Fabrication of polymeric membrane with high salt rejection by embedding poly(amidoamine) dendrimer-functionalized halloysite nano tubes

The fabrication of membranes for water desalination and wastewater treatment is an important strategy to overcome water quality problems worldwide. Herein, the influence of the presence of the poly(amidoamine) dendrimer (PAMAM) on the performance of the cellulose acetate (CA) membrane was investigated. Halloysite nanotube (HNT) was functionalized with the third generation of PAMAM dendrimer and incorporated into the CA membrane by phase inversion process to improve the properties of the membrane. The composite membranes' characterization was determined by Fourier-transform infrared spectroscopy (FTIR), Atomic force microscopy (AFM), contact angle measurements, zeta potential, thermogravimetric (TGA) analyses, and field emission scanning electron microscopy (FE-SEM). A dead-end filtration system examined the performance of the incorporated cellulose acetate membrane. Based on the results, the membrane permeability, salt rejection, and antifouling properties were improved due to the addition of hydrophilic HNTs-PAMAM nanocomposite in the membrane casting solution. The loading of 1wt% of HNTs/PAMAM was optimized as filler composition. It showed salt rejection of 91% and 75.6% for sodium sulfate and magnesium chloride respectively along with the maximum porosity (82%), antifouling performance (85%), reversible fouling ratio (45%), and the lowest contact angle (~32).


Introduction
Growing modern industries and population lead to the emission of pollution into the environment and increase the demand for clean water [1].Desalination is one of the processes that are employed to generate pure water which can be reused.Various technologies have been presented to produce and improve freshwater quality.Membrane technology in comparison to other separation techniques, has been utilized in water and wastewater treatment processes all over the world due to its ease of use, high effectiveness, and environmentally friendly.Nanofiltration (NF) is the prevailing membrane process for salt rejection.These methods contrary to reverse osmosis (RO) reveal reasonable permeating flux and rejection under low operating pressure.Incorporating hydrophilic nanoparticles (NP) in the NF membrane process effectively increases water flux and salt rejection.Therefore, preparing NF membranes with different NP for water treatment has been considered a research subject.[2][3][4] Cellulose acetate (CA), a biodegradable polymer, due to eco-friendly, high biocompatibility, and permeability has been studied for aqueous-based separation [5].Despite its advantages, low porosity and dense surface layer cause lower water flux and limit its application [6].Therefore, some procedures such as chemical grafting, surface modification, and blending have been achieved to overcome limitations and improve the characteristics of the NF membrane [7].Blending organic and inorganic nanomaterials to modify the membranes can increase filtration performance and improve mechanical properties.The combination of hydrophilic inorganic materials in the membrane process can enhance performance because these particles increase the porous and stability of the membrane [8].Various nanoparticles such as titanium dioxide [9], graphene oxide [10], and silica [11] have been employed as fillers in the polymeric matrix.
Halloysite nanotube (HNT) is naturally clay material composed of multilayer aluminosilicate as a tubular shape with a chemical formula of Al2Si2O5(OH)4•nH2O.HNTs have a onedimensional geometry, abundant nature, unique structure, high surface area, and environmentally friendly and reactive functional groups.HNT was employed in the synthesis of various separation processes like Nanofiltration (NF) [9], Ultrafiltration (UF) [12], and Forward Osmosis (FO) [13] membranes.The presence of hydrophilic siloxane groups (Si-O-Si), and Al-OH groups which are on the exterior layer of HNTs, enable functionalization and modification of HNT.Low dispersibility and weak interfacial interaction between HNTs and polymers are some drawbacks that limit its application.However, aggregation and leaching out of NP in casting solution can be decreased by modification of the surface of nanoparticles [12].Zeng et al [13] synthesized GObased composite membranes and fabricated halloysite nanotubes with dopamine (D-HNTs) by vacuum filtration for oils and dye removal.They found the incorporation of D-HNTs improves both wettability and water flux and, structure of GO.HNTs with grafted amino groups significantly enhanced the membrane's properties.Hebbar et al [14], prepared polyetherimide (PEI) membranes by immersion precipitation method.They used polydopamine-modified halloysite (DHNTs) nanotube as a hydrophilic additive in a mixture solution.The results indicated that introducing the PEI-DHNTs improves water flux, antifouling, and the filtration efficiency of the membrane.PAMAM-dendrimers are a hyperbranched three-dimensional and symmetrical polymer with a remarkable number of amidoamine terminal functional groups, [15] that are synthesized in two repeating Michael addition and amidation reaction.Owing to abundant functional groups and large specific surface area, can be considered as a desirable candidate to be grafted onto the surface of HNTs nanotube.The further amino and amide groups could be as sacrificial agents for enhancing salt rejection and permeation flux.
The purpose of this work is the fabrication of the HNT nanotube with poly amidoamine dendrimer third generation and employed as a hydrophilic filler to amend the CA membrane.The membrane is mixed with modifying agents to prevent the leaching out of NP.The effect of the embedding amide and amino group on the performance of the membrane was investigated by fabricating HNTs.HNT/PAMAM membranes with different amounts of filler (0.5, 1, 2 wt.%) were synthesized by phase inversion technique.The influence of the addition of nanocomposite on the performance of the membrane was examined as to antifouling, flux, and hydrophilicity properties.All the prepared membrane was characterized by FE-SEM, FT-IR, AFM, TGA, zeta potential, and contact angle.The membrane efficiency was determined by Na2SO4 and MgCl2 rejection.

Functionalization of HNTs
PAMAM dendrimer (G3) was synthesized as mentioned in our earlier study [16].The HNTs were modified as stated by the presented study with slight modifications [17].In brief, 0.2 g of HNTs was added to an aqueous solution of NaOH (0.3 M) and stirred for 60 min.Next, the mixture was centrifuged, the solution was removed, and the residue solid was dispersed in deionized water (DIW) to remove NaOH residue.HNTs-Na product was dried in a vacuum oven, overnight and was utilized for further modification with PAMAM dendrimer using ECH.The prepared HNT-Na was added in 5.8 mL DMF and stirred at 90 °C.Next, 0.3 g of ECH was transferred through a syringe and stirred for 1 h.After that, PAMAM/DMF solution was added drop by drop into the mixture and stirred for 30 min.The obtained product was centrifuged and rinsed with a mixture of DMF and ethanol/water (1:1 v/v) to remove impurities.Next, it was placed in a vacuum oven and denoted as HNT-PAMAM (HNP).

Preparation of the Membrane
The embedding HNT-PAMAM membranes were synthesized by the phase inversion process as reported in the literature [18].The specified value of HNP (0.5, 1.0, 2.0 wt % of CA content) was added to DMF and exposed to ultra-sonication to obtain homogenous dispersion.Next CA powder (15.0 wt%) was added to the suspension and continued stirring at 60 ℃ for ∼ 5 h.The resultant mixture was cast on a glass plate with a casting knife and adjusted to the thickness of 100 μm.Then the film was immersed in a coagulation bath (DIW) at a standard temperature of 1 h.Finally, the prepared film was labeled concerning the amount of HNP introducing compositions as CA/ HNP 0.5, CA/HNP 1.0, and CA/HNP 2.0.(Table 1)

Membrane characterization
Fourier-transform infrared spectroscopy (FTIR) (Ettlingen, Germany) was utilized to study the chemical structure of the prepared nanoparticles (NP).Thermogravimetric analysis was used to characterize the thermal stability of the pristine and functionalized membrane by a TGA analyzer (TA Instruments Ltd, USA).A field emission scanning electron microscope (FE-SEM) model MIRA3 (Brno, Czech) was employed to investigate the top surface and cross-section morphology of the prepared membranes.Furthermore, mapping scanning spectra and energydispersive spectroscopy (EDS) demonstrate the uniform dispersion and presence of NP on the surface.Also, the surface roughness of the prepared membrane was determined by atomic force microscopy (AFM) model ICON (Brucker, America).The surface charge of the CA and modified CA was determined with the Electro Kinetic Analyzer Model SurPASS 3 (Anton Paar, Austria).Also, the hydrophilicity of the modified membranes was confirmed by measuring the water contact angle.The contact angle was measured by the sessile droplet method CAG-20 (Iran) dynamic contact angle analyzer.The modified film was tested in the dead-end separation cell.The efficient permeation area of the cell was about 10.17 cm 2 .MgCl2 and Na2SO4 solution with 2000 ppm concentration was used as feed solution and inside pressure of the cell at 5~8 bar was provided by nitrogen (N2) gas.Firstly, all the prepared membranes were packed down at 1 bar for ~30 min.The conductivity of the solution was measured by the (98129 Model, Hanna Co., Italy) conductivity meter.Water flux (PWF) (L.m −2 .h −1 , LMH) was determined by collecting a mass of permeating water (M, Kg) through the surface area (A, m 2 ) in a period (t, h).The value of flux and salt rejection (R%) were evaluated based on the following equations:

Membrane
Where CP and CF are concentrations of permeate and feed, respectively.

Membrane porosity measurement
Membrane porosity was calculated as explained by Zhang et al. [19].The dry weight (Wd) of the prepared membrane (1*1 cm 2 size) was measured before soaking in DI water overnight.Next, the wet membrane was weighted and noted (Ww).The porosity of all membranes (ɛ) was calculated as following equation (3): Here, ρ is the density of pure water (0.998 g/cm 3 ), surface area (cm 2 ), and thickness (cm 2 ) of the membrane.

Antifouling properties
The antifouling parameters of the modified membrane were determined by the rejection of albumin protein (BSA) solution as to flux recovery ratio (FRR).Incipiently, the film was subjected to a dead-end cell, and the flux of pure water (Jw1) was measured at 5 bars for 60 min.Then BSA solution (250 mg/L) was moved through the film for 1h (Jp).After filtration, the membrane was cleaned and immersed in DI for 20 min to remove the remaining adhered BSA.Pure water was moved through the cleaned membrane and noted as (Jw2).Eventually, the FRR (%,), reversible fouling ratio (Rr), total fouling ratio (Rt), and irreversible fouling ratio (Rir) were measured based on the equations below (4-7):

Characterization of HNP membranes
The FT-IR spectra of HNT, PAMAM, and HNT-PAMAM were characterized in Fig. 1ac.Based on the data, a characteristic peak at 1027 cm -1 corresponded to the C-O vibration, and the peaks at 1357-1562 cm -1 attributed to the stretching C-H, N-H/ C-N, C=O bonds were found in the spectra of PAMAM (Fig. 1a).The spectra of HNT and functionalized HNT with PAMAM (Fig. 1 b, c) exhibited peaks at 3300-3886 cm −1 attributed to the stretching vibration of the NH and OH groups.The peaks related to the bending vibration of the Al-OH and stretching vibration of Si-O bands can be observed at 912 and 1034 cm -1 respectively.Furthermore, the appearance of stretching carbonyl and C-N group at 1678 cm -1 and N-H bending vibration band at 1590 cm-1 of the PAMAM confirmed the surface modification of the HNTs with PAMAM [2,17].hydrophilicity of the membranes, owing to that the functional groups of the HNT changed to amine groups and enhancing hydrophilicity, embedding HNT-PAMAM into the casting solution reduced the average contact angle for HNT-PAMAM (1w%) compared with pristine CA membrane from 51.4° to 32.5 °.These results confirmed that during phase inversion NP migrated toward the surface.Hence, the modified membrane has surface hydrophilicity properties with lower WCA values [20,21].The addition of NP in casting polymer solution had a considerable influence on the structure and morphology of the membrane and consequently the flux and rejection performance.To investigate the influence, FE-SEM images of the surface and cross-section of bare CA and modified membranes are illustrated in Figs. 3. Based on the surface membrane images (Fig. 3. a, c, e, g, i), alteration in morphology could be seen after loading of HNT and HNP nanoparticles.The halloysite with tubular structure distributed on the membrane surface.As shown, by enhancing the concentration of HNP from 0.5 to 2 wt% the membrane surface became smoother due to the homogenous distribution of the HNP, especially for membrane mixed with 1 wt% NP.The cross-section of the all-grafted CA membranes (Fig. 3 b, d, f, h, j) presented a uniform structure.In contrast with the pure membrane, the modified membranes with HNP, containing hydrophilic groups show finger-like and symmetric macrovoides which leads to enhanced fouling performance of the membrane.Zeta potential measurements of the bare CA and CA incorporated with HNP were measured at pH=6 and tabulated in Table 2.According to the obtained data, the hydroxyl groups are the cause i) j) of the negative surface of the CA membrane.Introducing the HNP, due to containing terminal amine group caused the zeta potential to become more negative from -0.773 mV to -13.27 mV.These results indicated the successful functionalization of HNT by PAMAM and confirmed FTIR analysis results.To investigate the percentage of the organic content of the modified membrane, TGA analyses were carried out and the thermogram is presented in Fig. 5.As can be observed, the weight loss of the bare CA at 257.52℃ corresponds to the degradation of CA, for the HNP membrane, the first weight loss occurred at 114.84℃ attributed to the water molecules which is adsorbed in HNP structure [22].These results approve the hydrophilic structure of the functionalized membrane.The AFM images and the value of average roughness (Ra) and root-mean-square roughness (Rq) of pristine and modified membranes are presented in Fig. 6 a, b, and Table 3.The membrane incorporated with HNP revealed higher surface roughness (74.73 nm) compared with the neat CA (60.06 nm) membrane.This promotion indicated the exchange rate between solvent and nonsolvent was accelerated during the phase inversion which is dependent on the hydrophilicity of HNP.Furthermore, this HNP NP increases surface area and antifouling properties [23].Rejection behavior depends on the membrane porosity which determines membrane performance.The amount of the total porosity is shown in Fig. 7.As can be seen, loading of NP changes porosity properties so that modified membrane with 0.5, 1.00, and 2.00 wt% exhibited higher porosity (32.45, 82.27 and 83.26) respectively, compared with the neat CA (31%) membrane.Systematically, increasing the loading of HNP enhanced the membrane porosity.Increasing the concentration of hydrophilic HNP due to the abundant structure of HNP formed pathway for the transport of water also facilitates the exchange of solvent/non-solvent during the phase separation process.The results of anti-fouling properties including FRR, Rr, Rir, and Rt of the membrane were presented in Table 4.The FRR for neat CA was 20%.After loading hydrophilic NP (0.50, 1.0 wt%) into matrix solution, the FRR value reached (81.39%), (85%) respectively, and slightly decreased to (80%) for (2.0 wt%).These results revealed that the loading of NP (1.0 wt%) exhibited better flux recovery after BSA filtration, owing to their hydrophilic groups which showed more anti-fouling ability.These results indicated that NP was adsorbed or stuck on the surface or inside the pores leading to hard removal by hydraulic cleaning.Hence, reversible membrane fouling could happen.Pristine CA showed the highest Rir (80%) along with Rr (30%) and Rt (50%) due to its hydrophobic surface that can influence the membrane flux while these parameters are for the membrane with loading (0.5, 1, 2 wt%), (18.60%, 16.27 % and 34.88%), (15%, 45% and 60%), (20%, 40% and 60%), respectively.

Membrane Separation Performance
The performance of the fabricated membrane on the salt rejection and flux was investigated by filtration Na2SO4 and MgCl2 solution (Fig. 8).The salt rejection of HNP 1.0 wt% for Na2SO4 (91%) and MgCl2 (76.5%) was higher than rejection for HNP 0.5 and HNP 2.0 wt% (64.13% and 33.445%) and (51.12% and 65.45%) respectively.The CA membrane with loading 1 wt% HNP exhibited high rejection and flux due to the high hydrophilicity nature of HNP that creates a water path comparison to the control CA. when the amount of NP loading was increased, the viscosity of the polymer mixture increased and favored the formation of less porous structures during the phase inversion.Therefore, the flux had no obvious change with a significant increasing trend.Overall, tubular morphology, porosity structures, and the presence of a hydrophilicity functional group of PAMAM as nanofiller in the membrane leads to the formation of a finger-like structure that facilitates water transport [10].Also, the salt rejection and water flux performance of the CA/HNP 1.0wt% membranes on the rejection of Na2SO4 and MgCl2 were evaluated in Fig. 9.The desalination order follows R (Na2SO4) > R (MgCl2).This phenomenon can be justified by the Donnan effect.The presence of the amine terminal groups on the surface of the functionalized HNT creates a negative surface membrane, as shown in the zeta potential result (Table .2).Hence, this salt rejection order revealed the high tendency of the NF membrane for the rejection of the multivalent anion SO4 2- with hydrated radius 3.79 Å compared to monovalent anion Cl -with hydrated radius 3.32Å.

Conclusion
In this study, Membrane with different amounts (0.5, 1, 2 wt%) of hydrophilic nanocomposite containing PAMAM dendrimer (G3) as a modifying agent was fabricated by phase inversion technique to investigate the influence of introducing HNP on the desalination and flux properties.The presence of negative charge groups such as amino, carboxyl, hydroxyl, and ether groups improved desalination performance.The morphology, thermal stability, and surface properties of the functionalized membranes were evaluated.Based on the result, increasing the amount of HNP in the matrix solution improves surface hydrophilicity and porosity characteristics.Performance and anti-fouling parameters of functionalized membrane compared with the bare CA membrane were determined through filtration of salt and BSA by dead-end cell.Achieved results indicate that the mixed matrix membrane exhibited better performance compared to the CA membrane, which is consistent with the contact angle results.The contact angle for the pristine CA membrane decreased from (51.4°) to (32.5°) compared with CA membranes incorporated with 1.0 wt% HNP.Comparing other fabricated membranes, CA/HNP 1.0 wt% membrane revealed favorable rejection performance for Na2SO4 (91%) and MgCl2 (75.6%) and indicated smaller Rir (15%), Rt (60%) and Rr (45%) fouling ratios and acceptable porosity (82.27%) of all other modified membranes.

Fig. 2 .
Fig. 2. Water contact angles of the pure CA (a) and modified (b) membranes.

Fig. 5 .
Fig.5.Thermal stability of the bare and modified CA membrane

Fig. 7 .
Fig. 7. Membrane porosity of CA and modified membrane

Fig. 8 .
Fig. 8.The effect of NP wt% loading on salt rejection and flux performance

4 Fig. 9 .
Fig.9.The effect of the modified membrane surface charge on the salt rejection performance

Figure 4 .Figure 5 .
Figure 4.The EDS and elemental mapping of CA/HNP 1.0 wt% membrane Figure 5. Thermal stability of the bare and modified CA membrane

Figure 6 .Figure 7 .Figure 8 Figure 9 .
Figure 6.Surface AFM micrographs of CA(a), and CA/HNP 1.00 membrane (b) Figure 7. Membrane porosity of CA and modified membrane Figure 8 The effect of NP wt% loading on salt rejection and flux performance Figure 9.The effect of the modified membrane surface charge on the desalination performance

Table 2 .
Zeta potential results of bare HNT and functionalized HNT.

Table 3 .
The surface roughness parameters of the bare and fabricated membranes

Table 4 .
Antifouling Parameters of the modified membranes