Removal of E. coli from urban stormwater using antimicrobial-modified filter media

Journal of Hazardous Materials 271 (2014) 73–81

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Removal of E. coli from urban stormwater using antimicrobial-modified filter media Ya L. Li a,b,∗ , Ana Deletic a,b , David T. McCarthy a,b a Environmental and Public Health Microbiology Lab (EPHM LAB), Monash Water for Liveability, Department of Civil Engineering, Monash University, Melbourne, Vic 3800, Australia b CRC for Water Sensitive Cities, Melbourne, Vic 3800, Australia

h i g h l i g h t s • • • • •

15 antibacterial filter media were prepared for enhanced bacterial removal from urban stormwater. Their performances were evaluated over 24 weeks under typical stormwater operational conditions. Filter media modified with copper compounds exhibited robust antibacterial efficiency. Filter media modified with Cu2+ and Cu(OH)2 showed effective bacteria removal during wet events. Filter media modified with Cu(OH)2 showed very good stability in stormwater.

a r t i c l e

i n f o

Article history: Received 5 November 2013 Received in revised form 15 January 2014 Accepted 26 January 2014 Available online 12 February 2014 Keywords: Stormwater Antibacterial Filtration Pathogen Water sensitive urban design

a b s t r a c t Stormwater filters featuring traditional sand filter media cannot reliably treat indicator bacteria for stormwater harvesting. In this work, copper-modified zeolite and granular activated carbon (GAC) were prepared through Cu2+ impregnation and in situ Cu(OH)2 precipitation. Their antibacterial properties and stability in natural stormwater were studied in gravity-fed columns for 24 weeks, under typical stormwater operational conditions. 11 types of other filter media, prepared using zinc, iron, titanium and quaternary ammonium salts as antibacterial agents, were tested in parallel by way of comparison. Cu2+ -immobilised zeolite and Cu(OH)2 -coated GAC yielded an estimated 2-log reduction of E. coli within 40 min with the presence of other native microbial communities in natural stormwater. Even at high flow velocity (effective contact time of 4.5 min), both media demonstrated 0.8 log removal. Both media and Cu2+ -treated GAC showed effective inactivation of the removed E. coli during dry periods. Copper leaching from Cu(OH)2 -coated GAC was found to be below the NHMRC specified drinking water standard, while that from Cu2+ -immobilised zeolite varied with the salinity in stormwater. These findings could provide useful information for further development of passive stormwater harvesting systems. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Stormwater filters and biofilters are gravity-fed filter beds, vegetated or non-vegetated, placed within urban landscapes that are never back-washed [1]. They remove nutrients and metals effec-

∗ Corresponding author at: Monash Water for Liveability, Department of Civil Engineering, Monash University, Melbourne, Vic 3800, Australia. Tel.: +61 399056202; fax: +61 399054944. E-mail addresses: [email protected] (Y.L. Li), [email protected] (A. Deletic), [email protected] (D.T. McCarthy). http://dx.doi.org/10.1016/j.jhazmat.2014.01.057 0304-3894/© 2014 Elsevier B.V. All rights reserved.

tively by means of biological uptake, straining and adsorption [1–3]. However, field and laboratory investigations have shown that their effluent seldom meets bacterial indicator targets for outdoor irrigation [4–7]. This is partially due to the inadequate microbial removal capacity of sand media used in these filters, as well as survival and detachment of microbes under intermittent stormwater inflows [8]. Therefore, stormwater filters and biofilters require novel media in order to ensure effective pathogen removal [9]. As inorganic antimicrobials, zeolites and activated carbon containing Ag, Cu or Zn have proliferated for food preservation, self-disinfection fabrics, etc. [10,11]. Ag-activated carbon has been

74

Y.L. Li et al. / Journal of Hazardous Materials 271 (2014) 73–81

tested for water treatment and demonstrated very good bacterial removal efficiency [12,13]. An investigation of Cu2+ -treated zeolite for water treatment, subjected to controlled conditions and 6 h hydraulic residence time, showed around 1–3 log removal for a wide range of microbes [14,15]. However, longevity in efficiency and stability – yet to be addressed – is essential for field application. Compared with the metal ion treatment of filter media, metal hydr(oxide) coating exhibits better stability in water due to its low solubility constant. In addition, metal hydr(oxide) coating shows effective bacteria removal through electrostatic attraction [12,16,17]. More importantly, the slow release of metal ions from the coating layer in solution at an effective level may exert an antimicrobial effect from the media. Kennedy et al. [18], for example, examined the bactericidal effects of CuO/Cu2 O-coated carbon and showed 4 log removal of E. coli. However, the coating processes required expensive organometallic precursor and toxic organic solvent, prohibiting large scale field applications. In addition, Cu(OH)2 , an effective component in Bordeaux mixture as a pesticide and fungicide, has not been investigated as antimicrobial coating for any water treatment media. Consequently, CuO coating through Cu(OH)2 precipitation has not been investigated for antimicrobial use. The above findings cannot be used directly for advancements of stormwater filters and biofilters, as urban runoff is a unique water source. For example, stormwater biofilters are located within an urban environment (often being an amenity feature), thus are exposed to highly variable hydraulic and pollutant loadings, intermittent wetting and drying conditions, and highly seasonal variations [19]. The distinct characteristics of stormwater will pose questions for the aforementioned modified materials: for example, will variability, and sometimes high salinity, of stormwater lead to excessive heavy metal leaching from antimicrobial media? This study therefore aims to develop and evaluate antibacterial media for urban stormwater treatment. Specifically, this work’s objectives are to:

• Develop simple yet scalable processes to modify GAC and zeolite with Cu(OH)2 , CuO, and Cu2+ producing 4 types of copper modified media; • Evaluate their stability and bacteria removal performance, inactivation efficiency over semi-long term experimental duration, under typical stormwater operational conditions including relatively high filtration rates, intermittent operation patterns, and variable salinity in stormwater; • Investigate the main bacterial removal mechanisms subjected to stormwater conditions; • Prepare 11 other antibacterial media using zinc, iron, titanium, and quaternary ammonium salts (consulted in literature for other types of water treatment), evaluating their stability and bacteria removal efficiency for stormwater operational conditions (in comparison with copper treated media).

2. Experiments 2.1. Modification of zeolite and granular activated carbon with antibacterial agents The chemicals (CAS number in parentheses) and their sources, featured in this study, comprise zinc sulfate heptahydrate (7446-20-0), copper(II) chloride (7447-39-4), iron(III) chloride (7705-08-0), hexadecyltrimethylammonium chloride (QAC) (11202-7), sodium hydroxide (1310-73-2) and ethylenediaminetetraacetic acid disodium salt (EDTA) (6381-92-6), Merck Chemicals, Australia; dimethyloctadecyl[3(trimethoxysilyl)propyl]ammonium chloride (Si-QAC) (27668-52-6), Sigma–Aldrich; TiO2 sol–gel [20]. Natural zeolite (particle size 0.3–0.6 mm), Zeolite Australia; and granular activated carbon (GAC) (particle size 0.3–0.6 mm), Activated Carbon Technologies Pty Ltd, constituted base media. The basic physicochemical properties of natural zeolite and GAC were listed in [21,22]. Modification of zeolite by Cu2+ , Fe3+ , Zn2+ : Zeolite was mixed with 2 M NaCl for 72 h to produce Na-zeolite. NaCl solution was replaced every 24 h. After being washed with deionised (DI) water, Na-zeolite was mixed with 0.015 M CuCl2 solution (metal content 5% by weight of zeolite) for 48 h. The CuCl2 solution was replaced every 24 h. This sample was then washed and dried at 105 ◦ C overnight to produce Cu2+ modified zeolite, denoted as Cu-Z. Following a similar procedure, Fe-Z and Zn-Z were prepared. Zn/Cu/Fe-Z was concocted by impregnating Na-zeolite in 0.015 M solutions of ZnSO4 , CuCl2 and FeCl3 for 24 h each sequentially. All mixing was under agitation from a rotary platform at 150 rpm. Modification of zeolite by CuO: 3 g CuCl2 was added into aqueous slurry of zeolite (80 g zeolite in 300 mL DI water). The mixture was rotated gently for 1 h, after which the pH of the slurry was adjusted to 8 with 2 M NaOH. After mixing for 1 h, the particles were separated, washed with DI water and dried at 60 ◦ C overnight. The dry media was then heat-treated at a rate of 5 ◦ C/min to 400 ◦ C and maintained at that temperature for 1 h before cooling by means of a temperature controlled programmer. After cooling, the CuO impregnated particles were washed five times with water. The washed sample was dried at 105 ◦ C to attain the final product CuOZ. Modification of GAC by Cu2+ : 10 g GAC was mixed with 500 mL of 0.015 M CuCl2. The slurry was gently rotated for 24 h. Particles were then separated, washed and dried at 105 ◦ C for use (Cu-G). Modification of GAC by Cu(OH)2 : The preparation procedure was similar to that of CuO-Z, excepting a lower heat treatment temperature of 180 ◦ C for preparing Cu(OH)2 -G. Modification by other antibacterial media: Zn(OH)2 , Fe(OH)3 , TiO2 , QAC, and Si-QAC modified media were prepared following the aforementioned methods [12,23–26]. Prepared media were denoted as Zn(OH)2 -Z, Zn(OH)2 -G, Fe(OH)3 -Z, TiO2 -Z, TiO2 -G, QACZ, SiQAC-Z and SiQAC-G.

Table 1 Combinations of base media with antibacterial agents. Antibacterial agent

Nil Metal ions Metal hydroxide Metal oxide Quatsa a b

Modified filter media Zeolite-based media

GAC-based media

Z0 Cu-Z, Zn-Z, Fe-Z, Zn/Cu/Fe-Zb Fe(OH)3 -Zb , Zn(OH)2 -Zb CuO-Zb , TiO2 -Zb SiQAC-Zb , QAC-Z

G0 Cu-Gb Zn(OH)2 -G, Cu(OH)2 -Gb TiO2 -Gb SiQAC-Gb

Quats – Quaternary ammonium salts. Media have never been tested for water treatment, while all the listed media have never been tested for stormwater applications.

Y.L. Li et al. / Journal of Hazardous Materials 271 (2014) 73–81

In total, zeolite and GAC were modified with 10 types of antibacterial materials, while only 15 combinations were examined (Table 1) due to resource limitation. A greater number of zeolite configurations were examined due to low cost and high cation exchange capacities [27]. GAC was investigated to immobilise Cu2+ , Cu(OH)2 etc. due to its vast surface area, while some combinations, for example G-CuO could adduce reported work [18]. Untreated zeolite (Z0) and GAC (G0) were used as controls. The metal content of antibacterial media was measured by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) in a NATA-accredited laboratory. The morphology and surface elements of antibacterial media were examined using Scanning Electron Microscope (SEM, JEOL JSM-7001F) equipped with Energydispersive X-ray spectroscopy (EDS). 2.2. Test water Natural stormwater (SW) was collected from outlets of two stormwater sedimentation basins, Melbourne (details in Table 2) and utilised on the same day. The water collected during wet weather operated in its raw state as a natural source of E. coli, whereas water collected during dry weather was spiked with labstrain E. coli (ATCC#11775) (ALS Environmental, Melbourne) as a source of mixture comprising natural E. coli and lab-strain E. coli. The test water was stored in a 60 L water tank with continuous stirring during dosing, to avoid sedimentation. DI water was used as microbial contamination-free test water to differentiate bacterial survival and detachment within filter media. The quality of test water – measured using a multi-parameter water quality probe (U50, Horiba Ltd., Japan) – reflected the variability of natural stormwater in terms of E. coli concentration, presence of other native microbial communities (as in natural stormwater), salinity, turbidity and temperature [5] (Table 2). 2.3. Filtration tests Stability and bacterial removal efficiency of the 15 types of modified media and two untreated controls were testedin columns at

75

two stages: Stage 1 pure media assessment to understand performance and several processes of each individual medium under stormwater operational conditions and differences between media types. This was to identify promising antibacterial media for stormwater treatment; Stage 2 sand filter application is to apply the promising media in stormwater sand filters which are common in stormwater treatment [1–3]. The experimental arrangement and operational conditions are summarised in Table 2. Experimental set up – Stage 1: The filter media were packed into columns in accordance with the method previously described [17]: three replicates for all media types except for Cu(OH)2 -G, which had one replicate, due to experimental difficulties. Once packed, all columns were flushed using 18 × 80 mL pulses DI water to remove the fine particles produced during packing and to allow the media to settle. Experimental set up – Stage 2: After completion of Stage 1 experiments, 3 best performing filter media and their control media from Stage 1 were emptied from columns and dried at 60 ◦ C for 24 h. The dried media was then mixed with fine sand (particle size 0.15–0.30 mm) in a ratio of 1:1.1 by volume, repacked into initial columns as per Stage 1. Thereby depth of media was increased from 100 mm to 210 mm, yet actual contact time with modified media remained unchanged. The columns were then flushed using 25 × 80 mL pulses DI water. As shown in Table 2, Stage 1 and 2 columns were dosed with test water intermittently over 24 weeks, covering 10 simulated rain events and 9 dry periods between rain events of varied duration 1–4 weeks. During rain events, columns were dosed with test water by maintaining a constant head, whereas during dry periods, they received no inflow. Columns’ outlets were restricted to attain the desired superficial flow velocity (Table 2): 860 mm/h in Event 1, 720 mm/h in Event 7 and 86 mm/h in other Events, translating to effective contact time with modified media of 4 min, 4.5 min and 40 min, respectively. This mimicked a hydraulic loading range typical for stormwater filters and biofilters. 2500 mL stormwater was applied to each column during Event 1 (65 pore volumes of modified media). 350 mL of water (9 pore volumes) was dosed during other Events, which is equivalent to the inflow into a filter sized at 2.5% of its impervious catchment area during a 14 mm rain event

Table 2 Experimental setup and operational conditions during filtration test. (a) Variables Filter media E. coli inflow concentration Column design (b) Experimental set-up

Zeolite based media (10 types + control); GAC based media (5 types + control) Natural stormwater (SW); microbial-contamination free water (DI water) Stage 1 pure media assessment; Stage 2 sand filter application

(c) Experimental conditions Stage 1 pure media assessment Rain event Time (week) Test water E. coli (MPN/100 mL) Flow velocity (mm/h) EC (␮S/cm) T (◦ C) pH Turbidity (NTU)

1 1 SW 2965 860 626 21 7.1 30

2 3 DI <1 86 4 19 6.5 0

3 6 SW 14,260 86 699 17 7.1 20

4 10 SWa 13,730 86 1330 18 7.6 NA

Repacking 5 13 SWa 48,390 86 143 12 7.0 6

6 16 SWa 2195 86 92 13 6.9 2

Stage 2 sand filter application 7 21 SWa 8136 720 170 12 7.6 3

8 22 DI <1 86 4 12 6.5 NA

9 23 SW 845 86 114 11 6.6 4

Filter media harvesting

10 24 SW 52 86 95 13 7.0 NA

Note: SW – Natural stormwater collected from outlets of two stormwater sedimentation basins in Melbourne: Events 1, 3 and 4 from Hampton Park (medium density residential development); Events 5, 6, 7, 9, and 10 from Jock Marshall reserve, Clayton (Low density residential development); SWa – SW spiked with lab-strain E. coli (ATCC#11775); DI – Deionised water; NA – no data.

76

Y.L. Li et al. / Journal of Hazardous Materials 271 (2014) 73–81

and close to the ‘design storm’ for typical stormwater biofilters in Melbourne’s climate [1]. During each Event, duplicate composite inflow samples were collected, while the entire outflow was collected from each column. For Events 1–5, sampling bottles were sterilised without additional treatment. However, during Events 6–10, the sampling bottles were pre-treated with 0.01 M EDTA in a ratio of 0.3:100 sample volumes to inactivate metal ions, thereby eliminating the impact of post filtration inactivation. E. coli concentrations in all inflow and outflow samples were analysed using ColilertTM method (IDEXXLaboratories, 2007). Total metal concentrations in these samples were measured using ICP-MS in a NATA-accredited laboratory. 2.4. Inactivation of E. coli by Cu2+ in aqueous phase

Table 3 Content of antibacterial agents of modified media. Modified filter media

Heavy metal

Heavy metal concentration (mg/g media)a

Cu-Z Zn-Z Fe-Z Zn/Cu/Fe-Z Zn(OH)2 -Z Fe(OH)3 -Z CuO-Z TiO2 -Z Cu-G Zn(OH)2 -G Cu(OH)2 -G TiO2 -G

Cu Zn Fe Zn/Cu/Fe Zn Fe Cu Ti Cu Zn Cu Ti

13 11 8.3 4.8/2.4/5.1 2.5 5.7 5.6 0.26 5.9 3.7 16 0.29

a

Inactivation of E. coli by Cu2+ in an aqueous phase was investigated in a batch system as in previous study [16]. Three conical tubes, each containing 50 mL of natural stormwater, were spiked with CuCl2 (final Cu2+ concentration 1 mg/L). The tubes were agitated at 250 rpm in a shaking incubator at 20 ◦ C. At specified times, 2 mL of the solution was collected from each tube into a conical tube containing 6.7 ␮l 0.01 M EDTA and mixed thoroughly. E. coli concentration was then analysed in accordance with the ColilertTM method. Moreover, a control experiment in the absence of Cu2+ was performed under identical conditions. 2.5. Bacterial survival on modified media—filter media harvesting Immediately following completion of Event 10, the mixed media with sand was excavated in three layers of equal length from each column used in Stage 2. A known weight of media from each depth was placed into 100 mL DI water containing 0.02% Tween 80 and agitated at 15 rpm for 10 min. Concentration of E. coli in the supernatant solution was analysed using the ColilertTM method. Soil moisture content in the filter media was also measured at each depth; hence the E. coli concentration was provided in MPN/g dry media. Total amount of viable E. coli within each column was calculated as being MPN/column. 2.6. Data analysis Where the outflow concentration was lower than the detection limit, half the detectable limit was factored as the concentration for statistical analysis. For each type of filter media, median E. coli concentrations in inflow and outflow across rain events during Stage 1 were calculated. E. coli log removal rate was calculated as the difference between log concentrations in inflows and outflows for each media type. Log removal rates were checked for normality using Kolmogorov–Smirnov test. Normality could not be confirmed for Cu-Z, Zn-Z and Zn(OH)2 -Z, possibly because of the high number of below detection limit measurements. Cu-Z and Zn-Z, for example, had 75% and 33% samples below detection limit. One-way ANOVA analysis of log removal rates along with post-hoc tests – Tukey was performed to test the significant difference between antibacterial media and untreated controls. For selected filter media, the changes in E. coli outflow concentrations over 24 weeks of testing were also examined. 3. Results and discussion 3.1. Characterisation of modified media Concentration of antibacterial agents and their accessibility to bacteria solutions are two of the governing factors for effective bacterial inactivation by antibacterial media. Both factors were con-

Net increase in metal content as opposed to untreated control.

trolled by the physical/chemical characteristics of base material, antibacterial chemicals and modification conditions. Table 3 shows concentration of antimicrobial agents on newly prepared filter media. The net increase in content of Fe3+ , Zn2+ , Cu2+ in treated zeolite (as opposed to untreated controls) ranged between 8 to 13 mg/g media. Heavy metal ion immobilisation on zeolite is mainly through cation-exchange in pores and lattice to balance their net negative charges, while hydrolysis might also play a role due to low precipitation pH of these metal ions [28]. In situ precipitation of Cu(OH)2 on zeolite followed by calcination, immobilised 5.6 mg Cu/g zeolite. Loading of Cu(OH)2 on GAC was much higher (16 mg Cu/g media) resulting from its extensive surface area. The amount of Cu2+ adsorbed on per gram GAC was 5.9 mg. SEM examination of Cu-Z and Cu-G revealed that the surface morphology of the base media was not significantly changed after Cu2+ immobilisation (Fig. 1). A deposited layer manifest on zeolite and GAC after in situ precipitation of Cu(OH)2 followed by heat treatment. The precipitate on zeolite was uniform and continuous with agglomerate size ranging from 100 to 200 nm, while that on GAC seemed mainly located in macropores. EDS analysis of single agglomerates on CuO-Z showed only two elements, Cu and O in an atomic ratio of 1:1 estimating to be CuO. The agglomerates observed on Cu(OH)2 -G showed mainly Cu, O, Cl in an atomic ratio of 1.8:2.6:1 proposing the existence of a partial hydrolysis product CuCl2 ·3Cu(OH)2 (or written as Cu2 (OH)3 Cl, a widely used fungicide) and lightly residual CuCl2 . These observations are in agreement with earlier findings about preparing CuO through homogeneous or heterogeneous precipitation from CuCl2 and NaOH solutions followed by calcination, where the yields were characterised using X-ray diffraction (XRD) and EDS [29,30]. The consistency between EDS and XRD for chemical composition analysis has also been proved by a few other works [31–33]. Although the precipitate on Cu(OH)2 -G was identified to be mainly CuCl2 ·3Cu(OH)2 , the notion for this media remained unchanged. 3.2. Removal performance of pure media-stage 1 Table 4 summarises statistics on E. coli concentrations in inflow and outflow samples, log removals and heavy metal concentrations in outflow recorded during Stage 1. The average inflow E. coli concentration of 13,730 MPN/100 mL in stormwater was within the range of untreated urban stormwater quality [5,19]. log removals by untreated zeolite and GAC were 0.40 and 0.58, respectively, while those by the modified media ranged from 0.37 to 3.44. Statistically significant differences in E. coli removal performance between modified and control media were observed only in samples from Cu-Z, Zn-Z, Cu(OH)2 -G columns, which showed log

Y.L. Li et al. / Journal of Hazardous Materials 271 (2014) 73–81

77

Fig. 1. SEM images of modified and unmodified filter media.

removals of 3.44, 0.92, 1.93, respectively. The bacterial treatment performance of Cu-Z and Cu(OH)2 -G met the unrestricted irrigation standard regarding log removals (i.e. >1.5 log), while the former met the same standard regarding E. coli concentration (i.e. <10 E. coli/100 mL) [5].

Both Cu-Z and Cu(OH)2 -G showed notably higher removal performance compared with all the other filter media. The outstanding performance of Cu-Z can be attributed to adequate loading of Cu2+ and, more importantly, superior antibacterial efficiency of Cu2+ . At a similar level of metal loading, Zn-Z showed only one log removal,

78

Y.L. Li et al. / Journal of Hazardous Materials 271 (2014) 73–81

Table 4 Average E. coli removal performances and release of heavy metals by modified and unmodified filter media – Stage 1 pure media assessment. Pollutants

E. coli

Inflow

Stormwatera 13730 (2580, 31330) MPN/100 mL Number of Median concentrationa (MPN/100 mL) samples

Outflow

Z0-control Cu-Z Zn-Z Fe-Z Zn/Cu/Fe-Z Zn(OH)2 -Z Fe(OH)3 -Z CuO-Z TiO2 -Z SiQAC-Z QAC-Z G0-control Cu-G Zn(OH)2 -G Cu(OH)2 -G TiO2 -G SiQAC-G Non-restricted irrigation standard [5] a b c d e

15 15 12 11 15 10 9 15 9 14 10 15 15 12 5 9 10

Heavy metal

5475 (1866, 17330) 5 (5, 40) 1637 (153, 3733) 4352 (2187, 6488) 1222 (834, 7270) 5323 (64, 7657) 5794 (1730, 7955) 3654 (1204, 8212) 5172 (1710, 8225) 2151 (1153, 6276) 3061 (1931, 7620) 3654 (1850, 9222) 1022 (426, 1616) 5550 (2026, 14550) 160 (5, 1493) 5172 (1631, 6703) 4692 (1979, 11940) <10

Median log removala 0.40 (−0.10, 0.87) 3.44 (2.53, 3.44) 0.92 (0.57, 2.92) 0.50 (0.33, 0.80) 1.05 (0.28, 1.22) 0.41 (0.26, 2.36) 0.37 (0.24, 0.90) 0.58 (0.22, 1.06) 0.42 (0.22, 0.91) 0.81 (0.37, 1.08) 0.65 (0.27, 0.85) 0.58 (0.17, 0.87) 1.13 (0.93, 1.51) 0.40 (-0.02, 0.85) 1.93 (1.04, 3.44) 0.42 (0.31, 0.93) 0.47 (0.06, 0.84) >1.5

Significanceb

p < 0.001e p = 0.004e p = 1.000 p = 0.968 p = 0.786 p = 1.000 p = 0.973 p = 1.000 p = 0.992 p = 1.000 p = 0.171 p = 1.000 p < 0.001 p = 1.000 p = 1.000

DI water 0 MPN/100 mL Median concentrationc (MPN/100 mL) 69 (58, 71) 1 (0.5, 2) 59 (30. 69) 141 (128, 216) 30 (27, 52) 54 (43, 97) 93 (85, 101) 15 (10, 43) 52 (32, 56) 141 (1, 158) 37 (32, 62) 62 (54, 83) 10 (6, 35) 8 (6, 15) 8 84 (72, 97) 98 (43, 114)

Stormwater

DI water

Mean concentration (ppm)d NA 35 26 0.15 10/6.0/0.15 17 0.05 0.60 0.00 NA NA NA 0.72 8.2 2.0 0.004 NA

NA 1.7 0.1 NA NA 0.1 NA 0.5 NA NA NA NA 1.0 NA 1.0 NA NA

25th and 75th percentiles in parenthesis. p < 0.05 indicates significant difference between modified filter media and unmodified control media. Three replicates in one event (min and max in parenthesis). Three replicates over two events for stormwater while one event for DI water. Although the data were not normal distributed, the high significance of the results was not affected since the p values are sufficiently low [34].

indicating lower antibacterial efficiency. In contrast, E. coli removal by Cu(OH)2 -G was due more to surface properties of modified media. Untreated GAC (G0) has a negatively charged surface, e.g. zeta potential −36.7 mV as reported by Pal et al. [12], which is not beneficial for removal of negatively charged bacteria. Metal hydroxide coating can alter zeta potential of the media to be more positive, e.g. GAC treated by 10% aluminum hydroxychloride solution showed zeta potential of +30.6 mV [12], enhancing bacteria adsorption. Moreover, the slightly soluble component of copper could inactivate the adsorbed bacteria, thus achieving synergistic adsorption and inactivation. However, these mechanisms were not dominant on Zn(OH)2 modified media, partly because of much less hydroxide coating – only a quarter of that on Cu(OH)2 -G (Table 3). Less antibacterial efficiency of Zn2+ and poor adhesion of zinc hydroxide flocs to the base media may play more significant roles [12,14,35]. Relatively low removal by Cu-G was observed, possibly due to the limited surface density of Cu2+ on GAC, which has much larger surface area than zeolite [21,22,36,37], thus less interaction with bacteria. Although both QAC and Si-QAC could be effective antibacterial coating [25,38], their robustness and stability in aqueous environment need to be further explored since our results and some previous studies [24,39,40] contradict the above. The other modified filter media did not show good E. coli removal in stormwater, although their effectiveness was evidently high in previous studies [14,16,18]. This might be attributed to more favourable conditions used in literature, i.e. fine media size (<0.15 mm), low hydraulic loading (160 mL test water), extended contact time (6 h), and/or high experimental temperature (30–35 ◦ C). Two weeks after Event 1, around 70 MPN/100 mL E. coli were detected in the DI water out of Z0 and G0, while that out of Cu-Z, CuG, CuO-Z, Cu(OH)2 -G were much lower (1, 10, 15 and 8 MPN/100 mL respectively) indicating the superior antibacterial efficiency of the copper modified media during dry periods. Interestingly, although Zn(OH)2 -G could not remove bacteria instantaneously during rain events, it could inactivate the retained bacteria through extensive contact between events. All the other modified media showed

effluent E. coli concentrations at a level similar to, or even higher than, their controls, illustrating their inefficient antibacterial activity. In summary, the copper modified media, especially Cu-Z, CuG and Cu(OH)2 -G, showed promising results for instantaneous removal during rain events and/or bacterial inactivation during dry periods. All these types of media were selected for further study to investigate longevity with respect to bacterial removal, bacterial inactivation, and metal leaching. Moreover, their bacterial removal mechanisms and the effect of flow velocity were also explored. 3.3. Removal performance of selected media over time (Stage 1 and 2) Bacteria removal by modified media during rain events: E. coli concentrations in inflow and outflow from the five selected modified and unmodified media (i.e. two controls) over time are shown in Fig. 2. Z0 and G0 showed minimal E. coli removal and even net leaching of E. coli, indicating limited straining and adsorption capacity in their raw state. Cu-G showed relatively better E. coli removal than unmodified GAC. Cu(OH)2 -G showed much better E. coli removal than unmodified GAC in Stage 1, which was evident even after mixing with fine sand in Stage 2. Bacterial removal by Cu(OH)2 -G seemed to be negatively affected by salinity in test water, notwithstanding the copper release was not obviously affected by the same factor (Fig. 2 and 3): log removal rates 0.77, 1.33, 2.48 and 2.64 in test water of EC of 1330, 699, 143, 92 ␮S/cm respectively. Although this trend has not been reported in literature, the reduced removal in high salinity water might be explained by the competition for adsorption sites by the presence of high concentration of ions, which may interfere with the electronic attraction between negatively charged bacteria cells and a positively charged media surface. In addition, reduced bacterial removal performance was observed after mixing the media with fine sand, i.e. 0.78 log removal in Event 9 (week 23), in contrast with 2.64 log removal in Event 6 (week 16). The reason for this observation might be – though is by no means

Y.L. Li et al. / Journal of Hazardous Materials 271 (2014) 73–81

79

Fig. 2. Median and 95% confidence interval of E. coli concentrations in inflow and outflow from selected filter media over time and salinity in test water (two composite samples for inflow, one replicate for Cu(OH)2 -G and three replicates for other media types).

limited to – degradation of copper hydroxide into copper oxide during the drying process between Stage 1 and 2, in the context of copper hydroxide loss during the drying, mixing and repacking undertaken in Stage 2. Cu-Z showed consistently good removal of E. coli over all sampling events. Its performance was unaffected by mixing with fine sand: 1.9 log removal in Event 6 (week 16) of Stage 1, and 1.7 log removal manifest in Event 9 (week 23) of Stage 2. This media was also effective in treating stormwater having very low E. coli influent concentration Event 10 (week 24). The median effluent concentration of MPN/100 mL over Events 6, 8, 9 and 10 met the non-restricted irrigation guideline value (10 MPN/100 mL) specified in Australian Guidelines for Water Recycling (Phase 2): Stormwater Harvesting and Reuse 2009 [5]. Literature has shown that the biocidal effect of Cu2+ is mainly due to deteriorated cell membrane integrity, damaged protein, denatured DNA through accumulated Cu2+ and their redox reaction products, i.e. highly reactive hydroxyl radicals [41]. Bacterial inactivation by Cu-Z could occur both at the solid/liquid interface and in the liquid phase. However, the inactivation at the solid/liquid interface was hypothesized to play a major role in Stage 2 due to the high local concentration of Cu2+ i.e. at the solid–liquid interface. Nonetheless, the contribution by leached copper in aqueous solution was discussed in Section 3.4. During Stage 2, bacterial removal performance by the modified media was examined at two superficial velocities: 720 mm/h in

Event 7 (unrestricted flow), 86 mm/h in Event 9 (restricted, typical biofilter flow velocity). As depicted in Table 5, untreated media could not remove bacteria effectively; even net leaching of bacteria was observed. Cu(OH)2 -G showed effective bacteria removal of 0.8 log at both velocities. This is indicative of the dominant role of electronic attraction in removing bacteria during rain events. Removal by Cu-Z at low flow velocity was 1 log higher than that at high velocity. As discussed above, the dominant removal process in Cu-Z columns was inactivation, resulting from contact with Cu2+ at the solid–liquid interface, the effectiveness of which required extensive contact. Bacterial inactivation during dry periods (in between rain events): After extensive use, bacterial inactivation efficiency of the modified media was evaluated in Stage 2 through DI water rinsing of the columns in week 22 and filter media harvesting in week 24. The latter was to investigate the possibility of attachment of E. coli to the filter media, rendering the cells undetectable. As shown in Fig. 4, the E. coli concentrations out of Z0 columns and that residing in Z0 columns, were 345 MPN/100 mL and 65 MPN/column respectively, which for G0 columns were 41 MPN/100 mL and 146 MPN/column respectively. In contrast, little or no E. coli were detected for Cu-Z, Cu-G and Cu(OH)2 -G, suggesting that their antibacterial efficiency remained after extensive use and after being mixed into sand filters. 3.4. Metal leaching and possible implications Metal leaching during rain events: Fig. 3 shows the copper concentration in effluent samples from selected media, including Cu-Z, Cu-G, Cu(OH)2 -G recorded during Stage 1 and 2. As expected, the impregnated Cu2+ in Cu-Z was ion-exchangeable; very low copper leaching was found in DI water (1.7 mg/L, Event 2), while significant

Table 5 Median E. coli log removals by selected modified and unmodified filter media at two flow velocities (min and max in parenthesis). Filter media

Fig. 3. (a) Elution of E. coli in DI water (week 22); (b) E. coli survival on filter media (filter media harvesting) (three replicates for each media type while only one replicate for Cu(OH)2 -G).

Z0 Cu-Z G0 Cu-G Cu(OH)2 -G

Superficial velocity (mm/h) 720

86

0.25 (0.19, 0.31) 0.65 (0.60, 0.91) 0.21 (0.15, 0.89) 0.50 (0.44, 0.53) 0.82

−0.06 (−0.12, 0.20) 1.66 (1.60, 2.93) −0.06 (−0.06, 0.12) 0.28 (−0.01, 0.39) 0.78

80

Y.L. Li et al. / Journal of Hazardous Materials 271 (2014) 73–81

Fig. 4. Median and 95% confidence interval of copper leaching from modified filter media over time and salinity in test water.

leaching occurred in stormwater of salinity 600 ␮S/cm (35 mg/L, Events 1 and 3). Copper leaching from Cu-Z at Stage 2 (0.8–1.1 mg/L) was below Australian drinking water guidelines [42] and considerably lower than that in Stage 1. However, these levels exceeded Australian and New Zealand guidelines for receiving fresh and marine water quality [43]. This reduced level of copper leaching can be mainly attributed to low salinity stormwater used in this stage, as well as system maturity. The low stormwater temperature in Stage 2 might also play an important role, resulting in slow reaction kinetics [44,45]. Adsorption of leached copper by fine sand proved negligible in a separate experiment. Despite copper leaching, the media may remain effective over a reasonable duration, particularly in catchments of medium to low salinity. For example in [14], zeolite loaded with 2 mg Cu/g, worked effectively for wastewater treatment. Therefore, at the level of copper leaching of around 1 mg/L, at least 1100 L water can be effectively treated by 100 g Cu-Z initially loaded with 13 mg Cu/g media. That means, the Cu-Z column used in this study can last for 50 years if being operated as typical stormwater filter in Melbourne’s climate. The immobilised Cu2+ in Cu-G was relatively stable and the leaching was not obviously affected by operational conditions, although the copper leaching in high salinity water (0.70–0.76 mg/L) was slightly lower than in DI water (1.0 mg/L). Cu(OH)2 -G showed very pleasing stability, and copper leaching

Fig. 5. Inactivation of native E. coli at 20 ◦ C in a batch system by 0 ppm Cu2+ (solid line, circle) and 1 ppm Cu2+ (dashed line, triangle), using natural stormwater runoff collected after a 19 mm rain event as the test water (three replicate stormwater solutions at each copper concentration).

from Cu(OH)2 -G remained almost constant throughout the range of test conditions. Inactivation in effluent (aqueous phase) by Cu2+ (implications of copper leaching for this study): E. coli concentrations in copper contaminated natural stormwater were monitored (1 mg/L Cu2+ ) and results are shown in Fig. 5. There was only a slight reduction in E. coli concentration after 40 min (the maximum contact time of bacteria with 1 mg/L Cu2+ in column before being collected into EDTA-pretreated sampling bottles), which was consistent with the observation by Straub et al. [46]. In addition, this batch test was performed at a temperature (20 ◦ C) higher than the operational temperature during Event 9 (11 ◦ C) (week 23), which further ensured the minor role of inactivation in aqueous phase, since it has been reported that the rate of inactivation was less intense at lower temperatures [45]. It was concluded that the inactivation by leached copper in aqueous phase played a minor role in the observed E. coli removal by Cu-Z in Stage 2. 4. Conclusions Removal of pathogens from stormwater discharges is of high importance for effective stormwater harvesting. Filter media, after modification with antibacterial agents, may potentially improve pathogen removal efficiency of current biofilters. Benefits of using antibacterial media include instantaneous improved bacterial removal during rain events and inhibition of bacterial survival during dry events. The findings in this study can be summarised thus: Among the 15 types of antibacterial media investigated herein, only copper compound modified media exhibited robust antibacterial efficiency over 5 months of exposure to stormwater; • Cu-Z provided the best bacteria removal rates, while enough hydraulic residence time was a pre-requisite for its effectiveness. Leaching of Cu2+ from Cu-Z was positively affected by the salinity level in stormwater limiting the application of Cu-Z to catchments having medium to low salinity (i.e. below 150 ␮S/cm). • In situ precipitation of copper hydroxide on filter media, followed by heat treatment, delivered stable Cu(OH)2 coating on GAC; stable and uniform CuO coating on zeolite. More importantly, Cu(OH)2 coating on filter media enhanced bacteria removal through a synergistic effect of electrostatic attraction and inactivation; the removal performance of the media was unaffected by flow velocity; but performance in high salinity water was inferior to that in low salinity water;

Y.L. Li et al. / Journal of Hazardous Materials 271 (2014) 73–81

• Future studies should integrate benefits of both Cu2+ treated zeolite and Cu(OH)2 coating, as they can achieve efficient microbial removal with controlled leaching at variable stormwater treatment conditions pertaining to flow velocity and salinity. Acknowledgements CRC for Water Sensitive Cities are acknowledged for supporting this study. Dr. Walid Daoud and Lorena Lopez-Vanegas are greatly acknowledged for their time and effort in preparing TiO2 modified media. Louisa John-Krol is greatly acknowledged for editing grammatical errors in English. The staff from Civil Engineering, Monash University, especially Richard Williamson, Frank Winston, Peter Kolotelo, Christelle Schang, Javier Neira and Peter Poelsma, are gratefully acknowledged for their active involvement in this project. The authors acknowledge use of the facilities and the assistance of Xiya Fang at the Monash Centre for Electron Microscopy. References [1] FAWB, Guidelines for Soil Filter Media in Bioretention Systems, Facility for Advancing Water Biofiltration Monash University, Melbourne, Australia, 2009. [2] A.P. Davis, M. Shokouhian, H. Sharma, C. Minami, Laboratory study of biological retention for urban stormwater management, Water Environ. Res. 73 (2001) 5–14. [3] K. Bratieres, T.D. Fletcher, A. Deletic, Y. Zinger, Nutrient and sediment removal by stormwater biofilters: a large-scale design optimisation study, Water Res. 42 (2008) 3930–3940. [4] G.I. Chandrasena, A. Deletic, J. Ellerton, D.T. McCarthy, Evaluating Escherichia coli removal performance in stormwater biofilters: a laboratory-scale study, Water Sci. Technol. 66 (2012) 1132–1138. [5] NRMMC, EPHC, NHMRC, Australian Guidelines for Water Recycling (Phase 2): Stormwater Harvesting and Reuse Natural Resource Management Ministerial Council Environment Protection and Heritage Council, National Health and Medical Research Council Canberra, Australia 2009. [6] J.M. Hathaway, W.F. Hunt, A.K. Graves, J.D. Wright, Field evaluation of bioretention indicator bacteria sequestration in Wilmington, North Carolina, J. Environ. Eng.- ASCE 137 (2011) 1103–1113. [7] Y.L. Li, A. Deletic, L. Alcazar, K. Bratieres, T.D. Fletcher, D.T. McCarthy, Removal of Clostridium perfringens, Escherichia coli and F-RNA coliphages by stormwater biofilters, Ecol. Eng. 49 (2012) 137–145. [8] M. Auset, A.A. Keller, F. Brissaud, V. Lazarova, Intermittent filtration of bacteria and colloids in porous media, Water Resour. Res. 41 (2005). [9] K. Bratieres, C. Schang, A. Deletic, D.T. McCarthy, Performance of enviss (TM) stormwater filters: results of a laboratory trial, Water Sci. Technol. 66 (2012) 719–727. [10] Y. Matsumura, K. Yoshikata, S. Kunisaki, T. Tsuchido, Mode of bactericidal action of silver zeolite and its comparison with that of silver nitrate, Appl. Environ. Microbiol. 69 (2003) 4278–4281. [11] H. Nakashima, N. Miyano, T. Takatuka, Elution of metals with artificial sweat/saliva from inorganic antimicrobials/processed cloths and evaluation of antimicrobial activity of cloths, J. Health Sci. 54 (2008) 390–399. [12] S. Pal, J. Joardar, J.M. Song, Removal of E. coli from water using surface-modified activated carbon filter media and its performance over an extended use, Environ. Sci. Technol. 40 (2006) 6091–6097. [13] Q.T. Tran, V.S. Nguyen, T.K.D. Hoang, H.L. Nguyen, T.T. Bui, T.V.A. Nguyen, D.H. Nguyen, H.H. Nguyen, Preparation and properties of silver nanoparticles loaded in activated carbon for biological and environmental applications, J. Hazard. Mater. 192 (2011) 1321–1329. [14] Z. Milan, C. de las Pozas, M. Cruz, R. Borja, E. Sanchez, K. Ilangovan, Y. Espinosa, B. Luna, The removal of bacteria by modified natural zeolites, J. Environ. Sci. Health Pt. A: Toxic Hazard. Subst. Environ. Eng. 36 (2001) 1073–1087. [15] J. Hrenovic, J. Milenkovic, T. Ivankovic, N. Rajic, Antibacterial activity of heavy metal-loaded natural zeolite, J. Hazard. Mater. 201 (2012) 260–264. [16] J. Lukasik, Y.F. Cheng, F.H. Lu, M. Tamplin, S.R. Farrah, Removal of microorganisms from water by columns containing sand coated with ferric and aluminum hydroxides, Water Res. 33 (1999) 769–777. [17] M.M. Ahammed, V. Meera, Metal oxide/hydroxide-coated dual-media filter for simultaneous removal of bacteria and heavy metals from natural waters, J. Hazard. Mater. 181 (2010) 788–793.

81

[18] L.J. Kennedy, A.G. Kumar, B. Ravindran, G. Sekaran, Copper impregnated mesoporous activated carbon as a high efficient catalyst for the complete destruction of pathogens in water, Environ. Prog. 27 (2008) 40–50. [19] D.T. McCarthy, J.M. Hathaway, W.F. Hunt, A. Deletic, Intra-event variability of Escherichia coli and total suspended solids in urban stormwater runoff, Water Res. 46 (2012) 6661–6670. [20] W.A. Daoud, J.H. Xin, Y.H. Zhang, Surface functionalization of cellulose fibers with titanium dioxide nanoparticles and their combined bactericidal activities, Surf. Sci. 599 (2005) 69–75. [21] S.B. Wang, Z.H. Zhu, Characterisation and environmental application of an Australian natural zeolite for basic dye removal from aqueous solution, J. Hazard. Mater. 136 (2006) 946–952. [22] D. Xu, P. Xiao, J. Zhang, G. Li, G. Xiao, P.A. Webley, Y. Zhai, Effects of water vapour on CO2 capture with vacuum swing adsorption using activated carbon, Chem. Eng. J. 230 (2013) 64–72. [23] A. Gupta, M. Chaudhuri, Enteric virus removal inactivation by coal-based media, Water Res. 29 (1995) 511–516. [24] Z.H. Li, S.J. Roy, Y.Q. Zou, R.S. Bowman, Long-term chemical and biological stability of surfactant modified zeolite, Environ. Sci. Technol. 32 (1998) 2628–2632. [25] A.J. Isquith, E.A. Abbott, P.A. Walters, Surface-bonded antimicrobial activity of an organosilicon quaternary ammonium chloride, Appl. Microbiol. 24 (1972) 859–863. [26] L. Lopez, W.A. Daoud, D. Dutta, Preparation of large scale photocatalytic TiO2 films by the sol–gel process, Surf. Coat. Technol. 205 (2010) 251–257. [27] S. Babel, T.A. Kurniawan, Low-cost adsorbents for heavy metals uptake from contaminated water: a review, J. Hazard. Mater. 97 (2003) 219–243. [28] D. Stojakovic, J. Hrenovic, M. Mazaj, N. Rajic, On the zinc sorption by the Serbian natural clinoptilolite and the disinfecting ability and phosphate affinity of the exhausted sorbent, J. Hazard. Mater. 185 (2011) 408–415. [29] S. Kratohvil, E. Matijevic, Preparation of copper-compounds of different compositions and particle morphologies, J. Mater. Res. 6 (1991) 766–777. [30] J.W.H. Smith, P. Westreich, A.J. Smith, H. Fortier, L.M. Croll, J.H. Reynolds, J.R. Dahn, Investigation of copper oxide impregnants prepared from various precursors for respirator carbons, J. Colloid Interface Sci. 341 (2010) 162–170. [31] R.S. Razavi, M.R. Loghman-Estarki, Synthesis and characterizations of copper oxide nanoparticles within zeolite Y, J. Cluster Sci. 23 (2012) 1097–1106. [32] B.G. Kutchko, A.G. Kim, Fly ash characterization by SEM-EDS, Fuel 85 (2006) 2537–2544. [33] G.G. Ren, D.W. Hu, E.W.C. Cheng, M.A. Vargas-Reus, P. Reip, R.P. Allaker, Characterisation of copper oxide nanoparticles for antimicrobial applications, Int. J. Antimicrob. Agents 33 (2009) 587–590. [34] R.E. Kirk, Statistics: An Introduction, fifth ed., Thomson/Wadsworth, Belmont, CA, 2008. [35] L.J. Albright, J.W. Wentwort, E.M. Wilson, Technique for measuring metallic salt effects upon indigenous heterotrophic microflora of a natural water, Water Res. 6 (1972) 1589–1596. [36] Z.M. Gu, J. Fang, B.L. Deng, Preparation and evaluation of GAC-based ironcontaining adsorbents for arsenic removal, Environ. Sci. Technol. 39 (2005) 3833–3843. [37] J.P. Chen, S.N. Wu, K.H. Chong, Surface modification of a granular activated carbon by citric acid for enhancement of copper adsorption, Carbon 41 (2003) 1979–1986. [38] R.S. Bowman, Applications of surfactant-modified zeolites to environmental remediation, Microporous Mesoporous Mater. 61 (2003) 43–56. [39] O.M. Virgadamo, L. Johnson, J.L. Darby, Evaluation of antimicrobial coatings for cloth media filtration: case study, J. Environ. Eng.- ASCE 133 (2007) 117–120. [40] E. Grabinska-Sota, Genotoxicity and biodegradation of quaternary ammonium salts in aquatic environments, J. Hazard. Mater. 195 (2011) 182–187. [41] G. Borkow, J. Gabbay, Copper as a biocidal tool, Curr. Med. Chem. 12 (2005) 2163–2175. [42] NRMMC–EPHC–NHMRC, Australian Guidelines for Water Recycling Augmentation of Drinking Water Supplies, 2008. [43] ANZECC/ARMCANZ, Australian and New Zealand Guidelines for Fresh and Marine Water Quality, Australian and New Zealand Environmental Conservation Council and Agriculture and Resource Management Council of Australia and New Zealand, Editor 2000. [44] A.Z. Woinarski, I. Snape, G.W. Stevens, S.C. Stark, The effects of cold temperature on copper ion exchange by natural zeolite for use in a permeable reactive barrier in Antarctica, Cold Regions Sci. Technol. 37 (2003) 159–168. [45] G. Faundez, M. Troncoso, P. Navarrete, G. Figueroa, Antimicrobial activity of copper surfaces against suspensions of Salmonella enterica and Campylobacter jejuni, BMC Microbiol. 4 (2004). [46] T.M. Straub, C.P. Gerba, X. Zhou, R. Price, M.T. Yahya, Synergistic inactivation of Escherichia coli and MS-2 coliphage by chloramine and cupric chloride, Water Res. 29 (1995) 811–818.