Abstract
Low-cost granular filter media with hybrid bacterial adsorption and survival inhibition capability is highly desired for the development of a low-impact water filtration system. In addition to overall removal, a deeper understanding of the fate and transport behaviour of bacteria in such systems should also be obtained to guide system operation. In this study, copper(II) hydroxide nanoparticles-modified granular activated carbon via a single-step in situ coating was prepared and denoted as CuH-G. Copper release behaviour and Escherichia coli removal efficiency of CuH-G were studied in saturated columns as a function of salinity, flow rate, and hydraulic loading. Copper release decreased exponentially on increasing salinity in test water, which potentiates controlled copper release for desired bacteria inhibition efficiency. With an effective contact time of 3.7 min, CuH-G provided 3.0 and 1.6 log E. coli removal in test water of salinity 237 and 680 μS/cm, respectively. Copper leaching at these two salinities were 1.7 and 0.74 mg/l, respectively below the Australian Guidelines for Water Recycling: Augmentation of Drinking Water Supplies. Further study of E. coli transport and deposition behaviour in heat-treated CuH-G at 160 °C revealed that the observed removal was largely attributed to enhanced attachment during filtration and survival inhibition post filtration.
HIGHLIGHTS
Stable copper hydroxide coating on porous media through simple in situ precipitation.
Filters packed with the modified media removed bacteria at a fast flow rate.
Copper release was controlled by salinity in test water and the below guideline value.
Breakthrough curve modelling revealed attachment as the key process during filtration.
Intermittent operation enabled adsorbed bacteria inactivated by locally released copper.
Graphical Abstract
INTRODUCTION
The availability of safe drinking water has been one of the biggest challenges worldwide. Despite many advances in technologies, this challenge is expanding due to ongoing population growth, more frequent extreme weather, and new water contamination opportunities including the spread of notorious pathogenic organisms. This issue must be addressed in part by harnessing various water resources by removing physical, chemical, and microbial contaminants in order to properly deliver safe water for human consumption (Persaud et al. 2019; Shen et al. 2020; Parker et al. 2021; Zhang et al. 2021; Feng et al. 2022; Galbraith et al. 2022). Centralised water treatment systems include conventional water treatment, disinfection, filtration, and post-disinfection. Granular activated carbon (GAC) filtration has been widely used in such systems to eliminate disinfection by-products, heavy metals, micro-pollutants, and natural organic matters largely due to its extremely large surface area providing high adsorption rates and capabilities. For the same reason, activated carbon has been used in most household drinking water filters. However, in general, the elimination capacity for microbes by unmodified activated carbon has been assumed to be negligible (Hijnen et al. 2010) partly due to the unfavourable electrostatic interaction between its negatively charged surface and negatively charged microbes. Furthermore, its inherent excellent biocompatibility with microbes can support adsorbed microbes' survival, accumulation, growth, and biofilm formation, thus eventually becoming a source of microbial contaminants leading to filtration system clogging.
Modification using metal hydroxide or oxide has proven to be effective in reducing or reversing the net negative surface charges of granular media sand, coal, activated carbon, zeolite, and significantly improving microbial removal through electrostatic attraction (Ahammed & Meera 2010). Sand coated with separate or combined ferric and aluminium hydroxide removed more than 99% Escherichia coli (Lukasik et al. 1999). Zeolite coated with ZnO had a measured zeta potential of about 20 mV at a neutral pH and isoelectric point of 11.2 resulting in a positive surface charge when the solution pH was below 11.2 (Wang et al. 2016). Sand coated with iron oxide had a net positive charged surface and retained 98–99% E. coli in the biofiltration system which was much higher than that from uncoated sand columns (70% retention) (Mohanty et al. 2013). Activated carbon modified with 1% aluminium hydroxychloride exhibited over 114% higher E. coli removal efficiency than that modified with 0.5% aluminium hydroxychloride (Pal et al. 2006). One potential challenge of such a system is the accumulation, saturation, biofilm formation, clogging, and loss of function over time due to the lack of adequate inactivation of adsorbed microbes. Furthermore, accumulative deposition of bacteria cells in filtrations system either on filter media or at pore constrictions may lead to aggregation and when reaching a critical concentration be released into aqueous solution as a result of hydrodynamic shearing forces (Bradford et al. 2006).
A hybrid of biocide and metal oxide/hydroxide coating on granular media has thus been studied to achieve the inactivation of adsorbed microbes (Gupta & Chaudhuri 1995; Zhang et al. 2016a; Wang et al. 2018; Bahcelioglu et al. 2021). Silver-based nanoparticles impregnated onto filter media for water disinfection have been well investigated with promising outcomes (Gupta & Chaudhuri 1995; Biswas & Bandyopadhyaya 2016; Zhang et al. 2016a; Jung et al. 2019; Bahcelioglu et al. 2021). For example, maintaining a contact time of 25 min, a column packed with silver nanoparticles treated with activated carbon can completely disinfect E. coli-contaminated water for up to 16 days of continuous operation (Biswas & Bandyopadhyaya 2016). Alum-silver-modified coal packed in downflow columns with a flowthrough time of 7.7 min removed 59.3–86.5% viruses along with more than 99% indigenous heterotrophic bacteria (Gupta & Chaudhuri 1995). Inactivation efficiency by silver-modified activated carbon in the column test was observed to be sensitive to contact time – complete E. coli removal at a contact time of 23 min while about 90% removal at a contact time of 19 min (Biswas & Bandyopadhyaya 2016). Impacting factors to the desired contact time included, but were not limited to, Ag loading (e.g. increasing mass content of Ag-activated carbon in cell suspension from 2 to 8 mg/ml decreased the required time to achieve the same level of bacterial killing (Biswas & Bandyopadhyaya 2016)), microbial species, and concentration (e.g. B. subtilis and E. coli were completely inhibited within 10 and 60 min, respectively (Yoon et al. 2008), and temperature. This suggests the necessity of the hybrid design – combined microbial adsorption and inactivation process.
Copper, the second most effective bactericidal metal after silver, has been a US Environmental Protection Agency-registered pesticide for agricultural crops (USEPA 2008). Furthermore, due to its low cost, copper has found wide commercial applications in water, agriculture, fabrics, marine industry, etc. (Vincent et al. 2016, 2018). Despite its promising antimicrobial efficiency, the application of copper in particular its oxide or hydroxide form in water filters has rarely been explored. Kennedy et al. (2008) examined the bactericidal effects of CuO/Cu2O-coated carbon and showed a 4 log removal of E. coli. However, the expensive organometallic precursor and toxic organic solvent involved in the fabrication could be a barrier for large-scale field applications. Li et al. (2014a) prepared Cu(OH)2-activated carbon through simple in situ precipitation and tested its performance using natural stormwater at ambient conditions over 16 weeks (temperature range: 11–21 °C). About 2 log E. coli reduction was observed in the presence of other native microbial communities. However, the conclusion was based on only one filter. Furthermore, there is no reported research investigating bacteria transport and deposition behaviour in filtration systems using antimicrobial filter media.
This study aims to explore the potential of using copper hydroxide as a modifier of GAC to achieve combined enhanced bacteria attachment and inhibition in the water filtration system. Copper hydroxide-modified media was developed through simple in situ precipitation at room temperature. The composition of coating layer was further adjusted through heat treatment at various temperatures. Copper release from the various modified media was investigated in terms of hydraulic loading, flow rate, and test water salinity. The dynamic bacteria removal performance and the impacting factors were investigated against E. coli, chosen as an indicator of faecal contamination through a downward filtration test. The fate and transport behaviour of bacteria in antimicrobial media-packed filtration system was examined through breakthrough curve monitoring and modelling. The research findings could benefit development of passive water filtration systems.
MATERIALS AND METHODS
Modification of GAC with copper hydroxide/oxide
The chemicals (CAS number in parentheses) and their sources, featured in this study, comprise copper(II) chloride anhydrous (7447-39-4), sodium hydroxide (1310-73-2), and ethylenediaminetetraacetic acid disodium salt (EDTA) (6381-92-6), Merck Chemicals, Australia. GAC (milled and sieved to grain size 0.3–0.6 mm), Activated Carbon Technologies Pty Ltd, constituted base media. The basic physicochemical properties of GAC were detailed by Xu et al. (2013).
Coating of GAC by copper hydroxide
The preparation procedure was similar to that in Li et al. (2014a). In brief, GAC was gently mixed with 1% (wt.) CuCl2 solution for 1 h, after which the pH of the slurry was adjusted to 8 using 2 M NaOH. After mixing for 1 h, the granular media were separated, washed with deionised water (DI water), and dried at 60 °C overnight. The media so-prepared was denoted as CuH-G while unmodified GAC as G0.
Heat treatment of copper hydroxide-coated GAC
A thermal treatment of CuH-G was made to tailor the composition of Cu(OH)2 coating on GAC introducing mixed copper hydroxide and copper oxide. This is based on the fact that Cu(OH)2 is a metastable phase and can be transformed into copper oxide through dehydration beginning at 150 °C and the dehydration rate is influenced by temperature (Cudennec & Lecerf 2003). Filter media CuH-G was heated at a rate of 5 °C/min to 160 °C and maintained at that temperature for 1 h before cooling naturally to room temperature using LABEC Muffle Furnace CEMLL. The filter media were then washed five times with DI water and dried at 105 °C overnight to attain the final product denoted as CuH-G1601h, CuH-G1801h, CuH-G1803h, and CuH-G2201h were prepared similarly but at raised temperatures and varied times. G01801h was prepared by heat treatment of G0 at 180 °C for 1 h and served as a control media to evaluate the influence of thermal treatment on coating stability and bacterial removal.
The copper 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 Energy-dispersive X-ray spectroscopy (EDS).
Copper hydroxide coating stability and impacting factors
Copper hydroxide coating stability on GAC was examined using DI water with varied salinities. The five types of copper hydroxide-modified GAC and untreated GAC as control (G0) were wet packed in polyvinyl chloride (PVC) columns (internal diameter 29.5 mm with sand-blasted inner surface, media depth 70 mm) with three replicates of each media type. Once packed, all columns were flushed using 21 × 80 ml pulses of DI water (top-to-bottom) to remove the fines produced during packing and to allow the media to settle. During flushing, total copper in the effluent of each column was monitored and the coefficient of variation of each type of media was at or below 10% (8, 10, 5, 5, and 5% for CuH-G, CuH-G1601h. CuH-G1801h, CuH-G1803h, and CuH-G2201h, respectively) indicating excellent replication. Hence, during the following coating stability tests, all columns were dosed equally; however, to save cost, the effluent from one replicate of each media was monitored for copper concentration. The experiment was designed to examine the impact factors on coating stability including (a) flow rate – eight pore volumes of DI water were applied at two flow rates: near unrestricted flow, 5.27 m/h and restricted flow 0.57 m/h (restricting the column outlet); (b) salinity – eight pore volumes of DI water with decreasing salinity (231, 117, 73, and 5 μS/cm) was applied to the columns at a flow rate of 0.57 m/h; and (c) hydraulic loading – accumulative 76 pore volumes of DI water with salinity 680 μS/cm were applied to the columns at a flow rate of 0.57 m/h through four intermittent events with 16, 20, 20, and 20 pore volumes, respectively, and composite effluent was collected during each event. DI water with desired salinity was obtained by dissolving reagent grade NaCl in DI water without adjusting pH. Salinity was measured as electrical conductivity using a multi-parameter water quality probe (U50, Horiba Ltd, Japan). DI water before addition of NaCl was measured to have a salinity of 5 μS/cm. Copper concentration in effluent samples was measured using ICP-MS in a NATA-accredited laboratory. All experiments were conducted at room temperature.
E. coli removal by modified media and impacting factors
Preparation of test water
E. coli ATCC#11775 was provided by ALS Environmental, Melbourne for use in all experiments. Test water, i.e. DI water inoculated with E. coli at a concentration of about 104 MPN/100 ml was prepared through two rounds of dilution using DI water and was used on the same day. Salinity of the test water was adjusted using NaCl.
Effect of salinity
Test water of two levels of salinity 267 and 680 μS/cm was applied to the columns at a flow rate of 0.57 m/h over two intermittent events: 8 and 16 pore volumes of test water, respectively. During each event, duplicate composite inflow samples were taken, while the entire outflow from each column was collected. The sterilised bottles, used for sample collection, were pre-treated with EDTA in order to eliminate post-filtration inactivation (Li et al. 2014b). E. coli concentrations in all inflow and outflow samples were analysed using the Colilert™ method (IDEXX-Laboratories, 2007) as per manufacturer's instructions.
Effect of hydraulic loading
The DI water used in hydraulic loading test (see Section 2.2) was inoculated with E. coli at a concentration of about 104 MPN/100 ml. During each intermittent event, duplicate composite inflow samples were taken, while the entire outflow from each column was collected for E. coli concentration analysis using the Colilert™ method (IDEXX-Laboratories, 2007). The intermittent mode of operation (i.e. 16–20 pore volumes of doing followed by a dry period) may mobilise the previously attached bacteria representing challenging operational conditions (Mohanty et al. 2013).
E. coli transport and deposition behaviour in modified GAC media
G0 and CuH-G1601h were packed into clear plastic columns (internal diameter: 31 mm, media depth: 145 mm which were higher than those mentioned in Section 2.3 aiming for E. coli deposition profile observation) and four replicates for each media type. Once packed, all columns were flushed (top-to-bottom) using 18 × 80 ml pulses of DI water without restricted outflow. One G0 column and one CuH-G1601h column were then used for the tracer test and the other columns were tested for E. coli breakthrough following the protocol developed by Bradford et al. (2006). During the tests, the columns were dosed with DI water or test water from the top by maintaining a constant head at the upper media level and the column's outlet was restricted to attain the desired flow of rate 0.8 m/h.
Tracer test
The columns were dosed with 2.5 pore volumes of DI water with a salinity of 667 μS/cm followed by four pore volumes of DI water. Effluent samples were continuously collected at 20-ml intervals and were then analysed for electrical conductivity and sodium concentration. The accumulative time to collect the samples was recorded.
E. coli breakthrough test
The columns were first equilibrated with eight pore volumes of DI water with a salinity of 667 μS/cm. Test water with salinity 667 μS/cm and E. coli concentration of about 105 MPN/100 ml, was then applied using a peristaltic pump at a flow rate of 0.8 m/h for two pore volumes, followed by four pore volumes of E. coli-free test water. Column's outlet valves were adjusted to attain the desired constant head and effluent samples were continuously collected at 20-ml intervals and the accumulative time to collect the samples were recorded. E. coli concentrations in all inflow and outflow samples were analysed using the ColilertTM method.
Subject to boundary conditions C = C0 at x = 0 and = 0 at x =
.
where C is the microbe concentration (MPN/m3) in an aqueous phase at a distance x and time t; S is the microbe concentration on porous media (MPN/kg); t is the time (h); x is the distance from the top of the filter media in each column (m); is the filter media dry bulk density (kg/m3);
is the porosity =
where Q is the flow rate (m3/h) and A is the cross-sectional area (m2); D is the hydrodynamic dispersion coefficient (m2/h); v is the average pore water velocity (m/h); dispersivity
,
are the first-order attachment and detachment rate coefficients (h−1), respectively;
and
are the inactivation rates of the free and attached cells (h−1), respectively.
Tracer breakthrough curve was used to find out the best-fit value D, . μ1 was found by measuring E. coli inactivation in column effluent suspensions. All other parameters including
were found by simulating the microbe transport data using Equations (1) and (2) through HYDRUS 1D computer program (Bradford et al. 2003; Schijven et al. 2003; Hijnen et al. 2005; Zhang et al. 2010).
Deposition profile
Deposition profile of bacteria across the depth of the filter media of each column was determined following the completion of breakthrough sampling. Filter media was excavated in six layers of equal length. E. coli concentration on media was obtained for each layer following a similar procedure developed by Li et al. (2014a) except that DI water of salinity 667 μS/cm with dispersant was used in this study. The E. coli concentration was expressed as MPN/g dry media.
Data analysis
Where the outflow concentration was below the detection limit, half the detection limit was factored as the concentration for analysis. The E. coli log removal rate was calculated as the difference between log concentrations in inflows and outflows. Nonlinear exponential regression tests of copper leaching versus salinity in DI water, and linear regression tests of copper leaching versus hydraulic loading were utilised for correlation analysis. One-way ANOVA of log removal rates and copper leaching along with post hoc tests – Tukey was performed using SPSS Statistics v28 to test the significant difference between various media.
RESULTS AND DISCUSSION
Characterisation of copper hydroxide/oxide-modified GAC
The initial copper content on freshly prepared media was summarised in Table 1. 16 mg of copper was loaded onto per gram GAC through precipitation in preparing CuH-G, i.e. Cu(OH)2-coated GAC. The post-thermal treatment and rinsing of CuH-G did not significantly detach the immobilised copper and samples showed comparable level of copper content indicating strong adhesion of Cu(OH)2 on GAC. Post-thermal treatment of CuH-G at high temperature or for prolonged time, i.e CuH-G2201h, CuH-G1803h turned the blue particle into black or partially black (observed by naked eyes) indicating the transformation of Cu(OH)2 into CuO.
Copper-modified GAC, their stability, and Escherichia coli removal efficiency during the filtration test
Filter media . | Performance and stability . | |||
---|---|---|---|---|
Media . | Copper hydroxide/oxide modification on GAC . | Copper content (mg/g dry media)a . | Median E. coli log removalb . | Mean copper leaching (mg/l)c . |
G0 | Unmodified GAC | 0.66 ( ± 0.09) | ||
G01801h | G0 treated at 180 °C for 1 h | 0.86 ( ± 0.12) | ||
CuH-G | Cu(OH)2in situ coating | 16 | 1.58 ( ± 0.21) | 0.51 ( ± 0.19) |
CuH-G1801h | CuH-G treated at 180 °C for 1 h | 13 | 1.52 ( ± 0.23) | 0.56 ( ± 0.23) |
CuH-G1601h | CuH-G treated at 160 °C for 1 h | 14 | 1.55 ( ± 0.15) | 0.52 ( ± 0.15) |
CuH-G1803h | CuH-G treated at 180 °C for 3 h | 16 | 1.55 ( ± 0.18) | 0.71 ( ± 0.22) |
CuH-G2201h | CuH-G treated at 220 °C for 1 h | 13 | 1.08 ( ± 0.10) | 0.47 ( ± 0.17) |
Filter media . | Performance and stability . | |||
---|---|---|---|---|
Media . | Copper hydroxide/oxide modification on GAC . | Copper content (mg/g dry media)a . | Median E. coli log removalb . | Mean copper leaching (mg/l)c . |
G0 | Unmodified GAC | 0.66 ( ± 0.09) | ||
G01801h | G0 treated at 180 °C for 1 h | 0.86 ( ± 0.12) | ||
CuH-G | Cu(OH)2in situ coating | 16 | 1.58 ( ± 0.21) | 0.51 ( ± 0.19) |
CuH-G1801h | CuH-G treated at 180 °C for 1 h | 13 | 1.52 ( ± 0.23) | 0.56 ( ± 0.23) |
CuH-G1601h | CuH-G treated at 160 °C for 1 h | 14 | 1.55 ( ± 0.15) | 0.52 ( ± 0.15) |
CuH-G1803h | CuH-G treated at 180 °C for 3 h | 16 | 1.55 ( ± 0.18) | 0.71 ( ± 0.22) |
CuH-G2201h | CuH-G treated at 220 °C for 1 h | 13 | 1.08 ( ± 0.10) | 0.47 ( ± 0.17) |
aTotal copper content in a freshly prepared media.
bMedian of 12 data (±standard deviation) – E. coli log removal rate during accumulative 76 pore volumes of E. coli-contaminated test water (salinity 680 μS/cm, E. coli concentration about 104 MPN/100 ml) applied to each column at a flow rate of 0.57 m/h through four intermittent events.
cMean of 4 data (±standard deviation) copper leaching during the 76 pore volumes of E. coli-contaminated test water dosing.
SEM images of (a) G0 and (b) CuH-G1601h (inlet at higher magnification: scale bar − 100 nm).
SEM images of (a) G0 and (b) CuH-G1601h (inlet at higher magnification: scale bar − 100 nm).
Coating stability of copper hydroxide/oxide-modified GAC and impacting factors
The dynamic stability of five types of copper hydroxide/oxide-coated GAC was investigated using PVC columns (top-to-bottom) regarding the impact of DI water on flow rate, salinity, and hydraulic loading.
Effect of the flow rate
Effect of the flow rate on effluent copper concentration in DI water at two flow rates (5.27 and 0.57 m/h as indicated in the figure legends). The various filter media types are indicated in x-axis and Table 1.
Effect of the flow rate on effluent copper concentration in DI water at two flow rates (5.27 and 0.57 m/h as indicated in the figure legends). The various filter media types are indicated in x-axis and Table 1.
Effect of salinity
Effect of the salinity level in test water on effluent copper concentration at a flow rate of 0.57 m/h. The various filter media types are indicated in the figure legends and Table 1. Nonlinear exponential regression coefficients (R2): CuH-G = 0.997; CuH-G1601h = 0.987; CuH-G1801h = 1.0; CuH-G1803h = 0.998; CuH-G2201h = 0.997.
Effect of the salinity level in test water on effluent copper concentration at a flow rate of 0.57 m/h. The various filter media types are indicated in the figure legends and Table 1. Nonlinear exponential regression coefficients (R2): CuH-G = 0.997; CuH-G1601h = 0.987; CuH-G1801h = 1.0; CuH-G1803h = 0.998; CuH-G2201h = 0.997.
Effect of hydraulic loading
Effect of hydraulic loading of test water (salinity 680 μS/cm and flow rate 0.57 m/h) on (a) effluent copper concentration and (b) Escherichia coli logarithmic removal. The various filter media types are indicated in the figure legends and Table 1. Each data point in E. coli logarithmic removal rate curves represents three replicates and is expressed as median and 95% confidence interval.
Effect of hydraulic loading of test water (salinity 680 μS/cm and flow rate 0.57 m/h) on (a) effluent copper concentration and (b) Escherichia coli logarithmic removal. The various filter media types are indicated in the figure legends and Table 1. Each data point in E. coli logarithmic removal rate curves represents three replicates and is expressed as median and 95% confidence interval.
E. coli removal by modified media and impacting factors
Effect of the filter media type
The median E. coli log removal rate by each type of media is summarised in Table 1. In general, copper-modified GAC columns showed significantly better E. coli removal than G0 (p < 0.001) supporting the hypothesis of improved effectiveness due to combined contact-killing (at solid–liquid interface and in aqueous phases) and favourable electrostatic adsorption of negatively charged bacterial cells (zeta potential of copper hydroxide at neutral pH is +20 mV (Albrecht et al. 2011)). While all other copper hydroxide-modified GAC performed similarly, CuH-G2201h showed significantly less E. coli removal (p < 0.001), which is hypothesised to be due to the less electrostatic interaction of CuO (present on CuH-G2201h) with E. coli cells than that of copper hydroxide (present on other copper-modified GAC) (Li et al. 2014a; Zhang et al. 2016b). In the following sections, the performance of the various designs was further explored as a response to varied test water salinities and hydraulic loadings.
Effect of test water salinity
Effect of two salinity levels (267 and 680 μS/cm as indicated in the figure legends) on (a) Escherichia coli logarithmic removal and (b) effluent copper concentration at a flow rate of 0.57 m/h. Various filter media types are indicated in x-axis and Table 1. Each boxplot in (a) represents three replicates and each bar in (b) represents one measurement.
Effect of two salinity levels (267 and 680 μS/cm as indicated in the figure legends) on (a) Escherichia coli logarithmic removal and (b) effluent copper concentration at a flow rate of 0.57 m/h. Various filter media types are indicated in x-axis and Table 1. Each boxplot in (a) represents three replicates and each bar in (b) represents one measurement.
Effect of hydraulic loading
The changes in E. coli log removal rates during 76 pore volumes of dosing over four intermittent events are graphed in Figure 4(b). Linear regression analysis showed that E. coli removal by CuH-G2201h, CuH-G, and CuH-G1803h decreased with increasing pore volumes (p < 0.05) while that by CuH-G1601h and CuH-G1801h remained constant throughout the testing period. As discussed earlier, copper leaching decreased significantly over hydraulic loading (Figure 4(a)). In this regard, linear correlations between E. coli removal rates and copper leaching were analysed for each media design using SPSS to investigate the impact of such decreased copper leaching at fixed salinity 680 μS/cm. Lack of correlation between the two was observed for every media type (p > 0.05) suggesting that the leached copper was at a low level (0.3–1.0 mg/l) and did not play a significant role in contributing to E. coli removal during filtration due to the very short residence time of test water in the modified media (3.7 min based on porosity 51%). In the aqueous phase, the true contact time with the effluent level of copper was much shorter assuming that copper concentration linearly increased from inlet to outlet. An earlier study showed that 1-ppm copper would not induce significant E. coli die-off within 40 min (Li et al. 2014a). Therefore, for such stable copper-modified media, the inactivation is assumed to occur through prolonged contact on the solid phase due to the locally available copper ions (Kennedy et al. 2008; Li et al. 2014a; Wang et al. 2018). This hypothesis was proven and quantified through breakthrough data modelling in the following section.
E. coli transport and deposition behaviour in copper hydroxide-modified GAC columns
Transport behaviour
Breakthrough curves of Escherichia coli in the G0 and CuH-G1601h columns and Hydrus-1D simulation at flow rate 0.8 m/h and test water salinity 667 μS/cm (the points of three symbols are the observed values of three replicates of each design and the line indicates the Hydrus-1D model fit).
Breakthrough curves of Escherichia coli in the G0 and CuH-G1601h columns and Hydrus-1D simulation at flow rate 0.8 m/h and test water salinity 667 μS/cm (the points of three symbols are the observed values of three replicates of each design and the line indicates the Hydrus-1D model fit).
Table 2 provides a summary of the estimated parameter values from fitting the breakthrough curves for the two different filter media fitted models. A good correlation was observed for G0 (R2 = 0.85) but not for CuH-G1601h (R2 = 0.12). One explanation for the poorer correlation for CuH-G1601h is that the effluent E.coli concentrations are much lower (sometimes be near or below detection limits), which comes with higher uncertainties (harder to measure low concentrations).
Hydrodynamic parameters of the columns and Hydrus-1D parameters for Escherichia coli removal by GAC filters with and without copper
. | G0 . | CuH-G1601h . | Test column (ID 31 mm) . |
---|---|---|---|
D50 | 0.450 mm | 0.475 mm | ![]() |
Pore velocity, ν (m/h) | 0.972 | 1.566 | |
Porosity, θ (%) | 81.9 | 52.4 | |
Hydrodynamic dispersion coefficient, D (m2/h) | 0.00602 | 0.00235 | |
Dispersivity αL (m) | 0.0062 | 0.0015 | |
μ1 (h−1) | 0.03 | 0.27 | |
μs (h−1) | 0 | 0.16 | |
katt (h−1) | 16.4 | 159 | |
kdett (h−1) | 0.024 | 0.15 | |
R2 | 0.85 | 0.12 |
. | G0 . | CuH-G1601h . | Test column (ID 31 mm) . |
---|---|---|---|
D50 | 0.450 mm | 0.475 mm | ![]() |
Pore velocity, ν (m/h) | 0.972 | 1.566 | |
Porosity, θ (%) | 81.9 | 52.4 | |
Hydrodynamic dispersion coefficient, D (m2/h) | 0.00602 | 0.00235 | |
Dispersivity αL (m) | 0.0062 | 0.0015 | |
μ1 (h−1) | 0.03 | 0.27 | |
μs (h−1) | 0 | 0.16 | |
katt (h−1) | 16.4 | 159 | |
kdett (h−1) | 0.024 | 0.15 | |
R2 | 0.85 | 0.12 |
The values for porosity (θ), dispersivity (αL), and pore velocity (ν) in Table 2 were obtained by simulating the tracer breakthrough curves. In this study, CuH-G1601h columns had much smaller dispersivity than control columns indicating that Cu(OH)2 coating modified the surface morphology of GAC porous media (Figure 1) and reduced pore-scale distribution of flow velocity and/or concentration. The estimated porosity of CuH-G1601h columns was 52.4% which was much lower than that of control columns 81.9% due to the coverage of macro-mesopores thus reduced accessibility to the internal porous structure.
For control columns (G0), it was obvious that attachment played a dominant role in removing E. coli with estimated attachment rate of 16.4 h−1, while the other processes were deemed negligible as seen from the estimated parameter coefficients. Attachment is dependent on the cell–cell, cell–water, cell–porous media interactions including electrostatic, van der Waals, hydrodynamic, hydration, hydrophobic, and steric interactions. Both G0 and E. coli cells carry net negative charges and thus are not an ideal couple for attachment based on the DLVO theory. The observed E. coli attachment herein, was likely a result of the high salinity in the test water (electrical conductivity 667 μS/cm) causing compression of the electrical double layer and a reduction in electrostatic repulsive force (Stevik et al. 2004; Zhang et al. 2010). Non-DLVO factors such as hydrophobic interaction and steric interactions may also contribute to the attachment of bacteria on activated carbon due to its hydrophobicity (Biswas & Bandyopadhyaya 2016), very large surface area, presence of macropores, etc. (Kim et al. 2009). In addition, the presence of a trace amount of multivalent metal species in G0 (aluminium, iron, copper, and zinc, 0.03% by weight, ICP-MS analysis) may also contribute to the deposition of bacteria cells through favourable electrostatic association. As reported, even small amount of metal oxide in granular media proved to be effective in enhancing virus and bacteria attachment and vice versa (Rivera-Utrilla et al. 2001; Hijnen et al. 2010). As most bacteria are negatively charged because of the predominance of the anionic groups present within the cell wall, a positively charged collector should enhance bacteria deposition. Therefore, as observed in this study, alteration of the surface charges of the granular media to be positive is expected to remove the repulsive double-layer interaction and result in better bacteria removal efficiency.




Escherichia coli deposition profile in G0 and CuH-G1601h columns: E. coli concentration (MPN/g dry media) changes from top layer to bottom layer. Each data point represents three replicates and is expressed as median and 95% confidence interval.
Escherichia coli deposition profile in G0 and CuH-G1601h columns: E. coli concentration (MPN/g dry media) changes from top layer to bottom layer. Each data point represents three replicates and is expressed as median and 95% confidence interval.
Deposition profile
The deposition curves of both G0 and CuH-G1601h columns followed an almost exponential decrease in cell concentration with increasing travel distance until the detection limit was reached in the sixth layer for G0 and third layer for CuH-G1601h columns (Figure 7). The absence of a sharp decrease from the column inlet confirmed the hypothesis of negligible role of straining in removing E. coli in the column experiment. Obtained under uniform colloid and collector characteristics and stable filtration conditions, this observation was consistent with the prediction by irreversible attachment models, and confirmed the dominant role of attachment in E. coli removal processes. In such system, E. coli cells were transported by advection through filter media and their attachment on filter media firstly took place near the inlet and moved along the depth of the filter. Hence, increasing the travel distance, i.e. the lengths of the columns imply improved E. coli removal performance. However, designs using copper-modified media may achieve higher removal performance with much shorter filed depth.
The retained E. coli in CuH-G1601h columns was about one order of magnitude lower than that in G0 columns. This represents an apparent contradiction to the measured 0.8 log higher E. coli removal by CuH-G1601h columns (Figures 6 and 7), which confirmed the contribution of bacterial inactivation by the copper-modified media CuH-G1601h. The recovery from both G0 and CuH-G1601h columns were very low, which is not surprising considering the greatly low ratio of detachment/attachment rate coefficient, i.e. 600–1,000. Future research may apply a combined surfactants and/or through cumulative extraction (Lukasik et al. 1999; Hijnen et al. 2010). Nevertheless, despite measurement uncertainties due to detection limits for E. coli, relative differences between configurations were observed (Hijnen et al. 2010). Viability of the cells (or viable but non-culturable state) could be investigated by alternative methods to understand the extent of damage (Pal et al. 2006).
CONCLUSIONS
Modifying GAC with copper hydroxide/oxide significantly improved E. coli removal from test water during high flow rate filtration (effective contact time: 3.7 min) and effectively inhibited the survival of retained bacteria during prolonged contact. Copper is low cost and can be easily coated onto GAC and copper leaching is controllable by adjusting the salinity of the test water. Therefore, large-scale application of the copper-modified GAC for drinking water treatment is economically viable. The following are the major conclusions of this study:
E. coli removal in the copper hydroxide/oxide-modified GAC columns were controllable by adjusting the salinity of test water.
The performance remained stable during intermittent application of 76 pore volumes of test water.
A breakthrough curve study revealed that E. coli removal by the modified GAC was mainly attributed to attachment during filtration and inactivation during the intermittent dry period.
Copper hydroxide/oxide coating on GAC showed good stability – when test water has higher than 250 μS/cm salinity, the leached copper is below drinking water guideline; the copper leaching decreased during continuous operation.
ACKNOWLEDGEMENTS
Support from Australian Commonwealth Government's Cooperative Research Centre (CRC) Programme through CRC for Water Sensitive Cities and from Australian Research Council through Future Fellowship grant FT200100985 is sincerely acknowledged. Support from Josh Kamil, Anna Lintern, Gayani Chandrasena, Catherine Osborne, Tony Bisada, Christelle Schang, Peter Kolotelo, Richard Williamson, Frank Winston, Javier Neira Bravo, Tracey Pham, and Louisa John-Krol is also greatly acknowledged with gratitude. The authors acknowledge the use of the facilities and the assistance of Judy Callaghan at the Monash Micro Imaging.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.