Impact of microparticles on UV disinfection of indigenous aerobic spores

ARTICLE IN PRESS WAT E R R E S E A R C H

41 (2007) 4546– 4556

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Impact of microparticles on UV disinfection of indigenous aerobic spores Eric Carona, Gabriel Chevrefils Jr.a, Benoit Barbeaua, Pierre Paymentb, Miche`le Pre´vosta, NSERC Industrial Chair on Drinking Water, E´cole Polytechnique de Montreal, Department of Civil, Geologic and Mining Engineering, P.O. Box 6079, Succ. Centre Ville, Montreal, Que´., Canada H3C 3A7 b INRS—Institut Armand-Frappier, 531 boul. des Prairies, Laval, Que´., Canada H7V 1B7 a

ar t ic l e i n f o

abs tra ct

Article history:

Numerous studies have shown that the efficacy of ultraviolet (UV) disinfection can be

Received 16 August 2006

hindered by the presence of particles that can shield microorganisms. The main objective

Received in revised form

of this study was to determine to what extent natural particulate matter can shield

17 May 2007

indigenous spores of aerobic spore-forming bacteria (ASFB) from UV rays. The extent of the

Accepted 13 June 2007

protective shielding was assessed by comparing the inactivation rates in three water

Available online 17 June 2007 Keywords: UV disinfection Drinking water Particles Dispersion Spores

fractions (untreated, dispersed and filtered on an 8 mm membrane) using a collimated beam apparatus with a low-pressure lamp emitting at 254 nm. Levels of inactivation were then related to the distribution and abundance of particles as measured by microflow imaging. Disinfection assays were completed on two source waters of different quality and particle content. A protocol was developed to break down particles and disperse aggregates (addition of 100 mg/L of Zwittergent 3–12 and blending at 8000 rpm for 4 min). Particle size distribution (PSD) analysis confirmed a statistically significant decrease in the number of particles for diameter ranges above 5 mm following the dispersion protocol and 8 mm filtration. The fluence required to reach 1-log inactivation of ASFB spores was independent of particle concentration, while that required to reach 2-log inactivation or more was correlated with the concentration of particles larger than 8 mm (R240.61). Results suggest that natural particulate matter can protect indigenous organisms from UV radiation in waters with elevated particle content, while source water with low particle counts may not be subject to this interference. & 2007 Elsevier Ltd. All rights reserved.

1.

Introduction

Ultraviolet (UV) treatment is considered to be the silver bullet for drinking water disinfection. Pioneer work has shown that the oocysts of Cryptosporidium, which are among the organisms most resistant to conventional chemical disinfection processes, can be readily inactivated with relatively low doses of UV radiation relative to other microorganisms (Clancy et al., 1998; Craik et al., 2001; Shin et al., 2001; Mofidi et al.,

2001; Zimmer et al., 2003). Cryptosporidium is of particular interest, since it was this organism that was responsible for a major waterborne outbreak in Milwaukee in 1993 that caused over 400,000 cases of cryptosporidiosis (Mac Kenzie et al., 1994). Water quality is known to significantly influence the transmitted fluence and the resulting level of inactivation (Qualls et al., 1983; Batch et al., 2004; Mamane and Linden, 2006b). Dissolved and suspended compounds absorb UV light,

Corresponding author. Tel.: +1 514 340 4711 5924; fax: +1 514 340 5918.

E-mail addresses: [email protected] (E. Caron), [email protected] (G. Chevrefils Jr.), [email protected] (B. Barbeau), [email protected] (P. Payment), [email protected] (M. Pre´vost). 0043-1354/$ - see front matter & 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2007.06.032

ARTICLE IN PRESS WAT E R R E S E A R C H

4 1 (200 7) 454 6 – 455 6

and particulate matter can cause light scattering (Qualls et al., 1983; Batch et al., 2004). Light scattering may lead to an overestimation of UV absorbance and thus to an underestimation of the calculated fluence in typical collimated beam UV experiments (Qualls et al., 1983). The scattering of UV rays increases the light’s path length and the overall resulting absorbance by the water, and thus reduces the energy available for disinfection (Mamane-Gravetz and Linden, 2004a). Christensen and Linden (2003) report that a turbidity of 10 NTU can cause the irradiance calculated from absorbance measurements made with a conventional spectrophotometer to be underestimated by 5%. In a cylindrical reactor 18 cm in diameter with a 50 W low-pressure UV lamp, this underestimation can reach up to 20% (Christensen and Linden, 2003). Since turbidity is a regulated, easy to use and widely used particle indicator in water treatment plants, attempts have been made to determine the threshold turbidity value under which no significant interference with UV disinfection can be anticipated. Various experimental approaches have relied on seeded microorganisms and added natural or synthetic turbidity (inorganic, organic and chemical flocs), either combined by simple mixing or by forming particles to embed the microorganisms. Results provide insight into the effects of the dispersion of UV light by particles of various compositions, the adsorption of organisms onto particles in drinking water sources and the inclusion of microorganisms in chemical flocs. In general, results suggest that the impact on the inactivation of mineral or organic turbidity under 10 NTU is not significant using seeded organisms, if adequate correction is made to account for light scattering (Wobma et al., 2004; Passantino and Malley, 2001; Passantino et al., 2004; Batch et al., 2004). Recent studies have investigated the protective effect of chemical flocs, and results suggest that turbidity below 10 NTU may impact the inactivation of coliform bacteria and the spores of Bacillus (Mamane-Gravetz and Linden, 2004a; Craik and Uvbiama, 2005; Ormeci and Linden, 2002). The general consensus is that considering turbidity alone is probably insufficient to account for the complex phenomena involved in particle shielding during UV disinfection. UV disinfection studies suggest that the number, size distribution and chemical nature of particles are better predictors of the potential for shielding microorganisms (Qualls et al., 1985; Mamane and Linden, 2006b; Templeton et al., 2005). Particle shielding can significantly decrease the level of inactivation of coliform bacteria (Ormeci and Linden, 2002; Qualls et al., 1983, 1985; Loge et al., 1999, 2001; Emerick et al., 1999; Parker and Darby, 1995; Emerick and Darby, 1993; Scheible, 1987; Emerick et al., 2000; Jolis et al., 2001) and aerobic spores (Mamane and Linden, 2006a; Craik and Uvbiama, 2005). Three main factors may influence the extent of the influence of particles on UV disinfection: (1) the number and size of the particles; (2) the degree of association of microorganisms with particles; and (3) the nature of the particles. (1) The number and size of the particles: Wastewater results suggest that coliform shielding is mainly attributable to particles of a diameter of 7–10 mm (Qualls et al., 1983; Jolis

4547

et al., 2001). Tailing is commonly observed for fluences greater than 30 mJ/cm2, and is considered to be the result of the protection of bacteria by the large flocs present in wastewater effluents (Loge et al., 2001; Parker and Darby, 1995; Scheible, 1987). Small particles less than 2 mm in diameter may shield smaller microorganisms such as MS2 and T4 bacteriophages (Templeton et al., 2005); however, large particles (larger than 41 mm in diameter) can compromise UV inactivation of Mycobacterium terrae (Bohrerova and Linden, 2006). No effect of smaller particles (o20 mm) was noted for the inactivation of M. terrae. (2) The degree of association of microorganisms with particles: The extent of the association of organisms with particles is a key factor in determining the significance for health and the potential for shielding. The potential for protection may not be significant if the particles do not harbor pathogenic microorganisms. Although some experiments have been conducted to reveal the extent of the aggregation effect on UV disinfection performances (Craik and Uvbiama, 2005; Templeton et al., 2005; Mamane-Gravetz and Linden, 2004a), most results have been obtained with spiked microorganisms and seeded particulate matter with or without coagulation. These conditions may not be indicative of the level of association of particles and microorganisms in source waters. (3) The nature of the particles: The nature of particles, inorganic versus organic, of natural and synthetic waters may affect the level of shielding (Mamane-Gravetz and Linden, 2004a; Templeton et al., 2005, 2006; Mamane and Linden, 2006a). Mamane and Linden (2006a, b) observed that 30–50% of seeded Bacillus subtilis spores in chemical flocs were protected from UV light, while Craik and Uvbiama (2005) found that coagulated and filtered particles formed with B. subtilis spores were partially shielded from UV light. Moreover, the presence of particles in organic wastewater flocs and elevated concentrations of humic acids reduced viral inactivation by 1 to 42.5 log, depending on UV fluence and the organisms tested (MS2 or T4 phages), while kaolin clay particles provided no significant protection (Templeton et al., 2005). When investigating the impact of particles on UV disinfection of unfiltered water, the impact of natural particles found in the source water can be assessed by measuring the inactivation of particle-associated indigenous organisms such as spores of aerobic spore-forming bacteria (ASFB). ASFB spores are naturally present in surface water, and enumerating them is easy, fast, economical and reliable (Barbeau et al., 1997). Pure strains of B. subtilis are very resistant to UV light (Mamane-Gravetz et al., 2005; Nicholson and Law, 1999; Sommer et al., 1998, 1999; Chang et al., 1985), while spores indigenous to surface waters (Mamane-Gravetz and Linden, 2004b, 2005) and to soils (Nicholson and Law, 1999) are even more resistant. If ASFB spores are associated with particles, they can serve as a model organism without the need for seeding or the addition of synthetic turbidity or organic matter. The relatively high resistance of ASFB spores to UV is an advantage, because of the expanded scale of their response when compared with that of Cryptosporidium. This allows smaller but significant effects of water quality to be

ARTICLE IN PRESS

480,723 284,405 243,145 441,020 283,305 80 88 46 N/D N/D 100 94 64 N/D N/D N/D N/D 7.0 N/D N/D 6.05 6.05 5.82 N/D N/D 6.19 6.08 5.95 N/D N/D 39.9 52.7 53.8 52.9 54.8 9.6E+02 8.3E+02 9.2E+02 N/A N/A Des Mille-Iles River 17-10-2005 2.3E+04 18-10-2005 8.0E+03 24-01-2006 3.0E+03 12-04-2006 3.7E+03 18-04-2006 1.8E+03

36.2 16.7 11.1 28.0 13.1

32,557 67,084 65,047 49,801 59,314 85 82 85 81 84 120 120 120 118 120 7.8 7.6 7.8 7.6 7.8 2.31 2.18 2.33 2.22 2.17

Alcalinity (mg CaCO3/L) Hardness (mg CaCO3/L) pH DOC (mg C/L) TOC (mg C/L)

2.34 2.20 2.33 2.24 2.19 89.4 90.0 94.4 92.4 87.9 1.47 1.04 0.75 1.40 1.31

Two surface water sources were tested during this project: the Mille Iˆles River (intake of the Re´gie Intermunicipale des Moulins Water Treatment Plant (WTP)) and the St. Lawrence River (intake of the Charles-DesBaillets WTP), both of which are in the Montreal area, in the province of Que´bec, Canada. The former is characterized by a high organic content (TOCE6 mg C/L) and varying turbidity (10 to 435 NTU), and is impacted by wastewater effluents and combined sewer overflows, as indicated by the rather high fecal coliform concentrations measured there (see Table 1). A survey of 100 source waters revealed a median concentration of particles larger than 3 mm of 9600 particles/mL (McTigue et al., 1998). Water from the Mille Iˆles River contains significantly higher concentrations of particles than this median, with an average value of 61,130 particles 43 mm/mL. The Charles-DesBaillets WTP (Montreal, Quc.) draws its water from the St. Lawrence River and is considered a high-quality surface water source with low turbidity (o1.5 NTU) and low TOC (o2.4 mg C/L). Median particle concentration for this second source is 9500 particles 43 mm/mL, below the reported median value. Detailed water quality parameters are presented in Table 1. Five water samples were collected at each site from October 2005 to February 2006. Samples were collected, shipped on

1.0E+01 3.0E+00 2.0E+01 1.0E+00 1.0E+00

Water samples and experimental approach

St. Lawrence River 22-11-2005 2.0E+02 19-12-2005 8.0E+01 13-01-2006 7.0E+01 18-01-2006 4.6E+02 20-01-2006 3.9E+02

2.1.

Transmittance 254 nm (%)

Materials and methods

Turbidity (NTU)

2.

Fecal coliforms (CFU/100 mL)

measured. Apart from being naturally present in surface water, ASFB spores represent an interesting model for their size, which is similar to that of Cryptosporidium oocysts. These spores have a diameter of 1–2 mm (Craik and Uvbiama, 2005), while oocysts have a diameter of about 4–6 mm (Feng et al., 2003). This similarity in size is relevant for determining the range of particle size that will potentially provide a shield from UV light. The primary objective of this project is to study the inactivation of indigenous ASFB spores in two source waters, in order to evaluate whether or not either their natural state of aggregation or their association with particles provides significant protection from 254 nm UV light. The impact of particles is estimated by (1) measuring the effect of physicochemical dispersion and membrane filtration (8 mm) on the inactivation rates of ASFB spores and (2) relating the variations in inactivation to the distribution of particles as measured by microflow imaging. Dispersion causes partial break-ups of aggregates and particles. Membrane filtration removes large particles and provides an estimation of the proportion of the spore population associated with particles. This study is original in that disinfection assays were performed on water containing indigenous spores at their usual environmental concentrations. More importantly, this approach allows for the assessment of the effect of particles and aggregation on the UV inactivation of spores in their natural state, as opposed to studying these phenomena using synthetic coagulated particles or by the addition of natural or synthetic turbidity. To the best of our knowledge, it is the first investigation of the impact of natural particles on the UV disinfection of naturally occurring spores.

Particles41 mm [ ]/ml

41 (2007) 4546– 4556

ASFB spores (CFU/100 mL)

WAT E R R E S E A R C H

Table 1 – Biological, physical and chemical parameters for each sampling campaign in both the Des Mille Iˆles River and the St-Lawrence River

4548

ARTICLE IN PRESS WAT E R R E S E A R C H

4 1 (200 7) 454 6 – 455 6

ice, maintained at 4 1C upon reception and analyzed or used for assays within 24 h. The protocol used in this study is a modified version of the one proposed by Parker and Darby (1995) for wastewater. The protocol involves performing UV disinfection assays on three fractions of water: (i) non-dispersed spores—water samples are irradiated without any pre-treatment, after which they are blended to disperse the naturally occurring particles or aggregates prior to enumeration; (ii) dispersed spores—samples are dispersed prior to irradiation to evaluate the impact of spore aggregation and/or embedding in the particles; and (iii) filtered spores—samples are filtered through 8 mm pore membranes, after which the filtered samples are irradiated and then blended to disperse spores prior to enumeration. All irradiated samples contained the chemical dispersion mix.

2.2.

Dispersion of particulate matter

A dispersion protocol was first developed to induce particle breakdown and the dispersion of aggregated ASFB spores. The final procedure selected was derived from two sets of trials conducted using various surfactant solutions and different blending intensities and durations. The surfactant solutions tested were Camper’s solution, which was first proposed for the extraction of Escherichia coli bacteria from granularactivated carbon (GAC) (Camper et al., 1985) and variations of this solution. Camper’s solution is a mix of 106 M (0.336 mg/L) of Zwittergent 3–12, Tris-buffer (0.01 M), EGTA (103 M) and 0.1% peptone. Additional variations included (i) Camper’s solution, but with a higher level of Zwittergent (10 mg/L); (ii) Zwittergent 3–12 only at 10 mg/L; and (iii) Zwittergent 3–12 only at 100 mg/L. The optimal chemical blend was selected as the one maximizing aerobic spore counts for the same blending condition. Then, the optimal blending conditions were determined by testing combinations of four blending speeds (8000, 15,400, 18,700 and 21,000 rpm) and four durations (2, 3, 4 and 5 min). This experimental design was replicated six times over a 2-month period, in order to consider water quality variability. Blending was performed on a volume of 300 mL using a Waring laboratory blender, Model 7012 (Waring, Torrington, CT, USA). Volumes of 200–300 mL were blended. Blended and unblended samples were irradiated up to fluences of 135 mJ/cm2. Between each blending phase, to avoid crosscontamination, the blending container was thoroughly washed with soap and distilled water, rinsed with Ultrapure MilliQ water, filled with 500 mL of MilliQ water and blended for 1 min at high speed (22,000 rpm), and finally exposed to UV for 10 min under the UV lamp in a biological hood at a fluence rate at 254 nm of 0.40 mJ/cm2 s. Once optimal conditions had been identified, dispersion was achieved by adding 100 mg/L of Zwittergent 3–12 and blending for 4 min at 8000 rpm. Two-minute rest intervals followed each minute of blending to minimize any increase in water temperature and foaming.

2.3.

Sample filtration

To remove particles by filtration, samples were filtered under vacuum on 47 mm, 8 mm pore size nitrocellulose filters

4549

(Millipore, SCWPO4700). A maximum of 100 mL was filtered in order to limit cake filtration. Filtration was performed immediately prior to UV exposure and before particulate analysis, in order to minimize post-filtration aggregation of particulate matter in the filtrate. Direct evidence of association and/or aggregation is provided by the removal of spores by filtration on an 8 mm pore size filter. On average, 60% (St. Lawrence River) to 75% (Mille Iˆles River) of the spores were retained by the 8 mm filtration.

2.4.

Particle size distribution analysis

Particle size distribution (PSD) analysis was carried out with a Brightwell Dynamic Particle Analyzer (DPA) (Brightwell Technologies, Ottawa, Ont.) based on direct magnified photographic image analysis. A volume of 2 mL was analyzed for each sample, which resulted in a minimum total of 6000 counted particles to more than 100,000 counted particles. The detection limit of this apparatus is 1 mm.

2.5.

Fluence measurements and UV treatment

The UV collimated beam apparatus was equipped with two low-pressure UV lamps emitting at 254 nm (Trojan UV Technologies, London, Ont., Canada). The fluence applied in the reactors was calculated as prescribed by the standard method using an International Light radiometer (Model IL1400A) coupled to a NIST calibrated sensor for a 254 nm wavelength. The collimated beams measured fluence rate was 0.040 mJ/cm2. The duration of exposure was adjusted according to the desired fluence. UV absorbance was measured using a spectrophotometer equipped with an integrating sphere (Variant, Cary 100, Victoria, Australia). Samples of 50 mL were irradiated in Petri dishes (9 cm). Three different irradiation sequences were applied to water samples: blending at 8000 rpm prior to UV exposure, blending at 8000 rpm following UV exposure, and 8 mm filtration prior to exposure followed by blending at 8000 rpm.

2.6.

Enumeration of spores

Indigenous spores of ASFB were enumerated using the Barbeau et al. (1997) method. Briefly, this method consists of filtration of the samples on a 0.45 mm membrane which is laid on a pad saturated with 1.5 mL of TSB. The Petri dishes are pasteurized at 75 1C for 15 min by immersing them in a water bath, and incubated for 24 h at 35 1C. Results are expressed as colony-forming units per specified volume. Analyses were conducted in triplicate.

2.7.

Data analysis

Spore counts resulting from various dispersion protocols are presented in the form of a release ratio normalized relative to the initial counts of a given control sample: blended with no chemicals to determine the best chemical mix, and not blended with the addition of the chemical mix for the trials on the optimization of the blending conditions. The use of a ratio allows the waters from experiments repeated over several sampling dates, during which the initial indigenous

ARTICLE IN PRESS 4550

WAT E R R E S E A R C H

41 (2007) 4546– 4556

ASFB spore concentrations varied, to be compared. Release ratios are calculated from each daily untreated control sample. Inactivation data are presented as the log function of N/N0, where N corresponds to the concentration of aerobic spores per 100 mL after treatment, dispersion and UV exposure. N0 is the concentration of spores per 100 mL prior to UV exposure, but after having been mixed with chemical solution (Zwittergent) and dispersed by blending. For the 8 mm filtered fraction, the ASFB spore concentration in the filtrate after blending is used as N0. Statistical analyses were performed using ANOVA Statistica, version 7, for the ANOVA Dunnet, general linear model (GLM) and w2. Differences are considered statistically significant for p-values o0.05, unless otherwise stated.

3.

Results and discussion

3.1.

Optimization of dispersion

The efficacy of various dispersion procedures was compared using water from the Mille Iˆles River, because of the abundance of ASFB spores and particulate matter found in that source water. Results for the selection of the chemical dispersant trials are displayed in Fig. 1. The original Camper procedure of chemical addition combined with blending yielded average ASFB spore counts slightly higher (by a factor 1.5) than those observed in non-dispersed controls. Using ANOVA statistical analysis, a significant increase in the ASFB spore release ratio of about 3 was confirmed using the higher concentration of Zwittergent 3–12 solution (100 mg/L), while differences between the other treatments were not found to be significant. Some of the ingredients of the Camper mix, particularly the peptone, affected the UV transmittance of the

4

Release ratio

Median 25%-75% Non-Outlier Range

3

3.2.

2

1

0 1

2

3 4 Chemical mix

water sample, which increased in absorbance by nearly fourfold. This interference and the lack of significant improvement in the release ratio led to the decision not to use these additional ingredients. The second step consisted of evaluating the impact of the duration and intensity of mechanical mixing. Various combinations of blending speeds and times were investigated using the optimal chemical mix (100 mg/L of Zwittergent 3–12). As shown in Fig. 2, no obvious trends were observed as a function of the intensity and duration of the mechanical shear stress between 8000 and 21,000 rpm. According to an ANOVA analysis, the only significant increases in colony counts from the control were observed for the samples blended at 8000 rpm for 4 and 5 min and at 21,000 rpm for 5 min. The highest release ratio of 1.5 suggests that at least 50% of the spores were aggregated. In Fig. 2, some of the average release ratios are slightly under 1, which is to be expected, considering the precision of the membrane filtration method for measuring ASFB. As described in Section 3.3, the dispersion protocol was not shown to affect the viability of B. subtilis spores. The results of our effort to determine the optimal blending speed are consistent in terms of range of release (from under 1 to 3), and their variability in terms of published results on the release of coliform bacteria from wastewater flocs. Parker and Darby (1995) reported a high release ratio of about 3 obtained with 1.5–3 min of blending at 19,000 rpm. Ormeci and Linden (2005) found that increases in the mixing speed to over 3500 rpm or extending the mixing time to more than 1 min did not result in greater release of coliform bacteria and could even decrease counts from wastewater flocs and particles. The variability of published dispersion results and the data presented here suggest that optimal dispersion conditions are highly source specific, and that the best protocol should be confirmed by experimental work. Although the results are not consistent for all rotational speeds tested, a blending duration of 4 min at 8000 rpm was chosen for optimal physical dispersion because it gave the highest recorded release ratio. Choosing the lower rotational speed also minimized foaming and temperature changes.

5

Fig. 1 – Release ratio of spores of aerobic spore-forming bacteria for different dispersion treatments: (1) water plus Camper’s mix not blended, (2) water plus Camper’s mix blended, (3) water plus Camper’s mix blended with 10 mg/L of Zwittergent 3–12, (4) water blended with 10 mg/L of Zwittergent 3–12 and (5) water blended with 100 mg/L of Zwittergent 3–12. Results constitute an aggregation from four sampling campaigns (n ¼ 12).

Particle breakdown

The effect of blending on PSD in water fractions was investigated using Microflow DPA (Brightwell Technologies Inc.). The effects of filtration and dispersion are presented in Figs. 3a and b, which show the PSD for the three treatments and the two water types (Mille Iˆles River and St. Lawrence River). To facilitate the comparison of particle concentrations in the two source waters, counts were combined into three classes (5–10, 10–20 and 420 mm), which are plotted in Fig. 4. The majority of particles were smaller than 5 mm (90–98%). Particles of this size, even if very numerous, are unlikely to shield organisms of the size of spores or oocysts. The cleaner source, the St. Lawrence River water, contained fewer particles of all size ranges, with an average total particle count of 51,886 particles/mL, as compared with an average total count of 357,076 particles/mL for the Mille Iˆles water. A statistical analysis was conducted on the data aggregated in 18 classes of counts from a diameter of o3 mm up to diameter of 420 mm, by increments of 1 mm. Because the distribution of

ARTICLE IN PRESS WAT E R R E S E A R C H

4551

4 1 (200 7) 454 6 – 455 6

Release ratio

2

1

Median 25%-75% Non-Outlier Range

8,000 RPM

15,400 RPM

18,700 RPM

-5min

-4min

-3min

-2min

-5min

-4min

-3min

-2min

-5min

-4min

-3min

-2min

-5min

-4min

-3min

-2min

0

21,000 RPM

Treatment Fig. 2 – Release ratio of spores of aerobic spore-forming bacteria for different blending treatments. Four speeds were tested: 8000, 15,400, 18,400 and 21,000 rpm for 2, 3, 4 and 5 min. Results represent four replicate sampling campaigns (n ¼ 12).

aggregated PSD is non-parametric, a w2 analysis was used, which revealed that blending and filtration resulted in statistically significant differences in PSD for each class of particles larger than 5 mm in diameter. Blending had a more pronounced impact on large particles in the St. Lawrence River source, with reductions greater than 85% for particles larger than 10 mm. Filtration reduced the number of particles exceeding 10 mm by about 96.3%. Filtration also removed particles smaller than 8 mm in diameter (filter pore size), most probably because of the decreased porosity of the cake formed on the filter as suspended material accumulated. By contrast, even though the number of particles larger than 8 mm was significantly reduced, the filtrate still contained particles of diameters above 8 mm, possibly formed through post-filtration orthokinetic aggregation (Gregory, 2006). These results are in agreement with previous studies in which the effect of blending on PSD was observed, and it was reported that blending caused fragmentation of large particles of 10–40 mm (Parker and Darby, 1995).

3.3.

Effect of blending

According to Ormeci and Linden (2005), two phenomena occur simultaneously during physical extraction: detachment as a result of shear stress, which increases the counts, and the disturbance of bacteria, which reduces the viable counts. Of primary importance is verification of the impact of the dispersion procedure on the viability of the spores, and also on their susceptibility to UV. A suspension of B. subtilis spores (ATCC 6633) was subjected to the dispersion procedure. B. subtilis spores can be considered a good surrogate for environmental ASFB spores, since B. subtilis is the most common genus among indigenous ASFB (Nieminski et al., 2000). Results demonstrate the absence of significant differences between counts of blended and non-blended samples,

whether or not they are exposed to UV light, at a fluence of 20 mJ/cm2. Thus, the dispersion protocol did not impact spore survival, nor did it reduce the resistance of spores to UV radiation. No spores survived at higher fluences (40 and 60 mJ/cm2), confirming previous observations that B. subtilis (ATCC 6633) is more susceptible to UV light than indigenous aerobic spores in surface water samples. Variability in spore sensitivity has been observed: (1) within the same strain of ASFB spores, depending on the sporulation conditions (Severin et al., 1983; Nicholson and Law, 1999) and (2) between various isolated and cultivated environmental strains (Mamane-Gravetz and Linden, 2005). A 2-log inactivation was observed at a fluence of 20 mJ/cm2, whereas indigenous spores required much higher fluences for a 2-log inactivation (Figs. 5a and b).

3.4.

Impact of particles on UV inactivation

Figs. 5a and b present the kinetic UV inactivation curves for the three water fractions (raw, dispersed and filtered) combined for all sampling dates. The inactivation curves are described by linear models that follow the Chick–Watson law. Results reveal significant differences between the inactivation rates of the raw (0.018 log cm2/mJ), dispersed (0.026 log cm2/mJ) and filtered (0.033 log cm2/mJ) samples of the Mille Iˆles River water. Figs. 3 and 4 reveal the extent of the reduction in the number of particles of all sizes following blending or filtration, and a significant decrease in the number of particles larger than 5 mm. The shift brought about by blending (loss of 46–52% of particles 410 mm) causes an increase in the rate of inactivation, which translates into a significant decrease in the required fluence for a given log inactivation level. Differences between treatments were not significant for the St. Lawrence water, and the inactivation rates for this water varied from 0.034 to 0.040 log cm2/mJ. Although not

ARTICLE IN PRESS 4552

WAT E R R E S E A R C H

41 (2007) 4546– 4556

nation of the two. This was estimated through the parameter ‘‘m’’ from a parallel Chick–Watson model:

1000000

h i log10 ðN=N0 Þ ¼ log10 ð1  mÞ10ks H þ m10ka H ,

Particles / mL

100000 Non-dispersed water Dispersed water 8 μm Filtered water

10000 1000 100 10 1 <3

5

7

9 11 13 15 17 Particle diameter (μm)

19

>20

19

>20

1000000

Particles / mL

100000 10000

Non-dispersed water Dispersed water 8 μm Filtered water

1000 100 10 1 <3

5

7

9 11 13 15 17 Particle diameter (μm)

Fig. 3 – Distribution of particle size in (1) untreated raw water, (2) blended water at a speed of 8000 rpm with 100 mg/L of Zwittergent 3–12 and (3) filtered on 8 lm. Graph (a) shows averaged data for five Mille Iˆles River sampling dates; graph (b) shows averaged data for five St. Lawrence River sampling dates.

significantly different, the inactivation rate for the filtered fraction was 15% higher (0.039 log/mJ/cm2). Several factors might explain the limited impact of particles in this water: (1) fewer particles: particle concentrations are much lower in the St. Lawrence River source water than in the Mille Iˆles River source water, especially particles larger than 5 mm (21% of the counts for the Mille Iˆles source); (2) fewer ASFB spores: levels of indigenous ASFB spores in the St. Lawrence River water are almost ten times lower than in the Mille Iˆles River water. The lower spore concentration limits the maximum fluence before the detection limit reaches about 50 mJ/cm2. It is possible that some tailing could occur at higher fluences, but this could not be observed because of experimental limitations; (3) the smaller proportion of ASFB spores associated with particles: Chevrefils et al. (2006) evaluated the proportion of the spore population resistant to UV, either through their association with particles in the test waters or through increased resistance of subpopulations, or a combi-

where ks is the rate of inactivation of single spores (cm2/mJ), ka the rate of inactivation of aggregated or resistant spores (cm2/mJ), m the fraction of the resistant spore population or present as aggregates, and H the fluence (mJ/cm2). They report values between nil for the St. Lawrence River water and an average of 3%, with maximum values of 9%, for the Mille Iˆles River water. These values may correspond to a level of association estimated through a disinfection model if the lowered ASFB spore inactivation rate is a direct consequence of its association with particles. While this may be true, not all associated spores will be equally protected from the UV rays; a spore located on the surface of a larger particle that is rotating during exposure will be partially protected, as opposed to a spore that is completely inserted into a particle. These values are lower than the 3–24% values in particles (mean diameter 11–80 mm) reported for Enterobacteriacae in wastewater by Emerick et al. (1999). As the Mille Iˆles River is heavily impacted by sewage outflows and CSO discharge (Payment et al., 2000), it is possible that the higher level of association is the direct result of the influence of wastewater. Apart from the protective effect of particles and/or aggregation, tailing has also been considered as (i) an artifact of the heterogeneity of the population and (ii) a lack of precision in the estimation of the small number of survivors (Cerf, 1997; Loge et al., 2001). However, since indigenous spores are composed of a mix of species with variable sensitivities to UV light (Mamane-Gravetz and Linden, 2005), it is possible that tailing is associated with the inactivation of more resistant subpopulations. Population composition may also vary from day to day. More importantly, the conditions of sporulation for a given strain affect their sensitivity to UV (Severin et al., 1983). More recently, sporulation conditions (liquid versus solid) have been shown to induce significant differences in the UV sensitivity of B. subtilis spores (Bohrerova et al., 2006). These authors also report a large difference in UV sensitivity between batches of the same strain subjected to identical sporulation conditions. Although no data on spore speciation and sensitivity profiles were available, the examination of the data for 8 mm filtered samples reveals no tailing in any of the samples. This observation does not support the hypothesis related to the presence of a resistant subpopulation. The extent of this conclusion is limited by the fact that the spore concentrations in the filtrate were not sufficient to measure more than 3 log of inactivation, at which point the detection limit was reached. Therefore, we cannot rule out the possibility that tailing associated with a resistant subpopulation occurs, although it would only be beyond the maximal fluence tested in this project (4100 mJ/cm2). Fig. 6 illustrates the positive correlation observed between the number of particles with a diameter greater than 8 mm in the Mille Iˆles River water and the required fluence for 2-log inactivation (based on the kinetic model described by Chevrefils et al., 2006). The 8 mm diameter was chosen in accordance with reports that the threshold particle size beyond which bacteria could benefit from some level of

ARTICLE IN PRESS WAT E R R E S E A R C H

4553

4 1 (200 7) 454 6 – 455 6

30000

25000

20000

15000

10000

5000

0

> 20 μm

Untreated Blended MI MI

Filtered 8μm MI

Untreated SL

Blended SL

Filtered 8μm SL

383

184

24

334

491

7

10 - 20 μm

3 204

1712

72

1 428

209

49

5 - 10 μm

19 124

13 414

330

3148

1 543

543

Fig. 4 – Number of particles measured by DPA in (1) untreated raw water, (2) blended water at a speed of 8000 rpm with 100 mg/L of Zwittergent 3–12 and (3) filtered on 8 lm. MI—Mille Iˆles River water, SL—St. Lawrence River water. Data represent average values from five sampling dates.

protection ranges between 7 and 10 mm (Qualls et al., 1983; Jolis et al., 2001; Emerick et al., 1999; Cairns et al., 1993) and taking into account the size of the spores and Cryptosporidium oocysts. The observed correlation (r2 ¼ 0.63) value suggests that PSD is probably not the only factor influencing the inactivation kinetics. This correlation could also be impacted by the level of association of the spores with the particulate matter, which in turn is dependent on the nature of the particles present in the source water, their surface charges and their hydrophobic nature (Faille et al., 2002). Specifically, it is likely to express variability from (1) the innate resistance of indigenous spores and (2) the impact of the nature of the particles. The chemical and physical nature of the particle influences how UV light is absorbed or scattered. Thus, different particles may protect harbored organisms unequally (Cairns et al., 1993; Loge et al., 1999; Templeton et al., 2006). An identical analysis was performed using the required fluence for 1-log inactivation (instead of 2 log). Results indicate that there was no correlation between the fluence and the particle counts above 8 mm (data not shown). Most fluence values for 1-log inactivation are between 20 and 40 mJ/cm2, independent of particle counts above 8 mm. As more than 91% of the spores are not protected by particles, the lack of correlation is consistent with the inactivation of the fraction of the spore population not associated with particles in this source. However, reaching more than 2 log of inactivation most likely involves the reduction of a significant fraction of the spores associated with particles. The inactivation kinetics would then be dependent on the concentration of particles large enough to harbor ASFB bacteria. This conclusion is also consistent with published results suggesting that non-particle-associated coliform bacteria constitute

the dominant fraction in wastewater, and can therefore be readily inactivated by UV, even in the presence of particles (Ridgway and Olson, 1982; Ormeci and Linden, 2002). In this experiment, indigenous ASFB spores were used as surrogates for Cryptosporidium, given their similarity in size and because of the greater abundance of spores. The observed effects of particles on the inactivation of these spores cannot be directly transferred to the prediction of UV inactivation of Cryptosporidium. Two major factors must be taken into consideration when attempting to relate this study to the inactivation of Cryptosporidium: (1) The level of association of the organisms with the particulate matter and the likelihood of this association. This information appears critical to evaluating the potential health risks associated with shielding by particles in unfiltered water. The level of association probably varies from one source water to another, as well as among microorganisms. A significant impact of particles was observed for waters originating in the polluted Mille Iˆles River. A water source with such an elevated particle content would necessarily be coagulated and filtered before UV disinfection. Thus, the issue of particles protecting oocysts would not apply. However, particles from unfiltered protected sources are less likely to be significantly colonized by pathogenic organisms when compared with those from surface waters receiving wastewater discharges. Available results on the attachment of Cryptosporidium oocysts are contradictory. After mixing with secondary effluents of sewage water for 24 h, 70% of oocysts were found to be associated with particles (Medema et al., 1998), whereas another study investigating

ARTICLE IN PRESS 4554

WAT E R R E S E A R C H

8μm filtered Log I = -0.281-0.0307∗x r2 = 0.88 ∗ 2 Dispersed Log I = -0.2939-0.0254 x r = 0.90 Non-dispersed Log I = -0.2894-0.0178∗x r2 = 0.85

4.4 (Log # particles > 8μm) / mL

0

Inactivation (Log)

41 (2007) 4546– 4556

-1

-2

-3

2

R : 0.63, p = 0.006

4.2 4.0 3.8 3.6 3.4 3.2 3.0 0

-4 0

20

40

60

80

100

120

Inactivation (Log)

-1

-3

-4 60

80

80

100

120

140

160

The second safety factor relates to the experimental uncertainty associated with the validation assay. For a 3-log Cryptosporidium inactivation, the REDbias can vary from 1.0 to 3.94 (USEPA, 2006). Therefore, the apparent safety factor provided by the typical fluence of 40 mJ/cm2 used in the drinking water industry might be required to cover the uncertainty of the validation procedure, rather than the potential impacts of particles on UV efficacy. Optimizing the validation procedure can minimize the RED value, which should in turn allow the inclusion of a safety factor to account for the impact of particles on UV disinfection for unfiltered supplies.

-2

40

60

Fig. 6 – Linear regression between the log value of the concentration of particles larger than 8-lm in diameter, and the required fluence (mJ/cm2) to achieve 2-log of inactivation of ASFB spores in the Mille Iˆles River water. Dotted lines are 95% confidence intervals.

0

20

40

Fluence for 2 log inactivation (mJ/cm2)

140

Fluence (mJ/cm2)

0

20

100

120

140

Fluence (mJ/cm2) Fig. 5 – ASFB spore inactivation curves for water from (a) the Mille Iˆles River and (b) the St. Lawrence River. Data from five sampling dates were aggregated by treatment.

4. the association of oocysts with soil particles concluded that attachment was limited because of surface charge (Dai and Boll, 2003). The shielding potential of coagulant flocs could be significant because of the nature of the chemical particles and the fact that microorganisms are adsorbed, swept and imbedded into these flocs during coagulation (2) The spores of ASFB are very resistant to UV, making the investigation of a subtle but significant shielding effect possible over a wide range of fluences. However, Cryptosporidium oocysts are more sensitive than aerobic spores. For an inactivation goal of 3 log, the UV sensitivity of indigenous spores was around 10–12 mJ/cm2/log, while that of Cryptosporidium is 4.0 mJ/cm2/log (United States Environmental Protection Agency (USEPA), 2006). Using a higher fluence than the requirements based on UV sensitivity can provide a margin of safety. However, it is also important to mention that the recent USEPA UV validation protocol (USEPA, 2006) requires the inclusion of two safety factors. The first (REDbias), is intended to account for the differences in sensitivity between the target biodosimetric organism and the target pathogen.

Conclusions

Based on our findings, these conclusions can be inferred:

 A combination of low-speed mechanical blending









(8000 rpm for 4 min) and chemical dispersion (addition of 100 mg/L of Zwittergent 3–12) resulted in significant release of the spores from aggregates, with values up to 50% in water from the Mille Iˆles River. Increasing the blending speed did not improve the release of particleassociated spores. The dispersion protocol and the filtration step on 8 mm membranes significantly decreased the number of natural particles of all size ranges larger than 5 mm in diameter. The proposed dispersion protocol did not have an effect on the survival of B. subtilis ATCC 6633 spores and their susceptibility to UV at 254 nm. Particulate matter and aggregates significantly lowered the fluence-based inactivation rate of indigenous spores of ASFB in the source water with elevated particle content. When the desired level of inactivation of indigenous spores of ASFB exceeds 1 log, the necessary fluence is significantly correlated to the concentration of particles with a diameter greater than 8 mm.

ARTICLE IN PRESS WAT E R R E S E A R C H

4 1 (200 7) 454 6 – 455 6

The results bear some relevance to the inactivation of Cryptosporidium associated with natural particles. However, the extent of protection of Cryptosporidium oocysts will most likely depend on (i) the number of large particles in the source; (ii) the level of aggregation and/or association of the oocysts with the particulate matter; and (iii) the desired level of inactivation targeted. These variables are probably sourcewater dependent, and future work should address the characterization of the natural state of Cryptosporidium for various types of waters and investigate the possible link between these parameters and the observed inactivation kinetics.

Acknowledgments The authors would like to acknowledge the scientific contributions of Jacinthe Mailly at the NSERC Chair on Drinking Water and Dr. Bernard Cle´ment, professor at the E´cole Polytechnique de Montreal, for their support in the laboratory work and statistical analyses. They would also like to thank the staff of the Re´gie Intermunicipale Des Moulins municipal WTP and the Charles-DesBaillets WTP in Montreal, who provided the water samples. The project was funded by the NSERC Industrial Chair on Drinking Water and the Canadian Water Network. Chair partners included the City of Montreal, the City of Laval, John Meunier Inc. and the Natural Sciences and Engineering Council (NSERC). R E F E R E N C E S

Barbeau, B., Boulos, L., Desjardins, R., Coallier, J., Pre´vost, M., Duchesne, D., 1997. A modified method for the enumeration of aerobic spore-forming bacteria. Can. J. Microbiol. 43, 976–980. Batch, L.F., Schulz, C.R., Linden, K.G., 2004. Evaluating water quality effects on UV disinfection of MS2 coliphage. J. Am. Water Works Assoc. 96 (7), 75–87. Bohrerova, Z., Linden, K.G., 2006. Ultraviolet and chlorine disinfection of Mycobacterium in wastewater: effect of aggregation. Water Environ. Res. 78 (6), 565–571. Bohrerova, Z., Mamane, H., Ducoste, J.J., Linden, K.G., 2006. Comparative inactivation of bacillus subtilis spores and MS-2 coliphage in a UV reactor: implications for validation. J. Environ. Eng. 132 (12), 1554–1561. Cairns, W.L., Sakamoto, G., Comair, C.B., Gehr, R., 1993. Assessing UV disinfection of a physico-chemical effluent by medium pressure lamps using a collimated beam and pilot plant. In: Water Environment Federation Specialty Conference: Planning, Design and Operation of Effluent Disinfection Systems, Whippany, NJ, USA. Camper, A.K., LeChevallier, M.W., Broadaway, S.C., McFeters, G.A., 1985. Evaluation of procedures to desorb bacteria from granular activated carbon. J. Microbiol. Methods 3 (3–4), 187–198. Chang, J.C.H., Ossoff, S.F., Lobe, D.C., Dorfman, M.H., Dumais, C.M., Qualls, R.G., Johnson, J.D., 1985. UV inactivation of pathogenic and indicator microorganisms. Appl. Environ. Microbiol. 49 (6), 1361–1365. Christensen, J., Linden, K.G., 2003. How particles affect UV light in the UV disinfection of unfiltered drinking water. J. Am. Water Works Assoc. 95 (4), 179–189.

4555

Clancy, J.L., Hargy, T.M., Marshall, M.M., Dyksen, J.E., 1998. UV light inactivation of Cryptosporidium oocysts. J. Am. Water Works Assoc. 90 (9), 92–102. Craik, S.A., Uvbiama, R.D., 2005. Effect of aggregation on UV inactivation of microorganisms in filtered drinking water. In: American Water Works Association—Water Quality Technology Conference, Quebec City, Que., Canada. Craik, S.A., Weldon, D., Finch, G.R., Bolton, J.R., Belosevic, M., 2001. Inactivation of Cryptosporidium parvum oocysts using mediumand low-pressure ultraviolet radiation. Water Res. 35 (6), 1387–1398. Dai, X., Boll, J., 2003. Evaluation of attachment of Cryptosporidium parvum and Giardia lamblia to soil particles. J. Environ. Qual. 32 (1), 296–304. Emerick, R.W., Darby, J.L., 1993. Ultraviolet light disinfection of secondary effluents: predicting performance based on water quality parameters. In: Planning, Design and Operation of Effluent Disinfection Systems. Specialty Conference Series, Whippany, NJ, USA. Emerick, R.W., Loge, F.J., Thompson, D., Darby, J.L., 1999. Factors influencing ultraviolet disinfection performance. Part II: Association of coliform bacteria with wastewater particles. Water Environ. Res. 71 (6), 1178–1187. Emerick, R.W., Loge, F.J., Ginn, T., Darby, J.L., 2000. Modeling the inactivation of particle-associated coliform bacteria. Water Environ. Res. 72 (4), 432–438. Faille, C., Jullien, C., Fontaine, F., Bellon-Fontaine, M.-N., Slomianny, C., Benezech, T., 2002. Adhesion of Bacillus spores and Escherichia coli cells to inert surfaces: role of surface hydrophobicity. Can. J. Microbiol. 48 (8), 728–738. Feng, Y.Y., Ong, S.L., Hu, J.Y., Song, L.F., Tan, X.L., Ng, W.J., 2003. Effect of particles on the recovery of Cryptosporidium oocysts from source water samples of various turbidities. Appl. Environ. Microbiol. 69 (4), 1898–1903. Gregory, J., 2006. Particles in Water. Properties and Processes. Taylor & Francis Group, Boca Raton, FL, USA. Jolis, D., Lam, C., Pitt, P., 2001. Particle effects on ultraviolet disinfection of coliform bacteria in recycled water. Water Environ. Res. 73 (2), 233–236. Loge, F.J., Emerick, R.W., Thompson, D.E., Nelson, D.C., Darby, J.L., 1999. Factors influencing ultraviolet disinfection performance Part I: Light penetration to wastewater particles. Water Environ. Res. 71 (3), 377–381. Loge, F.J., Bourgeous, K., Emerick, R.W., Darby, J.L., 2001. Variations in wastewater quality parameters influencing UV disinfection performance: relative impact of filtration. Journal of Environ. Eng. 127 (9), 832–836. Mac Kenzie, W.R., Hoxie, N.J., Proctor, M.E., Gradus, M.S., Blair, K.A., Peterson, D.E., Kazmierczak, J.J., Addiss, D.G., Fox, K.R., Rose, J.B., Davis, J.P., 1994. A massive outbreak in Milwaukee of Cryptosporidium infection tansmitted through the public water supply. New Engl. J. Med. 331 (3), 161–167. Mamane, H., Linden, K.G., 2006a. Impact of particle aggregated microbes on UV disinfection. I: Evaluation of spore–clay aggregates and suspended spores. J. Environ. Eng. 132 (6), 596–606. Mamane, H., Linden, K.G., 2006b. Impact of particle aggregated microbes on UV disinfection. II: Proper absorbance measurement for UV fluence. J. Environ. Eng. 132 (6), 607–615. Mamane-Gravetz, H., Linden, K.G., 2004a. Impact of particle aggregated microbes on UV disinfection. In: American Water Works Association-Water Quality Technology Conference, San Antonio, Texas, USA. Mamane-Gravetz, H., Linden, K.G., 2004b. UV disinfection of indigenous aerobic spores: implications for UV reactor validation in unfiltered waters. Water Res. 38 (12), 2898–2906. Mamane-Gravetz, H., Linden, K.G., 2005. Relationship between physiochemical properties, aggregation and u.v. inactivation

ARTICLE IN PRESS 4556

WAT E R R E S E A R C H

41 (2007) 4546– 4556

of isolated indigenous spores in water. J. Appl. Microbiol. 98 (2), 351–363. Mamane-Gravetz, H., Linden, K.G., Cabaj, A., Sommer, R., 2005. Spectral sensitivity of Bacillus subtilis spores and MS2 coliphage for validation testing of ultraviolet reactors for water disinfection. Environ. Sci. Technol. 39 (20), 7845–7852. McTigue, N.E., Lechevallier, M., Arora, H., Clancy, J., 1998. National Assessment of Particle Removal by Filtration. American Water Works Association Research Foundation and American Water Works Association, Denver, CO, USA. Medema, G.J., Schets, F.M., Teunis, P.F.M., Havelaar, A.H., 1998. Sedimentation of free and attached Cryptosporidium oocysts and Giardia cysts in water. Appl. Environ. Microbiol. 64 (11), 4460–4466. Mofidi, A.A., Baribeau, H., Rochelle, P.A., De Leon, R., Coffey, B.M., Green, J.F., 2001. Disinfection of Cryptosporidium parvum with polychromatic UV light. J. Am. Water Works Assoc. 93 (6), 95–109. Nicholson, W.L., Law, J.F., 1999. Method for purification of bacterial endospores from soils: UV resistance of natural Sonoran desert soil populations of Bacillus spp. with reference to B. subtilis strain 168. J. Microbiol. Methods 35 (1), 13–21. Nieminski, E.C., Bellamy, W.D., Moss, L.R., 2000. Using surrogates to improve plant performance. J. Am. Water Works Assoc. 92 (3), 67–78. Ormeci, B., Linden, K.G., 2002. Comparison of UV and chlorine inactivation of particle and non-particle associated coliform. Water Sci. Technol.: Water Supply 2 (5-6), 403–410. Parker, J.A., Darby, J.L., 1995. Particle-associated coliform in secondary effluents: shielding from ultraviolet light disinfection. Water Environ. Res. 67 (7), 1065–1075. Passantino, L., Malley Jr., J.P., 2001. Impacts of turbidity and algal content of unfiltered drinking water supplies on the ultraviolet disinfection process. In: American Water Works Association— Annual Conference, Washington, DC, USA. Passantino, L., Malley Jr., J., Knudson, M., Ward, R., Kim, J., 2004. Effect of low turbidity and algae on UV disinfection performance. J. Am. Water Works Assoc. 96 (6), 128–137. Payment, P., Berte, A., Pre´vost, M., Me´nard, B., Barbeau, B., 2000. Occurrence of pathogenic microorganisms in the Saint Lawrence River (Canada) and comparison of health risks for populations using it as their source of drinking water. Can. J. Microbiol. 46 (6), 565–576. Qualls, R.G., Flynn, M.P., Johnson, J.D., 1983. The role of suspended particles in ultraviolet disinfection. J. Water Pollut. Control Fed. 55 (10), 1280–1285.

Qualls, R.G., Ossoff, S.F., Chang, J.C.H., Dorfman, M.H., Dumais, C.M., Lobe, D.C., Johnson, J.D., 1985. Factors controlling sensitivity in ultraviolet disinfection of secondary effluents. J. Water Pollut. Control Fed. 57 (10), 1006–1011. Ridgway, H.F., Olson, B.H., 1982. Chlorine resistance patterns of bacteria from two drinking water distribution systems. Appl. Environ. Microbiol. 44 (4), 972–987. Scheible, O.K., 1987. Development of a rationally based design protocol for the ultraviolet light disinfection process. J. WPCF 59 (1), 25–31. Severin, B.F., Suidan, M.T., Engelbrecht, R.S., 1983. Kinetic modeling of U.V. disinfection of water. Water Res. 17 (11), 1669–1678. Shin, G.-A., Linden, K.G., Arrowood, M.J., Sobsey, M.D., 2001. Lowpressure UV inactivation and DNA repair potential of Cryptosporidium parvum oocysts. Appl. Environ. Microbiol. 67 (7), 3029–3032. Sommer, R., Haider, T., Cabaj, A., Pribil, W., Lhotsky, M., 1998. Time dose reciprocity in UV disinfection of water. In: Water Science and Technology: Proceedings of the 1998 19th Biennial Conference of the International Association on Water Quality, Part 10 (of 10), vol. 38, no. 12, June 21–26 1998, pp. 145–150. Sommer, R., Cabaj, A., Sandu, T., Lhotsky, M., 1999. Measurement of UV radiation using suspensions of microorganisms. J. Photochem. Photobiol. B: Biol. 53 (1–3), 1–6. Templeton, M.R., Andrews, R.C., Hofmann, R., 2005. Inactivation of particle-associated viral surrogates by ultraviolet light. Water Res. 39 (15), 3487–3500. Templeton, M.R., Andrews, R.C., Hofmann, R., 2006. Impact of iron particles in groundwater on the UV inactivation of bacteriophages MS2 and T4. J. Appl. Microbiol. 101 (3), 732–741. United States Environmental Protection Agency (USEPA), 2006. Ultraviolet Disinfection Guidance Manual for Final Long Term 2 Enhanced Surface Water Treatment Rule. Office of Water (4601), Washington, DC, USA. Wobma, P.C., Bellamy, W.D., Malley, J.P., Reckhow, D.A., 2004. UV Disinfection and Disinfection By-products Characteristics of Unfiltered Water. American Water Works Association Research Foundation and City of Winnipeg, Denver, CO, USA. Zimmer, J.L., Slawson, R.M., Huck, P.M., 2003. Inactivation and potential repair of Cryptosporidium parvum following low- and medium-pressure ultraviolet irradiation. Water Res. 37 (14), 3517–3523.