The goal herein was to examine the influence of different experimental parameters on the BiOCl0.875Br0.125 photocatalytic process, through the degradation of carbamazepine (CBZ), a widely used antiepileptic drug. Studies have already determined that CBZ resist conventional water and wastewater treatments, with removal efficiencies by wastewater treatment plants (WWTPs) below 10% (e.g. Clara et al., 2004). In the classification scheme for pharmaceutical biodegradation, CBZ has the status of “no-removal” (Joss et al., 2006). Catalyst reactivity The degradation of CBZ versus time at a concentrations of 0 (no catalyst) and 500 mg/L BiOCl0.875Br0.125, is presented in Figure 3.2. Direct photodegradation of CBZ after 5 min exposure was negligible (0 mg/L catalyst). Adding 500 mg/L BiOCl0.875Br0.125 to the sample dramatically increased its removal rate, demonstrating the high photo-reactivity of BiOCl0.875Br0.125 under simulated solar light. In all cases, the photocatalytic degradation of CBZ followed apparent first-order kinetics. 1 0 mg/L 0.8 CBZ (C/C0) Bi-500 mg/L 0.6 P25-500 mg/L 0.4 0.2 0 0 1 2 3 4 5 Time (min) Figure 3.2- Carbamazepine (CBZ) degradation (initial concentration 1 mg/L) under simulated solar light vs. time with 0 and 500 mg/L BiOCl0.875Br0.125 (and comparison to P25 500 mg/L), in DI water at pH 5. 07 The BiOCl0.875Br0.125 photo-reactivity was compared to Degussa P25, a commercially available TiO2, using 500 mg/L concentration of either catalyst, at pH 5 (Figure 3.2). The photocatalytic removal of CBZ after 5 min of irradiation reached ~66% and 80% with P25 and BiOCl0.875Br0.125 respectively. These results were obtained under specific experimental conditions, and a detailed comparison between BiOCl0.875Br0.125 and P25 is beyond the scope of this study. Reactivity of the catalyst was fully maintained for five consecutive irradiation cycles (data not shown). Where, with each new cycle, 1 mg/L of CBZ was freshly added to the same catalyst suspension after the former batch was 99% degraded (after 10 min). Light absorption by the catalyst and effect of irradiation wavelength The standard method used to measure absorbance relies on transmittance of light captured by a detector that is placed in line with the sample, using a spectrophotometer. The drawback of this method is that particles which scatter light at angles outside the reception angle of the detector will result in significant error in absorbance measurements (e.g. Du and Rabani, 2004). True UV absorbance, can be measured using integrating sphere (IS) spectrophotometers, optical devices that integrate the radiant flux of most reflected and transmitted radiation simultaneously (Mamane et al., 2006). 1.2 120 Ab… 100 Abssorption coefficient (1/cm) 1 80 0.6 60 0.4 40 0.2 20 0 200 0 250 300 350 400 Wavelength (nm) Irradiance (uW/cm2) 0.8 450 Figure 3.3- Irradiance of the solar simulator lamp and absorption coefficient spectra of 500 mg/L aqueous suspensions of BiOCl0.875Br0.125 (in DI water, pH 5), measured in the integrating sphere. 05 Figure 3.3 illustrates the spectra obtained from the solar simulator and the absorption coefficient spectra of 500 mg/L aqueous suspensions of BiOCl0.875Br0.125, measured using the integrating sphere (between 200-450 nm). It is evident that there is a general decrease in the catalyst’s light absorption with increase in wavelength. Where, above approximately 380 nm, absorption coefficient is negligible for the specific bismuth formulation studied. In light of these newly acquired absorption data, we estimate that the degradation of Rhodamine B under visible light, previously reported by Gnayem and Sasson (2012), was initiated through the absorption of light by the dye, rather than by the catalyst. Figure 3.4- Photo-catalytic removal of CBZ with 500 mg/L BiOCl0.875Br0.125 (in DI water, pH 5) at different wavelengths, after light exposure of 10 J/cm2. The effect of the irradiation wavelength on the photocatalytic degradation of CBZ was further investigated, using two long pass filters which screened all light below 320 and 400 nm, respectively. Specifically, the examined wavelength ranges were: λ<280 nm (no filter), λ<320 nm and λ<400 nm. Photo-degradation experiments were carried out in the presence of 500 mg/L BiOCl0.875Br0.125 (DI water, pH 5). The exposure time at each wavelength range was determined in order to achieve an identical light power density of 10 J/cm2, which was 09 calculated as the product of the exposure time (sec) and the measured incident irradiance (for each tested range), integrated over the entire UV-visible range. Carbamazepine (CBZ) was only marginally degraded under visible light irradiation (~5%), while extending the irradiation spectrum to the UVA and UVB range increased the removal of CBZ to 68% and 85%, respectively (Figure 3.4). These results were expected since BiOCl0.875Br0.125 does not absorb light at wavelengths longer than ~380 nm (Figure 3.3). Quantum yield (ф) calculation Quantum yield (ф) for the degradation of CBZ with 500 mg/L BiOCl0.875Br0.125 (DI water at pH 5) was calculated as the ratio between its initial degradation rate and the rate of photon absorbed by the catalyst, as follows (Dalrymple et al. 2010): r '0 ( E ( ) (1 10 a( )z ) / z) (3.1) 0 p where, r’0 is the initial degradation rate of CBZ (M/s), Ep0 is the measured photon incident irradiance on the surface of the sample (millieinstein/s cm2), z is the depth of the sample (cm) and a is the absorption coefficient of the solution (1/cm), measured with the integrating sphere (Figure 3.3) (assuming light absorption by CBZ is negligible at these wavelength; Lester et al., 2012). The summation of Equation 3.1 was taken over the wavelength range between 280–380nm, accounting for the overlap between the catalyst absorbance and the lamp emission (Figure 3.3). Quantum yield for CBZ degradation was found to be 0.75% (±0.05). A similar quantum yield was calculated when a long pass filter was utilized, screening all light below 320 nm, thus integrating wavelengths between 320-380 nm (data not shown). This implies that the efficiency of an absorbed photon to excite an electron in the BiOCl 0.875Br0.125 catalyst is similar in the UVA and UVB range. The reason for the higher degradation rate of CBZ at λ>280 nm (than at λ>320 nm), presented in Figure 3.4, is therefore the increase in light 45 absorption by the catalyst at shorter wavelengths (Figure 3.3), rather than an increase in its quantum yield. Little information exists in the literature on quantum yields for the photocatalytic degradation of organic pollutants. The main reason is that absorbed photons are experimentally difficult to measure, due to light scattering by the suspended particulates. Therefore, most studies use the term quantum yield (or photonic efficiency) referenced to incident photons, which merely represents the lower limit of the actual quantum yields (Serpone, 1997). For example, incident irradiance was used to calculate the quantum yield for the degradation of the antibiotic sulfamethoxazole (0.21-0.75%; Xekoukoulotakis et al., 2011) and the pesticide formetanate (0.5%; Marinas et al., 2001). Effect of catalyst and CBZ concentrations The catalyst concentration is an important operating parameter in a suspended photocatalysis system. Carbamazepine (CBZ) first-order degradation rate constant (k, 1/min) steadily increased with addition of BiOCl0.875Br0.125 at concentrations up to 500 mg/L (Table 3.1). With increasing catalyst concentration above 500 mg/L, a slightly reduced rate constant was observed. Different studies found optimal concentrations for suspended catalysts in the range of 75-1000 mg/L, depending on the type of catalyst and target pollutant, on the pollutant's concentration and on the spectral photon flux (Yang et al., 2008; Achilleos et al., 2010; Hapeshi et al., 2010; Martinez et al., 2011). Table 3.1- First-order rate constants for CBZ degradation at different catalyst concentrations R2 Catalyst concentration k (mg/L) (1/min) 0 - 50 0.062 (±0.011) 0.996 100 0.107 (±0.016) 0.998 250 0.206 (±0.016) 0.994 500 0.303 (±0.007) 0.986 750 0.274 (±0.008) 0.989 41 At low catalyst concentrations, increase in the degradation rate is commonly explained by increase in available active catalysts sites and light absorption. While at high concentration, the catalyst provides a screening effect that reduces light penetration into the irradiated solution. Figure 3.5 demonstrates the screening effect of BiOCl0.875Br0.125, where, absorption coefficient of the catalyst solution (exemplified for λ = 300 nm) increased with its concentration up to 750 mg/L, when practically all the light other than the backscattered is absorbed. The discrepancy between the optimal catalyst concentration for CBZ degradation (i.e. 500 mg/L) and the catalyst concentration with the highest light absorption (i.e. 750 mg/L) can be explained by other mechanisms such as internal scattering, deactivation of sites through collision with ground-state catalysts (Neppolian et al., 2002) and agglomeration of the catalyst at high concentrations (So et al., 2002). Absorption coeff. at 300 nm (1/cm) 1.6 1.2 0.8 0.4 0 0 100 200 300 400 500 600 700 800 900 1000 BiOCl0.875Br0.125 concentration (mg/L) Figure 3.5- Absorption coefficient measured at 300 nm with the integrating sphere as a function of BiOCl0.875Br0.125 concentration. Measurements were taken in DI water at pH 5. The influence of initial CBZ concentration on its initial degradation rate is demonstrated in Figure 3.6. Increasing CBZ concentration increased its degradation rate until it approaches a 42 plateau. This behavior is in good agreement with the Langmuir–Hinshelwood (L–H) model, commonly used to describe photo-catalytic degradation of organic compounds: r0 ( k K ads [CBZ ]0 d CBZ ) 0 in dt 1 K ads [CBZ ]0 (3.2) where, r0 is the initial degradation rate of CBZ (mM/min), kin is the intrinsic reaction rate constant (mM/min), Kads is the L-H adsorption constant of CBZ on the catalyst surface (1/mM) and [CBZ]0 is the initial concentration of CBZ in the aqueous solution. Figure 3.6- Initial degradation rate of CBZ (r0) as function of its initial concentration (C0). The inset presents 1/r0 vs. 1/[CBZ]0 for different initial CBZ concentrations The L-H model is usually plotted as the inverse of the pollutant’s initial degradation rate (1/r0) vs. the inverse of its initial concentration in the solution (1/C0), as follows: 1 1 1 1 r0 k in k in K ads [CBZ ]0 (3.3) The linear reletionship between 1/r0 and 1/[CBZ]0 (inset of Figure 3.6) suggests that the L-H model adequatly describes the kinetics of the examined photocatalytic process. From the 43 slope and intercept of the linear line, kin and Kads were calculated to be 0.25 mM/min and 261.7 1/mM, respectively. A much lower Kads of 28.9 1/mM was found by Martinez et al. (2011) for the degradation of CBZ using P25 TiO2 and near UV–vis light, suggesting that adsorption is more important in the degradation of CBZ with BiOCl 0.875Br0.125 than with P25. This topic is further evaluated in the following chapter. Effect of pH Figure 3.7 illustrates the degradation rate of CBZ, in the presence of 500 mg/L BiOCl0.875Br0.125, at different pH values in the range of 4-9. Obviously, photocatalysis rate decreases with increasing pH up to pH 9, with negligible photodegradation rate at a pH value of 9. 0.5 0.4 k (1/min) 0.3 0.2 0.1 0 3 4 5 6 7 8 9 10 pH Figure 3.7- First-order rate constant for CBZ degradation with BiOCl0.875Br0.125 (500 mg/L) vs. pH. Experiments were conducted in borate buffer (5 mM). pH may alter the pollutant’s charge and the catalyst surface charge which can further impact the adsorption kinetics and degradation mechanism of the target pollutant (e.g. Hapeshi et al., 2010; Wang et al., 2010). Furthermore, with TiO2, variation in pH can influence the production rate of hydroxyl radicals (•OH) (e.g. Lin et al., 2008). For example, in the 40 presence of TiO2 and UVA irradiation, paracetamol degradation rate slightly increased with increasing pH from 3.5 to 9.5, presumably due to enhanced •OH formation (Yang et al., 2008). On the other hand, the reaction rate significantly decreased at pH 11.0, mainly due to increase in electrostatic repulsion between the TiO2 surface (negatively charged at pH > 6.3) and paracetamol (negatively charged at pH above 9.5). Fu et al. (2005) showed that raising solution pH from 6.5-10, reduced the adsorption degree and correspondingly the degradation rate of Rhodamine B using BiWO3 photo-catalysts under visible light irradiation. To understand the influence of pH on the photodegradation of CBZ, the % adsorption of CBZ at different pH values was determined (Table 3.2). The charge of the BiOCl 0.875Br0.125 surface (zeta potential), taken from Gnayem and Sasson (2012), is also presented. Adsorption of CBZ (1 mg/L) on BiOCl0.875Br0.125 (500 mg/L) was measured in the dark after 60 min of equilibration time. Table 3.2- Influence of pH on CBZ degradation rate constant (k, 1/min), (dark) adsorption of CBZ and zeta potential of the catalyst pH 4 k (1/min) 0.43 Adsorption (%) 26.5 *Zeta potential (mV) -0.01 5 0.32 19.5 -4.64 6 0.23 16.1 -9.13 7 0.15 14.5 -12.47 8 0.03 8.2 -14.55 9 0.01 7.0 -17.36 *Adapted from Gnayem and Sasson (2012) The zeta potential was negative at all pH values from 5 to 9. Decreasing solution pH from 9 to 4 shifts the charge on the catalyst surface from negative to neutral (point of zero charge – PZC at pH 4) and is correlated to the increase in CBZ (dark) adsorption. Carbamazepine (CBZ) is neutrally charged and relatively hydrophobic (log Kow = 2.45, Trenholm et al., 2006) at the entire tested pH range, while the catalyst is most hydrophobic at pH 4 (PZC). Therefore, decreasing the solution pH from a value of 9 to 4 will increase the hydrophobic 44 attraction between CBZ and the catalyst, increasing CBZ adsorption and, ultimately, enhancing its degradation. Effect of water constituents It is generally accepted that non-target organic solutes in water may inhibit the photodegradation of a target pollutant by the combination of light attenuation (for light absorbing solutes), competition for reactive species and surface deactivation of the catalyst by adsorption (Doll and Frimmel, 2005b). To evaluate the influence of natural water constituents on the BiOCl0.875Br0.125 photocatalytic process, degradation of CBZ was examined in the presence of different inorganic ions and natural organic matter. Experiments in this section were conducted by adding different concentrations of various model compounds to a solution of 500 mg/L BiOCl0.875Br0.125 and 1 mg/L CBZ (in borate buffer 5 mM, pH 7). Influence of natural organic matter Suwannee River fulvic acid (SRFA) was used to simulate natural organic matter (International Humic Substances Society-IHSS; St. Paul, MN). While addition of SRFA to the solution (up to 20 mg/L) clearly inhibited the degradation of CBZ (Figure 3.8), it did not affect the CBZ adsorption to the catalyst, which remained unchanged even at 20 mg/L SRFA (results not shown). Dissolved fulvic acid have a yellow-brown color and strong spectral absorbance in the UV range (Frimmel, 1994), therefore, we assume that the inhibiting effect of SRFA on the photodegradation of CBZ results mainly from its role as an inner filter for incoming photons (rather than competition for active sites). 45 1 0.8 CBZ (Ct/C0) 0.6 0 mg/L 2 mg/L 0.4 6 mg/L 12 mg/L 0.2 20 mg/L 0 0 2 4 6 8 10 Time (min) Figure 3.8- Photodegradation of CBZ with BiOCl0.875Br0.125 (500 mg/L) at different concentrations of SRFA. Experiments conducted in borate buffer at pH 7. Influence of main ions Different inorganic ions were added (separately) to the irradiated solution at typical natural water concentrations (Table 3.3). Table 3.3- Examined ions Examined ions Compound added Ion concentration range (mg/L) Ca2+ Ca(OH)2 0-150 Mg2+ Mg(OH) 2 0-75 Na+ NaCl 0-400 Cl- NaCl 0-600 SO42- Na2SO4 0-500 NO3- NaNO3 0-100 Na2HPO4 0-10 NaHCO3 0-370 PO43HCO3 - 47 While addition of Ca2+, Mg2+, Na+, Cl-, SO42- and NO3- did not affect the degradation rate of CBZ (results not shown), increasing the concentration of PO43- up to 10 mg/L highly reduced the CBZ removal rate (Figure 3.9). The influence of HCO3- could not be determined due to high pH variability between samples of different HCO3- concentrations and variations in pH that occurred during irradiation. Abdullah et al. (1990) already demonstrated the inhibiting effect of phosphate (PO43-) on TiO2, where, the photodegradation rate of salicylic acid, aniline, and ethanol was markedly decreased with the addition of PO43- to the solution (at ~100 mg/L). The effect of phosphate on TiO2 was mainly attributed to its high adsorption degree to the catalyst. 1.2 1 CBZ (Ct/C0) 0.8 0.6 [PO4] = 0 mg/L [PO4] = 2 mg/L 0.4 [PO4] = 6 mg/L 0.2 [PO4] = 10 mg/L 0 0 2 4 6 8 10 time (min) Figure 3.9- Photodegradation of CBZ with BiOCl0.875Br0.125 (500 mg/L) at different concentrations of PO43-. Experiments conducted in borate buffer at pH 7. To further examine the influence of PO43- on BiOCl0.875Br0.125 photocatalysis, the extent of CBZ (dark) adsorption (%) was measured in the presence of different PO43- concentrations (Figure 3.10). Clearly, increasing the concentration of PO43- reduced the adsorption of CBZ to the catalyst (by competing over adsorption sites), presumably reducing its degradation rate. 45 The mechanism for PO43- adsorption is further evaluated at section 3.3.2. It should be noted that PO43- does not absorb light in the UV range, thus, its inhibiting effect cannot be attributed to light attenuation. 14% CBZ Adsorption 12% 10% 8% 6% 4% 2% 0% 0 2 4 6 8 10 12 14 16 [PO43-] mg/L Figure 3.10- Adsorption of CBZ in the dark vs. PO43- concentration. Experiments conducted in borate buffer at pH 7. Reactive species It is well known that photo-excitation of a semiconductor produces electron-hole pairs which are highly mobile and quickly recombine, in competition to their trapping (localization) at the semiconductor surface. Oxidation of a target pollutant occurs either through direct reaction with valance band holes (h+) or through the intermediate of reactive species such as •OH radicals (e.g. Hoffmann et al., 1995). The purpose of this section is not to determine the specific reactive species generated by BiOCl0.875Br0.125 under solar irradiation, but rather to assess their "oxidative power" and their similarity to those generated by TiO2. Oxidation of methanol (CH3OH) to formaldehyde (HCHO) (Equations 3.4-3.5) was used as an indicative step for the photogeneration of highly reactive species (Goldstein et al., 2008). Oxidizing the aliphatic methanol may occur by either 49 holes or radicals, as long as their oxidation potential is high enough (redox potential of methanol ~ 1.34V; Goldstein et al., 2008). Experiments were carried out in a 30 mL aerated suspension (DI, pH 5) of BiOCl0.875Br0.125 (500 mg/L) and methanol (2 M). h+/•OH + CH3OH •CH2OH + O2 H+/H2O + •CH2OH •OOCH2OH HCHO + H+ + O2•- (3.4) (3.5) The results clearly show a linear accumulation of formaldehyde with irradiation time (Figure 3.11). The linear relationship also suggests that formaldehyde does not react with the reactive species at the high methanol concentration used. A slower rate of formaldehyde formation is observed when CBZ and methanol are tested together (as opposed to methanol alone), due to competition between the two target compounds for the reactive species. A near linear accumulation of formaldehyde was previously demonstrated for different catalysts (Marugan et al., 2006; Goldstein et al., 2008), indicating the presence of highly reactive oxidizing species. In bismuth based photocatalysis, photo-excitation of valence band Bi+3 generates Bi+5 (or Bi+4) holes, with an estimated standard redox potential of ~1.59V (at pH 0) (Fu et al., 2005). These holes are usually considerd as the main oxidative species in the photodegradation of organic pollutants (e.g. Shenawi-Khalil et al., 2011), and we estimate they are also responsible for the degradation of methanol (and CBZ) in the present study. 55 180 with carbamazepine [Formaldehyde] uM 150 w/o carbamazepine 120 90 60 30 0 0 5 10 15 20 25 Irradiation time (min) Figure 3.11- Formaldehyde formation vs. irradiation time in a BiOCl0.875Br0.125 (500 mg/L) suspensions (DI, pH 5), in the presence of 2M methanol, with and without CBZ (1 mg/L). Since oxidation of the aliphatic methanol (CH3OH) to •CH2OH is known to be relatively slow compared to other aliphatic and aromatic compounds (as cited by Goldstein et al., 2008), our results imply that the BiOCl0.875Br0.125 photocatalytic process will not be limited to a small number of specific organic compounds. Addition of H2O2 Photocatalytic degradation of micropollutants can be accelerated by the addition of hydrogen peroxide (H2O2), as demonstrated for TiO2 by several researchers (So et al., 2002; Hapeshi et al., 2010). Hydrogen peroxide (H2O2) can promote the formation of reactive species through direct reaction with the photogenerated electrons or indirectly by reacting with superoxide ions: e- + O2 H2O2 + O2-• H2O2 + e- O2-• HO2• (3.6) •OH + OH- + O2 (3.7) •OH + OH- (3.8) 51 Figure 3.12 demonstrates the influence of added H2O2 (in the range of 0.05-2.5 mM) on the BiOCl0.875Br0.125 (500 mg/L) photodegradation rate of CBZ (DI, pH 5). Results are presented as the rate constant for CBZ degradation with H2O2 relative to its degradation rate without H2O2. Relative CBZ rate constant (kwith H2O2/kw/o H2O2 ) 1 0.8 0.6 0.4 0.2 0 0 0.5 1 1.5 2 2.5 3 Added H2O2 (mM) Figure 3.12- BiOCl0.875Br0.125 photodegradation of CBZ (presented as the degradation rate constant with H2O2 relative to the rate constant without H2O2) vs. added H2O2 concentration. Experiments were conducted in DI at pH 5. Addition of H2O2 at low concentrations (≤ 0.1 mM) did not significantly affect the degradation rate of CBZ. Whereas, increasing H2O2 concentration to 0.5 mM sharply decreased the CBZ degradation rate, followed by a plateau up to 2.5 mM H2O2. Apparently, the mechanism for enhanced •OH production by H2O2, previously suggested for TiO2, does not apply for BiOCl0.875Br0.125. The decrease is CBZ degradation rate at H2O2 concentrations ≥ 0.5 mM may be explained by H2O2 competition with CBZ over BiOCl0.875Br0.125 adsorption sites and/or oxidizing species, as demonstrated below (Goldstein et al., 2009): H2O2 + h+ 2H+ + O2•- (3.9) 52 Indeed, measurements during the experiments showed up to 30% adsorption of H2O2 to the catalyst after 60 min in the dark (with no clear correlation to H2O2 initial concentration). Moreover, concentration of H2O2 in the solution was reduced by up to 50% following 5 min of irradiation. Degradation of additional PPCPs: Ibuprofen, Bezafibrate and Propranolol BiOCl0.875Br0.125 was further evaluated for its ability to degrade other pharmaceuticals of different therapeutic classes, namely: bezafibrate-BZF (lipid regulator), ibuprofen-IBF (antiinflammatory) and propranolol-PPL (beta blocker). To better simulate real water treatment, experiments were conducted in borate buffer at pH 7 (typical for natural waters) and catalyst concentration of 200 mg/L (to minimize particles interferences). The target pharmaceuticals were selected due to their frequent detection in WWTPs and in the aquatic environment (e.g. Ternes, 1998), and their different charges at the pH examined. At pH 7, BZF (pKa 3.6) and IBF (pKa 4.4) were negatively charged, and PPL (pKa 9.67) was positively charged. Figure 3.13a presents the first-order degradation rate constant for the selected compounds, (tested separately at C0 = 1 mg/L), while Figure 3.13b shows the degree of the compounds’ adsorption (%) to the catalyst, measured in the dark prior to irradiation. Results for CBZ are presented for comparison, and direct photolysis of all compounds was negligible (data not shown). 53 (a) (b) Figure 3.13- (a) First-order rate constant for the degradation of CBZ, PPL, BZF and IBF with 200 mg/L BiOCl0.875Br0.125 (in borate buffer 5mM, pH 7) and (b) Adsorption degree of the compounds measured in the dark prior to irradiation. All examined compounds were efficiently degraded by the BiOCl0.875Br0.125, validating its potential to treat pharmaceutical contaminated water. Degradation rate of the different compounds followed a decreasing order of: PPL > BZF > IBF > CBZ (highest removal for PPL). Interestingly, while BZF and IBF were highly adsorbed to the catalyst, adsorption of PPL was negligible (Figure 3.13b). Thus, the importance of (dark) adsorption to the photodegradation process is compound-dependent. Another observation relates to the different adsorption mechanism of the examined compounds. The negatively charged BZF and IBF are most likely adsorbed to the positively charged [Bi2O2]+2, suggesting an ion 50 exchange mechanism; while PPL does not adsorb due to its positive charge. The ion exchange mechanism may also explain the adsorption of PO43- to the catalyst (observed at section 3.3.1.6), and its inhibiting effect on the CBZ degradation rate. References Achilleos, A., E. Hapeshi, N. P. Xekoukoulotakis, D. Mantzavinos, and D. FattaKassinos, "Factors affecting diclofenac decomposition in water by UV-A/TiO(2) photocatalysis", Chemical Engineering Journal, 161 (1-2): 53-59 (2010). Batt AL, Kim S, Aga DS (2007) Comparison of the occurrence of antibiotics in four full-scale wastewater treatment plants with varying designs and operations. Chemosphere, 68, 428-435. Bolton, J.R.and Mihaela, I.S. (2002). Fundamental photochemical approach to the concepts of fluence (UV dose) and electrical energy efficiency in photochemical degradation reactions, Res. Chem. Intermed., 28, 857–870. Buxton, B. V., Greenstock, C.L., Helman, W.P and Ross, A.B. (1988). Critical review of rate constants for reactions of hydrated electrons, hydrogen atoms, and hydroxyl radicals in aqueous solution. J. Phys. Chem. Ref. Data 17, 513–886. Hapeshi, E.A. Achilleos, M. I. Vasquez, C. Michael, N. P. Xekoukoulotakis, D. Mantzavinos, and D. Kassinos, (2010) Drugs degrading photocatalytically: Kinetics and mechanisms of ofloxacin and atenolol removal on titania suspensions", Water Research, 44 (6): 1737-1746. Huber, M.M., Canonica, S., Park, G.Y. and Gunten, U.V. (2003). Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environ. Sci.Technol. 37, 1016–1024. Mamane, H. Shemer, H. and Linden, K.G. (2007) Inactivation of E.coli, B. subtilis spores, and MS2, T4, and T7 phage using UV/H2O2 advanced oxidation, Journal of Hazardous Materials, 146, 479-486. Martinez, C., M. Canle, M. I. Fernandez, J. A. Santaballa, and J. Faria (2011) Kinetics and mechanism of aqueous degradation of carbamazepine by heterogeneous photocatalysis using nanocrystalline TiO(2), ZnO and multi-walled carbon nanotubes-anatase composites, Applied Catalysis B-Environmental, 102 (34): 563-571 54 CHAPTER IV The Interaction between Ozonation and Wastewater Particles Introduction Ozone was shown to be an effective barrier against emerging trace organic contaminants (TrOCs), and is increasingly used to remove these compounds from municipal and industrial wastewater effluent (Ternes et al. 2003; Huber et al. 2005; Bahr et al. 2007; Hollender et al. 2009; Wert et al. 2009; Lester et al. 2013). The extent of TrOCs removal is typically determined by the applied ozone dose, the (second-order) reaction rate constant between ozone and the target contaminant (kO3,TrOC, M-1s-1) and the wastewater composition (Huber et al. 2003; Wert et al. 2009; Nöthe et al. 2009). Among the different wastewater constituents, dissolved organic matter and nitrite (which readily react with ozone) are considered the main influencing parameters (Staehelin & Hoigne 1985; Paraskeva & Graham 2002). The influence of particulate matter on the other hand is less clear. Interactions of ozone with particles during drinking water application were investigated for several years (Bourgine et al. 1998; Chandrakanth & Amy 1996; Jasim et al. 2008; Currie et al. 2003). Several researches have reported that following ozonation, particle count decreases, particle size increases, zeta potential increases and turbidity removal increases (Jekel 1983; Georgeson & Karimi 1988; Farvardin & Collins 1989; Chandrakanth et al. 1996; Jasim et al. 2008). Few mechanisms were suggested for those phenomena such as polymerization of NOM, reduction in adsorbed organics molecular weight and/or lysis of algae followed by release of coagulating biopolymers (Grosso & Weber 1988; Chandrakanth & Amy 1996). Other studies found that interaction between ozone and wastewater particles may also take place (Jekel 1994; Zhu et al. 2008; Genz et al. 2011). For example, reaction of dissolved ozone with wastewater particles was suggested to contribute to the particles destabilization, polymerization and subsequent removal by coagulation (Jekel 1994). 55 Ozone-particles interaction mainly depends on the properties of the particles. For example, biopolymers (proteins and polysaccharides) particles were found to be transformed to smaller biopolymers fragments following ozonation (Genz et al. 2011). The particle size may also affect the interaction since ozone can aggregate fine particles and break down large ones (Yan et al. 2007). Other particles characteristics such as electrostatic interaction, dispersion forces and hydrophobic bonding were also found to play an important role in the of the particles are also can be readily affected by ozonation, (Zhu et al. 2008). Other influencing mechanisms may include ozone absorption by particles, sorption of some compounds to particles and/or interaction with colloids (Holbrook et al. 2004; Huber et al. 2005). Very little studies are available on ozone-particles interactions during wastewater ozonation, and their influence on the removal of TrOCs. It was shown that wastewater suspended material only marginally affected ozone degradation of TrOCs (Huber et al. 2005); however, total suspended solids (TSS) bulk parameter was used, with no additional fractionation. The overall goal of this study was to examine the interaction of ozone with particles of different sizes, during wastewater ozonation. This was done by evaluating the influence of different size particles on ozone degradation of selected TrOCs, as well as the changes in wastewater bulk parameters and the counter-effect of ozone on wastewater particles of different sizes. Materials and methods Standards and Reagents Six trace organic compounds (TrOCs) and were analyzed during experiments and can be categorized in two groups; fast (kO3 > 104 M-1s-1), and slowly reacting (kO3 < 10 M-1s-1) compounds. Diclofenac, Sulfamethoxazole, Carbamazepine, and Iohexol (>99% purity) were obtained from Sigma-Aldrich. Iopromide and Iopamidol (>99% purity) were provided from Holland-Moran, Israel. Pharmaceutical subclass, CAS numbers, rate constants and analytical data are summarized in table 1. LC-grade methanol and water were purchased from Bio-Lab (Jerusalem, Israel). All chemicals were used as obtained and working solutions were prepared with deionized (DI) water (Direct-Q3 UV system, Millipore). 57 Experimental Procedure To determine the interaction of ozone with particles of different size, secondary wastewater effluent from the Shafdan site (the largest wastewater treatment plant in Israel) was filtered using filters with different pore-size (11 μm, 6 μm, 2.5 μm paper filters and 1.2 μm, 0.45 μm glass filters, Whatman). Particle analysis showed that approximately 80% of particles above the pore size were removed by filtration. To ensure detection before and after ozonation, the TrOCs standards were spiked before starting the experiments at the initial concentration of 100 µg/L. Bench-scale ozonation experiments were conducted in 500 mL stirred glass vessels. Ozone stock solution (~40 mg/L) was prepared by continuously bubbling ozone gas, produced from an oxygen fed generator (up to 4 g/h, BMT 802N, Germany), in a glass reactor filled with chilled deionized water. Ozone dose is described as specific ozone consumption (Zspec) by normalizing applied ozone doses (mg/L) to the initial dissolved organic carbon (DOC0) concentration. Effluents with different sizes of particulate content were ozonated using three different Zspec - 0.17, 0.69, and 0.93 mg O3/mg DOC0. After complete reaction with ozone, ozonated samples were analyzed for bulk parameters, TrOCs, and particle size distribution. Experiments were repeated in three different sampling events during the period of October 12 to June 13. Analytical Methods Ozone measurements Ozone concentration in the bubbling ozone gas was measured by an ozone gas analyzer (BMT 963, Germany). Ozone concentration in the stock solution was measured by the indigo method (APHA, method 4500B). TrOCs sample analysis The target compounds were detected and quantified by HPLC (Agilent 1100 series; ACE-RP phenyl column 2.1 mm×250 mm) equipped with a UV diode array detector and a mass spectrometer (QTof MS, Waters Premier). The column temperature was 40 ºC, the flow rate was 0.5 mL/min and the injected volume was 100 µL. The HPLC mobile phase consisted of water (A) and methanol (B), adjusted to pH 3 by the addition of formic acid. The mobile 55 phase eluent gradient started with 10% eluent B for 1 min, followed by an 4 min linear gradient to 90% B, a 5 min isocratic elution at 90% B and a 2 min linear gradient back to 10% B, maintained for 4 min for equilibration prior to next run. The flow from the HPLC was passed through a split connector with 60 µL/min of effluent introduced into the MS interface. The mass spectrometer was used in ESI positive mode. Removal efficiencies during ozonation were calculated from the concentrations in spiked secondary and ozonated effluent. Analysis of wastewater-quality parameters Total organic carbon (TOC) of the wastewater was measured using a TOC analyzer (Torch, Teledyne Tekmar, OH, USA). To measure DOC, samples were filtered at 0.45 μm (APHA, method 5310B). Ultraviolet absorbance at 254 nm (UVA) was measured via UV-Vis spectrophotometer (Varian, Cary 100 BIO, Victoria, Australia) for 0.45 μm filtered samples. The turbidity was measured by a conventional 90º side-scatter instrument. Other wastewater parameters were analyzed using standard methods (APHA et al. 2005). Particle analysis Particles suspended in liquid were analyzed by the “Micro Flow Imaging” technology (DPA 4100, ProteinSimple Inc, Ottawa Ontario, Canada). This apparatus employs a digital camera with an illumination and magnification system to capture in-situ images of suspended particles in a flowing sample. Basically, a sample fluid is drawn through a flow cell and sections of the fluid are illuminated with light emitting diode light source at 470 nm wavelength, magnified and imaged onto a digital camera. These captured images are automatically analyzed to determine various size and shape parameters that represent the two dimensional projection of the particles. Analysis was conducted for particle size between 2.25–400 µm. Table 1. Pharmaceutical subclass, second-order rate constants for reaction with ozone (kO3) at pH 7 and CAS registry No. of spiked target compounds 59 MS kOH kO3 [M-1s-1] CAS Registry No. fragmentation [109 M-1s-1] Pharmaceutical Subclass 296.023 7.5 6.8*105 1530779-6 anti-inflammatory Diclofenac 254.059 5.5 5.5*105 723-46-6 sulfonamide antibiotic Sulfamethoxazole 237.102 8.8 5*105 298-46-4 antiepileptic Carbamazepine 66108Iohexol 95-0 95551791.877 3.1 0.8 contrast media Iopromide 09-5 60166777.861 2.8 4.1 Iopamidol 93-0 References: (Sein et al. 2008; Dodd et al. 2006; Huber et al. 2003; Baus et al. 2004) 821.888 5.5 4.1 Results and Discussion Particle size and distribution in Shafdan secondary effluents Turbidity is the most common parameter used to monitor particles. However, it does not provide information on the size, shape, and concentration of particles and thus may not be the suitable measurement of particle removal efficiency (Mamane et al. 2008). In order to extent the understanding of particle distribution, a dynamic image analysis of particles was conducted. The particle size distribution (PSD) was measured during October 2012-June 2013 in 10 different trails. The size of natural particles may be simplified to one parameter by defining particles as spheres and isolating the Equivalent Circular Diameter (ECD). The PSD is shown in table 2 in seven different size ranges as well as the total count in the PSD instrument analysis range (2.25–400 µm). Correlation of turbidity and particle counts has been widely studied (e.g. (Bourgine et al. 1998)) and a direct relationship between turbidity and 2-3 um particle concentration was seen here as well (data not shown). It can be seen that during the Israeli winter (December 2012 to March 2013), the total particle concentration was higher due to when the biological activated sludge treatment is not in optimum conditions. However, in February the particle concentration decreased, probably due to dilution of rainfalls. Similar trend was seen in the different size ranges up to 30 µm. the concentration of higher particle size was hence the accuracy reduced and comparison was difficult to operate. 75 Opposite phenomenon was found in UK as particle concentration wad higher during the summer which was attributed to high algae concentration during summer (Bourgine et al. 1998). Table 2. NFE’s PSD in different dates during Oct 2012-June 2013 3.6.13 2.6.13 6.3.13 12.2.13 24.1.13 17.1.13 26.12.12 29.11.12 5.11.12 15.10.12 5,521 2,077 4,021 912 322 60 0 12,912 5,233 1,871 3,429 888 247 34 0 11,702 28,189 12,100 10,662 5,049 4,185 692 134 61,011 12,471 4,199 14,493 2,071 680 0 0 33,915 29,763 10,362 10,253 5,877 1,790 266 62 58,372 26,909 9,907 6,473 2,391 744 105 28 46,558 28,285 11,786 8,716 2,244 1,028 190 30 52,279 19,315 4,350 6,925 1,957 1,438 308 63 34,356 19,759 3,533 4,872 1,023 654 70 15 29,926 21,563 7,694 5,665 1,463 780 135 45 37,346 Influence of Particles on the Ozone Treatment – Degradation of the Target Pollutants The degradation of the target TrOCs during ozonation was determined for filtered and non-filtered effluent (NFE). Carbamazepine, diclofenac and sulfametaxazole, with ozone second-order rate constants higher than 104 M-1s-1 (fast-reacting), were removed below the limit of detection (~100 ng/L) at Zspec of 0.69 mg O3/mg DOC0. Other studies also showed that compounds with kO3>104 M-1s-1 were eliminated to concentrations below the detection limit (~1-100 ng/L) for an ozone dose of 0.47 mg O3/mg DOC0 (Hollender et al. 2009). Figure 1 presents the removal of the three slow-reacting iodinated contrast media by ozone, at 0.69 and 0.93 mg O3/mg DOC0, for different pore size filtered effluents. It can be seen that filtration with decreased pore-size filters improved the ozone removal rate of the iodinated contrast media with similar trend for both ozone dosages. For example, iopamidol was eliminated by 71% at NFE compared to 80% at 11 µm and 93% at 0.45 µm filtered effluents in 0.93 mg O3/mg DOC0. With any filter pore size, increasing Zspec from 0.69 to 0.93 mg O3/mg DOC0 enhanced iodinated contrast media removal. Furthermore, experiments with high dosage (0.93 mg O3/mg DOC0) of ozone showed smaller error values in comparison to low dosage (0.69 mg O3/mg DOC0). From studies that investigated the fate of the selected TrOCs during activated sludge treatment, it was clear that sorption onto suspended solids is 71 ECD size (µm) 2-3 3-5 5-10 10-15 15-30 30-50 50+ Total not a relevant process (Ternes et al. 2004). The improved removal with decreasing pore-size filtration indicates that particles interfere with ozone treatment, probably by competing with dissolved matter over ozone. The three iodinated contrast media have negligible rate constants for the reaction with ozone (table 1) and are mainly oxidized by hydroxyl radicals. Therefore, it can be concluded that hydroxyl radical exposure increases with filtration, most likely since ozone reaction with dissolved organic matter generates more hydroxyl radicals (higher quantum yield) than ozone reaction with particles (on surface or via ozone intra-particle diffusion). Possibly, scavenging effect of fast-reacting hydroxyl radicals by the particles through particle-surface reactions may also influence the removal of micropollutants. 72 Figure 1. Removal of iodinated contrast media by 0.69 and 0.93 mg O3/mg DOC0 ozonation in different pore size filtered effluents 73 Influence of Particles on the Ozone Treatment – Water Quality Parameters and Ozone Depletion The effect of particles on water quality parameters during ozonation was tested. The effluent had a pH of ~7.5 and a DOC concentration of ~9 mg/L which was not influenced by filtration or by ozone treatment. However, DOC changes might not be detected due to low sensitivity of the TOC analyzer or due to formation of smaller DOC organic fractions (not mineralized). Ozone depletion in secondary effluents can be described as a reaction in two phases; a very fast depletion from 0-20 seconds, operationally defined as instantaneous ozone demand (IOD), followed by first order ozone depletion (Buffle et al. 2006; Nöthe et al. 2009). Figure 2 presents the decrease in UVA following ozonation at 0.93 mg O3/mg DOC0, for the different pore-size filtration as well as IOD. Decrease in effluent UVA following ozonation is expected due to ozone attack on conjugated systems (Wert et al. 2009). Direct correlation can be seen between particle filtration and UVA decrease by ozone; where, increasing particle fractions removal (by decreasing filter pore-size) increases UVA removal. It can also be seen from figure 2 that increasing particle fractions removal decreases IOD. This result fits well with the UVA and TrOCs removal. With decreasing pore size filtration, particle load decreases, less ozone consumed by particles and ozone reacts more with conjugated system and TrOCs. This strengthens the conclusion that particles in wastewater indeed compete with dissolved matter over applied ozone (or react with generated hydroxyl radicals), thus influencing the ozone treatment efficacy. Table 2 shows water quality parameters such as filtered biochemical oxygen demand (BODf filtered at 0.45 µm), filtered chemical oxygen demand (CODf -filtered at 0.45 µm), total suspended solids (TSS), and volatile suspended solids (VSS). BOD increased after ozonation as well as TSS and VSS. It was shown before that following ozonation, effluent's BOD and the BOD/CODf ratio increases, which implies of an increase in the effluent's biodegradability, which is highly desirable if ozonation is followed by a further biological treatment (Lester et al. 2013). Change in particle count after ozonation as a function of pore size filtration will be discussed in the next section. 70 Table 3. water quality parameters before and after moderate ozonation, in NFE and 1.2 µm filtered samples VSS TSS COD BOD O3 treatment 0.5 3.3 32 3.0 Before ozonation 1.4 3.5 32 6.1 0.69 mg O3/mg DOC0 0.5 0.3 29 1.5 Before ozonation 1.1 1.4 26 4.5 0.69 mg O3/mg DOC0 NFE 1.2 µm Filtration NFE 11u 6u 1.2uNFE 11u 6u 1.2u 0.45u, 6 µm, 0.45u, 1.2 µm, 51.04% NFE 11u 6u 1.2u 50.72% 0.45u, 11 µm, NFE 11u 6u 1.2u 48.35% 0.45u, NFE, 45.75% IOD (mg/L) % Removal UVA NFE 11 µm 6 µm NFE 11 µm 6 µm NFE 11 µm 6 µm NFE 11 µm NFE6 11 µmµm 6 µm 1.2 µm 0.45 µm1.2 , µm 0.45 µm1.2 , µm 0.45 µmNFE 1.2 , µm 1.21.2u µm 0.45 , µm , 11u0.45 6u 1, 7.0623809522, 6.8076190483, 6.612380952 4, 6.464761905 5, µm 5.990952381 0.45u, 0.45 , 56.11% Figure 2. UVA removal (red bars) and IOD (blue markers) with standard deviations by 0.93 mg O3/mg DOC0 ozonation in different pore size filtered effluents 74 Influence of Ozone Treatment on the Particles – Size Distribution Non-filtered secondary effluent were ozonated with a dose of 0.93 mg O3/mg DOC0. Figure 3 shows the PSD before and after ozonation for NFE samples during summer and winter experiments. The change in concentration (C) for each size range due to ozonation was calculated. Image analyzer is not capable of analyzing particles smaller than 2 µm (e.g. colloids, oxides, viruses, and bacteria), but can efficiently analyze larger particles (e.g. bacteria, bioflocs, organic debris, algae, cysts and silt). After ozonation, the total particle concentration in all of examined samples decreased. The most significant particle concentration decrease (~80%) was observed for particles with ECD range of 5-15 µm (medium size particles). A lower decrease was observed for 15-30 and 2-5 µm particle ranges. Similar trends were observed in summer and winter. Since O3 can act as a flocculent-aid (Jekel 1994) by destabilizing particles, there may be formation of larger particles (30–50+ µm) during ozonation process, as observed herein. The population balance for specific sized particles in a suspension can be described by integration of few mechanisms. Particles of each size class can be formed by flocculation of smaller particles and lost by flocculating to larger particles. Moreover, particles of each size class can be also formed by larger particles breakage and lost by breaking into smaller particles making them more mineralized and easier to remove or even dissolute into dissolved constituents (Zhu et al. 2008; Li et al. 2009). Surface dissociation, i.e., departure of ions from the surface and their transfer to a bulk electrolyte solution (Ofir et al. 2007) was not detected in DOC measurements possibly as the TOC analyzer is not sensitive to these changes. This PSD analysis shows that both flocculation and particle-breakage can co-exist simultaneously and their influence may be determined by the particles distribution among other factors. 75 Winter BO, AO, 2.75, 12,471 11,748 Winter BO Winter AO Summer BO Summer AO Particle Concentraion (#/ml) Summer BO, 2.75, 5,233 Summer AO, 2.75, 4,615 Winter BO, 7.75, 4,644 Winter BO, 8.75, 3,782 Winter Winter BO, 6.75, 2,311BO, 9.75, 2,326 Winter Winter BO, BO, 2,100 Winter AO, 4.75, 4.75, 2,041 Winter AO, 3.75, 3.75, 1,9962,100 Winter BO, 5.75, 1,429 Winter 5.75, Summer BO,AO, 4.75, 9921,077 Summer BO, 7.75, 884BO, 10.75, 879 Summer BO, 3.75, 879 Winter Summer AO, 4.75, 7996.75, Summer AO, 3.75, 754 Winter AO, 711 Summer BO,535 8.75, 654 617 Summer BO, 5.75, 653 Summer BO, 6.75, 622 Summer BO, 9.75, Winter AO, 7.75, Winter BO, 11.75, 478 Summer AO, 5.75, 415 Winter AO, 8.75, 410 Summer BO, 10.75, 393 Summer AO, 8.75, 353 Winter AO, 9.75, 352 Summer AO, 7.75, 304 Summer AO, 9.75, 283 Summer AO, 6.75, 279 Winter BO, 12.75, 276 Winter AO, 10.75, 264 Summer BO, 11.75, 235 Winter BO, 13.75, 225 Winter BO, 14.75, 213 Winter AO, 11.75, 186 Winter BO, 176 Winter BO, 149 Winter AO, 12.75, 142 Summer BO, 12.75, 127 Summer AO, 10.75, 124 Winter AO, 112 Winter BO, 100 Winter AO, 92 BO, 18.75, 87 Winter AO, 15.75, 87 Summer AO, 11.75, 8613.75, BO, 85 BO, 13.75, 74 Winter AO, 16.75, 71 AO, 65 Summer AO, 12.75, 5914.75, BO, 14.75, 58 Summer AO, 5215.75, Winter AO, 49 BO, 15.75, 45 Winter AO, 4236 40 BO, 16.75, 37 BO, BO, 19.75, Summer AO, 3616.75, BO, 17.75, 36 35 21.75, 32 Winter AO, 30 BO, 18.75, 30 Summer Summer AO, AO, 2817.75, 27 Summer AO, 2619.75, 26 22.75, 24 23 22 24.75, 21 23.75, 20 Summer AO, 1820.75, 18 Summer AO, 25.75, 27.75, 17 37 17 26.75, 16 15 28.75, 14 13 29.75, 11 10 30.75, 24.75, 25.75, 26.75, 9 23.75, 27.75, 8 31.75, 32.75, 22.75, 7 33.75, 29.75, 6 38.75, 28.75, 34.75, 35.75, 5 36.75, 39.75, 4 37.75, 40.75, 3 52.75, Winter AO, 41.75, 42.75, 2 50.75, 60.75, 80.75, 1 Summer Summer BO, 100.75, 101.75, 102.75, 103.75, 104.75, 105.75, 106.75, 107.75, 108.75, 109.75, 110.75, 43.75, 44.75, 45.75, 46.75, 47.75, 48.75, 49.75, 51.75, 53.75, 54.75, 55.75, 56.75, 57.75, 58.75, 59.75, 61.75, 62.75, 63.75, 64.75, 65.75, 66.75, 67.75, 68.75, 69.75, 70.75, 71.75, 72.75, 73.75, 74.75, 75.75, 76.75, 77.75, 78.75, 79.75, 81.75, 82.75, 83.75, 84.75, 85.75, 86.75, 87.75, 88.75, 89.75, 90.75, 91.75, 92.75, 93.75, 94.75, 95.75, 96.75, 97.75, 98.75, 99.75, 0 ECD Size (µm) Figure 3. Particle concentration distribution (PSD) during winter and summer experiments before ozonation (BO) and after 0.93 mg O3/mg DOC0 ozonation (AO) 77 Influence of Ozone Treatment on the Particles – effect of pre-filtration & ozone dose Filtered and non-filtered secondary effluent were ozonated with three different Zspec of 0.17, 0.69, and 0.93 mg O3/mg DOC0. The PSD analysis before and after ozonation is shown in figure 4. When effluent was pre-filtered with 0.45 and 1.2 µm filtration, particle concentration increased for any ECD size as function of Zspec. However, for 11 µm filtered effluent and NFE, particle concentration increased only at 0.17 mg O3/mg DOC0. In 0.69 mg O3/mg DOC0 ozonation, a decrease was observed in particle concentration and for 0.93 mg O3/mg DOC0 ozonation, different effects seen in 11 µm filtered effluent and NFE. One possible reason might be that in effluents with low particle concentration, the lower probability of ozoneparticle interaction results in efficient coagulation effect of dissolved matter while for nonfiltered samples, particle brakeage is the main mechanism. Another reason might be difference is particle properties such as morphology, hydrophobic / hydrophilic nature and molecular weight. It was reported before that hydrophobic NOM with intermediate molecular weight increases at lower ozone dosage, while at higher ozone dosages, NOM becomes more hydrophilic and its molecular weight becomes smaller, decreasing NOM removal (Yan et al. 2007). Colloidal stability is often measured by zeta potential of the particles. However, the average zeta potential was difficult to determine for a highly dispersed system containing an unknown number of colloidal particles (Farvardin & Collins 1989). 75 Figure 4. PSD analysis before and after 0.17, 0.69 and 0.93 mg O3/mg DOC0 ozonation in different pore size filtered effluents 79 Potential application Pre-filtration before wastewater ozonation is a possible application for enhancing the removal of micropollutants. This was further examined by using an on-site deep-bed filtration field pilot in the Shafdan WWTP. Particle analysis revealed similar filtration properties for both deep-bed field filtration and 11 µm laboratory filtration. The removal of Carbamazepine (CBZ, an antiepileptic drug) by ozonation from the Shafdan non-filtered and deep-bed filtered effluent was examined. CBZ concentration in both effluents, as determined by an SPE-HPLCMS/MS method, was approximately 850 ng/L. CBZ decomposition by ozonation process is presented in figure 5 for secondary and deep-bed filtered effluents. CBZ was efficiently removed in both effluents, with improved removal achieved in the deep-bed filtered effluent. Thus, pre-filtration may promote micropollutant removal via ozonation, by removing particulate matter. 1 C/Co 0.8 Secondary effluent Co = 840 ± 200 ng/L Deep-bed filtered effluent Co = 850 ± 200 ng/L 0.6 0.4 Not detected 0.2 0 0 0.3 0.5 Specific Ozone Consumption (mg O3 / mg DOC0) 0 0.3 0.5 Specific Ozone Consumption (mg O3 / mg DOC0) Figure 5. CBZ decomposition by ozonation in different specific ozone consumption for secondary (left) and deep-bed-filtered effluents OH-radical exposure was back-calculated using the oxidation of ozone resist TrOCs (iodinated contrast media). OH radical exposure can be described as a linear function of the ozone consumption (figure 6). This function is characterized by the slope indicating the efficiency of OH-radical formation and lag ozone consumption without significant radical formation, which can be calculated from intercept and slope as reported before (Hübner et al. 2013). It can be seen that the radical exposure for pre-filtered effluents is higher. 55 0.04 ʃ[OH]dt (109 Ms) 500 0.03 OH exposure y = 0.0031x - 0.0017 R² = 0.8764 ASF 0.02 y = 0.0028x - 0.007 R² = 0.7763 0.01 0 0 2 4 O3 dose6 [mg/L] 8 10 12 Figure 6. OH-radical exposure determination using 3 iodinated contrast media as trace substances in Shafdan secondary and deep-bed-filtered effluents (ASF) In figure 7, relative residual concentration of fast and moderate reacting TrOCs is presented as function of rate constants. in low applied dosages, a difference could be seen between the different TrOCs relative residual concentration. C/Co 0.8 Secondary effluents 0.6 0.4 0.2 0.0 0 0.5 30 ASF effluents 0.4 C/Co 10 20 kO3 (105 1/Ms) 0.3 0.2 0.1 0.0 0 10 20 kO3 (105 1/Ms) 51 30 Figure 7. Relative residual concentration of fast and moderate reacting TrOCs as function of rate constants with ozone in Shafdan secondary and deep-bed-filtered effluents (ASF) The attempt to calculate ozone exposure from reduction of different TrOCs is illustrated in figure 8 as a function of second order rate constant with ozone (kO3). Data show a strong correlation of exposure with rate constants (high data points from benzafibrate are not shown). These results confirm that ozone exposures cannot be calculated from tracer substances under the examined conditions. Thus, simple modeling of compound removal calculated ozone exposure (10-5 Ms) from second order rate kinetics is not possible. 5 10 mg/L 500 4 3 8.57 mg/L 500 2 1 0 0 10 20 kO3 (105 1/Ms) 30 Figure 8. Calculation of ozone exposure using fast reacting TrOCs as function of rate constant with ozone in Shafdan secondary effluents Conclusions The present study demonstrates the influence of wastewater effluent particles with different size distribution on ozonation and the influence of ozone on the particles’ concentration. Particle analysis showed a decrease in particles' count after ozonation for non-filtered samples, which may indicate both flocculation and decomposition of organic particles into smaller particles (not detected by the analyzer) or to dissolved organic matter. Medium size particles concentration was mainly affected by the ozone treatment. Filtered samples showed an increase in particle concentration after ozonation in at ECD sizes. UVA absorbance and micropollutant concentration decreased more substantially after ozonation as function of pore- 52 size filtration, with increased removal for the 0.45 µm filtered effluent. It can be concluded that the mutuality between particle size distribution and ozone treatment efficiency exist. 53 References APHA, A.P.H.A., AWWA, A.W.W.A. & WEF, W.E.F., 2005. Standard Methods for the Examination of Water and Wastewater. In E. W. Rice et al., eds. Washington, D.C: APHA, American Public Health Association AWWA, American Water Works Association WEF, Water Environment Federation. Bahr, C. et al., 2007. SUVA as control parameter for the effective ozonation of organic pollutants in secondary effluent. Water Science & Technology, 55(12), p.267. Available at: http://www.iwaponline.com/wst/05512/wst055120267.htm [Accessed August 26, 2013]. Baus, C., Sacher, F. & Brauch, D., 2004. Oxidation of iodinated X-ray contrast media in water samples – results of laboratory scale experiments. Proceedings to the 70th annual conference of the Water Chemical Society, Division of the German Chemical Society, Bad Saarow, pp.1–4. Bourgine, F.P. et al., 1998. The Effect of Ozonation on Particle Removal in Drinking-Water Treatment. JOURNAL OF THE CHARTERED INSTITUTION OF WATER AND ENVIRONMENTAL MANAGEMENT, 12(3), pp.170–174. Buffle, M.-O. et al., 2006. Measurement of the initial phase of ozone decomposition in water and wastewater by means of a continuous quench-flow system: application to disinfection and pharmaceutical oxidation. Water research, 40(9), pp.1884–94. Available at: http://www.ncbi.nlm.nih.gov/pubmed/16624368 [Accessed June 14, 2013]. Chandrakanth, M.S. & Amy, G.L., 1996. Effects of Ozone on the Colloidal Stability and Aggregation of Particles Coated with Natural Organic Matter. Environmental Science & Technology, 30(2), pp.431–443. Available at: http://pubs.acs.org/doi/abs/10.1021/es9500567. Chandrakanth, M.S., Honeyman, B.D. & Amy, G.L., 1996. Modeling the interactions between ozone, natural organic matter, and particles in water treatment. COLLOIDS AND SURFACES APHYSICOCHEMICAL AND ENGINEERING ASPECTS, 107(1), pp.321–342. Currie, M. et al., 2003. The Effect of Bicarbonate on Ozone-Enhanced Particle Removal in Water Treatment. Ozone: Science & Engineering, 25(4), pp.285–293. Available at: http://www.tandfonline.com/doi/abs/10.1080/01919510390481603 [Accessed August 26, 2013]. Dodd, M.C., Buffle, marc olivier & von Gunten, U., 2006. Oxidation of Antibacterial Molecules by Aqueous Ozone : Moiety-Specific Reaction Kinetics and Application to Ozone-Based Wastewater Treatment. Environmental science & technology, 40(6), pp.1969–1977. Farvardin, M.R. & Collins, A.G., 1989. Preozonation as an aid in the coagulation of humic substances – optimum preozonation dose. Water research, 23(3), pp.307–316. Genz, C. et al., 2011. The effect of pre-ozonation and subsequent coagulation on the filtration of WWTP effluent with low-pressure membranes. Water Science & Technology, 64(6), p.1270. Available at: http://www.iwaponline.com/wst/06406/wst064061270.htm [Accessed September 9, 2013]. Georgeson, D.L. & Karimi, A.A., 1988. Water quality improvements with the use of ozone at the los angeles water treatment plant.pdf. Ozone: Science & Engineering, 10(3), pp.255–76. Grosso, D. & Weber, W.J., 1988. Ozone-Induced Particle Destabilization.pdf. american water works association, 80(8), pp.73–81. Holbrook, R.D., Love, N.G. & Novak, J.T., 2004. Sorption of 17beta-estradiol and 17alphaethinylestradiol by colloidal organic carbon derived from biological wastewater treatment 50 systems. Environmental science & technology, 38(12), pp.3322–9. Available at: http://www.ncbi.nlm.nih.gov/pubmed/15260331. Hollender, J. et al., 2009. Elimination of organic micropollutants in a municipal wastewater treatment plant upgraded with a full-scale post-ozonation followed by sand filtration. Environmental science & technology, 43(20), pp.7862–9. Available at: http://www.ncbi.nlm.nih.gov/pubmed/19921906. Huber, M.M. et al., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environmental science & technology, 37(5), pp.1016–24. Available at: http://www.ncbi.nlm.nih.gov/pubmed/12666935. Huber, M.M. et al., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environmental science & technology, 39(11), pp.4290–9. Available at: http://www.ncbi.nlm.nih.gov/pubmed/15984812. Hübner, U., Keller, S. & Jekel, M., 2013. Evaluation of the prediction of trace organic compound removal during ozonation of secondary effluents using tracer substances and second order rate kinetics. Water research, pp.1–8. Jasim, S.Y. et al., 2008. The Effect of Ozone on Cold Water Coagulation. Ozone: Science & Engineering, 30(1), pp.27–33. Available at: http://www.tandfonline.com/doi/abs/10.1080/01919510701753366 [Accessed August 26, 2013]. Jekel, M.R., 1994. Flocculation Effects of Ozone. Ozone: Science & Engineering, 16(1), pp.55–66. Available at: http://www.tandfonline.com/doi/abs/10.1080/01919519408552380. Jekel, M.R., 1983. The benefits of ozone treatment prior to fllocculation process. Ozone: Science & Engineering, 5, pp.21–35. Lester, Y. et al., 2013. Treating wastewater from a pharmaceutical formulation facility by biological process and ozone. Water research, 47(13), pp.4349–56. Available at: http://www.ncbi.nlm.nih.gov/pubmed/23764586 [Accessed August 26, 2013]. Li, T. et al., 2009. Impact of preozonation on the performance of coagulated flocs. Chemosphere, 75(2), pp.187–92. Available at: http://www.ncbi.nlm.nih.gov/pubmed/19157497 [Accessed September 1, 2013]. Mamane, H., Kohn, C. & Adin, a., 2008. Characterizing Shape of Effluent Particles by Image Analysis. Separation Science and Technology, 43(7), pp.1737–1753. Available at: http://www.tandfonline.com/doi/abs/10.1080/01496390801974662 [Accessed September 10, 2013]. Nöthe, T., Fahlenkamp, H. & von Sonntag, C., 2009. Ozonation of wastewater: rate of ozone consumption and hydroxyl radical yield. Environmental science & technology, 43(15), pp.5990– 5. Available at: http://www.ncbi.nlm.nih.gov/pubmed/19731708. Ofir, E., Oren, Y. & Adin, a., 2007. Electroflocculation: the effect of zeta-potential on particle size. Desalination, 204(1-3), pp.33–38. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0011916406013038 [Accessed September 10, 2013]. Paraskeva, P. & Graham, N.J.D., 2002. Ozonation of municipal wastewater effluents. Water environment research : a research publication of the Water Environment Federation, 74(6), pp.569–81. Available at: http://www.ncbi.nlm.nih.gov/pubmed/12540098. Sein, myint et al., 2008. Oxidation of Diclofenac with Ozone in Aqueous Solution. Environmental science & technology, 42(17), pp.6656–6662. 54 Staehelin, J. & Hoigne, J., 1985. Decomposition of ozone in water in the presence of organic solutes acting as promoters and inhibitors of radical chain reactions. Environmental science & technology, 19(12), pp.1206–13. Available at: http://www.ncbi.nlm.nih.gov/pubmed/22280139. Ternes, T. a et al., 2003. Ozonation: a tool for removal of pharmaceuticals, contrast media and musk fragrances from wastewater? Water research, 37(8), pp.1976–82. Available at: http://www.ncbi.nlm.nih.gov/pubmed/12697241. Wert, E.C., Rosario-Ortiz, F.L. & Snyder, S. a, 2009. Effect of ozone exposure on the oxidation of trace organic contaminants in wastewater. Water research, 43(4), pp.1005–14. Available at: http://www.ncbi.nlm.nih.gov/pubmed/19135696 [Accessed August 8, 2013]. Yan, M. et al., 2007. Effect of pre-ozonation on optimized coagulation of a typical North-China source water. Chemosphere, 69(11), pp.1695–702. Available at: http://www.ncbi.nlm.nih.gov/pubmed/17644153 [Accessed August 26, 2013]. Zhu, H.T., Wen, X.H. & Huang, X., 2008. Pre-ozonation for dead-end microfiltration of the secondary effluent: suspended particles and membrane fouling. Desalination, 231(1-3), pp.166–174. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0011916408003585. 55