ARTICLE IN PRESS

Water Research 37 (2003) 4573–4586

Disinfection efficiency of peracetic acid, UVand after enhanced primary treatment of municipal wastewater Ronald Gehra,*, Monika Wagnera, Priya Veerasubramaniana, Pierre Paymentb

a Department of Civil Engineering, McGill University, Montreal, Canada H3A 2K6 b INRS-Institut Armand-Frappier, UniversiteduQu! ebec,! Laval, Canada H7V 1B7

Received 6 August 2002; received in revised form 3 April 2003; accepted 17 June 2003

Abstract

The City of Montreal Wastewater Treatment Plant uses enhanced physicochemical processes (ferric and/or alum coagulation) for suspended solids and phosphorus removal. The objective of this study was to assess the ability of peracetic acid (PAA), UV, or ozone to inactivate the indicator organisms fecal coliforms, Enterococci, MS-2 coliphage, or Clostridium perfringens in the effluent from this plant. PAA doses to reach the target fecal coliform level of 9000 CFU/100 mL exceeded 6 mg/L; similar results were obtained for enterococci, and no inactivation of Clostridium perfringens was observed. However a 1-log reduction of MS-2 occurred at PAA doses of 1.5 mg/L and higher. It was expected that this effluent would have a high ozone demand, and would require relatively high UVfluences, because of relatively high effluent COD, iron and suspended solids concentrations, and low UVtransmittance. This was confirmed herein. For UV,the inactivation curve for fecal coliforms showed the typical two-stage shape, with the target of 1000 CFU/100 mL (to account for photoreactivation) occurring in the asymptote zone at fluences >20 mJ/cm2. In contrast, inactivation curves for MS-2 and Clostridium perfringens were linear. Clostridium perfringens was the most resistant organism. For ozone, inactivation was already observed before any residuals could be measured. The transferred ozone doses to reach target fecal coliform levels (B2-log reduction) were 30–50 mg/L. MS-2 was less resistant, but Clostridium perfringens was more resistant than fecal coliforms. The different behaviour of the four indicator organisms studied, depending on the , suggests that a single indicator organism might not be appropriate. The required dose of any of the is unlikely to be economically viable, and upstream changes to the plant will be needed. r 2003 Elsevier Ltd. All rights reserved.

Keywords: Enhanced primary treatment; Disinfection; Peracetic acid; UV; Ozone

1. Introduction as a polyelectrolyte to precipitate phosphorus and improve solids settling, and sedimentation prior to The City of Montreal Wastewater Treatment Plant discharge into the St. Lawrence River. At present there (CMWTP) uses physicochemical processes to treat up to is no disinfection. Since the use of chlorine for waste- 7.6 106 m3/d of combined domestic and industrial water disinfection has been banned in the Province of wastewater. These processes include screening, grit Quebec, the City of Montreal is exploring alternative removal, addition of ferric chloride and/or alum as well disinfectants in order to produce water which could be suitable for contact aquatic sports, and as a raw potable *Corresponding author. Tel.: +1-514-398-6861; fax: +1- water source to communities downstream. The dis- 514-398-7361. charge permit for this wastewater treatment plant would E-mail address: [email protected] (R. Gehr). allow for an effluent containing 9000 CFU (colony

0043-1354/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0043-1354(03)00394-4 ARTICLE IN PRESS 4574 R. Gehr et al. / Water Research 37 (2003) 4573–4586 forming units) of fecal coliforms (FC) per 100 mL, and disinfection in a synergistic fashion (although Lubello 1000 CFU/100 mL to allow for photoreactivation if et al. [10], dispute this), and in the long term because it ultraviolet radiation (UV) is used [1]. However concerns persists longer than PAA. have been raised regarding the adequacy of indicator The disinfecting mechanism of PAA is still subject to organisms such as FC to predict the performance of some speculation. Lefevre et al. [11] and Liberti et al. [9] various disinfection processes against pathogens. Ac- suggested that the disinfectant property is due to the cordingly, the objectives of this study were (a) to assess release of ‘‘active’’ , which in turn disrupts the doses required for three disinfection processes— sulfhydryl (–SH) and sulfur (S–S) bonds within enzymes peracetic acid (PAA), UVor ozone—to reach the contained in the cell membrane. Thus transport across target FC standard, and (b) to compare the responses the cell membrane is affected, which impedes cellular of three other indicators—Enterococci (EC), Clostridium activity. perfringens (CP) and MS-2 coliphage—to these A second suggested disinfectant mechanism is the disinfectants. The data will be used to determine the release of hydroxyl radicals [10]. By varying the level of disinfection required at this wastewater treat- proportion of PAA to H2O2, Lubello et al. [10] also ment plant to attain a level of risk (using a model determined that it was the PAA, and not the H2O2, proposed by Haas et al. [2]) that is acceptable to public which was responsible for the biocidal action. health and to the population, and is economically Gehr et al. [8] presented results of batch screening feasible. The risks examined will be those associated tests where PAA was assessed as a disinfectant for with recreational activities during which direct and physicochemical as well as biological effluents. For a indirect contact occurs with river water impacted by this target FC level of 1000 CFU/100 mL, and contact times effluent. of 30–120 min, PAA doses for the physicochemical effluents ranged between 2 mg/L and greater than 6 mg/L, whereas for biological effluents lower doses of 2. Background 0.6–4 mg/L were required. Another recent study [10] assessed the possibility of using PAA in conjunction 2.1. Peracetic acid (PAA) with UVas a type of advanced oxidation process. A UV fluence of 120 mJ/cm2 applied simultaneously with a Of the three disinfectants being studied, PAA is PAA dose of 8 mg/L (30 min contact time) was able to the newest alternative for applications in North Amer- achieve over 4-log reduction of total coliforms, or less ica, though it has been used in Europe for waste- than 2 CFU total coliforms/100 mL, which is Italy’s water disinfection for many years. Interest in the use standard for unrestricted wastewater reuse in agricul- of PAA as a disinfectant for wastewaters began in ture. Such levels of inactivation were not possible with the late 1980s with publications by Baldry and cow- PAA or UVused separately. orkers [3,4]. Grantham [5] presented results from PAA reacts with organic matter in the sewage. If many internal studies of the National River Authority there is little organic matter, the disinfection reaction (NRA) of England and Wales. He pointed out that will be fast, and the additional disinfection after since PAA is an equilibrium mixture, the PAA dose 30 min contact will be insignificant. However, at stated in terms of the initial concentration of the high concentration levels of organic matter, dis- mixture will actually underestimate the dose of infection could also occur after this time, provided PAA delivered. The EPA [6] and Atasi et al. [7] have that the initial PAA dose was high enough to satisfy suggested that PAA could be particularly suited to the PAA demand of the sewage, and to establish a disinfection of combined sewer overflows (CSOs), and residual [8]. Gehr et al. [8] reported that PAA might be more As with all disinfectants, PAA has varying effective- appropriate as a disinfectant for biologically treated ness depending on the organism. Baldry et al. [4] found effluents, rather than those from physicochemical that E. coli and MS-2 coliphages had similar (low) effluents. resistance, but poliovirus, echovirus and coxsackievirus Commercially available PAA (also known as ethane- were considerably more resistant. Studies by Lazarova peroxoic acid or peroxyacetic acid) is available in a et al. [12] showed that different bacteriophages had quaternary equilibrium mixture containing , vastly different sensitivities to PAA: using the same hydrogen , peracetic acid and water, as shown wastewater effluent and 120 min contact time, 10 mg/L in the equation below: PAA was able to reduce f 174 bacteriophage by 7.5 logs, whereas 5 mg/L PAA was needed to reduce MS-2 CH CO H þ H O 2CH CO H þ H O: ð1Þ 3 2 2 2 3 3 2 coliphage by only 3.5 logs. Liberti et al. [9] determined The biocidal form is considered to be the undissociated that although PAA was effective against total coliforms, acid (i.e. CH3CO3H) which is predominant at pHo4:7 it was ineffective towards Giardia and Cryptosporidium [9]. However the H2O2 may also contribute directly to parasites. ARTICLE IN PRESS R. Gehr et al. / Water Research 37 (2003) 4573–4586 4575

2.2. Ultraviolet radiation (UV) used for wastewater disinfection due to operation and maintenance problems of first generation systems, as UVradiation is now the most common alternative to well as the high ozone demand of many effluents [22,23]. chlorination for wastewater disinfection in North Grantham [5] notes that water and wastewater utilities America [13]. UVlamps emit significant radiation in in England and Wales do not use ozone due to the high the range 240–260 nm, exactly that range over which costs involved. We have not been able to find examples nucleic acids (such as DNA and RNA) absorb energy. of current usage of ozone for large scale wastewater When exposed to UVlight, adjacent thymine bases on disinfection in North America, but several smaller plants the nucleic acid strands dimerize. Thus accurate exist in the US, Canada, Japan, Korea and Europe transcription of this DNA strand cannot occur, and [24,25]. the bacterial cell cannot divide [14]. Similar mechanisms The kinetics of ozone reactions have received the apply to inactivation of viruses. attention of many researchers. The kinetics are complex UVdose–response curves typically show an initial steep because of the different reaction rates with different decline (attributed to inactivation of the free-swimming chemicals in solution, speciation of the ozone and its organisms), followed by a second stage with a much decomposition products, and the interaction of these shallower slope. This latter stage is generally accepted to with the microorganisms. According to the literature, be due to the shielding effect of particulates [15], although ozone decomposes in three phases, although there is Blatchley et al. [16] suggest that it might also be due to disagreement as to the exact pathways [13]. The type and clumping of bacterial cells. If the target inactivation level kinetics of ozone decomposition has implications on the is in this region, much higher UVdoses—now referred to mechanism of disinfection. If ozone decomposition is as ‘‘fluences’’—will be required [17]. The critical particle slow, then chemicals (and microorganisms) will undergo size is in the range of 9–10 mm [15,18];belowthissize direct ozone attack. These reactions are selective and particles cannot shield or embed bacteria. Emerick et al. slow. On the other hand, if ozone decomposition is [15] claimed that above this size there is no difference in rapid, which occurs when the alkalinity is low, and/or the particles in terms of their ability to harbour coliform organics concentration is high, then oxidation will occur bacteria, but this result is disputed [19]. by means of the OH radical, which is very reactive and Early studies at the CMWTP [18] had established that non-selective. Studies with E. coli and G. muris have a UVfluence of 35 mJ/cm 2 would be required to achieve suggested that these organisms undergo greater inacti- a 3-log or greater reduction in FC counts. This would be vation when ozone residuals persist, rather than when sufficient to meet a target level of 2000 CFU/100 mL ozone is rapidly decomposed. before dilution and possible photoreactivation. At that Hermanowicz et al. [26] showed that the reaction rates time the plant was operating suboptimally; suspended for individual batch tests could be roughly described by solids (SS) was approximately 40 mg/L, unfiltered UV first-order kinetics, but that both the rate constant and transmittance at 254 nm was 40%, and Fe concentrations the reaction order decreased with increasing ozone dose were frequently above 1 mg/L. More recent collimated transferred. Rates of ozone consumption were much beam tests [8] determined that fluences of 8–12 mJ/cm2 higher in the continuous-flow experiments than in the could achieve target FC levels of 1000 CFU/100 mL for batch tests. For their batch tests, Oke et al. [27] mixed a SS of 20–30 mg/L, when the facility was operated with a saturated ozone solution with the test sample, and found mixture of ferric chloride and alum as coagulants. that first-order kinetics were appropriate for ‘‘clean’’ There have been many studies on the effect of UVon waters, but for natural lake waters there was a gradually different organisms (including Angehrn [14], Masschelein decreasing rate constant. Comparison of batch contact [20], and Oppenheimer et al. [21]). The typical indicator reactors (where the ozone gas is added at the start of the aerobic bacteria such as fecal coliforms show a relatively reaction) with continuous-flow reactors is made more low resistance (such as a 1-log inactivation requiring difficult because of the simultaneous increase of approximately 2.5–3 mJ/cm2), bacteriophages and many dissolved ozone and decrease of oxidizable species as viruses are slightly more resistant, and the anaerobic well as microorganisms. If Ct values are used, the spore-formers such as Clostridium perfringens are the assumption is that C will be constant throughout the most resistant (over 10 mJ/cm2 per log inactivation). contact time t, and that the kinetics (order and rate However none of the studies that we are aware of have constant) are also constant. In the present work, C was commented on whether the presence of particles may not constant, but the kinetics were assumed to be have different effects on different types of organisms. constant, and the Ct values were calculated as the area under the Ct curve. 2.3. Ozone (O3) There is a need to satisfy the ozone demand before disinfection can proceed. Ozone demand can be caused Although ozone has been a popular and successful by certain inorganics, organics, and suspended solids. disinfectant for drinking water, it has not been widely This may lead to unrealistically high doses, especially for ARTICLE IN PRESS 4576 R. Gehr et al. / Water Research 37 (2003) 4573–4586 the CMWTP, which uses only physicochemical treat- It is well accepted that ozone is effective against all ment [28]. However Gehr and Nicell [28], Lazarova et al. organisms likely to be encountered in wastewaters, [12] and Xu et al. [23] show that up to 3-logs of including viruses and protozoan cysts. Even organisms inactivation of E. coli or fecal coliforms could be resistant to chlorination, such as poliovirus Type 3, as obtained even before the ozone demand was met. This well as Cryptosporidium and Giardia protozoa, can be calls into question the concept of the product of residual inactivated by ozone at residual concentrations of 1 mg/ concentration and contact time (Ct) as a governing L or even less, and sufficient contact time (several parameter for wastewater disinfection performance. minutes) [12,13]. Xu et al. [23] found that at The actual mode of action of ozone disinfection is a transferred ozone dose (TOD) of 15.2 mg/L, the poorly understood, with various researchers suggesting order of resistance of the microorganisms tested was that ozone alters proteins and the unsaturated bonds of F-coliphage (>2.09 log reduction), FC and EC (2.48 fatty acids in the cell membrane, or that it affects cell log), and CP (0.31 log). To reach the WHO standard DNA. Hunt and Marin˜ as [29] showed that noticeable for irrigation ðFC ¼ 1000 CFU=100 mLÞ; TODs of changes in the interior of E. coli cells did not take place 2–15 mg/L were needed, depending on the quality of until most of the cells in the sample were non-viable. the wastewater. This confirms the hypothesis that in most cases inactivation was due to damage of the cell membrane, and that DNA damage might occur, but only if ozone 3. Materials and methods dosages were very high. Absi et al. [30] and Gehr and Nicell [28] reported on 3.1. Materials for disinfection continuous-flow pilot studies of ozone disinfection at the CMWTP. Disinfection could not be well correlated with Materials for disinfection and analysis of disinfectants wastewater quality parameters, although it was clear are shown in Table 1. that higher COD levels resulted in higher required ozone doses. The target FC level in that study was 5000 CFU/ 3.2. Analytical procedures for disinfection 100 mL. When FeCl3 was used as the upstream coagulant, the required ozone dose to achieve the target 3.2.1. PAA 90% of the time was 17 mg/L; when alum was used, a Residual concentrations of peroxycompounds (sum of limited number of tests indicated that this could be H2O2 and PAA) were measured in terms of the reduced to perhaps as low as 10 mg/L. absorbance of the radical cation of ABTS, formed when

Table 1 Materials for disinfection and analysis of disinfectants

Disinfectant Reagent/apparatus Source

PAA (12% w/w) Solvay Interox, Houston, TX Peracetic acid Horseradish peroxidase (EC 1.11.1.7, RZ 1.1) Sigma Chemical Co, St. Louis, MO Catalase (EC 1.11.1.6) Sigma Chemical Co, St. Louis, MO ABTS (98%) Sigma Chemical Co, St. Louis, MO Dibasic sodium phosphate Fisher Scientific Co., Montreal, QC Monobasic potassium phosphate Fisher Scientific Co., Montreal, QC Sodium thiosulfate Fisher Scientific Co., Montreal, QC Spectrophotometer Beckman DU-65

UVLow pressure mercury lamp collimator Radiometer International Light, Newsburyport, MA

Ozone OzoTitanTM ozone generator Hankin Ozone Systems Ltd., Scarborough, ON Oxygen feed gas MEGS Inc., Ville St. Laurent, QC Magnehelics differential pressure gauge Dwyer Instruments, Inc., Michigan City, IN Top-Trak digital gas flow meter Sierra Instruments Inc., Monterey, CA AFXTM model H1 high concentration UV-ozone analyser IN USA, Inc., Needham, MA Potassium iodide (ACS reagent) Fisher Scientific, Nepean, ON Concentrated sulfuric acid Anachemia Canada Inc., Montreal, QC Starch indicator (1%) Lab Chem Inc., Pittsburgh, PA Potassium indigo trisulfonate for indigo Sigma Chemical Co, St. Louis, MO Reagents I and II ARTICLE IN PRESS R. Gehr et al. / Water Research 37 (2003) 4573–4586 4577

H2O2 or PAA interact with ABTS and horseradish served as a guard detector. Ozonated gas leaving the peroxidase at pH 6.0 [31]. The colorimetric assay was ozone generator was bubbled at a flow rate of 0.46 L/ composed of 1.7 mM ABTS and 4.2 mg/L horseradish min through the gas washing bottles for a period of peroxidase (HRP) in 67 mM sodium phosphate buffer at 1 min. During this time period, the ozone concentration pH 6.0 [32]. When peroxycompounds (peracetic acid in the gas stream, as indicated by the UV-ozone and/or H2O2) are added to the assay solution, ABTS is analyser, as well as the gas flow rate were recorded. oxidized to its radical cation ðABTSþÞ under the Iodine formed only in the first gas washing bottle, since catalytic action of HRP. The amount of ABTSþ formed in the second gas washing bottle the KI trapping was recorded by measuring the absorbance at 405 nm, solution did not turn yellow. The KI trapping solution 6 min after sample addition. The absorbance was was immediately transferred into a broad-necked converted to peroxycompound concentrations using Erlenmeyer flask and acidified with 6.25 mL of 4.5 M the calibration line shown in the following equation: sulfuric acid. Following the addition of 5 mL of 0.1 N Na S O solution, the sample was titrated with 0.005 N Total peroxy-compounds concentration 2 2 3 Na S O . After titration to a pale yellow colour, 0.5 mL d 2 2 3 ¼ 0:0276 Absorbance405: ð2Þ of starch indicator solution was added and titration was All ABTS-HRP assays were performed in duplicate. The continued until the disappearance of the violet colour. recovery of peroxycompounds added to the wastewater The ozone production rate (OPR) was calculated was better than 75% for doses>3.0 mg/L PAA and according to 50–80% (depending on the sample) for doses of 3.0 and 1.5 mg/L PAA. ðVtitrant 1 N1 þ Vtitrant 2 N2Þ24 OPR ¼ ; ð3Þ t 3.2.2. Ozone Determination of the ozone production rate of the ozone where OPR is the ozone production rate (mg/min); generator: The ozone production rate of the ozone Vtitrant 1 is the volume of the 0.1 N Na2S2O3 solution generator was determined using an iodometric method added ðVtitrant 1¼ 5mLÞ; N1 is the normality of titrant 1 [33] as well as the readings of the UV-ozone analyser and ðN1¼ 0:1NÞ; Vtitrant 2 is the volume of the 0.005 N the gas flow meter. For the iodometric method Na2S2O3 solution added; N2 is the normality of titrant (considered herein as the benchmark) 250 mL of KI 2 ðN2¼ 0:005 NÞ; t is the ozonation time (min). trapping solution (containing 20 g/L KI, 7.3 g/L Na2- The ozone production rate at the specified setting HPO4 and 3.5 g/L KH2PO4) were added to a gas (15 dcVand a flow rate of 0.46 L/min) was on average washing bottle equipped with a fritted-glass diffuser 15.070.6 mg/min. (see Fig. 1). This bottle was connected to the outlet of The ozone production rate obtained using the the UV-ozone analyser. A second identical gas washing iodometric method was compared to the one based on bottle was connected to the outlet of the first bottle and the readings of the UV-ozone analyser and the gas flow

O2 Ozone generator P F O3 UV-ozone analyzer

Ozone reactor bottle

Magnetic stirrer Ozone trapping bottles P = pressure gauge F = flow meter Fig. 1. Diagram of ozone contacting apparatus. ARTICLE IN PRESS 4578 R. Gehr et al. / Water Research 37 (2003) 4573–4586 meter, which was calculated according to expected ozone concentration) were withdrawn through the septum at regular intervals, mixed with the appro- Pstandard OPR ¼ CO3 F; ð4Þ priate ozone reagent and the absorbances were imme- Pactual diately recorded. The concentration of dissolved ozone where OPR is the ozone production rate (mg/min); CO3 in an ozonated wastewater sample CO3 was calculated is the ozone concentration in the gas stream according to according to the UV-ozone analyser ðg=m3¼ mg=LÞ; P is the standard V ðA A Þ standard pressure used by the UV-ozone analyser to total sample blank CO3 ¼ ; ð6Þ f c Vsample calculate the ozone concentration ðPstandard¼ 101:3 kPaÞ; P is the actual pressure in the gas stream (kPa); F is actual where Asample is the absorbance of the indigo assay the gas flow rate (L/min). solution containing the ozonated wastewater sample at The differential pressure in the ozone reactor 600 nm; Ablank is the absorbance of the indigo assay ðPactual PstandardÞ was 3.2 kPa when one trapping bottle solution containing the wastewater before the start of was used and approximately 5.7 kPa when both trapping the ozonation at 600 nm; f is the proportionality bottles were connected in line. Thus, the value of the constant [ f ¼ 0:42 (cm mg/L)1]; l is the path length of pressure correction factor was 0.97 and 0.95, respec- the cell ðl¼ 5cmÞ; Vsample is the sample volume in the tively. Despite pressure correction, the ozone production indigo assay solution (mL); Vtotal is the total volume of rate based on the iodometric method yielded a value that the indigo assay solution ðVtotal¼ 14 mLÞ: was 10% lower than the one obtained based on the UV- ozone analyser and the gas flow rate. 3.3. Experimental procedure for disinfection Determination of the ozone mass flow rate leaving the reactor: During ozonation experiments the concentra- 3.3.1. PAA tion of ozone in the gas leaving the reactor bottle was Sample aliquots were added to 250 mL dark glass monitored using the UV-ozone analyser. These values bottles, and stirred using magnetic stirring bars during were corrected by a factor of 0.9 and multiplied by the disinfection. The experiment was started by adding the gas flow rate in order to obtain the mass flow rate of PAA disinfectant to yield PAA doses of 1.5, 3, 4.5 and ozone leaving the reactor over the course of the 6.0 mg/L. After 1 h incubation, samples were withdrawn ozonation experiment. from the bottles for residual peroxycompound measure- Determination of the ozone transfer rate and the ments and added to the ABTS-HRP assay solution. quantity of ozone transferred: The ozone transfer rate Immediately after this, catalase was added at 50 mg/L to into the ozonated sample was calculated according to quickly decompose the residual H2O2 following the Ozone transfer rate ðmg=L minÞ addition of sodium thiosulfate (Na S O ) at 100 mg/L to  2 2 3 eliminate residual PAA. Two control bottles were Pstandard OPR F CO3out 0:9 included in the experiment: one bottle was supplied Pactual ¼ ; ð5Þ with 6.0 mg/L PAA, but the disinfectant was quenched V reactor immediately by the addition of catalase and sodium where CO3out is the ozone concentration in the gas stream thiosulfate (quench control); the other bottle contained leaving the ozone reactor according to the UV-ozone the raw sample (0-sample). 3 analyser ðg=m ¼ mg=LÞ; Vreactor is the volume of ozonated sample in the ozone reactor bottle 3.3.2. UV ðVreactor¼ 2:3LÞ: The quantity of ozone transferred UVfluence–response curves were obtained according within a specified time period (in mg/L) was calculated to standard procedures for collimated beam tests (see, by multiplying the ozone transfer rate by the ozonation for example, Masschelein [20]). Ultraviolet light irradia- time. tion was carried out with a mercury low vapour pressure Determination of dissolved ozone concentrations (ozone lamp emitting light mainly at 253.7 nm, which was residuals)[indigo colorimetric method]: The determina- mounted over a collimating tube. The incident light tion of dissolved ozone concentrations was essentially intensity (I0) was recorded using a radiometer IL 1400A carried out according to Standard Method # 4500-ozone with ‘‘cosine diffuser lens’’ (International Light, New- [34]. Before the onset of ozonation 12.5 mL of sample buryport, MA). Fifty millilitres of wastewater sample were withdrawn through a septum installed into the side were poured into a crystallization dish containing a wall of the reactor bottle (see Fig. 1) using a syringe and magnetic stirring bar and placed on a magnetic stirrer mixed with 1.4 mL of indigo reagent I or reagent II under the collimated beam lamp. The irradiation time (depending on the ozone concentration). The absor- required to obtain the predetermined fluence was bances of these solutions were immediately recorded as calculated according to the absorbance blank values. After ozonation was initiated, samples (12.5 mL or less depending on the tir ¼ fluence=Iavg; ð7Þ ARTICLE IN PRESS R. Gehr et al. / Water Research 37 (2003) 4573–4586 4579 where tir is the irradiation time (s); fluence is the intended C. perfringens was prepared according to a method 2 UVfluence (mJ/cm ); Iavg is the average radiation described by Armon and Payment [36]. Briefly, it intensity inside an irradiated sample (mW/cm2). contained 71.14 g/L media mCP (Difco, QC), 0.4 g/L The average radiation intensity inside an irradiated d-cycloserine, 0.025 g/L polymyxin-B-sulphate, 0.06 g/L volume of a stirred wastewater medium was calculated indoxyl B-d-glucoside, 0.1 g/L phenolphthalene dipho- according to Morowitz [35]: sphate and 0.9 g/L ferric chloride (all from Sigma  1 edln 1=T Chemical Co, St. Louis, MO) in distilled water. Strains I ¼ I ; ð8Þ of E. coli F+ and MS-2 coliphage were obtained from avg 0 d ln 1=T Payment (Armand Frappier Institute, Laval, QC) and 2 where I0 is the incident radiation intensity (mW/cm ); d stored at 80 C. is the depth of wastewater sample under UVirradiation ðd¼ 2:2cmÞ; T is the transmittance at 254 nm [with a cell path length of 1 cm] (as a decimal fraction of 1). 4. Results and discussion The sample was irradiated for the specified time and kept in the dark and cold until microbiological analysis 4.1. Wastewater characteristics was performed. Due to the wide variability in the quality of the 3.3.3. Ozone effluent samples from the wastewater treatment plant, The wastewater sample (2.3 L) was added to a gas the data are presented in Fig. 2 on a log scale showing washing bottle equipped with a glass-fritted diffuser the mean as well as the range. An attempt was made to (ozone reactor bottle) and placed on a magnetic stirrer correlate these data with disinfection performance, but (Fig. 1). During ozonation the gas flow rate, the this was largely unsuccessful. It is clear, however, that pressure, the concentration of ozone in the outgoing this wastewater would not be ‘‘easy’’ to disinfect, and gas and the ozone residuals were monitored. Samples for this is typical of wastewaters from purely physicochem- microbiological analysis were withdrawn at 5, 10, 15, 30 ical treatment plants [17]. In particular, the high and 45 min. Residual ozone was eliminated using concentrations of COD (123–240 mg/L), SS (16–45 mg/ sodium thiosulfate at a concentration of 50 mg/L. The L), turbidity (16–31 NTU), and total iron (0.2–7.5 mg/ temperature of the wastewater was 20–21C throughout L), and low UVtransmittance (4.6–29.5%) are noted. It the experiments. is known from previous work that high COD will compromise the performance of ozone and PAA ([8,28], 3.4. Wastewater analysis respectively), whereas high SS, turbidity and iron, as well as low UVtransmittance will compromise UV Most wastewater analyses were performed according performance [15,17]. to Standard Methods [34]. Particle size distribution was analysed using a Lasentec M100 F Particle System 4.2. PAA disinfection efficiency Characterization Monitor (Lasentec, Redmond, WA). Fig. 3 shows the inactivation of fecal coliforms vs. 3.5. Microbial analyses PAA dose for a contact time of 1 h. Due to the high day- to-day variability of the wastewater quality, hence the Samples of wastewater effluent, following disinfec- disinfection efficiency of the PAA, results from all five tion, were assayed for three groups of bacteria. Samples test days are shown. Fig. 4 shows the residuals measured were also spiked with MS-2 coliphage stock solution so after 1 h for various doses. Although a contact time of as to obtain a final titre of B105 plaque forming units approximately 2 h would be available in the outfall (PFU)/mL, then disinfected. Procedures performed are tunnel at average flow, earlier studies had shown that the summarized in Table 2. Medium for selective culture of additional contact time was not beneficial [8] and this is

Table 2 Microbiological analytical procedures

Microorganisms Analytical procedures

Fecal coliforms Membrane filtration (Method 9222D, [34]) Enterococci Membrane filtration (Method 1600, [37]) Clostridium perfringens Membrane filtration and m-CP medium [36] incubated in BBLsGasPak (Fisher Co., Mtl, QC) at 44.5C for 18–24 h MS-2 coliphage Single-agar layer procedure with E. coli F+ as host strain: Method 1602 [38] ARTICLE IN PRESS 4580 R. Gehr et al. / Water Research 37 (2003) 4573–4586

1,000

Min 100 Mean Max

10

1

0 Kurtosis (mg/L) SS (mg/L) (%/cm) COD (mg/L) size (µm) Calcium total BOD5 (mg/L) Fe filtrable COT non- Al total (mg/L) Fe total (mg/L) Mg total (mg/L) Median particle Turbidity (NTU) UVT @ 254 nm <0,45µm (mg/L) purgeable (mg/L) PO4 total (mg P/L) Fig. 2. Summary of effluent quality data.

10,000,000

1,000,000

100,000

10,000

1,000

100

Fecal coliforms (CFU/100 mL) Aug. 15 Aug. 21 Aug. 30 Sept. 6 Sept. 12 10

1 0123456 PAA dose (mg/L) Fig. 3. Fecal coliform inactivation by PAA.

understandable in light of the low residual concentra- For a 1-log removal of EC, 4.5 mg/L PAA were tions of peroxycompounds remaining after 1 h. The required, and there was no inactivation of CP (results not target FC level (9000 CFU/100 mL) was reached on shown). Only two dose curves were obtained for MS-2, only 2 days (August 21 and September 12) at a and up to 1-log inactivation was achieved at relatively PAA dose of 4.5–6 mg/L. No significant disinfection low doses of 1.5–3 mg/L PAA (Fig. 5), but this did not was observed on the other days. The samples from improve at higher doses. It is possible that the mechan- August 21 and September 12 had higher than ism of inactivation for the bacteriophages is different to average UVtransmittance, lower COD, BOD and that for bacteria. Given that a dose of B1 mg/L was turbidity, and were more ‘‘dilute’’ than the other considered to be economically viable, and that there was samples. It appears therefore that the high level of large variability in the results—pointing to high sensi- organics in most samples was responsible for the poor tivity of PAA to variable effluent quality—the use of effectiveness of PAA. PAA would have to be ruled out for this application. ARTICLE IN PRESS R. Gehr et al. / Water Research 37 (2003) 4573–4586 4581

Fig. 4. PAA residuals.

100,000 30-Aug 12-Sep

10,000

1,000

100 MS-2 (PFU/ mL)

10

1 0123456 PAA dose (mg/L) Fig. 5. MS-2 inactivation by PAA.

4.3. UV disinfection efficiency a 1-log reduction was achieved for each 10 mJ/cm2 (Fig. 8). Several interesting points emerge from these As expected, the UVinactivation curve for FC results. First, as already noted, there was no tailing showed the familiar steep decline in numbers at fluences effect (as is also evident in results obtained by Shin et al. up to 20 mJ/cm2 (‘‘free-swimming portion’’) followed by [39]). Hence, it must be surmised that neither CP nor an asymptote beyond that fluence (Fig. 6). This type of MS-2 were incorporated into flocs. Nieuwstad and curve is generally found for wastewaters with high Havelaar [40] did find tailing effects with bacterio- suspended solids concentrations, and the results agree phages, but they suggested that since all phages are with earlier work [18]. In most cases the target FC level supposed to be identical, the tailing was due to could be reached in the asymptote zone (i.e. at clustering. fluences>20 mJ/cm2), but there were instances where Shielding (in the sense of an umbrella effect) appears even at 60 mJ/cm2 the FC count was above 1000 CFU/ not to have taken place either, which is understandable 100 mL. As with PAA, similar results were obtained for given the relatively long exposure times (10–15 min), EC inactivation. hence the fact that a solids shield would not likely The curves for CP and MS-2 did not show the 2 stage remain in place so long. Confirmation of this hypothesis behaviour. The CP counts showed a steady decline with (i.e. organisms not incorporated into solids or flocs) was increasing fluence (Fig. 7), with an approximate 1-log given in the case of the MS-2 by the fact that there was reduction for an increment of 30 mJ/cm2. For the MS-2, no difference in the results for seeds which were added ARTICLE IN PRESS 4582 R. Gehr et al. / Water Research 37 (2003) 4573–4586

10,000,000

1,000,000

100,000

10,000

1,000

100 Fecal coliforms (CFU/100 mL) Bars are 95% c.I; n = 13 10

1 0 102030405060 UV fluence (mJ/cm2) Fig. 6. Fecal coliform inactivation by UV.

10,000

1,000

100

Bars are 95% c.I; n = 11

10 Clostridium perfringens (CFU/100 mL)

1 0 102030405060

UV fluence ( mJ/cm2) Fig. 7. Clostridium perfringens inactivation by UV.

100,000

10,000

1,000

100 MS-2 (PFU/ mL)

Bars are 95% c.I; n = 10 10

1 0 102030405060 UV fluence (mJ/cm2) Fig. 8. MS-2 coliphage inactivation by UV.

1 h (7 tests) or 16 h (3 tests) before exposure to UV. This deserves further exploration, as it could have implica- fundamental difference in behaviour between the two tions in interpreting them as indicators of inactivation of sets of organisms (FC and EC vs. CP and MS-2) pathogens (bacterial, protozoan or viral). In the former ARTICLE IN PRESS R. Gehr et al. / Water Research 37 (2003) 4573–4586 4583 case (FC and EC), nothing would be gained by count had decreased by as little as 1-log. For practical increasing UVfluences substantially, whereas in the purposes (mainly economic), an ozone dose of 20 mg/L latter case (CP and MS-2) there would be a benefit or less was considered reasonable. Based on Figs. 9 and shown at higher fluences. 10, the ozone demand of this wastewater (being the The second point to emerge is that CP is by far the transferred ozone dose when a residual first appears) is most resistant to UVlight of the four organisms tested. approximately 25 mg/L. Therefore for the CMWTP Thus it could be considered to be a conservative effluent in its current state, ozone is inappropriate. indicator of UVperformance. Although not tested explicitly, this very high ozone dose requirement was assumed to be due to the presence of 4.4. Ozone disinfection efficiency organics in the wastewater, as shown by earlier studies using a continuous-flow system [30]. Fig. 9 shows a typical curve for ozone residual vs. time Interestingly, the inactivation of MS-2 by ozone was in the batch test, and each successive point in Figs. 10 very successful, as indicated in Fig. 11. For a transferred and 11 occurs at contact times of 5, 10, 15, 30 and ozone dose of approximately 17 mg/L, counts had 45 min, respectively. Pertinent data for ozone residuals, decreased by over 3-logs. Hence it could be expected etc., are shown in Table 3 below. The Ct values were that, notwithstanding the high ozone demand of the calculated using the area under the residual vs. time wastewater, ozone would be an effective disinfectant for curve (such as Fig. 9). viruses in this wastewater. This would suggest that the These values are within the same range as those given inactivation mechanism of the bacteriophage was much by the EPA [41] for drinking water, viz. 1.0 for 4-log different to that of the bacteria, as it had to be almost inactivation of viruses (at 10C), 1.2 for 2.5-log instantaneous. The tailing effect is likely an artifact of inactivation of Giardia cysts, and 2.4–10 for 2-log the MS-2 enumeration procedure, since MS-2 numbers inactivation of Cryptosporidium oocysts. However the cannot be accurately measured below 1 PFU/mL, but it implication of Ct being equivalent to a ‘‘dose’’ in this could also be due to clustering. case is false, since it took close to 20 min to develop a residual (hence any Ct value), yet inactivation did occur during this period. It follows that reaction products of 5. Summary and conclusions ozone (notably the radicals) are implicated in the inactivation process. PAA was not a suitable disinfectant for the City of Results of inactivation of FC and CP by ozone are Montreal’s Wastewater Treatment Plant. Based on a shown in Fig. 10. Inactivation of EC was very similar to fecal coliform (FC) target level of 9000 CFU/100 mL, that of FC. Although ozone effectiveness was approxi- over 6 mg/L would be required, whereas 1.5–2 mg/L is mately linear over the entire dose range tested, the target considered economically viable. For ozone, the required FC level (equivalent to approximately 2-log inactiva- dose exceeded 30 mg/L; this was also considered exces- tion) was reached only at a transferred ozone dose of sive. On the other hand, for a more stringent FC target between 30 and 50 mg/L, and at these doses the CP of 1000 CFU/100 mL to account for photoreactivation,

Sample - 14-Nov 0.6

0.5 Bac-Ozonation MS2-Ozonation 0.4

0.3

0.2 Ozone residual (mg/L)

0.1

0 0 5 10 15 20 25 30 35 40 45 50 Ozonation time (min) Fig. 9. Development of ozone residuals. ARTICLE IN PRESS 4584 R. Gehr et al. / Water Research 37 (2003) 4573–4586

10,000,000 6-Nov 1,000,000 8-Nov

100,000 14-Nov

10,000

1,000

100 Fecal coliforms (CFU/100 mL)

10

1 0 10203040506070 (a) Transferred ozone (mg/L)

10,000

1,000

100

6-Nov

10 8-Nov

Clostridium perfringens (CFU/100 mL) 14-Nov

1 0 10203040506070 (b) Transferred ozone (mg/L) Fig. 10. (a) Inactivation of fecal coliforms by ozone and (b) inactivation of C. perfringens by ozone.

100,000.00

10,000.00 6-Nov

1,000.00 8-Nov

14-Nov

100.00 MS-2 (CFU/ mL) 10.00

1.00

0.10 0 1020304050607080 Transferred ozone (mg/L) Fig. 11. Inactivation of MS-2 coliphage by ozone. the required UVfluence was 20 mJ/cm 2, which may be solids, and/or ferric concentrations in the effluent. economically acceptable. Based on previous studies as Lower concentrations of these parameters as a result well as on the current effluent quality, the high dose of changes to upstream processes or operations may requirements are likely due to the high COD, suspended alter the picture and render PAA or ozone more viable. ARTICLE IN PRESS R. Gehr et al. / Water Research 37 (2003) 4573–4586 4585

Table 3 Ozone residual kinetics data and Ct

Date Test org. Time to Slope after first Ct (mg-min/L) after Log inactivation after first appearance appearance given contact time (min) given contact time (min) of a residual (mg/L min) (min) 5–15 30 45 5 10 15 30 45 Nov. 6 FC 21 0.0173 — 0.6 4.8 0 1 1 2 3 Nov. 7 MS-2 20 0.0037 — 0.2 1.3 1 NA NA NA 3 Nov. 8 FC 16 0.0157 — 1.6 7 1 2 2 2 3 Nov. 9 MS-2 17 0.0103 — 1 4.4 1 3 4 4 5 Nov. 14 FC 18 0.02 — 1.4 7.3 0.3 1.1 1.2 2.1 2.8 Nov. 15 MS-2 NA NA — NA NA 1.2 4.0 3.8 3.9 4.1

NA = not available.

When other organisms were assessed, the picture [6] EPA. Combined sewer overflow technology fact sheet. changed considerably. At a dose of only 1.5 mg/L, PAA Alternative disinfection methods. 1999a, EPA 832-F-99-033. was able to reduce MS-2 coliphage counts by 1-log, but [7] Atasi K, Rabbaig M, Chen C. Alternative disinfectants there was no effect on C. perfringens (CP). Compared to evaluation for combined sewage overflow (CSO): Detroit FC, CP and MS-2 were much more resistant to UV. Baby Creek CSO case study. Proceedings of the 74th Their inactivation curves did not show the classic two- Annual Conference, Water Environment Federation, Atlanta, GA, 2001. stage shape, hence their inactivation was not signifi- [8] Gehr R, Cochrane D, French M. Peracetic acid (PAA) as a cantly affected by the presence of solids. In contrast, disinfectant for municipal wastewaters: encouraging per- ozone was highly effective against MS-2, achieving a formance results from physicochemical as well as biologi- reduction of over 3-logs for a dose of 17 mg/L. Thus it cal effluents. Proceedings of the Disinfection 2002 would likely be considered the most appropriate Conference, Water Environment Federation, St. Peters- disinfectant if viruses were the target organisms. burg, FL, February 17–20, 2002. Further research on the mechanisms of disinfection as [9] Liberti L, Lopez A, Notarnicola M. Disinfection with well as on the interaction between certain organisms and peracetic acid for domestic sewage reuse in agriculture. suspended solids (chemical flocs) in the effluent is CIWEM J 1999;13:262–9. warranted to explain the varying behaviour of these [10] Lubello C, Caretti C, Gori R. Comparison between PAA/ organisms when subject to different disinfectants. A risk UVand H 2O2/UVdisinfection for wastewater reuse. management approach is being undertaken to determine Water Sci Technol: Water Suppl 2002;2(1):205–12. whether some forms of recreational activities could be [11] Lefevre F, Audic JM, Ferrand F. Peracetic acid disinfec- tion of secondary effluents discharged off coastal seawater. permitted in the area impacted by a disinfected effluent. Water Sci Technol 1992;25(12):155–64. [12] Lazarova V, Janex ML, Fiksdal L, Oberg C, Barcina I, Pommepuy M. Advanced wastewater disinfection technol- References ogies: short and long term efficiency. Water Sci Technol 1998;38(12):109–17. [1] MEQ. Criteres" de qualite! de l’eau de surface au Quebec.! [13] WEF Wastewater disinfection. Manual of Practice FD-10. Ministere" de l’environnement du Quebec! (Canada), Alexandra, VA: Water Environment Federation, 1996. 1998. [14] Angehrn M. Ultraviolet disinfection of water. Aqua [2] Haas CN, Rose JB, Gerba CP. Quantitative microbial risk 1984;2:109–15. management. New York: Wiley; 1999. [15] Emerick RW, Loge FJ, Ginn T, Darby JL. Modeling the [3] Baldry MGC, French MS. Disinfection of sewage effluent inactivation of particle-associated coliform bacteria. Water with peracetic acid. Water Sci Technol 1989;21(3):203–6. Environ Res 2000;72(4):432–8. [4] Baldry MGC, French MS, Slater D. The activity of [16] Blatchley III ER, Dumoutier N, Halaby TN, Levi Y, La#ıne! peracetic acid on sewage indicator bacteria and viruses. JM. Bacterial responses to ultraviolet radiation. Water Sci Water Sci Technol 1991;24(2):353–7. Technol 2001;43(10):179–86. [5] Grantham R. Research by the National Rivers Authority [17] Gehr R, Wright H. UVdisinfection of wastewater (NRA) in England and Wales in the efficacy and side coagulated with ferric chloride: recalcitrance and fouling effects of a variety of techniques for the disinfection of problems. Water Sci Technol 1998;38(3):15–23. sewage effluents. Proceedings of the conference ‘‘Disin- [18] Cairns WL, Sakamoto G, Comair CB, Gehr R. Assessing fecting Wastewater for Discharge and Reuse’’. Water UVdisinfection of a physico-chemical effluent by medium Environment Federation, Portland, OR, March 17–20, pressure lamps using a collimated beam and pilot plant. 1996. p. 6-1–6-12. Proceedings of the Water Environment Federation ARTICLE IN PRESS 4586 R. Gehr et al. / Water Research 37 (2003) 4573–4586

Specialty Conference: Planning, Design and Operation the 11th World Ozone Congress, San Francisco, of Effluent Disinfection Systems, Whippany, NJ, 1993. Aug 29–Sept 3, 1993. p. S-7-33-42. p. 433–44. [31] Wagner M, Brumelis D, Gehr R. Disinfection of waste-

[19] Madge BA, Jensen JN. Discussion of: modelling the water by H2O2 or PAA: development of procedures for inactivation of particle-associated coliform bacteria. Water measurement of residual disinfectant, and application to Environ Res 2001;73(4):504–5. a physico-chemically treated municipal effluent. Water [20] Masschelein WJ. Ultraviolet light in water and wastewater Environ Res 2002;74(1):33–50. sanitation. Boca Raton: Lewis Publishers; 2002. [32] Putter. J, Becker R. Peroxidases. In: Bergmeyer J, GraXlM, [21] Oppenheimer JA, La#ıne! J-M, Jacangelo JG, Bhamrah A, editors. Methods of enzymatic analysis., vol. 3. Weinheim: Hoagland JE. Chlorine and UVdisinfection of tertiary Verlag Chemie; 1983. p. 286–93. effluent: a comparative study of bacterial and viral [33] IOA (International Ozone Association). Iodometric meth- inactivation and effluent by-products. Proceedings of the od for the determination of ozone in a process gas, 1998. 66th Annual Conference, Water Environment Federation, [34] APHA, AWWA, WEF. Standard methods for the Anaheim, CA, 1993. p. 557–68. examination of water and wastewater, 20th ed. Washing- [22] Robson CM, Rice RG. Wastewater ozonation in the ton, DC: APHA; 1998. USA—history and current status—1989. Ozone Sci Eng [35] Morowitz HJ. Absorption effects in volume irradiation of 1991;13:23–40. microorganisms. Science 1950;111:229–30. [23] Xu P, Janex M-L, Savoye P, Cockx A, Lazarova V. [36] Armon R, Payment P. A modified m-CP medium for Wastewater disinfection by ozone: main parameters for enumerating Clostridium perfringens from water samples. process design. Water Res 2002;36:1043–55. Can J Microbiol 1988;34:78–9. [24] Paraskeva P, Graham NJD. Ozonation of municipal [37] EPA. Method 1600: membrane filter test method for wastewater effluents. Water Environ Res 2002;74(6): enterococci in water. US Environmental Protection 569–81. Agency, Office of Water, Washington DC, 1997, EPA- [25] Larocque R. Ozone applications in Canada—a state of the 821-R-97-004. art review. Ozone Sci Eng 1999;21(2):119–26. [38] EPA. Method 1602: Male-specific (F+) and somatic [26] Hermanowicz SW, Bellamy WD, Fung LC. Variability of coliphage in water by single agar layer (SAL) procedure. ozone reaction kinetics in batch and continuous flow US Environmental Protection Agency, Office of Water, reactors. Water Res 1999;33(9):2130–8. Washington, DC, 2001, EPA 821-R-01-029. [27] Oke NJ, Smith DW, Zhou H. An empirical analysis of [39] Shin G-A, Linden K, Handzel T, Sobsey MD. Low- ozone decay kinetics in natural waters. Ozone Sci Eng pressure UVinactivation of Cryptosporidium parvum based 1998;2:361–79. on cell culture infectivity. Proceedings, American Water [28] Gehr R, Nicell J. Pilot studies and assessment of down- Works Association Water Quality and Technology Con- stream effects of UVand ozone disinfection of a ference, Tampa, FL, 1999, October 31–November 3. physicochemical wastewater. Water Qual Res J Can 1996; [40] Nieuwstad ThJ, Havelaar AH. The kinetics of batch 31(2):263–81. ultraviolet inactivation of bacteriophage MS2 and [29] Hunt NK, Marin˜ as BJ. Kinetics of Eschericia coli microbiological calibration of an ultraviolet pilot plant. inactivation with ozone. Water Res 1997;31(6):1355–62. J Environ Sci Health A 1994;29(9):1993–2007. [30] Absi F, Gamache F, Gehr R, Liechti P, Nicell J. Pilot [41] EPA. Alternative disinfectants and oxidants guidance plant investigation of ozone disinfection of a physico- manual. US Environmental Protection Agency, Office of chemically treated municipal wastewater. Proceedings of Water, Washington, 1999b, EPA 815-R-99-014.