Protein mediated textile dye filtration using graphene oxide–polysulfone composite membranes

Abstract

Here we report graphene oxide (GO) concentration dependent protein binding (BSA) and dye filtration (RO-16) capabilities of polysulfone–GO composite membranes under different pH conditions (2, 7 and 10). The membranes were fabricated with different GO concentrations (1, 2, 4 and 8% w/w) and were successfully characterized for their physical and chemical properties, as well as for their performance ability. The best BSA binding and dye rejection rates were observed with 2% GO membrane at pH = 10, which were 95% and 78.26% respectively, suggesting that 2% is the optimal concentration. Further, considering the fact that RO-16 dye is acidic friendly, contact time studies were carried out with 2% GO membranes at pH = 2 and pH = 10. It was observed that 2% GO–polysulfone membrane at pH = 2 shows the highest dye rejection rate of 87.4%, supporting the importance of contact time in filtration technology.

---

Introduction

Polymeric membranes are most widely used in filtration applications. Although they have existed for decades with continuous development through time, they are associated with some drawbacks such as fouling and mechanical instability; roughness and high essential protein binding are also factors that should be taken into account. Several studies have been carried out for various applications in order to investigate such above mentioned characteristics of the polymeric membranes by preparation of composites.^1–7 However, it is still a challenge to produce ideal membranes which possess none of the above mentioned drawbacks.

Since the discovery of graphene^8,9 there have been numerous studies related to its applications in almost all known fields of science. Though graphene has widely been studied for its electrical properties, transparency, flexibility and ease of production at large scales,^10–15 it is graphene oxide (GO) that has mainly been studied for filtration applications. GO is hydrophilic in nature, easy to disperse in water and other solvents and readily undergoes functionalization.

There are several methods to prepare GO^16–18 though it is generally prepared by modified Hummer’s method.¹⁹ Its use for filtration studies has been explored by integration with different polymeric components such as polysulfone (PSf), PVDF, PES,^20–22 each enhancing one or more properties. Several studies have been reported on graphene oxide (GO)–polymer composite membranes which are related to their physical and thermal properties, and performance abilities,^23–26 while a few others are related to their applications in anti-fouling, filtration and hydrophilicity.^27–31 While protein binding capability of GO integrated polymeric membranes has also been studied and reported,^32,33 limited literature is available related to the filtration of dyes using GO composite membranes. Other applications of graphene oxide–polymer composite membranes include biosensors, fuel cells and in electrochemistry.^34–36

Most dyes used today are synthetic. These dyes are stable and have more complex aromatic structures that makes them resistant to biodegradation.^37,38 Several types of dyes are used in various industries such as leather, rubber, plastics, pharmaceuticals, cosmetics and food industries for coloring products. The residues are discharged in to the environment which is hazardous.^39,40 Most of the environmental discharge consists of a combination of dyes instead of a single dye which makes removal more complicated as some dyes are positively charged and some are negatively charged. Hence a combination of sorbents is needed in order to remove these dyes. Several literature studies suggest the removal of different types of dyes using sugarcane bagasse,³⁷ protonated waste biomass,⁴¹ microorganisms,⁴² fungi⁴³ among others.

BSA (Bovine Serum Albumin), a standard protein with numerous biochemical applications is known to interact effectively with different carbon nanomaterials⁴⁴ and also can serve as a protein glue if chemically modified.⁴⁵ In this work we report the protein (bovine serum albumin) mediated textile dye filtration capability of different concentrations of GO–PSf composite membranes. Initially, experiments were carried out with bare PSf and BSA coated membranes to study the dye filtration mechanism, however, UV spectroscopy results between the inlet and outlet concentrations showed non-significant results. Though PSf–GO composite membranes were previously fabricated and thoroughly studied for different applications, to our knowledge this is the first time where dye rejection studies have been carried out with GO–PSf composite membranes using BSA as an additive. Addition of GO to bare PSf membranes highly alters the mechanical strength, contact angle, pore size, surface charge, roughness and Young’s modulus and in turn creates an internal platform for BSA binding. The chemical structure and property of BSA to bind to both GO and RO-16 (Reactive Orange) textile dye makes it feasible for dye filtration. After fabrication and prior to testing for protein binding and dye filtration, the membranes were characterized for their physical and mechanical properties.

Experimental

Polysulfone (PSf) was purchased from BASF chemical company and polyvinylpyrrolidone (PVP) (M_w = 35 000) from ISP (USA). N,N-Methylpyrrolidone (NMP) was purchased from Sigma and used as a solvent. Graphene oxide (GO) was acquired from Graphene Supermarket Inc. (USA). All chemicals in this study were used without further purification.

Fabrication of bare and graphene oxide nanocomposite membranes

For bare PSf membrane preparation, first PVP (6%, w/w) was added to NMP and was stirred until it was completely dissolved followed by adding PSf (16%, w/w) into this solution. The dope solution was stirred at room temperature in order to form a homogeneous mixture. For dope solutions with GO (to prepare GO-nanocomposite membranes), four different concentrations of GO (1, 2, 4 and 8 w/w) were initially dispersed in the solvent (NMP) using an ultrasonication probe for 30 min, followed by adding PVP (6%, w/w) and PSf polymer (16%, w/w) as previously described. Prior to membrane casting the formed dope solutions were degassed using an ultrasonication bath. The dope solutions were then cast using a casting knife with gap setting of 200 μm on a glass plate and casted with 100 mm s⁻¹ velocity using a lab scale casting machine (Cambridge, UK, Sheen automatic film applicator). The cast films were left for 10 s for solvent evaporation followed by immediate immersion in a de-ionized water bath to obtain polymer precipitation. The membranes were stored in de-ionized water for 2 weeks.

Membrane characterization techniques

The membranes were characterized for permeability, contact angle, scanning electron microscopy (SEM), mechanical properties, zeta potential and porosity (further details are provided in ESI†).

Preparation of protein-coated membranes

For protein coating, 100 mL of protein solution (100 mg L⁻¹) was prepared by adding BSA to phosphate buffer solution (0.05 mol L⁻¹, pH 6.2). A dead-end filtration cell with a magnetic stirring apparatus was used for the coating process. Bare and GO nanocomposite flat-sheet membranes were fixed into the filtration cell and initially rinsed with distilled water under pressure (2 bar) for 1 h. Further, the protein solution was filtrated from the same membrane at 1 bar and the protein concentrations in the inlet and permeate suspensions after filtration were measured to determine the coating efficiency. Concentrations of BSA solutions were determined by a HachLange DR500 UV Spectrophotometer. Retention performances of protein were calculated by eqn (1);where, R: protein rejection (%), C_p: protein concentration at permeate (mg L⁻¹), C_f: protein concentration at feed (mg L⁻¹).

Dye filtration

The dye filtration performances of bare and GO nanocomposite membranes coated with protein layer were studied with feed containing an azo reactive dye solution, 100 mg L⁻¹ of Reactive Orange (RO-16). RO-16 was chosen since it is one of the most commonly used dyes in the textile industry throughout the world. It has a molecular weight of 617.53 g mol⁻¹ and shows a maximum absorbance at λ_max = 496 nm. Dye solution was prepared by dissolving 100 mg of the dye powder in a 1 L volumetric flask of de-ionized water at room temperature.

The filtration experiments of RO-16 dye solution were carried out at 2 bar for 1 h using a dead-end stirred filtration cell at room temperature. The flux profile over time was monitored online gravimetrically by use of eqn (2):

where V is the volume of permeate water (L), A is the active membrane area (m²) and t is the permeation time (h).

Further, the feed and permeate samples were taken for color analysis.

Results and discussion

Membrane characterization

The fabricated membranes were characterized for roughness, contact angle, pore size, Young’s modulus, surface charge, and further characterized by scanning electron microscopy (SEM) and Fourier-transform infrared spectroscopy (FTIR). Table 1 details some morphological and surface characterization results. The contact angle of bare PSf membrane was measured as 71° and after the introduction of GO the contact angles were measured as 91.49, 87.86, 79.81 and 69.23° for GO levels of 1, 2, 4 and 8%, respectively. Zhao et al.⁴⁶ fabricated PVDF nanocomposite membranes with GO nanosheets and observed that composite membranes showed decreasing contact angles with increasing GO concentration. The decrease in contact angle (increase of hydrophilicity) was attributed to the large amount of oxygen-containing groups of GO which are dispersed on the membrane surface.

Table 1 Characterization values of the membranes

Membrane type	Roughness (RMS, nm)	Contact angle (°)	Average pore size (μm)	Young’s modulus (×10^7 Pa)	Surface charge (mV at pH 6.2)	Permeability
Bare PSf	70	71.0 ± 6.5	0.038	1.13	−21.0	157
1.0 GO–PSf	170	91.49 ± 4.6	0.047	1.105	−23.4	150
2.0 GO–PSf	90	87.86 ± 5.7	0.054	1.58	−15.1	171
4.0 GO–PSf	140	79.81 ± 4.8	0.071	3.145	−8.7	213
8.0 GO–PSf	240	69.23 ± 3.4	0.127	3.78	−13.2	235

---

Addition of GO increases the pore size of bare PSf membrane which further increased with increasing GO concentration, which is a general trend for nanocomposite membranes.

Characterization values of mechanical strength and Young’s modulus are also shown in Table 1. In comparison with the bare PSf membrane, the Young’s modulus of the composite membranes seems to improve, especially for the 4 and 8% GO–PSf membranes. Young’s modulus is an intrinsic property under small elastic deformations. Unlike ultimate strength and ultimate strain, Young’s modulus values only reflect the stress–strain behavior in the initial state of the loading process.⁴⁷ Thus, the obtained results highlight that the coexistence of an efficient GO dispersion and PSf/GO covalent interactions may lead to the development of PSf/GO composite membranes with better mechanical performance abilities. It is also observed that the surface charge of nanocomposite membranes decreases with increasing GO concentrations.

Scanning electron microscopy (SEM)

The membranes were characterized by scanning electron microscopy (SEM). As seen in Fig. 1A, bare PSf shows a typical dense layer on the top followed by huge macrovoids. On the other hand GO membrane cross section images show that the macrovoids are replaced by a polymer matrix with noticeable changes (Fig. 1B–E) upon increasing GO content.

Fig. 1 SEM cross-section images of bare polysulfone and GO nanocomposite membranes: A: bare PSf, B–E: 1, 2, 4 and 8% GO–PSf composite membranes, respectively.

GO as a hydrophilic additive can have an effect on the rate of exchange between solvent and non-solvent during the phase inversion process where it can increase the de-mixing by enhancing thermodynamic instability.²⁸ As a result the pores that are formed during the phase inversion could grow larger because of the stress that is being induced on the polymer surface which in turn could be due to the rapid solidification of the polymer.⁴⁸ According to the Hagen–Poiseuille relationship, under the same pressure and membrane thickness the dynamic viscosity of larger pores lead to larger water flux.⁴⁹

This phenomenon is also well supported by the contact angle and surface roughness measurements in Table 1. With increase in GO concentration the contact angle is reduced which shows an increase in hydrophilicity of the membranes.

Also as mentioned above, there is a fast exchange of solvent during the phase inversion and due to this there are some nodules which were formed on the polymer, ultimately resulting in the increase of surface roughness of the membranes with increasing GO concentration (except for 2% and 4%).³⁰

From the SEM cross sectional images of 2% and 4% GO it can be observed that the dense top layer is well organized and relatively less rough (Table 1) compared to 1% and 8% composites. During the phase inversion process, hydrophilic GO sheets tend to move to the top layer and settle there (this also has been proved by contact angle measurements). The dense top layer of 2% and 4% GO membranes show that the GO sheets organized themselves well with the available space. In comparison, 1% which had low concentration of GO sheets and 8% which had high concentration of GO sheets tend to stack and aggregate, increasing the surface roughness of the membrane. This low surface roughness has a role to play in the dye rejection process which will be further discussed below.

Fourier-transform infrared spectroscopy (FTIR)

Prior to FTIR analysis the composite membranes were completely air dried. From the spectra shown in Fig. 2 it can be observed that the peak intensities at 3340 and 1712 cm⁻¹ increase with the increase in GO concentration. These are characteristic peaks of GO and it is seen that GO is well dispersed. The band at 1712 cm⁻¹ is attributed to C=O.⁵⁰ The broad band between 3000 and 3650 cm⁻¹ is attributed to O–H functional group stretching from the graphene oxide surface. The absorption band of PSf at 1293 cm⁻¹ corresponds to the O=S=O asymmetric stretching while the peak at 1148 cm⁻¹ corresponds to symmetric stretching of O=S=O.³⁰ The weak peaks between 2850 and 3200 cm⁻¹ correspond to aliphatic and aromatic groups. The absorption band at 1241 cm⁻¹ is attributed to asymmetric stretching of C–O–C groups.³⁰

Fig. 2 FTIR spectra of polysulfone and different PSf + GO composite membranes.

Permeability

Permeability tests were carried out using the membranes and the results are shown in Fig. 3. The permeability of bare PSf membrane was found to be 154 ± 15 L m⁻² h⁻¹ bar⁻¹, while after GO introduction the permeability values were measured to be 150 ± 20, 171 ± 14, 213 ± 17 and 235 ± 20 L m⁻² h⁻¹ bar⁻¹ for 1, 2, 4 and 8% GO concentrations, respectively. As seen in the graph there is no difference between bare PSf and 1% GO membranes but with further increase in GO concentration, the permeabilities of the nanocomposite membranes increase. GO increases the pore size of the membranes (Table 1) and the permeabilities of the phase inversed membranes are generally related to the porosity and pore size of the membranes. From Table 1, it is clearly evident that with increased GO concentration there is an increase in the pore size of the membranes, further increasing the permeability.

Fig. 3 Permeability graphs of bare PSf and graphene oxide nanocomposite membranes.

Membrane flux

Membrane flux can be increased by increasing the hydrophilicity of the membrane. As seen in Table 1, with increasing GO concentration the contact angle reduces which means that the membranes are more hydrophilic (excluding bare PSf). Fig. 4 shows the flux comparison between bare PSf and PSf–GO membranes of different GO concentrations. From the collected data and the trend it can be seen that 8% GO–PSf membrane has higher flux compared to other membranes. So, the higher the hydrophilicity of the membranes, the higher is the flux.

Fig. 4 Flux comparison between bare PSf and different PSf–GO composite membranes.

In this case it has to be noted that the contact angles of bare PSf and 8.0% PSf–GO composite membranes are almost the same but 8.0% PSf–GO membrane has a higher flux. While the hydrophilicity and flux are directly proportional, it is evident from Table 1 that the average pore size of 8.0% PSf–GO membranes is 0.124 μm compared to that of bare PSf which is only 0.038 μm, which directly contributes to the increase in flux.

Dye filtration mechanism and performance of the membranes

Initially, experiments were carried out with bare PSf, 1, 2, 4 and 8% GO membranes (without BSA) for dye filtration. UV spectroscopy results revealed that these membranes are not effective enough to filter the dye molecules. Further, experiments were thus performed in two different stages:

1. Studying the protein binding ability of PSf + GO composite membranes and

2. Studying the dye rejection capability of protein bound GO + PSf membranes.

These studies were carried out at acidic (2), neutral (7) and basic pH (10).

Bovine serum albumin (BSA), a standard protein was chosen as a protein of interest because of its numerous biochemical applications, low cost and stability.

Dye sorption is mainly pH dependent and one of the most important factors that is to be considered in terms of the filtration mechanism. Experiments were carried out with all the membranes (bare PSf, 1, 2, 4 and 8% GO) at different pH values (2, 7 and 10). The pH of the solutions were set using HCl for acidic pH and NaOH for basic pH. A schematic representation of protein coating on to the composite membranes and dye binding to these protein coated membranes is shown in Fig. 5.

Fig. 5 Schematic illustration of protein coating onto membranes with dye attachment (note: the figures are only a structural representation and not to scale).

BSA binding

In our studies 2% GO membranes at pH = 7 seem to be optimal for BSA binding. After permeability test results with different concentrations of GO + PSf membranes using BSA, UV spectroscopy results show that BSA binds to GO. The absorbance efficiency is calculated by:

Absorption efficiency = ((C_{BSA in}− C_{BSA out})/C_{BSA in}) × 100

In our experiments, constant BSA concentration was used throughout and BSA was found to covalently bind to GO.⁵¹ The amine groups of BSA bind to the carboxyl groups of GO (Fig. 5).

As seen in Fig. 6, the highest BSA binding efficiency was observed with 2% GO membranes at pH = 7 at a level of about 95%, though bare PSf, 2% GO at pH = 10 and 4% GO at pH = 2 come close to this value.

Fig. 6 Graphical representation of the calculated data for protein (BSA) binding.

If we observe the trend at pH = 7, after 2% GO the binding reduced only to increase again with 8% GO membrane. We presume that with the availability of more carboxyl groups, BSA naturally tends to bind further reaching a saturation level after 2%. With 4% GO there were more free carboxyl groups available compared to amine groups of BSA (as mentioned above the BSA concentration is constant all through the experiment) and so there is a decrease in binding efficiency.

It was observed that BSA binding efficiency again increases with 8% GO, which we presume is due to the stacking of GO flakes on top of each other above a certain concentration. Due to this stacking, there were relatively more bonding sites realized for BSA to bind to GO.

In the case of pH = 10 after 2% we can observe a complete saturation and this can be attributed to the increase in surface charge. At higher pH, the zeta potential of the system is high, which means higher negative charge. As GO itself is negative and BSA is zwitterionic (change in pH effects the protein form and structure), at high pH the electrostatic repulsion is also high.

In the case of pH = 2 there was a decrease in trend after 4% which means that there was no stacking of GO sheets till this concentration. If we observe the above graph, pH = 2 and pH = 10 values of 8% GO membrane are almost similar in comparison to 4% GO membranes which shows a large difference. At pH = 2 the system tends to be protonated due to H⁺ ions which means it attracts more asymmetrically charged BSA molecules.

On the other hand BSA is non-uniformly charged at its primary structure, though it is more stable within the tertiary structure and these changes are caused due to the change in pH.^52–54 Due to this asymmetrical charge distribution it might be possible that BSA binds to GO (negative surface charge) at certain functional points (cationic) and repels itself from other locations. It has also been theoretically reported that only certain binding sites on BSA can be occupied by the dye molecules and the affinity varies from site to site due to the difference in polarity.⁵⁵

Due to this property of BSA it would be hard to achieve 100% binding even with significant increase in GO concentrations as the form of BSA also keeps changing with change in pH.⁵⁶

Dye rejection

Permeability tests with the membranes of different GO concentrations with covalently bonded BSA were carried out. UV spectroscopy results suggest that the dye molecules were adsorbed on to the BSA (Fig. 5). A constant RO-16 concentration was used in our experiments and the dye rejection rate is calculated by:

Removal efficiency = ((C_{dye in} − C_{dye out})/C_{dye in}) × 100

The calculated values and plotted graphs are shown in Fig. 7. In our studies, maximum dye rejection was observed with 2% GO membranes at pH = 10 though 2% GO at pH = 2 comes close. If we observe the trend at pH = 2, above 2% the rejection rate gradually dropped. In case of pH = 10 though the trend drops after pH = 2, it stays at a constant value for 4% and 8%.

Fig. 7 Graphical representation of the calculated data for dye (RO-16) rejection.

Results at different pH (2, 7 and 10) show that dye rejection rate increases till 2% GO and then reduces with 4 and 8% GO, indicating that 2% GO concentration was optimal (Fig. 7).

It should be noted that initial permeability tests on GO membranes with direct dye solution did not show any significant variation according to UV spectroscopy results. UV absorbance is same for the stock and the filtrate because the size of the RO-16 molecules was so small that they could easily pass through the intermolecular spaces within the membrane. However, by using GO–BSA bonded membranes, significant change in absorbance was observed.

In Fig. 7, as seen in the graph, the rejection rate was found to decrease beyond 2% GO. Certain parameters are taken into consideration to explain this reduction phenomenon as detailed below.

Size of the dye molecules in comparison to BSA. With increasing GO concentrations up to 2% most of the free amine sites on BSA were occupied and along with this the intermolecular spaces is blocked. BSA molecules are large compared to RO-16 molecules (BSA M_w 66 430 Da cf. RO-16 617.53 g mol⁻¹) so the dye molecules are blocked.

Further with the increase in GO concentrations there was an increase in BSA binding (except for 4%) which means the availability of more binding sites for RO-16 up to 2%. With increase in GO concentration, we presume that the GO flakes stack on top of each other followed by BSA binding to them. This widens the intermolecular spaces, through which the dye molecules could easily permeate and because of this, there was a reduction in dye rejection.

Anionic property of dye. RO-16 is anionic⁵⁷ as is GO which carries negative surface charge and due to coulombic repulsions the probability of GO binding to the dye molecules is reduced. So, with increasing GO concentrations, dye molecule rejection above 2% GO concentration is reduced. GO flakes tend to stack on top of each other with increased concentration of GO, thereby hindering the net negative surface charge and in turn broadening the intermolecular spacing which provides an easy access for the dye molecules to pass through.

As observed, the change in the pH of the system affects the surface charge of the protein molecule and so the adsorption of charged dye molecules.⁵⁸ It has been reported that acidic pH is favourable for RO-16 removal.⁵⁷ A lower percentage of dye removal with increase of GO concentration is because of the presence of excess carboxyl groups which compete with the binding sites of the dye,⁵⁹ resulting in electrostatic repulsion between anionic dye molecules and negatively charged GO sites. Table 2 summarizes the best performing membranes in terms of BSA binding and dye rejection.

Table 2 Comparison of membrane performance at different pH

pH	Best BSA binding efficiency	Best dye removal efficiency
2	88% (4% GO)	74% (2% GO)
7	95% (2% GO)	34% (2% GO)
10	88% (Control)	78% (2% GO)

---

As can be seen in Table 2, 2% GO membrane is found to be the best performing membrane in terms of BSA binding and dye rejection (except for bare PSf at pH = 10). Though bare PSf at pH = 10 has 88% BSA binding ability, it fared quite poorly in terms of dye rejection.

With the above experimental data we can conclude that 2% GO membranes are optimal in order to carry out highly efficient filtration for textile dye RO-16. The maximum dye rejection capability observed was 78% at pH = 10. Along with pH there are also certain other factors such as initial concentration, agitation time and contact time which influence the filtration mechanism.

Effect of initial concentration. Initially when BSA or dye molecules were introduced in to the membrane there was a rapid adsorption which then slowed down to be a gradual process; this is due to the availability of more free binding sites at the initial phase.

Agitation time. Agitation rate is also found to play an important role in the whole process as the increase in agitation rate decreases the film resistance and facilitates mass transfer.

Contact time. The most important factor is the contact time between the protein and the dye molecules. It has previously been reported that with increase in contact time the rejection capability increases.^60–63 During the adsorption process the dye molecules first have to overcome boundary layer effects and then adsorb on to the sorbent.⁵⁸ This process relatively takes a longer time.

Considering the above experimental results and facts, further experiments were carried out to study the effect of contact time between the protein and dye molecules. We have chosen 2% GO membrane based on its performance in terms of protein binding and dye rejection at difference pH values. Flux values were acquired for three different time intervals (45, 90 and 135 min).

As explained above, for bare PSf, 1, 2, 4 and 8% GO–PSf membranes, there is either an increasing (roughness, pore size, Young’s modulus) or decreasing (contact angle and surface charge) trends with increase in GO concentration.

Contact time experiments

Taking into consideration the importance of contact time between the protein and the dye molecules for higher dye rejection rate,^60–63 contact time studies were carried out with the best performing membranes from the above experimental results.

Contact time experiments with 2% GO membranes at pH = 2 and pH = 10 were carried out in accordance with the above results. BSA solutions at the two different pH was prepared separately as mentioned in the previous section. Two 2% GO membranes were compressed for an hour at 2 bar before running the experiment. First BSA solution at pH = 2 was taken in the dead-end stirred cell filtration system (Steriltech), allowed to wait for 15 min for the solution to settle, followed by the addition of dye solution. The BSA-dye solution was left under stirring for 45 min before acquiring a first flux profile. After acquiring the first data the solution was left under continuous stirring for another 45 min (total 90 min) before acquiring the second flux profile, followed by a third flux profile after 135 min. The same process was repeated for pH = 10 BSA solution using a different 2% GO membrane.

The calculated values and plotted graph for contact time experiments are shown in Fig. 8. As predicted, 2% GO at acidic conditions with 45 min contact time turned out to be most efficient in terms of dye rejection with 87.4% rejection rate. Previously it has also been stated that acidic conditions favour RO-16 removal.⁵⁵ As observed in Fig. 8, the trend line after 45 min keeps reducing which shows that the contact time between the BSA and dye molecules turned out to be inefficient beyond 45 min. An explanation for this decreasing trend line lies within the BSA which will be further discussed in the next section. Table 3 shows the comparison of final results in terms of dye rejection.

Fig. 8 Graphical representation of the calculated data for dye (RO-16) rejection after contact time studies.

Table 3 Comparison of final results after contact time studies

	Dye removal efficiency (%)
Contact time/min	45	90	135
pH 2	87.4	56.2	45.5
pH 10	64.9	50.0	46.0

---

Discussion

Though the binding and rejection mechanisms are not completely clear at the molecular level, there are certain factors which directly or indirectly affect the whole filtration process. In our studies, standard BSA concentration was used but increase in BSA concentration increases the binding and rejection capabilities of the membranes till a saturation point.⁵⁵ A further parameter is the temperature which affects certain proteins and dyes. In the case of RO-16 the removal is favoured at lower temperatures.⁵⁷

One of the most important parameters that has to be addressed is the advantage of using different concentrations of GO in the whole process. As shown in the above mentioned data it can be clearly seen that GO–PSf composite membranes produce far better results than bare PSf membranes. Incorporation of GO significantly affects the mechanical strength, contact angle, pore size, surface charge, roughness, Young’s modulus and permeability relative to the bare PSf membrane (Table 1).

UV spectroscopy results suggest that the binding between bare PSf membranes and BSA was negligible in comparison to PSf–GO. BSA coated membranes fared better in terms of dye filtration, which means that GO is creating an internal platform for BSA binding. The relation between size and surface charge of BSA molecules and GO plays an important role in increasing the BSA binding efficiency which in turn increases the dye rejection capability. The chemical structure and property of BSA to bind to both GO and RO-16 (Reactive Orange) textile dye makes it feasible for dye filtration.

To understand the complete mechanism involved in the dye rejection process it is important to know the effect of pH on GO, BSA and RO-16.

Effect of pH on GO

GO sheets primarily contain hydroxyl and epoxy groups which makes them hydrophilic⁶⁴ but recently reported studies state that small quantities of COOH groups at the edges of GO sheets are the ones which actually determine the solution behaviour of GO sheets.^65,66

According to MD simulation results it was figured out that the basal plane of GO is much more hydrophobic compared to the COOH edges. It is this COOH which determines the solution behaviour of GO.⁶⁶ It has also been reported that at lower pH the COOH groups are protonated (Fig. 9) and the sheets become less hydrophilic and form aggregates but do not precipitate due to the formation of GO–water–GO sandwich structures which keeps it stable and surface active.⁶⁷ However at high pH, GO sheets are more hydrophilic due to the deprotonated carboxyl groups (Fig. 9) . GO thus behaves as an amphiphile.⁶⁷

Fig. 9 Schematic representation of GO protonation and deprotonation at acidic and basic pH, respectively.

Particles with zeta potential ranging from −30 to +30 mV are considered to be stable due to the electrostatic repulsions.^68,69 Also at lower pH the zeta potential is drastically less compared to at higher pH which means that at low pH the surface negative charge is less as the carboxyl groups are mainly protonated.^67,68

GO sheets at higher pH act like surfactants and also as calculated by Chih et al.,⁶⁷ the surface tension of GO at pH = 14 is around 72 dyn cm⁻¹ which also suggests that GO concentration does not significantly effect the overall film or solution properties at high pH.

Interestingly at lower pH values the surface tension is reduced drastically from 70 to 52 dyn cm⁻¹ with increase in GO concentration which shows that GO sheets are surface active at lower pH.

Though our experiments are related to GO composite membranes it is also important to understand the behaviour of GO sheets in solution. An interesting fact regarding the effect of pH on the sheet size of GO is that with increase in pH the sheet size and stability increases. Swarnima et al. have examined this phenomenon⁶⁸ by DLS (dynamic light scattering) technique. Generally, to increase the pH, NaOH is used (also in our case) which acts as a hydrogenating agent for GO.⁷⁰ On the other hand when HCl is used to reduce the pH, very large graphene oxide sheets with poor stability are observed. This can be attributed to the increase in H⁺ ion concentration within the solution which gradually increases the sheet size⁶⁶ (increase in H⁺ ions also increases protonation which causes a decrease in electrostatic repulsion) and in turn attracting more BSA and dye molecules to bind to the membrane.

Though the surface charge and contact angle decrease with the increase of protonation at lower pH, the degree of change varies from dispersed GO to integrated GO. The work of El Kadi et al.⁷¹ gives more insight to this where supernatant GO (SGO) and remanent GO (RGO) were compared at different pH values. Overall, the degree of change in terms of protonation, surface charge, contact angle and stability is in the order of SGO > RGO > composite membranes.

Effect of pH on BSA

Bovine serum albumin (BSA) undergoes conformational changes when the pH is altered from neutral to 2^72,73 though these changes are not yet known at atomic resolution. There is a partial unfolding of the protein^74,75 and decrease in the ellipticity, losing approximately 40% of its initial helix.⁷¹ The isoelectric point of BSA is 4.8 and below this value several structural alterations are caused due to the repulsive forces acting below the isoelectric point. By exposure to acid environment partial unfolding of BSA helices starts,^76–78 leading to progressively exposing the protein surface to the acid environment. This leads to an increase in hydration where more and more H⁺ ions are accommodated thus leading to protonation.⁷¹ In other words, under acidic conditions ionizable groups on BSA are protonated and due to charge repulsion, unfolding takes place.⁷⁹ Due to the protonation, BSA molecules tend to attract more dye molecules which carry surface negative charge. The higher the protonation, higher is the dye binding to BSA molecules.

At pH = 7, BSA is relatively stable irrespective of the concentration and ionic strength.⁸⁰

Changes, however, occur to BSA during incubation at pH = 2 and pH = 10.

At pH = 2, BSA loses its monomers and starts to form aggregates (this is contact time dependent) which was not observed in our case as our BSA concentration and contact times are quite low (there might be a chance of formation of non-native aggregates which were undetectable).

At alkaline pH, there are certain conformational changes in BSA transforming from one form to another⁸¹ but at high alkalinity (pH above 12) the secondary structure of BSA is completely lost.⁸²

The rate of degradation depends on the BSA concentration in the solution. The higher the concentration the faster is the degradation. In comparison to the study by Estey et al.⁸⁰ which is related to the pH dependent degradation of BSA with time, in our case we presume that the degradation of BSA is quite low. Estey et al. concluded that 100 mg ml⁻¹ of BSA looses 50% of the monomers in one day which is equal to 10 mg ml⁻¹ in 5 days. Considering our BSA concentration which is 100 mg/1000 ml, the degradation rate at pH = 2 is calculated to be 0.05%.

Effect of pH on RO-16

Due to the presence of sulfonate groups, reactive dyes generally tend to ionize in aqueous solutions to form anions. This happens by the dissociation of sulfonate (–SO₃–) groups and the dyes possess negative charge. When the pH of the system decreases, protonation occurs, creating more binding sites on the membranes which can easily attract more negatively charged dye molecules.^57,83

When the pH of the system is increased, deprotonation takes place which leads to more surface negative charge further resulting in electrostatic repulsion between the anionic dye molecules and negatively charged sites of the system due to which the sorption of the dye molecules at alkaline pH is less compared to acidic conditions.

Contradicting the above theory, in the experimental results presented above, the highest dye rejection value of 78% (before contact time studies) was observed with 2% GO membrane at pH = 10 followed by 74% with 2% GO membrane at pH = 2. We attribute this to the effect of initial concentration.

Contact time study results are in correlation with the literature where there is an increase in dye rejection from 78% (pH = 10) and 74% (pH = 2) to 87% (pH = 2) with 2% GO membrane. Due to the increase in contact time between BSA and dye molecules, the anionic dye molecules tend to occupy most of the available binding sites within the acidic system.

Conclusions

Protein mediated textile dye filtration using GO–PSf composite membranes has been studied and presented experimentally. GO integrated polysulfone membranes were fabricated and characterized. The effect of GO concentration on protein binding and dye rejection capabilities at acidic and basic pH was experimentally elucidated.

It was observed that 2% GO membrane at pH = 10 is optimal in terms of BSA binding and dye rejection capabilities. Considering the fact that RO-16 is acidic friendly, contact time studies were carried out where the BSA and dye molecules were left in contact for different time intervals (45, 90 and 135 min) before acquiring flux profiles. The highest dye rejection rate observed was 87.4% (Table 3) after 45 min. Higher contact times (more than 45 min) did not render any significant advantage.

Realizing these parameters helps in developing a new class of composite membranes where the pore size more or less is irrelevant but performance is highly dependent on binding and adsorption properties.

Acknowledgements

The authors would like to thank EC-Marie Curie Co-fund Circulation Scheme and TUBITAK (project no. 114C032) for providing financial support. We would like to extend our special thanks to Reyhan Sengur, Serkan Güçlü, Recep Kaya and Yusuf Keskin from Prof. Dr Dincer Topacık National Membrane Research Center (MEM-TEK) at ITU for rendering their help in characterizing the membrane samples and Merve Senem Avaz from Faculty of Engineering and Natural Sciences, Sabanci University, Istanbul, for FTIR data acquisition.

References

1. S. Kim, F. Fornasiero, H. Gyu Park, J. B. In, E. Meshot, G. Giraldo, M. Stadermann, M. Fireman, J. Shan, C. P. Grigoropoulos and O. Bakajin, J. Membr. Sci., 2014, 460, 91–98.
2. G. Li, J. Pan, J. Han, C. Chen, J. Lu and L. Zhuang, J. Mater. Chem. A, 2013, 1, 12497–12502.
3. D. Y. Koseoglu Imer, Desalination, 2013, 316, 110–119.
4. A. F. Bushell, P. M. Budd, M. P. Attfield, J. T. A. Jones, T. Hasell, A. I. Cooper, P. Bernardo, F. Bazzarelli, G. Clarizia and J. C. Jansen, Angew. Chem., Int. Ed., 2013, 52, 1253–1256.
5. R. Du, X. Feng and A. Chakma, J. Membr. Sci., 2006, 279, 76–85.
6. E. Fontananova, G. D. Profio, F. Artusa and E. Drioli, J. Appl. Polym. Sci., 2013, 129, 1653–1659.
7. D. Y. Köseoglu, B. Kose, M. Altinbas and İ. Koyuncu, J. Membr. Sci., 2013, 428, 620–628.
8. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Science, 2004, 306, 666.
9. A. K. Geim and K. S. Novoselov, Nat. Mater., 2007, 6, 183.
10. N. M. Gabor, J. C. W. Song, Q. Ma, N. L. Nair, T. Taychatanapat, K. Watanabe, T. Taniguchi, L. S. Levitov and P. Jarillo-Herrero, Science, 2011, 334, 648.
11. A. K. M. Newaz, D. A. Markov, D. Prasai and K. I. Bolotin, Nano Lett., 2012, 12, 2931.
12. C. Lee, X. Wei, J. W. Kysar and J. Hone, Science, 2008, 321, 385.
13. K. S. Kim, Y. Zhao, H. Jang, S. Y. Lee, J. M. Kim, K. S. Kim, J.-H. Ahn, P. Kim, J.-Y. Choi and B. H. Hong, Nature, 2009, 457, 706.
14. L. Gomez de Arco, Y. Zhang, C. W. Schlenker, K. Ryu, M. E. Thompson and C. Zhou, ACS Nano, 2010, 4, 2865.
15. S. Bae, H. Kim, Y. Lee, X. Xu, J.-S. Park, Y. Zheng, J. Balakrishnan, T. Lei, H. Ri Kim, Y. I. Song, Y.-J. Kim, K. S. Kim, B. Özyilmaz, J.-H. Ahn, B. H. Hong and S. Iijima, Nat. Nanotechnol., 2010, 5, 574.
16. N. Tsuyoshi and M. Yoshiaki, Carbon, 1994, 32, 469–475.
17. G. I. Titelman, V. Gelman, S. Bron, R. L. Khalfin, Y. Cohen and H. Bianco-Peled, Carbon, 2005, 43, 641–649.
18. H. Masukazu, G. Takuya, H. Shigeo, F. Masahiro and O. Michio, Carbon, 2004, 42, 2929–2937.
19. W. S. Hummers and R. E. Offeman, J. Am. Chem. Soc., 1958, 80, 1339.
20. H. Zhao, L. Wu, Z. Zhou, L. Zhang and H. Chen, Phys. Chem. Chem. Phys., 2013, 15, 9084.
21. C. Zhao, X. Xu, J. Chen and F. Yang, J. Environ. Chem. Eng., 2013, 1, 349–354.
22. S. Zinadini, A. Akbar Zinatizadeh, M. Rahimi, V. Vatanpour and H. Zangeneh, J. Membr. Sci., 2014, 453, 292–301.
23. F. Jin, W. Lv, C. Zhang, Z. Li, R. Su, W. Qi, Q.-H. Yang and Z. He, RSC Adv., 2013, 3, 21394–21397.
24. M. Ionita, A. Madalina Pandele, L. Crica and L. Pilan, Composites, Part B, 2014, 59, 133–139.
25. D. Manzu, A. Ficai, G. Voicu, B. Stefan Vaasile, C. Guran and E. Andronescu, Mater. Plast., 2010, 47, 24.
26. S. I. Voicu, M. A. Pandele, E. Vasile, R. Rughinis, L. Crica, L. Pilan and M. Ionita, Dig. J. Nanomater. Bios., 2013, 8, 1389–1394.
27. J. Zhang, Z. Xu, M. Shan, B. Zhou, Y. Li, B. Li, J. Niu and X. Qian, J. Membr. Sci., 2013, 448, 81–92.
28. J. Lee, H.-R. Chae, Y. J. Won, K. Lee, C.-H. Lee, H. H. Lee, I.-C. Kim and J. M. Lee, J. Membr. Sci., 2013, 448, 223–230.
29. C. Zhao, X. Xu, J. Chen, G. Wang and F. Yang, Desalination, 2014, 340, 59–66.
30. B. M. Ganesh, A. M. Isloor and A. F. Ismail, Desalination, 2013, 313, 199–207.
31. S. Gahlot, P. P. Sharma, H. Gupta, V. Kulshrestha and P. K. Jha, RSC Adv., 2014, 4, 24662–24670.
32. J. Shen, M. Shi, B. Yan, H. Ma, N. Li, Y. Hu and M. Ye, Colloids Surf., B, 2010, 81, 434–438.
33. X. Xu, J. Huang, J. Li, J. Yan, J. Qin and Z. Li, Chem. Commun., 2011, 47, 12385–12387.
34. B. G. Choi, H. S. Park, T. J. Park, M. H. Yang, J. S. Kim and S. Y. Jang, ACS Nano, 2010, 4, 2910.
35. H. Tateishi, K. Hatakeyama, C. Ogata, K. Gezuhara, J. Kuroda, A. Funatsu, M. Koinuma, T. Taniguchi, S. Hayami and Y. Matsumoto, J. Electrochem. Soc., 2013, 160, F1175–F1178.
36. D. Chen, H. Feng and J. Li, Chem. Rev., 2012, 112, 6027.
37. S. Y. Wong, Y. P. Tan, A. H. Abdullah and S. T. Ong, Malaysian Journal of Analytical Sciences, 2009, 13, 185–193.
38. R. Gong, Y. Sun, J. Chen, H. Liu and C. Yang, Dyes Pigm., 2005, 67, 175–181.
39. V. Fu and T. Viraraghava, Water SA, 2003, 29, 465.
40. P. K. Gill, D. S. Arora and M. Chander, J. Ind. Microbiol. Biotechnol., 2002, 28, 201.
41. S. W. Won, H. J. Yun and Y.-S. Yun, J. Colloid Interface Sci., 2009, 331, 83–89.
42. M. M. Sahasrabudhe and G. R. Pathade, Eur. J. Exp. Biol., 2011, 1, 163–173.
43. S. Ilyas and A. Rehman, Iran. J. Environ. Health Sci. Eng., 2013, 10, 12,  DOI:10.1186/1735-2746-10-12.
44. G. Fan, J. Ge, H.-Y. Kim, B. Ding, S. S. Al-Deyab, M. El-Newehy and J. Yu, RSC Adv., 2015, 5, 64318–64325.
45. A. Pattammattel, M. Puglia, S. Chakraborty, I. K. Deshapriya, P. K. Dutta and C. V. Kumar, Langmuir, 2013, 29, 15643–15654.
46. C. Zhao, X. Xu, J. Chen, G. Wang and F. Yang, Desalination, 2014, 340, 59–66.
47. K. Hu, D. D. Kulkarni, I. Choi and V. V. Tsukruk, Prog. Polym. Sci., 2014, 39, 1934–1972.
48. H. Strathmann and K. Kock, Desalination, 1977, 21, 241–255.
49. M. Cheryan, Ultrafiltration and Microffiltration handbook, CRC Press, Lancaster, USA, 1998.
50. H. Wu, B. Tang and P. Wu, J. Membr. Sci., 2010, 362, 374–383.
51. J. Shen, M. Shi, B. Yan, H. Ma, N. Li, Y. Hu and M. Ye, Colloids Surf., B, 2010, 81, 434–438.
52. T. Peters Jr, All About Albumin: Biochemistry, Genetics, 1995, ISBN: 978-0-12-552110-9.
53. M. Dockal, D. C. Carter and F. Ruker, J. Biol. Chem., 2000, 275, 3042.
54. V. M. Rosenoer, M. Oratz and M. A. Rothschild, Albumin structure, function & uses, Pergamon Press, NY, 1977, pp. 53–84.
55. J. Sereikaite, Z. Bumeliene and V. A. Bumelis, Acta Chromatogr., 2005, 15, 298–307.
56. U. Bohme and U. Scheler, Chem. Phys. Lett., 2007, 435, 342–345.
57. S. Y. Wong, Y. P. Tan, A. H. Abdullah and S. T. Ong, J. Phys. Sci., 2009, 20, 59–74.
58. M. Ozacar and I. A. Sengil, J. Hazard. Mater., 2003, 98, 211–224.
59. M. S. Chiou and H. Y. Li, Chemosphere, 2003, 50, 1095–1105.
60. T. Ohashi, A. M. Jara, A. C. Batista, L. O. Franco, M. A. Barbosa Lima, M. Benachour, C. A. Alves da Silva and G. M. Campos-Takaki, Molecules, 2012, 17, 14219–14229.
61. N. Holm, Biochim. Biophys. Acta, Proteins Proteomics, 2007, 1774, 1128–1138.
62. D. Gao, Y. Tian, F. Liang, D. Jim, Y. Chen, H. Zhang and A. Yu, J. Lumin., 2007, 127, 515–522.
63. K. G. Bhattacharyya and A. Sharma, Dyes Pigm., 2003, 57, 211–222.
64. S. Stankovich, D. A. Dikin, R. D. Piner, K. A. Kohlhaas, A. Kleinhammes, Y. Jia, Y. Wu, S. T. Nguyen and R. S. Ruoff, Carbon, 2007, 45, 1558–1565.
65. A. Lerf, A. H. Y. He, M. Forster and J. Klinowski, J. Phys. Chem. B, 1998, 102, 4477–4482.
66. X. L. Li, G. Y. Zhang, X. D. Bai, X. M. Sun, X. R. Wang, E. Wang and H. J. Dai, Nat. Nanotechnol., 2008, 3, 538–542.
67. C.-J. Shih, S. Lin, R. Sharma, M. S. Strano and D. Blankschtein, Langmuir, 2012, 28, 235–241.
68. S. Kashyap, S. Mishra and S. K. Behera, J. Nanopart., 2014, 640281.
69. R. J. Hunter, Foundations of colloid science, Oxford University Press, 2001, vol. 2, pp. 376–377.
70. T. Kulia, P. Khanra, N. H. Kim, J. K. Lim and J. H. Lee, J. Mater. Chem. A, 2013, 1, 9294–9302.
71. N. El Kadi, N. Taulierm, J. Y. le Huerou, M. Gindre, W. Urbach, I. Nwigwe, P. C. Kahn and M. Waks, Biophys. J., 2006, 91, 3397–3404.
72. J. F. Foster, Plasma albumin, in The plasma proteins, Academic Press, New York, 1960, pp. 179–239.
73. J. F. Foster, Binding properties of albumin, in Albumin Structure, Function and Uses, Pergamon Press, Oxford, 1977, pp. 53–84.
74. T. J. Peters, Serum albumin, Adv. Protein. Chem., Academic Press, New York, 1985, pp. 161–245.
75. C. D. Carter and J. X. Ho, Structure of serum albumin, Adv. Protein. Chem., Academic Press, New York, 1994, pp. 153–203.
76. K. Foygel, S. Spector, S. Chatterjee and P. C. Kahn, Protein Sci., 1995, 4, 1426–1429.
77. M. D. Dockal, C. Carter and F. Ruker, J. Biol. Chem., 2000, 275, 3042–3050.
78. K. J. Frye and C. Royer, Protein Sci., 1998, 7, 2217–2222.
79. C. Tanford, Adv. Protein Chem., 1968, 23, 131–282.
80. T. Estey, J. Kang, S. P. Schwendeman and J. F. Carpenter, J. Pharm. Sci., 2006, 95, 1626–1639.
81. P. J. Sadler and A. Tucker, Eur. J. Biochem., 1993, 212, 811–817.
82. B. Ahmad, M. Z. Kamal and R. H. Khan, Protein Pept. Lett., 2004, 11, 307–315.
83. M. S. Chiou and H. Y. Li, Chemosphere, 2003, 50, 1095–1105.

---

Footnote

† Electronic supplementary information (ESI) available: Membrane characterization techniques. See DOI: 10.1039/c5ra13131f

---