Functionalization of reverse osmosis membrane with graphene oxide to reduce both membrane scaling and biofouling

In this paper, commercially available graphene oxide (GO) was used as nanomaterials to modify poly-amide reverse osmosis (RO) membrane surface. After functionalization, the membrane surface properties were improved as the water contact angle was reduced from 41.7 ± 4.5 ⁰ to 26.9 ± 0.9 ⁰ demonstrating improvement in hydrophilic properties besides decreasing membrane surface roughness. Membrane scaling tests were done in the presence of CaSO 4 solution at 20 mM concentration as feedwater showed that the modi ﬁ ed membranes (GO@RO) had superior anti-scaling properties. Results indicated that the permeate ﬂ ux was reduced by only 15% at the end of the scaling experiment as opposed to the bared RO membrane which faced up to 22% decline in ﬂ ux. The results of scanning electron microscopy, Fourier transform infrared, and X-ray diffraction con ﬁ rmed that the bared RO membrane was extensively scaled by the CaSO 4 precipitates. The results of antibacterial tests showed that GO functionalization caused an inhibition of bacterial growth by 81.7% as compared to the unmodi ﬁ ed RO membrane. Moreover, experiments were also performed to investigate microbially induced calcium sulfate precipitation on membranes. Overall, it was found that GO functionalization not only reduces biofouling but can also inhibit mineral scaling and microbially induced mineral formation. © 2020 Elsevier Ltd. All rights reserved.


Introduction
Qatar is characterized by hot summer exceeding 40 C with an evaporation rate of more than 2200 mm and annual precipitation as low as ~80 mm.Due to limited renewable water resources, the seawater desalination industry has faced rapid growth to fulfil an increase in water demands resulting from both population and economic growth [1].Seawater reverse osmosis (SWRO) is being suggested to be utilized over thermal desalination techniques owing to its lower environmental impacts [2].However, membrane biofouling and mineral scaling are affecting the performance of SWRO [3].Other types of membrane fouling that can affect the membrane performance include organic fouling (caused by humic acids and other organics) and colloidal fouling (suspended matter etc.).Although, more than one type of foulants may be present in the feedwater and cause membrane fouling, most of the researches still deal with the prevention of one of the fouling types [2].While the prevention of one type of membrane fouling may not result in significant improvement of membrane performance due to its sensitivity to suffering from other foulants.
Mineral scaling is mainly triggered by the precipitation of calcium carbonates, calcium sulfates, barium sulfates, and silicates in reverse osmosis (RO) technique [3,4].As the % recovery of RO membrane increases, the concentration of these sparingly soluble inorganic salts also increases above saturation level on the feed side of the membrane which results in the crystallization of these salts on the membrane causing mineral scaling [5,6].Thus, the scaling causes a decline in permeate flux and a reduction in membrane life [7].The scaling caused by calcium sulfate is expected to be most dominant because the precipitation of other salts such as calcium carbonates can be controlled by pH adjustments [8].In the literature, calcium sulfate scaling has been thoroughly investigated such as the effect of operating conditions [9,10], and organics [11], interaction with antiscalants [12], the kinetics of calcium sulfate precipitation [13], antagonistic gypsum and calcium carbonate scaling [14], the onset of scaling [15] and interaction of calcium sulfate with different type of membranes [16].
Membrane biofouling results from the attachment of microorganisms to the membrane surface followed by their exponential growth [17].The biofouling occurs frequently in SWRO systems, especially in Middle Eastern countries like Qatar, where the temperature of feed water entering into the system often exceeds 25 C, which is the optimum temperature for microbial growth [18].While the other types of fouling can be controlled by reducing the concentration of their causative agents, biofouling cannot be prevented even after the removal of 99.9% of the microorganisms from feed water.This is due to the ubiquitous nature of microorganisms as well as their ability to multiply at a faster rate [19].The presence of microorganisms not only cause biofouling but also plays a role in causing other fouling types like organic fouling; through secretion of carbohydrates and extracellular polymeric substances (EPS) on the membrane surface.Moreover, the microorganisms have been found to induce the formation of a variety of minerals such as carbonates, sulfates, nitrates, and others [20e22].Thus, the influence of this biologically induced mineralization (or biomineralization) can also result in mineral scaling on RO membranes.Since, both mineral scalants and microorganisms are usually present in feedwater, therefore, it is important to investigate the interaction between them and the simultaneous removal of both types of foulants.
In the literature, several nanomaterials such as titanium oxide, multiwalled carbon nanotubes and graphene oxide (GO) have been used to functionalize RO membranes to reduce biofouling [23e26].Graphene oxide (GO) has been catching special attention in the field of research and industries due to its unique properties such as antimicrobial activity, hydrophilicity, smoothness, negative charge, and its functionalization with carboxyl, hydroxyl, epoxy, and ether groups [23].It has been used as nanomaterials for its potential against biofouling [26].Moreover, GO possess superior mechanical, electrical, and optical properties and has found its applications in many fields such as water treatment, aerogels preparation, and electronic textiles [27e29].For example: GO was used to produce GO-CNT (carbon nanotube) hybrid hydrogels and was found to have 15.8 times better desalination capacity as compared to other materials for NaCl (concentration up to 35 g/L) and also demonstrated to have higher binding capacities for metals like lead and others [27].However, its ability to reduce both biofouling and scaling simultaneously are not yet reported.
Therefore, in this research, the polyamide reverse osmosis membrane was functionalized with commercially acquired GO for investigating its ability for simultaneous reduction of membrane biofouling and scaling.The effect of GO functionalization on membrane surface properties was investigated through contact angle measurements, scanning electron microscopy-energydispersive X-ray spectroscopy (SEM-EDX), Atomic force microscopy (AFM), Fourier transform infra-red (FTIR), and Raman Spectroscopic techniques.Additionally, both the mechanical and structural properties of the membranes were also studied.The effect of GO functionalization on mineral scaling was investigated by comparing the decline of permeate flux during the experimental time with that of bared RO membrane in addition to the characterization of scaled membranes using SEM-EDX, FTIR, and X-ray diffraction (XRD) techniques.The biofouling tests were performed through bacteriostasis rate determination.Besides, the effect of microorganism-calcium sulfate interaction on membranes was investigated through laboratory-designed experiments.The results of this research will help to demonstrate the potential of GO to control both mineral scaling and biofouling in RO membranes.

Materials and chemicals
Thin-film composite RO polyamide membranes (ESPA e Energysaving polyamide) were obtained from Hydranautics Inc. (USA).

Functionalization of RO membranes with graphene oxide (GO) nanomaterials
The functionalization of the RO membrane with GO was done using a previously reported procedure [30,31].In brief, the carboxyl groups on the membranes were converted to amine-reactive esters by exposing the membrane surface to coating solution "A" which contained 4 mM of N-(3-Dimethylaminopropyl)-N 0 -ethyl carbodiimide hydrochloride (EDC, 98%), 10 mM of N-hydroxysuccinimide (NHS, 98%) and 0.5 M Sodium chloride (NaCl) prepared in 10 mM MES monohydrate (>99.0%,BioXtra) at pH ¼ 5 for 1 h.The amide bond between ethylenediamine (ED, BioXtra) and activated esters were then formed by reacting the membrane with coating solution "B" which contained 10 mM of ED, 0.15 M NaCl prepared in 10 mM of HEPES(4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) buffer at pH ¼ 7.5 for 0.5 h.To convert the carboxyl groups of the commercially available GO to amine-reactive esters, GO was first dispersed in MES buffer at pH ¼ 6, and probe sonicated for 20 min to ensure good suspension.Then, centrifugation was done for 30 min and the supernatant was diluted in MES buffer. 2 mM of EDC and 5 mM of NHS were then added to the solution (named as coating solution "C") and the reaction was allowed to proceed for 15e20 min.The free amine groups of ED on the RO membrane surface were then bonded to amine-reactive esters of GO by exposing the membrane surface to coating solution "C" for 1 h at pH ¼ 7.2.In the end, the membrane was washed with DI water twice and bath sonicated for 2 min to detach any non-reacted GO nanoparticles.The GO functionalized RO membrane was designated as GO@RO.

Separation experiments
The experiments were performed to measure the pure water permeability (PWP) and salt rejection of RO and GO@RO membranes.The bench-scale filtration setup was constructed (Fig. 1) to carry out crossflow membrane filtration experiments as previously reported [32].The feedwater was supplied from 20 L tank using a high-pressure feed pump (Baldor Reliance Industrial Motor) to the inlet situated at the base of the crossflow membrane filtration cell (CF042, Sterlitech Corporation, USA).The transmembrane pressure (TMP) was controlled through valves at the feedwater and concentrate water lines.The temperature in the feedwater was controlled by recirculating the concentrate water line through the circulating water chiller to the feed tank.The operating conditions were set at 25 bar (pressure), 25 C (temperature), 1 L/h (flow rate), and the PWP (L/m 2 /h/bar) was measured using Equation (1).
Where Q p represents permeate water flow rate (L/h), A is the effective membrane area (m 2 ), and DP is the trans-membrane pressure (bar).The membrane salt rejection (%R) was also measured using NaCl (1 g/L) solution as feedwater by using Equation (2), as follows.
Where, C f is the conductivity of feedwater, and C p is the conductivity of permeate measured using an electrical conductivity meter (HACH, HQ440d, multi).

Membrane scaling experiments
In this research, calcium chloride (CaCl 2 ) and sodium sulfate (Na 2 SO 4 ) salts were used to prepare solutions with Ca 2þ and SO 4 2À concentrations at 20 mM i.e. 800 mg/L (Table 1).The prepared CaCl 2 and Na 2 SO 4 solutions were allowed to equilibrate for 24 h with the atmospheric CO 2 before being mixed.After mixing, the pH, and saturation index (SI ] are the activities of calcium and sulfate ions, respectively, and K sp is the solubility product) of the solutions with respect to gypsum, were calculated using geochemists' workbench (GWB, V11.0).For each scaling experiment, the membranes were cut into appropriate sizes and were rinsed using deionized (DI) water thoroughly before use.To maintain the feedwater conditions, the experiments were performed in total recycle mode at fixed operating conditions as summarized in Table 1.
The experiments were performed in two steps.In the first step, the membranes were conditioned at desired operating conditions for 1e2 h using ultrapure water.In the second step, the feedwater was switched to the model scaling solution (Table 1) and the resulting effect on flux decline over time was monitored until the stable permeate flux was obtained.The effect of membrane scaling on salt rejection was calculated by measuring the conductivity of permeate and feedwater (scaling solution) towards the end of the scaling experiment using Equation (2).The membrane flux (J, L/m 2 / h) was calculated using Equation (3), After the filtration experiment, the normalized flux was calculated using Equation ( 4), J o is an initial filtration flux and J is filtration flux observed at a given time

Membrane biofouling studies
The strain of Halomonas aquamarina (H.aquamarina), previously isolated from seawater of the Arabian Gulf and demonstrated its ability to biodegrade antiscalants and cause membrane biofouling [3], was also used for biofouling studies in this research.The strain was stored in 30% glycerol at À80 C in the Microbial strains bank at Qatar University and was revived before experimentation.For quantitative analysis of the antibacterial activity of the membranes, the bacteriostasis rate is often used in the literature [33,34].Briefly, 0.05 g of membrane samples (both modified and bared) were washed thoroughly with running distilled water several times before soaking them in 70% ethanol for 30 min.After soaking, the membranes were washed again with autoclaved distilled water to remove residual ethanol from their surfaces.The membranes were then added to the 10 ml of Luria Bertani (LB) liquid medium (containing in g/L: 10 tryptone, 5 yeast extract, and 10 NaCl) inoculated with the appropriate volume of freshly prepared bacterial inoculum to obtain initial optical density at 600 nm (OD 600 ) equal to 0.1, followed by incubation of samples at 30 C for 18 h.The standard serial dilution method was used to get the actual number of cells at the beginning of the experiment (t ¼ 0), as well as at the end of the experiment (t ¼ 18 h).The antibacterial activity (bacteriostasis rate, BR) of the modified membranes was then  calculated using Equation (5).
Where n o is the number of colonies on the plates treated with the control membrane (bared RO), while n 1 is the number of colonies on the plates treated with the modified membrane (GO@RO).

Simultaneous reduction of microbial growth and biomineralization
Luria Bertani (LB) medium was modified to investigate both the formation of microbial growth and biomineralization, simultaneously.For this purpose, salts of Na 2 SO 4 and CaCl 2 were added to the LB medium to achieve the concentration of Ca 2þ and SO 4 2À equal to 20 mM in the "modified LB medium" which helped to explore the formation of calcium sulfate crystals in suspended growth cultures.Briefly, 25 ml of freshly prepared autoclaved modified LB medium was added to a 50 ml sterile centrifuge tube.The medium was then inoculated with an appropriate volume of freshly prepared bacterial inoculum to achieve the initial optical density at 600 nm (OD 600 ) equal to 0.1.Both RO and GO@RO membrane samples were sterilized by thoroughly washing with running distilled water several times before soaking them in ethanol for 30 min.After soaking, the membranes were washed again with autoclaved distilled water to remove residual ethanol from their surfaces.Two pieces of membrane (1 cm Â 1 cm) were then added to the cultures to induce the growth of microorganisms and precipitation of calcium sulfate at the membrane surface.Then the cultures were incubated for up to 3 days at 30 C and 150 rpm in shaking incubator (SSI10R-2, Sheldon Manufacturing Inc., USA).After the end of the experiment, membrane pieces were removed, dried, and then immediately analyzed using scanning electron microscopy-energy-dispersive x-ray spectroscopy (SEM-EDX) (Nova™ NanoSEM 50 Series, FEI Company) for the formation of biofilm and calcium sulfate precipitates.

Membrane surface characterization
There were various techniques used for membrane surface characterization after modification with GO as well as after scaling and biofouling experiments.To characterize the membrane surface layer after functionalization; SEM-EDX, AFM, and FTIR techniques were used.The hydrophilicity of the modified membranes was measured through contact angle analysis using the sessile drop method.The water droplet of 2 mL was released onto the membrane surface placed in contact angle device (OCA15Pro, Germany).The contact angle was measured using SCA20 software after 5 s, at the minimum of 10 locations and the average water contact angle was then calculated.The SEM-EDX analysis was done using Nova™ NanoSEM 50 Series (FEI Company) and the images of the membrane surface were captured at different magnifications.The AFM was done in conjunction with a Nano indenter (AFM-MFP-3D, Asylum Research).The values of root mean square roughness (RMS) and average roughness (Ra) were determined from various locations on the membrane surface and reported as average.During FTIR analysis, the spectra of 400e4000 cm À1 were obtained using a Shimadzu FTIR spectrum instrument and the spectra of the control membrane was compared with that of the treated membrane.
The mechanical characterization of membranes was done using a Universal tensile machine (LFplus 1 kN Lloyd/USA) instrument following the Uniaxial tensile testing method.The samples were prepared in a dog-bone shape and the test was carried out at a rate of 5 mm/min.The parameters like Young's modulus (MPa), Tensile strength (MPa), Load at maximum load (N), and elongation at fracture (%) were measured.The pore size distribution and surface area of the membrane were measured using Brunauer-Emmett-Teller (BET) following the nitrogen adsorption/desorption technique.The dried membrane samples (0.6 g) were degassed under vacuum at 65 C for 10 h before analyzing in Surface area Analyzer (AS-3012 AimSizer/China).
In addition to SEM-EDX and FTIR techniques, XRD analysis using PANalytical, Empyrean/Netherland was also done for the scaled membrane to identify polymorphs of calcium sulfate precipitated on the membranes.Before scale layer characterization, the membrane was removed from the crossflow cell and was subsequently air-dried for 1 h.During sampling, the samples were taken from various locations across the membrane surface to investigate the formation of precipitates at both less saturated zone, LSZ (water flow entry region), and high saturated zone, HSZ (water flow exit regions).

Scanning electron and atomic force microscopy
SEM results showed a typical leaf-like ridge and valley structure of the bared polyamide RO membranes in Fig. 2a.The brighter parts on the surface represent ridges, while the darker parts on the surface demonstrate valleys because protruding ridges are generally brighter in the SEM image due to the "edge effect" [35].The presence of GO on the RO membrane can be observed from the darker regions on the surface (Fig. 2b), as previously shown in the literature [36e38].These regions tend to become denser with an increase in the content of GO [36].
AFM results showed a decrease in membrane surface roughness after modification with GO i.e. 74.709 nm (RO) > 70.812 nm (GO@RO).The AFM images clearly show the functionalization of GO on RO reduced the depths of valleys resulting in a reduction in overall roughness of the membrane surface.These findings are also in consistent with the SEM images in Fig. 2a and b.Previous research has also shown a reduction in membrane roughness after functionalizing the RO membrane surface with GO [31,32].

Raman and FTIR spectroscopy
To confirm the binding of GO on the RO membrane surface, Raman spectroscopy was used.This is done by comparing the ratio of peak at 1147 to peak at 1585 cm À1 as previously reported in the literature [30,39].The symmetric CeOeC stretching of the polyamide is shown at 1147 cm À1 , whereas, the phenyl ring vibration at 1585 cm À1 in Raman spectra.On the other hand, the GO also exhibits its characteristic peak at 1590 cm À1 .Hence, the binding of GO to the polyamide RO membrane surface can be seen through the intensity ratio of the two peaks.A decrease in the ratio will show the increase in the intensity at 1590 cm À1 demonstrating the presence of GO.The I1147/I1585 ratios of RO and GO@RO membranes were 1.14 ± 0.02 and 1.01 ± 0.02, respectively (Fig. 3a) which confirmed the successful coating of GO on the membrane in consistent with SEM observations.Furthermore, the peak around 3027 cm À1 shows the presence of CeH and hydroxyl (-OH) groups, and an increase in intensity for GO@RO membrane showed the increase in the number/density of these functional groups due to the presence of GO.
It has been noted that there are no significant differences results in the FTIR spectra of the GO functionalized polyamide RO membrane which can be ascribed to no chemical changes in the membrane surface with GO incorporation as well as the difficulty to detect a small amount of GO dispersed in the matrix [40,41].
However, few variations in the FTIR spectra were noted in this research.The broad peak centered around 3300 cm À1 is a complex peak due to the overlapping of stretching vibration of NeH and carboxylic (-COOH) groups of the polyamide layer [42].For bared RO polyamide membranes, it splits into two bands; one at around 3426 cm À1 which shows free NeH stretching groups, while the other at 3326 cm À1 which represents hydrogen-bonded NeH stretching modes [42].After functionalization with GO, only one broadband centered around 3300 cm À1 was noted with improved intensity due to the over-abundance of hydroxyl (-OH) functional groups of GO.Similar observations have also been noted previously [43e45].Furthermore, an enhancement in the peak intensity around 1680 cm À1 (C]O) has also been noted due to the formation of new amide linkages during the reaction between eNH 2 groups of polyamide membrane and carboxylic (-COOH) groups of the GO (Fig. 3c), as explained previously [36].

Mechanical and structural characterization of membranes
The results of mechanical and structural characterization are provided in Table 2 and Fig. 4. It is evident that the rigidity of the membranes shown by Young's modulus increased after GO functionalization with also increase in tensile strength demonstrating higher stretch resistance (Table 2).Thus, the results of these strength parameters showed that the mechanical properties of membranes improved significantly after functionalization with GO, which could be attributed to the excellent mechanical properties of the material.Wang et al. [46], also noted an increase in membrane hardness and Young's modulus measured by nanoindentation technique when Poly(ethyleneimine)-modified GO and polyacrylic acid was assembled onto a hydrolyzed polyacrylonitrile ultrafiltration supporting membrane.The improvement in the mechanical properties of the membrane was also attributed to the properties of GO.Whereas, the ductility given by elongation at break and the area under the stress-strain curve decreased slightly for GO@RO membrane which could have resulted from a process-induced defects such as cracks from drying step.
The structural properties of the membrane such as the average pore size of the membrane and surface area were studied through nitrogen adsorption-desorption measurements using the BET method (Fig. 4).Results showed that both the membrane pore size and surface area reduced after the functionalization of GO.The binding of GO and GO-silver nanocomposites to polyamide membrane surface has also shown to slightly alter the transport and structural properties of membranes which can be attributed to the atomic thickness of GO [39].In addition to the gas adsorptiondesorption method, there are various other techniques used to study the structural properties of the membranes such as mercury intrusion porosimetry, and permporometry with similar capability of measuring up to 4 nm [47].However, there are some advanced techniques like nano-permporometry and evaporporometry which can measure the pore size up to 0.6 and 0.5 nm, respectively [48].Although, these techniques are not widely available, but they have the potential to be utilized in the structural characterization of RO membranes in the future.

Membrane hydrophilicity and permeation properties
Generally, it is difficult to ensure retention of membrane intrinsic permeability during the modification procedure as it may sometimes result in a decrease in pore size [39].In this research, it was noted that the GO functionalization also caused a slight reduction in pure water membrane permeability consistent with the results of the structural characterization of membranes.On the other hand, the %salt rejection measured using NaCl as feedwater showed that the GO surface modification improved the membrane's salt rejection properties.Hence, it can be concluded that the decrease in pore size of the membrane (Table 2) resulted in a slight loss of water permeability and improvement in salt rejection (Fig. 5a).However, fortunately, there was no significant reduction in permeability was noted.
Modified RO membranes were also characterized for water contact angle to elucidate the effect of GO coating on membrane hydrophilicity.It was noted that the contact angle ( ) of the modified membrane reduced significantly from 41.7 ± 4.5⁰ to 26.9 ± 0.9⁰ (Fig. 5b), in comparison to unmodified RO membranes.This is consistent with the FTIR results, as the presence of additional carboxylic and hydroxyl groups of GO could have resulted in the improvement of membrane hydrophilicity.Previous researches have also reported similar results [31].

Permeate flux decline
The performance of membranes against mineral scaling was done by recirculating the synthetic gypsum solution across the membranes and measuring the impact of scaling on flux reduction with time.For each set of experiments, there were two steps.Initial preconditioning of membrane for 1e2 h at 25 C, 1LPM with DI water.Once, the stable flux was obtained, the second step was initiated in which the mixture of CaCl 2 and Na 2 SO 4 solutions (gypsum solution) was used as feedwater.The effect of membrane scaling was evident from the results of flux decline during the time period of the experiment as shown in Fig. 6a.It was noted that the   normalized permeate flux reduced to 0.78 ± 0.01, representing 22% flux decline as a result of scaling.Whereas, the permeate flux reduced by only 15% without any significant effect on the membrane's salt rejection capability by the end of the scaling experiment when GO@RO membrane was used (Fig. 6b).This showed that the GO functionalization reduced the extent of membrane scaling in RO membranes.Nevertheless, the results were further confirmed with membrane scale characterization techniques (SEM-EDX, FTIR, and XRD).

Scale layer characterization
To characterize the scale layer that may have formed on the membrane, SEM-EDX, FTIR, and XRD techniques were used.It has been demonstrated previously that the membrane fouling/scaling tends to increase towards the exit region as a result of a higher concentration polarization (CP) effect [49].Therefore, various scaled membrane samples were taken from different entry and exit regions to detect the formation of the scale layer.The results of SEM in Fig. 7a showed the formation of needlelike and floral structures on the membrane surface, with the EDX spectrum in Fig. 7b is showing the presence of calcium, sulfur, and oxygen.In previous researches [32,50,51], similar morphology of calcium sulfate crystals has been reported.Surface scaling occurs near the membrane surface as a result of supersaturation or due to the presence of substances responsible for crystals formation in the solution or due to the presence of conditions responsible for nucleation [52].Such conditions lead to the formation of scale on the membrane surface.Generally, the needle-like crystals originating from the core growth region on the membrane surface as noted in Fig. 7a, suggests that surface crystallization was dominated on the RO membrane [53].Interestingly, SEM-EDX did not detect the formation of these precipitates on the GO@RO membranes (Fig. 7c and d) which further confirms the results that membrane scaling was reduced significantly in the presence of GO.However, the presence of sulfur and oxygen was still observed in the EDX spectrum (Fig. 7d.).Since, the X-rays have a penetration depth from 1 to 2 mm, the source of these ions can be the polysulfone support layer situated below 20e200 nm thick polyamide active layer [54].
FTIR samples taken at various locations across the membrane surfaces helped to detect the extent of scale layer formation on the membranes.In general, the strong bands centered around 1140 cm À1 splitting into two components at around 1146, 1116 cm À1 , and 669, 662 cm À1 represents stretching and bending modes of SO 4 from pure gypsum [55,56].In this research, similar spectra were obtained in all the samples (n ¼ 5) for the RO membrane showing the extensive scale layer formation almost throughout the membrane surface.Whereas, gypsum was detected only at the very thin layer towards the most concentrated zone at the flow exit region (highly saturated zone, HSZ) in the case of GO@RO membranes (Fig. 8a).Thus, the results showed that the membrane scaling was reduced to only HSZ in the water channel after functionalization with GO.The results of XRD (Fig. 8b) further confirmed that the precipitates detected in HSZ for GO@RO membrane and the precipitates formed throughout the RO membrane surface belong to gypsum as the peaks at 11.5, 20.6, 23.1, 29.0 [57] attributes to gypsum polymorph of calcium sulfate.Better anti- scaling performance of GO@RO membranes could have resulted from improved hydrophilicity and surface smoothness.For gypsum to precipitate on the membrane, it will face greater energy barrier for surface nucleation or deposition.Moreover, the repulsive forces between negative charged GO and gypsum particles [31] may also have played an important role in discouraging mineral scaling on the membrane.

Biofouling tests
The inhibition of microbial growth by the modified membranes was used as an indicator to quantitatively evaluate the antibiofouling characteristics of the membranes.In the presence of the RO membrane, the growth of bacterial cells was not inhibited as the exponential increase in the number of cells resulted.However, after functionalization with GO, the growth of bacterial cells reduced significantly as shown in Fig. 9. Thus, the higher bacteriostasis rate of 81.7% (calculated using Equation ( 5)) could be attributed to the antibacterial activity of GO present on RO membranes.
During the last decade, the activity of GO and GO coated surfaces against microorganisms has been studied extensively [58e60].It has been noted that the size of graphene, its orientation, and the ability to produce reactive oxygen species (ROS) is associated with its antimicrobial properties [61,62].Besides, the electrostatic repulsion between GO sheet edges and bacterial cell membranes has also been reported previously [63].Besides, surface roughness and topography is also important in the interaction between GO and the bacterial cells.Generally, bacterial adhesion and subsequent growth are reduced on smoother surfaces [60].In this research, the SEM and AFM results showed that the reduction in membrane surface roughness.Therefore, it is expected that both the antimicrobial property of GO as well as modified membrane surface properties may have bestowed the antimicrobial property to the membrane.

Simultaneous reduction of bacterial growth and biomineralization
Microorganisms have the ability to induce mineralization in soil and other environments [22].Since both microorganisms and calcium sulfate are also present in RO systems, it is expected that biologically induced mineralization can intensify mineral scaling on the membranes [64].In this research, additional experiments were conducted to investigate this microorganisms-calcium sulfate interaction in the membranes and performance of modified  membranes to discourage this interaction.For this purpose, experiments were performed using a modified bacterial medium (to study bacterial growth) containing 20 mM of calcium and sulfate ions (to investigate CaSO 4 precipitation) on the membranes.The SEM-EDX technique was used to study the biofilm formation and CaSO 4 crystallization and the results are presented in Fig. 10.
It is evident that CaSO 4 precipitation did not take place in control samples i.e.RO membrane without bacteria (Fig. 10a and b).This is because the experiments were performed in unsaturated conditions and therefore, the abiotic precipitation due to the supersaturation did not take place.Whereas, SEM-EDX images in Figs.10c and d showed that the presence of bacteria resulted in both biofilm formation and mineral precipitation on RO membranes.The biofilm layer formed on the membrane was homogenous showing the attachment of rod-shaped bacteria (H.aquamarina, as inoculated in the samples) to the membrane surface (Fig. 10c).These adherent cells were found to be embedded with a slimy extracellular matrix which is composed of EPS.This EPS is produced by the cells present inside the biofilm and is generally composed of proteins, polysaccharides, lipids, and nucleic acids [64].
Since, the experiments were performed in non-crystallizing conditions; the presence of bacteria may have enhanced/induced the precipitation (Fig. 10c and d).There are two mechanisms through which bacteria can induce/mediate mineralization, (i) Adsorption of cations such as calcium in the case of CaSO 4 , around the cell membrane surface/cell wall or on the layers of EPS are the ways through which bacteria can serve as a nucleus for mineral precipitation [65].With the presence of sulfate ions in the medium, gypsum formation may result within the biofilm [66].Thus, the precipitation of calcium sulfate due to the adsorption of cations on a foreign surface (bacterial cell/EPS/biofilm, in this case) can be regarded as heterogeneous nucleation as it reduces the energy barrier needed for nucleation, causing acceleration of nucleation kinetics and formation of minerals on the surface [67], and (ii) Changes in solution chemistry such as sulfur-oxidizing bacteria can convert hydrogen sulfide into sulfate resulting in a decrease in pH and release of calcium ions due to dissolution of carbonates.Thus, the release of sulfate and calcium ions can result in gypsum formation.This phenomenon has been described in the literature [67,68].For example, under the activity of sulfur-oxidizing bacterium (Acidithiobacillus thiooxidans) and at low pH, gypsum crystallization has been demonstrated.The use of isotopes of calcium in the experimentation helped to explain the biologically mediated/ induced mineralization of calcium sulfate [67].
From Fig. 10e, a small number of bacterial cells attached to the GO@RO membrane can be seen and a trace amount of calcium ions adsorbed within them is also evident from Fig. 10f.Nevertheless, in comparison to the unmodified RO membrane (Fig. 10c and d), it can be concluded that both the biofilm growth and calcium sulfate precipitation were significantly reduced in the presence of GO@RO membrane.This further confirmed that the functionalization with antibacterial nanomaterial (GO) boosted the performance of RO membranes against membrane fouling and significantly reduced both biofouling and biomineralization, simultaneously.

Conclusion
In this research, polyamide RO membranes were coated with commercially available GO nanomaterials.The functionalization of the RO membrane surface with GO helped to improve membrane surface properties in terms of surface smoothness and hydrophilicity.The results of Raman and FTIR spectroscopy helped to confirm the presence of GO on the membrane surface.The modified membranes (GO@RO) were studied for their ability to reduce both scaling and biofouling.The results of scaling experiments showed that the flux decline of only 15% occurred in the presence of GO@RO membrane as compared to a 22% flux decline in the case of the bared RO membrane.Moreover, no precipitates were observed in SEM-EDX on the GO@RO membrane.The results of FTIR and XRD for the scaled membrane confirmed the presence of gypsum precipitates was limited to only highly saturated zones (towards the end of the water flow channel) on GO@RO membrane, whereas, the precipitates were noted throughout the surface of the RO membrane demonstrating extensive scaling on the bared RO membranes.Surface modification of RO with GO also inhibited microbial growth as 81.7% bacteriostasis rate was determined in comparison to the bared RO membrane.Further experiments were conducted to demonstrate the biomineralization of calcium sulfate on RO membranes and how GO coating helped to discourage both biofilm growth and biomineralization.In conclusion, the results showed that GO has the potential to inhibit simultaneously various types of fouling i.e. scaling, biofouling, and biologically induced scaling in SWRO systems.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2 .
Fig. 2. SEM images of (a) RO; (b) GO@RO membranes and AFM results of (c) RO; (d) GO@RO membranes (RMS e Root mean square, AR e Average roughness).(A colour version of this figure can be viewed online.)

Fig. 3 .
Fig. 3. Spectroscopic analysis of the membrane surface (a) Raman spectra (b,c) FTIR spectra.(A colour version of this figure can be viewed online.)

Fig. 5 .
Fig. 5. Results of (a) Membrane permeability and % salt rejection; (b) water contact angle.(A colour version of this figure can be viewed online.)

Fig. 6 .
Fig. 6.(a) Decline of normalized flux with time; (b) Comparison of %salt rejection and % flux decline between RO and GO@RO membranes.(A colour version of this figure can be viewed online.)

Fig. 7 .
Fig. 7. SEM-EDX analysis of the scaled membrane (a, b) RO; (c, d) GO@RO.(A colour version of this figure can be viewed online.)

Fig. 8 .
Fig. 8. (a) FTIR spectra of the scaled membrane both in high saturated zone (HSZ) and less saturated zone (LSZ) on RO and GO@RO membranes, (b) XRD spectra of the precipitates.

Table 1
Feedwater composition and operating conditions of the scaling experiment.

Table 2
Mechanical and structural characterization of membranes.