ach to grafting biofouling resistant coatings from polymeric membrane surfaces using an adhesive macroinitiator

Biofouling is a serious problem for any wetted structure, having a negative influence on applications as diverse as marine transport, implanted medical devices and water treatment. Here, we address this issue by creating a polydopamine-based coating on desalination reverse osmosis membranes incorporating a bromomacroinitiator for subsequent polymerisation of sulfobetaine monomers into anti-biofouling polymer brushes. Surface characterisation using attenuated total reflectance-Fourier transform infrared spectroscopy and the water contact angle demonstrated the attachment of the polysulfobetaine brushes and that the hydrophilicity increased for the coated membranes. Using a macroinitiator formation time of ten minutes followed by polyzwitterion coating of one hour resulted in a 17% increase in water flux without significant effect on the salt rejection performance. These membranes also exhibited substantial suppression of protein and bacterial attachment of 69% and 88% respectively compared to unmodified membranes.


Introduction
Biofouling is generated on surfaces in contact with water by the adhesion of biomacromolecules followed by the accumulation and growth of microorganisms that create a biolm, which can ultimately allow colonisation by macrofouling organisms. 13][4] To address surface biofouling while maintaining the bulk properties of the materials used, biofouling resistant coatings have been developed. 4Anti-biofouling coatings can incorporate biocides for the active prevention of biological growth and are oen employed on manufactured surfaces in the marine environment. 1,2Hydrophobic and superhydrophobic surface coatings have also found use on marine structures, preventing adhesion of hydrophilic cells and proteins and promoting the release of any biofouling that occurs. 1,4Despite these benets, both biocidal and hydrophobic coatings tend to lose activity over time 2,5,6 and so are not ideal where reapplication of the coating is impractical.As such, many recent anti-biofouling coatings for biomedical devices are based on highly hydrophilic coatings with a neutral overall charge such as polyethylene glycol (PEG) 2 and other polyethers, 7 polysaccharides 8 or polyzwitterions such as polybetaines 3,9 and polyampholytes. 10The low interfacial energies between the surface coating and water 4 combined with the formation of a hydration layer barrier at the surface 11 means these surfaces exhibit low affinity both to biomacromolecules and to cells.
For polymer membranes used in ltration of aqueous solutions poor hydrophilicity and fouling have been major problems affecting ltration performance, [12][13][14][15] with biofouling the most widespread and difficult to address fouling problem. 12,15As such, a highly hydrophilic biofouling resistant coating would be of great benet as it also has potential to improve ux.However, one weakness is the attachment of the coating as many approaches taken so far have signicant limitations.For example, physical absorption methods result in relatively unstable coatings. 13While chemical modications can provide a more stable coating, the use of complex and aggressive chemistries, oen specic to particular membrane material and not easily scalable, means these methods are oen not practical additions to the membrane manufacturing process. 13This issue can be addressed using melanin-like polydopamine (pDA), a biomimetic material developed from observation of the almost universally adhesive properties of mussel bers. 16The dopamine monomer is deposited through selfassembly and oxidative cross-linking from mild pH aqueous solution in a rapid reaction.This results in a durable, nanoscale, hydrophilic coating that can be readily further modied as desired 17,18 to introduce specic anti-biofouling measures for a particular membrane application.
For membrane ltration the increased hydrophilicity provided by an unmodied pDA surface can correlate with both increased ux and improved membrane fouling resistance in laboratory testing, [19][20][21] although optimising the surface to achieve both together is difficult. 20,22Further, membrane ltration tests designed to mimic industrial use have revealed unmodi-ed polydopamine coatings were not successful in reducing biofouling over extended time frames. 23As such, neutral PEG modied pDA coatings have been tried, but were also not found to reduce biofouling during long term tests, 23 which may be caused by the susceptibility of PEG to oxidative degradation. 10,249][30] Zwitterionic coatings provide a hydrophilic, neutrally charged surface, useful to both increase ux and reduce biofouling while maintaining or enhancing salt rejection, 31 but can be difficult to apply. 28Recently, the ease of application of pDA and the surface properties of zwitterionic compounds have been combined, the hybrid surface reducing biofouling and increasing ux while maintaining salt rejection for reverse osmosis polyamide membranes (PAM). 19This method utilized attachment of carboxybetaine polymers through amine groups to pDA lms on membranes using a 'graing to' approach. 19Concomitantly, our own research presented here has investigated a 'graing from' method.In this work the PAMs were coated with a 2-bromoisobutyryl bromide (BiBBr) modied dopamine, 18,32 subsequently acting to initiate polymerisation of sulfobetaine monomers from the surface using activators regenerated by electron transfer (ARGET) atom transfer radical polymerisation (ATRP) as shown in Scheme 1.
This process generates polymers of controllable and narrow molecular weight distribution without the stringent experimental conditions required for other ATRP processes. 33It also enables higher polymer densities than the 'graing to' approach, 34 which has been identied as an important parameter in maximising the anti-biofouling properties. 9This created a hydrophilic surface that improved ux while retaining salt rejection and enabled signicantly reduced biofouling by both proteins and bacteria.
Commercially available polyamide membranes (PAM) for seawater reverse osmosis were provided by Toray Industries (Japan).Prior to use, coupons of PAMs (15 cm Â 15 cm) were immersed in isopropanol for at least 30 min to remove components such as preservatives and to fully wet the pores, followed by immersion in pure water for 2 h.

Synthesis of polydopamine-initiator (pDA-BiBBr) modied PAMs
A dopamine-BiBBr solution was prepared following the procedure outlined by Zhu et al. 18 Briey, to make up a quantity of Scheme 1 Proposed membrane modification; step 1 reaction of dopamine hydrochloride and BiBBr to give a dopamine-BiBBr initiator.
Step 2 reaction of the dopamine-BiBBr initiator from step (i) in Tris buffer (pH ¼ 8.5) to form macroinitiator coating creating PAM-pDA-BiBBr intermediate.Step 3 surface initiated ARGET-ATRP was used to graft 3SBMA from the pDA-BiBBr initiator coating to produce a pDAg-p3SBMA modified PAM.solution for a single coating, dopamine hydrochloride (400 mg, 2.10 mmol) was purged with nitrogen for 5 min before the addition of DMF (20 mL).Triethylamine (0.15 mL, 1.05 mmol) and BiBBr (0.13 mL, 1.05 mmol) were then added.The reaction was stirred for 3 h at room temperature, aer which the dopamine-BiBBr initiator solution was used immediately for coating.A proposed scheme of the reaction is shown in Scheme 1 (step 1).
In order to coat the polyamide surface of the PAMs, prepared membranes were sandwiched between a poly-(methylmethacrylate) support plate and a neoprene rubber mat with a central hole cut out.A stainless steel ring was secured on top of the rubber frame, thus creating a well open to air at the top.Tris buffer (37 mM, 100 mL, pH 8.5) was added to the dopamine-BiBBr initiator solution and then immediately poured onto the polyamide surface of the PAM coupon.The coating apparatus was constantly agitated using a rocking platform (Ratek ERPM4) at ambient conditions.Aer 10 min the initiator modied membrane (PAM-pDA-BiBBr) was removed from the coating apparatus and thoroughly rinsed with deionized water.All PAM-pDA-BiBBr coupons were stored in methanol/pure water (1 : 10 v/v) in brown glass containers until permeation measurements or further coating modications.A proposed scheme of the reaction is shown in Scheme 1 (step 2).Membranes coated only with pDA were created from a solution of dopamine hydrochloride (2 mg mL À1 ) in Tris buffer (15 mM, 100 mL, pH 8.5) using the same coating apparatus under the same conditions and reaction time.PAM-pDA and control uncoated membranes were stored in methanol/pure water (1 : 10 v/v) in brown glass containers for consistency with the PAM-pDA-BiBBr samples.

Growth of p3SBMA from PAM-pDA-BiBBr using surface initiated ARGET-ATRP
A brown glass container containing a PAM-pDA-BiBBr modi-ed membranes was charged with 3SBMA (10 g, 34.2 mmol), copper(II) chloride (0.001 g, 6.84 Â 10 À3 mmol) and Tris(2pyridylmethyl)amine ligand (0.02 g, 6.84 Â 10 À2 mmol) in methanol/water (1 : 1 v/v; 155 mL).The container was sealed and the solution was stirred under nitrogen for 20 min.A solution of ascorbic acid reducing agent (0.6 g, 3.42 mmol) in methanol/water (1 : 1 v/v, 5 mL) was then added to the reaction mixture and polymerisation was conducted at room temperature for 1, 3, 6 or 24 h.Opening the container to air terminated the polymerisation reaction.Aer polymerisation, the pDA-g-p3SBMA modied PAM coupons were thoroughly washed with deionized water, and stored in methanol/pure water (1 : 10 v/v) until further analysis.A proposed scheme of the reaction is shown in Scheme 1 (step 3).

Attenuated total reectance (ATR)-Fourier transform infrared (FTIR) spectroscopy
FTIR spectra were obtained using a Nicolet Nexus 8700 FTIR Spectrophotometer tted with a 'Smart Orbit' ATR accessory containing a diamond crystal internal reection element.PAM samples were placed active-face down on the ATR crystal and held in place by a clamp and the data was collected in air.For each sample 128 scans were taken at a resolution of 4 cm À1 .A background of air was run before each sample set, and automatic baseline correction and scale normalisation were performed for each set of data.The data analysis was manipulated using Omnic Spectra soware.

Water contact angle
The hydrophilicity of the modied and unmodied PAMs was evaluated by performing water contact angle measurements in air using a water droplet (static sessile drop method), and analysed using Sinterface PAT1 soware.Membrane samples (2 cm Â 4 cm) were attached to a glass slide with double sided tape and placed on a horizontal platform.A water droplet (0.5 mL) was gently placed manually onto the membrane surface and an image captured by the camera.The internal angle of both sides of the water droplet was determined for 3 droplets at 3 different locations per sample, and the mean value AE SD of three different samples was reported.

Membrane permeation tests
All permeation tests of modied and unmodied PAMs were conducted using a cross-ow ltration system (Sterlitech CF042, six units) with pure water and standard saline solution (NaCl, 2000 ppm) at 25 C. A Hydracell Pump (M-03S) delivered the feed suspension through a Swagelock inline particulate lter (15 mm) to the cross-ow cells at a constant pressure of 2.75 MPa.The feed pressure was adjusted using a bypass needle valve (Swagelok) before the channel inlet and a back-pressure regulator at the channel outlet.All PAMs were cut to the size appropriate for the cross-ow units (14 cm Â 8 cm; active membrane surface area for each cell ¼ 42 cm 2 ).Prior to all ux measurements, pure water was cycled for 1 h at >2.75 MPa in order to compact the membranes.Permeate was collected in glass beakers and weighed to determine ux.All balances were connected to a computer and weight measurements were collected every 5 min using a LabVIEW (National Instruments, USA) soware program.Pure water ux (J w ) was calculated using eqn (1): where, V is the volume of permeated water (L), A is the effective membrane area (m 2 ) and Dt is the change in time (h).
For salt rejection (SR) analysis, conductivities of the feed solution, concentrate and permeate were measured using a conductivity meter (Extech Equipment, Australia), and converted to concentration units (mg L À1 ) using a calibration curve.Salt concentration measurements (mg L À1 ) were used to calculate salt rejection using eqn (2): 35 where, C perm is the permeate concentration and C feed is the average of the concentrate and feed concentrations.
To eliminate the effect of the differences between each PAM, relative water ux (h) was used to characterize the variation of water ux due to coating modication type. 36The relative water ux (h) was calculated by eqn (3): where, J 0 and J w (L m À2 h À2 ) are the pure water ux of membranes before and aer coating modication, respectively.
Similarly, relative salt rejection (s) was used to characterize the variation of salt rejection due to coating modication type.The relative salt rejection (s) was calculated using eqn (4): where, SR 0 and SR coated are the calculated salt rejection of membranes before and aer coating modication, respectively.

Protein absorption tests
BSA tests were performed following the procedure outlined by McCloskey et al. 21A BSA tetramethyl-tagged-rhodamine conjugate suspension was diluted to 0.1 mg mL À1 in pure water.Unmodied PAMs, pDA, pDA-BiBBr and pDA-g-p3SBMAX (where X ¼ 1, 3, 6 or 24 h) modied PAMs were cut into 1 cm 2 samples and placed into glass vials containing the BSA suspension (1 mL).The vials were kept in the dark for 1 h at room temperature.The samples were then rinsed thoroughly with pure water and allowed to dry overnight by placing the samples between two lter papers.The uorescence intensities (I) were determined using uorescence microscopy and analysed using Analysis 5 soware.The uorescence measured on the unmodied PAM samples immersed in pure water was designated as I 0 , i.e., with no BSA exposure.This background was then subtracted from the modied sample uorescence readings.

Bacteria resistance tests
The resistance of the unmodied and modied PAMs to bacterial adhesion and biolm formation was studied by exposing the membrane samples to a biodegradable organic substrate (nutrient solution) made up of acetate, nitrate and dihydrogen phosphate in a saline solution (NaCl, 2000 ppm) to achieve a C : N : P ratio of 100 : 20 : 10. 37 The nutrient solution was made by dissolving sodium chloride (2 g, 3.4 Â 10 À2 mol), anhydrous sodium acetate (200 mg, 2.43 Â 10 À3 mol), sodium nitrate (40 mg, 5.7 Â 10 À4 mol) and sodium phosphate monobasic (20 mg, 1.66 Â 10 À4 mol) in pure water (1 L).Aer 48 h of accelerated test, the nutrient exposed PAMs were treated with a cell xative solution and prepared for scanning electron microscopy (SEM) imaging.

Cell-xing solution and dehydration process
To prepare the xative solution, paraformaldehyde (4 g) was dissolved in phosphate buffered saline (60 mL) at 60 C. Sucrose (4 g) was then added and the solution was allowed to cool to room temperature.Glutaraldehyde solution (25% solution in water, 2 mL) was added and the nal volume adjusted to 100 mL using phosphate buffered saline.Nutrient solution exposed PAM samples were gently rinsed with phosphate buffered saline (prepared from tablets dissolved in 200 mL deionized water resulting in a solution having 0.01 M phosphate, 0.137 M potassium chloride and 0.137 M sodium chloride with pH of 7.4) to remove unbound organic matter.Three replicate 1 cm 2 samples per modied and unmodied PAM type were placed in vials and soaked overnight in enough xative solution to cover the sample.Aer xation for 24 h the samples were rinsed in phosphate buffered saline prior to dehydration by immersion for 15 min each in a series of aqueous ethanol baths (ethanol concentrations were 50%, 70%, 85%, 95% and 100% v/v of ethanol).The samples were then dried between lter paper overnight before preparation for SEM imaging.

Scanning electron microscopy
The dried membrane samples were delaminated and mounted face-up onto SEM stubs with double-sided carbon tape.To reduce charging effects, samples were sputter-coated with platinum (5 nm) using a Cressington 208HR sputter coater, using the lm thickness monitor.Samples were imaged using a FEI Quanta 450 FEG SEM.For analysis of biofouling 3 images were taken at different locations on each sample resulting in 9 images (at 10 000Â magnication) for each type of membrane.

Biofouling quantication
The number of bacteria was counted on all images as follows.SEM images were inspected visually, and a count was made of the number of bacteria cells per image.The mean was calculated for a minimum of 9 images -3 images for each of 3 areas on 3 replicate sample.

Statistical analysis
Statistical analyses were performed using GraphPad Prism (version 6.0; GraphPad Soware, San Diego, USA).Water contact angle and bacterial count data were analysed to identify within group changes from the unmodied membrane using a one-way ANOVA followed by Tukey's post hoc test.Results were considered statistically signicant if p < 0.05.

Results and discussion
Optimisation and graing of pDA-g-p3SBMA to PAMs In previous work 28,38 we demonstrated that initiator groups could be covalently linked to the PAM surface via reaction between the acid bromide of 2-bromoisobutyryl bromide and any free amine groups in the PAMs, followed by subsequent growth of p3SBMA coatings.However, this process is dependent on the availability of free amines on the membrane surface, uses aggressive organic chemistries and is not easily scalable, thereby limiting its use.Furthermore, polyamide membrane defects are reported when using direct covalent attachment of the BiBBr surface initiator. 39Therefore, we developed a method to incorporate a universal polydopamine macroinitiator coating that is independent of membrane substrate 16,18,40 and easily scalable.Also, the surface attachment of oligomeric species 41 using both covalent and non-covalent bonds under gentle conditions should prevent membrane defects created using standard BiBBr techniques.The pDA-BiBBr macroinitiator also has advantages compared to other similar cathecholamine adhesive bromoisobutyl initiators, 42 due to the simplicity of preparation and very short deposition time.
The application of the pDA-g-p3SBMA coating was achieved in a three-step process.In the rst step, the amine and/or hydroxyl groups of dopamine hydrochloride were allowed to react with the BiBBr initiator, 18 the mole ratio of BiBBr to dopamine hydrochloride of 0.5 statistically leaving much of the dopamine monomer unmodied to support rapid self-polymerisation. 18 Subsequently, the second step was performed simply by exposing PAM surfaces to aqueous solutions of Tris buffer and dopamine-BiBBr solution, leading to the deposition of the macroinitiator.The thickness of such pDA-based coatings can be varied experimentally by varying the time of deposition, 21 surpassing 10 nm in 1 h and a maximum thicknesses of 45-50 nm being achieved over 24 h. 16,43The minimum effective time of deposition was crucial as very thin layers of pDA-BiBBr are critical to avoid signicant deterioration in permeation properties of membranes due to the pDA coating. 16,21Therefore, in this work the dopamine-BiBBr solution was deposited onto the active side (polyamide layer) of a membrane for 10 min to form the pDA-BiBBr macroinitiator coated PAM.
In previous work, 44 this pDA-BiBBr macroinitiator was used successfully to modify a commercial PAM with subsequent surface-initiated ARGET ATRP from the surface with the antimicrobial agent, [2-(methacryloyloxy)ethyl]trimethylammonium chloride (MTAC).This positively charged monomer contains an ester bond that is relatively unstable to hydrolysis.Consequently, in this work we are using the neutrally charged and more stable 3SBMA monomer to create a coating with more general anti-biofouling application and better long-term stability.Furthermore, the zwitterionic 3SBMA has good biocompatibility, 45,46 and so this material also has potential for use in biomedical applications.

Surface characterisation of the modied PAMs
ATR-FTIR measurements were performed to characterize the chemical composition of the surfaces of both unmodied and modied PAMs.The ATR-FTIR spectrum in Fig. 1a shows the typical peaks of an aromatic polyamide layer.The peaks at 1662 cm À1 and 1539 cm À1 represent the amide I (C]O stretch) and amide II (N-H bending) respectively, while the peak at 1611 cm À1 represents the C]C stretching of an aromatic amine. 47In addition, peaks from the supporting polysulfone layer were also observed at 1484 cm À1 , 1502 cm À1 and 1581 cm À1 , a result of the depth of penetration of the IR beam using the ATR technique ($1 mm) being greater than the thickness of the polyamide layer ($200 nm). 47haracteristic peaks in the spectra of pDA and pDA-BiBBr modied PAMs were not discernably different to the unmodi-ed PAM (spectra not shown) due to the very thin layer of pDA and pDA-BiBBr (<10 nm) aer 10 min of deposition. 16The successful graing of p3SBMA from the PAM, as shown in the ATR-FTIR spectrum in Fig. 1b, provides evidence of the macroinitiator attachment.The spectrum of pDA-g-p3SBMA24 shows peaks attributable to p3SBMA at 1535 cm À1 , 1481 cm À1 , 1207 cm À1 , and 1033 cm À1 corresponding to the N-H bending, quaternary ammonium, S]O asymmetric stretching and S]O symmetric stretching respectively. 48n order to evaluate the hydrophilicity of the modied and unmodied membranes, water contact angle measurements were carried out in air with a water droplet (static sessile drop method).Water contact angle measurements for the unmodi-ed, pDA modied, pDA-BiBBr modied, pDA-g-p3SBMA1, pDA-g-p3SBMA3, pDA-g-p3SBMA6, and pDA-g-p3SBMA24 RO PAMs were measured and the results are shown in Fig. 2. Unmodied PAMs show the highest measurement of water contact angle at (39 AE 2 ).pDA coatings are considered to increase the hydrophilicity of coated surfaces due to the presence of hydroxyl and amine functional groups, hence a lower contact angle (25.8 AE 1 ) was observed aer deposition of the pDA coating relative to the unmodied membrane surface. 21he pDA-BiBBr coating had a larger contact angle (36 AE 3 ) than the pDA coating, which provides evidence of the immobilisation of the hydrophobic initiator groups. 44he increase in the hydrophilicity aer graing p3SBMA to the membrane surface for 24 h compared to both unmodied and pDA-BiBBr membranes was statistically signicant (Fig. 2).However, there was no statistically signicant difference between reaction times for the p3SBMA coated membranes, with all contact angles between 9 and 11 .The improved hydrophilicity of the pDA-g-p3SBMA coated PAMs was due to the formation of a hydration layer through electrostatic bonds and hydrogen bonding between the water and the coating. 11It was shown by Azzaroni et al. 49 that when the thickness of a sulfobetaine layer is #50 nm the surface is hydrophilic (contact angle < 15 ), but when the thickness is >50 nm the surface is more hydrophobic (contact angle > 40 ), ascribed to conformational variations between chains of different lengths.Consequently, as all pDA-g-p3SBMA have contact angles between 9 and 11 , the thickness was less than 50 nm for all modied membranes.

Permeation properties of unmodied and modied PAMs
Modied and unmodied PAMs were analysed using a cross-ow ltration apparatus.Table 1 shows the inuence of coating treatment on the fractional gain or loss of ux, which is reported as the ratio of pure water ux of modied PAMs to that of an unmodied control PAM.The pDA modied PAMs show an increase in water ux relative to the unmodied PAMs by 24% at 2.75 MPa and 10 min of deposition, which is in good agreement with reported results for similar systems. 36These increases are attributed to the increase in hydrophilicity due to the deposition of a very thin layer of pDA.Following the same logic, the slight decrease of water ux detected with a pDA-BiBBr coating relative to the unmodied PAM is as a result of the presence of hydrophobic initiator groups on the surface.
The polyzwitterionic PAM coatings led to mixed results depending on reaction time.The pDA-g-p3SBMA1 sample had increased water ux of 17% compared to the unmodied PAMs.For longer polymerisation times there was decreased water permeability from 21% to 27% relative to the unmodied PAMs, presumably due to an increase in thickness of the pSBMA coating.Despite this, all membrane coatings were still better performers than a PEG based anti-biofouling coating that reduced ux by 55%. 21he salt rejection of the unmodied and modied PAMs was measured under a transmembrane pressure of 2.75 MPa with 2000 ppm NaCl solution as the feed.Table 1 shows that the coating treatments had no signicant effect on the salt rejection performance of all modied PAMs relative to the unmodied control PAM.The overall salt rejection of the modied and unmodied PAMs was >97%.

Resistance to protein adsorption and bacterial adhesion
Protein adsorption tests.Aer the validation of permeability and selectivity of modied PAMs, testing was undertaken to examine their anti-biofouling properties.Fluorescence measurement of surfaces exposed to uorophore (rhodamine)tagged protein is a well-established means of comparing surface susceptibility to protein absorption, 50 and so was used here.Fig. 3 shows a signicant reduction in the amount of adsorbed BSA protein on all modied PAMs relative to the unmodied PAM.The variation in protein absorption between modied PAMs is likely explained by the fact that pDA coatings present a negative charge under neutral conditions, 51 as does the BSA conjugate and so there was repulsion between the pDA surface and protein.The addition of the neutral p3SBMA coating shields that charge allowing greater absorption in this instance.In general all modied PAMs revealed typical characteristics of protein-repellent surfaces due to the changes in their hydrophobicity, surface charges, and surface topology, 16 occurring as a result of surface modication.
Bacterial biofouling resistance tests.Studies have revealed that the ability of a surface to reduce protein adsorption does not necessarily correlate with its ability to reduce bacterial   adhesion. 52For this reason the bacterial biofouling resistance of modied and unmodied PAMs was also investigated in this study.For this test, all membrane coupons were exposed to nutrient solution for 48 h, which provided a food source for bacteria existing in the environment.Vrouwenvelder et al. 37 recently showed that the C : N : P ratio used in the nutrient solution provides an excess dose of phosphorous, which assists to exacerbate biofouling in pure water containing naturally occurring bacteria. 37Examples of the SEM images for the unmodied, pDA, pDA-BiBBr, and p3SBMA for 1 h coated PAMs aer exposure to nutrient solution are shown in Fig. 4a-d and the bacterial abundance for each sample is charted in Fig. 5.
Aer exposing the unmodied PAMs to nutrient solution a large number of bacteria were observed on the surface (Fig. 4a).There was a signicant decrease in the number of bacteria adhered to the surface for pDA and pDA-BiBBr (Fig. 4b and c), and also signicant reductions in bacterial attachment observed for p3SBMA coated samples (Fig. 4d).The difference in bacterial attachment between coating times was not statistically signicant.For all pDA-g-p3SBMA modied PAMs a signicant reduction was observed for both short-term protein adsorption and longer-term bacterial biofouling relative to unmodied PAMs (Fig. 3 and 5), in agreement with results observed for p3SBMA coated gold surfaces in the literature. 45This is because the pDAg-p3SBMA possesses pendant side groups containing anionic sulfate and cationic quaternary ammonium groups that produce a hydration layer through electrostatic interaction in addition to hydrogen bonding.Therefore, pDA-g-p3SBMA coatings can bind a signicant amount of water on the surface, as observed in the contact angle measurements (Fig. 2).This means there is a strong repulsive force preventing the hydrophobic protein and bacteria from adhering to the surface.

Conclusions
This work has demonstrated successful surface modication of commercially available SWRO PAMs with a macroinitiator followed by ARGET-ATRP of 3SBMA monomers, generating a hydrophilic, zwitterionic coating.A polymerisation time of only 1 h resulted in an increase in relative water ux of 17% over commercially available membranes, while still maintaining competitive rejection properties.Further, signicant reductions in biofouling were demonstrated with protein and bacteria attachment decreased by 69% and 88% respectively for 3SBMA coated PAMs compared to unmodied PAMs.The gentle application conditions and near universal adhesive properties of the macroinitiator means that this coating has general utility in membrane ltration.Furthermore, the biocompatibility of the coating components presents opportunities in biomedical applications such as: reverse osmosis membranes for blood and kidney dialysis, drug delivery and biodevices.

Fig. 2
Fig.2Water contact angle measurements of unmodified PAMs and modified PAMs from 1 to 24 h.Results are presented as the mean AE SD and analysed using a one-way ANOVA followed by Tukey's post ad hoc test; *represents a significant difference (p < 0.05).

Fig. 3
Fig. 3 Normalized fluorescence intensity of modified and unmodified PAMs after contact with a 0.1 mg mL À1 solution of rhodamine-tagged bovine serum albumin.Values normalized to fluorescence of the unmodified PAM, which was assigned an intensity of 100.Error bars are standard deviation over 3 replicates.

Fig. 5
Fig.5Plot of bacterial count of SEM images analysed using Image J software.Results are presented as the mean AE SD and data was analysed using a one-way ANOVA followed by Tukey's post ad hoc test; *represents a significant difference to unmodified sample (p < 0.05).

Table 1
Relative water flux and salt rejection properties of unmodified PAMs and modified PAMs (transmembrane pressure ¼ 2.75 MPa) a a The NaCl rejection (SR) for all membranes was on average $97.5%.