- Email: [email protected]
Feasibility of the silver-UV process for drinking water disinfection
ARTICLE IN PRESS
Water Research 39 (2005) 4925–4932 www.elsevier.com/locate/watres
Feasibility of the silver-UV process for drinking water disinfection Michael A. Butkusa,, Mark Talbota, Michael P. Labareb a
Department of Geography and Environmental Engineering, Environmental Engineering Program, US Military Academy, West Point, New York 10996, USA b Department of Chemistry and Life Sciences, US Military Academy, West Point, New York 10996, USA Received 10 February 2005; received in revised form 13 September 2005; accepted 21 September 2005
Abstract A synergistic effect between cationic silver and UV radiation (silver-UV disinfection) has been observed that can appreciably enhance inactivation of viruses. The purpose of this work was to assess the feasibility of this technique for drinking water disinfection and evaluate the effects of selected impurities, found in fresh water, and common parameters on inactivation of the coliphage MS-2 with the silver-UV process. Turbidity (kaolin), calcium hardness, carbonate alkalinity, and pH did not significantly degrade inactivation. Inactivation was reduced in the presence of chloride, at concentrations greater than 30 mg/L, and in water samples with UV-254 absorbance values greater than ca. 0.1 cm1. Inactivation of MS-2 with silver-UV disinfection was also reduced at high phosphate concentrations (above ca. 5 mM). Silver-UV inactivation of MS-2 increased with increases in temperature between 10 and 20 1C. Silver-UV inactivation of MS-2 was increased by greater than 1-log over UV alone, in two untreated fresh water sources, which indicates that silver-UV may be a viable treatment technology. An assessment of operation and management costs suggests that an increase in inactivation of MS-2 with silver-UV disinfection could be economically beneficial. Published by Elsevier Ltd. Keywords: Disinfection; Silver; Ultraviolet radiation; MS-2; Silver-UV
1. Introduction Viruses are reported to be the most resistant pathogens to UV disinfection, they will likely impact future regulatory fluence requirements of this technology (Brown et al., 2002; Linden et al., 2002). A reduction in UV design fluence and subsequent capital and operating costs will make UV disinfection more appealing to municipalities that may wish to simultaneously reduce disinfection byproducts and improve inactivation Corresponding author. Tel.: +845 938 2820; fax: +845 938 3339. E-mail address: [email protected] (M.A. Butkus).
0043-1354/$ - see front matter Published by Elsevier Ltd. doi:10.1016/j.watres.2005.09.037
of protozoa. Methods for improving inactivation of viruses with UV radiation are likely to help the attainment of these objectives. Recently, Butkus et al. (2004) reported a synergistic effect between cationic silver (hereafter referred to as ‘‘silver’’) and UV radiation that can appreciably enhance inactivation of the coliphage MS-2, a viable surrogate for pathogenic viruses, (hereafter referred to as ‘‘MS-2’’). They reported that inactivation of MS-2 by UV radiation was increased by greater than 1.5 orders of magnitude (to a value of ca. 3.3-log) in the presence of silver (0.1 mg/L) in a phosphate buffered suspension. Should future regulatory goals for UV disinfection become focused on a 2log inactivation of viruses (Mackey et al., 2002), then a
ARTICLE IN PRESS 4926
M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
reduction in fluence may be possible if silver were used in conjunction with UV disinfection (Butkus et al., 2004). In water, at concentrations sufficient for bactericidal activity, silver does not impart taste, color, or odor and has no apparent detrimental effects on mammalian cells (Yahya et al., 1992). The only known negative health effect is argyria, an irreversible darkening of the skin and mucous membranes, which has been caused by prolonged exposure to high silver concentration (e.g. silver therapy). Accordingly, there is no United States Environmental Protection Agency (USEPA) primary drinking water standard for silver. Although preliminary findings are encouraging, previous studies were conducted in a 20 mM phosphate buffered system free of other impurities. It was hypothesized that some impurities commonly found in raw and treated drinking water may influence the efficacy of silver-UV disinfection. The purpose of this work was to assess the feasibility of this technique for drinking water disinfection and evaluate the effects of selected impurities, found in fresh water, and variations in common water quality parameters (e.g. pH and temperature) on inactivation of MS-2 with silver-UV disinfection. MS-2 was used in this study because it has been proposed as the benchmark for validation of fullscale UV reactors (Linden et al., 2002; Mackey et al., 2002) and as a surrogate for pathogenic enteric viruses such as Noroviruses (see Thurston-Enriquez et al., 2003).
2. Methods and materials 2.1. Reagents Silver stock solutions were prepared with silver nitrate (Alfa Aesar, Ward Hill, MA) or electrochemically (Superior Aqua Enterprises, Inc., Sarasota, FL). Unless stated otherwise, silver nitrate was used to prepare silver stock solutions and all reagents were purchased from Fisher Scientific (Fair Lawn, NJ). Silver neutralizer solution was prepared by combining 11.5 g sodium thiosulfate (J.T. Baker, Phillipsburg, NJ) and 5.0 g of sodium thioglycolate (Sigma, St. Louis, MO) with 50.0 mL of deionized water (DI) (Yahya et al., 1992). Total silver concentration was quantified using a colorimetic procedure (Hach, Loveland Colo.) with a method detection limit of 0.05 mg Ag/L. Fresh silver and silver neutralizer stock solutions were prepared on the day of each experiment. All glassware used to handle silver was soaked in 10% nitric acid overnight and rinsed in DI water to remove adsorbed silver (Chambers, 1956) and other contaminants prior to use. Phosphate buffer solutions (20 mM) were prepared by combining 5.678 g of Na2HPO4 with 2 L of distilled water and adding sufficient NaH2PO4 to obtain a pH of 7.2. Other
phosphate buffer concentrations were obtained using the same methodology. Stock solutions of humic acid were prepared by adding 0.1 g of humic acid, sodium salt (Acros Organics, NJ) to 100 mL phosphate buffer (20 mM, pH 7.2). Humic acid stock solutions were sonicated for several minutes and diluted with phosphate buffer to obtain the desired absorbance value. Kaolin and sodium chloride solutions were prepared in phosphate buffer or in DI water. Solutions containing calcium and magnesium hardness were prepared by adding calcium sulfate or magnesium sulfate to DI water. Samples containing alkalinity were prepared by adding sodium bicarbonate and disodium carbonate to DI water. Alkalinity and hardness solutions were measured by a colorimetic titration (Hach, Loveland Colo.). The influence of pH on inactivation was examined by adjusting the pH of DI water with sodium hydroxide (Hach, Loveland Colo.) or sulfuric acid (Hach, Loveland Colo.), recording pH, and sterilizing before addition of MS-2 and silver. Nutrients in raw water were quantified by filtering samples through 2 mm nylon filters (Millipore, Bedford, MA) and ion chromatography (Dionex, Sunnyvale, CA). Absorbance (254 nm) was quantified with a spectrophotometer (Varian, Walnut Creek, CA) using 1 cm path length, quartz cuvettes (Fisher Scientific, Fair Lawn, NJ). 2.2. Preparation of purified MS-2 A culture of Escherichia coli (ATCC 15597) was grown in tryptic soy broth (TSB; Difco Laboratories, Detroit, MI) at 37 1C and 150 rpm. Freeze dried MS-2 (ATCC 15597-B1) was mixed with 1.5 mL of a 24 h culture of E. coli and 3.0 mL of melted (45 1C) TSB soft agar (0.5% agar, w:v). The mixture was overlaid on TSB agar (1.5% agar, w:v) plates and incubated at 37 1C for 24 h. Six milliliters of 20 mM phosphate buffer was added to the plate and incubated for 1 h. The phosphate buffer was removed, passed through a 0.22 mm filter and the filtrate was used as the MS-2 stock suspension having an initial density of ca. 109 plaque forming units (pfu)/mL. 2.3. Collimated beam setup The collimated beam apparatus used in this study (Suntec Environmental, Concord, Ont., Canada) was modified to hold a stir plate and to allow for easy and reproducible vertical and horizontal adjustment. The two low-pressure mercury vapor lamps in the instrument were warmed up for at least 30 min before all experiments. Lamp irradiance was quantified with a UV detector (IL1400A, International Light, Newburyport, MA) by placing the detector at the same height as the sample surface. Fluence was determined by placing the detector in the integration mode following the removal
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
2.4. Irradiation of samples Samples were prepared by combining 100076 mL of MS-2 viral stock suspensions with 970.1 mL (8.970.1 mL for samples containing silver) of phosphate buffer or DI water (containing the impurity to be studied e.g. sodium chloride) in acid-washed, sterile Pyrex glass Petri dishes. All samples containing silver were prepared such that 10070.6 mL of a particular silver stock solution was added to the MS-2 suspensions in the Petri dishes to minimize effects of dilution. The total volume (without stir bar) and depth (with stir bar) of the viral stock suspensions in the Petri dishes were 10 mL and 0.6 cm. The incubation period for silver and MS-2, prior to and including the time of UV radiation exposure, was held at 10 min unless stated otherwise. Samples were irradiated at the end of the incubation period. For example, because a fluence of 40 mJ/cm2 was obtained at the typical lamp intensity (ca. 190 mW/cm2) by exposing the samples to the UV radiation for ca. 4 min, the viral suspensions were placed under the collimated beam following ca. 6 min of incubation with silver. Samples containing silver were neutralized (to halt disinfection) following the incubation period by adding 30 mL of neutralizer solution (Yahya et al., 1992). Samples were stirred slowly to prevent forming a vortex in the water (Bolton and Linden, 2003). Samples containing silver were incubated at ambient temperature (ca. 20 1C) or in an incubator for a predetermined time. Each experiment was conducted at least in triplicate. For all samples, a minimum of three dilutions were plated in triplicate using the standard double agar overlay technique as described above and elsewhere (Adams, 1959; Linden et al., 2002) using an E. coli (ATCC 15597) host grown at 3770.1 1C for 3–6 h (Yahya et al., 1992). Plates were incubated at 3770.1 1C and enumerated at 2471 h. The dilution giving the highest number of pfu less than 300 was averaged and used to obtain the MS-2 survival. Controls were conducted in triplicate and plated at various times during each experiment to ensure that conditions during the course of an
experiment did not influence the number of pfu in the stock suspensions.
3. Equilibrium modeling and curve fitting The predominant species of silver in the viral suspensions were determined via the MINEQL+ chemical equilibrium modeling system (version 4.5, Environmental Research Software, Hallowell, ME). This model employs a thermodynamic data base to compute equilibrium speciation in complex aqueous systems. All model calculations were conducted at 20 1C, in an open system, and included the influence of ionic strength. TableCurve 2D (Systat, Software, Inc., Richmond, CA) was used to obtain approximating functions for interpolation of X–Y data.
4. Results and discussion Microbial studies are frequently conducted in phosphate-buffered systems to mitigate changes in pH. Indeed, some of the work reported here and elsewhere (Butkus et al., 2004) on silver-UV disinfection was conducted in a 20 mM phosphate buffer system. The influence of phosphate and carbonate buffers on silverUV disinfection of MS-2 is presented in Fig. 1. The inactivation values plotted on the ordinate represent the log survival ratio, where the survival ratio is defined as the number of pfu following exposure to silver and/or UV light (Nt) divided by the number of pfu in the control samples (No). The data imply that inactivation of MS-2 with silver-UV disinfection was influenced by
6 5
-Log (Nt/No)
of a shutter and recording the required exposure time. It was determined that variations in lamp irradiance across the surface of the samples were negligible by moving the detector in the horizontal plane at distances equivalent to the sample surfaces, which resulted in a Petri factor (PF) of unity (Butkus et al., 2004). Variations in fluence resulting from drift in lamp output were typically less than ca. 0.5%. However, lamp output was verified periodically during the course of every experiment to compensate for slight changes resulting from drift. The average fluence ðmW s=cm2 Þ or ðmJ=cm2 Þ was determined as described elsewhere (Bolton and Linden, 2003; Butkus et al., 2004).
4927
4 3 2 phosphate buffer carbonate buffer
1 0 0.00001
0.0001
0.001 0.01 Buffer Concentration (M )
0.1
1
Fig. 1. Inactivation of MS-2 as a function of buffer concentration. All samples were exposed to silver (0.1 mg/L for ca. 6 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 10 min of total silver exposure. Error bars represent one standard deviation.
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
the presence of phosphate. At phosphate concentrations below ca. 5 mM, inactivation was comparable to silverUV disinfection in DI water (4.3-log). However, inactivation was reduced by ca. 50% in a 20 mM phosphate buffer. Weak aqueous complexes and precipitates of silver and phosphate (Ag(H2PO4) 2, 2 AgHPO 4 , Ag(HPO4H2PO4) , AgH2PO4, and Ag3PO4(s)) have been reported in the literature (Baldwin, 1969; Hietanen et al., 1973; Ciavatta et al., 1996). Predictions from the MINEQL+ chemical equilibrium modeling system suggest that silver forms the Ag(HPO4H2PO4)2 complex, but other silver–phosphate and silver–carbonate species (not shown) are negligible under these conditions. Carbonate did not reduce inactivation at similar concentrations and conductivities (data not shown), which suggests that the effect of phosphate was not caused by ionic strength effects. The mechanism for the observed effect of phosphate and weak silver–phosphate complexes on silver-UV disinfection remains unknown. Phosphate concentrations in drinking water are likely to be much lower than 20 mM with the greatest phosphate concentrations expected in systems where it is added for corrosion control. Kawamura (2000) reported that typical zinc phosphate dosages are on the order of 1 mg/L (6 mM phosphate) when it is used for corrosion control. Thus, the negative influence of phosphate on silver-UV disinfection is not expected to be significant under typical water treatment conditions. Moreover, use of concentrated phosphate buffers in bench scale studies designed to evaluate silver-UV disinfection may result in inactivation values that are substantially lower than those produced with typical raw water sources. Silver-UV disinfection was evaluated in DI water and phosphate buffer to demonstrate the influence that phosphate may on future bench scale studies. Silver-UV inactivation was relatively unaffected by carbonate and bicarbonate, which suggests that carbonate alkalinity should not influence the efficacy of silver-UV disinfection. It was hypothesized that natural organic matter might influence the efficacy of silver-UV disinfection. Humic acid was used as a surrogate for natural organic matter as a first approximation. Fig. 2 illustrates the influence of humic acid, reported as absorbance, on inactivation by silver-UV disinfection and UV radiation alone in a 20 mM phosphate buffer. Due to the complex nature of humic acid, it was hypothesized that if a reaction between silver and humic acid were to occur, its rate might be slower than silver and other common inorganic drinking water impurities (e.g. silver and chloride). Consequently, samples containing silver and humic acid were incubated for 60 min. Because trends in data for silver-UV inactivation and UV radiation alone are congruent, it was presumed that absorbance of UV light (and subsequent reduction in fluence) was the
4.5 Ag-UV UV Silver-UV Model UV Model
4 3.5 -Log (Nt/No)
4928
3 2.5 2 1.5 1 0.5 0 0.01
0.1
1
10
Absorbance (cm-1)
Fig. 2. Inactivation of MS-2 as a function of humic acid concentration. All samples were exposed to silver (0.1 mg/L for ca. 56 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 60 min of total silver exposure. The predicted log survival ratio as a function of absorbance was obtained by entering the corrected fluence from Eq (1) into an approximating function fit to log survival ratio vs. fluence, for a silver concentration of 0.1 mg/L. Error bars represent one standard deviation.
predominant factor responsible for the observed reduction in inactivation as a function of absorbance (humic acid concentration). The influence of UV light absorbance on average fluence ðmW s=cm2 Þ or ðmJ=cm2 Þ was modeled as follows (Bolton and Linden, 2003): 1 10a L Fluence ¼ 0:975 E o PF DF t (1) a L lnð10Þ where, Eo is the lamp intensity at the center of the sample dish (mW/cm2), a the sample absorption coefficient (cm1), L the depth of sample (cm), t the exposure time (s), and DF is the divergence factor. The divergence factor was determined as follows (Bolton and Linden, 2003): DF ¼
Z , ZþL
(2)
where Z is the distance from the lamp to the sample surface (cm). According to the set up of the collimated beam apparatus: L ¼ 0.6 cm and Z ¼ 86 cm. The predicted log survival ratio as a function of absorbance was obtained by entering the corrected fluence (from Eq. (1)) into an approximating function fit to log survival ratio vs. fluence, for a silver concentration of 0.1 mg/L (see Butkus et al., 2004 for similar data). Because predictions made with this approach fit the data well at typical drinking water absorbance values (abso0.5 cm1), absorbance of UV light energy appears to be the predominant mechanism of reduction in inactivation at typical drinking water treatment absorbance values.
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
Differences between selected samples incubated for 10 and 60 min were not statistically different (po0:054). This suggests that incubation time between silver and humic acid did not have a salient effect on inactivation, which supports the mechanism proposed above. Although humic acid does not appear to have a significant effect on silver-UV disinfection, additional studies with other forms of natural organic matter may be warranted. Precipitation of silver chloride is well known and it was presumed that aqueous chloride would reduce the effectiveness of silver-UV disinfection. Fig. 3 illustrates the influence of chloride (added as sodium chloride) in both DI water and in 20 mM phosphate buffer. Trends in the data support the hypothesis that chloride reduced the efficacy of the synergistic effect between silver and UV radiation at chloride concentrations greater than ca. 30 mg/L (in DI water). Nevertheless, silver-UV inactivation was ca. 0.5-log greater than for UV alone (2.1-log in DI water) at the USEPA Secondary Standard for chloride (250 mg/L). This finding suggests that chloride concentrations in typical drinking water supplies should not limit this disinfection technology. Reduction in MS2 inactivation, as a function of chloride concentration, was more pronounced in the concentrated phosphate buffer. Data from controls (not shown) demonstrated that similar chloride concentrations did not influence the infectivity of MS-2. The effect of silver–chloride precipitation, on cationic silver concentration and subsequent silver-UV inactivation, was modeled with the MINEQL+ chemical equilibrium modeling system in DI water and phosphate buffer. Because silver chloride precipitation is consid-
ered rapid, it was assumed that equilibrium was achieved under the conditions of this study. The predicted log survival ratio as a function of chloride (aqueous silver concentration of 0.1 mg/L prior to precipitation) in Fig. 3, was obtained by entering the cationic silver concentration predicted by MINEQL+, following precipitation, into an approximating function fit to log survival ratio vs. silver concentration (UV fluence of ca. 40 mJ/cm2) (see Butkus et al., 2004 for similar data). This modeling approach generally underestimated inactivation of MS-2 at chloride concentrations less than ca. 150 mg/L. Because model predictions underestimated inactivation, it is presumed that cationic silver may not be the only silver species that contributed to silver-UV inactivation of MS-2 under the conditions evaluated here. The bioaccumulation and toxicity of silver complexes have been reported elsewhere (Butkus et al., 2003; Gupta et al., 1998; Reinfelder and Chang, 1999; Bury and Hogstrand, 2002). The data in Fig. 4 demonstrates that increased temperature in the mesophilic range increased inactivation of MS-2 by silver-UV disinfection. For reasons that are not yet known, reduction in MS-2 inactivation at lower temperatures was more pronounced in the concentrated phosphate buffer. Inactivation was reduced by approximately 0.7-log at the lowest temperature evaluated (10 1C) in DI water, which suggests that the combination of silver and UV radiation in cold water was still more effective than the sum individual effects of silver and UV radiation alone reported by Butkus et al. (2004). The influence of additional impurities and parameters on the synergistic effect between UV light and silver are presented in Table 1. Turbidity (10.2 NTU) resulting
5
5
Exp (PO4) Exp (DI) Eqlb Model (phosphate) Eqlb Model (DI)
4.5 4
DI phosphate
4.5 4
3.5
3.5
3
-Log (Nt/No)
-Log (Nt/No)
4929
2.5 2 1.5 1
3 2.5 2 1.5
0.5
1
0 0
100 200 300 Chloride Concentration (mg/L)
400
0.5 0 8
Fig. 3. Inactivation of MS-2 as a function of chloride concentration. All samples were exposed to silver (0.1 mg/L for ca. 6 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 10 min of total silver exposure. The model predicts the effect of chloride on cationic silver concentration and subsequent silver-UV disinfection. Error bars represent one standard deviation.
10
12
14 16 18 Temperature (°C)
20
22
Fig. 4. Inactivation of MS-2 as a function of temperature. All samples were exposed to silver (0.1 mg/L for ca. 6 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 10 min of total silver exposure. Error bars represent one standard deviation.
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
3.5
0.17
3.1
0.17
4.0
0.10
No difference in inactivation (po0.05)
from a suspension of kaolin in 20mM phosphate buffer did not influence the synergistic effect between UV radiation and silver. Passantino et al. (2004) reported that comparable turbidity values (10.5 NTU) did not influence inactivation of MS-2 resulting from exposure to UV radiation. Calcium hardness did not influence inactivation of MS-2, yet magnesium hardness decreased the inactivation of MS-2. The influence of pH on inactivation was examined by adjusting the pH of DI water with NaOH or H2SO4. Initial pH values ranged from 4.2 to 10.2 and no significant difference (po0:05) in log survival ratio was observed when compared with inactivation in DI water (pH ca. 5.7). Similar experiments in phosphate buffer indicated that inactivation was not significantly reduced (po0:05) when pH was varied between 6 and 9 when compared to inactivation in phosphate buffer at pH 7.2. The influence of impurities on silver-UV disinfection was evaluated with two fresh water sources: Lusk Reservoir, a source for drinking water at the US Military Academy, West Point, NY, and Hudson River water, sampled at West Point, NY. Both samples were adjusted to a temperature of 20 1C, amended with MS-2, and used as challenge water without additional pretreatment. Use of untreated samples was considered a conservative approach because conventional treatment processes would likely reduce some impurities before disinfection. The characteristics of these sources are presented in Table 2. Inactivation of MS-2 in the raw water suspensions, as a function of disinfection method, is presented in Fig. 5. Controls for each source indicated that the MS-2 was not affected by impurities in the raw water prior to disinfection. Disinfection of each suspension with UV alone resulted in comparable results for DI water, Lusk Reservoir, and Hudson River sources. Addition of silver before exposure to UV resulted in
7.13 54.0 248 76.0 23.3 0.149
6.67 17.0 103 24.0 0.76 0.065
0.087
0.057
29.0 3.2
Sulfatea (mg/L) Fluoridea Nitritea Phosphatea Bromidea
8.1 ND ND ND ND
13.8 Not detected (ND) 13.7 ND ND ND ND
a
Samples were filtered through a 2 mm nylon filter prior to analysis.
5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 silver-UV in Hudson
0.47 0.21
pH Alkalinity (mg CaCO3/L) Conductivity (mS/cm) Hardness (mg CaCO3/L) Turbidity (NTU) Absorbance (254 nm; unfiltered) Absorbance (254 nm; 0.45 mm filter) Chloridea (mg/L) Nitratea (mg/L)
silver-UV in Lusk
4.3 3.4
Lusk Reservoir
UV in Hudson
Standard deviation
Hudson River
UV in Lusk
DI water Phosphate buffer (20 mM, pH 7.2) Turbidity (10.2 NTU in 20 mM phosphate buffer) Mg hardness (300 mg/L as CaCO3 in DI water) Ca hardness (300 mg/L as CaCO3 in DI water) pH (6–9)
Log (N/No)
Parameter
silver-UV in DI
Treatment
Table 2 Challenge water characteristics
UV in DI
Table 1 Influence of selected water quality parameters on inactivation of MS-2 by silver (0.1 mg/L) and UV light (40 mJ/cm2) given 10 min of contact time
-Log (Nt/No)
4930
Treatment
Fig. 5. Inactivation of MS-2 in Lusk Reservoir water, West Point, NY, and Hudson River water, sampled from the shoreline in West Point, NY. Both samples were adjusted to a temperature of 20 1C, spiked with MS-2, and used as challenge water without additional pretreatment. Water quality characteristics are presented in Table 1. All samples were exposed to silver (0.1 mg/L for ca. 6 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 10 min of total silver exposure. Error bars represent one standard deviation.
substantially improved inactivation in all suspensions. (Using 0.1 mg/L of silver without UV radiation, Butkus et al. (2004) reported that a measurable inactivation of MS-2 was not observed for 10 min of contact time in phosphate buffer.) Silver-UV inactivation of MS-2 in
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
4.1. Economic analysis An economic analysis of silver-UV disinfection has not yet been reported. A brief evaluation of operation and maintenance (O&M) cost is presented here to provide a jumping-off point for a more comprehensive assessment on this topic. Because O&M costs associated with electricity and silver electrodes are frequently the predominate cost for silver disinfection systems (personal communication: Mr. James Muha, Superior Aqua Enterprises, Inc., Sarasota, FL) other costs were not considered in this preliminary analysis. It was assumed that 1 kw-h of electricity could produce enough silver electrochemically, at a concentration of 0.1 mg/L, to treat 79,040 gallons of water (personal communication: Mr. James Muha, Superior Aqua Enterprises, Inc., Sarasota, FL); the cost of silver is $318/kg; and, the cost of electricity is $0.08/kw-h. Based on these figures, the silver electrode replacement cost was determined to be more than an order of magnitude greater than the electricity cost. This suggests that the price of silver generally controls the economics of disinfection with silver. A volumetric flow rate of 0.12 ML/d was used in
4000 Annual O&M Cost ($/yr)
the Lusk Reservoir water was consistent with inactivation results for DI water. This was expected because chloride (based on data obtained in DI water), phosphate, and absorbance values were not sufficient to appreciably reduce inactivation (cf. Figs. 1–3). However, inactivation of MS-2 in the Hudson River sample was lower than what was expected based on absorbance (inactivation was reduced by ca. 0.1-log at absorbance values of 0.1 cm1, cf. Fig. 2), chloride, and phosphate. It is presumed that other impurities in the water may have reduced the synergistic effect of silverUV disinfection. For example, natural organic matter may have formed complexes with silver or high levels of magnesium may have been present. Identification of additional impurities that influence silver-UV disinfection could facilitate the development of a pretreatment regime that could further optimize this technology. The upshot is that there was a substantial improvement in inactivation of MS-2 with silver-UV disinfection in untreated Lusk Reservoir water and Hudson River water, which suggests that this technique could be effective in numerous systems. Because there are health concerns associated with nitrate, silver is often generated electrochemically for disinfection. A 3.4-log inactivation of MS-2 was observed when silver was generated electrochemically in Lusk Reservoir water and used at the same concentration as silver formed from silver nitrate (0.1 mg/L), given a total of 10 min contact time with silver and a fluence of ca. 40 mJ/cm2. This finding suggests that silver-UV disinfection can be accomplished without increasing nitrate levels in the treated water.
4931
3500 3000 2500 2000 1500 1000 500 0 2
2.5
3
3.5
4
4.5
-Log (Nt/No)
Fig. 6. Increase in O&M cost as a function of increase in log survival ratio resulting from increased silver concentration given a UV fluence of ca. 40 mJ/cm2 and annual average volumetric flow rate of 0.12 ML/d. The O&M cost for disinfection with UV radiation was adapted from Cotton et al. (2001).
this analysis because UV O&M cost data was available for this flow rate in the literature (Cotton et al., 2001) and silver-UV disinfection appears to be most advantageous at lower flow rates (see below). Fig. 6 presents increase in annual O&M cost as a function of increase in log survival ratio, resulting from increased silver concentration, for silver-UV disinfection (UV fluence of 40 mJ/cm2 and an annual average volumetric flow rate of 0.12 ML/d). The annual O&M cost for disinfection with UV radiation ($1800) was adapted from Cotton et al. (2001) for UV fluence of 40 mJ/cm2 and the same average volumetric flow rate and electricity cost. The figure illustrates that inactivation can be increased from 2-log to 3-log with a concomitant increase in silver-UV O&M cost of ca. $200 (11% increase in cost). Butkus et al. (2004) reported that 2-log inactivation of MS-2, using UV only, required ca. 46 mJ/ cm2 and 3-log inactivation required ca. 70 mJ/cm2 (57% increase in fluence). Although a 57% increase in fluence does not necessarily imply that there will be a 57% increase in UV power cost (Cotton et al., 2001), this limited evaluation suggests it is plausible to assume that silver-UV disinfection could be economically viable at some flow rates. Due to the non-linear behavior of UV O&M costs (Cotton et al., 2001), the O&M cost savings of silver-UV may be less favorable at higher volumetric flow rates. In addition to lowering UV fluence requirements for a target inactivation, residual silver might also allow for lower residual chlorine concentrations in distribution systems because of the reported synergy between silver and chlorine (Yahya et al., 1992). Lower required chlorine concentrations in distributions systems might result in lower O&M costs for chlorine (and fewer
ARTICLE IN PRESS 4932
M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
disinfection byproducts) making silver-UV disinfection more attractive. A more detailed economic assessment considering capital costs; ranges of volumetric flow rates, silver concentrations, silver costs, electricity costs and chlorine costs; and, UV cost as a function of fluence is needed before silver-UV disinfection can be recommended as an economically viable alternative. ‘‘Because silver-UV disinfection has been reported to inactivate a DNA virus, Haemophilus influenzae phage T4D (Rahn and Landry, 1973), it is expected that the synergistic effect between silver and UV radiation might also exist for the inactivation of pathogenic viruses such as poliovirus, Noroviruses viruses, and the enteric adenoviruses 40 and 41.’’ However, the efficacy of this technique for treatment of other pathogenic viruses warrants further investigation.
5. Conclusions The results presented here suggest that the silver-UV disinfection technique may be effective for many fresh water supplies (as well as for use in domestic wastewater treatment, hospital hot water systems, pools, and spas). Silver-UV inactivation of MS-2 was influenced by chloride, absorbance, magnesium, temperature, and phosphate. Because appreciable inactivation of MS-2 was observed with two untreated (except for temperature) fresh water sources, silver-UV disinfection may be a viable treatment technology. An assessment of O&M costs suggests that an increase in inactivation of MS-2 with silver-UV disinfection could be economically beneficial.
Acknowledgments The authors gratefully acknowledge the cost information and insight provided by Richard Peters, Hazen and Sawyer, New York, NY; Christine Cotton, Malcolm Pirnie, Tucson, AZ; and Mr. James Muha, Superior Aqua Enterprises, Inc., Sarasoto, FL. The technical assistance of Anand Shetty, Department of Geography and Environmental Engineering, United States Military Academy is also greatly appreciated. This project was supported by grants from The Army’s PM Soldier Systems Program.
References Adams, M.H., 1959. Bacteriophages. Intersciences, New York, N.Y. Baldwin, W.G., 1969. Phosphate equilibria. II. Studies on the silver-phosphate electrode. Arkiv foer Kemi 31, 407–414.
Bolton, J.R., Linden, K.G., 2003. Standardization of methods for fluence (UV dose) determination in bench-scale UV experiments. J. Environ. Eng. 129, 209–215. Brown, N.P., Jacangelo, J.G., Haas, C.N., Gerba, C.P., 2002. Assessment of existing disinfection practices for inactivation of emerging pathogens. In: American Water Works Association Annual Conference Proceedings, New Orleans, LA. Bury, N.R., Hogstrand, C., 2002. Influence of chloride and metals on silver bioavailability to Atlantic salmon (Salmo salar) and Rainbow trout (Onchorhynchus mykiss) yolk-sac fry. Environ. Sci. Technol. 36 (13), 2884–2888. Butkus, M.A., Edling, L., Labare, M.P, 2003. The efficacy of silver as a bactericidal agent: advantages, limitations, and considerations for future use. J. Water Supply: Res. Technol.—AQUA 52 (6), 407–415. Butkus, M.A., Labare, M.P., Starke, J.A., Moon, K., Talbot, M., 2004. Use of aqueous silver to enhance the inactivation of coliphage MS-2 with UV disinfection. Appl. Environ. Microbiol. 70 (5), 2848–2853. Chambers, C.W., 1956. A procedure for evaluating the efficiency of bactericidal agents. J. Milk Food Technol. 19, 183–187. Ciavatta, L., Iuliano, M., Porto, R., 1996. Complex formation equilibria between silver(I) and orthophosphate ions. Ann. Chim. 86, 411–427. Cotton, C.A., Owen, D.M., Cline, G.C., Brodeur, T.P., 2001. UV disinfection costs for inactivating cryptosporidium. J. AWWA 93 (6), 82–94. Gupta, A., Maynes, M., Silver, S., 1998. Effects of halides on plasmid-mediated silver resistance in E. coli. Appl. Environ. Microbiol. 64, 5042–5045. Hietanen, S., Sille´n, L.G., Ho¨gfeldt, E., 1973. On Some phosphate equilibria. Chem. Scripta 3, 23–29. Kawamura, S., 2000. Integrated Design and Operation of Water Treatment Facilities. Wiley, New York. Linden, K.G., Batch, L., Schulz, C., 2002. UV disinfection of filtered water supplies: water quality impacts on MS2 dose–response curves. In: American Water Works Association Annual Conference Proceedings, New Orleans, LA. Mackey, E.D., Hargy, T.M., Wright, H.B., Malley Jr., J.P., Cushing, R.S., 2002. Comparing Cryptosporidium and MS2 bioassays–implications for UV reactor validation. J. AWWA 94, 62–69. Passantino, L., Malley Jr., J., Knudson, M., Ward, R., Kim, J., 2004. Effect of low turbidity and algae on UV disinfection performance. J. AWWA 96 (6), 128–137. Rahn, R.O., Landry, L.C., 1973. Ultraviolet irradiation of nucleic acids complexed with heavy atoms—III. Influence of Ag+ and Hg2+ on the sensitivity of phage and of transforming DNA to ultraviolet radiation. Photochem. Photobiol. 18, 39–41. Reinfelder, J.R., Chang, S.I., 1999. Speciation and microalgal bioavailability of inorganic silver. Environ. Sci. Technol. 33 (11), 1860–1863. Thurston-Enriquez, J.A., Haas, C.N., Jacangelo, J., Riley, K., Gerba, C.P., 2003. Inactivation of feline calicivirus and adenovirus type 40 by UV radiation. Appl. Environ. Microbiol. 69, 577–582. Yahya, M.T., Straub, T.M., Gerba, C.P., 1992. Inactivation of coliphage MS-2 and poliovirus by copper, silver, and chlorine. Can. J. Microbiol. 38, 430–435.
Water Research 39 (2005) 4925–4932 www.elsevier.com/locate/watres
Feasibility of the silver-UV process for drinking water disinfection Michael A. Butkusa,, Mark Talbota, Michael P. Labareb a
Department of Geography and Environmental Engineering, Environmental Engineering Program, US Military Academy, West Point, New York 10996, USA b Department of Chemistry and Life Sciences, US Military Academy, West Point, New York 10996, USA Received 10 February 2005; received in revised form 13 September 2005; accepted 21 September 2005
Abstract A synergistic effect between cationic silver and UV radiation (silver-UV disinfection) has been observed that can appreciably enhance inactivation of viruses. The purpose of this work was to assess the feasibility of this technique for drinking water disinfection and evaluate the effects of selected impurities, found in fresh water, and common parameters on inactivation of the coliphage MS-2 with the silver-UV process. Turbidity (kaolin), calcium hardness, carbonate alkalinity, and pH did not significantly degrade inactivation. Inactivation was reduced in the presence of chloride, at concentrations greater than 30 mg/L, and in water samples with UV-254 absorbance values greater than ca. 0.1 cm1. Inactivation of MS-2 with silver-UV disinfection was also reduced at high phosphate concentrations (above ca. 5 mM). Silver-UV inactivation of MS-2 increased with increases in temperature between 10 and 20 1C. Silver-UV inactivation of MS-2 was increased by greater than 1-log over UV alone, in two untreated fresh water sources, which indicates that silver-UV may be a viable treatment technology. An assessment of operation and management costs suggests that an increase in inactivation of MS-2 with silver-UV disinfection could be economically beneficial. Published by Elsevier Ltd. Keywords: Disinfection; Silver; Ultraviolet radiation; MS-2; Silver-UV
1. Introduction Viruses are reported to be the most resistant pathogens to UV disinfection, they will likely impact future regulatory fluence requirements of this technology (Brown et al., 2002; Linden et al., 2002). A reduction in UV design fluence and subsequent capital and operating costs will make UV disinfection more appealing to municipalities that may wish to simultaneously reduce disinfection byproducts and improve inactivation Corresponding author. Tel.: +845 938 2820; fax: +845 938 3339. E-mail address: [email protected] (M.A. Butkus).
0043-1354/$ - see front matter Published by Elsevier Ltd. doi:10.1016/j.watres.2005.09.037
of protozoa. Methods for improving inactivation of viruses with UV radiation are likely to help the attainment of these objectives. Recently, Butkus et al. (2004) reported a synergistic effect between cationic silver (hereafter referred to as ‘‘silver’’) and UV radiation that can appreciably enhance inactivation of the coliphage MS-2, a viable surrogate for pathogenic viruses, (hereafter referred to as ‘‘MS-2’’). They reported that inactivation of MS-2 by UV radiation was increased by greater than 1.5 orders of magnitude (to a value of ca. 3.3-log) in the presence of silver (0.1 mg/L) in a phosphate buffered suspension. Should future regulatory goals for UV disinfection become focused on a 2log inactivation of viruses (Mackey et al., 2002), then a
ARTICLE IN PRESS 4926
M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
reduction in fluence may be possible if silver were used in conjunction with UV disinfection (Butkus et al., 2004). In water, at concentrations sufficient for bactericidal activity, silver does not impart taste, color, or odor and has no apparent detrimental effects on mammalian cells (Yahya et al., 1992). The only known negative health effect is argyria, an irreversible darkening of the skin and mucous membranes, which has been caused by prolonged exposure to high silver concentration (e.g. silver therapy). Accordingly, there is no United States Environmental Protection Agency (USEPA) primary drinking water standard for silver. Although preliminary findings are encouraging, previous studies were conducted in a 20 mM phosphate buffered system free of other impurities. It was hypothesized that some impurities commonly found in raw and treated drinking water may influence the efficacy of silver-UV disinfection. The purpose of this work was to assess the feasibility of this technique for drinking water disinfection and evaluate the effects of selected impurities, found in fresh water, and variations in common water quality parameters (e.g. pH and temperature) on inactivation of MS-2 with silver-UV disinfection. MS-2 was used in this study because it has been proposed as the benchmark for validation of fullscale UV reactors (Linden et al., 2002; Mackey et al., 2002) and as a surrogate for pathogenic enteric viruses such as Noroviruses (see Thurston-Enriquez et al., 2003).
2. Methods and materials 2.1. Reagents Silver stock solutions were prepared with silver nitrate (Alfa Aesar, Ward Hill, MA) or electrochemically (Superior Aqua Enterprises, Inc., Sarasota, FL). Unless stated otherwise, silver nitrate was used to prepare silver stock solutions and all reagents were purchased from Fisher Scientific (Fair Lawn, NJ). Silver neutralizer solution was prepared by combining 11.5 g sodium thiosulfate (J.T. Baker, Phillipsburg, NJ) and 5.0 g of sodium thioglycolate (Sigma, St. Louis, MO) with 50.0 mL of deionized water (DI) (Yahya et al., 1992). Total silver concentration was quantified using a colorimetic procedure (Hach, Loveland Colo.) with a method detection limit of 0.05 mg Ag/L. Fresh silver and silver neutralizer stock solutions were prepared on the day of each experiment. All glassware used to handle silver was soaked in 10% nitric acid overnight and rinsed in DI water to remove adsorbed silver (Chambers, 1956) and other contaminants prior to use. Phosphate buffer solutions (20 mM) were prepared by combining 5.678 g of Na2HPO4 with 2 L of distilled water and adding sufficient NaH2PO4 to obtain a pH of 7.2. Other
phosphate buffer concentrations were obtained using the same methodology. Stock solutions of humic acid were prepared by adding 0.1 g of humic acid, sodium salt (Acros Organics, NJ) to 100 mL phosphate buffer (20 mM, pH 7.2). Humic acid stock solutions were sonicated for several minutes and diluted with phosphate buffer to obtain the desired absorbance value. Kaolin and sodium chloride solutions were prepared in phosphate buffer or in DI water. Solutions containing calcium and magnesium hardness were prepared by adding calcium sulfate or magnesium sulfate to DI water. Samples containing alkalinity were prepared by adding sodium bicarbonate and disodium carbonate to DI water. Alkalinity and hardness solutions were measured by a colorimetic titration (Hach, Loveland Colo.). The influence of pH on inactivation was examined by adjusting the pH of DI water with sodium hydroxide (Hach, Loveland Colo.) or sulfuric acid (Hach, Loveland Colo.), recording pH, and sterilizing before addition of MS-2 and silver. Nutrients in raw water were quantified by filtering samples through 2 mm nylon filters (Millipore, Bedford, MA) and ion chromatography (Dionex, Sunnyvale, CA). Absorbance (254 nm) was quantified with a spectrophotometer (Varian, Walnut Creek, CA) using 1 cm path length, quartz cuvettes (Fisher Scientific, Fair Lawn, NJ). 2.2. Preparation of purified MS-2 A culture of Escherichia coli (ATCC 15597) was grown in tryptic soy broth (TSB; Difco Laboratories, Detroit, MI) at 37 1C and 150 rpm. Freeze dried MS-2 (ATCC 15597-B1) was mixed with 1.5 mL of a 24 h culture of E. coli and 3.0 mL of melted (45 1C) TSB soft agar (0.5% agar, w:v). The mixture was overlaid on TSB agar (1.5% agar, w:v) plates and incubated at 37 1C for 24 h. Six milliliters of 20 mM phosphate buffer was added to the plate and incubated for 1 h. The phosphate buffer was removed, passed through a 0.22 mm filter and the filtrate was used as the MS-2 stock suspension having an initial density of ca. 109 plaque forming units (pfu)/mL. 2.3. Collimated beam setup The collimated beam apparatus used in this study (Suntec Environmental, Concord, Ont., Canada) was modified to hold a stir plate and to allow for easy and reproducible vertical and horizontal adjustment. The two low-pressure mercury vapor lamps in the instrument were warmed up for at least 30 min before all experiments. Lamp irradiance was quantified with a UV detector (IL1400A, International Light, Newburyport, MA) by placing the detector at the same height as the sample surface. Fluence was determined by placing the detector in the integration mode following the removal
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
2.4. Irradiation of samples Samples were prepared by combining 100076 mL of MS-2 viral stock suspensions with 970.1 mL (8.970.1 mL for samples containing silver) of phosphate buffer or DI water (containing the impurity to be studied e.g. sodium chloride) in acid-washed, sterile Pyrex glass Petri dishes. All samples containing silver were prepared such that 10070.6 mL of a particular silver stock solution was added to the MS-2 suspensions in the Petri dishes to minimize effects of dilution. The total volume (without stir bar) and depth (with stir bar) of the viral stock suspensions in the Petri dishes were 10 mL and 0.6 cm. The incubation period for silver and MS-2, prior to and including the time of UV radiation exposure, was held at 10 min unless stated otherwise. Samples were irradiated at the end of the incubation period. For example, because a fluence of 40 mJ/cm2 was obtained at the typical lamp intensity (ca. 190 mW/cm2) by exposing the samples to the UV radiation for ca. 4 min, the viral suspensions were placed under the collimated beam following ca. 6 min of incubation with silver. Samples containing silver were neutralized (to halt disinfection) following the incubation period by adding 30 mL of neutralizer solution (Yahya et al., 1992). Samples were stirred slowly to prevent forming a vortex in the water (Bolton and Linden, 2003). Samples containing silver were incubated at ambient temperature (ca. 20 1C) or in an incubator for a predetermined time. Each experiment was conducted at least in triplicate. For all samples, a minimum of three dilutions were plated in triplicate using the standard double agar overlay technique as described above and elsewhere (Adams, 1959; Linden et al., 2002) using an E. coli (ATCC 15597) host grown at 3770.1 1C for 3–6 h (Yahya et al., 1992). Plates were incubated at 3770.1 1C and enumerated at 2471 h. The dilution giving the highest number of pfu less than 300 was averaged and used to obtain the MS-2 survival. Controls were conducted in triplicate and plated at various times during each experiment to ensure that conditions during the course of an
experiment did not influence the number of pfu in the stock suspensions.
3. Equilibrium modeling and curve fitting The predominant species of silver in the viral suspensions were determined via the MINEQL+ chemical equilibrium modeling system (version 4.5, Environmental Research Software, Hallowell, ME). This model employs a thermodynamic data base to compute equilibrium speciation in complex aqueous systems. All model calculations were conducted at 20 1C, in an open system, and included the influence of ionic strength. TableCurve 2D (Systat, Software, Inc., Richmond, CA) was used to obtain approximating functions for interpolation of X–Y data.
4. Results and discussion Microbial studies are frequently conducted in phosphate-buffered systems to mitigate changes in pH. Indeed, some of the work reported here and elsewhere (Butkus et al., 2004) on silver-UV disinfection was conducted in a 20 mM phosphate buffer system. The influence of phosphate and carbonate buffers on silverUV disinfection of MS-2 is presented in Fig. 1. The inactivation values plotted on the ordinate represent the log survival ratio, where the survival ratio is defined as the number of pfu following exposure to silver and/or UV light (Nt) divided by the number of pfu in the control samples (No). The data imply that inactivation of MS-2 with silver-UV disinfection was influenced by
6 5
-Log (Nt/No)
of a shutter and recording the required exposure time. It was determined that variations in lamp irradiance across the surface of the samples were negligible by moving the detector in the horizontal plane at distances equivalent to the sample surfaces, which resulted in a Petri factor (PF) of unity (Butkus et al., 2004). Variations in fluence resulting from drift in lamp output were typically less than ca. 0.5%. However, lamp output was verified periodically during the course of every experiment to compensate for slight changes resulting from drift. The average fluence ðmW s=cm2 Þ or ðmJ=cm2 Þ was determined as described elsewhere (Bolton and Linden, 2003; Butkus et al., 2004).
4927
4 3 2 phosphate buffer carbonate buffer
1 0 0.00001
0.0001
0.001 0.01 Buffer Concentration (M )
0.1
1
Fig. 1. Inactivation of MS-2 as a function of buffer concentration. All samples were exposed to silver (0.1 mg/L for ca. 6 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 10 min of total silver exposure. Error bars represent one standard deviation.
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
the presence of phosphate. At phosphate concentrations below ca. 5 mM, inactivation was comparable to silverUV disinfection in DI water (4.3-log). However, inactivation was reduced by ca. 50% in a 20 mM phosphate buffer. Weak aqueous complexes and precipitates of silver and phosphate (Ag(H2PO4) 2, 2 AgHPO 4 , Ag(HPO4H2PO4) , AgH2PO4, and Ag3PO4(s)) have been reported in the literature (Baldwin, 1969; Hietanen et al., 1973; Ciavatta et al., 1996). Predictions from the MINEQL+ chemical equilibrium modeling system suggest that silver forms the Ag(HPO4H2PO4)2 complex, but other silver–phosphate and silver–carbonate species (not shown) are negligible under these conditions. Carbonate did not reduce inactivation at similar concentrations and conductivities (data not shown), which suggests that the effect of phosphate was not caused by ionic strength effects. The mechanism for the observed effect of phosphate and weak silver–phosphate complexes on silver-UV disinfection remains unknown. Phosphate concentrations in drinking water are likely to be much lower than 20 mM with the greatest phosphate concentrations expected in systems where it is added for corrosion control. Kawamura (2000) reported that typical zinc phosphate dosages are on the order of 1 mg/L (6 mM phosphate) when it is used for corrosion control. Thus, the negative influence of phosphate on silver-UV disinfection is not expected to be significant under typical water treatment conditions. Moreover, use of concentrated phosphate buffers in bench scale studies designed to evaluate silver-UV disinfection may result in inactivation values that are substantially lower than those produced with typical raw water sources. Silver-UV disinfection was evaluated in DI water and phosphate buffer to demonstrate the influence that phosphate may on future bench scale studies. Silver-UV inactivation was relatively unaffected by carbonate and bicarbonate, which suggests that carbonate alkalinity should not influence the efficacy of silver-UV disinfection. It was hypothesized that natural organic matter might influence the efficacy of silver-UV disinfection. Humic acid was used as a surrogate for natural organic matter as a first approximation. Fig. 2 illustrates the influence of humic acid, reported as absorbance, on inactivation by silver-UV disinfection and UV radiation alone in a 20 mM phosphate buffer. Due to the complex nature of humic acid, it was hypothesized that if a reaction between silver and humic acid were to occur, its rate might be slower than silver and other common inorganic drinking water impurities (e.g. silver and chloride). Consequently, samples containing silver and humic acid were incubated for 60 min. Because trends in data for silver-UV inactivation and UV radiation alone are congruent, it was presumed that absorbance of UV light (and subsequent reduction in fluence) was the
4.5 Ag-UV UV Silver-UV Model UV Model
4 3.5 -Log (Nt/No)
4928
3 2.5 2 1.5 1 0.5 0 0.01
0.1
1
10
Absorbance (cm-1)
Fig. 2. Inactivation of MS-2 as a function of humic acid concentration. All samples were exposed to silver (0.1 mg/L for ca. 56 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 60 min of total silver exposure. The predicted log survival ratio as a function of absorbance was obtained by entering the corrected fluence from Eq (1) into an approximating function fit to log survival ratio vs. fluence, for a silver concentration of 0.1 mg/L. Error bars represent one standard deviation.
predominant factor responsible for the observed reduction in inactivation as a function of absorbance (humic acid concentration). The influence of UV light absorbance on average fluence ðmW s=cm2 Þ or ðmJ=cm2 Þ was modeled as follows (Bolton and Linden, 2003): 1 10a L Fluence ¼ 0:975 E o PF DF t (1) a L lnð10Þ where, Eo is the lamp intensity at the center of the sample dish (mW/cm2), a the sample absorption coefficient (cm1), L the depth of sample (cm), t the exposure time (s), and DF is the divergence factor. The divergence factor was determined as follows (Bolton and Linden, 2003): DF ¼
Z , ZþL
(2)
where Z is the distance from the lamp to the sample surface (cm). According to the set up of the collimated beam apparatus: L ¼ 0.6 cm and Z ¼ 86 cm. The predicted log survival ratio as a function of absorbance was obtained by entering the corrected fluence (from Eq. (1)) into an approximating function fit to log survival ratio vs. fluence, for a silver concentration of 0.1 mg/L (see Butkus et al., 2004 for similar data). Because predictions made with this approach fit the data well at typical drinking water absorbance values (abso0.5 cm1), absorbance of UV light energy appears to be the predominant mechanism of reduction in inactivation at typical drinking water treatment absorbance values.
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
Differences between selected samples incubated for 10 and 60 min were not statistically different (po0:054). This suggests that incubation time between silver and humic acid did not have a salient effect on inactivation, which supports the mechanism proposed above. Although humic acid does not appear to have a significant effect on silver-UV disinfection, additional studies with other forms of natural organic matter may be warranted. Precipitation of silver chloride is well known and it was presumed that aqueous chloride would reduce the effectiveness of silver-UV disinfection. Fig. 3 illustrates the influence of chloride (added as sodium chloride) in both DI water and in 20 mM phosphate buffer. Trends in the data support the hypothesis that chloride reduced the efficacy of the synergistic effect between silver and UV radiation at chloride concentrations greater than ca. 30 mg/L (in DI water). Nevertheless, silver-UV inactivation was ca. 0.5-log greater than for UV alone (2.1-log in DI water) at the USEPA Secondary Standard for chloride (250 mg/L). This finding suggests that chloride concentrations in typical drinking water supplies should not limit this disinfection technology. Reduction in MS2 inactivation, as a function of chloride concentration, was more pronounced in the concentrated phosphate buffer. Data from controls (not shown) demonstrated that similar chloride concentrations did not influence the infectivity of MS-2. The effect of silver–chloride precipitation, on cationic silver concentration and subsequent silver-UV inactivation, was modeled with the MINEQL+ chemical equilibrium modeling system in DI water and phosphate buffer. Because silver chloride precipitation is consid-
ered rapid, it was assumed that equilibrium was achieved under the conditions of this study. The predicted log survival ratio as a function of chloride (aqueous silver concentration of 0.1 mg/L prior to precipitation) in Fig. 3, was obtained by entering the cationic silver concentration predicted by MINEQL+, following precipitation, into an approximating function fit to log survival ratio vs. silver concentration (UV fluence of ca. 40 mJ/cm2) (see Butkus et al., 2004 for similar data). This modeling approach generally underestimated inactivation of MS-2 at chloride concentrations less than ca. 150 mg/L. Because model predictions underestimated inactivation, it is presumed that cationic silver may not be the only silver species that contributed to silver-UV inactivation of MS-2 under the conditions evaluated here. The bioaccumulation and toxicity of silver complexes have been reported elsewhere (Butkus et al., 2003; Gupta et al., 1998; Reinfelder and Chang, 1999; Bury and Hogstrand, 2002). The data in Fig. 4 demonstrates that increased temperature in the mesophilic range increased inactivation of MS-2 by silver-UV disinfection. For reasons that are not yet known, reduction in MS-2 inactivation at lower temperatures was more pronounced in the concentrated phosphate buffer. Inactivation was reduced by approximately 0.7-log at the lowest temperature evaluated (10 1C) in DI water, which suggests that the combination of silver and UV radiation in cold water was still more effective than the sum individual effects of silver and UV radiation alone reported by Butkus et al. (2004). The influence of additional impurities and parameters on the synergistic effect between UV light and silver are presented in Table 1. Turbidity (10.2 NTU) resulting
5
5
Exp (PO4) Exp (DI) Eqlb Model (phosphate) Eqlb Model (DI)
4.5 4
DI phosphate
4.5 4
3.5
3.5
3
-Log (Nt/No)
-Log (Nt/No)
4929
2.5 2 1.5 1
3 2.5 2 1.5
0.5
1
0 0
100 200 300 Chloride Concentration (mg/L)
400
0.5 0 8
Fig. 3. Inactivation of MS-2 as a function of chloride concentration. All samples were exposed to silver (0.1 mg/L for ca. 6 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 10 min of total silver exposure. The model predicts the effect of chloride on cationic silver concentration and subsequent silver-UV disinfection. Error bars represent one standard deviation.
10
12
14 16 18 Temperature (°C)
20
22
Fig. 4. Inactivation of MS-2 as a function of temperature. All samples were exposed to silver (0.1 mg/L for ca. 6 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 10 min of total silver exposure. Error bars represent one standard deviation.
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
3.5
0.17
3.1
0.17
4.0
0.10
No difference in inactivation (po0.05)
from a suspension of kaolin in 20mM phosphate buffer did not influence the synergistic effect between UV radiation and silver. Passantino et al. (2004) reported that comparable turbidity values (10.5 NTU) did not influence inactivation of MS-2 resulting from exposure to UV radiation. Calcium hardness did not influence inactivation of MS-2, yet magnesium hardness decreased the inactivation of MS-2. The influence of pH on inactivation was examined by adjusting the pH of DI water with NaOH or H2SO4. Initial pH values ranged from 4.2 to 10.2 and no significant difference (po0:05) in log survival ratio was observed when compared with inactivation in DI water (pH ca. 5.7). Similar experiments in phosphate buffer indicated that inactivation was not significantly reduced (po0:05) when pH was varied between 6 and 9 when compared to inactivation in phosphate buffer at pH 7.2. The influence of impurities on silver-UV disinfection was evaluated with two fresh water sources: Lusk Reservoir, a source for drinking water at the US Military Academy, West Point, NY, and Hudson River water, sampled at West Point, NY. Both samples were adjusted to a temperature of 20 1C, amended with MS-2, and used as challenge water without additional pretreatment. Use of untreated samples was considered a conservative approach because conventional treatment processes would likely reduce some impurities before disinfection. The characteristics of these sources are presented in Table 2. Inactivation of MS-2 in the raw water suspensions, as a function of disinfection method, is presented in Fig. 5. Controls for each source indicated that the MS-2 was not affected by impurities in the raw water prior to disinfection. Disinfection of each suspension with UV alone resulted in comparable results for DI water, Lusk Reservoir, and Hudson River sources. Addition of silver before exposure to UV resulted in
7.13 54.0 248 76.0 23.3 0.149
6.67 17.0 103 24.0 0.76 0.065
0.087
0.057
29.0 3.2
Sulfatea (mg/L) Fluoridea Nitritea Phosphatea Bromidea
8.1 ND ND ND ND
13.8 Not detected (ND) 13.7 ND ND ND ND
a
Samples were filtered through a 2 mm nylon filter prior to analysis.
5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 silver-UV in Hudson
0.47 0.21
pH Alkalinity (mg CaCO3/L) Conductivity (mS/cm) Hardness (mg CaCO3/L) Turbidity (NTU) Absorbance (254 nm; unfiltered) Absorbance (254 nm; 0.45 mm filter) Chloridea (mg/L) Nitratea (mg/L)
silver-UV in Lusk
4.3 3.4
Lusk Reservoir
UV in Hudson
Standard deviation
Hudson River
UV in Lusk
DI water Phosphate buffer (20 mM, pH 7.2) Turbidity (10.2 NTU in 20 mM phosphate buffer) Mg hardness (300 mg/L as CaCO3 in DI water) Ca hardness (300 mg/L as CaCO3 in DI water) pH (6–9)
Log (N/No)
Parameter
silver-UV in DI
Treatment
Table 2 Challenge water characteristics
UV in DI
Table 1 Influence of selected water quality parameters on inactivation of MS-2 by silver (0.1 mg/L) and UV light (40 mJ/cm2) given 10 min of contact time
-Log (Nt/No)
4930
Treatment
Fig. 5. Inactivation of MS-2 in Lusk Reservoir water, West Point, NY, and Hudson River water, sampled from the shoreline in West Point, NY. Both samples were adjusted to a temperature of 20 1C, spiked with MS-2, and used as challenge water without additional pretreatment. Water quality characteristics are presented in Table 1. All samples were exposed to silver (0.1 mg/L for ca. 6 min) followed by UV radiation (ca. 40 mJ/cm2) and then neutralized resulting in 10 min of total silver exposure. Error bars represent one standard deviation.
substantially improved inactivation in all suspensions. (Using 0.1 mg/L of silver without UV radiation, Butkus et al. (2004) reported that a measurable inactivation of MS-2 was not observed for 10 min of contact time in phosphate buffer.) Silver-UV inactivation of MS-2 in
ARTICLE IN PRESS M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
4.1. Economic analysis An economic analysis of silver-UV disinfection has not yet been reported. A brief evaluation of operation and maintenance (O&M) cost is presented here to provide a jumping-off point for a more comprehensive assessment on this topic. Because O&M costs associated with electricity and silver electrodes are frequently the predominate cost for silver disinfection systems (personal communication: Mr. James Muha, Superior Aqua Enterprises, Inc., Sarasota, FL) other costs were not considered in this preliminary analysis. It was assumed that 1 kw-h of electricity could produce enough silver electrochemically, at a concentration of 0.1 mg/L, to treat 79,040 gallons of water (personal communication: Mr. James Muha, Superior Aqua Enterprises, Inc., Sarasota, FL); the cost of silver is $318/kg; and, the cost of electricity is $0.08/kw-h. Based on these figures, the silver electrode replacement cost was determined to be more than an order of magnitude greater than the electricity cost. This suggests that the price of silver generally controls the economics of disinfection with silver. A volumetric flow rate of 0.12 ML/d was used in
4000 Annual O&M Cost ($/yr)
the Lusk Reservoir water was consistent with inactivation results for DI water. This was expected because chloride (based on data obtained in DI water), phosphate, and absorbance values were not sufficient to appreciably reduce inactivation (cf. Figs. 1–3). However, inactivation of MS-2 in the Hudson River sample was lower than what was expected based on absorbance (inactivation was reduced by ca. 0.1-log at absorbance values of 0.1 cm1, cf. Fig. 2), chloride, and phosphate. It is presumed that other impurities in the water may have reduced the synergistic effect of silverUV disinfection. For example, natural organic matter may have formed complexes with silver or high levels of magnesium may have been present. Identification of additional impurities that influence silver-UV disinfection could facilitate the development of a pretreatment regime that could further optimize this technology. The upshot is that there was a substantial improvement in inactivation of MS-2 with silver-UV disinfection in untreated Lusk Reservoir water and Hudson River water, which suggests that this technique could be effective in numerous systems. Because there are health concerns associated with nitrate, silver is often generated electrochemically for disinfection. A 3.4-log inactivation of MS-2 was observed when silver was generated electrochemically in Lusk Reservoir water and used at the same concentration as silver formed from silver nitrate (0.1 mg/L), given a total of 10 min contact time with silver and a fluence of ca. 40 mJ/cm2. This finding suggests that silver-UV disinfection can be accomplished without increasing nitrate levels in the treated water.
4931
3500 3000 2500 2000 1500 1000 500 0 2
2.5
3
3.5
4
4.5
-Log (Nt/No)
Fig. 6. Increase in O&M cost as a function of increase in log survival ratio resulting from increased silver concentration given a UV fluence of ca. 40 mJ/cm2 and annual average volumetric flow rate of 0.12 ML/d. The O&M cost for disinfection with UV radiation was adapted from Cotton et al. (2001).
this analysis because UV O&M cost data was available for this flow rate in the literature (Cotton et al., 2001) and silver-UV disinfection appears to be most advantageous at lower flow rates (see below). Fig. 6 presents increase in annual O&M cost as a function of increase in log survival ratio, resulting from increased silver concentration, for silver-UV disinfection (UV fluence of 40 mJ/cm2 and an annual average volumetric flow rate of 0.12 ML/d). The annual O&M cost for disinfection with UV radiation ($1800) was adapted from Cotton et al. (2001) for UV fluence of 40 mJ/cm2 and the same average volumetric flow rate and electricity cost. The figure illustrates that inactivation can be increased from 2-log to 3-log with a concomitant increase in silver-UV O&M cost of ca. $200 (11% increase in cost). Butkus et al. (2004) reported that 2-log inactivation of MS-2, using UV only, required ca. 46 mJ/ cm2 and 3-log inactivation required ca. 70 mJ/cm2 (57% increase in fluence). Although a 57% increase in fluence does not necessarily imply that there will be a 57% increase in UV power cost (Cotton et al., 2001), this limited evaluation suggests it is plausible to assume that silver-UV disinfection could be economically viable at some flow rates. Due to the non-linear behavior of UV O&M costs (Cotton et al., 2001), the O&M cost savings of silver-UV may be less favorable at higher volumetric flow rates. In addition to lowering UV fluence requirements for a target inactivation, residual silver might also allow for lower residual chlorine concentrations in distribution systems because of the reported synergy between silver and chlorine (Yahya et al., 1992). Lower required chlorine concentrations in distributions systems might result in lower O&M costs for chlorine (and fewer
ARTICLE IN PRESS 4932
M.A. Butkus et al. / Water Research 39 (2005) 4925–4932
disinfection byproducts) making silver-UV disinfection more attractive. A more detailed economic assessment considering capital costs; ranges of volumetric flow rates, silver concentrations, silver costs, electricity costs and chlorine costs; and, UV cost as a function of fluence is needed before silver-UV disinfection can be recommended as an economically viable alternative. ‘‘Because silver-UV disinfection has been reported to inactivate a DNA virus, Haemophilus influenzae phage T4D (Rahn and Landry, 1973), it is expected that the synergistic effect between silver and UV radiation might also exist for the inactivation of pathogenic viruses such as poliovirus, Noroviruses viruses, and the enteric adenoviruses 40 and 41.’’ However, the efficacy of this technique for treatment of other pathogenic viruses warrants further investigation.
5. Conclusions The results presented here suggest that the silver-UV disinfection technique may be effective for many fresh water supplies (as well as for use in domestic wastewater treatment, hospital hot water systems, pools, and spas). Silver-UV inactivation of MS-2 was influenced by chloride, absorbance, magnesium, temperature, and phosphate. Because appreciable inactivation of MS-2 was observed with two untreated (except for temperature) fresh water sources, silver-UV disinfection may be a viable treatment technology. An assessment of O&M costs suggests that an increase in inactivation of MS-2 with silver-UV disinfection could be economically beneficial.
Acknowledgments The authors gratefully acknowledge the cost information and insight provided by Richard Peters, Hazen and Sawyer, New York, NY; Christine Cotton, Malcolm Pirnie, Tucson, AZ; and Mr. James Muha, Superior Aqua Enterprises, Inc., Sarasoto, FL. The technical assistance of Anand Shetty, Department of Geography and Environmental Engineering, United States Military Academy is also greatly appreciated. This project was supported by grants from The Army’s PM Soldier Systems Program.
References Adams, M.H., 1959. Bacteriophages. Intersciences, New York, N.Y. Baldwin, W.G., 1969. Phosphate equilibria. II. Studies on the silver-phosphate electrode. Arkiv foer Kemi 31, 407–414.
Bolton, J.R., Linden, K.G., 2003. Standardization of methods for fluence (UV dose) determination in bench-scale UV experiments. J. Environ. Eng. 129, 209–215. Brown, N.P., Jacangelo, J.G., Haas, C.N., Gerba, C.P., 2002. Assessment of existing disinfection practices for inactivation of emerging pathogens. In: American Water Works Association Annual Conference Proceedings, New Orleans, LA. Bury, N.R., Hogstrand, C., 2002. Influence of chloride and metals on silver bioavailability to Atlantic salmon (Salmo salar) and Rainbow trout (Onchorhynchus mykiss) yolk-sac fry. Environ. Sci. Technol. 36 (13), 2884–2888. Butkus, M.A., Edling, L., Labare, M.P, 2003. The efficacy of silver as a bactericidal agent: advantages, limitations, and considerations for future use. J. Water Supply: Res. Technol.—AQUA 52 (6), 407–415. Butkus, M.A., Labare, M.P., Starke, J.A., Moon, K., Talbot, M., 2004. Use of aqueous silver to enhance the inactivation of coliphage MS-2 with UV disinfection. Appl. Environ. Microbiol. 70 (5), 2848–2853. Chambers, C.W., 1956. A procedure for evaluating the efficiency of bactericidal agents. J. Milk Food Technol. 19, 183–187. Ciavatta, L., Iuliano, M., Porto, R., 1996. Complex formation equilibria between silver(I) and orthophosphate ions. Ann. Chim. 86, 411–427. Cotton, C.A., Owen, D.M., Cline, G.C., Brodeur, T.P., 2001. UV disinfection costs for inactivating cryptosporidium. J. AWWA 93 (6), 82–94. Gupta, A., Maynes, M., Silver, S., 1998. Effects of halides on plasmid-mediated silver resistance in E. coli. Appl. Environ. Microbiol. 64, 5042–5045. Hietanen, S., Sille´n, L.G., Ho¨gfeldt, E., 1973. On Some phosphate equilibria. Chem. Scripta 3, 23–29. Kawamura, S., 2000. Integrated Design and Operation of Water Treatment Facilities. Wiley, New York. Linden, K.G., Batch, L., Schulz, C., 2002. UV disinfection of filtered water supplies: water quality impacts on MS2 dose–response curves. In: American Water Works Association Annual Conference Proceedings, New Orleans, LA. Mackey, E.D., Hargy, T.M., Wright, H.B., Malley Jr., J.P., Cushing, R.S., 2002. Comparing Cryptosporidium and MS2 bioassays–implications for UV reactor validation. J. AWWA 94, 62–69. Passantino, L., Malley Jr., J., Knudson, M., Ward, R., Kim, J., 2004. Effect of low turbidity and algae on UV disinfection performance. J. AWWA 96 (6), 128–137. Rahn, R.O., Landry, L.C., 1973. Ultraviolet irradiation of nucleic acids complexed with heavy atoms—III. Influence of Ag+ and Hg2+ on the sensitivity of phage and of transforming DNA to ultraviolet radiation. Photochem. Photobiol. 18, 39–41. Reinfelder, J.R., Chang, S.I., 1999. Speciation and microalgal bioavailability of inorganic silver. Environ. Sci. Technol. 33 (11), 1860–1863. Thurston-Enriquez, J.A., Haas, C.N., Jacangelo, J., Riley, K., Gerba, C.P., 2003. Inactivation of feline calicivirus and adenovirus type 40 by UV radiation. Appl. Environ. Microbiol. 69, 577–582. Yahya, M.T., Straub, T.M., Gerba, C.P., 1992. Inactivation of coliphage MS-2 and poliovirus by copper, silver, and chlorine. Can. J. Microbiol. 38, 430–435.