Abstract
Reverse Osmosis (RO) is a promising technology that will increase access to clean and safe water sources throughout the world. However, the impact of RO filtration of natural waters is severely hindered by biofouling. Formation of complex biofilms on RO membranes dramatically decreases output due to release of extracellular polymeric substances (EPS) by the microorganisms. We present a polydopamine-copper (PD-Cu) coating for RO feed spacer materials to prevent biofouling and enhance longevity of Cu ions. The following spacers were tested in a continuous flow bench scale RO system: (1) Polypropylene (PP) feed spacers coated with PD-Cu, (2) a pristine PP, control spacer, (3), a PD control spacer and (4) a Cu control spacer. Results showed the PD-Cu spacers exhibited higher Cu ion chelation, retaining 71 ± 2% more Cu ions compared to a Cu-only spacer after 13 h. In a stirring beaker, PD-Cu spacers lost loosely attached Cu ions until the optimum Cu concentration was achieved, approximately 30.6 ± 0.3% of total composition, within 6 h, and the remaining Cu ions bonded with PD covalently. In addition, PD-Cu spacers showed a 17.5% higher permeate flux and a 58% biofilm biovolume decrease as compared to a pristine spacer over 24 h.
HIGHLIGHTS
Polypropylene spacers were fabricated by chemical bath deposition.
Polydopamine chelates with copper ions enhancing their longevity in polypropylene spacers.
Hydrogen peroxide provides both physical and chemical biofouling control.
PD-Cu spacers prevent biofilm growth on the membrane by 58% compared to the pristine spacer.
Graphical Abstract
INTRODUCTION
Reverse Osmosis (RO) is a promising technology to combat clean water shortages facing 41% of the world population (Greenlee et al. 2009). However, biofouling is a persistent challenge in membrane-based water treatment practices because it minimizes water output and decreases water quality (Kim et al. 2009). Biofouling is the unwanted deposition and growth of microorganisms on wetted membrane and spacer surfaces as biofilm, leading to a decrease in membrane flux (Flemming 2002). Biofouling persists as an issue because bacterial biofilms are formed by colonies that are rooted in a self-produced matrix of extracellular polymeric substances (EPS) (Flemming et al. 2016). EPS adhere to the membrane and spacer surfaces as well as to each other making them difficult to remove (Bucs et al. 2018).
Biofouling results in inefficient water production due to increased downtimes with estimated added costs for desalination plants near 15 billion US$ yearly worldwide (Kochkodan & Hilal 2015). Despite different ventures into membrane and spacer modifications for antifouling, membrane fouling is still inevitable with time (Ang et al. 2011). The complex nature of biofilm formation on RO membranes has left most existing anti-fouling strategies less effective (Vanysacker et al. 2014).
Past and current strategies to mitigate biofouling are dominated by hydrophilic, bactericidal, and biocidal coatings (Yang et al. 2009; Araújo et al. 2012; Reid et al. 2014) and chemical treatments (Yu et al. 2020). Immobilization of antimicrobial nanoparticles (NPs) onto spacers is essential for surface stability, and immobilized enzymes have demonstrated a higher thermal stability, low agglomeration and an increase in repulsive energies (Kriegseis et al. 2020). Previous studies have demonstrated that polydopamine (PD) covalently bonds to any surface and can immobilize metal ions and nanoparticles (He et al. 2015). PD increases surface hydrophilicity making spacers and membranes susceptible to biofouling (Yang et al. 2017). Metal ions chelate to PD effectively, which also attaches to materials permanently (Im et al. 2017). Therefore, the use of PD on RO spacers and membranes lessens biofouling and the loss of metal-ions (Gao et al. 2021). Metal based ions (i.e. Ag and Cu) are strong biocides used extensively in biofilm eradication (Becerra et al. 2020). In this work, for the first time ever, we modified the surface of RO spacers using chemical bath deposition to coat the spacers with PD and Cu.
In our design, we used a Cu2+-coating on PD due to previous studies that showed this metal ion in particular could exhibit ‘contact killing’ of viruses and bacteria, a phenomenon which dates back to the ancient times, Cu was registered by the U.S. Environ-mental Protection Agency (EPA) as the first antimicrobial metal (Grass et al. 2011).
In addition to the composite coating, hydrogen peroxide (H2O2) was employed to compound the effect of this anti-biofouling surface modification. A study by W. Yang et al. observed a 63% (CuO-coated spacer) and 54% (PP-coated spacer) decrease in differential pressure across the feed channel once hydrogen peroxide is dosed into the system on day 12 (Yang et al. 2019), this observation supported the ideology of coupling biocidal surface modifications with disinfectants (Pichardo-Romero et al. 2020).
Hydrogen peroxide has been studied as an effective antifouling agent due to its un-stable nature which readily decomposes to water and oxygen as bubbles once in contact with membrane biofilm (Olsen et al. 2009). Oxygen bubbles induce some turbulence that sheers off loosely attached biofilm from the membrane and spacer, reducing biofouling as a result. H2O2 is more effective in biofouling mitigation than many tested compounds and majority bacterial strains are sensitive to H2O2 (Kristensen et al. 2010). The use of H2O2 has previously been reported to be cost effective in both sludge dewatering and RO because of its ability to destroy EPS (Liang et al. 2020).
In this study, a Cu-PD-coated spacer and H2O2 biofouling mitigation method was introduced in a bench-scale reverse osmosis system to monitor the interdependence of these parameters in biofilm formation. Cu2+ O NPs and PD were deposited onto the surface of a PP spacer by chemical bath deposition. Flux data were compared between a pristine spacer, a Cu- coated spacer, and a PD-Cu spacer. Data recorded was collected on a bench scale RO system and used to determine the success of PD on metal-based ions longevity and its effect on biofilm formation.
METHODS
Spacer modification
Pristine polypropylene mesh to use as spacers were purchased from McMaster CARR (9275T34; Ohio, USA). These were cut to 8.5 cm × 4 cm before use. Chemical bath deposition (CBD) adapted from P. Hajipour et al. (Hajipour et al. 2020) was used to fabricate the spacer with polydopamine (PD) and Cu (ii) oxide nanoparticles (n-CuO). First, a tris-Buffer solution was prepared containing 100 mM tris (hydroxymethyl) aminomethane (99.8%, Acros). Dopamine hydrochloride (99%, Fisher Scientific) was dissolved into the tris-buffer solution and pH was adjusted to 8.5 using hydrochloric acid (1.0 M HCl, Fisher Scientific). Polydopamine has a high affinity to water, therefore, it is important to completely oxidize dopamine for better adhesion to spacer (Mulyati et al. 2020).
Spacer characterization
A Mira 3 Field Emission Scanning Electron Microscope (FESEM, Mira 3 Tescan, Czech Republic) with a Schottky source was used to analyse the surface morphology of each RO spacer. To observe the spacer surfaces, the spacers were sputter coated with gold to inhibit charging by enhancing surface conductivity of the plastic polymer improving the secondary electron signal. Energy Dispersive X-ray Spectrometer analysis (EDS, Octane Elect, Ametek, EDAX, USA) was employed to characterize the success of Cu deposition on the PP spacer as well as elemental composition of each spacer at accelerating voltage of 5 kV.
Water contact angle (Attension Theta Lite, Biolin Scientific, Sweden) measurements were performed using ultrapure water droplets to determine hydrophilicity. Eight water droplets of 100 μL were placed at different locations of modified PP plastics and contact angles determined.
The surface charge was measured zeta potential to modified and unmodified PP spacer (Pristine, PD-Cu, Cu and PD). Streaming potential was conducted in a 10 mM KCl (Fisher Scientific) used as the background solution at pH 7 in an electro kinetic analyzer for solid surface analysis (SURPASSTM3, Anton Parr, Austria) equipped with a cylindrical gap cell. The different spacers were sliced to thin pieces and placed in the gap cell. Successful spacer dip coating should result in minimal coagulation and a higher or equal zeta potential compared to the pristine. The SURPASS3 was used to extract surface charges. Modified and pristine spacers were cut accordingly to fit the cylindrical gap cell with gap height adjusted to 103 μm for each test. Zeta potential values were compared between three different spacers; higher values indicated a more stable material; low values indicated the spacer recorded instances of coagulation. Each modified spacer was tested and rinsed three times.
Feed water & copper loss analysis
Tap water was filtered with an activated carbon filter to remove some chlorides and organic carbons present and used as it is as feed water. E-coli was added into the feed once the membranes were stabilized. Feed water solution was filtered with 0.2 μm nylon syringe filter and collected into 10 mL glass cuvettes (Hannah Instrument) containing 0.1 mL 70% nitric acid before and after a fouling experiment then sent to Montana Bureau of Mines and Geology (Montana Technological University, MBMG Laboratory) for elemental composition analysis. The samples were stored in our aqueous technologies laboratory (ATL) fridge (4 °C) overnight before they were analyzed. A Thermo Scientific iCAP 6000 Series inductively coupled plasma optical emission spectrometer (ICPOES) was used to determine major cations and a Two Metrohm Compact IC Plus instrument performed the anion analyses in the feed water. The initial composition of feed water is shown on Table 1, initial Cu concentrations were below detection levels.
Pre-treated feed water composition
Ion . | Concentration (mg/L) . |
---|---|
Calcium | 11.1 |
Potassium | 2.46 |
Sodium | 562 |
Phosphorous | 36.8 |
Copper | <0.0975 |
Chloride | 11 |
Nitrite | 97.6 |
Ion . | Concentration (mg/L) . |
---|---|
Calcium | 11.1 |
Potassium | 2.46 |
Sodium | 562 |
Phosphorous | 36.8 |
Copper | <0.0975 |
Chloride | 11 |
Nitrite | 97.6 |
A Feed water anion and cation composition was determined using the Thermo Scientific iCAP 6000 Series inductively coupled plasma optical emission spectrometer (ICPOES) and a Two Metrohm Compact IC Plus at the Montana Bureau of Mines and Geology.
Comprehensive Cu loss analysis was performed at room temperature with separate 1,500 mL beakers containing ultrapure Millipore water. Each beaker contained an 8.5 cm × 4.0 cm Cu and PD-Cu spacers were immersed separately. The beakers solution was stirred continuously at 350 rpm for 13 h. Water samples were collected and averaged in triplets at hour intervals and measured with a portable Hanna Multiparameter Benchtop Photometer and pH meter (Hanna Instruments). A CuVer 1 Cu Reagent powder pillow was used for both Low range (0–1.5 mg/L Cu, ±0.010 mg/L) and high range (0–5 mg/L of Cu, ±0.020 mg/L), all samples were measured in low range. Concentrations of Cu in each sample were measured using Bicinchoninate method. Initial Cu in the Cu-spacer was determined by soaking a coated spacer in ultrapure water for 48 h and decolorization of the spacer from dark observed with Cu concentrations lower than 0.010 mg/L. In PD-Cu spacers, initial Cu concentrations were determined by Equation (1). Sample volumes were collected at most 10% of the initial solution volume to minimize measurement errors.
Bench-scale reverse osmosis system
Bench Scale Reverse Osmosis System. A Teledyne syringe pump was used as the H2O2 dosing pump with a constant flowrate of 30 mL per min. The pressure gauges were maintained at 17 bars for all membrane cells and the chiller set at 20 °C. Permeate flowrates were measured with Sensirion Liquid Flowmeters (Sensirion, Switzerland) and recorded to a laptop via USB Sensor Viewer software in μL/min.
Bench Scale Reverse Osmosis System. A Teledyne syringe pump was used as the H2O2 dosing pump with a constant flowrate of 30 mL per min. The pressure gauges were maintained at 17 bars for all membrane cells and the chiller set at 20 °C. Permeate flowrates were measured with Sensirion Liquid Flowmeters (Sensirion, Switzerland) and recorded to a laptop via USB Sensor Viewer software in μL/min.
Hydrogen peroxide at a concentration of 0.2% was dosed into the running system using a D Series Syringe Pump (Teledyne Technologies) at a constant flow rate of 30 mL min−1 for 6 min to remove biofilm after significant flux decline. Our bench scale system was designed to incorporate a high-pressure pump (Hydra-Cell, Wanner Engineering Inc.) that maintained a constant feed flow at 17.2 ± 0.1 bar. Feed water was maintained at a constant temperature of 20 °C by an Isotemp Recirculating Chiller (Thermofisher, USA) to avoid inconsistent flux changes as mentioned in earlier studies (Robinson et al. 2019).
Reverse Osmosis experiments were done in this closed loop RO system. Flat sheet polyamide membranes (BW30LE, Sterlitech) with an active area of 42 cm2 were used in all experiments. The flat sheet membranes were installed in the membrane test cell and the feed water was applied at 17 ± 0.2 bars (250 psi), monitored with an inline AMETEK USG 1500 psi gauge. The system was operated with the filtered tap water feed until the permeate flux stabilized after ∼2 h before adding bacteria.
Biofouling simulation
To simulate biofouling feed water is continuously run through the membrane cells. A final concentration of 2,000 ppm sodium acetate, 680 ppm sodium nitrate and 200 ppm sodium phosphate were added at pH 8 to induce ions into the feed water.
Escherichia coli (Migula) Castellani and Chalmers purchased from America Type Culture Collection (ATCC 10798, U.S.A). E. coli were grown on a tryptic soy agar plate and stored in the refrigerator (≤4 °C) for up to 2 weeks. A single colony was taken from the agar plated and placed in 10 mL of tryptic soy broth (TSB) (Fisher Scientific) and incubated in MAXQ6000 Stackable Shaker Packages (Thermo Scientific, USA) overnight at 20 °C. The feed water was used in three bacteria washing using a Vortex-Genie 2 mixer (Scientific Industries, Inc.) until a final concentration of 106 cells per mL was obtained and added to the feed reservoir. Absorbance measurements were used to determine the E. coli concentrations by a GENEYS UV-VIS (Thermo Scientific, USA) set at a wavelength of 600 nm.
The PD-Cu, Cu, and PD coated spacers were tested separately with the same membrane with a constant feed pressure of 17 bar and an initial crossflow rate of 1,100 μL min−1. Low concentration hydrogen peroxide was dosed into the system to provide both chemical and physical cleaning as investigated in previous biofouling control studies.
Membrane biofilm characterization
Biofilm formation was quantified using confocal laser scanning microscope Leica CMS GmbH, CLMS (Leica Microsystems, Wetzlar, Germany) equipped with an HCX APO L U-V-I 20x/0.50 water numerical aperture objective and an 84.9 μm pinhole. A standard LIVE/DEAD Backlight working solution was prepared by dissolving a pipet containing SYTO9 (L13152, Invitrogen), and another containing propidium iodide in a common 5 mL volume of ultrapure water (LIVE/DEAD® BacLightTM Bacterial Viability Kits Manual, Invitrogen). The working solution was then used to stain the membrane for 40 min in the dark. At least 10 random z stack fields were collected from each membrane and live biovolumes analysed using COMSTAT2. Confocal images were captured and modified according to Otsu's thresholding method.
Biofouled and used membranes were dehydrated using the modified glutaraldehyde technique before SEM characterization. The membranes were immersed in 2% glutaraldehyde at pH 7 for 2 h at 4 °C (fridge temperature). Next, they were washed with phosphate buffer saline (PBS) containing 8 g NaCl, 0.2 g KCl, 1.44 g Na2HPO4 and 0.24 g KH2PO4 in one litre. Finally, the samples were slowly dehydrated with graded alcohol 10, 50, 70, 95 and 100% (twice), each time 10 min. A Mira 3 Field Emission Scanning Electron Microscope (FESEM, Mira 3 Tescan, Czech Republic) was used to study membrane biofilm morphology before and after H2O2 dosing.
Membrane stability and hydrogen peroxide
A continuous ‘constant’ flux after dosing has previously been suggested as evidence of membrane damage, which was avoided by injecting H2O2 at low concentrations of 0.2% (Ling et al. 2017). The effects of H2O2 on membrane biofilm was determined by viewing SEM images before and after dosing and flux measurements recorded over 24 h.
Statistical analysis
Biofilm biovolume, contact angles and zeta potential were compared statistically between pristine, Cu, PD and PD-Cu spacers. Data was averaged for at least three samples expressed as mean ± standard deviation. Significance difference between samples were determined with a one-way analysis of variance (ANOVA) post hoc comparisons between groups were made with a Tukey test. All statistical analyses were performed on Minitab 20 Statistical Software, version 20.0) at 95% confidence level (Minitab, State College, PA).
RESULTS AND DISCUSSION
PD-Cu membrane modification results in improved RO membrane output over time
As compared to the pristine control spacer (white), the Cu-only spacer was a black colour, while PD and PD-Cu spacers appeared dark brown (Supplementary Figure 1). A loose agglomeration of CuO was present on the Cu-only spacer.
SEMs of a (a) pristine spacer, (b) Cu-coated spacer, (c) PD coated spacer, and a (d) PD-Cu coated spacer.
SEMs of a (a) pristine spacer, (b) Cu-coated spacer, (c) PD coated spacer, and a (d) PD-Cu coated spacer.
In order to confirm the elemental composition of the fabricated spacers EDX analysis was performed. Different areas of the spacers were chosen, and the corresponding peaks shown in (Supplementary Figure 2). Supplementary Figure 2A-C shows quantitative results are evident of the presence of C, O and N were on PD spacer (Supplementary Figure 2A), C, O and Cu on Cu spacer (Supplementary Figure 2B) and C, O, N and Cu were observed on the PD-Cu spacer (Supplementary Figure 2C).
Permeate flux and salt rejection during RO filtration of (a) pristine spacer, (b) Cu spacer, (c) PD spacer, and (d) PD-Cu spacer. Initial flux was 19.186 ± 0.001 L·m−2·h−1, a cross flow velocity of 0.18 m·s−1. Feed water was maintained at 20 ± 1 °C. Each data is an average of three experiments.
Permeate flux and salt rejection during RO filtration of (a) pristine spacer, (b) Cu spacer, (c) PD spacer, and (d) PD-Cu spacer. Initial flux was 19.186 ± 0.001 L·m−2·h−1, a cross flow velocity of 0.18 m·s−1. Feed water was maintained at 20 ± 1 °C. Each data is an average of three experiments.
The pristine spacer showed an immediate flux decline (Figure 3(a)), flux declined rapidly in Cu spacer, this could be explained by loose Cu nanoparticles from the spacer depositing on the membrane. Cu spacers showed a high flux decline between times 6 and 21 h, this was followed by a sudden constant flux to the end (Figure 3(b)). When pristine spacers were used, the flux started to decline in under 5 h, while PD spacers began to foul at 7 h (Figure 3(c)), and PD-Cu spacers did not show a significant flux decline until 12 h (Figure 3(d)). Flow through the PD-Cu coated spacer was recorded as having the highest flux from all spacers in 24 h.
Polydopamine increases spacer longevity
Cu loss analysis. Beaker ultrapure water Cu concentration over time per spacer's unit area. Deposited density (DD) of the spacers was 0.184 ± 0.004 mg/cm2 for Cu and 0.200 ± 0.002 mg/cm2 for PD-Cu.
Cu loss analysis. Beaker ultrapure water Cu concentration over time per spacer's unit area. Deposited density (DD) of the spacers was 0.184 ± 0.004 mg/cm2 for Cu and 0.200 ± 0.002 mg/cm2 for PD-Cu.
Cu longevity test on spacers. (a) Images of PD-Cu spacer (dark coloured) vs Cu spacer after a soaking in HCl for at least 24 h. (b) An SEM micrograph of a PD-Cu spacer after soaking in 0.1 M HCl for 24 h; polydopamine and Cu particles are still present on the spacer.
Cu longevity test on spacers. (a) Images of PD-Cu spacer (dark coloured) vs Cu spacer after a soaking in HCl for at least 24 h. (b) An SEM micrograph of a PD-Cu spacer after soaking in 0.1 M HCl for 24 h; polydopamine and Cu particles are still present on the spacer.
Ultrapure water contact angles of all spacers were determined and compared. Stability of Cu particles on the PP surface were determined by zeta potential using a SURPASSTM3. These values were recorded and compared between controls and the PD-Cu spacer.
Ultrapure water droplets contacts angles. Flat polypropylene plastics were viewed under a SURPASS3 at pH 7.
Ultrapure water droplets contacts angles. Flat polypropylene plastics were viewed under a SURPASS3 at pH 7.
Biocidal copper mitigates biofilm formation
Biofouling was observed on the polyamide membranes under each of the modified spacers. To minimize effects of other factors not being investigated in this study, PD-Cu spacers were compared to three different controls, PD spacers, Cu spacers and pristine PP spacers. For all membranes, biofilms were quantified as biovolume (μm3/μm2) with COMSTAT2 software.
Membrane ‘live’ biovolumes. E. coli biofilm images were captured 24 h after the initiation of the experiment. Biofilms were stained with SYTO 9, (‘live’) for live biofilm biovolume. Biovolumes were quantified using COMSTAT2 in quadruplets.
Membrane ‘live’ biovolumes. E. coli biofilm images were captured 24 h after the initiation of the experiment. Biofilms were stained with SYTO 9, (‘live’) for live biofilm biovolume. Biovolumes were quantified using COMSTAT2 in quadruplets.
Hydrogen-peroxide as biocide and disinfectant
(a) Permeate Flux with periodic hydrogen peroxide dosing. Flux data was collected from one control experiment with a pristine spacer. (b) SEM image before hydrogen peroxide dosing. (c) SEM image after hydrogen peroxide dosing.
(a) Permeate Flux with periodic hydrogen peroxide dosing. Flux data was collected from one control experiment with a pristine spacer. (b) SEM image before hydrogen peroxide dosing. (c) SEM image after hydrogen peroxide dosing.
The biocidal effects of H2O2 were investigated on a polyamide membrane under a pristine spacer. Permeate flux was restored back to the initial rate immediately H2O2 was injected into the system, however, permeate flux declined back to previous values almost immediately. H2O2 dosing provided an effective short-term in situ biofilm removal mechanism. SEM imaging showed E. coli biofilms on a control membrane without H2O2 dosing (Figure 8(b)) and a clear surface matrix on the membrane 15 min after dosing H2O2 (Figure 8(c)). To evaluate the effects of hydrogen peroxide on biofilm, membrane biofilms were investigated under a confocal microscope, the volume of EPS and life cells were calculated on COMSTAT 2.1. The biovolumes observed in the same membrane (Figure 8(c)) averaged ∼1.02 μm3/ μm2 immediately after H2O2 was dosed. This explains the spontaneous spike in permeate flux immediately after H2O2 is dosed. Previously, it has been established that H2O2 affects biofilm to a higher degree within the first 5 min and little or no change in the next 25 min. H2O2 was observed to provide short-term biofilm mitigation.
CONCLUSIONS
Biofouling persists as a costly issue in RO it affects the quality and quality of drinking water processed. PD-Cu modified spacers can alleviate this problem being a simple and cheap fabrication mechanism. PD-Cu spacers can efficiently reduce biofouling of RO membranes mainly through two mechanisms: the biocidal effect of Cu deposited on the spacer and the inhibitory effect of polydopamine (PD) on Cu particles desorption. These experiments indicated that PD-Cu spacers increased the antibiofouling properties of the membrane compared to PD, Cu and pristine controls with no need to modify commercially available polyamide membranes. Further studies will be needed to mitigate possible life cycle environmental effects of Cu leaking into the RO effluent.
ACKNOWLEDGEMENTS
The authors sincerely acknowledge Gary Wyss, Nancy Oyer and Stefanescu Cristina for technical laboratory advice and assistance. Sincere thanks to the Montana NSF EPSCoR (Established Program to Stimulate Competitive Research) for contributing to the laser scanning microscope (LSM) that helped in this study. This material is based upon work supported in part by the National Science Foundation EPSCoR Cooperative Agreement OIA-1757351. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.
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.